# ALTERNATIVE THERAPEUTICS AGAINST ANTIMICROBIAL-RESISTANT PATHOGENS

EDITED BY : Rebecca Thombre, Kamlesh Jangid, Ravi Shukla and Noton Kumar Dutta PUBLISHED IN : Frontiers in Microbiology

### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-216-9 DOI 10.3389/978-2-88963-216-9

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# ALTERNATIVE THERAPEUTICS AGAINST ANTIMICROBIAL-RESISTANT PATHOGENS

Topic Editors: Rebecca Thombre, University of Kent, United Kingdom Kamlesh Jangid, National Centre for Cell Science Pune, India Ravi Shukla, RMIT University, Australia Noton Kumar Dutta, Johns Hopkins University Baltimore, United States

Citation: Thombre, R., Jangid, K., Shukla, R., Dutta, N. K., eds. (2019). Alternative Therapeutics Against Antimicrobial-Resistant Pathogens. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-216-9

# Table of Contents

*06 Editorial: Alternative Therapeutics Against Antimicrobial-Resistant Pathogens*

Rebecca Thombre, Kamlesh Jangid, Ravi Shukla and Noton Kumar Dutta


Samsher Singh, Nitin P. Kalia, Prashant Joshi, Ajay Kumar, Parduman R. Sharma, Ashok Kumar, Sandip B. Bharate and Inshad A. Khan

*28 Structure–Function Relationship of Aminopeptidase P From*  Pseudomonas aeruginosa

Cui-Ting Peng, Li Liu, Chang-Cheng Li, Li-Hui He, Tao Li, Ya-Lin Shen, Chao Gao, Ning-Yu Wang, Yong Xia, Yi-Bo Zhu, Ying-Jie Song, Qian Lei, Luo-Ting Yu and Rui Bao

*40 Repurposing and Revival of the Drugs: A New Approach to Combat the Drug Resistant Tuberculosis*

Divakar Sharma, Yogesh K. Dhuriya, Nirmala Deo and Deepa Bisht


Sanjay Chhibber, Jasjeet Kaur and Sandeep Kaur

*88 Antimicrobial Potential of Epiphytic Bacteria Associated With Seaweeds of Little Andaman, India*

Perumal Karthick and Raju Mohanraju


Mitali Mishra, Satish Kumar, Rakesh K. Majhi, Luna Goswami, Chandan Goswami and Harapriya Mohapatra

### *118 Efficacy of Colistin and its Combination With Rifampin* in Vitro *and in Experimental Models of Infection Caused by Carbapenemase-Producing Clinical Isolates of* Klebsiella pneumoniae

María E. Pachón-Ibáñez, Gema Labrador-Herrera, Tania Cebrero-Cangueiro, Caridad Díaz, Younes Smani, José P. del Palacio, Jesús Rodríguez-Baño, Alvaro Pascual, Jerónimo Pachón and M. Carmen Conejo

*127* Trans*-Cinnamaldehyde and Eugenol Increase* Acinetobacter baumannii *Sensitivity to Beta-Lactam Antibiotics*

Deepti P. Karumathil, Meera Surendran Nair, James Gaffney, Anup Kollanoor-Johny and Kumar Venkitanarayanan

*137 Advance in Research on* Mycobacterium tuberculosis *FabG4 and its Inhibitor*

Debajyoti Dutta

*143 Herring Oil and Omega Fatty Acids Inhibit* Staphylococcus aureus *Biofilm Formation and Virulence*

Yong-Guy Kim, Jin-Hyung Lee, Chaitany J. Raorane, Seong T. Oh, Jae G. Park and Jintae Lee

*153 Nano-Strategies to Fight Multidrug Resistant Bacteria—"A Battle of the Titans"*

Pedro V. Baptista, Matthew P. McCusker, Andreia Carvalho, Daniela A. Ferreira, Niamh M. Mohan, Marta Martins and Alexandra R. Fernandes

*179* In Vitro *Antimicrobial Activity of Green Synthesized Silver Nanoparticles Against Selected Gram-negative Foodborne Pathogens* Yuet Ying Loo, Yaya Rukayadi, Mahmud-Ab-Rashid Nor-Khaizura, Chee Hao Kuan, Buong Woei Chieng, Mitsuaki Nishibuchi and Son Radu

*186 Prevention of Dermal Abscess Formation Caused by* Staphylococcus aureus *Using Phage JD007 in Nude Mice*

Bingyu Ding, Qingtian Li, Mingquan Guo, Ke Dong, Yan Zhang, Xiaokui Guo, Qingzhong Liu, Li Li and Zelin Cui

*193 Non-toxigenic* Clostridioides *(Formerly Clostridium)* difficile *for Prevention of* C. difficile *Infection: From Bench to Bedside Back to Bench and Back to Bedside*

Dale N. Gerding, Susan P. Sambol and Stuart Johnson


Min Lu, Tianhong Dai, Clinton K. Murray and Mei X. Wu

*243 Effects of Monolaurin on Oral Microbe–Host Transcriptome and Metabolome*

Viviam de Oliveira Silva, Luciano José Pereira, Silvana Pasetto, Maike Paulino da Silva, Jered Cope Meyers and Ramiro Mendonça Murata *252 Purification and Characterization of an Active Principle, Lawsone, Responsible for the Plasmid Curing Activity of* Plumbago zeylanica *Root Extracts*

Rajashree Bhalchandra Patwardhan, Prashant Kamalakar Dhakephalkar, Balu Ananda Chopade, Dilip D. Dhavale and Ramesh R. Bhonde

*262 Inhibiting Bacterial Drug Efflux Pumps via Phyto-Therapeutics to Combat Threatening Antimicrobial Resistance*

Varsha Shriram, Tushar Khare, Rohit Bhagwat, Ravi Shukla and Vinay Kumar

*280 Emerging Strategies to Combat ESKAPE Pathogens in the Era of Antimicrobial Resistance: A Review*

Mansura S. Mulani, Ekta E. Kamble, Shital N. Kumkar, Madhumita S. Tawre and Karishma R. Pardesi

# Editorial: Alternative Therapeutics Against Antimicrobial-Resistant Pathogens

### Rebecca Thombre1,2 \*, Kamlesh Jangid<sup>3</sup> , Ravi Shukla<sup>4</sup> and Noton Kumar Dutta<sup>5</sup>

*<sup>1</sup> Department of Biotechnology, Modern College of Arts, Science and Commerce, Pune, India, <sup>2</sup> School of Physical Sciences, University of Kent, Canterbury, United Kingdom, <sup>3</sup> National Centre for Microbial Resource, National Centre for Cell Science, Savitribai Phule Pune University Campus, Pune, India, <sup>4</sup> NanoBiotechnology Research Laboratory, RMIT University, Melbourne, VIC, Australia, <sup>5</sup> Center for Tuberculosis Research, Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, United States*

Keywords: antimicrobial resistance (AMR), multidrug-resistance (MDR), nanomaterials, nanoparticles, plant based compounds, antimicrobial agents, alternative therapy, novel compounds

**Editorial on the Research Topic**

### **Alternative Therapeutics Against Antimicrobial-Resistant Pathogens**

### Edited by:

*Henrietta Venter, University of South Australia, Australia*

Reviewed by: *Jianhua Wang, Feed Research Institute (CAS), China*

> \*Correspondence: *Rebecca Thombre rebecca.thombre@gmail.com*

### Specialty section:

*This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology*

Received: *26 July 2019* Accepted: *05 September 2019* Published: *19 September 2019*

### Citation:

*Thombre R, Jangid K, Shukla R and Dutta NK (2019) Editorial: Alternative Therapeutics Against Antimicrobial-Resistant Pathogens. Front. Microbiol. 10:2173. doi: 10.3389/fmicb.2019.02173* Antimicrobial resistance (AMR) has emerged as one of the greatest global challenge to public health in the twenty-first century. The use of antibiotics is associated with the serendipitous discovery of Penicillin by Sir Alexander Fleming in 1928 (Fleming, 1929). However, Emmerich and Löw had demonstrated the first application of the antibiotic "Pyocyanase" in hospital in 1899. The golden era of antibiotic was ushered between 1950s and 1970s which was marked by rapid developments in discoveries of many classes and types of antibiotics (Emmerich and Löw, 1899). The increased use and abuse of antibiotics caused the emergence of multidrug resistant bacteria that caused hard-to-treat infections (Aminov, 2010). It is imperative to search alternative therapeutics and strategies to combat AMR and diminish the exacerbated use of antibiotics. In this special issue, we present 26 articles that highlight the use of novel peptides, phage-based therapies, nanomedicine, contemporary and alternative medicines, plant (herbal), and bacteria based antimicrobials as potential alternatives to combat multidrug resistant (MDR) bacteria. The articles are categorized in different groups, including (i) Antimicrobial nanoparticles against drug resistant bacteria, (ii) Bacteriophages: A promising approach to fight MDR, (iii) Anti biofilm agents, (iv) Antimicrobial peptides, (v) Efflux pump inhibitors, and (vi) Host /Pathogen directed therapies.

A variety of repurposing FDA-approved drugs (Sharma et al.) that are employed in the management of pathological conditions of non-infectious etiology have been shown to exhibit broad spectrum antimicrobial activity in vitro and in vivo. Such compounds including marine eukaryotes like seaweeds (Karthick and Mohanraju), phytochemicals (Kim et al.; Lu et al.), antimicrobial peptides and proteins (Kumar et al.), termed "non-antibiotics" (Dutta et al., 2007; Mazumdar et al., 2009, 2010), possess antibacterial properties, acting through mechanisms different from those of existing drugs, by enhancement of combination-therapy effective (Pachon-Ibanez et al.; Shriram et al.), reversal of drug resistance (Guo et al.; Patwardhan et al.) or re-sensitizing activities (Dutta et al., 2014; Shriram et al.), inhibition of biofilm formation (Guo et al.; Karumathil et al.; Kaur et al.; Khalifa et al.; Kim et al.; Lu et al.; Punjabi et al.), as well as by induction and control of efflux pumps (Dutta et al., 2010; Baptista et al.; Karumathil et al.; Lu et al.; Shriram et al.).

This broad group of antimicrobial agents has two subgroups, each with distinctly different adjunct activities, either pathogen-directed or host directed. The first group is that of the antimicrobial non-antibiotics—drugs that have direct antimicrobial activity and the proposed path for compounds targeting microbial factors. The second group can generally be classified into two categories: those that enhance the antimicrobial activity of the host immune system, and those which dampen the inflammatory response preventing tissue damage (Karumathil et al.; Singh and Subbian). Hostdirected therapies are attractive options as they are not prone to the resistance associated with antibiotics (Dutta et al., 2016; Frank et al., 2019). Currently, the following HDT agents are being evaluated in phase 2 clinical trials as adjuncts to rifabutin-modified standard therapy in adults with drug-sensitive, smear-positive pulmonary TB: (1) the mammalian target of rapamycin (mTOR) inhibitor, everolimus (0.5 mg), (2) auranofin (6 mg), (3) vitamin D3, and (4) the phosphodiesterase-4 (PDE4) inhibitor, CC-11050 (ClinicalTrials.gov Identifier: NCT02968927). A randomized clinical trial, Statins as Adjunctive Therapy for TB (StAT-TB), is currently underway to determine if pravastatin adjunctive therapy shortens the median time to sputum-culture negativity and improves lung function outcomes among HIV-infected and uninfected patients with drug-susceptible pulmonary TB (NCT03456102).

In summary, these articles cover a vast expanse of research findings based on emerging trends in combatting antimicrobial resistance using traditional and natural antimicrobials, plant and microbial derivatives and nanomaterials. Currently, AMR is a

### REFERENCES


constantly growing global threat to public health worldwide and has been declared as a thrust area by World Health Organization (WHO). AMR is mediated via various mechanisms such as enzymatic degradation of drugs, alteration of antimicrobial targets, efflux of drugs, alteration of microbial membrane permeability, formation of biofilms, and persister cell states. Most of the AMR resistance genes (ARG) are disseminated via horizontal gene transfer mediated by genetic elements like plasmids, transposons, bacteriophages, and other genetic elements (Thombre et al., 2016). One of the challenges of AMR is annihilating the spread and prevalence of ARGs in the environmental resistome via plasmids. Future strategies and new lines of research need to be undertaken using conjugation inhibitors, plasmid incompatibility systems, and CRISPR/Casbased approaches to tackle the incredibly profound multidrug resistant bacteria (Buckner et al., 2018).

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

We would like to thank the contributing authors for submission of their papers and reviewers for their valuable time. The Editorial guidance and suggestions of Dr. Rustam Aminov for handling this special issue is highly appreciated. RT thanks the Principal, Modern College, Shivajinagar, Pune, for providing necessary facilities.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Thombre, Jangid, Shukla and Dutta. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antimicrobial and Antibiofilm Potential of Acyclic Amines and Diamines against Multi-Drug Resistant Staphylococcus aureus

### Gurmeet Kaur, P. Balamurugan, Sahana Vasudevan, Saikiran Jadav and S. A. Princy\*

Quorum Sensing Laboratory, Centre for Research in Infectious Diseases, School of Chemical and Biotechnology, SASTRA University, Thanjavur, India

Multi-drug resistant Staphylococcus aureus (MDRSA) remains a great challenge despite a decade of research on antimicrobial compounds against their infections. In the present study, various acyclic amines and diamines were chemically synthesized and tested for their antimicrobial as well as antibiofilm activity against MDRSA. Among all the synthesized compounds, an acyclic diamine, (2,2<sup>0</sup> -((butane-1,4-diylbis(azanediyl)bis(methylene))diphenol) designated as ADM 3, showed better antimicrobial activity (minimum inhibitory concentration at 50 µg/mL) and antibiofilm activity (MBIC<sup>50</sup> at 5 µg/mL). In addition, ADM 3 was capable of reducing the virulence factors expression (anti-virulence). Confocal laser scanning microscope analysis of the in vitro tested urinary catheters showed biofilm reduction as well as bacterial killing by ADM 3. On the whole, our data suggest that acyclic diamines, especially ADM 3 can be a potent lead for the further studies in alternative therapeutic approaches.

Keywords: biofilm, Staphylococcus aureus, multi-drug resistance, antibiofilm, diamines, antibacterial

## INTRODUCTION

Staphylococcus aureus, a Gram-positive, facultative anaerobic cocci bacterium, is one of the most notorious pathogen, causing infections in humans. Their abilities to evade the host immune defense mechanism and resistance to the first and second generation antibiotics has made the pathogen a subject of interest in the scientific community (Fedtke et al., 2004). S. aureus is an opportunistic pathogen related to the various types of infectious diseases such as wound infections, catheterrelated bloodstream infections (CRBSI), musculoskeletal infections, toxic shock syndrome and, about 20% of population worldwide found to be the long-term carrier as a part of their normal flora (Kluytmans et al., 1997; Cole et al., 2001). Various factors associated with S. aureus such as virulence gene expressions, cell to cell signaling mechanism, inactivation of antibiotics, alteration in target sites, efflux pumps, and biofilm formation have led to the emergence of multi-drug resistant S. aureus (MDRSA) (Dinges et al., 2000; Becker et al., 2003; Zhu et al., 2008; Arya et al., 2011; Qayyum et al., 2016).

Planktonic microbes attach to a particular substratum and produce an anchoring polymer called as an extracellular polysaccharide (EPS) which leads to the formation of the multicellular microbial community known as biofilm (Flemming et al., 2007). The altered metabolic activity of the cells that are associated with biofilm formation have high rates of EPS production, activation of specific genes associated with biofilm formation and virulence, reduction in the growth rate than their planktonic

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Pankaj Kumar, Johns Hopkins School of Medicine, United States Divakar Sharma, Interdisciplinary Biotechnology Unit, Aligarh Muslim University, India

\*Correspondence:

S. A. Princy adlineprinzy@biotech.sastra.edu

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 12 July 2017 Accepted: 31 August 2017 Published: 15 September 2017

### Citation:

Kaur G, Balamurugan P, Vasudevan S, Jadav S and Princy SA (2017) Antimicrobial and Antibiofilm Potential of Acyclic Amines and Diamines against Multi-Drug Resistant Staphylococcus aureus. Front. Microbiol. 8:1767. doi: 10.3389/fmicb.2017.01767

**8**

counterparts (Flemming et al., 2007). Biofilm formation accounts for one of the major reasons for the emergence of multi-drug resistance (MDR) in various pathogenic microbes and is common in bacteria. In the biofilm mode of lifestyle, bacterial population acquires adaptation to tolerate and survive a wide range of diverse environmental stress such as scarce nutritional availability, antibiotics exposure, competition for survival in a multi-species environment (Parsek and Singh, 2003). The antibiotics tolerance ability of the biofilm cells complicates the treatment of various infections in humans, such as, cystic fibrosis, endocarditis, which includes biofilm formation on various biological implants such as, urinary catheters, heart catheters, various joint implants, and replacement of heart valves (Singh et al., 2000). Biofilms pose a threat to the human race because of their persistent nature and plays a major role in certain pathogenic infections (Singh et al., 2000; Ravichandiran et al., 2012; Arya and Princy, 2013; Otto, 2013). Several incidences have been reported with MDRSA strain infections such as CRBSI which is primarily due to either bacterial colonization on a device that may be intraluminal, i.e., formation of biofilm inside the lumen and causing persistent infection (Safdar and Maki, 2004). Similarly, musculoskeletal infections, wound infections, and nosocomial infections have also been reported by MDRSA which are very difficult to treat with the existing drugs (Bascones-Martinez et al., 2011; Abad et al., 2014; Vincze et al., 2014). Antibiofilm compounds are one such alternative option in the recent research focus. These small chemical ligands can independently inhibit bacterial biofilm or disrupt the biofilm matrix at the molecular level via, disturbing their metabolic pathways.

Medicinally important chemical compounds play a crucial role in treating various diseases which pose a threat to human survival at times. From our earlier reports, we have understood that acyclic amines could be promising leads against S. aureus, as one of our compounds (SarABI-12, 2-[(methylamino)methyl]phenol) was shown to have target specific interaction with the staphylococcal accessory regulator, SarA (Arya and Princy, 2013). In addition, our in vitro studies (Balamurugan et al., 2017) have confirmed the antibiofilm and anti-virulence properties against clinical S. aureus strains. The data also showed a significant reduction in the expression of SarA regulated virulence genes like fnbA, hla, and hld. It is also interesting to note that several researchers have reported the potential of cyclic amines and diamines to inhibit MDRSA, Enterococcus sp., Clostridium difficile, Escherichia coli, Aspergillus oryzae, Aspergillus niger (He et al., 2003; Abdel-Rahman et al., 2004; Hensler et al., 2006). The focus on the use of cyclic amines and diamines for antibacterial, antifungal and anti-proliferative activity has also been implicated (He et al., 2003; Juranic, ´ 2014; Kazakova et al., 2014). Previous reports by Chtchigrovsky et al. (2013) shows the anti-proliferative activity of trans-Nheterocyclic carbene–amine–Pt(II) complexes. These complexes have the ability to bind to DNA, leading to intrastrand crosslinks between two adjacent guanines and minor interstrand cross links, resulting in DNA damage and cell death (Chtchigrovsky et al., 2013). Similarly, Subík et al. (1977) report the antibacterial and antifungal activity of amine oxides due to their ability to disorganize the cellular membranes leading to cidal effects.

It is expected that similar to cyclic amines, acyclic amines can also have medicinally important biological potentials, supported by our previous works (Arya and Princy, 2013; Balamurugan et al., 2017). Hence, we have investigated the antibiofilm and antibacterial properties of acyclic amines which have not yet been explored against pathogenic bacteria.

## MATERIALS AND METHODS

### Synthesis of Acyclic Amine and Diamine Compounds

The acyclic amines (AAM 1–5) and acyclic diamines (ADM 1–8) compounds were synthesized by a reductive amination process (Ramachandran et al., 2010). The synthesis process of acyclic amine and diamines is shown in the **Figures 1A,B**. For the synthesis of acyclic amines, a solution of the suitable aryl aldehyde (10 mmol) and the suitable acyclic amine (12 mmol) in methanol (5 mL) was stirred at room temperature for 1 h. The reactants

were shown in Supplementary Table S1. The intermediate imines were reduced to the corresponding acyclic amines and diamines by the addition of sodium borohydride (15 mmol) at the icecold condition. After stirring, the reaction mixture was incubated at room temperature overnight and further diluted with water and extracted thrice with dichloromethane. The mixed organic phases were dried with anhydrous sodium sulfate, evaporated to dryness under reduced pressure to obtain the acyclic amines and diamines. The structures of the synthesized products were confirmed by <sup>1</sup>H-NMR spectroscopy.

### Bacterial Strains and Growth Conditions

Clinical isolates of S. aureus were received from the rejected urinary catheters, JSS Medical College Mysore with prior approval from the Institutional Ethical Committee (IEC/JSSMC/PG/1050/2015-16). The strains were initially subjected to screening for MDR against the antibiotics: levofloxacin, ampicillin, ciprofloxacin, and methicillin. The S. aureus strain (MDRSA - QSL2040) that showed resistance to methicillin as well as other tested antibiotics (**Table 1**) was used as a test strain and the S. aureus (ATCC 25923) was used as reference strain (Kumar et al., 2013; Khatkar et al., 2014). These strains were cultured in the cation-adjusted Mueller–Hinton broth (CAMHB) for the minimal drug dose response against the growth and bactericidal actions. For the biofilm assays, the tryptic soy broth (TSB) supplemented with 3% NaCl and 0.5% glucose was used as the biofilm medium.

### Evaluation of Acyclic Amine and Diamine Compounds for MIC, MBC, and MBIC

The minimum inhibitory concentration (MIC), the minimum bactericidal concentration (MBC), and the minimum biofilm inhibitory concentrations (MBIC) of the acyclic amines and diamines were determined according to the Clinical and Laboratory Standards Institute (CLSI) with slight modifications. The S. aureus clinical isolate (MDRSA - QSL2040) was grown overnight in TSB. The overnight culture was diluted 1:100 in



Sensitivity/resistance pattern was concluded as per the CLSI guidelines.

physiological saline and the bacterial suspensions were adjusted to a final inoculum size of 1 × 10<sup>6</sup> cells (Issam et al., 2015). The adjusted inoculum was added to the wells of a 96-well microtiter plate containing twofold serial dilutions of the compound (varying concentrations from 2 to 400 µg/mL as individual sets keeping different high concentrations) in CAMHB medium. The plates were incubated at 37◦C for 24 h. The setup was carried out as independent experiments (in triplicates) to calculate the effective concentration.

The optical density (OD600) was read immediately after inoculation and again after 18 h of incubation at 37◦C, in a microtitre plate reader (iMark, Bio-Rad, Japan). The lowest concentration that inhibited growth when compared to the untreated control culture was taken as the MIC. Similarly, MBC was determined by counting the number of colonies (CFU/mL) after 24 h of incubation at 37◦C and was defined as the lowest concentration at which the viable cells were reduced to a level by ≥90% in comparison to the untreated control cultures. All independent assays were carried out in triplicates.

For biofilm assay, diluted overnight culture (1:100) was inoculated in TSB and allowed to grow till it reached exponential growth phase (5 h). The culture was tested for 0.5 McFarland unit and followed by inoculation along with varying concentrations of the compounds (2–400 µg/mL) in 96-well microtitre plates. Plates were incubated for 24 h at 37◦C without shaking. After incubation, the wells were washed twice with 200 µL of phosphate-buffered saline (PBS) gently to remove the nonadherent cells. Adherent cells in the biofilm were fixed by adding 200 µL of 100% methanol prior to staining with 200 µL of 0.2% (w/v) crystal violet (CV) for 20 min. The excess stain was washed twice with PBS and the plates were air dried. The bound CV in the air dried plates were eluted with 200 µL of 33% acetic acid. The biofilm was quantitatively determined by measuring the absorbance at OD595 nm in a microtitre plate reader (iMark, Bio-Rad, Japan). The concentration at which the formation of biofilm is inhibited ≥50% when compared to the untreated control culture is defined as MBIC<sup>50</sup> and ≥90% is MBIC90. All the assays were carried out in triplicates.

### Anti-virulence Assays

Among the synthesized compounds, ADM 3 was selected as it showed higher antibiofilm activity at minimal concentration. The compound, ADM 3 was tested against the clinical isolate of S. aureus with two sub-MBIC<sup>50</sup> concentrations (i.e., <sup>1</sup>/4× MBIC50, <sup>1</sup>/2× MBIC50) and at a higher MBIC<sup>50</sup> concentration (2× MBIC50). S. aureus reference strain, untreated with ADM 3 was kept as control. After 24 h, the cultures were centrifuged at 6000 rpm for 10 min at 4◦C and filter sterilized in a 0.22 µm filter paper to collect the cell free supernatant. The collected supernatant was stored at 4◦C and further used to quantify hemolysin and protease (Pietrow et al., 2013) among the various secreted exotoxins by S. aureus.

### Hemolytic Assay

Hemolysin in the culture supernatant was quantified according to the procedure described earlier with slight modifications

(Cheung and Otto, 2012). Briefly, 10 mL of sheep blood was centrifuged at 2400 rpm for 5 min and the pellet obtained was washed twice with 10 mL of PBS. Ten microliters of this erythrocyte suspension was incubated with the cell free supernatant for 1 h at 37◦C. Finally, the incubated sample was centrifuged at 2400 rpm for 5 min. Erythrocyte suspension treated with 1% Triton X-100 was used as a positive control. The optical density of the supernatant was read at 540 nm. Water along with the erythrocyte suspension was considered as blank. The percentage (%) hemolysis was calculated using the following formula:

%Hemolysis = absorbance (sample) − absorbance (blank)/ absorbance (positive control)

### Proteolysis Assay

Quantitative estimation of the protease was carried out with azocasein assay. A total of 200 µL of the cell free supernatant was incubated with 800 µL of azocasein for 30 min at 37◦C. To this, 1200 µL of 1% trichloroacetic acid was added to arrest the enzymatic reaction. The contents were incubated on ice for 30 min and centrifuged at 12,000 rpm for 5 min. To 1600 µL of the supernatant 400 µL of 1.8 N NaOH was added and the optical density was read at 420 nm against the blank (azocasein + TCA + NaOH). The amount of enzyme required to digest 1 mg of azocasein per minute is known as 1 unit of protein activity (Pietrow et al., 2013). The percentage (%) proteolysis was calculated using the following formula:

%Proteolysis = absorbance (sample) − absorbance (blank)/ absorbance (control)

### Appraisal of the Therapeutic Challenge of ADM 3 on S. aureus Using an In Vitro Catheter Model

An in vitro catheter model was used to evaluate the effect of ADM 3 on biofilm formation in hydrodynamic conditions (Hancock et al., 2010). Silicone catheter segments of 20 mm length were cut vertically into two halves, sterilized in 0.5% sodium hypochlorite solution followed by washing with sterilized water. The sterilized catheters were placed in a six-well microtitre plate and the surface was coated with human blood plasma by incubation at 37◦C for 24 h. Subsequently, the plasma was removed from the wells and the coated catheters were subjected for the establishment of S. aureus infection followed by treatment with ADM 3 for 7 days with concentrations of <sup>1</sup>/4× MBIC50, <sup>1</sup>/2× MBIC50, MBIC50, and 2× MBIC50. The plates were incubated at 37◦C for 24 h with shaking at 120 rpm. After 24 h of incubation, the medium was changed along with a fresh dosage of the drug each time and this was repeated for 7 days. Also, at every 24 h interval, the catheters from the wells were removed and analyzed for the viability of cells (Weiss et al., 2009). Briefly, the viability of cells from the catheters was processed by immersing the catheters individually into sterile PBS and followed by sonication. The adherent cells recovered in PBS were plated for colony count in TSB agar plates. Microscopy imaging was used to visually analyze the effect of ADM 3 on biofilm removal. The cells were stained with fluorescein isothiocyanate (5 mg/mL) and ethidium bromide (1.25 mg/mL) prepared by mixing 5 µL each of the dyes in 1 mL of cold 0.9% NaCl solution. The samples were incubated for 10 min, and then the excess dye was removed by washing with 0.9% NaCl. All the stained samples were imaged using Olympus FV 1000 confocal microscope with a 10× objective of numerical aperture 0.3. The stained samples were excited at 488 nm using Multi Argon LASER and images were collected from four randomly chosen spots from the sample surface.

## Cytotoxicity Analysis of ADM 3 on HEp-2 Cells

Cell viability was assessed by MTT assay. The reduction of MTT is catalyzed by mitochondrial dehydrogenase enzymes and thus considered to be a measure of cell viability. HEp-2 cells were seeded to each well in a 96-well microtiter plate (1 × 10<sup>5</sup> cells), in 100 µL of DMEM growth medium and the plate was incubated at 37◦C for 24 h. After incubation, the fresh growth medium was replaced in the wells. ADM 3 was diluted from an original stock to obtain a concentration of 200 µg/mL. The HEp-2 cells were incubated with ADM 3 and allowed to adhere for a period of 72 h. A total of 10 µL of MTT solution (5 mg/mL in 1× PBS) was added to each well and the plate was incubated in dark for 4 h at 37◦C. Further, the content of the each well was removed and the formazan crystals were dissolved in 200 µL of dimethyl sulfoxide (DMSO) solution per well. The absorbance was measured at 590 nm using ELISA plate reader. Only the interior rows of the microtiter plate were used for these experiments to minimize the variations in cell viability due to medium evaporation at the periphery site. The percentage cell viability was calculated with reference to the untreated control cells. The experiment was conducted twice in quadruplicates.

Percentage cell viability

$$=\frac{\text{OD of drug treated sample} - \text{OD of blank}}{\text{OD of control} - \text{OD of blank}} \times 100\%$$

### Statistical Analysis

Graph pad prism software (version 6.01) was used for statistical analysis. One-way ANOVA and multiple comparisons were done. The minimum level of significance was set at P ≤ 0.05. All the assays were conducted in triplicates and the results were expressed as mean ± SD.

## RESULTS

### Synthesis of Acyclic Amines and Diamines

The spectral data of all the synthesized compounds were in full agreement with the proposed structures (Supplementary Table S1). The <sup>1</sup>H-NMR data for the synthesized compounds are given below.

### **AAM 1: N-(naphthalene-2-ylmethyl)butan-1-amine**

Isolated yield = 78%,1H NMR (300 MHz, CDCl3) δ 8.12 (d, J = 8.3 Hz, 1H), 7.90–7.83 (m, 1H), 7.77 (d, J = 7.9 Hz, 1H), 7.59–7.38 (m, 4H), 4.24 (s, 2H), 2.81–2.71 (m, 2H), 1.61 (s, 1H), 1.52 (dd, J = 8.2, 6.1 Hz, 2H), 1.38 (dq, J = 14.1, 7.1 Hz, 2H), 0.93 (t, J = 7.3 Hz, 3H).

### **AAM 2: (N-benzylbutan-1-amine)**

fmicb-08-01767 September 13, 2017 Time: 15:45 # 5

Isolated yield = 55%, <sup>1</sup>H NMR (300 MHz, CDCl3) δ 7.38–7.19 (m, 5H), 3.79 (s, 2H), 2.63 (t, J = 7.2 Hz, 2H), 1.76 (s, 1H), 1.50 (dt, J = 14.4, 7.0 Hz, 2H), 1.40–1.24 (m, 2H), 0.91 (t, J = 7.2 Hz, 3H).

### **AAM 3: (N-benzylcyclohexanamine)**

Isolated yield = 71%, <sup>1</sup>H NMR (300 MHz, CDCl3) δ 7.36–7.22 (m, 5H), 3.81 (s, 2H), 2.53–2.44 (m, 1H), 1.31–1.06 (m, 11H).

### **AAM 4: 4(N-(4-methoxybenzyl)butan-1-amine)**

Isolated yield = 65%, <sup>1</sup>H NMR (300 MHz, CDCl3) δ 7.28 – 7.22 (m, 2H), 6.91 – 6.81 (m, 2H), 3.79 (s, 3H), 3.74 (s, 2H), 2.66 – 2.58 (m, 2H), 2.49 (s, 1H), 1.52 (ddd, J = 14.4, 8.3, 5.9 Hz, 2H), 1.34 (dd, J = 15.1, 7.4 Hz, 2H), 0.90 (t, J = 7.3 Hz, 3H).

### **AAM 5: (N-(4-chlorobenzyl)butan-1-amine)**

Isolated yield = 73%, <sup>1</sup>H NMR (300 MHz, CDCl3) δ 7.33–7.21 (m, 4H), 3.75 (s, 2H), 2.64–2.57 (m, 2H), 1.68 (s, 1H), 1.55–1.43 (m, 2H), 1.40–1.27 (m, 2H), 0.91 (t, J = 7.3 Hz, 3H).

### **ADM 1: 2,2**<sup>1</sup> **-((Ethane-1,2-diylbis(azanediyl)bis(methylene)) diphenol**

Yield 72%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 2.51 (s, 4H), 3.79 (s, 4H), 6.68–6.73 (m, 4H), 7.03–7.08 (m, 4H).

### **ADM 2: 2,2**<sup>0</sup> **-((Propane-1,3-diylbis(azanediyl)bis(methylene)) diphenol**

Yield 79%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 1.63 (quin, J = 6.9 Hz, 2H), 2.55 (t, J = 6.9 Hz, 4H), 3.80 (s, 4H), 6.67–6.72 (m, 4H), 7.03–7.08 (m, 4H).

### **ADM 3: 2,2**<sup>1</sup> **-((Butane-1,4-diylbis(azanediyl)bis(methylene)) diphenol**

Yield 69%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 1.47 (s, 4H), 2.50 (q, J = 1.8 Hz, 4H), 3.81 (s, 4H), 6.66–6.72 (m, 4H), 7.03–7.09 (m, 4H).

### **ADM 4: N**<sup>1</sup> **,N**<sup>2</sup> **-dibenzylethane-1,2-diamine**

Yield 82%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 2.51 (s, 4H), 3.66 (s, 4H), 7.20–7.32 (m, 10H).

### **ADM 5: N**<sup>1</sup> **,N**<sup>2</sup> **-dibenzylpropane-1, 3-diamine**

Yield 73%, <sup>1</sup>H-NMR (CDCl3, 300 MHz): δ 1.72 (quin, J = 6.9 Hz, 2H), 2.69 (t, J = 6.9 Hz, 4H), 3.77 (s, 4H), 7.24–7.32 (m, 10H).

### **ADM 6: N**<sup>1</sup> **,N**<sup>2</sup> **-bis(naphthalene-2-ylmethyl)ethane-1,2 diamine**

Yield 76%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 2.75 (s, 4H), 4.11 (s, 4H), 7.40–7.52 (m, 8H), 7.80 (d, J = 7.5 Hz, 2H), 7.89–7.92 (m, 2H), 8.13–8.17 (m, 2H).

TABLE 2 | Antimicrobial and antibiofilm values of acyclic amines and diamines.


### **ADM 7: N**<sup>1</sup> **,N**<sup>3</sup> **-bis(naphthalene-2-ylmethyl)propane-1,3 diamine**

Yield 78%, <sup>1</sup>H-NMR (CDCl3, 300 MHz): δ 1.61 (t, J = 6.3 Hz, 4H), 2.76 (t, J = 6.3 Hz, 4H), 4.22 (s, 4H), 7.41–7.55 (m, 8H), 7.76 (d, J = 7.5 Hz, 2H), 7.86 (d, J = 7.5 Hz, 2H), 8.10 (d, J = 7.8 Hz, 2H).

### **ADM 8: N**<sup>1</sup> **-(naphthalene-2-yl)–N**<sup>4</sup> **-(naphthalene-2-ylmethyl) butane-1,4-diamine**

Yield 80%, <sup>1</sup>H-NMR (DMSO-d6, 300 MHz): δ 1.83 (quin, J = 6.9 Hz, 2H), 2.83 (t, J = 6.9 Hz, 4H), 4.21 (s, 4H), 7.37– 7.52 (m, 8H), 7.76 (dd, J = 7.5, 1.8 Hz, 2H), 7.85–7.88 (m, 2H), 8.07–8.11 (m, 2H).

## Antimicrobial and Antibiofilm Activity of Acyclic Amines and Diamines

The MICs, MBCs, and MBICs of acyclic amines and diamines were shown in **Table 2**. In general, acyclic amines were found to have high MIC and MBC values but no antibiofilm activity. Alternatively, the acyclic diamine compounds, particularly ADM 3 was observed to exhibit good antimicrobial as well as antibiofilm activity. The concentration of antimicrobial and antibiofilm activity of ADM 3 was found to be 50 and 5 µg/mL (MBIC50) respectively. The biofilm inhibitory concentration for compound ADM 3 was found to be 10-fold lesser than the MIC. Thus, compound ADM 3 acts as a potential antibacterial as well as antibiofilm agent and was considered for further studies.

### Anti-virulence Assays

fmicb-08-01767 September 13, 2017 Time: 15:45 # 6

Percentage hemolysis and proteolysis by ADM 3 on S. aureus clinical isolate, MDRSA - QSL2040 are shown in **Figures 2**, **3** respectively. ADM 3 exhibited a similar effect on protease as well as on hemolysin production and a dose-dependent reduction was observed in both the cases. Clinical isolate (MDRSA - QSL2040) treated with ADM 3 showed a significant reduction in proteolytic and hemolytic activity when compared with the untreated control.

## In Vitro Catheter Model

Live dead staining of S. aureus in the presence and absence of ADM 3 revealed that, ADM 3 treatment induced significant loss of viability and colonization of biofilm on catheters as evidenced by the red fluorescence observed in treated samples at day 7. All the captured images were uniform and a representative image is shown in **Figure 4**. Thus, in consonance with our antibiofilm results in microtitre plate experiments, ADM 3 was effective in inhibition of S. aureus biofilm in catheters. The treatment of these catheters showed a decrease in the cell count from day 4 onward compared to untreated control (Supplementary Figure S1).

## Cytotoxicity Analysis on HEp-2 Cell Lines

Percentage cell viability of each experimental group was calculated with reference to the untreated control cells and it was observed that cell viability was higher than 85% in all drug doses (**Figure 5**). Statistical analysis was performed using the one way ANOVA and P ≤ 0.001 was considered significant.

FIGURE 4 | Confocal laser scanning microscope images of biofilm formed on urinary catheters. I—FITC labeled (green/viable cells); II—ethidium bromide labeled (red/non-viable cells); and III—superimposed images of I and II. (A) Control, untreated with ADM 3, (B) treated with ADM 3 (MBIC<sup>50</sup> at 5 µg/mL), and (C) treated with ADM 3 (2× MBIC<sup>50</sup> at 10 µg/mL).

## DISCUSSION

fmicb-08-01767 September 13, 2017 Time: 15:45 # 7

The key consequences of biofilm based infections are resistant to antibiotics as well as many other conventional antimicrobial agents. This eventually leads to the development of MDR pathogens with an extreme capacity for evading the host defense mechanisms. It is well known that bacterial biofilm exhibits antibiotic diffusion barrier (Cloete, 2003). In such cases, a large portion of the cells embedded in biofilms might escape from both antibiotic treatments as well as host's immune system responses. Such bacterial cells lead to the replenishment of biofilm at a new site causing are occurrence of infection with high resistance to antibiotics. Consequently, it becomes difficult to treat the infection using bactericidal drugs via systemic administration and demands the need for the evaluation of novel antibiofilm drugs. As an alternative to conventional antibiotic therapy and to control MDR, we hereby report a series of acyclic amines and diamines as potential drug candidates in the light of our earlier studies (Arya and Princy, 2013; Balamurugan et al., 2017). Among these, ADM 3 (2,2<sup>0</sup> -((butane-1,4-diylbis(azanediyl)bis(methylene))diphenol) showed a significant reduction in bacterial cell count as well as biofilm inhibition.

Targeting virulence factors in bacteria is an alternative approach to antimicrobial therapy that offers promising insights and opportunities to inhibit bacterial pathogenesis (Kauppi et al., 2003; Åberg and Almqvist, 2007). Certain virulence factors have been shown as potential leads for drug design and therapeutic intervention, whereas new possible insights are crucial for exploring others (Projan, 2002). Among the various virulence factors expressed in S. aureus, protease, as well as hemolysin production, plays a crucial role in the interaction of bacteria with the host cell for establishing an infection (Kupferwasser et al., 2003). In the present study, ADM 3 that showed effective MBIC<sup>50</sup> and MIC was also effective in reducing protease as well as hemolysin production.

In vitro catheter model showed decreased biofilm as well as increased red fluorescence in the confocal laser scanning microscope images at day 7. This suggests that ADM 3 has both antimicrobial and antibiofilm activities. In addition, decreased cell count observed after day 4 till the experimental study period also suggests the potential of ADM 3. This implies that the ADM 3 was proven to show a potent biofilm inhibiting agent as well as an antimicrobial agent. It is already shown through our in silico

### REFERENCES


studies that the hydroxyl and amine groups of the compound SarABI-12, interact with the oxygen atoms of E89 and R90 of SarA protein respectively, to form hydrogen bonds (Arya and Princy, 2013). Moreover, our in vitro studies also have confirmed the antibiofilm activities by the target specific interaction with SarA, which is a quorum regulator of S. aureus. Since ADM 3 is a derivative of SarABI-12, which carries both the hydroxyl and amine groups, it is speculated that the biofilm inhibition activity also could be of the similar mechanism.

Furthermore, a therapeutic molecule should not have cytotoxic effects if it has to be taken for further clinical trials. Generally, a compound is usually considered to have in vitro cytotoxicity if the particular concentration of the drug causes a 50% cell killing. Our data suggest that the compound ADM 3 may act as a potential drug which does not cause any cytotoxicity. To the best of our knowledge, our study is the first to report the antimicrobial and antibiofilm activity of acyclic amines and diamines. Our data also provide an insight that these compounds can act as a potential drug candidate to treat MDRSA. It would be interesting to explore the activity of these organic molecules at the molecular level to depict their mode of action involved in antibiofilm and antibacterial activity.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

We sincerely acknowledge the management and Dr. S. Swaminathan, Dean Sponsored Research of SASTRA University for their support and encouragement throughout this project. The authors are also grateful to the TRR in-house funding scheme of SASTRA University for the purchase of consumables.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.01767/full#supplementary-material


disease and diabetes-review of the Literature. Med. Oral Patol. Oral Cir. Bucal 16, e722–e729. doi: 10.4317/medoral.17032


evaluation of 4-(1-aryl-5-halo-2-oxo-1, 2-dihydro-indol-3-ylideneamino)- N-substituted benzene sulphonamides. Arab. J. Chem. 10, S2845–S2852. doi: 10.1016/j.arabjc.2013.11.009


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Kaur, Balamurugan, Vasudevan, Jadav and Princy. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fmicb-08-01767 September 13, 2017 Time: 15:45 # 8

# Boeravinone B, A Novel Dual Inhibitor of NorA Bacterial Efflux Pump of Staphylococcus aureus and Human P-Glycoprotein, Reduces the Biofilm Formation and Intracellular Invasion of Bacteria

### Edited by: Noton Kumar Dutta,

Johns Hopkins University, United States

### Reviewed by:

Henrietta Venter, University of South Australia, Australia Xian-Zhi Li, Health Canada, Canada

### \*Correspondence:

Inshad A. Khan inshad@rediffmail.com; iakhan@iiim.ac.in

### †Present address:

Nitin P. Kalia, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 05 July 2017 Accepted: 13 September 2017 Published: 04 October 2017

### Citation:

Singh S, Kalia NP, Joshi P, Kumar A, Sharma PR, Kumar A, Bharate SB and Khan IA (2017) Boeravinone B, A Novel Dual Inhibitor of NorA Bacterial Efflux Pump of Staphylococcus aureus and Human P-Glycoprotein, Reduces the Biofilm Formation and Intracellular Invasion of Bacteria. Front. Microbiol. 8:1868. doi: 10.3389/fmicb.2017.01868 Samsher Singh1,2, Nitin P. Kalia<sup>1</sup>† , Prashant Joshi2,3, Ajay Kumar<sup>4</sup> , Parduman R. Sharma2,4, Ashok Kumar2,4, Sandip B. Bharate2,3 and Inshad A. Khan1,2 \*

<sup>1</sup> Clinical Microbiology Division, Indian Institute of Integrative Medicine (CSIR), Jammu, India, <sup>2</sup> Academy of Scientific and Innovative Research (AcSIR), Indian Institute of Integrative Medicine (CSIR), Jammu, India, <sup>3</sup> Medicinal Chemistry Division, Indian Institute of Integrative Medicine (CSIR), Jammu, India, <sup>4</sup> Cancer Pharmacology Division, Indian Institute of Integrative Medicine (CSIR), Jammu, India

This study elucidated the role of boeravinone B, a NorA multidrug efflux pump inhibitor, in biofilm inhibition. The effects of boeravinone B plus ciprofloxacin, a NorA substrate, were evaluated in NorA-overexpressing, wild-type, and knocked-out Staphylococcus aureus (SA-1199B, SA-1199, and SA-K1758, respectively). The mechanism of action was confirmed using the ethidium bromide accumulation and efflux assay. The role of boeravinone B as a human P-glycoprotein (P-gp) inhibitor was examined in the LS-180 (colon cancer) cell line. Moreover, its role in the inhibition of biofilm formation and intracellular invasion of S. aureus in macrophages was studied. Boeravinone B reduced the minimum inhibitory concentration (MIC) of ciprofloxacin against S. aureus and its methicillin-resistant strains; the effect was stronger in SA-1199B. Furthermore, time–kill kinetics revealed that boeravinone B plus ciprofloxacin, at subinhibitory concentration (0.25 × MIC), is as equipotent as that at the MIC level. This combination also had a reduced mutation prevention concentration. Boeravinone B reduced the efflux of ethidium bromide and increased the accumulation, thus strengthening the role as a NorA inhibitor. Biofilm formation was reduced by four–eightfold of the minimal biofilm inhibitory concentration of ciprofloxacin, effectively preventing bacterial entry into macrophages. Boeravinone B effectively inhibited P-gp with half maximal inhibitory concentration (IC50) of 64.85 µM. The study concluded that boeravinone B not only inhibits the NorAmediated efflux of fluoroquinolones but also considerably inhibits the biofilm formation of S. aureus. Its P-gp inhibition activity demonstrates its potential as a bioavailability and bioefficacy enhancer.

Keywords: NorA, P-glycoprotein, biofilm, Staphylococcus aureus, efflux inhibition

## INTRODUCTION

fmicb-08-01868 October 3, 2017 Time: 15:45 # 2

The development of multidrug resistance (MDR) upon chronic exposure to chemotherapeutic agents is considered the major reason for the failure of chemotherapy in cancer and infectious diseases (Poole, 2005; Leitner et al., 2011; Joshi et al., 2014). Efflux pumps are transmembrane proteins, which are the major contributors to multidrug resistance (MDR) in patients with cancer and infectious diseases. P-glycoprotein (gp), a multidrug-resistance protein (also called MDR1), accounts for the development of resistance to various drugs as well as for the transport of endogenous and exogenous substrate ligands, such as various toxins, hormones, and drugs (Chen et al., 1986; Joshi et al., 2014). Moreover, P-gp alters the pharmacokinetics and pharmacodynamics of its substrates by modulating their absorption, distribution, metabolism, elimination, and toxicity properties (Lin, 2003). Bacterial multidrug efflux pumps are the major contributors to microbial resistance to several classes of antibiotics (Kaatz et al., 2003; Tegos et al., 2011; Kalia et al., 2012).

Presence or augmented expression of efflux pumps is responsible for reduced drug availability to inhibit the specific target. Due to continuous efflux, comparatively lower concentration of antimicrobial agent reaches the target, which may increase the rate of new mutations thus generating newer types resistant mutants exhibiting novel resistance mechanisms (Poole, 2005; Piddock, 2006; Sun et al., 2014). With the continuous emergence of pathogenic resistance to conventional drugs because of efflux through MDR pumps, increasing efforts are directed toward discovering efflux inhibitory molecules. NorA efflux pump is the most widely studied among efflux pumps of Staphylococcus aureus. According to a study NorA efflux pump was found to be overexpressed in 43% of resistant strains of S. aureus (Patel et al., 2010; Astolfi et al., 2017). NorA, an established multidrug efflux pump model in S. aureus, accounts for extruding quinolones, quaternary ammonium salts, acridines, rhodamines, verapamil, and ethidium bromide (Kaatz et al., 2003; Khan et al., 2006; Leitner et al., 2011; Tegos et al., 2011; Holler et al., 2012; Schindler et al., 2013; Fontaine et al., 2015; Tintino et al., 2016). Recently some non-antibiotics derivatives as potent NorA efflux pump inhibitors have been reported utilizing pharmacophore based modeling and drug repurposing utilizing in silico approach (Astolfi et al., 2017; Sabatini et al., 2017). In addition, some less toxic benzochromene derivatives and dithiazole thione derivatives have also been reported as NorA efflux inhibitors (Ganesan et al., 2016; Lowrence et al., 2016). A recently published comprehensive review summarized the applications of nanoparticles are also being explored for efflux inhibitory activity in addition to biofilm inhibition (Gupta et al., 2017).

Despite their varying structures, the efflux pumps of bacterial and mammalian systems have sufficient substrate homology. Studies have reported many dual types of natural plant products that act as inhibitors, such as verapamil, reserpine, piperine, and capsaicin (**Figure 1**), as well as osthol and curcumin (Bhardwaj et al., 2002; Khan et al., 2006; Kalia et al., 2012; Joshi et al., 2014). These inhibitors offer advantages, such as enhanced gastrointestinal absorption, improved permeation through the blood–brain barrier, increased drug concentrations in mammalian cells for improved killing of invasive pathogens, and the use of lower concentrations of drugs to eliminate side effects (Khan et al., 2006).

In addition to drug efflux, a critical problem is biofilm formation, which poses a major hindrance in the evolution of novel chemotherapeutic agents (Bjarnsholt et al., 2013). Nearly 80% of human bacterial infections are caused by biofilm formation, and 60% of nosocomial infections are caused by biofilm formation on medical devices and implants (Darouiche, 2004; Hou et al., 2012). Biofilm eradication requires a much higher concentration, namely 10–1000 times the minimum inhibitory concentration (MIC), of antibiotics. Active efflux has been reported to contribute to increased resistance in bacterial biofilms (Lewis, 2001; Kvist et al., 2008; Baugh et al., 2014; Van Acker and Coenye, 2016; Sabatini et al., 2017). Biofilms have slow metabolism and growth, the main reasons for increased resistance, because most antibiotics target metabolic or cell division pathways (Lewis, 2007, 2008). Researchers have reported that the synergistic action of molecules abolishes biofilm formation (Baugh et al., 2012).

Staphylococcus aureus is considered an extracellular pathogen, but there are accumulating evidences to reveal that S. aureus causes severe intracellular infections, such as endocarditis and septic shock (Hooper, 2005). This organism, on getting an opportunity to invade, survives and promotes diseases. Hirakata et al. (2009) and Kalia et al. (2012) concluded that efflux pumpoverexpressing strains can easily invade cells.

Several natural plant products act as efflux pump inhibitors (Stavri et al., 2007). Boeravinone B, the compound of interest in the present study, was isolated from the roots of Boerhavia diffusa, a plant belonging to the Nyctaginaceae family (Patil and Bhalsing, 2016). This plant is also known as Punarnava in the Indian traditional system of medicine, Ayurveda, and is established for its medicinal properties. In ancient times, plants were used to cure various physiological conditions, such as inflammation, jaundice, dyspepsia, nephrotic syndrome, convulsions, spleen enlargement, abdominal pain and stress, nematodal and microbial infections, and asthma (Ahmed-Belkacem et al., 2007; Khan et al., 2013; Bairwa et al., 2014). Boeravinone C and G have been reported to possess P-gp

FIGURE 1 | Reserpine, piperine, capsaicin and boeravinone B (molecule used in this study).



inhibitory activity (Ahmed-Belkacem et al., 2007; Stacy et al., 2013).

In this study, we described the role of boeravinone B in ciprofloxacin potentiating activity against NorA-overexpressing strains of S. aureus and P-gp inhibition in the colon cancer cell line.

## MATERIALS AND METHODS

### Chemicals

All fluoroquinolones, ethidium bromide reserpine, piperine, fluorescein isothiocyanate (FITC) and 4<sup>0</sup> ,6-diamidino- 2 phenylindole (DAPI) were purchased from Sigma Chemical Co. (St. Louis, MO, United States). Microbiological media were purchased BD Microbiology Products (Sparks, MD, United States). All other chemicals were of analytical grade.

### Bacterial Strains

Staphylococcus aureus ATCC 29213, SA-1199B, SA-1199, SA-K1758, MT23142, and MRSA15187 were cultivated on trypticase soya agar (TSA). Important characteristics are described in **Table 1**.

### Boeravinone B Isolation

Authentic samples of B. diffusa roots were collected from Jammu region (Voucher Specimen No. 21713), dried and powdered followed by solvent extraction in dichloromethane: methanol mixture (1:1 v/v) using cold maceration method. Crude extract was concentrated and purified by repeated silica gel column chromatography using hexane: ethyl acetate (95: 5 to 75: 25 v/v).

## Drug Potentiation by Broth Checkerboard Microdilution

Drug potentiation or synergistic action of compound was assessed by broth checkerboard microdilution technique, which is most known method (Eliopoulos and Wennersten, 2002). Combination of ciprofloxacin and boeravinone B was tested in Mueller Hinton Broth (MHB; pH 7.0) against all bacterial strains using 96 well microtitre plates at a concentration range of 0.06–32 µg/ml and 0.8–50 µM, respectively. Additionally, combination of ethidium bromide and boeravinone B was assessed in similar way using S. aureus SA-1199B. Cell suspension of a density of 0.5 McFarland (≈1.5 × 10<sup>8</sup> cfu per ml of Escherichia coli) was used as inoculum after diluting 1:100, a volume of 0.1 ml (5 × 10<sup>5</sup> ) was added to plates. The plates were incubated at 37◦C for 18–20 h. Reserpine and piperine (known efflux pump blocker) were used as the control in this study. Minimum effective concentration (MEC) of boeravinone B was determined which is defined as concentration at which there is atleast fourfold reductions in MIC of ciprofloxacin. In a similar way various other fluoroquinolones and ethidium bromide were also assessed for synergistic potential of boeravinone B.

### Time Kill Curve

Time-kill study of ciprofloxacin alone and in combination with boeravinone B was performed in 50 ml volume conical flasks containing 25 ml of MHB using the previously described method (Eliopoulos and Moellering, 1996). Ciprofloxacin at 2 µg/ml (0.25 × MIC) was tested alone and in combination with boeravinone B at the MEC concentration (12.5 µM) as determined above. Ciprofloxacin was also tested alone at an MIC of 8 µg/ml. NorA overexpressing S. aureus SA-1199B was used as the test bacterium. Ten folds serial dilutions of these aliquots were spotted on TSA plates.

### Impact on Frequency of Resistance

Frequency of resistance (FOR) or Mutation prevention concentration (MPC) of ciprofloxacin against S. aureus ATCC 29213 was determined by method described elsewhere (Drugeon et al., 1999). MHA plates were supplemented with MIC, 2×, 4×, and 8× MIC of ciprofloxacin's concentrations with or without boeravinone B at MEC. A bacterial suspension of 0.1 mL (10<sup>9</sup> cfu) was plated on to each ciprofloxacin containing plates without and with boeravinone B. MPC was determined as the ratio of colonies appearing on ciprofloxacin or ciprofloxacin with boeravinone B combination to total colonies plated on a drug free plate.

## Ethidium Bromide Efflux and Accumulation Studies

Fluorescence based method as described by Brenwald et al. (1998) was used to determine the ethidium bromide efflux inhibitory as well accumulation studies. For this purpose a freshly grown NorA overexpressing S. aureus SA-1199B was suspended to an optical density of 0.2 in uptake buffer (110 mM NaCl, 7 mM KCl, 50 mM NH4Cl, 0.4 mM Na2HPO4, 52 mM Tris base and 0.2% glucose, adjusted to pH 7.5 with HCl). Cells were loaded with 10 µg/ml ethidium bromide for 30 min at 37◦C. Extracellular ethidium bromide was removed by centrifugation and pellet was resuspended in same volume of uptake buffer with and without boeravinone B (12.5 µM). Reserpine at 25 µg/ml was used as a known efflux pump inhibitor. For accumulation studies bacterial cells loaded with ethidium bromide for 30 min after which EPIs (boeravinone B and reserpine) were added. Loss or gain of fluorescence in presence of compound and known EPI was deduced at 530 and 600 nm for excitation and emission wavelength, respectively, using multimode reader Infinite 200 Pro (Tecan Mannedorf, Switzerland).

## Biofilm Inhibition

fmicb-08-01868 October 3, 2017 Time: 15:45 # 4

Effect of boeravinone B on the biofilm inhibitory potential of ciprofloxacin was studied with drug combination on biofilm formation, as previously described (Schillaci et al., 2010). For biofilm formation, cultures (S. aureus SA-1199 and SA-1199B) were grown overnight using TSB supplemented with 2% glucose (TSBG) and diluted to an optical density (570 nm) of 0.015 after diluting 1:200 (1 × 10<sup>9</sup> cells/mL). Microtitre plates for inhibitory studies were prepared as described for synergistic activity with ciprofloxacin by checkerboard microdilution method. After 24 h, wells were washed thrice with PBS followed by fixation with methanol for 15 min and finally air-dried in inverted position at 37◦C. The wells of the dried plates were stained with 0.1% (w/v) crystal violet for 10 min and rinsed thoroughly with water until the negative control wells (without biofilms) appeared colorless. To quantify biofilm formation, 0.2 mL of 95% ethanol was added to wells of plate that were stained with crystal violet. Immediately absorbance read at 595 nm using microplate reader. In a separate plate the contents were decanted and washed with PBS to remove the planktonic cells and again filled with fresh TSB. After incubation for 30 min, 15 µl of 0.04% resazurin and 12.5 µl of 20% Tween80 were added in each well of the plate to access the viability of the cells within biofilm. The plates were incubated for 15 min and fluorescence was determined at an excitation at 560 nm and emission at 590 nm (Tecan Infinite M200 microplate reader) as described previously (Bauer et al., 2013). Minimal biofilm inhibitory concentration (MBIC) is defined as the concentration at which absorbance was reduced by ≥90% or was nearly equivalent to media control wells.

## Confocal Microscopic Studies for Biofilm Inhibition

The potentiating effect of boeravinone B was further confirmed by staining the biofilms with FITC and DAPI which stains protein (amine reactive) and nucleic acid (DNA), respectively, by modified method as described by Yang (using DAPI in place of Hoechst) followed by visualization by confocal laser scanning microscopy (Yang et al., 2015). The biofilm of S. aureus ATCC 29213 was grown in six well poly styrene plate (Nunc) containing sterile 18 mm glass cover slips. Ciprofloxacin was used at 4 µg/ml which is MBIC and at 1 µg/ml which is subMBIC (0.25× MBIC) as determined above. Boeravinone B at 12.5 µM (MBIC) was added to sub-inhibitory concentration of ciprofloxacin (0.25× MBIC). The plates were incubated for 24 h at 37◦C. PBS washes were given to eliminate planktonic cells followed by fixation of biofilm with 5% para-formaldehyde for 1 h at 50◦C. Fixed biofilm was flooded with 0.001% (w/v) FITC and 1 µg/ml DAPI at kept at room temperature for 1 h. Images were seen and captured on 40× oil immersion lens using CLSM (Olympus flou View 1000).

## In Vitro Screening of P-Glycoprotein Inhibitory Activity

P-glycoprotein inhibitory activity was assessed by the method described earlier using colorectal cell line LS180 (Jin et al., 2012; Joshi et al., 2014). Cell monolayer was prepared in 96 well plate using Dulbecco's Modified Eagle's medium (DMEM) media. The media was decanted and replaced with Hank's buffer containing 10 mM of Rh123 as a P-gp substrate. Twofold serial dilutions of boeravinone B (80–5 µM) were added and plate was incubated at 37◦C in CO<sup>2</sup> incubator for 90 min. A known P-gp inhibitor elacridar (10 µM) was used as positive control. The contents of the wells were decanted and washed four times with cold PBS followed by cell lysis for 1 h using 200 µl of lysis buffer (0.1% Triton X 100 and 0.2 N NaOH). A total of 100 µl of lysate was used for reading fluorescence of Rh123 at an excitation at 485 nm and emission at 590 nm. All samples were normalized by dividing fluorescence of each sample with total protein present in the lysate. Half minimal inhibitory concentration (IC50) value of boeravinone B was calculated using Graphpad Prism 5 software (GraphPad Software, Inc., La Jolla, CA, United States). Data is expressed as mean ± SD or representative of one of three similar experiments. Comparisons were made between control and treated groups or the entire intra group using one way ANOVA with post Bonferroni test through GraphPad Prism 5.00.288 statistical analysis software. <sup>∗</sup>p-values <0.05 were considered significant.

## Molecular Docking Studies against P-Glycoprotein and NorA Efflux Pumps

Due to unavailability of crystal structures of human P-gp and S. aureus NorA, molecular docking studies were carried out using the homology models. The homology models for human P-gp (Jin et al., 2012) was prepared from Caenorhabditis elegans P-gp (PDB ID: 4AZF) and S. aureus NorA efflux pump (Aller et al., 2009) was prepared from glycerol 3-phosphate transporter pump from E. coli (PDB ID: IPW4) using Glide Induced Fit Docking in default setting. For Nor A docking studies, ciprofloxacin/reserpine binding site was used to construct grid file whereas for P-gp docking, verapamil binding site was used (Aller et al., 2009).

### Macrophage Invasion Assay

To assess the impact of boeravinone B on invasion of S. aureus in macrophage, previously described method was used (Cheung and Bayles, 2007). Murine macrophage cell line J774 maintained on Roswell Park Memorial Institute medium (RPMI) supplemented with 10% (v/v) fetal calf serum was used as host. The macrophages were grown in monolayer to a density to 10<sup>5</sup> cells/well in 24 well plate in physiological conditions, i.e., 37◦C and 5% CO<sup>2</sup> for 24 h. 10<sup>6</sup> cfu/well of overnight grown S. aureus strains SA-1199B, SA-1199, and SA-K1758 were used to infect macrophages (MOI of 10) in presence and absence of boeravinone B at its MEC (12.5 µM) for 2 h. Extracellular and adherent bacteria washed with PBS (pH 7.4) followed by treatment with 50 µg/ml gentamicin for 30 min. Intracellular bacteria were quantified by host cell lysis by brief treatment of 0.07% SDS which followed neutralization with 6% BSA. Viable bacteria were counted by plating on TSA. In a similar experiment RPMI devoid of FCS was used to assess the role of FCS.

TABLE 2 | Ciprofloxacin potentiation by boeravinone B against various strains of S. aureus with different levels of NorA expressions and MRSA.


#All EPIs tested have no self-antimicrobial activity up to 100 µM.

TABLE 3 | Activity of boeravinone B on the MICs of ethidium bromide and other fluoroquinolones in susceptible and resistant S. aureus strains.


### RESULTS

### Isolation and Purity of Boeravinone B

By using the cold maceration method, 6,9,11-trihydroxy-10 methylchromeno[3,4-b]chromen-12(6H)-one (boeravinone B; **Figure 1**) was isolated from the roots of B. diffusa. It is a yellow solid compound, with a melting point of 201◦C– 203◦C. The structure was confirmed through high-resolution mass spectroscopy (m/z 313.0708 [M + H+], as calculated for C17H12O<sup>6</sup> + H+: 313.0707) and comparison of the <sup>1</sup>H and <sup>13</sup>C NMR data. Boeravinone B used in this study had 96% purity, as determined through high performance liquid chromatography. Data related to the purity and structure of the isolated boeravinone B was characterized by comparison of spectral data with described reported values (Kadota et al., 1989) (Supplementary Data S1).

### Effect of NorA Efflux Pump Inhibition on MIC of Fluoroquinolones and Ethidium Bromide

The resistance reversal activity of boeravinone B was quantified by calculating the fold reduction in the MIC of ciprofloxacin. Boeravinone B lack anti-staphylococcal activity when assessed up to 100 µM against all tested strains. This MIC of ciprofloxacin alone and in combination with boeravinone B was determined (**Table 2**). Boeravinone B reduced the MIC of ciprofloxacin by twofold in S. aureus SA-1199 and ATCC 29213, whereas the effect was stronger in the NorA-overexpressing S. aureus strains SA-1199B and MT-23142, where the MIC was reduced by eightfold. Conversely, SA-K1758 (NorA knocked out) did not show any reduction in the MIC of ciprofloxacin. Boeravinone B was also active against a methicillin-resistant S. aureus (MRSA) strain, which showed a fourfold reduction in the ciprofloxacin MIC.

In addition to ciprofloxacin, spectrum of activity was also assessed using other fluoroquinolones like ofloxacin and norfloxacin which are well known substrates for NorA efflux pump and newer fluoroquinolones like gatifloxacin and moxifloxacin which are poor efflux substrates for NorA. Boeravinone B was used at 4× MEC for ethidium bromide (6.25 µM or 2 µg/ml). Potential of boeravinone B to inhibit the efflux of other fluoroquinolones was investigated using S. aureus strains SA-1199B, SA-1199, SA-K1758, and MT23142 (**Table 3**). MIC potentiation using ethidium bromide yielded similar effects of eightfold reduction in the MIC of ethidium bromide (2 µg/mL) compared with that of ethidium bromide alone, which exhibited an MIC of 16 µg/mL against SA-1199B. In NorA overexpressing strains (SA-1199B and MT23142), boeravinone B reduced the MIC of norfloxacin by eightfold while only twofold reduction in MIC of ofloxacin, moxifloxacin, and gatifloxacin was observed. It was observed that ciprofloxacin MIC reduction was relatively less (two– fourfold) in case of wild type S. aureus SA-1199 against these fluoroquinolones. There was no synergistic effect against S. aureus SA-K1758 due to absence of NorA efflux pump. Significant reduction in MICs against NorA overexpressing strains of S. aureus concludes that ciprofloxacin and norfloxacin

diamond) and ciprofloxacin at subMIC in combination with boeravinone B at MEC (closed square) including boeravinone B alone (Open circle). Growth control having plain media without ciprofloxacin or EPI served control (Open square). Each time point is mean Log<sup>10</sup> of three independent readings.

TABLE 4 | Mutation prevention concentration (MPC) of ciprofloxacin with boeravinone B against S. aureus ATCC 29213.


have stronger affinity for NorA and efflux can be inhibited using boeravinone B.

## Effect of Boeravinone B on Time–Kill Kinetics of Ciprofloxacin

The effect of boeravinone B on the time–kill kinetics of ciprofloxacin was assessed in S. aureus SA-1199B. Ciprofloxacin alone (MIC: 8 µg/mL) showed 3-log reduction in growth, whereas the same was accomplished at a subinhibitory concentration of the ciprofloxacin (0.25 × MIC) in combination with boeravinone B at a minimum effective concentration (MEC; 12.5 µM). The sub MIC of ciprofloxacin alone could not reduce the growth. Ciprofloxacin alone at an MIC effectively reduced the colony-forming units (CFU) in 6 h, but cell growth reoccurred after 24 h. The combination of ciprofloxacin (at subinhibitory concentration) and boeravinone B prevented the growth of resistant colonies (**Figure 2**).

### Frequency of Ciprofloxacin Resistance in Presence of Boeravinone B

For effective clinical applications, drugs should be able to suppress mutations and resistance mechanisms. A resistant

against S. aureus SA-1199B. Symbols represents control which have no addition of any EPI (closed diamond), in presence of boeravinone B (open square) and in presence of known EPI, i.e., reserpine (open triangle).

mutant selection study was performed on wild-type S. aureus ATCC 29213 because this strain does not have any known mutations in the regulatory region of NorA and drug targets (DNA gyrase and topoisomerase IV). The minimum concentration of a drug at which no resistant mutant is selected is defined as its MPC. Ciprofloxacin alone showed an MPC of 4 µg/ml whereas, in combination with 12.5 and 25 µg/ml of boeravinone B, the MPC of ciprofloxacin reduced to 2 and 1 µg/ml, respectively (**Table 4**). The MPC of this combination was lower than the Cmax of ciprofloxacin (3–4 µg/ml) in human plasma, indicating the clinical relevance of these combinations in restricting the selection of resistant mutants.

absence of boeravinone B.

## Effect of Boeravinone B on Ethidium Bromide Efflux Inhibition and Accumulation

Ethidium bromide is a well-known substrate for the NorA MDR efflux pump. SA-1199B cells were allowed to be loaded with ethidium bromide with and without boeravinone B and placed in a fluorometer cuvette containing fresh medium. Ethidium bromide fluoresces only when bound to nucleic acids in cells; therefore, a rapid decrease occurred in the fluorescence because of the NorA-mediated efflux of ethidium bromide. As shown in **Figure 3**, only the control cells without boeravinone B maximally extruded ethidium bromide, resulting in significantly decreased fluorescence in the assay period. Conversely, in cells treated with boeravinone B, the efflux of ethidium bromide was prevented, resulting in the prolonged retention of fluorescence. In the accumulation studies, ethidium bromide uptake was allowed in bacterial cells for 30 min. A counterbalance effect occurred because of the simultaneous efflux of ethidium bromide, which effectively increased the fluorescence. The addition of boeravinone B at this stage resulted in a surge in the fluorescence because of the inhibition of efflux pumps, resulting in a higher retention of fluorescence (**Figure 3**). Reserpine, a known efflux pump inhibitor, was used as the positive control in both studies. When compared, ethidium bromide accumulation in SA-1199B, SA-1199, and SA-K1758, accumulation was maximum in knockout followed by wild type and overexpressing which is in correlation with the phenotypes of the tested strains. In presence of boeravinone B, accumulation was significantly increased in NorA overexpressing, while minor effect was seen for wild type. S. aureus SA-1199. There was no effect on the accumulation efficiency of SA-K1758 which is devoid of functional norA gene (**Figure 4**).

TABLE 5 | Effect on biofilm inhibitory potential of ciprofloxacin when combined with boeravinone B at different concentrations against various strains of S. aureus.


## Biofilm Inhibition

The minimum concentration of a drug for inhibiting a biofilm is defined and expressed as its MBIC. The MBIC of a drug is typically higher than its MIC because of the lesser permeability of the drug within the matrix of a biofilm. In this study, ciprofloxacin showed an MBIC of 4 and 16 µg/mL for S. aureus SA-1199 and SA-1199B, respectively. A concentration-dependent decrease occurred in the MBIC of ciprofloxacin against both strains in the presence of boeravinone B (6.25–25 µM; **Table 5**). To ascertain that the biofilm inhibition was caused by the killing of the bacteria embedded in the matrix of biofilm, the resazurin staining of a parallel set of plates was performed. The change in the color of this redox dye from blue to red indicated the viability of the cells in the well. The biofilm inhibition correlated well with the cell viability (Supplementary Figure S1).

### Microscopic Studies

Effect of boeravinone B plus ciprofloxacin was evaluated by microscopic studies of the S. aureus ATCC 29213 generated biofilm. Fluorescein isothiocyanate stains proteins

and imparts green fluorescence, whereas 4<sup>0</sup> ,6-diamidino-2 phenylindole binds to nucleic acids and yields blue fluorescence when observed in the whole image field. As shown in **Figure 5**, the untreated biofilms were associated with a large amount of proteins and DNA. The sub MBIC of ciprofloxacin did not inhibit the biofilms; however, in combination with the MEC of boeravinone B, ciprofloxacin inhibited the biofilm, as did the MBIC of ciprofloxacin, as evident by the reduced intensity of green and blue fluorescence.

### Molecular Docking on P-gp

Boeravinone B was observed to interact with P-gp at the verapamil binding site, particularly with the residues Met69, Phe72, Tyr307, Tyr310, Phe314, Leu332, Phe335, Phe336, Leu339, Phe728, Phe759, Phe957, Phe978, and Val982 through hydrophobic π–π interactions (**Figure 6A**). Consequently, it led to an increase in the intracellular accumulation of the substrate Rh123. Similarly, the NorA efflux pump is a transmembrane pump, in which ciprofloxacin or reserpine optimally bind to site 1 through strong hydrogen bonding of carboxyl functionality with the Arg98 cationic guanido group. The hydrophobic and hydrogen interactions of ciprofloxacin were mimicked by boeravinone B. Molecular modeling studies with the bacterial NorA efflux pump homology model revealed that boeravinone B binds to the Ile23 and Glu222 residues of the pump through strong hydrogen bonding formed by phenolic hydroxyl groups and thus block the interaction of substrate quinolones with the efflux binding cavity (**Figure 6B**).

## Invasion Inhibition and Role of Fetal Calf Serum

The intracellular invasion of S. aureus SA-1199B was nearly 1.5 log<sup>10</sup> higher than that of S. aureus SA-1199 in the murine macrophage cell line J774. Boeravinone B at its MEC (12.5 µM) was not active when the experiment was performed in the presence of 10% fetal calf serum (FCS). When the experiment was repeated using RPMI media without FCS, the invasiveness of S. aureus had reduced, suggesting the binding of boeravinone B to FCS. The invasiveness of SA-1199B showed greater reduction (1.6 log10) than did that of S. aureus SA-1199 (0.5 log10). S. aureus SA-K1758 exhibited the least penetration into macrophages, and boeravinone B did not affect the invasiveness of this strain (**Figure 7**).

### DISCUSSION

Efflux pumps are membrane-bound proteinaceous transporters known for extruding chemotherapeutic agents or xenobiotics, which are otherwise harmful for bacterial survival. During evolution, microbes were selected and evolved through antibiotic pressure with the overexpression of efflux pumps, which renders the antibiotics inefficient and hence results in the poor efficacy of chemotherapeutic agents. Efflux pumps can export a large pool of structurally unrelated substrates out of cells, which hampers their clinical efficacy. NorA is an MDR efflux pump of the major facilitator superfamily pumps and has 12 transmembrane segments. It is established for expelling various fluoroquinolones. NorA has been well established as a model system for studying efflux pump inhibition in Gram-positive pathogens (Kaatz, 2005; Kumar et al., 2008).

In the present study, boeravinone B obtained from B. diffusa was assessed for its NorA inhibitory activity in SA-1199B, which is a model organism for NorA inhibition studies because it mimics the resistance induced by fluoroquinolone pressure. Ethidium bromide MIC potentiation by 16-fold in NorA overexpressing strains in presence of boeravinone B shows inhibition of efflux mediated resistance. Presence of boeravinone B also exhibited two–eightfold reduction in MIC of ciprofloxacin and norfloxacin in these strains. The potentiating effect was less prominent (twofold) in wild type strain S. aureus SA-1199. Additionally there was no reduction in MIC of S. aureus SA-K1758 which is lacking functional norA gene. This concludes that ciprofloxacin and norfloxacin fluoroquinolone are favorable substrate for NorA. Similar observation about the enhanced affinity of NorA for these fluoroquinolones have been reported previously (Gibbons et al., 2003; Mullin et al., 2004). Boeravinone B increased the intrinsic susceptibility of S. aureus to ciprofloxacin and significantly reduced the emergence of ciprofloxacin-resistant S. aureus. A subinhibitory concentration of ciprofloxacin (2 µg/ml) with the MEC of boeravinone B (12.5 µM) exhibited a bactericidal effect and a 3-log<sup>10</sup> reduction in CFU in 6 h, with no emergence of mutants even after 24 h. Conversely, ciprofloxacin alone exhibited the same effect at 8 µg/mL, but resistant mutants were observed after 24 h (**Figure 2**). Ethidium bromide efflux inhibition and accumulation

are well-known mechanism-based descriptors for the specificity of NorA inhibition. The fluorescence-based efflux accumulation studies of ethidium bromide-preloaded NorA-overproducing S. aureus cells revealed reduced efflux in the presence of boeravinone B, indicating the inhibition of the ethidium bromide efflux mechanism. Furthermore, its effect on the retention of Rh123 in P-gp-overexpressing human adenocarcinoma LS-180 cells confirmed its role as a dual inhibitor of P-gp and the S. aureus efflux pump inhibitor. Studies have reported several similar dual inhibitors, such as osthol, curcumin, piperine, and capsaicin (Khan et al., 2006; Kalia et al., 2012). Docking results with the NorA homologous structure and P-gp obtained using Glide clearly elaborated that boeravinone B interacts with both pumps through core hydrophobic π–π interactions and hydrogen bonding.

Both methicillin susceptible S. aureus and MRSA are common pathogenic concerns in the hospital environment (Livermore et al., 2015). MRSA remains a major concern in the clinical treatment of S. aureus-related infections since half a century. Vancomycin was the only drug of choice for treating the

increased incidence of MRSA, but resistance abolished the drug's mechanism of action. With time, novel candidates were observed to be active against these strains, such as daptomycin, linezolid, and some glycopeptides, but resistance to these drugs was reported (Nannini et al., 2010). Because of frequent reports on resistance to antimicrobials, researchers are now using adjuvant therapy as a treatment strategy (Drugeon et al., 1999).

Active efflux has been reported to be associated with increased resistance in bacterial biofilms (Lewis, 2007; Van Acker and Coenye, 2016). The treatment of biofilm-related infections is a major unmet clinical concern because currently used antibiotics are insufficient to treat these highly persistent infections; some candidate molecules are under early development. Evidently, biofilms have slow metabolism and cell growth, which contribute to increased resistance because most currently used antibiotics have been discovered by targeting metabolic or cell division pathways. The synergistic action of molecules was reported to abolish the formation of biofilms (Baugh et al., 2014). These ideas prompted the assessment of the role of boeravinone B in the prevention and eradication of biofilms, a major clinical problem in the current era of anti-infective drug discovery (Bharate et al., 2015). Current studies have revealed that in combination with boeravinone B, the sub MBIC of ciprofloxacin inhibited the formation of biofilms through enhanced drug accumulation. Therefore, boeravinone B could penetrate the complex matrix of S. aureus-generated biofilms to inhibit viable bacteria, as observed in the resazurin assay (Supplementary Figure S1). The involvement of efflux pump inhibitors, such as certain diamines, was reported in reducing the invasiveness of Pseudomonas aeruginosa in eukaryotic cells (Hirakata et al., 2009). We obtained the similar results for capsaicin in a previous study (Kalia et al., 2012). This observation shows similar effects of boeravinone B, as expected, after the removal of FCS from the assay media (RPMI).

### CONCLUSION

Boeravinone B is a novel dual inhibitor of bacterial NorA efflux pump and human P-gp. Such inhibitors play potential role in

### REFERENCES


drug combinations by retaining higher concentrations of drugs in bacterial and host cells.

### AUTHOR CONTRIBUTIONS

SS and NK performed microbiological studies. PJ performed isolation of compound and in silico studies. AsK and PS performed microscopic studies. SS and IK participated in design and execution of microbiological studies. SB participated in isolation and identification of compound used in this study. PJ and AjK participated in P-glycoprotein efflux inhibition studies. IK and SS prepared the manuscript and all authors participated in revising manuscript. All authors approved the study. All authors are accountable for all aspects of work in ensuring that questions related to the accuracy or integrity of the part of work are appropriately investigated and resolved.

## FUNDING

SS received a Senior Research Fellowship from the Indian Council of Medical Research, New Delhi, India (80/993/2015-ECD-1). This work was supported by Council of Scientific and Industrial Research (CSIR), New Delhi, India (Grant no. BSC0205).

## ACKNOWLEDGMENTS

We are grateful to Professor G. W. Kaatz (Wayne State University School of Medicine, Detroit, MI, United States) for providing S. aureus SA-1199, SA-1199B and SA-K1758. MT23142, a NorA over expressing strain was a kind gift from Professor David Hooper (Harvard Medical School, Boston, MA, United States).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.01868/full#supplementary-material

activity of daptomycin and delafloxacin against Staphylococcus aureus biofilms. Antimicrob. Agents Chemother. 57, 2726–2737. doi: 10.1128/AAC.00181-13




against the NorA efflux pump of Staphylococcus aureus. Microb. Pathog. 97, 9–13. doi: 10.1016/j.micpath.2016.04.003


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Singh, Kalia, Joshi, Kumar, Sharma, Kumar, Bharate and Khan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Structure–Function Relationship of Aminopeptidase P from Pseudomonas aeruginosa

Cui-Ting Peng1,2† , Li Liu1,2† , Chang-Cheng Li<sup>2</sup> , Li-Hui He<sup>2</sup> , Tao Li<sup>2</sup> , Ya-Lin Shen<sup>2</sup> , Chao Gao<sup>2</sup> , Ning-Yu Wang2,3, Yong Xia<sup>2</sup> , Yi-Bo Zhu<sup>2</sup> , Ying-Jie Song<sup>2</sup> , Qian Lei<sup>2</sup> , Luo-Ting Yu1,2 \* and Rui Bao<sup>2</sup> \*

<sup>1</sup> Pharmaceutical and Biological Engineering Department, School of Chemical Engineering, Sichuan University, Chengdu, China, <sup>2</sup> Center of Infectious Diseases, State Key Laboratory of Biotherapy, West China Hospital, Sichuan University and Collaborative Innovation Center, Chengdu, China, <sup>3</sup> School of Life Sciences and Engineering, Southwest Jiaotong University, Chengdu, China

### Edited by:

Teresa M. Coque, Instituto Ramón y Cajal de Investigación Sanitaria, Spain

### Reviewed by:

Jose L. Martinez, Consejo Superior de Investigaciones Científicas (CSIC), Spain Francesco Imperi, Sapienza Università di Roma, Italy

### \*Correspondence:

Rui Bao baorui@scu.edu.cn Luo-Ting Yu yuluot@scu.edu.cn †These authors have contributed equally to this work.

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 19 July 2017 Accepted: 20 November 2017 Published: 05 December 2017

### Citation:

Peng C-T, Liu L, Li C-C, He L-H, Li T, Shen Y-L, Gao C, Wang N-Y, Xia Y, Zhu Y-B, Song Y-J, Lei Q, Yu L-T and Bao R (2017) Structure–Function Relationship of Aminopeptidase P from Pseudomonas aeruginosa. Front. Microbiol. 8:2385. doi: 10.3389/fmicb.2017.02385 PepP is a virulence-associated gene in Pseudomonas aeruginosa, making it an attractive target for anti-P. aeruginosa drug development. The encoded protein, aminopeptidases P (Pa-PepP), is a type of X-prolyl peptidase that possesses diverse biological functions. The crystal structure verified its canonical pita-bread fold and functional tetrameric assembly, and the functional studies measured the influences of different metal ions on the activity. A trimetal manganese cluster was observed at the active site, elucidating the mechanism of inhibition by metal ions. Additionally, a loop extending from the active site appeared to be important for specific large-substrate binding. Based on the structural comparison and bacterial invasion assays, we showed that this non-conserved surface loop was critical for P. aeruginosa virulence. Taken together, these findings can extend our understanding of the catalytic mechanism and virulence-related functions of Pa-PepP and provide a solid foundation for the design of specific inhibitors against pathogenic-bacterial infections.

Keywords: Pseudomonas aeruginosa, aminopeptidase P, virulence, tri-nuclear form, X-ray crystallography

## INTRODUCTION

Pseudomonas aeruginosa is a common nosocomial pathogen and is notoriously difficult to treat due to its high intrinsic and acquired drug resistance (Hancock, 1998; Stover et al., 2000). To address the challenges of P. aeruginosa infections, alternative approaches other than conventional antibiotic therapy were undertaken, with an emphasis on anti-virulence strategies (Hentzer et al., 2003; Luckett et al., 2012; Gi et al., 2014; Hwang et al., 2016; Maura et al., 2016; Johnson and Abramovitch, 2017; Rampioni et al., 2017). The targets for most anti-virulence strategies are those well-studied virulence factors (adhesins, toxins, effector secretion system components) that directly participate in pathogen-host cell interactions (Woods et al., 1982; Hahn, 1997; Hood et al., 2010). However, not all of these virulence factors play significant roles in P. aeruginosa infection (Miyata et al., 2003). Instead, there are many genes with unidentified roles in P. aeruginosa virulence,

**Abbreviations:** DMEM, Dulbecco's Modified Eagle's Medium; FBS, fetal bovine serum; IPTG, isopropyl β-D-1 thiogalactopyranoside; LB, Luria–Bertani; MOI, multiplicity of infection; NCPSS, National Center for Protein Sciences Shanghai; OD600, optical cell densities at 600 nm; PBS, phosphate-buffered saline; P. aeruginosa, Pseudomonas aeruginosa; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; SSRF, Shanghai Synchrotron Radiation Facility.

while the corresponding mutants resulted in obvious virulence attenuation, suggesting potential applications for these genes in anti-virulence therapy discovery.

PepP encodes an enzyme belonging to the aminopeptidases P (APPro) family (E.C.3.4.11.9), a type of metalloprotease that catalyzes the removal of the N-terminal residue from a polypeptide that has proline as the second residue (Taylor, 1993; Gonzales and Robert-Baudouy, 1996; Wilce et al., 1998). It is one of the critical virulence-associated genes identified using a P. aeruginosa-C. elegans infection mode (Feinbaum et al., 2012). In this model, transposon mutants of pepP in P. aeruginosa PA14 have attenuated virulence, leading to reduced C. elegans survival. PepP is highly conserved in all P. aeruginosa genomes sequenced to date and with high similarity to this gene from other Pseudomonas species (82.4%–100% identity). In contrast to other metalloaminopeptidases, APPro is a cytoplasmic aminopeptidase and recognizes a restricted scissile Xaa-Pro bond of those polypeptide or protein substrates that may be linked to virulenceassociated phenotypes or other biological processes beyond the general protein degradation ability (Lowther and Matthews, 2002). APPro is widely distributed among bacteria, fungi, plants and mammals. It is thought to play a role in many important biological pathways, including hormone regulation in mammals (Simmons and Orawski, 1992) and the terminal degradation of proline-containing peptides and proteins (Wilce et al., 1998), or organophosphate compounds in bacterial (Jao et al., 2004). The gene coding Escherichia coli aminopeptidases P (Ec-PepP) was even identified as a factor involved in outer membrane vesicles (OMV) production (McBroom et al., 2006; Vanaja et al., 2016). While structural and biochemical studies on APPro have visualized a general catalytic mechanism and revealed a conserved binding pocket for N-terminal substrates (Wilce et al., 1998; Graham et al., 2005, 2006; Liu et al., 2007; Graham and Guss, 2008; Jeyakanthan et al., 2009; Weaver et al., 2014), further investigations are needed to elucidate the structural basis for APPro's specific substrate recognition and diverse functions. Thus, a better understanding of the structure and function of P. aeruginosa APPro (Pa-PepP) would enable us to propose the possible mechanisms involved in bacterial virulence.

In this study, we solved the X-ray crystal structure of Pa-PepP with a resolution of 1.8 Å. A biochemical analysis verified the residues critical for catalysis. The presence of a trinuclear manganese cluster in the reaction center suggests a mechanism for inhibition by excessive metal ion binding. Furthermore, an extended substrate binding site was identified to be responsible for virulence-related protein recognition.

### RESULTS

### Pa-PepP Adopts a Canonical Pita-Bread Fold and Assembles as a Tetramer in Crystal

Refinement of the Pa-PepP structure resulted in a final model with a free R-factor (Rfree) of 0.2206 at 1.8 Å. Sufficient electron density allowed us to model all residues from 1 to 444. The crystallographic statistics are summarized in **Table 1**. The monomer structure displays a two-domain organization where the N-domain (1–175) is composed of a mainly parallel beta-sheet core (B1–B6) flanked by seven alpha helices (A–G). In contrast, the catalytic Cdomain (176–444) adopts a conserved "pita-bread" fold with six beta sheets (B7-B12) in antiparallel configurations (**Figure 1A**). This pita-bread fold is commonly found in N-terminal amido-, imido-, and amidino-scissile bond-cleaving enzymes, and serves as a structural basis for the metaldependent catalysis (Lowther and Matthews, 2000; Besio et al., 2010).

Although the recombinant Pa-PepP purified from E. coli strain BL21 (DE3) exists as a monomer in solution (**Figure 1B**), tetrameric oligomerization was observed in crystal packing (**Figure 1C**). The monomers were arranged as a dimer with an extended loop contributing to the active site of the adjacent subunit and an average interface area of approximately 2050.9 A<sup>2</sup> per subunit. The dimer-of-dimers was generated by crystallographic symmetry operation (x–1/2, –y–1/2, –z), and the major interactions are contributed by the C-domain of each monomer, resulting in an ∼815.6 A<sup>2</sup> buried area per subunit. For most metalloaminopeptidases, the oligomeric state is essential for their biological functions. The loop from the adjacent monomer extends the substrate-binding site and thus enables the enzyme to cleave larger substrates.


(http://www.pymol.org). (B) The gel filtration curve for the purified recombinant Pa-PepP in solution. The AUC results are presented as the ultraviolet absorption in 280 nm. The black solid curve refers to Pa-PepP eluted from GE HiLoad 16/600 Superdex 200 column. The red dash curve refers to the sample elution volume and corresponding molecule mass of the sample when compared with the standard sample proteins. (C) 3D structure of Pa-PepP showed a detailed tetramer of four subunits or a homodimer of Pa-PepP. Each subunit is presented as different color with Mn2<sup>+</sup> ions are shown as orange spheres in the metal binding site.

Thus, the antiparallel dimerization would be the basic active polymerization forms to its biological activity, and the different quaternary arrangement may reflect a possible regulatory mechanism for its activity.

## Effect of the Metal Ion in Activity Modulation

Metalloaminopeptidases are well known for their metal-dependent catalysis mechanisms. Proline-specific

human XPNPEP1 (hAPP1, 172-618), X-prolyl aminopeptidase from Caenorhabditis elegans (CeAPP, 174–616), and aminopeptidase P from Plasmodium falciparum (PfAPP, 305–774). Residues for chelating metal ion are highlighted using the red circle under the sequences; Residues contributed to S1 pockets are highlighted with a blue background and residues for S' sites are presented with a green background; The red color displayed residues or short sequences are non-conserved among prokaryotic and eukaryotic APPros, which are also proved to be important to APPros' function. Sequences were aligned using the program ClustalX (Li et al., 2008), and the alignment was presented using the online ESPript 3.0 server (http://espript.ibcp.fr/ESPript/ESPript/).

aminopeptidases such as APPro and prolidase prefer dinuclear Mn(II) cluster as cofactor (Wilce et al., 1998; Graham et al., 2005). The residues coordinating the two Mn atoms (Mn<sup>A</sup> and MnB) are highly conserved among APPro family members (**Figure 2**), and the non-equivalent roles of the two metal ions in catalysis have been extensively studied (Graham et al., 2005; Hu et al., 2007). To test the role of divalent metal cluster on Pa-PepP, we measured the relative activities of Pa-PepP in the presence of 1 mM Mn2+,Ca2+,Mg2+, Ni2+, Zn2+, and EDTA (**Figure 3A**). The addition of Mn2<sup>+</sup> significantly restored the Pa-PepP activity, while the limited enhancement of activity was observed upon addition of Ca2<sup>+</sup> or Mg2+. The exhibited basal activity may be a result of the partially pre-bound metal ions from the cell, and the limited effects of Ca2<sup>+</sup> and Mg2<sup>+</sup> may be due to their weak binding affinity to the active site. Previous studies on metal selection in Ec-PepP revealed that the Zn2<sup>+</sup> ion has high affinity for APPro and inhibits the hydrolysis reaction by occupying a third metal binding site (Graham et al., 2005; Hu et al., 2007). Although we could not measure the effect of Zn2<sup>+</sup> ion on Pa-PepP because of the formation of zinc hydroxide precipitates upon adding ZnCl<sup>2</sup> to the alkaline enzyme buffer, we observed the complete inhibition of enzyme activity in the presence of both the Ni2<sup>+</sup> ion and the chelating agent EDTA. These results suggested that the Ni2<sup>+</sup> may also possess high

affinity to APPro and tends to bind to those positions equivalent to Zn2<sup>+</sup> ions in Ec-PepP.

To identify the binding state of Mn2<sup>+</sup> ions in active site, 10 mM MnCl<sup>2</sup> was added during the crystallization experiment, along with 10 mM proline. In the final structure, there are three Mn atoms in each monomer (**Figure 4**). Mn<sup>A</sup> is liganded by the side chains of His354, Asp271, Glu384, Glu408, and Mn<sup>B</sup> interacts with Asp260, Asp271, Glu408. These interactions are highly conserved among those reported APPro structures that resulted in a distorted trigonal-pyramidal coordination network. The smaller anomalous signals and 0.5 occupancies of the Mn<sup>B</sup> site confirmed one of the structural features in metal-dependent aminopeptidase: the metal binding site B has relatively weaker binding affinity comparing to the site A (Hu et al., 2007). Additionally, the third Mn atom (MnC) is located in a position previously identified as being occupied by water in Zn-load Ec-PepP structures (Graham et al., 2005). The strong anomalous peak at this position verified its identity as a tightly bound Mn2<sup>+</sup> ion rather than a water molecule. It is interesting to note that this Mn<sup>C</sup> position is different from the third metal binding site found

in Ec-PepP, in which the Zn<sup>C</sup> is coordinated by His243, His361, and the hydrolysis product Pro residue (Graham et al., 2005). In Pa-PepP, Mn<sup>C</sup> interacts through one solvent molecular (W) with Mn<sup>A</sup> and MnB, with a metal-separation distance of 3.1–3.7 Å. The Pro residue lies immediately above the solvent molecular W but does not participate in the interaction with MnC. Remarkably, Mn<sup>C</sup> occupies the S1 subsite that ordinarily binds the substrate, providing an explanation for the inhibition effect of excessive metal binding. To verify the inhibitory effect of Mn2<sup>+</sup> ion, we performed an activity assay in the presence of different amounts of Mn2+. The result is consistent with our structural analysis (**Figure 3B**): there is a Mn2<sup>+</sup> ion concentration-dependent activity regulation pattern, in which 0.5–1 mM of Mn ion is

the optimal concentration for 20 nM (1 µg/ml) Pa-PepP, and inhibitory effects are observed when Mn2<sup>+</sup> ion exceeded 5 mM. Inhibition was not observed at ion concentrations up to 20 mM in the assays with Ca2<sup>+</sup> and Mg2<sup>+</sup> (**Supplementary Figure S1**).

## Conserved Active-Site Motif and Extended Binding Site for Specific Substrate Recognition

Several APPro structures have been reported to date, including APPro from Escherichia coli (Ec-PepP, 1M35) (Wilce et al., 1998), Streptococcus thermophilus (3IL0), Bacillus anthracis (3IG4), Yersinia pestis (4PV4), Thermotoga maritima (2ZSG), Streptococcus pyogenes (3OVK), human (hAPP1, 3CTZ) (Li et al., 2008), Caenorhabditis elegans (CeAPP1, 4S2R) (Iyer et al., 2015) and Plasmodium falciparum (PfAPP, 5JQK) (Drinkwater et al., 2016). Ec-PepP shares the highest sequence identity with Pa-PepP and both of them are prokaryotic APPro (Gonzales and Robert-Baudouy, 1996). All of these APPros share a common catalytic domain structure that contains a metal center flanked by the S1'-S1 pockets (**Figures 2**, **5A**). Previous structural studies on Ec-PepP in complex with the substrate or inhibitor have revealed those pockets (Graham et al., 2004; Graham and Guss, 2008). We modeled the Val-Pro-Leu bound Pa-PepP complex by superposing Pa-PepP structure with the substratebound Ec-PepP structure (PDB code: 2BN7) (**Figure 5A**). The shallow hydrophobic S1' pocket could be defined by the pre-bound Pro residue, Arg406, Tyr388, and His350. S1' sites are conserved in APPro and prolidase but not in other metalloaminopeptidases, which implies that they play critical roles in regulating the activity of proline-specific peptidases. Additionally, in prokaryotic APPros and prolidases, a tryptophan (Trp88' in Pa-PepP) from an adjacent subunit extends the boundary of S1' site and contributes to the P1' Pro binding (Yoshimoto et al., 1994; Jao et al., 2006). At the S1'-S1 junction, His243 and His361 interact with the main-chain carbonyl of P1' Pro and P1 Val, respectively, indicating their specific roles in recognizing and stabilizing the stereospecific scissile X-Pro bond. In the S1 pocket, the side chain of P1 Val faces Tyr229, Ile232, Arg245, while its main chain replaces the Mn<sup>C</sup> ion and is involved in the Mn<sup>A</sup> and Mn<sup>B</sup> ion mediated interactions. In metallomainopeptidases, the dinuclear Mn center is thought to be essential for nucleophilic attack (Graham et al., 2005). To verify the impacts of the critical motifs revealed by structural analysis, we constructed Ala substitutions on those motifs (Trp88 for S1' pocket, His243 for scissile bond stabilizing, Glu384 for metal binding) and verified that the expression level and stability are comparable between the wild type and all the mutants (**Supplementary Figure S2**). The Pa-PepP activity assay shows that all the mutations would eliminate enzymatic activity, which further confirmed that the evolutionary conservation in those regions is strongly associated with the essential functional roles of APPro family (**Figure 5B**).

In contrast to the prolidases that cleave dipeptides only, APPro is capable of modifying protein substrates. The relatively larger substrate binding ability of APPro is partly due to the extension of the substrate binding site beyond the S1'–S1 region (Drinkwater et al., 2016). The modeled tripeptide-bound Pa-PepP structure indicated that Arg153 and Arg351 accommodate the carboxylate group of the P2' Leu, and guide the C-terminus of the substrate toward the tetramer surface (**Figure 5A**). Furthermore, the R351G mutation results in an increase in enzyme activity. The similar Arg153/Arg370 basic patch is also found in Ec-PepP, in which the non-basic residue substitution on either of them leads to an obvious decrease in K<sup>m</sup> and an increase in kcat/K<sup>m</sup> (Jao et al., 2006). Thus, this basic patch, which is located at the entrance of catalytic cave, can play an important role in specificities of protein substrate recognition and orientation.

## Structural Analysis Identified a Surface Loop Participating in P. aeruginosa Virulence

The conserved structural properties in active site suggest a common catalytic mechanism for APPro family members, while the differences in the substrate-binding regions indicate the specific functions for each APPro member. Next to the P2' binding site, a surface loop (residue 365–377 in Pa-PepP) between helix α<sup>15</sup> and sheet β<sup>10</sup> was identified as a candidate for specific functions due to the diversity in its sequences between APPro members (**Figure 2**). To investigate its impact on enzyme catalytic activity, we mutated residues RVGGEW (Pa-PepP 367- 372) to GQDRS (Ec-PepP 368-372). Equivalent activities were observed between the corresponding mutant and the wild-type (**Figure 5B**). Since, in the activity assay, we used a non-specific tripeptide analog instead of the native substrates, these results may not accurately reflect the actual substrate-binding situation.

To further elucidate the impacts of Pa-PepP mutations on the bacterium virulence, three mutants with different effects on enzymatic activity were made in P. aeruginos PA14 for use in a Hela-cell invasion assay (**Figure 5C**). The PA14 bacterial growth in vitro was not affected by pepP deletion or any mutations in pepP (data not shown). The results showed that Pa-PepP knockout PA14 (PA14-1pepP) strain lost half of its ability to enter mammalian cells compared to wild-type PA14, and the cell invasion ability for PA14-1pepP strain was restored when introducing a complementary plasmid pRK415 for Pa-PepP expression, which confirmed the contribution of Pa-PepP to P. aeruginos virulence (Feinbaum et al., 2012). The enzymatic inactive mutant (H243A) exhibited an attenuated invasion ability, suggesting a role for the cleavage activity of Pa-PepP in its biological function. Strikingly, the PA14-R351G mutant also displayed weak cell internalization efficiency comparable to PA14-1pepP, revealing that the specificity of protein-substrate recognition is indispensable to the virulence-related function. The mutant PA14-Pa-PepP-367-372 also presented an attenuated invasion ability, verifying the critical role of 367-372 loop in the biological function of Pa-PepP, even though it does not have an obvious impact on enzymatic activity.

## DISCUSSION

Aminopeptidases hydrolyze the N-terminal residues of peptide/protein substrate and generally have broad specificity.

They are associated with various essential biological functions and have been suggested as suitable therapeutic targets for corresponding diseases (Griffith et al., 1997; Pasqualini et al., 2000; Gardiner et al., 2006; Stack et al., 2007; Trenholme et al., 2010; Skinner-Adams et al., 2010). There are a number of aminopeptidases inhibitors that have already been developed as potent drug candidates (Prechel et al., 1995; Maggiora et al., 1999; Aozuka et al., 2004; Krige et al., 2008). Given the rare occurrence of proline as the second residue in protein, the proline-specific APPro exhibits even narrower substrate specificities, making it an attractive target for specific therapeutic application. In Feinbaum et al.'s (2012) C. elegans model, aminopeptidase PepP gene was identified as a virulence related factor and was involved in the highly attenuated C. elegans deaths, which suggested that Pa-PepP may be a potential drug target for the treatment of P. aeruginosa infections. However, other infection models are still needed to further confirm the role of pepP in P. aeruginosa virulence, and mechanisms account for Pa-PepP's correlation with bacterium virulence remain unknown. In the C. elegans model, the PA14 strain with transposon insertions in gene pepP display increased pyocyanin levels, whereas pyocyanin has been demonstrated to play an important role in

P. aeruginosa virulence in many models of infections (Lau et al., 2004; Dietrich et al., 2006). Thus, we predict that there may be other mechanisms linked to the effect of pepP on the virulence associated phenotypes. In this work, we analyzed the specific substrate binding site of Pa-PepP based on structural studies, as well as its consensus signatures and mechanisms, which provided clues for better understanding its unique physiologic function.

Like all other metalloaminopeptidase, APPro requires metal ion for its enzyme activity, and the manganese might be the most preferred cation. The dinuclear metal center structure was consistent among all APPro members, with the M<sup>B</sup> ion binds more loosely than MA. Different anomalous map peaks of the bound Mn in Pa-PepP structure verified the inequivalent occupancy of two metal-binding sites. It should be noted that the different binding affinities of the metal sites allow for the sequential addition of multiple types of metal ions in other metalloaminopeptidases, resulting in diverse enzymatic activities (Carpenter and Vahl, 1973; Van Wart and Lin, 1981; Allen et al., 1983; Bayliss and Prescott, 1986; Lin and Van Wart, 1988). Stimulatory or inhibitory effects of various divalent metal ions were also observed in both Ec-PepP and Pa-PepP, suggesting that the metal selectivity contributes to activityregulating mechanisms in APPro. Furthermore, the additional M<sup>C</sup> atom found in Pa-PepP provides a structural basis for the manganese concentration-dependent modulation of activity. In conclusion, the delicate structure of the metal binding sites enables the APPro to precisely sense the type and concentration of metal ions, facilitating its scalability in specific function performances.

The individual domain superposition of the APPros shows a close agreement between the catalytic domain structures (rmsd values of 0.847–1.846) and the nearly identical active sites allow them to maintain the unique catalytic mechanism. However, the extensive sequence diversity between prokaryotic and eukaryotic APPros, especially with respect to the domain assignments, domain motions, and oligomeric assemblies, are the breakthrough points for understanding the particular functions of each member (**Supplementary Figure S3**). For instance, the antiparallel dimerization observed in prokaryotic APPro resulted in an inward extending loop region that reached into the cavity of the neighboring subunits and participated in the P1' residue binding. Instead, in eukaryotic species, this part was replaced by the additional N-domain to ensure the maintenance of the active site pocket (Li et al., 2008). Moreover, those different oligomerization interfaces may affect the substrate selectivity in APPros.

It has been proposed that the tetrameric assembly is indispensable for Ec-PepP's function because the increase of the binding surface area can accommodate larger protein substrates (Yoshimoto et al., 1994). Pa-PepP possesses more positively charged groups on its solvent-accessible area compared to Ec-PepP (**Supplementary Figure S4**), implicating a preference for substrate selection. Notably, the 367:372 loop identified from Pa-PepP is located at the entrance region of the catalytic cave and replacement of this loop by corresponding fragment from Ec-PepP attenuated the cell invasion ability of PA14. These observations revealed that residue variations and charge distribution differences around the substrate binding interfaces can lead to distinct functions for Pa-PepP and Ec-PepP, even though they shared high sequence similarity.

## CONCLUSION

APPros exert specific physiologic functions, and many of them are relate to virulence in bacteria. However, due to the insufficiency of information about its biological substrate, the knowledge of its biology functions are limited for now. Here, we discussed the ion modulation mechanism in Pa-PepP and unveiled the differences in the regions surrounding substrate binding site, which extend our understanding about the catalytic mechanism and virulence-related functions of Pa-PepP. Most importantly, the structural analysis provides us a solid foundation for designing specific inhibitors against pathogenic bacterial infection by blocking the particular protein substrate binding site instead of directly interfering the common catalytic center.

## MATERIALS AND METHODS

## Protein Expression and Purification

Full-length pepP was amplified by PCR using gene-specific primers (Supplementary Table S1) from the P. aeruginosa PA14 genome DNA on Gene amplification Machine (Gene Touch, BIOER, HangZhou, China). Full-length pepP containing six C-terminal histidine residues (LEHHHHHH) was homologous recombined with the linearized pET-22b (+) using a ClonExpressTM II One Step Cloning Kit (Vazyme). All point mutants were generated using the QuickChange (I-5TM2∗High Fidelity Master Mix, MCLAB) PCR-based method, on the pET-22b (+) construct. The recombinant plasmid was transformed into E. coli strain BL21 (DE3) for protein expression. The bacterial culture was grown in LB medium in the presence of 100 µg mL−<sup>1</sup> ampicillin and incubated with shaking at 310 K until the OD600 reached 0.9 (ZhiChu, ShangHai). The culture was cooled to 289 K before protein expression was induced with 0.2 mM IPTG for 20 h (Bao et al., 2013). Following induction, the bacteria were collected and resuspended in a lysis buffer consisting of 25 mM Tris–HCl pH 8.5, 10 mM NaCl, 5 % glycerol and 1 mM phenylmethanesulfonyl fluoride (20 g of cells/100 mL buffer, Sigma–Aldrich) and lysed by sonication. The lysate was cleared by centrifugation at 11000 g for 45 min and then the supernatant was loaded onto a 2 mL Ni–NTA affinity resin (Qiagen) for 2 L culture. The Ni–NTA column was washed with ten column volumes of the lysis buffer supplemented with 20 mM imidazole. The target protein was eluted with the same buffer in the presence of 200 mM imidazole. Fractions were pooled and determined by SDS–PAGE, followed by further purification on size-exclusion chromatography Superdex 200 column (GE Healthcare), which was pre-equilibrated with the buffer consisting of 25 mM Tris– HCl pH 8.5, 10 mM NaCl. Fractions containing Pa-PepP were pooled and concentrated to a concentration of approximately

16 mg mL−<sup>1</sup> using a Centricon filter (10 kDa cutoff; Millipore, Billerica).

## Crystallization and Data Collection

Initial crystallization experiments were carried out using four commercial crystallization screens from Hampton Research and Rigaku (Index HT, Crystal Screen HT, WIZARD HT, XTAL QUEST HT). Crystallization screens were conducted as previously described with some modifications (Bao et al., 2009). Briefly, crystallization initially screens were carried out using a Mosquito liquid dispenser by hanging-drop vapor-diffusion method at 291 K in 96-well plates. The 200 nL mixing drop containing protein solution and reservoir buffer by 1:1, with a final protein concentration of 16 mg ml−<sup>1</sup> in 25 mM Tris– HCl pH 8.5, 10 mM NaCl. The final optimized crystals for the recombinant protein were obtained by mixing 2 µl of the protein sample containing Mn2<sup>+</sup> and proline with an equal volume of the reservoir solution containing 30% PEG400, 100 mM sodium cacodylate pH 6.5, 200 mM lithium sulfate. Crystals grew in approximately 2–3 days and were transferred to a cryo-protectant solution (reservoir solution with 6% PEG400) prior to flashcooling in liquid nitrogen. X-ray data were collected with a CCD camera on BL-17U stations of the SSRF, China. The diffraction data were indexed, integrated, and scaled using the HKL2000 program suite (Otwinowski and Minor, 1997).

## Structure Determination and Refinement

Data processing and scaling were carried out using the HKL2000 software package. The data were processed to a resolution limit of 1.847–1.783 Å (Rmerge = 0.099) in space group P2, with unit-cell parameters a = 111.197, b = 123.432, c = 149.485 Å. The phase problem was solved by molecular replacement using PHENIX with aminopeptidase P from E. coli (PDB entry 1az9; space group P 6422; resolution 1.9 Å) as a template. The process of structure building and refinement was monitored using the COOT (Emsley et al., 2010). Water molecules were automatically added by PHENIX (Adams et al., 2010).

## Differential Scanning Calorimetry (DSC) Assays

Differential scanning calorimetry (DSC) is a promising thermoanalytical technique used for evaluating the stability of proteins as well as other biomolecules. The enthalpy value determined by DSC can provide direct information about the energetics of thermally induced processes, and the measured melting temperature(Tm) reflects thermal stability of proteins. The assays were conducted on the MicroCal VP-Capillary DSC System (Malvern). Protein samples were prepared with a final concentration of 1 mg/ml in 400 µl buffer and each sample has another 400 µl blank buffer as reference which were loaded in pairs. Operating parameters (such as pre-scan equilibration time, scan rate, temperature programming) were set and at least three buffer-buffer scans were performed before the sample was scanned. And DSC automated data analysis was conducted after experiments ran out. The DSC curves can read by Origin (Origin Pro 7.5).

## Construction of P. aeruginosa pepP Gene Mutant

To construct pepP mutants of P. aeruginosa, a two-step allelic exchange bacterial genome engineering strategy was employed (Hmelo et al., 2015; Sana et al., 2015). Briefly, in the first step of allelic exchange, the suicide vector pEX18Gm was integrated sitespecifically into the chromosome of P. aeruginosa by homologous recombination, resulting in antibiotic-resistant single-crossover mutants. Then, in the second step of allelic exchange, a double crossover event occurred through a second homologous recombination, and the mutant was isolated using sucrosemediated counter-selection. The corresponding mutants were finally identified by PCR and DNA sequencing. Specifically, PCRs were performed to amplify the target fragment sequences with upstream (800 bp) and downstream (800 bp) from P. aeruginosa chromosomal DNA, while the suicide plasmid pEX18Gm was linearized with gene-specific primers. The two PCR products were recombined with ClonExpress <sup>R</sup> II One Step Cloning Kit (Vazyme), The resulting plasmid, pEX18-Gm-pepP, was then performed Site-directed mutagenesis or deletion. All these primers are listed in Supplementary Table S1. These vectors were then transformed into E. coli S17-1 and then mobilized into P. aeruginosa strains PA14 by conjugation, in order to transfer suicide plasmids from the E. coli donor S17-1 to the P. aeruginosa recipient PA14. Colonies were first screened using gentamicin resistance plates to get single-crossover mutants. And then the double-crossover mutants were screened by Nosalt LB (NSLB) agar with 15% (wt/vol) sucrose. The pepP gene replacement mutant strain was further confirmed by PCR and DNA sequencing.

## Construction of Complementation Plasmid pRK415-pepP

To construct the complementation plasmid pRK415-pepP, PCRamplified pepP was cloned into the EcoRI and HindIII sites of plasmid pRK415, giving rise to the plasmid pRK415- pepP (Lin et al., 2017). The recombinant plasmid and plasmid pRK415 were transformed into E. coli S17-1, respectively, and then mobilized into P. aeruginosa PA14-MpepP by conjugation to transfer pRK415-pepP from the S17-1 to PA14-MpepP. The PA14- MpepP strain carried pRK415-pepP or pRK415 plasmid were screened by Pseudomonas isolation agar (PIA) with 150 µg/ml tetracycline. For expression of PepP in the P. aeruginosa strains, the PA14-MpepP strain carried pRK415-pepP were induced by addition of 1 mM IPTG and then conducted cell invasion assays.

## HeLa Cell Invasion Assays

To enumerate bacteria internalized by HeLa cells to verify bacterium virulence, gentamicin survival assays were conducted with slight modifications (Chi et al., 1991). Briefly, mammalian HeLa cells (obtained from ATCC) were grown in DMEM medium, containing 10% (v/v) FBS (Gibco, Auckland, New Zealand) and 1 % antibiotics (penicillin and streptomycin) in 5% CO<sup>2</sup> at 37◦C. Suspension cultures of HeLa cells were seeded at 2–5 × 10<sup>5</sup> cells per well in 12-well tissue culture plates for the overnight at 37◦C. Cells were washed three times with PBS (pH 7.2) and changed to antibiotic-free medium immediately before infection. Cells were infected with exponential phase P. aeruginosa strains PA14 and mutant strains at a MOI of 10 for 1 h in 5% CO<sup>2</sup> at 37◦C. The cells were washed twice with PBS and incubated for an additional 1 h in DMEM medium containing 150 µg/ml of gentamicin in order to kill extracellular bacteria. The monolayers cells were then washed three times with PBS and lysed with 0.5% Triton X-100 for 10–20 min, and appropriate dilutions were plated on LB plates to determine the number of viable intracellular bacteria.

### Enzyme Activity Assays

fmicb-08-02385 December 5, 2017 Time: 12:6 # 10

The enzyme activity of wild type Pa-PepP and the mutants were spectrophotometrically determined using the quenched fluorescent substrate Lys (Abz) – Pro–Pro–pNA (synthesized by GL Biochem Shanghai Ltd., China) as the substrate. Prior to the assay, wild type Pa-PepP and the mutant enzyme are dialyzed with EDTA overnight to fully remove the pre-bound metal ions. And all enzyme preparations were freshly diluted with ice-cold 50 mM Tris (pH 8.5) and 100 mM NaCl and treated with MnCl<sup>2</sup> for 10 min at 37◦C. After addition of substrate, the final reaction conditions were 50 mM Tris (pH 8.5), 100 mM NaCl, 1 mM MnCl2, 0–250 µM substrate, and 20 nM Pa-PepP (final assay volume 100 µL). The assay was allowed to proceed for 5 min at 37◦C in the Thermo Scientific Varioskan Flash plate reader. The appearance of fluorescent product (λex = 301 nm, λem = 410 nm) was monitored at 10 s intervals. Since the pure fluorescent product was not available to quantitate the changes in fluorescence, the kinetic results are presented as activities in relative fluorescence units. The kinetic parameters Km (Michaelis constant) for all the assays were obtained by fitting experimental data to the Michaelis-Menten equation by non-linear regression using the program OriginPro 7.5 (OriginLab Software). The data were showed in Supplementary Tables S2 and S3.

### Statistical Analysis

Statistical analysis was analyzed by 2-tailed Student's t-test. In all statistical analysis, P-values < 0.05 were considered to be statistically significant.

### ACCESSION NUMBERS

Atomic coordinates of the refined structures have been deposited in the Protein Data Bank (www.pdb.org) with the PDB code 5WZE.

## AUTHOR CONTRIBUTIONS

C-TP performed the experiments and wrote this manuscript. LL planned experiments and analyzed data. C-CL conducted structural analysis. L-HH made modifications of the manuscript. TL performed the structure determination and refinement. Y-LS performed the HeLa cell invasion assays. CG planned the experiments. N-YW contributed the essential experiment materials. YX guided the experiments. Y-BZ constructed the P. aeruginosa pepP gene mutants. Y-JS also performed the HeLa cell invasion assays. QL made modifications of the manuscript and conducted the PCR experiments. L-TY guided the experiments and made modifications of the manuscript. RB wrote part of the paper and performed the drawing.

## FUNDING

The work was supported by National Key Research and Development Plan (Grant No. SQ2016YFJC040104), National Natural Science Foundation of China (Grant Nos. 81670008 and 81501787).

## ACKNOWLEDGMENTS

We thank Shanghai Synchrotron Radiation Facility (SSRF) beamline BL17U for beamtime allowance (Wang et al., 2015). We thank the staffs of National Center for Protein Sciences Shanghai (NCPSS) beamlines BL18U and BL19U and SSRF, Shanghai, People's Republic of China, for assistance during data collection. We thank Dr. Derek Holman for carefully reading the manuscript.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2017. 02385/full#supplementary-material

FIGURE S1 | Relative activity of Pa-PepP in presence of different amount of Ca2<sup>+</sup> (A) and Mg2<sup>+</sup> (B). All assays were performed at 37◦C for 5 min in the presence of 50 mM Tris (pH 8.5), 100 mM NaCl, 50 µM Lyz(Abz)–Pro–Pro–pNA quenched fluorescent substrate, and 1 µg/mL−<sup>1</sup> Pa-PepP that had been incubated with different concentrations of Ca2<sup>+</sup> (A) and Mg2<sup>+</sup> (B) at 37◦C for 10 min. Each bar represents the mean of three independent measurements (SEM).

FIGURE S2 | The differential scanning calorimetry (DSC) curves of wide type Pa-PepP and all the mutants. DSC assays were conducted to investigate the thermal stability of wild type Pa-PepP and the mutants. The enthalpy value and the melting temperature(Tm) value determined by DSC were showed in the DSC curves. As it showed in the DSC curves, all these protein samples shared the similar Tm value (66–68◦C), with the only exceptional of Pa-PepP-E384A which has a acceptable shift of Tm value of 63◦C.

FIGURE S3 | Superposition of the monomer Pa-PepP and other APPros. (A) Superposition of the monomer Pa-PepP (red) with Ec-PepP (blue), hAPP1 (light gray), PfAPP (light blue), and CeAPP (light pink). The N domain and C domain of each APPro were circled with black frame and circle, respectively. (B) Superposition of the monomer Pa-PepP (red) with Ec-PepP (blue). The black circle refers to the non-conservative surface loop identified as an important motif in Pa-PepP's function. (C) Schematic diagrams for the prokaryotic (right) and eukaryotic APPros dimers (left). N and C are labels for their N-terminal and C-terminal domains, respectively. And the plus N-terminal domain is labeled as N'.

FIGURE S4 | Surface charge distribution of Pa-PepP and Ec-PepP. Surface charge distribution of Pa-PepP (A) and Ec-PepP (B) generated by CCP4mg (McCoy et al., 2007). Pa-PepP possesses more positively charged groups on its solvent-accessible area compared to Ec-PepP. The green circle refers to the predicted substrate entrance of Pa-PepP and Ec-PepP, and the entrance is also indicated by the red arrow near the black colored loop in the ribbon view of tetramer (C).

## REFERENCES

fmicb-08-02385 December 5, 2017 Time: 12:6 # 11


aminopeptidase P: insights into substrate recognition and the mechanism of catalysis. Biochemistry 45, 964–975. doi: 10.1021/bi0518904



targets for anti-malarials. Trends Biochem. Sci. 35, 53–61. doi: 10.1016/j.tibs. 2009.08.004


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Peng, Liu, Li, He, Li, Shen, Gao, Wang, Xia, Zhu, Song, Lei, Yu and Bao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Repurposing and Revival of the Drugs: A New Approach to Combat the Drug Resistant Tuberculosis

Divakar Sharma1,2 \*, Yogesh K. Dhuriya<sup>3</sup> , Nirmala Deo<sup>1</sup> and Deepa Bisht<sup>1</sup>

<sup>1</sup> Department of Biochemistry, National JALMA Institute for Leprosy and Other Mycobacterial Diseases, Agra, India, 2 Interdisciplinary Biotechnology Unit, Aligarh Muslim University, Aligarh, India, <sup>3</sup> Developmental Toxicology Laboratory, Systems Toxicology and Health Risk Assessment Group, CSIR-Indian Institute of Toxicology Research (CSIR-IITR), Lucknow, India

Emergence of drug resistant tuberculosis like multi drug resistant tuberculosis (MDR-TB), extensively drug-resistant tuberculosis (XDR-TB) and totally drug resistant tuberculosis (TDR-TB) has created a new challenge to fight against these bad bugs of Mycobacterium tuberculosis. Repurposing and revival of the drugs are the new trends/options to combat these worsen situations of tuberculosis in the antibiotics resistance era or in the situation of global emergency. Bactericidal and synergistic effect of repurposed/revived drugs along with the latest drugs bedaquiline and delamanid used in the treatment of MDR-TB, XDR-TB, and TDR-TB might be the choice for future promising combinatorial chemotherapy against these bad bugs.

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Sandeep Sharma, University of Pennsylvania, United States Shashank Gupta, Brown University, United States

### \*Correspondence:

Divakar Sharma divakarsharma88@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 29 September 2017 Accepted: 27 November 2017 Published: 11 December 2017

### Citation:

Sharma D, Dhuriya YK, Deo N and Bisht D (2017) Repurposing and Revival of the Drugs: A New Approach to Combat the Drug Resistant Tuberculosis. Front. Microbiol. 8:2452. doi: 10.3389/fmicb.2017.02452 Keywords: drug resistance tuberculosis, repurposing, revival of drugs, synergistic effect, proteomics and bioinformatics

## INTRODUCTION

### Current Scenario

Mycobacterium tuberculosis is a deadly infectious pathogen causing tuberculosis (TB) worldwide. According to WHO (2016) 10.4 million new cases with 1.5 million deaths including 0.4 million individuals with HIV-TB co-infection were reported globally (WHO, 2016). Although TB can be cured by chemotherapy, but the emergence of drug resistant tuberculosis [such as multidrug-resistant tuberculosis (MDR-TB), extensively drug-resistant tuberculosis (XDR-TB) and totally drug resistant tuberculosis (TDR-TB)] has created a new challenge to combat the adverse situation of the disease. Since the decade rates of antibiotic resistance to first and second line anti-TB drugs are dramatically increasing (Cambau et al., 2015). Due to various complexities and high burden of HIV-TB co-infection treatments of MDR-TB, XDR-TB, and TDR-TB are problematic. Apart from the genomics studies various proteomics as well as bioinformatics studies regarding to the drug resistance tuberculosis were accumulated in the last decade (Jiang et al., 2007; Zhang and Yew, 2009; Sharma et al., 2010, 2014, 2015a,b, 2016a,b, 2017; Kumar et al., 2013; Lata et al., 2015a,b; Singh et al., 2015; Sharma and Bisht, 2016, 2017a,b,c;) and suggested the novel drug resistance mechanisms/markers/targets for potential therapeutics in near future. Proteomics and bioinformatics play a crucial role in the diagnostics and therapeutics against the emerging bad bug of tuberculosis. Apart from the expression proteomics studies cited above we have reported that Rv0148 (Putative shortchain type dehydrogenase/reductase) and Rv3841 (ferritin protein) have potentially involved in aminoglycosides resistance (Sharma et al., 2015b, 2016a) and could be the potential diagnostics and

therapeutics against the second line of drug resistance. To combat these worsened situations (MDR-TB, XDR-TB, and TDR-TB) new and effective drugs and diagnostics are urgently needed. However, as we know that development of new drugs and diagnostics are enormously expensive and time consuming process.

## Repurposing and Revival of Drugs: A Chemotherapeutic Option against Drug Resistant Tuberculosis

Repurposing and revival of the drugs are the new trends/options or one of the pharmaceutical strategies to treat the particular disease that are already FDA approved for other diseases (Tsukamura, 1980; Van Deun et al., 2010) and also for tuberculosis earlier, in this antibiotics resistance era or in the situation of global emergency. Many compounds in TB advanced clinical trials are the molecules that were formerly used to treat other infectious diseases/tuberculosis earlier and now they have been repurposed for treatment of TB (Nayer and Steinbach, 1939; Tsukamura, 1980; Van Deun et al., 2010; Hasse et al., 2014).

Sulfonamides and sulphanilamide were first used in 1930s as an anti-TB drug (Nayer and Steinbach, 1939) but its use was discontinued due to its lesser efficacy as compared to first line drugs (streptomycin and isoniazid). Revival of sulfamethoxazole (SMX) in TB was first pointed out by its efficacy to prevent the TB infection in HIV patients whose receiving trimethoprim/sulfamethoxazole (TMP/SMX) to prevent other infection such as Pneumocystis jirovecii (Hasse et al., 2014). In a Nigerian trial study on patients of HIV-MDR-TB coinfetion, efficiency of MDR-TB treatment by TMP/SMX confirmed a significantly shorter time to sputum conversion in these patients (Oladimeji et al., 2014). Sulfadiazine is an anti-leprosy drug which is repurposed in the treatment of MDR-TB and XDR-TB. Brouqui et al. (2013) suggested that sulfadiazine regimen is safe and effective against MDR-TB and TDR-TB treatment (Ameen and Drancourt, 2013; Brouqui et al., 2013).

Clofazimine (CZM) is one of the repurposed molecules, has been initially used as an anti-leprosy drug since half the century. It was recently repurposed for managing the treatment of MDR-TB (Van Deun et al., 2010). CZM is recommended as a secondline anti-TB drug and used in combination with other anti-TB drugs for the treatment of drug-resistant tuberculosis. CZMcontaining regimen can cure MDR-TB cases in 9–12 months. In M. tuberculosis, CZM appears to act as a prodrug, which is reduced by type 2-NADH dehydrogenase, to release reactive oxygen species (ROS) upon reoxidation by oxygen (O2) (Yano et al., 2011). CZM, exhibits noticeable anti-mycobacterial and anti-inflammatory activity by inhibition of phospholipase and effects on potassium transporters, respectively (Steele et al., 1999; Cholo et al., 2006). Previous published studies have reported that CZM has good quality efficacy and little toxicity against drug-resistant mycobacterial strains in animal models, which suggested, CZM as a promising anti-TB drug for the management of MDR-TB (Van Deun et al., 2010; Cholo et al., 2012). Recently, numerous observational studies have reported that CZM including regimens provided a useful role in the treatment of patients with MDR-TB (Cholo et al., 2006; Van Deun et al., 2010).

Linezolid, an oxazolidinone antibiotic which is used for treatment of gram-positive bacterial infections (Till et al., 2002; Yanagihara et al., 2002), has now potentially repurposed for the treatment of drug resistant TB (MDR-TB and XDR-TB) (Fortún et al., 2005). Linezolid is an effective anti-TB drug for treating MDR-TB and XDR-TB with various side effects such as neurotoxicity and hematologic toxicity (Tang et al., 2015). Schecter et al. (2010) suggested that linezolid had low rates of discontinuation, well tolerated and good efficacy in the treatment of MDR-TB. Most recently in a case study Jaspard et al. (2017) reported, bedaquiline and linezolid drug combination might be safe for XDR-TB in the late third trimester of pregnancy or pregnant woman. Pregnant woman gave birth to a child without abnormalities follow-up of the fetal showed that no fetal toxicities upto 2 years after the delivery (Jaspard et al., 2017).

Minocycline is also one of the repurposed molecules, has been initially used in the treatment of leprosy since the 1980s (Tsukamura, 1980). In 2008 it was repurposed for managing the treatment of XDR-TB patient in Japan (Kawada et al., 2008). Combinatorial therapy of amoxicillin/clavulanic acid along with other second-line drugs has been used in the treatment of MDR-TB. It's cheaper cost and less toxicity has made the drug of choice in WHO group five drugs (Cassir et al., 2014). Recently, combinatorial treatment by amoxicillin/clavulanic acid and carbapenems has reduced the M. tuberculosis load (Diacon et al., 2016). Hugonnet et al. (2009) reported the in vitro activity of meropenem combined with clavulanate against XDR strains and paying attention to repurpose these beta-lactams as new anti-TB drugs (Hugonnet et al., 2009). However, carbapenems have been used successfully as part of salvage therapies for XDR patients, they have to be administered intravenously (Tiberi et al., 2016). Recently, an early bactericidal activity-Phase II (EBA Phase II) clinical trial has validated the promising potential of a carbapenem combined with amoxicillin and clavulanic acid for TB treatment (Diacon et al., 2016).

Singhal et al. (2014) reported that the FDA-approved antidiabetic drug metformin (MET) inhibits the intracellular mycobacterial growth by inducing mitochondrial ROS production, restricts disease immunopathology, enhances the efficacy of other anti-TB drugs and could be used as combinatorial therapy against the drug resistant tuberculosis. Gupta et al. (2013) suggested that tuberculosis treatment shortening by verapamil as an adjunctive therapy in mice has opened the direction for future research on verapamil and other efflux pump inhibitors in human tuberculosis. Gupta et al. (2014) also reported that efflux inhibition with verapamil tremendously decreases the MIC of bedaquiline and CZM to M. tuberculosis and suggested the synergistic effects of verapamil and bedaquiline in an animal model of TB infection. Recently Dutta et al. (2016) reported that statin a lipid-lowering drug (repurposed for TB treatment) when added to the first-line antitubercular drugs, reduces the lung bacillary burden in chronically infected mice.

Host-directed therapies are important adjunctive therapies for tuberculosis treatment that expanded by host immune effector mechanisms. Recently Gupta et al. (2017) showed that denileukin diftitox potentiates standard TB treatment in the mouse model, which might be due to depletion of T-regulatory and myeloidderived suppressor cells during TB infection.

## Synergistic Effects of Repurposed/Revived Drugs: With Others Anti-TB Drugs Used in Treatments of Tuberculosis Bad Bug

Synergistic effect of repurposed/revived drugs with other anti-TB drugs has been used for the future selection of these drugs in the WHO regimen which could be used in the treatment of MDR-TB, XDR-TB, and TDR-TB. In a study Zhang et al. (2015) suggested that CZM in combination with ethambutol (EMB) and moxifloxacin (MOX) may be a potential drug regimen for the treatment of MDR-TB. Synergistic effect of SMX has been reported in vitro with rifampicin (Macingwana et al., 2012). Tasneen et al. (2011) suggested CZM was the best third drug in combination with bedaquiline and pyrazinamide for the treatment of MDR-TB and a good example of drug synergism. Partial synergistic effect was oveserved between linezolid and capreomycin and suggested the efficacy of this combinatorial therapy against M. tuberculosis (Zhao et al., 2016). Most recently synergistic effect of bedaquiline and linezolid combinatorial therapy has been reported in XDR-TB of pregnant woman which is a good symbol of synergy for the last line of drugs (Jaspard et al., 2017). In a study synergistic effect carbapenems with rifampicin have been reported against M. tuberculosis (Kaushik et al., 2015).

### REFERENCES


## CONCLUSION AND FUTURE PERSPECTIVE

On the basis of the reported literature and studies (regarding to proteomics, bioinformatics, and repurposing/revival of drugs) cited above we suggest that proteomics and bioinformatics play a crucial role in the exploration of diagnostics, therapeutics and mechanism of resistance against drug resistance tuberculosis. Repurposing and revival of the drugs is an alternative strategy to combat from this inadequate situation of drug resistance tuberculosis in this antibiotic resistance era. Synergistic effect of repurposed/revived drugs (sulfamethoxazole, sulfadiazine, clofazimine, linezolid, minocycline, amoxicillin/clavulanic acid, and carbapenems like meropenem) along with the latest drugs (bedaquiline and delamanid) used in the treatment of MDR-TB and XDR-TB might be one of the promising combinatorial chemotherapy for the treatment of MDR-TB, XDR-TB, and TDR-TB. Synergistic effect of these repurposed/revived drugs with bedaquiline and delamanid combinatorial therapy could be added a different perspective over the existed literature.

## AUTHOR CONTRIBUTIONS

DS design the concept and wrote the manuscript. DS, YD, ND, and DB finalized the manuscript.

## ACKNOWLEDGMENTS

The authors are grateful to Director, NJIL & OMD for the support. DS is ICMR-PDFs (ICMR, New Delhi).

Tuberculosis–new trick for an old dog? N. Engl. J. Med. 375, 393–394. doi: 10.1056/NEJMc1513236



potential drug targets. PLOS ONE 10:e0139414. doi: 10.1371/journal.pone.01 39414


Zhang, Y., and Yew, W. W. (2009). Mechanisms of drug resistance in Mycobacterium tuberculosis. Int. J. Tuberc. Lung Dis. 13, 1320–1330.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Sharma, Dhuriya, Deo and Bisht. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Harnessing the mTOR Pathway for Tuberculosis Treatment

Pooja Singh and Selvakumar Subbian\*

Public Health Research Institute at New Jersey Medical School, Rutgers Biomedical and Health Sciences Rutgers, The State University of New Jersey, Newark, NJ, United States

Tuberculosis (TB) remains as one of the leading killer infectious diseases of humans. At present, the standard therapeutic regimen to treat TB comprised of multiple antibiotics administered for a minimum of six months. Although these drugs are useful in controlling TB burden globally, they have not eliminated the disease. In addition, the lengthy duration of treatment with multiple drugs contributes to patient non-compliance that can result in the development of drug resistant strains (MDR and XDR) of Mycobacterium tuberculosis (Mtb), the causative agent of TB. Therefore, new and improved therapeutic strategies are urgently needed for effective control of TB worldwide. The intracellular survival of Mtb is regarded as a cumulative effect of the host immune response and the bacterial ability to resist or subvert this response. When the host innate defensive system is manipulated by Mtb for its survival and dissemination, the host develops disease conditions that are hard to overcome. The host intrinsic factors also contributes to the poor efficacy of anti-mycobacterial drugs and to the emergence of drug resistance. Hence, strengthening the immune repertoire involved in combating Mtb through host-directed therapeutics (HDT) can be one of the approaches for effective bacterial killing and clearance of infection/disease. Recently, more scientific research has been focused toward HDT strategies that empowers host cells for effective killing of Mtb, reduce the duration of treatment and/or alleviates the development of MDR/XDR, since Mtb cannot develop resistance against a drug that targets the host cell function. Autophagy is a conserved cellular process critical for maintaining cellular integrity and function. Autophagy is regulated by multiple pathways that are either dependent or independent of mTOR (mechanistic target of rapamycin; a.k.a. mammalian target of rapamycin), a master regulatory molecules that impacts several cellular functions. In this review, we summarize the role of autophagy in Mtb pathogenesis, the mTOR pathway and, modulating the mTOR pathway with inhibitors as potential adjunctive HDT, in combination with standard anti-TB antibiotics, to improve the outcome of current TB treatment.

Keywords: mTOR, autophagy, everolimus, host directed therapy, tuberculosis, drug resistance, adjunct therapy, phagocytosis

## INTRODUCTION

Tuberculosis (TB) is one of the leading killer among infectious diseases of humans, accounting for about 10.4 million new cases and 1.8 million deaths in 2015 (World Health Organization, 2016). The global burden of TB has also been exacerbated by other co-morbid conditions, including diabetes and HIV-infection, and TB is a leading cause of mortality among HIV infected

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Osmar Nascimento Silva, Universidade Católica Dom Bosco, Brazil Vikram Saini, University of Alabama at Birmingham, United States

> \*Correspondence: Selvakumar Subbian

subbiase@njms.rutgers.edu

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 23 October 2017 Accepted: 11 January 2018 Published: 30 January 2018

### Citation:

Singh P and Subbian S (2018) Harnessing the mTOR Pathway for Tuberculosis Treatment. Front. Microbiol. 9:70. doi: 10.3389/fmicb.2018.00070

**45**

individuals with nearly 400,000 deaths reported in 2015. The standard therapeutic regimen recommended by the WHO for treating drug-sensitive pulmonary TB, known as DOTS (Directly Observed Treatment, Short course), is comprised of four antibiotics: isoniazid (INH), rifampicin (RIF), pyrazinamide (PZA) and ethambutol (ETH) for 2 months (initial phase) followed by INH, and RIF for 4 months (continuation phase). This multi-drug regimen is essential and necessary to ensure successful bacteriological cure in patients with TB. Although these drugs are useful in controlling the overall disease burden at the level of individual patients as well as global TB control measures, they have not eliminated the disease at both these levels (Ryan, 1992). This is in part due to the lengthy duration of treatment with multiple drugs, which promotes the fear of drug dependency and doubts of not getting cured and contributes to drug-induced tissue toxicity issues. Adverse effects, ranging from serious ones, like hepatitis and pneumonia, to minor ones, like vomiting, acne and nausea, have been reported to be associated with DOTs therapy (Michael et al., 2016). Thus, high dropout rate of TB patients from treatment regimens (a.k.a. patient non-compliance) is a serious issue contributing directly to the development of drug resistance in Mycobacterium tuberculosis (Mtb), the causative agent of TB.

Development of drug resistance in a single bacterium has been suggested to be sufficient to create an outbreak of drug resistant bacteria (Borrell and Gagneux, 2009). In 2015, nearly 4.8 million cases of isoniazid- and rifampicin-resistant [a.k.a. multidrugresistant TB (MDR-TB)] cases were reported. In addition to INH and RIF (the first line drugs), Mtb can develop resistance to PZA and ETH (second line drugs) and other injectable aminoglycosides, leading to extensively drug-resistant TB (XDR-TB) cases. Nearly 9.5 % of all MDR-TB cases in 2015 were estimated to be XDR-TB. A recent study aimed at predicting the future burden of TB suggests an increased prevalence of MDR and XDR cases due mainly to person-to-person transmission of drug-resistant Mtb, rather than the pathogen acquiring drug resistance within the infected host (Sharma et al., 2017). Hence, current treatment strategies demand intense patient monitoring during and after drug treatment, which poses major strategical and economical challenges for the global TB control programs conducted by various health agencies. Therefore, it is imperative that new anti-TB therapies are developed and implemented to shorten the number of antibiotics taken and/or duration of treatment, to lower the drug- induced toxicities, and to improve the drug efficacy among TB patients with co-morbid conditions, such as HIV-infection and/or patients with MDR/XDR-TB.

Development of drug resistance among infecting Mtb is also dependent on host intrinsic factors, such as genetic makeup, health, and well-being, all of which impact the immune response against the bacteria. A key component of the host innate defense system are macrophages, phagocytic cells that engulf and destroy infecting microorganisms. However, Mtb can "invade" macrophages (and other host cells), where it is able to survive, proliferate and cause infection/disease. Invasion of macrophages by Mtb brings changes to the normal phagocytosis events, such as calcium ion homeostasis, membrane protein distribution and phagosome-lysosome fusion. If/when Mtb survives, it continues to multiply intracellularly and induce a pro-inflammatory response, leading to the onset of cell mediated/adaptive immunity and granuloma formation, which is generally thought of as a region of equilibrium between the host and the bacterium. For Mtb, the granuloma serves as an environment where the bacteria can exist in a dormant, semiand/or non-replicating state. For the host, the granulomas restrict the spread of Mtb to other tissues/organs because the diseased area is cordoned-off by the activated immune cells (Guirado et al., 2013). The host-pathogen interactions in the granuloma are highly complex, where the bacteria may get killed or able to survive and persist (Flynn and Chan, 2003). Taken together, the intracellular survival of Mtb is regarded as a cumulative effect of the host immune response and the bacterial ability to resist or subvert this response. Hence, strengthening the immune repertoire involved in combating Mtb through host-directed therapeutics (HDT) can be one of the approaches for effective bacterial killing and clearance of infection/disease.

Host directed therapy (HDT) aims at manipulating the metabolism and/or immune cell function to optimize the proinflammatory response or to modify the tissue physiology (Subbian et al., 2011a,b; Tobin et al., 2012). Recently, research on HDT as potential therapeutic strategy for infectious diseases has gained significant momentum due to the possibility of re-purposing drugs that have been already approved to treat chronic ailments and the advantage that pathogenic bacteria, such as Mtb, cannot develop resistance against a HDT, which targets host cell functions (Zumla et al., 2015). Autophagy is a homeostatic cellular process that removes intracellular debris derived from endo-and exo-genous sources, thus ensuring efficient functioning of cells. It is also a key innate immune response of the host cells to protect against invading pathogens. Therefore, targeting the autophagy machinery using small molecules and drugs to improve the host cell effector functions is an emerging concept in the treatment of several chronic diseases (Rubinsztein et al., 2012). Autophagy is regulated by multiple, complex networks and pathways that are either dependent or independent of mTOR (mechanistic target of rapamycin; a.k.a. mammalian target of rapamycin), a master regulatory molecule that impacts several cellular functions (**Figure 1**). In this review, we focus mainly on the role of autophagy in Mtb pathogenesis and modulating the mTOR pathway as potential adjunctive HDT to improve current, antibiotic-based treatment for pulmonary TB.

## MODULATION OF PHAGOCYTE FUNCTION BY MYCOBACTERIUM TUBERCULOSIS

Successful intracellular pathogens inhibit host cell antimicrobial processes involved in restricting their survival (Flynn and Chan, 2003; Kim et al., 2012). In that context, Mtb is known to inhibit killing within the phagolysosome of macrophages and other antigen presenting cells (APC) by modulating phagosome maturation and its fusion with the lysosome. In the infected APC, pathogenic Mtb inhibits actin assembly around the phagosome, thereby inhibiting host lipid molecules from interacting with

phagosomal proteins necessary for further maturation and fusion with the lysosome (Vergne et al., 2003; Rohde et al., 2007; Ehrt and Schnappinger, 2009; Shui et al., 2011; Seto et al., 2012). When bacteria are phagocytosed by APC, the phagosome acquires early endosomal protein markers, such as EEA1 (early endosomal antigen 1) and Rab5, which are gradually replaced with Rab7 during the maturation of the phagosome (Chandra et al., 2015); ultimately, LAMP1 (lysosome-associated membrane protein 1) and acid hydrolases mark the late phagosome for fusion with lysosome (Huynh et al., 2007). It has been reported that phagosomes containing live Mtb do not acquire Rab5 due to the presence of tryptophan aspartate coat protein (TACO). Phagosomal association with TACO is also reported in macrophages that can engulf other pathogenic mycobacteria, which also result in the inhibition of phagosomal maturation (Pieters and Gatfield, 2002). Proper maturation of phagosomes is the key to its fusion with lysosomes, which can kill the bacteria by delivering toxic molecules. However, due to the absence of proton-ATPase molecules in Mtb-containing phagosomes, the phagosome-lysosome fusion does not take place and the bacteria survive intracellularly (Vergne et al., 2005).

Another mechanism used by Mtb to manipulate APC involves perturbation of intracellular calcium ion (Ca2+) levels (Kusner and Barton, 2001). Several studies have demonstrated fluctuations in intracellular Ca2<sup>+</sup> levels in Mtb-infected macrophages (Vergne et al., 2003; Jayachandran et al., 2007). During phagocytosis of opsonized or heatkilled Mtb, intracellular Ca2<sup>+</sup> concentrations increase, while macrophages infected with live pathogenic Mtb have reduced calcium ion level, which in turn significantly reduce the levels of Ca2<sup>+</sup> associated-calmodulin and the phosphorylated Ca2+/calmodulin-dependent protein kinase II (CaMKII) (Jayachandran et al., 2007). Reduction in CaKMII level also blocks the delivery of lysosomal components to phagosome. Mtb reportedly prevents intracellular Ca2<sup>+</sup> increase through its cell wall glycolipid, ManLAM (mannose-capped lipoarabinomannan), and by inhibiting host sphingosine kinase (SK). ManLAM also inhibits ionophore-induced increase in Ca2<sup>+</sup> levels in macrophages. Reduced Ca2+/Calmodulin association impairs PI3K signaling, which inhibits recruitment of EEA1 to phagosomes (Rojas et al., 2000). Inhibition of SK abrogates phosphorylation of sphingosine, which is required for G-protein coupled receptor (GPCR) signaling that regulates Ca2<sup>+</sup> homeostasis. Thus, ManLAM triggers a sequence of events leading to Ca2<sup>+</sup> signaling disruption and phagosomal maturation arrest, which facilitate successful intracellular survival of infecting Mtb (Chan et al., 1991; Rojas et al., 2000).

Apart from its function in maintaining cellular homeostasis, autophagy is also known to sense and destroy intracellular bacteria in innate immune cells, such as macrophages. Although intracellular Mtb can efficiently modulate the bactericidal mechanisms of phagocytes, autophagy has been shown to be effective in killing Mtb (Gutierrez et al., 2004; Maiuri et al., 2007). Xenophagy, a type of autophagy whereby microorganisms can be sequestered and subject to lysosomal degradation, has been proposed to play an important role in elimination of bacteria (Gutierrez et al., 2004; Rubinsztein et al., 2012; Songane et al., 2012). In Mtb-infected host cells, the autophagosome collects ubiquitin while maturing, which then ultimately fuses with lysosome, thereby enhancing the lysosome-mediated bacterial killing. Survival of Mtb in macrophages has been reported to be dependent on the autophagosome delivery to the lysosome. However, in vivo and in vitro results have shown disparity in Mtb survival following inhibition of autophagy markers (**Table 1**) (Levine and Deretic, 2007; Lerena et al., 2008; Levine et al., 2011). Mutation or knockdown of autophagy associated host genes, such as Unc-51-like kinase 1 (Ulk1), Beclin1, Atg5, Atg7 or p62, has been reported to increase the survival of intracellular bacteria (Kim et al., 2011; Mizushima et al., 2011; Shang et al., 2011; Alers et al., 2012). However, although xenophagy is reported to restrict the survival of Mtb and BCG (bacille Calmette-Guerin) within macrophages, there are studies suggesting that intracellular pathogens such as Shigella flexneri, Listeria monocytogenes, Burkholderia pseudomallei, Orientia tsutsugamushi, Porphyromonas gingivalis, Staphylococcus, Brucella abortus, and Salmonella typhimurium are capable of blocking induction of autophagy by downregulating the colocalization of LC3 (microtubule-associated protein 1A/1B-light chain 3), restoring activation of mTOR, and utilizing nutrients for their growth and survival (Thurston et al., 2009; Yoshikawa et al., 2009; Zheng et al., 2009; Choy et al., 2012; Fraunholz and Sinha, 2012; Asrat et al., 2014; Yu et al., 2014).

Pathogenic Mtb also inhibits phagosome function in infected macrophages by releasing vesicle-bound lipids and glycolipids, which accumulate in lysosomes and interfere with the phagosome-lysosome fusion (Beatty et al., 2000). Taken together, pathogenic intracellular Mtb uses multiple strategies to manipulate the host defense machinery of APC for its own survival. Manipulation of APC function by Mtb impacts subsequent downstream events, including autophagy, antigen presentation, apoptosis, and activation of various signaling pathways involved in the production of cytokines, chemokines and other effector molecules that are crucial for controlling bacterial growth and replication (Briken et al., 2004; Cooper, 2009; Guenin et al., 2009; Rajaram et al., 2010).

## AUTOPHAGY AND mTOR SIGNALING

Autophagy, a Greek word meaning "eating of self," is a conserved cellular process critical for maintaining cellular integrity and function. This catabolic process is activated in cells due to lack of nutrient availability or cellular damage or stress, and involves degradation of damaged organelles and misfolded or abnormal proteins. During starvation, cytosolic components of cells are sequestered by autophagy to release nutrients for de novo biosynthesis of molecules (Laplante and Sabatini, 2012). Autophagy can also be activated by pathological factors, such as infections and other diseases. In these cases, normal cellular functions are facilitated by the elimination of pathogens through autophagy-dependent mechanisms, such as surface antigen presentation (Rubinsztein et al., 2012; Songane et al., 2012). Moreover, autophagy is one of the macrophage defense mechanisms against Mtb infection.

Autophagy is characterized by phagophore formation, elongation and maturation of the autophagosome, which ultimately fuse with the lysosome for the degradation of its contents. Formation of the autophagosome begins with a double membranous structure derived from the lipid bilayers of the endoplasmic reticulum (ER) or Golgi apparatus and conjugated with autophagy related (ATG) proteins (Maiuri et al., 2007; Alers et al., 2012). The three main components of autophagosome generation are: PI3KC3 (class III phosphoinositide 3-kinase

TABLE 1 | Major differences and similarities between mTOR complexes- mTORC1 and mTORC2.


complex 3), ULK1 (unc-51-like kinase 1) complex and ATG complex (**Figure 1**). This process is negatively regulated by mTOR kinase, which, when activated, blocks the ULK1 complex (Kim et al., 2011; Shang et al., 2011). Under stress conditions, such as nutrient deprivation or bacterial invasion, mTOR gets inactivated, enabling the ULK1 complex to recruit and activate PI3KC3 (Dibble and Cantley, 2015). This initiation complex, formed on the ER, leads to the nucleation of cell membrane, which is followed by recruitment of an ubiquitinlike molecule, LC3. In the final step, LC3 conjugates with phosphotidylethanolamine, resulting in self-fusion of the double membrane to form the autophagosome, which subsequently fuses with lysosome to degrade the engulfed contents.

Apart from mTOR signaling pathway (Lipinski et al., 2010), autophagy is also regulated by the inositol signaling pathway (Sarkar et al., 2005), Ca2<sup>+</sup> /Calpain signaling pathway (Gordon et al., 1993) and cAMP (cyclic adenosine monophosphate) (Noda and Ohsumi, 1998). Inhibition of these mTOR-independent pathways for promoting autophagy has been studied under different disease conditions (Floto et al., 2007; Grumati et al., 2010; Hidvegi et al., 2010). Promising outcomes of autophagy induction via mTOR-independent pathway have been observed only with a combination therapy strategy, where the small molecules enhancers (SMERs) or inhibitors (SMIRs) of mTORindependent pathway are used in combination with an mTOR inhibitor. For example, lithium, an inositol (1,4,5)-triphosphate inhibitor, when administered with rapamycin results in a stronger induction of autophagy (Sarkar et al., 2008). On the other hand, rapamycin alone can induce autophagy even at high intracellular inositol (1,4,5)-triphosphate levels, which has autophagy inhibitory effects. Since targeting ULK complex formation or ATG complex, rather than affecting the upstream pathways, seems to have a specific and stronger impact on autophagy, mTOR has been the target of interest for promoting autophagy upon infection with Mycobacteria (Gutierrez et al., 2004).

## THE mTOR COMPLEX

In addition to its role in autophagy, mTOR is also a master regulator of cell metabolism, growth, proliferation, translation initiation, and cytoskeletal organization. It belongs to the family of phosphoinoside 3-kinase- (PI3K-) related kinase and is a highly conserved serine/threonine protein kinase, which exists in host cells as part of two protein complexes—mTORC1 and mTORC2 (Laplante and Sabatini, 2012; Singh and Cuervo, 2012) (**Figure 2**). Theses complexes differ in their structure and activity, in part due to the difference in mTOR regulatory proteins such as RAPTOR (regulatory associated protein of mTOR; rapamycin sensitive) in mTORC1 and RICTOR (rapamycininsensitive companion of mTOR) in mTORC2, as well as other accessory proteins (Laplante and Sabatini, 2009). The proteins that are common to both mTORC1 and C2 complexes are the mammalian lethal with Sec13 protein 8 (mLST8) and DEP domain containing mTOR interacting protein (DEPTOR). While mLST8 acts as a positive regulator, DEPTOR functions as a

gets shut down or inhibited when cell encounters reducing nutrient level and decrease in energy. Inhibition of mTORC1 leads to inhibition of cellular metabolic processes. mTORC2; A six component complex has DEPTOR as its negative regulator and RICTOR and mSLT8 as positive regulators. This complex influences activation of mTORC1 by phosphorylation of AKT. How nutrient level influences mTORC2 is not known yet.

negative regulator of mTOR signaling. The mTORC1 is activated by RAPTOR, PRAS40 (proline-rich AKT substrate 40 kDa) and by phosphorylation of tuberous sclerosis protein 2 (TSC2) (Huang et al., 2008). The PI3K/AKT pathway is a positive regulator of mTOR signaling (Kim et al., 2011; Ng et al., 2011; Pan et al., 2012). Apart from PI3K/AKT, arginine, DNA damage, AMPK (AMP-activated protein kinase) and ERK1/2 (extracellular signal-regulated protein kinases 1 and 2) signaling were also reported to regulate mTORC1 activation (Kim et al., 2002; Inoki et al., 2003; Fingar and Blenis, 2004; Laplante and Sabatini, 2009). Importantly, mTORC1 activation inhibits autophagy (Jung et al., 2010). Deactivation of mTORC1 in cells under nutrition depletion or treatment with rapamycin leads to initiation of autophagy. (Seto et al., 2013). Similarly, dephosphorylating ULK1 by inactivation of mTORC1 induces autophagy (Egan et al., 2011; Kim et al., 2011; Shang et al., 2011).

The binding of mTORC2 with RICTOR facilitates the interaction of these proteins with TSC2 and mammalian stressactivated protein kinase interacting protein (mSIN-1); another protein found in association with RICTOR is PROTOR-1, which promotes activation of serum and glucorticoid-induced kinase 1 (SGK1). Interaction of all these proteins ultimately promotes mTORC2 complex formation and phosphorylation of AKT. Therefore, mTORC2 activation also regulates mTORC1 activation via AKT phosphorylation. Similarly, while mTORC1 activation is mediated by PI3K-AKT/PKB pathway in response to nutrient availability and mitogenic stimulation of the cell, phosphorylation of growth factors by autophosphorylation of their receptor tyrosine kinases activates mTORC2 complex, which also activates class I PI3K-AKT/PKB pathway (Dibble and Cantley, 2015). mTOR complexes also differ in the nature of their stimulant, for example, mTORC1 is activated by low levels of amino acids and growth factors, energy molecules and stress, while mTORC2 remains unaffected by the changing levels of these mTORC1 stimulants. However, role of mTORC2 is important for the regulation of AKT, which in turn governs mTORC1 functions. With use of TSC deficient cells importance of autophagy for cell survival was validated. In conditions like TSC (tuberous sclerosis complex), mTOR inhibition by rapamycin and pro-survival due to autophagy may have beneficial effects (Parkhitko et al., 2011).

In addition to regulating autophagy, activation of mTORC1 also promotes cellular metabolic pathways, such as glucose metabolism, protein and lipid synthesis, all of which contributes to cell growth and proliferation. S6Ks (p70 ribosomal protein S6 kinase 1/2) and 4E-BPs (eukaryotic initiation factor4 binding protein) are the two major proteins interacting with mTORC1 and play a major role in protein synthesis. mTORC1 phosphorylates 4E-BP1 thereby inhibiting its interaction with elF4E, which is then able to promote cap-dependent translation. Similarly, mTORC1 interaction with S6K1 stimulates capdependent translation of ribosomal proteins. Phosphorylation of S6K by mTORC1 also activates glucose transporter protein (Glut1) which activates glycolysis, lipogenesis and increases glucose uptake (Zeng et al., 2016). This increased glycolysis due to Glut1 is also reported to elevate T cell function and proliferation (Macintyre et al., 2014). Likewise, lipid synthesis is influenced by positive regulation of SREBP1 (sterol regulatory element binding protein 1) and PPARγ (peroxisome proliferatoractivated receptor-γ) (Kim and Chen, 2004), which are regulated by mTORC1. The mTOR inhibitor, rapamycin, reduced phosphatidic acid phosphatase (lipin-1) phosphorylation, which is essential for glycerolipid synthesis; lipin-1 also activates PPARγ and other proteins associated with lipid synthesis (Huffman et al., 2002). Oxidative metabolism is also influenced by mTOR signaling. In a mouse model, inhibition of mTORC1 reduced the muscle mass and oxidative metabolism, leading to early death. It has been shown that PGC1-α is associated with the oxidative metabolism and that mTOR directly interacts with this regulatory protein (Laplante and Sabatini, 2012). Other proteins interacting with mTORC1 are HIF-1α (hypoxia-inducible factor 1-alpha) and STAT3 (signal transducer and activator of transcription 3), which are involved in a plethora of cellular functions, ranging from angiogenesis to inflammation and cytokine response (Laughner et al., 2001).

### mTOR INHIBITORS AS POTENTIAL HDT FOR TB

Since mTOR signaling pathway regulates several cellular processes, including autophagy, that are linked to the host immune response to pathogens, it is an attractive target for developing/testing small molecules to modulate host immunity for better protection against infectious agents. Moreover, the peripheral blood mononuclear cells (PBMCs) and CD4+CD25+FoxP3+Treg cells isolated from active tuberculosis patients demonstrated mTOR inhibition during infection (Zhang et al., 2017). In contrast, mTOR activation, by deletion of Tsc1 in hematopoietic stem cells, induces accumulation CDK (cyclindependent kinase) inhibitors p16ink4a, p19Arf, p21Cip1 leading to impaired hematopoietic system and decreased lymphopoiesis (Chen et al., 2009). These observations establish that mTOR inhibition improves cell survival and the understanding that mTOR inhibition may be promoting host cell defense mechanisms against invading pathogens (Harrison et al., 2009).

The following are some of the key mTOR inhibitors in use to treat chronic conditions in humans.

### Rapamycin

Rapamycin, specifically known for its mTOR inhibitory activity, was first isolated from Streptomyces hygroscopicus. Despite of its antifungal and antibacterial properties, Rapamycin is wellknown for its immunosuppressant activity, which led to its use in organ transplant cases to reduce graft rejection. Similar to temsirolimus, rapamycin, which is also known as sirolimus, targets FKBP12 (FK506-binding protein 1A, 12 kDa) and inhibits the formation of active mTOR complex. Thus all of the currently known sirolimus derivatives target FKBP12 and inhibit mTOR complex.

In a zebrafish model of M. marinum infection, mTOR was shown to be associated with the host resistance to infection. In this model, mTOR mutants were hyper-susceptible to M. marinum at higher infection dose; however, when the inoculum size was decreased, the mTOR-deficient zebrafish cleared infection early (Pagan et al., 2016). Inhibition of mTOR in mice by rapamycin treatment at early age did not significantly affect the life expectancy or susceptibility to disease, but administration at an old age improved the survival expectancy (Harrison et al., 2009). In another study, administration of rapamycin to BCG-vaccinated mice has been shown to elicit better vaccination efficacy against Mtb infection, which is associated with induced autophagy, increased antigen presentation on dendritic cells and elevated Th1-type immune response (Gutierrez et al., 2004; Jagannath et al., 2009). Results from a low dose Mtb infection (MOI = 1) of human monocyte-derived-macrophages pre-infected with HIV, showed elevated bacterial load upon administration of rapamycin (1µM) (Andersson et al., 2016). This study described mTOR inhibition as an advantage for the intracellular survival of Mtb; however, in an already immunocompromised cell (due to HIV infection), it is difficult to assess the impact of mTOR inhibition on Mtb growth. Although rapamycin used to be the popular drug of interest to achieve cellular mTOR inhibition, poor solubility and long intracellular half-life complicates the consideration of this molecule as potential HDT for TB therapy.

### Temsirolimus

Temsirolimus, commercially known as CCI-779 or Torisel, is currently approved by the US-FDA for use in renal cell carcinoma (RCC) treatment. This prodrug can transform to sirolimus when dihydroxymethyl propionic acid ester group at C40 position is removed. Temsirolimus is metabolized by the enzyme CYP3A4 (cytochrome P450 3A4) and has a half-life of 9–27 h (MacKeigan and Krueger, 2015). Intravenous administration of temsirolimus increases its bioavailability and dose intensity (Boni et al., 2009). Mechanistically, temsirolimus targets host FKBP-12 protein. The drug-FKBP-12 interaction inhibits the formation of mTOR-FKBP-12 complex, leading to the inactivation of mTOR complex and inhibition of p70S6k and S6 phosphorylation. These effects cumulatively results in arrested cell growth, proliferation and survival in RCC patients. Nonspecific pneumonitis and gastrointestinal disorders are major side effects in RCC patients treated with this drug. In addition, metabolic diseases such as hyperglycemia, and hypercholesterolemia are associated with temsirolimus administration in these patients (Malizzia and Hsu, 2008). Importantly, temsirolimus treatment has been widely associated with reactivation of latent Mtb infection among RCC patients. Also, progression of tumor was noted in these patients when temsirolimus was administered in combination with rifampicin, a first-line anti-TB drug (Bossé et al., 2016).

### Ridaforolimus

Ridaforolimus (AP23573 or MK-8669) is an analog of sirolimus with improved bioavailability, solubility and half-life (30–75 h) (Rivera et al., 2011). It is administered orally or intravenously for the treatment of solid tumors of soft tissues, bone and other hematologic malignancies (Huang et al., 2015). In a phase I clinical trial with 87 ER+/high-proliferative breast cancer cases, majority of patients treated with ridaforolimus demonstrated reduced tumor activity (Di Cosimo et al., 2015). Similarly, a phase II clinical trial showed promising results for ridaforolimus to treat patients with endometrial, soft tissue and bone cancers (Palavra et al., 2017). The effect of ridaforolimus on cell metabolism and growth is largely dependent on the dose of drug used for treatment. This is due to its varying effect on mTOR inhibition with variation in dosage (Rivera et al., 2011). However, no reactivation of latent Mtb infection has been reported in these studies (Huang et al., 2015; Palavra et al., 2017).

### Everolimus

Everolimus (40-O-(2-Hydroxy)-ethyl-rapamycin), commercially known as SDZ-RAD, RAD001, Certican and Afinitor, is a derivative of rapamycin bearing a stable 2-hydroxy ethyl chain substitution at position 40. This agent has a better solubility, oral availability, and decreased mean elimination half-life (∼18–30 h), leading to early removal of drug from the body compared to the parent compound (rapamycin). Because of the better absorption, it has higher bio-availability (30–60%) and a Tmax of 1–2 h. Everolimus is an immunosuppressive, anti-inflammatory drug that inhibits host cell proliferation by arresting the progression of cell cycle from G1 to S phase; the immune suppressive function is exerted by inhibiting IL-2 and IL-15 mediated lymphocyte proliferation (Kovarik et al., 2002; Lingaraju et al., 2010; Ahya et al., 2011). In addition, everolimus promotes autophagy by inhibiting mTORC1 (Saran et al., 2015).

Inhibition of mTOR pathway by everolimus treatment has been reported to improve cellular immune response in both animal models and human studies. In a study performed with 218 healthy volunteers of >65 years of age, everolimus treatment had beneficial effects over aging-related issues (Mannick et al., 2014). Specifically, these elderly volunteers treated with a low dose of everolimus showed about 20% improvement in their protective response after influenza vaccination. This improvement was associated with reduced expression of programmed death-1 receptor, which is otherwise highly expressed in aging individuals, on CD4 and CD8 T cells, thus increasing T cell antigen processing and expression. This low dose administration (0.5 mg daily or 5 mg weekly) of eveolimus demonstrated minimum number of adverse events, (35 adverse events) compared to a higher dose administration (20 mg weekly), which resulted in 109 adverse events amongst 53 elderly individuals (Mannick et al., 2014). This study clearly highlights the importance of optimizing the dose of mTOR inhibitors, such as everolimus in this case, for better efficacy with minimal adverse effects. In contrast, case studies with organ transplant patients have mentioned a higher risk of Mtb infection and reactivation of LTBI as possible side-effects of everolimus administration (Kovarik et al., 2002; Fijałkowska-Morawska et al., 2011). Although patients in this study were treated with a higher dose of everolimus, than the influenza vaccine study mentioned above, the mechanism underlying the connection between dose of everolimus and reactivation of LTBI is not clearly understood. However, the negative consequences of high dose administration of everolimus can be overcome by co-administration with CYP3A4 enzyme inducers, such as rifampicin, which are used to treat opportunistic TB infections in organ transplant patients (Eisen et al., 2003). Thus, it is important to understand the doseresponse of everolimus in the context of host cellular functions and how the drug influences phagocytosis and autophagymediated elimination of Mtb during infection.

All the mTOR inhibitors described above are also frequently used in the treatment of various forms of cancer, including breast cancer, renal cell carcinoma and tuberous sclerosis complex, due to their ability to inhibit host cell proliferation and growth (Pohanka, 2006; Koh et al., 2013). However, the idea of using these mTOR inhibitors as potential adjunct HDT for TB therapy needs to be substantiated through experimental evidences related to dosage, pharmacokinetics and pharmacodynamics (PK/PD) parameters and, cost vs. benefit effects on the host immunity. Such evidences need to be reinforced by series of studies on reliable and relevant pre-clinical animal models of Mtb infection. In addition, metabolic dysfunctions, such as hyperglycemia is a common side effect in cancer patients treated with mTOR inhibitors, including everolimus (Porta et al., 2011). Although the impact of such inhibitors in the context of TB treatment remains to be determined, serious side effects of HDT drugs preclude their potential use in any treatment. Moreover, as immune-suppressing agents, the application of mTOR inhibitors as a stand-alone HDT therapy for TB holds a significant risk of reactivation of latent Mtb infection, similar to the situation observed in rheumatoid arthritis patients treated with anti-TNF-α antibody (Kovarik et al., 2002). However, when used at an immune modulating-, as opposed to immune suppressingdose, these mTOR inhibitors can be potential candidates to serve as an adjunct therapeutic molecule, along with standard anti-TB drugs, in improving the treatment outcome. Thus, fine-tuning the dose of mTOR inhibitors is an important and necessary step toward application of these HDT compounds for TB treatment. Importantly, since Mtb cannot develop resistance to a drug that targets host signaling pathway, such as mTOR or cellular processes, such as autophagy, HDT drugs has the potential to alleviate the development of MDR- and XDR-Mtb strains and their transmission in the community. Analogous to the trend in cancer treatment that have shifted from chemotherapeutic and radiologic regimens to more-host targeted treatment approaches, Mtb infection and/or disease can benefit from specific HDT drugs that targets, for example, the mTOR pathway and/or autophagy.

### SUMMARY AND CONCLUSION

Pathogenic Mtb possess several virulent determinants, such as the unusually lipid-rich cell wall, that serve as permeability barriers and protects the bacteria from the harsh intracellular environment within phagocytes, and from the bactericidal activities of anti-TB drugs. Additionally, these mycobacteria-derived molecules interact with the host immune cells and modulates their function, promoting bacterial survival/persistence, causing disease within the host and enabling the development of bacterial drug resistance. Thus, when the innate defensive mechanisms of phagocytes are manipulated by the pathogen to promote its survival, the host develops active disease, which is hard to overcome. This can be one of the reasons for the inefficiency of current anti-mycobacterial drugs to eliminate TB, and for the emergence of drug-resistant Mtb strains. Perturbing host cell functions through HDT molecules

### REFERENCES


has the potential to enhance the effector functions of these cells, which are the ultimate arsenals in combating bacterial infection. Moreover, these immune modulating drugs do not contribute to the emergence of drug resistance by the infecting bacteria. This criterion is crucial when considering therapy, particularly for patients with MDR, XDR-TB, as well as those patients with co-existing chronic conditions, such as diabetes or HIV infection, in which conventional antibiotics therapy has been shown to be complex, complicated, toxic and insufficient in achieving a bacteriological cure. Host cell autophagy, regulated by mTOR pathway, plays an important role in cellular homeostasis as well as in antibacterial defense mechanism. Therefore, targeting mTOR pathway with small molecules, such as everolimus, has the potential to develop novel and better combination drug therapy, along with standard anti-TB drugs to combat various forms of TB in patients with/without other co-morbid conditions. This approach can also enhance bacterial killing, reduce treatment duration, and/or improve clinical outcome. Clearly, more research and experimental evidence is warranted on these and other HDT molecules, for their efficacy, toxicity and other properties, through extensive pre-clinical studies using appropriate animal models of TB, before they are tried as therapeutic intervention for TB in human clinical trial.

### AUTHOR CONTRIBUTIONS

SS: conceived the concept; PS and SS: wrote/edited the manuscript and agreed for submission.

### FUNDING

This study was supported by a grant from the Bill and Melinda Gates Foundation (OPP1157210) to SS.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Singh and Subbian. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Defeating Antibiotic- and Phage-Resistant Enterococcus faecalis Using a Phage Cocktail in Vitro and in a Clot Model

Leron Khalifa<sup>1</sup> , Daniel Gelman<sup>1</sup> , Mor Shlezinger1,2, Axel Lionel Dessal<sup>1</sup> , Shunit Coppenhagen-Glazer<sup>1</sup> , Nurit Beyth1,2† and Ronen Hazan<sup>1</sup> \* †

<sup>1</sup> Faculty of Dental Sciences, Hadassah School of Dental Medicine, Hebrew University of Jerusalem, Jerusalem, Israel, <sup>2</sup> Department of Prosthodontics, Hadassah School of Dental Medicine, Hebrew University of Jerusalem, Jerusalem, Israel

### Edited by:

Rebecca Thombre, Savitribai Phule Pune University, India

### Reviewed by:

Andrei A. Zimin, Institute of Biochemistry and Physiology of Microorganisms (RAS), Russia Stephen Tobias Abedon, The Ohio State University, United States

### \*Correspondence:

Ronen Hazan ronenh@ekmd.huji.ac.il †These authors have contributed equally to this work.

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 08 June 2017 Accepted: 12 February 2018 Published: 28 February 2018

### Citation:

Khalifa L, Gelman D, Shlezinger M, Dessal AL, Coppenhagen-Glazer S, Beyth N and Hazan R (2018) Defeating Antibiotic- and Phage-Resistant Enterococcus faecalis Using a Phage Cocktail in Vitro and in a Clot Model. Front. Microbiol. 9:326. doi: 10.3389/fmicb.2018.00326 The deteriorating effectiveness of antibiotics is propelling researchers worldwide towards alternative techniques such as phage therapy: curing infectious diseases using viruses of bacteria called bacteriophages. In a previous paper, we isolated phage EFDG1, highly effective against both planktonic and biofilm cultures of one of the most challenging pathogenic species, the vancomycin-resistant Enterococcus (VRE). Thus, it is a promising phage to be used in phage therapy. Further experimentation revealed the emergence of a mutant resistant to EFDG1 phage: EFDG1<sup>r</sup> . This kind of spontaneous resistance to antibiotics would be disastrous occurrence, however for phage-therapy it is only a minor hindrance. We quickly and successfully isolated a new phage, EFLK1, which proved effective against both the resistant mutant EFDG1<sup>r</sup> and its parental VRE, Enterococcus faecalis V583. Furthermore, combining both phages in a cocktail produced an additive effect against E. faecalis V583 strains regardless of their antibiotic or phage-resistance profile. An analysis of the differences in genome sequence, genes, mutations, and tRNA content of both phages is presented. This work is a proof-ofconcept of one of the most significant advantages of phage therapy, namely the ability to easily overcome emerging resistant bacteria.

Keywords: bacteriophages, Enterococcus faecalis, antibiotic-resistance, phage therapy, phage-resistance, phage cocktail

## INTRODUCTION

In 1967, William H. Stewart, the Surgeon General, said in the White House: "It's time to close the books on infectious diseases, declare the war against pestilence won. . .," expressing the general opinion, at the time, that antibiotics succeeded in solving all bacteria-related problems. Ironically, this very success has proved dangerous, as the frequent reliance on antibiotics has resulted in dissemination of resistant variants (Charles and Grayson, 2004; Davies and Davies, 2010).

One of the antibiotic – resistant, most frequently isolated species from hospital-associated infections, is vancomycin-resistant enterococci (VRE) (O'Driscoll and Crank, 2015; Khalifa et al., 2016). While mostly a commensal, when inhabiting the gastro-intestinal tract, Gram-positive Enterococcus faecalis bacterium can cause fatal infections (Rice, 2008; Orsi and Ciorba, 2013). In many cases, neither antibiotics (Erlandson et al., 2008; Chong et al., 2010) nor other antibacterial means like endodontic treatments and rinsing with calcium hydroxide, sodium hypochlorite,

and chlorohexidine (Khalifa et al., 2016) have successfully prevented the morbidity or mortality resulting from Enterococcus infections especially in cases of biofilm (Donlan, 2001; de la Fuente-Nunez et al., 2013). The hazardous potential of these bacteria is exacerbated by the emergence of antibiotic-resistant strains, primarily VRE (Linden, 2002). So much so, that in 2013 the Center for Disease Control and Prevention (CDC) has declared VRE enterococci to be one amongst the top 18 drug-resistant threats to treat today<sup>1</sup> .

With the limitations of chemical antibiotics and the emergence of antibiotic-resistant strains, the use of bacteriophages has been regaining interest (Golkar et al., 2014). Their effective potential is magnified as they are highly strain-specific, targeting only their pathogen and tend to only minimally disrupt normal flora (Loc-Carrillo and Abedon, 2011). Moreover, bacteriophages have evolved naturally to be become the most proficient bacterial killers (Srinivasiah et al., 2008) in all forms, including hard-to-eradicate biofilms (Donlan, 2009; Khalifa et al., 2016). Use of bacteriophages has numerous other advantages over antibiotics (Loc-Carrillo and Abedon, 2011; Khalifa et al., 2016), including ease of isolation. Additionally, recent human and animal trials have shown that phages are safe for use (Burrowes et al., 2011).

We recently demonstrated the treatment potential of phages against vancomycin-resistant E. faecalis infections. EFDG1, an anti-E. faecalis phage, was isolated from sewage and its genome was characterized. This phage proved highly effective against E. faecalis V583 in planktonic and biofilm cultures in vitro and in an ex vivo human root canal model (Khalifa et al., 2015a).

Successful though phage therapy may be, bacterial cells resistant to phages will no doubt emerge, just as in the case of antibiotics (Duerkop et al., 2016), as indeed has occurred in our case too. However, in contrast to the grave challenges facing antibiotics, phage therapy offers several simple approaches to combat phage-resistant bacteria. One strategy would be to isolate new phages and to use them in cocktails. Based on the number of previously isolated phages and phage sequences in microbiomes, it is estimated that a bacterial strain may have as many as several 100 phages capable of combating it Khalifa et al. (2016). A second approach is to improve current phages by random or directed mutagenesis: using UV radiation (Drake, 1966) or gamma rays (Bertram, 1988), by means of chemical mutagens such as methyl methanesulfonate (MMS) (Drake, 1982) or through genetic engineering (Kiro et al., 2014; Pires et al., 2016). In any case, multiple-phage cocktails are preferable to single-phage therapy, as they are better equipped to deal with phage-resistant bacterial mutants and multispecies bacterial cultures (polyphage therapy) (Chan and Abedon, 2012; Gu et al., 2012; Ormala and Jalasvuori, 2013).

In the current article, we demonstrate the first approach: isolating a new phage against phage-resistant bacteria, constructing a phage cocktail and then utilizing it successfully against both naïve and phage-resistant mutant strains in in vitro models. The phage EFLK1 was effective against the EFDG1 phage-resistant E. faecalis strain (EFDG1<sup>r</sup> ). Moreover, a cocktail of EFDG1 and EFLK1 phages had an additive effect on planktonic and biofilm cultures.

## MATERIALS AND METHODS

### Emergence and Isolation of Phage-Resistant Bacteria

Following all experiments of EFV583 with phage, survivors were routinely tested for resistance to the phage. The survivors were plated and the resulting single colonies were collected and grown. The bacteria were then tested for susceptibility to the phage, by either spotting phages on them or by cross streak agar assay. For the spotting method, about 100–200 µl of bacterial culture was spread uniformly on a BHI agar plate and once it dried; 10 µl spots of phages were spotted, which was done in duplicates or triplicates. While the cross streak agar assays were performed by horizontally streaking the phage lysate in the middle of the plate and then vertically streaking the bacterial colonies over the phage in unidirectional streaks (Moore, 2011; Tiruvadi Krishnan et al., 2015).

Bacterial colonies that grew in presence of phage were considered to exhibit resistance to the phage. These colonies were then collected and grown at 37◦C and checked numerous times for their resistant characteristics against the phage. Once their resistant property was confirmed, these bacteria were frozen with 25% glycerol at −80◦C and store for future use.

### Bacterial Strains and Materials

Enterococcus faecalis V583 (ATCC 700802) VRE and EFDG1<sup>r</sup> , EFDG1 phage-resistant bacterial strain were grown in brain heart infusion (BHI) broth (Difco, Detroit, MI, United States) aerobically in an incubator shaker with shaking at 220 RPM at 37◦C. BHI agar (1.5%) was used for isolation streaks and isolations of phage. For top agar, 0.6% of agar (soft agar) was used, with BHI as nutrient supplement.

### Isolation and Storage of Phage

In order to combat the phage-resistant mutant strain, a new lytic phage was isolated from the sewage effluents collected from the "Nahal Sorek" decontamination facility located in West Jerusalem, as described previously (Khalifa et al., 2015a). The effluents were centrifuged using a 5430R centrifuge, and a fixed angle rotorFA-45-24-11HS; Eppendorf at 10,000 × g for 10 min, to pellet out the debris. The supernatant was filtered twice, initially through 0.45-µm-pore-size filters (Merck Millipore, Ltd., Ireland) to filter out larger particles and then through 0.22-µm-pore-size filters (Merck Millipore) to achieve lysate free from any residues. Different concentrations of the filtrates (between 0.1 and 1 ml) were then added to bacterial broth culture diluted 1:1000 in BHI. Once lysis was obtained, the lysate was centrifuged and filtered using the aforementioned method.

The phage lysate was then applied on bacterial lawn by the method of double layer agar to obtain clear plaques that

<sup>1</sup>https://www.cdc.gov/drugresistance/biggest\_threats.html

symbolize the presence of phages. About 100 µl of bacteria was mixed with 100 µl of phage lysate and incubated for 20 min at 37◦C to achieve better adsorption of the phage to the bacteria. The entire mixture was then added to 5 ml of melted 0.6% BHI top agar, poured evenly onto a BHI agar plate and incubated overnight at 37◦C. The resultant single plaques were collected and added to 1 ml BHI and incubated at 4◦C for diffusion. The liquid was centrifuged at 5000 RPM for 10 min and filtered through 0.22-µm-pore-size filters to obtain a clear phage lysate that was further used to grow large quantities of the phage. The phage was grown in BHI broth with EFV583 bacteria until a titer of 10<sup>9</sup> or higher was achieved. The phage titer of the phage lysate was determined by counting the Plaque forming units per ml (PFU/ml). For this; the phage lysate was serially diluted in BHI broth and 5 µl of these dilutions were spotted onto an overlayer of molten 0.6% agarose with 100 µl of bacteria. This plate was incubated overnight at 37◦C and the numbers of plaques were counted to finally calculate the PFU/ml. A large volume of EFLK1 phage lysate with a high titer (1010) was prepared and stored at 4◦C.

## Characterization of Phage Activity against E. faecalis Naïve and Phage-Resistant Form in Planktonic and Biofilm Cultures

To assess the phage effectiveness against planktonic bacteria, the growth kinetics were analyzed using a 96 well plate reader. The phage was added initially to both logarithmic cultures (10<sup>4</sup> to 10<sup>5</sup> CFU/ml) at MOI of 100–1000 (10<sup>8</sup> PFU/ml) and stationary cultures (10<sup>8</sup> to 10<sup>9</sup> CFU/ml) of E. faecalis at MOI of 0.1–1 (10<sup>8</sup> PFU/ml). Cell viability was also checked by determining CFU/ml for each well.

The biofilm eradication efficiency of the phage was tested by adding it in a MOI of 0.1 (10<sup>8</sup> PFU/ml) to a 2-week old stationary biofilm of E. faecalis. To obtain a 2-week old biofilm, overnight grown bacteria were diluted with BHI to obtain 10<sup>4</sup> to 10<sup>5</sup> CFU/ml of bacteria, which were then plated in a 96-well plate that was incubated for 2 weeks without any change of media or addition of new bacteria. Killing efficacy was validated using viable cell counting by calculating the CFU/ml of the biofilm as described previously (Khalifa et al., 2015a). Briefly, aspirated spent media was removed from the wells followed by washing the wells twice with sterile PBS (100 ul) carefully to not disrupt the biofilm. Sterile PBS (100 ul) was added and the biofilm was scraped and collected. The cells were sonicated in a water-bath for 5–10 min followed by several dilutions (1:10) and plating on BHI plates for CFU count.

## One Step Growth

We based our protocol on the one step growth method of Ellis and Delbrück (Ellis and Delbrück, 1939). Briefly, phages were added to stationary-phase bacteria (10<sup>9</sup> CFU/ml) in a final concentration of 1 × 10<sup>7</sup> PFU/ml followed by incubation in 37◦C for 10 min. The samples were diluted at a ratio of 1:10<sup>4</sup> in fresh BHI medium, followed again by incubation at 37◦C for 10 more minutes. Starting at 20 min after the introduction of the bacteriophages to the bacterial cultures, the diluted mixtures were enumerated through a double-layered agar plaque assay. To this end, 3 ml of Soft BHI Agar (0.6% Agar) were mixed with 100 µL of the diluted mixture and with 100 µL of stationaryphase bacteria, according to the tested bacterial strain. The plaque assays were conducted in the following time points after the introduction of the bacteriophages: 20, 35, 50, 60, 70, 85, 100, 120 min, in biological duplicates. Latent time and burst size were calculated from this growth curve for each of the bacteriophages described, with each of the bacterial hosts.

### Genome Sequencing

To verify that the phage is indeed new, its DNA was isolated using the Norgen Biotek Phage DNA Isolation Kit (Cat. # 46800) and the resulting genome was sequenced and analyzed (Khalifa et al., 2015a,b). Analysis and comparison to other E. faecalis phages was performed using the GENIOUS 10.0.6 software and its plugins<sup>2</sup> . Annotation of genes was carried out using PHAST<sup>3</sup> .

## TEM Visualization

To observe the structure of the isolated phage accurately transmission electron microscopy (TEM) by the classic method of Gill as described in OpenWetWare<sup>4</sup> , was conducted following the procedure mention in our previous paper (Khalifa et al., 2015a). Briefly, 1 ml of phage lysate with 10<sup>9</sup> PFU/ml was centrifuged at 19,283 × g (centrifuge 5430R, rotor FA-45-24-11HS; Eppendorf) for 2 h at room temperature. The supernatant was discarded, and the pellet was resuspended in 200 µl of 5 mM MgSO<sup>4</sup> and incubated overnight at 4◦C. 30 µl of 5 mM MgSO<sup>4</sup> and 10 µl of the phage sample were mixed gently on a parafilm strip and 30 µl of 2% uranyl acetate was pipetted on it. On these drops of phage samples grids were then placed carefully using forceps, with the carbon side facing down. After about a minute the grids were dried and stored in the desiccator until further use. A TEM (Joel, TEM 1400 plus) with a charge-coupled device camera (Gatan Orius 600) was used to capture images.

## Naïve and Resistant Bacterial Mixture

The naïve and resistant E. faecalis were mixed in 1:1 ratio (v/v) and the efficacy of phages was checked individually and in cocktail against them. The experiment mirrored the one described above (see section "Characterization of Phage Activity against E. faecalis Naïve and Phage-Resistant Form in Planktonic and Biofilm Cultures") for assessing phage activity on planktonic bacteria and on biofilm.

## In Vitro Fibrin Clot Model

The in vitro fibrin clot model was prepared according to the protocol described by McGrath et al. (1994) and Entenza et al. (2009). Overnight cultures (10<sup>9</sup> CFU/ml) of E. faecalis individually or in a mixed cultures of naïve and resistant bacteria were diluted 1:10 with citrated plasma. Clots were formed by triggering coagulation by adding 20 µl bovine thrombin

<sup>2</sup>https://www.geneious.com

<sup>3</sup>http://www.phantome.org

<sup>4</sup>http://openwetware.org/wiki/Gill:Preparing\_phage\_specimens\_for\_TEM

(5000 U/ml) and 20 µl CaCl<sup>2</sup> (50 mmol). The resultant clots were then re-suspended in 500 µl of bacteriophage lysate of 10<sup>8</sup> PFU/ml and for the control, 500 µl of BHI were added instead. The tubes were kept in an incubator-shaker for 6 h at 37◦C. After incubation, the clots were washed with sterile PBS and lysed with 32 µl of 0.25% of Trypsin EDTA. After a 5-min centrifugation at 14,000 rpm, the cell pellet was re-suspended in 100 µl of PBS and used for calculating CFU/ml.

## RESULTS

## Emergence of the Phage-Resistant E. faecalis Mutant EFDG1<sup>r</sup> and Isolation of EFLK1 Phage to Combat It

The lytic phage EFDG1 was isolated against vancomycin-resistant strains of E. faecalis V583 planktonic and biofilms cultures (Khalifa et al., 2015a). When strain EFDG1<sup>r</sup> (resistant to EFDG1 phage) evolved during the experiments (**Figure 1A**), we screened various sewage effluent samples containing potential anti-E. faecalis phages. Within several weeks, a new lytic phage was isolated, EFLK1, from a sample collected from the sewage water of the "Nahal Sorek" decontamination facility. The new phage exhibited clear plaques on double-layered agar lawn of EFDG1<sup>r</sup> . It also prevented the growth of EFDG1<sup>r</sup> , as observed both in optical density growth curve (**Figure 1A**) and colony forming units (CFUs) count (**Figure 1B**). In agreement, a one-step growth experiment revealed that phage EFLK1 propagated on an EFDG1<sup>r</sup> while EFDG1 phage did not (**Figure 1C**).

## Phenotypic Differences Between EFDG1 and EFLK1 Phages

Despite the similarity between the two phages (32.187%), the efficiency and kinetics of EFLK1 phage infectivity against the parental strain V583 was distinct from that of EFDG1 phage. The lysis caused by EFDG1 phage during logarithmic growth was observed within 5 h (**Figure 1A**) and it reduced the viable cells number by four logs (**Figure 2A**) and (Khalifa et al., 2015a). In these conditions, EFLK1 phage was markedly slower; its lysis started after 12 h (**Figure 2A**) and it reduced cell viability by three logs (**Figure 2C**). In contrast, EFLK1 phage faired better against stationary cultures of V583 strain: EFLK1 phage lysis was observed after 10 h compared to much less significant lysis by EFDG1 phage (**Figure 2B**). These differences in lysis are reflected in the reduction of cell viability from 10<sup>11</sup> to ∼10<sup>5</sup> by EFLK1 phage and only to ∼10<sup>8</sup> by EFDG1 phage (**Figure 2D**).

## In Vitro Characterization of Combinatorial Cocktails of EFDG1 and EFLK1 Phages Aimed for Therapy

Since we aim to use EFDG1 and EFLK1 phages as therapeutic agents, we tested various combinations of phage cocktails,

(CFU/ml) for logarithmic (C) and stationary phase (D) bacteria with and without treatment of the phages and their cocktails. (E) Biofilm biomass quantified by crystal violet staining of 2-weeks of E. faecalis V583 biofilm. (F) CFU/ml for phage treated and untreated 2-weeks old E. faecalis V583 biofilm.

initially on naïve E. faecalis V583. To this end, we combined EFDG1 and EFLK1 phages in various ratios: 1:1 (cocktail#1), 1:2 (cocktail#2), 2:1 (cocktail#3), 1:3 (cocktail#4), and 3:1 (cocktail#5) correspondingly (**Figure 2**). The best cocktail was found to be Cocktail#1 (1:1) as it killed the logarithmic culture of E. faecalis V583 as efficiently as EFDG1 phage (**Figures 2A–C**), and the stationary culture like EFLK1 phage (**Figures 2B–D**). Thus, cocktail#1 equaled the better phage in each case and outperformed the other cocktails against the challenging biofilm of E. faecalis V583 (**Figure 2E**).

When cocktail#1 was used against EFDG1<sup>r</sup> mutants, it was found to be as effective as EFLK1 phage against logarithmic, stationary and biofilm cultures of the resistant mutants (**Figure 3**).

Finally, in order to simulate the evolution of resistant mutants within a culture, we mixed the WT E. faecalis V583 with its resistant mutant EFDG1<sup>r</sup> and treated them with individual phages and cocktail #1, separately (**Figure 4**). We found that cocktail #1 and EFLK1 phage were effective against the bacterial mixture while EFDG1 phage was not.

### Demonstrating Phage Efficacy in a Fibrin Clot Model

A recognized tool for the study of bacterial endocarditis (Hershberger et al., 2000), we used the in vitro fibrin-clot infection model (McGrath et al., 1994; Entenza et al., 2009) to assess the efficacy of the cocktail and the individual phages EFDG1 and EFLK1 against E. faecalis V583 and EFDG1<sup>r</sup> (**Figure 5**). CFU/ml results following a 6-h incubation of E. faecalis V583 with EFLK1 phage and cocktail#1 showed a CFU reduction of three logs with EFDG1 phage and six logs with EFLK1 phage and cocktail#1. EFLK1 phage and the cocktail also reduced the CFU by five and six logs respectively, against EFDG1<sup>r</sup> , while, EFDG1 phage failed as expected.

### Genotypic Differences between EFDG1 and EFLK1 Phages

To explain the phenotypic differences between EFLK1 and EFDG1 phages, their genomes were compared (Khalifa et al., 2015a). In addition to EFLK1 phage, three additional E. faecalis phages of the Spounavirinae subfamily of the Myoviridae phage

family of lytic phages (Lavigne et al., 2009), were described. The highest resemblance (∼45%) was found with EFDG1 phage, previously isolated by us (Khalifa et al., 2015a), and less with the other two phages phiEF24c and ECP3 (**Figures 6A–C**). All four phages share 60 core genes which are ∼28% of their genes (**Figure 6A**).

The most significant difference between EFDG1 and EFLK1 phages genome sequences found was the number of tRNA genes (**Figure 6D**): EFDG1 phage contains 24 of them, EFLK1 phage carries none and the other two phages (phiEF24c and ECP3) contain five each which are all clustered in one region as in EFDG1 phage (**Figure 6D**). An analysis of the distribution of tRNA genes among all 2,053 fully sequenced tailed phages (NCBI, Caudovirales, taxid: 28883) revealed that nearly 60% (1,227) lacked tRNA genes as well (**Figure 6D**). Of the remainder, the majority carried only 1–5 genes, just like ECP3 and phiEF24c (**Figure 6D**). The highest number of tRNA genes (33) was found in the Streptomyces phage Jay2Jay (accession: KM652554.1). The cassette of five tRNAs in phiEF24c is identical to that in ECP3. EFDG1 phage contains the same five tRNAs and 19 more, all in a different order. Blast of these cassettes shows that they are unique and do not correspond to the ones on the E. faecalis genome.

One explanation for the presence of tRNA genes in phages is variance in codon usage between their own genome and that of the host. Consequently, one would expect EFDG1 phage codon usage to correlate less with that of its E. faecalis host than

the codon usage of the other phages and the large amount of tRNAs to compensate for that extra bias. However, COR function (R Bioconductor package) analysis disproved this theory, as no significant differences were found between the codon usage of the phages and E. faecalis. EFDG1 has 0.921 ECP3; 0.89522, EFLK1; 0.8954, and phiEF24c share 0.8948 correlation with the E. faecalis codon usage (**Supplementary Table S1**).

Next, we looked at the specific tRNA genes carried by these phages in accordance with their codon usage (**Supplementary Table S1**). The three anti-E. faecalis Spounavirinae phages with tRNA's (EFDG1, ECP3, and phiEF24c) carry tRNA genes corresponding to codons CTA, AGA, GAC, and TGG of leucine, arginine, aspartic acid, and tryptophan respectively. Besides these, EFDG1 phage encodes for two while ECP and phiEF24c phages for single tRNA corresponding to the initiation and sole methionine codon AUG. Indeed, in the three phages, the usage of all these codons is higher than their usage in E. faecalis (**Supplementary Table S1**), perhaps justifying the corresponding tRNA gene's existence. Surprisingly, the codon usage of these five codons is also higher in EFLK1 phage, which does not have any tRNA genes (**Supplementary Table S1**). Moreover, the codon usage should lead the four phages to carry the tRNA gene corresponding to codon TAC (tyrosine) while EFLK1, ECP3, and phiEF24c phages should carry tRNA's for AAG (lysine), AAC (asparagine), ACA (threonine) and GTA (valine) like EFDG1 phage, however this is not the case. On the other hand, EFDG1 phage carries several tRNA genes, such as TTA (leucine) which, according to the codon usage, seems to be not required. Thus, the question of the effect of codon usage on the presence of these tRNA and its relation to the phenotypic differences remains open.

We also analyzed the P2 mutation described by Uchiyama et al. (2011) in phage phiEF24c. The P2 mutation, located in Orf31, a conserved putative tail fiber, dramatically improves the ability of phiEF24c to adsorb and lyse the cells (Uchiyama et al., 2011). Nevertheless, EFDG1, EFLK1, and ECP3 phages harbor homologs for Orf31 but none of them has the P2 mutant allele (**Figure 6E**).

## DISCUSSION

Phage therapy holds the promise of succeeding where antibiotics have failed against resistant bacteria. In some cases, phage resistance actually has led to loss of bacterial virulence (Scott et al., 2007; Zahid et al., 2008; Capparelli et al., 2010). In addition, while production of new antibiotics is slowing down (Silver, 2011; Laxminarayan et al., 2013), the rate of new phages isolation is increasing and new phages are being discovered daily as can be seen in phage databases such as the European Nucleotide Archive (ENA<sup>5</sup> ). It is speculated that phages are the most prevalent replicating form on Earth and their number is estimated to be ∼1031, which is about 10 times more than prokaryotes (Diaz-Munoz and Koskella, 2014). Thus, in case of resistance, the likelihood of isolating an appropriate phage is indeed high. Furthermore, the variety of phages supports the construction of an almost unlimited number of combinations of cocktails to combat various strains or multispecies infections.

In this work, we demonstrated the isolation of a new phage when facing phage-resistant mutants and the advantages of a two-phage cocktail against VRE E. faecalis and its phage-resistant mutant.

Although similar, phages EFDG1 (Khalifa et al., 2015a) and EFLK1 have interesting phenotypic differences. While EFDG1 phage is more efficient against logarithmic E. faecalis V583 cultures, EFLK1 phage is a better killer of stationary (**Figure 2**). Joined in cocktail, they are even better at killing the bacteria than individually. Additionally, where EFDG1 phage failed to kill EFDG1<sup>r</sup> , EFLK1 phage and the cocktail were effective against the WT and the phage-resistant mutant alone or in a mixed culture. This simulates the evolution of such mutants. We found the 1:1 ratio phage cocktail to be most effective, but we expect that ratios will vary and should therefore be determined on a case by case basis.

The reason for the phenotypic differences between EFDG1 and EFLK1 phages remains shrouded. One significant feature is the presence of 24 tRNA genes in EFDG1 phage, and their absence in EFLK1 phage. Theoretically, phages effective in the logarithmic stage (EFDG1), which benefits from elevated levels of host tRNA, should conceivably need to encode a small number of tRNA genes while phages effective in the stationary phage (EFLK1) should conceivably need a greater number. However, our findings indicated the opposite. The logarithmic efficient EFDG1 phage has more tRNA than the stationary efficient EFLK1 phage. Furthermore, the phages exhibited similar codon usage. In addition, the tRNA genes of EFDG1, ECP3, and phiEF24c phages are all encoded in a conserved region which is absent from EFLK1's genome. This suggests that the tRNA's are required for the efficient translation of the proteins encoded in that region.

<sup>5</sup>http://www.ebi.ac.uk/genomes/phage.html

Though one would expect to find a higher codon usage bias with respect to E. faecalis in this region, this is not the case. Despite a correlation between tRNA abundancy and virulence against bacteria (Bailly-Bechet et al., 2007) and evidence that deletion of tRNA genes reduced burst sizes and protein synthesis rates in the case of phage T4 of E. coli (Wilson, 1973), the role of tRNA in phages remains enigmatic.

P2 mutation in Orf31 (**Figure 6**), found to enhance adsorption in phiEF24c (Uchiyama et al., 2011), is absent in our phages. Nevertheless, Orf31 of EFDG1 phage appears to be less conserved in general compared to the other three phages of E. faecalis. This could mean that EFDG1 phage binds with a different affinity, or binds to another receptor all together. Further research is needed to determine whether the differences in Orf31 are responsible for the phenotypic differences between EFDG1 and EFLK1 phages. The effectiveness and complementary nature of EFDG1 and EFLK1 phages marks these phages as suitable for in vivo experiments toward the development of a high-efficacy anti-E. faecalis phage therapy.

## AUTHOR CONTRIBUTIONS

fmicb-09-00326 February 26, 2018 Time: 17:56 # 10

RH initiated the research, designed the experiments, analyzed the data, and wrote the paper. LK designed and performed the experiments and wrote the paper. MS performed some of the experiments. AD performed the bioinformatic analysis. SC-G designed the experiments and wrote the paper. DG helped to perform some of the experiments. NB initiated the research, designed the experiments, analyzed the data, and wrote the paper.

## FUNDING

We are grateful to the funding committee of the Rosetrees Trust for grant #0364821. Additionally, LK and MS are the recipients of the Ariane de Rothschild fellowship for women.

## REFERENCES


## ACKNOWLEDGMENTS

We would like to thank the following: Hebrew University's interdepartmental core unit at the Ein Kerem Campus for help in deep sequencing, Dr. Violeta Temper of Hadassah Hospital's Infectious Diseases Unit at Ein Kerem for bacterial isolates; Nahal Sorek Sewage Decontamination Institute's Yuri Veinstein for samples for phage isolation, and finally to Dr. Dov Glazer and Elisheva Dorfman for critical editing of this manuscript.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00326/full#supplementary-material

TABLE S1 | tRNA codon usage in EFDG1, EFLK1 and the other two E. faecalis phages as compared to that in E. faecalis. This is depicted by the frequency occurrence of the amino acid codon and the number of tRNA encoded by them in each case.



response to bacteriophage predation. PLoS Pathog. 3:e119. doi: 10.1371/journal. ppat.0030119


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Khalifa, Gelman, Shlezinger, Dessal, Coppenhagen-Glazer, Beyth and Hazan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Gas Plasma Pre-treatment Increases Antibiotic Sensitivity and Persister Eradication in Methicillin-Resistant Staphylococcus aureus

Li Guo<sup>1</sup> \*, Ruobing Xu<sup>2</sup> , Yiming Zhao<sup>2</sup> , Dingxin Liu<sup>1</sup> , Zhijie Liu<sup>1</sup> , Xiaohua Wang<sup>1</sup> , Hailan Chen<sup>3</sup> and Michael G. Kong1,3,4 \*

*<sup>1</sup> State Key Laboratory of Electrical Insulation and Power Equipment, Center for Plasma Biomedicine, Xi'an Jiaotong University, Xi'an, China, <sup>2</sup> School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, China, <sup>3</sup> Frank Reidy Center for Bioelectrics, Old Dominion University, Norfolk, VA, United States, <sup>4</sup> Department of Electrical and Computer Engineering, Old Dominion University, Norfolk, VA, United States*

### Edited by:

*Noton Kumar Dutta, Johns Hopkins University, United States*

### Reviewed by:

*Airat R. Kayumov, Kazan Federal University, Russia Xiancai Rao, Army Medical University, China*

### \*Correspondence:

*Li Guo guoli35@mail.xjtu.edu.cn Michael G. Kong mglin5g@gmail.com*

### Specialty section:

*This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology*

Received: *14 December 2017* Accepted: *08 March 2018* Published: *23 March 2018*

### Citation:

*Guo L, Xu R, Zhao Y, Liu D, Liu Z, Wang X, Chen H and Kong MG (2018) Gas Plasma Pre-treatment Increases Antibiotic Sensitivity and Persister Eradication in Methicillin-Resistant Staphylococcus aureus. Front. Microbiol. 9:537. doi: 10.3389/fmicb.2018.00537* Methicillin-resistant *Staphylococcus aureus* (MRSA) is a major cause of serious nosocomial infections, and recurrent MRSA infections primarily result from the survival of persister cells after antibiotic treatment. Gas plasma, a novel source of ROS (reactive oxygen species) and RNS (reactive nitrogen species) generation, not only inactivates pathogenic microbes but also restore the sensitivity of MRSA to antibiotics. This study further found that sublethal treatment of MRSA with both plasma and plasma-activated saline increased the antibiotic sensitivity and promoted the eradication of persister cells by tetracycline, gentamycin, clindamycin, chloramphenicol, ciprofloxacin, rifampicin, and vancomycin. The short-lived ROS and RNS generated by plasma played a primary role in the process and induced the increase of many species of ROS and RNS in MRSA cells. Thus, our data indicated that the plasma treatment could promote the effects of many different classes of antibiotics and act as an antibiotic sensitizer for the treatment of antibiotic-resistant bacteria involved in infectious diseases.

Keywords: cold atmospheric-pressure plasma, antibiotics resistance, methicillin-resistant Staphylococcus aureus, reactive oxygen species, reactive nitrogen species

## INTRODUCTION

Antibiotics are the primary treatment for infectious bacterial diseases (Li et al., 2017). Antibiotic resistance typically emerges several years after the development and use of a new antibiotic, usually within an average of 8 years after introduction (Schmieder and Edwards, 2012). Furthermore, infections with multidrug-resistant bacteria are occurring more frequently, and few or no drugs are available to combat them (Boucher et al., 2009; Wright, 2016; Gonzalez-Bello, 2017). One of the major multidrug-resistant bacteria is methicillin-resistant Staphylococcus aureus (MRSA), which is recognized as a leading cause of nosocomial infections (Shahsavan et al., 2012; Emaneini et al., 2017). Persisters are a small non-growing population of bacteria which could also escape from different antibiotics (Levin and Rozen, 2006). Persisters extend the duration of antibiotic treatment, cause the recurrence of infectious diseases, and the generation and ascent of antibiotic resistance (Levin and Rozen, 2006). Increased antibiotic resistance and the shortage of new antibiotics threaten global public health (Boucher et al., 2009; Dwyer et al., 2014). Therefore, strategies to combat antibiotic resistance, such as the development of new antibiotics or prolonging the lifespan of current antibiotics, are in high demand (Melander and Melander, 2017).

One strategy that has been explored recently is developing approaches that induce bacterial killing via the same mechanism as existing antibiotics. Antibiotics with different targets have been proposed to share a common mechanism of bactericidal activity—enhancing intracellular reactive oxygen species (ROS) in bacterial cells (Kohanski et al., 2007, 2010). Although ROS could be developed as antimicrobials to treat infectious diseases, treatment with these highly reactive molecules is problematic because ROS react non-selectively with such a range of critical, macromolecular targets, which could lead to "off-target" effects (Dharmaraja, 2017). Given that oxidative stress is also associated with antibiotic treatment, low levels of ROS could be used as an adjuvant to potentiate the antibacterial activity of commercial antibiotics (Brynildsen et al., 2013; Morones-Ramirez et al., 2013; Shen et al., 2016). A previous study showed that •OH induced in Ag+-treated bacteria potentiated the bactericidal activity of antibiotics against bacterial persisters and biofilms (Morones-Ramirez et al., 2013). Furthermore, fosfomycin, which acts by generating O •− 2 , was used to successfully treat MRSA infections when combined with many commercial antibiotics (Shen et al., 2016). These strategies relied on the addition of inorganic or organic chemicals, which would bring residues after the treatment.

Cold atmospheric-pressure plasma (referred to as "plasma") generates many reactive oxygen and nitrogen species (ROS and RNS), such as H2O2, <sup>1</sup>O2, O3, •NO, and •OH as well as electrons, ions, and photons. This form of plasma is atmospheric-pressure and near room temperature, thus it treats cells and tissues without thermal damage, making it attractive for a range of biomedical applications, such as bacteria inactivation (Fridman et al., 2008; Moreau et al., 2008; Kong et al., 2009; Kang et al., 2014). The FDA has authorized the use of at least three gas plasma-based products using "plasma biomedicine" technology, such as blood coagulation. In addition, there have been several phase-II clinical trials for plasma-based therapies, such as chronic wound healing, which is the promising applications of plasma in medicine (Isbary et al., 2010, 2012; Heinlin et al., 2013). In the treatment of wounds, the reactive species of plasma could not only kill the microorganisms in the infectious wounds and burns but also increase proliferation of fibroblasts and other cells (Lloyd et al., 2010). Bayliss et al. (2013) found that treating MRSA with sublethal doses of plasma restored the sensitivity of MRSA to trimethoprim, kanamycin, and oxacillin, but the utility of plasma treatment with other types of antibiotics was not studied.

In the present study, in order to further our understanding of the effect of plasma treatment on antibiotic sensitivity, MRSA was treated with both plasma and plasma-activated saline prior to exposure to multiple antibiotics. Sublethal treatment of MRSA with plasma increased the sensitivity of MRSA to seven antibiotics whilst also reducing the numbers of persisters. These results support the use of plasma as an antibiotic sensitizer for the treatment of antibiotic-resistant bacteria involved in infectious diseases.

## MATERIALS AND METHODS

### Plasma Device and Plasma Treatments

As shown in **Figure 1A**, the surface discharge structure of the plasma consisted of a high-voltage plane electrode, a liquid-facing grounded mesh electrode of a hexagonal shape and a dielectric layer (made of polytetrafluoroethylene) sandwiched between the two electrodes. The surface plasma was generated when a sinusoidal high voltage was applied and the discharge power density was 0.2 W/cm<sup>2</sup> with good mesh-to-mesh homogeneity (showed in the front view of the plasmas in **Figure 1A**). One milliliter MRSA suspension or saline in a Petri dish (diameter 35 mm) with the depth of 1 mm was placed under the plasma, whereas the air gap between the plasma and the liquid surface was about 8 mm. The gas plasma system was housed in a sealed organic glass box with a gas flow of helium and 1% artificial air (79% N<sup>2</sup> + 21% O2) at a constant rate of 4 L/min. The artificial air was used as the source of ROS and RNS, while the helium was used to enhance the production efficiency of those species as well as their fluxes on the treated samples via diffusion.

## Measurement of ROS and RNS Generated by the Plasma

The concentrations of H2O<sup>2</sup> and NO<sup>−</sup> 2 /NO<sup>−</sup> 3 in 0.9% NaCl were measured using hydrogen peroxide/peroxidase assay kit (Thermo Fisher Scientific) and nitrite/nitrate colorimetric assay kit (Cayman), respectively. •OH, <sup>1</sup>O2, •NO, O•− 2 , •NO2, and ONOO<sup>−</sup> were measured using an electron spin-resonance (ESR) spectroscopy (Bruker) together with relevant spin traps. The latter include: 100 mM DMPO (5,5-Dimethyl-1-pyrroline N-oxide, Dojindo), 5 mM MGD (N-(Dithiocarbamoyl)- N-methyl-D-glucamine, Dojindo), 10 mM TEMP (2,2,6,6- Tetramethylpiperidine, TCI), and 10 mM TEMPONE-H (1-Hydroxy-2,2,6,6-tetramethyl-4-oxo-piperidine, Enzo).

### Treatment of Staphylococcus aureus With Plasma or Plasma-Activated Saline

A single S. aureus ATCC33591 colony was grown in 4 ml of Mueller-Hinton (MH) broth (Oxoid) at 225 r.p.m. at 37◦C overnight, and the resulting overnight culture was diluted 1:100 in MH and incubated at 37◦C at 225 r.p.m. until an OD<sup>600</sup> of 0.6 was reached. The bacterial cells were collected by centrifugation, washed once with saline (0.9% NaCl) and resuspended in saline at an OD<sup>600</sup> of 2.0. S. aureus suspensions were treated with plasma directly, or the suspensions were centrifuged and resuspended in plasma-activated saline (saline treated with plasma for the indicated time), saline with 250µM H2O<sup>2</sup> <sup>+</sup> <sup>125</sup>µM NO<sup>−</sup> 2 + 375µM NO<sup>−</sup> 3 , or saline with 500µM H2O<sup>2</sup> + 250µM NO<sup>−</sup> <sup>2</sup> <sup>+</sup> <sup>750</sup>µM NO<sup>−</sup> 3 and then incubated for 30 min at room temperature. Samples were serially diluted with 0.9% NaCl, and 10 µl of each dilution was spotted onto MH agar plates and incubated overnight at 37◦C and the numbers of surviving bacteria were determined by counting the resulting CFUs.

### Determination of Antibiotic Sensitivity

The S. aureus suspensions prepared as described above were either untreated, treated with plasma directly for 30 or 40 s, or

centrifuged and resuspended in plasma-activated saline (saline treated by plasma for 30 or 40 s), followed by incubation for 30 min at room temperature. Untreated S. aureus, or cells treated with either plasma or plasma-activated saline (100 µl), were plated on MH agar plates. Next, antibiotic susceptibility tests were performed using Etest paper (Biomerieux) for tetracycline (0.016–256µg/ml), gentamycin (0.016–256µg/ml), clindamycin (0.016–256µg/ml), chloramphenicol (0.016–256µg/ml), ciprofloxacin (0.02–32µg/ml), rifampicin (0.02–32µg/ml), and vancomycin (0.016–256µg/ml) or the Kirby-Bauer method for clindamycin (250, 500, 1,000, and 2,000 µg). Then the plates were cultured overnight at 37◦C and bacteriostatic rings were analyzed. The MICs of antibiotics were measured by the microdilution assay using 8–0.016 × MIC of untreated MRSA as described (Lepe et al., 2014).

### Quantification of Persister Survival

S. aureus treated with plasma or plasma-activated saline was diluted six times with MH medium containing antibiotics. The final antibiotic concentrations, corresponding to 10× the minimum inhibitory concentration (MIC), were as follows: tetracycline, 500µg/ml (MP Biomedicals); gentamycin, 200µg/ml (Sigma); clindamycin, 5,000µg/ml (TCI); chloramphenicol, 500µg/ml (MP Biomedicals); ciprofloxacin, 50µg/ml (Sigma); rifampicin, 4µg/ml (Sigma); and vancomycin, 50µg/ml (Sigma). The numbers of surviving bacteria were determined at the time points indicated by harvesting 100 µl of bacterial culture, which was centrifuged at 10,000× g for 1 min. The resulting pellet was resuspended in 100 µl 0.9% NaCl. Samples were serially diluted, and 10 µl of each dilution was spotted onto MH agar plates and incubated overnight at 37◦C and the numbers of surviving bacteria were determined by counting the resulting CFUs.

## Detection of Reactive Species in MRSA

The probes 3′ -(p-aminophenyl) fluorescein (APF, Sigma), 3 ′ -(p-hydroxyphenyl) fluorescein (HPF, Sigma), and MitoSOXTM Red mitochondrial superoxide indicator (ThermoFisher) were incubated with S. aureus cultures, at a final concentration of 5µM, and trans-1-(2′ -methoxyvinyl)pyrene (tMVP, J&K Scientific) was used at a final concentration of 10µM at 37◦C for 30 min. Cultures were collected by centrifugation, washed with saline (0.9% NaCl) three times and resuspended in saline at an OD<sup>600</sup> of 2.0. S. aureus suspensions were untreated, treated with plasma directly for 40 s, or the suspensions were centrifuged and resuspended in plasma-activated saline (saline treated with plasma for 40 s). Immediately after the treatments, the fluorescence intensities were detected using a microplate reader (Thermo Scientific Varioskan Flash) at the excitation and emission wavelengths [APF and HPF: 490/515 nm; superoxide indicator: 510/580 nm; trans-1-(2′ -methoxyvinyl)pyrene: 405/460 nm] of each probe.

## RESULTS

### Plasma-Induced Aqueous ROS/RNS

MRSA or saline treated with the plasma device was shown in **Figure 1A**. Gaseous ROS and RNS were generated by the surface discharge, and some of them would diffuse across the air gap and then dissolved into the liquids. The plasma-induced aqueous ROS and RNS dissolved into saline were measured after the plasma treatment for 30 s and 40 s. Long-lived species <sup>H</sup>2O2, NO<sup>−</sup> 2 , and NO<sup>−</sup> 3 , as well as the short-lived species •OH, <sup>1</sup>O2, •NO, and ONOO−, were detected in the saline. The concentrations of aqueous H2O2, NO<sup>−</sup> 2 , and NO<sup>−</sup> 3 after the plasma treatment for 40 s were 23, 14, and 102µM, respectively (**Figure 1B**). Electron spin resonance (ESR) spectroscopy was used for the measurement of short-lived species, in which the results were the concentrations of spin adducts, only reflecting the relative concentrations of the specific ROS and/or RNS. The concentrations of spin adducts DMPO-OH, TEMPO, Nitrocyl-Fe, and TEMPONE after the plasma treatment for 40 s were 0.1, 69, 21, and 37µM (**Figure 1B**). The spin adducts concentrations after the plasma treatment for 30 s were lower, such as TEMPO was 30% lower than that after the plasma treatment for 40 s. These results suggested that the concentration of aqueous •OH was very low, and the concentration of aqueous <sup>1</sup>O<sup>2</sup> should be much larger. The plasma generated many species ROS and RNS, which were thought to play a crucial role in the biological effects.

## Sublethal Treatment With Plasma Increased the Sensitivity of MRSA to Antibiotics

The sublethal dose (reduction about 50% viability of MRSA) of plasma and plasma-activated saline were determined by exposing the bacteria to plasma for increasing times or incubating with saline treated with plasma for increasing times and subsequently determining bacterial viability. The treatment of plasma for 40 s or the saline activated by plasma for 40 s lead to about 50% death of MRSA cells (Figure S1). For safety, a lower dose was also employed, so a slightly lower dose (plasma treatment for 30 s) that lead to about 40% death of MRSA cells and the LD<sup>50</sup> dose (plasma treatment for 40 s) were both used to detect the effects on antibiotic sensitivity.

To study the effect of plasma on the sensitivity of MRSA to the antibiotics, MRSA were sublethally pre-treated with plasma or plasma-activated saline followed by incubation with a range of antibiotics including tetracycline, gentamycin, clindamycin, chloramphenicol, ciprofloxacin, rifampicin, and vancomycin. The minimal inhibitory concentration (MIC) of tetracycline for untreated MRSA measured by Etest paper was ∼32µg/ml, whilst the MIC of tetracycline for MRSA treated with plasma for 30 or 40 s was 10-fold lower, at ∼3µg/ml (**Figure 2A**). In the second approach, the MICs were measured by the microdilution assay, and the MIC of tetracycline against MRSA treated with plasma for 40 s was 8-fold lower than that against untreated MRSA (**Table 1**). The effects of plasma pre-treatment on MRSA sensitivity to antibiotics were independent of the particular antibiotic used, and the increase in sensitivity was observed for most of the antibiotics tested. The MICs of gentamycin against untreated MRSA and MRSA treated with plasma for 30 and 40 s measured by Etest paper were ∼1.5, 1.0, and 0.5µg/ml, respectively, and the MIC of gentamycin against MRSA treated with plasma for 40 s was 16-fold lower than that against untreated MRSA as indicated by the microdilution assay (**Figure 2A**; **Table 1**). The MICs of chloramphenicol, ciprofloxacin, rifampicin, and vancomycin against MRSA treated with plasma also decreased in varying degrees (**Figure 2A**; **Table 1**). Plasma-treated MRSA did not appear to be susceptible to treatment with clindamycin at the tested concentration (**Figure 2A**). Two possible explanations for this were that either the plasma treatment did not change the antibiotic sensitivity of MRSA to clindamycin or the MIC of clindamycin against MRSA was beyond the range of the Etest. Thus, MRSA susceptibility to clindamycin was measured at a higher antibiotic dose using the Kirby-Bauer method. The diameter of the bacteriostatic ring of untreated MRSA to 2,000 µg clindamycin was about 16 mm, and that of the MRSA treated for 30 and 40 s were 23 and 26 mm, respectively (Figures S2A,B). The microdilution assay also demonstrated that the MICs of clindamycin against MRSA treated with plasma for 30 and 40 s was 2- and 4-fold lower, respectively, than that against untreated MRSA (**Table 1**). These data indicated that sublethal plasma treatment to MRSA restored the sensitivity to five different classes of antibiotics.

Next, the effects of plasma-activated saline upon changing the sensitivity of MRSA to these antibiotics were detected. The minimal inhibitory concentration (MIC) of tetracycline against untreated MRSA was ∼32µg/ml, whilst that for MRSA treated with saline activated by plasma for 30 or 40 s was 10 fold lower, at ∼3µg/ml (**Figure 2B**). The MIC of tetracycline against MRSA treated with saline activated by plasma for 40 s was also 8-fold lower than that against untreated MRSA as indicated by the microdilution assay (**Table 1**). The MICs of gentamycin, clindamycin, chloramphenicol, ciprofloxacin, rifampicin, and vancomycin against MRSA treated with plasmaactivated saline were also lower than those against the untreated MRSA, demonstrating that plasma-activated saline exhibited the same effect in increasing the sensitivity to five different classes of antibiotics (**Figures 2B**, Figures S2C,D; **Table 1**).

### Plasma Pre-treatment Promoted Eradication of MRSA Persisters

Given that pre-treatment of MRSA with plasma or plasmaactivated saline decreased the MIC of multiple antibiotics, we next asked whether either of these pre-treatment conditions could also promote the killing of persisters. To test this, untreated and sublethal dose plasma-treated MRSA were incubated with different antibiotics, as indicated (**Figure 3A**). Incubation of untreated and double-diluted untreated MRSA with the different antibiotics generally exhibited typical killing curves, with bacterial numbers decreasing from 2 × 10<sup>8</sup> to 10<sup>4</sup> -10<sup>6</sup> CFUs, with a small fraction of persisters detected after 3 days of antibiotic exposure. Of note, the numbers of untreated MRSA and double-diluted untreated MRSA that incubated with rifampicin increased after 1 day. These results indicated that the decrease of the initial MRSA numbers did not greatly influence the killing

TABLE 1 | Antibiotic susceptibilities of *Staphylococcus aureus* treated with plasma and plasma-activated saline.


*<sup>a</sup>MICs were determined by broth microdilution method. The results were read after 24 h of incubation at 37*◦*C.*

curves of antibiotics. In contrast, plasma-treatment reduced the number of MRSA persisters to below the limit of detection when incubated with tetracycline, gentamycin, clindamycin, chloramphenicol, and rifampicin within only 1 day of exposure (**Figure 3A**). A similar decrease was also observed within 2 and 3 days of incubation with ciprofloxacin and chloramphenicol, respectively (**Figure 3A**). For vancomycin, plasma treatment for 30 and 40 s effectively reduced the numbers of remaining MRSA recovered during exposure to this antibiotic to 10<sup>4</sup> and 10<sup>3</sup> CFUs, respectively, though these did not drop below the detection limit (**Figure 3A**). Treatment with plasma-activated saline exhibited the same effect of inactivating persisters as observed by direct plasma treatment (**Figure 3B**). The numbers of plasma-activated saline treated MRSA recovered upon

exposure to antibiotics tetracycline, gentamycin, clindamycin, and rifampicin dropped to below the detection limit, but plasmaactivated saline similarly only decreased the numbers of MRSA to 10<sup>3</sup> CFUs upon exposure to vancomycin (**Figure 3B**). These results indicated that both plasma and plasma-activated saline treatment promoted the eradication of MRSA persisters by antibiotics.

### Reactive Species in MRSA

Plasma-treated solutions contain many active species, such as the long-lived species H2O2, NO<sup>−</sup> 2 , and NO<sup>−</sup> 3 as well as the short-lived species •OH, <sup>1</sup>O2, •NO, O•− 2 , and ONOO<sup>−</sup> (**Figure 1B**). To evaluate the impact of long-lived species, a mixture of H2O<sup>2</sup> (500µM), NO<sup>−</sup> 2 (250µM), and NO<sup>−</sup> 3 (750µM) was used to treat the MRSA. H2O2, NO<sup>−</sup> 2 , and NO<sup>−</sup> 3 treatment did not change the antibiotic susceptibility of MRSA and the inactivation of persisters by antibiotics, indicating that the three long-lived species are not likely the direct and driving factor (Figure S3). Hence, short-lived species were considered more important, and the ROS and RNS in MRSA cells were measured. MRSA were incubated with 3′ -(p-aminophenyl) fluorescein (APF) for •OH, ClO−, and ONOO−, HPF for •OH and ONOO−, MitoSOXTM Red mitochondrial superoxide indicator for O•− 2 and trans-1-(2′ methoxyvinyl)pyrene (tMVP) for <sup>1</sup>O2, then exposed to plasma or plasma-activated saline for 40 s, and the levels of different ROS/RNS determined by changes in the fluorescence intensities (**Figure 4**). After plasma treatment, the fluorescence intensities of the four probes increased. The fluorescence intensities of MRSA incubated with APF and HPF increased slightly after 30 min of plasma treatment, by 28 and 25%, respectively, which indicates an increase in the levels of •OH, ClO−, and ONOO<sup>−</sup> (**Figures 4A,B**). Fluorescence of MRSA incubated with MitoSOXTM Red mitochondrial superoxide indicator increased by 77%, and the trans-1-(2′ -methoxyvinyl)pyrene (tMVP) for <sup>1</sup>O<sup>2</sup> increased by 130% after 30 min of plasma treatment (**Figures 4C,D**). The fluorescence intensities of the probes which incubated with plasma-activated saline-treated MRSA were slightly weaker than those of the direct plasma-treated samples. These data suggested that the plasma and plasma-activated saline treatment induced the increases of many species of ROS and RNS in MRSA cells.

FIGURE 4 | ROS and RNS levels increased in plasma-treated and plasma-activated saline-treated MRSA. MRSA incorporated with APF (A), HPF (B), superoxide indicator (C), and tMVP (D), which measured the ROS or RNS as indicated, treated with plasma for 40 s or saline treated with plasma for 40 s. Then the fluorescence intensities of were measured in plasma-treated, plasma-activated saline-treated, and untreated MRSA. Data are representative of three independent experiments. Error bars represent s.d.

## DISCUSSION

In this study, we demonstrated that treating MRSA sublethally with plasma-generated ROS and RNS decreased the MICs of several antibiotics and increased persister eradication, along with increases in the levels of ROS and RNS in MRSA cells. Plasmaactivated saline had the same effect upon the antibiotic sensitivity of MRSA and persister inactivation as direct plasma treatment, suggesting that the plasma-generated ROS and RNS could be applied in both gaseous or aqueous form depending on the mode of application. The short-lived species in the plasma-activated saline had short half-lives, but they could react and generate longlived species, which also could generate to short-lived species reversibly, such as ONOO<sup>−</sup> generated from NO<sup>−</sup> 3 (Oehmigen et al., 2011; Liu et al., 2017). Plasma generated various ROS, which are involved in a great many chemical reactions (Oehmigen et al., 2011). The underlying reactions and detailed mechanisms of plasma-activated saline are still not well understand and require further study.

Unlike other treatments that have previously been used as antibiotic adjuvants to enhance ROS production in bacterial cells, plasma induces the production of a wider range of ROS species (Morones-Ramirez et al., 2013; Shen et al., 2016). The ROS and RNS generated by the plasma constitute a complex mix of products, including •OH, <sup>1</sup>O2, O•− 2 , ONOO−, and ClO<sup>−</sup> (Liu et al., 2016). Subsequently, these many different ROS or RNS species were also detected in plasma-treated bacterial cells. The intracellular ROS and RNS could induce oxidative stresses, such as damaging lipids, proteins and DNA by •OH, <sup>1</sup>O2, and O•− 2 , as well as protein damages by ONOO<sup>−</sup> and ClO<sup>−</sup> (Davies, 2016). O•− 2 could be detoxified by endogenous antioxidants of the oxidative response, but no enzyme can detoxify •OH or <sup>1</sup>O2, and MRSA could not detoxify all the multiple reactive species induced by plasma (Vatansever et al., 2013). It was speculated that the compound damages stimulated multiple response pathways and kept MRSA busy with repairing, which contributed to the effects of antibiotics.

Reactive species generated by plasma could also induce damages in eukaryotic cells, subsequently, the safety and toxicity of this application should be considered. The CC<sup>50</sup> of the plasma treatment used in this study for human primary dermal fibroblasts was about 40 s treatment, which was close to the LD<sup>50</sup> on MRSA (Figure S4A). Comparing with bacterial cells, the cultured cells are more easily to be inactivated in vitro because of the lack of cell wall. However, the plasma treatment did not increase micronuclei formation in fibroblast cells (Figures S4B,C). Coincidently, a newly published paper found similar results on lymphocyte TK6 cells (Bekeschus et al., 2018). Further, the olive tail moment of plasma-treated cells exhibited little difference with that of untreated cells (Figures S4D,E). So the plasma treatment did not remarkably increase the mutagenicity of fibroblasts as demonstrated by both micronucleus assay and comet assay. Besides, Maisch et al. (2012) showed that gas plasma could efficiently inactivate S. aureus and Escherichia coli on pig skin without inducing morphological changes or damage-related apoptosis. Clinical trials also demonstrated that 5 min daily treatment with plasma decreased bacteria in chronic wounds of patients without side effects (Isbary et al., 2010). These studies demonstrated the safety of plasma treatment under limited conditions. Many topical biocides are toxic, and the plasma could be developed as alternative, especially the plasma-activated saline could be the save alternative (Wales and Davies, 2015). When used as a sensitizer with antibiotics as was done in this study, the doses of plasma were much lower than that used for bacteria

### REFERENCES


inactivation, which would reduce the risk of toxicity and improve the safety.

In conclusion, sublethal treatment with plasma-generated ROS and RNS decreased the MICs of several antibiotics and increased persister eradication, along with increases in the levels of ROS and RNS in MRSA. Plasma and plasma-activated saline could be explored as a novel antibiotic sensitizer to generate oxidative stress to combat the increasing problem of antibiotic resistance. Further studies are needed to test these methods against other multidrug-resistant bacteria and to elucidate the underlying mechanism.

## AUTHOR CONTRIBUTIONS

LG, RX, and YZ designed and executed most of the experiments and analyzed the data. ZL assisted in the execution of some experiments. LG, DL, and MK wrote the manuscript. MK, LG, DL, XW, and HC reviewed and approved the final version of the manuscript.

### ACKNOWLEDGMENTS

This work was supported by National Natural Science Foundation of China (51477136 and 51521065), Doctoral Fund of Ministry of Education of China (2017M613106), Doctoral Fund of Ministry of Education of Shaanxi Province, State Key Laboratory of Agricultural Microbiology (AMLKF201705), State Key Laboratory of Electrical Insulation and Power Equipment (EIPE14123), and Fundamental Research Funds for the Central Universities.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00537/full#supplementary-material


chronic wounds: results of a randomized controlled trial. Br. J. Dermatol. 167, 404–410. doi: 10.1111/j.1365-2133.2012.10923.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guo, Xu, Zhao, Liu, Liu, Wang, Chen and Kong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Liposome Entrapment of Bacteriophages Improves Wound Healing in a Diabetic Mouse MRSA Infection

### Sanjay Chhibber\*, Jasjeet Kaur and Sandeep Kaur

Department of Microbiology, Panjab University, Chandigarh, India

Diabetic populations are more prone to developing wound infections which results in poor and delayed wound healing. Infection with drug resistant organisms further worsen the situation, driving searches for alternative treatment approaches such as phage therapy. Major drawback of phage therapy, however, is low phage persistence in situ, suggesting further refinement of the approach. In the present work we address this issue by employing liposomes as delivery vehicles. A liposome entrapped phage cocktail was evaluated for its ability to resolve a Staphylococcus aureus-induced diabetic excission wound infection. Two characterized S. aureus specific lytic phages, MR-5 and MR-10 alone, in combination (cocktail), or entrapped in liposomes (versus as free phages) were assesed for their therapeutic efficacy in resolving diabetic wound infection. Mice treated with free phage cocktail showed significant reduction in wound bioburden, greater wound contraction and faster tissue healing than with free monophage therapy. However, to further enhance the availability of viable phages the encapsulation of phage cocktail in the liposomes was done. Results of in vitro stability studies and in vivo phage titer determination, suggests that liposomal entrapment of phage cocktail can lead to better phage persistence at the wound site. A 2 log increase in phage titre, however, was observed at the wound site with liposome entrapped as compared to the free phage cocktail, and this was associaed with increased rates of infection resolution and wound healing. Entrapment of phage cocktails within liposomes thus could represent an attractive approach for treatment of bacterial infections, not responding to antibiotis as increased phage persistence in vitro and in vivo at the wound site was observed.

Keywords: bacteriophage, diabetes, phage cocktail, liposome, alloxan

## INTRODUCTION

Diabetes is a chronic disease that manifests in the form of a chronic hyperglycemia which is associated with a plethora of complications including visual impairment, blindness, kidney disease, nerve damage, amputation, heart disease, and stroke (Loghmani, 2005). WHO has projected diabetes as the 7th leading contributor to death by 2030 (Mathera and Loncar, 2006). In addition, the total number of people with diabetes is projected to rise from 171 million in 2000 to 366 million in 2030 (Wild et al., 2004). Diabetic patients are five times more susceptible to fungal and bacterial infections (Axelrod, 1985). This is due to several factors that include a high glucose level which

### Edited by:

Stephen Tobias Abedon, The Ohio State University, United States

### Reviewed by:

Rodolfo García-Contreras, Universidad Nacional Autónoma de México, Mexico Xiancai Rao, Army Medical University, China

> \*Correspondence: Sanjay Chhibber sanjaychhibber8@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 12 June 2017 Accepted: 12 March 2018 Published: 29 March 2018

### Citation:

Chhibber S, Kaur J and Kaur S (2018) Liposome Entrapment of Bacteriophages Improves Wound Healing in a Diabetic Mouse MRSA Infection. Front. Microbiol. 9:561. doi: 10.3389/fmicb.2018.00561

encourages bacterial growth. Secondly, diabetes decreases a body's blood flow, slowing healing of abrasions, open wounds and other injuries, thus making diabetic individuals more prone to infection (King, 2001). Finally, peripheral neuropathy, a primary complication of diabetes, results in loss of sensation (Pendsey, 2010) and has a central role in the development of foot and wound infections (Bader, 2008; Powlson and Coll, 2010). According to the American Diabetes Association (ADA), 25% of people with diabetes suffer from a wound problem during their lifetime, including infections. Staphylococcus aureus is one of the most common pathogens isolated from wounds in diabetic patients. Further, with the spread of methicillin resistant strains of S. aureus (MRSA), management of such infections in diabetic population has become challenging. Diabetics who contract MRSA are at a greater risk of harboring more serious, slow-healing wounds (Rogers and Bevilacqua, 2009).

With time, the bacteria in diabetic infections tend to evolve greater resistance to antibiotics (Xavier et al., 2014). Consequently, having a comprehensive treatment plan for MRSA is crucial. Phage therapy seems to be a good option for addressing these problems (Barrow and Soothill, 1997; Bull et al., 2002; Wills et al., 2005; Capparelli et al., 2007; Sunagar et al., 2010; Hsieh et al., 2011; Kumari et al., 2011; Chhibber et al., 2013; Chadha et al., 2016). Fish et al. (2016), on the basis of the success obtained with phage therapy of a diabetic foot infected with S. aureus, have suggested that in the future controlled clinical trials can be conducted. The ability of phages to rapidly lyse infected bacteria, their specificity and effectiveness against multidrug-resistant pathogenic bacteria, their ability to selfreplicate (auto dosing) (Kumari et al., 2009; Chhibber et al., 2013; Kaur et al., 2014), their natural abundance and their proven clinical safety makes this therapy worth considering in immunocompromised individuals such as diabetic patients (Carlton et al., 2005; Hanlon, 2007). Phage therapy in many cases nevertheless can benefit from approaches that can increase phage retention and persistence at the target sites.

Toward improving phage persistence and retention at diabetic wound sites, we propose here the use of liposome entrapment of phages. Liposomes are composed of natural lipids which are biodegradable, non-immunogenic and non-toxic. Along with these advantages, their structural versatility enables us to design a number of liposome-based formulations (Gregoriadis, 1995). Liposomes also mimic biological membranes in terms of their structure and behavior, which enables them to penetrate the epidermal barrier to a greater extent and thus can be used effectively in treating skin infections (Hua, 2015). Use of drug delivery systems (DDSs) such as liposomes also helps to improve the pharmacological properties of conventional ("free") drugs by altering their pharmacokinetics (PK) and bio-distribution (BD) (Zhang et al., 2013; Zylberberg and Matosevic, 2016). Workers in the past have reported the use of liposomes with entrapped antibiotics or other wound healing agent in treating wound infections (Roesken et al., 2000; Kurilko et al., 2009; Fukui et al., 2012). In recent studies, entrapment of phages in liposomes has also been reported (Colom et al., 2015; Nieth et al., 2015; Singla et al., 2015; Chadha et al., 2017). However, none of these workers have studied the use of liposomes to enhance the stability and persistence of phages for the better therapeutic outcome after treatment of diabetic wound infection caused by MRSA. The present study therefore investigated the therapeutic use of liposome entrapped phage cocktail in treating MRSA mediated skin wound infection in diabetic mice.

## MATERIALS AND METHODS

## Bacterial Strains and Phage Used

Staphylococcus aureus ATCC 43300 (MRSA) from ATCC, Manassas, VA, United States was used in this study. S. aureus specific phages, MR-5 and MR-10, both of which have been isolated and characterized in our laboratory as lytic, dsDNA, tailed phages belonging to Myoviridae family, order Caudovirales were used in the present study (Kaur et al., 2012; Chhibber et al., 2013).

## Preparation of Purified High Titer Phage Stocks (MR-5 and MR-10)

High-titer MR-5 and MR-10 suspension was prepared according to the method described by Langley et al. (2003). The inoculated broth was centrifuged and supernatant containing phage collected and passed through a 0.22-µm-pore-size filter. The filtrate was concentrated by using Millipore Labscale TFF system (Pellicon) and 10 K (polyethersulfone) membrane until the sample volume was reduced to 25–30 ml. DNase I (0.25 mg/ml) was added and kept for 1 h at 37◦C to digest the free DNA. NaCl (final concentration of 1 M) and polyethylene glycol (PEG) 8000 were added to the concentrated lysate and kept at 4◦C overnight. The precipitate was collected, dialyzed against phosphate buffered saline (PBS) overnight at 4◦C, passed through an0.22-µm-pore-size filter and stored at 4◦C till further use. The phage titer was determined in the final product.

### Liposomes

Free phage preparations were entrapped within liposomes to increase their persistence and efficacy. For this a series of optimization experiments were carried to select the exact liposomal formulation for in vivo use.

### **Preparation**

The method of Bhatia et al. (2004) was used for the preparation of phage cocktail-loaded liposomes. For preparation of cationic liposomal formulation of phage, phosphatidylcholine:cholesterol: tween 80:stearylamine (7:3:1:0.5) was dissolved in chloroformmethanol mixture (2:1 v/v). Thin film was prepared by a rota-evaporator [Heidolph, Hei-VAP Advantage (ML)] using a hydration temperature of 40◦C. Phage suspension (10 ml prepared in PBS [pH 7.2]) was added to the thin film at 40◦C and rotated for 10 min to detach the film from the glass wall. The suspension was left overnight at room temperature for swelling. The next day the dispersion is sonicated in a water bath sonicator for 30 min.

### **Characterization**

The morphological characteristics of liposomes were monitored viz. shape, uniformity, and structure by using optical microscope

(Olympus CH20i) at suitable magnification (100X). Morphological analysis of liposomes loaded with phage cocktail was then done using TEM following the procedure of Goodridge et al. (2003). Drops of ultra-centrifuged phage samples (1,00,000 g for 2 h, 4◦C; L-80, Beckman Instrument, Switzerland) were dropped on nitrocellulose coated grids (diameter, 3 mm; 300 meshes), stained with 2% (w/v) potassium phosphotungstate (pH: 6.8–7.2) for 10 s, and examined under a transmission electron microscope (TEM) (Hitachi H 7500, Tokyo, Japan) at Sophisticated Analytic Instrumentation Facility (SAIF), Panjab University, Chandigarh, India. Particle size and polydispersity index of liposomes was measured by dynamic light scattering (DLS) technique on Beckman coulter instrument (Delsa Nano C) at Department of Chemistry, Panjab University, Chandigarh.

The phage entrapment efficiency of liposomes was determined by modified protamine aggregation method (Gulati et al., 1998). Total phage count was determined by adding Triton × (0.02%) in a ratio of 1:1 v/v so as to disrupt liposomes and release entrapped phage particles. The mixture was then centrifuged (10,000 × g for 10 min) and total number of phages in the supernatant was determined using plaque assay. Free phages were separated by protamine aggregation method where protamine sulfate was added at a final concentration of 20 mg/ml. The liposomal phage preparation with added protamine sulfate was allowed to stand overnight. Supernatant was collected and the number of free phage particles was determined using plaque assay. Phage entrapment efficiency was then calculated as total number of non-free phages divided by total number of phages, presented as a percentage.

### **Animal experiments**

Animal experimental protocols were approved by the Institutional Animal Ethics Committee (Approval ID: IAEC/411) of Panjab University, Chandigarh, India and performed in accordance with the guidelines of Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Government of India, on animal experimentation. All efforts were made to minimize the suffering of animals.

BALB/c female mice, 4–6 weeks old weighing 20–25 g were used in this study. The animals, obtained from Central Animal House, Panjab University, Chandigarh, India, were kept in polycarbonate cages housed in well-aerated rooms with a 12-h light/12-h dark cycle at 25 ± 2 ◦C, and fed with standard rodent diet and water ad libitum. Diabetes as per the protocol of Pan and Liu (2008) was induced by giving two intra-peritoneal injections of alloxan monohydrate (150 mg/kg body weight) at 48-h intervals in mice fasted overnight. One week after the second injection, blood glucose levels (both random as well as fasting) were recorded. The blood sample was obtained by tail clipping method of Oladeinde et al. (2007) and glucose levels were checked using pre-calibrated Glucometer (Optium Xceed).

### Wound Model

A diabetic excision wound model was developed following the method of Mendes et al. (2012). S. aureus 43300 cultivated overnight in brain heart infusion broth was pelleted and washed twice with PBS. Bacterial suspension was adjusted to achieve a cell density corresponding to a range of bacterial inoculums (10<sup>8</sup> , 10<sup>9</sup> , and 10<sup>10</sup> CFU/ml). The number of CFU/ml was confirmed by quantitative plate count. Diabetic BALB/c mice were taken and distributed in four different groups of 14 animals (n = 14) each. The mice were anesthetized by giving 0.1 ml of ketamine-xylazine (50–5 mpk) intraperitoneally. Hair was removed by using commercially available hair removing cream followed by disinfection of the dorsal surface with 70% alcohol. A punch biopsy instrument (diameter 5 mm, accu punch) was used to create a full thickness round wound extending through the panniculus carnosus. The wounds created in all the animals were left open and animals were given ibuprofen suspension (20 mg/ml) in drinking water. Each of the three groups received different doses of inoculum. Animals of the fourth group received same volume of PBS injected into their wound. Two mice from each group were sacrificed on days 1, 3, 5, 7, 10, 15, and 20 post-bacterial challenge by cervical dislocation. Complete epithelial and dermal compartments of wound margins, granulation tissue and adjacent muscle and subcutaneous fat tissue was excised from each individual wound and processed. The tissue was homogenized and dilutions of the homogenates were plated to determine the bacterial burden.

### Free Phage Efficacy (Study 1)

In vivo work was divided into two studies. In the first study the therapeutic potential of free phages (monophage therapy as well as cocktail of phages) not involving any delivery system was assessed. For further clarity, phage preparations not entrapped within liposomes have been referred as free monophage (FMP) for, e.g., FMP-MR-5 and FMP-MR-10 and free cocktail of phages (FCP).

For the first study, the therapeutic potential of S. aureus 43300 specific phages FMP-MR-5, FMP-MR-10 and their cocktail, FCP, was evaluated in diabetic BALB/c mice. The phages were administered locally at a dose of 10<sup>9</sup> PFU/50 µl, at MOI10 as based on a titer of applied phages of 2 × 10<sup>10</sup> PFU/ml, into the wound on its dorsal surface. Diabetic BALB/c mice were randomly divided into five groups (n = 14 per group; as described immediately below) and seven extra animals were put in the infection control group to check the bacterial load on days 1 and 3. In all cases diabetic mice were first infected with S. aureus 43300 (10<sup>8</sup> CFU/50 µl) with treatments following 30-min later.

Group 1 (untreated infection control): No phages or clarithromycin applied.

Group 2: Local administration of phage FMP-MR-5 at MOI10.

Group 3: Local administration of phage FMP-MR-10 at MOI10.

Group 4: Local administration of the phage cocktail, FCP (each phage at MOI10 mixed in an equal ratio).

Group 5: Two I/P doses of clarithromycin (10 mpk) given in 12-h intervals.

All of the animals involved in Study 1 were monitored daily for mortality post-bacterial challenge. In addition to mortality,

wound bioburden, and phage titer was assessed in surviving animals on different days.

### Liposome Entrapped Phage Efficacy (Study 2)

In the second study, liposome entrapped phage cocktail was evaluated for its potential in treatment of diabetic wound infection. Phage cocktail (MR-5 + MR-10, 1:1) entrapped within liposomes shall be referred as liposomes entrapped cocktail of phages (LCP). Results were compared with the groups instead treated with FCP and with clarithromycin. Diabetic BALB/c mice with dermal wounds in this study were randomly divided into a total of four groups (n = 14). As above, in all cases diabetic mice were first infected with S. aureus 43300 (10<sup>8</sup> CFU/50 µl) with treatments following 30-min later.

Group 1 (untreated infection control): No phages or clarithromycin applied.

Group 2: Local administration of FCP (each phage at MOI10 mixed in an equal ratio).

Group 3: Local administration of LCP (each phage at MOI10 mixed in an equal ratio).

Group 4: Two I/P doses of clarithromycin (10 mpk) given in 12-h intervals.

All animals in Study 2 were monitored daily for mortality post-bacterial challenge. In addition to mortality, wound bioburden, phage titer, wound contraction rate, MPO, Masson's trichrome staining, histopathological examination, and period of epithelization was also assessed in surviving animals on different days.

### **Wound microbiology and rate of healing**

In both Studies 1 and 2, two mice from each group were sacrificed on days 1, 3, 5, 7, 10, and 15 post-infection by cervical dislocation. Bacterial load (wound bioburden) was assessed as per the method of Park et al. (2009). After disinfection, the wound tissue was carefully excised and homogenized in 1 ml of PBS. This tissue homogenate was then serially diluted and plated for CFUs. Phage titer was determined (in terms of PFU/ml) by centrifuging the homogenate (10,000 g for 10 min at 4◦C). The supernatant was filtered through 0.45 µm pore sized filter and the filtrate so obtained was serially diluted and plated on a lawn of S. aureus 43300.

Rate of wound contraction, as a percentage, was calculated by measuring the size of the wound (diameter in mm) on each day of all the animals involved in Study 2 using vernier caliper. This data is presented as percentage decrease in wound size. Thus, (initial wound size – wound size on a specific day) × 100. The number of days required for wound healing and eschar to fall off leaving no raw wound behind was taken as period for epithelization.

### **Myeloperoxidase (MPO) estimation**

Two mice from each group (same groups as those described for liposome entrapped phage cocktail protection study, i.e., Study 2 with 14 animals per group) were sacrificed and their wound tissue homogenized. Homogenized samples were processed for MPO determination as per the method of Greenberger et al. (1996). The absorbance was read immediately at 490 nm over a period of 4 min. MPO was calculated as the change in optical density (O.D) × dilution factor (D.F).

### **Masson's trichrome staining**

To assess the wound healing after treatment with LCP, FCP and clarithromycin, collagen formation was checked in each animal group of Study 2. Collagen is used as an essential marker in wound healing studies as it restores the integrity of the skin. Therefore, collagen formation was observed in the skin wound tissue on different days. Samples were fixed on different glass slides, stained as per the protocol<sup>1</sup> and examined under an Olympus microscope (at 100X) to evaluate the amount of collagen formation.

### **Histopathological examination**

Tissue damage and healing following treatment with LCP, FCP, and clarithromycin in Study 2 was assessed on the basis of histopathological examination of infected wound tissue according to the method of Brans et al. (1994). The sections were picked on separate slides, stained with hematoxylin and eosin (Hi-Media, Mumbai) and then examined under a microscope to evaluate the extent of tissue damage and recovery.

### Statistical Methods

All the data is expressed as mean ± standard deviation of more than two experimental values for every variable. The statistical significance of difference between groups was determined by Student's t-test (two groups), one-way ANOVA followed by a Tukey test using Sigma Stat, Graph pad prism (GraphPad Software, San Diego, CA, United States). P-value of less than 0.05 was considered statistically significant.

## RESULTS

### Optimization of Liposomal Formulation and Stability

For phage cocktail entrapment, three different liposomal formulations (7:3:1, 8:2:1, 9:1:1) were made using different ratios of phospholipid phosphatidylcholine (PC), cholesterol (CHOL), and Tween-80. Further, to increase the physical stability of liposomal preparation; positive charge inducer stearylamine (SA) was added in a ratio of 7:3:1:0.5. Liposomal formulation 7:3:1:0.5 showed a minimum liposome size of 230 nm, low PDI of 0.220, uniform liposomes in terms of shape and lamellarity. Both the phages MR-5 and MR-10 (1:1) were entrapped in this liposomal preparation which had an entrapment efficiency of 87% and liposomal size of 212 nm. The photomicrographs (**Figure 1**) obtained after TEM clearly demonstrates presence of phages in the cationic liposomes. LCP was made using 7:3:1:0.5 formulation and tested for stability (drop in phage titer) over a 9 week period. This liposomal preparation had a PDI below 0.3 with minimum size variation and was found to be stable at 4◦C in terms of fusion size, aggregation and number of entrapped phages during the entire storage period. Thus, this ratio was selected for further in vivo experiments.

<sup>1</sup>http://www.ihcworld.com/\_protocols/special\_stains/masson\_trichrome.htm

### Induction of Diabetes in Mice

The average random blood glucose level of normal BALB/c mice was found to be 139 ± 3.8 mg/dl whereas the fasting blood glucose level (after overnight fasting) was 96.2 ± 7.9 mg/dl. Fasting glucose levels were checked daily post-alloxan injection. Mice with fasting blood glucose levels in the range of 150–180 mg/dl, 4–5 days post-injection were termed as moderately diabetic. However, 15 days after alloxan administration, the fasting blood glucose level was found to be 194.6 ± 17.1 mg/dl, whereas the random blood glucose levels were ≥400 mg/dl. Such animals were considered severely diabetic and selected for infection studies.

### Excision Wound Infection Model

The excision wound model was developed by inflicting full thickness wound of 5 mm extending through panniculus carnosus with the help of punch biopsy on the dorsal surface of mouse. Since the aim of this study was to assess the efficacy of phages in reducing the mortality and clearing off infection; a high dose of 10<sup>8</sup> CFU/50 µl of S. aureus 43300 was chosen as the optimal infectious dose. At this bacterial dose, an acute infection of the inflicted wound was established with 70% of untreated mice exhibiting mortality within 24–48 h post-infection.

### Phage Protection Studies

### **Mortality vs. survival**

Animals in all groups were monitored daily for mortality. Untreated mice of Study 1 (group 1) and Study 2 (group 1) exhibited 70% mortality within 24–48 h post-inoculation. However, all treated animals showed no mortality during the experimental period and all infections resolved. This clearly indicated that a single injection of free phages, alone or as cocktail, or liposome-entrapped phages were able to protect the test animals from death.

### **Wound microbiology**

The wound bioburden (bacterial density) and phage titer was also assessed in all the mice groups of Study 1 and Study 2. In Study 1, untreated animals showed an increase in bacterial load that reached ∼9 log CFUs/ml by day 3 (peak day). Mice receiving FMP-MR-5, FMP-MR-10, i.e., groups 2 and 3 respectively, showed bacterial loads which instead were 7 log CFUs/ml on day 1 and day 3 followed by a significant decline (P < 0.05) to 3 logs on day 10 and sterile wounds by day 15 (**Table 1**). However, mice treated with cocktail, i.e., FCP showed higher reduction in bacterial burden as compared to infection control as well as monophage treated mice on all days. A significant decrease of 3 log (P < 0.05) was observed in cocktail treated group on days 3 and 5 as compared to infection control. Clarithromycin treated mice also showed significant reduction on all days with sterile wounds obtained by day 10.

Phage titer for Study 1 is shown in **Figure 2** was also determined on subsequent days post-treatment. Although an initial titer of 10<sup>9</sup> PFU/50 µl (MOI 10) was administered at the wound site, maximum titer of 10<sup>5</sup> PFU/ml was estimated on day 1, which shows a 4 log reduction from the initial phage titer. This phage count was consistent (approximately 10<sup>5</sup> PFU/ml) in all the groups on days 1 and 3, followed by gradual decline on subsequent days with minimum count of 10<sup>2</sup> PFU/ml present on day 10. However, phage titer was higher on all days in the wound of mice receiving FCP as compared to monophage treated animals. No phage particles were detected in the wound of mice receiving either single or cocktail of phages by day 15.

Results in **Table 2** for Study 2 show that the wounds of mice in the infection control (group 1) also had consistently higher bacterial burden (8–9 log CFU/ml) till day 5 followed by decline from day 7 onwards. In FCP treated mice, a significant decrease of 3 log CFU/ml occurred on day 3 and day 5 itself as compared to untreated mice. However, bacterial load (2.21 log CFU/ml) was detected even by day 10. In case of LCP treated mice maximum reduction in bacterial load equivalent to ∼4 log CFU/ml (P < 0.05) was seen on day 3 (peak day of infection). Negligible counts were obtained on day 7 and this correlated well with visible reduction in wound size by day 7 (**Figure 3**). The wound showed complete healing within 9 days unlike 20 days and less than 15 days required for untreated surviving diabetic mice and FCP treated group, respectively. Animals receiving clarithromycin showed higher reduction in bacterial burden on all days as compared to FCP treated mice

TABLE 1 | Bacterial load (in terms of Log CFU/ml) in skin wound of S. aureus 43300 (10<sup>8</sup> CFU/50 µl) infected diabetic BALB/c mice following treatment with FMP (MR-5 and MR-10), FCP and clarithromycin (10 mg/kg/ per i.p.).


Treatment was given after 30 min of wound inoculation. (−): zero bacterial count; NA: no mice available for determining the wound bio-burden. All values represent mean ± standard error of the mean calculated from three independent values.

43300 (10<sup>8</sup> CFU/50 µl) infectedmice (n = 14, each group) following treatment with FMP-MR-10, FMP-MR-5, and FCP on different days post-infection. Error bars represent the standard deviation (SD) from three independent values. <sup>∗</sup>P < 0.05 indicate statistical significant differences (Student's t-test) between FMP and FCP treated groups.

and reduction in bacterial load was comparable to that seen in LCP treated animals.

Phage titer in Study 2 in terms of log PFU/ml (**Figure 4**) was maximum on day 1 and thereafter a decrease was noticed on subsequent days in both FCP and LCP treated groups. Maximum titer of 7.45 and 7.12 log PFU/ml on day 1 and day 3 was seen in liposome phage entrapped cocktail (LCP) treated mice as compared to 5.80 and 5.68 log PFU/ml present on day 1 and day 3 in FCP treated group. A decline was seen thereafter due to clearance of its host bacteria as no phage particles were obtained on day 15 and after day 10 with FCP and LCP treated groups, respectively. This shows that entrapment of liposomes led to a significant increase in phage titer by 2 log (P < 0.05) at the wound site, which explains the faster wound healing process.

### Additional Parameters (Study 2)

### **Wound contraction**

Wound area of all the animal groups was measured with the help of vernier caliper (**Figure 5**). In untreated mice (group 1), an initial wound size of 5 mm persisted until day 3 and it was finally resolved by day 20. In FCP treated animals wound contraction was comparatively slower with wound of 3 mm size evident by day 5 and even by day 12 a wound size of 0.5 mm was observed. However, in case of both LCP treated as well as clarithromycin treated animals, significant decrease in wound size was seen by day 3 onwards as compared to both infection control and FCP treated mice. Minimal wound size of less than 1 mm was observed by day 7 with almost complete closure of wound by day 9 in clarithromycin as well as LCP treated mice. Thus wound closure rates were greater given when phages were encapsulated than when they were not.

### **Myeloperoxidase (MPO) activity**

Neutrophils are the first cells to reach the site of infection or inflammation. MPO is then produced by neutrophils into the extracellular space during degranulation. So as to assess the levels of inflammation or bacterial load MPO levels were measured, and results are presented in **Figure 6**. Highest MPO

TABLE 2 | Bacterial load (in terms of Log CFU/ml) in skin wound of S. aureus 43300 (10<sup>8</sup> CFU/50 µl) infected diabetic BALB/c mice following treatment with FCP, LCP, and clarithromycin (10 mg/kg/per i.p.).


Treatment was given after 30 min of wound inoculation. (−): zero bacterial count; NA: no mice available for determining the wound bio-burden. All values represent mean ± standard error of the mean calculated from three independent values.

levels were present in untreated animals on all days with peak MPO activity of 2.81 units/ml recorded on day 3, that decreased on subsequent days. In case of FCP treated mice, peak MPO activity of 2.41 units/ml was observed on day 1, followed by significant decline thereafter as compared to untreated mice with minimal value of less than 1 unit/ml measured on day 10. However, maximum reduction in MPO levels was seen in mice receiving LCP as it showed peak MPO levels of 1.77 units/ml on day 1 with minimum MPO activity of 0.211 units/ml recorded on day 7 which is significantly less (P < 0.05) than activity present in untreated and FCP treated mice Lower levels of MPO activity observed on subsequent days in LCP treated group correlated well with lesser bacterial load seen after treatment. Clarithromycin treated mice also showed significant reduction (P < 0.05) in MPO enzyme activity as compared to infection control group and FCP treated group. Thus, these results indicate that neutrophil migration was maximum in untreated animals, while neutrophil migration upon phage and antibiotic treatment decreased due to the decrease in bacterial infection at the site.

### **Masson's trichrome staining**

Wound tissue was taken on day 5 post-infection from all the groups. The tissue of untreated diabetic mice showed little collagen formation as seen in **Figure 7A**. The faint green area seen in the figure depicts early stages of collagen formation. Mostly, growing fibroblast was present in between the majority of blood vessels, capillaries and inflammatory cells, but much less mature collagen fibers were visible with fibroblastic condensation, showing delayed and early stage of wound healing. FCP treated mice (**Figure 7B**) showed growing fibroblasts present between blood vessels and inflammatory cells and thin collagen fibrils, which is an intermediate stage of collagen formation, were common. Mature collagen was also present. The amount of collagen formed was comparatively greater than the untreated mice and less than the cocktail and antibiotic treated groups. LCP treated animals showed maximum formation of mature collagen among all the groups (**Figure 7C**). As is clearly evident that collagen recovered from individual thin fibrils in the dermis into thick bundles or fibers, indicating that mostly mature collagen was present. This correlated well with the less bacterial bioburden on day 5. Together with decreased vascularity and reduced inflammatory cells, the results suggest late stage of wound healing because collagen formation is one of the essential markers of wound healing. Clarithromycin treated animal group also showed presence of mature thick collagen fibers but thin collagen fibrils were comparatively more than LCP treated mice (**Figure 7D**).

### **Histopathological examination**

For histopathological examination, wound tissue was taken on day 5 in order to study tissue damage and extent of wound healing in all the groups. **Figure 8A** depicts wound tissue of untreated animals (group 1). Acute inflammation associated with surface ulceration and breached epidermal layer was seen. A zone of acute inflammatory (AI) reaction with large numbers of infiltrating inflammatory cells and development of vascular granulation tissue (GT) showing large number of capillaries, blood vessels, and inflammatory cells is also visible. But mice receiving FCP (group 2) showed formation of epidermal buds extending toward dermis (**Figure 8B**), growing fibroblasts

in the granulation tissue and collagen formation, indicating healing along with presence of less inflammatory cells. Hair follicles were also visible indicating regeneration of damaged skin. Similarly, clarithromycin treated mice (group 4) showed progressively reduced inflammation with few neutrophils present in the wound tissue (**Figure 8E**). Presence of granulation tissue with fibroblasts, collagen formation, and epidermal buds indicated rapid and significant wound healing in clarithromycin treated group. However, maximum healing with minimum tissue infiltration and damage was clearly visible in the wound tissue of mice treated with LCP as compared to free phage cocktail treated groups (**Figures 8C,D**). In LCP treated group a thick eschar (Ec) over the surface consisting of necrosed cells in dense eosinophillic coagulated protein background was visible. Epidermal healing was accompanied with hyperplasia of epidermis with buds and streaks extending deep into the dermis. Fewer growing fibroblasts and mostly collagen was seen in the granulation tissue with decreased vascularity and reduced inflammatory cells. Presence of hair follicles was also evident. These results indicate significant wound healing by day 5. This effect might be due to the higher number of phages present at the wound site as compared to phage titre in case of FCP treated animals.

### DISCUSSION

Bacteriophage therapy represents an attractive treatment option for the eradication of S. aureus induced diabetic wound infections. Sunagar et al. (2010) reported significant protection in diabetic mice (90% survival rate) from lethal bacteremia following single injection of S. aureus specific phage (phage GRCS). Similarly, VinodKumar et al. (2011) demonstrated that wounds of diabetic rats which were infected with S. aureus were

sterilized within 8 days after receiving monophage treatment. Chhibber et al. (2013) focused on the synergistic use of phage and antibiotic in the effective resolution of hindpaw infection in diabetic mice. However, in all of these studies monophage therapy was employed. Monophage based therapies suffer from the drawback of easy emergence of phage resistant bacterial mutants as well as their inability to cover a broad range of possible host targets (Scott et al., 2007; Labrie et al., 2010; Chadha et al., 2016). Thus, use of phage cocktail by incorporating different phages covering the broad host range can provide an effective solution that can be readily delievered without any delay, making it potentialy more effective clinically than monophage therapy (Chan et al., 2013).

In the present study, two already characterized S. aureus specific lytic phages, i.e., MR-5 and MR-10 (Kaur et al., 2012; Chhibber et al., 2013) belonging to Myoviridae family, were used to study therapeutic efficacy in resolving diabetic wound infection. The main focus of this study was to overcome two major drawbacks of phage therapy. Firstly, therapeutic efficacy of phage cocktail therapy (FCP) was studied and compared with monophage therapy in treatment of MRSA mediated wound infection. The results indicated that when phages were used singly or in cocktail (FCP) there was 100% protection as compared to untreated controls. However, more rapid reduction in bioburden and faster wound healing with FCP treated animals.

Secondly, the problem of potentially low phage retention and persistence in vivo was addressed by the use of lipid based carriers, i.e., liposomes. On studying the phage titer postinfection, it was found that phage titer dropped to 10<sup>5</sup> PFU/ml from an initial titer of 10<sup>9</sup> PFU/50 µl within 24 h of phage administration, resulting in presence of low phage number around the wound site. This low titer could be due to phage inactivation by skin's extensive armamentarium of immunecompetent cells, that contributed toward delayed phage effect

(Merril et al., 1996; Keen, 2012; Abedon et al., 2014). The wound bio-burden initially increased as only 10<sup>5</sup> phages were present at the wound site on day 1. The lower number of phage particles probably was unable to tackle the high bacterial load of actively multiplying bacterial population. But as phage growth occurred at the expense of host bacteria, this prevented further increases in bacterial count. This delayed protective response needs to be addressed by preventing rapid phage inactivation and clearance from infection site. Although past workers have made attempts to overcome this problem, however, each approach has its own limitations.

## Liposome Entrapment

Our laboratory has for the first time reported successful entrapment of phages within suitable lipid based delivery system, i.e., liposomes that are biocompatible and help to maintain phage titer within in vivo system (Singla et al., 2015; Chadha et al., 2017). In another study, encapsulation of Salmonella specific phages led to increased stability, prolonged intestinal residence and enhanced therapeutic efficacy (Colom et al., 2015). Liposomes have been regarded as effective DDS due to their GRAS (generally regarded as safe) status, high biocompatibility and high diffusivity in skin as compared to bare drugs (Verma et al., 2003). For drug delivery application liposomes are usually unilamellar and range in diameter from 100 to 300 nm. Large liposomes are rapidly removed from the blood circulation (Barenholz, 1998). Therefore, it is very important to optimize the liposome formulation in terms of size so as to maximize drug loading and minimize leakage and rapid clearance from body. Finally, the cocktail of phages MR-5 and MR-10 (10<sup>9</sup> PFU/50 µl) (1:1) was entrapped within the liposome after optimizing the conditions required for their production and stability. Cocktail entrapped liposomal formulation (LCP) was of unilamellar nature, devoid of any observed aggregation, having a size of 212 nm and exhibited high entrapment efficiency of 87%. Stability studies in terms of size were conducted over a 9 week long period at different temperatures. The liposomal formulation was found to be most stable at 4◦C without any physical changes or reduction in number of entrapped phages during storage period. However, at 37◦C, an increase in particle size and aggregation was seen in the suspension. Leakage of entrapped drug from the liposomal formulation during storage before administration can lead to poor action of the administered drug (Nallamothu et al., 2006). The results are in consonance with the previous findings of Parmar et al. (2010) who showed that high temperature caused increase in size and aggregation

as well as leakage of drug from liposomal formulation of budesonide.

The therapeutic efficacy of LCP was evaluated in Study 2. The wound bio-burden correlated well with the visible healing of the wound. No mortality occurred in any of the treatment groups. Maximum reduction was obtained on day 3 in mice treated with liposome entrapped cocktail as compared to infection control group. The entire process of wound healing and re-epithelization was faster and quicker in mice receiving LCP as compared to both untreated diabetic mice as well as animals that received FCP. No phage was present after day 10 in the wound tissue. However, phage titer was higher on all subsequent days in mice treated with liposome entrapped phage cocktail. A significant enhancement of 2 log was observed in LCP mice, clearly indicating that drop in initial titer and phage viability could be prevented by phage entrapment within liposomes. Also, as noted, workers in the past have shown that use of DDS improves the pharmacokinetics and bio-distribution of the associated drug (Allen and Cullis, 2004; Zhang et al., 2013). In addition, efficient uptake of encapsulated bacteriophage by eukaryotic cells has also been reported (Nieth et al., 2015; Singla et al., 2016).

## Faster Wound Healing

Wound contraction also demonstrated comparatively rapid healing (measured in terms of wound size) in all the treated groups as compared to infection control mice. Less number of days required for healing and obtaining sterile wounds was due to increased phage titer at wound site in LCP treated mice. Liposomes provided a depot effect and prevented phages from getting cleared off rapidly from the site. This rapid wound healing by day 9 also correlated well with the maximum deposition of mature collagen fiber in wound tissue by day 5 in animals treated with LCP. Mature collagen is an essential marker of wound healing and represents later stages of wound healing (Braiman-Wilksman et al., 2007; Suvik and Effendy, 2012). Histopathological examination of wound tissue of animals also supported these findings, and confirmed that those mice that

received LCP showed minimal tissue damage and inflammation as compared to other groups. Epidermal healing and good amount of granulation tissue indicated significant wound healing by day 5. Besides this, maximum decrease in tissue MPO levels was also seen in group receiving LCP. This could be due to the broad spectrum effect of phage cocktail against MRSA, which efficiently arrested growth of invading bacteria at the site of injury, leading to decrease in neutrophil accumulation that correlated well with tissue healing.

From these observations it is concluded that although FCPs is a better option than monophage therapy, rapid inactivation of phages subsequently affects their ability in clearing the infectious agent. Liposomes are suitable delivery systems that help to increase the therapeutic efficacy of phages by improving their stability and delayed clearance by allowing timely release of phages at the infection site, leading to faster healing. The findings of this study thus provide new insights in the treatment of

### REFERENCES


diabetic wound infection not explored by past workers. The use of liposome entrapped phage preparation is an attractive option.

### AUTHOR CONTRIBUTIONS

SC: conceived and designed the experiments/contributed reagents/materials/analysis tools. JK and SK: performed the experiments. SC and SK: analyzed the data. SC, JK, and SK: wrote the paper.

## FUNDING

This work was supported by Department of Microbiology, Panjab University, Chandigarh, DST-PURSE and UGC-SAP, New Delhi.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Chhibber, Kaur and Kaur. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antimicrobial Potential of Epiphytic Bacteria Associated With Seaweeds of Little Andaman, India

Perumal Karthick\* and Raju Mohanraju

Department of Ocean Studies and Marine Biology, Pondicherry University, Port Blair, India

Seaweeds of the intertidal regions are a rich source of surface associated bacteria and are potential source of antimicrobial molecules. In the present study, 77 epiphytic isolates from eight different algae collected from Little Andaman were enumerated. On testing for their antimicrobial activities against certain pathogens twelve isolates showed positive and six of them showed significant antimicrobial inhibition zone against Shigella boydii type 1, Shigella flexneri type 2a, Shigella dysenteriae type 5, Enterotoxigenic Escherichia coli O115, Enteropathogenic E. coli serotype O114, Vibrio cholera; O1 Ogawa, Aeromonas hydrophila, Klebsiella pneumoniae, Staphylococcus aureus. Based on the activity these six isolates (G1C, G2C, G3C, UK, UVAD, and Tor1) were identified by 16S rRNA gene sequence and were found to belong to the phyla Firmicutes and Proteobacteria. Purified antimicrobial compounds obtained from these isolates were identified by GC-MS. Furan derivatives were identified from G2C Pseudomonas stutzeri KJ849834, UVAD Alcanivorax dieselolei KJ849833, UK Vibrio sp. KJ849837, Tor1 Exiguobacterium profundum KJ849838. While 2-Pyrrolidinone, Phenol, 2, 4-bis (1, 1-dimethylethyl) were from G3C Vibrio owensii KJ849836 and (1-Allylcyclopropyl) methanol from the extracts of G1C Bacillus sp. KJ849835. The results of the present study shows that these six potent isolates isolated from the seaweeds are found to be a source of antimicrobial compounds.

Keywords: Alcanivorax dieselolei, Little Andaman, Furan, Gracilaria corticata, seaweeds

## INTRODUCTION

Marine eukaryotes such as seaweeds are one of the primary producers which offers nutrient rich environment for microbial communities (Egan et al., 2008; Wahl, 2008). Biofilm forming bacteria isolated from the surface of seaweeds release certain compounds (Zheng et al., 2005) which serve as nutrient supplement for the algae (Croft et al., 2005), such compounds protect the host plant from the fouling communities (Rao et al., 2007). Surface associated marine organisms such as bacteria, fungi, diatoms, larval forms of marine invertebrate's have been reported to be associated with the thallus of seaweeds (Goecke et al., 2010; Burke et al., 2011; Murthy et al., 2016; Karthick and Mohan, 2017). Such host association particularly epiphytic bacteria are sources of certain natural compounds (Singh et al., 2011; Ali et al., 2012; Martin et al., 2014). Importance of microbial diversity of seaweeds, particularly bacterial genus are highly host specific with novel species, which have emerged from these algal environment (Goecke et al., 2013).

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Vinay Kumar, Savitribai Phule Pune University, India Osmar Nascimento Silva, Universidade Católica Dom Bosco, Brazil Pedro Ismael Da Silva Junior, Instituto Butantan, Brazil

> \*Correspondence: Perumal Karthick

karthickmicrobes@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 30 June 2017 Accepted: 16 March 2018 Published: 04 April 2018

### Citation:

Karthick P and Mohanraju R (2018) Antimicrobial Potential of Epiphytic Bacteria Associated With Seaweeds of Little Andaman, India. Front. Microbiol. 9:611. doi: 10.3389/fmicb.2018.00611

The secondary metabolites produced by these bacteria are highly recognized for their importance in the field of biomedical applications (Armstrong et al., 2001; Kelecom, 2002; Burgess et al., 2003). Antimicrobial activity of the epiphytic bacterial communities from seaweeds have been reported (Kanagasabhapathy et al., 2006, 2008; Vijayalakshmi et al., 2008; Ravisankar et al., 2013; Horta et al., 2014; Martin et al., 2014; Karthick et al., 2015a), similarly anti-diatom activity have also been observed by Kumar et al. (2011). In the Andaman Islands, luxuriant growth of all the three groups of seaweeds are available throughout the year. Some studies on taxonomy of seaweeds have been carried out in this region but information on the epiphytic interaction and its potentiality has not been undertaken. Based on the occurrence of seaweeds in Little Andaman and their bacterial association, the present study has been undertaken to describe the isolation of epiphytic bacterial, screening, optimization, evaluation and identification of potential isolates and their antimicrobial activity against different pathogens as test organisms.

## MATERIALS AND METHODS

### Isolation of Marine Bacteria

Eight different seaweeds representing all the three groups were handpicked from the intertidal region of Harminder Bay Bridge, Little Andaman, Andaman Islands, India. Among these, six species Gracilaria corticata, Acanthophora spicifera (red algae), Ulva lactuca (green algae), Sargassum swartzii, Turbinaria ornata, and Padina tetrastromatica (brown algae) are common and other two species Mastophora rosea (red algae) and Caulerpa microphysa (green algae) were found to be rare in occurrence in these islands. The collected samples were placed in sterile plastic bags and transported to the laboratory. These were washed thrice with autoclaved seawater to remove loosely bounded epiphytes, sand particles and other attached settlements on the surface of thallus. After rinsing, firmly attached epiphytic bacteria from thallus region were swabbed with sterile cotton buds and these were then swabbed on Zobell marine agar plate (Himedia). Plates were incubated for 5 days at 32◦C (Lemos et al., 1985). After incubation, colonies were picked and restreaked for the isolation of individual colonies and the purity of the isolates were checked under the microscope for single morphology. These pure cultures obtained were stored at −20◦C in marine broth supplemented with 20% glycerol.

### Antimicrobial Assay of Epiphytic Bacteria

The antagonistic activity of epiphytic bacteria obtained were studied on solid media by cross streaking and double-layer method described Lemos et al. (1985) and agar well diffusion method by Karthick et al. (2013b).

### Extraction of Antimicrobial Compounds

All the 77 isolates were cultured on 100ml marine broth, Luria broth and minimal media by modifying the methodology slightly by decreasing the incubation time and by increasing the temperature for obtaining better results. The culture broth was centrifuged at 10000 rpm for 30 s to remove the cells and cell free broth was extracted thrice with 100 ml of ethyl acetate. All the solvents were removed under reduced pressure at 40◦C (Zheng et al., 2005). Crude extracts obtained were stored at −20◦C until usage for the antimicrobial assay against targeted pathogens. Sterile media without culture being adjusted to pH 7 were used as control.

## Minimal Medium

All the potential isolates were cultured in inorganic salt medium referred to as minimal medium for the extraction of secondary metabolites (Jafarzade et al., 2013).

### Inoculum Preparation

Potential cultures were cultivated in 100 ml minimal medium supplemented with 3% NaCl, 1% glucose and 1% yeast extract as carbon and nitrogen sources in a 250 ml Erlenmeyer flask and incubated at 32◦C for 24 h in an incubator shaker. Five milliliter of this culture was used as bacterial (Starter) culture (Jafarzade et al., 2013).

## Effect of pH

1 ml of starter cultures were grown with minimal media supplemented with 3% NaCl, 1% glucose, and 1% yeast extract prepared and inoculated with minimal media supplemented with 0.75% of sodium chloride, 1% of glucose and yeast extract for the production of antimicrobial compounds with various pH levels (6–8) at 32◦C for 5 days. After incubation, supernatant was extracted three times with ethyl acetate (EtOAc). The sterile media without the culture adjusted to pH was used as control. The extracts were then tested for antimicrobial activity.

### Effect of Sodium Chloride Concentration

100 ml of minimal medium was dispensed into 250 ml Erlenmeyer flasks and sterilized. Yeast extract (1%) and glucose (1%) were filter sterilized and added as nitrogen and carbon sources just prior to inoculation. One milliliter of the starter culture was inoculated into the sterilized medium. Effect of salinity in the production of antimicrobial properties at various concentrations of sodium chloride ranging from 1 to 3% with constant pH of 7 at 32◦C for 5 days was experimented. After incubation cell free supernatant was extracted three times with ethyl acetate (EtOAc). Sterile media without the inoculum adjusted with various concentration of sodium chloride was used as control (Jafarzade et al., 2013). The extracts were then tested for antimicrobial activity.

## Effect of Different Concentrations of Glucose and Yeast Extract

Effect of different concentration (1–3%) of glucose and yeast extract for the production of antimicrobial compound by the epiphytic bacterial isolates was studied using 1 ml of the starter culture inoculated into the minimal medium. Other parameters such as pH 7 and 0.75% sodium chloride were maintained at optimum level during the primary screening at 32◦C for 5 days. After incubation supernatant was extracted three times with ethyl acetate (EtOAc). The sterile medium containing glucose, yeast extract and sodium chloride was used as control. The extracts were then tested for antimicrobial activity.

## Test Microorganisms

fmicb-09-00611 March 30, 2018 Time: 19:12 # 3

Eighteen bacterial pathogens namely Escherichia coli MTCC 443, Klebsiella pneumoniae MTCC 109, Salmonella typhi MTCC 733, Staphylococcus aureus MTCC 96, Shigella flexneri MTCC 1457, Shigella flexneri type2a 503004, Shigella boydii type 1 NK2379, Shigella sonnei NK4010, Shigella dysenteriae type 5 NK2440, Enterotoxic E. coli serotype 0115, Enteropathogenic E. coli serotype 0114, Shiga toxin producing E. coli serotype O157:H7 VT3, Vibrio fluvialis IDH 02036, Vibrio parahaemolyticus serovar O3: K6 K5030, Vibrio cholera O139, Vibrio cholera 01, Ogawa Eltor, Aeromonas hydrophila IDH1585, Salmonella enterica serovar typhi C6953 and three fungal strains Aspergillus niger, Aspergillus flavus and Rhizopus sp. were used for studying the antibacterial and antifungal assay.

## Determine Minimum Inhibitory Concentration of TLC Purified Metabolites

Minimum inhibitory concentration (MIC) of TLC purified metabolites of potential six isolates (G1C, G2C, G3C, UK, UVAD, and Tor1) was tested against Klebsiella pneumoniae and Staphylococcus aureus and it was determined by well diffusion assay. 50 mg of TLC purified extracts was dissolved in 1 ml DMSO. 50 and 100 µl/ml concentration of 50 mg/ml concentration of purified extracts was transferred into the well prepared (9 mm) with well cutter. Gentamicin was used as a positive control and Dimethyl sulfoxide (DMSO) was used as a negative control. Growth inhibition zone formed after the incubation was examined with measuring the diameter (mm) and results were recorded. All the assay was performed in triplicates.

### Partial Purification and GC-MS Analysis

Concentrated fractions were fractioned by Thin Layer Chromatography (TLC) using Silica gel plates with different solvents in a ratio of 2:2:1 ethyl acetate, chloroform and methanol. Bands was scraped from the plates and screened for antimicrobial assay. Active fraction was collected and analyzed by Gas Chromatography and Mass spectrometry GC-MS QP 2010 Shimadzu Corp (Japan). One µl of purified fractioned extract was loaded into the DB-5 Column with Helium as a carrier gas at a flow rate of 1 ml/min. Split Injection mode of the ratio of 1:20 was adopted. Temperature programming was from 75◦C for 2 min further increased to 175◦C with 15◦C/min and then increased up to 280◦C at the rate of 5◦C/min. Sample run time was maintained upto 10 min. The peaks representing mass to charge ratio characteristic of the antimicrobial fractions were compared with those in the mass spectrum of NIST library identifying the corresponding organic antimicrobial compounds.

## Phenotypic Characterization

Phenotypic characterization of all the seventy seven bacterial isolates were identified following as described in the Bergey's manual of systemic Bacteriology (Brenner et al., 2005).

## Molecular Identification by 16S rDNA Sequencing

Genomic DNA was prepared from the bacterial isolates by following the method of Mohandass et al. (2012). PCR amplification of 16S rRNA gene was conducted in a final volume of 25 µl with the bacterial consensus universal forward and reverse 16S rDNA primers 27F and 1492R (Lane, 1991). The reaction mixture contained 1x PCR buffer (Sigma, United States), 2.5 mM MgCl2, 200 µM DNTP's, 1U of Taq DNA polymerase, 25 picomol of each forward and reverse oligonucleotide primers and approximately 20 ng of genomic DNA. The amplification profile consisted of an initial denaturation at 94◦C for 3 min, followed by 35 cycles at 94◦C for 1 min, 55◦C for 1 min and 72◦C for 1 min. This was followed by a final extension step of 72◦C for 5 min. The samples were held at 4◦C until further analysis. The PCR products were sequenced by an automated Sequencer (Applied Biosystems, Foster City, CA, United States) at the National Institute of Oceanography, Goa, India. The sequences were submitted to Gen Bank for which accession numbers were assigned.

## BLAST Search and Phylogenetic Analysis

The PINTAIL program (Ashelford et al., 2005) was used to check chimera formations. The partial 16S rRNA gene sequences of the potential isolates were compared with those available in the public databases. Identification upto the species level was determined by a 16S rDNA sequence similarity of more than 99% with that of the prototype sequence in GenBank. Sequence alignment and comparison were performed using the multiple sequence alignment program Clustal X 1.81 (Thompson et al., 1997). Sequences were edited manually to remove the gaps. Neighbor-joining method was employed to construct the Phylogenetic tree using MEGA4 software (Tamura et al., 2007) and the maximum likelihood method was adopted for calculating the evolutionary distance (Tamura et al., 2004).

## Bacterial Identification Based on Fatty Acid Methyl Ester (FAME)

Young pure cultures of SG107, 108, 114, 115, 120, and Tor6 were grown on Trypticase Soy Broth Agar (TSBA) for 24 or 48 h at 28◦C. The Gas chromatographic analysis of whole cell fatty acid methyl ester (FAME) was performed for further identification and grouping of isolates. FAME extraction were performed using the standard procedures of extraction, purification, and methylation (Sasser, 1990). Fatty acid profiles generated were compared against an inbuilt Sherlock TSBA Library Version 6.0B [S/N 160284] (MIDI Inc., Newark, DE, United States). A similarity index of more than 0.500 was used for clustering of isolates at species level. Cellular fatty acid composition analysis

was done at Regional center of Kochi, National Institute of Oceanography.

### RESULTS

## Identification of Epiphytic Bacterial Isolates

All the seventy seven cultivable epiphytic bacterial isolates obtained from the thallus of eight different seaweeds were plated on Marine agar. These isolates were purified and based on by phenotypic characterization were assigned to belong to the phylum Firmicutes and Proteobacteria (**Table 1**). Among these isolates, six of them (G1C, G2C, G3C, UK, UVAD, and Tor1) showed wide range of activities against pathogens with an range of 10–30 mm zone of inhibition and were identified by partial 16S rRNA gene sequences (**Table 2**). The isolate G1C showed 99% similarity as Bacillus sp., G2C as Pseudomonas stutzeri, G3C and UK were identified as Vibrio owensii and Vibrio sp., respectively. The Isolate UVAD was identified as Alcanivorax dieselolei with 99% similarity and Torl strain was identified with 99.7% similarity as Exiguobacterium profundum (**Figure 1**).

Six other isolates showed moderate to less activity activities against the pathogens with a range of 5–10 mm zone of inhibition and were identified by FAME analysis. Among them three isolates SG107 as Bacillus sp., SG108 Paenibacillus lentimorbus and SG115 as Bacillus sphaericus belonged to phylum Firmicutes. The other three isolates Pantoea agglomerans, SG 120 was Vibrio aestuarianus and TR was identified as Klebsiella pneumoniae

TABLE 1 | List of bacterial isolates obtained from different seaweeds.


ozaena and were assigned to phylum Proteobacteria, SG114 (**Table 3**).

## Antimicrobial Activity of Epiphytic Bacteria

Antimicrobial activity for all the 77 bacterial isolates were tested by adopting three different methods (Agar overlay, cross streaking and agar well diffusion technique) against 21 pathogens. Among these larger zones of inhibition were observed in agar well diffusion assay and this assay was chosen for further antimicrobial activity test. All the isolates were cultured in three different media (marine broth, luria broth and minimal medium) among the medium used minimal media showed broad range of antimicrobial activity (**Table 4**). Based on the preliminary activity only six potential isolates (G1C, G2C, G3C, UK, UVAD, and Tor1) were optimized in minimal medium for the production of antimicrobial compounds.

## Zone of Inhibitory Activity With Optimized Bacterial Isolates

Isolate G1C showed strong inhibitory activity against Shigella boydii (31 mm), Enterotoxigenic E. coli (28 mm), Enteropathogenic E. coli and Aeromonas hydrophila (23 mm) and this higher inhibition zones were obtained with minimal media supplemented with 1 and 2% Sodium chloride, 1% of glucose and yeast extract with pH in the range of 7–8. Isolate G2C was more effective against Shigatoxin E. coli (26 mm), Aeromonas hydrophila and Salmonella typhi (24 mm) observed from minimal medium containing only 1% of sodium chloride, glucose and yeast extract with pH in the range of 6–7. Isolate G3C was effective against Salmonella enterica serovar typhimurium (26 mm), Vibrio cholerae Eltor and Shigella dysenteriae (24 mm) and antimicrobial activity was observed with minimal media with 1 and 2% of sodium chloride, glucose and 1% of yeast extract with pH in the range of 6–7. Isolate TOR1 exhibited broad range of antibacterial activity against Klebsiella pneumoniae, Shigella dysenteriae (31 mm), Shigella sonnei (25 mm). UVAD isolate showed maximum zone of inhibitory activity against Staphylococcus aureus (30 mm), Salmonella enterica serovar typhimurium (29 mm) and Shigella dysenteriae (27 mm) and isolate UK displayed maximum zone of inhibition against Aeromonas hydrophila (29 mm), Shigella flexneri 2A and Shigella flexneri (24 mm), respectively. All these six bacterial isolates showed maximum inhibitory activity against most of the tested pathogens (**Table 4**).

TABLE 2 | 16S rRNA gene sequence identity of six potential bacterial isolates obtained from different seaweeds.


FIGURE 1 | Neighbor-joining phylogenetic tree based on 16S rRNA gene sequences showing the relationship of six potential strains with its closest neighbors. Bootstrap values (50%) are shown at branch points in value and S. hanedai X82132 were used as outgroup.



TABLE 4 | Antimicrobial activity of six potential epiphytic bacterial isolates.


(+) >10 mm, (++) 10–20 mm, (+++), 20–30 mm, (−) No inhibition zone observed.

## Identification of Compounds From Purified Extract

Cell free supernatant of the six potential isolates (G1C, G2C, G3C, UK, UVAD, and Tor1) that showed antimicrobial activity were purified by TLC. Results of minimum inhibitory concentration of TLC purified metabolites of all the six potential isolates at 50 mg/mL concentration diluted in 50 and 100 µl/ml concentration showed the positive results against K. pneumoniae and S. aureus. 50 µl/ml concentration of potential extracts exhibited inhibitory activity against the tested pathogens. Maximum zone of inhibition 10 mm was measured against K. pneumoniae by UK, UVAD, Tor1, G1C, and G3C extract at

100 µl concentration. 9 mm zone of inhibition was measured against S. aureus by 100 µl concentration of G2C, G3C, UK, and Tor1. Based on the inhibitory growth of bacteria at 50 µl concentration of purified metabolites, it can be concluded that minimum inhibitory concentration was observed at 2.5 mg/mL. Based on the activity purified compounds obtained from these six isolates were characterized and identified. Furan derivatives were found to be present in four of the isolates namely G2C Pseudomonas stutzeri, Tor1 Exiguobacterium profundum. UVAD Alcanivorax dieselolei and UK Vibrio sp. While 2-Pyrrolidinone, Phenol, 2, 4-bis (1, 1-dimethylethyl) were identified from the isolate G3C Vibrio owensii and (1-Allylcyclopropyl) methanol from G1C Bacillus sp.

### DISCUSSION

Seaweed biomass were found in large quantities in both intertidal and subtidal regions of all the regions of Andaman Island and in Little Andaman's (Karthick et al., 2013a,b). Good hemolytic activity in certain seaweeds of these Island has been reported recently (Punnam Chander et al., 2014). Besides these studies Karthick et al. (2015b) had reported on antimicrobial activity of certain seaweeds against pathogenic bacterial and fungal stains. Several authors suggested that macro algal associated bacteria were found to be an efficient producer of antimicrobial compounds (Burgess et al., 1999; Lee et al., 2006; Kanagasabhapathy et al., 2008; Karthick et al., 2015a; Ismail et al., 2016). On the other hand, certain brown algae also produced biologically active compounds which inhibited the settlement of bacterial colonies on the thallus (Nagayama et al., 2002).

In the present study it was observed that higher number of epiphytic bacteria were isolated from brown and red algae, certainly the proportion of higher isolates were from brown rather than green and red algae. On surface colonization nonpigmented bacterial isolates were found dominant in most of the seaweeds used in this study. Epiphytic bacteria from marine macro algae have been well studied in reference to their ecological importance with host organisms (Croft et al., 2005) with a dominance of Gram-negative bacteria. Similarly in the present study 46 Gram-negative bacterial isolates were isolated in comparison to 31 being Gram-positive. Bacteria belonging to genus the Bacillus were dominant with 20 isolates followed by other genus such as Vibrio, Aeromonas, and Pseudomonas.

Ravisankar et al. (2013) observed that the surface of the brown algae Hypnea valentiae and Padina tetrastromatica contained more number of non-pigmented bacterial colonies which are similar to our studies wherein 10 isolates were obtained from Padina tetrastromatica. Similar observations were observed in Tunisian waters, where 17 isolates were obtained from the thallus of Ulva intestinalis (Ali et al., 2010) and 10 isolates were reported from Ulva lactuca in Fiji waters, of which majority of the isolates were efficient antimicrobial producers (Kumar et al., 2011).

In the present study twelve isolates (15.7%) of the total 77 isolated exhibited antimicrobial activity and six isolates showed broad spectrum of activity against both bacterial and fungal pathogens. Similarly (Jayanth et al., 2002) isolated 14.52% of associated bacteria from the red algae Gracilaria with antagonistic properties against certain human pathogens. On the other hand 11% of associated bacteria isolated from seaweeds were reported to have antagonistic nature against Bacillus subtilis, E. coli, S. aureus, Agrobacterium tumefaciens, and Saccharomyces cerevisiae (Zheng et al., 2005).

The 16S rRNA sequences of bacterial isolates obtained from the surface of green algae Ulva australis and Delisea pulchra, belonged to the representative's classes of Alpha and Gammaproteobacteria and interestingly Actinobacteria, Firmicutes, and Bacteroidetes were observed as antimicrobial producers (Penesyan et al., 2009). Ali et al. (2012) on 16S rRNA sequence of the isolates obtained from the surface of coralline red algae Jania rubens found them belong to the group Proteobacteria. Similar observation made by Singh et al. (2015) also highlighted that bacterial isolates belonged to the order Bacillales, Pseudomonadales, Alteromonadales, and Vibrionales were dominant in green algae Ulva lactuca, U. fasciata, and red algae Gracilaria corticata and G. dura. In this study also it's evident that all the 77 bacterial isolates were closely related to the phylum Proteobacteria and Firmicutes. These findings substantiate that these groups are more specific to the macro algal surface. Similarly species belonging to the genera Bacillus and Vibrio were found to be strong antimicrobial producers colonizing more on the surface of seaweeds.

Genus Bacillus predominantly colonizes on the surface of marine niche and several studies have been reported Bacillus from different marine sources particularly associated with brown algae (Thakur and Anil, 2000) and from the thallus surface of different red algae (Kanagasabhapathy et al., 2008). Apart from their association with seaweeds Bacillus were previously isolated from sediments and seaweeds with antimicrobial properties (Prieto et al., 2012). So far, more than 800 metabolites have been reported with various biological activities from the Bacillus genera. Recently, cell free supernatant extracted from Bacillus associated with a nematode were found to be very effective against multidrug resistant Staphylococcus aureus (Susilowati et al., 2015). As observed in the present study one potential isolate G1C identified as Bacillus sp. showed remarkable activity against most of the tested pathogens, in particular against toxin producing pathogens like Shigella boydii, Enterotoxigenic E. coli, Shigatoxin E. coli Enteropathogenic E. coli, and Aeromonas hydrophila etc. Similarly SG107 Bacillus sp. and SG115 Bacillus sphaericus obtained from the brown algae Sargassum swartzii also showed moderate to less activity against few pathogens.

Vibrios being truly marine and they are widespread in various marine niches and are known to produce secondary metabolites for their survival. Earlier genus Vibrio sp., Pseudomonas sp., and Bacillus pumilus were reported to be a probiotic bacteria used in aquaculture (Hill et al., 2009). Kanagasabhapathy et al. (2008) reported that Vibrio strain isolated from red algae showed certain biological activities. In the present study, two potential Vibrio isolates GC3 Vibrio owensii and UK Vibrio sp. were obtained from the surface of red algae Gracilaria corticata and Mastophora rosea, respectively, and these isolates exhibited wider range of antimicrobial activity against most of the tested pathogens like Salmonella typhi, Shigella dysenteriae, Vibrio cholerae, and

Staphylococcus aureus. Similarly, Pawar et al. (2015) extracted antibacterial compounds from marine Vibrio sp. which were found to be active against numerous pathogens.

In our study of 14 bacterial isolates were obtained green algae Ulva lactuca among these one isolate UVAD Alcanivorax dieselolei was found to possess higher range of antimicrobial activity. Previously this species Alcanivorax dieselolei has been reported to be isolated from the deep sea sediment involved in degrading alcanes (Liu and Shao, 2005), and petroleum products (Brito et al., 2006). Ali et al. (2010) reported that two epiphytic bacteria obtained from green alga U. intestinalis showed potent antimicrobial activity. These studies suggest that green algae Ulva species attracts novel bacterial colonization on their surface with potential microbial communities and these isolates produce various compounds to protect the host from the predators and other micro and macro fouling colonization.

Pseudomonas stutzeri has been reported to have wide range of biological activity by the production of secondary metabolites. Previously Pseudomonas stutzeri isolated from fish gut exhibited antimicrobial activity (Uzair et al., 2008), hydrocarbon degradation (Vazquez et al., 2009), and reported as uncommon opportunistic pathogen (Park et al., 2013), controlling biofilm formation (Wu et al., 2016). In this study Pseudomonas stutzeri isolated from the red algae Gracilaria corticata produced antimicrobial compounds which showing potent activity against numerous toxin producing pathogens S. aureus, Shigella boydii, S. flexneri 2A, S. dysenteriae, K. pneumoniae, Et. E. coli, St. E. coli, V. cholerae Eltor, A. hydrophila.

Earlier Exiguobacterium sp. showed antimicrobial properties (Shatila et al., 2013) and this bacterium also known to produce antifouling compound and thus protected the host organisms from fouling communities (Jain et al., 2013). In the present study Tor1 Exiguobacterium profundum isolate obtained from Turbinaria ornata, showed antibacterial activity against clinical pathogens S. aureus, Shigella boydii, S. flexneri, S. flexneri 2A, S. dysenteriae, K. pneumoniae, Et. E. coli, and St. E. coli. The same genus was identified in earlier studies from different seaweeds occurring in different geographical locations showing various biological activities. In the present study 6 potential isolates obtained from seaweeds were found to be good antimicrobial producers. The same genus was identified in earlier studies from

TABLE 5 | Biological activities of associated bacteria isolated from seaweeds.


∗ Indicates the species identified in this study.

different seaweeds occurring in different geographical locations showing various biological activities (**Table 5**).

In earlier studies Furan derivatives were reported to have antimicrobial properties (Kirilmis et al., 2009; Joshi et al., 2010; Ramasamy and Balasubramanian, 2012), cytotoxic agent (Wang et al., 2008) and were observed to have a wide range of biological activities like antiproliferative, antiviral, antifungal, immunosuppressive, anti-platelet, anti-oxidative, insecticidal, anti-inflammatory, anti-feedant, and cancer preventative activity (Venkateshwarlu et al., 2013). In our present study we have identified Furan compounds from four potential isolates (G2C, UVAD, Tor1, and UK). Apart from antimicrobial properties, these compounds are being used for other pharmacological properties (Bober et al., 2012). In this study G3C Vibrio owensii produced antimicrobial compounds such as 2-Pyrrolidinone, Phenol, 2, 4-bisdimetyl ethyl)-ester and Pyrrolo [1,2-a] pyrazine-1,4-dione. Earlier, these compounds were reported to have antimicrobial properties (Sutariya et al., 2012; Khatiwora et al., 2013; Padmavati et al., 2014; Dhanya et al., 2016). Marine Vibrio sp. is highly capable of producing Phenol, pyrrolo [1,2-a]pyrazine-1,4-dione, Pyrrolidinone derivative compounds containing pharmacological properties (Pawar et al., 2015). Based on earlier findings and as observed in the present study Vibrios are efficient producer of Phenol and Pyrrolidinone derivatives. In conclusion based on the findings of the present study, the compounds produced from six potential isolates (G1C, G2C,

## REFERENCES


G3C, UK, UVAD, and Tor1) having effective antimicrobial properties, could be further studied for other activities. These isolates could prove to be potential candidates for the production of novel antimicrobial compounds in order to control the pathogens.

### AUTHOR CONTRIBUTIONS

PK designed the work, performed all the experiments, analyzed and wrote the manuscript. RM contributed the designation of research work and evaluated the manuscript.

### ACKNOWLEDGMENTS

The authors are greatful to Dr. C. Mohandass, Chief Scientist, National Institute of Oceanography (NIO), Goa for sequencing studies, Dr. Anas Abdulaziz, Scientist, Marine Microbial Reference Facility (NIO) RC, Kochi, Dr. R. Babu Rajendran, Professor, Department of Environmental Biotechnology, Bharathidasan University for GC-MS facilities. The authors also thank personally to their colleagues Dr. K. N. Murthy, Dr. C. H. Ramesh, Dr. Sumantha Narayana for their support during field collection and Pondicherry University authorities for providing the facilities.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Karthick and Mohanraju. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Prediction of Cell-Penetrating Potential of Modified Peptides Containing Natural and Chemically Modified Residues

Vinod Kumar 1,2†, Piyush Agrawal 1,2†, Rajesh Kumar 1,2†, Sherry Bhalla<sup>1</sup> , Salman Sadullah Usmani 1,2, Grish C. Varshney <sup>3</sup> and Gajendra P. S. Raghava1,2 \*

<sup>1</sup> Center for Computational Biology, Indraprastha Institute of Information Technology, Okhla, India, <sup>2</sup> Bioinformatics Centre, CSIR-Institute of Microbial Technology, Sector-39A, Chandigarh, India, <sup>3</sup> Cell Biology and Immunology, CSIR-Institute of Microbial Technology, Sector-39A, Chandigarh, India

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Tikam Chand Dakal, Hospital Maisonneuve-Rosemont and University of Montreal, Canada William Farias Porto, Universidade Católica Dom Bosco, Brazil

### \*Correspondence:

Gajendra P. S. Raghava raghava@iiitd.ac.in

†These authors have contributed equally to this work.

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 08 January 2018 Accepted: 28 March 2018 Published: 12 April 2018

### Citation:

Kumar V, Agrawal P, Kumar R, Bhalla S, Usmani SS, Varshney GC and Raghava GPS (2018) Prediction of Cell-Penetrating Potential of Modified Peptides Containing Natural and Chemically Modified Residues. Front. Microbiol. 9:725. doi: 10.3389/fmicb.2018.00725 Designing drug delivery vehicles using cell-penetrating peptides is a hot area of research in the field of medicine. In the past, number of in silico methods have been developed for predicting cell-penetrating property of peptides containing natural residues. In this study, first time attempt has been made to predict cell-penetrating property of peptides containing natural and modified residues. The dataset used to develop prediction models, include structure and sequence of 732 chemically modified cell-penetrating peptides and an equal number of non-cell penetrating peptides. We analyzed the structure of both class of peptides and observed that positive charge groups, atoms, and residues are preferred in cell-penetrating peptides. In this study, models were developed to predict cell-penetrating peptides from its tertiary structure using a wide range of descriptors (2D, 3D descriptors, and fingerprints). Random Forest model developed by using PaDEL descriptors (combination of 2D, 3D, and fingerprints) achieved maximum accuracy of 95.10%, MCC of 0.90 and AUROC of 0.99 on the main dataset. The performance of model was also evaluated on validation/independent dataset which achieved AUROC of 0.98. In order to assist the scientific community, we have developed a web server "CellPPDMod" for predicting the cell-penetrating property of modified peptides (http:// webs.iiitd.edu.in/raghava/cellppdmod/).

Keywords: modified cell-penetrating peptides, machine learning, Random Forest, SVM, in silico method, chemical descriptors, antimicrobial peptide

## INTRODUCTION

Since the existence of human race, therapeutic molecules have been used to cure human illness and to extend lives (Tosato et al., 2007). In past, thousands of molecules have been studied to combat deadly diseases. The ideal molecule must attain the desired therapeutic effect without causing side effects. A large number of promising therapeutic molecules disparage before reaching to its target (Gupta and Jhawat, 2017). In order to overcome this, several delivery vehicles have been discovered in last three decades, such as nanoparticle (Wang et al., 2018) and lipid carrier conjugate (Xu et al., 2017). Cell-penetrating peptide (CPP) is one of the most emergent and widely accepted drug delivery vehicle, having ability to internalize even into eukaryotic cells in non-disruptive way. These are short peptides of 3 to approximately 40 amino acids, mostly cationic followed by amphipathic in nature (Agrawal et al., 2016). CPPs can transport various biologically active molecules inside microbes as well as mammalian cells (Gao et al., 2014; Kurrikoff et al., 2016). CPPs such as TP10 and pVEC had been shown to significantly inhibit growth of few microbes as Candida albicans, Staphylococcus aureus as well as Mycobacterium smegmatis (Nekhotiaeva et al., 2004). CPPs and cationic antibacterial peptides have similar physicochemical properties, so many CPPs have shown antimicrobial activity (Splith and Neundorf, 2011; Bahnsen et al., 2013; Rodriguez Plaza et al., 2014). The poor membrane permeability of drug molecule always remains a concern in drug designing. In the era of drug resistance, where pathogen membrane provides a significant barrier, intracellular delivery of antibiotics/drugs by the virtue of CPP, proved to be a vital step in combating drug resistance to some extent (Sparr et al., 2013). CPP based conjugates (Ganguly et al., 2008; Jain et al., 2015) and combination therapy has been explored against several resistant pathogens (Randhawa et al., 2016). They have been proved effective against intracellular pathogens too (Gomarasca et al., 2017).

A universal mechanism of CPP internalization is always proved to be an exploring question, as the involved pathways are not fully clarified yet. The difficulty arises due to differing size, physicochemical properties, as well as concentration of diverse CPP and CPP-conjugates (Guidotti et al., 2017). Several mechanisms have been shown by various CPPs to translocate in to the cell, as micelle formation (Derossi et al., 1996), pore formation (Matsuzaki et al., 1996), membrane thinning (Pouny et al., 1992), endocytosis (Ferreira and Boucrot, 2018) and micropinocytosis (Jones, 2007). Majority of CPP internalization occurs via endocytosis, but several evidences suggest that at a threshold concentration direct penetration does occur (Palm-Apergi et al., 2012). CPPs can be used for intracellular delivery of small molecule-based drug (Lindgren et al., 2006), oligonucleotide (Margus et al., 2012), peptide and protein (Morris et al., 2001) and trans-epithelial delivery of peptides (Tan et al., 2014).

Despite, numerous properties and potential applications of CPPs, still there use in real life is limited. The primary limitation associated with CPP is endosomal compartment entrapment which reduces the bioavailability of the drug several times. In literature, it has been shown that bioavailability of CPPs can be increased several times by introducing a chemical modification in a CPP (Postlethwaite et al., 1996; Kim et al., 2006; Lundberg et al., 2007; Koppelhus et al., 2008; Aubry et al., 2009). N-terminal stearylation of Arg8 peptide improved the delivery of siRNA (Futaki et al., 2001), C-terminal cysteamidation of MPG peptide improved the delivery of siRNA (Simeoni et al., 2003), cysteine residue modification improved the stability of Tat peptide and thus enhances the plasmid delivery (Lo and Wang, 2008), Poly-L-ornithine modification in PepFect 14 peptide increases transfection efficiency of oligonucleotide in HeLa pLuc 705 (Ezzat et al., 2011). Thus, it is important to understand chemical modification of residues in a peptide and its impact on cell-penetrating property of peptides.

In the last few years, several computational methods have been developed for the prediction of CPPs. These methods have been developed on various features like amino acid composition (Sanders et al., 2011), dipeptide composition (Tang et al., 2016), binary profile, physiochemical properties and motifs (Gautam et al., 2013). They have also applied Z-scale based method (Sandberg et al., 1998), feature selection techniques (Tang et al., 2016)**,** classifiers like Random Forest (RF) (Wei et al., 2017), Support Vector Machine (SVM) (Sanders et al., 2011). Beside this, few more methods have been developed in recent years for predicting CPPs with high accuracy (Chen et al., 2015; Tang et al., 2016; Wei et al., 2017). Best of authors knowledge, all methods developed so far for predicting CPPs are suitable for peptides containing natural residues only, but no method has been developed for predicting cell penetration property of peptides with non-natural and modified residues. In this study, a systematic attempt has been made to develop a machine learning method for predicting cell penetration ability of peptides containing non-natural and modified residues. Machine learning technique derive features/rules from the experimentally validated modified CPPs and Non-CPPs are used to predict cell penetration ability of a modified peptide. We hope this method will be useful for researchers working in the field of drug delivery.

### MATERIALS AND METHODS

### Creation of Dataset for CPPs and Non-CPPs

Cell-penetrating peptides were extracted from CPPsite2.0 database (Agrawal et al., 2016), which provides comprehensive information on wide-range of CPPs. It consists of 1,850 experimentally validated natural and modified CPPs. We remove CPPs that does not contain any modified residue; we also remove peptides whose tertiary structure is not available in the database. Finally, we got 732 chemically modified CPPs whose structure is available in CPPsite 2.0. We assign this set of 732 CPPs as positive set or set of CPPs. To develop any method, we also need equal number of negative examples. In this study, we extracted non-CPPs from SATPdb (Singh et al., 2016) database which maintains information of 19,192 peptides having several properties. We extracted structures of 732 peptides, which may exhibit any characteristic other than cell penetrating property. This set of peptides were assigned as negative set or set of non-CPPs. Finally, we built the dataset that contains 732 CPPs and 732 non-CPPs whose sequence and tertiary structure is available in CPPsites 2.0 or SATPdb.

## Datasets for Internal and External Validation

The dataset was divided into two datasets namely training (main) and validation dataset (Bhalla et al., 2017). The training dataset consists of 80% of peptides, 582 CPPs, and 582 non-CPPs. The validation dataset consists of remaining 20% of peptides, 150 CPPs, and 150 non-CPPs. We used training dataset for developing models and for internal validation. In internal validation, models were trained and tested using commonly used five-fold cross-validation technique (Nagpal et al., 2017). Performance of best model achieved on training dataset, was evaluated on validation dataset. The evaluation of the performance of model on validation or independent dataset is called external validation.

## Model Development

### Computation of Features From Peptide Structures **Composition Based Features**

Atom composition is computed from CPPs and non-CPPs by converting peptide structures in SMILES format using openbabel (O'Boyle et al., 2011). These SMILES were further used to compute atom composition of following atoms C, H, O, N, S, Cl, Br, and F. The atomic composition provided the fixed length of 8 vectors.

$$\text{Fraction of atom} \left( a \right) = \frac{\text{Total number of atom} \left( a \right)}{\text{Total number of all possible diameters}} \times 100 \quad \text{(1)}$$

Where atom (a) is one out of 8 atoms.

### **Diatom Composition**

We computed diatom composition of amino acids just like the atomic composition for CPPs and non-CPPs. The diatomic composition provides the composition of the pair of atoms in each residue (e.g., C-C, C-O, etc.) of the peptide, and used to convert the variable length of modified peptides to fixed length feature vectors. The diatomic composition provided the fixed length of 64 (8 × 8) vectors.

$$Fraction\ of\ Diatom(a) = \frac{Total\ number\ of\ Diatom\ (a)}{Total\ number\ of\ all\ possible\ diatoms} \times 100\ \{2\}$$

Where diatom (a) is one out of 64 diatoms.

### **Chemical Descriptors**

A biological property of any chemical molecule is determined by its chemical descriptors, which have been used in the past to develop QSAR based molecules (Kumar et al., 2015). PaDEL software, a freely available software was used for the calculation of chemical descriptors (Yap, 2011). We calculated 15,537 different types of descriptors, including 2D, 3D, and 10 different types of fingerprints. As all descriptors don't correlate with biological activity, we have done feature selection using "CsfSubsetEval" function present in WEKA software (Smith and Frank, 2016) to remove unnecessary descriptors hence reduced noise from dataset.

## Computation of Features From Amino Acid Sequence of Peptide

### Amino Acid Composition

We substitute the symbol of the modified residue with its original natural amino acid, for calculating amino acid composition for the positive and negative dataset. This left us with the sequence having 20 natural amino acids which generated the vector of 20.

$$AAC\left(a\right) = \frac{Ra}{N}x100\tag{3}$$

Here, AAC (a) is the percent composition of amino acid (a); R<sup>a</sup> is the numbers of residues of type a, and N represents the total number of peptide's residues.

### Dipeptide Composition

We also calculated dipeptide composition of the peptides since it provides global information of the peptide. The dipeptide composition was calculated using the formula 4, and it generated the vector of 400 (20 × 20).

$$Fraction\text{ of Dipeptile}\,(a) = \frac{Total\text{ number of Dipeptile}\,(a)}{Total\text{ number of all possible hipeptile}} \times 100\tag{4}$$

Where dipeptide (a) is one out of 400 dipeptides.

### Terminus Composition-Based Model

We also calculated N and C terminus amino acid composition as well as dipeptide composition for developing prediction models. The composition of 5, 10, and 15 residues from N-terminus as well as C-terminus was taken into account. Also, we joined the terminal residues like N5C5, N10C10, and N15C15 and for developing models.

### Residue Preference

In order to observe the residue preference at a particular position in the peptide, web-logos were prepared for first 15 N and 15 C-terminals along with their modifications using online WebLogo software (Crooks et al., 2004). These logos provide the position specific frequency of amino acids in a peptide. Each logo consists of stacks of symbols, one stack for each position in the sequence. The overall height of the stack indicates the sequence conservation at that position while the height of symbols within the stack indicates the relative frequency of each amino acid at that position.

### Statistical Analysis

To check whether is there any significance difference between modified CPPs and non-CPPs, we performed Welch t-test on the selected features of 2D, 3D and Fingerprints descriptors using in house R-script. Adjusted p-values were calculated using Boneferroni adjustment.

### Performance Measure

Different parameters were used to check the performance of various models developed in this study. These parameters are divided into two groups.

### Threshold Dependent Parameters

This category includes Sensitivity (Sen), Specificity (Spc), Accuracy (Acc), and Matthews's correlation coefficient (MCC), where Sensitivity is true positive rate, Specificity is true negative rate, accuracy is ability to differentiate true positive and true negative and MCC is a correlation coefficient between observed and predicted. These can be calculated using the following equations.

$$\text{Sensitivity} = \frac{TP}{\underset{m\text{th}}{\text{PS}}} \times 100\tag{5}$$

$$\text{Specificity} = \frac{TN}{\text{NS}} \times 100\tag{6}$$

$$\text{ann} \quad \text{mm}$$

$$\text{Accuracy} = \frac{TP + TN}{\text{PS} + \text{NS}} \times 100\tag{7}$$

$$\text{MCC} = \frac{1 - \left(\frac{FN}{PS} \times \frac{FP}{NS}\right)}{\sqrt{\left(1 + \frac{FP - FN}{PS}\right) \times \left(1 + \frac{FN - FP}{NS}\right)}}\tag{8}$$

Where TP represents correctly predicted positive, TN represents the correctly predicted negative examples, PS represents total sequences in positive set, NS represents total sequences in negative set, FP represents actual negative examples which have been wrongly predicted as positive and FN represents wrongly predicted positive examples. This is a well-established method of measuring performance and has been used earlier in many studies (Porto et al., 2017; Agrawal et al., 2018).

### Threshold Independent Parameters

In this study, we also used threshold independent measure to evaluate the performance of models. In case of threshold independent measures, Receiver Operating Characteristics (ROC) curve is drawn between false positive and false negative rates. In order to measure performance, Area Under Curve ROC curve is computed called AUROC.

### RESULTS

## Analysis

We compute percent average composition of atoms in CPPs and non-CPPs to understand the preference of certain types of atoms present in the CPPs and non-CPPs. Overall, the profile is more or less same in both CPPs and non-CPPs.CPPs are slightly rich in H and N atoms whereas non-CPPs are slightly rich in C, O, and S (**Figure S1**). We analyzed the amino acid composition of both positive (CPPs) and negative (Non-CPPs) dataset. It has been observed that certain type of residues like R, K, and Q are more prominent in CPPs; in contrast residues are like C, L, V, P, and G are not preferred in CPPs (**Figure 1**). In the same manner, we also calculated the average amino acid composition of the first 15 N and 15 C- terminal amino acid residues (**Figure S2**). At the N terminal R, Q, I and M are more prominent in CPP as compared to Non-CPP (**Figure S2A**). Similarly at C terminal, R, K and Q are more prominent (**Figure S2B**).

In addition to compositional preference, we also computed preference of different types of residues in CPPs. It was revealed that some specific type of residues was preferred in the positive dataset contain CPPs as compared to the negative dataset contain non-CPPs. Residues like Rand K are highly preferred at various positions CPPs particularly at N terminal (**Figure 2**)**.** Similarly, K and R are mostly preferred at C terminal also (**Figure 3**)**.**

### Machine Learning Based Prediction Model

We used various machine-learning approaches like SVM, Random Forest, Naive Bayes, J48 and SMO for developing the prediction model. These models utilize different features or descriptors to discriminate or classify CPPs and non-CPPs. The results are explained in details in the following sections.

### Model Based on Peptide Structure

Tertiary structure of a peptide can present all type of modifications. Thus structure of peptide is used to predict cell penetration ability of modified peptide. In this study,

we got structure of peptides from databases CPPsite 2.0 and SATPdb. The models were developed using various features of peptide structures. First, we developed model using atomic composition of peptides. In order to obtain atomic composition of peptides from its structure, we convert structure from sdf format to SMILES. The atomic composition of peptides was calculated from SMILES of peptide. Prediction models were developed using different classifiers like SVM, RF, Naive Bayes, SMO and J48 using atomic composition as an input feature. Random Forest based classification model provided the highest accuracy of 84.02%, MCC of 0.68 and AUROC of 0.91 on the training dataset. On validation dataset, we achieved maximum accuracy of 78.33%, MCC of 0.57 and AUROC of 0.88. Performance of different classifiers given in **Table 1.** We also developed model using diatom composition of peptides and obtained the highest accuracy of 88.40% with MCC of 0.77. On validation dataset, we achieved maximum accuracy of 91.00% with MCC 0.83. Here SVM based model performed best among all the classifiers used for prediction (**Table 2**)**.**

We developed models individually for 2D descriptors, 3D descriptors, and Fingerprints as well as the single model by combining 2D, 3D descriptors, and Fingerprints. The descriptors were computed using PaDEL software from tertiary structure of peptides (sdf format). The models were developed on the features, selected after performing feature selection, by attribute evaluator named, "CfsSubsetEval" with search method of "BestFirst" at default parameters in the forward direction (amount of backtracking, N = 5 and lookup size D = 1). In case of 2D descriptors, total 144 descriptors were calculated initially and were reduced to 17 after feature selection. List of the selected features is provided in **Table S1**. We applied different machine learning techniques on these selected features and observed that Random Forest based model achieved the maximum accuracy

of 92.34%, MCC of 0.85 and AUROC of 0.97 for the main dataset and 91.67% accuracy, 0.83 MCC and 0.97 AUROC for the validation dataset (**Table 3**).

In case of 3D descriptors, total 47 features were calculated and was reduced to 6 after applying feature selection (**Table S2**). On these features, Random Forest model performed better than other models and achieved maximum accuracy of 76.55%, MCC of 0.53 and AUROC value of 0.85 on the main dataset and 73.49% accuracy, 0.47 MCC and 0.83 AUROC on validation dataset (**Table 4**)**.** The different types of fingerprints generated 14,532 features, which were reduced to 27 after feature selection (**Table S3**)**.** Performance of different classifiers were evaluated on these features (**Table 5**) and once again Random Forest showed the best performance with maximum accuracy of 92.25%, MCC of 0.85 and AUROC of 0.98 on main dataset and accuracy of 92.33%, MCC of 0.85 and AUROC of 0.98 on validation dataset.

Finally, we calculated all the 2D, 3D descriptors and fingerprints at the same time, which generated 15,204 features. Feature selection reduced it down to 48 important features on which different machine learning classifiers were evaluated. Here we observe the maximum accuracy of 95.10%, MCC of 0.90 and AUROC of 0.99 on main dataset and 92.33% accuracy, 0.85 MCC and 0.98 AUROC on validation dataset by Random Forest model (**Table 6**). **Figure 4** shows the AUROC curve as well as AUROC values of different models.

### **Significance of features**

We obtained significant difference between the positive and negative features based on adjusted p-values. P-values were found to be less than 0.05 for most of the features. Therefore, we can say that these features can be used to discriminate modified CPPs and non-CPPs. Mean value of positive and negative features along with their p-value for 2D, 3D, and fingerprint descriptors is provided in **Tables S1**–**S3**.


TABLE 2 | Performance of different machine learning methods on diatom composition.


TABLE 3 | Performance of different machine learning methods on 2D descriptors.


TABLE 4 | Performance of different machine learning methods on 3D descriptors.


TABLE 5 | Performance of different machine learning methods on fingerprints.



TABLE 6 | Performance of different machine learning methods on 2D, 3D and fingerprints collectively.

### Model Based on Peptide Sequence

It is nearly impossible to present a modified peptide by amino acid sequence. Thus, prediction of modified peptide from there sequence is not possible. Same time generating tertiary structure of a peptide is a tedious job for a biologist. We made an attempt to develop prediction model for cell penetration peptides of modified peptides from their amino acid sequence only by ignoring modifications in peptide. First, we developed simple composition-based models using various machine learning techniques. The SVM based model showed the best performance among all the classifiers used in the study. The accuracy of 91.67%, MCC of 0.83 and AUROC of 0.96 was achieved for the main dataset. On validation dataset, we obtained accuracy of 89.67%, MCC of 0.79 and AUROC of 0.96 (**Table S4**)**.** We also developed SVM based model on first 5, 10, and 15 N and C-terminus residues. Results are given in **Table S5**.

Secondly, we developed models using dipeptide composition, SVM classifier showed the highest accuracy of 91.84%, MCC of 0.84 and AUROC of 0.96 for the main dataset. For independent dataset, the accuracy of 92.33%, MCC of 0.85 and AUROC of 0.97 was achieved (**Table S6**). Results of SVM based models on terminus residues for dipeptide composition is provided in **Table S7**. It is important for users to understand that sequence based model is not alternate to structure based models or alternate to past sequence based models developed for natural peptides. This sequence based is just approximate cell penetration potential of a modified peptide from its amino acid sequence.

### Implementation of Webserver

To assist the scientific community, the best models are provided freely at http://webs.iiitd.edu.in/raghava/cellppdmod/. The "PREDICTION" module, consider tertiary structure (PDB format) of the modified peptide as an input and does the prediction. If a user has no structural information, he/she can generate PDB structure of their peptide up to 25 residues in length using server "PEPstrMOD" (Singh et al., 2015) (http:// webs.iiitd.edu.in/raghava/pepstrmod/) developed by our group specifically for predicting the structure of the modified peptide. In case of natural peptide user can also use following servers PEP-FOLD (Thevenet et al., 2012) (http://bioserv.rpbs.univparis-diderot.fr/services/PEP-FOLD/) and QUARK (Xu and Zhang, 2012) (https://zhanglab.ccmb.med.umich.edu/QUARK/) for predicting structure of peptides. Multiple modification options are provided there, and the user can choose the desired modification. After generating the structure, user can do the prediction on "PREDICTION" module, whether the given modified PDB structure is CPP or non-CPP. Beside the main model, we have also implemented model based on peptide sequence (Subsidiary model). We have also provided a "DOWNLOAD" module from where the user can download the dataset used in this study.

## DISCUSSION

CPPs has shown a promising impact in the field of therapeutics or for targeting a specific disease (Bechara and Sagan, 2013). However, the major limitations associated with some of these CPPs is their entrapment of CPP-cargo in endosomal compartments followed by endocytosis and therefore their bioavailability and half-life is severely reduced (Mäe et al., 2009). To overcome this limitation, people have tried to modify the CPP chemically. For example, to increase the delivery of nucleic acid more efficiently, people have introduced chemical modifications like N terminal stearylation (Futaki et al., 2001; Khalil et al., 2004), C-terminal cysteamidation (Simeoni et al., 2003; Morris et al., 2007), residue modifications (Lundberg et al., 2007).Tat is one of the first CPP, discovered from protein of HIV and various studies showed that it enhances the uptake of various drug and protein (Brooks et al., 2005). But DNA delivery by Tat is limited, because of the instability of Tat-DNA complex (Lo and Wang, 2008). Lo and Wang (2008) showed that Cysteine makes the Tat-DNA complex more stable. Incorporation of two cysteine residues results into interpeptide disulphide bond, form by air oxidation once bind to DNA. This enhance the stability of Tat-DNA complex, as well as protect DNA in extracellular environment. Therefore, gene transfection efficiency is more in modified Tat than simple Tat.

Computational algorithms have been proved a wide success in designing therapeutic peptides (Dhanda et al., 2017), therefore a large number of sequence-based model to design CPP has been developed in past. But all of these models have one limitation in common that they can only handle peptides with natural residues. Due to the huge therapeutic importance of modified CPP, prediction and designing of modified CPPs is the need of hour. So, we have developed a computational method, which is based on structural features, can handle the natural as well as modified peptides both. Beside this we have also incorporated a subsidiary model based on the sequence of peptides which consider only natural residues, to handle large number of peptides simultaneously. Here, sequence-based model is not alternate to the methods developed in past to predict natural CPPs.

We have developed various models using machine learning techniques such as SVM, Random Forest, J48, naïve bayes, SMO; individually for atom composition, 2D descriptors, 3D descriptors, and Fingerprints as well as the single model by combining 2D, 3D descriptors, and Fingerprints. We obtain best performance by Random Forest for both combined (2D, 3D, and Fingerprint descriptors) as well as fingerprint with accuracy 92.33% and AUROC 0.98 on validation dataset. As fingerprint alone will be computationally more feasible as compared to the combined method, so we have implemented this model on webserver.

We believe this work will prove a great assist to the researchers aim to design cell penetrating peptide, as well as incorporate different modification and to check their effect on cell penetration ability. In future, we can improve this method, if better art of structure prediction will be developed, as right now PEPstrMOD could tackle only 7–25 amino acid length and other best model I-TASSER only deals with natural residues. So, in conclusion this field must grow simultaneously with the betterment of art-ofstructure prediction.

### AUTHOR CONTRIBUTIONS

VK and PA generated the dataset. VK, PA, RK, and SB performed the experiments. VK, PA, and RK performed data analysis and prepared the tables and figures. VK, PA, RK, SB, and SU developed the web interface. VK, RK, PA, SU, and GR write the manuscript. GR and GV conceived the idea and coordinated the project.

### ACKNOWLEDGMENTS

Authors are thankful to funding agencies J. C. Bose National Fellowship (DST), Council of Scientific and Industrial Research (CSIR), Department of Science and Technology (DST-INSPIRE), Indian Council of Medical Research (ICMR), University Grant Commission (UGC) and Department of Biotechnology (DBT) for fellowships and financial support.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00725/full#supplementary-material

Figure S1 | Percentage atomic composition of modified CPPs and non-CPPs.

Figure S2 | Percentage amino acid composition of CPPs and non-CPPs (A) 15 N-terminal residues and (B) 15 C-terminal residues.

Table S1 | List of 2D features with their positive mean value, negative mean value and p-value.

Table S2 | List of 3D features with their positive mean value, negative mean value and p-value.

Table S3 | List of fingerprints with their positive mean value, negative mean value and p-value.

Table S4 | Performance of different machine learning methods on amino acid composition.

Table S5 | Performance of SVM method on amino acid composition features of terminus residues.

Table S6 | Performance of different machine learning methods on dipeptide composition.

Table S7 | Performance of SVM method on dipeptide composition features of terminus residues.

### REFERENCES


Bahnsen, J. S., Franzyk, H., Sandberg-Schaal, A., and Nielsen, H. M. (2013). Antimicrobial and cell-penetrating properties of penetratin analogs: effect of sequence and secondary structure. Biochim. Biophys. Acta 1828, 223–232. doi: 10.1016/j.bbamem.2012.10.010


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Kumar, Agrawal, Kumar, Bhalla, Usmani, Varshney and Raghava. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Antibacterial Efficacy of Polysaccharide Capped Silver Nanoparticles Is Not Compromised by AcrAB-TolC Efflux Pump

Mitali Mishra<sup>1</sup> , Satish Kumar<sup>2</sup> , Rakesh K. Majhi<sup>1</sup> , Luna Goswami<sup>2</sup> , Chandan Goswami<sup>1</sup> and Harapriya Mohapatra<sup>1</sup> \*

<sup>1</sup> School of Biological Sciences, National Institute of Science Education and Research, Homi Bhabha National Institute, Bhubaneswar, India, <sup>2</sup> School of Biotechnology, Kalinga Institute of Industrial Technology, Bhubaneswar, India

### Edited by:

Rebecca Thombre, Savitribai Phule Pune University, India

### Reviewed by:

Amit Kumar Mandal, Raiganj University, India Rajashree Bhalchandra Patwardhan, Savitribai Phule Pune University, India

### \*Correspondence:

Harapriya Mohapatra hm@niser.ac.in; hmsbsniser@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 26 November 2017 Accepted: 11 April 2018 Published: 04 May 2018

### Citation:

Mishra M, Kumar S, Majhi RK, Goswami L, Goswami C and Mohapatra H (2018) Antibacterial Efficacy of Polysaccharide Capped Silver Nanoparticles Is Not Compromised by AcrAB-TolC Efflux Pump. Front. Microbiol. 9:823. doi: 10.3389/fmicb.2018.00823 Antibacterial therapy is of paramount importance in treatment of several acute and chronic infectious diseases caused by pathogens. Over the years extensive use and misuse of antimicrobial agents has led to emergence of multidrug resistant (MDR) and extensive drug resistant (XDR) pathogens. This drastic escalation in resistant phenotype has limited the efficacy of available therapeutic options. Thus, the need of the hour is to look for alternative therapeutic approaches to mitigate healthcare concerns caused due to MDR bacterial infections. Nanoparticles have gathered much attention as potential candidates for antibacterial therapy. Equipped with advantages of, wide spectrum bactericidal activity at very low dosage, inhibitor of biofilm formation and ease of permeability, nanoparticles have been considered as leading therapeutic candidates to curtail infections resulting from MDR bacteria. However, substrate non-specificity of efflux pumps, particularly those belonging to resistance nodulation division super family, have been reported to reduce efficacy of many potent antibacterial therapeutic drugs. Previously, we had reported antibacterial activity of polysaccharide-capped silver nanoparticles (AgNPs) toward MDR bacteria. We showed that AgNPs inhibits biofilm formation and alters expression of cytoskeletal proteins FtsZ and FtsA, with minimal cytotoxicity toward mammalian cells. In the present study, we report no reduction in antibacterial efficacy of silver nanoparticles in presence of AcrAB-TolC efflux pump proteins. Antibacterial tests were performed according to CLSI macrobroth dilution method, which revealed that both silver nanoparticles exhibited bactericidal activity at very low concentrations. Further, immunoblotting results indicated that both the nanoparticles modulate the transporter AcrB protein expression. However, expression of the membrane fusion protein AcrA did show a significant increase after exposure to AgNPs. Our results indicate that both silver nanoparticles are effective in eliminating MDR Enterobacter cloacae isolates and their action was not inhibited by AcrAB-TolC efflux protein expression. As such, the above nanoparticles have strong potential to be used as effective and alternate therapeutic candidates to combat MDR gram-negative Enterobacterial pathogens.

Keywords: nanoparticles, efflux pump, AcrAB-TolC, Enterobacter cloacae, antibiotic resistance

## INTRODUCTION

fmicb-09-00823 May 2, 2018 Time: 14:48 # 2

The drastic escalation in the proportion of multiple antibiotic resistant bacteria is recognized as a serious health care concern globally (WHO, 2014). The evolution of drug resistant microbes has challenged the success of antimicrobial agents for combating infectious diseases. The group of opportunistic pathogens, belonging to genera Enterococcus, Staphylococcus, Klebsiella, Acinetobacter, Pseudomonas, and Enterobacter (acronymed as ESKAPE) are exhibiting resistance to virtually all available antimicrobial agents (Boucher et al., 2009). Indeed, the acronym 'ESKAPE' is perfect choice to reflect these 'bad bugs.' To complicate the scenario, many of the above-mentioned bacteria are increasingly becoming responsible for spread of multidrug resistant (MDR) community acquired infections (Pendleton et al., 2013).

The Enterobacter spp., particularly E. aerogenes and E. cloacae are often associated with nosocomial infections like urinary tract infection (specifically catheter related), abdominal cavity/intestinal infections, wound infections, pneumonia, and septicemia (Sanders and Sanders, 1997; Fraser and Arnett, 2010). These two clinically predominant species display one or more resistance mechanisms toward antimicrobials like β-lactams, cephalosporins, aminoglycosides. Ironically, E. aerogenes and E. cloacae are also exhibiting trends of increased resistance toward the last resort antimicrobial agents – carbapenems and colistin (Davin-Regli and Pagès, 2015). Well-equipped with armory of antibiotic resistance strategies such as, outer membrane permeability barrier, efflux pumps and antibiotic degrading enzymes, Enterobacter spp. form an interesting infectious model to explore (Grimont and Grimont, 2006; Mezzatesta et al., 2012; Davin-Regli and Pagès, 2015).

The rise in antimicrobial resistance coupled with diminishing available options has necessitated searching for alternative therapeutic options (Martins and McCusker, 2016). One such lucrative direction was use of combinatorial therapy for treatment of infections caused by gram-negative pathogens, such as Pseudomonas aeruginosa. Typically, the combinatorial therapies for gram-negative pathogens included a β-lactam with aminoglycoside or fluoroquinolone (Tamma et al., 2012). The strategy was to extract their synergistic potential and broader spectrum of activity. But, in many clinical trials, it developed adverse effects leading to nephrotoxicity and even greater multiple antibiotic resistance (Kerantzas and Jacobs, 2017). The limitations faced by combinatorial therapies led to further improvisation and antimicrobial peptides were the more sought after. These host-defense peptides were investigated due to their antimicrobial properties against both gram-positive and gram-negative pathogens and immunomodulatory roles. However, low metabolic stability and discrepancies between in vivo and in vitro antimicrobial effectiveness limited the scope of antimicrobial peptides (Mahlapuu et al., 2016). Other interesting alternatives were more conceptual driven, such as bacteriophage therapy (Yang et al., 2014) and/or repurposing the existing drugs for better antibacterial and anti-virulence features (Rangel-Vega et al., 2015). Although both of them appeared as promising candidates, but development of such therapy at large scale remains challenging and yet to be explored.

All of the above-developed strategies faced severe obstacles in getting translated for sustainable, practical and large-scale usage, as demanded by the present situation. Much of the aforementioned limitations were circumvented by development of nanoparticle as drugs (Salata, 2004; Singh and Lillard, 2009). Over the years, nanoparticle-based approaches have gathered much reputation due to their fast and effective antibacterial properties for various pathogens (Gelperina et al., 2005; Tran and Webster, 2011). Nanoparticles endowed with wide bactericidal spectrum, ability to inhibit biofilm, increased permeability and efficacy at low dose, form a powerful weapon to combat MDR bacteria (Kim et al., 2007; Pelgrift and Friedman, 2013; Chen et al., 2014; Sanyasi et al., 2016). Review of the literature reveals that presence of efflux pumps [particularly those belonging to resistance nodulation division (RND) super family, e.g., tripartite efflux pump AcrAB-TolC] in gram-negative bacteria compromises the antibacterial efficacy of the drugs by extruding them out of the cytoplasm (Zgurskaya, 2009). Efflux of drugs reduces their intracellular concentration, leading to increased resistance in pathogens toward the drug. Further, it has been reported that substrates to be effluxed (antibacterial drugs) induce overexpression of the efflux proteins (Rosenberg et al., 2003).

Previously we had reported antibacterial activity of polysaccharide-capped silver nanoparticles (AgNPs) toward several MDR bacteria viz. E. coli, Klebsiella spp., Enterobacter spp., and S. aureus (Sanyasi et al., 2016). We had demonstrated AgNPs to inhibit biofilm formation and alter expression of cytoskeletal proteins FtsZ and FtsA, with minimal cytotoxicity toward mammalian cells. In the present study, we have further investigated antibacterial effects of the aforementioned AgNPs and that of a newly developed silver-metal-carbohydrate nanoparticle (Ag-MCNP) particularly on MDR E. cloacae isolates. Further, with the objective to consolidate our findings, we also investigated the effect of these silver nanoparticles on expression of AcrAB-TolC efflux pump proteins in MDR E. cloacae isolates.

## MATERIALS AND METHODS

### Bacterial Strains and Media

A wild type E. cloacae (EspIMS6) urinary tract infection isolate from a tertiary care hospital at Bhubaneswar, India, and E. cloacae subsp. cloacae ATCC-13047 type strain were used in the present study. The bacterial isolates were identified by biochemical tests followed by 16S rRNA gene sequencing. As prescribed for Enterobacteriaceae (CLSI, 2016), E. coli strain ATCC 25922 was used as control during antibiotic susceptibility testing. Unless otherwise mentioned, bacterial cultures were grown on nutrient agar/broth (Hi-Media, India) and incubated at 37◦C in a controlled environment shaker (New Brunswick, NJ, United States). Muller Hinton broth (MHB)/agar (MHA) medium (Hi-Media, India) was used for antibiotic susceptibility test. Following sub-culture, the strains were preserved at −80◦C in 20% glycerol for further use. All experiments were performed in triplicate and repeated at least three times.

### Resistance Profiling

fmicb-09-00823 May 2, 2018 Time: 14:48 # 3

Antibiotic disks used in resistance profiling were obtained from Himedia, India. Each strain was tested for susceptibility to different antibiotics by disk diffusion method (Bauer et al., 1966). The diameter of the inhibition zones was interpreted following CLSI standard as proposed for Enterobacteriaceae (CLSI, 2016).

### Characterization of Silver Nanoparticles Used

### Physical Characterization of the Nanoparticles Used

Particle size and zeta potential (ζ) of the polysaccharide capped silver nanoparticles were measured by Zetasizer Nano ZS instrument (Malvern Instruments, United Kingdom) at a constant temperature of 25 ± 1 ◦C. The samples (0.1 mg/ml) were suspended in Milli-Q water and sonicated for 1 min. The mean hydrodynamic diameter and zeta potential for each sample was measured in triplicate and the results were measured as mean size ± SD. The surface plasmon resonance (SPR) was recorded at different time points (10–150 min) for polysaccharide-capped silver nanoparticles (AgNP) against the only carboxyl methyl tamarind (CMT) polysaccharide solution as blank.

### Determining Hemocompatibility of Silver Nanoparticle

Hemocompatibility of polysaccharide capped silver nanoparticle was determined in terms of percent hemolysis using sheep blood (Mangalaraj and Devi, 2017). The diluted suspension of extracted RBCs (0.2 ml) was mixed with varied concentrations (1, 3, 6, 9, and 12 µg/ml) of AgNPs in PBS (0.8 ml). Diluted suspension of RBCs mixed with 0.8 ml PBS and 0.8 ml double distilled water were used as negative and positive control, respectively. The mixture was gently vortexed and incubated at room temperature for 3 h. After centrifugation (1600 rpm, 5 min) of the incubated mixture, absorbance of the supernatant was recorded at 541 nm by UV-Vis spectrophotometer (Agilent Cary 100 UV-Vis, Germany). Finally, hemocompatibility was evaluated in terms of percent hemolysis using the formula: (A<sup>S</sup> − AN)/(A<sup>P</sup> − AN) × 100; where "AS" is the sample absorbance, "AN" is the absorbance of negative control and "AP" is the absorbance of positive control (Jana et al., 2016).

### Determining Effect of CMT-Capped Ag-MCNP on Mammalian Cells

The cytotoxic effect of polysaccharide capped nanoparticles against Mouse macrophage (RAW 264.7) cells were evaluated by MTT [3-(4,5-Dimethylthiazol-2-Yl)-2,5-Diphenyltetrazolium Bromide] assay. Approximately, 1 × 10<sup>4</sup> cells were seeded in 96-well plates (Tarsons, India) and were incubated for 24 h at 37◦C in a 5% CO<sup>2</sup> incubator with different concentration of Ag-MCNP's (1, 2, 3, 4, and 6 µg/ml, respectively). Raw cells seeded without nanoparticles were used as control. To determine the cell viability, MTT dye (100 µl from 0.1 mg/ml stock) was added to each well and incubated for 4 h at 37◦C, 5% CO<sup>2</sup> in dark. In this assay, metabolically active cells reduce the MTT salt into water insoluble purple MTT formazan crystal by mitochondrial dehydrogenase. The formazan crystals formed as a result of cellular reduction of MTT were dissolved in buffer solution (4 g NP-40 detergent in 50 ml 0.02 M HCl and 50 ml isopropanol) and incubated for 1 h at 37◦C and absorbance was measured at 570 nm in an ELISA reader (BioTek, Germany). The percentage of cell viability at different doses of Ag-MCNPs was obtained by the following formula: % cell viability = [OD sample – OD control] – 100/OD control.

### Determination of Minimum Inhibitory Concentration (MIC) Breakpoints for Nanoparticles

Minimum inhibitory concentration (MIC) values were determined for both the nanoparticles viz. silver nanoparticle (AgNP) and silver-metal-carbohydrate nanoparticles (Ag-MCNP's) by macrobroth double dilution method based on the guidelines of Clinical Laboratory Standard Institute (CLSI) for Enterobacteriaceae (CLSI, 2016). Ten microliters of mid log bacterial cultures (OD600 nm: 0.6–0.8) were inoculated into MHB with different concentration (µg/ml) of nanoparticles. Inoculated tubes were incubated at 37◦C, 220 rpm in bacteriological incubator shaker (New Brunswick, NJ, United States). MIC break point values (in µg/ml) were noted after overnight incubation at the above-mentioned conditions. Further, we checked survival of the bacteria by dilution plating onto MHA plates to enumerate viable colonies.

### Determining Effect of Silver Nanoparticles on AcrAB-TolC Efflux Pump Protein Expression

The bacterial cells of (OD600 nm: 0.6–0.8), were treated with sub-MIC concentration of nanoparticles, i.e., AgNPs (6 µg/ml) and Ag-MCNPs (0.75 µg/ml), and incubated for specified time points (e.g., 0, 30, 60, 120, 180 min). Both the treated and control bacterial cells were pelleted down, washed once with 1X PBS (pH 7.4), dissolved in 1X PBS and kept on ice. The protein samples were preserved by adding proteinase-K, lysed by treating with 5X Lamelli buffer and heat denatured at 95◦C for 5 min. Extracted protein samples were then loaded onto 12% polyacrylamide gels, transferred to polyvinylidene fluoride (PVDF) membrane (Millipore, United States) for 1 h at 17 V, blocked by 5% skimmed milk in Tris buffered Saline with 0.05% Tween-20 (1X TBST). Blots were incubated with custom synthesized anti-rabbit polyclonal anti-AcrA, anti-AcrB, and anti-TolC primary antibodies (GenScript, United States) at 1:1000 dilutions overnight at 4◦C. After incubation, blots were washed thrice with 1X TBST. Further blots were incubated with goat anti-rabbit immunoglobulin secondary antibodies (1:10,000) conjugated to horseradish peroxidase. Specific bands were visualized by Chemiluminescence method using SuperSignalTM West Femto maximum sensitivity substrate kit (Thermo Scientific, United States). The images were acquired

using ChemiDoc and analyzed using Quantity-one software (Bio-Rad, United States).

## RESULTS

## Resistance Profile of Enterobacter Isolates

The wild type E. cloacae clinical isolate (EspIMS6) exhibited extreme drug resistant (XDR) phenotype with resistance to β-lactams, cephalosporins belonging to second and thirdgeneration, imipenem, fluoroquinolones (**Table 1**). The type strain, E. cloacae ATCC 13047 displayed MDR phenotype. Both of these isolates harbor multidrug efflux system AcrAB-TolC belonging to RND superfamily (data not shown).

## Characterization of Silver Nanoparticles Used

### Physical Characterization of the Nanoparticles Used

The size of nanoparticles has direct relevance with the stability, surface charge, bio distribution, cellular uptake, and drug release of nanomedicine (Lamprecht et al., 2001). The average size measured from photon correlation spectroscopy was found to be 140 d.nm in terms of number percent distribution (**Figures 1A,B**). The poly-dispersity index (PDI) of 0.208 indicated the monodispersed pattern of nanoparticles (Chandni et al., 2013). The surface charge of nanoparticles is another vital physical characteristic affecting its stability. The surface characteristics of polysaccharide capped AgNPs was measured upon immediate synthesis and 6 months postsynthesis. The mean zeta potential of AgNP was measured to be −34.5 ± 2.2 mV (**Figures 1C,D**). Nanoparticles with zeta potential values outside the range of +30 to −30 mV are considered stable for a longer period in suspended state (Yue

TABLE 1 | Antibiotic susceptibility profile of clinical Enterobacter cloacae isolates.


Antibiotic resistance profile of EspIMS6 and E. cloacae type strain ATCC 13047 control using DoDeca Enterobacteriaceae disks 1 and 2 (Himedia, India). Disk 1: Ampicillin-AMP (10 mcg), Gentamicin-GEN (10 mcg), Amikacin-AK (30 mcg), Ciprofloxacin-CIP (5 mcg), Ofloxacin-OF (5 mcg), Co-Trimoxazole-COT (25 mcg), Amoxyclav-AMC (30 mcg), Cefuroxime-CXM (30 mcg), Ceftazidime-CAZ (30 mcg), Ceftazidime/Clavulanic acid-CAC (30/10 mcg), Cefepime-CPM (30 mcg), Imipenem-IPM (10 mcg). Disk 2: Cefotaxime-CTX (30 mcg), Cetriaxone-CTR (30 mcg), Cefoxitin-CX (30 mcg), Meropenem-MRP (10 mcg), Piperacillin/Tazobactam-PIT (100/10 mcg), Aztreonam-AT (30 mcg), Gatifloxacin-GAT (5 mcg), Ampicillin/Sulbactam-A/S (10/10 mcg), Cefoperazone-CPZ (75 mcg), Levofloxacin-LE (5 mcg), Ceftizoxime-CZX (30 mcg), Ticarcillin/Clavulanic acid-TCC (75/10 mcg).

et al., 2008). To validate long-term stability of suspended AgNPs, hydrodynamic diameter, PDI and surface charge were measured 6 months post-synthesis by photon correlation spectroscopy and results compared with those obtained for initially synthesized AgNPs. Our data suggest that there is slight increase in hydrodynamic diameter (161 d.nm), with PDI (0.426) and surface charge (−28.6) (**Figures 1A,B**). However, PDI smaller than 0.5 suggested that there is no agglomeration of nanoparticles and thus these particles are stable at room temperature for longer period of time (Yue et al., 2008). Further CMT-capped AgNP was characterized by SPR observed from UV-Vis spectral analysis. The occurrence of single maximum peak at wavelength 420 nm corresponded to the SPR of AgNPs (**Figure 1E**), and reflected about the size of AgNPs to be around 30–40 nm (Oldenburg, 2012). Further characterization of this AgNP such as size, dispersion, crystallinity and presence of silver has been performed by TEM, FE-SEM, XRD and EDAX methods, which was reported in our previous article (Sanyasi et al., 2016).

### Hemocompatibility of Silver Nanoparticles

Extent of hemolysis of red blood cells was determined to check hemocompatibility of silver nanoparticles. It was observed that AgNPs at concentrations of 1, 3 and 6 µg/ml, exhibited only 0.063, 0.31, and 0.63% of hemolysis, respectively (**Figure 2**). At higher concentrations of AgNP (i.e., 9 and 12 µg/ml) hemolysis was approximately 2%. Nanoparticles exhibiting hemolysis below 5% are considered hemocompatible (Jana et al., 2016). Hence, our results indicated that AgNPs to be hemocompatible.

### Effect of Ag-MCNPs on Cell Viability

We had previously shown that AgNP's were non-toxic to mammalian cells (Sanyasi et al., 2016). Here, we have assessed the cytotoxic effect of Ag-MCNPs on macrophage RAW 264.7 cell line using MTT assay. The MTT assay is rapid, sensitive, and inexpensive method to determine cell viability. The dose dependent viability of RAW 264.7 cells in different concentration of the Ag-MCNPs is shown in **Figure 3**. Cell viability result revealed that 99% cells were viable in presence of Ag-MCNPs at concentration of up to 3 µg/ml, while at 4 µg/ml concentration of Ag-MCNPs, 82% cells remained viable (**Figure 3**). However, at higher concentration (6 µg/ml) of Ag-MCNP, cell viability was reduced to 55% suggesting that Ag-MCNP was toxic to mammalian cells at this dosage. This indicated that Ag-MCNPs below concentrations of 6 µg/ml are non-toxic to mammalian cells. Notably, the concentration of Ag-MCNPs that was antibacterial in nature (i.e., 1.5 and 3 µg/ml), are non-toxic to mammalian cells and hence could be suitable for biomedical applications.

## Antimicrobial Efficacy of AgNP and Ag-MCNP

We determined the MIC breakpoints for sliver nanoparticle (AgNP) and Ag-MCNP. We observed that both the nanoparticles tested significantly inhibited bacterial growth in a dose dependent manner (**Table 2** and Supplementary Figure S1). However, MIC

6 months post-synthesis (B). Stability of AgNPs in terms of Zeta potential (–34 mV) was measured immediately after synthesis (C) and 6 months post-synthesis (D). Determination of stability of AgNPs by SPR was conducted at λmax 420 over 10–150 min (E).

breakpoint for AgNPs was 12 µg/ml for both EspIMS6 and E. cloacae ATCC 13047. Interestingly, Ag-MCNPs had a much lower MIC breakpoint of 1.5 and 3 µg/ml for isolates EspIMS6 and E. cloacae ATCC 13047, respectively. The results clearly indicated that Ag-MCNPs was more efficient in killing the MDR E. cloacae compared to AgNPs alone.

### Effect of Silver Nanoparticles on AcrAB-TolC Efflux Pump Protein Expression

As reported from the literature, success of potential antibacterial agents rested upon their efficacy to be retained in the cell vis-àvis being effluxed out. We investigated this by studying the effect of silver nanoparticles on expression of multidrug efflux pump proteins AcrAB-TolC.

Overall immunoblotting results revealed that presence of silver nanoparticles did not significantly affect AcrA and TolC efflux pump protein expression (**Figures 4A,C**). However, in the E. cloacae ATCC 13047, AcrB, and TolC protein expression showed slight albeit insignificant decrease in presence of Ag-MCNPs (**Figures 4B,C** and Supplementary Figure S2). Inhibition of AcrB monomeric protein expression was more prominent in clinical isolate EspIMS6, where only 38 kDa AcrB protein was consistently expressed at all the time points of exposure to AgNP (at 6 µg/ml) as well as Ag-MCNP (at 0.75 µg/ml) (**Figure 4B** and Supplementary Figure S2). Though, ∼112 kDa

TABLE 2 | Antibacterial activity of silver and silver-metal composite nanoparticles.


band was less expressed in E. cloacae ATCC 13047 in response to AgNP (at 6 µg/ml), a 38 kDa band of AcrB was observed post 30 min of treatment with Ag-MCNP at a concentration of 0.75 µg/ml in E. cloacae ATCC 13047 (**Figure 4B**, panels 3 and 4 and Supplementary Figure S2). It is pertinent to mention here that, AcrB is a transmembrane RND transporter protein of the tripartite efflux pump assembly. The functional membrane embedded protein is a homotrimer with molecular masses of ∼380 kDa, with each monomer of approximately 112 kDa (Tikhonova et al., 2011). However, in both isolates EspIMS6 and E. cloacae ATCC 13047 treated with the AgNP nanoparticles, it was shown to inhibit monomeric 112 kDa bands, and showed bands corresponding to ∼38 kDa (Supplementary Figure S2). This ∼38 kDa band perhaps represents the truncated by product of the AcrB protein complex. It is pertinent to mention that the antibodies used were beyond doubt.

Expression of outer membrane protein TolC decreased post 120 min in E. cloacae ATCC 13047, in presence both the silver nanoparticles tested viz. AgNPs and Ag-MCNPs (**Figure 4C**, panels 3 and 4). Here too we did not observe any significant changes in TolC expression for the wild type isolate EspIMS6 in presence of either of the nanoparticles (**Figure 4C**, panels 1 and 2).

## DISCUSSION

The present study reports no decrease in antibacterial efficacy of silver nanoparticles viz. AgNP and Ag-MCNP, despite expression of AcrAB-TolC efflux protein in MDR bacteria. This finding is significant as it reports no noticeable effect of nanoparticles on AcrAB-TolC efflux pump protein expression that are implicated in MDR phenotype. Secondly, the results corroborate with our previous report and confirm sustainability of using the said silver nanoparticles as antibacterial agents.

During the past decade, silver nanoparticles owing to their low toxicity, antibacterial and wound healing properties, have emerged as promising therapeutic alternatives (Ambrožová et al., 2017). To ensure feasibility of large-scale use of nanoparticles for medicinal purposes, eco-friendly approaches have been widely adapted to generate silver nanoparticles. Such approaches circumvent the use of toxic chemicals and hazardous equipment. Eco-friendly green synthesis approach utilizes natural costeffective reductants such as microorganisms and plant products (Rao and Tang, 2017). We had earlier reported utilization of green synthesis approach to generate polysaccharide-capped AgNPs (Sanyasi et al., 2016).

Few of the important physical parameters essential for development of nanoparticles include size and surface charge that affects their stability (Lamprecht et al., 2001). In the present study, photon correlation spectroscopy results revealed that the CMT-capped AgNPs were smaller in size and are monodispersed in nature (Chandni et al., 2013). The mean zeta potential of AgNPs used was measured to be −34.5 ± 2.2 mV, suggesting higher electrical charge on their surface and thus, approved their long-term stability (Yue et al., 2008). SPR characterization of AgNPs demonstrated their size to be around 30–40 nm, which corroborated well with previous report (Oldenburg, 2012). Biocompatibility is extremely important feature for nanoparticles to be used as therapeutic agents. For this, we performed hemolysis assay that described the AgNPs to be non-hemolytic. We had previously reported AgNPs to be less cytotoxic toward mammalian cells even at higher concentrations (Sanyasi et al., 2016). The newly developed Ag-MCNPs reported here, too displayed minimal cytotoxic activity toward mammalian cells. Both AgNPs and Ag-MCNPs revealed excellent antimicrobial activity against multiple drug resistant E. cloacae isolates at low dose. Our results corroborated well with previous reports suggesting antimicrobial properties of silver nanoparticles (Kim et al., 2007; Chen et al., 2014), that boosted their candidature as potential antimicrobial agents. Durán et al. (2016) proposed cell membrane disruption by the silver ions as a plausible mechanism of antibacterial activity by silver nanoparticles.

However, one of the less addressed aspect has been the effect of AgNPs on bacterial efflux systems. Kovács et al. (2016) reported AgNPs to inhibit activity of p-glycoproteins efflux pump (belonging to ABC transporter family) in MDR cancer cells, thereby enhancing effectiveness of chemotherapy. Christena et al. (2015) have established the efflux inhibitory properties of copper nanoparticles (CuNPs) in tackling MDR Staphylococcus aureus and Pseudomonas aeruginosa isolates. However, effect of AgNPs on expression of bacterial

MDR efflux pumps has not been elucidated. AcrAB-TolC efflux pumps have been found to mediate MDR phenotype in many clinically significant gram-negative pathogens including E. cloacae (Piddock, 2006; Pérez et al., 2012; Amaral et al., 2013). These efflux systems efficiently extrude out of the bacterial cell, structurally divergent classes of drugs, resulting in multiple antibiotic resistance phenotype (Poole, 2005; Sun et al., 2014). This prompted us to study the

effect of AgNPs and Ag-MCNPs on AcrAB-TolC efflux pump protein expression. Our results indicated that AgNPs affected the expression of AcrB transporter protein to a greater extent as compared to AcrA and TolC. Functional AcrB is a trimeric RND transporter protein with each monomer of ∼112 kDa (Du et al., 2015). Review of literature reveals antimicrobial drugs to induce over expression of AcrAB-TolC efflux pump protein, which in turn facilitates their rapid extrusion (Rosenberg et al., 2003). In the present study, expression of functional AcrB protein was reduced in E. cloacae isolates upon addition of silver nanoparticles (Supplementary Figure S2). It is pertinent to mention that AcrB protein is responsible for substrate recognition and binding of diverse structures (Sun et al., 2014). The results provide circumstantial evidence indicative of AgNPs functioning as efflux inhibitor. Efflux inhibitory role of metal nanoparticles have been hypothesized to be either by direct binding to active site of the efflux pumps or hindering efflux kinetics (Gupta et al., 2017). However, owing to the small size of nanoparticles and substrate non-specific nature of efflux pumps proteins these explicit mechanisms still remain elusive. Earlier studies have shown that expression of AcrB transporter is affected by cellular metabolites (Ruiz and Levy, 2014) and hence forms an ideal target for efflux and virulence inhibitor design (Blair et al., 2015). Also, AcrB is a proton motive force (PMF) dependent RND transporter. Interestingly, Dibrov et al. (2002) in a study on mechanism of action of nanoparticles, observed that metal nanoparticles such as AgNPs, dissipate PMF by interfering with bacterial respiration. Hence, energy required for this tripartite efflux system may be hindered by the addition of AgNPs. In the present study, in presence of AgNPs, though TolC and AcrA protein expression remained more or less unaffected, the absence of a functional monomeric AcrB perhaps renders the bacterial cells impaired for efflux activity. Change in efflux activity in Enterobacter spp. by AgNPs has been reported in context of EmmDR efflux pump [belonging to multiple antibiotic and toxic extrusion (MATE) superfamily] (Gui et al., 2013). However, to the best of our knowledge, this is the first report on effect of

### REFERENCES


silver nanoparticles on AcrAB-TolC efflux protein expression. This study indicates potential efflux inhibitor property of the AgNPs.

To summarize, the reported silver nanoparticles generated by green synthesis method, are highly stable, small in size, non-hemolytic with minimal cytotoxic effect. Further, their antimicrobial activity is not affected by expression of AcrAB-TolC efflux proteins. Such versatile repertoire of qualities enables us to report the said silver nanoparticles as strong candidates as antibacterial agents. The results of the present study support further development of silver nanoparticle formulations as alternate therapeutic agents to combat infections caused by MDR gram-negative Enterobacterial pathogens.

## AUTHOR CONTRIBUTIONS

The present study was conceived by HM and executed by MM. SK, LG, RM, and CG developed and characterized the nanoparticles.

## FUNDING

MM received INSPIRE fellowship from the Department of Science and Technology, New Delhi (Grant No. IF110739).

### ACKNOWLEDGMENTS

HM and MM thankfully acknowledge intramural financial support from NISER for the study.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00823/full#supplementary-material

infectious diseases society of America. Clin. Infect. Dis. 48, 1–12. doi: 10.1086/ 595011


cholerae. Antimicrob. Agents Chemother. 46, 2668–2670. doi: 10.1128/AAC.46.8. 2668


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer RP and handling Editor declared their shared affiliation.

Copyright © 2018 Mishra, Kumar, Majhi, Goswami, Goswami and Mohapatra. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Efficacy of Colistin and Its Combination With Rifampin in Vitro and in Experimental Models of Infection Caused by Carbapenemase-Producing Clinical Isolates of Klebsiella pneumoniae

María E. Pachón-Ibáñez<sup>1</sup> \*, Gema Labrador-Herrera<sup>1</sup> , Tania Cebrero-Cangueiro<sup>1</sup> , Caridad Díaz<sup>2</sup> , Younes Smani<sup>1</sup> , José P. del Palacio<sup>2</sup> , Jesús Rodríguez-Baño3,4 , Alvaro Pascual3,5, Jerónimo Pachón1,4 and M. Carmen Conejo<sup>5</sup>

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Maria Bagattini, University of Naples Federico II, Italy Remy A. Bonnin, Université Paris-Saclay, France

> \*Correspondence: María E. Pachón-Ibáñez mpachon-ibis@us.es

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 25 January 2018 Accepted: 19 April 2018 Published: 15 May 2018

### Citation:

Pachón-Ibáñez ME, Labrador-Herrera G, Cebrero-Cangueiro T, Díaz C, Smani Y, del Palacio JP, Rodríguez-Baño J, Pascual A, Pachón J and Conejo MC (2018) Efficacy of Colistin and Its Combination With Rifampin in Vitro and in Experimental Models of Infection Caused by Carbapenemase-Producing Clinical Isolates of Klebsiella pneumoniae. Front. Microbiol. 9:912. doi: 10.3389/fmicb.2018.00912 <sup>1</sup> Clinical Unit of Infectious Diseases, Microbiology, and Preventive Medicine, Institute of Biomedicine of Seville, University Hospital Virgen del Rocío/CSIC/University of Seville, Seville, Spain, <sup>2</sup> Fundacion Centro de Excelencia en Investigación de Medicamentos Innovadores en Andalucía, MEDINA Foundation, Granada, Spain, <sup>3</sup> Clinical Unit of Infectious Diseases, Microbiology, and Preventive Medicine, Institute of Biomedicine of Seville, University Hospital Virgen de Macarena/CSIC/University of Seville, Seville, Spain, <sup>4</sup> Department of Medicine, University of Seville, Seville, Spain, <sup>5</sup> Department of Microbiology, University of Seville, Seville, Spain

Despite the relevance of carbapenemase-producing Klebsiella pneumoniae (CP-Kp) infections there are a scarce number of studies to evaluate in vivo the efficacy of combinations therapies. The bactericidal activity of colistin, rifampin, and its combination was studied (time–kill curves) against four clonally unrelated clinical isolates of CP-Kp, producing VIM-1, VIM-1 plus DHA-1(acquired AmpC β-lactamase), OXA-48 plus CTX-M-15 (extended spectrum β-lactamase) and KPC-3, respectively, with colistin MICs of 0.5, 64, 0.5, and 32 mg/L, respectively. The efficacies of antimicrobials in monotherapy and in combination were tested in a murine peritoneal sepsis model, against all the CP-Kp. Their efficacies were tested in the pneumonia model against the OXA-48 plus CTX-M-15 producers. The development of colistin-resistance was analyzed for the colistin-susceptible strains in vitro and in vivo. In vitro, colistin plus rifampin was synergistic against all the strains at 24 h. In vivo, compared to the controls, rifampin alone reduced tissue bacterial concentrations against VIM-1 and OXA-48 plus CTX-M-15 strains; CMS plus rifampin reduced tissue bacterial concentrations of these two CP-Kp and of the KPC-3 strain. Rifampin and the combination increased the survival against the KPC-3 strain; in the pneumonia model, the combination also improved the survival. No resistant mutants appeared with the combination. In conclusion, CMS plus rifampin had a low and heterogeneous efficacy in the treatment of severe peritoneal sepsis model due to CP-Kp producing different carbapenemases, increasing survival only against the KPC-3 strain. The combination showed efficacy in the less severe pneumonia model. The combination prevented in vitro and in vivo the development of colistin resistant mutants.

Keywords: Klebsiella pneumoniae, animal models, carbapenemase producers, colistin, rifampin

## INTRODUCTION

fmicb-09-00912 May 11, 2018 Time: 15:54 # 2

Carbapenem-resistant Klebsiella pneumoniae strains are spreading worldwide, representing an urgent threat to public health, as stressed by the Center for Disease Control and Prevention (CDC) of United States, the European Centre for Disease Prevention and Control (ECDC) and the World Health Organization (WHO). The rapid spread, mostly in hospital settings, is transforming many common health care-associated complications into infections that are sometimes untreatable with the currently available antimicrobials (Nordmann et al., 2012).

The carbapenem resistance in K. pneumoniae is mainly due to the production of acquired carbapenemases (Falagas et al., 2014). The most important carbapenemases found in this species may belong to the Ambler classes A (mainly KPC), B (the most frequent are VIM and IMP) and D (OXA-48-like enzymes). Invasive infections by isolates producing VIM and KPC are associated with high death rates (Tzouvelekis et al., 2012). The information about infections caused by OXA-48 producers is scarce, mostly because of its difficult identification (Nordmann et al., 2011; Canton et al., 2012; Tzouvelekis et al., 2012). Nevertheless, OXA-48 is the most frequent carbapenemase produced by Enterobacteriaceae isolated in many European countries (Canton et al., 2012; Palacios-Baena et al., 2016; De Laveleye et al., 2017).

Although the hydrolysis spectrum of these enzymes may vary, they hydrolyze most beta-lactams, including carbapenems. Moreover, carbapenemase producers often show co-resistance to other antimicrobial agents, leaving very few treatment options, such as tigecycline, colistin, and some aminoglycosides (de Oliveira et al., 2015). So, in an attempt to improve the poor clinical efficacy of the available drugs, combination therapy is often used as definitive therapy for infections caused by carbapenemase-producing K. pneumoniae (CP-Kp). To date, recommendations are based on few retrospective clinical studies and in vitro studies (Tzouvelekis et al., 2012). In addition, a scarce number of in vivo studies have assessed the efficacy of antimicrobial combinations against CP-Kp (Tzouvelekis et al., 2012). Clinical studies have reported favorable outcomes for patients treated with combinations of colistin and a carbapenem, tigecycline, fosfomycin, or an aminoglycoside (Michalopoulos et al., 2010; Lee and Burgess, 2012; Munoz-Price et al., 2013; Daikos et al., 2014). However, little data exists on which combination therapy is superior.

Several studies have reported synergistic activity of colistin and rifampin against colistin-resistant and colistin-susceptible KPC-producing K. pneumoniae clinical strains, using the checkerboard method. Similarly, studies based on time-kill experiments detected synergy with this combination against colistin-resistant KPC or NDM producers clinical strains, but no synergistic effect against VIM producers. To our knowledge, there is no data regarding the efficacy of this combination against K. pneumoniae producing OXA-48, nor in vivo studies to validate the in vitro results previously mentioned (Elemam et al., 2010; Tascini et al., 2013; Nastro et al., 2014; Tangden et al., 2014). Thus, the aim of this study was to evaluate the in vitro and in vivo efficacy of CMS plus rifampin against CP-Kp clinical strains producing different carbapenemases.

## MATERIALS AND METHODS

## Bacterial Strains, Beta-lactamase Characterization and Molecular Typing

Four genetically unrelated clinical isolates of CP-Kp were studied: Kp07, a VIM-1 ST 1603 clone producer (Miro et al., 2013); Kp21, which co-produced VIM-1 and the acquired AmpC type beta-lactamase DHA-1 ST 11 clone (Miro et al., 2013); Kp28, co-producing OXA-48 ST11 clone and the extended spectrum beta-lactamase (ESBL) CTX-M-15 (Oteo et al., 2015); and Kp29, co-producing KPC-3 ST512 clone with the broad spectrum beta-lactamases TEM-1 and SHV-11 (Lopez-Cerero et al., 2014), thereinafter VIM-1, VIM-1/DHA-1, OXA-48 plus CTX-M-15, and KPC-3 producers, respectively. Identification of these isolates was confirmed by a Microflex LT-MALDI Biotyper mass spectrometer (Bruker Daltonics GmbH, Bremen, Germany). The presence of carbapenemase genes, and genes coding for other beta-lactamases was confirmed by PCR and sequencing as described previously. The absence of genetic relation among the isolates was confirmed by PFGE analysis of chromosomal restriction fragments obtained after XbaI cleavage following the criteria of Tenover et al. (1995). Two of the strains, KPC-3 and VIM-1 (DHA-1) were multidrug-resistant while the other two were not. The antibiotic susceptibility profiles are included in the Supplementary Information.

### Antimicrobials

For the in vitro assays, antimicrobials, colistin sulfate salt and rifampin, were used as standard laboratory powders (Sigma-Aldrich, Madrid, Spain). For in vivo experiments, clinical formulations were used: colistimethate sodium (CMS) (Genéricos Españoles S.A., Madrid, Spain) and rifampin (Sanofi-Aventis, Madrid, Spain).

## In Vitro Studies

### Antimicrobial Susceptibility Testing

MICs of antibiotics were determined by broth microdilution as recommended by the Clinical and Laboratory Standards Institute [CLSI] (2012), using Mueller Hinton broth II (MHB) (Becton Dickinson & Co., Sparks, MD, United States) and agar dilution method for fosfomycin. MIC results were interpreted according to the European Committee on Antimicrobial Susceptibility Testing [EUCAST] (2016)<sup>1</sup> breakpoints. Studies were performed in triplicate.

### Time-Kill Curves

The concentrations of colistin used for susceptible strains corresponded to the MIC value obtained by microdilution,

<sup>1</sup>http://www.eucast.org/clinical\_breakpoints/

whereas the concentration used for resistant strains was 2 mg/L, the susceptibility breakpoint recommended by European Committee on Antimicrobial Susceptibility Testing [EUCAST] (2016). Rifampin was used at a fixed concentration of 2 mg/L. Experiments were carried out with a starting inoculum of 1 × 10<sup>6</sup> cfu/mL and the antibiotics alone or in combination. Tubes were incubated at 37◦C, with shaking, and samples were taken at 0, 1, 3, 6, and 24 h, serially diluted and plated using a spiral platter (Eddy Jet, IUL S.A., Barcelona, Spain). Bacterial colonies were counted using an automatic counter (Flash & Go, IUL S.A., Barcelona, Spain) (Pournaras et al., 2011; Souli et al., 2011). Experiments were performed three times on separate occasions. Synergy was defined as a decrease ≥ 2 log<sup>10</sup> cfu/mL for the antimicrobial combination compared with the most active single agent. Bactericidal activities of single antibiotics or combination were defined as a decrease ≥ 3 log<sup>10</sup> cfu/mL from the starting inoculum (Souli et al., 2009). Studies were performed in triplicate.

### Animals

Immunocompetent C57BL/6 female mice weighing approximately 20 g (Production and Experimentation Animal Center, University of Seville, Seville, Spain) were used; they had a sanitary status of murine pathogen free and were assessed for genetic authenticity. Mice were housed in an individually ventilated cage system under specific pathogen-free conditions, and water and food supplied ad libitum. This study was carried out following the recommendations in the Guide for the Care and Use of Laboratory Animals (Souli et al., 2011). Experiments were approved by the Committee on the Ethics of Animal Experiments of the University Hospital Virgen Macarena, Seville, Spain (CI 1961). All procedures were performed under sodium thiopental (B. Braun Medical S.A., Spain) anesthesia, and all efforts were made to minimize suffering.

### Pharmacokinetic/Pharmacodynamic Analysis

Serum antibiotic concentrations were determined in healthy mice after a single intraperitoneal (ip) administration of CMS (20 mg/kg) or rifampin (25 mg/kg). In sets of three anesthetized mice, blood samples from the periorbital plexus were obtained at different time points after the administration of CMS and rifampin. Blood samples were immediately centrifuged (4500 rpm, 15 min at 4◦C), and serum samples stored at −80◦C until the analysis. Both serum free and total antibiotic concentrations were measured using HPLC-tandem mass spectrometry (LC-MS/MS) (Waters et al., 2008; Gobin et al., 2010). Free fractions of the drugs were calculated as described previously (Waters et al., 2008; Gonzalez et al., 2013; Cheah et al., 2015).

Maximum concentration of drug in serum (Cmax), elimination half-life (t1/2), area under the concentration-time curve from 0 to 24 h (AUC0−24), free AUC0−<sup>24</sup> (fAUC0−24), AUC0−24/MIC and fAUC0−24/MIC ratios were calculated using the PKSOLVER program (Zhang et al., 2010). The pharmacodynamic targets to assess the efficacy were fAUC0−24/MIC and AUC0−24/MIC for CMS and rifampin, respectively.

# Experimental Models

## Peritoneal Sepsis Model

A previously described murine peritoneal sepsis model was used (Parra Millan et al., 2016). Briefly, groups of un-anesthetized C57BL/6 mice were infected by ip injection of 0.5 mL of the Minimal Lethal Dose (MLD) of each strain, corresponding to (log<sup>10</sup> cfu/mL): 8.97, 8.94 9.35, and 8.38 for VIM-1, VIM-1/DHA-1, OXA-48 plus CTX-M-15, and KPC-3 strains, respectively. Treatments were initiated 4 h post-inoculation. For each one of the strains, mice were randomly included into four different therapeutic groups: (1) controls (untreated), (2) CMS, 20 mg/kg/8 h/ip, (3) rifampin, 25 mg/kg/6 h/ip, and (4) CMS plus rifampin, using the same dosing schedule as in monotherapy. The sample size for the combination groups and CMS monotherapy for the colistin-susceptible strains (VIM-1 and OXA-48 plus CTX-M-15) was 15 mice; nevertheless, for the monotherapies in case of resistance to colistin (VIM-1/DHA-1 and KPC-3) and for rifampin, the sample size was 10 to accomplish the 3Rs (Replacement, Reduction and Refinement) rules for performing animal research. Afterward, mice were treated and monitored for 72 h. The antimicrobial dosages were based on the PK/PD data, and their proven efficacy, alone and in combination, in previous experimental murine models of infection (Pachon-Ibanez et al., 2010). Samples were extracted and processed immediately after the death of mice; the survivor mice were sacrificed (sodium thiopental) at the 72 h. Aseptic thoracotomies were carried out, and blood samples were obtained for qualitative blood cultures; results were expressed as positive (≥1 cfu present in the plate) or negative. Spleens were aseptically extracted, weighed, and homogenized in sterile saline (Stomacher 80; Tekmar Co., Cincinnati, OH, United States) before quantitative cultures (log<sup>10</sup> cfu/g) in Columbia agar with 5% sheep blood plates.

### Pneumonia Model

A previously characterized murine pneumonia model was used (Pachon-Ibanez et al., 2010; Docobo-Perez et al., 2012; Parra Millan et al., 2016) with the OXA-48 plus CTX-M-15 strain, as OXA-48 is the most prevalent carbapenemase in Spain nowadays (Palacios-Baena et al., 2016). Anesthetized mice were infected intratracheally, using 50 µl of a final inoculum of 8.18 log<sup>10</sup> cfu/mL. Therapies were initiated 4 h post-infection and treatment groups and dosages were the same as in the peritoneal sepsis model. Mice were treated and monitored over 72 h. After death or sacrifice of the mice at 72 h, blood and lungs samples were aseptically obtained and processed as detailed above.

### In Vitro and in Vivo Selection of Colistin-Resistant and Rifampin-Resistant Mutants

Up to 10 colonies, of colistin-susceptible strains exposed during 24 h to colistin at the MIC, alone or in combination with rifampin, in time-kill curves, and from controls (non-exposed to

the antimicrobials), were sub-cultured twice in antimicrobial free medium and frozen at −80◦C until MIC testing.

In the experimental murine pneumonia model, after processing the lungs as described above, the remaining homogenized tissue was vortexed and centrifuged, and the pellet was resuspended in 2 mL of sterile saline. All the volume was spread on agar and incubated for 48 h at 37◦C. A maximum of ten colonies recovered from each lung were selected, sub cultured in antibiotic free medium twice and frozen at −80◦C until MIC testing.

MICs determinations of colistin and rifampin were carried out in triplicate.

### Statistical Analysis

Mortality and bacteremia rates are expressed as percentages and bacterial tissue concentrations (log<sup>10</sup> cfu/g) as means ± SD. The two-tailed Fisher's test, analysis of variance (ANOVA), and the Dunnet and Tukey post hoc tests were used. A P < 0.05 was considered significant. The SPSS v22.0 statistical package was used (SPSS Inc).

### RESULTS

### In Vitro Results

### Antimicrobial Susceptibility Testing

MICs of each antibiotic for the four clinical isolates are shown in **Table 1**. Two strains were resistant to colistin (VIM-1/DHA-1 and KPC-3 producers).

### Time-Kill Curves

The results are shown in **Figure 1**. Colistin was bactericidal at 3 h against the VIM-1 producer, but an important regrowth at 24 h. In addition, colistin reduced bacterial concentration (2.5 log<sup>10</sup> decrease in cfu/mL) of the OXA-48 plus CTX-M-15 producing isolate, again with an important regrowth at 24 h. Colistin did not display any bactericidal activity against the colistin-resistant isolates (VIM-1/DHA-1 and KPC-3 producers). Rifampin did not show bactericidal activity against any strain. The combination was synergistic against all the strains at 24 h, but achieving only a bacteriostatic effect against the VIM-1, VIM-1/DHA-1, and OXA-48 plus CTX-M-15 isolates; on the contrary, the combination was bactericidal against the KPC-3 isolate.

TABLE 1 | MICs of colistin and rifampin for the four carbapenemase-producing K. pneumoniae clinical strains.


<sup>∗</sup>Susceptible, MIC ≤ 2 mg/L. ∗∗Resistant, MIC > 2 mg/L. ∗∗∗Breakpoints are not defined (intrinsically resistant).

## Pharmacokinetics and Pharmacodynamics

Pharmacokinetic parameters of each antimicrobial are shown in **Table 2**. Pharmacodynamics profiles are shown in **Tables 3**, **4**.

### In Vivo Results: Peritoneal Sepsis Model

The efficacies of the antimicrobials are shown in **Table 3**. Mortality, bacterial clearance on spleen and bacteremia, are analyzed immediately after the death of mice or at the end of the experiment (72 h of treatment).

### Mortality

Mortality in all control groups (non-treated) was 100% within the first 24 h post-infection. Rifampin and its combination with CMS reduced mortality in animals infected with the KPC-3 strain to a 66.67% and a 40%, respectively, but not CMS alone. There was no reduction of mortality with any treatment when the peritoneal sepsis was produced by the other strains.

### Bacterial Clearance From Spleen

Rifampin alone improved significantly the clearance of bacteria from spleen (CFU/g of tissue) compared with the control and the CMS groups in mice infected with either of the two colistinsusceptible strains producing VIM-1 (7.14 ± 0.48 vs. 8.92 ± 0.46 and 7.14 ± 0.48 vs. 8.60 ± 0.30, respectively) or OXA-48 plus CTX-M-15 (8.24 ± 0.76 vs. 9.56 ± 0.47, and 8.24 ± 0.76 vs. 9.52 ± 0.55, respectively). CMS plus rifampin reduced the bacterial concentration compared with the controls for both colistin-susceptible strains, producing VIM-1 or OXA-48 plus CTX-M-15 and for the colistin-resistant strain producing KPC-3, (6.01 ± 1.77 vs. 8.92 ± 0.46, 7.06 ± 2.50 vs. 9.56 ± 0.47, and 6.76 ± 2.48 vs. 10.19 ± 0.29, respectively). The combination was also better than CMS alone for both colistin-susceptible strains (6.01 ± 1.77 vs. 8.60 ± 0.30, 7.06 ± 2.50 vs. 9.52 ± 0.55, respectively).

### Bacteremia

CMS plus rifampin showed the best effect among all the treatment groups in sterilizing blood cultures of mice infected with any of the four strains. Nevertheless, this reduction was only significant in mice infected with the VIM-1 producer (53.33% vs. 100%).

### Pneumonia Model Results

**Table 4** summarizes the results for the pneumonia model.

### Mortality

The severity of this model was lower, with a mortality of 40% at the end of the experiment (72 h) using the MLD in the control group. Only the combination therapy decreased significantly mortality compared with the control group (6.26% vs. 40%; P < 0.05).

### Bacterial Clearance From Lungs

Rifampin alone and its combination with CMS decreased significantly bacterial lung concentration compared with the

FIGURE 1 | Time-kill curves for colistin (CST) and rifampin (RIF) alone and in combination against four clinical strains of carbapenemase producing strains (VIM-1, VIM-1/DHA-1, OXA-48/CTX-M-15 and KPC3). The CST concentration used for colistin-susceptible strains corresponded to the value of their MICs (colistin 0.5 mg/L for VIM-1 and OXA-48/CTX-M-15); for colistin-resistant isolates the concentration of CST was that corresponding to the susceptibility breakpoint recommended by EUCAST (colistin 2 mg/L for VIM-1/DHA-1 and KPC-3). The RIF concentration used for all strains was 2 mg/L. RIF, filled circles; CST, filled squares; combination of both antimicrobials, filled triangles.

control (3.07 ± 1.63 vs. 8.05 ± 2.36, and 2.99 ± 1.36 vs. 8.05 ± 2.36, respectively).

### Bacteremia

None of the antimicrobial treatments achieved a significant reduction of the bacteremia in comparison with the control group.

### In Vitro and in Vivo Selection of Colistin-Resistant Mutants and Rifampin-Resistant Mutants

Colistin MICs of colonies recovered from susceptible strains (producing VIM-1 or OXA-48 plus CTX-M-15) after exposure at the MIC during 24 h in time-kill curves increased from 0.5 mg/L to > 8 mg/L. No mutants resistant to colistin were detected when these strains were exposed to the combination of colistin plus rifampin.

The MIC of rifampin for the colonies recovered from the lungs of mice challenged with OXA-48 plus CTX-M-15 producer and treated with rifampin alone increased from 16 to > 256 mg/L (16-folds) and the MIC of colistin from the colonies recovered from the lungs of mice treated with CMS as monotherapy increased from 1 to > 32 mg/L (32 folds). Nevertheless, the MICs of rifampin and colistin remained unchanged for those recovered from the control and the combination group.

## DISCUSSION

The results of this study show that CMS alone has no significant efficacy in terms of bacterial clearance either from tissue and blood or in the survival rate in the peritoneal sepsis model, even in animals infected with strains susceptible to this antimicrobial. To investigate whether these disappointing results were due to a model effect, additional experiments were made in a less severe murine pneumonia model using the strain producing OXA-48 plus CTX-M-15. Once again, the bacterial lung and blood concentrations as well as mortality were no different from those

TABLE 2 | Pharmacokinetic profiles of CMS and rifampin in mice serum.


CST, colistin; RIF, rifampin; ip, intraperitoneal; tCST, total colistin; fCST, free colistin; tRIF, total rifampin; fRIF, free rifampin; ip., intraperitoneal; Cmax, maximum concentration of drug in serum; t1/2, half-life; AUC0−24, area under the concentration-time curve from 0 to 24 h; fAUC0−24, the unbound fraction AUC0−24.


TABLE 3 | In vivo efficacy and pharmacodynamics of CMS and rifampin, alone and in combination, for the experimental peritoneal sepsis model.

CTL, control (no antimicrobial treatment); CST, colistin; RIF, rifampin; cfu, colony-forming unit; AUC0−24, area under the concentration-time curve from 0 to 24 h; fAUC0−24, the unbound fraction AUC0−24. <sup>∗</sup>PD parameters were determined in healthy mice after CMS or rifampin injection. <sup>a</sup>P ≤ 0.02 compared to CTL and CST groups. <sup>b</sup>P < 0.05 compared to CTL, CST, and RIF groups. <sup>c</sup>P < 0.05 compared to CTL and CST groups. <sup>d</sup>P < 0.05 compared to CTL group.


CTL, control (no antimicrobial treatment); CST, colistin; RIF, rifampin; cfu, colony-forming unit, AUC0−24, area under the concentration-time curve from 0 to 24 h; fAUC0−24, the unbound fraction AUC0−24. <sup>∗</sup>PD parameters were determined in healthy mice after CMS or rifampin injection. <sup>a</sup>P < 0.05 compared to CTL group.

of the control group. We believe that these results are related with the development of colistin resistance during the CMS monotherapy found in the pneumonia model.

The CMS dosage used in this study has been proven to be effective in several murine experimental studies using clinical multidrug-resistant (MDR) Acinetobacter baumannii strains susceptible to colistin (MIC = 0.5 mg/L) (Pachon-Ibanez et al., 2010; Docobo-Perez et al., 2012; Parra Millan et al., 2016). Nevertheless, when this CMS dosage was used in a murine pneumonia model by a NDM-1-producing K. pneumoniae clinical strain, CMS monotherapy was not effective as we found in the present study (Docobo-Perez et al., 2012). Similar results were reported by de Oliveira et al. (2015) in a retrospective clinical study, observing suboptimal efficacy of polymyxins in the treatment of KPCproducing Enterobacteriaceae infections, even in combination with imipenem or meropenem.

One possible explanation to the lack of efficacy found with CMS alone, could be that the PD parameter predictive of colistin efficacy, fAUC0−24/MIC ratio (Cheah et al., 2015), has been described in infections caused by other Gram-negative bacilli, such as Pseudomonas aeruginosa and A. baumannii. So, the present results would suggest that the described fAUC0−24/MIC values to achieve various magnitudes of bacterial killing in these Gram-negative bacilli are not the appropriate for the treatment of CP-Kp infections, even for isolates with the same MIC values. However, in these studies the end-point to define the fAUC0−24/MIC values predicting colistin efficacy was to achieve a 2 log<sup>10</sup> decrease in bacterial concentrations in lungs or thigh, but not to evaluate the mice survival rates.

Colistimethate sodium monotherapy did not reduce significantly the bacterial lung and blood results or the survival in the less severe pneumonia model by the strain producing OXA-48 plus CTX-M-15. However, it is noteworthy that CMS achieved a bacterial lung decrease of 2.91 log<sup>10</sup> cfu/g, a value defined as optimal for P. aeruginosa and A. baumannii (Cheah et al., 2015), although the wide SD precluded the significance of these data.

In addition, the results in the pneumonia model demonstrate that the severity of the chosen animal model is important when studying the efficacy of antimicrobial treatments. Thus, new colistin fAUC0−24/MIC values have to be defined for CP-Kp and, in our opinion, for the treatment of other Gram-negative bacilli infections.

The activity of rifampin in monotherapy in the peritoneal sepsis model was also limited, reducing only bacterial spleen concentration compared to control and CMS groups, infected with VIM-1 or OXA-48 plus CTX-M-15 strains, for which an AUC0−<sup>24</sup> <sup>h</sup>/MIC = 34.5, predictor of efficacy was achieved. In the case of animals infected with the KPC-3 strain, rifampin did not decrease significantly the bacterial load in spleen or blood, but reduced the mortality. However, the survival mice conditions suggested that if the experiment were longer than 72 h, all animals included in this treatment group would have died shortly after. Moreover, a huge increase in MIC of rifampin was observed after rifampin in vivo monotherapy. In the less severe pneumonia model by OXA-48 plus CTX-M-15 strain, rifampin monotherapy increased significantly the bacterial clearing from lungs and blood, and also increased the mice survival. The efficacy of rifampin against other Gram-negative bacteria has been showed in experimental pneumonia models in mice (Wolff et al., 1999; Montero et al., 2004; Pachon-Ibanez et al., 2010, 2011). Nevertheless, rifampin alone cannot be an alternative due to the generation of rifampin-resistant mutants (Pachon-Ibanez et al., 2006).

Colistimethate sodium in combination with rifampin demonstrated in vitro a synergistic effect against the four strains, but the effect was bactericidal only against KPC-3 strain. These results are in accordance with other studies showing synergy of this combination against KPC-producing K. pneumoniae strains. Nastro et al. (2014) reported a synergistic effect at 24 h against 27 colistin-resistant KPC-2-producing K. pneumoniae clinical strains. Similarly, Elemam et al. (2010) found synergy with polymyxin B plus rifampin against 12 KPC-producing K. pneumoniae clinical strains. Tascini et al. (2013) also reported that CMS plus rifampin exhibited synergy against 13 colistin-resistant KPC-producing K. pneumoniae clinical strains, moreover being bactericidal against the 62%. On the contrary, in the case of K. pneumoniae producing VIM-1 the results of the present study differ from those by Tangden et al. (2014), where CMS plus rifampin did not exhibit synergy against two colistin-susceptible VIM-1-producing K. pneumoniae clinical strains.

With regard to the in vivo efficacy of CMS plus rifampin in the experimental murine peritoneal sepsis model by the colistin-susceptible VIM-1, the combination was better than the control and CMS groups taking into account the bacterial clearance from spleen and blood. When using the colistinsusceptible OXA-48 plus CTX-M-15 strain, the combination only reduced the bacterial spleen concentration. For these two colistin-susceptible strains, CMS and rifampin achieved the pharmacodynamic values described as optimal, fAUC0−24/MIC and AUC0−24/MIC, respectively (Jayaram et al., 2003; Gumbo et al., 2007; Landersdorfer et al., 2017). Nevertheless, is worth mentioning that the efficacy of the combination for both colistin-susceptible strains was not optimal, if we consider that mortality rates remained 93.3 and 73.3%, respectively.

Against the colistin-resistant KPC-3 strain, this combination reduced the bacterial spleen concentration and mortality, in accordance with the in vitro synergy studies. Nastro et al. (2014) reported a favorable outcome in five patients with colistin-resistant KPC-2-producing K. pneumoniae infections when treated with a combination of CMS plus rifampin. Against the other colistin-resistant strain, producing VIM-1/DHA-1, this combination was not efficacious, in contrast with the synergy observed in the time-kill studies.

In the pneumonia model caused by the OXA-48 plus CTX-M-15 strain, the activity of CMS plus rifampin was similar to that of rifampin in monotherapy, decreasing the bacterial lung concentration and increasing the mice survival compared with the control group. This combination has been successfully used against other Gram-negative bacilli (Pachon-Ibanez et al., 2010). In this pneumonia model, the strain of K. pneumoniae developed colistin resistance and a considerable increase in the MIC of rifampin when used as monotherapies, which was prevented with the combination. This prevention of colistin resistant mutant with this combination have been reported previously as in the in vitro study published by Rodriguez et al. (2010) in which they proved that the association of colistin plus rifampin was synergistic against heteroresistant A. baumannii isolates and prevented the development of colistin-resistant mutants (Rodriguez et al., 2010).

Limitations of the study have to be done, first that even though we wanted to evaluate the in vitro and in vivo effect of colistin plus rifampin combination against clonally unrelated clinical isolates of K. pneumoniae producing other carbapenemases, the number of the tested strains is low. Moreover, the less severe pneumonia model was only performed with one of the strains, and we believe it will be very interesting to do it with all of the strains.

In summary, the results obtained suggest that CMS plus rifampin has a low and heterogeneous efficacy in the treatment of severe infections, such as a peritoneal sepsis infection, caused by different CP-Kp strains, increasing only the mice survival in the infection caused by the KPC-3 strain. CMS plus rifampin combination prevents in vivo the development of mutants resistant to colistin. Because of the lack of available alternatives for these colistin-resistant KPC-3-producing strains, the combination of CMS and rifampin might be considered for further evaluation. Moreover, this combination showed efficacy in the less severe pneumonia model by the OXA-48 plus CTX-M-15 strain. Finally, an optimal PD index value for CMS efficacy needs to be defined for K. pneumoniae. Overall, the results of the present study suggest that the efficacy of CMS plus rifampin depends on the class of carbapenemase produced by K. pneumoniae and on the severity of the infection.

### AUTHOR CONTRIBUTIONS

MP-I has planned and coordinated the experiments, analyzed the results, and written the manuscript. GL-H and TC-C had performed the in vitro and in vivo experiments. CD and JPP

had performed the antibiotic concentrations studies by HPLCtandem mass spectrometry (LC-MS/MS). YS had reviewed the manuscript and the experiments. JR-B, AP, and JP had reviewed the manuscript and the experiments. MCC obtained the funds to perform the studies and wrote the project, contributed to the performance of the in vitro experiment, and reviewed the results and the manuscript.

### FUNDING

This study was supported by the Consejería de Salud of the Junta de Andalucía (PI-0622-2012) and supported by Plan Nacional de I+D+i and Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación

### REFERENCES


Cooperativa, Ministerio de Economía y Competitividad, Spanish Network for Research in Infectious Diseases (REIPI RD12/0015/0001) – co-financed by European Development Regional Fund "A way to achieve Europe" ERDF. The MEDINA authors disclosed the receipt of financial support from Fundación MEDINA, a public-private partnership of Merck Sharp & Dohme de España S.A./Universidad de Granada/Junta de Andalucía.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00912/full#supplementary-material

chromatography-tandem mass spectrometry. Antimicrob. Agents Chemother. 54, 1941–1948. doi: 10.1128/AAC.01367-09



combinations with fosfomycin against KPC-2-producing Klebsiella pneumoniae and protection of resistance development. Antimicrob. Agents Chemother. 55, 2395–2397. doi: 10.1128/AAC.01086-10


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Pachón-Ibáñez, Labrador-Herrera, Cebrero-Cangueiro, Díaz, Smani, del Palacio, Rodríguez-Baño, Pascual, Pachón and Conejo. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Trans-Cinnamaldehyde and Eugenol Increase Acinetobacter baumannii Sensitivity to Beta-Lactam Antibiotics

Deepti P. Karumathil<sup>1</sup> , Meera Surendran Nair<sup>1</sup> , James Gaffney<sup>2</sup> , Anup Kollanoor-Johny<sup>3</sup> and Kumar Venkitanarayanan<sup>1</sup> \*

<sup>1</sup> Department of Animal Science, University of Connecticut, Storrs, CT, United States, <sup>2</sup> College of Veterinary Medicine, Cornell University, Ithaca, NY, United States, <sup>3</sup> Department of Animal Science, University of Minnesota, Saint Paul, MN, United States

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Vishvanath Tiwari, Central University of Rajasthan, India Krassimira Hristova, Marquette University, United States Nagendran Tharmalingam, Brown University, United States

### \*Correspondence:

Kumar Venkitanarayanan kumar.venkitanarayanan@uconn.edu

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 22 August 2017 Accepted: 30 April 2018 Published: 23 May 2018

### Citation:

Karumathil DP, Nair MS, Gaffney J, Kollanoor-Johny A and Venkitanarayanan K (2018) Trans-Cinnamaldehyde and Eugenol Increase Acinetobacter baumannii Sensitivity to Beta-Lactam Antibiotics. Front. Microbiol. 9:1011. doi: 10.3389/fmicb.2018.01011 Multi-drug resistant (MDR) Acinetobacter baumannii is a major nosocomial pathogen causing a wide range of clinical conditions with significant mortality rates. A. baumannii strains are equipped with a multitude of antibiotic resistance mechanisms, rendering them resistant to most of the currently available antibiotics. Thus, there is a critical need to explore novel strategies for controlling antibiotic resistance in A. baumannii. This study investigated the efficacy of two food-grade, plant-derived antimicrobials (PDAs), namely trans-cinnamaldehyde (TC) and eugenol (EG) in decreasing A. baumannii's resistance to seven β-lactam antibiotics, including ampicillin, methicillin, meropenem, penicillin, aztreonam, amoxicillin, and piperacillin. Two MDR A. baumannii isolates (ATCC 17978 and AB 251847) were separately cultured in tryptic soy broth (∼6 log CFU/ml) containing the minimum inhibitory concentration (MIC) of TC or EG with or without the MIC of each antibiotic at 37◦C for 18 h. A. baumannii strains not exposed to the PDAs or antibiotics served as controls. Following incubation, A. baumannii counts were determined by broth dilution assay. In addition, the effect of PDAs on the permeability of outer membrane and efflux pumps in A. baumannii was measured. Further, the effect of TC and EG on the expression of A. baumannii genes encoding resistance to β-lactam antibiotics (blaP), efflux pumps (adeABC), and multi-drug resistant protein (mdrp) was studied using real-time quantitative PCR (RT-qPCR). The experiment was replicated three times with duplicate samples of each treatment and control. The results from broth dilution assay indicated that both TC and EG in combination with antibiotics increased the sensitivity of A. baumannii to all the tested antibiotics (P < 0.05). The two PDAs inhibited the function of A. baumannii efflux pump, (AdeABC), but did not increase the permeability of its outer membrane. Moreover, RT-qPCR data revealed that TC and EG down-regulated the expression of majority of the genes associated with β-lactam antibiotic resistance, especially blaP and adeABC (P < 0.05). The results suggest that TC and EG could potentially be used along with β-lactam antibiotics for controlling MDR A. baumannii infections; however, their clinical significance needs to be determined using in vivo studies.

Keywords: A. baumannii, trans-cinnamaldehyde, eugenol, antibiotic resistance

## INTRODUCTION

fmicb-09-01011 May 18, 2018 Time: 16:54 # 2

Acinetobacter baumannii is a multi-drug resistant (MDR) Gram-negative, aerobic bacillus that has emerged as a major cause of nosocomial infections with mortality rates ranging from 34 to 61% (Karageorgopoulos and Falagas, 2008; Wieczorek et al., 2008; Esterly et al., 2011). In humans, MDR A. baumannii causes a wide-spectrum of clinical conditions, including pneumonia (Leung et al., 2005), blood-stream infections (Wisplinghoff et al., 2004; Centers for Disease Control and Prevention [CDC], 2004), meningitis (Metan et al., 2007), urinary tract infections (Sunenshine et al., 2007), and wound infections (Scott et al., 2007) In addition, reports of other manifestations such as endocarditis, peritonitis, and osteomyelitis associated with A. baumannii have been reported (Olut and Erkek, 2005; Menon et al., 2006) A. baumannii is ranked as one of the most common bacteria associated with intensive care units (Garnacho-Montero and Amaya-Villar, 2010; Ulu-Kilic et al., 2013), and is difficult to treat due to its resistance to most of the currently available antibiotics (Maragakis and Perl, 2008; Doi et al., 2009; Neonakis et al., 2011; Al Mobarak et al., 2014; Ellis et al., 2015) For example, carbapenems, which were once the antimicrobials of choice for treating A. baumannii, are no longer completely effective due to resistance development by the bacterium (Falagas et al., 2006; Bassetti et al., 2008; Abbott et al., 2013; Fonseca et al., 2013). Similarly, although polymyxins have been successfully used to treat A. baumannii infections, strains resistant to these drugs have appeared (Hernan et al., 2009; Lean et al., 2014; Pogue et al., 2015). In light of these reports, the Infectious Diseases Society of America ranked A. baumannii as one of the top priority, antibiotic-resistant pathogens to target due to its rapid propensity to develop drug resistance, and a limited choice of antibiotics available to treat infections caused by this bacterium (Talbot et al., 2006; Shlaes et al., 2013).

A. baumannii is considered to be multidrug resistant, if it exhibits resistance to more than three classes of antibiotics (Falagas et al., 2006). The resistance of A. baumannii to antibiotics has been attributed to multiple mechanisms, including reduced permeability of its outer membrane to antibiotics, constitutive expression of drug efflux pumps, and its ability to acquire and incorporate genetic elements such as plasmids, transposons, and integrons (Giamarellou et al., 2008; Cai et al., 2012). In addition, A. baumannii has a significant ability to produce biofilms on various surfaces (Espinal et al., 2012; Longo et al., 2014), which not only increases the potential of A. baumannii for nosocomial spread, but also contributes to its resistance to antibiotics and virulence (Lee et al., 2008; Rao et al., 2008). Thus, there is a critical need to explore novel strategies for treating A. baumannii infections.

Traditionally, plants have served as a source of novel drugs for treating a variety of diseases in humans (Cowan, 1999). A variety of plant-derived compounds possessing antimicrobial properties against a wide range of microorganisms have been documented (Kon and Rai, 2012; Upadhyay et al., 2014). The antimicrobial effects of four components of ginger against MDR strains of A. baumannii has been reported (Wang et al., 2010). Similarly, the essential oil from coriander was found to exert either synergistic or additive effects with antibiotics such as tetracycline, chloramphenicol, ciprofloxacin, gentamicin, piperacillin, and cefoperazone against A. baumannii (Duarte et al., 2012). Recently, aqueous extract of kiwi (Actinidia deliciosa) and clove (Syzygium aromaticum) were found to exert antibiofilm activity against A. baumannii (Tiwari et al., 2017). Similarly, it was shown that plant extract from Aegle marmelos and imipenem had synergistic effect against carbapenem resistant strain of Acinetobacter baumannii (Tiwari et al., 2016). In another study, the essential oil from Origanum vulgare possessed potent antimicrobial activity against MDR A. baumannii (Saghi et al., 2015). In addition, previous research from our laboratory indicated that several plant-derived antimicrobials (PDAs), including trans-cinnamaldehyde (TC), an ingredient in cinnamon, eugenol (present in clove) and carvacrol and thymol obtained from oregano oil and oil of thyme, respectively, decreased antibiotic resistance in MDR S. Typhimurium DT 104 (Johny et al., 2010). These investigators observed that TC reduced DT 104's resistance to five antibiotics, where thymol and carvacrol decreased resistance to three antibiotics.

The β-lactam group of antibiotics are the most commonly prescribed antibiotics for the treatment of bacterial infections worldwide (Pitout et al., 1997; Thakuria and Lahon, 2013). A. baumannii is capable of producing β-lactamases that can hydrolyze the β-lactam ring of penicillins, cephalosporins, and carbapenems, thereby conferring resistance to these antibiotics. Therefore, this study investigated the efficacy of TC and eugenol (EG) in increasing the sensitivity of A. baumannii to seven β-lactam antibiotics. In addition, the effect of these PDAs on genes conferring resistance to β-lactam antibiotics in A. baumannii was determined.

### MATERIALS AND METHODS

## A. baumannii Cultures and Growth Conditions

Two clinical isolates of A. baumannii, namely 251847 (International Health Management Associates, IL), and 17978 (ATCC; fatal meningitis isolate) were used in the study. All bacteriological media used in the study, except Leeds MDR Acinetobacter agar, were purchased from Difco (Becton Dickinson, Sparks, MD, United States). Leeds MDR agar was procured from Hardy Diagnostics (Santa Maria, CA, United States). The bacterial isolates were cultured separately overnight in 10 ml tryptic soy broth (TSB), followed by streaking on Leeds MDR Acinetobacter agar plates and incubation at 37◦C for 24 h. An individual colony from Leeds MDR Acinetobacter agar was sub-cultured twice in 10 ml of TSB at 37◦C for 24 h with agitation to reach ∼8 log<sup>10</sup> CFU/ml. The cultures were sedimented by centrifugation (3,700 × g, 15 min, 4◦C), and the pellet was washed twice and re-suspended in sterile phosphate buffered saline (PBS; pH 7.2). The cultures were then diluted appropriately in PBS to obtain ∼5 to 6 log<sup>10</sup> CFU/ml to be used as the inoculum. The bacterial population in the inoculum was confirmed by plating on tryptic soy agar (TSA) with incubation at 37◦C for 24 h.

## Plant-Derived Antimicrobials (PDAs) and Chemicals

Trans-cinnamaldehyde (≥98%; TC, trans-3-phenyl-2 propenal), eugenol (≥98%; EG,4-allyl-2-methoxyphenol), 1-N-phenylnaphthylamine (NPN), EDTA, ethidium bromide (EtBr), carbonyl cyanide m-chlorophenylhydrazone (CCP) and pyronin Y were purchased from Sigma-Aldrich (St. Louis, MO, United States).

## Determination of Sub-inhibitory Concentration (SIC) and Minimum Inhibitory Concentration (MIC) of PDAs Against A. baumannii

The sub-inhibitory concentration (SIC) and minimum inhibitory concentration (MIC) of TC and EG against A. baumannii were determined as previously reported (Johny et al., 2010; Amalaradjou and Venkitanarayanan, 2011) TSB (10 ml) tubes containing 0.75–7.5 µM (TC) and 0.61–6.1 µM (EG) in increments of 0.375 µM (TC) and 0.305 µM (EG were inoculated separately with A. baumannii at ∼ 6 log<sup>10</sup> CFU/ml, and incubated at 37◦C for 24 h. Tubes without any added PDAs served as controls. After incubation, the samples were serially diluted (1:10) in PBS, plated on TSA, and incubated at 37◦C for 24 h before counting the colonies. The highest concentration of TC or EG that did not inhibit A. baumannii growth after 24 h of incubation was selected as the SIC, while the lowest concentration of the antimicrobial that inhibited visible growth of the bacteria after 24 h incubation was taken as the MIC of that treatment. The experiment was done in duplicates and repeated three times.

## Effect of PDAs on Antibiotic Resistance in A. baumannii

The β-lactam antibiotics tested in the current study included Ampicillin, Meropenem, Methicillin, Penicillin, Aztreonam, Amoxicillin, and Piperacillin (Sigma-Aldrich). Previously published MIC of each aforementioned antibiotic against A. baumannii was used in the study (Vashist et al., 2011; Morfin-Otero and Dowzicky, 2012; Malone and Kwon, 2013) (**Table 1**). To determine the effect of combination of PDAs and antibiotics on A. baumannii, the MIC of each antibiotic and that of TC/EG were added to 10 ml TSB containing A. baumannii (∼ 6 log<sup>10</sup> CFU/ml), and incubated at 37◦C for 24 h (Johny et al., 2010). The bacterial counts were determined after broth dilution assay and

TABLE 1 | MIC of antibiotics used in testing against A. baumannii.


surface plating of appropriate dilutions on TSA. The treatments included only A. baumannii (positive control), A. baumannii + antibiotic, A. baumannii + PDA and A. baumannii + antibiotic + PDA. In addition, suitable controls, including A. baumannii + diluent (ethanol) and A. baumannii + ethanol + antibiotic were also included. Duplicate samples were included for each treatment, and the experiment was replicated three times.

### Efflux Pump Inhibition Assay

To study the effect of TC and EG on inhibiting the action of efflux pumps in A. baumannii, EtBr and pyronin Y efflux assays were performed according to a published protocol (Chusri et al., 2009). Overnight cultures of A. baumannii were washed twice and resuspended in PBS containing 0.4% glucose to an OD<sup>600</sup> of ∼0.5. The bacterial suspension was added with the MIC of TC/EG or CCP (positive control, 100 µM), and incubated at 37◦C for 5 h. A. baumannii suspension in PBS + 0.4% glucose served as control. After incubation, 200 µl of the treatments/control was separately transferred to a 96-well microtiter plate, followed by addition of EtBr (Sigma) to a final concentration of 4 mg/l, and the fluorescence was measured at excitation 530 nm and emission 645 nm. The assay was repeated with pyronin Y (Sigma) at a final concentration of 5 mg/l at 530 nm and emission 645 nm. The experiment was repeated three times with duplicates for A. baumannii 17978 and A. baumannii 251847.

### Outer Membrane Permeabilization Assay

To study the effect of TC and EG on the outer membrane of A. baumannii, NPN uptake assay was performed using a published protocol (Chusri et al., 2009). Overnight A. baumannii 17978 and A. baumannii 251847 cultures were separately washed and resuspended in 5 mM HEPES buffer to an OD<sup>600</sup> ∼ 0.5. Aliquots of 100 µl of A. baumannii suspension were transferred to a microtiter plate along with the MIC of TC/EG or EDTA 1 mM/HEPES buffer. This was followed by addition of 40 µM of NPN to make the total volume to 200 µl, the fluorescence was measured within 3 min at excitation 355 nm and emission 460 nm, and continuously recorded for 3 h every 10 min. The experiment was repeated three times with duplicates in each treatment.

### Effect of PDAs on Antibiotic Resistance Gene Expression in A. baumannii RNA Extraction and cDNA Synthesis

To study the effect of TC and EG on genes associated with resistance to β-lactam antibiotics in A. baumannii, bacterial cultures were grown separately with or without the SIC of TC/EG at 37◦C in TSB to mid-log phase. A. baumannii grown with either antibiotic alone or PDA alone were also included as controls. After incubation, the cultures were subjected to centrifugation (12,000 × g, 15 min, 4◦C) and the resultant pellet was added with 0.5 ml of RNAse free, sterile water and 1 ml of RNA protect reagent (Qiagen, Valencia, CA, United States). The total RNA from each sample was extracted using the RNeasy mini kit (Qiagen), and the manufacturer's instructions were followed

in estimating the total RNA using Nanodrop (Thermo Fisher Scientific, Waltham, MA, United States). Super-script II reverse transcriptase kit (Invitrogen, Carlsbad, CA, United States) was used for cDNA synthesis, and the resultant cDNA was used as a template for RT-qPCR. The amplified product was detected using SYBR Green reagents.

### Real-Time Quantitative PCR (RT-qPCR)

The antibiotic resistance genes of A. baumannii assayed for expression analysis included efflux pump genes, namely adeA, adeB, adeC; β-lactam resistance gene, blaP; and the multidrug resistance protein gene, mdrp. Primer Express software <sup>R</sup> (Applied Biosystems, Foster city, CA, United States) was used for designing the primers specific for the genes and for the endogenous control (16S rRNA). The primers were designed from A. baumannii AB0057 genome (CP001182.1) published in the NCBI database (Adams et al., 2008) and their sequences are provided in **Table 2**. The custom synthesized primers were obtained from Integrated DNA Technologies (Foster City, CA, United States). RT-qPCR was performed with StepOnePlusTM Real Time PCR system (Applied Biosystems) using the SYBR green assay (Applied Biosystems) under custom thermal cycling conditions with the normalized RNA as the template (Bookout and Mangelsdorf, 2003). Duplicate samples were analyzed and standardized against 16S rRNA gene expression. The relative changes in mRNA expression levels were determined using comparative threshold cycle (CT) method (2−11C<sup>T</sup> ) between the control and the treatments.

### Statistical Analysis

A completely randomized design with a factorial treatment structure was used. The factors included two A. baumannii strains, seven antibiotics, and two PDAs. The data were analyzed using the PROC GENMOD procedure of Statistical Analysis Software (SAS ver. 9.4; SAS Institute Inc., Cary, NC, United States). Least square means were considered significant at P < 0.05. The data comparisons for gene expression study

TABLE 2 | List of primers used in detecting A. baumannii antibiotic resistance genes.


F, forward; R, reverse.

were made using Students t-test. The difference was considered significant at P < 0.05.

## RESULTS

The SICs of TC and EG against A. baumannii were found to be 1.1 mM (0.015%) and 1.8 mM (0.03%), respectively, while the MIC was 4 mM (0.05% TC and 0.065% EG) for both PDAs. Although a previously reported MIC of each antibiotic against A. baumannii was used in the study, both A. baumannii isolates grew ∼ by 2.0 log<sup>10</sup> CFU/ml after 24 h (antibiotic control), as observed in **Figures 1**, **2** and Supplementary Figures 1, 2. In the presence of the MIC of TC, the bacterial count after 24 h did not change significantly from the inoculation of level of 6.0 log<sup>10</sup> CFU/ml (**Figure 1**) as expected. However, when A. baumannii was grown in the presence of each antibiotic and TC, its growth after 24 h was significantly decreased in comparison to that in the positive control, antibiotic control and TC control (P < 0.05). In both A. baumannii isolates, the greatest sensitivity was observed to methicillin and lowest sensitivity was seen against ampicillin and meropenem (**Figure 1** and Supplementary Figure 1). Similarly, the sensitivity of both A. baumannii isolates to all seven antibiotics was significantly increased in the presence of EG, as indicated by the reduced growth after 24 h (P < 0.05) (**Figure 2** and Supplementary Figure 2). In A. baumannii 17978, the greatest sensitivity was observed against ampicillin (**Figure 2**), whereas A. baumannii 251847 was maximally sensitive to amoxicillin (Supplementary Figure 2).

The results of the efflux pump inhibition assay using EtBr and pyronin Y in A. baumannii ATCC 17978 and A. baumannii 251847 are presented in **Figures 3**, **4** and Supplementary Figures 3, 4. EtBr and pyronin Y are known substrates for AdeABC and AdeIJK efflux pumps, respectively (Magnet et al., 2001; Xing et al., 2014). The MIC of TC and EG resulted in an increased accumulation of EtBr inside bacterial cells, as indicated by an increase in fluorescence compared to the PBS control (P < 0.05) (**Figures 5A–G**). CCP, an efflux pump inhibitor used as a positive control (Magnet et al., 2001) also resulted in an increase in fluorescence indicating suppression of efflux pump in A. baumannii. However, a similar increase in the fluorescence

exposed to any treatments (PC) and bacteria exposed to only TC (TC Ctrl) served as controls for the experiment.

FIGURE 2 | Effect of EG in combination with β-lactam antibiotics in A. baumannii 17978. Bars with different superscripts differ from each other within a cluster (P < 0.05). A. baumannii 17978 was grown with each β-lactam antibiotic either alone or in combination with EG. Bacteria not exposed to any treatments (PC) and bacteria exposed to only EG (EG Ctrl) served as controls for the experiment.

was not observed for pyronin Y compared to the PBS control (**Figure 4** and Supplementary Figure 4). The results of NPN uptake assay in A. baumannii ATCC 17978 and A. baumannii 251847 are presented in Supplementary Figures 5, 6, respectively. Neither TC nor EG increased NPN uptake, while the EDTA control did show an increase in fluorescence. Further, the samples treated with TC or EG demonstrated a decrease in fluorescence over time (data not shown).

The effect of seven antibiotics and two PDAs and their combination on the expression of various antibiotic resistance genes in A. baumannii 17978 is depicted in **Figures 5A–G**. It was observed that compared to control, the expression of all tested genes was up-regulated (P < 0.05) following A. baumannii growth in the presence of the antibiotics. However, TC significantly down-regulated the expression of major genes conferring resistance to β-lactam antibiotics compared to control (P < 0.05). The genes encoding efflux pump adeA and adeB were down regulated by ∼6- and 9-fold, respectively. Moreover, the expression of genes encoding β-lactamase (blaP) and the multiple drug resistance protein (mdrp) was decreased by ∼3 fold (P < 0.05). Similar to the results observed with TC, all antibiotic resistance genes were down-regulated on exposure

to EG compared to control (P < 0.05). The expression of genes encoding efflux pumps, adeA and adeB were downexpressed by 3- and 14-fold, respectively. Similarly, blaP and mdrp were also down-regulated in both strains (P < 0.05). The combination of TC or EG with the antibiotics also resulted in a down-regulation of majority of the tested genes (P < 0.05). Among the combinations of TC or EG with the seven antibiotics, the combination containing aztreonam resulted in a significant down-regulation of all the tested genes compared to control (**Figure 5D**). The combination of EG with piperacillin (**Figure 5C**) or penicillin (**Figure 5F**) also significantly reduced the expression of all the antibiotic resistance genes (P < 0.05).

### DISCUSSION

Rapid emergence of antibiotic resistance in pathogenic microorganisms, especially to multiple antibiotics has ignited research efforts to discover novel antibiotics and develop effective derivatives of currently available antibiotics. However, no promising antibiotics are under development, and the rapidity and complexity of resistance development in pathogens have further exacerbated the situation. In light of this, a potential viable approach was explored to reduce bacterial antibiotic resistance in the development of inhibitors of resistance mechanisms in bacteria (Renau et al., 1999; Walsh and Fanning, 2008). This strategy involves the co-administration of an antibiotic with an "inhibitor," which counteracts bacterial resistance mechanism(s), thereby rendering the resistant pathogen sensitive to the drug. The advantage of this approach is that it makes it possible to continue the use of current antibiotics, for which in-depth pharmacological and toxicological data are already available. In this regard, PDAs represent a potential natural group of "inhibitors" of bacterial antibiotic resistance mechanisms.

Abundant literature exists on the antimicrobial properties of a variety of plant compounds against a wide range of microorganisms (Burt, 2004; Upadhyay et al., 2014). However, only a handful of studies have addressed their effect on bacterial

FIGURE 5 | Effect of TC and EG on antibiotic resistance genes in A. baumannii 17978. (A) A. baumannii 17978 was grown with the SIC of TC/EG either alone or in combination with amoxicillin, and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to amoxicillin alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (B) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with ampicillin and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to ampicillin alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (C) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with piperacillin and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to piperacillin alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (D) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with aztreonam and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to aztreonam alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (E) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with meropenem and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to meropenem alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (F) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with penicillin and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to penicillin alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05). (G) A. baumannii 17978 was treated with SICs of TC/EG either alone or in combination with methicillin and RT-qPCR was done to test the effect of the treatments on major antibiotic genes in A. baumannii. Bacteria exposed to methicillin alone and bacteria not exposed to any treatments served as controls. Bars with <sup>∗</sup> are significantly different from control (P < 0.05).

antibiotic resistance, especially in Gram negative bacteria (Gallucci et al., 2006; Chusri et al., 2009; Johny et al., 2010; Ilic´ et al., 2014). In the current study, both TC and EG enhanced the sensitivity of A. baumannii to all seven β-lactam antibiotics tested (**Figures 1**, **2** and Supplementary Figures 1, 2). This is evident from the significant reductions in bacterial counts observed in the samples containing PDA and antibiotics as compared to that in the treatments containing each PDA or antibiotic alone. A checkerboard assay testing a wide range of concentrations of two antimicrobials is commonly used for determining their combinatorial effects on a bacterium (Jayamani et al., 2017; Tharmalingam et al., 2018). However, since our preliminary studies revealed that concentrations of TC and EG below the MIC were not significantly effective in increasing A. baumannii sensitivity to antibiotics, we did not use a checkerboard assay, but instead tested the combinatorial effect of MIC of plant molecules and antibiotics on the pathogen.

A recent study also reported synergistic effect of Eucalyptus camaldulensis with polymyxin B against MDR Acinetobacter baumannii (Knezevic et al., 2016). In another study, Enrofloxacin and cinnamon were observed to possess synergistic effect against Salmonella enterica (Solarte et al., 2017). Further, extracts of medicinal plant Holarrhena antidysenterica with Conessine increased susceptibility of extensively drug resistant A. baumannii to novobiocin and rifampicin (Siriyong et al., 2016). Additionally, no cytotoxic effects on human epithelial cells at the tested concentrations were reported previously with TC and EG (Amalaradjou et al., 2010; Karumathil et al., 2016). Moreover our previous studies revealed that in-feed supplementation of TC and EG in broiler chicks (Kollanoor-Johny et al., 2012) and TC in mice (Narayanan et al., 2017) for 10 days did not result in any toxicity.

A. baumannii has been reported to exhibit resistance to β-lactam antibiotics through several mechanisms, including the production of β-lactamases, changes in penicillin-binding proteins (PBPs), altering the structure and number of porin proteins, decreased membrane permeability, and by use of efflux pumps that exit antibiotics out of the bacterial cell (Vila et al., 2007; Manchanda et al., 2010; Bonnin et al., 2013; Tang et al., 2014). Moreover, the presence of efflux pumps and MDR proteins in A. baumannii contribute significantly to both intrinsic and acquired resistance to antibiotics (Lomovskaya and Bostian, 2006). A. baumannii genome encodes a wide array of multidrug efflux systems, including AdeABC, a resistancenodulation-division (RND) family-type pump (Damier-Piolle et al., 2008; Wieczorek et al., 2008; Yoon et al., 2013; Sun et al., 2014). The substrates for this pump include beta-lactams, aminoglycosides, erythromycin, chloramphenicol, tetracycline, fluoroquinolone, trimethoprim, and EtBr (Magnet et al., 2001; Higgins et al., 2004; Heritier et al., 2005; Peleg et al., 2007). The three component structures of AdeABC include the inner membrane fusion protein (AdeA), transmembrane component (AdeB) and an outer membrane protein (AdeC), with the inactivation of adeB resulting in the loss of pump function and multidrug resistance (Magnet et al., 2001).

In order to determine if TC or EG exerted an inhibitory effect on the aforementioned efflux pumps in A. baumannii, an efflux pump inhibition assay was performed with EtBr and pyronin Y along with CCP, a documented efflux pump inhibitor (Magnet et al., 2001; Chusri et al., 2009). EtBr and pyronin Y are known substrates for AdeABC and AdeIJK efflux pumps, respectively (Magnet et al., 2001; Damier-Piolle et al., 2008; Chusri et al., 2009; Xing et al., 2014). Both TC and EG resulted in the inhibition of AdeABC efflux pump in the two tested A. baumannii strains, as evident from the increase in fluorescence in the treated samples compared to PBS control (**Figure 3** and Supplementary Figure 3). However, the PDAs did not exert any inhibitory effect on the action of AdeIJK efflux pump in A. baumannii, as evident from the lack of difference in fluorescence units between PDA-treated and PBS control samples (**Figure 4** and Supplementary Figure 4). Similarly, the results from the NPN uptake assay revealed that TC and EG did not increase A. baumannii's outer membrane permeability, as seen from Supplementary Figures 5, 6, where no increase in fluorescence was observed in PDA-treated A. baumannii as against the samples treated with EDTA, a known outer membrane permeabilizer in Gram-negative bacteria (Helander and Mattila-Sandholm, 2000; Alakomi et al., 2006). These results indicate that TC and EG increased the sensitivity of A. baumannii to the tested antibiotics at least in part by inhibiting the efflux pump, AdeABC.

For ascertaining if TC or EG exerted an inhibitory effect on any of the antibiotic resistance genes conferring resistance to β-lactam antibiotics in A. baumannii, we performed a RT-qPCR on mRNA extracted from A. baumannii following growth in the presence and absence of the PDAs. The results from the RT-qPCR revealed that TC and EG in combination with or without each antibiotic significantly down-regulated the expression of the majority of genes that confer resistance to β-lactam antibiotics (**Figures 5A–G**). Among the various genes screened, those encoding efflux pumps, adeA and adeB were maximally down-regulated by both PDAs. These results concur with the results from the EtBr efflux pump inhibition assay, and suggest that TC and EG enhanced the efficacy of the seven β-lactam antibiotics against A. baumannii by thwarting the various resistance mechanisms, especially those involving the efflux pumps.

### CONCLUSION

The results of this study suggest the potential use of TC and EG in conjunction with the currently available β-lactam antibiotics for the treatment for MDR A. baumannii infections and could lead to development of new treatment options with reduced antibiotic dosage. However, efficacy studies in suitable animal models are warranted before recommending their clinical usage.

## AUTHOR CONTRIBUTIONS

KV is the corresponding author and primary contact during manuscript submission, review, and publication process. The work was done under his supervision as the principal investigator. He significantly contributed to the design, drafting,

revisions, and interpretation of data. The manuscript is being submitted with his final approval for publication. MN is the submitting author. DK and AK-J designed the study. DK and JG conducted the experiments. DK and MN prepared the manuscript for submission.

### REFERENCES


### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.01011/full#supplementary-material



literature. Scand. J. Infect. Dis. 37, 919–921. doi: 10.1080/003655405002 62567



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Karumathil, Nair, Gaffney, Kollanoor-Johny and Venkitanarayanan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Advance in Research on Mycobacterium tuberculosis FabG4 and Its Inhibitor

### Debajyoti Dutta\*

Department of Biochemistry, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, AB, Canada

Increasing evidence from recent reports of drug-resistant mycobacterial strains poses a challenge worldwide. Drug-resistant strains often undergo mutations, adopt alternative pathways, and express drug efflux pumps to reduce or eliminate drug doses. Besides these intrinsic resistance mechanisms, bacteria can evade drug doses by forming biofilms. Biofilms are the concerted growth of adherent microorganisms, which can also be formed at the air-water interface. The growth is supported by the extracellular polymer matrix which is self-produced by the microorganisms. Reduced metabolic activity in a nutrient-deficient environment in the biofilm may cause the microorganisms to take alternative pathways that can make the microorganisms recalcitrant to the drug doses. Recent works have shown that Mycobacterium tuberculosis expresses several proteins during its growth in biofilm, those when deleted, did not show any effect on mycobacterial growth in normal nutrient-sufficient conditions. Studying these unconventional proteins in mycobacterial biofilms is therefore of utmost importance. In this article, I will discuss one such mycobacterial biofilm-related protein FabG4 that is recently shown to be important for mycobacterial survival in the presence of antibiotic stressors and limited nutrient condition. In an attempt to find more effective FabG4 inhibitors and its importance in biofilm forming M. tuberculosis, present knowledge about FabG4 and its known inhibitors are discussed. Based on the existing data, a putative role of FabG4 is also suggested.

Keywords: FabG4, Mycobacterium tuberculosis, biofilm, β-oxo acyl-ACP reductase, inhibitor

## INTRODUCTION

The ability of Mycobacterium tuberculosis to form biofilms was noted almost 120 years ago (Jones, 1896). However, the physiological and molecular basis of biofilm is only beginning to unravel until recently. The first protein that was found to be involved in M. smegmatis maturation of biofilm is the chaperone GroEL-1 (Ojha et al., 2005). GroEL-1 deficient mutant of M. smegmatis was suggested to lack GroEL-1 interaction with fatty acid synthesis type -II complex thereby reducing the mycolic acid content in biofilms. Mycolic acids are the predominant components of mycobacterial cell envelope that are produced by fatty acid synthesis type-II pathway in mycobacteria. The reason that mycobacterial cell envelope largely contributes to the biofilm attachment endorses mycolic acids may be involved in biofilm formation (Marrakchi et al., 2014). However the physiology of mycobacteria changes as it shifts from planktonic growth to biofilm

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Luiz Augusto Basso, Pontifícia Universidade Católica do Rio Grande do Sul, Brazil Li Yuan, Shihezi University, China

\*Correspondence:

Debajyoti Dutta ddutta@ualberta.ca; debajyoti.47@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 26 March 2018 Accepted: 16 May 2018 Published: 06 June 2018

### Citation:

Dutta D (2018) Advance in Research on Mycobacterium tuberculosis FabG4 and Its Inhibitor. Front. Microbiol. 9:1184. doi: 10.3389/fmicb.2018.01184

**137**

dependent growth and results in several modifications in the expression level of protein and molecules pertaining to the cell envelope (Ojha et al., 2008; Sambandan et al., 2013; Rastogi et al., 2017). Fatty acid synthesis and its associated pathways for mycobacterial cell envelope synthesis are one of the major areas for developing antitubercular drugs (Zumla et al., 2013). Because of this altered phenotypes and a waxy extra-cellular matrix of biofilm, Mycobacteria become resilient to drug doses (Islam et al., 2012). Therefore the conventional TB drugs may not be as effective for biofilm-forming mycobacteria.

Membrane and cell envelope-associated biofilm-related proteins are particularly of interest as these proteins are likely to be involved in the making of cellular attachment to the biofilms. For example, recent studies have shown that a lipid transporter MmpL11 is specifically required for biofilm formation (Pacheco et al., 2013). Other works have identified a number of proteins specific to mycobacterial growth in the biofilm at air-water interface (Kerns et al., 2014). One of the proteins that are conserved among mycobacterial species is FabG4. The protein was proposed to possess host antigenic property and has a potential to be a biofilm-specific marker (Kerns et al., 2014). In addition to that, FabG4 was recognized as one of the crucial protein for mycobacterial survival in a stressed condition. This article will discuss the known facts about FabG4, its inhibitors, and discuss its possibility to serve as a candidate to study and treat biofilm-related mycobacteria.

### IMPORTANCE OF FABG4 IN MYCOBACTERIA

FabG commits the second step of fatty acid synthesis that is to convert β-oxo acyl-ACP to β-hydroxy acyl-ACP. M. tuberculosis genome contains multiple FabG homologs. Two of them are conserved among all mycobacterial species, FabG1 and FabG4. FabG1 remain at the focus of attention because it takes part in fatty acid synthesis type-II (Marrakchi et al., 2002). On the other hand, several studies have indicated that FabG4 is not an inactive gene in the M. tuberculosis genome. Its expression was first documented in the proteome by using 2-D gel electrophoresis accompanying MALDI-MS analysis (Jungblut et al., 1999) and later verified by others (Rosenkrands et al., 2000; Sinha et al., 2002). Gu et al. (2003), first provide the evidence that the protein is expressed in the mycobacterial membrane fraction. However, it's requirement in mycobacterial physiology was not shown until the comprehensive work done by McFadden and coworkers (Beste et al., 2009). The authors showed that the protein is uniquely required for mycobacterial growth in Roisin's minimal media, which contains limited carbon source (Beste et al., 2009). Proteomics studies have further identified that FabG4 is one of the major proteins which expression is induced by the antibiotic stressor (Sharma et al., 2010). The protein expression was also detected in the drug-resistant strains (Singh et al., 2016; Verma et al., 2017). Most interestingly, FabG4 can functionally complement eukaryotic β-oxoacyl-ACP reductase activity. FabG4, when expressed in the knockout of ora1 yeast cells can restore the mitochondrial fatty acid synthesis type-II (Gurvitz et al., 2009). Because yeast ORA1 is a mitochondrial FabG1/FabG4 ortholog (Hiltunen et al., 2009), functional complementation of ORA1 with FabG4 is indicative of an active FabG4 protein that can utilize ACP/Coenzyme A tagged β-oxoacyl substrates and is utilized by mycobacteria during limiting resource or antibiotic stressed condition.

### UNIQUE FEATURES OF FabG4

Characterization of recombinantly expressed FabG4 has shown that the protein is highly specific for NADH (Dutta et al., 2011). This is intriguingly different from its close relative (33% homology) FabG1 that utilizes NADPH. Crystal structure of the FabG4 complex with NADH reveals that the specificity comes from a single aspartate residue that reduces the volume of the phosphate binding groove of adenosine ribose phosphate moiety of NADPH (Dutta et al., 2013). Crystal structure further shows that FabG4 contains a "flavodoxin-type" structural domain in its N-terminal and a typical ketoreductase domain in its C-terminal (**Figure 1A**). The N-terminal structural domain is also found in modular polyketide synthase ketoreductase domain PlmKR1 (Bonnett et al., 2013). PlmKR1 exists as a monomer but FabG4 is different in the sense that the protein is a dimer, where the N-terminal domain of a protomer makes an extensive interaction with the C-terminal domain of the other protomer. Therefore, FabG4 was designated as "High Molecular weight FabG" (HMwFabG) (Dutta et al., 2011). Typical FabG does not contain this structural domain.

Many of the N-terminal residues (1–19) of FabG4 cannot be traced in crystal structure suggesting that the region has no rigid secondary structure. Most of the N-terminal residues were traced in FabG4 complex with NAD<sup>+</sup> structure (PDB 4FW8) showing a possibility of short helix containing seven residues (21–27) (**Figure 1A**). Long N-terminal sequences in FabGs are also found among eukaryotes, which were suggested to be the signal sequence (Wickramasinghe et al., 2006). Whether it is true for FabG4 is not known, but sequence alignment of HMwFabGs from actinobacteria and many proteobacteria reveals conserved residues Pro30, Leu33, and Arg35 in N-terminal sequence. Truncation of the N-terminal residues does not have any effect on catalytic activity. On the contrary, the C-terminal residues of FabG4 are apparently involved in catalytic activity since its truncation yields with a defective protein (Dutta et al., 2011). This is because the conserved C-terminal residues are engaged in electrostatic interaction with the active site proximal loops (Dutta et al., 2011).

### INHIBITORS OF FabG4

Structure of FabG4 and its complex with a substrate mimic hexanoyl-CoA (PDB 3V1U) accelerated the work on structurebased inhibitor design against FabG4 (Dutta et al., 2013) (**Figure 1B**). The particular structure also provides a platform to design FabG specific inhibitors because the structure provided the first evidence of FabG-substrate complex (**Table 1**). The

first reported FabG4 inhibitors are based on the polyphenol compounds (Banerjee et al., 2014, 2015b). Synthesized compounds are also tested against M. smegmatis showing the MIC of 5 µg mL−<sup>1</sup> . Docking studies had revealed that the compounds potentially occupy the NADH binding region with a few hydrogen-bonding interactions with the loop-residues responsible for CoA substrate binding. Another study to design and synthesize the inhibitors based on the common pharmacophores like β-lactam and Isoniazid was also successful inhibiting FabG4 with IC<sup>50</sup> as low as 15.2 ± 0.5 µM (Banerjee et al., 2015a). Similar to the polyphenol based compounds β-lactam and Isoniazid-based compounds primarily dock on the NADH binding site. Isothermal titration calorimetry indicates

coenzyme A (HXC) and NAD<sup>+</sup> bound structure shows the putative substrate

two sequential binding sites showing positive cooperativity. Inhibitory effects of β-lactam and Isoniazid-based compounds have shown to be inhibitory toward mycobacterial biofilm formation. Recently, thiophene containing trisubstituted methane compound, S-S006-830 was found to have a substantial inhibitory effect on M. tuberculosis biofilm formation and possesses antitubercular activity (Singh et al., 2017). FabG4 is one of the three membrane-associated proteins identified to strongly interact with S-S006-830. Computational docking study predicts two possible binding sites. One comprises of the region overlapping NADH and CoA substrate binding and the other site is on the N-terminal domain. S-S006-830 is special because apart from the NADH-substrate binding site it also targets to the N-terminal structural domain. Two unique FabG inhibitors are worth mentioning in this regard. The first, Pyridomycin analog specifically targets NADH binding proteins and proposed to exhibit inhibitory effects on FabG4 by bridging both NADH and substrate-binding region (Hartkoorn et al., 2014). The second inhibitor was identified using library screening that uniquely binds to the helical dimeric interface of FabG (Cukier et al., 2013). The helical dimeric interface in FabG is responsible of NAD(P)H binding cooperativity (Dutta et al., 2012). All FabG4 and related inhibitors are summarized in **Table 1**.

## POSSIBLE ROLE OF THE ENZYME

High Molecular weight FabG are exclusively found among bacteria. Homologous genes of FabG4 are found in actinobacteria and many proteobacteria (Dutta et al., 2011). Genomic analysis of HMwFabG containing bacteria shows another protein coexisting downstream to FabG4 ORF. In M. tuberculosis this protein is HtdX (Rv0241c). Both the proteins are conserved among mycobacterial species. Even in M. leprae genome where the genomic reduction had happened, retained both fabG4 and its downstream htdX (Cole et al., 2001). However, deletion of htdX or fabG4 does not show any apparent effect on bacterial survival in vivo (Sassetti and Rubin, 2003). Nonetheless, the protein is shown to be essential for survival in minimal media (Beste et al., 2009). Presumably, the conditional dependency of M. tuberculosis on FabG4 indicates an alternative pathway during limiting resource condition. Another clue is the FabG4 dependency on NADH. The NADH utilizing β-ketoacyl reductase activity of FabG4 is four times higher to the reverse reaction that utilizes NAD<sup>+</sup> for β-hydroxyacyl dehydrogenase activity (Dutta et al., 2013). Compared to NADPH, NADH is the low energy molecule and mostly associates to the catabolic pathways (Cantó et al., 2015). Shifting the coenzyme specificity toward low energy NADH possibly indicates that the bacteria are rewiring its metabolism toward the energy saving mode. Furthermore, FabG4 is isolated from the membrane fraction, which is indicative of its membrane-associated role (Gu et al., 2003; Singh et al., 2017). The role could be either fatty acid synthesis or modification. Notably, the other gene htdX was found to be responsive to the inhibitors of cell wall synthesis (Boshoff et al., 2004). Since both fabG4 and htdX are situated in a conserved cluster it is tempting to predict that they are involved in the same

binding modes.


pathway. Both sequence analysis and preliminary enzymatic activities are corroborated that HtdX can commit a dehydratase reaction that is to convert β-hydroxyacyl-CoA to enoylacyl-CoA (Gurvitz, 2009; Sacco et al., 2010; Biswas et al., 2013; Banerjee et al., 2015a). The reaction is theoretically the successive step of FabG4 enzymatic reaction for fatty acid elongation cycle. Altogether the data suggests that during limiting resource condition FabG4 participates into some low-energy requiring pathway that is related to membrane biosynthesis, remodeling or recycling.

### FUTURE SCOPE

FabG4 is a host antigenic protein reportedly expressed in mycobacterial biofilms. The protein expression was consistently found in membrane fraction and stressor-induced. As a conserved gene in all mycobacterial species, FabG4 can be a biomarker for biofilm. Although, FabG4 inhibitors are manifested antitubercular activity more experiments are needed to call it a drug target. Role of FabG4 and its connection with the membrane in mycobacteria also require further experimentations. To study FabG4 importance in biofilm forming M. tuberculosis it is, therefore, necessary to find out its targeting and interaction with other proteins. Identifying the role of FabG4 in mycobacteria might provide a hint of a new aspect of a survival strategy. The second aspect is the strategy to design more specific inhibitors of FabG4. While the C-terminal domain of FabG4 is common among FabGs, inhibitors against its N-terminal structural domain might be a way to target the protein more specifically. Designing the fluorophore inhibitors against its structural domain could be a way to find out its role in vivo.

## AUTHOR CONTRIBUTIONS

DD is solely responsible for the conception or design, drafting, and revising the work.

### ACKNOWLEDGMENTS

fmicb-09-01184 June 4, 2018 Time: 14:18 # 5

The author thanks Professor Amit Kumar Das, Department of Biotechnology, Indian Institute of Technology Kharagpur for his guidance during Ph.D. The author also thanks

### REFERENCES


Professor Amit Basak, Department of Chemistry, Indian Institute of Technology Kharagpur and Dr. Deb Ranjan Banerjee and Rupam Biswas for their contribution to the FabG4 work from Indian Institute of Technology Kharagpur.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Dutta. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fmicb-09-01184 June 4, 2018 Time: 14:18 # 6

# Herring Oil and Omega Fatty Acids Inhibit Staphylococcus aureus Biofilm Formation and Virulence

Yong-Guy Kim<sup>1</sup>† , Jin-Hyung Lee<sup>1</sup>† , Chaitany J. Raorane<sup>1</sup> , Seong T. Oh<sup>2</sup> , Jae G. Park<sup>3</sup> \* and Jintae Lee<sup>1</sup> \*

<sup>1</sup> School of Chemical Engineering, Yeungnam University, Gyeongsan, South Korea, <sup>2</sup> College of Pharmacy, Yeungnam University, Gyeongsan, South Korea, <sup>3</sup> Advanced Bio Convergence Center, Pohang Technopark Foundation, Pohang, South Korea

### Edited by:

Kamlesh Jangid, National Centre for Cell Science (NCCS), India

### Reviewed by:

Nagendran Tharmalingam, Alpert Medical School, United States Dane Parker, Columbia University, United States

### \*Correspondence:

Jae G. Park jaepark@pohangtp.org Jintae Lee jtlee@ynu.ac.kr †These authors have contributed equally to this work.

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 20 February 2018 Accepted: 23 May 2018 Published: 15 June 2018

### Citation:

Kim Y-G, Lee J-H, Raorane CJ, Oh ST, Park JG and Lee J (2018) Herring Oil and Omega Fatty Acids Inhibit Staphylococcus aureus Biofilm Formation and Virulence. Front. Microbiol. 9:1241. doi: 10.3389/fmicb.2018.01241 Staphylococcus aureus is notorious for its ability to become resistant to antibiotics and biofilms play a critical role in antibiotic tolerance. S. aureus is also capable of secreting several exotoxins associated with the pathogenesis of sepsis and pneumonia. Thus, the objectives of the study were to examine S. aureus biofilm formation in vitro, and the effects of herring oil and its main components, omega fatty acids [cis-4,7,10,13,16,19-docosahexaenoic acid (DHA) and cis-5,8,11,14,17-eicosapentaenoic acid (EPA)], on virulence factor production and transcriptional changes in S. aureus. Herring oil decreased biofilm formation by two S. aureus strains. GC-MS analysis revealed the presence of several polyunsaturated fatty acids in herring oil, and of these, two omega-3 fatty acids, DHA and EPA, significantly inhibited S. aureus biofilm formation. In addition, herring oil, DHA, and EPA at 20 µg/ml significantly decreased the hemolytic effect of S. aureus on human red blood cells, and when pre-treated to S. aureus, the bacterium was more easily killed by human whole blood. Transcriptional analysis showed that herring oil, DHA, and EPA repressed the expression of the α-hemolysin hla gene. Furthermore, in a Caenorhabditis elegans nematode model, all three prolonged nematode survival in the presence of S. aureus. These findings suggest that herring oil, DHA, and EPA are potentially useful for controlling persistent S. aureus infection.

### Keywords: biofilm, hemolysis, herring oil, omega fatty acids, Staphylococcus aureus, virulence

## INTRODUCTION

The sticky conglomerations of bacteria that adhere to diverse medical devices or damaged body tissues can become causes of persistent infections. These bacteria encase themselves in a slime layer or biofilm, which poses serious problems to human health because of their abilities to tolerate conventional antibiotic chemotherapies, host immune systems, and external stresses (Potera, 1999; Hoffman et al., 2005).

Staphylococcus aureus is an important etiologic agent and a major cause of a diverse array of acute life-threatening bloodstream infections in man, and is often held responsible for worldwide outbreaks of nosocomial infections (Lowy, 1998). S. aureus biofilms play a critical role in antibiotic tolerance (Stewart and Costerton, 2001), and methicillin-resistant S. aureus (MRSA) and vancomycin-resistant S. aureus have become major nosocomial threats

(Speller et al., 1997). S. aureus secretes many exotoxins, including coagulase, enterotoxins, α-hemolysin, protein A, and TSST-1, which damage biological membranes and eventually cause cell death (Ohlsen et al., 1997; Otto, 2014). Also, S. aureus biofilms are found in medical devices and food surfaces, and are responsible for food-poisoning and toxic shock syndrome (Chambers and Deleo, 2009). In particular, α-hemolysin (Hla) is a major cytotoxic agent that has been associated with the pathogeneses of pneumonia, sepsis, and severe skin infections (Menzies and Kernodle, 1996; Wilke and Bubeck Wardenburg, 2010) and with biofilm formation (Caiazza and O'Toole, 2003). S. aureus biofilms also play a dominant role in the determination of disease severity and postoperative course (Singhal et al., 2011). Therefore, it is important to find means to inhibit biofilm formation and the virulent characteristics of S. aureus. For this purpose, novel and non-toxic compounds that prevent the development of drug tolerance are urgently required.

Semi-dried raw Pacific herring (Clupea pallasi), also known as Gwamegi (a Korean traditional food), is an important fishery product with a unique flavor, taste, and texture (Lee et al., 2002; Kang et al., 2011). Herring oil contains an abundance of two rare omega-3 polyunsaturated fatty acids (PUFAs), namely, docosahexaenoic acid (DHA; C22:6, ω-3) and eicosapentaenoic acid (EPA; C20:5, ω-3), which are commonly present at low concentrations in non-marine animals and are beneficial to the human body (Hirai et al., 1980; Schmidt and Dyerberg, 1994). Furthermore, it has been reported that two PUFAs act as bacteriocides on Gram-positive and Gram-negative bacteria, such as, Helicobacter pylori (Correia et al., 2012), Burkholderia cenocepacia (Mil-Homens et al., 2012), S. aureus (Desbois et al., 2009), Fusobacterium nucleatum, Porphyromonas gingivalis, and Streptococcus mutans (Huang and Ebersole, 2010; Sun et al., 2016, 2017). These two PUFAs have also been reported to possess significant anti-inflammatory, antitumorigenic (Bougnoux, 1999; Van Dyke, 2008), and antioxidant activities (Giordano and Visioli, 2014). Huang and Ebersole (2010), Sun et al. (2016, 2017) and Thibane et al. (2010) suggested that DHA and EPA could be considered as potential supplementary therapeutic agents due to their anti-biofilm activities on Candida species and periodontopathic bacteria. However, the abilities of DHA and EPA to inhibit biofilm formation and virulence production by S. aureus have not been assessed.

Therefore, the phenotypic effects of herring oil were studied and its active constituents were identified by gas chromatography-mass spectrometry (GC-MS). Two of its major constituents (DHA and EPA) were further investigated by confocal laser scanning microscopy (CSLM) and using a human blood assay to determine their effects on biofilm formation and toxin production by S. aureus. A Caenorhabditis elegans model was used to investigate the anti-virulent properties of herring oil, DHA, and EPA. qRT-PCR (quantitative realtime reverse transcription polymerase chain reaction) was used to investigate their effects on the transcriptional profiles of genes related to biofilm formation and virulence production in vitro.

## MATERIALS AND METHODS

## Ethics Statement

Human blood assays were authorized by the Ethical Committee of Yeungnam University (Gyeongsan, South Korea). The study was conducted according to the guidelines issued by the Ethical Committee of Yeungnam University. All blood donors provided written consent before blood collection.

## Bacterial Strains, Materials, and Growth Assay

A methicillin-sensitive S. aureusstrain (MSSA; ATCC 6538) and a methicillin-resistant S. aureus strain (MRSA; ATCC 33591) were used in the present study. All experiments were conducted at 37◦C in Luria-Bertani (LB) broth for the MSSA strain and in LB broth containing 0.2% glucose for the MRSA strain. Herring (Clupea pallasii) oil was prepared from semi-dried fish bodies without organs (so called Gwamegi, Korean traditional food). Herring was dried at open air condition for 20 days in winter at Pohang, Korea. The semi-dried herring (764 g) was further dried in a freezing drier for 3 days to gain 586 g of dried herring. Then it was chopped, and the oil was extracted by soxhlet apparatus that was refluxed with n-hexane for a day at 70◦C. After refluxing, the residual solution was collected and washed with 1% AcOH solution and brine to remove excess protein. Washed oil was dried over anhydrous MgSO<sup>4</sup> to gain clear yellow viscous oil (267 g, 45.6% yield). The omega fatty acids (cis-4,7,10,13,16,19-DHA,cis-5,8,11,14,17-EPA,cis-11-eicosenoic acid, and erucic acid), crystal violet, ethanol, ciprofloxacin, vancomycin and glucose were purchased from Sigma-Aldrich (St. Louis, MO, United States). For cell growth assays, optical densities were measured at 600 nm using a spectrophotometer (Optizen 2120UV, Mecasys, South Korea). To determine MIC (minimum inhibitory concentration), cells were inoculated into LB broth and cultured overnight at a dilution of 1:100 at 37◦C, and then incubated for 24 h in the presence of a test substance. After sequential dilutions, cultures were spread on LB agar plates, incubated for 24 h at 37◦C, and numbers of colonies were counted. Each experiment was performed using at least four independent cultures.

## Gas Chromatograph/Mass Spectroscopy (GC-MS)

The chemical composition of air-dried herring oil was determined by GC using an Agilent 6890N GC and SP-2560 (Supelco, Sigma-Aldrich, St. Louis, MO, United States) fused silica capillary column (100 m × 0.25 mm i.d., film thickness 0.25 µm). Capillary column details, temperature conditions, and the derivatization (methylation) of fatty acids were as previously described (Lee et al., 2017). In brief, helium was used as the carrier at 0.75 ml/min and the GC injector was held at 225◦C. The GC oven temperature was programmed as follows; 100◦C for 4 min, increased to 240◦C at 3◦C/min, and followed by 15 min at 250◦C. The split ratio was controlled at 200:1. Triundecanoin (C11:00) was used as the internal standard and quantifications were performed by integrating areas and correcting for fatty acid

methylation. Supelco 37 components FAME Mix (Supelco) were used as the reference standard.

### Crystal-Violet Biofilm Assay

The two bacterial strains (MSSA 6538, MRSA 33591) were subjected to a static biofilm formation assay on 96-well polystyrene plates (SPL Life Sciences, South Korea), as previously described (Kim et al., 2015). Briefly, cells were inoculated into LB broth (300 µl) at an initial turbidity of 0.05 at 600 nm and herring oil, cis-11-eicosenoic acid, DHA, EPA, or erucic acid were added at different concentrations and incubated for 24 h without shaking at 37◦C. To quantify biofilm formation, cell cultures were washed three times with water and then biofilms were stained with 0.1% crystal violet for 20 min. Biofilms were then dissolved in 300 µl of 95% ethanol and absorbances were measured at 570 nm (OD570). For the biofilm dispersion assay, S. aureus was cultured in 96-well plates for 24 h without shaking at 37◦C. Then, herring oil, DHA, or EPA was added to the cultures and incubated for another 10 h before the biofilm assay. Static biofilm formation results are the averages of four independent cultures of twelve replicate wells.

## Confocal Laser Scanning Microscopy and COMSTAT Analysis

Static biofilm formation by S. aureus (MSSA 6538) in 96-well plates (without shaking) in the presence or absence of herring oil, DHA, or EPA were assessed by confocal laser scanning microscopy (Nikon Eclipse Ti, Tokyo, Japan). Cells were stained with carboxyfluorescein diacetate succinimidyl ester (Invitrogen, Molecular Probes Inc., Eugene, OR, United States), which stains viable cells in biofilms, as previously reported (Lee et al., 2016). Stained S. aureus ATCC 6538 biofilms were visualized by confocal laser scanning microscopy using an Ar laser (excitation wavelength 488 nm, and emission wavelength 500–550 nm) and a 20× objective. Color confocal images were constructed using NIS-Elements C version 3.2 (Nikon eclipse) under the same conditions. For each experiment, at least 10 random positions in two independent cultures were observed, and 20 planar images were analyzed per position. To quantify biofilm formation, COMSTAT biofilm software (Heydorn et al., 2000) was used to measure biovolumes (µm<sup>3</sup> per µm<sup>2</sup> ), mean thicknesses (µm), and substratum coverages (%). Thresholds were fixed for all image stacks, and at least 4 positions and 20 planar images were analyzed per position.

### Hemolysis Assay

Human red blood cell hemolysis efficacies were assayed using whole cultures of S. aureus, as described previously (Lee et al., 2013; Tharmalingam et al., 2018). Briefly, S. aureus cells (MSSA 6538) were diluted 1:100 in LB broth (9 × 10<sup>9</sup> CFU/ml) with overnight culture and incubated with or without herring oil, DHA, or EPA for 20 h with shaking at 250 rpm. Separately,

human whole blood samples were centrifuged at 900g for 2 min and then washed three times with PBS and diluted in PBS (330 µl of red blood cells per 10 ml of PBS buffer). Bacterial culture (200 µl) was then added to 1 ml of diluted human red blood cells (3.3% in PBS). To determine hemolytic activities, mixtures of blood and S. aureus were incubated at 250 rpm for 2 h at 37◦C. Supernatants were collected by centrifugation at 16,600 g for 10 min and optical densities were measured at 543 nm.

### Whole Blood Bacterial Cell Killing Assay

The whole blood cell killing assay used was as previously described (Liu et al., 2008). Briefly, S. aureus (MSSA 6538) cells were inoculated (1:100 dilution, OD<sup>600</sup> ∼0.05) in LB broth with overnight culture and incubated at 37◦C for 20 h with shaking at 250 rpm with herring oil, DHA, or EPA (1, 5, or 20 µg/ml) or DMSO (the negative control). Freshly drawn human whole blood (0.3 ml) was then mixed with the S. aureus cultures (0.1 ml), and mixtures were incubated at 37◦C for 3 h with shaking at 250 rpm. S. aureus (MSSA 6538) survival was measured by counting CFUs on LB agar plates.

### Caenorhabditis elegans Survival Assay

To investigate the effects of herring oil and unsaturated fatty acids on the virulence of S. aureus MSSA 6538, we used a nematode survival assay as previously described (Kim et al., 2016) with slight modification. In brief, S. aureus cells were incubated with or without herring oil, DHA, or EPA (2, 5, or 20 µg/ml) at 37◦C for 24 h and synchronized adult C. elegans fer-15;fem-1 nematodes were added into single wells of 96-well plate containing cultivated S. aureus cells. Approximately, 30 nematodes were allowed to feed on the cultured S. aureus MSSA 6538 at 25◦C for 1 day.

For the cytotoxicity assay, 110 ± 10 nematodes were added into single well of 96-well plates containing M9 buffer and solutions of the compounds were added to final concentrations of 20 or 100 µg/ml at 25◦C for 1 day. Then, nematodes were scored as alive or dead using an iRiSTM Digital Cell Imaging System (Logos Bio Systems, South Korea). At least three independent experiments were conducted using quadruplicate wells.

### RNA Isolation and qRT-PCR

Staphylococcus aureus MSSA 6538 cells were inoculated into 25 ml of LB broth at 37◦C in 250 ml flasks at a starting OD<sup>600</sup> of 0.05, and then incubated for 5 h with shaking at 250 rpm in the presence or absence of herring oil (100 µg/ml), DHA, or EPA (20 µg/ml). RNase inhibitor (RNAlater, Ambion, TX, United States) was then added and cells were immediately chilled for 30 s in dry ice bath having 95% ethanol to prevent RNA degradation. Cells were then harvested by centrifugation at 16,600 g for 1 min and total RNA was isolated using a Qiagen RNeasy mini Kit (Valencia, CA, United States).

Quantitative real-time reverse transcription polymerase chain reaction was used to investigate the transcriptional levels of 10 biofilm-related genes, that are, agrA, arlR, arlS, aur, hla, icaA, nuc1, rbf, RNAIII, saeR, sarA, sarZ, seb, sigB, and spa in S. aureus MSSA 6538 cells. Gene specific primers were used and 16s rRNA was used as the housekeeping control (Supplementary Table S1) to normalize the expressions of genes of interest. The qRT-PCR technique employed was an adaptation of a previously described method (Lee et al., 2016). qRT-PCR was performed using a SYBR Green master mix (Applied Biosystems, Foster City, CA, United States) and an ABI StepOne Real-Time PCR System (Applied Biosystems). Expression levels were determined using two independent cultures and six qRT-PCR reactions per gene.

## Statistical Analysis

Values were expressed as means ± standard deviations, and data were analyzed by one-way ANOVA followed by Dunnett's test using SPSS version 23 (SPSS Inc., Chicago, IL, United States) to determine the significances of differences. Statistical significance was accepted for p-values of <0.05.

TABLE 1 | Herring oil – GC-MS analysis.


Components present at levels >5% by weight are shown in blue/bold font.

### RESULTS AND DISCUSSION

## Effect of Herring Oil on S. aureus Biofilm Formation

The anti-biofilm activity of herring oil was investigated against two S. aureus strains (MSSA 6538 and MRSA 33591) in 96-well polystyrene plates. The addition of herring oil at the beginning of bacterial culture dose-dependently inhibited S. aureus biofilm formation (**Figures 1A,B**). Specifically, herring oil at 100 µg/ml reduced biofilm formation by MSSA 6538 by >75%, whereas 20 µg/ml was required to inhibit biofilm formation by MRSA 33591 by >65%.

Herring oil did not decrease the growth of S. aureus strains (MSSA 6538 and MRSA 33591) at concentrations up

to 1000 µg/ml (**Figures 1C,D**), and did not affect planktonic cell numbers of either S. aureus strain at concentrations up to 1000 µg/ml (data not shown). These findings show the antibiofilm activity of herring oil was not due to any bactericidal effects.

## Identification of Main Components in Herring Oil

Gas chromatography-mass spectrometry identified 28 fatty acids in herring oil (**Table 1**). Components that constituted more than 5% of herring oil were; erucic acid (17.49%), cis-4,7,10,13,16,19- DHA (11.31%), cis-11-eicosenoic acid (9.75%), palmitic acid (8.18%), myristic acid (5.78%), and cis-5,8,11,14,17-EPA (5.78%) (**Table 1**). The compositions of herring oil identified by GC-MS analysis concur with previous studies (Lambertsen and Braekkan, 1965). Previously, unsaturated fatty acids, such as, cis-11-eicosenoic acid and oleic acid showed strong anti-biofilm effect while saturated fatty acids (myristic acid and palmitic acid) did not have anti-biofilm activity against S. aureus (Lee et al., 2017). Therefore, this study was focused on the anti-biofilm activities of three unsaturated predominant omega fatty acids (DHA, EPA, and erucic acid).

## Effect of DHA and EPA on Biofilm Formation by S. aureus

The anti-biofilm potencies of four major PUFAs (cis-11 eicosenoic acid, DHA, EPA, and erucic acid) were tested against the two S. aureus strains (MSSA 6538 and MRSA 33591) in 96-well plates. cis-11-Eicosenoic acid, DHA, and EPA at concentrations of 20 µg/ml significantly inhibited biofilm formation by both S. aureus strains (**Figure 2A**). DHA and EPA both inhibited biofilm formation by >90% at 50 µg/ml, which was greater than that previously reported for cis-11 eicosenoic acid (Lee et al., 2017). In this previous report, relationships between free fatty acids and bacterial biofilm formation were studied. In particular, oleic acid and cis-2 decenoic acid were found to suppress biofilm formation of S. aureus by blocking bacterial adhesion and dispersion from established biofilms, respectively (Stenz et al., 2008; Davies and Marques, 2009). In addition, it has been suggested that long chain unsaturated fatty acids like cis-11-eicosenoic acid (Lee et al., 2017) inhibit the biofilm forming ability of S. aureus. However, the present study is the first to report the antibiofilm activities of DHA and EPA against S. aureus. The unsaturated fatty acid erucic acid, which was present in herring oil at high levels (**Table 1**), at 20 µg/ml inhibited biofilm formation by MRSA 33591 by >85%, but did not inhibit biofilm formation by MSSA 6538 (**Figure 2A**). However, higher concentrations of erucic acid at 500 and 1000 µg/ml significantly reduced biofilm formation of MSSA 6538 without affecting planktonic growth (data not shown). Hence, it appears that the effects of erucic acid are dose-dependent and strain dependent.

Confocal laser scanning microscopy was used to quantify biofilm formation by S. aureus in the presence of herring oil, DHA, or EPA. Fluorescent stacking 3D images indicated that herring oil (50 µg/ml), DHA (20 µg/ml), or EPA (20 µg/ml) markedly inhibited S. aureus biofilm formation (**Figure 2B**), and this, was confirmed by COMSTAT. More specifically, DHA, EPA, and herring oil reduced all three S. aureus MSSA 6538 parameters measured (biovolume, mean

thickness, and substratum coverage) (**Figure 2C**), actually, DHA and EPA reduced biomass and mean thickness by >90%, respectively. These results show herring oil, DHA, and EPA all effectively reduced biofilm formation on the bottom of 96-well plate.

Mature biofilm detachment by antibiotics, enzymes, and fatty acids is interest due to their resistance to antimicrobial agents (Flemming and Wingender, 2010; Gwisai et al., 2017; Lee et al., 2018). However, herring oil (1000 µg/ml), DHA (100 µg/ml), and EPA (100 µg/ml) could not disperse matured S. aureus ATCC 6538 biofilms (data not shown).

## Anti-microbial Effects of DHA and EPA on S. aureus

The anti-microbial activities of DHA and EPA were evaluated by determining their minimum inhibitory concentrations (MICs) against S. aureus MSSA 6538 and MRSA 33591. Planktonic growth was completely inhibited by DHA and EPA at MICs (200 µg/ml), while MICs of ciprofloxacin and vancomycin are 0.5–1 µg/ml and 4 µg/ml, respectively, which concurs with previously reported values (Desbois and Lawlor, 2013; Gwisai et al., 2017). Notably, these MICs of DHA and EPA were 10 times higher than the concentration (20 µg/ml) required for anti-biofilm activities (**Figure 2A**).

Planktonic cell growths of S. aureus MSSA 6538 and MRSA 33591 in the presence of 5 µg/ml DHA or 20 µg/ml EPA were slightly inhibited (**Figures 2D,E**). However, the cell growths of both strains were more than 68% suppressed by DHA or EPA at 50 µg/ml, which indicates both DHA and EPA have anti-bacterial effects at relatively high concentrations (**Figures 2D,E**). These observations suggest DHA and EPA at low concentrations have anti-biofilm effects, but at higher levels these can also suppress the growth of planktonic cells.

## Effects of Herring Oil, DHA, and EPA on Hemolysis by S. aureus

Staphylococcus aureus produces α-toxin, which is one of the most investigated S. aureus cytotoxins. α-Toxin integrates into the membranes of erythrocytes causing hemolysis (Bhakdi et al., 1984; Song et al., 1996), which contributes to biofilm formation (Caiazza and O'Toole, 2003). Thus, the effects of herring oil, DHA, and EPA on human red blood cell hemolysis by S. aureus were investigated. As a control, herring oil, DHA, and EPA without S. aureus cells did not show any hemolytic activity (data not shown). Interestingly, all three dose-dependently inhibited the hemolytic activity of S. aureus (**Figure 3**), and DHA and EPA at 20 µg/ml completely abolished its hemolytic activity, whereas herring oil at 20 µg/ml reduced its hemolytic activity by >75%. Also, herring oil, DHA, and EPA all significantly reduced the hemolytic activity by S. aureus even at 1 µg/ml, which is much lower than the concentrations required for anti-biofilm and antimicrobial activities.

These results support that the inhibitions of S. aureus biofilm formation by herring oil, DHA, and EPA are associated with the inhibition of hemolysis, and suggest that the genes related to biofilm formation and hemolysis might be responsible for these events. The present study is the first to report that herring oil, DHA, and EPA suppress the hemolysis of human red blood cells by S. aureus, which has important implications regarding the control of S. aureus virulence.

## Effects of DHA, EPA, and Herring Oil on S. aureus Survival in the Presence of Human Whole Blood

Staphylococcus aureus characteristically exhibits resistance to phagocytes in humans and other animals, and thus, we used a survival test to investigate the effects of herring oil, DHA, and EPA on S. aureus survival in the presence of human whole blood. Herring oil, DHA, or EPA at 1, 5, or 20 µg/ml significantly reduced the survival of S. aureus exposed to fresh human whole blood (**Figure 4**), indicating that they weaken the ability of S. aureus to resist the human innate immune system.

## Effects of DHA, EPA, and Herring Oil on S. aureus Virulence in the Nematode Model

Staphylococcus aureus colonizes and replicates in the digestive tract of C. elegans and has the ability to kill C. elegans by an infection-like process that exhibits remarkable overlap with that observed in mammals (Garsin et al., 2003). Hence, the effects of herring oil, DHA, and EPA on S. aureus virulence were investigated by analyzing C. elegans survival in the presence of S. aureus. Herring oil, DHA, and EPA were found to markedly prolong C. elegans survival in the presence of S. aureus (**Figure 5A**). For example, nematode survival was only ∼10% in the presence of untreated S. aureus, whereas in the presence of herring oil, DHA, or EPA at 20 µg/ml more than 55, 60, and 70%, respectively, of worms survived. In other words, the survival rate of C. elegans was increased more than seven and ninefold when worms were exposed to S. aureus in the presence of DHA or EPA at 20 µg/ml, respectively (**Figure 5A**). In addition, no toxic effects were observed when non-infected worms were exposed to herring oil, DHA, or EPA at concentrations up to 100 µg/ml (**Figure 5B**).

## Transcriptional Changes Induced by DHA, EPA, and by Herring Oil in S. aureus

To investigate the molecular mechanisms underlying the inhibitory effects of herring oil, DHA, and EPA against S. aureus, expressions of selected biofilm- and virulence-related genes, and important regulatory genes were investigated in S. aureus cells by qRT-PCR. Herring oil, DHA, and EPA dramatically reduced the expression of α-hemolysin (hla) by 27-, 24-, and 20-fold, respectively (**Figure 6**). Also, herring oil, DHA, and EPA downregulated regulatory RNA molecule (RNAIII) that is present upstream of the agr operon while the expressions of other genes were not affected (**Figure 6**). It has been reported that RNAIII stimulate hla translation (Morfeldt et al., 1995). Hence, hla repression is partially due to the down-regulation of RNAIII by herring oil, DHA, and EPA.

α-Hemolysin is known to play a critical role in cell-to-cell interactions during biofilm formation (Caiazza and O'Toole, 2003; Anderson et al., 2012; Scherr et al., 2015). This reduction of α-hemolysin expression has been discovered in recently published articles showing that alizarin (Lee et al., 2016), azithromycin (Gui et al., 2014), flavonoids (Cho et al., 2015), norlichexanthone (Baldry et al., 2016), and stilbenoids (Lee et al., 2014) inhibit both biofilm formation and hemolysis by S. aureus. Our findings support that α-hemolysin has a positive relationship with biofilm formation in S. aureus. In addition, it has been shown hla is essential for S. aureus pathogenicity in a nematode model (Sifri et al., 2003, 2006). Notably, neither herring oil, DHA, or EPA significantly affected the expressions of the biofilm-related genes namely, agrA, arlR, arlS, aur, icaA, nuc1, rbf, saeR, sarA, sarZ, seb, sigB, or spa. Thus, the present study indicates that herring oil, DHA, and EPA attenuate S. aureus virulence, as evidenced by reductions in its antibiofilm (**Figures 1**, **2**) and anti-hemolytic (**Figure 3**) activities, and partially by down-regulating the expression of the hla gene (**Figure 6**).

These results indicate that herring oil, DHA, and EPA could be used to treat S. aureus infections as lone drug or in combination with traditional antibiotics (**Figure 5**). Furthermore, because all three are considered intrinsically safe in animal, they could be used for medicinal purposes or for example, to surface treat in food processing facilities without undue restriction.

## CONCLUSION

The DHA and EPA are long chain omega-3 fatty acids, which have major health benefits in man. More specifically, circulating levels of DHA and EPA are essential part of the body's defense system and play key roles in the inhibitions of inflammation and host immune responses (Calder, 2013). Herring oil was found to contain DHA and EPA at levels of 11.3 and 5.8% by weight, respectively, and inhibit biofilm formation and hemolysis by and the virulence of S. aureus. Interestingly, the ability of S. aureus to survive exposure to human whole blood was reduced in the presence of DHA or EPA, and exposure to DHA or EPA also increased infected nematode survival. These findings show for the first time that DHA and EPA suppress S. aureus biofilm formation and its virulence. Also, herring oil and its constituent omega-3 fatty acids (DHA and EPA) might be useful for the treatment and/or prevention of surface-associated biofilm formation by S. aureus (including MRSA strains) and for suppressing its virulence.

## AUTHOR CONTRIBUTIONS

Y-GK, J-HL, CR, and SO performed the experiments and analyzed the data. Y-GK, J-HL, JP, and JL designed the study and wrote the paper. All authors read and approved the final manuscript.

## FUNDING

This research was conducted under the industrial infrastructure program for fundamental technologies (N0000885) which is funded by the Ministry of Trade, Industry and Energy (MOTIE, South Korea). This research was also supported by the Basic Science Research Program of a National Research Foundation of Korea (NRF) funded by the Korean Ministry of Education (2018R1D1A3B07040699 to J-HL) and (2017R1A6A3A01076089 to Y-GK), and a Priority Research Centers Program (#2014R1A6A1031189).

### REFERENCES

fmicb-09-01241 June 15, 2018 Time: 12:0 # 9


### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.01241/full#supplementary-material

program COMSTAT. Microbiology 146, 2395–2407. doi: 10.1099/00221287- 146-10-2395


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Kim, Lee, Raorane, Oh, Park and Lee. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fmicb-09-01241 June 15, 2018 Time: 12:0 # 10

# Nano-Strategies to Fight Multidrug Resistant Bacteria—"A Battle of the Titans"

Pedro V. Baptista<sup>1</sup> \*, Matthew P. McCusker 2†, Andreia Carvalho<sup>1</sup> , Daniela A. Ferreira<sup>3</sup> , Niamh M. Mohan3,4, Marta Martins <sup>3</sup> \* and Alexandra R. Fernandes <sup>1</sup> \*

### Edited by:

Rebecca Thombre, Pune University, India

### Reviewed by:

Shahper Nazeer Khan, Aligarh Muslim University, India Rodolfo García-Contreras, Universidad Nacional Autónoma de México, Mexico

\*Correspondence:

Pedro V. Baptista pmvb@fct.unl.pt Marta Martins mmartins@tcd.ie Alexandra R. Fernandes ma.fernandes@fct.unl.pt

### †Present Address:

Matthew P. McCusker, Kerry Europe, Global Technology & Innovation Centre, Naas, Ireland

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

> Received: 31 March 2018 Accepted: 11 June 2018 Published: 02 July 2018

### Citation:

Baptista PV, McCusker MP, Carvalho A, Ferreira DA, Mohan NM, Martins M and Fernandes AR (2018) Nano-Strategies to Fight Multidrug Resistant Bacteria—"A Battle of the Titans". Front. Microbiol. 9:1441. doi: 10.3389/fmicb.2018.01441 <sup>1</sup> UCIBIO, Departamento de Ciências da Vida, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Caparica, Portugal, <sup>2</sup> School of Food Science and Environmental Health, College of Sciences and Health, Dublin Institute of Technology, Dublin, Ireland, <sup>3</sup> Department of Microbiology, Moyne Institute of Preventive Medicine, Schools of Genetics and Microbiology, Trinity College Dublin, University of Dublin, Dublin, Ireland, <sup>4</sup> Nuritas Limited, Dublin, Ireland

Infectious diseases remain one of the leading causes of morbidity and mortality worldwide. The WHO and CDC have expressed serious concern regarding the continued increase in the development of multidrug resistance among bacteria. Therefore, the antibiotic resistance crisis is one of the most pressing issues in global public health. Associated with the rise in antibiotic resistance is the lack of new antimicrobials. This has triggered initiatives worldwide to develop novel and more effective antimicrobial compounds as well as to develop novel delivery and targeting strategies. Bacteria have developed many ways by which they become resistant to antimicrobials. Among those are enzyme inactivation, decreased cell permeability, target protection, target overproduction, altered target site/enzyme, increased efflux due to over-expression of efflux pumps, among others. Other more complex phenotypes, such as biofilm formation and quorum sensing do not appear as a result of the exposure of bacteria to antibiotics although, it is known that biofilm formation can be induced by antibiotics. These phenotypes are related to tolerance to antibiotics in bacteria. Different strategies, such as the use of nanostructured materials, are being developed to overcome these and other types of resistance. Nanostructured materials can be used to convey antimicrobials, to assist in the delivery of novel drugs or ultimately, possess antimicrobial activity by themselves. Additionally, nanoparticles (e.g., metallic, organic, carbon nanotubes, etc.) may circumvent drug resistance mechanisms in bacteria and, associated with their antimicrobial potential, inhibit biofilm formation or other important processes. Other strategies, including the combined use of plant-based antimicrobials and nanoparticles to overcome toxicity issues, are also being investigated. Coupling nanoparticles and natural-based antimicrobials (or other repurposed compounds) to inhibit the activity of bacterial efflux pumps; formation of biofilms; interference of quorum sensing; and possibly plasmid curing, are just some of the strategies to combat multidrug resistant bacteria. However,

**153**

the use of nanoparticles still presents a challenge to therapy and much more research is needed in order to overcome this. In this review, we will summarize the current research on nanoparticles and other nanomaterials and how these are or can be applied in the future to fight multidrug resistant bacteria.

Keywords: antimicrobial resistance, multidrug resistance, nanomaterials, nanoparticles, plant-based compounds, novel antimicrobial agents, nanotheranostics

### INTRODUCTION

Multidrug resistant (MDR) bacteria remain the greatest challenge in public health care. The numbers of infections produced by such resistant strains are increasing globally. This acquired resistance of pathogens presents a key challenge for many antimicrobial drugs. Recent advances in nanotechnology offer new prospects to develop novel formulations based on distinct types of nanoparticles (NPs) with different sizes and shapes and flexible antimicrobial properties.

NPs may offer a promising solution as they can not only combat bacteria themselves but can also act as carriers for antibiotics and natural antimicrobial compounds (Wang et al., 2017a). While various materials have been explored from liposomal to polymer based nano-drug carriers, metallic vectors, such as gold NPs, are attractive as core materials due to their essentially inert and nontoxic nature (Burygin et al., 2009). Arguably the most attractive aspect of NPs drug delivery systems is their ability to introduce a wide range of therapeutics, either bound to their large surface area or contained within the structure, to the site of infection effectively and safely by having a controlled rate of targeted delivery (Pissuwan et al., 2011; Gholipourmalekabadi et al., 2017). By improving the pharmacokinetic profile and therapeutic index of encapsulated drugs compared to free drug equivalents, the dose required to achieve clinical effects can be dramatically decreased (Gao et al., 2018). This in turn, can reduce the toxicity and the adverse side effects associated with high systemic drug concentrations and frequent dosing (Liu et al., 2009).

This review covers the latest approaches in the development of new nanobiotechnology approaches that may challenge the medical practice to fight bacteria and particularly MDR bacteria.

### NANOMATERIALS AGAINST BACTERIA

Nanomaterials have at least one dimension in the nanometer scale range (1–100 nm) that convey particular physical and chemical properties considerably different from those of bulk materials (Wang et al., 2017a). Among the wide range of nanomaterials, particular interest has been directed toward NPs. NPs have a number of features, which make them favorable as vectors for drugs to combat disease-causing pathogens. These include their enhancement of drug solubility and stability (Huh and Kwon, 2011); their ease of synthesis (Gholipourmalekabadi et al., 2017); their biocompatibility with target agents; and their modulated release, which can be controlled by stimuli, such as light, pH and heat (Wang Z. et al., 2017). Their distinctive functionality in drug delivery is achieved by their ultra-small size and vast surface to volume ratios. This is a key competitive advantage over conventional therapies in the treatment of infections caused by intracellular pathogens and MDR strains. Their functionalization with different (bio)molecules is another important feature. These comprise NPs containing Ag, Au, Al, Cu, Ce, Cd, Mg, Ni, Se, Pd, Ti, Zn, and super-paramagnetic Fe (Hemeg, 2017; Slavin et al., 2017). AgNPs are considered the most effective nanomaterial against bacteria but other metallic NPs, such as CuONPs, TiONPs, AuNPs, and Fe3O2NPs, have also demonstrated bactericidal effects (Dakal et al., 2016; Hemeg, 2017; Slavin et al., 2017).

While poor membrane transport limits the potency of many antibiotics (Andrade et al., 2013), drug loaded NPs vehicles can enter host cells via endocytosis, facilitating their intracellular entry (Wang Z. et al., 2017). Membrane penetration can also be achieved through interactions with surface lipids, for example, using gold NPs in the co-administration of protein-based drugs (Huang et al., 2010). The therapeutic appeal of NPs is enhanced by their ability to confer physical protection against bacterial resistance mechanisms (Huh and Kwon, 2011). Furthermore, the potential to load multiple drug combinations into NPs presents a highly complex antimicrobial mechanism of action, to which, bacteria are unlikely to develop resistance (Huh and Kwon, 2011). Although, this is usually believed to be the case, there are some studies reporting development of bacterial resistance against silver NPs (Panácek et al., 2018 ˇ ). There is also evidence that exposure of bacteria to this type of NPs may increase its antibiotic tolerance (Kaweeteerawat et al., 2017).

NPs can exert their antibacterial activity via a multitude of mechanisms, such as: (1) direct interaction with the bacterial cell wall; (2) inhibition of biofilm formation; (3) triggering of innate as well as adaptive host immune responses; (4) generation of reactive oxygen species (ROS); and (5) induction of intracellular effects (e.g., interactions with DNA and/or proteins). Because they do not present the same mechanisms of action of standard antibiotics (**Figure 1**), they can be of extreme use against MDR bacteria (Singh K. et al., 2014; Aderibigbe, 2017; AlMatar et al., 2017; Hemeg, 2017; Natan and Banin, 2017; Rai et al., 2017; Slavin et al., 2017; Zaidi et al., 2017; Bassegoda et al., 2018; Katva et al., 2018; Siddiqi et al., 2018).

Besides the broad-spectrum antibacterial properties that NPs have against Gram-positive and -negative bacteria, NPs have been used as vectors for the delivery of antimicrobial moieties that greatly improve their biocidal properties (Beyth et al., 2015; Rai A. et al., 2016; Singh J. et al., 2016; Esmaeillou et al.,

2017; Wang et al., 2017a; Zaidi et al., 2017; Hadiya et al., 2018). Some of the advantages of using NPs as vectors are due to their small and controllable size; their protective action against enzymes that would otherwise destroy antimicrobial compounds; their ability to actively deliver antibiotics; and their capability to combine several therapeutic modalities onto a single nanomaterial (e.g., several antibiotics/compounds onto the same NPs for combined action; combining silencing agents and drugs, etc.) (Turos et al., 2007; Huh and Kwon, 2011; Mohammed Fayaz et al., 2011; Liu et al., 2013; Qi et al., 2013; Li et al., 2014; Ranghar et al., 2014; Thomas et al., 2014; Wang et al., 2014; Payne et al., 2016; Rai A. et al., 2016; Singh J. et al., 2016; Yeom et al., 2016; Esmaeillou et al., 2017; Zaidi et al., 2017; Zong et al., 2017; Hadiya et al., 2018).

NPs carriers can tackle bacterial threats "passively," through prolonged drug retention at the specific infection site, or "actively," through surface conjugation with active molecules that bind a certain target (Wang Z. et al., 2017). The balance between the surface modification interaction strength, the compound release rate and the stability of the conjugate should be carefully considered for the design of an effective "active" delivery strategy (Burygin et al., 2009; Pissuwan et al., 2011). In an attempt to overcome their therapeutic limitations, various research groups have investigated the conjugation of antibiotics to NPs (Tiwari et al., 2011). For example, Saha et al. describe the direct conjugation of ampicillin, streptomycin and kanamycin to gold NPs (Saha et al., 2007). The resulting complexes were shown to have lower minimum inhibitory concentration (MIC) than the free drug counterparts against both Gram -negative and -positive bacteria. While the detailed mechanism of these effects are not explained by the authors in the above case, Fayaz et al. has attempted to uncover how their vancomycin functionalized gold NPs demonstrated activity against strains which are usually vancomycin resistant based either on mutations (vancomycin resistant Staphylococcus aureus), or membrane structure (Escherichia coli) (Mohammed Fayaz et al., 2011). They propose that only when the antibiotic was complexed with the NPs could this result in nonspecific, multivalent interactions and anchoring of the carrier to the cell wall synthesis proteins. Based on the presence of pits in the cells, which was observed using transmission electron microscopy, the authors concluded that the consequence of the non-specific binding was compromised membrane integrity, and subsequent cell death (Mohammed Fayaz et al., 2011; Gao et al., 2018).

## ANTIBACTERIAL MECHANISM OF NPS

The antibacterial activity of NPs against MDR bacteria and biofilms depends on a number of factors, namely, their large surface area in contact with bacteria through electrostatic attraction, van der walls forces or hydrophobic interactions; on the nanoparticle size and stability; together with the drug concentration (Chen et al., 2014; Gao et al., 2014; Li et al., 2015). The interaction of NPs with bacteria generally triggers oxidative stress mechanisms, enzymatic inhibition, protein deactivation and changes in gene expression. Still, the most common antibacterial mechanisms are related to oxidative stress, metal ion release, and non-oxidative mechanisms (Wang et al., 2017a; Zaidi et al., 2017 see **Figure 1**).

Oxidative stress induced by ROS is one of the most important mechanisms assisting the antibacterial activity of NPs (Dwivedi et al., 2014; Rudramurthy et al., 2016). ROS are natural byproducts of cellular oxidative metabolism and have significant important roles in the modulation of cell survival and death, differentiation and cell signaling. In bacteria, ROS are formed from aerobic respiration, and their production is balanced by the cell antioxidant machinery but upon an additional ROS insult, oxidation of biomolecules, and cell components result in severe cellular damage (Li et al., 2012b). The excessive production of ROS leads to a disturbed redox homeostasis resulting in oxidative stress, affecting membrane lipids and altering the structure of DNA and proteins (Dwivedi et al., 2014). It has been shown that while O<sup>−</sup> 2 and H2O<sup>2</sup> can be neutralized by endogenous antioxidants, ·OH and singlet oxygen (<sup>1</sup> [O2]) lead to acute microbial death (Zaidi et al., 2017). Different NPs may generate distinctive ROS, such as superoxide (O<sup>−</sup> 2 ) or hydroxyl radical (·OH), hydrogen peroxide (H2O2), and 1 [O2]) (Wang et al., 2017a). In this manner, the level of ROS generated by NPs is dependent on the chemical nature of the NPs themselves. Application of metallic NPs is currently being considered to overcome bacterial infections since they have shown antimicrobial efficacy due to their high surfaceto-volume ratio. An increase ratio is usually accompanied by increased production of ROS, including free radicals. Zhang et al. (2013) demonstrated that ROS generation and metal ion release significantly enhanced the antibacterial activity through uncoated AuNPs in aqueous suspension under UV irradiation (365 nm). Umamaheswari (Umamaheswari et al., 2014) demonstrated that the antibacterial activity of AuNPs against E. coli, Salmonella Typhi, Pseudomonas aeruginosa and Klebsiella pneumoniae were due to oxidative stress caused by increased intracellular ROS. A recent study (Zhang et al., 2013) evaluated AuNPs and AuNPs -laser combined therapy against C. pseudotuberculosis and suggested that the mechanism of action is related with ROS production, that causes an increase of oxidative stress of microbial cells in the form of vacuole formation as an indication of potent activity. This effect was higher with AuNPs-laser, causing a rapid loss of bacterial cell membrane integrity due to the fact that laser light enhances at least one fold antimicrobial activity of AuNPs. Several other studies have addressed the role of metal NPs to induce MDR bacteria death via oxidative stress (**Table 1**) (Foster et al., 2011; Li et al., 2012b; Rai et al., 2012; Zhang et al., 2013; Reddy L. S. et al., 2014; Singh R. et al., 2014; Pan et al., 2016; Courtney et al., 2017; Ulloa-Ogaz et al., 2017; Zaidi et al., 2017). Indeed, titanium dioxide NPs were shown to adhere to the surface of the bacterial cell and trigger the production of ROS, which in turns lead to damage of the structure of cellular components and consequent cell death (Foster et al., 2011). In another important study using different metal NPs, AgNPs were shown to generate superoxide radicals and hydroxyl radicals, whereas Au, Ni, and Si NPs generated only singlet oxygen, which upon entering the cell produced an antibacterial effect (Zhang et al., 2013). More recently, Reddy and co-workers demonstrated that ZnONPs alone can also act as an effective antibacterial agent via the generation of ROS (Reddy L. S. et al., 2014). Exposure to UV irradiation may also potentiate the action of NPs. Li et al. (2012b) reported the augmented antibacterial effects of zinc oxide (ZnO) and titanium oxide (TiO) NPs triggered by UV irradiation as the results of the increased production of superoxide, hydroxyl and singlet oxygen radicals that potentiated bacteria mortality by severe oxidative stress. Graphene oxide– iron oxide NPs have also demonstrated maximum antibacterial activity due to the generation of hydroxyl radicals and diffusion into bacterial cells (Pan et al., 2016). More recently, Ulloa-Ogaz and collaborators demonstrated that copper oxide NPs interact with bacteria, generating an intracellular signaling cascade that trigger oxidative stress and, thus, an antibacterial effect (Ulloa-Ogaz et al., 2017).

Metal oxides slowly release metal ions that are up taken by the cell, reaching the intracellular compartment where they can interact with functional groups of proteins and nucleic acids, such as amino (–NH), mercapto (–SH), and carboxyl (– COOH) groups (Wang et al., 2017a). This interaction alters the cell structure, hampers enzymatic activity and interferes with the normal physiological processes in the bacterial cell. It has been shown that copper oxide (CuO) NPs cause a significant alteration of the expression of key proteins and may inhibit denitrification. Proteomic analysis showed that CuONPs cause an alteration of proteins involved in nitrogen metabolism, electron transfer and transport (Su et al., 2015). Also, the interaction of gold–superparamagnetic iron oxide NPs with bacterial proteins via disulfide bonds affects the metabolism and redox system of bacterial cells (Niemirowicz et al., 2014). NPs may also enter bacteria through absorption, releasing metal ions to the surrounding medium and/or binding to the negatively charged functional groups of the bacterial cell membrane. For example, silver ions (from silver NPs) are adsorbed on the cell membrane, leading to protein coagulation (Jung et al., 2008). Jankauskaitl and collaborators described the bactericidal effect of graphene oxide/Cu/Ag NPs against E. coli, P. aeruginosa, K. pneumoniae, S. aureus, and Methicillin-resistant S. aureus (MRSA) through a possible synergy between multiple toxic mechanisms (Jankauskaite et al., 2016).

Non-oxidative mechanisms involve interaction of the NPs with the cell wall. In bacteria, the cell wall and membrane behave as defensive barriers that protect against environmental insults. Cell membrane components provide different adsorption pathways for the NPs (Lesniak et al., 2013). The cell wall of Gramnegative bacteria is composed of lipoproteins, phospholipids and lipid polysaccharides (LPS), which form a barrier only allowing the entry of certain macromolecules (Zaidi et al., 2017). In Grampositive bacteria, the cell wall is composed of a thin layer of TABLE 1 | Nanoparticles against MDR pathogens and their mechanisms of action.



(Continued)


(Continued)


(Continued)


peptidoglycans and abundant pores that allow the penetration of foreign molecules, leading to covalent binding with proteins and cellular components, interrupting the proper functioning of the bacterial cell (Sarwar et al., 2015). In addition, Grampositive bacteria have a highly negative charge on the surface of the cell wall. For example, LPS provides negatively charged regions on the cell wall of Gram-negative bacteria that attracts NPs; and, since teichoic acid is only expressed in Gram-positive bacteria, the NPs are distributed along the phosphate chain. As such, the antimicrobial effect is more foreshadowed in Grampositive than -negative bacteria (Wang et al., 2017a). Indeed, Yu and colleagues synthesized a novel hydroxyapatite whisker (HAPw)/zinc oxide (ZnO) NPs and evaluated the antimicrobial effect against S. aureus, E. coli, and Streptococcus mutans. The authors demonstrate that the antibacterial effect depends on the components and structure of the bacterial cell wall. The antibacterial action of these NPs could be improved for Grampositive bacteria and certain components could prevent the adhesion of ZnO NPs to the bacterial cell barrier (Yu et al., 2014). Ansari et al. reported that the accumulation on NPs in the bacterial cell wall causes irregularly shaped pit, perforation and disturbs metabolic processes (Ansari et al., 2014). In a study carried out by Joost and co-workers, it was demonstrated that a treatment with TiO<sup>2</sup> NPs increased the bacterial cell volume, resulting in membrane leakage (Joost et al., 2015).

### BIOFILM FORMATION AND QUORUM-SENSING

Biofilm formation plays an important role in bacterial resistance protecting bacteria and allowing then to evade the action of antibiotics (Lebeaux et al., 2014; Khameneh et al., 2016). The most active fractions of bacteria are now recognized to occur as biofilms, where cells are adhered to each other on surfaces within a self-produced matrix of extracellular polymeric substance (EPS). EPS provide a barrier allowing to inhibit the penetration of antibiotics and further promote antibiotic resistance leading to a serious health threat worldwide since biofilms are resistant to antibiotics penetration and escape innate immune system by phagocytes (Hall-Stoodley et al., 2004; Bjarnsholt, 2013). Numerous experimental evidence show that NPs are capable of disrupting the bacterial membranes and can hinder biofilm formation thus reducing the survival of the microorganism (Peulen and Wilkinson, 2011; Leuba et al., 2013; Pelgrift and Friedman, 2013; Slomberg et al., 2013; Chen et al., 2014; Miao et al., 2016; Yu et al., 2016; Kulshrestha et al., 2017). This way, NPs provide an alternative strategy to target bacterial biofilms with potential to use both antibiotic-free and antibiotic-coated approaches (Gu et al., 2003; Li et al., 2012a; Sathyanarayanan et al., 2013). Earlier reports demonstrated that NPs are able to interfere with biofilm integrity by interacting with EPS and with the bacterial communication - quorum sensing (QS). The properties of NPs must be designed to be able to inhibit biofilm formation namely through size and surface chemistry. The size of NPs is important to it since they must be able to penetrate the EPS matrix and surface chemistry will command the amount of interactions with the EPS (Lundqvist et al., 2008). The majority of the strategies to achieve inhibition of biofilm formation are to target and interfere with QS molecules (Singh et al., 2017).

QS systems in bacterial populations act as major regulatory mechanisms of pathogenesis, namely in the formation of biofilm structures. These systems help bacteria to "communicate" with each other, through the production and detection of signal molecules (Rutherford and Bassler, 2012; Papenfort and Bassler, 2016). Using this cell-to-cell communication, bacterial populations are able to synchronize the expression of their genes, acquiring competitive advantage to respond to changes in the environment (Rutherford and Bassler, 2012). Therefore, QS systems are known to promote the formation of antibiotic tolerant biofilm communities. It is known that biofilm structures are a recalcitrant mode of bacterial growth that increases bacterial resistance to conventional antibiotics (Reen et al., 2018). This way, bacterial biofilms pose a significant challenge to the efficacy of conventional antibiotics being considered an essential platform for antibiotic resistance (Høiby et al., 2011). Taking this into account, it isn't surprising that the targeting and disruption of QS signaling systems and consequently, of the biofilm production, set the pillar for future next-generation antivirulence therapies to be developed (LaSarre and Federle, 2013; Venkatesan et al., 2015; Jakobsen et al., 2017).

Surface-functionalized NPs with β-cyclodextrin (β-CD) or N-acylated homoserine lactonase proteins (AiiA) are able to interfere with signaling molecules preventing these molecules from reaching its cognate receptor, therefore inhibiting the signal/receptor interaction. This process will "turn off " QS and obstructing the bacterial communication (Kato et al., 2006; Ortíz-Castro et al., 2008). Several papers reported inhibition of biofilm formation namely by gold NPs (AuNPs). Acyl homoserine lactones (AHL) are signaling molecules with a role in bacterial QS and bind directly to transcription factors to regulate gene expression Recently, Gopalakrishnan and colleges synthesized (Vinoj et al., 2015) AuNPs coated AiiA purified from Bacillus licheniformis. These AiiA AuNPs inhibited EPs production and demonstrated potent antibiofilm activity against Proteus species at 2–8µM concentrations without being harmful for the host cells at the 2µM concentration. Sathyanarayanan et al. (2013) demonstrated that using AuNPs there is a significant reduction of S. aureus and P. aeruginosa biofilms applied in high concentration (exceeding 50 mg/L). A recent study by Yu et al. (2016) demonstrated that AuNPs were able to strongly attenuate biofilm formation of P. aeruginosa. The inhibition observed in this study was related with interruption of adhesin- mediated interaction between the bacteria and the substrate surface due to electrostatic attractions between the AuNPs and cell wall surface of P. aeruginosa, instead of QS-related molecules. Positive charge AuNPs inhibited significantly S. aureus and P. aeruginosa biofilm formation (while minimizing mammalian cytotoxicity) (Ramasamy et al., 2016). The use of NPs demonstrates an exclusive approach to penetrate infectious biofilms and target bacterial communication, overcoming this major health issue related with biofilm infections.

Because most of these NPs-based platforms exert their action via distinct mechanisms/structures/pathways of those used by traditional antibiotics, combined therapeutic regimens are promising strategies to tackle the surge of multidrug resistant (MDR) bacteria bypassing their defense mechanisms (Pelgrift and Friedman, 2013; Singh K. et al., 2014; Hemeg, 2017; Zaidi et al., 2017). Additionally, NPs have been shown to activate macrophages in a dose dependent manner (Patel and Janjic, 2015) which promotes the host defenses (Hemeg, 2017; Jagtap et al., 2017).

This multi-target action of NPs may overcome multidrug resistance by circumventing several obstacles encountered by traditional antibiotics (Pelgrift and Friedman, 2013; Chen et al., 2014; Singh K. et al., 2014; Hemeg, 2017; Jagtap et al., 2017; Rai et al., 2017; Zaidi et al., 2017). **Table 1** highlights several types of NPs that have shown effective bactericidal activity when administered isolated; combined with standard antibiotics; and/or radiation or as vectors for biocidal delivery allowing killing of MDR bacteria, and in some cases also inhibiting biofilm production.

We will now focus on the different types of metal NPs highlighting their most relevant mechanism/effects against MDR bacteria and/or biofilms structures.

## SILVER NANOPARTICLES (AGNPS)

Since the ancient times, silver has been recognized as having antimicrobial effects (Rai et al., 2009; Reidy et al., 2013). Based on all the evidence to date, AgNPs are probably one of the most promising inorganic NPs that can be used for the treatment of bacterial infections (Natan and Banin, 2017). These NPs may be synthesized by traditional chemical reduction or via "green" chemistry approaches using plant and/or microbial extracts (Iravani et al., 2014; Ribeiro et al., 2018).

Several mechanisms have been proposed to understand how AgNPs mediate cell death, including cell wall disruption (Lok et al., 2007; Bondarenko et al., 2013), oxidation of cellular components, inactivation of the respiratory chain enzymes, production of ROS, and decomposition of the cellular components (Chen et al., 2014; Rizzello and Pompa, 2014; Dakal et al., 2016). The permeability of the membrane increases after incorporation of AgNPs into the cell membrane. The adsorption of the NPs leads to the depolarization of the cell wall, altering the negative charge of the cell wall to become more permeable. It was demonstrated disruption of the cell wall with subsequent penetration of the NPs. The entry of AgNPs induces ROS that will inhibit ATP production and DNA replication (Zhang et al., 2013; Dakal et al., 2016; Durán et al., 2016; Ramalingam et al., 2016). However, there is evidence that AgNPs can release Ag+, known to exhibit antimicrobial activity, when interacting with thiol-containing proteins, which weaken their functions (Durán et al., 2010). The precise method of the antibacterial mechanism of AgNPs is still not completely understood (Franci et al., 2015; Durán et al., 2016). All the existing data indicates that AgNPs exert several bactericidal mechanisms in parallel, which may explain why bacterial resistance to silver is rare (Karimi et al., 2016). Concerns regarding the cytotoxicity and genotoxicity of AgNPs have been raised (Chopra, 2007) but various authors have conducted clinical trials based on AgNPs and no important clinical alterations have been detected (Munger et al., 2014a,b; Smock et al., 2014). Interestingly, AgNPs have been found to exhibit higher antimicrobial activity than antibiotics like gentamicin or vancomycin against P. aeruginosa and MRSA (Saeb et al., 2014). Lara et al. showed the potential bactericidal effect of AgNPs against MDR P. aeruginosa, ampicillin-resistant E. coli O157:H7 and erythromycin-resistant Streptococcus pyogenes (Lara et al., 2010). Nagy et al., reported that AgNPs were capable of inhibiting the growth of S. aureus and E. coli via the up-regulation of the expression of several antioxidant genes and ATPase pumps (Nagy et al., 2011). Dolman et al. also showed that the Ag-containing Hydrofiber <sup>R</sup> dressing and nanocrystalline Ag-containing dressing are effective agents against antibiotic sensitive Gram-negative and -positive bacteria as well as antibiotic resistant bacteria, such as MRSA, Vancomycinresistant Enterococci (VRE) and Serratia marcescens, avoiding the formation of biofilms on biomaterials (Percival et al., 2007). Su and collaborators showed that AgNPs immobilized on the surface of nanoscale silicate platelets (AgNP/NSPs) have strong antibacterial activity against MRSA and silver-resistant E. coli via generation of ROS (Su et al., 2011). Singh and collaborators showed that AgNPs from P. amarus extract exhibited excellent antibacterial potential against MDR strains of P. aeruginosa (Singh K. et al., 2014). Recently, two different shaped AgNPs (spheres and rods) were used against Gram-positive and negative bacteria, both showing promising antibacterial activity against different strains (Acharya et al., 2018).

An emerging practice is to combine AgNPs with antibiotics to enhance antimicrobial potency. Recently, Katya and collaborators showed that the combination of gentamicin and chloramphenicol with AgNPs has a better antibacterial effect in MDR E. faecalis than both antibiotics alone (Katva et al., 2018). McShan et al. described that AgNPs combined with either one of two-different class of antibiotics (tetracycline and neomycin) can exhibit a synergistic effect, showing an enhanced antibacterial activity at concentrations below the MIC of either the NPs or the antibiotic (McShan et al., 2015). Other authors also reported similar results (Thomas et al., 2014; Panácek et ˇ al., 2016a,b; Salomoni et al., 2017). Djafari and collaborators described the synthesis of water-soluble AgNPs using the antibiotic tetracycline as co-reducing and stabilizing agent (AgNPs@TC) and demonstrated their effectiveness against tetracycline-resistant bacteria (Djafari et al., 2016).

Antimicrobial peptides (AMPs) represent one of the forms of defense strategy against infections in living organisms and are emerging as essential tools to kill pathogenic bacteria, since they exhibit broad-spectrum activity and low resistance development (Yeaman, 2003). Lytic peptides are AMPs produced by all organisms. In mammals, they are an innate host defense mechanism against pathogens (Bahar and Ren, 2013). The mechanism of action of AMPs relies on the ability to interact with bacterial membranes or the cell wall, thus inhibiting cellular biochemical pathways and ultimately killing the bacteria (Zhang and Gallo, 2016). Defensins and cathelicidin are two of the larger families of lytic peptides that kill bacteria by disrupting the membrane. Unfortunately, AMPs have poor enzymatic stability, low permeability across biological barriers and may be rapidly degraded in the human body by proteases, which greatly limits their application (Wang, 2014). Immobilization of the peptides onto NPs can increase their stability, enhancing the antimicrobial properties of the NPs and therefore, has the potential to be used as a new tool to tackle antibiotic resistant bacteria (Brandelli, 2012; Rai A. et al., 2016). Indeed, the first author to demonstrate that functionalized AgNPs with peptides increased their antibacterial activity was Ruden and co-workers (Ruden et al., 2009). Based on this strategy several researchers functionalized AgNPs with AMPs (AgNP@AMP) with increases in the antimicrobial activity compared with free AMPs (Ruden et al., 2009; Liu et al., 2013; Mohanty et al., 2013). Polymyxin B is the most used AMP and exhibits antibacterial activity via interaction with the endotoxin LPS in the outer membrane of Gram-negative bacteria (Morrison and Jacobs, 1976; Lambadi et al., 2015). It was proved that AgNPs functionalized with polymyxin-B removed almost completely endotoxins from solutions and hindered the formation of biofilm onto surgical blades (Jaiswal et al., 2015; Lambadi et al., 2015). Liu et al., demonstrated that the immobilization of peptides with AgNPs enhanced their antimicrobial activity compared to an unbound peptide and also minimized toxicity of AgNPs compared to using the AgNPs alone (Liu et al., 2013). A recent study by Pal et al. describes a system consisting of a cysteine containing AMP conjugated with AgNPs, which demonstrated that the Ag-S bonds increased stability and enhanced antimicrobial activity than conjugation using electrostatic interactions (Pal et al., 2016).

Other methods have been used to improve the antibacterial activity of AgNPs. One of these methods relies on the use of visible blue light, which was previously shown to exhibit strong antibacterial activity (Dai T. et al., 2013; Maclean et al., 2014). El Din and collaborators demonstrated that blue light combined with AgNPs exhibits therapeutic potential to treat MDR infections and can represent an alternative to conventional antibiotic therapy, since the antimicrobial activity of the combination was greater than the components alone. Moreover, this approach proved to be synergistic in the treatment of an unresponsive antibiotic-resistant bacteria responsible for a wound in a horse (El Din et al., 2016). Spherical shaped thioglycolic acid-stabilized AgNPs (TGA-AgNPs) conjugated with vancomycin were used as drug delivery systems and demonstrated to possess increased antimicrobial activity against MDR bacteria such as MRSA and VRE (Esmaeillou et al., 2017).

## GOLD NANOPARTICLES (AuNPs)

Metallic gold is considered inert and non-toxic, which may vary when it shifts form metallic bulk to oxidation states (I and II) (Merchant, 1998). Gold NPs (AuNPs) may be synthesized by traditional chemical reduction of a gold salt or via "green" chemistry approaches using plant and/or microbial extracts (Shah et al., 2014). The most used and described method is the chemical synthesis based on the reduction of chloroauric acid by citrate (Lee and Meisel, 1982; Fernandes and Baptista, 2017). Some studies have addressed the potential of using AuNPs as antibacterial agents, but some controversy still exists (Cui et al., 2012; Bresee et al., 2014; Shah et al., 2014; Shareena Dasari et al., 2015; Zhang et al., 2015a; Shamaila et al., 2016).

According to Yu H and collaborators, AuNPs are usually not bactericidal at low concentrations and weakly bactericidal at high concentrations (Shareena Dasari et al., 2015; Zhang et al., 2015a). This is possibly due to the effect of co-existing chemicals, such as gold ions, surface coating agents, and chemicals involved in the synthesis that were not completely removed (Shareena Dasari et al., 2015; Zhang et al., 2015a). However, other authors describe that the antibacterial mechanism of AuNPs is associated to (i) the collapse in the membrane potential, hindering ATPase activity causing a deterioration of the cell metabolism; (ii) hindering of the binding subunit of the ribosome to tRNA (Cui et al., 2012); and (iii) Shamaila and co-workers showed that AuNPs may affect the bacterial respiratory chain by attacking nicotinamide (Shamaila et al., 2016). Since AuNPs are non-toxic to the host (Conde et al., 2014; Li et al., 2014; Rajchakit and Sarojini, 2017), the possibility of fine tuning their conjugation chemistry to act as carriers or delivery vehicles of antibiotics or other antibacterial moieties may enhance their bactericidal effect and potentiate the effect of antibiotics (Zhao and Jiang, 2013; Conde et al., 2014; Li et al., 2014; Uma Suganya et al., 2015; Zhang et al., 2015a; Fernandes et al., 2017).

Cationic and hydrophobic functionalized AuNPs were shown to be effective against both Gram-negative and positive uropathogens, including MRSA. These AuNPs exhibited low toxicity to mammalian cells (biocompatibility) and the development of resistance to these NPs was very low (Li et al., 2014). Vinoj et al. demonstrated that coating AuNPs with N-acylated homoserine lactonase proteins (AiiA AuNPs) resulted in a nanocomposite with activity against MDR species compared with AiiA proteins alone (Vinoj et al., 2015). Other approaches were also studied, as adsorbing AuNPs to PVAlysozyme micro bubbles potentiate the antibacterial activity due to the interaction of AuNPs with cells membranes causing bacterial lysis (Mahalingam et al., 2015). Galic acid capped AuNPs have also been found to be active against Gram-negative and -positive bacteria (Kim D. et al., 2017). Recently, Ramasamy and collaborators described the direct one-pot synthesis of cinnamaldehyde immobilized on gold nanoparticles (CGNPs) with effective biofilm inhibition of more than 80% against Grampositive bacteria (methicillin-sensitive and -resistant strains of S. aureus, MSSA and MRSA, respectively) and Gram-negative (E. coli and P. aeruginosa) in vitro and in vivo (Ramasamy et al., 2017a,b). Also, the integration of AuNPs with ultrathin graphitic carbon nitride was described as having high bactericidal performance against both MDR Gram-negative and -positive bacteria, and a high effectiveness in eliminating existing MDRbiofilms and preventing the formation of new biofilms in vitro (Wang Z. et al., 2017). Also, conjugation of antibiotics to AuNPs, such as vancomycin, methicillin, etc., increases their intrinsic activity against MDR strains (Mohammed Fayaz et al., 2011; Lai et al., 2015; Roshmi et al., 2015; Payne et al., 2016). Recently Payne

and collaborators develop a single-step synthesis of kanamycincapped AuNPs (Kan-AuNPs) with high antibacterial activity against both Gram-positive and -negative bacteria, including kanamycin resistant bacteria. The authors observed a significant reduction in the MIC against all the bacterial strains tested for Kan-AuNPs when compared to the free drug. This higher efficacy was due to the disruption of the bacterial envelope, resulting in leakage of the cytoplasmic content and consequent cell death (Payne et al., 2016). Pradeepa and collaborators synthesized AuNPs with bacterial exopolysaccharide (EPS) and functionalized them with antibiotics (levofloxacin, cefotaxime, ceftriaxone and ciprofloxacin). They observed that these AuNPs exhibited excellent bactericidal activity against MDR Grampositive and -negative bacteria compared to free drugs. E. coli was the most susceptible MDR bacteria followed by K. pneumoniae and S. aureus (Pradeepa et al., 2016). Recently, Yang and collaborators described the effect of small molecule (6-aminopenicillanic acid, APA)-coated AuNPs to inhibit MDR bacteria (Yang et al., 2017). They doped AuNPs into electrospun fibers of poly(ε-caprolactone) (PCL)/gelatin to produce materials that avoid wound infection by MDR bacteria and demonstrated in vitro and in vivo that APA-AuNPs reduced MDR bacterial infections (Yang et al., 2017). Shaker and Shaaban evaluated the surface functionalization of AuNPs with carbapenems [imipenem (Ipm) and meropenem (Mem)] as a delivering strategy against carbapenem resistant Gram-negative bacteria isolated from an infected human. Both Ipm-AuNPs and Mem-AuNPs, with 35 nm diameter showed a significant increase in antibacterial activity against all the tested isolates (Shaker and Shaaban, 2017). Also, Shaikh and collaborators described recently the synthesis and characterization of cefotaxime conjugated AuNPs to target drug-resistant CTX-M-producing bacteria. The authors could invert resistance in cefotaxime resistant bacterial strains (i.e., E. coli and K. pneumoniae) by using cefotaxime-AuNPs. Their results reinforce the efficacy of conjugating an unresponsive antibiotic with AuNPs to restore its efficacy against otherwise resistant bacterial pathogens (Shaikh et al., 2017).

Combination of AuNPs with other approaches has also been demonstrated. Indeed, one of the most extraordinary properties of AuNPs is the capability to transform light into heat under laser irradiation (Mendes et al., 2017; Mocan et al., 2017). This property is extremely important because it can be exploited to develop photothermal nanovectors to destroy MDR bacteria at a molecular level (for a complete review see Mocan et al., 2017). For example, Khan and collaborators showed that the combination of Concanavalin-A (ConA) directed dextran capped AuNPs (GNPDEX-ConA) conjugated with methylene blue (MB) (MB@GNPDEX-ConA) and photodynamic therapy (PDT) enhanced the efficacy and selectivity of MB induced killing of MDR clinical isolates, including E. coli, K. pneumoniae, and E. cloacae (Khan et al., 2017). Gil-Tomas and collaborators described that the functionalization of AuNPs covalently with toluidine blue O– tiopronin forms an enhanced, exceptionally potent antimicrobial agent when activated by white light or 632 nm laser light (Gil-Tomás et al., 2007). Hu and collaborators prepared a mixed charged zwitterion-modified AuNPs consisting of a weak electrolytic 11-mercaptoundecanoic acid (HS-C10-COOH) and a strong electrolytic (10-mercaptodecyl)trimethylammonium bromide (HS-C10-N4) that exhibited in vivo and under nearinfrared (NIR) light irradiation an enhanced photothermal ablation of MRSA biofilm with no damage to the healthy tissues around the biofilm (Hu et al., 2017). Also, the antibacterial activity of glucosamine-gold nanoparticle-graphene oxide (GlcN-AuNP-GO) and UV-irradiated GlcN-AuNP-GO was evaluated against E. coli and E. faecalis. Results show that UV irradiation of GlcN-AuNP-GO results in higher antibacterial activity than the standard drug kanamycin (Govindaraju et al., 2016). Ocsoy et al. reported the development of DNA aptamerfunctionalized AuNPs (Apt@AuNPs) and gold nanorods (Apt@AuNRs) for inactivation of MRSA with targeted PTT (Ocsoy et al., 2017). The authors showed that although both NPs could specifically bind to MRSA cells, Apt@AuNPs and Apt@AuNRs increased resistant cell death for 5% and for more than 95%, respectively through PTT. This difference in induction of cell death was based on the relatively high longitudinal absorption of NIR radiation and strong photothermal conversion capability for the Apt@AuNRs compared to the Apt@AuNPs. However, with the new developments of using AuNPs for hyperthermia in the visible light (Mendes et al., 2017) might additionally potentiate the Apt@AuNPs results observed for these authors (Ocsoy et al., 2017). Recently, Mocan et al. also described the synthesis of AuNPs by wet chemistry, their functionalization with IgG molecules following laser irradiation. Their results indicate that administration of IgG-AuNPs following laser irradiation provided an extended and selective bacterial death in a dose dependent manner (Mocan et al., 2016).

In recent years, a new approach relying on the conjugation of AuNPs with AMPs has shown interesting results (Rajchakit and Sarojini, 2017). Indeed, Kuo and collaborators mixed syntheticpeptides containing arginine, tryptophan and cysteine termini [(DVFLG)2REEW4C and (DVFLG)2REEW2C], with aqueous tetrachloroauric acid to generate peptide-immobilized AuNPs [i.e., (DVFLG)2REEW4C-AuNPs and (DVFLG)2REEW2C-AuNPs] that were effective antibacterial agents against Staphylococci, Enterococci, and antibiotic-resistant bacterial strains (Kuo et al., 2016). Rai and co-workers demonstrated that the use of cecropin-melittin (CM-SH) a known peptide with antibacterial properties (Boman et al., 1989), functionalized in the surface of AuNPs through Au-S bond, showed higher antimicrobial activity and higher stability in media compared with an in vitro and in vivo infection animal model with the MIC of CM-SH AuNPs four times lower than free CM-SH (Rai A. et al., 2016). Conjugation of AMP with AuNPs usually involves the formation of the Au-S coordinate covalent bond, relying on the amine or thiol groups in peptides or conjugating specific linkers to AMPs with a terminal (N- or C-terminal) cysteine which helps conjugation with gold (Tielens and Santos, 2010; Xue et al., 2014). However, there is one example where covalent conjugation of an AMP to AuNPs has been achieved via Au-O bond (Lai et al., 2015). Other approaches use a linker for the conjugation to AuNPs, Poly(ethylene glycol) carboxylic acid (PEGCOOH) covalently bound to AMP showed a significantly increase of the antibacterial and antibiofilm activity for resistant Gram-negative bacteria (Casciaro et al., 2017). Yeom and co-workers demonstrated the most advanced in vivo clinical application for AuNPs@AMP using infected mice and resulting in the inhibition of Salmonella Typhimurium colonization in the organs of the animals (Yeom et al., 2016). The reason behind the increased antimicrobial activity of AuNPs@AMP over the free components is that AuNPs can get a higher concentration of the antibiotic at the site of action. These NPs can interact with LPS, proteins in the membrane of the bacteria and in some cases, penetrate the bacterial membrane through the porin channel. This way they can interact with the inner membrane making the AuNPs@AMP conjugate more efficient than the non-conjugated form (Katz and Willner, 2004; Wangoo et al., 2008; Chen J. et al., 2009).

## BIMETALLIC NPS

Ag and Au may be used in a single NP to enhance the effects of a drug and reduce the required dose. Alternatively, they can be used alone since they possess antimicrobial properties that are enhanced when combined in the form of bimetallic NPs (Arvizo et al., 2010; Singh R. et al., 2016). The role of Ag against MDR pathogens has been previously described. However, AgNPs are difficult to functionalize with biomolecules and drugs. Such limitation may be circumvented by means of alloy/bimetallic NPs that excel their monometallic counterparts providing improved electronic, optical and catalytic properties (Cho et al., 2005; Shah et al., 2012). As reported above, AuNPs constitute good vectors to the delivery of pharmacologic compounds. Gold(Au)-silver(Ag) alloys are an optimal solution since they combine the antimicrobial effect of silver with the ease of functionalization and improved stability in complex biological media provided by gold (Doria et al., 2010; dos Santos et al., 2012). Fakhri and co-workers synthetized and functionalized AgAuNPs with a tetracycline and concluded that there exists a synergetic effect of the antibiotic with the bimetallic nanoparticle, with greater bactericidal activity of this form in detriment of its free forms. The mechanism of action was established as being the generation of ROS (Fakhri et al., 2017). Also recently, Baker and collaborators described the synthesis and antimicrobial activity of bimetallic AgAuNPs from the cell free supernatant of Pseudomonas veronii strain AS41G inhabiting Annona squamosa L. The authors showed their synergistic effect with standard antibiotics with 87.5, 18.5, 11.15, 10, 9.7, and 9.4% fold increased activity with bacitracin, kanamycin, gentamicin, streptomycin, erythromycin and chloramphenicol, respectively, against bacitracin resistant strains of Bacillus subtilis, E. coli, and K. pneumoniae (Baker et al., 2017). Zhao and collaborators have demonstrated the antibacterial activity of AuPtNPs bimetallic NPs against sensitive and drug-resistant bacteria via the dissipation of the bacterial membrane potential and the elevation of adenosine triphosphate (ATP) levels (Zhao et al., 2014).

Other types of bimetallic NPs have been studied and their antibacterial activity explored, but in most cases as coating agents and not as a delivery approach and antibacterial activity (Argueta-Figueroa et al., 2014).

## METAL OXIDES

Metal oxides NPs are among one of the most explored and studied family of NPs and are known to effectively inhibit the growth of a wide range of sensitive and resistant Grampositive and -negative bacteria, emerging as hopeful candidates to challenge antimicrobial resistance (Raghunath and Perumal, 2017; Reshma et al., 2017; Kadiyala et al., 2018). Iron oxide (Fe3O4), Zinc oxide (ZnO), and Copper oxide (CuO) possess antimicrobial properties and can be applied in clinical care (Sinha et al., 2011). Due to the intrinsic photocatalytic activity of the metal oxides they generate ROS and become powerful agents against bacteria (Tong et al., 2013; Singh R. et al., 2014). These will be described in more detail on the following sections.

## IRON OXIDE (FE3O4)

The synthesis of iron oxide NPs may be achieved via different routes (Babes et al., 1999; Berry and Curtis, 2003). The antibacterial mechanism of these NPs is mainly attributed to dissolved metal ions and the generation of ROS (Wang et al., 2017a). It was shown that superparamagnetic iron oxide NPs interact with microbial cells by penetrating the membrane and interfering with the electron transfer (Behera et al., 2012; El-Zowalaty et al., 2015). Additionally, it has been described that iron oxide NPs can damage macromolecules, including DNA and proteins, through the formation of ROS (Leuba et al., 2013). Pan et al. developed a system of reduced graphene oxide (rGO)-iron oxide nanoparticles (rGO-IONP) by the chemical deposition of Fe2+/Fe3<sup>+</sup> ions on nanosheets of rGO in aqueous ammonia. The in vivo results showed maximum antibacterial activity due to the generation of hydroxyl radicals that can cause physical and chemical damage, which inactivated MRSA (Pan et al., 2016).

## ZINC OXIDE (ZnO)

ZnO NPs are often used to restrict microorganism growth, being effective against planktonic bacteria, and also inhibiting the formation of biofilms (Hsueh et al., 2015; Sarwar et al., 2016) (Espitia et al., 2012). These NPs can be synthesized by various methods, from green chemistry to sonochemistry (Salem et al., 2015; Ali et al., 2016; Nagvenkar et al., 2016). The antibacterial mechanism of the NPs is partially attributed to two principal factors, the dissolution of metal ion and the generation of ROS (Gelabert et al., 2016; Nagvenkar et al., 2016; Sarwar et al., 2016). ZnO releases Zn2<sup>+</sup> in liquid medium and is adsorbed on the surface of bacteria or may entry the cell, where it interacts with functional groups in proteins and nucleic acids, hindering enzyme activity and the normal physiological processes (Yu et al., 2014). However, some authors demonstrated that Zn ions have little antimicrobial activity, implying that dissolution of Zn2<sup>+</sup> might not be the main mechanism of action (Aydin Sevinç and Hanley, 2010). Sarwar and coworkers demonstrated that nanosized ZnO caused significant oxidative stress to Vibrio cholera, the damage inflicted was DNA degradation, protein leakage, membrane depolarization and fluidity (Sarwar et al., 2016). Ehsan and Sajjad, described that ZnO NPs impregnated with antibiotics showed good antibacterial activities against S. aureus, Proteus, Acinetobacter, P. aeruginosa, and E. coli, being that these were resistant to antibiotics but became sensitive in the presence of these NPs with antibiotics (Ehsan and Sajjad, 2017). It was also discovered that these NPs induce the production of ROS even in the dark, and this happens due to the surface defects on the NPs. The different shapes function as enzyme inhibitors, where nanopyramids are the most effective (Cha et al., 2015; Lakshmi Prasanna and Vijayaraghavan, 2015). Recently, Aswathanarayan and Vittal described the antimicrobial effect of ZnO NPs against MDR Gram-positive and -negative pathogens in comparison to gold and iron NPs and these could be used at concentrations less toxic to mammalian cells (Aswathanarayan and Vittal, 2017). ZnO NPs are also known for inhibiting biofilm formation and production of quorum-sensing-dependent virulence factors in P. aeruginosa (Lee et al., 2014; García-Lara et al., 2015).

## COPPER OXIDE (CuO)

Copper containing NPs have been shown effective against animal and plant pathogens (LewisOscar et al., 2015), impeding formation of MDR biofilms, and showing the potential to serve as antimicrobial coating agents (LewisOscar et al., 2015). Kruk et al. and Zhang et al. showed that copper NPs are capable of inhibiting the growth of MDR bacteria, namely, P. aeruginosa and MRSA (Zhang et al., 2014, 2015b; Kruk et al., 2015). The antimicrobial activity of these NPs is comparable to that of AgNPs but at a lower cost (Kruk et al., 2015). Copper oxide NPs generate ROS that often leads to chromosomal DNA degradation, which seems to highlight a "particle-specific" action rather than resulting from the release of metallic ions (Chakraborty et al., 2015). Su and collaborators investigated the effects of CuONPs on bacterial denitrification and explored the effect on the expression of intracellular proteins. When CuONPs entry into bacteria metabolic functions are affected, such as active transport, electron transfer, and nitrogen metabolism (Su et al., 2015).

NPs can also be complexed with other metals, like gallium. Gallium NPs have been described to facilitate phagosome maturation of macrophages infected with virulent M. tuberculosis and therefore being able to inhibit growth of this pathogen (Choi et al., 2017).

## THE POTENTIAL FOR NANOTHERANOSTICS

NPs applications in biodetection is huge and more insights on pathogen detection using NPs platforms can be seen in Veigas et al. (2013, 2014, 2015); Costa et al. (2014); Weng et al. (2015); Kim J. et al. (2017); Wang et al. (2017b); Galvan and Yu (2018), and Yang et al. (2018).

Theranostics is a combination of diagnosis and therapy onto a single platform, which allow for timely biodetection and/or real-time monitoring of therapy. By using NPs, this can be translated to the nanoscale—Nanotheranostics. NPs have been applied for multiplex high-throughput diagnostics to assist precision therapy. For example, Verigene <sup>R</sup> is an AuNPs test commercialized for diagnosis. It is an automated microarraybased system that identifies Gram-negative pathogens from positive blood cultures. Verigene <sup>R</sup> BC-GN also detects key resistance mechanisms (Walker et al., 2016; Claeys et al., 2018). Others have used, magnetic and functionalized magnetic iron oxide NPs as affinity probes to capture Grampositive and -negative bacteria. The analyses of captured bacteria using matrix-assisted laser desorption/ionization mass spectrometry was <1 h (Reddy P. M. et al., 2014). One pioneer work on nanotheranostics against bacterial infection was the development of a method for in vivo photoacoustic detection and photothermal eradication of S. aureus. Twocolor gold and multilayer magnetic nanoparticles were functionalized with an antibody cocktail for the targeting of S. aureus. These platform demonstrated ultrasensitive detections for circulating bacterial cells (CBCs), in vivo magnetic enrichment and PT eradication of CBCs (Galanzha et al., 2012). Recently, Zhou and collaborators developed a silicon 2,3-naphthalocyanine dihydroxide (Nc) and Vancomycin functionalized silica-encapsulated, silver-coated gold NPs (Au@AgNP@SiO2@Nc-Van) as a novel theranostic system for surface-enhanced Raman scattering (SERS) detection and antimicrobial photodynamic therapy (aPDT) of vancomycin (Van)-resistant enterococci (VRE) strains (Zhou et al., 2018). These authors observed a 4–5 logs reduction of bacteria upon in vitro aPDT of VRE treated with a nanomolar concentration of the Au@AgNP@SiO2@Nc-Van and an infection regression and even complete eradication of VRE in vivo using infected mice (Zhou et al., 2018).

A selenium nanoplatform (Se@PEP-Ru) was designed with excellent fluorescent properties for imaging bacteria and with high antimicrobial properties (Huang et al., 2017). Zhao and co-workers developed an activated theranostics nanoprobe for near-infrared fluorescence imaging and photothermal therapy of MRSA infections, based on SiO2/PAH-cypate nanosystems modified with PEG and Vancomycin-conjugated poly(acrylic acid) molecules (PAAPEG-Van). This probe is activated by bacteria-responsive polyelectrolyte dissociation from silica NPs. The authors believe that this concept can be used as an approach to design and for production of bacteria responsive multifunctional nanomaterials and constitute their ultimate functions in the treatment of drug-resistant bacterial infections (Zhao et al., 2017). Kuo and collaborators developed a nanotheranostics system using Au nanorods conjugated with a hydrophilic photosensitizer, toluidine blue O, that acted as dual-function agents in photodynamic inactivation and hyperthermia against MRSA (Kuo et al., 2009).

## CLINICAL TRANSLATION

At present, there are a few metal NPs-based strategies against bacterial infections undergoing clinical trials. The costs associated to the use of nanotechnology platforms are very high, and therefore conventional treatments are preferred. However, these platforms might be preferable in specific situations, with direct impact on the quality of patients life (Caster et al., 2017).

Bio-kil <sup>R</sup> [3-(Trimethoxysilyl) propyloctadecyldimethyl ammonium chloride] (Cargico Group, Taiwan) is a patented technology that is based on affixing nano-sized antimicrobials onto a large surface area through covalent chemical bonding to form a durable polymer. Bio-kil <sup>R</sup> eliminates microorganism through a physical biocide process. This type of nanomaterial consists in inorganic metal components and organic quaternary ammonium components. Recently, Bio-Kil <sup>R</sup> has been shown to reduce the environmental bacterial burden and MDR organisms (Lee et al., 2017).

AgTive (NCT00337714) is a silver-impregnated central venous catheter and has been marketed with the claim to improved bactericidal activity. AgTive catheters are made of polyurethanes impregnated with silver NPs, and their interaction with body fluids and intravenous solutions results in the release of significantly larger amounts of silver ions from the catheter reducing bloodstream infection (Antonelli et al., 2012).

Acticoat is a nanocrystalline silver dressing that acts as an antimicrobial topical, releasing silver into the wound. This nanoformulation has been shown to inhibit in vitro biofilms formation in P. aeruginosa and Acinetobacter baumannii by more than 90% (Potgieter and Meidany, 2017). Madigan Army Medical Center is studying the efficacy of a silver NPs gel SilvaSorb (NCT00659204) and currently is in phase III of the clinical trials. The aim of this study is to compare the antimicrobial efficacy of a one-time application of SilverSorb (AcryMed, Inc., Portland) against the standard antibacterial hand gel Purell (GoJo Industries, Akron), in reducing transient bacterial counts isolated from the hands of 40 patients seeded with S. marcescens.

Nano Silver Fluoride is a new formulation that combines silver NPs, chitosan and fluoride and was developed with antimicrobial properties. This nanoformulation has excellent results as antibacterial agent against S. mutans and Lactobacilli. Currently, is used to prevent dental caries in children (Dos Santos et al., 2014).

Despite this review do not concern liposomal formulations since it refers to clinical translation other formulations involving NPs, such as liposomal formulations, have been also identified as antimicrobial agents. Most of these formulations rely on the incorporation of traditional antibiotics into nanoliposomes to improve distribution and circulation times (Caster et al., 2017). **Table 2** summarizes antimicrobial liposomes, which are undergoing clinical trials. For example, Amikacin (NCT01315691) is a potent aminoglycoside antibiotic that is useful for the treatment of MDR Gram-negative bacteria. Arikace is an inhaled liposomal formulation that encapsulates amikacin composed of dipalmitoyl-phosphatidylcholine (DPPC) and cholesterol (Meers et al., 2008). These formulation have high drug loading and stability when administrated and in phase II trial, there was no notable difference in toxicity between liposomal drug treatment and placebo (Clancy et al., 2013). Another two-inhaled liposomal formulation are currently in clinical trials. Linhaliq (NCT02104245) is a combination of liposomal and aqueous phase ciprofloxacin, whereas Lipoquin (NCT00889967) is a liposomal ciprofloxacin that allows prolonged drug release. Both of these nanoformulation were developed for the treatment of non-cystic fibrosis bronchiectasis (NCFBE) patients with chronic lung infections with P. aeruginosa. Phase II in patients with both CF and non-CF bronchiectasis have been completed. After analysis of clinical data from the two different formulations, Linhaliq showed better performance. The Food and Drug Administration (FDA) has designated Linhaliq as a qualified infection disease product and made it eligible for Fast track designation. In 2016, Pulmanic completed two phases III clinical trials, but has not yet been approved by the FDA. The Hadassah Medical Organization (Jerusalem, Israel) has incorporated quaternary ammonium polyethyleneimine (QA-PEI) based polymers into dental composites. The bacterial membrane may be disturbed by the charged quaternary moiety, it also has potent activity against a series of Gram-positive and -negative pathogens (Ortega et al., 2015). In 2013, these nanoformulation completed phase II trials but no data on outcome have been released to date.

MAT2501 is designed to targeted delivery of the antibiotic amikacin while providing an improved safety and tolerability profile. Currently, Matinas Biopharma has reported positive data from the Phase I study in healthy volunteers for the treatment of MDR Gram-negative bacterial infections and is in preparation for a phase II in patients.

## OTHER POTENTIAL APPLICATIONS OF NPS

In the case of non-antibiotic therapy, combinations of NPs with essential oils, peptides and other natural compounds have featured as promising antimicrobial strategies. The therapeutic applications of these substances are often limited by their toxicity and volatility (Chen F. et al., 2009; Allahverdiyev et al., 2011). A recent study has shown that chitosan NPs vectors, modified with eugenol and carvacrol essential oils on their surface, were active against E. coli and S. aureus at concentrations better or equal to unmodified NPs versions (Chen F. et al., 2009). Furthermore, the toxicity of the conjugates toward mouse fibroblasts was significantly less than the pure oils alone. With regards to peptides, the active sequences can be vulnerable to denaturation, aggregation or hydrolysis within end products or in the human body. Colloidal systems containing NPs are at the forefront of peptide research, as they can be designed to encapsulate and protect peptides during biological transit. Water in oil micelles have been successfully used to increase the potency of antimicrobial peptides against E. coli (Gontsarik et al., 2016). In another example, liposomes have been used to improve the stability of encapsulated nisin against pH and temperature extremes thereby increasing its potential in food processing (Taylor et al., 2007). Popular NPs vehicle materials


TABLE 2 | Antimicrobial liposomal nanoformulation in clinical development.

for peptides include phytoglycogen NPs (Bi et al., 2011), chitosan (Wu et al., 2017), pectin (Krivorotova et al., 2017), and alginate (Khaksar et al., 2014).

NPs have also been applied with tremendous success in biodetection systems, namely as sensors and diagnostics platforms with increased sensitivity and selectivity. Due to the decrease in size of the transduction mechanisms provided by NPs, most of these platforms have found applications at pointof-need and/or point-of-care (Costa et al., 2014; Veigas et al., 2014; Weng et al., 2015; Kim J. et al., 2017; Wang et al., 2017b; Galvan and Yu, 2018; Yang et al., 2018). In some cases, diagnostics/sensing and therapeutic properties have been combined onto single NPs, providing for innovative tools – Nanotheranostics. Recently, several nanotheranostics strategies against bacteria have been described (Kuo et al., 2009; DeGrasse, 2012; Dai X. et al., 2013; Khlebtsov et al., 2013; Kim et al., 2013; Gamella et al., 2014; Pei et al., 2014; Setyawati et al., 2014; Patel and Janjic, 2015; Thompson et al., 2015; Jagtap et al., 2017; Mocan et al., 2017; Zhao et al., 2017).

### BOTTLENECKS AND FUTURE CHALLENGES OF NPS

Despite the foreseen potential of NPs for medical applications, there are still several bottlenecks related with their acute and long-term exposure in humans. Several routes of exposure must be considered when evaluating NPs exposure, such as oral and gastrointestinal tract, dermal, respiratory system, and endovenous administration directly to the bloodstream (De Matteis, 2017). It is well known also that the physicochemical properties of NPs (e.g., size, shape and surface chemistry) affect their interaction with biological systems, influencing cellular uptake, pharmacokinetics, biodistribution, all of them with direct impact on final biological effects (for recent reviews see Bakand and Hayes, 2016; Xia et al., 2016; De Matteis, 2017; Warheit, 2018). These aspects have been addressed over the past years via the evaluation of the in vitro and in vivo toxicity of metal and metal oxide NPs (Dobrovolskaia et al., 2007; Asharani et al., 2010; Li et al., 2010; Baek and An, 2011; Hackenberg et al., 2011; Conde et al., 2012, 2014; Bondarenko et al., 2013; Ivask et al., 2014; Larsen et al., 2016; Sukwong et al., 2016; Rai et al., 2017), whose conclusions concerning their nanosafety differ depending on the type of assessment. This poses a major concern to effectively draw critical conclusions on NPs safety due to the vast number of different types/shapes/surface modified nanoparticles, the different methods used to evaluate their safety and environmental effects, and also by the fact most of these in vitro/in vivo studies present acute studies rather than long-term exposure (Bakand and Hayes, 2016; Xia et al., 2016; De Matteis, 2017; Warheit, 2018). Nevertheless, these in vivo and in vitro studies have been providing clues to the specific mechanisms by which NPs trigger an adverse effect enabling future surface modification of NPs to make them safer and less toxic (De Matteis, 2017). These concerns relating to nanosafety have been addressed and implemented via European Commission FP7 and H2020 sponsored programs followed by some relevant conclusions issued by the US National Academy of Science Committee on Research Progress of Environmental Health and Safety Aspects of Engineered Nanomaterials (Warheit, 2018).

Due to the 3Rs (Replacement, Reduction and Refinement) policies of in vivo studies, the future challenge of Regulatory Agencies is the standardization of the in vitro methodologies to establish the toxicology profile of NPs based on good laboratory practice (GLP) and the construction of flexible and reliable databases in which NPs are classified according to the data derived from these toxicological investigations. Together, these efforts might provide information on the dosage at which a particular NP can be considered safe and thus appropriate for medical use.

## CHALLENGES OF CURRENT RESEARCH

As mentioned above, nanomaterials have great potential to prevent and treat bacterial infection, but several challenges remain for the translation to the clinics. Some of these include assessing the interactions of nanoantibiotics with cells, tissues and organs, for dose recalibration and identification of appropriate routes of administration (Sandhiya et al., 2009). The biocompatibility of NPs is generally evaluated through in vitro assays, using cell culture. Because NPs, used as antimicrobial agents can enter through skin contact, ingestion, inhalation, oral and intravenous injection, in vivo models must also be applied to better understand their effects, including potential toxicity, clearance and metabolism (Beyth et al., 2015). Several studies have shown that intravenously injected NPs accumulate in the colon, lung, bone marrow, liver, spleen and lymphatics (Hagens et al., 2007). Inhalation has also been shown to cause cytotoxicity at the lung, and in the liver, heart and spleen through systemic circulation (Poma and Di Giorgio, 2008; Leucuta, 2013). This is of particular relevance for small NPs because of efficient cellular uptake and transcytosis across epithelial and endothelial cells into the blood and lymphatic circulation. Several NPs systems have demonstrated toxicity in multiple organs, such as free radical-mediated oxidative stress generated by the interaction of antimicrobial NPs with cell components that can result in hepatotoxicity and nephrotoxicity (De Jong and Borm, 2008; Lei et al., 2008).

The effective translation to the clinics will require appropriate guidelines for production and scale-up of manufacturing these nanomaterials, for characterization of the physicochemical properties and their impact on biocompatibility, for standardization of nanotoxicology assays and protocols to assist easy comparison of data originating from in vitro and in vivo studies, for the evaluation of their metabolism and mode of action (Duncan and Gaspar, 2011; Bertrand and Leroux, 2012; Beyth et al., 2015; Cordeiro et al., 2016; Rai M. et al., 2016; Zazo et al., 2016). Finally, the community still needs to address the economic impact of translation of these nanomaterials to the clinics.

### CONCLUSIONS

Given their vast therapeutic potential, it is becoming increasingly important to understand the mechanisms by which NPs complexes can impact bacterial viability. While one of the beneficial aspects of NPs drug carriers involves "macrotargeting," i.e., specific delivery to the site of infection, understanding the "micro-targeting" of bacterial mechanisms is imperative for the widespread future use of these vectors. Their impact of cell functions such as cell wall permeability,

### REFERENCES


efflux activity, formation of reactive species, and inhibition of essential cellular metabolism and reproduction is of utmost importance.

### AUTHOR CONTRIBUTIONS

PB supervision and correction of Nanoparticles part of the manuscript; MPM supervision and correction of MDR bacteria part of the manuscript; AC draft Nanoparticles part of the manuscript, figure draw, tables design; DF draft MDR bacteria part of the manuscript; NM draft MDR bacteria part of the manuscript; MM coordination of MDR bacteria part of the manuscript and final correction and integration; AF coordination of Nanoparticles part of the manuscript and final correction and integration.

### ACKNOWLEDGMENTS

This work was supported by the Unidade de Ciências Biomoleculares Aplicadas-UCIBIO, which is financed by National Funds from FCT/MEC (PTDC\_CVT-EPI\_6685\_2014; UID/Multi/04378/2013) and co-financed by the ERDF under the PT2020 Partnership Agreement (POCI-01-0145- FEDER-007728). DF is funded by the Trinity College Dublin Postgraduate Research (1252) Studentship. NM is funded by the Irish Research Council under the employment-based programme EBPPG/2015/233 in conjunction with Nuritas limited. We would like to acknowledge Ana Sofia Santos for preliminary revision of the manuscript.


subtilis, and Streptococcus aureus. Sci. Total Environ. 409, 1603–1608. doi: 10.1016/j.scitotenv.2011.01.014


nanoparticles and their antibacterial activity. ACS Nano 9, 9097–9105. doi: 10.1021/acsnano.5b03247


functional impairment in human mesenchymal stem cells. Toxicol. Lett. 201, 27–33. doi: 10.1016/j.toxlet.2010.12.001


low concentrations showing no cytotoxic effect. Molecules 21, 1–17. doi: 10.3390/molecules21010026


aureus ATCC 24213 induced by CuO-NPs. Environ. Sci. Pollut. Res. 24, 22048–22060. doi: 10.1007/s11356-017-9718-6


**Conflict of Interest Statement:** NM was employed by the company Nuritas limited.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Baptista, McCusker, Carvalho, Ferreira, Mohan, Martins and Fernandes. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# In Vitro Antimicrobial Activity of Green Synthesized Silver Nanoparticles Against Selected Gram-negative Foodborne Pathogens

Yuet Ying Loo<sup>1</sup> \*, Yaya Rukayadi<sup>1</sup> , Mahmud-Ab-Rashid Nor-Khaizura<sup>1</sup> , Chee Hao Kuan<sup>2</sup> , Buong Woei Chieng3,4, Mitsuaki Nishibuchi<sup>5</sup> and Son Radu1,6

<sup>1</sup> Department of Food Science, Faculty of Food Science and Technology, Universiti Putra Malaysia, Selangor, Malaysia, <sup>2</sup> Department of Agricultural and Food Science, Faculty of Science, Universiti Tunku Abdul Rahman, Kampar, Malaysia, <sup>3</sup> Department of Chemistry, Faculty of Science, Universiti Putra Malaysia, Selangor, Malaysia, <sup>4</sup> Materials Processing and Technology Laboratory, Institute of Advanced Technology, Universiti Putra Malaysia, Selangor, Malaysia, <sup>5</sup> Center for Southeast Asian Studies, Kyoto University, Kyoto, Japan, <sup>6</sup> Laboratory of Food Safety and Food Integrity, Institute of Tropical Agriculture and Food Security (ITAFoS), Universiti Putra Malaysia, Selangor, Malaysia

### Edited by:

Rebecca Thombre, Pune University, India

### Reviewed by:

Leda Giannuzzi, National University of La Plata, Argentina M. Oves, King Abdulaziz University, Saudi Arabia

### \*Correspondence:

Yuet Ying Loo yuetying88@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 28 March 2018 Accepted: 22 June 2018 Published: 16 July 2018

### Citation:

Loo YY, Rukayadi Y, Nor-Khaizura M-A-R, Kuan CH, Chieng BW, Nishibuchi M and Radu S (2018) In Vitro Antimicrobial Activity of Green Synthesized Silver Nanoparticles Against Selected Gram-negative Foodborne Pathogens. Front. Microbiol. 9:1555. doi: 10.3389/fmicb.2018.01555 Silver nanoparticles (AgNPs) used in this study were synthesized using pu-erh tea leaves extract with particle size of 4.06 nm. The antibacterial activity of green synthesized AgNPs against a diverse range of Gram-negative foodborne pathogens was determined using disk diffusion method, resazurin microtitre-plate assay (minimum inhibitory concentration, MIC), and minimum bactericidal concentration test (MBC). The MIC and MBC of AgNPs against Escherichia coli, Klebsiella pneumoniae, Salmonella Typhimurium, and Salmonella Enteritidis were 7.8, 3.9, 3.9, 3.9 and 7.8, 3.9, 7.8, 3.9 µg/mL, respectively. Time-kill curves were used to evaluate the concentration between MIC and bactericidal activity of AgNPs at concentrations ranging from 0×MIC to 8×MIC. The killing activity of AgNPs was fast acting against all the Gram-negative bacteria tested; the reduction in the number of CFU mL−<sup>1</sup> was >3 Log<sup>10</sup> units (99.9%) in 1–2 h. This study indicates that AgNPs exhibit a strong antimicrobial activity and thus might be developed as a new type of antimicrobial agents for the treatment of bacterial infection including multidrug resistant bacterial infection.

Keywords: silver nanoparticles, tea leave extracts, antimicrobial activity, foodborne pathogens, Gram-negative, time-kill curves

## INTRODUCTION

Recently, nanotechnology has emerged as a dynamically developing area of scientific interest in the world. Nanoparticles are defined as a nanoscale particle of size ranging from 1 to 100 nm. Among the metallic nanoparticles, silver nanoparticles (AgNPs) have gained increasingly attention due to its unique physical, biological and chemical properties. AgNPs are well-known to exhibit a strong antimicrobial activity against various microorganisms such as bacteria, viruses, and fungi due to its smaller in size and large surface area (Franci et al., 2015). AgNPs are also widely used as anti-fungal (Medda et al., 2015), anti-inflammatory (Hebeish et al., 2014), and anti-viral properties (Bekele et al., 2016).

Green synthesis of AgNPs employing either biological microorganisms or plant extracts has emerged as a simple and alternative to chemical synthesis. Green synthesis method provides advancements over chemical methods as it is environmental friendly and cost effective. Plant extractsmediated synthesis of AgNPs can be advantageous compared with other biological processes as it does not require the process of maintaining the cell cultures and aseptic environments (Loo et al., 2012). Several studies on the green synthesis of AgNPs using plant extracts have been reported (Medda et al., 2015; Ahmed et al., 2016; Dhand et al., 2016; Selvam et al., 2017).

Foodborne illnesses have emerged as a major public health concern around the world. WHO (2014) reported that there is about 30% of the population in industrialized countries affected by foodborne diseases every year. The consumption of foods contaminated with foodborne pathogens such as bacteria, fungi, viruses, and toxins are often recognized as the main source of foodborne illness in humans. Food especially minimalprocessed food can be contaminated during pre-harvesting, post-harvesting, processing, transport, handling, or preparation. The most common foodborne pathogens found in food are Salmonella spp. (Lee et al., 2015; D'Ostuni et al., 2016), Listeria spp. (Ferreira et al., 2014; Välimaa et al., 2015), Escherichia coli O157 (Heiman et al., 2015), Campylobacter spp. (Kaakoush et al., 2015), and Clostridia spp. (Chukwu et al., 2016).

The presence of multidrug resistance pathogens have increased the number of infectious disease and became the main cause of death in the world (WHO, 2000; Tanwar et al., 2014). Widely misuse and abuse of antibiotics are the leading cause of antibiotic resistance in the bacteria (O'Bryan et al., 2018). Multidrug resistant bacteria infection may lead to several impacts including increase of mortality and morbidity rates, prolong of hospitalization period, and economic loss (Patel et al., 2008). Woh et al. (2017) detected multi-drug resistant non-typhoidal Salmonella among migrant food handlers, which may cause cross-contamination to the food products. Thus, the development of a new and natural antimicrobial agent is needed as there is a growing concern in multidrug resistant foodborne pathogens.

The aim of this study is to determine the antibacterial activity of green synthesized AgNPs against a diverse range of Gram-negative foodborne pathogens by using disk diffusion method, resazurin microtitre-plate assay minimum inhibitory concentration (MIC), minimum bactericidal concentration test (MBC), and time-kill curve assay.

## MATERIALS AND METHODS

### Preparation of Silver Nanoparticles

The synthesis of AgNPs using pu-erh tea leaves extracts was done using the method as described previously (Loo et al., 2012). Ten gram of tea leaves was weighed in a beaker. The tea leaves was added with 100 mL of distilled water and maintained at 60◦C for 10 min. After 10 min, the tea extract was filtered using 0.45 µm Millipore membrane filter and followed by 0.2 µm Millipore membrane filter. For synthesis of AgNPs, 12 mL of tea extracts was added into 100 mL of AgNO<sup>3</sup> (1 mM) in Erlenmeyer flask at room temperature. Color changes of the solution were observed. The synthesized AgNPs were characterized by UV-vis spectroscopy, X-ray diffraction (XRD), Fourier transform infrared (FTIR) spectroscopy, and transmission electron microscopy (TEM).

### Bacteria Strains Preparation

Escherichia coli ATCC 25922 (E. coli), Klebsiella pneumoniae ATCC 13773 (K. pneumoniae), Salmonella Typhimurium ATCC 14028 (S. Typhimurium), and Salmonella Enteritidis ATCC 13076 (S. Enteritidis) were obtained from the American Type Culture Collection (Rockville, MD, United States). All the bacteria strains were cultured in Mueller Hinton broth (MHB) (Merck, Germany) at 37◦C for 24 h with 200 rpm agitation.

## Preparation of Resazurin Solution

The resazurin solution was prepared at 0.02% (wt/vol) according to Khalifa et al. (2013). A 0.002 g of resazurin salt powder was dissolved in 10 mL of distilled water and vortexed. The mixture was filtered by Millipore membrane filter (0.2 µm). The resazurin solution can be kept at 4◦C for 2 weeks.

### In Vitro Susceptibility Test Disk Diffusion Method

The antibacterial activity of AgNPs against the selected Gramnegative foodborne pathogens was carried out using Kirby–Bauer Disk Diffusion Susceptibility Test method (Bauer et al., 1966). The bacteria strains were spread on the Mueller-Hinton agar (MHA) (Merck, Germany) using sterile cotton swab. Sterile blank antimicrobial susceptibility disk was used in the test. The disks were loaded with 10 µL of tea leaves extracts, silver nitrate solution (1 mM), and solution containing tea leaves mediated synthesized AgNPs separately. The disks were then placed on the agar plate and incubated at 37◦C for 24 h. The zone of inhibition was observed after 24 h of incubation.

### Minimum Inhibitory Concentration (MIC) and Minimum Bactericidal Concentration (MBC) Evaluation

The MIC and MBC of green synthesized AgNPs were done using the method described in the guideline of CLSI (2012). The MIC test was performed in 96-well round bottom microtiter plate using standard broth microdilution methods while the MBC test was performed on the MHA plates. The bacterial inoculums were adjusted to the concentration of 10<sup>6</sup> CFU/mL. For the MIC test, 100 µL of the synthesized AgNPs stock solution (500 µg/mL) was added and diluted twofold with the bacterial inoculums in 100 µL of MHB started from column 12 to column 3. Column 12 of the microtiter plate contained the highest concentration of AgNPs, while column 3 contained the lowest concentration. Column 1 served as negative control (only medium) and the column 2 served as positive control (medium and bacterial inoculums). Each well of the microtiter plate was added with 30 µL of the resazurin solution and incubated at 37◦C for 24 h. Any color changes were observed. Blue/purple color indicated no bacterial growth while pink/colorless indicated bacterial growth.

The MIC value was taken at the lowest concentration of antibacterial agents that inhibits the growth of bacteria (color remained in blue).

The MBC was defined as the lowest concentration of the antibacterial agents that completely kill the bacteria. MBC test was performed by plating the suspension from each well of microtiter plates into MHA plate. The plates were incubated at 37◦C for 24 h. The lowest concentration with no visible growths on the MHA plate was taken as MBC value.

### Time-Kill Curve

Time-kill assay was done in MHB medium as described by Zainin et al. (2013) and Lau et al. (2014). The bacterial inoculums were adjusted to 10<sup>6</sup> CFU/mL. The AgNPs solution was diluted with MHB media containing bacterial inoculums to obtain the final concentration of 0× MIC, 0.5× MIC, 1× MIC, 2× MIC, 4× MIC, and 8× MIC for each type of bacteria in the total final volume of 1 mL. The cultures were then incubated at 37◦C with 150 rpm agitation. The cultures (100 µL) were spread on MHA plates at time 0, 0.25, 0.5, 1, 2, and 4 h. The experiment was carried out in triplicate. The number of colonies on the MHA plates was quantified in CFU/mL after incubation at 37◦C for 24 h. For statistical analysis, SPSS (v.19) statistical package was used to determine the significant (P < 0.05) difference among the tested foodborne pathogens.

## RESULTS AND DISCUSSION

This study was aimed to determine the antibacterial effect of green synthesized AgNPs. The green synthesized AgNPs used in this study were characterized by UV-vis spectroscopy, XRD, FTIR spectroscopy, and TEM. The XRD patterns for synthesized AgNPs showed that five main characteristic diffraction peaks for Ag were observed at 2θ = 38.4, 44.5, 64.8, 77.7, and 81.7, which correspond to the (111), (200), (220), (311), and (222) crystallographic planes of face-centered cubic (fcc) Ag crystals. The UV-vis absorption spectrum of the synthesized AgNPs showed a broad peak at 436 nm which is a characteristic band for Ag. Three infrared bands were observed at 3,271, 1,637, and 386 cm−<sup>1</sup> in FTIR measurement. The band at 3,271 and 1,637 cm−<sup>1</sup> indicated that the presence of protein as capping agent for AgNPs which increases the stability of the nanoparticles, while the broad peak at 386 cm−<sup>1</sup> corresponded to the Ag metal. TEM image revealed that the AgNPs is spherical with the particle size of 4.06 nm (Loo et al., 2012).

The antibacterial activity of AgNPs was determined against four species of Gram-negative foodborne pathogens: E. coli ATCC 25922, K. pneumoniae ATCC 13773, S. Typhimurium ATCC 14028, and S. Enteritidis ATCC 13076. The results for disk diffusion test, MIC and MBC of the AgNPs are summarized in **Table 1**. For the disk diffusion test, the presence of clear zone around the AgNPs disk suggesting that the AgNPs possessed antibacterial activity which is able to inhibit the growth of the Gram-negative foodborne pathogens. As previous study by Guzman et al. (2012), reported that AgNPs employed antibacterial activity on Gram-negative bacteria. The visible clear zone produced by AgNPs against four different species of Gramnegative bacteria is showed in **Figure 1**.

Disk diffusion test was described as the preliminary study in screening the antibacterial activity of an antimicrobial agent; therefore, a further evaluation in determining the antibacterial activity of AgNPs using MIC value was needed (Burt, 2004). MIC was defined as the lowest concentration of the antibacterial agent to inhibit the growth of bacteria by serial dilution. As showed in **Table 1**, the MIC values of AgNPs against the foodborne pathogens were ranged from 3.9 to 7.8 µg/mL. K. pneumonia, S. Typhimurium and S. Enteritidis showed the MIC value of 3.9 µg/mL while E. coli showed the MIC value of 7.8 µg/mL. MBC is the lowest concentration of antibacterial agent to kill the bacteria (showed no growth on the agar plate). In the study, MBC for K. pneumoniae and S. Enteritidis were 3.9 µg/mL while S. Typhimurium and E. coli showed the MBC value of 7.8 µg/mL. The MIC and MBC value of E. coli showed that E. coli was less susceptible to AgNPs. This may due to the positive charges of AgNPs trapped and blocked by lipopolysaccharide, thus make E. coli less susceptible to AgNPs (Lara et al., 2010b).

Resazurin dye was used in the study as an indicator in the determination of cell growth, especially in cytotoxicity assays (McNicholl et al., 2007). Oxidoreductases within viable cells reduced the resazurin salt to resorufin and changed the color from blue non-fluorescent to pink and fluorescent. According to McNicholl et al. (2007), resazurin dye has been applied for decades to check for the bacterial and yeast contamination in milk.

The time kill activity of four foodborne pathogens is shown in **Figure 2**. The bactericidal activity of AgNPs is effective against the selected Gram-negative pathogens; the reduction in the number of CFU/mL was ≥3 Log units (99%). The bactericidal endpoint of AgNPs for E. coli was reached after 2 h of incubation at 4× MIC (31.2 µg/mL) and 8× MIC (62.4 µg/mL); while for K. pneumoniae, the bacteria was killed after 2 h of incubation at 2× MIC (7.8 µg/mL), 4× MIC (15.6 µg/mL), and 8× MIC (31.2 µg/mL). S. Typhimurium was killed after 1 h of incubation at 4× MIC (15.6 µg/mL) and 8× MIC (31.2 µg/mL). The bactericidal endpoint of AgNP for S. Enteritidis was reached after 2 h of incubation at 2× MIC (7.8 µg/mL) and 4× MIC (15.6 µg/mL); however, the end point reached faster after 1 h of incubation at 8× MIC (31.2 µg/mL). No significant differences (P > 0.05) were found among the tested Gramnegative foodborne pathogens. This indicates that AgNPs are

TABLE 1 | The diameter of zone inhibition (mm), MIC value (µg/mL), and MBC value (µg/mL).


broad spectrum antimicrobial agents which exert the same effect to all Gram-negative bacteria strains.

Silver nanoparticles are well-known as the most universal antimicrobial substances due to their strong biocidal effect against microorganisms, which has been used for over the past decades to prevent and treat various diseases (Oei et al., 2012). AgNPs are also widely used as anti-fungal (Kim et al., 2009), antiinflammatory (Nadworny et al., 2010), and anti-viral properties (Lara et al., 2010a). Recently, non-hazardous AgNPs can easily be synthesized using a cost-effective method and tested as a new type of antimicrobial agents.

In this study, the application of AgNPs as an antimicrobial agent was tested against selected Gram-negative bacteria on agar plate and liquid medium. The results showed that the tested bacteria could completely inhibit by AgNPs. The inhibition of bacteria growth was reported affected by the concentration of AgNPs and bacteria used in the experiments (Sondi and Salopek-Sondi, 2004). The green synthesized AgNPs in this study are able to inhibit the high concentration of bacteria (approximately 10<sup>6</sup> CFU/mL). This indicated that AgNPs showed an excellent antimicrobial effect as the high CFU concentration of bacteria used in this study are rarely appeared in real-life systems.

The antibacterial activity of AgNPs has been reported by many researchers. However, the MIC values from the previous studies showed the range through a large extent of variation. Therefore, the comparison of the results is difficult as there is no standard method for determination of antibacterial activity of AgNPs and different methods have been applied by the researchers (Zarei et al., 2014). In this study, AgNPs exhibit a good antibacterial activity against the tested foodborne pathogens. Based on the results, the tested bacteria were able to kill in a shorter time at low concentration of AgNPs. This may due to the cell wall structure of Gram-negative bacteria. The characteristic cell wall structure of Gram-negative bacteria is different from Gram-positive bacteria. Gram-negative bacteria have a cytoplasmic membrane, a thin peptidoglycan layer, and an outer membrane containing lipopolysaccharide. There is a space between the cytoplasmic membrane and the outer membrane called the periplasmic space or periplasm. The periplasmic space contains the loose network of peptidoglycan chains known as the peptidoglycan layer.

The rapid reproduction time of bacteria is one of the main causes of bacteria's infectivity (Lara et al., 2010b). However, the reproduction time of the bacteria could be an ideal way to prevent the viable infection as AgNPs were effective in inhibiting

and killing the bacteria in a dose and time dependent manner as shown in the time-kill assays. Zhang et al. (2016) reported that the smaller size of AgNPs could cause more toxicity to the bacteria and show better bactericidal effect compared to the larger particles as they have larger surface area. Previous study by Agnihotri et al. (2014) found that the antibacterial efficacy was increased for the AgNPs with less than 10 nm size. They also concluded that AgNPs with the size of 5 nm have the fastest antibacterial activity compared to others size of AgNPs.

Silver nanoparticles have emerged as antimicrobial agents against multidrug resistant bacteria due to their high surfacearea-to-volume ration and unique chemical and physical properties. AgNPs have particle size ranging from 1 to 100 nm. The surface area-to-volume ratio of AgNPs increases as the particles size decreases. Morones et al. (2005) reported that AgNPs with the size of 10-100 nm showed strong antimicrobial effect against both Gram-positive and negative bacteria. The small particle size enables AgNPs to adhere to the cell wall and penetrate into the bacteria cell easily, which in turn improves their antimicrobial activity against bacteria. The antimicrobial effects of AgNPs against multidrug resistant bacteria have been studied by many researchers and it was proved that AgNPs are effective against multidrug resistant bacteria such as multidrug resistant E. coli (Paredes et al., 2014; Kar et al., 2016), multidrug resistant strain of Pseudomonas aeruginosa (Durairaj et al., 2012), methicillin-resistant Staphylococcus aureus (MRSA) (Paredes et al., 2014; Yuan et al., 2017), and extended-spectrum β-lactam (ESBL) producing bacteria (Doudi et al., 2013; Subashini et al., 2014).

On the other hands, AgNPs are advantageous compared to conventional chemical antimicrobial agents as the major problem caused by the conventional chemical antimicrobial agents is multidrug resistance. The effectiveness of the chemical antimicrobial agents depends on the specific binding of the microorganisms with the surface and metabolites of the antimicrobial agents. However, the chemical antimicrobial agents are limited to use especially in medical field as various microorganisms have developed multiple resistance traits over a period of generations. Thus, the development of AgNPs could be an alternative way to overcome the multidrug resistance microorganisms as bacteria are less likely to develop resistance to metal nanoparticles compared to the conventional antibiotics.

The exact mechanisms of AgNPs against bacteria still remain unknown. However, there are some researchers proposed that the action of AgNPs on bacteria may due to its ability to penetrate into the cell (Sondi and Salopek-Sondi, 2004), the formation of free radicals (Danilczuk et al., 2006; Kim et al., 2007), the inactivation of proteins in the cell by silver ions (Rai et al., 2012) and the production of reactive oxygen species (ROS) (Dakal et al., 2016). Besides that, there are also some factors in affecting the bactericidal mechanisms of AgNPs such as the concentration of AgNPs and bacteria class (Kim et al., 2007; Zhang et al., 2014), shape (Pal et al., 2007; Meire et al., 2012), size (Martinez-Castanon et al., 2008), and the combination of various antibiotics (Fayaz et al., 2010; Singh et al., 2013).

### CONCLUSION

fmicb-09-01555 July 12, 2018 Time: 18:28 # 6

Silver nanoparticles showed significant antibacterial activity against the selected Gram-negative foodborne pathogens. Thus, AgNPs might be a good alternative to develop as antibacterial agent against the multidrug-resistant strains of bacteria. The applications of AgNPs may lead to valuable findings in various fields such as medical devices and antimicrobial systems.

### AUTHOR CONTRIBUTIONS

YL, BC, YR, and RS developed the study design. YL and CK carried out the confirmation for the selected foodborne

### REFERENCES


pathogens. MN provided culture media and technical advice in the study. YL interpreted the data, drafted the manuscript, and revised the manuscript. YR, M-A-RN-K, CK, BC, and RS checked on the manuscript. All authors read and approved the final version of the manuscript.

## FUNDING

This research was funded by a Research University Grant Scheme Initiative Six (RUGS 6) of Universiti Putra Malaysia (GP-IPS 9438703) and Fundamental Research Grant Scheme (FRGS) of Ministry of Higher Education (MOHE), Malaysia (02-01-14- 1475FR) and, in part, by the Kakenhi Grant-in-Aid for Scientific Research (KAKENHI 24249038), Japan Society for the Promotion of Sciences and grant-in-aid of Ministry of Health, Labor and Welfare, Japan.

resistant Pseudomonas aeruginosa. World Acad. Sci. Eng. Technol. 6, 210–213. doi: 10.12659/MSMBR.883835


of silver nanoparticles using Tinospora cordifolia (Thunb.) Miers and evaluate its antibacterial, antioxidant potential. J. Radiat. Res. Appl. Sci. 10, 6–12. doi: 10.1016/j.jrras.2016.02.005


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Loo, Rukayadi, Nor-Khaizura, Kuan, Chieng, Nishibuchi and Radu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fmicb-09-01555 July 12, 2018 Time: 18:28 # 7

# Prevention of Dermal Abscess Formation Caused by Staphylococcus aureus Using Phage JD007 in Nude Mice

Bingyu Ding1,2, Qingtian Li<sup>2</sup> , Mingquan Guo<sup>3</sup> , Ke Dong<sup>3</sup> , Yan Zhang<sup>3</sup> , Xiaokui Guo<sup>3</sup> , Qingzhong Liu<sup>1</sup> , Li Li<sup>1</sup> \* and Zelin Cui<sup>1</sup> \*

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Ananda Shankar Bhattacharjee, Bigelow Laboratory for Ocean Sciences, United States Junji Xing, Houston Methodist Research Institute, United States

### \*Correspondence:

Li Li annylish@126.com Zelin Cui czl\_phage@126.com; czl@sjtu.edu.cn

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 11 October 2017 Accepted: 22 June 2018 Published: 23 July 2018

### Citation:

Ding B, Li Q, Guo M, Dong K, Zhang Y, Guo X, Liu Q, Li L and Cui Z (2018) Prevention of Dermal Abscess Formation Caused by Staphylococcus aureus Using Phage JD007 in Nude Mice. Front. Microbiol. 9:1553. doi: 10.3389/fmicb.2018.01553 <sup>1</sup> Department of Laboratory Medicine, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China, <sup>2</sup> Department of Laboratory Medicine, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China, <sup>3</sup> Department of Immunology and Microbiology, Shanghai Jiao Tong University School of Medicine, Shanghai, China

Aim: In this study, Staphylococcus phage JD007 bactericidal activity and induced immune responses during treatment were assessed in a dermal abscess model.

Materials and Methods: Dermal abscesses in nude mice were established by injecting a clinical isolate of S. aureus SA325 isolated from the back under-dermal abscess of an in-patient.

Results: Phage JD007 was able to inhibit the growth of S. aureus SA325 at MOI = 1 or 10, significantly preventing the formation of dermal abscesses. Moderate immune responses were observed in the prevention group through detection of cytokines.

Conclusion: Phage JD007 inhibits the formation of dermal abscesses caused by a clinical S. aureus strain in nude mice without robust immune responses.

Keywords: Staphylococcus aureus, dermal abscess, inhibition, nude mouse, phage JD007

## INTRODUCTION

With the increasing prevalence antibiotic resistant infections, including MRSA (methicillinresistant Staphylococcus aureus) and the emergence of VRSA (vancomycin-resistant S. aureus) (Xiao et al., 2013; Limbago et al., 2014), innovative treatments are urgently needed. Recently, there has been renewed interest in phage therapy (Thiel, 2004; Salmond and Fineran, 2015), which has been proposed as a potential antimicrobial therapy as far back as the 1920's (Salmond and Fineran, 2015). Advances in biotechnology have led to the identification of numerous classes of bacteriophages with different host specificities such as K, GH15, Twort, phiIPLA-RODI, and phiIPLA-C1C (O'Flaherty et al., 2004; Kwan et al., 2005; Gu et al., 2012b; Gutierrez et al., 2015). Furthermore, whole genome sequences of these phages enables us to understand their biology at a much deeper level (Cui et al., 2017b).

Phages have been used to treat infectious disease in various animal models. A single dose of phage MR10 exhibited efficacy similar to linezolid in resolving the course of hindpaw infection in diabetic animals, suggesting this approach could serve as an effective strategy in treating MRSA mediated foot infections in diabetic individuals (Chhibber et al., 2013). Phage LS2a has been shown to prevent abscess formation in rabbits when injected simultaneously with S. aureus into the subcutaneous site (Wills et al., 2005). Phage Kpn5 could potentially be used as stand-alone therapy for the Klebsiella pneumoniae induced burn wound infection, effective even against antibiotic-resistant strains (Kumari et al., 2009). Additionally, in clinical trials, phage products have showed efficacy according to some reports (Marza et al., 2006).

Phage host specificity narrows their use with lysing efficacies that can vary significantly depending on the phage strain. Reports have also documented failures of phage therapy in vivo (Sarker and Brussow, 2016), so, it is necessary to evaluate phages' bacteria killing efficacy both in vitro and in an animal model. Staphylococcus phage JD007, which belongs to family of Myoviridae was identified in 2012 and has been shown to have wide host range (Cui et al., 2012), capable of killing S. aureus of different MLST Types (Cui et al., 2017a). The aim of this study is to evaluate the treatment efficacy of JD007 in dermal abscesses caused by a clinical strain of MRSA and the immune responses elicited during therapy.

## MATERIALS AND METHODS

### Animals

BALB/c nude mice (WT, Shanghai Super B&K laboratory animal Corp. Ltd., [SCXK(HU)2005—0001], Shanghai, China) were used in this study. Dermal abscesses were stereo tactically induced as described previously (Capparelli et al., 2007). Uninfected mice served as controls. Mice were euthanized on days 1, 3 and 10 after infection. Ethics eligibility of this study was approved by the Shanghai General Hospital Ethics Committee (Shanghai Jiao Tong University School of Medicine). All experiments were performed in accordance with the national animal protection guidelines approved by the local animal ethics committee. Fifty mice were randomly divided into five groups, 10 mice in each group, including control group (no treatment), phage JD007 group (injected with purified phage JD007), infection group (infected by S. aureus SA325), prevention group (phage JD007 were injected 1 h before mice was infected by S. aureus SA325), and treatment group (phage JD007 was injected 1 h after mice were infected by S. aureus SA325).

### Bacteriophage Purification

High-titre phage stocks were obtained through amplification in liquid LB (Luria-Bertani) medium supplemented with 10 µM MgCl<sup>2</sup> and 5 µM CaCl2. In culture, S. aureus SA325 was infected phage JD007 at a MOI of 0.1 and incubated at 37◦C overnight. The visible bacterial lysate in liquid culture were obtained, and then incubated with chloroform (final concentration was 2%) for 30 min with gentle shaking to kill residual bacteria. Remaining bacterial products were removed by centrifugation at 6,500 rpm (Beckman, JA18.0, United States) for 15 min. Phage contained in the supernatant was enriched at 4◦C overnight using polyethylene glycol (PEG) 8000 (final concentration 10% w/v), and precipitated at 8,500 rpm for 20 min (Beckman, JA18.0, United States). Afterwards, the pellet was dissolved in TM buffer [(Tris-Mg2<sup>+</sup> Buffer) 10 mM Tris–HCl (pH 7.2–7.5), 100 mM NaCl, 10 mM MgCl2.5 mM CaCl2] and vortexed. PEG 8000 was removed by adding the same volume of chloroform after vortexing. The solution was centrifuged at 4,000 g/min for 10 min, and the supernatant containing the phage was

phage group, infection group, prevention group, and treatment group. The blue arrows showed the skin sites injected with S. aureus SA325.

isolated. CsCl was added at a concentration of 0.5 g per 1 mL, and the phages were purified by discontinuous centrifugation in a CsCl gradient (1.33, 1.45, 1.50, and 1.70g/cm<sup>3</sup> ) in TM buffer in Ultra-Clear tubes (Beckman Coulter, Inc., Fullerton, CA, United States) by centrifuging at 35000 rpm/min for 4 h. Finally, the layer with enriched phage was obtained with a syringe, dialyzed against TM buffer and stored at 4◦C (Boulanger, 2009).

### S. aureus Strain and Culture Condition

Staphylococcus aureus strain SA325 (MRSA, confirmed by antibiotic resistant profile) was isolated from an in-patient with a re-occurring back under-dermal abscess due to long-term bed rest. SA325 concentration was calculated by serial dilutions when grown to OD600 nm = 0.4, and plated uniformly using the glass spreading rod, and cultured overnight at 37◦C, the value of SA325 per OD600 nm was 9.5 × 10<sup>9</sup> CPU/mL.

### S. aureus Growth Inhibition Assay

Drop tests were conducted using two-layer agar plates. S. aureus SA325 was cultured to OD600 nm = 0.4 in TSB (Tryptic soy broth) supplemented with 10 µM MgCl<sup>2</sup> and 5 µM CaCl2. Bacteria grown to logarithmic phase were mixed with 0.7% agar LB and then poured on the plate of LB agar (1.5%) uniformly, incubating at room temperature for 30 min. Finally, 3 µL of serial dilutions of phage JD007 was dropped on the plate. The inhibition zones were observed after an overnight culture.

The inhibition growth curves for phage JD007 infected S. aureus SA325 were determined using MOIs of 0, 0.01, 0.1, and 1 inoculated in 96 well plates separately and incubated at 37◦C. OD600 nm of these cultures was measured every 30 min extending for 8 h. Assay was performed three times (n = 3) with three biological replicates.

### Mice Dermal Abscess Infection Model

Staphylococcus aureus was prepared by culturing at 37◦C in TSB to exponential phase at which time the bacteria was collected and washed twice with PBS with immediate centrifugation. SA325 was re-suspended at final concentration of 1 × 10<sup>9</sup> CFUs/100µL, and kept on ice before injecting nude mice intradermally.

All animal experiments were approved by the Shanghai General Hospital Ethics Committee (Shanghai Jiao Tong University, China), and conducted according to the Chinese Law for Animal Protection. To investigate the antimicrobial efficacy of phage JD007 for subcutaneous abscesses, BALB/c nude mice (6–8 weeks old) were anesthetized by intraperitoneal injection of a saline solution containing fentanyl (0.05 mg/kg), midazolam (5 mg/kg), and medetomidine (0.5 mg/kg). The backs of the animals were disinfected with 70% ethanol, and 50 µL suspension containing 5 × 10<sup>8</sup> CFU S. aureus SA325 in PBS were inoculated subcutaneously; 50 µL of phage JD007 was injected into the infected region. Mice were weighed before inoculation. Weight and abscess formation were measured daily extending for 10 days. The length (L) and width (W) of abscesses were determined using a caliper. The size of the abscesses was then calculated with the standard formula for area: V = [π/2] × L × W (Weinandy et al., 2014).

### Immune Response During Treatment Using Staphylococcus Phage JD007

Three mice in each described above were randomly enrolled and euthanized on the third day of the experimental protocol. Serum from each mouse was obtained, and immune cytokines

(IL-1β, IL-6, IL-8, IFN-γ, and TNF-α) were measured using ELISA (Beijing 4A Biotech Co., Ltd.).

### Statistical Analysis

A two-tailed unpaired Student's t-test was used for statistical analysis with GraphPad Prism Software. P-values of less than 0.05 were considered significant unless specifically indicated otherwise.

## RESULTS

## S. aureus SA325 Growth Is Inhibited by Phage JD007 in Vitro

Staphylococcus aureus SA325 was co-cultured with phage JD007 at 37◦C with co-culture OD600 nm measurements made at 30 min intervals for 8 h. Results showed that phage JD007 inhibited the

Frontiers in Microbiology | www.frontiersin.org

growth of S. aureus SA325 as early as 1 h after co-incubation at MOI = 1 (**Figure 1**). Whereas, S. aureus SA325 grew normally when infected by phage JD007 at lowers MOIs = 0.1 and 0.01. The drop test results showed that phage JD007 formed clear inhibition zones in two-layer agar plates.

## Phage JD007 Prevents S. aureus SA325-Mediated Dermal Abscess Formation in Nude BALB/c Mice in Vivo

Dermal abscess formation was established in the back of nude mice and observed daily for 10 days. Results showed that the maximum sizes of dermal abscesses were achieved on the second day following infection. On the seventh day, all of the abscesses were scarred. As shown in **Figures 2**, **3**, we could see no visible dermal abscesses formed in control group, phage group, and prevention group during the entire observation period. On the third day, the abscess sizes tended to decrease in size, and mice's dermal abscess size in the treatment group were significantly smaller than those in the infection group (p < 0.05).

## Inflammatory Cytokines Responses of Nude BALB/c Mice During Phage Therapy

Cytokines of IL-1β, IL-6, IL-8, IFN-γ, and TNF-α in the sera were measured using ELISA on the third day. As shown in **Figure 4**, there were no significant differences between IFN-γ and TNF-α among those groups. IL-1β, IL-6, and IL-8 in treatment groups were higher than those in control group, whereas there were no significant differences between the negative control and prevention groups.

## Weight of Nude BALB/c Mice During Phage Therapy

Weights of mice in each group were measured daily, encompassing the entire observation period. Body weights gain were calculated by subtracting original body weight from weights on the tenth day. As shown in **Figure 5**, weight gains in infection group and phage treatment were much lower than that compared with control group. The weight gain in the prevention and phage JD007 groups had no significant difference compared with the PBS control group during treatment.

## DISCUSSION

Staphylococcus aureus SA325 was isolated from a re-occurring back under-dermal abscess of an in-patient with long-term bed rest. In vitro bactericidal activity tests showed that phage JD007 could form a clear inhibition zones using a two-layer soft agar plate assay, and could inhibit growth of SA325 at MOI = 1 or 10.

fmicb-09-01553 July 19, 2018 Time: 16:25 # 5

In order to evaluate the ability of phage JD007 to prevent or treat of abscesses caused by S. aureus, a dermal abscess model was established in nude mice model using clinical isolate SA325, as reported previously (Malachowa et al., 2013). In this study, we successfully established the dermal abscess infection in nude BALB/c mice using S. aureus SA325. Furthermore, the efficacies of treatment of the abscesses using phage JD007 were evaluated. There were no observed abscess formation in the prevention group. The average abscess sizes of phage treatment group were smaller than those of the infection group. The results indicate that phage JD007 prevents formation of abscess. Also, it supports earlier observations on phage MSa inhibition S. aureus dermal abscess formation (Capparelli et al., 2007).

To evaluate the response of mice during prevention and treatment after phage JD007 inoculation, the weight changes were observed during the whole process. As bacteriophages are viruses containing nucleic acid and proteins, they may bring some risk to humans when used for the treatment of bacteremia, as the complex composition of the phage may induce unpredictable immune responses when administered to the blood. The average weights gains in the infection and treatment groups were significantly lower than those in the control, PBS, phage JD007, and prevention groups. Furthermore, immune responses of the mice during treatment were evaluated, mice sera was taken on the third day, and the inflammatory cytokines of IL-1β, IL-6, IL-8, IFN-γ, and TNF-α in serum were measured using ELISA. The cytokines IL-1β and IL-6 both belong to endogenous pyrogens, and they can be applied to the hypothalamus regulating center to cause fever. IL-1β and IL-6 levels were significantly higher in the treatment group, while there were no significant differences among prevention, infection, or control groups, indicating that phage JD007 treatment may destroy the cells of S. aureus robustly, leading to increase of IL-1β and IL-6 that may cause fever. It's reported that TRIM29 could negatively regulate the proinflammatory cytokine IL-6 and TNF-α production in response to LPS and bacteria Haemophilus influenzae (Xing et al., 2016). Whether S. aureus infection suppresses the TRIM 29 expression and leads to increased expression of IL-6 should be further examined. IL-8 was significantly elevated in infection and treatment groups. IL-8 can signal chemotactic neutrophil leukocytes and activate T cells in the infection sites, clearing the pathogen. IFN-γ and TNF-α can activate mononuclear macrophages, which can enhance the phagocytosis and killing functions; TNF-α can directly kill the cell infected by virus. There were no significant changes in any of these groups.

The specificity of bacteriophages is one of important reason hindering clinical use of phage for therapy, though cocktails

### REFERENCES


may overcome this fault (Gu et al., 2012a; Mendes et al., 2014). There are numerous public reports describing the success of phage therapy inhibiting infectious diseases caused by dermal associated infections (Vieira et al., 2012; Trigo et al., 2013). We postulate that phage therapy could become an important choice for the treatment of wound infections. As shown in **Figure 6**, the pool of different phages should be established before using them for therapy; at the same time, the bacteria causing infection should be isolated and its sensitivity to the phages in the pool confirmed. The formula of sensitive phages must be prepared at first; once the bacteria causing the infection is confirmed to be sensitive to phages in the pool, the formula containing the sensitive phage should be chosen for treatment of the infections. The success of phage therapy depends upon the pool of different kinds of bacteriophages.

In summary, phage JD007 could inhibit the growth of S. aureus both in vitro and in vivo. Phage JD007 can prevent abscess formation caused by S. aureus isolated from an inpatient's abscess samples through under-dermal injection, and can also significantly reduce the severity of dermal abscess caused by S. aureus. In conclusion, phage JD007 is a promising candidate phage for use in preventing dermal abscess.

## AUTHOR CONTRIBUTIONS

ZC, QTL, and BD carried out the experiments and drafted the manuscript. QTL, KD, BD, QZL, YZ, and MG participated in the mice model experiments. ZC, QTL, XG, and LL participated in the design of the study and performed the statistical analysis. ZC, QTL, and LL conceived the study, participated in its design, and coordinated to help draft the manuscript.

## FUNDING

This work was sponsored in part by the National Natural Science Foundation of China (No. 31500154), the fund of Shanghai Health and Family Planning Committee (No. 201440289), and the outstanding medical youth program A of Shanghai General Hospital (No. 06N1702002).

## ACKNOWLEDGMENTS

Thanks Luis J. Cocka for suggestions of modification manuscript.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ding, Li, Guo, Dong, Zhang, Guo, Liu, Li and Cui. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Non-toxigenic Clostridioides (Formerly Clostridium) difficile for Prevention of C. difficile Infection: From Bench to Bedside Back to Bench and Back to Bedside

Dale N. Gerding<sup>1</sup> \*, Susan P. Sambol 1,2 and Stuart Johnson1,2

*<sup>1</sup> Research Service, Edward Hines Jr. VA Hospital, Hines, IL, United States, <sup>2</sup> Department of Medicine, Loyola University Chicago Medical Center, Maywood, IL, United States*

### Edited by:

*Noton Kumar Dutta, Johns Hopkins University, United States*

### Reviewed by:

*Sarah J. Kuhl, VA Northern California Health Care System, United States Xingmin Sun, University of South Florida, United States*

> \*Correspondence: *Dale N. Gerding dale.gerding2@va.gov*

### Specialty section:

*This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology*

> Received: *02 March 2018* Accepted: *09 July 2018* Published: *26 July 2018*

### Citation:

*Gerding DN, Sambol SP and Johnson S (2018) Non-toxigenic Clostridioides (Formerly Clostridium) difficile for Prevention of C. difficile Infection: From Bench to Bedside Back to Bench and Back to Bedside. Front. Microbiol. 9:1700. doi: 10.3389/fmicb.2018.01700* The beneficial effect of colonization of the gastrointestinal tract by non-toxigenic *Clostridioides difficile* (NTCD) strains as a preventive of toxigenic *C. difficile* infection (CDI) has been known since the early 1980s. Investigators in both the USA and United Kingdom demonstrated that prior colonization by randomly selected NTCD strains provided prevention against infection by toxigenic *C. difficile* in hamsters, albeit with limited durability. In the 1980s two patients with multiply recurrent CDI in the UK were treated with vancomycin followed by NTCD to prevent further recurrences, with one success and one failure. Epidemiologic studies of hospitalized patients using weekly rectal swab cultures demonstrated that asymptomatic colonization of patients by toxigenic *C. difficile* was much more common than CDI, but also that the rate of asymptomatic NTCD colonization of patients was unexpectedly high. Development of molecular strain typing of *C. difficile* was instrumental in characterizing different strains of both toxigenic *C. difficile* and NTCD leading to identification of NTCD strains that were effective human colonizers. These strains were reintroduced in hamsters in the 1990s and shown to prevent CDI efficiently and durably when challenged with epidemic toxigenic *C. difficile* strains. One strain of NTCD, NTCD-M3, was manufactured under cGMP standards and was demonstrated to be safe in a phase 1 volunteer trial. NTCD-M3 was then tested in a phase 2 double-blind placebo controlled trial for the prevention of recurrent CDI in patients experiencing their first CDI episode or first CDI recurrence. NTCD-M3 was given at doses of 10<sup>4</sup> or 10<sup>7</sup> spores per day orally for 7 or 14 days following successful treatment of CDI with vancomycin and/or metronidazole. CDI recurred in 30% of placebo patients and 11% of all NTCD-M3 patients (*p* = 0.006); recurrence rate for the best dose, 10<sup>7</sup> spores/d × 7 days, was 5% (*p* = 0.01 vs. placebo). Detection of colonization predicted prevention success; among the 86 patients who were colonized with NTCD-M3 the recurrence rate was 2% vs. 31% in patients who received NTCD-M3 but were not colonized (*p* < 0.001). Additional trials of NTCD-M3 for primary prevention of CDI and prevention of CDI recurrence seem warranted by these promising results.

Keywords: non-toxigenic, Clostridium difficile, NTCD-M3, prevention, NTCD, colonization, spores, Clostridioides difficile

## INTRODUCTION

The purpose of this review is to summarize the experimental, pre-clinical, and clinical data for the use of non-toxigenic Clostridioides difficile (NTCD) for prevention of C. difficile infection (CDI) caused by toxigenic strains of C. difficile. NTCD strains in which the pathogenicity locus (PaLoc) for the main C. difficile virulence factors, toxin A and B is replaced by a 115 bp sequence occur naturally. Multiple investigators have published convincing results of experiments over the last 35 years that demonstrate the efficacy of NTCD as a colonizing bacterial agent that prevents infection with toxigenic strains of C. difficile in animals and humans. Although a definitive mechanism for this protection remains to be determined, the effectiveness of this approach in multiple animal models and human clinical observations and trials suggests that it is a safe and effective biotherapeutic strategy that could be used for both primary prevention of CDI and prevention of CDI recurrence.

## Background Bench Discoveries in the 1980s

The concept of NTCD inhibiting or interfering with the infectivity of a toxigenic strain of C. difficile was first reported by Wilson and Sheagren (1983) in the hamster model in 1983. They demonstrated that following cefoxitin administration, colonization with NTCD prior to challenge with toxigenic C. difficile increased hamster survival from 21% to 93%, however, durability of prevention was not determined. Simultaneous administration of the NTCD strain and a toxigenic C. difficile strain did not increase survival. Borriello and Barclay (1985) extended these observations by colonization of hamsters with three different strains of NTCD, M-1, D-1 and S-1 administered at doses ranging from 3 × 10<sup>6</sup> to 3 × 10<sup>9</sup> organisms followed by challenge 5 days later with a toxigenic strain B-1 at a dose of 3 × 10<sup>7</sup> organisms. Results of these experiments showed at 25 days post toxigenic challenge survival of 13 of 18 (72%) vs. 0/21 (0%) hamsters not given an NTCD strain. Additional experiments showed (1) that protection was dependent upon live NTCD organisms; heat-killed NTCD did not protect hamsters, (2) that protection was dependent upon NTCD colonization; decolonization with vancomycin resulted in no protection against toxigenic challenge, and (3) that protection was specific to C. difficile; attempts to colonize with C. perfringens, C. bifermentans, and C. beijerincki, failed, and C. sporogenes colonized but failed to protect against toxigenic C. difficile challenge.

Borriello and Barclay (1985) also attempted to identify the mechanism by which NTCD protected against toxigenic C. difficile. They showed that NTCD did not degrade or alter the potency of C. difficile toxin. They investigated the mucosal association of NTCD by harvesting cecum samples from three NTCD-colonized hamsters at day 51 of colonization, washing them eight times with brain heart infusion (BHI) broth and saline, then homogenizing the tissue and doing quantitative culture of the homogenate. They found association of NTCD with the cecal mucosa ranging from 6 × 10<sup>3</sup> to 3 × 10<sup>5</sup> cfu/g tissue, suggesting there may be association of NTCD with the cecal mucosa or alternatively, they suggested competition for essential nutrients. However, in the discussion they report that after day 25 death occurred in most animals and that toxigenic C. difficile and cytotoxin could be detected in the cecum of these animals. Sequential analysis of the fecal pellets in these animals detected a change over time from only NTCD recovery to recovery of toxigenic C. difficile or both. This was of concern and the authors speculated that the presence of toxin enabled toxigenic strains of C. difficile to have a colonization advantage over NTCD. In a subsequent publication, Borriello et al. (1988) showed that a virulent toxigenic strain B-1 of C. difficile was more adherent to mucosa of small and large bowel in the hamster than NTCD strain M-1. They also showed that injection of toxin into the cecum of hamsters enhanced colonization of M-1 in the colon, but not in the small bowel or cecum when compared to toxigenic strain B-1.

Similar observations of protection against toxigenic C. difficile by NTCD were shown in gnotobiotic mice by Corthier and Muller (1988). NTCD protected against toxigenic challenge from 18h to 10 days following NTCD in 100% of gnotobiotic mice. If there was no delay in administration of the toxigenic strain following NTCD, 60% of mice survived. Survival of mice beyond 10 days was not recorded.

These early bench and animal studies were not further pursued beyond the 1980s, most likely because the protective effect in hamsters did not appear to be durable. However, these observations provide a valuable early insight into the potential utility of NTCD as a preventive strategy for CDI.

## Bedside Clinical Observations

The first published report of the use of NTCD in patients was by Seal et al. (1987). Two patients with multiple recurrences of CDI were treated orally with a BHI culture of NTCD strain M-1 previously shown by Borriello and Barclay (1985) to prevent CDI in hamsters transiently. One ml of the 24h culture was diluted in 50 ml of milk (final concentration ∼1 × 10<sup>7</sup> cfu/ml) which was administered once daily for 3 days. **Patient A**, age 88 years, was treated for her first CDI episode with metronidazole and for episodes two and three with oral vancomycin. Although successfully treated with resolution of symptoms, she had persistent C. difficile in her stool for at least 3 days following the last vancomycin treatment but was negative on stool culture just prior to receiving NTCD. Following NTCD her stool cultures were positive for NTCD for 14 days with no toxigenic C. difficile detected, and she had no further CDI recurrences for at least 4 months. **Patient B**, age 76, was also treated for her first CDI episode with metronidazole and for episodes two and three with vancomycin. She was treated successfully for her diarrhea symptoms with vancomycin, but had high counts of toxigenic C. difficile in her stool for the first 3 days of NTCD administration (5.9–7.5 log10/gm stool) with no NTCD detected. NTCD did not appear in the stool until day 6 at which time there were ∼10<sup>7</sup> cfu/gm stool of both NTCD and toxigenic C. difficile, but on days 9 and 10 only toxigenic C. difficile was found at a concentration of ∼10<sup>5</sup> cfu/gm stool. Starting on day 17 she had 2 days of self-limited diarrhea with both toxigenic C. difficile and NTCD at ∼10<sup>3</sup> cfu/gm stool and stool toxin detected. On day

28 counts of both toxigenic C. difficile and NTCD had declined to ∼10<sup>2</sup> cfu/gm and on day 45 neither were detected in stool. The patient died 1 month later of an unrelated cause but had no further diarrhea. In view of the mild recurrence of diarrhea with stool toxin detected at day 17 it is likely that this patient failed preventive treatment with NTCD, whereas patient A was a success. No further patient treatments with NTCD were reported until well into the 21st century, however, considerable clinical information on detection of NTCD in patients was obtained in the interim.

In the 1980s it became apparent that CDI was a healthcare associated infection (Gerding et al., 1986; McFarland et al., 1989), and typing systems for identification of specific C. difficile strains were developed (Kuijper et al., 1987; McFarland et al., 1989; Clabots et al., 1993) based not only on isolates from patients with CDI, but also from specimens collected from asymptomatic patients using rectal swabs (McFarland et al., 1989; Johnson et al., 1990; Clabots et al., 1992). McFarland et al. (1989) were the first to detect the presence of large numbers of patients in hospital that were asymptomatically colonized with C. difficile. They demonstrated that transmission of C. difficile was frequent in the hospital setting using immunoblot organism typing and space/time relations. They found that nearly 2/3 of patients that acquired C. difficile remained asymptomatic whereas just over 1/3 had symptomatic CDI (McFarland et al., 1989).

The use of restriction endonuclease analysis (REA) typing of C. difficile markedly enhanced the epidemiologic investigation of CDI in hospitals (Kuijper et al., 1987; Clabots et al., 1993). Early in the development of REA there were 75 distinct groups identified by letters of the alphabet, 43 cytotoxin positive groups, 28 NTCD groups, and 4 groups that included both cytotoxin positive and negative isolates. Isolates for typing were obtained from CDI patients, rectal swabs of asymptomatic patients and multiple environmental sites. In comparison with other typing systems then in use REA demonstrated greater sensitivity than immunoblot typing, bacteriophage-bacteriocin typing and ribotyping (Clabots et al., 1993). Remarkably, the most frequently isolated REA group was the NTCD group M which was found more commonly than any toxigenic groups and may be a reflection of the inclusion of large numbers of isolates collected from asymptomatic patients using rectal swabs.

Using a similar weekly rectal swab methodology used by McFarland et al. (1989) to detect C. difficile asymptomatic colonization coupled with REA typing of C. difficile (Kuijper et al., 1987; Clabots et al., 1993), Johnson et al. (1990) showed that 21% of patients hospitalized on three wards were culture-positive for C. difficile, the majority 51 of 60 (85%) were asymptomatic. All of the 9 cases of CDI were caused by two closely related toxigenic REA types, B and B2, and none occurred among the 51 asymptomatic colonized patients, 9 of whom were colonized with the B or B2 strain types but remained asymptomatic. Fifteen of the 17 patients with the B or B2 isolates were located on the same surgical ward suggesting transmission on that ward. Among the 18 different REA type C. difficile isolates found on the 3 study wards, 10 were toxigenic and 8 (44%) were nontoxigenic including 7 distinct REA groups of NTCD, groups A, C, M, P, S, T, and U. This was the first indication that NTCD strains were carried frequently by hospital patients. In addition, it was found that asymptomatic C. difficile colonization occurred in approximately the same frequency on all 3 hospital wards, ranging from 12.2 to 19.3% of patients, all of whom had relatively long hospital stays. All the patients with toxigenic C. difficile or NTCD in their stool remained asymptomatic.

A larger similar study using rectal swab cultures was conducted on a single surgical ward with a historically high rate of CDI by Clabots et al. (1992). The study was conducted over 9 months and enrolled 634 (94%) of 678 admissions to this ward, obtaining rectal swab cultures on admission and weekly thereafter. Of the 634 ward admissions, 65 (10.3%) were either asymptomatically culture-positive for C. difficile (61 patients) or had CDI (4 patients). Twelve additional asymptomatic admissions were positive for C difficile when first sampled, but were not sampled within 48 h of admission and were indeterminate as to whether they acquired C. difficile on the ward or were already colonized when admitted. The majority of admissions (355, 56%) were admitted from home and had not been hospitalized within the previous 30 days. Twenty-two (7%) of these home admissions were culture-positive for C. difficile indicating the high potential for introduction of C. difficile strains to the hospital from the community. Of the 557 admissions to the ward that were not culture-positive for C. difficile, 54 (9.7%) acquired C difficile while on the ward, but only 3 had CDI. REA typing identified 58 different distinct types during the study. Of the top seven most frequently isolated REA types, three were NTCD types and constituted 23/68 (34%) of these isolates. The most commonly isolated NTCD strain type was M3 which subsequently, together with types M23 and T7, was selected for laboratory trials in hamsters for prevention of CDI.

Low rates of CDI among patients previously asymptomatically colonized with C. difficile were observed in several trials utilizing weekly rectal swab cultures and suggested that colonization might be protective against CDI if symptoms did not develop relatively quickly following toxigenic C. difficile exposure, or if patients were colonized by NTCD. Data from four such prospective studies were combined in a random effects model by Shim et al. (1998). All four studies employed similar weekly rectal swab cultures for C. difficile in asymptomatic hospitalized patients and utilized stool cultures for C. difficile if CDI symptoms developed. Patients recovering from prior CDI were excluded as they have a high colonization rate with toxigenic C. difficile post treatment and are prone to recurrences of CDI. Asymptomatic patients were included if they had at least two consecutive weekly C. difficile rectal swab cultures and were classified as either colonized or not colonized. Colonized patients were further categorized as colonized with a toxigenic C. difficile strain or NTCD. All isolates underwent REA typing as well as toxin testing. Among the 618 non-colonized patients, 22 (3.6%) developed CDI whereas only 2 (1%) of 192 colonized patients developed CDI (pooled risk difference −2.3%, p = 0.021, **Figure 1**). When the analysis was confined only to patients who had received antibiotics within 14–21 days of the start of the sampling or during sampling, the pooled risk difference was −3.2%, p = 0.024. Greater exposure to antibiotics in noncolonized patients could explain the increased rate of CDI, but

studies with a random effects model is shown in bottom row (Total). From Shim et al. (1998).

antibiotic exposure was lower (79%) in non-colonized patients compared to colonized patients (92%). The mechanism by which colonization prevents CDI is unknown, and toxigenic strains of C. difficile as typed by REA that caused CDI were also found to be carried asymptomatically suggesting that strain type did not explain prevention. In these 4 studies there was a very high proportion (44%) of patients that were colonized by NTCD. There were 15 different REA groups of NTCD found among the 76 NTCD isolates, but the majority of NTCD were in two REA groups, M (49%) and T (19%). No CDI occurred among NTCD colonized patients and may account at least partially for the reduced CDI rates in colonized patients.

Taken together these clinical observations documented the relatively high rate of colonization of hospitalized patients by non-toxigenic strains of C. difficile and documented a very low rate of CDI among all colonized patients. Furthermore, REA typing demonstrated the ability to group and type NTCD strains in a manner similar to toxigenic C. difficile and identified two dominant groups of NTCD, the M and T groups by REA typing.

### Back to Bench Investigation: Pre-clinical NTCD Development

Bedside clinical and epidemiologic observations coupled with the evolution of molecular typing of C. difficile prompted further laboratory testing of C. difficile, particularly in animal models. Of specific interest was whether strains of C. difficile responsible for hospital outbreaks were more virulent than other strains that were found to asymptomatically colonize patients more often than they caused CDI. Sambol et al. (2001) tested three toxigenic epidemic C. difficile strains, REA types B1, J9, and K14 in the hamster model and compared them to toxigenic REA type Y2 which was found to commonly colonize patients but caused little CDI. A fifth strain, toxin variant REA type CF2, which is toxin A-negative, toxin B-positive and associated with CDI outbreaks was also included. The hamster protocol was similar to that of Borriello and Barclay (Borriello and Barclay, 1985), but differed in that only spores of C. difficile were administered by gavage 5 days following a single oral clindamycin dose of 30 mg/kg. A minimum infective dose for each C. difficile strain was predetermined to be 100 spores to mimic a postulated low inoculum exposure. The epidemic strains B1, J9, and K14 infected hamsters within 24–48 h of administration and caused 100% mortality by 48 h. REA type Y2 infected 9 of 10 hamsters over a range of 1–4 days, somewhat more slowly than the epidemic strains, but eventually all 9 infected hamsters died. Toxin variant strain CF2 infected only 6 of 10 hamsters over a range of 1 to 10 days and only 3 of the 6 died; the remaining 3 hamsters were colonized through study termination at day 79 and appeared to have no symptoms. The study documented differences in virulence of C. difficile strains in the hamster model and served as a basis on which to test preventive strategies with NTCD in the hamster model.

### NTCD Prevention

To test the effectiveness of NTCD as a preventive of infection with toxigenic C. difficile, the above described hamster model was used (Sambol et al., 2002). Three NTCD strains were selected from among isolates collected with rectal swabs from asymptomatic hospitalized patients; M3, M23, and T7, all of which were isolated at high frequency from patients (**Figure 2**). The NTCD inoculum was 10<sup>6</sup> spores by oral gavage on day 2 following clindamycin 30 mg/kg orally. Toxigenic C. difficile challenge with strains B1, J9, or K14 with 100 spores by oral gavage occurred on day 5 following clindamycin (3 days after the NTCD inoculum). Prevention of CDI mortality ranged from 87 to 97% (**Figure 3**) and correlated closely with detection of NTCD colonization of the stool prior to challenge with toxigenic C. difficile strains. Colonization with NTCD persisted in these surviving hamsters until end of study which varied from 55 to 106 days and there was no late mortality from toxigenic C. difficile.

### NTCD Prevention Durability

Durability of protection against late challenge with toxigenic C. difficile was performed with M3 and M23 NTCD. In the first of these studies M3 colonized 5 of 5 hamsters on day 4 following inoculation on day 2 following clindamycin. All 5 hamsters lost colonization between days 23 and 44. The uncolonized hamsters were challenged with 100 spores of toxigenic B1 C. difficile on day 62 and survived until day 153 end of study. In a second M3 experiment 3 of 4 hamsters were colonized on day 3 following inoculation on day 2 post clindamycin and were challenged on day 63 with 100 spores of toxigenic J9 and all survived with 3 of 4 remaining colonized with M3 until end of study day 125. In a third experiment, 12 hamsters were given clindamycin on day 0; 10 received NTCD strain M23 on day 2 and became colonized by day 5. All 12 hamsters were challenged with toxigenic strain B1 on day 41 and all survived (10 colonized with M23 and 2 uncolonized). All 10 remained colonized until day 49 when the study ended. These studies suggest that hamsters apparently recover colonization resistance following clindamycin by days

41–62 and are protected from late challenge with toxigenic C. difficile whether colonized with NTCD or not.

Patients are occasionally treated with a single antibiotic dose, as for example when undergoing surgical prophylaxis, and would be representative of the experiments with hamsters following a single dose of antibiotic. However, most infection treatment is carried out using multiple doses, often on the same day and for multiple days at a time. This raises the question of how NTCD might be employed to prevent CDI since many of the antibiotics used systemically will be active against NTCD and presumably prevent it from colonizing the patient. NTCD REA types M3, M23, and T7 are susceptible to most antibiotics including ampicillin, clindamycin, erythromycin, tetracycline, trimethoprim-sulfamethoxazole and metronidazole, but resistant to ciprofloxacin and ceftriaxone (Sambol et al., 2002), patterns that are typical of most C. difficile isolates. Resistance to clindamycin, tetracyclines, carbapenems and erythromycins as well as very high-level resistance to fluoroquinolones is known to occur in C. difficile and may be transferrable. Whereas use of a highly antibiotic resistant strain of NTCD would favor more frequent colonization, it would also increase the risk of antibiotic resistance transfer to other bacteria in the gastrointestinal tract.

### Effect of Daily Clindamycin and Resistance in NTCD

To assess the effect of antibiotic resistance of NTCD on colonization and prevention of infection by toxigenic C. difficile, Merrigan et al. (2003) conducted experiments in hamsters treated with daily oral doses (30 mg/kg) of clindamycin for 5 consecutive days. Two strains of NTCD, REA type M3 (clindamycin susceptible at 0.5 mcg/ml), REA type M13 (clindamycin resistant at >256 mcg/ml) and a toxigenic strain of C. difficile, REA type B1 (clindamycin resistant at >256 mcg/ml) were used. Clindamycin resistance in type M13 and B1 was determined to be due to presence of the erm(B) gene. M3 was negative for erm(B). Colonization experiments with M3 and M13 were conducted in groups of 5 hamsters by administering (1) a single dose of 1 × 10<sup>6</sup> spores on day 3 of clindamycin, (2) a daily dose of 1 × 10<sup>6</sup> spores on days 3–5 of clindamycin or (3) a daily dose of 1 × 10<sup>6</sup> spores on days 3–7. Colonization occurred with all doses of resistant strain M13 within 1.25 days of the first dose. No colonization occurred following the single dose of susceptible strain M3, but following the 3-day and 5-day administrations all 10 hamsters eventually became colonized but after a mean lag time of 6.4 to 6.6 days following the first dose. When the experiment was repeated with challenge with toxigenic strain B1, 100 spores orogastrically, on day 5, day 7, or day 9, all hamsters were protected by all doses of clindamycin resistant M13. Clindamycin susceptible M3 gave no protection against challenge with B1 on day 5, but protected 1 of 5 hamsters challenged on day 7, and 5 of 5 challenged on day 9. The results of the experiment clearly show the advantage of an antibiotic resistant NTCD strain in preventing CDI, but also suggest that an antibiotic susceptible strain can be effective if administered daily past the end of antibiotic administration,

albeit without as many days of prevention due to the longer time required to establish colonization in the presence of an antibiotic to which it is susceptible. Obviously, a toxigenic C. difficile strain resistant to the antibiotic being given has an advantage over susceptible NTCD while the antibiotic is being given, an advantage that is lost within a few days of stopping the antibiotic.

### NTCD Prevention After Daily Exposure to Antibiotics Other Than Clindamycin

Similar experiments were conducted to determine NTCD protection during continuous daily administration of ampicillin or ceftriaxone (Merrigan et al., 2009) with the exception that the dose of the toxigenic challenge strain was raised to 1 × 10<sup>6</sup> spores to further test NTCD prevention. Most strains of C. difficile are resistant to cephalosporins including ceftriaxone, and are susceptible to penicillins such as ampicillin. In the ceftriaxone experiments the antibiotic was given intraperitoneally (60 mg/kg) daily for 5 days and NTCD strains M3, M23, or T7 were given to groups of 5 hamsters orogastrically at a dose of 1 × 10<sup>6</sup> spores 3 h following the first ceftriaxone dose. The toxigenic challenge strain (REA type J9, 1 × 10<sup>6</sup> spores orogastrically) was responsible for multiple outbreaks of CDI in U.S. hospitals in the 1990s (Johnson et al., 1999). All the strains used were resistant to ceftriaxone; M3 and M23 MIC = 128 mg/mL; T7 MIC = 96 mg/mL; and J9 MIC = >256 mg/mL. Results are shown in **Figure 4** and demonstrate colonization within 48h for all NTCD strains and protection from J9 challenge on day 3 in all hamsters. A single control animal died within 48h as expected with J9 challenge.

Protection against daily administration of an antibiotic such as ampicillin to which NTCD and toxigenic C. difficile are both susceptible poses a different challenge (Merrigan et al., 2009). In these experiments NTCD REA type M3 (ampicillin MIC=2.0 mcg/ml) was utilized for prevention (1 × 10<sup>6</sup> spores daily) and challenged at intervals of 2 days with 1 × 10<sup>6</sup> spores of toxigenic

administration. Animals are indicated by ovals: open ovals indicate no detection of *C. difficile* in stool; gray-shaded ovals indicate detection of non-toxigenic *C. difficile* in stool; black ovals indicate detection of toxigenic strain J9 in stool. An X superimposed over an oval indicates the day of death. IP, intraperitoneally. From Merrigan et al. (2009).

strain J9 (ampicillin MIC = 0.75 mcg/ml). Ampicillin was given orogastrically (60 mg/kg daily) to groups of 4 hamsters for 5 days. As expected based on pilot studies, neither M3 or J7 colonized during ampicillin administration, but were able to establish colonization by day 8 if given near the end of the ampicillin course, indicating that following fecal excretion of the inhibitory antibiotic colonization could be established. Experiments were then designed to determine if NTCD M3 administered daily for 5 days to 3 groups of 4 hamsters beginning on day 1, day 3, and day 5 could protect against toxigenic J9 challenges on the middle day of each 5-day M3 course, days 3, 5, and 7, with both strains administered orogastrically at 1 × 10<sup>6</sup> spores. The study was designed to simulate periodic exposure to toxigenic C. difficile while being given NTCD daily. The experiment is shown schematically in **Figure 5** and demonstrates absence of J9 infection despite no colonization by either strain during the 5 days of ampicillin, followed by progressive colonization by M3 on days 7 and 8 in all hamsters. Results of these experiments suggest that in the presence of daily administration of an antibiotic to which NTCD is susceptible, that NTCD should be given daily and for several days after cessation of the antibiotic to attain successful colonization. Toxigenic C. difficile is not likely to infect during antibiotic administration and daily NTCD prophylaxis unless the toxigenic strain is resistant to the antibiotic and NTCD is susceptible.

### NTCD Prevention Against the Highly Virulent BI Strain of C. difficile

Early in the twenty first century it became apparent that CDI rates, morbidity and mortality were increasing in North America and were attributed at least in part to the presence of an epidemic strain characterized as BI group by REA, NAP1 by pulsed field gel electrophoresis, and 027 by PCR ribotyping and referred to as NAP1/BI/027 (Loo et al., 2005; McDonald et al., 2005). This strain spread throughout the United States and Canada as well as the United Kingdom and much of Europe as CDI rates rose precipitously in these countries. NAP1/BI/027 carries a third toxin, binary toxin and when compared to historic isolates of the same type it was apparent that high-level fluoroquinolone resistance had been acquired (McDonald et al., 2005). This prompted study of these strains in the hamster model by Razaq et al. (2007) who compared epidemic BI strains BI6 and BI17 to historic non-epidemic BI strain BI1 and 2 standard toxigenic strains, K14 and 630 (REA type R23). In these studies groups of 10 hamsters were given 100 spores of each strain of C. difficile orogastrically 5 days following a single clindamycin dose (30 mg/kg orally). Although group BI strains were not more rapidly fatal than standard toxinotype 0 strain K14, they were more rapidly fatal than strain 630 and epidemic strain BI6 was the most rapidly fatal of all the strains tested.

The widespread clinical presence of epidemic strain NAP1/BI/027 raised the question of whether NTCD strains would be effective in preventing disease due to BI strains in the hamster model. Nagaro et al. (2013) gave groups of 12 hamsters oral clindamycin (30 mg/kg orally) followed 2 days later by 1 × 10<sup>6</sup> spores of NTCD strains M3 or T7 in 10 hamsters, and on day 5 challenged all 12 with 100 spores of historic strain BI1 or epidemic strain BI6. All hamsters colonized with M3 or T7 and were protected against challenge with historic strain BI1. NTCD strain M3 colonized 9 of 10 hamsters and protected against BI6 challenge in the 9 colonized hamsters, but not the uncolonized animal. However, epidemic strain BI6 proved more difficult for protection by NTCD strain T7. For the first time in these hamster experiments we found NTCD colonization

FIGURE 5 | Colonization with non-toxigenic *Clostridium difficile* M3 and challenge with toxigenic strain J9 during daily oral ampicillin administration for 5 days. Both J9 and M3 are susceptible to ampicillin. Animals are indicated by ovals: open ovals indicate no detection of *C. difficile* in stool; gray-shaded ovals indicate detection of non-toxigenic M3 in stool. PO, orally. From Merrigan et al. (2009).


FIGURE 6 | Hamsters (*n* = 10/group) challenged with epidemic toxigenic BI6 *Clostridium difficile*. Day 2 = 2 days post-clindamycin treatment. *White ovals*; uncolonized hamsters. *Gray ovals*; non-toxigenic colonized hamsters. *Striped ovals*; hamsters co-colonized with non-toxigenic and toxigenic *C. difficile*. *Black ovals with* "*X*"; hamster death from toxigenic BI6. From Nagaro et al. (2013).

displaced by the toxigenic strain (**Figure 6**). Five of 10 hamsters were co-colonized with T7 and BI6 on day 7 and 4 of the 5 died. The remaining co-colonized hamster lost BI6 colonization on day 21 and survived. These results suggest that toxigenic and NTCD strains of C. difficile compete for colonization and that certain strains such as BI6 (but not its historic predecessor BI1) despite their lower inoculum and later administration are able to compete more effectively than other toxigenic strains against NTCD strain T7, but not against strain M3. Thus, differences in colonization efficiency are evident for both toxigenic strains of C. difficile as well as NTCD strains.

### Possible Mechanism of NTCD Prevention

The mechanism by which C. difficile strains establish colonization is not established. One possibility is that strains differ in their ability to adhere to colonic mucosal cells or mucus. Merrigan et al. (2013) utilized a derivative of the Caco-2-derived human intestinal epithelial cell-line to assess adherence of 33 strains of toxigenic C. difficile (with an emphasis on epidemic NAP1/BI/027 and ribotype 078 strains) and 3 NTCD strains (REA types M3, M23, and T7) in this assay with a focus on differences in surface layer protein A (SlpA) in the strains. The average adherence range of vegetative cells was approximately 8–11% for

FIGURE 7 | Prevention of recurrence with NTCD-M3 following vancomycin treatment of toxigenic B1 *C. difficile* infection in hamsters. Day 5 = 5 days post-clindamycin treatment. *White ovals*: uncolonized hamsters. *Black ovals*: toxigenic infected hamsters. *Gray ovals*: non-toxigenic colonized hamsters. *Black ovals with* "*X*": hamster death from toxigenic B1. From Sambol et al. (2004).

strain 630, 2–4% for strain K14, 5–9% for strain BI17, 8–14% for NTCD strain M3, and 9–12% for NTCD strain T7. As a group, BI/NAP1/027 strains exhibited a mean adherence value of 4.6%. Further experiments showed that pretreatment of cell layers with surface layer protein (SLP) preparations inhibited C. difficile adherence up to 80% with the maximum SLP dose. SLP inhibition was not strain specific; NTCD M3 SLP inhibited M3 and BI17 adherence equally, and BI17 SLP inhibited BI17 and M3 adherence equally. Sequence analysis of the SlpA proteins for both the high molecular weight and low molecular weight subunits was compared among strains. One might postulate that there would be a high degree of conservation of amino acid sequences among highly competitive strains, yet NTCD strain M3 displayed the least sequence conservation, especially in the low molecular weight subunit, compared not only to epidemic toxigenic strains, but also when compared to NTCD strain T7.

### NTCD Prevention of Recurrent CDI

Patients who have CDI incur about a 20–30% risk of having a recurrence of their symptoms following successful treatment. If

such patients could be colonized with NTCD following antibiotic treatment of CDI, it could prevent subsequent recurrence. To test this possibility, the CDI treatment model in hamsters that demonstrated late recurrence of CDI mortality following treatment with vancomycin was modified to administer NTCD M3 following vancomycin treatment (Swanson et al., 1991; Sambol et al., 2004). Sambol et al. (2004) gave a group of 15 hamsters oral clindamycin (30 mg/kg) followed 5 days later by 1 × 10<sup>4</sup> spores of epidemic strain B1. Vancomycin was administered orogastrically in doses of 100 mg/kg daily for 3 consecutive days on Days 6, 7, and 8 with the first dose administered within 14 h of the initial B1 challenge. On Days 9, 10, and 11, NTCD strain M3 was orally administered to 10 of the 15 hamsters at 1 × 10<sup>6</sup> spores/day per hamster (5 control hamsters did not receive NTCD). On Day 25, all surviving hamsters were given a re-challenge dose of 1 × 10<sup>4</sup> toxigenic B1 spores to simulate re-exposure to the toxigenic strain (**Figure 7**). Of the control hamsters, one out of 5 was infected with B1 three days after vancomycin treatment, and died on Day 15, a spontaneous B1 recurrence. After the B1 re-challenge on Day 25, 4 of 4 surviving control hamsters became infected with B1 between Days 27 and 30, and died by Day 30.

Of the hamsters given NTCD, 9 of 10 colonized with M3 between Days 10–16; one hamster remained uncolonized until Day 17, when it became infected with epidemic strain B1 and died, a spontaneous CDI recurrence. After the B1 re-challenge on Day 25, all 9 surviving hamsters remained colonized with M3 until study end on Day 70 with no evidence of B1 infection. These results demonstrate that NTCD colonization following vancomycin treatment is successful and will prevent recurrent C. difficile infection.

### Potential for NTCD Acquisition of Toxin Genes

Concern has been raised that in clinical use NTCD strains may acquire the toxin A and B pathogenicity locus (PaLoc) and become pathogenic. Brouwer et al. (2013) have shown in vitro that toxigenic strain 6301erm can transfer the PaLoc to NTCD strain CD37, which is REA type T18. PaLoc-containing transconjugants were obtained at a frequency of 7.5 × 10−<sup>9</sup> transconjugants per donor in filter matings. DNA fragments containing the PaLoc ranged in size from 66,034 to 272,977 bp.



\**p-values adjusted for pre-specified parameters: use of metronidazole vs. vancomycin, and primary episode vs. first recurrence.*

Transfers from 6301erm to two other NTCD isolates, PCR ribotypes 138 and 140, were also successful demonstrating that PaLoc transfer is not restricted to a specific donor-recipient pair. Supernatants of recipient strain CD37 following PaLoc transfer were found to cause cell rounding in the cell cytotoxicity assay that was neutralized by toxin B antibodies indicating that the PaLoc transfer resulted in functional toxin production. The mechanism by which PaLoc transfer takes place has not been precisely defined, but transfer via a cell-to-cell conjugationlike transfer mechanism has been proposed (Brouwer et al., 2016). The likelihood of such transfers occurring in vitro is unknown, but would presumably require high concentrations of a toxigenic C. difficile strain and NTCD in close proximity in the gastrointestinal tract of a host. In the hamster model, we have demonstrated that vancomycin administered daily for 3 days will eliminate NTCD detection in stool, and would presumably be effective in managing CDI if PaLoc transfers were to occur in patients.

### Back Again to Bedside: Clinical Trials

The above pre-clinical data were used to file investigational new drug applications with agencies in the US, Canada and Europe. Manufacturing of spores of NTCD-M3 under cGMP was undertaken successfully by ViroPharma, Inc. and the spores prepared in liquid suspension for safety testing in healthy volunteers at doses of 10<sup>4</sup> , 10<sup>6</sup> , and 10<sup>8</sup> spores/dose or placebo (Villano et al., 2012). NTCD-M3 was found to be safe in younger volunteers age 18–45 years at all three single doses given in the absence of any antibiotic pre-treatment. To better simulate the likely age of patients with CDI, volunteers age >60 years (age range 60–73 years) were given single doses of 10<sup>4</sup> , 10<sup>6</sup> , and 10<sup>8</sup> spores or placebo safely. Volunteers >60 years were then given the highest dose, 10<sup>8</sup> spores twice daily for 5 days and tolerated this safely. This was the only group in which NTCD could be detected in the stools of the volunteers in the absence of prior antibiotic administration. Stool cultures first became positive on days 2–4 and were positive in 6 of 8 volunteers at day 7 but in none at day 14 and 21 suggesting the possibility that this may have been pass-through of NTCD spores and not colonization.

Finally, groups of 12 >60 year old volunteers were pre-treated with 5 days of oral vancomycin 125 mg 4 × daily to simulate patients being treated for CDI to determine if colonization with NTCD would occur post vancomycin treatment. Each group of 9 NTCD and 3 placebo subjects was given 10<sup>4</sup> or 10<sup>6</sup> or 10<sup>8</sup> spores of NTCD-M3 or placebo daily for 14 days beginning the day following the last vancomycin dose. In each group the 9 volunteers given NTCD-M3 had NTCD detected in their stools during the 14-day administration and in 4 to 5 volunteers on days 21 and 28 following NTCD. In the 3 volunteers receiving placebo in the 10<sup>4</sup> and 10<sup>6</sup> NTCD-M3 spore cohorts one volunteer in each cohort became colonized with a toxigenic strain of C. difficile but remained asymptomatic. In the cohort receiving 10<sup>8</sup> spores of NTCD-M3 the 3 placebo patients all became colonized with NTCD-M3, indicating that at this highest dose of NTCD-M3 there was transmission of NTCD-M3 from the active recipient group or the environment to the placebo group who were housed on the same clinical trial facility. No safety issues were identified in the volunteers receiving NTCD-M3.

For the first time since Seal et al. (1987) reported the treatment of 2 CDI patients with NTCD in 1987, NTCD was again administered in a Phase 2 double blind, randomized, doseranging, placebo controlled trial of NTCD for prevention of recurrent CDI in patients who have had their first episode or first recurrence of CDI. The primary outcome of the study was safety, with fecal colonization rate, CDI recurrence rate, and optimal dosing of NTCD-M3 as exploratory secondary outcomes. The study design shown in **Figure 8** compared 2 doses, 10<sup>4</sup> and 10<sup>7</sup> spores given orally in liquid once daily for 2 durations, 7 days (10<sup>7</sup> spores) and 14 days (10<sup>4</sup> and 10<sup>7</sup> spores) compared to placebo for 14 days. NTCD-M3 was begun 1–2 days following completion of successful treatment of CDI with vancomycin or metronidazole (Gerding et al., 2015). A total of 168 patients were randomized and treated and 157 completed treatment. The four arms of the study were well balanced in terms of patient age (range 18– 94 years) with 39% >65 years old. Most patients were having their primary CDI episode (83%), were outpatients (76%), and were treated with metronidazole (60%). Vancomycin was used for treatment in 20% of subjects and 20% were treated with both vancomycin and metronidazole.

Safety results indicated that participants who received NTCD-M3 had fewer treatment emergent adverse events and fewer serious adverse events than placebo patients with the exception of headache which occurred in 10% of NTCD-M3 recipients and 2% of placebo (Gerding et al., 2015). No serious adverse events were considered treatment related. Colonization by NTCD-M3 was defined as detection of NTCD-M3 in stool at any time following completion of treatment. **Figure 9** demonstrates that colonization with toxigenic C. difficile following antibiotic treatment of CDI is common (63% of placebo subjects at week 2) whereas with administration of NTCD-M3, much of the toxigenic C. difficile colonization was replaced by NTCD-M3; 71% of participants who received 10<sup>7</sup> spores and 63% who received 10<sup>4</sup> spores. Colonization with NTCD-M3 was maximal in weeks 1–3 and slowly declined thereafter until last detected in week 22. A small number of participants still harbored toxigenic C. difficile at 26 weeks, the last time stools were sampled.

Prevention of recurrence of CDI is summarized in **Table 1**. Using the protocol definition of recurrence of CDI at 6 weeks, recurrence rate was 13 (30%) of 43 placebo subjects and 14 (11%) of 125 NTCD-M3 subjects (odds ratio [OR], 0.28; 95%CI, 0.11– 0.69; P = 0.006); the lowest recurrence was in 2 of 43 (5%) patients receiving 10<sup>7</sup> spores/day for 7 days (OR, 0.1; 95%CI, 0.0–0.6; P = 0.01 vs. placebo). Using an alternate definition of recurrent CDI defined as use of antibacterial treatment for CDI, the recurrence rate was 14 of 43 (33%) placebo subjects and 17 of 125 (14%) NTCD-M3 subjects (p = 0.009). NTCD-M3 also reduced all-cause diarrhea from 77% in placebo subjects to 57% of NTCD-M3 subjects (p = 0.020). There was good correlation of CDI prevention with colonization by NTCD-M3; among subjects who received NTCD-M3 and became colonized, the CDI

REFERENCES


recurrence rate was 2 of 86 (2%) vs. 12 of 39 (31%) subjects who received NTCD-M3 and were not colonized (p < 0.001). Doseranging indicated that 10<sup>7</sup> spores were more effective than 10<sup>4</sup> spores, and there was no advantage to extending treatment to 14 days vs. 7 days for the 10<sup>7</sup> dose. In summary, NTCD-M3 was safe and effective in preventing CDI in patients who were experiencing their first CDI episode or first recurrence of CDI.

### SUMMARY

NTCD has a long history of basic laboratory and clinical experience that suggests it is a safe and very effective preventive for recurrent CDI (hamster and human data) and as a preventive for first episode CDI if used in subjects receiving antibiotics (hamster data only). Further development is required with phase 3 clinical trials for recurrent CDI prevention and for phase 2 and 3 primary CDI prevention clinical trials.

### AUTHOR CONTRIBUTIONS

DG wrote the first manuscript draft and designed the experiments described from his laboratory and provided funding. SJ and SS edited the manuscript, contributed text and figures, and were instrumental in designing and carrying out the described experiments in the DG laboratory.

### FUNDING

This work has been supported by multiple Merit Review grants from the Department of Veterans Affairs Research Service.


administration of a non-toxigenic strain. Eur. J. Clin. Microbiol. 6, 51–53. doi: 10.1007/BF02097191


**Conflict of Interest Statement:** DG holds patents (unlicensed, no royalties) and technology for the use of NTCD for prevention of CDI and is Scientific Advisory Board Member for Rebiotix, Actelion, Merck, Summit and DaVolterra, is a consultant for Pfizer, MGB Pharma, and Sanofi Pasteur, and holds a research grant from Seres Therapeutics. SJ is Scientific Advisory Board Member for Bio-K+, Synthetic Biologics, Summit and CutisPharma.

The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Gerding, Sambol and Johnson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Efficiency of Biosynthesized Silver and Zinc Nanoparticles Against Multi-Drug Resistant Pathogens

Kapil Punjabi<sup>1</sup> \*, Sourabh Mehta<sup>2</sup> , Rujuta Chavan<sup>3</sup> , Vidushi Chitalia<sup>1</sup> , Dhanashree Deogharkar<sup>1</sup> and Sunita Deshpande<sup>1</sup>

<sup>1</sup> Department of Clinical Pathology, Haffkine Institute for Training, Research and Testing, Mumbai, India, <sup>2</sup> National Centre for Nanoscience and Nanotechnology, University of Mumbai, Mumbai, India, <sup>3</sup> Department of Microbiology, Guru Nanak Khalsa College, Mumbai, India

### Edited by:

Rebecca Thombre, Modern College, India

### Reviewed by:

M. Oves, King Abdulaziz University, Saudi Arabia Rajashree Bhalchandra Patwardhan, Savitribai Phule Pune University, India

> \*Correspondence: Kapil Punjabi kapil9372@gmail.com

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 19 March 2018 Accepted: 29 August 2018 Published: 20 September 2018

### Citation:

Punjabi K, Mehta S, Chavan R, Chitalia V, Deogharkar D and Deshpande S (2018) Efficiency of Biosynthesized Silver and Zinc Nanoparticles Against Multi-Drug Resistant Pathogens. Front. Microbiol. 9:2207. doi: 10.3389/fmicb.2018.02207 Biosynthesis of metallic nanoparticles has acquired particular attention due to its economic feasibility, low toxicity, and simplicity of the process. In this study, extracellular synthesis of silver and zinc nanoparticle was carried out by Pseudomonas hibiscicola isolated from the effluent of an electroplating industry in Mumbai. Characterization studies revealed synthesis of 40 and 60 nm nanoparticles of silver (AgNP) and zinc (ZnNP), respectively, with distinct morphology as observed in TEM and its crystalline nature confirmed by XRD. DLS, zeta potential, NTA, and FTIR studies further characterized nanoparticles giving data about its size, stability, and functional groups. Considering the toxicity of nanoparticles the evaluation of antimicrobial activity was studied in the range of non-toxic concentration for normal cell lines. Silver nanoparticles were found to be the most effective antimicrobial against all tested strains and drugresistant clinical isolates of MRSA, VRE, ESBL, MDR, Pseudomonas aeruginosa with MIC in the range of 1.25–5 mg/ml. Zinc nanoparticles were found to be specifically active against Gram-positive bacteria like Staphylococcus aureus including its drugresistant variant MRSA. Both AgNP and ZnNP were found to be effective against Mycobacterium tuberculosis and its MDR strain with MIC of 1.25 mg/ml. The synergistic action of nanoparticles assessed in combination with a common antibiotic gentamicin (590 µg/mg) used for the treatment of various bacterial infections by Checker board assay. Silver nanoparticles profoundly exhibited synergistic antimicrobial activity against drug-resistant strains of MRSA, ESBL, VRE, and MDR P. aeruginosa while ZnNP were found to give synergism with gentamicin only against MRSA. The MRSA, ESBL, and P. aeruginosa strains exhibited MIC of 2.5 mg/ml except VRE which was 10 mg/ml for both AgNPs and ZnNPs. These results prove the great antimicrobial potential of AgNP and ZnNP against drug-resistant strains of community and hospital-acquired infections and opens a new arena of antimicrobials for treatment, supplementary prophylaxis, and prevention therapy.

Keywords: silver nanoparticles, zinc nanoparticles, Pseudomonas hibiscicola, MRSA, VRE, MDR-TB

## INTRODUCTION

fmicb-09-02207 September 20, 2018 Time: 16:17 # 2

New-age researchers are dedicated to upgrading the current world to a better and easier habitation. To achieve this, nanotechnology will play an important role in newer and improved materials that are robust, economic, and sources of renewable energy and pollution-free environment. The future doctors will use devices made up of such enhanced materials to not only ease the process of diagnostics but also considerably reduce the time and help treatment at early stages. Simpler and faster diagnosis coupled with efficient and reliable treatments will help to cure cancer, diabetes, and various infections. This may seem easy but a difficult road-map lies in front of these researchers, but certainly leads with help of nanotechnology makes it look possible and a revolution seems to be around the corner (Bhattacharyya et al., 2009).

With many arms of nanotechnology being shaped, one major domain is the biomedical application. The biomedical applications of nanotechnology comprise of many sub areas predominantly therapeutic and diagnostic areas have received considerable attention. Different nanoparticles and nanomaterials owing to their small size can easily interact with biomolecules at cellular as well as sub cellular levels. This greatly enhances the signaling of markers in case of diagnostics and improved specificity for targeted therapeutics (Conde et al., 2012).

Multiple applications of nanotechnology across fields like medical, chemical, environmental, agricultural, industrial, energy sciences, and information technology. This attracts major attention from various organizations in both developed and developing nations throughout the world (Roco, 2007).

Many approaches for the synthesis of nanoparticles have been proposed principally categorized into three classes namely – physical, chemical, and biological. Both physical and chemical methods are the major source of nanoparticles in the current period. While these methods ensure continuous supply they also have price to pay in terms of high energy requirements, toxic byproducts. This highlights the requirement of sustainable and ecofriendly mode of synthesis. This brings into picture the biological methods with microorganisms playing a significant role. Thus, microbiologists around the world try to identify microorganisms for this capability and develop a simple and cost-effective method to synthesize nanoparticles of constant size, shape, and monodispersity (Babu et al., 2011; Horikoshi and Serpone, 2013).

Biosynthesized nanoparticles are ideal candidates for medical applications as they are biocompatible and naturally stable. One of the important property of such nanoparticles is antimicrobial activity. Many green silver nanoparticles have been reported for antibacterial and antibiofilm activity through mechanisms like ROS generation (Qayyum et al., 2017). Also supportive studies on plant derived silver nanoparticles conclude that biogenic nanomaterials are biocompatible and an effective therapeutic agent against bacterial, fungal infections, and cancer treatment (Oves et al., 2018).

Although nanoparticles are considered as future to many areas of science and technology, its probable environmental toxicity is a matter of concern. Thus, in this study, we synthesized nanoparticle, using bacteria isolated from a challenged environment like soil and effluent of electroplating industry and characterized to check potential therapeutic applications if rendered usable by evaluation of its in vitro toxicity.

## MATERIALS AND METHODS

## Synthesis and Characterization of Nanoparticles

The synthesis and characterization of AgNP was followed by our previously reported work using the bacterial culture of Pseudomonas hibiscicola isolated from effluent of an electroplating industry. The culture inoculated in sterile nutrient broth and incubated at 37◦C for 24 h in an orbital shaker. After incubation, it was centrifuged till clear supernatant was achieved. The supernatant was subjected to 1 mM concentration of silver nitrate AgNO3 (Sigma-Aldrich, United States) and incubated on a shaker with similar conditions. The bioreduction of silver ions was monitored by visual observation and measuring absorption spectrum of samples using UV–vis Spectrophotometer Thermo scientific Nanodrop – 1000 v3.7 scanning range from 200 to 800 nm (Punjabi et al., 2017). For synthesis of ZnNP, similar procedure was followed; briefly the culture supernatant of P. hibiscicola was challenged with 10 mM zinc acetate [Zn(O2CCH3)2] (Himedia labs Pvt. Ltd., India) in 1:2 v/v ratio and incubated in similar conditions. After 24 h of incubation, precipitate formation and turbidity indicated presence of ZnNP in the reaction mixture. Reduction of metal ion to nanoparticle was confirmed by observing plasmon peak in the range of 320–380 nm in UV–Vis spectra (Waghmare et al., 2011).

To determine the size, morphology, and other characters of synthesized ZnNPs, characterization studies were undertaken. Suspension of zinc nanoparticles was subjected to UV–Vis spectroscopy, nanoparticle tracking, and analysis (NTA), DLS – particle size analyzer, and zeta potential analysis of the liquid suspension was done using NanoPlus DLS (Micromeritics Instrument Corporation). While powdered material was used for X-ray diffraction **(**XRD) patterns that were recorded on normal focus diffractometer (Regaku Miniflex, Japan) at voltage of 30 kV and current 15 mA with scan rate of 3◦ /min. The data were recorded in the range of 20◦–50◦ and analyzed using Jade 6.0 software. Fourier transform infrared (FTIR) spectra obtained on a JASCO spectrometer (6100 type-A) instrument in the range of 400–4000 cm−<sup>1</sup> were identified spectroscopy and TEM analysis on a Philips CEM 200 Microscope (Pattanayak, 2013).

### Toxicity of Nanoparticles

Toxicity of nanoparticles was assessed on Vero cell line (Monkey kidney cell line). The cell line, Vero passage number 27 was obtained from National Centre for Cell Sciences, Pune, India, in Dulbecco's Modified Essential Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (complete medium).

To evaluate the effect of AgNP and ZnNP on cell toxicity colorimetric MTT (3-(4, 5-dimethylthiazol-2)-2, 5 diphenyl tetrazolium bromide) assay was performed. The MTT assay is a

colorimetric technique wherein amount of absorbance is directly proportional to number of living cells. The MTT tetrazolium compound is reduced by viable cells into an intensely colored formazan precipitate that subsequently is solubilized into a uniformly colored solution with a second procedural step before absorbance is measured using a plate reading spectrophotometer (Freshney, 2010).

Known number (1 × 10<sup>4</sup> cells) of cells was seeded into tissue culture grade 96-well plates (Riss et al., 2011). Cells were treated with various concentrations (10.0–0.6 mg/ml) of nanoparticles for 24 h and treated with 0. 5% MTT (HiMedia Laboratories Pvt. Ltd., India).

After 4 h, all the contents of wells were removed and 100 µl of dimethyl sulfoxide (DMSO) was added. The absorbance of colored solution was measured on the Multimode reader (Synergy HTX Multimode reader, BIOTEK, India) using a test wavelength of 490 nm (Freshney, 2010; Satyavani et al., 2011). Percentage toxicity was estimated for each concentration, and the IC<sup>50</sup> value was calculated for both nanoparticles (Freshney, 2005).

## Antimicrobial Activity of Nanoparticles Bacterial Strains

### **Gram-negative bacteria**

Proteus mirabilis ATCC 21100, Escherichia coli ATCC 25922, Klebsiella pneumoniae ATCC 10031, K. pneumoniae ATCC 700603 (extended spectrum β lactamases producer, ESBL), Pseudomonas aeruginosa ATCC 27853, Enterobacter aerogenes ATCC 13048, Salmonella typhimurium ATCC 14028, Salmonella abony ATCC 6017, and Shigella boydii ATCC 8700.

### **Gram-positive bacteria**

Staphylococcus aureus ATCC 25923, Bacillus subtillis ATCC 6633, S. aureus ATCC 43300 (methicillin-resistant S. aureus, MRSA), S. aureus ATCC 6538 Enterococcus faecalis ATCC 51299 (vancomycin-resistant Enterococci, VRE), Staphylococcus epidermidis ATCC 12228, and Bacillus cereus ATCC 11778.

### **Mycobacteria**

Mycobacterium tuberculosis H37Rv.

### **Clinical strains**

MDR P. aeruginosa and MDR M. tuberculosis.

ATCC strains were procured from Clinical Pathology department, Haffkine Institute, microbial culture collection (MCC) Pune, India, and Himedia laboratories Pvt. Ltd., Mumbai, India. The MDR strains of clinical isolates were procured from tertiary care hospital in Mumbai. The drug resistance of the isolates was confirmed before use in the study as per standard protocols of Clinical and Laboratory Standards Institute [CLSI] (2012).

### Minimum Inhibitory Concentration

Antimicrobial activity of the nanoparticles synthesized was determined by broth dilution method and confirmation by growth on solid media (Wiegand et al., 2008; Ataee et al., 2011). For determining the minimum concentration of nanoparticles required for the inhibition of bacterial growth, broth dilution method was used according to standard broth microdilution method M07 A9 of Clinical and Laboratory Standards Institute [CLSI] (2012). This method facilitates the testing of inhibitory activity at various nanoparticle concentrations (Clinical and Laboratory Standards Institute [CLSI], 2012).

### Anti-mycobacterial Activity – Broth Dilution Method

The minimum inhibitory concentrations (MICs) of silver and zinc nanoparticles were assessed by broth dilution method against M. tuberculosis H37Rv and MDR strain. Tenfold serial dilutions of the inoculum suspension were made in M7H9 broth supplemented with Middlebrook OADC enrichment. The stock concentrations of nanoparticles were serially diluted to obtain concentrations in the range 1.25–10 mg/ml. The prepared bacilli suspension (500 µl) was added to all tubes except the control tubes, all the tubes were incubated at 37◦C for 7 days, which is physiological temperature favorable for growth of pathogenic strains. To determine the MIC, 10 µl of each tube was transferred to Lowenstein Jensen Medium (LJ medium) and incubated at 37◦C for 3 weeks. After incubation, growth on LJ media was analyzed for each concentration of both nanoparticles. The observations were recorded and MIC values were determined (Andrews, 2001; Jadaun et al., 2007).

### Synergistic Activity of Gentamicin and Nanoparticles

### Disk Diffusion Test for Gentamicin

Antibiotic susceptibility test (AST) against gentamicin (10 µg) and high-level gentamicin (120 µg) (HiMedia labs Pvt. Ltd.) was performed by the Kirby–Bauer disk diffusion test to study the drug-resistant pattern of the aforementioned organisms (Bauer et al., 1966; Clinical and Laboratory Standards Institute [CLSI], 2012).

### E-Test

To determine the MIC of gentamicin for these isolates EzyMICTM Strip (GEN) (0.016–256 mcg/ml) and EzyMICTM Strip (HLG) (0.064–1024 mcg/ml) (HiMedia labs Pvt. Ltd.) were used. The test was performed on Mueller Hinton agar (HiMedia labs Pvt. Ltd.) swabbing a sterile cotton-swab soaked in inoculum prepared by comparing the turbidity of direct colony suspension to standard 0.5 McFarland. The EzyMICTM Strip were placed at the center of the agar plate and incubated at 37◦C for 24 h. After incubation, the plates were read at the point of complete inhibition of all growth, including hazes, micro-colonies, and isolated colonies as indicated (Nachnani et al., 1992).

### MIC of Gentamicin

To determine the MIC of gentamicin, broth dilution method was used according to standard broth microdilution method M07 A9 of Clinical and Laboratory Standards Institute [CLSI] (2012).

### Checkerboard Assay

The stock solutions of gentamicin used ranged 2.5 to 160 µg/ml. The stock solutions of silver and zinc nanoparticles used ranged from 0.3 to 20 mg/ml. Double dilutions of both the antibiotic as well as the nanoparticles was used for the assay according to the recommendations of CLSI. A total of 50 µl of sterile nutrient

broth was distributed aseptically in all the sterile microtiter plate (96-well plate). The antibiotic gentamicin was serially diluted along the abscissa, while the nanoparticle was diluted along the ordinate. An inoculum equal to 0.5 McFarland turbidity standards was prepared for each culture in sterile saline. Each well was inoculated with 10 µl of the culture. The microtiter plate was incubated at 37◦C for 24 h. After incubation, the growth was observed by streaking a loopful from each well in sterile nutrient agar plates. The plates were incubated at 37◦C for 24 h. The plates were observed for growth of the test organism. The combination of the drugs in which the growth is completely inhibited was considered as effective MIC for the combination (Breitinger, 2010).

The 6FICs (fractional inhibitory concentrations) were calculated as follows:

### 6FIC = FIC A + FIC B

where FIC A is the MIC of drug A (gentamicin) in the combination/MIC of drug A (gentamicin) alone, and FIC B is the MIC of drug B (nanoparticle) in the combination/MIC of drug B (nanoparticle) alone. The combination is considered synergistic when the 6FIC is ≤0.5, indifferent when the 6FIC is >0.5 to <2, and antagonistic when the 6FIC is ≥2 (Rabadia et al., 2013). The overall study period was 12 months from material synthesis to performing all in vitro activities.

### RESULTS

### Synthesis and Characterization of Nanoparticles

Synthesis of AgNP and ZnNP was achieved extracellularly using P. hibiscicola isolated from effluent of electroplating industry. Polydispersed silver nanoparticles of average 40 nm in size were obtained as mentioned in the previous work. In case of ZnNP, typical white precipitate formation as seen in **Figure 1** indicated the synthesis of zinc nanoparticles. The UV–Vis spectra (**Figure 1**) confirm the synthesis of zinc nanoparticles in the reaction mixture.

Further characterization studies gave details about the nanoparticle size, morphology, surface charge, and crystalline nature. The dynamic light scattering (DLS) and zeta potential of the zinc nanoparticles (**Figure 2**) indicated synthesis of stable nanoparticles of size 110 nm and polydispersity index of −0.2. The zeta potential of the zinc nanoparticles was found to be 24.64 mV.

Fourier transform infrared **Figure 3** shows absorption bands at 3513, 2931, and 1020 cm−<sup>1</sup> indicating the presence of capping agent with the nanoparticles. The spectra correspond to secondary alcohol, alkane, and amine regions. These functional groups have a role in stability/capping of nanoparticles. XRD pattern **Figure 4** shows 2θ values at 31.96◦ , 34.58◦ , 36.38◦ , 47.5◦ , 56.6◦ , 62.8◦ , and 67.9◦ . All evident peaks could be indexed as the zinc oxide wurtzite structure (JCPDS Data Card No: 36- 1451), which confirms the synthesized nanoparticles were free of impurities as it does not contain any characteristics XRD peaks other than zinc oxide peaks.

Nanoparticle tracking and analysis of AgNP and ZnNP (**Figure 5**) shows the particle size distribution which was in the range of 10–70 nm for silver nanoparticles while the mean size was found to be 39 nm with concentration of 3.79 × 10<sup>8</sup> particles/ml; whereas for ZnNP, the particle size distribution was in the range of 5–90 nm, and mean size was 62 nm with concentration of 4.72 × 10<sup>8</sup> particles/ml.

The size and morphology of silver nanoparticles synthesized were observed in TEM with average size of 40 nm and most of the population consisting of single crystalline nanoparticles. Spherical zinc nanoparticles of varying size were observed with average size of 60 nm (**Figure 6**).

## Toxicity of Nanoparticles

Cytotoxicity was evaluated in the range of 0.3 to 10 mg/ml concentration of nanoparticles. Toxicity of a compound is calculated by determining the IC<sup>50</sup> value, which is that concentration at which growth of cells is inhibited by 50%. The percentage cytotoxicity is plotted against the concentration of nanoparticle as seen in **Figure 7**.

Based on toxicity of nanoparticles to Vero cell line, the IC<sup>50</sup> values for silver and zinc nanoparticles were found to be 5.54 and 6.24 mg/ml, respectively.

## Antimicrobial Activity of Nanoparticles Minimum Inhibitory Concentration

Antimicrobial activity was assessed up to 10 mg/ml concentration of nanoparticle, which is almost double the concentration of cell toxicity to normal cells. For therapeutic consideration as an antimicrobial for synthesized nanoparticles only activity obtained at 5 mg/ml or below was considered. The MIC of both nanoparticles against range pathogenic bacterial strains is as mentioned in **Table 1**.

### Anti-mycobacterial Activity – Broth Dilution Method **Table 2** shows the MICs obtained for standard strain and MDR strain of M. tuberculosis against both nanoparticles.

## Synergistic Activity of Gentamicin and Nanoparticles

Nanoparticles were estimated for its synergistic action with gentamicin against MDR strains. For which susceptibility and MIC of gentamicin were checked, and results of that were

FIGURE 7 | Plot of cytotoxicity vs concentration of nanoparticles.

TABLE 1 | MIC of nanoparticles against range of pathogenic bacteria.



R, resistant; S, susceptible (Clinical and Laboratory Standards Institute [CLSI], 2014).

### TABLE 4 | E-test MIC for gentamicin.

TABLE 3 | Susceptibility for gentamicin.

TABLE 2 | MIC of nanoparticles against standard and MDR strain of Mycobacterium tuberculosis.


as summarized in **Tables 3**, **4**. Representative images in the **Supplementary Material**.

### MIC of Gentamicin

The MIC of gentamicin by microbroth dilution method was evaluated up to 320 µg/ml, considering the limit and toxicity of gentamicin. **Table 5** shows the MICs obtained for each test strain.


R, resistant; S, susceptible (Clinical and Laboratory Standards Institute [CLSI], 2014).

### Checkerboard Assay

For determination of synergistic activity of synthesized nanoparticles with gentamicin against test strains checkerboard

### TABLE 5 | MIC for gentamicin – microbroth dilution method.

fmicb-09-02207 September 20, 2018 Time: 16:17 # 8


TABLE 6 | Synergistic activity of nanoparticles and gentamicin.


AgNPS, silver nanoparticles; ZnNPs, zinc nanoparticles.

assay was performed. The FIC values and interpretation of synergism are as shown in **Table 6**.

## DISCUSSION

Bacterial synthesis of silver nanoparticles by P. hibiscicola was first reported in our previous work (Punjabi et al., 2017). Silver nanoparticles synthesized were characterized by various spectral and microscopic techniques to specifically confirm the synthesis and determine the details of synthesized nanoparticles.

In this study, the NTA revealed the average size of AgNP ranged 10 to 70 nm, but the mean size of synthesized silver nanoparticles was found to be 39 nm. This is in accordance to similar studies with biosynthesis of nanoparticles and its measurement by NTA. Birla et al. (2013) report the size distribution of nanoparticles obtained on a particleby-particle basis in case of NTA which enables particle population, size, intensity of nanoparticle in suspension, and simultaneously determining their Brownian motion. The NTA calculates the particles size by distance traveled by them, and calculation is based on Stokes–Einstein equation. The average size distribution of silver nanoparticles synthesized by Fusarium oxysporum in their study was 41 nm (Birla et al., 2013). TEM revealed an average size of 40.0 nm, which is in accordance with results obtained by NTA. The TEM analysis indicated that the particles were crystalline in nature.

The extracellular biosynthesis of silver and zinc oxide nanoparticles was investigated using Pichia fermentans JA2 by Chauhan et al. (2015). The biosynthesized silver nanoparticles were found to inhibit most Gram-negative clinical pathogens used in the study, while zinc oxide nanoparticles only inhibited P. aeruginosa. The silver and zinc oxide nanoparticles synthesized using Pichia fermentans JA2 had potent antibacterial effect that leads to notable contribution to pharmaceutical associations.

The above report boosts the idea of synthesizing zinc nanoparticles by P. hibiscicola. Thus, studies were undertaken to synthesize, optimize, and characterize zinc nanoparticles, which were found to be of variable shapes and the average size of 60 nm. Although DLS reports 110 nm, NTA shows 62 nm final size of nanoparticles was only considered as obtained in TEM micrographs, which were 60 nm. In most case, synthesis of zinc oxide nanoparticles often takes place which could highly be possible in this case as well. This could be alleged on basis of the spectra in UV–Vis spectrophotometer and XRD pattern, which does indicate the presence of oxide specific peaks. Different functional groups identified as secondary alcohols, amines, and alkanes by FTIR analysis, which could mainly be due to biological synthesis.

Jayaseelan et al. (2012) present a work for biosynthesis of zinc oxide nanoparticles (ZnONPs) using bacteria, Aeromonas hydrophila. The ZnO NPs were characterized, and a peak at 374 nm in the UV–vis spectrum was observed. Further, XRD confirmed the crystalline nature of the nanoparticles and AFM exhibited the morphology of the nanoparticle to be spherical, oval with an average size of 57.72 nm. The FTIR peaks summarized as alkynes, alkanes, alkenes, primary amines, alcohols, acetate, ethers, and carboxyl compounds. The results of zinc nanoparticles synthesized by P. hibiscicola are highly compatible with this study. It could well be said so that synthesis of zinc oxide nanoparticles formation takes place by P. hibiscicola also.

The most vital parameter to determine the therapeutic feasibility is toxicity of the material. Therefore, in vitro cytotoxicity of both nanoparticles synthesized were assessed on animal tissue. The technique employed to do so was MTT assay, which is a colorimetric estimation of living cells determined by absorbance values upon treatment with the drug for a stipulated amount of time. The cell line used to evaluate the toxicity of nanoparticles was Vero a non-cancerous epithelial cell line commonly used for the said purpose.

In one of the reports, use of AgNP as a therapeutic agent is limited due to its cytotoxicity against mammalian cells. Wherein cytotoxicity of P. putida synthesized silver nanoparticles against HEp-2 cells was determined by MTT assay. No significant cytotoxic effect at 25 µg/ml concentration was found, which was lethal for the bacteria tested for antibacterial efficacy (Gopinath et al., 2017). This study highlights and supports the results obtained for nanoparticles synthesized by P. hibiscicola although the range of concentration varies mainly due to the presence of protein in the nanomaterial. The purification of nanoparticles was not carried out and freezedried powders of as-synthesized nanoparticle were used for cytotoxicity and other activities, leading to the larger amount of material.

One of the most important ailments affecting thousands around the world is microbial infections and more to concern is its drug resistance happening at a rapid pace. Special emphasis needs to be given to drug-resistant infections of bacteria in the community and hospital-acquired infections like ESBL, MRSA, VRE, MDR, P. aeruginosa infections, and MDR M. tuberculosis infections. The resistance development

has been concerning, and world organizations have been raising alarms at regular intervals. This has also lead to screening a vast number of agents from both biological as well as the chemical cluster for antimicrobial potential. Nanoparticles have potentially all the requisites for qualification as an antimicrobial agent. Not only can nanoparticles have efficiency against this infection causing etiological agents but can also prevent the current therapeutic agents from being rendered useless due to resistance, by alternative or supplementary therapies with nanoparticles along with regular antibiotics. Thus, an interesting aspect was to evaluate nanoparticles synthesized by P. hibiscicola for its antimicrobial activity.

Only activity obtained at or below 5 mg/ml was considered suitable for antimicrobial potential against the test strain considering the toxicity of nanoparticles at higher concentrations. Zinc nanoparticle gave activity only against two Gram-negative and four Gram-positive strains, importantly against MRSA at 2.5 mg/ml. Silver nanoparticles were the most potent antimicrobial with activity against most bacteria. In case of the drug-resistant variants for both ESBL and MRSA, activity was observed at 2.5 mg/ml. Both AgNP and ZnNP had anti-TB activity at 1.25 mg/ml for both strains. These results definitely imply that nanoparticles synthesized by P. hibiscicola have great antimicrobial potential. This is encouraging to do further evaluations and assess its practical application with more specific in vivo studies.

Abd-Elnaby et al. (2016) reported an excellent antibacterial activity against Gram-negative and Gram-positive bacterial pathogens by microbiologically synthesized AgNPs. Mostafa et al. (2015) report activity of chemically synthesized silver nanoparticles against P. aeruginosa, S. aureus, Streptococcus pyogenes, and S. typhi. A report stating anti-tuberculosis activity of biosynthesized silver nanoparticles-mentioned spherical-shaped silver nanoparticles with the average particle size of 5 nm, had anti-mycobacterial activity against M. tuberculosis and clinical isolates of multi-drug resistant M. tuberculosis (MDR-TB) (Banu and Rathod, 2013).

Arrays of pathogens causing different infections were inhibited with the synthesized nanoparticles suggesting its efficiency in controlling infections. This also recommends for assessment of these nanoparticles for synergistic activity with common antibiotics. Thus, synergistic activity against drugresistant strains was evaluated with antibiotic gentamicin. Gentamicin is an aminoglycoside and commonly used antibiotic against both Gram-positive and Gram-negative bacteria. Currently, gentamicin is susceptible to some and resistant to some bacterial isolates, but it could fast be completely resistant if measures to control resistance are not in place soon. Thus, hypothetically nanoparticles could rescue this situation with its potential synergistic efficiency if found.

Antibacterial susceptibility against gentamicin was assessed by disk diffusion method followed by MIC of gentamicin by E – test and micro broth dilution method. The results of disk diffusion and E-test identify all the isolates as resistant to gentamicin as per CLSI guidelines of 2014. The MIC of gentamicin by micro broth dilution method was evaluated with a maximum concentration of 320 µg/ml, which is reportedly a toxic concentration. Synergistic activity of nanoparticles and gentamicin was evaluated by Checkerboard method, considering the MICs and toxicity of both gentamicin and nanoparticles.

Promising results were obtained for synergistic activity, zinc nanoparticles exhibited synergism with gentamicin in case of MRSA, so did silver nanoparticles for MRSA, ESBL, and MDR P. aeruginosa. A study on the fugal-mediated synthesis of silver nanoparticles reports synergistic activity of silver nanoparticles with amikacin, kanamycin, and streptomycin against E. coli, S. aureus, and P. aeruginosa. The report concluded synergistic action of antimicrobial agent can significantly curtail side effects by reducing dosages. Thus, use of nanoparticles combined with antibiotics can improve their efficacy against various pathogenic microbes (Barapatre et al., 2016). The results of synergistic activity of nanoparticles synthesized by P. hibiscicola are in accordance with this report and rightly suggested nanoparticles can indeed be one of the most effective agents in combating serious infections that are rendered untreatable with available antibiotics. There are limited studies with this extent of antimicrobial activity, with qualitative as well as quantitative data against wide range of ATCC and clinical strains. Further to this, determining the mode of action will be an important aspect. But it is well known that most metallic nanoparticles cause oxidative damage to microbes due to excessive generation of reactive oxygen species (ROS), this can be extended to the nanoparticles synthesized.

The extensive potential of silver and zinc nanoparticles against drug-resistant strains of community and hospitalacquired infections opens a new arena of antimicrobials for treatment, supplementary prophylaxis, and prevention therapy. Biologically synthesized nanoparticles can become future antiinfectives and prove to be one of the most potential therapeutic agents.

## CONCLUSION

This study covers certain major areas of nanoscience like biological synthesis, characterization, and biomedical application. It provides an extremely potential isolate capable of synthesizing silver and zinc nanoparticles. Characterization studies revealed synthesis of stable nanoparticles of average size 50 nm for silver and 60 nm for zinc nanoparticles. The cytotoxicity studies of nanoparticles against normal cells (non – cancerous) revealed the toxic concentrations for both. Toxicity is the most important factor for therapeutic application. Efficiency of these nanoparticles against drug-resistant pathogen makes it a viable candidate for therapeutics. Thus, biologically synthesized nanoparticles can become future anti-infective agents and can be formulated for drugs or topical agents.

## AUTHOR CONTRIBUTIONS

fmicb-09-02207 September 20, 2018 Time: 16:17 # 10

KP conceptualized and executed the entire experiment as well as the writing of this work. SM helped with the characterization studies, mainly the XRD analysis. RC, VC, and DD helped with the experimental work, including microbial assays and toxicity analysis. VC also helped write and format

### REFERENCES


the manuscript. SD managed the successful completion of the entire project and played an instrumental role in the manuscript.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.02207/full#supplementary-material



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Punjabi, Mehta, Chavan, Chitalia, Deogharkar and Deshpande. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Efficient Production of Recombinant Protegrin-1 From Pichia pastoris, and Its Antimicrobial and in vitro Cell Migration Activity

Evanna Huynh<sup>1</sup> , Nadeem Akhtar<sup>1</sup> and Julang Li1,2 \*

<sup>1</sup> Department of Animal Biosciences, University of Guelph, Guelph, ON, Canada, <sup>2</sup> College of Life Science and Engineering, Foshan University, Foshan, China

Protegrin (PG) belongs to the antimicrobial peptide cathelicidin family. To date, five protegrin sequences have been identified in pigs, PG-1 to PG-5. Of these, PG-1 exhibits potent antimicrobial activity against a broad range of antibiotic-resistant microorganisms as well as viruses. However, the other potential role(s) of PG beyond antimicrobial has largely been unexplored. The aim of this study was to use nonpathogenic yeast Pichia pastoris to express antimicrobially active recombinant protegrin (rPG-1). Additionally, the effect of PG-1 on cell migration and proliferation was also examined in vitro using pig intestinal epithelial cells as a model. Highest level of rPG-1 (104 ± 11 µg/mL) was detected at 24 h in fermentation culture medium. Similar to rPG-1, 0.8 ± 0.10 g/L of proform PG-1 (rProPG-1) and 0.2 ± 0.02 g/L of the PG-1 cathelin domain (rCath) was detected in fermentation culture medium. Resulting recombinant PG-1 and cleaved rProPG-1 exerted antimicrobial activity against Escherichia coli DH5α at the same level as chemically synthesized PG-1. Enhanced cell migration was observed (p < 0.05) in groups treated with rProPG-1, rCath, and rPG-1 compared to the control. Furthermore, rPG-1 was stable at temperatures ranging from 25◦C to 80◦C. In summary, biologically active recombinant protegrin in its pro-, cathelin-, and mature- forms were successfully expressed in P. pastoris suggesting potential feasibility for future therapeutic applications.

### Keywords: antimicrobial peptide, cell migration, fermentation, Pichia pastoris, protegrin

### INTRODUCTION

Tissue repair, including those happened after gastrointestinal infection, is an orchestrated process with inflammation and tissue re-epithelialization as critical approaches for successful healing. In addition to growth and repair factors (e.g., epidermal growth factor, trefoil factors), antimicrobial peptides (AMPs) are also expressed and secreted upon mucosal injury (Frohm et al., 1997; Hase et al., 2003; Kovach et al., 2012). AMPs, as well as the recently identified unusual amino acid L-cyclopropylalanine with anti-fungal and -bacterial activity (Ma et al., 2017) represent the first line of defense, or innate immunity, against foreign pathogens. These low-molecular weight molecules can exert antimicrobial activity at a nanomolar to micromolar range against a broad spectrum of microorganisms including bacteria, fungi, enveloped viruses, and parasites

### Edited by:

Kamlesh Jangid, National Centre for Cell Science (NCCS), India

### Reviewed by:

César de la Fuente, Massachusetts Institute of Technology, United States Pedro Ismael Da Silva Junior, Instituto Butantan, Brazil

> \*Correspondence: Julang Li jli@uoguelph.ca

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 12 November 2017 Accepted: 10 September 2018 Published: 27 September 2018

### Citation:

Huynh E, Akhtar N and Li J (2018) Efficient Production of Recombinant Protegrin-1 From Pichia pastoris, and Its Antimicrobial and in vitro Cell Migration Activity. Front. Microbiol. 9:2300. doi: 10.3389/fmicb.2018.02300

(Hancock and Sahl, 2006; Sang and Blecha, 2008; Dias et al., 2017). In recent years, AMPs appear to be functionally multifaceted, fulfilling a variety of functions beyond microbial killing, including immunomodulation, tissue regeneration and vascularization in inflammation, tissue repair and even tumor surveillance (Hirano et al., 2000; Ishihara and Hirano, 2002; Proudfoot, 2002; Choi et al., 2012; de la Fuente-Núñez et al., 2014). A cationic AMP P17 from ant venom induces an alternative phenotype of human monocyte-derived macrophages (h-MDMs) promoting its anti-fungal activities (Benmoussa et al., 2017).

Cathelicidin is an evolutionary distinct group of AMPs in mammals. This peptide group has a highly conserved N-terminal (cathelin) domain in the pro-peptide followed by a variable AMP domain on the C-terminus (Kovach et al., 2012). Protegrin (PG) belongs to the cathelicidin family, and to date, five protegrin sequences have been identified in pigs, PG-1, PG-2, PG-3, PG-4, and PG-5. Of these, PG-1 exhibits potent antimicrobial activity against a broad range of antibiotic-resistant microorganisms as well as viruses, including HIV (Hancock and Sahl, 2006; Liu and Stappenbeck, 2016). Increasing evidence of alterations of mucosal AMP levels has leaded to the suggestion on their influential role on the gut immune protection (Schauber et al., 2006; Termén et al., 2008; Inaba et al., 2010), although the immuno- and tissue repair-modulating roles of PGs have not been investigated.

PG-1 possesses potential therapeutic utility. However, chemical synthesis of peptides is associated with high production costs and increased peptide length often limits success of the synthesis, and thusly, potentially limiting therapeutic application. Pichia pastoris is a highly appealing expression system which has advantages including rapid growth rate, high levels of protein production and eukaryotic post-translational modifications. The commercial applications of recombinant protein of phospholipase C (Ciofalo et al., 2006) and biopharmaceutical product (Thompson, 2010) produced by P. pastoris have the FDA generally recognized as safe (GRAS) status in the United States. Furthermore, administration of P. pastoris cells as a vehicle for recombinant vaccine delivery via intramuscular injection and oral delivery have been shown to be safe and effective in chickens (Taghavian et al., 2013). Previous work on the PG-3 cathelin domain revealed that the pro-piece cathelin domain has a role in activating cathepsin L (Zhu et al., 2008), and the pro-piece of human hCAP-18 has been suggested to have a role in preventing cysteine-proteinase-mediated tissue damage during inflammation (Zaiou et al., 2003). However, the role of PG-1 and its cathelin domain in regulating other cellular activity is unclear. We hypothesized that the proform, pro-piece and mature form of PG-1 can be efficiently produced in yeast, and that they may process tissue repair function, in addition to the well known antimicrobial role of PG-1.

The first objectives of this investigation was to examine the feasibility of expressing recombinant PG-1 in its pro-, cathelinand mature- form in P. pastoris and characterize its antimicrobial activity in vitro. Considering the wound healing role, besides antimicrobial activity, of PG-1 has not been investigated to date, we will also examine the effect of PG on cell migration and proliferation in vitro using pig intestinal epithelial cells as a model.

## MATERIALS AND METHODS

## Construction of P. pastoris DNA Expression Vector

Expression vector designed to express pro-form PG-1 (pJ912 proPG-1) was codon-optimized (**Supplementary Material 1**) for optimal expression in P. pastoris and synthesized by DNA2.0 (Menlo Park, CA, United States). Native PG-1 sequence was used as a starting sequence (NCBI accession number: CAA56251). Enterokinase (EK) cleavage site (DDDDK) replaced the native neutrophil elastase site between the cathelin domain and the mature PG-1 domain (**Figure 1A**). Resultant plasmids were transformed into Escherichia coli DH5α (E. coli) for plasmid propagation and positive recombinant colonies were selected using 25 µg/mL zeocin (Invitrogen, CA, United States) in Luria-Bertani (LB) medium (Difco, MI, United States).

## Electroporation of P. pastoris X-33 and Screening for Recombinant Strains

Plasmid DNA was linearized with SwaI (New England Biolabs, Ipswich, MA, United States) and purified by phenol-chloroform extraction and ethanol precipitation. Approximately 3–5 µg of purified linearized plasmid DNA was dissolved in 10 µL of nuclease-free H2O. Electrocompetent P. pastoris X33 (Invitrogen, Carlsbad, CA, United States) was prepared according to the manufacturer's instruction. Competent cells (100 µL) were mixed with the dissolved purified DNA in a 0.4 cm electroporation cuvette (BTX, CA, United States). Electroporation was carried out in Electro Cell Manipulator 600 (BTX, CA, United States) at 1.5 kV. Immediately after electroporation, 1.0 mL of 1 M ice-cold sorbitol (Fisher Scientific, Pittsburgh, PA, United States) was added to the cuvette. Cells were transferred to a sterile 1.5 mL tube and incubated at 30◦C without shaking for 2 h. Transformed cells were plated on YPD agar [1% yeast extract (Difco, Detroit, MI, United States), 2% peptone (Difco, Detroit, MI, United States), 2% glucose (Fisher Scientific, Pittsburgh, PA, United States), 2% agar (Difco, Detroit, MI, United States)] supplemented with 1 M sorbitol and zeocin (100 µg/mL) and incubated for 3–4 days at 30◦C until colonies appeared. Colonies were purified by streaking onto YPD agar with zeocin (100 µg/mL).

## Genomic DNA Analysis to Determine P. pastoris Mut-Genotype

To confirm recombinant gene integration and P. pastoris transformant Mut phenotype, genomic DNA was isolated from P. pastoris transformants using the freeze-and-thaw method previously described in (Jafari et al., 2011). The following universal AOX1 primers were used for PCR: Sense: 5<sup>0</sup> -GACTGGTTCCAATTGACAAGC-3<sup>0</sup> , anti-sense: 5 <sup>0</sup>GCAAATGGCATTCTGACATCC-3<sup>0</sup> . Amplification was

carried out at 94◦C for 30 s, 55◦C for 30 s, and 72◦C for 1 min for 35 cycles. PCR was performed using Taq DNA polymerase (Invitrogen, Carlsbad, CA, United States).

## Expression Cassette Copy Number Determination in P. pastoris

The number of copies of pJ912-ProPG expression cassettes integrated into the P. pastoris genome was determined by quantitative real-time PCR as described by (Abad et al., 2010) and copy number of each clone was determined using the 2−11Ct method (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008). Blank P. pastoris X33, AOX1 promoter, and endogenous ARG4 gene were used as calibrator, target gene, and reference gene, respectively.

## Inducing Expression of Pro-form PG-1 in P. pastoris

Positive transformant colonies were inoculated into a 500 mL flask containing 50 mL BMGY medium [1% yeast extract, 2% peptone, 100 mM potassium phosphate (pH 6.0), 1.4% yeast nitrogen base (Difco, Detroit, MI, United States), 0.00004% biotin (Difco, Detroit, MI, United States), 1% glycerol (Fisher Scientific, Pittsburgh, PA, United States)] at 30◦C with shaking at 250 rpm. After 16–24 h, cultures reached an optical density at 600 nm (OD600) = 2.0–6.0 and were harvested by centrifugation and resuspended in BMMY medium (BMGY medium containing 0.5% v/v methanol instead of glycerol) to induce expression. Cells were grown for 72 h at 30◦C with shaking (250 rpm) and 100% methanol was added every 24 h to a final concentration

of 0.5% (v/v). At various time points, 100 µL aliquots were taken from the culture and replaced with an equal amount of fresh BMMY medium. Supernatant and cell pellet samples were centrifuged and stored separately at −80◦C for later analyses.

## Recombinant Protegrin Production in a Bioreactor

Pichia pastoris expressing recombinant ProPG-1, Cath, and PG-1were grown independently in a 3-L BioFlo 115 Benchtop Bioreactor (New Brunswick Scientific, CA, United States). Fermentation was carried out in a total volume of 1 L basal salt medium (BSM) (Stratton et al., 1998) with modifications. Pichia trace minerals 1 (PTM1) solution was prepared with (g/L) CuSO4·5H2O (6.0), NaI (0.08), MnSO4·H2O (3.0), Na2MoO4·2H2O (0.2), H3BO<sup>3</sup> (0.02), CoCl<sup>2</sup> (0.5), ZnCl<sup>2</sup> (7.0), FeSO4·7H2O (65.0), biotin (0.2), and 5 mL of concentrated H2SO4. A 100 mL recombinant P. pastoris inoculum prepared in YPD (OD at 600 = 15 ± 1.3) was transferred to the sterilized bioreactor containing 1 L of BSM. Antifoam A (Sigma, St. Louis, MO, United States) was added manually to control foaming in the bioreactor, if needed. Temperature was maintained at 30◦C and pH at 5.6 using 12.5% (v/v) NH4OH. Aeration rate of 5 L min−<sup>1</sup> was constant throughout the fermentation. The stirrer speed was controlled between 250 and 700 rpm aiming at dissolved oxygen (DO) concentration of 20% air saturation. After ∼24 h of fermentation, consumption of glycerol was indicated by an increase in DO concentration at an OD of 30 ± 2.1, a glycerol feed [50% (v/v) glycerol containing PTM1 (12 mL/L glycerol)] was then maintained for another12 h at 15 mL/L/h. After completion of the glycerol feeding OD reaches to 170 ± 14.6. This was followed by a methanol feed [100% methanol containing PTM1 (12 mL/L methanol)] and maintained at a constant rate of 3 mL/L/h. The OD of the culture medium after 24 of methanol induction was 288 ± 19.4, the fermentation was stopped and culture medium was centrifuged (10,000 rpm, 10 min, 4◦C) to obtain cell free supernatant and stored at −80◦C for further analyses.

## Protein Sample Preparation

Standard lyophilized PG-1 was synthesized at approximately 95% purity (EZbiolabs, Carmel, IN, United States) and was dissolved in and diluted to 200 µg/mL with filter-sterilized acidified water (0.01% acetic acid) supplemented with 0.1% bovine serum albumin (BSA) (Sigma, St. Louis, MO, United States) to reduce non-specific adsorption of PG-1 to plastic ware. Standard PG-1 was stored at −80◦C. Pro-form PG-1 expressed by P. pastoris in the supernatant was dialyzed against 25 mMTris-Cl (pH 7.0) and concentrated 60-foldby ultracentrifugation with a 10 kDa cut-off (Millipore Inc., Burlington, MA, United States). Ultracentrifugation retentate and supernatant was filter sterilized and stored at −80◦C prior to antimicrobial assay. EK at 0.8 U and 2.5 U (GenScript, Piscataway, NJ, United States) was used to cleave 5 µL of pro-form PG-1 overnight (final 10 µL reaction volume) to release the mature PG-1 from the cathelin domain according to manufacturer's instruction for subsequent antimicrobial activity assay.

## Two-Stage Radial Diffusion Assay

The antimicrobial activity of recombinant pro-form PG-1 and cleaved pro-form PG-1 were tested by a two-stage radial diffusion assay as previously described by (Steinberg and Lehrer, 1997) against E. coli and methicillin-resistant Staphylococcus aureus (MRSA). Positive control sample wells received 10 µL of serially diluted chemically synthesized standard PG-1 or its vehicle as a control. Sample wells received 10 µL of pro-form PG-1 containing 0.01% acetic acid and 0.1% BSA after EK digestion. Minimum inhibition concentration (MIC) was defined at a concentration of PG-1 that resulted in a clear inhibition zone around the well. To test whether BMMY can inhibit PG-1 antimicrobial activity, 5 µL pro-form PG-1 after the 2.5 U EK digestion reaction was resuspended with 5 µL of spent BMMY medium and the mixture was used to inoculate sample well.

## Antimicrobial Activity Assay Using Colony-Count Method

The MIC of rPG-1 against E. coli was tested according to the antimicrobial assay previously described (Pizzo et al., 2015) with slight modification. The stability of rPG-1 was monitored at different pH (2–11) and temperature (25–95◦C) conditions after a preincubation time of 30 min before being assayed. Briefly, overnight grown E. coli in LB medium was diluted to an OD 600 of 1 and further diluted 1:100. E. coli- peptide mixtures were prepared by adding 40 µL of bacterial cells to different concentration of rPG-1 (1–10 µg/mL) in separate eppendorf tubes and adjusted to a final volume of 1 mL using 20 mM acetate buffer (pH 5). Cells without rPG-1 and with 0.05 mg/mL ampicillin were taken as negative and positive controls, respectively. Experimental and control tubes were incubated for 4 h at 37◦C, and 50 µL of the diluted (1:100) mixtures were plated on LB agar media and incubated at 37◦C overnight for colony counts. The survived cells in each plate were compared with controls. All concentrations of rPG-1 at different pH and temperature were tested in triplicate experiments and <5% standard deviation were observed for each experiment.

## Western Blot

Culture supernatants were subjected to 12% SDS-PAGE and electrophorectically transferred to 0.45 µm pore size PVDF membrane (Millipore Inc., Burlington, CA, United States). The membrane was blocked in 5% (w/v) skim milk in TBST buffer (0.3% Tris, 0.8% NaCl, 0.02% KCl, 0.1% Tween 20) for 1 h and then incubated with affinity purified polyclonal rabbit anti-PG-1 antibody at 1:500 dilution, synthesized from Biomatik (Cambridge, ON, Canada) overnight at 4◦C. After washing the membrane three times at 5 min each in TBST, the membrane was incubated with anti-rabbit IgG secondary antibody (1:1000 dilution) conjugated to horseradish peroxidase (Cell Signaling, Danvers, MA, United States) for 1 h at room temperature. Bands were visualized using an ECL Plus Western blotting system according to manufacturer's instruction (Amersham Biosciences, Piscataway, NJ, United States).

### In vitro Cell Proliferation Assay

fmicb-09-02300 September 25, 2018 Time: 19:18 # 5

IPEC-J2 cells were grown in six-well plates (Corning, Corning, NY, United States) until 50% confluent in medium consisting of DMEM F-12 containing 5% fetal bovine serum at 37◦C in 5% CO<sup>2</sup> and were then serum starved for 24 h. The cells were then incubated with serum-free DMEM F-12 in the presence of 10 µL filter-sterilized fermentation supernatant from 24 h MeOH induced cultures from PP-ProPG-1, PP-Cath, PP-PG-1 and PP-wild type (PP-WT), and cultured for 24 h. Cells were then trypsinized and enumerated using a hemocytometer.

### Transwell Assay

To analyze migration of IPEC-J2 cells, 8-micron pore sized Millicell Cell culture transwell inserts (Millipore Inc., Burlington, CA, United States) were used. A total of 1 × 10<sup>5</sup> cells were plated in the upper inserts and the lower chamber contained serum-free DMEM F-12 in the presence of 10 µL filtersterilized fermentation supernatant from 24 h MeOH induced cultures from PP-ProPG-1, PP-Cath, PP-PG-1, and PP-WT. After incubation for 16 h, the cells were fixed with 4% (w/v) paraformaldehyde. Cells that did not migrate into the membrane were gently scraped off the upper surface of the transwell with a cotton swab. Migration was quantified by cell enumeration through Hoechst 33342 staining of cell nuclei (Life Technologies, Grand Island, NY, United States).

### Statistical Analysis

Results are expressed as mean ± SEM (standard error mean) from at least three independent experiments. The data were analyzed by two-factor analysis of variance (ANOVA) using a GraphPad Prism software version 5.0. Data sets were analyzed by Tukey's test for multiple comparisons to determine statistical differences between groups. The results were considered significant at a P value of <0.05.

### RESULTS

### Generation of P. pastoris Expressing Recombinant ProPG-1, Cath and PG-1

To generate P. pastoris that could express recombinant ProPG-1 (rProPG-1), Cath (rCath), and PG-1 (rPG-1), the cDNA sequence of codon-optimized porcine protegrin ProPG-1, Cath, and PG-1 were ligated into the respective pJexpress P. pastoris expression vector, respectively. The α-mating factor secretion signal sequence was placed at the N-terminal (**Figure 1A**) to facilitate secretion of the expressed recombinant proteins into the culture fermentation medium, also referred to as supernatant. In our study, we replaced the native elastase cleavage site of preformprotegrin with an EK site. The addition of the EK cleavage site can permit the potential application of recombinant PG-1 in animal feed, as EK is a serine protease and is naturally produced and secreted in the intestinal duodenum (Baratti et al., 1973; Fonseca and Light, 1983).

The DNA sequence-verified recombinant plasmids were linearized with SwaI and transformed into electrocompetent P. pastoris X-33. Transformants harboring the pJexpress-ProPG-1, pJexpress-Cath and pJexpress-PG-1 constructs are here on referred to as PP-ProPG-1, PP-Cath, and PP-PG-1, respectively. To confirm integration of the construct into the Pichia genome, isolated genomic DNA from clones were analyzed by PCR using the 50AOX1 primer paired with 30AOX1-TT (**Figure 1B**). PCR results showed chromosomal integration of the expression cassette resulting in a 760 bp, 458 bp, and 700 bp PCR product for positive PP-ProPG-1, PP-PG-1, and PP-Cath transformants, respectively. Mut<sup>+</sup> strains had a 2.2 kb PCR product corresponding to the intact AOX1 gene. Mut<sup>s</sup> strains do not have a 2.2 kb PCR product and only a PCR product corresponding to the AOX1 and insert part of the expression construct (**Figures 1C,D**).

## Effect of Copy Number on Recombinant Protegrin Expression

To employ the expression potential of the P. pastoris system, we examined the influence of gene copy number on the expression of rProPG-1, rCath, and rPG-1 from PP-ProPG-1, PP-Cath, and PP-PG-1, respectively. Clones with putative multiple copies of the rProPG, rCath, and rPG1 expression cassette from varying concentrations of G418 antibiotic selection were selected for copy number determination by qPCR. The purpose was to examine the effects of copy number on rProPG-1, rCath, and rPG-1 expression. Clones containing 1–14 copies of the rProPG-1 expression cassette were analyzed for the expression rProPG-1 under shake-flask conditions. As shown in **Table 1**, rProPG-1 expression level increased progressively as the copy number increased from 1 to 7, with seven copies resulting in earlier and highest expression of rProPG-1 at 24 h post-induction. As the number of copies increased from 7 to 11, the expression level did not further increase. However, 14 copies of the cassette yielded higher expression level than 11 copies, but 14 copies of the cassette yielded lower rProPG-1 than in the clone with seven copies at 24 h methanol induction. This indicates a potential maximal limit number of copies that can have a direct influence on rProPG-1 expression level. Similarly, rCath expressed by PP-Cath increased progressively from one copy to six, and decreased at eight copies (**Table 2**). PP-Cath transformant with six copies yielded the highest rCath (**Figure 2A**). Furthermore, gene copy number from 2 to 11 were evaluated in PP-PG-1 clones and clones with six and eight copies yielded highest


Supernatant samples at 24 and 72 h methanol induction from small-scale fermentations of different PP-ProPG-1 clones were assessed for rProPG-1 expression. (−) indicates no rProPG-1 was detected and (+), (++), (+++) indicate low, medium and high levels of rProPG-1 detected using dot blot analysis, respectively.


Supernatant samples at 24 and 72 h methanol induction from small-scale fermentations of different PP-Cath clones were assessed for rCath expression. (+), (++), (+++) indicate low, medium and high levels of rCath detected using dot blot analysis, respectively.

rPG-1 activity (**Figure 2B**, **Table 3**). Progressive increase in rPG-1 was observed as copy number increased up to eight and appeared to decrease at eleven copies. Interestingly, two clones both with six copies of rPG-1 construct expressed at different levels (**Table 3**), suggesting other factors, in addition to gene copy number, can influence recombinant protein expression level in P. pastoris.

## Pichia pastoris Recombinant Protegrin Expression in a Bioreactor

To potentially obtain increased yield of recombinant protegrin, high-density fermentation was performed in a 3-L bioreactor. Initially, the transition from shake-flask culturing to a bioreactor unexpectedly yielded much lower recombinant protein concentrations using the standard media composition (Invitrogen manual, Carlsbad, CA, United States). By reducing BSM concentration by 50% during culture media optimization, expression level increased. For rPG-1 expression, rPG-1 was not detected until yeast extract and peptone were added to the BSM. These improvements suggest that the media formulation could be optimized to improve the yield.

Expression with all the three recombinant P. pastoris strains was performed using three-step procedures, including glycerolbatch phase, glycerol fed-batch and methanol induction phase. The DO spike was found to be an effective indicator to initiate a new feeding profile and to monitor the overall health of the culture for successful recombinant protegrin expression. Aliquots of the fermentation supernatant were examined by Western blot and ELISA. Highest level of rPG-1 (104 ± 11 µg/mL) was



rPG-1 antimicrobial activity level correlates with amount of rPG1 produced. PP-PG-1 clones were assessed for antimicrobial activity against MRSA. (−) indicates no rPG-1 activity was detected and in the others, the activities of rPG-1 are indicated as + (clear zone 0.5–1.5 mm), ++ (clear zone 1.5–2.5 mm), and +++ (clear zone > 2.5 mm).

detected at 24 h of methanol induction in the fermentation culture medium. Post 24 h, rPG-1concentration decreased (**Figure 3A**). Similar to rPG-1, 0.8 ± 0.10 g/L rProPG-1, and 0.2 ± 0.02 rCath g/L was detected in fermentation culture medium at 24 h of methanol induction (**Figures 3B–D**).

### Enterokinase Cleavage of rProPG-1 and Antimicrobial Activity

To determine whether the rProPG-1 could yield antimicrobial activity upon removal of the N-terminus pro- domain, EK was used to cleave the introduced amino acid DDDDK site on rProPG-1 (**Table 4**). Cleavage of rProPG-1 was detected by Western blot using antibodies specific to mature PG-1 in the C-terminal domain. EK successfully cleaved rProPG-1 as the detection of mature PG-1 decreased from 13.4 kDa to 2.1 kDa, indicating the removal of the 11 kDa pro- domain (**Figure 4A**). The resulting cleaved rProPG-1 also conferred antimicrobial activity against E. coli DH5αat the same level as chemically synthesized PG-1 (**Figure 4B** and **Table 5**). The uncleaved rProPG-1 and rCath did not exhibit antimicrobial activity as expected.

### Protegrin Peptide Properties and Boman Index

Peptide properties such as molecular weight, pI value, net charge and Boman Index can be used to predict a peptide's antimicrobial potential under various conditions (Boman, 2003). rProPG, rCath, and rPG-1 were calculated using both Innovagen's Peptide Property Calculator and a predictive tool available at AMP Database v2.34, respectively. The rPG-1 showed highly basic pI value (>10), positive net charge (+6), and high Boman index (>2.5; **Table 4**). The peptides containing the cathelin domain, rProPG-1, and rCath, have pI values of 6.49 and 4.6, respectively. ProPG is neutral at pH 7.0 (net charge −0.3) and Cath has a negative net charge (−6.2). Both of these peptides share similar

FIGURE 3 | Quantification of rPro-PG-1, rCath, and rPG-1 produced by P. pastoris in separate bioreactor. (A) rPG-1 (from PP-PG-1) was detected with antibody against pig PG-1 in Western blot. Blank P. pastoris strain X33 and 300 ng of chemically synthesized rPG-1 served as assay negative and positive controls, respectively. (B) Indirect-HRP ELISA standard curve. Antibody detecting the cathelin domain of ProPG-1 and cathelin was used as the primary antibody. Cathelin peptide was used to generate the standard curve. (C) Recombinant ProPG-1 (from PP-ProPG-7) was detected with antibody against pig PG-1 in Western blot. 200 ng and 100 ng of chemically synthesized ProPG-1 served as positive controls (P200 and P100, respectively). (D) Recombinant cathelin (from PP-Cath) was detected with antibody against pig cathelin in Western blot. 200 ng and 150 ng of chemically synthesized cathelin served as positive controls (P200 and P150, respectively). Empty P. pastoris X33 strain served as a negative control.



Size, pI value, net charges were calculated by Innovagen's Peptide Property Calculator. Boman index was calculated by the predictive tool available at Antimicrobial Peptide Database v2.34. Native elastase site was replaced by enterokinase cleavage site (DDDDK<sup>∗</sup> ).

Boman Index (2.8–2.9). A high Boman Index in rProPG-1, rCath, and rPG-1 can indicate the potential of these peptides to interact with other proteins in vivo.

## Enhanced Cell Migration by PG-1 in Its Pro-, Cathelin-, and Mature- Form

We next sought to examine the effect of protegrin on modulating cell proliferation and migration in vitro using transwell cell migration assays with IPEC-J2 cells, a cell line derived from TABLE 5 | Minimum inhibitory concentration (MIC) values of the resulting mature PG-1 cleaved from rProPG-1 after enterokinase (EK) digestion, and rPG-1 expresses by PP-PG-1.


<sup>a</sup> Chemically synthesized mature PG-1 (CC-PG-1); <sup>b</sup>PG-1 cleaved from rProPG-1 after EK digestion; <sup>c</sup> rPG-1 expressed by PP-PG-1. MRSA: Methylene-resistant Staphylococcus aureus.

porcine intestine epithelium. As shown in **Figure 5A**, cell migration in groups treated with recombinant proform PG-1 (rProPG-1), pro-piece cathelin domain (rCath), and mature PG-1 (rPG-1) was enhanced. Migrated cell count analysis revealed that there was a two-fold (rPro-PG-1), 2.3-fold (rCath), and 2.5 fold (rPG-1) increase in cell migration when compared to the untreated negative control (p < 0.05, **Figure 5B**). To study if

these peptides (rProPG-1, rCath, and rPG-1) can increase cell proliferation, cell counting proliferation assay was performed. In comparison to the untreated control, groups treated with the peptides resulted in no significant change in cell proliferation (data not shown).

### Antimicrobial Activity of rPG-1 and Its Stability at Various pH and Temperature

The minimum inhibitory concentration (MIC) values of the resulting mature peptide (rPG-1) expressed by PP-PG-1 was determined by a two-stage radial diffusion assay. The lowest concentration of the rPG-1 that inhibited the visible growth of E. coli and MRSA after overnight incubation is the MIC (**Table 6**). rPG-1 conferred similar antimicrobial activity against E. coli at the same level as chemically synthesized PG-1 (CS-PG-1). The MIC value for rPG-1 expressed by PP-rPG-1 against MRSA is 37.5% lower than the CS-PG-1 at 8 µg/mL versus 5 µg/mL, respectively (**Table 5**). Differences in MIC may be attributed to differences in quantification methods of the chemically synthesized and recombinant peptides (e.g., empirical weighing

FIGURE 5 | Effect of proform PG-1, cathelin domain and mature PG-1 on cell migration. (A) Representative images of transwell migrated IPEC-J2 cells stained with Hoechst 33342. (B) Quantification of cell migration from transwell migration assay. Bars represent the mean ± SEM of three experiments. Asterisk <sup>∗</sup> (p < 0.05) and ∗∗ (p < 0.01) denote significant difference from untreated control group.

TABLE 6 | Stability of the rPG-1 at different pH and temperature in terms of MIC against E. coli after a pre-incubation time of 30 min, antimicrobial activity assay using colony count method was performed to determine the antimicrobial activity.


versus western blot densitometry). Various pH and temperature conditions were used to investigate the stability of rPG-1 after an incubation time of 30 min. As shown in **Table 6**, rPG1 retained potent activity in both acidic and alkaline (pH 2.0, 4.0, 6.0, and 11.0), and temperature conditions (25–80◦C). Antimicrobial activity was quite stable over a pH range of 2.0–11.0, although the MIC at a higher temperature (90◦C) increased by 25% compared to the MIC observed at 25◦C.

### DISCUSSION

The use of P. pastoris as an expression system has numerous advantages such as rapid and high-density growth in defined and relatively inexpensive media, high recombinant protein yield, and post-translational modification abilities similar to that of mammalian cells. P. pastoris also secrete a low number of endogenous proteins, making downstream protein recovery and purification more feasible. AMPs that have been expressed by P. pastoris such as hPAB-β, Ch-penaeidin, MP1102 (Li et al., 2005; Chen et al., 2011; Zhang et al., 2015) are driven by the AOX promoter system. These studies indicate methylotrophic yeast-inducible system is suitable for high-level expression of active AMPs. Pro-healing and immune-modulating aspect of cathelicidin AMPs are primarily based on the human and mouse cathelicidin AMP. In our work, we investigated the feasibility of expressing recombinant PG-1 in its pro-, cathelin-, and matureform in P. pastoris and characterized their antimicrobial activity in vitro. We then examined the effect of recombinant PG-1 in its various forms on cell migration and proliferation in vitro. To our knowledge, our finding is the first report on the role of PG-1 on cell migration and thus tissue repair potential. Interestingly, a recent report showed an increased expression of porcine β-Defensin2 (PBD2), pBD3, pBD114, pBD129, and protegrins (PG) 1-5 in IPEC-J2 cells when exposed to a host defense peptide (HDP)- synthesizing Lactobacillus reuteri I5007 (Liu et al., 2017). Similarly, butyrate and its analogs showed induction of porcine HDP gene expression with approximately 20-, 45-, 60-, and 80 fold induction was observed for pBD2, PG1-5, pEP2C, and pBD3, respectively, using IPEC-J2 intestinal epithelial cells (Zeng et al., 2013). These findings together with our findings on the PG-1 tissue damage healing potential suggest that PG-1 may be one of the components of innate defense during infection.

To investigate the feasibility of producing rProPG-1, rCath, and rPG-1 for downstream studies and application, we constructed plasmids for the expression and secretion of these three peptides in P. pastoris. Stable clones were generated via integration of the constructs into the P. pastoris genome. This will also eliminate the need for constant selective pressure during fermentation. With integration, it is possible to generate strains harboring more than one copy of the introduced expression cassette via repeated single crossover DNA recombinant events (Clare et al., 1991a; Nordén et al., 2011). Insertion of multiple recombinant gene copies is often desired to enhance recombinant protein yield, as low transcript levels can be a limiting factor in protein production (Nordén et al., 2011). To increase the

number of integrated expression cassettes, we initially attempted to construct a known number of concatemers of the expression cassette prior to transformation into P. pastoris, similar to Mansur et al. (2005). We found the stable generation of long concatemers of the expression construct difficult to achieve. Instead, we then enriched for strains with an increased number of vector copies by plating transformants on medium containing increasing levels of the selection drug (e.g., G418), where higher drug resistant clones can correlate with higher gene dosage (Nordén et al., 2011). Clones with putative multiple copies of the rProPG-1, rCath, and rPG-1 expression cassette from varying concentrations of G418 antibiotic selection were then selected to evaluate recombinant protein expression levels. Although it was evident that there is a positive correlation between gene dosage level and level of expression (**Tables 1–3**), an upper limit of gene dosage was observed. This bell-shaped correlation curve between gene copy number and protein yield has been reported in P. pastoris, where production and secretion level gradually decline with increased gene dosage (Zhu et al., 2009). This upper limit of gene dosage has been suggested to be related to endoplasmic reticulum (ER) stress with the accumulation of recombinant protein within the ERs, also further indicating there may be other bottleneck factors limiting recombinant protein production in addition to gene dosage.

To potentially obtain increased yield of recombinant protegrin (rProPG-1, rCath, and rPG-1), high-density fermentation was also performed. Recombinant protein was detected upon modification of the bioreactor BSM by reducing salt concentrations by 50% (BSM50). It has been reported that high salt concentration can increase cell death, possibly caused by high osmotic pressure in the medium (Brady et al., 2001; Jahic et al., 2006). Upon cell lysis, the secreted recombinant protein can undergo proteolysis as cellular proteases are released into the culture medium. The addition of casamino acids, yeast extract and peptone can help address the issue of protein degradation during the cultivation process (Jahic et al., 2003). During rPG-1 expression by PP-PG-1, rPG-1 was not detected until yeast extract and peptone was added to the BSM50 culture medium, suggesting modification of media formulation can improve yield. The yeast extract, peptone and casamino acids may serve as preferential substrates for proteolytic enzymes, thereby decreasing degradation of the expressed recombinant protein of interest (Clare et al., 1991b; Jafari et al., 2011). Interestingly, the addition of these substrates to the medium was more critical in PP-PG-1 fermentation than PP-ProPG-1 and PP-Cath. This could be due to the peptide properties of rPG-1 toxicity against the P. pastoris host cells. These substrates can potentially bind or interact with rPG-1 to limit its toxicity against P. pastoris. Components of microbial growth medium have been shown to exhibit binding activity to metal ions (Ramamoorthy and Kushner, 1975) and may also affect recombinant protein folding by modulating the expression of chaperone proteins (Rosano and Ceccarelli, 2014). Furthermore, fermentation medium composition can influence the physio-chemical surface properties of microbes. Microbial cell wall protein composition can differ when cultured in medium with and without peptone and yeast extract (Schär-Zammaretti et al., 2005). The expression

level of rPG-1 (104 ± 11 mg/L) is comparable to other AMPs expressed using the similar P. pastoris expression system ranging from 108 mg/L to 292 mg/L (Li et al., 2005; Chen et al., 2011; Zhang et al., 2015). Of the three protegrin peptides expressed, rPG-1 had the lowest yield compared to rProPG-1 and rCath (0.8 ± 0.10 g/L and 0.2 ± 0.02 rCath g/L, respectively). The isoelectric point (pI) of a recombinant protein can affect expression level, where high pI is associated with no detectable protein expression (Boettner et al., 2007). This is in line with the high pI value of rPG-1 (10.7; **Table 4**) and the lower expression level obtained. This further suggests pI can be a potential parameter influencing protein expression. Addition of fusion protein tags to aid purification can potentially affect recombinant protein expression level. To increase protein yield in future experiments, the theoretical pI of rPG-1 can be reduced from 10.6 to 8.3 upon the fusion of glutathione S-transferase (GST) tag.

To reduce the likelihood of AMPs exerting their toxicity effect against host cells, fusing AMPs to a foreign or native partner protein can aid in enhancing protein production and reduce proteolytic degradation. In our work, we replaced the native elastase cleavage site of proform protegrin with an EK cleavage site. Antimicrobial activity was not exhibited until cleavage occurred in the presence of EK, releasing the mature and active PG-1 (**Figure 4B**). EK is a serine protease and is naturally produced and secreted in the intestinal duodenum to convert trypsinogen into its active trypsin form for subsequent activation of pancreatic digestive enzymes in pigs (Baratti et al., 1973). The addition of the EK cleavage site can permit rProPG-1 to be cleaved by the endogenous intestinal EK to release the mature antimicrobial PG-1. This allows the potential application of rProPG-1 in animal feed to replace the use of conventional antibiotics that are contributing to the rapid emergence of antibiotic resistant bacteria. Interestingly, while the MIC for both rPG-1 and the EK cleaved rProPG-1 for E. coli is the same as that of the chemically synthesized PG-1 control, their MICs (10, and 8 µg/mL for the mature PG-1 cleaved from rProPG-1 and the mature rPG-1, respectively) for MSRA are higher than that of the control (5 µg/mL). Similar MIC results were found using protegrin-1 against E. coli (4 µg/mL), and MRSA (4 µg/mL) (Yan and Hancock, 2001). In another report, the MICs of PG-1 against Gram-positive and Gram-negative bacteria range from 0.12 to 2 µg/mL (Steinberg et al., 1997).

A high Boman index value (e.g., 2.50–3.00) can indicate an AMP to be multifunctional within the cell due to its ability to interact with a wide range of proteins (Boman, 2003; Nam et al., 2014). A lower index value (≤1) can suggest that the AMP has low side effects (e.g., low hemolytic activity). Since protegrin, in its pro-, cathelin-, and mature-form, is predicted to interact with other proteins, we sought to examine if these peptides have other functions besides antimicrobial activity. We found pro-, cathelin-, and mature-form PG-1 enhanced cell migration but not cell proliferation (**Figure 5**), suggesting these peptides may interact with cell-surface receptors. It was reported that cathelicidin LL-37 and mCRAMP trans-activated epidermal growth factor receptor (EGFR) (Tokumaru et al., 2005; Yang et al., 2006). Whether PG-1, and its proform and the cathelin domain

have their migration function via activating membrane receptor warrant further investigation.

For therapeutic applications, the stability of the recombinant AMPs produced is important. The stability test results indicated that the rPG-1 retain antimicrobial activity in a wide range of pH conditions ranging from pH 2.0 to 11.0. The pH in wounds ranges from 5.0 to 8.5, which is suitable for rPG-1 use. Oral application of rPG-1 may also be feasible in the digestive tract (pH 3.0 and 4.4 in mouse and pig, respectively). The activity of rPG-1 remained stable over a broad temperature range from 25◦C to 95◦C. This stability may be due to the high cysteine content (disulfide bonds) in rPG-1 (Lai and Gallo, 2009). In conclusion, recombinant protegrin in its pro-, cathelin-, and mature- forms were successfully expressed in P. pastoris using the AOX expression system. These peptides exhibit antimicrobial activity as well as cell migration activity. Further in vivo studies on the role of PG in modulating intestinal health would be of interest for intestinal inflammation application (Harwig et al., 1996).

### REFERENCES


## AUTHOR CONTRIBUTIONS

EH performed the major experiments and wrote the manuscript. NA performed parts of experiments and helped in writing. JL helped in designing the research, data interpretation and editing.

## FUNDING

This project has been funded by Ontario Pork (grant # 051379), Natural Sciences and Engineering Research Council of Canada (NSERC, grant # 400876), and Foshan University (grant # 053385).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.02300/full#supplementary-material

peptides that inhibit biofilms formed by pathogens isolated from cystic fibrosis patients. Antibiotics 3, 509–526. doi: 10.3390/antibiotics3040509



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Huynh, Akhtar and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Bactericidal Property of Oregano Oil Against Multidrug-Resistant Clinical Isolates

### Min Lu<sup>1</sup> , Tianhong Dai<sup>1</sup> \*, Clinton K. Murray<sup>2</sup> and Mei X. Wu<sup>1</sup> \*

<sup>1</sup> Wellman Center for Photomedicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, United States, <sup>2</sup> First Area Medical Laboratory, JBSA-Fort Sam Houston, Houston, TX, United States

Development of non-antibiotic alternatives to treat infections caused by multidrugresistant (MDR) microbes represents one of the top priorities in healthcare and community settings, especially in the care of combat trauma-associated wound infections. Here, we investigate efficacy of oregano oil against pathogenic bacteria including MDR isolates from the combat casualties in vitro and in a mouse burn model. Oregano oil showed a significant anti-bacterial activity against 11 MDR clinical isolates including four Acinetobacter baumannii, three Pseudomonas aeruginosa, and four methicillin-resistant Staphylococcus aureus (MRSA) obtained from combat casualties and two luminescent strains of PA01 and MRSA USA300, with a MIC ranging from 0.08 mg/ml to 0.64 mg/ml. Oregano oil also effectively eradicated biofilms formed by each of the 13 pathogens above at similar MICs. Transmission electron microscopy (TEM) and scanning electron microscopy (SEM) revealed that oregano oil damaged bacterial cells and altered the morphology of their biofilms. While efficiently inactivating bacteria, there was no evidence of resistance development after up to 20 consecutive passages of representative bacterial strains in the presence of sublethal doses of oregano oil. In vivo study using the third-degree burn wounds infected with PA01 or USA300 demonstrated that oregano oil, topically applied 24 h after bacterial inoculation, sufficiently reduced the bacterial load in the wounds by 3 log<sup>10</sup> in 1 h, as measured by drastic reduction of bacterial bioluminescence. This bactericidal activity of oregano oil concurred with no significant side effect on the skin histologically or genotoxicity after three topical applications of oregano oil at 10 mg/ml for three consecutive days. The investigation suggests potentials of oregano oil as an alternative to antibiotics for the treatment of wound-associated infections regardless of antibiotic susceptibility.

Keywords: oregano oil, Pseudomonas aeruginosa, Acinetobacter baumannii, MRSA, biofilms, burn wound, mouse model, bioluminescence imaging

### INTRODUCTION

Skin wound infection is a widespread problem in both civilian and military healthcare settings. Skin wounds are particularly prone to bacterial infections because the wounds provide an ideal medium for bacterial proliferation and a portal of entry into the bloodstream and are direct exposure to the "dirty" environment. The infections can be readily treated with a variety of antibiotics if the bacteria involved are susceptible. However, there are only extremely limited or no treatment options when

### Edited by:

Noton Kumar Dutta, Johns Hopkins University, United States

### Reviewed by:

Nagendran Tharmalingam, Brown University, United States Sougata Ghosh, RK University, India

\*Correspondence:

Tianhong Dai TDAI@mgh.harvard.edu Mei X. Wu MWU5@mgh.harvard.edu

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 30 March 2018 Accepted: 11 September 2018 Published: 05 October 2018

### Citation:

Lu M, Dai T, Murray CK and Wu MX (2018) Bactericidal Property of Oregano Oil Against Multidrug-Resistant Clinical Isolates. Front. Microbiol. 9:2329. doi: 10.3389/fmicb.2018.02329

**229**

antibiotic resistant strains are involved in the wound infections, which occurs at a worrying speed. If the wound infection cannot be eliminated in a timely fashion, the infection would alter cellular metabolisms and induce persistent inflammation systemically that can predispose the patients to various complications and life-threatening sepsis (Fitzwater et al., 2003; Wang et al., 2018). For instance, burn wound infection outbreaks caused by multidrug-resistant (MDR) organisms emerged as a serious problem early in the course of Iraq military operations despite that the United States military has provided rapid and highly effective care for wounded soldiers (Scott et al., 2007; Vento et al., 2013). As a matter of fact, skin infections caused by MDR bacteria are the most common cause of morbidity and mortality in patients infected with MDR microbes and represent almost 61% of deaths of this infected population (Gomez et al., 2009).

Extensive uses of broad spectrum antibiotics are the single most important factor in evolution of bacterial resistance (Hampton, 2013). A number of studies have shown that the most frequently identified MDR strains of bacteria in nosocomial infections and on the battlefield are Gram-negative bacteria Acinetobacter baumannii and Pseudomonas aeruginosa, and Gram-positive bacterium methicillin-resistant Staphylococcus aureus (MRSA) (Scott et al., 2007; Calhoun et al., 2008; Li et al., 2014; Levin-Reisman et al., 2017). In addition, bacterial biofilms formed by MDR bacteria are the major obstacles in treatment of burn wounds (Bloemsma et al., 2008; Jiang et al., 2017). Bacteria within biofilms can be as much as 1,000 times more resistant to antibiotics and are responsible for recurrent antibiotic-resistant infections elsewhere in the body upon dissemination from the site of the biofilm (Ceri et al., 1999; Caraher et al., 2007). Currently, the only effective treatments available to fight these infections are older drugs like colistin, which are highly toxic and detrimental to the overall health of the patients (Crane et al., 2009). There is a pressing need for the development of non-antibiotic approaches to combat MDR microbes.

Essential oils (EOs) are a mixture of volatile constituents produced by aromatic plant/herbs. There are about 3,000 wellrecognized EOs, of which 300 are generally recognized as safe (GRAS) to humans by the United States Food and Drug Administration (U.S. FDA) and have broad applications in food preservation, additives, and favors, perfume, cosmetic industries, antiseptic oral solutions, toothpastes, cleaner, and air fresheners for centuries (Pandey et al., 2017; Sakkas and Papadopoulou, 2017). These natural products are of particular interest as "green" antimicrobial agents because of their lowcost, biocompatibility, potential antibiofilm properties, and friendly to eukaryote cells and environment (Burt, 2004; Nostro et al., 2007; Kavanaugh and Ribbeck, 2012). Among these safe EOs, oregano oil has been shown to have a variety of activities such as antioxidant (Yan et al., 2016), antiinflammatory (Ocana-Fuentes et al., 2010; Shen et al., 2010), anti-fungal (Akgul and Kivanc, 1988; Soylu et al., 2007), and anti-allergic (Benito et al., 1996). Its antimicrobial effect has been demonstrated in vitro cell culture, food systems studies (Lopez-Reyes et al., 2010; Soylu et al., 2010; Munhuweyi et al., 2017), and in vivo systemic infections (Manohar et al., 2001; Preuss et al., 2005). In the present study, we investigate effectiveness of oregano oil in inactivation of MDR bacteria isolated from combat casualties in vitro and bioluminescent strains of P. aeruginosa (PA01) and MRSA (USA300) in mouse burn wounds. Our study showed that oregano oil effectively inactivated various pathogenic bacteria and their biofilms irrespective of their antibiotic susceptibility. The study is the first in vivo attempt on the use of oregano oil for the treatment of burn wounds infected with clinically important MDR bacteria.

## MATERIALS AND METHODS

## Chemical Constituents of Oregano Oil

Oregano oil was purchased from Bulk Apothecary (Aurora, OH, United States) and used throughout the study. To define the constituents of oregano oil, gas chromatography/mass spectrometry (GC-MS) analyses were carried out using an Agilent 6980 GC coupled to an Agilent 5973N MS and a fusedsilica capillary column (HP-5MS: 30 m × 0.25 mm i.d., film thickness 0.25 µm). The initial column temperature was set at 60◦C for 10 min and then increased at 3◦C/min till 220◦C. The temperature was held at 220◦C for 10 min and raised to 240◦C by increments of 1◦C/min; the injector port temperature was 250◦C with the carrier gas of helium at a flow rate of 0.8 ml/min. Ionization voltage of MS in the EI-mode was 70 eV and ionization source temperature at 250◦C with a mass range of 35–465 amu. The volatile components were identified by comparison of their retention indices relative to n-alkanes (C6- C28) and mass spectra with those of authentic compounds by means of NIST and Wiley databases and with the Adams library spectra.

### Bacterial Strains

The antibacterial activity of oregano oil was tested against a panel of MDR bacteria isolated from combat casualties, including seven Gram-negative strains A. baumannii (AF0004, AF0005, IQ0012, and IQ0013) and P. aeruginosa (AF0001, IQ0042, and IQ0046), and four Gram-positive MRSA (AF0003, IQ0064, IQ0103, and IQ0211). All the bacterial isolates were obtained from San Antonio Military Medical Center under a Materials Transfer Agreement and demonstrated MDR according to the microbiology tests performed at the United States Army Institute of Surgical Research (**Table 2**). Amongst the A. baumannii and P. aeruginosa tested, only strains of IQ0012, IQ0013, and IQ0042 were susceptible to imipenem; the rest of strains were resistant to all antibiotics tested. The five MRSA strains were resistant to amikacin, ampicillin, cefazolin, cefoxitin, and erythromycin, but susceptible to gentamicin, levofloxacin, nitrofurantoin, and tetracycline. In addition, luminescent strains of P. aeruginosa (PA01) and MRSA USA300 were used in the in vivo study, allowing real-time monitoring of infection in the mouse burn wounds in vivo via bioluminescence imaging (Dai et al., 2013; Zhang et al., 2014; Wang et al., 2016).

## Determinations of Minimum Inhibitory Concentration (MIC)

In order to determine a MIC, a broth microdilution assay was employed as previously described (Joshi et al., 2010; Gao et al., 2011). Stock solution of oregano oil was prepared at 40 mg/ml in DMSO and twofold dilutions (0.04–1.28 mg/ml) of the stock EO in brain heart infusion (BHI) medium were added into 96 well plates for bactericidal tests. In each well, 20 µL of the suspensions containing 10<sup>8</sup> CFU/ml of bacteria was added to 180 µL of the above medium containing oregano oil at varying concentrations. Medium supplemented with a similar amount of DMSO only severed as controls. The microplates were incubated at 37◦C for 24 h and examined for bacterial growth. The first well without turbidity was determined as a MIC value. All assays were performed in triplicate.

## Antibiofilm Activity

Bacteria were incubated in trypticase soy broth (TSB) with 0.1% glucose at 37◦C for 18 h, after which the cultures were harvested by centrifugation and washed twice with PBS. Bacterial suspensions with an optical density of OD<sup>600</sup> equal to 0.1 in TSB were added to 96-well plates at 100 µL/well, followed by incubation at 37◦C under static condition for 24 h to form biofilms. Oregano oil was added at indicated concentrations and incubated with the biofilms for 1 h, after which biofilms were washed twice with PBS and bacterial viability was determined using an Alamar Blue assay. TSB without bacteria was used as a negative control. All experiments were performed in triplicate.

## Assessment of Possible Resistance Development to Oregano Oil

To study any potential development of resistance to oregano oil, three representative strains (A. baumannii AF0005, P. aeruginosa IQ0042, and MRSA IQ0064) were propagated for 20 generations in the presence of sublethal doses of oregano oil as previously described (Li et al., 2014). Briefly, 200-µL aliquots of the bacterial suspensions (10<sup>7</sup> CFU/ml) were inoculated into 96-well plates and exposed to the sub-MIC (2/3 MIC) of oregano oil at 37◦C for 1 h, and the resultant bacteria were labeled as the first generation and tested for a MIC as above. The second generation was obtained by exposing the first generation to its sub-MIC for 1 h and determined for its MIC again. The procedure was repeated for up to 20 times. Oregano-resistance was determined by any significant increases in the MIC of successive generations.

## Transmission Electron Microscopy (TEM)

To determine bactericidal mechanism of oregano oil, A. baumannii AF0005 and P. aeruginosa IQ0042 were investigated as the representative strains for oregano-induced ultrastructural damages using transmission electron microscopy (TEM). Bacterial suspensions were fixed in 2.5% glutaraldehyde plus 2% paraformaldehyde overnight at 4◦C. Fixed cells were collected by centrifugation at 4,000 × g for 5 min and rinsed with 0.1 M sodium cacodylate buffer (pH 7.2) for three times. After the final wash, hot agar was added to each pellet and the cell pellets were post-fixed in 2% osmium tetroxide for 1 h, dehydrated with a graded ethanol series, embedded in fresh Epon, and then polymerized at 60◦C for 48 h. Ultra-thin sections were cut on ultramicrotome and collected onto 200 mesh bare copper grids. Samples were stained for 30 min with uranyl acetate and lead citrate and examined with a CM-10 TEM (Philips, Eindhoven, Netherlands).

## Scanning Electron Microscopy (SEM)

To investigate the ultrastructural changes of bacterial biofilms caused by oregano oil, scanning electron microscopy (SEM) was performed using the representative strains of P. aeruginosa IQ0042 and MRSA IQ0064. Briefly, biofilms of IQ0042 and IQ0064 were grown for 24 h on sterilized squares of ACLAR 33C (Electron Microscopy Sciences, Hatfield, PA, United States), and treated for 1 h with oregano oil at 0.75 mg/ml or 0.3 mg/ml, respectively. Untreated and oregano-treated biofilms were fixed at 4◦C for 24 h in 0.1 M sodium cacodylate buffer containing 2.5% glutaraldehyde, 0.15% alcian blue, and 0.15% safranin O. The fixed biofilms were washed with 0.1 M sodium cacodylate buffer, infiltrated with 2% osmium tetroxide for 2 h, and dehydrated to 100% ethanol. The biofilms were dried using a criticalpoint dryer (Tousimis Research Corporation, Rockville, MD, United States), mounted on specimen stubs, sputter-coated with 10 nm Cressington 208 platinum (Cressington Scientific Instruments, Watford, United Kingdom), and examined on a S4800 SEM (Hitachi Ltd., Tokyo, Japan). Micrographs were acquired under high vacuum using an accelerating voltage of 3.0 kV.

### Animal

Female BALB/c mice at 8 weeks of age and 17–19 g were purchased from Charles River Laboratories (Wilmington, MA, United States). All animal procedures were approved by the Institutional Animal Care and Used Committees (IACUC) of Massachusetts General Hospital (Protocol 2014N000009) and were in accordance with guidelines of the National Institutes of Health.

### Treatment of Burn Infection in Mice by Oregano Oil

Mice were anesthetized with an intraperitoneal injection of ketamine-xylazine cocktail and shaved on the lower dorsal skin. The burn was introduced by a brass block (1 cm<sup>2</sup> ) heated to thermal equilibration with boiling water prior to application of its extremity onto the shaved skin for 5 s, which generated a third-degree burn wound. Sterile saline was intraperitoneally administered at 0.5 ml/mouse to support fluid balance during recovery. Aliquots of 50 µL bacterial suspensions containing 5 × 10<sup>6</sup> CFU in PBS were inoculated onto the burn 30 min after the injury and remained in place while the mice recovered from anesthesia. Luminescent strains of P. aeruginosa PA01 and MRSA USA300 were used as the causative pathogens in the study. At 24 h after bacterial inoculation when the biofilms were formed in the wounds, oregano oil was diluted with grape seed oil, which is commonly used for EO dilution in aromatherapy, at a final

concentration of 5 or 10 mg/ml, and topically applied to the infected wounds.

Bioluminescence emission of the bacteria in the wounds was recorded in real time by a Lumina in vivo image system (IVIS) (PerkinElmer, Waltham, MA, United States). Bioluminescence images were acquired (60 s exposure, medium binning) at different time points after infection. During imaging, mice were anesthetized in chambers containing 2.0% isoflurane inhalant mixed with oxygen via an IVIS manifold placed within the imaging chamber. Bioluminescence was quantified with the Living Image software (Xenogen).

For measurement of bacterial burden, the infected burn wounds were collected after sacrifice of the mice on day 7 of bacterial inoculation. The collected tissues were homogenized in 2 ml sterile PBS. The resultant homogenate was serially diluted and spotted onto BHI agar plate containing Skirrow's supplement (10 µg/ml vancomycin, 5 µg/ml trimethoprim lactate, and 2500 IU/L polymyxin B). The plates were then incubated at 37◦C for 24 h and bacterial colonies were enumerated in a treatment-blind fashion.

## Gram Stain

Gram stain was carried out as previously described, with some modifications, to corroborate formation of PA01 biofilms in the infected wounds (Christensen et al., 2013; Wang et al., 2016). Briefly, at 24 h after bacterial inoculation, the infected wounds were excised, fixed in 10% phosphate-buffered formalin for 2 days, and then embedded in paraffin. Tissue sections were cut at 5 µm, de-paraffined, and rehydrated, followed by staining with 0.8% crystal violet in 1% sodium bicarbonate for 1 min and then in gram's iodine for 2 min. After decolorization with acetone/alcohol = 1:1 (v/v), the tissue sections were counterstained with 0.1% safranin O for 2 min, washed, air dried, and mounted with permount (Fisher Scientific, Waltham, MA, United States). Sections were visualized by Hamamatsu NanoZoomer 2.0 HT and the images were processed using NDP viewer software.

## Toxicity of Oregano Oil to Mouse Skin in vivo

To evaluate any possible toxicity of oregano oil to the skin in vivo, mice were shaved on the low dorsal skin 24 h prior to application of oregano oil. Oregano oil at 10 mg/ml was applied topically on the shaved area once a day for three consecutive days. Mice treated with PBS served as negative controls. The mice were sacrificed 24 h after the final oregano oil application, and the skin was cross-sectioned using 8 mm biopsy punch for standard histological examination. The tissue sections stained with hematoxylin and eosin (HE) were visualized by Hamamatsu Nanozommer 2.0 HT and the images were processed using NDP viewer software.

DNA damage in oregano oil-treated skin was assessed using the DeadEnd Fluorometric TUNEL system (Promega, Madison, WI, United States), in which damage DNA undergoes end labeling with fluorophore as per the manufacturer's instruction. Briefly, after deparaffinization and rehydration, the tissue sections were incubated with Proteinase K for 10 min to permeabilize the cells, washed, and stained with the TUNEL reaction mixture for 1 h at 37◦C in a humidifies chamber. The sections were counterstained with DAPI to mark cell nuclei. Fluorescence images were captured using a FluoView FV1000-MPE confocal microscopy (Olympus Corporation, Tokyo, Japan). For the positive control, tissue sections were pre-treated with 10 unit/ml of RQ1 RNase-free DNase I for 10 min to induce DNA fragmentation before the sections were assayed by the TUNEL staining kit.

## Statistical Analyses

Data are presented as means ± standard deviations (SDs). Statistical significance was assessed with two-tailed Student's t-test between two groups or one-way ANOVA for multiple group comparison. P-values of < 0.05 were considered statistically significant. All statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software).

## RESULTS

## GC/MS Analysis of Constituents in Oregano Oil

Chemical ingredients of commercial oregano oil were identified by GC/MS analysis (**Table 1**). The twenty constituents accounted for 97.45% of the total amount of ingredients in



<sup>a</sup>Retention index relative to C6-C28 n-alkanes on the HP-5MS capillary column. <sup>b</sup>Retention index identification by using the Kovats index. <sup>c</sup>Expressed as percentage of the total peak area of the chromatograms without correction.


oregano oil. The phenols carvacrol (72.25%) and thymol (6.62%), as well as the monoterpene hydrocarbons p-cymene (5.21%), γ-terpinene (4.12%), and α-pinene (1.21%) were the predominant components of oregano oil.

### Oregano Oil Effectively Inactivated Bacteria in vitro Irrespective of Antibiotic Sensitivity

The MIC values of oregano oil against the 13 bacterial strains are shown in **Table 2**. Oregano oil showed a significant antibacterial activity over PBS controls against A. baumannii strains of AF0004, AF0005, IQ0012, and IQ0013 and MRSA strains of AF0003 and IQ0211, with the MICs ranging from 0.08 to 0.16 mg/ml. The MICs were significantly lower than those MICs ranging from 0.32 to 0.64 mg/ml against strains of P. aeruginosa and MRSA strains IQ0064, IQ0103, and USA300 (**Table 1**). Oregano oil also exhibited similar antibacterial activities against established biofilms (24-h-old) formed by the 13 bacterial strains within 1 h, with complete inactivation of the biofilms of A. baumannii, P. aeruginosa, and MRSA at the concentrations of 0.3, 1.0, and 0.4 mg/ml, respectively, in good agreement with the MIC values for planktonic bacterial cells (**Table 1** and **Figure 1**). The results clearly suggest that oregano oil can overcome the obstacles of biofilms and kill bacteria within as sufficiently as planktonic bacteria, in contrast to antibiotics that kills bacterial biofilms poorly.

### TEM and SEM Illustrated Ultrastructural Damages of Bacteria

Transmission electron microscopy showed ultrastructural damages of A. baumannii AF0005 (**Figure 2B**) and P. aeruginosa IQ0042 cells (**Figure 2D**) after exposure to oregano oil for 1 h at 0.16 mg/ml and 0.56 mg/ml, respectively. The cell wall and membrane damages were apparent in A. baumannii AF0005 and P. aeruginosa IQ0042 cells with a severe leakage of intracellular substances resulting in cell membrane shrinking and separating from cell wall (**Figures 2B,D**, arrows). Moreover, cytoplasmic vacuoles in A. baumannii AF0005 (**Figure 2B**, asterisk) and many stainless-vesicles in P. aeruginosa IQ0042 (**Figure 2D**, asterisk) were observed. Intracellular structural discontinuation such as dissociation between cell wall and membrane was also seen in A. baumannii AF0005 (**Figure 2B**, oval). In comparison, untreated A. baumannii AF0005 (**Figure 2A**) and P. aeruginosa IQ0042 cells (**Figure 2C**) had intact, clear cell wall and membrane and dense and homogeneous cytoplasm. However, we did not find any significant differences in the ultrastructure between the control and oregano oil-treated MRSA USA300 by TEM (data not shown), which probably hints at different responses of MRSA from A. baumannii or P. aeruginosa. As with biofilms, dense and thick bacterial biofilm was observed on the dentin surface, comprised of numerous layers of densely concentrated cocci in 24-h-old P. aeruginosa IQ0042 (**Figure 2E**) and MRSA IQ0064 (**Figure 2G**) biofilms. The biofilms of P. aeruginosa IQ0042 (**Figure 2E**) and MRSA IQ0064 (**Figure 2G**) were treated with oregano oil at 1 mg/ml and 0.4 mg/ml for 1 h, respectively. The cells decohered in the extracellular polymeric matrix, dead

fmicb-09-02329 October 3, 2018 Time: 19:15 # 5

TABLE 2 | MIC

(mg/ml) of the oregano EO and results of antibiotic susceptibility

 testing for the pathogens.

bacteria were readily seen all over, and biofilms were destroyed completely in oregano oil-treated samples (**Figures 2F,H**, arrows). There were only a few bacteria scantly growing on the dentin surface owing to intensive cell death (**Figures 2F,H**).

## No Evidence of Resistant Development to Oregano Oil

Risk of resistant development was evaluated with three representative strains: A. baumannii AF0005, P. aeruginosa IQ0042, and MRSA IQ0064. The bacteria were cultured up to 20 successive passages in the presence of sub-lethal doses of oregano oil for bacterial inactivation. As shown in **Figure 3A**, A. baumannii AF0005, P. aeruginosa IQ0042, and MRSA IQ0064 retained susceptibility to the original MIC values of 0.16, 0.32, or 0.56 mg/ml, respectively, after 20 cycles of treatment. Moreover, there were no statistically significant differences in the survival rates of all the bacterial strains among the cycles 0, 1, and 20 in the three tested strains, after exposure to oregano oil at the MICs for 24 h (**Figures 3B–D**). The results indicate that resistance to oregano oil of the bacterial strains did not take place under this condition.

## Oregano Oil Significantly Reduced Bacterial Burden in Burn Wounds Infected With PA01

Gram stain of histological longitudinal section (**Figure 4A**) and crossing section (**Figure 4B**) of a representative skin specimen demonstrated the presence of PA01 biofilms at 24 h after bacterial inoculation, as evidenced by abundant bacteria densely clustered together (red) in a highly hydrated extracellular matrix on the surface of skin and in the epidermis (**Figure 4A**, outline in red). In addition, highly resilient microbial assemblies were readily found in the dermis suggesting that the PA01 could infect not only the epidermis but also the dermis within 24 h (**Figure 4B**). In successive bacterial luminescence images of representative wounds infected with 5 × 10<sup>6</sup> CFU of PA01, oregano oil treatment at 10 mg/ml almost completely eradicated bacterial luminescence in 60 min, while luminescence remained unchanged during the same period in untreated mice (**Figures 4C,D**). Moreover, there was no recurrence of infection in the following days in oregano oil-treated mice, whereas the mice remained significantly infected in untreated mice in the same experimental period (**Figures 4C,D**). An average reduction in bacterial luminescence of 2.9 log<sup>10</sup> and 3.5 log<sup>10</sup> were achieved in 60 min at a concentration of 5 or 10 mg/ml of oregano oil, respectively (**Figure 4E**). On the contrary, bacterial luminescence of the wounds in the absence of oregano oil treatment was almost unaltered with only 0.08 log<sup>10</sup> reduction during the equivalent period (**Figure 4E**, P < 0.0001). A time course study of the mean bacterial luminescence from days 2 to 7 after bacterial inoculation corroborated that the treatment consistently and significantly lowered the luminescence compared to untreated mice during the whole period of the experiment regardless of whether oregano oil was used at 5 or 10 mg/ml (**Figure 4F**). The mean areas under the curve (AUC) of the bioluminescence time course were 6.9 × 10<sup>9</sup> and 2.4 × 10<sup>9</sup> for oregano oil-treated groups at 5 and 10 mg/ml, respectively, but it was 5.9 × 10<sup>10</sup> for untreated mice (P < 0.0001; **Figure 4G**), representing an 8.6-fold or a 24.6-fold reduction of the AUC in infected burns by oregano oil. We next excised the infected wounds on day 7 to determine bacterial CFU remaining in the wounds. There were 6.7 × 10<sup>6</sup> and 2.4 × 10<sup>6</sup> CFU in the wounds treated with oregano oil at a concentration of 5 or 10 mg/ml, respectively, which was significantly lower than the bacterial burden of 6.6 × 10<sup>7</sup> CFU/wound in untreated mice (**Figure 4H**).

### Oregano Oil Significantly Reduced Bacterial Burden in Burn Wounds Infected With USA300

As shown in **Figure 5A**, an average reduction in bacterial luminescence of 2.9 log<sup>10</sup> was attained when the infected wounds were treated with oregano oil at 5 mg/ml for 40 min, which was highly significant compared to only 0.4 log<sup>10</sup> decline of the bioluminescence in the absence of oregano oil during the same period (P < 0.0001; **Figure 5A**). The diminished bacterial luminescence was persistent for 7 days after a single oregano

treatment for 1 h. Oregano oil was used at 0.16 mg/ml for A. baumannii AF0005 cells, 0.56 mg/ml for P. aeruginosa IQ0042 cells, 1.0 mg/ml for P. aeruginosa IQ0042 biofilms, and 0.4 mg/ml for MRSA IQ0064 biofilm. (A,B) A. baumannii AF0005 cells; (C,D) P. aeruginosa IQ0042 cells; (E,F) P. aeruginosa IQ0042 biofilms; and (G,H) MRSA IQ0064 biofilms. Shown are cell wall and membrane damages (B,D; arrows); dissociation between cell wall and membrane (B; oval), cytoplasmic vacuoles and bubbles (B,D; asterisk), and cell collapse (F,H, arrows). The number of bacteria was drastically reduced after oregano oil treatment in (F,H) as compared to untreated controls (E,G).

oil treatment at 5 mg/ml (**Figure 5B**). The mean AUC of the bioluminescence were 3.7 × 10<sup>7</sup> in oregano oil-treated group, but it was 1.8 × 10<sup>9</sup> for untreated mice (P < 0.0001; **Figure 5C**), a 48.6-fold reduction of the AUC by oregano oil treatment. In accordance with this, a single dose of oregano oil treatment diminished bacterial load to 2.7 × 10<sup>7</sup> CFU/mouse that was 18 fold lower than 4.8 × 10<sup>8</sup> CFU/mouse in the untreated group (P < 0.0001; **Figure 5D**).

### No Side Effects in Mouse Skin Caused by Oregano Oil

There was no noticeable skin reaction, as visualized with the naked eye, after three consecutive days of oregano oil treatment at 10 mg/ml (**Figure 6B**) when compared to untreated skin (**Figure 6A**). On the histological levels, the skin maintained an undisturbed structure with a clear layer of healthy epidermal cells on the top of the dermis, indistinguishable to mocktreated skins (**Figures 6C,D**). Genotoxicity was next evaluated by a TUNEL assay to examine any DNA damage induced by oregano oil. In comparison with mock-treated controls, there was no any apparent increase of DNA staining in oregano oiltreated skin (**Figures 6E,F**), while the staining was readily seen in the positive control in which skin section was treated with DNase I (**Figure 6G**). The absence of any oregano oil-induced skin reaction and DNA damage after a 3-day treatment suggests that topical application of oregano oil is neither cytotoxic nor genotoxic to the host.

## DISCUSSION

In seeking non-antibiotic microbicides, we have screened dozens of EOs from Chinese indigenous aromatic plants/spices because EOs have been long recognized as one of the most promising natural products for safe microbicides in folk medicines (Lu et al., 2013a,b,c). The selection was initially based on their antiseptic applications in the food industry and in agricultures after an extensive database search. Among the dozen EOs tested, about one third showed significant antibacterial activities against clinically and agricultural important microbes (Lu et al., 2013a,b,c). Oregano oil stood out as one of the best ones in terms of safety and efficacy. We thus detailed the bactericidal activity of oregano oil against 11 MDR clinical isolates of P. aeruginosa, A. baumannii and MRSA as well as two bioluminescent strains of P. aeruginosa PA01 and MRSA USA300 in the current study. Oregano oil effectively killed all the bacterial strains tested, with the MICs ranging from 0.08 to 0.64 mg/ml and at an order of sensitivity of A. baumannii > MRSA > P. aeruginosa (**Table 1**). The finding is in agreement with previous studies demonstrating that oregano oil and its main component carvacrol had a higher MIC against P. aeruginosa

FIGURE 4 | Oregano oil treatment of PA01 infections in the burn wounds. (A,B) Gram-stained longitudinal section (A) and crossing section (B) of a representative wound showing the presence of PA01 biofilms outlined in red. The skin sample was harvested 24 h after bacterial inoculation. (C,D) Successive bacterial luminescence images of representative wounds infected with 5 × 10<sup>6</sup> CFU of luminescent PA01 with (D) and without (C) oregano oil at 10 mg/ml. The oregano oil was topically applied onto the wounds at 24 h after bacterial inoculation. (E) A dose response of mean bacterial luminescence of the wounds infected with 5 × 10<sup>6</sup> CFU of PA01 in the presence or absence of oregano oil treatment at 5 or 10 mg/ml. (F) Time courses of mean bacterial luminescence of the infected wounds in the presence or absence of oregano oil treatment at 5 or 10 mg/ml from days 2 to 7. (G) Mean areas under the bacterial luminescence curves (F), representing the overall bacterial burden of infected wounds. (H). The wounds were treated with grape seed oil (control) or oregano oil 24 h after infection and bacterial CFU were quantified on day 7 after bacterial inoculation. RLU, relative luminescence units; A.U., arbitrary units. The data represent means ± SDs (n = 8). ∗∗p < 0.01, ### or ∗∗∗p < 0.001 and #### or ∗∗∗∗p < 0.0001 in the presence vs. absence of oregano oil. ns, no significance.

compared to other species, such as S. spp. (Nostro et al., 2007), Chromobacterium violaceum, Salmonella typhimurium, and S. aureus (Burt et al., 2014). Similar to the clinical isolates, oregano oil also inactivated standard strains of A. baumannii ATCC 19606 (Rosato et al., 2010), P. aeruginosa ATCC 27853 and S. aureus ATCC 29213 (Bouhdid et al., 2009), with the MICs 0.15 mg/mL, 1 mg/mL, and 0.33 mg/mL, respectively, which are comparable to our investigation. Previous studies suggested that Gram-negative bacteria appeared to be more resistant than Gram-positive bacteria in response to EO (Tepe et al., 2005; Longaray Delamare et al., 2007; Gilles et al., 2010). This relative resistance of Gram-negative over Gram-positive bacteria may be ascribed to their cell wall structure and outer membrane arrangement. The outer membrane of Gram-negative bacteria is rich in lipopolysaccharide molecules, relatively impermeable to lipophilic compounds, thereby presenting a barrier to penetration of EO antimicrobial substances (Gao et al., 2011). It may be also associated with the enzymes in the periplasmic space, which are capable of breaking down the antimicrobial substances upon their entrance of the cells (Nikaido, 1996). However, our studies disagreed with these observations and found Gram-negative A. baumannii was more sensitive to oregano oil than Grampositive MRSA. This different outcome suggests that antibacterial activity of oregano oil may not depend on the type of Gram reaction in contrast to other EOs, a possibility that is supported by the studies of Gao et al. (2011). In their studies, Gram-negative bacteria Klebsiella pneumoniae was the most sensitive bacteria whereas the Gram-positive bacteria Listeria monocytogenes was the most resistant strain to the Sphallerocarpus gracilis seed EO (Gao et al., 2011).

The possibility that the cell wall and membrane were primary targets of oregano oil was supported by TEM imaging of the ultrastructure of the bacteria. We found damages of the cell wall and membranes, occurrent with cytoplasmic vacuoles, stainless-vesicles, and disruption and discontinuation of the intracellular structures in a large number of bacterial cells after oregano oil treatment (**Figures 2B,E**). This finding is consistent with an association of the antibacterial activity of oregano oil/carvacrol with disturbance of membrane embedded proteins and disruption of lipids, RNA synthesis, ATPase activity, and efflux pump previously demonstrated (Simoes et al., 2009; Tapia-Rodriguez et al., 2017). Moreover, oregano oil may cause an imbalance in intracellular osmotic pressure owing to a leakage of cytoplasmic contents following cell wall and membrane damages, and formation of cytoplasmic vacuoles, eventually inducing cell necrosis, although more investigations are required to conclude the mechanism in detail.

also TUNEL stained (E,F). DNase I treated skin samples (G) were TUNEL stained in parallel as positive-staining controls.

Biofilms are sessile organizations of bacterial cells with a strong adherence to surfaces. Biofilm-associated microbial cells are well protected by an extracellular matrix that comprises exopolysaccharides, proteins and DNA and is poorly permeable (Donlan, 2002; Husain et al., 2015). Systemic antibiotics administered to treat bacterial infections frequently fails at least in part due to the poor permeability of biofilms. Interestingly, oregano oil was capable of biofilm-killing at least at an early stage (24-h-old biofilms) as efficiently as planktonic cells. Biofilms of A. baumannii, P. aeruginosa, and MRSA were eliminated by oregano oil at a concentration of 0.3, 1.0, or 0.4 mg/ml, respectively, similar to the corresponding MICs attained in planktonic cells. The similarity can be extended to the order of sensitivity with P. aeruginosa biofilms more resistant than MRSA biofilms than A. baumannii biofilms (**Figure 1** and **Table 1**). This may be attributed to the superior permeability and lipid solubility of oregano oil to bacterial cell membrane and wall (Magi et al., 2015; Khan et al., 2017). Likewise, the effectiveness of oregano oil to inactivate S. aureus, S. epidermidis, and P. aeruginosa biofilms was also found at similar MICs as those against planktonic cells (Nostro et al., 2007; dos Santos Rodrigues et al., 2017; Tapia-Rodriguez et al., 2017). SEM observations confirmed the physical damage and considerable morphological alteration in the P. aeruginosa IQ0042 (**Figures 2E,F**) and MRSA IQ0064 (**Figures 2G,H**) biofilms following oregano oil treatment. These observations raise an intriguing possibility that EO may have advantages over water soluble antibiotics in treatment of biofilms because bacteria living in the biofilms are well known to be more

resistant to antibiotics (up to 1,000 times) than their planktonic counterparts, in part owing to poor permeability of biofilms to the antibiotics (Ceri et al., 1999; Caraher et al., 2007).

One concern of using oregano oil as an alternative for the treatment of infections in clinics will be whether MDR bacteria can develop resistance to oregano oil. Although this remains largely unaddressed to date, our results suggest that resistance may not be readily developed because 20 passages in the presence of sublethal concentrations of oregano oil did not alter their susceptibility to the oil (**Figure 3A**). Moreover, oregano oil has been used in food prevention and other antiseptic application for centuries and no resistance has been reported so far. It is commonly believed that EOs act at multiple sites within bacterial cells (cell membrane, cell wall, structural proteins, enzymes, nucleic acids, unsaturated lipids, etc.) and would be less likely to induce the development of resistance (Burt, 2004; Simoes et al., 2009; Tapia-Rodriguez et al., 2017). On the contrary, the MIC of conventional antibiotics could gradually increase with a treatment length due to their single action to inactivate the bacteria (Baym et al., 2016; Levin-Reisman et al., 2017).

The bactericidal activity of oregano oil was corroborated in mouse burn models using model bioluminescent strains of Gram-negative P. aeruginosa PA01 and Gram-positive MRSA USA300. When applied at 24 h after bacterial inoculation forming early stage biofilms, oregano oil effectively reduced the bacterial burden by 25-folds for PA01 and 49-folds for USA300, respectively, in comparison to untreated wounds. While efficiently inactivating bacteria, oregano oil exhibited no cytotoxicity or genotoxicity to the skin, in good agreement with its long record of safety. Moreover, oregano oil did not adversely affect human keratinocytes (Babili et al., 2011) and was safe when administered orally in mice (Manohar et al., 2001; Preuss et al., 2005; Feng et al., 2017).

## REFERENCES


In summary, we reported here the effectiveness of oregano oil against a panel of MDR bacteria isolated from combating casualties and demonstrated for the first time efficacy of oregano oil for the treatment of burn infections in mice. The study serves as an initial effort in the pursuit of a novel therapeutic option for wound infections, especially those caused by MDR bacteria.

### AUTHOR CONTRIBUTIONS

ML designed and performed all the experiments, analyzed the data, and wrote the manuscript. TD supervised and designed the experiments, analyzed the data, and wrote the paper. CM isolated and characterized all the clinical bacteria and wrote the paper. and MW designed and supervised the study, analyzed the data, and wrote the manuscript.

## FUNDING

This study was supported in part by FA9550-16-1-00173, Department of Defense/Air Force Office of Scientific Research Military Photomedicine Program and Department funds to MW and TD.

### ACKNOWLEDGMENTS

We thank Drs. Michael R. Hamblin and Ji Wang for stimulating comments and discussions from Massachusetts General Hospital and the staff at the photopathology core at Wellman Center for Photomedicine for assisting transmission electron microscopy and histopathology.



baumannii infection in a mouse burn model: implications for prophylaxis and treatment of combat-related wound infections. J. Infect. Dis. 209, 1963–1971. doi: 10.1093/infdis/jit842

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Lu, Dai, Murray and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effects of Monolaurin on Oral Microbe–Host Transcriptome and Metabolome

Viviam de Oliveira Silva1,2† , Luciano José Pereira<sup>3</sup> , Silvana Pasetto<sup>2</sup> , Maike Paulino da Silva<sup>2</sup> , Jered Cope Meyers<sup>4</sup> and Ramiro Mendonça Murata4,5 \*

<sup>1</sup> Department of Veterinary Medicine, Federal University of Lavras, Lavras, Brazil, <sup>2</sup> Division of Periodontology, Diagnostic Sciences, Dental Hygiene and Biomedical Science, Herman Ostrow School of Dentistry, University of Southern California, Los Angeles, CA, United States, <sup>3</sup> Department of Health Sciences, Federal University of Lavras, Lavras, Brazil, <sup>4</sup> Department Foundational Sciences, School of Dental Medicine, East Carolina University, Greenville, NC, United States, <sup>5</sup> Department of Microbiology and Immunology, Brody School of Medicine, East Carolina University, Greenville, NC, United States

### Edited by:

Rebecca Thombre, Pune University-Shivajinagar, India

### Reviewed by:

Piyush Baindara, National Centre for Biological Sciences, India Karishma R. Pardesi, Savitribai Phule Pune University, India

> \*Correspondence: Ramiro Mendonça Murata murartar16@ecu.edu

### †Present address:

Viviam de Oliveira Silva, Department of Research and Scientific Initiation, Centro Universitário Atenas, Paracatu, Brazil

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 28 May 2018 Accepted: 16 October 2018 Published: 06 November 2018

### Citation:

Silva VO, Pereira LJ, Pasetto S, Silva MP, Meyers JC and Murata RM (2018) Effects of Monolaurin on Oral Microbe–Host Transcriptome and Metabolome. Front. Microbiol. 9:2638. doi: 10.3389/fmicb.2018.02638 The aim of this in vitro study was to evaluate the effects of monolaurin against Aggregatibacter actinomycetemcomitans (Aa) and determine their effects on the host transcriptome and metabolome, using an oral cell/bacteria co-culture dual-chamber model to mimic the human periodontium. For this, the Aa, was applied to cross the monolayer of epithelial keratinocytes (OBA-9) to reach the fibroblasts layer (HGF-1) in the basal chamber. The Monolaurin treatments (25 or 50 µM) were added immediately after the inoculation of the dual-chamber with Aa. After 24 h, the transcriptional factors and metabolites produced were quantified in the remaining cell layers (insert and basal chamber) and in supernatant released from the cells. The genes IL-1α, IL-6, IL-18, and TNF analyzed in HGF-1 concentrations showed a decreased expression when treated with both concentration of Monolaurin. In keratinocytes, the genes IL-6, IL-18, and TNF presented a higher expression and the expression of IL-1α decreased when treated with the two cited concentrations. The production of glycerol and pyruvic acid increased, and the 2-deoxytetronic acid NIST, 4-aminobutyric acid, pinitol and glyceric acid, presented lower concentrations because of the treatment with 25 and/or 50 µM of Monolaurin. Use of monolaurin modulated the immune response and metabolite production when administered for 24 h in a dual-chamber model inoculated with A. actinomycetemcomitans. In summary, this study indicates that monolaurin had antimicrobial activity and modulated the host immune response and metabolite production when administered for 24 h in a dual-chamber model inoculated with A. actinomycetemcomitans.

Keywords: Aggregatibacter actinomycetemcomitans, periodontal disease, immune system, fibroblast, keratinocyte, monolaurin

## INTRODUCTION

Periodontal disease is characterized by bacterial infection associated with the presence of biofilm, resulting in chronic inflammation of the tooth supporting tissues leading to a progressive destruction of periodontal tissue and alveolar bone. Signs and symptoms of periodontitis can include gingival swelling, bleeding during brushing, periodontal pockets, increased spacing

**243**

between the teeth and tooth loss (Irfan et al., 2001; Michaud et al., 2007; Savage et al., 2009; Gurav, 2014).

The most common form of periodontal disease, denominated periodontitis, affects approximately 50% of the adult population with its severe forms affecting 10–15% of these individuals (Chapple and Genco, 2013).

The oral biofilm is composed of at least 800 different species of bacteria (Paster et al., 2006; Filoche et al., 2010). Numerous periodontopathogenic bacteria (Porphyromonas gingivalis, Fusobacterium nucleatum, Streptococcus intermedius, and others) are responsible for the initiation and maintenance of periodontal inflammation (Coussens and Werb, 2002; Oringer, 2002; Loos, 2005; Moutsopoulos and Madianos, 2006; de Almeida et al., 2007). The Aggregatibacter actinomycetemcomitans is a fermentative, gram-negative and capnophilic coccobacillus. It is not only considered the most important agent of localized aggressive periodontitis lesions, but is associated with chronic periodontitis as well (Slots et al., 1980; Zambon et al., 1983; Haraszthy et al., 2000; Gafan et al., 2004; Yang et al., 2004; Cortelli et al., 2009; Rakic et al., 2010 ´ ; da Silva-Boghossian et al., 2011). The progression of periodontal disease is associated with the virulence factors of the microorganism, together with the vulnerability of the host. Aa produces leukotoxin, adhesins, bacteriocins, and lipopolysaccharide, which are responsible for interacting with the host cells triggering an inflammatory response in the periodontium (Cortelli et al., 2009). In addition, another critical factor for soft tissue inflammation and bone resorption is the immuneinflammatory host response to the bacterial biofilm (Salvi and Lang, 2005).

According to Aimetti (2014), non-surgical periodontal treatments possess limitations, such as the long-term maintainability of deep periodontal pockets, risk of disease recurrence, and skill of the operator. The development of new therapeutic agents that have the ability to inhibit the biofilm formation and modulate the inflammatory response can have a major impact in the prevention and treatment of periodontal disease. Monolaurin, also known as glycerol monolaurate or lauroyl, is a natural surfactant compound commonly used in cosmetics and the food industry. It is recognized by the Food and Drug Administration (FDA) as a GRAS (Generally Recognized as Safe) food additive since 1977. This natural compound is found in coconut oil and human breast milk (Fu et al., 2006; Peterson and Schlievert, 2006).

Monolaurin presents antibacterial and antiviral activity in vitro (Projan et al., 1994; Isaacs, 2001; Preuss et al., 2005; Peterson and Schlievert, 2006; Carpo et al., 2007), which may be of great interest in the treatment and/or prevention of various infections. However, the effect of monolaurin on periodontopathogenic bacteria has yet to be determined.

Therefore, the aims of this in vitro study were to evaluate the cytotoxicity, the antimicrobial effect of monolaurin against A. actinomycetemcomitans and to determine its effect on the host transcriptome and metabolome using a gum cell – bacteria co-culture model.

## MATERIALS AND METHODS

## Monolaurin

For carrying out the experiments, the monolaurin utilized was 1-lauroyl-rac-glycerol (Sigma-Aldrich, St. Louis, MO, United States). For dilution, sterilized deionized water was used as the vehicle.

## Cells and Bacterial Strain

Human gingival fibroblast – HGF-1 (ATCC <sup>R</sup> CRL-2014) were cultured in Dulbecco's Modified Eagle's Medium (DMEM) (Lonza, Walkersville, MD, United States) with 10% Fetal Bovine Serum (FBS) (Lonza, Walkersville, MD, United States). Human gingival epithelial cells (keratinocyte OBA-9) were cultured in specific medium for keratinocytes, Defined Keratinocyte-SFM (Life Technologies, Carlsbad, CA, United States) (Oda and Watson, 1990; Kusumoto et al., 2004).

Aggregatibacter actinomycetemcomitans (strain D7S-1) was cultivated from a subgingival plaque from an African-American female patient diagnosed with generalized aggressive periodontitis (Chen et al., 2010). The bacteria were grown in Trypticase Soy Broth (TSB) (Becton Dickinson, Franklin Lakes, NJ, United States).

## Antimicrobial Activity

Monolaurin antimicrobial activity was evaluated in A. actinomycetemcomitans after 24 h of treatment. Microorganisms were inoculated [1 × 10<sup>6</sup> colony-forming unit per milliliter (CFU/mL) in a 96-well microtiter plate with TSB (Becton Dickinson, Franklin Lakes, NJ, United States)] and monolaurin was immediately added in various concentrations (0.5–1,000 µM) to determine the minimum inhibitory concentration (MIC) (Branco-de-Almeida et al., 2011). Microplates were maintained in a humidified incubator at 37◦C and 5% CO2. After incubation, bacterial growth was assayed by measurement of absorbance at 660 nm. MIC was defined as the lowest concentration of monolaurin that had restricted growth to a level, 0.05 at 660 nm (no visible growth).

## Cytotoxicity Assay

In vitro cytotoxicity effect was measured by the fluorometric resazurin method (O'Brien et al., 2000). OBA-9 or HGF-1 cells were seeded (1 × 10<sup>5</sup> cells/mL) in 96-well microtiter plates and maintained in a humidified incubator at 37◦C and 5% CO2. After 24 h, cell morphology was observed under an inverted microscope (EVOS FL Life Technologies, Carlsbad, CA, United States) to confirm that they had adhered to the bottom of each well and were presenting proper morphology. The monolaurin (0.5–1,000 µM) was added to the cell cultures and incubated at 37◦C and 5% CO2. After 24 h, the medium was discarded, cells were washed with room temperature phosphate buffered saline (PBS) (Lonza, Walkersville, MD, United States), and fresh, room temperature medium was added with resazurin (Cell Titer Blue Viability Assay – Promega Corp., Madison, WI, United States). Subsequently, the plate was incubated at 37◦C and 5% CO2. After 4 h, the contents of the wells were transferred to a new microplate and the fluorescence was read in a microplate reader (SpectraMax M5, Molecular Devices, Sunnyvale, CA, United States) with excitation 550 nm, emission 585 nm, and cut off 570 nm (Pasetto et al., 2014). To calculate the results and obtain the LD<sup>50</sup> (Lethal Dose), we performed a non-linear regression the program MasterPlex ReaderFit (Hitachi Solutions America).

## Dual-Chamber Model

fmicb-09-02638 November 8, 2018 Time: 15:43 # 3

A dual-chamber, oral cell/bacteria co-culture was used to investigate the immunological effects of monolaurin (Silva et al., 2017). HGF-1 cells (1 × 10<sup>5</sup> ) were seed in the basal chamber of a 24-well plate. Transwell inserts (8 µm pore × 0.3 cm<sup>2</sup> of culture surface – Greiner Bio-One, Monroe, NC, United States) were positioned in each well and OBA-9 cells (1 × 10<sup>5</sup> ) were seeded into the inserts. DMEM (Lonza, Walkersville, MD, United States) with 10% FBS (Lonza, Walkersville, MD, United States) was used for the medium. The plates were incubated at 37◦C in humid air containing 5% CO<sup>2</sup> for 24 h (Silva et al., 2017). Trans Epithelial Electric Resistance (TEER) of each cell layer was measured with a Millicell-ERS Volt-Ohm Meter (Millipore, Bedford, MA, United States). Cell layer confluence in the Transwell insert was measured daily until reaching optimal TEER (>150 Ohm/cm<sup>2</sup> ) after 48 h. Once reaching optimal TEER, the media of the basal chamber and the insert were replaced with new media containing A. actinomycetemcomitans (1 × 10<sup>6</sup> CFU/mL). Medium containing the microorganism was added to the insert, passing through the upper layer of cells (OBA-9) and reaching the bottom cell layer (HGF-1). Immediately after the inoculation of oral cell/bacteria co-culture with A. actinomycetemcomitans the monolaurin treatments (25 and 50 µM) were added and the plate was incubated at 37◦C in humid air containing 5% CO2. The time of exposure of the microorganism to monolaurin was 24 h. Each experiment was repeated three times with two replicates per group (n = 6). The experiment was divided in three groups: (1) Control – oral cell/bacteria co-culture with Aa inoculated and no treatment; (2) Mono 25 – oral cell/bacteria co-culture with Aa inoculated and treated with 25 µM of monolaurin; (3) Mono 50 – oral cell/bacteria co-culture with Aa inoculated and treated with 50 µM of monolaurin.

### Collecting Samples for Analysis

After the treatment period, liquid contents from each well was collected and centrifuged at 1,200 rpm for 10 min. Following centrifugation, supernatant was divided into two aliquots for enzyme-linked immunosorbent assay (ELISA) and metabolome analysis. RNA was isolated from the remaining cell layers of HGF-1 and OBA-9 (surface of the wells and inserts) for gene analysis in quantitative real-time PCR.

### ELISA Assay

Inflammatory cytokines were determined using a commercial ELISA (Multi-Analyte ELISArray – Qiagen, Valencia, CA, United States). The cytokines were measured by standard ELISA protocol using a panel of 12 inflammatory cytokines: interleukin-1 alpha (IL-1α), interleukin-1 beta (IL-1β), interleukin-2 (IL-2), interleukin-4 (IL-4), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-10 (IL-10), interleukin-12 (IL-12), interleukin-17 alpha (IL-17α), interferon gamma (IFNγ), tumor necrosis factor alpha (TNFα), Granulocyte-Macrophage Colony Stimulating Factor (GM-CSF). Cytokines that presented the greatest change in concentration were selected for detailed analysis. These cytokines were analyzed using a standard ELISA protocol (Single Analyte ELISA Kit – Qiagen, Valencia, CA, United States).

### Gene Expression – Quantitative Real-Time PCR

Total RNA was extracted through a commercial RNA purification kit (Mini Kit Qiagen RNeasy Protocol – Qiagen, Valencia, CA, United States) and the purity and quantity were measured in a NanoPhotometer P360 (Implen, Westlake Village, CA, United States). Total RNA was converted into singlestranded cDNA using a high-capacity reverse transcription kit (QuantiTec Reverse Transcription Kit – Qiagen, Valencia, CA, United States). An array for evaluation of gene expression of the inflammatory response was completed by quantitative real-time PCR (Prime PCR Pathway Plate/Acute Inflammation Response – Bio-Rad, Hercules, CA, United States) from the cDNA obtained. Based on these results, six genes/primers were selected for detailed study: interleukin 1 alpha (IL-1α), interleukin 6 (IL-6), interleukin 18 (IL-18), caspase 3 (CASP3), matrix metallopeptidase 1 (MMP-1), and tumor necrosis factor (TNF) (QuantiTect Primer Assay – Qiagen, Valencia, CA, United States). QuantiTect SYBR Green PCR Kits (Qiagen, Valencia, CA, United States) were used to determine the gene expression of the selected primers. The reaction product was quantified by relative quantification using Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the reference gene. Data were analyzed as fold-change compared to control. The cycle threshold was calculated and interpreted using CFX Connect (Bio-Rad, Hercules, CA, United States) (Pfaffl, 2001).

### Metabolome Analysis

Samples for metabolome analysis were treated as described by Silva et al. (2017) and analyzed at the West Coast Metabolomics Center (UC Davis Genome Center, Davis, CA, United States) by means of gas chromatography/mass spectrometry (Agilent 6890, Santa Clara, CA/Leco Pegasus IV, St. Joseph, MI, United States). Then, the metabolites were compared with the standard library for proper identification (Fiehn and Kind, 2007).

## Statistical Analysis

Data were expressed as mean ± standard deviation. The comparison between the groups was made through analysis of variance (ANOVA) and when a difference was determined, the Bonferroni test was applied. Data were analyzed using GraphPad Prism (version 5.01, GraphPad Software, San Diego, United States). The level of significance was set at p < 0.05.

## RESULTS

fmicb-09-02638 November 8, 2018 Time: 15:43 # 4

## Antimicrobial Activity and Cytotoxicity Assay

In order to determine antimicrobial activity of monolaurin, a range of 0.5–1,000 µM was tested. Antibacterial activity was detected at 25–50 µM. The minimum inhibitory concentration (MIC) was established between 25 and 50 µM. The cytotoxicity assays were conducted in HGF-1 and OBA-9 cells and the results are shown in **Figure 1**. For HGF-1 the LD<sup>50</sup> was 146 µM of monolaurin and for OBA-9 the LD<sup>50</sup> was 69 µM of monolaurin.

## ELISA

Enzyme-linked immunosorbent assay (ELISA) results are presented in **Figure 2**. Both treatments (25 and 50 µM) of monolaurin increased the concentrations of GM-CSF and IL-1α. The concentrations of IL-6 and IL-10 increased only with the 50 µM treatment. On the other hand, IL-1β decreased as a function of monolaurin.

FIGURE 1 | Cytotoxicity assay of cells treated with different doses of monolaurin in an oral cell/bacteria co-culture inoculated with A. actinomycetemcomitans: (A) HGF-1 cell (fibroblasts); (B) OBA-9 cell (keratinocytes). Cell viability was presented in percentage (%) ± standard deviation. Control group = 100% viability, n = 6. Non-treated cells were considered as 100% viability; DMSO was used as positive control to demonstrate an appropriate system-cell response (100% cytotoxicity – data not shown).

inoculated with A. actinomycetemcomitans and treated with 25 µM of monolaurin (Mono 25) or 50 µM of monolaurin (Mono 50) compared with the control group. Control group is oral cell co-culture inoculated with A. actinomycetemcomitans and no treatment. The control group mean is expressed as 100% and treated groups have their mean relative to the control group. Mean ± standard deviation; n = 4.

## Gene Expression – Quantitative Real-Time PCR

Quantitative real-time PCR results are presented in **Figures 3**, **4**. The genes IL-1α, IL-6, IL-18, and TNF analyzed in HGF-1 showed a decreased expression (p < 0.05) when treated with 25 or 50 µM of monolaurin compared with group Control (**Figures 3A–C,F**). CASP3 and MMP-1 genes showed no difference (**Figures 3D,E**).

Keratinocytes presented a higher IL-6, IL-18, and TNF gene expression (p < 0.05) when treated with 25 or 50 µM of monolaurin compared with the control group (**Figures 4B,C,F**). IL-1α gene decreased expression (p < 0.05) in the group treated with both concentrations (**Figure 4A**). CASP-3 and MMP-1 genes showed no difference (**Figures 4D,E**).

### Metabolome

The metabolome study returned a total of 283 metabolites. Of these, 120 were identified and only 6 presented a statistical difference as a function of monolaurin treatment. Some metabolites presented significantly altered concentrations (**Figure 5**). It was observed that both treatments of monolaurin increased the production of glycerol (**Figure 5D**) and pyruvic acid (**Figure 5F**) when compared to control. On the other hand, the 2-deoxytetronic acid NIST (**Figure 5A**) showed the opposite result, lower concentrations of this metabolite, in function of the treatment with 25 and 50 µM of monolaurin. The 50 µM monolaurin treatment group presented lower concentrations of 4-aminobutyric acid (**Figure 5B**), glyceric acid (**Figure 5C**), and pinitol (**Figure 5E**), against the control group.

## DISCUSSION

Periodontal disease is known to be a widespread oral health problem across the world. Several methods for prevention and treatment of oral diseases have been used over time and new

technologies are being developed. However, little attention has been given to the understanding of how periodontal structures are affected and how these structures react to different types of treatments (Balloni et al., 2016). Currently, some mouthwashes and toothpastes have been used as topical antimicrobial agents (Barnett, 2003). However, monolaurin is not listed as an active ingredient in commercially available mouthwashes. Monolaurin has been shown to possess diverse biological activities, such as antibacterial and antiviral properties (Preuss et al., 2005; Li et al., 2009; Tangwatcharin and Khopaibool, 2012; Seleem et al., 2016). In our in vitro study, susceptibility assays of monolaurin showed inhibition of microorganism grow at 25– 50 µM with minimal cytotoxicity. The putative pathways by which monolaurin affects Aa may involve: (i) increased membrane permeability and cell lysis, (ii) disruption of electron transport chain and uncoupling oxidative phosphorylation, and (iii) inhibition of membrane enzymatic activities and nutrient uptake (Yoon et al., 2018). In addition, co-culture supernatant were assessed for expression of host pro-inflammatory cytokines, gene expression of inflammatory cytokines and metabolomic profile in human cells infected with A. actinomycetemcomitans.

The gingival tissue comprises the oral epithelium and superficial underlying connective tissue. These tissues are sites of initiation for inflammatory processes and the first to be affected by biofilm (Balloni et al., 2016). Gingival epithelial cells represent a physical barrier against bacteria and are involved in processes of innate immunity (Jang et al., 2015). Keratinocytes and fibroblasts represent the major cell types present in gingival epithelial tissues. They interact directly with pathogenic microorganisms and are able to express a variety of cytokines and chemokines such as IL-1α, IL-6, IL-8, TNF, among others (Bodet et al., 2007; Groeger and Meyle, 2015). In the present study, there was a decrease in the gene expression of IL-1α, IL-6, and IL-18 and TNF in HGF-1 and an increase in the expression of IL-6, IL-18, and TNF in OBA-9 cells treated with monolaurin indicating a positive modulation in the inflammatory response of these cells.

Interleukin-1 alpha is a polypeptide that plays several roles in tissue homeostasis, immunity and has a proinflammatory nature. It is produced mainly by macrophages and monocytes, but also epithelial cells and fibroblasts (Champagne et al., 2003; Balloni et al., 2016). IL-1α plays a critical role in protecting the body against invaders, such as bacteria and viruses, and is involved in extracellular matrix and bone metabolism (bone resorption, disintegration, and removal of bone tissue that is no longer needed). Therefore, IL-1α is considered a marker of periodontal disease (Govindarajan et al., 2015; Groeger and Meyle, 2015).

Here, we observed a decrease in the expression of IL-1α by HGF-1 and OBA-9 cells in the monolaurin-treated groups which may be related to a beneficial inflammatory response since, elevated levels of IL-1α and IL-1β in the gingival crevicular fluid are commonly found in patients with periodontal disease. This increased production of IL-1 is associated with a higher inflammatory response to the bacterial stimulus, resulting in more severe disease as well as an unfavorable response to treatments (Kornman et al., 1997; Engebretson et al., 1999; McGuire and Nunn, 1999; Shirodaria et al., 2000).

Interleukin-6 is associated with acute phase reactions and is usually secreted concomitantly with other proinflammatory cytokines. Some authors relate this cytokine to the destruction of periodontal tissue by favoring the accumulation of inflammatory cells and by activating and releasing inflammatory mediators that in turn accentuate the response to periodontal disease (de Lima Oliveira et al., 2012; Zhang et al., 2016).

The kinetics of production of IL-18 against the action of bacteria is rapid, according to its proinflammatory nature. It plays a central role in systemic and local inflammation (Biet et al., 2002). One study evaluated IL-18 levels in crevicular gingival fluid in four groups of patients: healthy periodontium, gingivitis, chronic periodontitis, and aggressive periodontitis. The results demonstrated that IL-18 concentration in the crevicular gingival fluid was low in the group with healthy periodontium and progressively increased in groups from gingivitis, to aggressive periodontitis, and finally chronic periodontitis. As the inflammation increased, there was a concomitant increase in IL-18 level (Nair et al., 2016). In the current study, treatment with

monolaurin decreased the IL-18 gene expression in fibroblasts, however, an opposite result was observed in keratinocytes, where IL-18 expression was increased in the group treated with 50 µM monolaurin.

Tumor necrosis factor plays a central role in periodontitis, being associated with the disease because of its ability to induce destruction of connective tissue and its effects on bone resorption through the activation of osteoclasts. Thus, TNF is one of the first proinflammatory cytokines produced in response to infection by pathogenic bacteria (Stashenko et al., 1987; de Lima Oliveira et al., 2012; Zhang et al., 2016). Here, the authors observed a reduction in the gene expression of TNF in monolaurin treated fibroblasts. On the other hand, in keratinocytes, TNF expression was higher in cells treated with 25 µM of this compound. Animal models challenged with P. gingivalis were treated with antimicrobial drugs (chlorhexidine, minocycline, and doxycycline) in order to evaluate its effect on the inflammatory response. Minocycline induced higher levels of TNF while chlorhexidine reduced TNF levels, this result led the authors to believe that these agents modified the inflammatory response to P. gingivalis regardless of its antimicrobial effect, each in its own way (Houri-Haddad et al., 2008). Overall, monolaurin modulate the expression of inflammatory cytokines, suggesting it may have a modulatory role on the host pro-inflammatory response to help eradicate the bacterial infection. Similarly, previous studies suggested that monolaurin play a significant role on T cell functions and signaling by altering TCR-induced LAT, PLC-γ, and AKT cluster formation PI3K-AKT signaling axis, and calcium influx which ultimately decrease the cytokine production (Zhang et al., 2016).

The knowledge of the metabolic profile can contribute to understanding of the periodontal disease course as well as identify possible metabolites as biomarkers. This study evaluated the effects caused by monolaurin on the cellular metabolome and found significant variations. Relative quantification revealed lower levels of 2-deoxytetronic acid NIST, 4-aminobutyric acid, glyceric acid and pinitol, and higher levels of glycerol and pyruvic acid in cells treated with 25 or 50 µM monolaurin. Analysis of fibroblast metabolome (HGF) treated with IL-1β in combination with titanium dioxide nanoparticles (TiO<sup>2</sup> NPs), demonstrated that IL-1β induction reduced concentrations of primary metabolites, especially those of urea cycle, polyamine, S-adenosylmethionine and glutathione synthetic pathways. The addition of TiO<sup>2</sup> NPs further increased these metabolic changes induced by IL-1β. These findings may be useful for the future establishment of new metabolic markers and therapeutic strategy for gingival inflammations (Garcia-Contreras et al., 2015). One study compared the salivary metabolic profile of healthy patients and patients with periodontal disease. Changes in several classes of metabolites have been observed in individuals with periodontal

### REFERENCES


disease. According to the authors, such changes reflected an increase in host–bacterial interactions in the diseased state (Barnes et al., 2011).

## CONCLUSION

In summary, this study indicates that monolaurin possesses antimicrobial activity and modulates the host immune response and modulates production when administered for 24 h in a dual-chamber model inoculated with A. actinomycetemcomitans. These findings demonstrate that monolaurin could be considered a potential candidate for in vivo studies, which may translate into its potential clinical use to treat pathogenic microbe–host interactions.

## AUTHOR CONTRIBUTIONS

VS conceived and designed the experiments, performed the experiments, analyzed the data, performed the statistical analysis, and wrote the paper. LP conceived and designed the experiments, contributed to reagents, materials, and analysis tools, analyzed and critically reviewed the data. MS and SP performed the experiments and analyzed the data. JM analyzed the data, and critically reviewed and interpreted the data. RM conceived and designed the experiments, contributed to reagents, materials, and analysis tools, performed the experiments, analyzed and critically reviewed the data. All authors contributed to manuscript revision, read, and approved the submitted version.

### ACKNOWLEDGMENTS

The authors thank to CAPES (Coordination of Improvement of Higher Education Personnel), FAPEMIG (Research Sponsoring Agency of the State of Minas Gerais), and CNPq (National Counsel of Technological and Scientific Development) for support. Research reported in this publication was also supported by the NIH/NCCIH under award number AT006507. The content is solely the responsibility of the authors and does not represent the official views of the National Institutes of Health.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.02638/full#supplementary-material

gingival cell behaviour. Toxicol. Vitr. 34, 88–96. doi: 10.1016/j.tiv.2016. 03.015

Barnes, V. M., Ciancio, S. G., Shibly, O., Xu, T., Devizio, W., Trivedi, H. M., et al. (2011). Metabolomics reveals elevated macromolecular degradation in periodontal disease. J. Dent. Res. 90, 1293–1297. doi: 10.1177/ 0022034511416240



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Silva, Pereira, Pasetto, da Silva, Meyers and Murata. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Purification and Characterization of an Active Principle, Lawsone, Responsible for the Plasmid Curing Activity of Plumbago zeylanica Root Extracts

Rajashree Bhalchandra Patwardhan<sup>1</sup> \*, Prashant Kamalakar Dhakephalkar<sup>2</sup> \*, Balu Ananda Chopade<sup>3</sup> , Dilip D. Dhavale<sup>4</sup> and Ramesh R. Bhonde<sup>5</sup>

### Edited by:

Rebecca Thombre, Pune University-Shivajinagar, India

### Reviewed by:

Sunil D. Saroj, Symbiosis International University, India Kapil Punjabi, Indian Institute of Technology Bombay, India

### \*Correspondence:

Rajashree Bhalchandra Patwardhan dr.rbpatwardhan@gmail.com Prashant Kamalakar Dhakephalkar pkdhakephalkar@aripune.org

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 31 March 2018 Accepted: 12 October 2018 Published: 08 November 2018

### Citation:

Patwardhan RB, Dhakephalkar PK, Chopade BA, Dhavale DD and Bhonde RR (2018) Purification and Characterization of an Active Principle, Lawsone, Responsible for the Plasmid Curing Activity of Plumbago zeylanica Root Extracts. Front. Microbiol. 9:2618. doi: 10.3389/fmicb.2018.02618 <sup>1</sup> Department of Microbiology, Haribhai V. Desai College of Arts, Science and Commerce, Pune University, Pune, India, <sup>2</sup> Bioenergy Group, Agharkar Research Institute, Pune, India, <sup>3</sup> Department of Microbiology, Savitribai Phule Pune University, Aurangabad, India, <sup>4</sup> Department of Chemistry, Savitribai Phule Pune University, Pune, India, <sup>5</sup> School of Regenerative Medicine – Manipal Academy of Higher Education, Bengaluru, India

Plasmid curing is the process of obviating the plasmid encoded functions such as antibiotic resistance, virulence, degradation of aromatic compounds, etc. in bacteria. Several plasmid curing agents have been reported in literature, however, no plasmid curing agent can eliminate all plasmids from different hosts. Hence, there is always a need for novel plasmid curing agents that can be effectively used for reversal of plasmid encoded functions such as virulence, antibiotic resistance, etc. In the present study, an active principle responsible for the plasmid curing activity was purified from roots of Plumbago zeylanica by bioassay guided fractionation and identified as 2-hydroxy-1,4-naphthoquinone (lawsone), on the basis of spectral and analytical data such as NMR, GCMS, FTIR. Plasmid curing activity of lawsone was observed against reference as well as wild plasmids (pBR322, pRK2013, R136, pUPI281, and pUPI282) residing in a range of hosts. Curing of plasmid was confirmed by agarose gel electrophoresis. MICs of antibiotics against A. baumannii A24 (pUPI281) and E. coli (pRK2013) decreased significantly in presence of lawsone suggesting synergy between lawsone and antibiotics. Lawsone also inhibited transfer of plasmid pRK2013 to E. coli either by transformation or conjugation. Viability assays (MTT) revealed that lawsone was not toxic to mammalian cells. Thus, the present investigation has revealed lawsone as an effective plasmid curing agent capable of suppressing development and spread of antibiotic resistance. Further, lawsone has important application in basic research to identify phenotypes encoded by the plasmids in plasmid curing experiments. To the best of our knowledge this is the first report of plasmid curing activity of lawsone isolated from roots of P. zeylanica.

Keywords: lawsone(2-hydroxy-1, 4 naphthoquinone), Plumbago zeylanica, extraction, purification, plasmid curing, antibiotic resistance

## INTRODUCTION

fmicb-09-02618 November 7, 2018 Time: 16:47 # 2

Plasmids are independent, circular, self-replicating extrachromosomal DNA elements with characteristic copy numbers within the host. Various properties encoded by plasmid include resistance to antibiotics and heavy metals, degradation of hydrocarbons, synthesis of bacteriocins and antibiotics, etc. Plasmid mediated antibiotic resistance can be transferred easily from one bacterium to another by transformation, conjugation or mobilization (Opal et al., 2000). Plasmid encoded resistance to multiple antibiotics has been increasingly recognized as a major challenge in the treatment of infections. In addition to the antibiotic resistance, some bacterial plasmids confer pathogenicity as well, to the host cell (Saunders, 1981).

Plasmid curing agents are the chemicals or physical agents that inhibit the replication of plasmid resulting in subsequent elimination of such plasmids from the host population after several replication cycles. Obtaining plasmid cured derivatives is desired in the investigation of bacterial harboring plasmids. Comparison of plasmid harboring strain and its cured derivative allows assigning phenotypic characters to genes located on plasmids. Simultaneous loss of a particular character by curing, gives a strong indication of its plasmid borne genetic character (Trevors, 1986). Further, plasmid curing converts the antibiotic resistant bacterial cells into sensitive ones (Molnar, 1988). Thus, elimination of R-plasmids makes the antibiotic therapy effective. Novel strategies to struggle antimicrobial multidrug resistance are required, and plasmid curing, and antiplasmid strategies could reduce antimicrobial resistance genes frequency and sensitize bacteria to antibiotics (Michelle et al., 2018).

Plasmid curing agents reported in the scientific literature include sodium lauryl sulfate, ethidium bromide, acridine orange, etc. (Bouanchaud et al., 1969; Chopade et al., 1994). However, it is known from earlier studies that acridine dyes and ethidium bromide cannot be used in vivo because of their mutagenicity, carcinogenicity, teratogenicity while SDS, because of its detergent action (Shriram et al., 2010). Hence, these curing agents are not useful in controlling the spread of antibiotic resistance in hospital environment. The curing agents have been effectively used to study plasmid encoded phenotypes in various laboratory studies. Till date no curing agent is known that can universally cure all plasmids from bacterial population. So non-toxic and highly effective plasmid eliminating agents are required to be developed constantly.

Plants are known to produce diverse bioactive substances of chemotherapeutic value (Chopra et al., 1992; Samy and Ignacimuthu, 2000). Ahmad et al. (2000) reported ability of aqueous extracts of Plumbago zeylanica to eliminate plasmid encoded antibiotic resistance in Escherichia coli. However, active principle responsible was not purified and identified.

In the view of this background, root extracts of P. zeylanica were explored for the presence of novel plasmid curing agent that was effective against antibiotic resistant plasmids in Escherichia coli, Salmonella Typhi as well as Acinetobacter baumannii. In the present investigation, lawsone was purified from the roots of P. zeylanica and identified as a plasmid curing agent capable of reversing multiple antibiotic resistance in broad range of clinical isolates as well as reference strains.

## MATERIALS AND METHODS

## Plant Material

P. zeylanica plants were collected from Western Ghat region of India. The roots were harvested, dried in shade, powdered and used for further extraction procedures. The plant material was authenticated by Botanical Survey of India, Ministry of Environments and forests, Government of India. A voucher specimen (RBPUP1) is deposited at the Herbarium of Botanical Survey of India, Office of Joint Director, Pune, India.

### Extraction and Purification of Plasmid Curing Agent From P. zeylanica Roots

Active principle responsible for the plasmid curing activity was purified by the bioassay guided fractionation procedure as described previously (Shriram et al., 2008). Air-dried and powdered roots (2.5 kg) of Plumbago zeylanica were successively extracted in Soxhlet apparatus with petroleum ether, cyclohexane, benzene, diethyl ether, chloroform, acetone, ethanol, and methanol (according to eluotropic series based on polarity) at boiling temperature for respective solvents (Shriram et al., 2008). Each extract was filtered and concentrated to dryness under vacuum on a rotary evaporator (Heidolph-Germany) and dissolved in 10 ml of dimethyl sulfoxide (DMSO). Isolation and purification of active compounds from P. zeylanica roots was performed by column chromatography with silica gel (100–200 mesh) (Sadasivam and Manickam, 1992). Ethanol extract showing antimicrobial and plasmid curing activity was then coarsely fractionated over silica gel using a stepwise gradient solvent system consisting of hexane: ethyl acetate (9:1, 8:2, 7:3, 6:4, 5:5, 4:5, 3:6) to separate the respective fractions (25–40 µm, 3<sup>0</sup> 50 cm, eluent hexane-ethyl acetate, flow rate 3 ml/min). TLC analysis was carried out on 0.25 mm precoated silica gel sheets (polygram sil Gluv 254). Chromatographic fractions as well as pure compounds were monitored by TLC, detected by UV light at 250 nm (UV GL-25 Mineralight lamp) and color reaction by spraying with a solution of 2% 2,4-dinitro phenyl hydrazine in methanolic sulphuric acid followed by 5 min heating at 100◦C. A total of 83 subfractions (10 ml) were collected and monitored by TLC. Subfractions 17–23 yielded (64 mg), a single pure compound detected by TLC (hexane-ethyl acetate 7:3, R<sup>f</sup> = 0.617). This compound was further purified by preparative TLC (Macherey-Nagel-Germany Precoated TLC plates SIL G-200 UV<sup>254</sup> 2 mm thickness) using hexane-ethyl acetate solvent system.

## Characterization of Purified Plasmid Curing Agent

Melting point was determined in degree Celsius (◦C) with Thomas Hoover Capillary Melting point apparatus (New Jersey,

United States). IR spectra were recorded with Perkin Elmer 1600 FTIR and Shimadzu FTIR spectrophotometer as a thin film or in nujol mull or using KBr pellets and were expressed in cm−<sup>1</sup> . Spectral data on GCMS was acquired with the direct insertion probe on a Shimadzu spectrometer at 70 eV. Isolated compound was analyzed for structure elucidation by techniques such as <sup>1</sup>H nuclear magnetic resonance (NMR) and <sup>13</sup>C NMR (Choudhary and Atta-ur-Rahman, 1997). The <sup>1</sup>H NMR spectra (300 MHz) were recorded in CDCl<sup>3</sup> as a solvent on a Varian (mercury) instrument. <sup>1</sup>H NMR chemical shifts were expressed in δ (ppm) units, downfield to internal standard TMS (Daniel, 1991). Assignment of signals was confirmed by decoupling experiments.

### Microbial Strains and Culture Conditions Used

Bacterial isolates and standard plasmids used in this study are enlisted in **Tables 1**, **2**. Clinical isolates of Acinetobacter were identified based on their morphological, cultural and biochemical characteristics according to the Bergey's Manual of Systematic Bacteriology (Kloos and Schleifer, 1986) and using API20NE system (Biomeraux, France) as per manufacturer's instructions. All strains were preserved as glycerol stocks at −20◦C or as agar slants at 4–8◦C. All bacterial cultures were grown on Luria agar at 37◦C (Hi Media, Mumbai, India).

## Determination of Resistance to Antibiotics

Antibiotic resistance profile was determined by the disk diffusion method (Bauer et al., 1966). Multi-disks containing antibiotics (Don Whitley Scientific Equipments, Mumbai, India) were placed on Muller Hinton agar plates (Hi Media, India) spread with ca. 10<sup>5</sup> cells of actively growing test culture and incubated at 37◦C. The inhibition zones measured after 24 h incubation and were interpreted according to the manufacturer's interpretation table. MICs of antibiotics were determined by

TABLE 1 | Antibacterial activity of lawsone from P. zeylanica ethanol root extract.


MIC (minimum inhibitory concentration) was detected by disk diffusion assay using 10<sup>5</sup> cells as inoculum on NA for bacteria. DMSO without lawsone, used as a solvent control, did not inhibit growth of any of the bacterial cultures tested. All the strains were isolated from clinical specimens like pus, stool, blood, urine, and sputum.

agar dilution method as described previously (Dhakephalkar and Chopade, 1994). Concentrations of each antibiotic used ranged from 1 µg/ml to 1,024 µg/ml. MIC was interpreted as the lowest concentration of the antibiotic inhibiting bacterial growth.

## Plasmid Isolation

Plasmid isolation was carried out by alkali lysis method as well as boiling method described by Sambrook et al. (1989). DNA was detected by horizontal agarose gel (0.7%) electrophoresis using Tris-acetate-EDTA buffer (pH 8.0).

## Curing of Antibiotic Resistance

The plasmid curing was performed as described earlier by Deshpande et al. (2001). In brief, microbial culture was exposed to different concentrations of curing agent during cultivation in Luria broth at 37◦C for 24 h. Subsequently, the culture was serially diluted and plated on Luria agar to obtain isolated colonies. A total of 100 colonies was replica plated on Luria agar with and without antibiotic. Cured derivatives were scored by their failure to grow in presence of antibiotics. Efficiency of curing was calculated as number of colonies showing reversal of antibiotic resistance per 100 colonies tested. Agarose gel electrophoresis was performed to confirm the removal of plasmid DNA from the plasmid cured strain. Effect of pH (range 5.5–8.0), lawsone concentrations (range 64–512 µg/ml) and inoculum densities (range 104–10<sup>7</sup> cells/ml) on curing efficiency of lawsone was studied in E. coli (pRK2013) and A. baumannii (pUPI281). All the experiments were performed at least in duplicate.

### Effect of Purified Active Principle (Lawsone) on Transformation of Plasmid

E. coli HB101was used as host for the transformation experiments. Calcium chloride method (Sambrook et al., 1989) was used for the preparation of competent cells. Transformation experiments were performed by "heat shock method" as described previously (Sambrook et al., 1989) using plasmid pRK2013 and competent cells of E. coli HB101 as recipient. Transformation efficiency was expressed as number of transformants obtained per µg of plasmid DNA. Purified active principle, at specified concentrations, was added to competent cells along with transforming DNA to evaluate its effect on plasmid transformation.

## Effect of Purified Active Principle (Lawsone) on Plasmid Transfer by Conjugation

Conjugation experiments were performed by membrane filter mating method as described by Towner and Chopade (1987). Effect of lawsone on plasmid transfer was studied by incubating both donor E. coli (pRK2013) and recipient E. coli HB101 (adsorbed on membrane filter) on nutrient agar containing varying concentrations of curing agent. Difference in the percentage efficiency of conjugation in the presence or absence

### TABLE 2 | Plasmid curing with purified lawsone.

fmicb-09-02618 November 7, 2018 Time: 16:47 # 4


MIC, minimum inhibitory concentration; SIC, sub inhibitory concentration. ND, none of the 300 colonies tested showed curing of plasmid. @Am, amoxicillin; Ak, amikacin; Ap, ampicillin; Cm, chloramphenicol; Gm, gentamicin; Km, kanamycin; Lf, lomefloxacin; Nm, neomycin; St, streptomycin; Tc, tetracycline; MTCC (Reference cultures), Microbial Type Culture Collection, Chandigarh, India.

of curing agent was calculated to evaluate effect of lawsone on transfer of plasmid by conjugation.

### Synergistic Action of Lawsone With the Antibiotic: Combination Studies

Synergism between lawsone and antibiotics such as streptomycin and kanamycin was tested against A. baumannii (pUPI281) and E. coli (pRK2013), respectively. Concentration of lawsone as well as the antibiotic tested was in the range 0–1,000 µg/ml. Actively growing culture was inoculated (10<sup>4</sup> cells/ml) in Luria broth supplemented with varying concentrations of lawsone and streptomycin or kanamycin and incubated at 37◦C for 18 h. Culture growth was monitored on spectrophotometer at 600 nm. Interactions of curing agents with antibiotics were assessed by a modified Chequerboard agar dilution method (Charles et al., 1997). Synergistic or otherwise effect of curing agent and antibiotic in combination was evaluated by determining the fractional inhibitory concentration index (minimum FIC index). The FIC was calculated for each combination using the following formula: **FIC A** + **FIC B = FICI** Where, FICA = MIC of drug A in combination/MIC of drug A alone, and FICB = MIC of drug B in combination/MIC of drug B alone. The FICI was interpreted as follows: synergy = FICI < 0.5; no interaction = FICI > 0.5–4; antagonism = FICI > 4 (Dorsthorst et al., 2002).

### Toxicity Testing of Lawsone

Actively growing BHK 21 (baby hamster kidney fibroblast cell line) and AV3 (human amniotic epithelial cell line) cells were obtained from the NCCS repository (Pune, India). These cells were subcultured using TPVG solution (2% trypsin, 0.1% EDTA, and 0.1% glucose in PBS) and seeded on to a 6-well tissue culture plate (Falcon, BD Biosciences, United States) at the concentration of 3 × 10<sup>5</sup> cells per well with Eagle's minimal essential medium (EMEM) supplemented with 10% fetal calf serum (FCS). Plates were incubated for 24 h for the monolayer formation. The monolayers of both, BHK 21 and AV3 were then exposed to different concentrations of lawsone. Plates were observed under inverted microscope (Olympus IX70) for the cell morphology and floating cell population after 24 h of incubation at 37◦C in 5% CO<sup>2</sup> environment. Control cells were grown in absence of lawsone.

### MTT Assay for Examination of the Viability of Cells

BHK21 and AV3 cells were seeded in a 96-well tissue culture plate (Falcon, BD Biosciences, United States) and incubated overnight. The cells were then exposed to varying concentrations of lawsone for 24 h. Control cells were not exposed to the lawsone treatment. MTT assay was performed to determine the toxicity of the plasmid curing agent to the cells upon exposure (Bahuguna et al., 2017) In this assay yellow colored 3-(4,5-dimethylazol-2-yl)-2,5 diphenyl-tetrazolium bromide (MTT) is enzymatically converted into purple formazan in the exposed cells. The intensity of purple formazan was measured at 570 nm. using Spectra Max 250 UV-vis microplate reader (Molecular Devices, Sunnyvale, CA, United States). Experiment was performed in duplicate and the average for all wells was shown as cell-viability percentage related to unexposed control cells.

### RESULTS

### Purification and Identification of Plasmid Curing Agent From P. zeylanica Roots

Crude ethanol extract of roots of P. zeylanica demonstrated significant antimicrobial as well as plasmid curing activity (Patwardhan, 2007). An active principle responsible for the plasmid curing activity of the ethanol extract of P. zeylanica was purified by bioassay guided fractionation. Molecular formula of the purified compound, isolated as yellow crystals, (melting point, 192◦C) was established as C10H6O<sup>3</sup> (**Figure 1**) based on GCMS spectral data (molecular ion peak at m/z = 174). Further structural characterization was made by UV, FTIR, <sup>1</sup>H NMR and <sup>13</sup>C NMR spectroscopic methods. UV λmax at 273 nm for the same compound clearly indicated conjugation in the compound. In the IR spectra, the peak observed at 1,587 cm−<sup>1</sup> suggested the presence of C=C. The peak at 1,641 cm−<sup>1</sup> indicated the presence of α, β-unsaturated carbonyl group. The broad band at 3,056–3,350 cm−<sup>1</sup> was suggestive of hydroxyl group.

In the <sup>1</sup>H-NMR spectrum, the singlet at 6.36 δ was assigned to H-3. The broad singlet 7.33 was due to hydroxyl group which was found to be exchangeable with D2O. The signal at δ 7.71 appeared as doublet of triplet with J = 6.3 and 1.5 Hz was assigned to either H6 or H7. Similarly, appearance of doublet of triplet at δ 7.79 with J = 6.3 and 1.5 Hz was assigned to either H6 or H7. The signal at δ 8.11, corresponding to two protons, appeared as doublet of doublet with J = 6.3 and 1.5 Hz was assigned to H5 and H8. The compound was identified as lawsone based on spectral data (**Figure 1** and **Table 3**). This data was found to be in good agreement with the spectral and analytical data of lawsone (Sigma, United States) thus confirming our assignment. To the best of our knowledge, this is the first report of lawsone extracted from roots of Plumbago zeylanica.

## Curing of Antibiotic Resistance by Lawsone

Lawsone inhibited the growth of A. baumannii, P. aeruginosa, S. Typhi and E. coli, in disk diffusion assay (200–800 µg/disk) (**Table 1**). Plasmid curing activity of lawsone against R-plasmids harbored in reference as well as clinical isolates is illustrated in **Table 2** and **Supplementary Figures S1**, **S2**. Lawsone cured plasmids pBR322 and pRK2013 at 11% and 20% efficiencies, respectively. Lawsone also cured plasmid R136 in S. Typhi at 4.2% curing efficiency. However, it was not able to cure plasmid RP4 in E. coli. Lawsone eliminated plasmids pUPI281 and pUPI282


in A. baumannii with 8.5% and 13.5% efficiency, respectively. Curing efficiencies of lawsone observed against plasmids in reference strains as well as clinical isolates of A. baumannii were comparable to those observed for conventional curing agents such as DNA-intercalating dyes, acridine orange and ethidium bromide (**Table 4**). Plasmid curing efficiency of lawsone was determined against plasmid pBR322, pRK2013, and pUPI281 at different cell densities of host harboring these plasmids. It was observed that curing efficiencies were higher at lower cell densities (10<sup>4</sup> cells/ml) than at higher cell densities (10<sup>5</sup> cells/ml or 10<sup>6</sup> cells/ml) (**Table 5**). Similarly, increasing concentrations of lawsone (64–512 µg/ml) resulted in higher curing efficiencies when initial inoculum density was maintained constant (**Table 6**). Lawsone could not cure either pUPI281 or pRK2013 at 64 µg/ml or lower concentrations. Plasmid curing activity of lawsone was studied at different pH ranging from 6.0 to 8.0 with plasmids pRK2013 and pUPI281. It was observed that lawsone cured pRK2013 and pUPI281 with curing efficiency of 27% and 17%, respectively at pH 6.0 and 8.0. There exists a possibility of reversal of antibiotic resistance if and when a bacterial culture is grown in absence of antibiotics. Hence, in the present study bacterial culture was grown in absence of antibiotics as well as plasmid curing agent (solvent control). Such culture was plated to obtain isolated colonies in absence of antibiotics. The isolated colonies were replicated on agar media with or without antibiotic to determine the frequency of spontaneous loss of plasmid or reversal of antibiotic resistance. None of the 100 colonies tested were spontaneously cured derivative of plasmid bearing host. However, in this technique only 100 colonies were tested by replica plating. Hence, in another modification, cultures were grown in absence of antibiotics or plasmid curing agents for several generations and then plated on agar media with and without antibiotics. The frequency of spontaneous reversal of antibiotic resistance was found to be less than one in 108. This observation was consistent with earlier observation (Inoue, 1997). In comparison, curing efficiencies reported in this study were extremely high (>10<sup>4</sup> times). Both these observations (reversal of antibiotic resistance or plasmid curing observed only upon exposure to plasmid curing agents and not spontaneously) clearly indicated that reversal of antibiotic resistance as a consequence of plasmid curing was not a spontaneous phenomenon but was caused by the exposure of plasmid harboring strains to plasmid curing agent, i.e., lawsone in the present study. In the present investigation, loss of plasmid in the cured derivatives was confirmed by agarose gel electrophoresis analysis (**Figure 2**). Loss of plasmid and not mutation was thus, confirmed as a cause of reversal of antibiotic resistance in the cured derivative. It may be noted here that even mutation could have caused reversal of antibiotic resistance. However, mutagenic activity of the plasmid curing agents is undesired for potential clinical application. Lawsone was able to cure plasmid and inhibit the transfer of plasmid by conjugation as well as by transformation which is significant in containing the spread of antibiotic resistance. Controls did not show loss of plasmids at various pH values, indicating that pH alone did not have any effect on the plasmid curing.

### TABLE 4 | Curing of R plasmids with conventional curing agents.

fmicb-09-02618 November 7, 2018 Time: 16:47 # 6


Three hundred clones were tested for each plasmid in the curing experiments. ND, none of the 300 colonies tested showed curing of plasmid.

TABLE 5 | Effect of cell density on plasmid curing efficiency of lawsone.


Above desired number of cells were incubated in Luria broth in the presence of 512 µg of lawsone/ml at 37◦C for 24 h. Number of clones tested – 300. ND, none of the 300 colonies tested showed curing of plasmid.

### Effect of Lawsone on Plasmid Transfer by Conjugation and Transformation

Effect of lawsone on plasmid transfer by conjugation or transformation was investigated with plasmid pRK2013. More than 90% decrease in the frequency of plasmid transfer was observed when conjugation was performed in the presence of lawsone (12.5 µg/ml) by membrane mating method. Plasmid pRK2013 was transferred from A. baumannii to E. coli with conjugation frequencies of 2.1 × 10−<sup>5</sup> and 2 × 10−<sup>4</sup> in the presence and absence of lawsone, respectively (**Table 7**). The similar inhibition was observed when the conjugation was performed by broth mating (8.8 × 10−<sup>6</sup> and below detection limit). Similarly, lawsone inhibited transfer of plasmid by transformation. Frequency of transformation of E. coli HB101

TABLE 6 | Effect of concentration of lawsone on curing efficiency.


Number of clones tested – 300. 10<sup>4</sup> cells/ml were incubated in the presence of above described concentrations of lawsone at 37◦C for 24 h. ND, none of the 300 colonies tested showed curing of plasmid.

TABLE 7 | Effect of lawsone on R plasmid transfer from ampicillin resistant E. coli to streptomycin resistant E. coli by conjugation.


10<sup>8</sup> cells of donor E. coli (pRK2013) and recipient E. coli (HB101) were mixed and incubated in absence or presence of 12.5 µg of lawsone/ml at 37◦C for 24 h. The growth was serially diluted and plated out to score number of transconjugants formed. The data represents average values of two parallel experiments.

with pRK2013 was observed to be 1.1 × 10<sup>4</sup> transformants/µg plasmid DNA. However, 63% decrease in the transformation efficiency was observed in the presence of lawsone.

### Synergistic Action of Lawsone With the Antibiotic: Combination Studies

Synergy between lawsone and antibiotics was investigated against A. baumannii (pUPI281) and E. coli (pRK2013). The MICs were determined for the curing agent alone and in combination with the different antibiotics by checkerboard assay method. A. baumannii (pUPI281) and E. coli (pRK2013) were resistant to streptomycin and kanamycin, respectively

(MIC > 1,000 µg/ml). Growth of A. baumannii (pUPI281) was inhibited when lawsone and streptomycin were added together at 250 µg/ml concentrations each. Similarly, growth of E. coli (pRK2013) was inhibited when lawsone and kanamycin were added together at 100 µg/ml and 250 µg/ml concentrations, respectively. The Fractional Inhibition Concentration (FIC) indices were calculated (**Table 8**). These results clearly indicated that FICI was synergistic (<0.5) for lawsone in combination with antibiotic when used against A. baumannii. Additive action of lawsone with antibiotic was observed against E. coli. These results suggested that the plasmid curing activity of naphthoquinone directly or indirectly rendered the clinical isolates susceptible to the inhibitory action of antibiotics at significantly lower concentration, and hence, were considered encouraging.

## Toxicity Testing and Viability Assay (MTT) of Lawsone

Toxicity testing of lawsone revealed that BHK 21 cells which are fibroblastic in nature are more affected with lawsone than AV3 cells which are epithelial indicating its selective toxicity to fibroblast only. In MTT assay BHK 21 cells showed more drop in % viability with increased concentration of lawsone compared to AV3 Cells (**Table 9** and **Figure 3**).

## DISCUSSION

The bioassay guided fractionation of the organic extracts of P. zeylanica roots resulted in identification of lawsone as an active principle exhibiting the plasmid curing activity. Lawsone

TABLE 8 | FIC index of individual curing agents and in combination with antibiotics.


<sup>∗</sup>Fractional Inhibition Concentration of curing agent (lawsone). @Fractional Inhibition Concentration of antibiotics. #Fractional Inhibition Concentration index.

TABLE 9 | Toxicity testing with MTT assay.


BHK 21, baby hamster kidney fibroblast cells; AV3, human amnionic epithelial cells. Formula used: % viable cells = (average of test OD/average of control OD) × 100.

has been reported to be a major constituent of Lawsonia inermis. However, to the best of our knowledge, this is the first report of lawsone extracted from roots of Plumbago zeylanica. Antimicrobial properties of lawsone have been documented in the literature (Ahmed and Beg, 2001).

Treatment of chronic infections caused by multidrug resistant bacteria is a serious challenge for the physicians. In most of the cases the resistance to multiple antibiotics is mediated by the plasmids. Curing of plasmid-mediated antibiotic resistance in pathogenic strains of bacteria is of great practical importance both in treatment of bacterial infection and in microbial genetics. In the present investigation, curing of plasmids effected by lawsone was confirmed on the basis of reversal of antibiotic resistance phenotype evident from the significant reduction in the MIC of antibiotics; as well as elimination of plasmids in the cured derivatives. Agarose gel electrophoresis revealed presence of plasmid DNA in wild host. However, gel electrophoresis of DNA from cured derivatives revealed loss of the corresponding plasmid band confirming plasmid curing. Plasmid curing was not observed in host exposed to DMSO, a solvent control used for lawsone. Spontaneous loss of plasmid DNA has been reported at low frequency of less than 1 in 10<sup>9</sup> cells. (Inoue, 1997). In comparison curing efficiencies observed in the present study were significantly higher (>10<sup>4</sup> times), thus attributing reversal of antibiotic resistance to the exposure of cells to the curing agent (lawsone). Cured strains were tested for the loss of plasmids by agarose gel electrophoresis to confirm that reversal of antibiotic resistance was due to elimination of plasmid carrying genes encoding resistance to antibiotics and not due to mutations. Physical loss of plasmid as evidenced by agarose gel electrophoresis of plasmid DNA preparation of cured strains (**Figure 2**) indicated that genes encoding antibiotic resistance

as well as clinical isolates harboring R-plasmids and their cured derivatives. Lane 1, Reference plasmids from E. coli MTCC 131; Lane 2, E. coli MTCC398 (pRK2013); Lane 3, Cured derivative of E. coli MTCC398; Lane 4, E. coli K12 (pBR322); Lane 5, Cured derivative of E. coli K12; Lane 6, Ac. baumannii A24 (pUPI281); Lane 7, Cured derivative of Ac. baumannii A24.

were located on plasmid and that plasmid loss resulted in subsequent loss of antibiotic resistance.

Ability of lawsone to cure plasmid encoded antibiotic resistance in Acinetobacter strains is particularly significant since Acinetobacter strains are known to act as a reservoir of natural or acquired antibiotic resistance genes in the nosocomial environment facilitating in the spread of antibiotic resistance genes to more pathogenic bacteria (Towner, 1997; Deshpande et al., 2001; Cisneros and Rodriguez-Bano, 2005). Bacteria have tendency to develop resistance against any antimicrobial agent used against them. Concentrations of curing agents used in the plasmid curing experiments were significantly lower than inhibitory concentrations. Hence, it is proposed that chances of bacteria developing any mechanism to inactivate plasmid curing agents or their activity are extremely low. It must be remembered here that plasmids are after all dispensable elements in a bacterial cell.

Plasmids are the extrachromosomal elements that are responsible for development and spread of antibiotic resistance in bacteria. Their role assumes even more significance in the nosocomial environment as plasmid encoded resistance to multiple antibiotics can be transferred from one host to another by inter species transfer modes such as conjugation and/or transformation. Such acquired resistance can make otherwise sensitive pathogens resistant to multiple antibiotics. Thus, making the treatment of infections more difficult and in some cases almost impossible. Ability of lawsone to interfere with interspecies plasmid transfer by conjugation and transformation and thus contain the spread of multi-resistance to antibiotics was investigated in the present study. It was observed that Lawsone not only cured plasmids but also inhibited the transfer of plasmid by both conjugation as well as by transformation. Membrane filter mating technique and broth mating are the two different techniques used to investigate plasmid transfer by conjugation in bacteria. Lawsone was able to inhibit the plasmid transfer by conjugation in both techniques. Acinetobacter has been reported as a reservoir of antibiotic resistance genes in nosocomial environment and is frequently reported to be involved in the transfer of multi-resistance to antibiotics. E. coli has been the most extensively investigated organism for the inter species gene transfer. Hence, in the present study both organisms were included to investigate the effect of lawsone on inter species plasmid transfer. Ability of lawsone to inhibit plasmid transfer among these organisms may thus assume special significance in nosocomial environment. Ability of lawsone to inhibit plasmid transfer by transformation or conjugation was reported for the first time in the present investigation.

FIC clearly indicated that combination of curing agent and antibiotic was synergistic against A. baumannii and additive against E. coli. This observation was considered encouraging as

it suggested that the plasmid curing activity of naphthoquinone rendered the clinical isolates susceptible to the inhibitory action of antibiotics at significantly lower concentration. Efficacy of extended antibiotic therapy in several microbial infections may be improved with the combination of plasmid curing agent lawsone, as a synergistic drug. Such synergistic combination of antibiotic and lawsone, may serve as a prospective device in selection of appropriate drug therapy that is likely to contribute to the ongoing crusade against microbial drugresistance.

Plant derived compounds have been previously reported as plasmid curing agents. Such compounds included 8-epidiosbulbin E acetate isolated from the bulbs of Dioscorea bulbifera (Shriram et al., 2008); 1'-acetoxychavicol acetate from Alpinia galanga (L.) Swartz, (Latha et al., 2009); Plumbagin (5-hydroxy-2-methyl-1,4- naphthoquinone) derived from the root of the tropical/subtropical Plumbago species (Patwardhan et al., 2015), etc. These plasmids curing agents have been proven to be effective at curing plasmids in vitro. Unsaturated fatty acids have been shown to be effective conjugation inhibitors in laboratory studies on a variety of plasmids (Lopatkin et al., 2017). Furthermore, they are associated with reduced toxicity on tissue culture cells. Lawsone, a plant derived compound reported in the present investigation has shown dual ability to cure plasmid and also inhibit interspecies plasmid transfer. However, more research is needed to confirm in vivo efficacy and to determine potential toxicity of these compounds if administered in vivo.

Sauriasari et al. (2007) reported that lawsone was not mutagenic to bacterial strains. However dose dependent cytotoxicity of lawsone was reported in the same study. Cytotoxicity of lawsone was reported to be significantly lower than that of other naphthoquinone derivatives including cisplatin, a widely used anticancer drug (Oliveira et al., 2017). Cytotoxicity of naphthoquinones such as lawsone could be attributed to generation of reactive oxygen species, disruption of mitochondrial functions, inhibition of thymidine incorporation into DNA and DNA intercalation (Aithal et al., 2009; Bonifazi et al., 2010; Klaus et al., 2010; Babula et al., 2012). A series of tests in the published literature have identified lawsone as non-genotoxic agent. Such tests included Ames test, V79 hprt test, Syrian hamster embryo cell transformation assay, bone marrow micronucleus tests in CD1 mice, and bone marrow chromosome aberration test in mice and hamsters (Kirkland and Marzin, 2003). An additional genetic toxicity program performed to clarify in vivo genotoxic potential of lawsone revealed that lawsone was devoid of clastogenic potential in vivo (Marzin and Kirkland, 2004). Thus, the review of literature illustrating evaluation of genotoxicity of lawsone in a series of in vivo and in vitro tests has revealed

### REFERENCES

Ahmad, I., Mehmood, Z., Mohammad, F., and Ahmad, S. (2000). Antimicrobial potency and synergistic activity of five traditionally used indian medicinal plants. J. Med. Arom. Plant Sci. 2223, 173–176.

that lawsone is not genotoxic when administered orally up to a dose of 300 mg/kg body weight. Ability of lawsone to intercalate with double stranded DNA has been reported in the published literature (Kirkland and Marzin, 2003). Also, the ability of lawsone to alter the cell membrane by generating free radicle stress has been reported earlier (Sauriasari et al., 2007). However, the plasmid curing activity reported in the present study was observed at concentration significantly lower than the inhibitory concentrations. Elucidation of exact mechanism by which lawsone effected plasmid curing in bacterial scale is not known at present and requires further extensive investigation.

Thus, in conclusion, this investigation has revealed a compound with an ability to eliminate antibiotic resistance and cure plasmids from pathogenic strains that are resistant to multiple antibiotics without any ill effect on mammalian cells at lower concentrations. The results obtained provide endorsement for investigating the potential value of lawsone in combating clinical drug resistance. The synergistic effect of lawsone with the antibiotic exhibits its tremendous potential in modern day therapeutics. The non-toxic, non-mutagenic, plasmid curing and plasmid transfer inhibiting role of lawsone demands further investigation to make it a potential drug of choice in the treatment of antibiotic resistant bacterial strains; thus, demonstrating a new dimension in antibiotic therapy.

### AUTHOR CONTRIBUTIONS

RP, PD, and BC designed and executed the experiments. RP performed the experiments. RP, PD, BC, DD, and RB interpreted the results and analyzed the data. RP, PD, DD, and RB contributed to the writing of the manuscript. RP and PD revised the manuscript.

## ACKNOWLEDGMENTS

RP is grateful to University Grants Commission (Teacher Fellowship for the period of two years under FIP according to Xth Plan by UGC; letter No. 34-8/2003 (WRO) dated 25th February, 2004) as well as Principal, Haribhai V. Desai College for teacher fellowship.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.02618/full#supplementary-material


against cultured melanoma tumor cells. Cell Biol. Int. 33, 1039–1049. doi: 10. 1016/j.cellbi.2009.06.018


galanga against multi-drug resistant bacteria. J. Ethnopharmacol. 123, 522–525. doi: 10.1016/j.jep.2009.03.028


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Patwardhan, Dhakephalkar, Chopade, Dhavale and Bhonde. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Inhibiting Bacterial Drug Efflux Pumps via Phyto-Therapeutics to Combat Threatening Antimicrobial Resistance

### Varsha Shriram<sup>1</sup> , Tushar Khare<sup>2</sup> , Rohit Bhagwat <sup>3</sup> , Ravi Shukla<sup>4</sup> and Vinay Kumar 2,3 \*

*<sup>1</sup> Department of Botany, Prof. Ramkrishna More College, Savitribai Phule Pune University, Pune, India, <sup>2</sup> Department of Biotechnology, Modern College of Arts, Science and Commerce (Savitribai Phule Pune University), Pune, India, <sup>3</sup> Department of Environmental Science, Savitribai Phule Pune University, Pune, India, <sup>4</sup> Centre for Advanced Materials and Industrial Chemistry, School of Science, RMIT University, Melbourne, VIC, Australia*

Antibiotics, once considered the lifeline for treating bacterial infections, are under threat due to the emergence of threatening antimicrobial resistance (AMR). These drug-resistant microbes (or superbugs) are non-responsive to most of the commonly used antibiotics leaving us with few treatment options and escalating mortality-rates and treatment costs. The problem is further aggravated by the drying-pipeline of new and potent antibiotics effective particularly against the drug-resistant strains. Multidrug efflux pumps (EPs) are established as principal determinants of AMR, extruding multiple antibiotics out of the cell, mostly in non-specific manner and have therefore emerged as potent drug-targets for combating AMR. Plants being the reservoir of bioactive compounds can serve as a source of potent EP inhibitors (EPIs). The phyto-therapeutics with noteworthy drug-resistance-reversal or re-sensitizing activities may prove significant for reviving the otherwise fading antibiotics arsenal and making this combination-therapy effective. Contemporary attempts to potentiate the antibiotics with plant extracts and pure phytomolecules have gained momentum though with relatively less success against Gram-negative bacteria. Plant-based EPIs hold promise as potent drug-leads to combat the EPI-mediated AMR. This review presents an account of major bacterial multidrug EPs, their roles in imparting AMR, effective strategies for inhibiting drug EPs with phytomolecules, and current account of research on developing novel and potent plant-based EPIs for reversing their AMR characteristics. Recent developments including emergence of *in silico* tools, major success stories, challenges and future prospects are also discussed.

Keywords: antimicrobial resistance, efflux pumps, efflux pump inhibitors, phyto-therapeutics, drug resistance reversal

## INTRODUCTION

Antimicrobial resistance (AMR) or ineffectiveness of commonly used drugs/antibiotics against specific bacteria has emerged as one of the most threatening human health concerns and a major challenge for global drug discovery programs. AMR (also known as drug resistance) has been reported at three increasing levels, multidrug resistance (MDR), extensive drug resistance (XDR)

### Edited by:

*Henrietta Venter, University of South Australia, Australia*

### Reviewed by:

*Daniel Pletzer, University of British Columbia, Canada Manuel Varela, Eastern New Mexico University, United States*

> \*Correspondence: *Vinay Kumar vinaymalik123@gmail.com*

### Specialty section:

*This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology*

Received: *22 May 2018* Accepted: *19 November 2018* Published: *10 December 2018*

### Citation:

*Shriram V, Khare T, Bhagwat R, Shukla R and Kumar V (2018) Inhibiting Bacterial Drug Efflux Pumps via Phyto-Therapeutics to Combat Threatening Antimicrobial Resistance. Front. Microbiol. 9:2990. doi: 10.3389/fmicb.2018.02990* and pan-drug resistance (PDR). By definition, MDR stands for acquisition of non-susceptibility to at least one agent in three or more antimicrobial classes, while XDR shows the nonsusceptibility to at least one agent in all, except two or fewer antimicrobial classes, while PDR implies non-susceptibility to all antimicrobial agents from all available classes (Exner et al., 2017; Spengler et al., 2017). AMR is threatening millions of lives worldwide, and is rightly declared as a global risk by the World Economic Forum (World Economic Forum, 2013). Since the very first report on AMR in Enterobacteria in 1950s (Watanabe, 1963; Levy, 2001), many drug-resistant strains have been reported and their number as well as the resistance level is on the rise. Though several classes of antibiotics were discovered in the antibiotic era (**Table 1**), we are heading to a post-antibiotic era, where an increasing number of previously curable infections are turning into non-curable and life-threatening (Spengler et al., 2017). Though development of AMR or antibiotic resistance is a natural phenomenon, irrational use of antibiotics speed-ups the emergence of drug-resistant strains (World Health Organization, 2014). Once the AMR is gained by the bacteria, it is successively transmitted to the next progeny via vertical gene transfer or other bacteria through horizontal gene transfer process, making their treatment more difficult (Chandra et al., 2017).

The drug resistance characteristics may be attributed to the abilities of such strains in fast altering their genetic makeup or inducing epigenetic changes (Davies and Davies, 2010; Motta et al., 2015; Rahman et al., 2017). Necessary adaptations are achieved by bacteria to-respond-to and to counteract the antibiotics either via procurement of foreign genetic material encoding resistance via horizontal gene transfer or mutations in drug-targets / antibiotics-degrading enzymes and alterations in permeability of the outer bacterial membrane. There is an unprecedented upsurge in bacterial strains with elevated AMR in both Gram-negative and Gram-positive phenotypes. The ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) pathogens have emerged with high degree of AMR and are major cause of life-threatening nosocomial infections (Santajit and Indrawattana, 2016). Even other strains like Escherichia coli, Shigella species, Neisseria gonorrhoeae, and Proteus mirabilis have shown significant levels of AMR (Fair and Tor, 2014; Prasch and Bucar, 2015; Cerceo et al., 2016). Development of resistance against the carbapenem, a class of highly effective antibiotics and regarded as the last line of defense against pathogenic Gram-negative bacteria hints at the alarming situation (Kumarasamy et al., 2010; Dwivedi et al., 2015).

Intrinsically, AMR is more prevalent and severe in Gramnegative bacteria than their Gram-positive counterparts due mainly to the outer membranes serving as permeability barrier for drug-influx into the Gram-negative bacteria (Silhavy et al., 2010; Exner et al., 2017). To attain low sensitivity against biocidal compounds, Gram-negative bacteria reduce their outer membrane permeability by reducing the number of porins and inducing drug efflux pumps (EPs) for outward transport of drug molecules, often in a non-specific manner making the bacterial cells resistant to multiple antibiotics (Masi et al., 2017). However, despite these morphological differences, Grampositive bacteria cannot be ignored or underestimated and noteworthy examples include methicillin resistant S. aureus (MRSA) and vancomycin resistant S. aureus (VRSA), coagulase negative Staphylococci members including S. epidermidis and S. haemolyticus, Streptococcus pneumonia, E. faecalis, E. faecium, and Clostridium difficile (Schindler and Kaatz, 2016). **Figure 1** shows a gradual upsurge in the number of research articles focused on most-prevalent MDR strains.

In recent years, EPs have emerged as key drivers for AMR in Gram-negative and Gram-positive bacteria, and therefore, are looked upon as potent and universal targets for containing the drug-resistant phenotypes. EPs are vital in other physiological processes also including stress-adaptations, virulence, pathogenicity and transportation of essential nutrients (Piddock, 2006; Fernandez and Hancock, 2012; Costa et al., 2013; Kourtesi et al., 2013; Sun et al., 2014). Identifying novel and potent EP inhibitors (EPIs) to revert the AMR is therefore gaining momentum. EPIs are the compounds with capability to reduce resistance or a complete reversal of AMR against otherwise ineffective antibiotics via inhibiting the EPs (Sun et al., 2014; Gill et al., 2015; Wright, 2016; Spengler et al., 2017). The first EPI against RND-type EPs was reported by Lomovskaya et al. (2001) a phenylalanine-arginine β-naphthylamide (PAβN) effective against Mex pumps in P. aeruginosa and AcrAB-TolC pump in E. coli. Since then, various synthetic and natural compounds have been screened for their EPI capabilities (reviewed by Prasch and Bucar, 2015; Spengler et al., 2017; Shin et al., 2018; Yang et al., 2018). **Figure 2** illustrates the inhibition of microbial drug efflux via synthetic and natural EPIs.

Medicinal plants with antimicrobial properties have the potential to serve as the reservoir of novel and effective EPIs (Newman and Cragg, 2012). Though there are several reports on assessing medicinal plants for their antimicrobial properties (crude extracts and occasionally pure molecules), few investigations were aimed against MDR/XDR strains and fewer on deciphering the underlying resistance mechanisms targeted by the plant products (Kumar et al., 2013). But newer studies are coming up for identifying novel phytomolecules capable of reversing EP-mediated AMR. Some striking phytochemicals which have recently been identified for their EPI potentials include catechol, pinene, gingerol, capsaicin, resveratrol and the number is increasing (Prasch and Bucar, 2015).

In this review, we are presenting an account of major EPs, their roles in imparting bacterial AMR, strategies for identifying plant-based EPIs emphasizing on the potent phyto-EPIs active at relatively lower concentrations, reported during the last 8 years. High throughput screening and in silico approaches for predicting the EPIs and their binding targets/sites are also discussed.

## PHYSIOLOGICAL ROLES PLAYED BY BACTERIAL EFFLUX PUMPS

Bacterial genome comprises of EP genes, expressed under tight regulation of global/local transcription factors (e.g., BmrR:


TABLE 1 | Classes of commonly used antibiotics along with their examples and corresponding modes of action.

transcriptional regulator of efflux pump Bmr in B. subtilis; QacR: transcriptional repressor of QacA transporter in S. aureus; AcrR: transcription repressor of acrB efflux pump in E. coli) proposing the important physiological roles the EPs play during cell development, stress adaptations and bacterial pathogenesis (Sun et al., 2014). The knowledge about these regulatory mechanisms may advance the understandings of physiologically originated AMR, frequently observed in nature (Sun et al., 2014). As discussed above, bacterial EPs have a tremendous capacity to extrude the variety of toxic compounds, needed for the cell survival in a given physiological niche and are vital for maintaining pathogenicity. This is further supported by the studies showing reduced pathogenicity in the bacterial strains lacking EPs. Buckley et al. (2006) showed that S. typhimurium acrB or tolC deficient mutant poorly colonized in the avian gut, highlighting the requirement of complete AcrAB-TolC system for virulence. The S. typhimurium strain lacking all the drug efflux assemblies became avirulent, when tested in a mouse model (Nishino et al., 2006). To confirm the role of EPs in bacterial pathogenesis, Hirakata et al. (2002) assessed the ability of EP (MexAB-OprM, MexCD-OprJ, MexEF-OprM, and MexXY-OprM) mutants of P. aeruginosa to invade Madin-Darby canine kidney cells. The findings revealed that except mexCD-OprJ, all other systems evidenced decreased bacterial invasion abilities.

EPs are also known to effect the bacterial cell communication during the stress-responses, especially in the quorum-sensing (González and Keshavan, 2006). As transportation of autoinducers (chemical signals generated during quorum sensing) is a key-event during cell-cell interactions via quorum-sensing, drug EPs assist their transport across the membrane (Liang et al., 2016). Moore et al. (2014) confirmed a vital role played by MexAB-OprM efflux system in the secretion of a major auto-inducer N-acylated L-homoserine lactone by P. aeruginosa cells. The study also postulated this auto-inducer as a substrate for MexAB-OprM system (Moore et al., 2014).

Further, Martinez et al. (2009) advocated the EP-mediated ceasing of quorum-sensing via augmented efflux of autoinducers, facilitating quick bacterial response to stress signals. One more physiological role attributed to EPs is in the biofilm formation. Recent studies confirm the involvement of many well-characterized efflux systems, AcrAB-TolC of E. coli, AcrD of S. enterica, AdeFGH of A. baumannii and MexAB-OpeM of P. aeruginosa in biofilm formation (Alav et al., 2018). Kvist et al. (2008) reported an up-regulation of 20 genes encoding EP-transporters in E. coli during the growth of biofilm. Similarly, the EP-mediated export of colanic acid for capsule-matrix formation was observed along with up-regulated TetA(C) in E. coli, facilitating the biofilm formation (May et al., 2009). Collectively, the physiological roles of EPs are vital for pathogenic stability and virulence maintenance in bacteria.

The synthetic EPIs namely carbonylcyanide mchlorophenylhydrazone (CCCP), chlorpromazine and PAβN were reported to prevent biofilm formation in E. coli, P. aeruginosa and S. aureus (Baugh et al., 2014). However, investigations on evaluating phytochemicals for their antibiofilm potencies via inhibiting EPs are few. Fiamegos et al. (2011) isolated 4,5-O-dicaffeoylquinic acid from Artemisia absinthium which was proved to be a potent inhibitor of MFS pumps in E. coli and E. faecalis and as an anti-biofilm agent (Fiamegos et al., 2011). Recent reports advocate that nanomaterials in combination with phyto-EPIs can also be an effective therapy for containing drug-resistant infections (Gupta et al., 2017).

## BACTERIAL EFFLUX PUMPS: THE WARHEADS IN AMR STRAINS

Though bacterial AMR has several origins and many adaptive mechanisms are employed by drug-resistant strains against the antibiotics, the intrinsic EPs hold the key. Recent clinical and laboratory data establish that bacterial EPs are not only critical for drug-extrusion but also contribute to their virulence and adaptive responses (Du et al., 2018). Often, antimicrobial drug exposure induces intricate bacterial reactions including altered expressions of several genes encoding the transporters, as revealed by the phenotypic profiling of E. coli (Nichols et al., 2011).

Bacterial EPs are acknowledged either as primary active transporters using ATPs as an energy source, or as secondary active transporters acquired due to electrochemical potential difference created by pumping out Na<sup>+</sup> and H<sup>+</sup> outside the membrane (Dwivedi et al., 2017a).

This may be considered for classifying EPs into two broad super-families namely; ATP-binding cassette (ABC) multidrug transporters and secondary transporters using proton motive force (PMF) as an energy source (Putman et al., 2000). The second super-family can again be categorized in four subclasses, the major facilitator superfamily (MFS), resistancenodulation-cell division (RND), multidrug and toxic compound extrusion (MATE) and small-MDR (SMR) family (Fernandez and Hancock, 2012; Sun et al., 2014). RND and MFS pumps are the most common in bacteria. With relatively narrow spectrum of specificity, MFS pumps are found in both Gram-negative and Gram-positive bacteria; while poly-selective RND pumps are exclusive to Gram-negative bacteria (Ward et al., 2001; Molnár et al., 2010). The SMR transporters show specificity for broad-spectrum polyaromatic cations convening resistance for compounds sharing similar chemical description. The MATE transporters are similar in size to MFS transporters but they do not share any sequence similarity with them (Jack et al., 2001). Apart from the classified super-families and sub-families, the MATE, SMR, and RND classes are distributed uniquely to prokaryotes whereas MFS and ABC transporters are dispersed in both prokaryotes and eukaryotes.

The large MFS is one of the most functionally diverse transporter families comprising multiple transportation types for drugs as well as sugars. These transporters comprise ∼400 amino acids arranged as membrane-spanning helices (Saier et al., 1998). Based on the helical structure, MFS transporters can be classified as either 12-[e.g., TatA(B): class B tetracyclin transporter from E. coli] or 14-helix transporters [e.g., TatA(K): class K tetracyclin transporter from S. aureus], and TetA(B) is one of the most extensively studied members of the family (Lynch, 2006).

The SMR pumps represent smallest multidrug transporters, possessing only four trans-membrane helices without any extra membrane domain. But single minimal functional SMR unit represents eight helices as these are functionally active in dimeric form (Higgins, 2007). Well-illustrated example from this family is the electromagnetic antiporter EmrE from E. coli, responsible for resistance to a range of cationic-hydrophobic entities including antibiotics.

The latest structurally characterized class of EPs is MATE, involved in various vital biological functions (He et al., 2010). These transporters are equivalent to MFS transporters with a typical composition of ∼450 amino acids putatively arranged in 12 helices; but with no sequence similarity with MFS counterparts (Jack et al., 2001). Some characterized MATE members include NorM from N. gonorrhoeae and N. meningitides and YdhE from E. coli. However, limited structural and functional knowledge is available about this family (Dwivedi et al., 2017a).

Though EPs from other families contribute to the AMR against certain antibiotics, RND pumps are the most potent drug efflux systems conferring resistance against clinically important antibiotics and biocides. Members of this family are known for their roles against a wide range of molecules with dissimilar structures including antibiotics, biocides, organic solvents, antimicrobial peptides, detergents, dyes, and bile salts (Poole, 2004). The tripartite complex of pumps from this family comprises inner membrane protein (IMP), outer membrane protein (OMP) along with a periplasmic membrane fusion protein (MFP) as a connector (Venter et al., 2015). The best understood tripartite complexes include AcrA-AcrB-TolC from E. coli and MexA-MexB-OprM from P. aeruginosa (Du et al., 2013, 2014). Greater structural and functional similarities between IMPs of these two systems are described by Welch et al. (2010). Exploration on biochemical and structural aspects of AcrB has revealed that these IMPs contain proximal and distal binding pockets, divided by G-loop (with 614–621 residues). Conformational flexibility of G-loop is crucial for movement of substrate along the binding site (Eicher et al., 2012; Cha et al., 2014). **Table 2** lists various EPs belonging to the major families from prevalent pathogenic bacteria.

Collectively, the complex EP assemblies are critical for bacterial pathogenesis, virulence, biofilm formation, and adaptive-responses ultimately conferring and defining bacterial AMR (Piddock, 2006; Martinez et al., 2009; Baugh et al., 2012, 2014; Du et al., 2018). EPs are critical for bacterial AMR as they exclude most of the unwanted entities


TABLE 2 | Examples of various efflux pumps belonging to major efflux pump families from prevalent pathogenic bacterial strains.

*ACR, acriflavine; AG, aminoglycosides; BAC, benzalkonium chloride; BAP, biaryl-piperazines; BL,* β*-lactams; BIO, biocides; BPD, bisanilino-pyridines; CAB, cetyl-trimethylammoniumbromide; CAP, cationic antibacterial peptides; CAZ, ceftazidime; CEF:cefepime; CHL, chloramphenicol; CHX, chlorhexidine; CIP, ciprofloxacin; CLA, clofazimine; DAU, daunomycin; DAPI, 4',6-diamidino-2-phenyl indole; DOR, doxorubicin; DOX, doxycycline; EB, ethidium bromide; ERY, erythromycin; FQ, fluoroquinolones; FUA, fusidic acid; LIN:lincosamide; LNZ, linezolid; ML, macrolides; NOR, norfloxacin; NOV, novobiocin; PAS, p-aminosalicylate; PLE, pleuromutilin; PRI, pridones; PYR, pyrroles; QAC, quaternary ammonium compounds; SPE, spectinomycin; STA, streptogramin A; STR, streptomycin; SUL, sulfonamides; TEL, telithromycin; TET, tetracycline; TGC, tigecycline; TMP, trimethoprim; TPP, tetraphenylphosphonium,TRI, triclosan.*

until the cell gets required time for acquiring resistance (Piddock, 2014; Venter et al., 2015).

## BACTERIAL EFFLUX PUMPS: THE URGENT THREATS REQUIRING IMMEDIATE REMEDY

It is well-established that EPs comprise one of the most vital systems in bacteria responsible for both innate and acquired AMR (Blair et al., 2015). There are reports of EPs from different superfamilies and the occurrence of various types of EPs from the same superfamily in a single bacterial species (Piddock, 2006). For instance, whole genome sequencing of the colistin resistant Enterobacter cloacae showed presence of multiple EPs (Norgan et al., 2016). Differential substrate profiles of EPs are also a characteristic feature which may diverge between or within the superfamily (Poole, 2005, 2007). Although the core motive of EPs related studies is focused on AMR, several reports however confirmed other but significant functions of bacterial EPs including quorum-sensing, biofilm formation, virulence, pathogenicity and bacterial behavior (Piddock, 2006; Yang et al., 2006; Fahmy et al., 2016).

Up-regulation of gene expression levels are one of the main drivers for chromosomally acquired AMR. This can be triggered due to the gene induction, activated transcription, or due to regulatory mutations (Grkovic et al., 2002). The coding region for an EP is usually found contiguous to the regulatory proteins controlling the expression levels of pump gene in response to substrates. For example, AdeL, an LTTR (LysR-type transcriptional regulator) family protein exists opposite to the adeFGH operon that regulates the expression of genes encoding RND efflux system in A. baumannii (Liu et al., 2018). The expression levels of EP-associated proteins along with porins is mutually controlled by several global regulatory elements, modifying the transcription patterns of EP-family transcripts either directly or through a cascade of regulatory events (Warner and Levy, 2010; Sun et al., 2014). Further, the expression of MexAB-OprM efflux system is governed by repressor protein mexR, encoded by a gene located upstream of the mexAB-oprM operon in P. aeruginosa (Suresh et al., 2018). Similarly, the acrAB operon system is regulated by regulator acrR in E. coli, located 140 bp up-stream of the acrAB operon (Ma et al., 1996).

Another striking bacterial character adding to the AMR nature is heteroresistance, the occurrence of differential responses to antibiotics by the bacterial cells from the same population, a phenomenon first reported in S. aureus (Kayser et al., 1970). Interestingly, drug resistant and sensitive bacterial cells may coexist in a single culture (Morand and Mühlemann, 2007). The mechanism underlying heteroresistance acquirements are yet to be fully understood, however, the active EPs are strongly linked to heteroresistance (Chen et al., 2017). Designing a treatment course against such strains is difficult as there are high chances of increase in the frequency of resistant-bacterial-population and stimulation of cross-resistance to antimicrobial lysozymes of the host system (Napier et al., 2014; Telke et al., 2017). Up-regulation of OpxAB gene in Salmonella typhimurium (Chen et al., 2017) and AdeABC gene in A. calcoaceticus-A. baumannii (Ruzin et al., 2007) are attributed for mediating the tigecyclin heteroresistance. Similarly, colistin associated heteroresistance is also reported in E. asburiae LH74 and E. cloacae NH52 (Telke et al., 2017), showing its association with overexpression of acrAB-tolC EPs under the regulation of soxRS genes.

AMR phenotypes may result from concurrent acquisition of several AMR mechanisms simultaneously. It may include a combination of phenomena like chromosomally acquired resistance, multiple chromosomal changes with time, and/or a single mutational event activating the AMR mechanisms including the EPs (Lister et al., 2009). The over-expressions of EPs and their corresponding genes have been reported to contribute to MDR in P. aeruginosa (Shigemura et al., 2015). Recent studies have confirmed the role of EPs in fluoroquinolone resistant E. coli (Amabile-Cuevas et al., 2010; Swick et al., 2011; Yasufuku et al., 2011). Similarly, two fluoroquinolone resistant clinical isolates of Shigella showed overexpression of the TolC channels, part of AcrAB-TolC tripartite responsive to ciprofloxacin (Kim et al., 2008). These findings confirm that the up-regulation of EP genes contribute significantly to diminish intracellular antibiotics level, with a selectivity of the efflux transporter.

Overall, the poly-specificity of EPs, their overexpression in response to drugs along with the phenomenon of heteroresistance seem key factors responsible for drug-resistance in a wide-range of bacterial species, especially in Gram-negative bacteria making them difficult to treat with conventional drug arsenal. The drug-efflux mediated bacterial AMR is a mounting threat to global healthcare, therefore EPs are gaining unprecedented attention not only from the perspectives of basic understandings that how they work and impart drug-resistance but also as emerging targets for development of novel and potent adjunct-therapies for combating AMR in community and nosocomial infections. As a result, inhibition of drug efflux from bacterial cells via inhibiting or disrupting the EPs is an emerging approach for combating the threatening AMR. Various approaches have been developed in recent past and a schematic for these strategies for inhibition or disruption of bacterial drug efflux is illustrated in **Figure 3**.

## PHYTOTHERAPEUTICS–THE POTENT EFFLUX PUMP INHIBITORS

Phytochemicals are critical for human health-care since ancient times. Medicinal plants are hailed as a reservoir for phytochemicals capable of providing new and potent drug leads to contain the AMR via targeting the principal determinants of drug-resistance including EPs (Newman and Cragg, 2012; Prasch and Bucar, 2015). This review focuses mainly on EPIs of plant origin (phyto-EPIs) reported in the running decade. We are discussing some important and successful case studies on phyto-EPIs effective against AMR phenotypes. **Table 3** summarizes the list of phytochemicals, their source and effective concentrations used for inhibiting the efflux pumps of AMR bacterial strains.

One of the potent EPIs, the anti-hypersensitive alkaloid reserpine was isolated from Rauwolfia vomitoria (Stavri et al., 2007). Similarly, EP inhibitory activity of gallotannin (1,2,6 tri-O-galloyl-b-D-glucopyranose) isolated from hydro-alcoholic extracts of Terminalia chebula fruits was demonstrated by Bag and Chattopadhyay (2014) against MDR uropathogenic E. coli. Gallotannin induced a 2- to 4-fold reduction in minimal inhibitory concentration (MIC) of test antibiotics via inhibiting the ethidium bromide (EtBr) pump (**Table 3**). As EtBr is a known EP-substrate, inhibition of EtBr efflux backs the postulated EPinhibitory activity of gallotannin (Bag and Chattopadhyay, 2014). Methanolic sap extracts of Acer saccharum was evaluated for its drug efflux inhibitory potentials against P. aeruginosa (ATCC 15692 and UCBPPPA14), E. coli (ATCC 700928) and P. mirabilis (HI4320) confirmed via monitoring the EtBr efflux (Maisuria et al., 2015).

A clavine alkaloid lysergol from Ipomoea muricata was evaluated against AMR E. coli strains (MTCC1652 and KG4) to test its EP inhibitory potentials, and strong activities (higher than the standard reserpine) were exhibited by lysergol and its derivative, 17-O-3′′,4′′,5′′-trimethoxybenzoyllysergol (Maurya

Venter et al., 2015).

et al., 2013). Authors also reported the inhibitory activities of this compound against ABC pump YojI in E. coli (Maurya et al., 2013). Similarly, falcarindiol, isolated from Levisticum officinale exhibited EPI activities against the Gram-negative strains (Garvey et al., 2011).

On the similar lines, Dwivedi et al. (2017b) reported the antibiotic-potentiating activities of catharanthine against superbug P. aeruginosa. The investigation involved in silico docking followed by the in vitro evaluation revealed that catharanthine potentiates the activity of tetracyclin and


TABLE 3 | A summarized list of phytochemicals, their source and effective concentrations for inhibiting efflux pumps from antimicrobial resistant bacteria.

streptomycin, as confirmed by a reduced MIC, and acts as a potent EPI (Dwivedi et al., 2017b). A pentacyclic triterpenoid ursolic acid from leaves of Eucalyptus tereticornis described as a precursor of putative EPI was evaluated against MDR E. coli (KG4), two promising semi-synthetic, esterified derivatives of ursolic acid, 3-O-acetyl-urs-12-en-28-isopropyl ester and 3-O-acetyl-urs-12-en-28-n-butyl ester and the parent compound exhibited better EP inhibitory potencies than the standard reserpine (Dwivedi et al., 2014a). The molecular docking confirmed the targets of these compounds as AcrA/B, MacB, TolC, and YojI (Dwivedi et al., 2014a). Similarly, two alkaloids isolated from roots and rhizomes of Berberis vulgaris, the barberine and palmatine showed potent EP inhibitory efficacies against P. aeruginosa isolated from burn infections (Aghayan et al., 2017).

Phenylpropanoids from the n-hexane and chloroform fractions of Alpinia galanga exhibited EP inhibitory activities against Mycobacterium smegmatis mc<sup>2</sup> 155ATCC 700084 (Roy et al., 2012). A dose-dependent EP inhibition was observed with 1′ -S-1′ -acetoxyeugenol acetate (Roy et al., 2012). Mukanganyama et al. (2012) examined another mycobacterial member Mycobacterium aurum A+ against a naphthoquinone diospyrine isolated from Diospyros montana along with its derivatives. Two derivatives proved highly potent EPI and allowed bacterial cells to accumulate high concentrations of ciprofloxacin (Mukanganyama et al., 2012).

The acylphloroglucinol isolated from n-hexane fractions from Hypericum olympicum, olympicin-A showed promising activities against S. aureus (Shiu et al., 2013). The radiometric accumulation assay of the strain overexpressing NorA pump indicated the enhanced accumulation of (14)C-enoxacin, thus confirming efflux inhibition (Shiu et al., 2013). Two coumarins [5,7-dihydroxy-6-(2-methylbutanoyl)-8-(3 methylbut-2-enyl)-4-phenyl-2H-chromen-2-one and 5,7 -dihydroxy-8-(2-methylbutanoyl)-6-(3-methylbut-2-enyl)-4 phenyl-2H-chromen-2-one] obtained from floral buds of Mesua ferrea were assessed against NorA-overexpressing S. aureus 1199B and the clinical isolate MRSA 831 (Roy et al., 2013). Linoleic acid isolated from ethanolic extracts of Portulaca oleracea showed efflux inhibitory potential at 64 mg L−<sup>1</sup> concentration, equivalent to reserpine when quantified against MRSA (RN4220/pUL5054: erythromycin resistant, overexpressing MsrA ABC EP, Chan et al., 2015). In search of the drug-resistance reversal agents, dos Santos et al. (2018) assessed caffeic acid and gallic acid against four strains of S. aureus; 1199 as a wild type strain, 1199B as NorA harboring fluroquinolone resistant, IS-58 possessing TetK pump and RN4220 possessing MrsA pump. The study confirmed caffeic acid as a potent AMR-reversal agent, as it effectively inhibited MrsA and NorA EPs from S. aureus strains RN4220 and 1190B, respectively (dos Santos et al., 2018). In another interesting study, Kakarla et al. (2016) reported LmrS inhibitory activities of Cuminum cyminum. The study revealed that the cumin inhibits the LmrS mediated transport of drugs resulting in growth inhibition of MRSA clinical isolate in a dose-dependent manner (Kakarla et al., 2016).

Traditionally, most of the investigations were aimed at identifying EPIs for Gram-positive strains for reversing their AMR characters with very few reports against Gram-negative members. This can be because Gram-negative bacteria are more difficult targets then their positive counterparts due mainly to the presence of powerful EPs and other effective membrane barriers (lipophilic layer) averting them from external impacts (Stavri et al., 2007; Prasch and Bucar, 2015). Though, some approaches have emerged in recent years for improving antibiotic-penetration across the permeability membranes of Gram-negative bacteria such as the inhibition of new accessible target, identification of uptake pathways and the "Trojan Horse" approach (achieving fast or facilitated antibiotics uptake), establishing the rules of permeation (for predicting whether elevated uptake or reduced efflux would be the most efficient way for increasing the potency of specific antimicrobial class) and identifying potent EPIs, last one being probably the most potent (Zgurskaya et al., 2015).

Recently, Bruns et al. (2017) successfully inhibited EmrD-3 pump-mediated drug efflux from a Gram-negative bacterium Vibrio cholerae by garlic extract and its bioactive compound, allyl sulfide. At relatively low concentrations, the extract seems to target the EmrD-3 pump, but at higher garlic extract concentrations, the respiratory chain was affected. This example confirms targeting the energization of the efflux system by plant compounds as a potential strategy for drug efflux inhibition (Bruns et al., 2017).

Further, the MFS conserved sequence motifs, present across the entire superfamily, provide vital information regarding alignments of MFS transporter sequences (at least motif containing region), which may help in understanding the structural templates and actual binding events achieved via these transporters. Molecular dynamic simulation (MDS) studies of VMAT2 multidrug transporter (MFS family) revealed the presence of two domains of six trans-membrane helices (Yaffe et al., 2013). The trans-membrane residues at anchoring sites are identified as hinge points, at which straightening and flexing movement of helices occur, required for transport. These anchor point residues are highly conserved throughout the MFS family (Yaffe et al., 2013) and are emerging targets for drug efflux inhibition. Recent advances in scientific and technological arena have added significant in-depth understandings of the structural and biochemical basis of drug efflux, substrate profiles, molecular regulation and inhibition of major EPs.

Active EPs play a critical role in intrinsic and elevated drug resistance acquired via overproduction or over-activation of pumps in Gram-negative bacteria, and the development of clinically useful EPIs or new antibiotics to bypass pump-effects continues to be a challenge in combating Gram-negative bacterial infections (Li et al., 2015). As practically all the antibiotics are susceptible to active drug efflux, the potent EPIs can target these pumps antagonistically and can make old antibiotics effective again (the phenomenon known as re-sensitization). Besides, considering the fact that several antimicrobial agents like lipophilic penicilines, many glycopeptides, oxazolidinones, macrolides and lipopeptide daptomycin are effective in treating only Gram-positive bacterial infections and their poor potencies against Gram-negative pathogens is at least partially due to their active drug efflux, novel and potent EPIs are needed to significantly broaden the range of these antimicrobial agents. All this clearly indicates that EPIs have tremendous potential in adjunctive therapies along with the known but otherwise ineffective antibiotics ultimately reducing the emergence of AMR and virulence (Opperman and Nguyen, 2015). But developing novel and potent EPI is difficult and needs to overcome several hurdles such as choice of antibiotics for potentiation and matching the pharmacological properties of EPI-antibiotics pair (Opperman and Nguyen, 2015; Zgurskaya et al., 2015).

Considering the serious threats posed by the Gram-negative bacteria and their drug-resistance nature, more investigations aiming to target them with the novel, alternative and effective approaches including exploration of natural products are coming up. Though there are limited success stories, but they may lay the foundation for developing potent EPIs to avert the AMR phenotypes with the help of natural sources.

### IN SILICO MOLECULAR DYNAMICS SIMULATIONS (MDS) TO SCREEN AND DEVELOP EPIS

The MDS is a commanding approach for computational validation and to support the hypothesized mechanisms of EPs and EPIs (Nikaido and Takatsuka, 2009). It has made possible to simulate the membrane protein complex structures with micro-second time-scales. The MDS approach along with molecular docking and other in silico tools are successfully utilized for screening and prediction of molecular interactions between potential EPs and their corresponding inhibitors (Jamshidi et al., 2016). This has led to the unraveling of the mechanism how drug efflux systems recognize and transfer specific molecules; thus helping researchers in challenging the efflux-mediated resistance and finding appropriate EPIs for improvement in antibiotics efficacy during AMR (MDR/XDR) infections (Collu et al., 2012; Nakashima et al., 2013). There are several recent reports describing the successful applications of these in silico approaches for screening and identifying potent EPIs of plant origin (Bhaskar et al., 2016; Jamshidi et al., 2016; Mangiaterra et al., 2017; Verma et al., 2017).

A general scheme for MDS-based approaches is depicted in **Figure 4**. Briefly, it starts with the identification of threedimensional structures of potential EP-binding sites (pockets). The next step is the prediction of trans-membrane segments

from protein sequences. The predicted structure can then be checked for its stereochemical properties by analyzing the overall and residue-by-residue geometry. The modeled protein structure is then reduced with the solvent implied by chimera programs (http://www.cgl.ucsf.edu/chimera/, Pettersen et al., 2004) and the projected protein structure can be used for its interaction with potential EPI molecules. The three-dimensional structure of the EPI is then explored to attain perfect and stable EPI-EP complex. Potent tools for docking studies include AutoDock (http://autodock.scripps.edu/, Morris et al., 2009), and SwissDock (http://www.swissdock.ch/, Grosdidier et al., 2011). Such automated docking tools can predict the exact binding position of the candidate drug-molecule to the receptors, and provides vital information about exact amino residues taking part in bond-formation with potential drug(s), their bond lengths and type and other interactions adding to the stability of the docking complexes.

Recently, Kesherwani et al. (2017) used high throughput virtual screening of natural compounds against NorM, a MATE transporter from N. gonorrhea followed by flexible docking. Authors performed molecular simulation in a membrane environment for investigating the stability and binding energy of top lead compounds, and identified a phytomolecule from Terminalia chebula with higher binding free energy than the substrates (rhodamine 6 g, ethidium). The compound successfully blocked the disruption of Na+-coordination along with an equilibrium state bias toward occlude state of NorM transporter, ultimately blocking the extrusion of antimicrobial drugs via inhibiting the NorM transporter in drug-resistant N. gonorrhea.

Similarly, Suriyanarayanan and Sarojini (2015) analyzed EPI potentials of plant-derived flavonoid quercetin in bringing down the drug efflux via inhibiting the EmrE, a transporter belonging to SMR family from E. coli. Authors used in silico approaches and molecule docking approaches. The docking analysis of quercetin with EP-protein showed the importance of residues for function and stability, and notably quercetin showed best interactions as compared to the compounds like verapamil, reserpine, chlorpromazine, and carbonyl cyanide m-chloro phenylhydrazone. MDS confirmed the stability of quercetin-Mmr complex, which insights the potential of quercetin as a non-antibiotics adjuvant for treatment of bacterial infection via reducing the drug efflux from bacteria.

Mangiaterra et al. (2017) identified two phyto-EPIs using in silico high-throughput virtual screening. Molecular docking revealed these two compounds morelloflavone and pregnan-20 one derivative as inhibitors of MexAB-OprM EPs of P. aeruginosa and supportive in vitro assays confirmed their synergism with ciprofloxacin (Mangiaterra et al., 2017).

Molecular docking plays a crucial role and help in defining drug-protein interactions which determines whether compounds act as substrates for EP proteins. Therefore, inhibitors or modulators of EPs are well-recognized along with their comparative binding efficiencies via detailed docking analyses (Collu et al., 2012). Putative EPI activities of quercetin, plumbagin, nordihydroguaretic acid, shikonion and mangiferin were confirmed (Ohene-Agyei et al., 2014). Similarly, docking of reserpine, salvin, totarol, ferruginol along with known antibiotics to NorA revealed that all the tested compounds showed binding at large hydrophobic cleft, suggesting the substantial interactions with key-residues (Bhaskar et al., 2016). Notably, all these investigations were backed by the bioassays confirming the validity of information generated via in silico screening.

Owing to the importance of an instantaneous requirement of curing the XDR/MDR strains with utmost specificity, a greater understanding of exact drug-identification and its transport by MDR-EPs is important. The in silico MDS approach along with virtual docking and wet laboratory validation therefore can be considered as an imperative path in identifying potent phyto-EPIs.

## MOLECULAR INTERACTIONS UNDERLYING INHIBITION OF EFFLUX PUMPS BY PHYTO-THERAPEUTICS

The inhibition of active drug efflux by EPIs results into the elevated intracellular antimicrobial concentrations, and lowered or complete reversal of efflux-mediated bacterial drug resistance, prevention of microbial invasiveness by inhibiting the export of virulence-factors and shortened adaptation-time required for bacteria, prohibiting the emergence of mutant strains with high AMR (Bhardwaj and Mohanty, 2012; Sun et al., 2014). Major strategies developed for drug efflux inactivation are, first, alterations in regulatory mechanisms for activation/repression of EP gene expressions (Purssell and Poole, 2013), second, deprivation of motive forces required for working of pumps by diminishing the proton gradient (Viveiros et al., 2005; Martins et al., 2008), third structural modifications in existing antimicrobials to bypass the chemophore recognition by the EPs (Chollet et al., 2004; Rice et al., 2005), fourth disrupting the pump-functionality by averting assembly of pumps by targeting protein interfaces (Tikhonova et al., 2011); interaction between protein motifs (Hobbs et al., 2012); obstructing the exit duct (Zeng et al., 2010), and fifth, the trapping of EPs in the inactivated form by competitive binding of EPIs and cytoplasmic membrane proteins (Nakashima et al., 2013; Opperman et al., 2014; Nguyen et al., 2015; Opperman and Nguyen, 2015; **Figure 3**). In addition, targeting the molecular hinge structures by the conserved sequence motifs is also an emerging strategy for EP inhibition (Abdali et al., 2017). The conserved sequence motifs (7–13 residues) are characteristics of MFS family and these motifs on c-terminal end of trans-membrane helix are rich in glycine and proline, vital for promotion of hinge formation. These conserved residues are considered as major contributors in binding and transportation of respective substrates (Luo and Parsons, 2010), and therefore targeting them holds significance for drug efflux inhibition.

To identify the phyto-EPIs, some authors tried to decipher the physiological and molecular interactions involved in EP dysfunction. Sharma et al. (2010) described piperine as an inhibitor of Rv1258c, an efflux protein transporter present on cytoplasmic membrane [encoding for tetracyclin/aminoglycoside resistance (TAP-2)-like EP] in M. tuberculosis H37Rv. After structural prediction of the protein, further investigation revealed the binding pocket of Rv1258c. Authors showed H-bond interaction (2.06 Å) with Arg141 residue and piperine provided stable protein-ligand interaction. The findings confirmed the role of piperine in augmenting rifampicin sensitivity in M. tuberculosis (Sharma et al., 2010). Capsaicin also proved a potent EPI, inhibiting NorA pump of S. aureus (Kalia et al., 2012). The study showed the involvement of Arg98 and Ile23 residues from active binding site in the key binding interactions. The stable interaction between capsaicin and active site at the proposed orientation allows an aliphatic chain of capsaicin, extending in a hydrophobic cleft (containing residues Pro24, Phe140, Ile244, and Phe303) permitting strong hydrophobic interactions due to a lesser distance between ligand and molecule (1.7–3.2 Å). A weak H-bond formation between OH-group (from aryl moiety of capsaicin) and Arg98 was attributed for providing extra-stability to capsaicin/NorA complex (Kalia et al., 2012). Another study by Zhang et al. (2014) described the interactions between ginsenoside 20(S)-Rh2 and NorA from S. aureus. The stable H-bond formation between ginsenoside 20(S)-Rh2 and Gln51/Asn340/Ser226 residues at active binding site in the central cavity of protein was attributed for the inhibition of NorA pump, thus promoting accumulation of ciprofloxacin inside the bacterial cell (Zhang et al., 2014). In the similar vein, Ohene-Agyei et al. (2014) assessed five phytochemicals (plumbagin, shikonin, quercetin, mangiferin and nordihydroguaretic acid) for their EPI potentials against AcrB protein from AcrAB-TolC drug transporter. The stable H-bond formation between T monomer of AcrB with minocycline attached to binding pocket and phytochemicals was responsible for efflux inhibition by the phytochemicals (Ohene-Agyei et al., 2014). Further, the authors also postulated that the considered natural compound act as a substrate and compete with the antibiotics for drug-resistance reversal (similar to PAβN). Therefore, these natural products act as high-affinity substrate inhibitors rather than substances for trapping the EPs in an inactive state.

## CONCLUDING REMARKS

Increasing AMR in community and nosocomial settings is a big threat to human healthcare and accounts for a large number of mortalities and morbidities globally. Bacterial EPs make up a major warhead of the drug-resistant pathogens and increase and maintain the AMR via extruding or reducing the intracellular concentrations of applied antibiotics, often in a non-specific manner. The drug EPs are also emerging as chemical tools to understand molecular mechanisms underlying drug extrusion from the bacterial cells. EPs play several important physiological and molecular roles in bacterial cell survival and stress-responses. The necessity to overcome AMR has encouraged investigators to characterize resistance-inhibiting or modulating EPIs to block the drug extrusion, restoring antibacterial susceptibility and returning existing antibiotics into the clinic. The severity of the AMR is higher in Gram-negative bacteria, owing to their superior capabilities in maintaining high drug efflux levels coupled with lower intracellular levels of toxic drugs including antibiotics. MDR/XDR strains maintain their intrinsic and acquired resistance via overproduction of pumps. The development of clinically useful EPIs to bypass pump effects continues to be a challenge. Though there are some noteworthy developments in recent past aimed at reversing the AMR phenotypes including facilitation of better drug-penetration across the outer membranes of Gram-negative bacteria, and establishing new rules of permeation, identifying new and powerful EPIs seems best approach that can be explored as drug leads or in adjunctive therapies.

Several recent developments in in-silico MDS approaches have enabled the researchers to computationally validate and support the hypothesized mechanisms of EPs and EPIs (Suriyanarayanan and Sarojini, 2015; Ramaswamy et al., 2017; Vargiu et al., 2018). It is now possible to simulate membrane protein complex structures with micro-second time-scales. The MDS approach along with molecular docking and other in silico tools are successfully utilized for screening and prediction of the molecular interactions between potential EPs and their corresponding EPIs, ultimately helping in identification of the potent EPIs particularly from plant origin. However, this field is yet to be explored fully.

Considering the fact that practically all the antibiotics are susceptible to active drug-efflux, use of the potent EPIs to target and block these pumps can help in potentiating the old antibiotics effective again against a range of drug-resistant bacteria. EPIs are being looked as promising adjunctive therapies with the known antibiotics to improve their antibacterial potency at low concentrations, reduce the emergence of AMR and virulence. But developing novel and potent EPI is not easy and needs to overcome several hurdles such as choice of antibiotics for potentiation and matching the pharmacological properties of EPI-antibiotic(s) pair. More comprehensive and

### REFERENCES


deeper investigations are therefore needed that involve the exploring the high-throughput screening assisted by in silico tools for identifying the potent EPI phytomolecules and their corresponding targets. Newer studies are being undertaken for identifying phytomolecules effective in inhibiting bacterial efflux pumps via potentiation of antibiotics against pathogenic bacteria including Gram-negative pathogens. This may pave the way for identification of phyto-EPIs that can head toward clinical phases and ultimately clinical practices with an aim to contain the AMR.

### AUTHOR CONTRIBUTIONS

VK conceived the idea. VS, TK, RB, RS, and VK wrote the manuscript. All the authors made substantial contribution to the work and approved it for publication.

### ACKNOWLEDGMENTS

The support under Star College Scheme of Department of Biotechnology, Government of India implemented at Modern College, Ganeshkhind, Pune is gratefully acknowledged. The use of facilities created under FIST program of Department of Science and Technology (DST), Government of India, implemented at Prof. Ramkrishna More College, Akurdi, Pune and Modern College, Ganeshkhind, Pune are gratefully acknowledged.

implications in drug discovery. Antimicrob. Agents Chemother. 54, 5167–5172. doi: 10.1128/AAC.00610-10


of Streptococcus pneumoniae. Antimicrob. Agents Chemother. 50, 685–693. doi: 10.1128/AAC.50.2.685-693.2006


decreased susceptibility to biocides. J. Antimicrob. Chemother. 65, 228–232. doi: 10.1093/jac/dkp427


expression, purification and characterisation. J. Mol. Microbiol. Biotechnol. 3, 193–200.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Shriram, Khare, Bhagwat, Shukla and Kumar. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Emerging Strategies to Combat ESKAPE Pathogens in the Era of Antimicrobial Resistance: A Review

Mansura S. Mulani, Ekta E. Kamble, Shital N. Kumkar, Madhumita S. Tawre and Karishma R. Pardesi\*

Department of Microbiology, Savitribai Phule Pune University, Pune, India

The acronym ESKAPE includes six nosocomial pathogens that exhibit multidrug resistance and virulence: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp. Persistent use of antibiotics has provoked the emergence of multidrug resistant (MDR) and extensively drug resistant (XDR) bacteria, which render even the most effective drugs ineffective. Extended spectrum β-lactamase (ESBL) and carbapenemase producing Gram negative bacteria have emerged as an important therapeutic challenge. Development of novel therapeutics to treat drug resistant infections, especially those caused by ESKAPE pathogens is the need of the hour. Alternative therapies such as use of antibiotics in combination or with adjuvants, bacteriophages, antimicrobial peptides, nanoparticles, and photodynamic light therapy are widely reported. Many reviews published till date describe these therapies with respect to the various agents used, their dosage details and mechanism of action against MDR pathogens but very few have focused specifically on ESKAPE. The objective of this review is to describe the alternative therapies reported to treat ESKAPE infections, their advantages and limitations, potential application in vivo, and status in clinical trials. The review further highlights the importance of a combinatorial approach, wherein two or more therapies are used in combination in order to overcome their individual limitations, additional studies on which are warranted, before translating them into clinical practice. These advances could possibly give an alternate solution or extend the lifetime of current antimicrobials.

Edited by:

Rebecca Thombre, Modern College of Arts, Science and Commerce, Pune University, India

### Reviewed by:

Vishvanath Tiwari, Central University of Rajasthan, India Rajashree Bhalchandra Patwardhan, Savitribai Phule Pune University, India

> \*Correspondence: Karishma R. Pardesi karishma@unipune.ac.in

### Specialty section:

This article was submitted to Antimicrobials, Resistance and Chemotherapy, a section of the journal Frontiers in Microbiology

Received: 31 May 2018 Accepted: 01 March 2019 Published: 01 April 2019

### Citation:

Mulani MS, Kamble EE, Kumkar SN, Tawre MS and Pardesi KR (2019) Emerging Strategies to Combat ESKAPE Pathogens in the Era of Antimicrobial Resistance: A Review. Front. Microbiol. 10:539. doi: 10.3389/fmicb.2019.00539 Keywords: ESKAPE, multidrug resistance, alternative therapy, combination therapy, phage therapy, antimicrobial peptides, silver nanoparticles, photodynamic light therapy

### INTRODUCTION

Wonder drug penicillin started the era of antibiotics in 1928 and since then it has tremendously developed modern medicine. Persistent use of antibiotics, self-medication, and exposure to infections in hospitals has provoked the emergence of multidrug resistant (MDR) bacteria responsible for 15.5% Hospital Acquired Infection (HAIs) in the world (Rice, 2008; Allegranzi et al., 2011; Ibrahim et al., 2012; Pendleton et al., 2013). The term "ESKAPE" encompasses six such pathogens with growing multidrug resistance and virulence: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp. (Rice, 2008). ESKAPE pathogens are responsible for majority of nosocomial infections and are capable of "escaping" the biocidal action of antimicrobial agents (Rice, 2008; Navidinia, 2016).

A systematic review of clinical and economic impact of antibiotic resistance reveals that, ESKAPE pathogens are associated with the highest risk of mortality thereby resulting in increased health care costs (Founou et al., 2017). World Health Organization (WHO) has also recently listed ESKAPE pathogens in the list of 12 bacteria against which new antibiotics are urgently needed (Tacconelli et al., 2018). They describe three categories of pathogens namely critical, high and medium priority, according to the urgency of need for new antibiotics. Carbapenem resistant A. baumannii and P. aeruginosa along with extended spectrum β-lactamase (ESBL) or carbapenem resistant K. pneumoniae and Enterobacter spp. are listed in the critical priority list of pathogens; whereas, vancomycin resistant E. faecium (VRE) and methicillin and vancomycin resistant S. aureus (MRSA and VRSA) are in the list of high priority group. The mechanisms of multidrug resistance exhibited by ESKAPE are broadly grouped into three categories namely, drug inactivation commonly by an irreversible cleavage catalyzed by an enzyme, modification of the target site where the antibiotic may bind, reduced accumulation of drug either due to reduced permeability or by increased efflux of the drug (Santajit and Indrawattana, 2016). They are also able to form biofilms that physically prevent the immune response cells of host as well as antibiotics to inhibit the pathogen. Moreover, biofilms protect specialized dormant cells called persister cells that are tolerant to antibiotics which cause difficult-to-treat recalcitrant infections (Lewis, 2007).

The general antimicrobial therapy to effectively treat infections involves the use of antibiotics either singly or in combination. With every passing year, the overall number of antibiotics effective against ESKAPE is declining, which is predisposing us toward a future with antibiotics that are ineffective. Analysis of the antibiotic lists recommended in the Clinical & Laboratory Standards Institute (CLSI) guidelines revealed that many antibiotics suggested against ESKAPE since 2010 have been deleted with addition of relatively few antibiotics/antibiotic combinations. Furthermore, there are incidences of resistance reported against some of these newly added antibiotics (**Table 1**). It is, therefore, imperative to find alternative ways to treat infections especially those caused by ESKAPE pathogens.

Alternative therapies that are currently in practice or under trial include the use of antibiotics in combination or with adjuvants, bacteriophage therapy, use of antimicrobial peptides, photodynamic therapy, antibacterial antibodies, phytochemicals and nanoparticles as antibacterial agents (Mandal et al., 2014; Kaur, 2016). There are several reviews published which describe alternative/novel therapies against MDR pathogens, however, not many have specifically focused on the ESKAPE group as a whole. These reviews describe the various therapeutic agents used with respect to their dosage, mode of action, pharmacokinetics and pharmacodynamics, stability, and toxicity. Although many alternative therapies reported have shown promising results in vitro, their efficacy when applied in vivo may not be the same due to one or several limitations. It therefore, becomes necessary to understand the action of these therapeutic agents in vivo. This review highlights therapies used to treat ESKAPE infections namely, antibiotics in combination or with adjuvants, bacteriophage therapy, antimicrobial peptides, photodynamic light therapy, and nanoparticles which have received significant attention in the last 5 years. Hereby, we have also tried to include studies especially focusing on in vivo efficacy of above mentioned therapy/therapeutic agents, their advantages and limitations. **Figure 1** summarizes the major limitations of each alternative therapy which has been elaborately discussed in the following sections. To improve the efficiency and/or minimize the limitations of any therapy/therapeutic agents, combinatorial approaches are suggested. It involves use of two or more monotherapies together (co-administered, functionalized, or conjugated) which is also further described. Altogether, combinatorial approach could be a prospect for exploring novel alternate solutions against ESKAPE.

### ANTIBIOTICS IN COMBINATION

Antibiotics in combination have been tested as a treatment method by a number of researchers because the possibility of a pathogen to develop resistance against a combination of two drugs is much less than that against a single drug. Similarly, the synergistic effect of combined antibiotics is more than that of the individual antibiotic. A combination of drugs also increases the spectrum of coverage (Vazquez-grande and Kumar, 2015) and has been found to be beneficial in severe infections caused by multiple pathogens (Ahmed et al., 2014). Some of these combinations tested against the ESKAPE are listed in **Table 2**. The problem of antibiotic resistance is so severe that it has become necessary to try combinations of the most recently synthesized antibiotics and/or last resort antibiotics to study their potential in antimicrobial therapy. The Gram positive members of the ESKAPE, E. faecium and S. aureus, have been tested against a combination of fosfomycin and daptomycin which has shown to successfully clear infection (Hall Snyder et al., 2016; Coronado-Álvarez et al., 2018). The former is a broad spectrum antibiotic that has shown promising results against Gram negative bacteria while the latter is a last resort antibiotic used to treat infections caused by E. faecium and S. aureus. Despite being a resistant pathogen there is a lack of substantial research in antibiotic combination therapy against E. faecium over the last 5 years. Most combinations tested against S. aureus in vitro include daptomycin or vancomycin with other antibiotics including ceftaroline, a newly added antibiotic to the CLSI guidelines. The effect of these and other such combinations have also been tested in various mouse models which cleared away the S. aureus infection with minimal to no toxicity. Colistin (polymyxin E) is the last resort antibiotic prescribed against Gram negative bacilli. In recent years, research has been conducted to treat infections caused by K. pneumoniae and A. baumannii using a combination of colistin or tigecycline with other antibiotics and has shown promising results in vitro and in cohort studies.

Some molecules when combined with antibiotics make an ineffective drug effective. These molecules, named "adjuvants" or "resistance breakers," have little to no antimicrobial activity of their own (González-Bello, 2017) but inhibit the mechanism

### TABLE 1 | Antibiotics added/revised and eliminated against ESKAPE from CLSI document M100 since 2010.


0 = Antibiotics deleted from CLSI guidelines between 2010 and 2018; 1, New antibiotics added in the CLSI guidelines since 2010; 1 = No resistance reported till date; 1 = Resistance reported; Ef, E. faecium; S, S. aureus; K, K. pneumoniae; A, A. baumannii; P, P. aeruginosa; E= Enterobacter spp.

S = (Long et al., 2014; Chan et al., 2015; Nigo et al., 2017); K = (Zowawi et al., 2015; Vuotto et al., 2017; Stanley et al., 2018); A = (Göttig et al., 2014; Goic-Barisic et al., 2017; Nowak et al., 2017; Caio et al., 2018; Chuang and Ratnayake, 2018); P = (Prakash et al., 2014; Gill et al., 2016; Alipour et al., 2017; Mohapatra et al., 2018; Palavutitotai et al., 2018); E = (Lee et al., 2015; Babouee Flury et al., 2016; Zeng et al., 2016; Kulengowski et al., 2018).

of resistance by increasing the uptake of the antibiotic through the bacterial membrane, blocking of efflux pumps, and changing the physiology of resistant cells (i.e., dispersal of biofilms to planktonic cells) (Kalan and Wright, 2011; Bernal et al., 2013). Essential oils and phenothiazines enhance the antimicrobial activity of drugs and also inhibit the transmission of resistance to other populations (Bueno, 2016).

The most popularly known adjuvants are β-lactamase inhibitors while the most recent adjuvant tested to restore meropenem activity is vaborbactam which inhibits K. pneumoniae carbapenemase activity (Jorgensen and Rybak, 2018). It has also undergone a clinical trial; registered on ClinicalTrials.gov under identifier NCT02020434; which proved the combination to be safe after testing 41 subjects (Rubino et al., 2018). However, it has a limited inhibition activity as it has been unable to do so with class B and class D β-lactamases. Other β-lactamase inhibitors include avibactam, nacubactam, and tazobactam (Monogue et al., 2018a,b). Metal chelators like EDTA, deferasirox, and deferoxamine also inhibit β-lactamases as these enzymes require metal ions for their activity (Aoki et al., 2010; Santos et al., 2012; Yoshizumi et al., 2013). These chelators have been tested in combination with antibiotics like imipenem, tobramycin, and vancomycin against S. aureus, P. aeruginosa, and E. coli in murine models with successful decrease in bacterial load. Quorum quenchers, molecules that inhibit quorum sensing thereby inhibiting biofilm formation, have also shown potential to cure infection in combination with certain antibiotics (Balamurugan et al., 2015; Chatterjee et al., 2016). 1-[(2,4-Dichlorophenethyl)amino]-3-Phenoxypropan-2-ol is so far the most promising antimicrobial agent as it has been reported to kill not only persister cells of P. aeruginosa but also non-persister cells of all the other ESKAPE members. It has also been shown to enhance killing of antibiotic resistant strains in both planktonic and biofilm forms. Its combination with a variety of antibiotics is shown to kill all of the ESKAPE making it an ideal candidate as an adjuvant (Defraine et al., 2017).

Even though there is increased activity of antibiotics when used in combination against pathogens in vitro, there are limited studies demonstrating the same in vivo and some among those have proven disadvantageous. If monotherapy selects for a narrow spectrum of resistance, a combination of two or more antibiotics selects for a broad spectrum of resistance defeating the purpose of combination therapy entirely (Vestergaard et al., 2016). Certain combinations that are meant to treat infections tend to have the opposite effect resulting in far worse damage. One antibiotic can lead to the induction of a resistance mechanism against a second antibiotic administered in combination leading to antagonistic effect (Fallah, 2018). A clinical trial conducted in Italy in which infections caused by XDR A. baumannii were treated with a combination of colistin and rifampin showed no improvement in survival rates. In fact, this combination led

to increased hepatic toxicity (Durante-Mangoni et al., 2013). A similar study using a combination of colistin, tigecycline and carbapenems against A. baumannii showed futile results (López-Cortés et al., 2014). Metal chelators have shown to sequester ions not only from bacterial cells but also from host tissue cells (Yoshizumi et al., 2013).

The most recent alternative to antibiotics or their combination that shows a promising future is antibiotic hybrids which have been defined by Domalaon et al. (2018) as synthetic constructs of two or more pharmacophores belonging to an established agent known to elicit a desired antimicrobial effect. This method provides the advantage of combination therapy through the mono therapy approach, where chances of resistance are curbed, while overcoming the problem of non-complementary pharmacodynamic profiles of the individual antibiotics.

The ESKAPE tend to become resistant to either or both antibiotics used in combination with every passing year due not only to natural selection of resistant strains but also horizontal gene transfer from them to sensitive strains. This warrants testing of still new combinations. The result is a never-ending cycle from which there is no escape. It can therefore be concluded that antibiotics in combination may not always be effective and that there is a need for extensive research of alternative strategies.

### BACTERIOPHAGE THERAPY

Phages are century old therapeutic agents that were used for the treatment of bacterial infections. The discovery of antibiotics was an influential factor in side-lining this ambition (Mann, 2005). The focus on phage therapy has sharpened ever since antimicrobial resistance has been on a dramatic rise. Lytic phages against ESKAPE pathogens have been isolated from hospital wastewater, making them easily available therapeutic agents (Latz et al., 2016). Bacteriophages used for therapy present many advantages such as high host specificity (prevent damage to normal flora, do not infect the eukaryotic cells), low dosages for treatment, rapid proliferation inside the host bacteria, etc. that make them ideal candidates to treat bacterial infections (Domingo-Calap and Delgado-Martínez, 2018). Unlike antibiotics, the advantage of using phages is that, they develop new infectivity and regain an upper hand over bacteria as they mutate alongside their host (Pirnay et al., 2018).

Several studies carried out in vitro have proven phages to be effective as antibacterial agents against biofilm and planktonic cells of ESKAPE (Pallavali et al., 2017; Dvoˇrácková et ˇ al., 2018; Jamal et al., 2019). **Table 2** gives information of phage therapy studied in various animal models as well as recent case studies and case reports of patients infected with ESKAPE pathogens. Phage therapy carried out in animal wound infection models have shown reduced mortality and enhanced wound healing. Additional studies carried out in vivo have also demonstrated efficacy and safety (non-toxic with reduced inflammatory responses) of phages used in treatment of bacterial infections.

Phage therapy, though promising, comes with some limitations. It can, however, be overcome by appropriate modifications (Wittebole et al., 2014; El-Shibiny and El-Sahhar,


TABLE

2


Alternative

 strategies against ESKAPE pathogens.



(Continued)

TABLE

2


Continued


TABLE

2


Continued

2017; Domingo-Calap and Delgado-Martínez, 2018). High specificity of the phages can be considered as both advantageous and a limiting factor. Monophage therapy involves the need to check the efficacy of the phage by testing it in vitro against the disease-causing bacteria before applying it to a patient which can be a difficult process. The use of phage cocktails, comprising of a combination of phages acting against different bacterial species or strains, can avoid these problems (Chan et al., 2013). International experts believe that an ideal phage cocktail should be prepared using phages belonging to different families or groups such as having broad host range, high adsorption ability to the highly conserved cell wall structures in bacteria. Using such phage cocktails may reduce the emergence of phage resistant bacterial population. However, others advocate strategies wherein individual active phages are applied sequentially to the patient. In clinical practice, however, it appears to be difficult (Rohde et al., 2018).

Genomic characterization of phages is very important so as to predict their "safety" in therapeutic applications as demonstrated by several experts in this field. Phages can be vectors for horizontal gene transfer in bacteria, sometimes being involved in exchange of virulence or antibiotic resistance genes making a microbe more pathogenic or resistant to an antibiotic (Chen and Novick, 2009). Phages reported for therapeutic applications should not harbor virulence or antibiotic resistance genes as well as integrases, site-specific recombinases, and repressors of the lytic cycle that may accelerate the transfer/integration of these genes in the host bacterial genome. Algorithms that can be used for predicting lifestyle of a phage, and its virulent traits are available but their database needs to be updated with more genome sequences of phages (Mcnair et al., 2012). Two recent reviews excellently describe the work flow to select ideal phage candidates for therapeutic purposes (Casey et al., 2018; Philipson et al., 2018). Recent studies demonstrating in vivo efficacy of phages against ESKAPE infections have used fully characterized phages that show no virulence or antibiotic resistance genes, are considered safe as they do not exhibit any allergic or immune response, and are also reported to remain stable at varied pH and temperature which make them ideal candidates for therapy (Fish et al., 2016; Kishor et al., 2016; Wang et al., 2016; Zhou et al., 2018).

Similarly, it has also been reported that the bacterial strains used for phage production should ideally be free of functional prophages. These prophages may get induced and contaminate the phage preparation. However, a recent report discusses the risk benefit evaluation that needs to be done in highly experimental treatments of patients infected with MDR pathogens such as ESKAPE (Rohde et al., 2018).

Another limitation reported is the stability of phages and their proper administration in order to reach the site of action. Phage formulations are ingested orally, administered nasally or applied topically (Malik et al., 2017; Cooper et al., 2018). Studies have shown improved efficacy of phage when entrapped with liposomes (Singla et al., 2016; Chadha et al., 2017; Malik et al., 2017; Chhibber et al., 2018). Phages can be targeted at the infection site in the form of powdered formulations (Chang et al., 2018). Phage derived product like phage encoded lytic enzymes showing function similar to lysozyme can also be used as an antibacterial agent or can be combined with other antimicrobials like antibiotics to improve efficacy of treatment (Lin et al., 2017). A phage derived protein, "endolysin" is reported for its antibacterial and antibiofilm activity against ESKAPE (Viertel et al., 2014; Gong et al., 2016; Rios et al., 2016; Lin et al., 2017; Zhang et al., 2018). V12CBD a recombinant protein derived from bacteriophage lysine, PlyV12, was able to attenuate virulence of S. aureus, and enhance its phagocytosis in mice (Yang et al., 2018b).

Several commercial phage preparations which can be used against ESKAPE are available some of which include, "Stafal," "Sextaphage," "PhagoBioDerm," and "Pyophage". Stafal (Bohemia Pharmaceuticals, Slovakia) is an antistaphylococcal phage preparation, Sextaphage (Microgen, ImBio Nizhny Novgorod, Russia) is a cocktail against P. aeruginosa and E. coli while, Pyophage (Georgian Eliava Institute of Bacteriophage, Microbiology, and Virology) contains bacteriophages specifically eliminating causative agents of pyoinflammatory and enteric diseases. PhagoBioDerm, a polymeric bandage impregnated with cocktail of phages, ciprofloxacin, and other active ingredients ensured for sustained release of phages to treat ulcers or wound infections caused by S. aureus and P. aeruginosa (Markoishvili et al., 2002). Clinical potential of these preparations has been investigated further to determine their broad spectrum activity against other strains in vitro, in vivo efficacy in animal models as well as through several case studies or clinical trials described below.

Recent human case studies involving treatment of ESKAPE associated infections with different phages are described in **Tables 2**, **3**. Readers may refer to previous reviews published on phage therapy which have described in numerous studies (Abedon et al., 2011; Chan et al., 2013; Górski et al., 2017; Lin et al., 2017; Sybesma et al., 2018). Most of the case studies report phage therapy given to patients on a compassionate care basis where antibiotic treatment fails.

Several case studies have demonstrated the efficacy of phage therapy in treating patients suffering from eye infections (Fadlallah et al., 2015), pancreatitis (Schooley et al., 2017), diabetic foot ulcer (Fish et al., 2016), and urinary tract infection (Ujmajuridze et al., 2018). Fadlallah et al. (2015), reported a case study of a 65 year old woman suffering from a secondary eye infection by VRSA, treated with a well-characterized commercially available lytic phage, SATA-8505 (ATCC PTA-9476) against MRSA. After 6 months of treatment, the patient was diagnosed as negative for cultures of VRSA. This case study suggests that bacteriophage eyedrops can be used as an alternative treatment of infectious keratitis by MDR pathogens. Fish et al. (2016), reported a case involving nine patients suffering from diabetic foot ulcer, who were treated with a preparation of anti-staphylococcal phages (Sb-1) on a compassionate care basis. The average healing time reported was ∼7 weeks. However, one ulcer, with poor vascularity required 18 weeks of treatment. Similarly, Ujmajuridze et al. (2018), reported a preclinical pilot study for a double blind RCT registered at ClinicalTrials: NCT03140085 (Leitner et al., 2017). Pyophage were used in this study to treat nine patients suffering from UTI infections. The first phase


**289**



**291**

of this study involved adaptation cycles of the commercial Pyophage preparation to increase its sensitivity toward the uropathogens. In the second phase, six of the nine patients responded to the adapted bacteriophage showing up to five log reduction in CFU of the infecting bacteria. Another study by Schooley et al. (2017) described the case of a 68 year old male diabetic patient having necrotizing pancreatitis which was complicated by an MDR A. baumannii. The patient responded to a phage therapy which consisted of a cocktail of nine phages against A. baumannii.

Few cases showing partial success of phage therapy in treating the infection are also reported (Jennes et al., 2017; Duplessis et al., 2018). Jennes et al. (2017) reported a case study of 61 year-old man diagnosed with septicaemia caused by a colistinonly-sensitive P. aeruginosa. Blood cultures turned negative immediately after BFC1 phage therapy but sores remained infected. No adverse side effects were observed during this therapy. In another study, a 2-year-old boy with a history of DiGeorge syndrome and complex congenital heart disease developed post-operative recalcitrant P. aeruginosa bacteraemia. A cocktail of two phages were used to target the infecting pathogen turning the blood cultures negative. However, the bacteraemia re-occurred after the discontinuation of phage therapy (Duplessis et al., 2018). It is however felt that commonly approved guidelines for application of phages is warranted in order to compare the efficacies of various phage treatments.

Several clinical trials have demonstrated the safety of phages and phage lytic enzymes which are in agreement with the studies carried out in animal models or as reported in numerous case studies. Although, many phase I and/or phase II clinical trials to demonstrate the efficacy of phages against ESKAPE infections have been registered in the past few years, the number of well-documented and completed trials are too low to draw meaningful conclusions (Sybesma et al., 2018). Moreover, the number of patients enrolled for the trials have severely limited the conclusions. A notable example of this is "PhagoBurn," a multicentric randomized single blind and controlled clinical study on phage therapy to treat burn wound infections caused by P. aeruginosa and E. coli (ClinicalTrials.gov registry, NCT02116010). Twenty seven patients were enrolled for the trial which resulted in being too low from the precalculated 220 patients expected to give statistically significant results. Restrictive patient inclusion criteria, shorter duration of patient enrolment, and low incidence of burn wound infection during the period of study were described as possible reasons for the very few eligible patients (Servick, 2016). The 27 patients were randomly assigned to the phage treatment group and safety control group for further investigation. A cocktail of 12 phages (PP1131) with lytic activity against P. aeruginosa were added to an alginate template that was applied directly to the wound of the treated group. A diluted phage cocktail was used due to its high endotoxin content. The control group received a topical application of the standard care treatment (1% sulfadiazine silver). Few adverse events were reported both in phage treated and safety control group. The trial, however, was terminated prematurely due to insufficient efficacy of PP1131. A supporting study showed that the bacteria isolated from patients with the failed treatment were resistant to low phage doses used for this study. This, however, was the first Randomised Clinical Trial (RCT) performed following both good manufacturing practices as well as good clinical practices and was approved by three national health regulators (Belgium, France, and Switzerland). Further studies with increased phage concentration and higher number of participants are needed (Jault et al., 2019). In case of phage lytic enzymes, the first-in-human phase I clinical trial study for phage endolysin-based candidate drug SAL200, after a single intravenous administration among healthy volunteers showed no clinically significant detrimental effect (Youn Jun et al., 2017).

The other recent ongoing trial includes a randomized placebo controlled double blind clinical trial (ClinicalTrials.gov registry, NCT03140085) to study the efficacy and safety of commercially available Pyophages for treating UTI infections in patients planned for transurethral resection of prostate. Eighty-one patients were expected to enroll for this interventional study design planned between Nov 2015–Dec 2018 that may provide necessary insights into this potentially transforming alternative treatment option (Leitner et al., 2017). Similarly, another multicentric, randomized, two-parallel group, double blind controlled trial expected to enroll 60 patients was registered (ClinicalTrials.gov registry, NCT02664740) to compare efficacy of anti-staphylococcal phages verses standard treatment and placebo for diabetic foot ulcers infected by S. aureus. This study has not yet begun recruiting patients.

To further overcome limitations of phage therapy, phages can be combined with antibiotics which may show synergistic action by making either a phage or antibiotic or both to act more effectively (**Table 3**). Reduction in the formation of bacterial biofilms has also been reported when antibiotic treatment is used in combination with phages (Jo et al., 2016; Chaudhry et al., 2017). Endolysin produced by bacteriophages, proves to be more beneficial than lysing cell wall and facilitating entry of the antibiotic inside the bacteria (Rios et al., 2016). Phage PEV20 and ciprofloxacin exhibited a synergistic effect in vitro against P. aeruginosa (Lin Y. et al., 2018). Another interesting study reported that P. aeruginosa, while developing phage resistance when under attack by OMKO1 phage, changed the efflux pump mechanism which ultimately increased the sensitivity of P. aeruginosa to antibiotics. Such an approach creates a win-win situation causing the killing of the bacteria either through phage or by antibiotic action (Chan et al., 2016). Similarly, it has been reviewed that such changes in the surface receptors of any bacterium may also reduce its virulence (Rios et al., 2016). A case study reported OMKO1 phage in combination with ceftazidime could successfully treat a complicated postoperative P. aeruginosa infection in a patient who underwent an aortic arch replacement surgery (Chan et al., 2018). The success or failure of phage antibiotic combination therapy is still in a state of immaturity as the mechanisms involved in synergy have not been fully understood and the data of in vivo models and case reports is scarce. Further investigations are warranted in order to make any conclusive remarks.

## ANTIMICROBIAL PEPTIDES (AMPs) IN THERAPY

Antimicrobial peptides (AMPs) are short, positively charged, diverse host defense oligopeptides produced by all living forms including protozoa, bacteria, archaea, fungi, plants, and animals (Wang et al., 2010). They show a broad spectrum of activity against a wide range of pathogens. The capacity of AMPs to interact with bacterial cell membrane and thereby cause cell lysis makes them a potential alternative to combat MDR pathogens (Berglund et al., 2015). Furthermore, in contrast to conventional antibiotics, AMPs physically damage the bacterial cell through electrostatic interactions thereby making it difficult for bacteria to develop resistance against AMPs (Pfalzgraff et al., 2018).

Considering the critical status of ESKAPE pathogens, several attempts have been made to find out AMP based effective therapeutics. To date, there are numerous natural as well as bioengineered AMPs reported to show in vitro (Björn et al., 2016; Cappiello et al., 2016; Liu et al., 2017; Gandt et al., 2018; Irani et al., 2018; Téllez et al., 2018) as well as in vivo (Björn et al., 2016; Liu et al., 2017) antimicrobial, antibiofilm, anti-inflammatory, and wound healing abilities with minimum cytotoxicity. Histatin 5, a natural histidine rich cationic human salivary peptide, is an example. This peptide shows a strong in vitro anti-biofilm as well as potent bactericidal activity (≥70%) against ESKAPE (Du et al., 2017). Similarly, a de novo-engineered cationic peptide, WLBU-2, and a natural AMP LL-37 at 1/3X MIC has demonstrated 90% biofilm inhibition as compared to that shown by antibiotics such as tobramycin, ciprofloxacin, ceftazidime, and vancomycin at 1X MIC (Lin Q. et al., 2018). In 2017, Gaglione et al. examined human ApoB derived recombinant peptides namely r(P)ApoB<sup>L</sup> and r(P)ApoBS. Both peptides showed effective in vitro wound healing, anti-inflammatory, antimicrobial, and antibiofilm properties against MDR strains of S. aureus and P. aeruginosa.

Similar to their remarkable in vitro properties, AMPs also exhibit promising in vivo activity against ESKAPE. For example, peptide HLR1r, a structural derivative of human milk protein, lactoferrin, at very low concentration (5 mg/kg) was found to show anti-infectivity against MRSA infected wound excision model in rat along with in vitro anti-inflammatory and noncytotoxic effects suggesting use of HLR1r in topical formulation to treat skin infections (Björn et al., 2016). PT-13 a peptide derived from seeds and leaves of Populus trichocarpa crude extract also demonstrated effective in vivo antibacterial activity in S. aureus infected G. mellonella model (Al Akeel et al., 2018). In another instance, a synthetic analog of Feleucin-K3 has shown to clear P. aeruginosa induced bacteremia in mice model with good stability and very low cytotoxicity (Xie et al., 2018). Also, a hydrogelformulation containing K-11, a hybrid peptide of melittin, cecropin A1 and magainin-2 has shown to possess wound healing ability against A. baumannii infected murine excision model proposing its possible use as a topical anti-infective therapeutic agent (Rishi et al., 2018).

Over the past decades, intense efforts taken by the scientific community and pharmaceutical industries together has made it possible to introduce certain peptides such as vancomycin, telavancin, telaprevir, teicoplanin, enfuvirtide, daptomycin, dalbavancin, bacitracin etc. for clinical use (Gomes et al., 2018). A clinical trial conducted on rabbits and humans using the peptide melamine proved it to be a stable and safe antibacterial coating for eye lenses (Dutta et al., 2014). Similarly, Pexiganan (analog of magainin), LL-37 (analog of human cathelicidin peptide), hLF1- 11, and PXL-01 (derivatives of human milk protein), Novexatin (derivative of human defense peptide), Iseganan (derivative of porcine leukocytes), PAC-113 (derivative of human saliva histatin-3 peptide) etc. are few examples of AMPs which also are under clinical trials (Mahlapuu et al., 2016).

Unfortunately, such a low number of AMPs seeking clinical approval is quite discouraging. Despite their successful in vitro and/or in vivo broad-spectrum activities, numerous AMPs have not yet crossed the hurdle of clinical trial. Amongst the few challenges that hamper the in vivo efficacy of AMPs are their cytotoxicity to mammalian cells, liability to degradation by tissue proteases, loss of activity at low salt concentrations or in presence of plasma proteins and higher production cost (Mahlapuu et al., 2016; Rios et al., 2016). The issue of peptide degradation can be solved by structural modification of AMPs such as addition of non-natural amino acids or their D-isomers, peptide cyclisation, acetylation, and amidation of N-terminus. Introduction of peptide mimetics or the use of suitable delivery system like liposome encapsulation can be done to improve their stability and reduce toxicity (Seo et al., 2012; Reinhardt and Neundorf, 2016). Additionally, efficiency of AMPs can be enhanced by combining AMPs with antibiotics (Gaglione et al., 2017; Zheng et al., 2017; Pletzer et al., 2018) or nanoparticles (Chaudhari et al., 2016; Kuo et al., 2016). Otvos et al. (2018), reported a synergestic effect of A3-APO, a proline-rich AMP, and colistin when studied in a K. pneumoniae infected bacteremia mice model. The basis for in vitro decrease in MIC of A3-APO when combined with colistin can be explained by considering the fact that colistin kills bacteria by interfering with bacterial membrane assembly and, therefore, slight reductions in bacterial membrane integrity potentiate A3- APO antibacterial action. Surprisingly, the same combination also showed a 100% survival in mice. This observation can be a direct consequence of A3-APO ability to induce immune augmentation or the deactivation of bacterial toxins. Enhanced in vitro bactericidal activity against S. aureus was also found in the case where LL-37, a human cathelicidin peptide, combined with gold nanoparticles as compared to vancomycin alone (Wang et al., 2018). In this case, gold nanoparticles increased the local density of positive charges and peptide mass and thereby enhanced the bactericidal properties of LL-37.

To summarize, owing to their in vitro and in vivo broadspectrum antibacterial activities AMPs offer a hopeful alternative to conventional therapeutics. However, to overcome challenges in developing a safe, stable and efficient commercial product, a thorough understanding of their structure and interaction with bacterial as well as host cells is still needed. It will also be helpful to find better AMP formulation strategies to obtain maximum therapeutic actions. Overall, considering the extensive research being carried out on different AMPs against various infectious

agents, the future of peptide based commercial drug formulations looks hopeful.

## PHOTODYNAMIC LIGHT THERAPY

Antimicrobial light therapy, either alone or combined with a photosensitizer (PS), results in a photooxidative stress response that leads to microbial death. Excitation of PS with light of an appropriate wavelength leads to formation of an excited triplet state. An excited PS can transfer electrons or energy to biomolecules or molecular oxygen, resulting in the formation of reactive oxygen species (ROS) or singlet oxygen radicals, which are toxic to cellular targets such as nucleic acids, proteins and lipids (Mai et al., 2017; Yang M.-Y. et al., 2018). Some of the most frequently used PSs include phenothiazinium derivatives (methylene blue, toluidine blue), xanthine derivatives (rose bengal), porphyrin, chlorin, or fullerene derivatives amongst many others (Abrahamse and Hamblin, 2016; Cieplik et al., 2018). Antimicrobial photodynamic therapy is widely used for treating dental, skin, and soft tissue infections. For a more detailed description of the current state and future prospects of light therapy with respect to the various photosensitisers, light sources, and methods used, mechanism of antimicrobial action or antibiofilm potential, the reader may be referred to the excellent reviews published recently (Cieplik et al., 2018; Hu et al., 2018; Tomb et al., 2018; Wozniak and Grinholc, 2018). However, none of these reviews have especially focused on in vivo studies of aPDT against ESKAPE pathogens.

There has been extensive research on designing the PSs so as to improve their pharmaceutical potential. An ideal PS used for antimicrobial therapy should have greater permeability to cross the microbial cell wall/cell membrane, selective toxicity toward the microbial cell with minimal or no damage to the host tissue and an absorption coefficient appropriate for effective penetration at the site of action. The PS chosen should not have a long half-life which causes prolonged photosensitization in the host cells even after the infection is cured. It should also not be effluxed out by the microbial efflux systems (Cieplik et al., 2018; Hu et al., 2018; Tomb et al., 2018; Wozniak and Grinholc, 2018). Efficacy of aPDT also depends on the light fluence, PS concentration and treatment time (Tomb et al., 2017; Sueoka et al., 2018; Ullah et al., 2018).

PSs chosen preferably have a large absorption coefficient in the visible spectrum, especially in the long wavelength (red near infrared) region, to allow effective penetration of light in the infected tissue (**Table 2**). Many researchers have attempted to improve the availability of PS by potentiating or functionalizing it with other molecules including galactose, amino acids, efflux pump inhibitors, potassium iodide, EDTA etc. A variety of PSs functionalized with addends are used to target ESKAPE pathogens. A boron-dipyrrolemethene (BODIPY) based polygalactose, named pGEMA-I (7.3 kDa) with increased water solubility was used to demonstrate antibacterial and antibiofilm activity against P. aeruginosa, without much affecting the viability of normal cells. It was demonstrated that the selective recognition of the pathogen was due its carbohydrate binding lectin protein (LecA) which interacted with the galactose moiety of the PS (Zhao et al., 2018). C60-fullerene (LC16) bearing deca-quaternary chain and deca-tertiary-amino groups facilitates electron-transfer reactions via the photoexcited fullerene for antimicrobial effect studied in A. baumannii and S. aureus (Huang et al., 2014; Zhang et al., 2015). Another drawback of aPDT is that the ROS generation may cease after the light irradiation is turned off thus allowing un-killed bacteria to re-grow. Potentiating aPDT with potassium iodide allows the formation of iodine/tri-iodide that may remain active in the wound for a longer duration sufficient enough to prevent bacterial re-growth (Zhang et al., 2015; Wen et al., 2017).

In vitro studies have shown that blue light (aBL) has a broad spectrum antibacterial and antibiofilm activity against all six ESKAPE members (Halstead et al., 2016). In vivo data also corroborated this finding and further confirmed that using a low penetrating blue light of 415 ± 10 nm should be a preferred choice of treatment in case of topical wound infections as it causes minimal damage to the uninfected tissue cells below (Amin et al., 2016; Wang et al., 2017; Katayama et al., 2018). Some studies additionally report that an exogenous PS may not be required (Amin et al., 2016; Wang et al., 2017). Their finding was supported by experimental data showing that the endogenous porphyrins present in the bacterial cell membrane play a role in triggering the photoxidative response (Amin et al., 2016). aBL using 5-aminolevulinic acid with disodium EDTA (ALA-EDTA/2Na) had antibacterial and antibiofilm potential thus showing significant wound healing of P. aeruginosa infected cutaneous ulcers in mice model (Katayama et al., 2018). However, the role of EDTA in increasing the antibacterial action of aBL needs further investigation. The physiological mechanism behind wound healing was studied in a S. aureus infected burn model in mice revealing that the enhanced levels of factors promoting angiogenesis and epithelial regeneration (bFGF, TGFβ1, and VEGF) led to the inhibition of inflammatory factors (TNFα and IL6) in the aPDT treated group as compared to the control (Mai et al., 2017).

Sueoka et al. (2018) studied the time dependant effect of aPDT-TON 504 on P. aeruginosa and showed that a repeated exposure to light emitting diode (LED) enhanced the inhibitory effect on bacterial growth. This enhanced effect was possibly because the bacteria that survived the initial aPDT were injured by singlet oxygen generated due to excitation of remaining photosensitizer. On the contrary there are also a few reports investigating the development of resistance/tolerance to aBL. It was observed that initial exposure to low doses of aBL increased the tolerance of methicillin susceptible S. aureus (MSSA) to subsequent doses of high intensity aBL-405 nm. It is likely that increased tolerance to high intensity light may be due to upregulation of bacterial stress responses which needs detailed investigations. However, a second set of experiment showed that repeated sub-lethal exposures of 405 nm light indicate no evidence of tolerance in S. aureus (Tomb et al., 2017).

A considerable amount of literature has been published on combined efficacy of aPDT and antibiotics demonstrated in vitro (El-Azizi and Khardori, 2016; Ronqui et al., 2016). Synergistic effects of aPDT-antibiotic combination resulted in inactivation of several virulence factors in P. aeruginosa isolates (Fila et al., 2017). aPDT when used in combination with vancomycin prolonged the survival of Galleria mellonella infected with a vancomycin resistant strain of E. faecium as compared to either of the two therapies used alone (Chibebe Junior et al., 2013). Wozniak and Grinholc (2018) in their comprehensive review analysis have, however, pointed out that most of the aPDT carried out in combination with antibiotics lack a standard methodology followed to evaluate the synergistic effect.

One of the strategies for improved PS delivery involves the use of nanoparticles which are co-administered to allow the PS entry across the membrane or for a synergistic ROS response resulting in an antimicrobial action. A combination of tetrasulfonated hydroxyl aluminum phthalocyanine (AlPcS4) and bimetallic gold/silver nanoparticles (Au/Ag-NPs) synthesized using a cell-free filtrate of Aureobasidium pullulans showed significantly higher killing as compared to the agents used individually (Maliszewska et al., 2018). Au/Ag-NPs possibly disrupted the cell membrane allowing enhanced uptake of the PS, AlPcS4. In vitro and in vivo studies demonstrated that AgNPs used in combination with blue light showed synergistic antimicrobial and antibiofilm activities against P. aeruginosa infection. Interestingly, this combination was also effective for the treatment of a chronic wound caused by mixed infection in a horse (Nour El Din et al., 2016). More recently, an in vitro study carried out using A. baumannii isolates, demonstrated that the antibacterial action of ZnO-NPs with blue light irradiation was due to their ability to damage the cytomembrane but not DNA (Yang M.-Y. et al., 2018).

Over expressing multidrug efflux pumps, commonly found in resistant pathogens, are reported to affect the intracellular concentration of PS used in aPDT, thus limiting its action (Tegos and Hamblin, 2006). Tegos et al. (2008) carried out in vitro experiments by co-incubating various combinations of PS with efflux pump inhibitors to select the best combination showing antibacterial activity. Photodynamic killing mediated by toluidine blue (TBO), when used in combination with PaβN, or INF271 (as EPIs) was most effective against P. aeruginosa and S. aureus isolates, respectively. Similarly, an in vitro study demonstrated that verapamil, an efflux pump inhibitor, when combined with aPDT, required a lower light dose for effective antibacterial, and antibiofilm action against S. aureus (de Aguiar Coletti et al., 2017). In further development of such an approach, it was recently demonstrated that INF55-(Ac)en–MB, synthetic antimicrobial hybrids designed by covalently linking a PS (methylene blue) to efflux pump inhibitors (INF55 and INF271) were more effective in treating wound infections caused by S. aureus or A. baumannii studied in mice models (Rineh et al., 2017, 2018).

It is also worth noting that synergistic effect of aPDT when used in combination with antimicrobial peptide was also demonstrated recently (de Freitas et al., 2018; Nakonieczna et al., 2018). Aurein 1.2 augmented the aPDT activity mediated by methylene blue or chlorin-e6 against strains of S. aureus, A. baumannii and more importantly against vancomycin resistant E. faecium, whereas the AMP aPDT combination with curcumin (as PS) had no effect thus revealing a PS-dependent mechanism (de Freitas et al., 2018). Later, Nakonieczna et al. (2018) carried out a similar study showing synergistic effect of rose bengalaPDT with two synthetic AMPs, CAMEL, and pexiganan against 35 isolates of P. aeruginosa. Notably, it was also shown that this combination was non-toxic to human keratinocytes. Conjugates such as ZnPc(Lys)<sup>5</sup> (a zinc phthalocyanine derivative coupled with pentalysine) also showed a synergistic antibacterial action which was sufficient to heal S. aureus wound infection in mice models (Ullah et al., 2018).

Overall, photodynamic therapy appears to be a promising option for treatment of infections caused due to ESKAPE pathogens, particularly effective in topical applications. aPDT co administered or conjugated with antibiotics, antimicrobial peptides, nanoparticles, or efflux pump inhibitors show a synergistic effect. However, it is difficult to compare efficacy between different combinatorial approaches due to lack of uniform methodologies. More studies on investigation of toxicity and biocompatibility of various combinations should be investigated using in vivo models for translating them into clinical practice.

## SILVER NANOPARTICLES IN THERAPY

Nanomedicine is one of the emerging branches for treating drug resistant pathogens. Metal nanoparticles have wide biomedical applications as antimicrobial agents due to their unique physical and chemical properties (Beyth et al., 2015; Hemeg, 2017). Amongst metal nanoparticles, silver nanoparticles (AgNPs) synthesized using physical, chemical or biological methods have shown promising antibacterial activity due to their multi-targeted approach which reduces the probability of resistance (Möhler et al., 2018; Siddiqi et al., 2018). AgNPs act by releasing Ag<sup>+</sup> ions which results in disruption of electron transport or signal transduction pathway or leads to generation of ROS, ultimately damaging important biomolecules such as cell wall, cell membrane, cellular DNA, and/or proteins (Dakal et al., 2016; Qayyum et al., 2017).

AgNPs act by inhibiting or disrupting planktonic cells as well as biofilms of MDR pathogens. Even though earlier reports have suggested the cytotoxic effects of AgNPs (Mohanty et al., 2012), recently in vitro and in vivo studies have demonstrated the safe usage of AgNPs (Möhler et al., 2018). AgNPs synthesized using aqueous leaf extract of Corchorus capsularis exhibited antibacterial activity against S. aureus and P. aeruginosa and were found to be non-toxic to mouse fibroblast cells (Kasithevar et al., 2017). Electrochemically synthesized AgNPs showing antimicrobial activity against planktonic and biofilm forming P. aeruginosa strain were non-toxic to G. mellonella larvae model (Pompilio et al., 2018). Recently, sunlight mediated AgNPs synthesized using Capsicum annuum was tested in S. aureus infected zebra fish model which proved to be effective in inhibiting biofilm formation. Histological studies revealed that they are non-toxic and hence can be tested for efficacy in higher mammalian in vivo models (Lotha et al., 2018). A singleblind clinical trial (Clinical Trial Registration: NCT01243320 and NCT01405794) carried out in 60 healthy human volunteers, showed that commercial AgNPs when administered orally at dose of 10 and 32 ppm and monitored over 14 days were found to be non-toxic. The study revealed no significant changes in metabolic, hematologic and pro-inflammatory responses as well as no morphological changes in vital organs (Munger et al., 2014).

One of the widely explored applications of AgNPs is their use in the form of composite dressings or hydrogels for treatment of topical wound infections. AgNPs incorporated in chitosan composite dressings offer sustained release of Ag<sup>+</sup> ions at low dosage which are non-toxic to fibroblast cells. Studies in mice models suggested that AgNPs/chitosan composite dressings and low molecular weight chitosan-coated silver nanoparticles were effective in reducing bacterial load, were non-toxic and biocompatible, had low absorption in body and promoted better wound healing against S. aureus and P. aeruginosa (Liang et al., 2016; Peng et al., 2017). Similarly, studies using three other polymer dressings made of chitosan, nylon, and collagen incorporated with AgNPs exhibited in vitro antibacterial activities against ESKAPE pathogens (Radulescu et al., 2016; Rath et al., 2016; Ding et al., 2017). These dressings did not exhibit inflammatory responses, showed re-epithelization of cells and better wound contraction leading to accelerated wound healing in mice models. Sodium carboxymethyl cellulose hydrogel loaded with polyethylene glycol coated AgNPs showed antibacterial activity, re-epithelization, and wound healing in MRSA infected mice model (Mekkawy et al., 2017). Similarly, topical application of nanosilvernanohydrogels in combination with Aloe vera accelerated wound contraction and enhanced wound healing due to the moist environment provided by Aloe vera (Anjum et al., 2016). In yet another formulation, AgNPs coated on to MCM-41 type mesoporous silica nanoparticles prevented their aggregation and allowed sustained release of Ag<sup>+</sup> ions displaying a long-term antibacterial activity against S. aureus. These antibacterial nanofibrous membranes could reduce inflammatory response and accelerate wound healing in wistar rats (Dong et al., 2016). A randomized clinical trial was carried out to test the antibacterial effect of two silver dressings and their healing time in burn patients. It was demonstrated that the hydrofiber silver dressing (Aquacel<sup>R</sup> ) was preferred over the nanocrystalline silver dressing (ActicoatTM) due to reduction of bioburden, quick wound healing, ease of using, comfort to the patients, and low cost (Verbelen et al., 2014). On the other hand ActicoatTM showed complete wound healing within 12 weeks in 64% of the patients with leg ulcers as compared to those who were treated with Iodosorb dressings (cadexomer iodine) (Miller et al., 2010).

Polymer-based nanomaterials and metal NPs are used in antimicrobial coatings on surface of medical devices, such as catheters and implants for prevention of infections. AgNPs when embedded in electrospun hyaluronic acid/polycaprolactonenanofibrous membranes coated on flexor tendon animal models prevented bacterial infection during the early postsurgical period (Chen et al., 2015; Shalumon et al., 2018). Likewise, implants coated with nanocomposite layer of polysaccharide 1-deoxylactit-1-yl chitosan and AgNPs in a mini-pig animal model showed good biocompatibility with the bone tissue (Marsich et al., 2013). All these studies demonstrated that the entrapped AgNPs allowed controlled release of Ag<sup>+</sup> ions displaying prolonged antibacterial and antibiofilm action as well as reduced inflammatory responses. A randomized clinical trial demonstrated the efficacy of triple-lumen central venous catheters impregnated with AgNPs (AgTive) which showed reduced bacterial colonization as compared to conventional catheters in intensive care unit patients (Antonelli et al., 2012).

Repeated exposure of AgNPs at sub inhibitory concentrations may lead to resistance in bacterial pathogens. To overcome this limitation, a combination of AgNPs with antibiotics has been suggested in order to increase the therapeutic efficacy of either, resulting in reduction of dose and hence toxicity. In vitro studies demonstrating AgNPs co-incubated with different antibiotics showed synergistic antibacterial activities against ESKAPE (Ghosh et al., 2012; Panácek et al., 2015; ˇ Golinska et al., ´ 2016; Habash et al., 2017; Singh et al., 2018; Wypij et al., 2018). Synergistic antibacterial activity of AgNPs in combination with polymyxin B was demonstrated in A. baumannii infected mouse model with 60% survival rate as compared to the controls treated with antibiotic or AgNPs alone (Wan et al., 2016).

Despite the use of AgNPs as a potential therapeutic agent, literature survey indicates a paucity of data obtained from in vivo studies carried out to test the toxicity, efficacy, pharmacokinetic, and immuno-modulatory response of the AgNPs. Further investigations through well-defined studies and clinical trials will lead to applications of AgNPs in wound dressings or medical devices.

## CONCLUDING REMARKS

There is an urgent need to restock our armamentarium of antimicrobials in order to stay ahead of the ever rising drug resistant ESKAPE pathogens. There is an insufficiency of effective antibiotic combinations in addition to the dry pipeline of new drugs. Huge efforts have been taken to use antibiotics in combination with adjuvants targeting important metabolic mechanisms/pathways contributing to drug resistance (permeablisers, lactamase inhibitors, efflux pump inhibitors, quorum sensing inhibitors, toxin inhibitors etc.) The modest success received to date with such antibiotic-adjuvant combinations has paved way to explore other alternative strategies to combat drug resistance. There is a significant rise in the interest shown by the scientific community to use novel therapeutic agents such as phages, antimicrobial peptides, metal nanoparticles, and photodynamic light which, although, have some limitations as discussed above. Some of the commonly described limitations of these therapies include stability and toxicity of the therapeutic agent, its targeted delivery at the site of infection, or immune response developed by the host against the therapeutic agent. Ongoing research has therefore led to further develop or modify these novel therapeutic agents or therapies so as to surmount the limitations as well as to overcome the barriers of bacterial resistance.

This review summarizes studies that demonstrate potential alternative therapies using in vivo models some of which have extended further to the level of clinical trials. The interest in phage therapy to treat bacterial infections is fast growing leading to development of commercial preparations such as "Stafal," "Sextaphage," "PhagoBioDerm," and "Pyophage" against MDR pathogens. Similarly, use of silver nanoparticles as antibiofilm coatings in surgical implants, antimicrobial agents in topical applications or as formulations in wound dressings has shown promising activities in animal models. Clinical trials using commercially available nanosilver coated dressings (ActicoatTM, Aquacel<sup>R</sup> ) or catheters (AgTive) is another noteworthy advancement. AMPs have received great attention due to their broad spectrum activity; however, they have shown limited pharmaceutical potential due to their toxicity, stability, and production costs. Photodynamic light therapy which is widely used for cancer therapy has also been demonstrated to be an effective strategy for clearing wound infections. However, additional studies demonstrating the efficacy and safety of these therapeutic agents against ESKAPE infections are desired. Similarly, randomized clinical trials would enable these therapeutic agents to cross the regulatory hurdles and find application in clinical practice.

It was observed that, majority of these studies have used animal models infected by S. aureus, A. baumannii, and P. aeruginosa to test the efficacy of the therapeutic agent. The probable reason for this could be that these pathogens mostly cause topical infections (wound, burn and abscess) and because majority of the limitations (targeted delivery, stability, immune response, toxicity etc.) described for each therapy can be minimized though not avoided in such models. It would be important to study the effect of these therapeutic agents against systemic infections caused by ESKAPE members. It was also observed that, the methods used for estimating efficacy of any therapeutic agent were not uniform. **Table 2** reveals that, the in vivo efficacy of various therapies is given either in terms of log or percent reduction of microbial load or as percent survival of the infected host (animal model). The methods followed to estimate the reduced pathogen loads as well as dosages used for treatment also vary. It is therefore not appropriate to compare these studies to identify the best therapeutic agent/therapy against any ESKAPE member.

In addition to the growing concern in searching and evaluating the clinical potential of the above discussed alternative therapies, research on combinatorial approach, based on the synergistic action of two or more therapies is also gaining attention. Most commonly studied combinations involve use of a therapeutic agent/ therapy (phage, aPDT, AMP, or AgNP) in combination with antibiotic/s or in some cases with an efflux pump inhibitor or quorum sensing inhibitor. Another interesting option used was the combination of the therapeutic

### REFERENCES


agents in a conjugate or hybrid (antibiotic-antibiotic, antibiotic-EPI, PS-AMP, PS-EPI etc.) for an increased efficacy against the pathogen. Most of these studies demonstrated that the combinatorial approach helped overcome the limitation caused by individual therapeutic agent. For example, an antibiotic combined with an efflux pump inhibitor or a photosensitizer conjugated with an AMP improved their entry and retention into the target pathogen for enhanced antimicrobial action. Similarly, combinations of antibiotics with nanoparticles or AMPS reduced the toxicity caused by these agents which were required at high dosages when used alone. The synergistic action allows for an increased bioavailability of the drug or therapeutic agent, broad antimicrobial spectrum, reduced toxicity, and decreased chances of development of resistance. However, **Table 3** shows that most of the studies demonstrating potential of combinatorial approach are currently based on in vitro evaluation only. Data supporting the potential of a combinatorial therapy with respect to the mechanism of synergy, its in vivo efficacy, toxicity, and immune response is scarce and needs further investigation.

Finally, cost effectiveness of the above described therapeutic agents/therapy over the conventional antimicrobial agents also play a crucial role in their clinical application. Production cost for any therapeutic agent will strongly depend on the various regulatory hurdles they pass to come into clinical practice. An agent or therapy which is too expensive may not be a preferred choice for treatment in under developed countries or weaker economies thus leading to over-use of conventional antimicrobials thus contributing to the growing drug resistance.

To conclude, a uniform research methodology used to test the efficacy of these therapeutic agents in accordance with welldefined standards will make it possible to reliably compare the data presented by various research groups. Well-performed clinical trials of these therapeutic agents used as monotherapy or as a combinatorial approach will allow us to derive the real potential of these therapeutic combinations for being translated into clinical practice.

### AUTHOR CONTRIBUTIONS

KP conceived the concept and edited. MM, EK, SK, and MT wrote and edited the manuscript and agreed for submission.

### FUNDING

MT is a Project Assistant under UPE-Phase II (UGC-262-A-2); EK is a recipient of CSIR-SRF; SK is a Project Assistant DST-PURSE (GOI-A-670).


of Pseudomonas aeruginosa internalized in bronchial cells of cystic fibrosis patients and lung cell migration: biochemical properties and a plausible mode of action. Antimicrob. Agents Chemother. 60, 7252–7262. doi: 10.1128/AAC.00904-16


against Gram-negative bacteria, including drug-resistant strains, relies on a nonmembranolytic mode of action. Infect. Drug Resist. 11, 969–979. doi: 10.2147/IDR.S165179


LED light. J. Dermatol. Sci. 90, 323–331. doi: 10.1016/j.jdermsci.2018. 03.001


planktonic growth on a zebrafish infection model. Microb. Pathog. 124, 291–300. doi: 10.1016/j.micpath.2018.08.053


multidrug-resistant strains and biofilms of Pseudomonas aeruginosa. Amino Acids 50, 1471–1483. doi: 10.1007/s00726-018-2625-4


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer RP declared a shared affiliation, with no collaboration, with the authors to the handling editor at the time of the review.

Copyright © 2019 Mulani, Kamble, Kumkar, Tawre and Pardesi. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.