# FUNCTIONALIZED NANOCARRIERS FOR THERANOSTICS

EDITED BY : Stefano Leporatti, Andrea Ragusa and Rawil Fakhrullin PUBLISHED IN : Frontiers in Bioengineering and Biotechnology

#### 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 978-2-88966-318-7 DOI 10.3389/978-2-88966-318-7

#### 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

# FUNCTIONALIZED NANOCARRIERS FOR THERANOSTICS

Topic Editors: Stefano Leporatti, Italian National Research Council, Italy Andrea Ragusa, University of Salento, Italy Rawil Fakhrullin, Kazan Federal University, Russia

Citation: Leporatti, S., Ragusa, A., Fakhrullin, R., eds. (2021). Functionalized Nanocarriers for Theranostics. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-318-7

# Table of Contents


Stella Siciliani, Amedea Manfredi, Marco Biggiogera, Michele Rossi, Paolo Ferruti, Elisabetta Ranucci and Livia Visai


Mahdi Yar Saleh, Neela Prajapati, Mark A. DeCoster and Yuri Lvov


Seema Gupta, Indu Singh, Ashwani K. Sharma and Pradeep Kumar

*132 Polylactic Acid-Based Patterned Matrixes for Site-Specific Delivery of Neuropeptides On-Demand: Functional NGF Effects on Human Neuronal Cells*

Olga A. Sindeeva, Olga Kopach, Maxim A. Kurochkin, Andrei Sapelkin, David J. Gould, Dmitri A. Rusakov and Gleb B. Sukhorukov


Serkan Yaman, Harish Ramachandramoorthy, Gizem Oter, Daria Zhukova, Tam Nguyen, Manoj K. Sabnani, Jon A. Weidanz and Kytai T. Nguyen

*175 The New Frontiers in Neurodegenerative Diseases Treatment: Liposomal-Based Strategies*

Mariafrancesca Cascione, Valeria De Matteis, Stefano Leporatti and Rosaria Rinaldi

# Editorial: Functionalized Nanocarriers for Theranostics

#### Stefano Leporatti <sup>1</sup> \*, Andrea Ragusa1,2 and Rawil Fakhrullin<sup>3</sup>

*<sup>1</sup> Consiglio Nazionale delle Ricerche CNR-Nanotec, Istituto di Nanotecnologia, c*\*o Campus Ecotekne, Lecce, Italy, <sup>2</sup> Department of Biological and Environmental Sciences and Technologies, University of Salento, Lecce, Italy, <sup>3</sup> Institute of Fundamental Medicine and Biology, Kazan Federal University, Kazan, Russia*

Keywords: nanocarriers, theranostics, nanomedicine, nanosystems, nanoplatforms

**Editorial on the Research Topic**

#### **Functionalized Nanocarriers for Theranostics**

Nanotechnology has led to the development of a variety of nanocarriers with applications in diagnosis and therapy, giving rise to novel theranostic nano-tools. They combine diagnostic and therapeutic moieties into a single nano-device, potentially allowing an early detection of the pathology together with targeted treatment.

Our Research Topic has gathered several contributions describing novel functionalized nano-transporters, such as targeted nanoparticles (NPs), nanovesicles (NVs), liposomes (LPs), and nano-clays, for combined diagnosis and therapy. Eleven articles have been collected with a well-balanced ratio: five are comprehensive reviews and the other six are original research articles.

Among a plethora of different nanomaterials available to fabricate drug delivery systems (DDSs) for cancer therapy and diagnosis, poly(D,L-lactic-co-glycolic acid) (PLGA) has been extensively used due to its biocompatibility and biodegradability. In a comprehensive review article, Shen et al. report the employment of PLGA-based DDSs for remotely triggered cancer therapy, including photo-triggered, ultrasound-triggered, magnetic field-triggered, and radiofrequencytriggered cancer therapy, involving photodynamic therapy (PDT), photothermal therapy (PTT), and photo-triggered chemotherapeutics release. The state-of-the-art in theranostic approaches involving mesoporous silica nanoparticles (MSNs) for treating hepatocellular carcinoma (HCC) has been reviewed by Tao et al. In this work, they outline the recent advances in MSNs-based systems for HCC therapy and diagnosis and they also discuss the precision delivery strategies of MSNs in liver cancer.

The signature service and insoluble network formation of the peptide self-assemblies as hydrogels have drawn a plethora of research activity among scientists all over the globe in the past decades. In this respect, Gupta et al. review the last 5-year efforts on novel approaches for the design and development of single molecule amino acids, ultra-short peptide self-assemblies (di- and tri-peptides only) and their derivatives as drug/gene carriers and tissue-engineering systems. Peptides and small molecule-based nanostructures can be convenient alternatives for therapeutic delivery due to their good biocompatibility and their easy design, synthesis, and functionalization. These properties render their self-assembled nanostructures smart tools suitable for biomedical applications (Gupta et al.).

Parodi et al. overview the recent development of smart nanotheranostic systems responsive to pathological stimuli, including oxidative stress, altered pH, enzymatic expression, and reactive biological molecules. Therapeutic and diagnostic properties can be included in the same molecule embedded in the nanocarrier structure that can be activated at the injury site, or they can independently derive from different chemicals loaded into and/or conjugated onto the surface of the NPs.

#### Edited and reviewed by:

*Gianni Ciofani, Italian Institute of Technology (IIT), Italy*

> \*Correspondence: *Stefano Leporatti stefano.leporatti@nanotec.cnr.it*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

Received: *12 October 2020* Accepted: *16 October 2020* Published: *10 November 2020*

#### Citation:

*Leporatti S, Ragusa A and Fakhrullin R (2020) Editorial: Functionalized Nanocarriers for Theranostics. Front. Bioeng. Biotechnol. 8:616574. doi: 10.3389/fbioe.2020.616574*

**4**

Currently, the onset of neurodegenerative diseases (NDs) is strongly widespread due to the increasing age of the world population and the lack of efficient therapies. In a state-ofthe-art review article, Cascione et al. overview the lipid-based drug delivery improvements in in vivo applications against NDs. Lipid carriers have a good ability to deliver both hydrophobic and hydrophilic molecules through the Blood-Brain Barrier (BBB), demonstrating an enhanced efficacy of the drug following liposomal encapsulation.

The six original research articles contributions describe novel tailored nano-carriers for drug delivery applied to cancer and neurodegenerative diseases treatment. Two of them are dealing with the use of functionalized halloysite nanotubes (HNTs) as efficient drugs/active molecules nano-transporters (Guryanov et al.; Saleh et al.). Guryanov et al. report on prodigiosin-loaded halloysite-based nanoformulation (p-HNTs) and its effects on cell viability. By comparing the effects of p-HNTs on malignant (Caco-2, HCT116) and non-malignant (MSC, HSF) cells, the authors demonstrate selective cytotoxic and genotoxic activity. Moreover, in another very interesting and prospective article, Saleh et al. proposed the use of HNT nanocarriers to penetrate the BBB and effectively deliver the payload over an extended time period. On the other hand, the Sukhorukov's group tried to address the delivery of neuropeptides on-demand, potentially suitable for the migratory or axonal guidance of human nerve cells, by using polylactic acid (PLA)-based microchamber arrays (MCAs) (Sindeeva et al.). Optical targeting of microchambers for drug release triggered functional cell response locally.

The last three contributed articles are reporting the use of different type of NPs for delivering and releasing drug against different type of tumors. In this respect, Zacheo et al. developed lipid NVs of size varying from 100 up to 300 nm and successfully loaded them with fluorophores molecules (DOP-F-DS and a fluorescent protein), inorganic nanoparticles (quantum dots and magnetic NPs), and anti-cancer drugs (SN-38 and doxorubicin).

The synthesis and the functionalization of gold NPs with cancer-specific biomolecules may represent a winner strategy for a selective and targeted tumor-phototherapy. With this in mind, Bloise et al. described a simple approach for the synthesis of extra-small gold NPs for breast cancer therapy. Extrasmall gold nanospheres stabilized with a thiol-functionalized polyamidoamine (AGMA1) and trastuzumab were synthesized and tested in vitro as nanovectors for breast cancer targeted drug delivery. They proposed to assemble small Au NPs into larger structures through controllable interparticle interactions by using AGMA1, with the final aim to enhance the absorption of NIR light (Bloise et al.).

To improve chemo-drug therapeutic efficiency and overcome issues such as inadequate response rates, high toxicity, severe side effects due to non-specific targeting of anti-cancer drugs, and the development of multidrug resistance during prolonged treatment, a multifunctional NP was developed to effectively target and treat melanoma by Yaman et al.. PLGA NPs were coated with a cellular membrane derived from the T-cell hybridoma, 19LF6 endowed with a melanoma-specific anti-gp100/HLA-A2 T-cell receptor (TCR) and loaded with trametinib, an FDA-approved melanoma chemotherapeutic drug.

### AUTHOR CONTRIBUTIONS

SL wrote the editorial, which was revised, proofed, and accepted by all the authors.

### ACKNOWLEDGMENTS

SL and AR are grateful to the Tecnopolo per la medicina di precisione (TecnoMed Puglia)—Regione Puglia: DGR no. 2117 del 21/11/2018, CUP: B84I18000540002 and Tecnopolo di Nanotecnologia e Fotonica per la medicina di precisione (TECNOMED)—FISR/MIUR-CNR: delibera CIPE no. 3449 del 7-08-2017, CUP: B83B17000010001. RF was funded by the Russian Foundation for Basic Research grant #18-29-11031 mk.

**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 © 2020 Leporatti, Ragusa and Fakhrullin. 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.

# Extra-Small Gold Nanospheres Decorated With a Thiol Functionalized Biodegradable and Biocompatible Linear Polyamidoamine as Nanovectors of Anticancer Molecules

Nora Bloise1,2, Alessio Massironi<sup>3</sup> , Cristina Della Pina<sup>4</sup> \*, Jenny Alongi<sup>5</sup> , Stella Siciliani<sup>6</sup> , Amedea Manfredi<sup>5</sup> , Marco Biggiogera<sup>6</sup> , Michele Rossi<sup>4</sup> , Paolo Ferruti<sup>5</sup> , Elisabetta Ranucci<sup>5</sup> \* and Livia Visai1,2 \*

#### Edited by:

Stefano Leporatti, Italian National Research Council, Italy

fbioe-08-00132 March 2, 2020 Time: 20:46 # 1

#### Reviewed by:

Martin J. D. Clift, Swansea University, United Kingdom Sílvia Castro Coelho, University of Porto, Portugal

#### \*Correspondence:

Cristina Della Pina cristina.dellapina@unimi.it Elisabetta Ranucci elisabetta.ranucci@unimi.it Livia Visai livia.visai@unipv.it

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology

Received: 29 November 2019 Accepted: 10 February 2020 Published: 04 March 2020

#### Citation:

Bloise N, Massironi A, Della Pina C, Alongi J, Siciliani S, Manfredi A, Biggiogera M, Rossi M, Ferruti P, Ranucci E and Visai L (2020) Extra-Small Gold Nanospheres Decorated With a Thiol Functionalized Biodegradable and Biocompatible Linear Polyamidoamine as Nanovectors of Anticancer Molecules. Front. Bioeng. Biotechnol. 8:132. doi: 10.3389/fbioe.2020.00132 <sup>1</sup> Department of Molecular Medicine (DMM), Biochemistry Unit, Center for Health Technologies (CHT), UdR INSTM University of Pavia, Pavia, Italy, <sup>2</sup> Department of Occupational Medicine, Toxicology and Environmental Risks, Istituti Clinici Scientifici Maugeri S.p.A, IRCCS, Pavia, Italy, <sup>3</sup> Department of Chemistry and Industrial Chemistry, University of Pisa, UdR INSTM PISA, Pisa, Italy, <sup>4</sup> Dipartimento di Chimica, Università degli Studi di Milano e CNR-ISTM, Milan, Italy, <sup>5</sup> Dipartimento di Chimica, Università degli Studi di Milano, Milan, Italy, <sup>6</sup> Department of Biology and Biotechnology, University of Pavia, Pavia, Italy

Gold nanoparticles are elective candidate for cancer therapy. Current efforts are devoted to developing innovative methods for their synthesis. Besides, understanding their interaction with cells have become increasingly important for their clinical application. This work aims to describe a simple approach for the synthesis of extrasmall gold nanoparticles for breast cancer therapy. In brief, a biocompatible and biodegradable polyamidoamine (named AGMA1-SH), bearing 20%, on a molar basis, thiol-functionalized repeat units, is employed to stabilize and coat extra-small gold nanospheres of different sizes (2.5, 3.5, and 5 nm in gold core), and to generate a nanoplatform for the link with Trastuzumab monoclonal antibody for HER2-positive breast cancer targeting. Dynamic light scattering, transmission electron microscopy, ultraviolet visible spectroscopy, X-ray powder diffraction, circular dichroism, protein quantification assays are used for the characterization. The targeting properties of the nanosystems are explored to achieve enhanced and selective uptake of AGMA1-SHgold nanoparticles by in vitro studies against HER-2 overexpressing cells, SKBR-3 and compared to HER-2 low expressing cells, MCF-7, and normal fibroblast cell line, NIH-3T3. In vitro physicochemical characterization demonstrates that gold nanoparticles modified with AGMA1-SH are more stable in aqueous solution than the unmodified ones. Additionally, the greater gold nanoparticles size (5-nm) is associated with a higher stability and conjugation efficiency with Trastuzumab, which retains its folding and anticancer activity after the conjugation. In particular, the larger Trastuzumab functionalized nanoparticles displays the highest efficacy (via the pro-apoptotic protein increase, anti-apoptotic components decrease, survival-proliferation pathways downregulation) and internalization (via the activation of the classical clathrin-mediated

endocytosis) in HER-2 overexpressing SKBR-3 cells, without eliciting significant effects on the other cell lines. The use of biocompatible AGMA1-SH for producing covalently stabilized gold nanoparticles to achieve selective targeting, cytotoxicity and uptake is completely novel, offering an important advancement for developing new anticancer conjugated-gold nanoparticles.

Keywords: cancer cell targeting, polyamidoamines, Trastuzumab, gold nanoparticles, breast cancer nanomedicine

#### INTRODUCTION

Breast cancer is the most frequent and invasive cancer type in women (Rojas and Stuckey, 2016). There is an increasing demand for novel, efficient and specific diagnostic and medical tools to face the adverse side-effects of the traditional ones. Some invasive forms of breast cancer (i.e. locally advanced breast cancer) presented a twofold higher incidence than the others, suggesting an urgent improvement of Trastuzumab (HerceptinTM) delivery toward erbB2 tyrosine kinase receptor (HER-2) found overexpressed in ∼25%–30% of breast cancers (Sawaki et al., 2006). Engineered delivery systems based on gold nanoparticles (AuNPs) are particularly promising to address this problem. AuNP formulations are employed for a wide range of medical applications, including contrast and photothermal agents for computed tomography (CT) and tumor photothermal ablation, respectively (Mieszawska et al., 2013). Their chemical and physical properties, that span the broader visible to nearinfrared, ensure them features such as low toxicity, high stability, easy synthesis and conjugation with cancer-specific biomolecules (Lee et al., 2014; Carnovale et al., 2016). It is generally agreed that physicochemical differences in nanoparticles such as particle size, shape, surface charge and surface coating could vary the way particles are recognized, processed and excreted by the body (Carnovale et al., 2016). Notably, the molecular rules governing these properties are poorly understood, and the different setup conditions used in the different laboratories (such as cell types, dosage schedule, measurement methods) make it difficult to drawing valid and conclusive information. Experimental and theoretical studies stated that the optimal size for AuNPs cellular uptake ranges from 25 to 50 nm, since it stimulates efficiently membrane wrapping and receptor-ligand interaction to drive the NPs into the cell (Gao et al., 2005; Jiang et al., 2008; Dreaden et al., 2012; Panariti et al., 2012; Saw et al., 2018). Surface charge can also have an effect on the cellular uptake (Zhang et al., 2015). Interacting with negative cell membranes, the electropositive nanoparticles exhibited a higher cellular uptake efficiency compared to electronegative ones. Conjugation of AuNPs with targeting molecules effectively improves the tumor target delivery (Paciotti et al., 2004). Several studies reported that Trastuzumab could be successfully immobilized on gold nanoparticles to improve their interactions with SKBR3 breast cancer cells, overcome Trastuzumab resistance and detect breast cancer (Hainfeld et al., 2006; Bickford et al., 2008; Chattopadhyay et al., 2010; Carpin et al., 2011; Lee et al., 2014). In vitro experiments indicated that, while human skin cells proliferated in the presence of Trastuzumab-conjugated gold nanoparticles, most of the breast cancer cells died (Rathinaraj et al., 2015). Despite the broad interest surrounding gold-based nanosystems, reproducibility, toxicity and excretion concerns limit their clinical translations (Choi et al., 2007; Lewinski et al., 2008; Tam et al., 2010). Indeed, currently no gold nanoparticles have yet been approved by the FDA agency. Different biodegradable polymers were tested for assembling and coating gold nanoparticles clusters (Tam et al., 2010), while minimizing immunogenicity reactions. Cheheltani et al. (2016) proposed a small, excretable AuNP-based platform, encapsulated into biodegradable poly di(carboxylatophenoxy)phosphazene (PCPP) nanospheres. A study by Tam et al. (2010) reported polymer/inorganic nanoclusters combining the imaging contrast and therapeutic capabilities with the biodegradability of a polymer stabilizer. Linear polyamidoamines (PAAs) have recently emerged as promising tools for drug delivery as they offer key advantages due to their ease of formulation and biodegradability (Ferruti et al., 2005; Jacchetti et al., 2008; Ferruti, 2013; Mauro et al., 2013). PAAs were previously investigated as anticancer drug carriers (Lavignac et al., 2009). In particular, the PAA nicknamed AGMA1 can be used as a potential non-viral, non-toxic and efficient vector for the intracellular delivery of siRNA and DNA (Cavalli et al., 2010; Cavalli et al., 2017). Interestingly, AGMA1, containing tert-amine, carboxyl and guanidine groups, whose repeat unit is reminiscent of the arggly-asp (RGD) peptide motif (Franchini et al., 2006), a wellknown fibronectin sequence mediating cell attachment, can act as an excellent cell adhesion and proliferation substrate (Gualandi et al., 2016). For in vivo applications, gold-based nanosystems should be larger than 6 nm in diameter to ensure long blood circulation, hence accumulation in diseased tissues but slowly breaking down into sub-6 nm components for rapidly excretion via the kidneys (Arruebo et al., 2007; Choi et al., 2007). The goal of the present study was to develop more efficient gold nanoparticles for therapeutic use. To this purpose, a biocompatible and biodegradable polyamidoamine bearing 20%, on a molar basis, randomly distributed SH pendants (AGMA1- SH, indicated also as "P") was employed to stabilize AuNPs of different sizes, that is 2.5, 3.5, and 5 nm in Au core (Au@P), decorated with Trastuzumab (Au@PT), whose hydrodynamic diameter was suitable for a cellular uptake (**Figure 1**). AGMA1, besides being a biocompatible and biodegradable polymer, was found to be easily internalized in cells and, therefore,

it can be expected to facilitate the AuNPs cellular uptake (Franchini et al., 2006; Cavalli et al., 2010), whereas thiol groups are supposed to stabilize the AGMA1-gold bond by the renowned S-Au soft-soft interaction. Once tested in vivo, it is reasonable to speculate that, because of the AGMA1 biodegradability, the small Au core component (ranging from 2.5 to 5 nm) may be the only one to be extruded by the urinary system. In this work, the impact size of Au@P nanoparticles on the interactions with both breast cancer cells, and healthy cells was systemically studied. The cytotoxicity and cellular uptake of all nanoparticles were studied by MTT and ICP-MS tests. Western blot analysis and cellular uptake studies by means of specific endocytosis inhibitors were further carried out to explore the interaction with and mechanism of internalization of Au@PT by SKBR-3 target cells.

### MATERIALS AND METHODS

#### Reagents

4-Aminobutylguanidine sulfate (97%), lithium hydroxide monohydrate (99%), cystamine hydrochloride (96%), 2 mercaptoethanol (99%) and calcium chloride (97%) were purchased from Sigma-Aldrich and used as provided. 2,2- Bis(acrylamido)acetic acid (96%) was synthesized as previously described (Ferruti et al., 1999). HAuCl<sup>4</sup> solution (composition: 17 wt.% Au; concentration: 30 wt.% in dilute HCl) was purchased from Sigma-Aldrich and diluted with water to achieve a final Au concentration of 10 mg/mL. NaBH<sup>4</sup> (>96% pure), purchased from Sigma-Aldrich, was employed as the reducing agent for Au3<sup>+</sup> to colloidal gold dispersion (10 mg/mL). Activated carbon Vulcan XC72R (specific area = 254 m<sup>2</sup> g −1 , pore volume = 0.19 mL g−<sup>1</sup> ) from CABOT was used as a supporting material for colloidal samples for XRPD analysis. MilliQ water obtained by an Academic A-10 Millipore apparatus was employed as a solvent for all aqueous solutions. Trastuzumab (Herceptin <sup>R</sup> , Roche) was kindly provided by the Chemotherapic section of Policlinico San Matteo (Pavia). For Western blot analyses, primary mouse monoclonal anti β-actin was purchased from Santa Cruz Biotechnology, mouse monoclonal anti-ERK1/2 and rabbit polyclonal anti-phospho ERK1/2 (Thr 202/204) from Abcam, rabbit monoclonal anti-AKT and rabbit polyclonal anti-phospho -AKT (Ser 473) from Cell Signaling Technologies (Danvers, MA, United States) and rabbit monoclonal anti-BAX and anti-BCL-XL from Thermo Fisher Scientific. Horseradish peroxidase (HRP)-conjugated secondary antibodies anti-mouse and anti-rabbit (Dako) anti-human (Tebu-bio) were used, and detection was performed by enhanced chemiluminescent substrate (ECL) solutions (Pierce Thermo Fisher Scientific, Rockford, IL, United States).

#### Instruments

<sup>1</sup>H and <sup>13</sup>C NMR spectra were run on a Brüker Advance 400 spectrometer operating at 400.132 (1H) and 100.623 (13C) MHz. Size exclusion chromatography (SEC) traces were obtained with a Knauer Pump 1000 equipped with a Knauer Autosampler 3800, TSKgel G4000 PW and G3000 PW TosoHaas columns connected in series, Light Scattering (LS) Viscotek 270 Dual Detector, UV detector Waters model 486, operating at 230 nm, and a refractive index detector Waters model 2410. The mobile phase was a 0.1 M Tris buffer pH 8.00 (0.05 M with 0.2 M sodium chloride (Bignotti et al., 1994). The operational conditions were: sample concentration 20 mg/mL; flow rate 1 µL/min; injection volume 20 µL; column dimensions 300 × 7.5 mm<sup>2</sup> , temperature 25◦C. The instrument constants were determined using PEO 19 kDa as a narrow polymer standard. All samples were filtered through a 0.2 µm syringe Whatman filter before measurements.

### Synthesis of AGMA1-SH

2,2-Bis(acrylamido)acetic acid (4.128 g, 20.0 mmol) and lithium hydroxide monohydrate (1.53 g, 36.0 mmol) were dissolved in water (18 mL) and 4-aminobutylguanidine (agmatine) sulfate (3.765 g, 16.0 mmol) was added under vigorous stirring. The reaction mixture was heated up to 40◦C until dissolution and

kept under nitrogen atmosphere at room temperature with gentle stirring for 1 day (d). After this time, the reaction mixture was cooled down to room temperature and cystamine dihydrochloride (0.469 g, 2.0 mmol) and lithium hydroxide monohydrate (0.170 g, 4 mmol) were added. The resultant mixture was kept at room temperature for 10 days until a soft transparent hydrogel was obtained. Then water (40 mL) and excess mercaptoethanol (8.0 mL) were added until a clear solution was obtained. This was diluted with water (100 mL), acidified to pH 4.5 with 37% hydrochloric acid and then ultrafiltered through a membrane with nominal molecular weight cut-off 3000. The final product was retrieved by freeze-drying the retained fraction. Yield 68%, M¯ n = 8400, PDI = 1.25.

<sup>1</sup>H NMR (D2O): δ (ppm) = 1.52–1.60 (br, NCH2CH2CH2); 1.71–1.77 (br, NCH2CH2); 2.64–2.76 (br, NHCOCH2); 2.92– 2.94 (m, CH2S); 3.08–3.20 (m, NCOCH2CH2N); 3.34–3.47 (br, NCH2CH2CH<sup>2</sup> and NCH2CH2S); 5.51–5.56 (s, CHCOOH).

<sup>13</sup>C NMR (D2O): d (ppm) = 18.5 (CH2S); 22.3 (NCH<sup>2</sup> CH2CH2); 25.0 (NCH2CH2); 29.0 (NHCOCH2); 40.4 (CH<sup>2</sup> CH2NH); 49.1 (NCH2CH2CONH); 52.5 (NCH2CH2CH2CH<sup>2</sup> NH); 54.8 (NCH2CH2S); 56.0 (CHCOOH); 155.1 (H2NC = NH); 171.3 (NHCO); 173.5 (COOH).

### Synthesis of Gold Nanoparticles Decorated With AGMA1-SH and Trastuzumab

Three differently sized colloidal samples of AGMA1-SH-coated gold nanoparticles decorated with Trastuzumab (Herceptin <sup>R</sup> ), were prepared by varying gold solution concentration from 20 ppm to 400 ppm. The samples were labeled as 2.5Au@PT, 3.5Au@PT, and 5Au@PT, where the number indicates the average diameter of gold core (2.5, 3.5, and 5.0 nm, respectively), P means the polymer (AGMA1-SH) and T the antibody drug (Trastuzumab). Accordingly, the following parameters were optimized to produce stable colloidal nanosystems over time: gold solution concentration (20 ppm for 2.5Au@PT, 200 ppm for 3.5Au@PT, and 400 ppm for 5Au@PT), AGMA1- SH:Au:Trastuzumab ratio (1wt.:1wt.:1wt.) and NaBH4:Au ratio (1wt.:1wt.). More in detail, proper aliquots of AGMA1-SH aqueous solution (10 mg/mL) were added under stirring to defined volumes of HAuCl<sup>4</sup> aqueous solution (10 mg/mL Au) and MilliQ <sup>R</sup> water at room temperature. Established amounts of freshly prepared NaBH<sup>4</sup> aqueous solution (10 mg/mL) were rapidly added under vigorous stirring to reduce Au3<sup>+</sup> to Au<sup>0</sup> and, hence, produce light-tea to reddish colored colloidal dispersions. Finally, proper volumes of Trastuzumab aqueous solution (20 mg/mL) were poured drop wise. Only in the case of the 5Au@PT sample, sodium borohydride was added prior to the polymer in order to allow gold nanoparticles to grow up to 5.0 nm. **Supplementary Table S1** reports the established aliquots for each component. At the end of reaction, to remove the excess of synthesis reagents or/and to eliminate unbound antibody, each AuNPs solution was purified by dialysis for 24 h in a Float-A-Lyzer Spectra/Por G2 (MWCO of 100 kDa for Au@PT) under stirring at 4◦C.

### Synthesis of Gold Nanoparticles Decorated With AGMA1-SH or Trastuzumab

Three differently sized AuNP@AGMA1-SH colloidal samples without the drug (Trastuzumab) and three differently sized AuNP@Trastuzumab colloidal samples without the polymer (AGMA1-SH) were synthesized adopting the aforementioned procedure. The following parameters were optimized in order to produce stable colloidal nanosystems: gold solution concentration (20 – 400 ppm), AGMA1- SH:Au or Trastuzumab:Au ratios (1wt.:1wt.) and NaBH4:Au ratio (1wt.:1wt.). The samples were labeled as 2.5Au@P, 3.5Au@P, 5Au@P and 2.5Au@T, 3.5Au@T, 5Au@T respectively. **Supplementary Table S1** reports the established aliquots for each component. At the end of reaction, to remove the excess of synthesis reagents each AuNPs solution was purified by dialysis for 24 hours (h) in a Float-A-Lyzer Spectra/Por G2 (MWCO 10 kDa for Au@P and 100 kDa for Au@T, respectively) under stirring at 4◦C. As evinced by microscopic analyses (TEM), AuNPs were synthesized in uniform spherical sizes. Hence, by using the simple geometrical model of spherical particles (Eq. 1–6), the theoretical Trastuzumab molecule to gold nanoparticle ratio (T/AuNP) could be determined for the three differently sized nanosystems. Accordingly, T/AuNP = 0.65 for '2.5Au' samples, 1.79 for '3.5Au' samples and 5.21 for '5Au' samples. In addition, extrapolating the fraction of gold atoms lying at the surface (gold dispersion from **Supplementary Figure S1**) led to the theoretical Trastuzumab molecule to gold surface atom ratio (T/surface Au) for the three differently sized samples:

T/surface Au = 0.0025 for '2.5Au' samples, 0.0039 for '3.5Au' samples and 0.0054 for '5Au' samples.

$$\text{Volume } Au \text{NP} = \frac{4}{3}\pi r^3 \quad \text{(\$r\$ is Au nonparticle radius = 1.25, \$)}$$

$$\text{...}$$

1.75, and 2.5 nm respectively) (1)

mass AuNP = volume AuNP × ρ

$$\text{(\rho is Au density)} = 19.3 \text{ g/cm}^3\text{)} \quad \text{(2)}$$

$$number\ AuNP = \frac{\text{mass Au} \text{ in the sample}}{\text{mass AuNP}} \tag{3}$$

$$\text{moles } AuNP = \text{mass } AuNP/atomic \text{ weight } Au \tag{4}$$

$$A \text{ atoms} \, A u \text{NP} = \text{moles} \, A u \text{NP} \times A \text{vogadro's} \, \text{number} \tag{5}$$

surface atoms AuNP = gold dispersion (%) × atoms AuNP

(Gold dispersion is 55% for '2.5Au', 35%

$$\text{for } \text{``3.5Au' and 25\% for } \text{``5Au'} \text{)}\tag{6}$$

#### Physicochemical Characterization

Dynamic light scattering (DLS) and Z-potential analyses were carried out on 1 mg/mL nanoparticle aqueous suspensions prepared using MilliQ water, with a Malvern Zetasizer NanoZS

instrument equipped with a laser fitted at 532 nm and fixed 173◦ scattering angle. Before analyses, samples were filtered through a 0.2 µm syringe Whatman filter. The solution pH was adjusted to the selected value, using 0.1 M HCl or 0.1 M NaOH aqueous solutions. Measurements were performed in triplicate, and the reported values averaged over of 10 runs. Regarding DLS measurements on AGMA1-SH-coated gold nanoparticles decorated with Trastuzumab (Au@PT), AGMA1-SH-coated gold nanoparticles (Au@P) and Trastuzumab-coated gold nanoparticles (Au@Trastuzumab), 20 mg/L aqueous suspensions were analyzed. X-ray powder diffraction (XRPD) analyses were performed employing a Rigaku D III-MAX horizontal-scan powder diffractometer with Cu Kα radiation (λ = 1.5418 Å) to determine the average size of gold crystallites using the Scherrer equation (Patterson, 1939). The Bragg angle was rotated in the 34◦– 43◦ 2θ interval and the mean diameter d of gold core was determined considering the peak at 2θ = 38.5◦ , typical of metallic gold. All the colloidal samples were immobilized on XC72R carbon to obtain the theoretical amount of 1% Au. UVvis spectra were collected in the 200–800 nm region using a HP 8453 diode array instrument on colloidal samples (20 ppm Au) for determining the plasmon resonance band (PRB) of gold nanoparticles detected near 500–530 nm in the 2–10-nmdiameter range.

#### Transmission Electron Microscopy

TEM analysis was performed by using a JEOL JEM 3010 operating at a 300-kV acceleration voltage. Briefly, 10 µL of each type of gold nanoparticle suspension in water (20 µg/mL) deposed on ParlodionTM membranes.

#### Efficiency of Trastuzumab Conjugation

The efficiency of Trastuzumab conjugation to nanoparticles was assessed by indirect (Bicinchoninic Acid, BCA), and direct (dotblot) approaches as previously described (Liu et al., 2010; Zhou et al., 2015; Di Francesco et al., 2018; Codullo et al., 2019). In short, both Au@PT and Au@T nanoparticles suspensions (20 µg/mL) were centrifuged at 3000 × g for 15 min, then the supernatants were collected for determine the unbound Trastuzumab concentration by BCA assay (Pierce Biotechnology Inc., Rockford, IL, United States), while the pellets were resuspended in 1 mL of MilliQ water for dot blot assay.

#### BCA Assay

The unbound content in the resulting solution was determined by BCA assay according to the specifications of the manufacturer. The unbound amounts were calculated from properly drawn Trastuzumab calibration curve. Trastuzumab amount conjugated to nanoparticles was thus obtained via reduction of the amount in the supernatant from the initial amount. The percentage of conjugation efficiency (CE) was calculated as follows (Martínez-Jothar et al., 2019):

$$\% \text{ CE} = \frac{(total\ amount\ of\ T\ added\text{-}unbound\ T)}{\text{total\ amount of\ T\ added}} \ast 100\text{ }$$

#### Dot Blot Assay

Different amounts of Trastuzumab-functionalized Au@P differently sized were spotted on nitrocellulose membrane Amersham Hybond ECL, GE Healthcare Life Sciences, Pittsburgh, PA, United States) and air-dried. 2.5Au@P, 3.5Au@P, 5Au@P as negative controls. In parallel, Au@T samples dot blot was also performed. The non-specific sites were blocked by soaking the membrane in 5% BSA in TBS containing 0.05% Tween 20 for 1 h at room temperature (RT). The membrane was then incubated using goat anti-human-HRP-conjugated antibody (1:1000) for 1 h at RT and after extensive washing developed with enhanced chemiluminescence reagents (LI-COR) and ImageQuant LAS4000 Imaging System (Ge Healthcare). The spots were then analyzed with ImageQuant TL software (Ge Healthcare) and the results were normalized to calibration curve containing known amounts of Trastuzumab.

#### Circular Dichroism

The secondary structure of the Trastuzumab free (20 µg/mL) and Trastuzumab-conjugated gold nanoparticles (20 µg/mL) were evaluated with circular dichroism (CD). CD spectra were measured using Jasco J710 spectropolarimeter (Jasco Corp., Tokyo, Japan) at 25◦C and in a 1 cm path-length quartz cell under the following conditions: 300–190 nm spectral range, 2 nm of bandwidth, 200 nm min-1 of scan speed and 3 accumulation. Data was processed using 10-point smoothing in Origin 6.0 (OriginLab Corporation, Northampton, MA, United States).

#### In vitro Trastuzumab Release From Gold Nanoparticles Decorated With AGMA1-SH

2.5Au@PT, 3.5Au@PT, and 5Au@PT (20 µg) were suspended in 1 mL of MilliQ water, in a low-protein binding centrifuge tube, and kept at room temperature. At select time points (1 – 15 days), each type of nanoparticles suspension was centrifuged, the supernatant was collected and replaced with the same volume of fresh release medium (Colzani et al., 2018). The concentration of antibody in the supernatants was determined using the BCA Kit as described above. Standard curve of Trastuzumab was used to determine antibody concentrations in each sample. The results are expressed as the percentage of cumulative fraction of Trastuzumab released at each time point compared to the conjugated amount. The in vitro release of 2.5Au@T, 3.5Au@T, and 5Au@T was not analyzed due to the higher aggregation recorded immediately within the first days after their synthesis.

#### Cell Types and Culture Conditions

HER2 positive breast adenocarcinoma cells (SKBR-3), ERα positive breast adenocarcinoma cells (MCF-7) used as cancer negative control, murine fibroblasts cells (NIH-3T3) used as negative control. All cell lines were obtained from the American Type Culture Collection (HTB85, ATCC, Manassas, VA, United States). After thawing, SKBR-3 cell line was cultured in McCoy's 5A Medium Modified (Sigma-Aldrich), 10% of Fetal Bovine Serum (FBS, EuroClone), 1% of L-Glutamine

(Lonza), 2% sodium pyruvate (Lonza), 0.4% antibiotics (Lonza) and 0.1% fungizone. MCF-7 cell line was cultured in Minimum Essential Medium Eagle (Sigma-Aldrich), 10% of Fetal Bovine Serum (EuroClone), 1% sodium pyruvate (Lonza), 1% Non-Essential Aminoacids (EuroClone), 1% Bovine Insulin (Sigma-Aldrich), 1% L-Glutamine (Lonza) and 0.4% antibiotics (Lonza). NIH-3T3 cell line was cultured in Dulbecco's Modified Eagle Medium High Glucose 4.5 mg/mL (Sigma-Aldrich), 10% of Bovine Calf Serum (BCS, EuroClone) and 1% of L-Glutamine (Lonza). All cell lines were incubated at 37◦C in 5% CO2, routinely trypsinized (Trypsin EDTA solution 1X, Lonza) after confluence, counted and seeded into wells.

### Stability of AuNPs Upon Incubation With Cell Culture Media by Dynamic Light Scattering

Prior to all biological investigations, each suspension of AGMA1- SH-coated gold nanoparticles decorated with Trastuzumab (2 µg/mL) was centrifuged and resuspended in the cell culture media appropriate for the corresponding cell line, containing 10% of serum, and incubated at physiological temperature (37◦C), without shaking, to mimic cell culture conditions (Halamoda-Kenzaoui et al., 2015; Mari¸sca and Leopold, 2019). Size change was tracked by serial DLS size measurements at 0 (directly after dispersion in the complex media) and after 4 h of incubation.

### Cell Viability

Cells were seeded at 1 × 10<sup>4</sup> viable cells/well on 96-well plates and incubated for 24 h (to allow cells to attach to the well). After the culture medium removal, cells were incubated with media containing increasing concentration of each types of gold nanoparticles. At fixed incubation times, the quantitative 3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide (MTT, Sigma Aldrich, St. Louis, MO, United States) assay was performed to assess the dehydrogenase activity, as an indicator of the metabolic state of cells, as previously described (Merli et al., 2018). The cell viability was expressed as percentage related to the control (untreated) set equal to 100%. At day 3 of incubation, a qualitative viability assay [fluorescein diacetate (FDA) assay] was performed on SKBR-3 untreated (ctrl), treated with T free (2 µg/mL), 2.5Au@PT (2 µg/mL) and 5Au@PT (2 µg/mL), as previously described (Bloise et al., 2013).

### Scanning Electron Microscopy (SEM)

For morphological evaluation all types of cells were seeded on plastic cell culture coverslip disks (Thermanox Plastic, Nalge Nunc International, Rochester, NY, United States), then incubated or not with nanoparticles suspensions (2 µg/mL) and treated as previously described (Bloise et al., 2013). Before observation, samples were sputter coated with gold and observed using a Field Emission Scanning Electron Microscope (FESEM Supra 25, Zeiss).

## Uptake Studies by Inductively Coupled Plasma Mass Spectrometry (ICP-MS)

All types of cell lines were plated (500000/well) on a 6-well plate and treated for 1 and 3 days with 2 µg/mL of each types of nanoparticles (2 mL volume added in each well). After treatments, cells were washed three times with PBS to eliminate the NPs which were not internalized in cells, trypsinized, counted and centrifuged in order to obtain the cellular pellet. The cellular pellet was treated with 1X RIPA buffer (diluting the 10X RIPA in sterile distilled water, EDM Millipore Corporation, Chemicon) in ice for 30 min, transferred in sterile Eppendorf and centrifuged at 13000 rpm for 15 min at 4◦C. The pellet was separated from the supernatant and treated with 750 µL of freshly prepared aqua regia. The samples were diluted to 3 mL with bidistilled water and analyzed for the Au content with ICP-MS (ELAN DRC, Perkin Elmer). The nanoparticles uptake efficiency per cell was calculated as previously described (Gunduz et al., 2017): Uptake efficiency (%) = (Number of NPs taken up by cells/Number of NPs incubated with cells) <sup>∗</sup> 100%. To obtain number of AuNP in each sample, total ppm determined by ICP-MS was divided to the mass of one AuNP. Total AuNPs number was divided to cells number as follow: number of AuNPs/n◦ of cells = (total mass of AuNPs by ICP-MS/mass of one AuNP)/n◦ of cells (where the mass of one AuNP = ρ 4/3 ρ <sup>3</sup>π with ρ = density of gold; r = radius of one AuNP).

### Confocal Laser Scanning Microscopy (CLSM)

SKBR-3, cells were seeded on cover glasses with a density of 50.000 cells in a 24 well culture plate. After 24 h incubation, cells were treated with Trastuzumab free (2 µg/mL), 2.5Au@PT, 5Au@PT nanoparticles (2 µg/mL) and without any control for 24 h. Untreated cells were used as control. At the end of the culture time, the cells were washed with PBS, fixed with 4% (w/v) paraformaldehyde solution for 15 min, permeabilized with 0.1% Triton X-100, and blocked with Bovine Serum Albumin (BSA 3% in 1X PBS) for 1 hour at room temperature. Finally, cells were incubated with Alexa-Fluor-488–conjugated secondary antibodies anti-human (diluted 1:1000) for 45 min at room temperature (Invitrogen; excitation/emission maxima ∼ 495/519 nm). At the end of the incubation, the cells were washed in PBS, counterstained with a solution of Hoechst 33342 (2 µg/mL in 1X PBS; excitation/emission maxima ∼361/497 nm) to target the cellular nuclei, and then washed. Finally, samples were observed with a confocal fluorescence microscope (Leica TCS SPII Microsystems, Wetzlar, Germany). The AuNPs were visualized in reflection bands after the excitation at 545 nm.

### Ultrastructural Analysis

For the ultrastructural analysis, cells underwent a Silver enhancement made with Silver Enhancer Kit (Sigma-Aldrich) in order to increase the signal of colloidal gold particles. Subsequently, cells were fixed with 2.5% glutaraldehyde in PBS and a post-fixation with 1% osmium tetroxide in dH2O was performed to fix lipids. The samples were gradually dehydrated in

acetone and embedded in epoxy resin. Thin sections of 70–80 nm were obtained with Reichert OM3 ultramicrotome and placed on formvar-carbon-coated nickel grids (300 Mesh). Finally, the grids were stained with uranyl and lead, using an unstained grid as control. All the samples were observed on a Zeiss EM900 electron microscope operating at 80 kV.

#### Endocytosis Pathway Determination

Cells were seeded in 6-well culture plates at a density of 500000 cells/well, after 24 h incubated with different endocytosis inhibitors [2.5 mM amiloride (Sigma-Aldrich), 2.5 µg/mL chlorpromazine (Sigma-Aldrich) and 100 µM indomethacin (Sigma-Aldrich)] and then maintained for 1 h at 37◦C with 5% of CO2-air. The concentrations and treatment times of each chemical inhibitors were optimized in a preliminary experiment to select the maximum non-toxic doses and treatment times. After 1 h incubation, the inhibitors were removed, and the cells exposed to 2 µg/mL AuNPs in complete medium for 24 h incubation. Then ICP-MS protocol was followed as described in the previous section.

#### Western Blot Analysis

Cells were scraped from all the samples, including T, and lysed with ice-cold lysis buffer (1X RIPA buffer containing 1 mM and 1X protease inhibitor (Protease Inhibitor Tablets, SIGMA) for 30 min on ice. The lysates were then used for western blot analysis according literature protocol (Cristofaro et al., 2018). Primary antibodies anti-phosphorylated AKT and anti-AKT (diluted 1:500), anti-ERK and anti-phosphorylated ERK (diluted 1:1000), anti-BAX and anti-BCL-XL (diluted 1:500), anti-β-actin (diluted 1:500) and appropriate secondary antibodies HRP-conjugated were used. Detection was performed as described in the dotblot section. Bands densitometry analysis was carried out with Image J software.

#### Statistical Analysis

All statistical calculations were carried out using GraphPad Prism 5.0 (GraphPad Inc., San Diego, CA, United States). Statistical analysis was performed using Student's unpaired t-test and through one-way variance analysis (ANOVA), followed by Bonferroni post hoc, for multiple comparisons (significance level of p ≤ 0.05).

### RESULTS AND DISCUSSION

#### Synthesis of AGMA1-SH

AGMA1-SH bearing 20% randomly distributed thiolfunctionalized repeat units was synthesized following a two-step procedure previously reported for obtaining thiolfunctionalized ISA23 (a cytobiocompatible, stealth-like PAA deriving from the polyaddition of 2,2-bisacrylamidoacetic acid with 2-methylpiperazine) (Donghi et al., 2009). According to this procedure (**Figure 2**), at 20% on a molar basis, cysteaminederiving units were introduced in AGMA1 by substituting in the preparation recipe 10% cystamine for 20% agmatine. Cystamine, being a tetrafunctional monomer in Michael-type polyaddition acted as a crosslinker, and a soft hydrogel was obtained. Cystamine disulfide groups were subsequently reductively cleaved with excess 2-mercaptoethanol. The major portion of AGMA1-SH repeat units still bore 4-aminobutylguanidine pendants, as AGMA1, therefore it maintained its favorable cell-adhesive properties, but also contained cysteamine moieties capable of ensuring strong interactions with gold.

### Synthesis of Gold Nanoparticles Decorated With AGMA1-SH and/or Trastuzumab

Three differently sized AGMA1-SH-coated AuNPs decorated with Trastuzumab were synthesized according to a previously established green and robust route (Comotti et al., 2004). The aim was to assess how variously sized nanosystems could affect the biological activity, in particular endocytosis and apoptosis. In order to evaluate the role of each component inside the nanosystem, three differently sized AGMA1-SHcoated AuNPs and Trastuzumab-coated AuNPs were also prepared for comparison. Tuning gold nanoparticles diameter was expected to change anti-cancer ability of the nanosystem as a consequence of different surface to volume ratios which might affect the gold interaction with the polymer and drug, as well as their surroundings. On this regard, Bond and Thompson published a plot able to determine gold dispersion per nanoparticle (percentage of surface gold atoms with respect to gold atoms per nanoparticle) as a function of nanoparticle diameter (Bond and Thompson, 1999). Surface area and gold atoms number versus diameter can be determined as well (**Supplementary Figure S1**). Accordingly, the smaller the particles, the higher the fraction of atoms on the surface thus more gold sites are available for interaction. In particular, the three tailored AuNPs sizes (2.5, 3.5, and 5 nm) would lead to marked different gold dispersions of around 55, 35, and 25%, respectively. All the nanosystems were stable over time due to the mentioned Au-S soft-soft interaction, since sulfur is present either in AGMA1-SH or Trastuzumab (Della Pina et al., 2009). Briefly, the AuNP core (produced by chloroauric acid reduction, see below) was coated with the biocompatible polymer AGMA1-SH in order to prevent the nanosystem either from aggregation or ingestion by phagocytes ('stealth effect'). The subsequent conjugation to an anticancer molecule (the antibody drug Trastuzumab) would enable gold nanovectors to recognize and inhibit only tumor cells. HAuCl<sup>4</sup> aqueous solution served as a gold source, whose concentration was tuned to achieve the desired size of AuNPs and, hence, of the whole nanosystem. Proper amounts of AGMA1-SH were then added and Au3<sup>+</sup> was instantly reduced to Au<sup>0</sup> by NaBH<sup>4</sup> addition without the presence of any conventional stabilizer (namely PVA, PVP or sodium citrate, often cytotoxic) due to the intrinsic protecting effect of the polymer. Shortly after, the polymer-coated AuNPs were decorated with the drug. Only in the case of 5Au@PT, 5Au@P, and 5Au@T samples, sodium borohydride was added prior to the polymer or antibody drug addition in order to allow gold nanoparticles to grow up to 5.0 nm. Both the polymer and the drug, in fact, were found

to hinder gold nanoparticles growth when present before the reducing agent addition, probably due to their inhibiting effect against aggregation (**Figure 1**). More importantly, the innovative protocol introduces some advantages with respect to other synthetic routes commonly adopted (Rossi et al., 2016). First, the use of thiol-functionalized biocompatible PAAs allows to avoid conventional stabilizers, often cytotoxic, via Au-S bonds, which ensure long-term stability of the nanosystem without cytotoxicity. Second, the use of an instant reducing reagent (NaBH4), whose excess can be easily removed by dialysis, leads to homogeneously sized nanoparticles, simple to be tuned by HAuCl<sup>4</sup> concentration.

### Physicochemical Characterization of Gold Nanoparticles Decorated With AGMA1-SH and/or Trastuzumab

Direct TEM observation of Au@PT complexes showed that all colloidal solutions contained monodispersed and sphericalshaped nanoparticles matching their designs (**Figure 3A**). As reported in the experimental section, the gold nanoparticle average diameter was determined by XRPD analyses using Scherrer equation. All colloidal samples were absorbed on XC72R carbon (1%Au/C) and the peak at 2θ = 38.5◦ , typical of metallic gold, was considered. A long experimental work allowed to reach the expected theoretical values for most of the samples, especially when the polymer and the drug were both present (**Figure 3B**). Size evaluation of the whole colloidal nanosystem (gold – polymer – drug) required DLS technique. DLS measures the time-dependent fluctuations in the intensity of scattered light from a suspension of particles undergoing random Brownian motion. Analysis of these intensity fluctuations allows the determination of diffusion coefficients, which in turn yield the particle size through the Stokes-Einstein equation. **Table 1** reports the hydrodynamic diameter average and Z-potential for all samples. Particles size and Z-potential are very important parameter to evaluate the stability of colloidal system (Sun et al., 2016). Absolute values of Z-potential above ± 30 mV are typically considered as an indicator of suspension stability against aggregation due to charge stabilization (i.e. the electrostatic repulsive forces are high enough to counteract aggregation) (Lowry et al., 2016). Meaningfully, Z-potential value can provide an explanation about the absorption mechanisms of drugs and biological ligands on nanoparticles surface. Likewise, it can be a key factor in determining the interaction with the cells. As listed in **Table 1**, all the AuNP spheres (with or without AGMA1-SH) showed a positive Z-potential value. However, some significant differences were observed. In absence of AGMA1-SH, the Z-potential values of Au@T nanoparticles decreased dramatically toward Z-values closed to neutrality when particle size increased, especially for the 5Au@T nanoparticles, which indicated a very low stability of the nanosystems. The reason can be attributed to the absence of the stabilizing effect of AGMA1-SH, that if present on the surface may impedes nanoparticle aggregation by steric hindrance to prevent surface interactions between AuNPs (Gao et al., 2012). After the addition of AGMA1-SH, the Z-potential value of the differently sized Au@P intensely increased. The most relevant increase was observed for 5Au@P (from 0.2 to 33.8 mW), confirming the impact of AGMA1-SH to prevent nanoparticles aggregation and deposition. After the Trastuzumab conjugation, a slight and not significant decrease on Z-values was assessed in 2.5 and 3.5Au@PT suspensions than 2.5 and 5Au@P counterparts (24.2 to 18.4 mV for 2.5Au@PT and 23.0 to 21 mW for 3.5Au@PT, respectively). With value higher than 30.0 mW, 5Au@P and 5Au@PT represent the most stable nanoparticles, underling a possible effect due to the size of the nanospheres. Numerous strategies explored the attachment of a targeting ligand (i.e. sugars and peptides) through S-Au bonds (Bertrand et al., 2014; Doulain et al., 2015). Although some drawbacks, particular promising are those exploiting these bonds for the functionalization with molecules for biosensing, anticancer drug delivery (i.e. antibodies), and bioimaging (Ghosh et al., 2008; Gautier and Cisnetti, 2012). Recently, Matos et al. (2018) used an engineered Trastuzumab presenting an additional free cysteine per light chain, obtaining a stable and homogeneous gold thiol-linked Thiomab conjugate with unmodified anticancer activities. Sulfur is a component of Trastuzumab itself (C6470H10012N1726O2013S42, hence 42 Sulfur atoms per molecule). Hence, it is reasonable to speculate that its free thiol groups may contribute on the stability enhancement of AGMA1-SH-coated gold nanoparticles interacting directly with the gold surface. The UV-vis spectra displayed a plasmon resonance band (PRB) in the strict range 498–504 nm for the smallest samples (2.5Au@PT, 2.5Au@P, and 2.5Au@T), 508– 515 nm for the intermediate-sized samples (3.5Au@PT, 3.5Au@P, and 3.5Au@T) and 517–525 nm for the largest ones (5Au@PT, 5Au@P, and 5Au@T) (**Figure 3C** and **Supplementary Figure S2**). The location of PRB peak strongly depends on the size of the metal nanoparticles, as well as on the surrounding medium and its dielectric constant. In particular, the larger the nanoparticles, the higher the wavelength. Hence, the detected PRB values are coherent with the expected theoretical trend (**Figure 3C**).

In addition to the physicochemical characteristics of all nanoparticle's formulations tested, the efficiency of the conjugation reaction was also determined (**Table 2** and

TABLE 1 | Size distribution of hydrodynamic diameters (dh) and Z-potential as determined by DLS.


TABLE 2 | Conjugation efficiency of AGMA1-SH-AuNPs.


<sup>a</sup>Conjugation efficiency% = (1 − ([Trastuzumab in the supernatant]/[Trastuzumab added in the conjugation reaction])) × 100. <sup>b</sup>µg of Trastuzumab per 20 µg nanoparticles = Trastuzumab added in the conjugation reaction − Trastuzumab in the supernatant.

**Supplementary Table S2**). Trastuzumab was conjugated to the Au@P with an efficiency of ∼50–70%. In general, the percentage of conjugation efficiency augmented by increasing the gold nanoparticles sizes. It was approximately of 50.2 ± 3.9% for 2.5Au@PT, 60.0 ± 6.3% for 3.5Au@PT and 69.4 ± 7.8% for 5Au@PT, respectively, denoting that the size of gold nanoparticles intensely impacts the binding with the targeting molecules. The relative amount of T (µg) per nanoparticles (µg) are also reported (**Table 2**). Beyond to visualize Trastuzumab on the nanoparticles as spots on the nitrocellulose membrane, dot blot assay confirmed the indirect values obtained by BCA analysis (**Supplementary Table S3** and **Supplementary Figure S3**). Size-related effect was also recorded in the Trastuzumabfunctionalized nanoparticles without AGMA1-SH. It is worth noting that, in all these nanoparticles, the amount of Trastuzumab per nanoparticles was lower than AGMA1- SH-coated gold nanoparticles (**Supplementary Table S2**). Gao et al. (2012) documented that a thiol-ending polyethylene glycol (PEG-SH), significantly improved to colloidal stability of gold nanoparticles in aqueous solution, facilitating their conjugation with epidermal growth factor receptor antibodies. Hence, presumably the AGMA1-SH presence, stabilizing the gold nanoparticles, may be crucial for the establishment of a successful interaction with the antibody (Gao et al., 2012). In nanobiotechnology applications, curvature of nanoparticles has a significant effect on protein structure and activities, although little is exactly known about molecular level mechanism. Likewise, hydration layer above the nanoparticles surface, the ionic strength of buffer solution and stabilizing polymers concentration can be critical factors for protein interaction with nanoparticles (Iijima and Kamiya, 2009; Kushida et al., 2014; Yu and Zhou, 2016; Lin et al., 2017). Kushida et al. (2014) showed that the interaction of the coagulation proteins with silica nanoparticles decreases with the decreasing NP size

FIGURE 3 | Characterization of Trastuzumab-conjugated gold nanoparticles with different sizes. (A) TEM images of 2.5 (a), 3.5 (b), and 5 (c) nm core diameter Au@PT nanoparticles (Scale bar = 200 nm), (B) XRPD, (C) UV-vis absorbance spectra (circle at 280 nm indicates the Trastuzumab peak), and (D) circular dichroism measurements of each Trastuzumab-conjugated gold nanoparticles families as function of core size.

TABLE 3 | Hydrodynamic diameter (nm) of Trastuzumab-AGMA1-SH-gold nanoparticles dispersed in complete cell culture media at two different time points.


Mean value and standard deviation of three measurements per condition are reported.

(higher curvature). Vertegel et al. (2004) observed a greater loss of lysozyme conformation adsorbed onto larger silica nanoparticles under otherwise similar conditions. About the gold nanoparticles, the effects of curvature on ligand interaction remains a complicate issue (Villarreal et al., 2017). Villarreal et al. (2017) CD spectroscopy method is commonly used to determine protein structure and its interaction with other molecules. CD spectroscopy confirmed the Trastuzumab conjugation to the gold nanoparticles (**Figure 3D** and **Supplementary Figure S4**), but also suggested a correlation between Trastuzumab-conjugated secondary structure maintenance and the particle hydrodynamic radius (R). Consistently with literature findings (Jiang et al., 2005, 2008; Worrall et al., 2006; Saptarshi et al., 2013), the correct folding increased with R growing from 2.5Au@PT to 5Au@PT (**Figure 3D**). Only the largest nanoparticles (5Au@PT) displayed CD spectra superimposable to the unconjugated antibody, indicating that the nanoparticles are a highly efficient and selective tool for HER-2 receptor. Previous studies hypothesized that the higher surface curvature of smaller nanoparticles may restrict the relative orientation between antibodies, resulting in a certain degree of conformational rigidity, that while blocking the correct protein folding, it impaired their docking receptor surface (Jiang et al., 2008). By contrast, for conjugates built on larger sizes, the antibody, once absorbed on nanoparticles surface, extended itself into the surrounding medium in a conformation allegedly better suited to interact with its receptors (Ghitescu and Bendayan, 1990; Jiang et al., 2008; Saptarshi et al., 2013). Concerning the in vitro release of Trastuzumab, each type of Au@PT presented a low released amount in MilliQ water up to 15 days of investigation (5.3 ± 0.7%, 4.4 ± 0.5%, and 3.5 ± 0.6% for 2.5, 3.5, and 5Au@PT, respectively) (**Supplementary Figure S5**). Interestingly, analyses up to 12 months showed that 5Au@PT complex was the most stable of storage at refrigerated conditions, without any aggregate's formation and with an approximately release of 20 ± 0.3%. By contrast, both 2.5 and 3.5Au@PT displayed a tendency to precipitate after few months from their synthesis. Challenging and crucial aspect of nanoparticles characterization is the measurement of their stability under condition resembling in vitro or in vivo environment. Numerous studies revealed that the proteins, salts, and antibiotics present in the different culture media can interact with AuNPs, determining radical changes in their size by forming aggregates and reducing their stability (Moore et al., 2015). Notably, all these components can confer the particles with a new biological identity and then dramatically alter their interaction with the cells (Gebauer et al., 2012). Therefore, it has been becoming increasingly evident that the nanoparticles toxicity and activity assessed in vitro entails the mixing nanoparticles of interest with the biological media for enhancing the comprehension of their interaction with the biological systems. Hence, to mimic the cell bioenvironment prior to cells treatment, freshly synthesized batches of Au@PT nanoparticles were re-suspended in the cell culture media (containing 10% of serum) proper for each cell lines used for biological investigation. Hydrodynamic changes were assessed by DLS analysis at 37◦C after 4 h of incubation. As shown in **Table 3** after serum incubation, the hydrodynamic diameter of the three types of Au@PTs (in particular for the smaller ones) increased suggesting the serum proteins adsorption on nanoparticle surface, that forming a "corona." Although with discrepancies in the results, numerous studies on different types of nanoparticles and on various cell lines highlighted the significance of protein corona (Hajipour et al., 2015). They profoundly influence the aggregation of nanoparticles and their cellular uptake (Dominguez-Medina et al., 2016). Also, it has been demonstrated that the protein corona composition affects the internalization of gold nanoparticle coated with different polymers (Walkey et al., 2014). Recently, it was found that the "stealth effect" of poly(ethylene glycol) may be explained by the presence of specific proteins in their protein corona (Schöttler et al., 2016). As summarized by Moore et al. (2015) the type of basal medium can dictate the size and stability of nanoparticles. Results from Strojan et al. (2017) demonstrated that the medium in which silica nanoparticles were dispersed, had significantly affected nanoparticles protein corona composition, suggesting an important implication on nanoparticles potential biological effects. Ji et al. (2010) have found that titanium dioxide (TiO2) nanoparticles increased their hydrodynamic in the presence of cell culture media without serum. By contrast, the addition of bovine serum protein stabilized nanoparticles, as there was only a little change in size following the addition of proteins (Spadavecchia et al., 2016). Moreover, as previously reviewed, different surface modifications (i.e. by using biocompatible polymers coating) can be used for lessening the protein corona formation then improving the stability of nanoparticles in liquid media (Iijima and Kamiya, 2009). Based on and in agreement with literature findings (Colombo et al., 2016), we inferred that the small increase observed on 5Au@PT nanoparticle size, in all complete media tested, might due to the higher surface functionalization with T, which decreased the extent of serum protein adsorption. By contrast, the lower amount of antibody per nanoparticle on 2.5Au@PT and 3.5Au@PT may favor protein absorption then resulting in an increase of hydrodynamic diameter. Overall, protein adsorption may be hindered by the functionalization of gold nanoparticles with AGMA1-SH and Trastuzumab, but it cannot be totally inhibited, and a certain amount of serum proteins do adsorb on the NP surfaces, irrespective to the presence of polymers coating, justifying then the small increase of the hydrodynamic size.

It is worth emphasizing that the NPs characterization in standard conditions for the in vitro biological experiments, i.e. in serum-containing cell culture medium, is still a questioned issue. Indeed, additional determining factors can regulate the nanoparticles amount reaching the cell monolayer and the type of interaction with cells. Cell types and its protein component, including the protein secreted by cells, contribute to affect the deposition of the nanoparticles in the cell-well (Halamoda-Kenzaoui et al., 2015). Recently, it was observed that nanoparticle administration method in vitro cell culture could alter particlecell interaction (i.e. cellular adsorption and uptake) (Moore et al., 2019). All the factors make hard to envisage the exact behavior of the nanoparticles once in contact with the cells and should to be taken in account during the biological experiments. Whilst bearing in mind that accurate experiments are required, overall

characterization proved the efficacy of the use of AGMA1-SH to produce stable gold nanoparticles. Consequently, their anticancer activities were subsequent evaluated by in vitro cell models as follows described.

### In vitro Cytotoxicity and Uptake: Evaluation in Breast Cancer and No-Neoplastic Cell Lines

In vitro and in vivo successful achievements employing delivery synthetic nanoplatforms need targeting molecules to recognize tumor cells specifically, without affecting the surrounding healthy cells.

Aiming to assess the activities of Au@P functionalized with Trastuzumab, SKBR-3, human breast cancer cell line was selected as target cell model. This cell line overexpresses HER2, itself the target of Trastuzumab, and has been used extensively as an in vitro model for evaluating the anticancer targeting of nanoparticles as tool to enhance the selectivity and specificity of therapies against cancer cells (Kim et al., 2011; Martínez-Jothar et al., 2019; Niza et al., 2019). Moreover, to confirm the selective effect of Au@PT toward the HER2-receptor, others two cell lines were included for comparative analysis: MCF-7, a human breast cancer expressing low HER-2 as control, and the murine fibroblasts cell line, NIH-3T3, as health cell model. Both are commonly used in experimental studies for assessing the selectivity of anticancer treatments and the nanomaterials toxicity (Xu et al., 2005; Holliday and Speirs, 2011; Shirshahi et al., 2013; Merli et al., 2018; Cruz and Kayser, 2019). All in vitro experiments were conducted treating all cell types (1 × 10<sup>4</sup> viable cells/well) with increasing concentrations (from 0 to 2 µg/mL) of each Au@PT size and free-Trastuzumab. The concentration range was fixed by determining the "no toxic" dose by incubating the same number of cells with increasing concentrations of the unfunctionalized Au@P counterparts (**Supplementary Figure S6**). A low cytotoxicity was observed up to 2 µg/mL, then the concentration range upper limit was set at this value. As reported in **Table 2**, 2 µg/mL of 2.5Au@PT, 3.5Au@PT, and 5Au@PT contain approximately 1– 1.4 µg of Trastuzumab, which was the concentration within the effective doses against breast cancer cells (Suarez et al., 2013). 2.5Au@PT, 3.5Au@PT, and 5Au@PT cytotoxicity was tested by measuring cell viability using MTT after incubation with Au@PT. 2.5Au@PT and 5Au@PT were particularly effective in impairing the SKBR-3 viability, whereas they had low or no toxicity on MCF-7 and NIH-3T3 respectively (**Figure 4A**). Interestingly, cells exposed to Au@PT underwent viability decrease after 3 days of incubation (around 50 and 20% with 5Au@PT and 2.5Au@PT respectively). It could be argued that the difference in cells response is due to the changes observed in the secondary structure of the antibody, which was able to affect the binding capacity of Trastuzumab-AuNPs with HER-2 receptors. 3.5Au@PT was comparable to that of 2.5Au@PT treated cells, nearby 25% at 2 µg/mL (data not shown). Targeting molecule may reduce their targeting capacity and activities as protein corona adsorption (Mirshafiee et al., 2013). Also surface rigidity, ligand density affects nanoparticles effect on cell (Gong et al., 2015). Consequentially, it is conceivable that lower activity detected after the treatment with 2.5Au@PT, the ones showing the highest increase of hydrodynamic size following the incubation with the complete cell media, may be related to protein corona absorption. However, other explanations cannot be excluded. Pan et al. (2007) have demonstrated that the cytotoxicity of modified gold nanoparticles is dependent on the size and not on a particular ligand attached to them. In addition, some researchers showed that because metallic, these nanoparticles have an effect on cell membrane integrity, and their size and charge both affect significantly cell viability (Wozniak ´ et al., 2017). Similarly, cellular uptake of gold nanoparticles was influenced by many factors including size, charge, coating, shape, incubation time and cell types (Dreaden et al., 2012; Panariti et al., 2012; Yue et al., 2017). Chithrani et al. (2006) observed that size and shape strongly influenced cellular uptake of citric acid ligand-modified Au nanospheres, and others reported a clear interrelationship between nanoparticles surface functionality an uptake (Jiang et al., 2015). Meanwhile, the adhesion of nanoparticles onto cell membrane can be reduced because of protein corona formation around nanoparticles surfaces, which in turn inhibit their cellular uptake. As reported by Cheng et al. (2015) protein from serum, modulating biomolecular corona profile, changed the AuNPs internalization in a sizeand cell type-dependent manner. Aiming to explore the size contribution in Au@PT uptaking, the succeeding experiments were performed using nanoparticles differing for about fourfold in hydrodynamic radius, 2.5Au@PT and 5Au@PT. Others have previously underlined that an equilibrium between multivalent crosslinking of membrane receptors and membrane wrapping involved in receptor-mediated endocytosis exists (Jiang et al., 2008; Yue et al., 2017). Jiang et al. (2008) showed that extremely small (2 nm) or large Trastuzumab-conjugated nanoparticles (>50 nm) would both produce an inefficient uptake. Aoyama et al. (2003) and Osaki et al. (2004) investigated size effects and receptor activation, concluding that receptormediated endocytosis is firmly size-dependent with an optimal size equal to ≈25 nm. Others reported a theoretical model based on size and receptor-mediated endocytosis that gives an optimal radius ≈27–30 nm for spherical particles (Gao et al., 2005). Besides, Wang et al. (2010) using a 3D image process identified the AuNPs distribution, revealing that the optimal size is 45 nm for cells uptake. The results here described seem to show a broader agreement with previous experimental observations. Indeed, as compared to 2.5Au@PT (d<sup>h</sup> ∼ 6 nm), a significant enhancement in the uptake of 5Au@PT (d<sup>h</sup> ∼ 22 nm) was detected in HER-2 overexpressing human breast cancer SKBR-3 cells (**Figure 4B**). Nonetheless, the internalization of both 2.5Au@PT and 5Au@PT was highly dependent on time (**Figure 4B**). Large nanoparticles aggregates are not suitable for cellular interaction and uptakes. Using SEM-BSE mode, it was easily appreciated smaller 5Au@PT cluster coating homogenously SKBR-3 cells surface, whereas a closely 2.5Au@PT were packed into bigger aggregates (**Supplementary Figure S7**), suggesting the difficult capture of the latter by cells. In addition, uptaking studies using unconjugated-Au@P confirmed the role played by the size, with the most efficient

uptake occurring with the largest nanoparticles (5Au@P; d<sup>h</sup> ∼ 22 nm) (**Supplementary Figure S8**).

The accumulation of Au@P and Au@PT inside the cells could be regulated also by the exocytosis process. It was found that Au NPs were exocytosed in a linear manner in size, with smaller AuNPs exhibiting a faster exocytosis rate and percent (after 8 h of incubation, 4 nm-AuNPs were exocytosed while > 90% of 74 nm remains into the cells) (Chithrani and Chan, 2007). Equally, the thermodynamic process could be the driving force for wrapping. In general, smaller particles must be clustered together to create enough driving for uptake, showing an uptake amount smaller than 50 nm AuNPs (Gong et al., 2015). Recently, Cruz E and Kayser V have found that Trastuzumabattachment increased nanoparticles cellular uptake in HER2 amplified cell lines selectively (Cruz and Kayser, 2019). Although higher uptake is required in delivery applications, internalization must be specific to the targeted tumor cells. In agreement, the observation that Au@PT internalization was higher in HER2 overexpressing cell line (SKBR-3) than in both MCF-7 and NIH-3T3, it was a further evidence that cellular uptake increased through the HER2 receptor crosslinking Trastuzumab-activated (**Figure 4B**). Likely, the low percentage of uptakes detected in MCF-7 and NIH-3T3 was due to passive internalization (**Figure 4B**) (Zhao and Stenzel, 2018). Later, in line with previous studies(van der Meel et al., 2012; Martínez-Jothar et al., 2019), SKBR-3 cell line showed a significantly increased uptake for Trastuzumab-conjugated AuNPs as compared to nontargeted nanoparticles (50% vs. 10%, respectively after 3 days of incubation); these data further proved that the uptake of antibody-targeted nanocarriers is truly mediated by specific ligand-receptor interactions.

### HER Targeting: Evaluation of Morphology, Viability and Intracellular Changes After Au@PT Treatment

The correct binding of a ligand molecule with cell surface receptors intimately controls the cell function and fate. Many works showed that the physical parameters of nanoparticles (i.e. size) strongly affect ligand binding and the subsequent activation of its membrane receptors (Chithrani et al., 2006; Brzóska et al., 2018). In line with quantitative results, FDA assays showed the highest number of death cells (red signal) in cultures exposed with 5Au@PT suspension (**Figure 5A**). SEM images clearly highlighted this important difference, displaying more spherical cells in shape with noticeable abnormal morphology in 5Au@PT-treated samples in comparison with control, free Trastuzumab or 2.5Au@PT (**Figure 5A**). The binding of

indicated as \*\*\*p < 0.001, \*\*p < 0.01, and \*p < 0.05.

the ligand to its surface receptors allows the formation of multivalent HER-crosslinking surface ErbB2 receptors, affecting intracellular signaling, consequently inhibiting downstream pathways involved in cell survival, proliferation, and metastasis (Klapper et al., 2000).

The strategy of combining the properties of AuNPs with anticancer molecules could be the future of cancer nanomedicine (Sun et al., 2018). Chattopadhyay et al. (2010) showed that Trastuzumab-PEG-AuNPs (30 nm) in combination with 300 kVp X-rays enhanced DNA double strand breaks (DSBs) in SKBR-3 cells. Mechanistic studies performed by Vemuri et al. (2019) revealed that the functionalization of naturally derived phytochemicals, such as Quercetin and Paclitaxel, to AuNPs were significantly effective in inhibiting cell proliferation, apoptosis, angiogenesis, colony formation and spheroid formation. AuNPs principally trigger apoptosis through intrinsic pathways, including mitochondria- and ER-related pathways. Green synthesized of AuNPs (10 – 42 nm) can induce the apoptosis activating the caspase-3 and 9 in human cervical cancer cells (Baharara et al., 2016) and the citrate-coated AuNPs (8 nm)

promoted apoptosis eliciting BAX translocation and cytochrome c release in human liver (HL7702) cells (Gao et al., 2011). In agreement with literature findings (Sun et al., 2018; Vemuri et al., 2019), Au@PT cytotoxic effects elicited significant changes in survival and proliferation (AKT/PI3K and ERK pathways, respectively) as well as apoptotic process (pro-apoptotic BAX, anti-apoptotic BCL-xL, respectively) (**Figure 5B**). AKT and ERK activation levels were significantly reduced in the treatment with 2.5Au@PT and 5Au@PT than with control (**Figures 5Ba–c**), with a slight decrease after the incubation with 5Au@PT. Additionally, both 2.5Au@PT and 5Au@PT treatments led to a substantial increase in the expression of BAX with concomitant decrease in expression of BCL-xL in comparison with free Trastuzumab and control (**Figures 5Bd–f**). As mentioned, the most marked changes were observed after 5Au@PT incubation, still proving the size-effect dependence. Overall, the results showed that Trastuzumab-AGMA1-SH-gold AuNPs conjugates retain great antimitotic and anti-apoptotic effectiveness after the conjugation to Au@PT, which could improve the cancer therapeutic effects of Trastuzumab itself.

### HER-2 Targeting: Evaluation of Internalization by Confocal Laser and Transmission Electron Microscopies

As a drug delivery platform, an efficient cell uptake is important to ensure sufficient therapeutic outcomes (Carnovale et al., 2016). Some studies suggest that Trastuzumab can induce a rapid internalization of HER-2 (Rubin and Yarden, 2001; Chattopadhyay et al., 2010; Han et al., 2014), a mechanism that could facilitate 5Au@PT uptaking by SKBR-3 cells. Colocalization of HER-2 and gold nanoparticles were then studied (**Figure 6**). In cells treated with free Trastuzumab, the antibody staining (in green) was predominant in cell membrane because of its interaction with HER-2 receptors (**Figures 6Ab,c**). On the other hand, cancer cells incubated with 5Au@PT exhibited a reduced Trastuzumab staining at the cell membrane (**Figures 6Ad,e**) but increased inside the cells and co-localized with 5Au@PTs as revealed by the scattering signals of gold (in red) (**Figures 6Ad,e**).

These data, besides proving the powerful and selective ability of Trastuzumab to promote the transport of 5Au@PT into HER-2 overexpressing cells, also suggest that receptors undergo endocytosis process when bound to Au@PT of a specific size range.

Transmission electron microscopy provides an excellent tool for in-depth biological sample analysis at the cellular and organ level. In the present work, it confirmed SKBR-3 cellular internalization of 5Au@PTs (**Figure 6B**). In (c) at higher magnification, the distribution of the NPs in close proximity of the vesicle's membrane seems to evidence the binding between the drug and its receptor: a cluster of 5Au@PT was approaching the cell membrane (d) and was trapped inside vesicles of ranging sizes (e). The electron dense material appears to have a low resolution due to the silver enhancement procedure. However, a series of control experiments in absence of silver, osmium, uranyl and lead ruled out the possibility of artifacts as shown in **Supplementary Figure S9**. Generally, AuNPs were localized within early endosomes and lysosomes (Boyoglu et al., 2013; Xie et al., 2017). Therefore, it is presumable the vesicles structures visualized by TEM were the result of the

TEM samples stained with uranyl and lead for cell morphology: (a) ctrl cells; (b) cells treated with 5Au@PT; uptake of large clusters inside vesicles is clearly visible after silver enhancement; (c) detail of a cell vesicle containing 5Au@PT; (d,e) high magnification of 5Au@PT putative uptake nearby the cell membrane; scale bars: 2 µm, 2 µm, 500 nm, 500 nm, and 500 nm respectively.

endosome/lysosome trafficking induced by Trastuzumab binding to its membrane receptor. Some of these structures may be also autophagosome. Previously, it was observed gold nanoparticles induces autophagosome accumulation in accordance with sizedependent nanoparticle uptake and lysosome impairment (Ma et al., 2011). A high-resolution technique may help to clarify the full pathway of Au@PT intake and leading to a precise subcellular localization.

### HER-2 Targeting: Molecular Mechanisms Underlying the 5Au@PT Capture

In conclusion, to underscore the sensitive endocytic route responsible for 5Au@PT internalization by target cells, three different endocytosis pathways inhibitors were employed (**Figure 7**): amiloride for micropinocytosis inhibition, chloropromazine for clathrin-mediated endocytosis (receptor dependent mechanism) and indomethacin for caveolae-mediated endocytosis inhibition. The uptake of 5Au@PT was dramatically affected by chloropromazine treatment (around 50% of reduction), whereas the other two inhibitors did not substantially modify the delivery into the cells (**Figure 7A**). Cross-section analysis confirmed the aforementioned results, revealing a strong scattering distribution along the cellular surfaces due to inhibition treatment (**Figure 7B**). Receptor-mediated clathrin carries out the uptake of specific macromolecules (ligands) or ligand-coated NPs following their binding to receptors on the surface of cell membrane, promoting the invagination of the plasma membrane to form area with a higher receptor concentration (Zhao and Stenzel, 2018).This type of mechanism has been widely used strategy for the drug delivery by nanocarriers (i.e. nanogel, liposomes). Saporin-loaded nanobody-targeted polymeric nanoparticles induced HER2 clustering, and promoting its internalization causes an efficient NPs receptor-mediated endocytosis with subsequent intracellular delivery of its cytotoxic cargo (saporin) (Martínez-Jothar et al., 2019). As regard AuNPs, Trastuzumabfunctionalized AuNPs mostly underwent endocytosis in breast cancer cells largely through a receptor-facilitated mechanism (Chattopadhyay et al., 2010; Han et al., 2014; Rathinaraj et al., 2015; Sun et al., 2018). Han et al. (2014) found that quantum dots-HER bound specifically to SKBR-3 membrane, inducing their internalization via a receptor-mediated mechanism. In accordance with the aforementioned findings, the present results showed that the most probable mechanism for 5Au@PT nanoparticle endocytosis in breast cancer cells (SKBR-3) was receptor-dependent, suggesting that the AuNPs-mediated delivery of monoclonal antibody toward its specific cell membrane receptor was achieved efficiently.

In summary, the conjugation of the 5Au@Ps with Trastuzumab allowed their specific binding to HER-2 overexpressing cells, and through a receptor-mediated internalization process, may trigger their deposition into the cytoplasm, where gold nanoparticles, in combination with free Trastuzumab released into the cells, likely contribute to breast cancer cell death.

### CONCLUSION

The synthesis and the functionalization of gold-nanoparticles with cancer-specific biomolecules may represent a winner strategy for a selective and targeted tumor-phototherapy. Currently, a considerable number of efforts have been taken in order to obtain gold-nanoparticles by green and safe method. In this work, extra-small gold nanospheres stabilized with thiolfunctionalized AGMA1 and Trastuzumab were for the first time synthesized and tested in vitro as nanovectors for breast cancer targeted drug delivery. The synthetic method followed an innovative protocol introducing significant advantages with respect to other routes presently adopted. In primis, tailoring thiol-functionalized biocompatible PAAs avoided the use of conventional stabilizers, often cytotoxic, via Au-S softsoft interactions, which ensured long-term stability of the hybrid nanomaterial without any cytotoxicity. Importantly, sulfur is even a component of Trastuzumab itself. This represented an intrinsic, powerful tool for further stabilizing

the composite nanosystem. Moreover, the employment of an instant reducing reagent (NaBH4) guaranteed homogeneously sized and tailored 'on demand' gold nanoparticles by properly tuning HAuCl<sup>4</sup> concentration. The cytotoxicity and cellular uptake behaviors have been investigated in no-neoplastic cells (NIH-3T3) and in two breast cancer cell lines (MCF-7 and SKBR-3) by MTT assay and inductively coupled plasma mass spectrometry (ICP-MS) respectively. It was found that 5Au@PT with the highest hydrodynamic diameter and positive charge showed much higher cell internalization ability and cytotoxicity than the smaller ones (2.5Au@PT and 3.5Au@PT). Moreover, the experimental results against Trastuzumab target cells (SKBR-3) demonstrated that the cytotoxicity of 5Au@PT was closely related to pro-apoptotic protein increase, anti-apoptotic components decrease, survival-proliferation pathways downregulation and uptaking by cells via the activation of the classical receptormediated endocytosis. This work thus suggests an utmost importance of size control at the AuNPs when designing molecular-cancer delivery systems. Indeed, nonetheless the present method results suitable for AuNPs synthesis, the stability and conjugation with anticancer molecules depends on particles size, which affects also the its interaction with cancer cells. AuNPs unfortunately are not the optimal candidates for photothermal therapy (PTT) due to their limited absorption in the near infrared (NIR) region. In accordance with other studies, to overcome the size dilemma is possible to assemble small AuNPs into larger structures through controllable interparticle interaction by using biodegradable AGMA1-SH, with the final aim to enhance the absorption of NIR light. Overall, this project should offer important data for particle design improvement and for speeding up AuNPs application in breast cancer therapy with enhanced and personalized therapeutic outcomes.

### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors to any qualified researcher on reasonable request.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

LV, ER, and CD conceived the study, were responsible for the correctness of analyses, and contributed to writing and editing of the manuscript. All authors contributed to critical experiments and insights, participated to this manuscript writing, and approved the final manuscript.

#### FUNDING

Fondazione Umberto Veronesi Fellowship - Grant 2017 and Grant 2018, Piano di Sostegno alla Ricerca – Linea 2 Azione A 2018 Università degli Studi di Milano, project PSR2018\_DIP\_005\_DELLA\_PINA\_CRISTINA, Pavia University's Crowdfunding on breast tumor studies (2015, https://universitiamo.eu/en/campaigns/tumore-seno-diagnosicura-nanoparticelle-doro) and a grant of the Italian Ministry of Education, University and Research (MIUR) to the Department of Molecular Medicine of the University of Pavia under the initiative "Dipartimenti di Eccellenza (2018–2022)."

### ACKNOWLEDGMENTS

We are grateful to A. Rocchi A for her contribution in the biological part, P. Vaghi (Centro Grandi Strumenti, https://cgs. unipv.it/, University of Pavia, Italy) for the technical assistance in the CLSM studies, to Lucia Cucca (University of Pavia) for ICP analysis, to Giovanna Bruni (University of Pavia) for SEM studies support, and to M. Bordoni (University of Pavia) for revision of the English text.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2020.00132/full#supplementary-material

immunotargeted gold nanoshells and two-photon excitation microscopy. Nanotechnology 19:315102. doi: 10.1088/0957-4484/19/31/315102


factor. Toxicol. Appl. Pharmacol. 356, 140–150. doi: 10.1016/j.taap.2018. 08.005


triggers nanoparticle aggregation. ACS Nano 10, 2103–2112. doi: 10.1021/ acsnano.5b06439


biological media and its influence on cellular Uptake. PLoS One 10:e0141593. doi: 10.1371/journal.pone.0141593


of pancreatic cancer: from synthesis to proof-of-concept in vitro studies. Int. J. Nanomed. 11, 791–822. doi: 10.2147/IJN.S97476


**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 © 2020 Bloise, Massironi, Della Pina, Alongi, Siciliani, Manfredi, Biggiogera, Rossi, Ferruti, Ranucci and Visai. 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.

# Emerging and Innovative Theranostic Approaches for Mesoporous Silica Nanoparticles in Hepatocellular Carcinoma: Current Status and Advances

#### Yaoye Tao1,2,3,4, Jianguo Wang1,2,3,4 and Xiao Xu1,2,3,4 \*

<sup>1</sup> Division of Hepatobiliary and Pancreatic Surgery, Department of Surgery First Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China, <sup>2</sup> National Health Commission (NHC) Key Laboratory of Combined Multi-Organ Transplantation, Hangzhou, China, <sup>3</sup> Key Laboratory of the Diagnosis and Treatment of Organ Transplantation, Chinese Academy of Medical Sciences (CAMS), Hangzhou, China, <sup>4</sup> Key Laboratory of Organ Transplantation, Hangzhou, China

#### Edited by:

Rawil Fakhrullin, Kazan Federal University, Russia

#### Reviewed by:

Nanasaheb D. Thorat, University of Limerick, Ireland Monica Terracciano, University of Naples Federico II, Italy

> \*Correspondence: Xiao Xu zjxu@zju.edu.cn

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology

Received: 15 October 2019 Accepted: 25 February 2020 Published: 10 March 2020

#### Citation:

Tao Y, Wang J and Xu X (2020) Emerging and Innovative Theranostic Approaches for Mesoporous Silica Nanoparticles in Hepatocellular Carcinoma: Current Status and Advances. Front. Bioeng. Biotechnol. 8:184. doi: 10.3389/fbioe.2020.00184 Hepatocellular carcinoma (HCC) is one of the most prevalent and lethal solid cancers globally. To improve diagnosis sensitivities and treatment efficacies, the development of new theranostic nanoplatforms for efficient HCC management is urgently needed. In the past decade, mesoporous silica nanoparticles (MSNs) with tailored structure, large surface area, high agents loading volume, abundant chemistry functionality, acceptable biocompatibility have received more and more attention in HCC theranostic. This review outlines the recent advances in MSNs-based systems for HCC therapy and diagnosis. The multifunctional hybrid nanostructures that have both of therapy and diagnosis abilities are highlighted. And the precision delivery strategies of MSNs in HCC are also discussed. Final, we conclude with our personal perspectives on the future development and challenges of MSNs.

Keywords: mesoporous silica nanoparticles, hepatocellular carcinoma, theranostic, precision delivery, biomedical applications

#### INTRODUCTION

Liver cancer is currently the fourth primary cause from cancer-related deaths and its incidence and mortality is still increasing, with an estimated 841,080 new cases and 781,631 deaths from this disease in 2018 (Llovet et al., 2016; Bray et al., 2018; World Health Organization [WHO], 2018). Among all primary liver cancers, hepatocellular carcinoma (HCC) represents approximately 90% of all cases. Surgical resection and liver transplantation are considered the curative therapies for long-term control of HCC, however, the majority of HCC patients are diagnosed at advanced stages beyond the standard of surgical treatment (GBD, 2013; Roberto et al., 2016; Yegin et al., 2016). A few molecular targeting drugs such as sorafenib (SO) approved for advanced HCC, which show merely a marginal survival benefit contrasting with conventional drugs. Unfortunately, its efficacy and adverse effects for HCC patients remained unsatisfactory (Bruix et al., 2015; Gao J. et al., 2015). Therefore, new treatment and diagnose modalities for the management of HCC are urgently warranted.

With the development of nanotechnology, nanomaterials modified as multifunctional nanoplatforms for cancer therapeutics, diagnostics, or both (known as "theranostics") attracted increasing attention (Wang J. et al., 2016; Xu et al., 2018; Kesse et al., 2019). Since the first report using silica nanoparticles (Cornell Dot) accepted by the United States Food and Drug Administration (FDA) for a stage-I human clinical trial in 2011 (Benezra et al., 2011), the recent decade has witnessed a steadily increase in research on biomedical application of mesoporous silica nanoparticles (MSNs) in the liver cancer (**Figure 1**). In general, MSNs show the following unique structural and biomedical properties:


In addition, MSNs synthesized with a hollow core called hollow-type MSNs are capable of encapsulating a superhigh dose of agent, typically more than 1000 mg agent per 1 g silica, which is obviously higher than those by other nanoparticles (Zhu et al., 2005a,b).


These distinctive features endow MSNs with unique advantages to encapsulate a variety of therapeutic and bioimaging agents and implement the desired functions. To give an overview of recent progress of MSNs in theranostic for HCC, this review is arranged as follow. Firstly, it outlines precision delivery strategies of the agents in MSNs to HCC sites and cells. Next, the current state of the research of using MSNs in the field of HCC theranostics are highlighted. Finally, we vision the future advancements for MSNs.

#### PRECISION DELIVERY STRATEGIES OF MSNS IN HCC

Nanoparticles, designed to deliver agents preferentially to the HCC tissues and cells, provide the precondition for overcoming the shortcomings of conventional treatment and diagnose approaches (Bae et al., 2011). The targeting ability to tumor not only enhances the effects of agents, but also controls dose-limiting side effects in other tissue. A well-designed nanosystem always includes multiple delivery strategies to reach a high accumulation in tumor. In this part, we discuss the delivery strategies developed in MSN-based HCC theranostics (**Figure 3**).

#### Passive Targeting

It is well recognized that liver tumors display remarkable extensive angiogenesis with defective vascular structure, accompanying with impaired lymphatic drainage system (Zhu et al., 2011). Therefore, the vascular networks of HCC have an increased permeability to circulating nanoparticles, while the lymphatic system has a reduced disposal rate to internalized nanoparticles, which allow MSNs to accumulate in HCC tumor

(D) Reproduced with permission from Liu et al. (2017). (E) Reproduced with permission from Lee et al. (2016). (F) Reproduced with permission from Lv et al. (2015).

interstitial space (Maeda et al., 2000). This so-called enhanced permeation and retention (EPR) effect has been considered as basics for achieving passive targeting in the nanosystems (Maeda et al., 2013).

Particle size, shape, and surface chemistry of MSNs could greatly influence the EPR effect of the nanoparticles (Lee et al., 2009; Maeda et al., 2013; Li et al., 2016). For instance, Meng et al. demonstrated 50 nm MSNs coated with PEI-PEG copolymer yield a intratumoral accumulation of about 12% of the total dose, which is significantly higher compared to 1% of 100 nm phosphonate-coated MSNs and 3% of 50 nm PEGylated MSNs. The additional cationic polymer coated on the MSNs ameliorated the potential downside of PEG surface. In conclusion, size tuning and decoration of the MSNs with PEI-PEG copolymer lead to an dramatic enhancement of EPR effect and sufficient accumulation in tumor (Meng et al., 2011a). Besides, a research from Harvard Medical School showed that combined radiation and cyclophosphamide could enhance tumor-associated vascular leaking, leading to a sixfold increase of nanoparticles accumulation in tumor (Miller et al., 2017).

#### Active Targeting

EPR-mediated passive targeting always lacks specificity for different tumor tissues and tumor development stages (Natfji et al., 2017). To improve the targeting efficiency, active targeting strategies have gained much attention recently.

Owing to the overgrowth and abnormality of HCC, many receptors are usually upregulated on the surface of HCC cells, compared to other normal cells. Through the recognition of these receptors by targeting ligands on MSNs, more smart targeting strategies have been achieved. The targeting ligands now used for MSN-based HCC theranostics include lactobionic acid (Zhang et al., 2012), folic acid (Chen et al., 2019), arginine-glycine-aspartate (RGD) (Chen et al., 2012), transferring (Hao et al., 2017), hyaluronic acid (Lee et al., 2018), low-density lipoprotein (LDL) (Ao et al., 2018), and others (**Table 1**).

In another way, magnetic mesoporous silica nanoparticles (M-MSNs) with superior magnetic properties maintaining the excellent advantages of MSNs could achieve magnetic-mediated targeting functions under external magnetic fields (EMFs) (Shao D. et al., 2016; Wang Y. et al., 2016; Tang et al., 2017). The targeting capacity guided by EMFs including two aspects: on the one hand, M-MSNs would mainly accumulation around the tumor site under the EMFs, which could apply a magnetic force on nanoparticles to enhance the EPR effect and overcome the drag experienced in the blood flow (Thorat et al., 2019a); on another hand, EMFs effect could enhance endocytosis of the tumor cells. Interestingly, under the EMFs about 15% higher fluorescence intensity of nanoparticles was detected in HepG2 cells than that of the non-magnetic field untreated HepG2 cells, while this phenomenon is absence in HL-7702

and cells. Final, stimuli-responsive release in tumor tissues and cells would realize by virtue of pH, redox, light, and so on.

cells, which means the effect of EMFs is selective for HCC cells (Xing et al., 2018).

### Stimuli-Responsive Release

Nano delivery systems are ideal with "zero premature release" before arriving the disease foci. The stimuli-responsive release can realize this point in response to internal stimuli in tumor microenvironment or external stimuli (Thorat et al., 2019b).

#### Internal Stimuli

Among internal stimuli, PH-responsive release plays the most promising role in HCC, since a more acidic extracellular (pH ≈ 6.8) environment is usually formed around solid tumor tissues than normal tissues and blood (pH ≈ 7.4) due to increased acid production resulting from high glycolysis (Lu et al., 2018). It has been found that electrostatic interactions between the positively charged agents [such as doxorubicin (DOX)] and negatively charged MSNs were greatly reduced by protonation in a low pH condition, which results in a pHresponsive release of DOX (Chen et al., 2018). Similarly, some gatekeepers over the pore entrance to reduce premature release of agents could be protonated under acidic pH, rapidly collapsing, thus the agents would be released from the nanoparticles (Chen et al., 2019). Besides, some chemical bonds connecting the gatekeepers and MSNs, which would be broken down by pH stimulation, are also be used in pH-responsive release (Liu et al., 2016). Differentiated from the common single pH-responsive systems, a novel cascade pH stimuli triggering nanosystem had been developed (**Figure 4A**). At first, benzoic-imine bonds would be dissociated in the tumor microenvironment pH signal (6.8) to release PEG and improve cellular uptake. Then, boronic acid-catechol ester bonds would be hydrolyzed in the endosome/lysosome pH signal (4.5–6.5) to release more drug in tumor cells, which leads to significant tumor growth inhibition. In tumor-bearing mice, the increased life span of mice treated with this nanoparticles was raised 42.4% than no pH-responsive nanoparticles (Liu et al., 2016).

In addition, HCC cells always undergo oxidative stress, which means the overproduction of various reactive oxygen species (ROS) in tumor cells. In the meantime, HCC cells also have an elevated glutathione (GSH) level to protect themselves under ROS cytotoxicity (Cheng S.B. et al., 2017; Jiang et al., 2017). Consequently, GSH-responsive become one of most popular redox-responsive precision delivery strategies because of the higher GSH level in intracellular matrix of HCC cells than that in extracellular matrix or intracellular matrix of normal


cells (Chen X. et al., 2017). The disulfide bond which could be cleaved by GSH is widely used to connect the gatekeepers and MSNs (Chen X. et al., 2017; Tang et al., 2017; Wang J.K. et al., 2017). In one study, the surfaces of MSNs were functionalized by cytochrome c (CytC) via disulfide bonds, which would be rapidly cleaved in HCC cells to release the loaded drug. Around 78.9% of agents was released from MSNs with stimulus of reductive signals after incubation for 3 h, whereas only 5.11% of agents was released for the group without reductive signals (Zhang et al., 2014). Moreover, Yue et al. presented mesoporous organosilica nanoparticles (MONs) containing disulfide bridges inside, synthesized on the framework of MSNs (**Figure 4B**). In cancer cells with the high concentration of GSH, MONs would be broken into small pieces leading to greater targeting drug release. The fluorescence intensity of DOX in the MONs treated cells was approximately 20% higher than MSNs treated cells (Yue et al., 2018a). In a similar way, the ROS-responsive MSNs are also developed for HCC theranostics (Pei et al., 2018).

In other reports, it has been well established that there is over expression of some enzymes in HCC, such as protease (Menard et al., 2019), glycosidase (Hakeem et al., 2016), and esterase (Xia et al., 2017). Once the enzyme is found at higher level in the tumor site, the MSNs can be programmed to targeting release of agents via enzymatic conversion of the carrier. Some microRNAs such as miR-122 are abundant in liver cancer. Therefore, one type of microRNAs-responsive MSNs in Huh7 cells by hybridization between antagomir-122 and endogenous miR-122 obtained a unequivocal success (Yu et al., 2015).

#### External Stimuli

Compared with the above microenvironment-responsive systems, external stimuli-responsive systems can be easily manipulated to precisely achieve spatiotemporal control and on-demand agents release (Lin et al., 2018). The main drawback of intrinsic stimuli is that internal environment of body is complicated and unmanageable, especially for the highly heterogeneous tumor tissues, which maybe results in an uncontrolled release (Thorat et al., 2019a).

Among the various light stimuli, near-infrared (NIR) has the great advantages of minimal absorption and deep penetration into tissue (Wu et al., 2015; Sun et al., 2017). The NIR-I (700– 950 nm) and NIR-II (1000–1350 nm) are the most widespread regions for light-responsive systems. The MSNs hybridized with photoabsorbing materials exploit the fine photothermal effect limited within tumor tissues through converting light energy to thermal energy, which is a particularly promising phenomenon

for use in light-responsive agents release (**Figure 4C**; Lee et al., 2016). NIR-thermal agents precision delivery can be reasonably attributed to the following reasons: first, the Brownian motion of agents would be accelerated by the photothermal effect; second, the photothermal effect could increase tumor cells membrane permeability; third, the photothermal effect may destabilize the membrane of endosome and thus facilitate escape of the agent from the endosome (Lee et al., 2016; Shao T. et al., 2016; Chen et al., 2018).

Besides aforementioned applications, M-MSNs are not only for magnetic-mediated targeting, but also for stimtli-responsive release. Alternating magnetic field (AMF) can heat M-MSNs by the magnetothermal effect, and then, the thermal fluctuations within M-MSNs trigger agents release (He and Shi, 2014; Wang et al., 2018). The mechanisms of the magnetothermal effect by these nanoparticles are related to brownian motion, neel relaxation, and hysterysisloss. And the release profile of agents can be regulated by changing the field strength and frequency of AMF. By the way, magnetic fluid hyperthermia will cause damage on tissues surrounding the nanoparticles to kill cancer cells. But for safety of healthy tissues, combination of field amplitude of about 10 kA m−<sup>1</sup> and a frequency of about 400 kHz was suggested for stimulation (Thorat et al., 2019a).

In the future, with advance of nanotechnology, more various stimuli-responsive will be realized in MSNs for precision delivery.

#### MSNS IN HCC DETECTION AND DIAGNOSIS

#### MSNs-Assisted Bioimaging

Ultrasound (US) is still the most common technique to screen the HCC since it is safe, economical and accessible (Verslype et al., 2012). However one study reported that the sensitivity of US for the small lesion (<2 cm), which is important for early detection and diagnosis of HCC, is only 27.3% (Kim et al., 2017).

To improve diagnostic accuracy and sensitivity for early stage tumor, photoacoustic (PA) imaging integrating optical and ultrasound advantages have been developed which has the highest resolution in deep tissue compared with any conventional imaging tools (Liu Y. et al., 2019). So in the integrated system, PA imaging easily combined with ultrasound imaging can strongly enhance the clinical diagnosis (Kim et al., 2016). To this end, Lee et al. synthesized a MSNs-based liver targeting PA contrast agent, hyaluronate–silica nanoparticle (HA–SiNP) conjugate. Because of strong photoacoustic signal of SiNP in NIR windows, the PA amplitude in liver after HA–SiNP conjugates injection was remarkably enhanced 95.9% compared to normal liver beyond other PA contrast agents, which provides more detail anatomical and functional information for HCC diagnosis (Lee et al., 2018).

Before treatment, dynamic contrast-enhanced magnetic resonance imaging (MRI) is considered as the best approach to define the tumor staging and assist in choosing a suitable treatment strategy. Non-specific contrast agents, Gd complexes, have been the most widely applied agents for liver MRI (Bellin et al., 2003). Nevertheless, for HCC diagnosis, liver-specific MRI contrast agents, which mainly target the liver tissue, maybe an alternative choice. Aim to make up the limitations of non-specific contrast agent, Kim et al. investigated the liver-specific MRI contrast agent, Mn2+-doped SiO<sup>2</sup> nanoparticles (Mn-SiO2), enhancing the visibility of HCC lesion. The nanoparticles engulfed in Kupffer cells would release the Mn2<sup>+</sup> ions, thus T1-weighted MRI shows hyperintense in healthy liver tissues with abundant Kupffer cells over lesions, which are always lack of Kupffer cells (Kim et al., 2013).

Furthermore, it has been reported that intraoperative fluorescent imaging for imaging-guided surgery by virtue of MSNs could dramatically improve surgical intervention of tumor (Zeng et al., 2016). This RGD-conjugated MSN highly loaded with ICG dye could precisely delineate the margins of HCC intraoperatively by NIR. Depend on the subjective experience, only the conspicuous tumors (5.09 ± 2.31 mm) could be visually discriminated by surgeons intraoperatively. This lesion is also confirmed by fluorescent imaging. The microtumor lesions (0.4 ± 0.21 mm), which could not be recognized with the nake eye, are accurately detected by fluorescent imaging. Currently in operations, surgeons mainly rely on conventional preoperative imaging methods and subjective experience. However, in most cases, tumor microfoci can't be discovered which is regarded as one of the etiology for tumor recurrence. But in the intraoperative fluorescent imaging by MSNs, tumor microfoci in liver could easily be distinguished and resected resulting in better surgical outcomes. It is helpful to reduce the high postoperative recurrence rate of HCC.

#### MSNs-Assisted Liquid Biopsy

Liquid biopsy refers to non-invasive tests analyzing the bodily fluids and is a promising option to detect early stage HCC (Zhou et al., 2016; Ye et al., 2019). The source from tumor for liquid biopsy covers circulating tumor cells (CTCs), nucleic acids, proteins and circulating exosomes (Lee et al., 2018). Currently, several new MSNs have been reported for detection of tumor cells and their associated molecules. Hu et al. developed functionalized MSNs for specifically detecting HCC cells with assist of a biotinlabeled aptamer. The binding rate with hepG2 cells could reach approximately 90%, while the binding rate with L02 cells was close to 2%, which means the nanoparticles established a sensitive detection system for HCC cells (Hu et al., 2017). In another research, MSNs are utilized to detect the apoptotic tumor cells for evaluation of treatment response. This detection system contains two steps: (1) the HCC cells among various cells would be immobilized on the nanotubes; (2) the apoptotic HCC cells would be quantitated through the specific interaction between antiphosphatidyl serine antibody and phosphatidylserine. This cytosensor has a high sensitivity, which even could respond as low as 800 cells mL−<sup>1</sup> (Wu et al., 2012). Besides, MSNs are also used to enrich phosphopeptides from serum of HCC patients. And then the phosphopeptides could be extract from MSNs for further analysis (Hu et al., 2009).

## MSNS IN HCC TREATMENT

### Drug Therapy

Systemic chemotherapy usually is the only option for patients in advanced cancer, however, no satisfactory results have been obtained in HCC (Bruix et al., 2015; Gao J. et al., 2015). Thus, numerous studies focusing on MSNs to improve the drug effect have been reported. Awing to the unique structure of MSNs, they are suitable for delivery of both hydrophobic/hydrophilic anticancer drugs. Moreover, drugs release in MSNs always experience a decrease in release rate, resulting in sustained release pharmacokinetics (Kumar et al., 2015; He et al., 2017). DOX, which is easily tracked through fluorescence effects, is widely used as a model drug for assessing drug loading and delivery capacity in MSNs (Rudzka et al., 2013; Xie et al., 2014; Yang et al., 2015). Although DOX in MSNs had showed a good antitumor effects, other drugs that is more sensitive for HCC should be explored. It's worth noting that many hydrophobic drugs loaded in MSNs overcome their poor water solubility, including paclitaxel (Li et al., 2010; He et al., 2017; Xu et al., 2017), curcumin (Lv et al., 2017; Xing et al., 2018), berberine (Yue et al., 2018b), and so on. Co-delivery multiple drugs have been recognized as a more efficient treatment than a single drug. Thus a MSNs-based nanoparticle had been developed for co-delivery of SO and ursolic acid (UA). Compared with the SO or UA respectively treated group, the expression of EGFR and VEGFR2 in SO + UA treated group decreased about 60%, which tremendously increase apoptosis of tumor cells and inhibit proliferation, adhesion, migration and angiogenesis. The further in vivo increased therapeutic efficacy of nanoparticles demonstrate the synergistic effect of SO and UA (Zhao et al., 2017). UA, which possesses significant antitumor activity, is limited in clinical application with its poor water solubility. By virtue of MSNs, UA can be delivered to tumor tissues, exhibiting a synergetic antitumor effect with SO. It suggested a promising approach for exploiting the potential of the drugs.

#### Protein Therapy

fbioe-08-00184 March 7, 2020 Time: 16:43 # 8

Proteins have been explored as potential therapeutic candidates in cancer therapy since they have a low amount of side effects and the immunity to multidrug resistance mechanism (Epler et al., 2012; Deodhar et al., 2017). However, due to their facile degradation and fragile structure in vivo, effective delivery of proteins is a great challenge in spite of the vehicles (Escoto, 2013; Deodhar et al., 2017). Thus, MSNs have been used as promising vehicles to deliver proteins in HCC therapy (Zhang et al., 2014; Liu et al., 2017). The porous and stable nature of MSNs allows proteins encapsulated inside nanoparticles and provides a stable shelter to protect proteins. Lipid bilayer-modified MSNs had been designed to deliver ricin toxin A-chain (RTA) in which a few of RTA prematurely release and the activity of the remanent proteins was retained. RTA-loaded MSNs induce apoptosis in Hep3B at picomolar concentrations of RTA, which is 3500-fold less than the IC50 values of free RTA. These excellent functions enable protein-based therapies to reach their full potential (Epler et al., 2012). Covalent modification of protein on the MSNs surface was another approach for protein delivery. The proteins in this MSNs release via cleavage of the covalent bond in the pH/redox stimulation. Moreover, the proteins immobilized on the surface can be used as gatekeepers mentioned previously since the hydrodynamic diameters of the proteins are sufficient to cover the pores of MSNs. Relatively, in this way, the protective effects for proteins may be weaker (Pei et al., 2018).

### Gene Therapy

Gene therapy has been regarded as a new opportunity to satisfy the needs in treatment of cancer (Das et al., 2015). With the advancement in RNA biology, gene therapies not only introduce the exogenous genes by DNA but also change the gene expression at the mRNA level through by virtue of short interfering RNAs (siRNAs), miRNAs and antisense oligonucleotides (ASOs) (Shim et al., 2018). Recently, Clustered regularly interspaced short palindromic repeats/CRISPR-associated nuclease 9 (CRISPR-Cas9) also opens a new avenue in gene therapy to correct the mutations of cancer (Karimian et al., 2019). However, the development of an efficient and safe vector for therapeutic genetic materials is still a major issue. MSNs are promising carriers for gene delivery for their versatile payload of various genetic materials without chemical modification. For forming a stable complex with electronegative nuclei acid, MSNs are often modified to possess net positive charges by methods including amination-modification (Xiao et al., 2010; Yu et al., 2015; Zheng et al., 2018) and cationic polymer functionalization (Xue et al., 2017; Wang et al., 2018). In these terms, the modified surface can not only increase the adsorption capacity of negatively charged nuclei acid molecules, but can also facilitate MSNs to escape from endosome/lyposomes by "proton sponge effect" (Tang et al., 2012; Wang et al., 2018).

To overcome multidrug resistance in HCC, Xue et al. prepared lipid-coated MSNs containing DOX and miR-375 which can inhibit P-glycoprotein (P-gp) expression via inhibition of astrocyte elevated gene-1 (AEG-1) expression in HCC. P-gp, which is overexpressed in multidrug resistance cells in HCC, could impede the effectiveness of chemotherapy. So augment the level of miR-375 in HCC cells would serve as a credible way to overcome multidrug resistance. Further evaluation of antitumor effect in the DOX-resistant HepG2 cells xenograft tumor mouse model showed the tumor volume in DOX and miR-375 nanoparticles treated group is only about half of that in DOX nanoparticles treated group in 1 month, suggesting an alternative option to overcome multidrug resistance in HCC by these nanoparticles (Xue et al., 2017).

### Phototherapy/Sonodynamic Therapy (SDT)

Recently, phototherapy has emerged as a promising strategy for HCC. Photothermaltherapy (PTT) and photodynamic therapy (PDT) are two main types of phototherapy. PTT destroies the tumor cells by light-induced photothermal effects, while PDT damages the tumor cells by light-induced cytotoxic singlet oxygen (1O2), one kind of the most representative ROS (Liu H. et al., 2011; Zhao et al., 2012). NIR is widely used in PTT and PDT because of its superiority compared to other lights. In recent reports, MSNs prepared with photothermal agents including MoS2, C, Au, CuS and indocyanine green (ICG) showed a strong photothermal effect (Wu et al., 2015; Lee et al., 2016; Wang Z. et al., 2016; Wang et al., 2017a; Chen et al., 2018). Usually normal cells possess a higher heat tolerance over cancer cells at elevated temperatures around 43◦C, so the heat generated (43◦C) would trigger the death of tumor cells only (El-Boubbou, 2018). Gao B. et al. (2015), reported MSNs with gold core acted as a radiation sensitizer, thereby inducing the more effective radiotherapy by iodine 125 seed. Furthermore, Wang et al. (2019) rethink that radiosensitization strategy is not sufficient because of the hypoxic microenvironment in HCC, so they fabricated Janus-structured gold triangle-mesoporous silica nanoparticles to ameliorates hypoxia through PTT generated by gold triangle. This research indicated synergistic radio-photothermal therapy is a reasonable combination scheme. And there are some photosensitizers, such as Zinc(II)-phthalocyanine, Chlorin e6, Photosan-II loaded in MSNs for PDT in HCC (Liu et al., 2014; Lv et al., 2015; Lin et al., 2018). The success of PDT depends on the potential of photosensitizers to transfer energy from light to tumor dissolved oxygen (O2) to generate <sup>1</sup>O2. Pre-existing hypoxia in HCC and O<sup>2</sup> consumption during PDT can remarkably lower down the PDT efficacy (Thorat et al., 2019a). To address this problem, in situ synthesized Pt nanoparticles as a catalyst to convert H2O<sup>2</sup> into O<sup>2</sup> on MSNs was constructed. And 4.2-fold O<sup>2</sup> and 1.6 fold <sup>1</sup>O<sup>2</sup> generation ability compared to normal MSNs greatly improved the PDT efficacy in HCC and exhibited threefold tumor suppression ability in HCC mice model (Lan et al., 2019).

However, when tumor locates in the deeper tissue, which is common for HCC, the lights arriving to the tumor are so weak that hardly produce an enough cytotoxic effects. Therefore, MSNs-based SDT has been developed account for the advantage of deeper tissue-penetrating of ultrasound (Li et al., 2018b). In spite of the ROS cytotoxic effects on the tumor cells, moreover, SDT can enhance the chemotherapeutic sensitivity of tumor cells by activating the cellular internalization and the

mitochondrial apoptotic pathway, and inhibiting ATP-binding cassette transporter (Xu et al., 2013; Qian et al., 2016). For example, Li et al. (2018b) chose MONs as nanocarriers for the delivery of both sonosensitizer (protoporphyrin, PpIX) and DOX. The tumor-inhibiting rates in mice increased about 21.6% with the US irradiation (Li et al., 2018b). So the combination of SDT and chemotherapy is a hopeful strategy for HCC treatment.

#### DUAL EFFECTS OF THERAPY AND DIAGNOSIS IN MSNS

As described, MSNs-based therapy and diagnosis for HCC have been investigated a lot separately. More importantly, MSNs-based theranostic nanostructures are capable of detecting the tumor not only before or after, but also during the treatment anytime. Recently, Chen H. et al. (2017) offered a proposal extending the concept of nanotheranostics from nanomedicine owning both diagnostic and therapeutic functions, to the approaches that use diagnosis to aid nanoparticle therapy procedures. With such approaches, medical treatments can be tuned promptly on the basis of the detection results, which means more specific therapies for individual patient (**Figure 5**; Xie et al., 2010).

The easiest way to achieve theranostics is co-delivery of therapeutic and imaging agents. Ashley et al. modified MSN with supported lipid bilayers resulting in nanostructures ("protocells"), which could be loaded with mixtures of therapeutic (drugs, proteins, genes) and diagnostic (quantum dots) agents (Ashley et al., 2011). Quantum dots as a fluorescent dyes could trace the biodistribution of content in nanostructures because mixtures of therapeutic agents and quantum dots will be released from nanostructures simultaneously. Within a certain period of time, quantum dots remaining in nanostructures could show the location of nanoparticles targeting to tumor for diagnosis. Therefore, in this way, the fluorescent dyes mainly tracked the therapeutic agents in tumor rather than diagnosis for HCC. The meaning of this monitoring is to investigate tumor accumulation and release of agents, which is an important factor for nanoparticle therapy procedures. On the contrary, if diagnostic agents stay in nanoparticles throughout, the diagnosis and therapy for tumor could be implemented at the same time (Fan et al., 2019).

In another way, because many nanomaterials are already imaging agents, hybrid nanostructures fabricated with these nanomaterials and MSNs could make diagnosis and therapy together. Among all hybrid nanostructures, the physically responsive (light/magnetic/ultrasonic) nanostructures widely explored as innovative "theranostics" in cancer have been described in excellent reviews (Thorat et al., 2019a,b). The greatest strength for these nanostructures is that their properties to physically stimulus accord with the diagnosis mode (MRI/US) in clinic. Li et al. successfully coat a mesoporous-silica layer onto the surface of Ti3C<sup>2</sup> (Ti3C2@mMSNs). Ti3C<sup>2</sup> has a high photothermal-conversion efficiency and enables Ti3C2@mMSNs to possess the potential contrast-enhanced PAimaging and heat production property. So these nanostructures can monitor the photothermal hyperthermia treatment process in real-time (Li et al., 2018c). In another study, Liu et al. prepared the capping MSNs-coated iron oxide nanoparticles with programmable DNA hairpin gates to form M-MSNs, which could decrease T2-weighted tumor signal in MRI for HCC diagnosis. Interestingly, BHQ1 (fluorescence quencher) and 6 carboxyfluorescein (FAM) were linked to the tail extension of the DNA hairpin structure. When DNA hairpin gates change the conformation after addition of HCC- specific miRNA-21, the fluorescence of FAM will significantly "ON" to monitor the release of DOX in nanostructures (Liu J. et al., 2019). In spite of the therapeutic functions, the above nanostructures have two diagnostic functions: positioning tumor tissues and monitoring drug release.

The most amazing theranostic nanostructure is developed using mesoporous silica layer as shells and up-conversion luminescent (UCL) GdOF:Ln (Ln = 10%Yb/1%Er/4%Mn) as cores by Lin et al. Under NIR irradiation, GdOF:Ln could efficiently transfer NIR energy to the conjugated PDT agent (ZnPc) and emit bright red up-conversion emission. The shell decorated with carbon dots also can generate photothermal effect at the same time. Gd/Yb has the strong X-ray attenuation endowing nanostructures for computed tomography (CT) contrast agents, meanwhile Gd-based particles can be harnessed as contrast agents for MRI. This nanostructure with excellent and rational design is appropriate for both various imaging (UCL, CT, MRI, PT) and various therapies (PDT, PTT, chemotherapy), thus achieving multimodal imaging guided combination therapies (Lv et al., 2015).

With the development of nanotheranostics, many researchers believe that the integration of diagnosis and therapy to single modality would really benefit patient over independently managed diagnosis and therapy (Kelkar and Reineke, 2011). However, recent development of theranostics in MSNs was limited to "how can theranostics modalities in nanoparticles be realized" rather than "why can theranostics modalities in nanoparticles benefit the patients." In hospital, patients have no opportunity to occupy the diagnostic tools all the time, so the concept of real-time monitoring translating to clinic may be only few hours monitoring time to examine patients totally. And multimodal imaging provides good alternatives for patients, but we should weigh advantages and disadvantages. It is because that the more imaging modes nanoparticles can be applied in, the more complicated structure nanoparticles possess. In most cases, four or more imaging modes in nanostructures make no benefits to the patient compared to one imaging mode. We suggest that development of theranostic nanostructures should be based on clinical demand and clinical practice, which is the key for translational medicine.

#### CONCLUSION AND OUTLOOK

In this review, we summarized the recent progress in HCC theranostic applications based on MSNs. We generalized the precision delivery strategies applied for HCC in recent researches, including passive targeting, active targeting and stimuli-responsive release. The multiple therapy and diagnosis

approaches had been realized in MSNs. Among a large amount of applications, we highlighted theranostic hybrid nanostructures, in which combination of the modalities of diagnostic imaging and therapy endowed MSNs-based nanostructures the ability to image and monitor the tumor tissue during treatment making possible more timely adjustment of therapy.

Despite the extensive researches, some emerging therapy and diagnosis approaches have not been used in MSNs for HCC. Most recently, cancer immune therapy is growing obviously, including cancer vaccinations, chimeric antigen receptor (CAR) T-cell therapy and immune checkpoint blockade therapy (Zhang and Chen, 2018). Actually, many MSNs-based immune therapies have been reported in other cancers (Ding et al., 2018; Li et al., 2018a; Xie et al., 2019). In addition, a lot of clinical trials about immune checkpoint blockade therapy such as PD-1 therapy for HCC are underway. Therefore, immune therapy is a potential development point for MSNs-based HCC therapy. Positron Emission Tomography (PET) is important for metastasis and prognostic assessment in patients with HCC (Filippi et al., 2019). Some Radiolabeling MSNs also have been reported for PET (Ni et al., 2018), thus it is hopeful to accomplish for more accurate diagnosis of HCC due to its smart recognition to tumor cells.

Although the preclinical trials of MSNs are successfully completed, currently, there are no MSNs that have been approved applied in clinic. There remain several critical challenges that need to be overcame for MSNs. First, the current small animal

models are not suitable to evaluate the delivery efficiency and long-term toxicity of nanoparticles in humans. Second, the laboratory scale production of MSNs cannot easily repeat in the industrial scale of production for clinical application, especially for the complicated modified MSNs. Third, there is no thorough evaluation criterion, which may confuse the researchers to improve the existing MSNs. Nevertheless, there are some clinical trials in silica nanoparticles, the prospect of clinic translation for MSNs was less than good. We should focus our efforts in the following aspects: (1) Try to use big animals to assess safety and efficacy of the MSNs; (2) Improve and simplify production process of MSNs; (3) Establish a homogenized evaluation system; (4) modulate the theranostic functions of MSNs closer to the clinic. The long-wind road for clinical translation and commercialization of MSNs still needs researchers going on.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

XX brought forward the subject and guided the writing. YT gathered the research data and wrote the manuscript. JW supervised the writing and revising. All authors critically revised the manuscript and approved it for publication.

#### FUNDING

This work was supported by the National Major Science and Technology Projects of China (No. 2017ZX10203205), National Natural Science Foundation of China (Nos. 81570589 and 81801824), and China National Funds for Distinguished Young Scientists (No. 81625003).


nanoparticle-supported lipid bilayers. Adv. Healthc. Mater. 1, 348–353. doi: 10.1002/adhm.201200022


fbioe-08-00184 March 7, 2020 Time: 16:43 # 12



nanoparticle encapsulation, and cancer cell-specific drug targeting. ACS Appl. Mater. Interfaces 10, 3229–3240. doi: 10.1021/acsami.7b14496


fbioe-08-00184 March 7, 2020 Time: 16:43 # 14


**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 © 2020 Tao, Wang and Xu. 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.

# PLGA-Based Drug Delivery Systems for Remotely Triggered Cancer Therapeutic and Diagnostic Applications

Xue Shen<sup>1</sup> , Tingting Li2,3, Xiaoxue Xie<sup>2</sup> , Yi Feng<sup>2</sup> , Zhongyuan Chen<sup>2</sup> , Hong Yang2,3 , Chunhui Wu2,3, Shengqi Deng<sup>1</sup> \* and Yiyao Liu2,4 \*

<sup>1</sup> Sichuan Industrial Institute of Antibiotics, Chengdu University, Chengdu, China, <sup>2</sup> School of Life Sciences and Technology, University of Electronic Science and Technology of China, Chengdu, China, <sup>3</sup> Center for Information in Biology, University of Electronic Science and Technology of China, Chengdu, China, <sup>4</sup> Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, China

Edited by:

Andrea Ragusa, University of Salento, Italy

#### Reviewed by:

Alessandra Quarta, Institute of Nanotechnology (CRN), Italy Shiyong Song, Henan University, China

> \*Correspondence: Shengqi Deng dq1155@sina.com Yiyao Liu liuyiyao@uestc.edu.cn

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology Received: 05 February 2020

Accepted: 06 April 2020 Published: 05 May 2020

#### Citation:

Shen X, Li T, Xie X, Feng Y, Chen Z, Yang H, Wu C, Deng S and Liu Y (2020) PLGA-Based Drug Delivery Systems for Remotely Triggered Cancer Therapeutic and Diagnostic Applications. Front. Bioeng. Biotechnol. 8:381. doi: 10.3389/fbioe.2020.00381 Intelligent drug delivery systems based on nanotechnology have been widely developed and investigated in the field of nanomedicine since they were able to maximize the therapeutic efficacy and minimize the undesirable adverse effects. Among a variety of organic or inorganic nanomaterials available to fabricate drug delivery systems (DDSs) for cancer therapy and diagnosis, poly(D,L-lactic-co-glycolic acid) (PLGA) has been extensively employed due to its biocompatibility and biodegradability. In this paper, we review the recent status of research on the application of PLGAbased drug delivery systems (DDSs) in remotely triggered cancer therapy and the strategies for tumor imaging provided by PLGA-based DDSs. We firstly discuss the employment of PLGA-based DDSs for remotely triggered cancer therapy, including photo-triggered, ultrasound-triggered, magnetic field-triggered, and radiofrequencytriggered cancer therapy. Photo-triggered cancer therapy involves photodynamic therapy (PDT), photothermal therapy (PTT), and photo-triggered chemotherapeutics release. Ultrasound-triggered cancer therapy involves high intensity focused ultrasound (HIFU) treatment, ultrasound-triggered chemotherapeutics release, and ultrasoundenhanced efficiency of gene transfection. The strategies which endows PLGA-based DDSs with imaging properties and the PLGA-based cancer theranostics are further discussed. Additionally, we also discuss the targeting strategies which provide PLGAbased DDSs with passive, active or magnetic tumor-targeting abilities. Numerous studies cited in our review demonstrate the great potential of PLGA-based DDSs as effective theranostic agent for cancer therapy and diagnosis.

Keywords: PLGA-based DDSs, targeting, remotely triggered cancer therapy, imaging, theranostics

### INTRODUCTION

The utilization of nanotechnology in drug delivery has been extensively supposed to alter the pattern of the pharmaceutical industries for the predictable future (Farokhzad and Langer, 2009; Lim et al., 2013; Farooq et al., 2019). Intravenous chemotherapy is one of the common cancer treatments. However, the normal tissues and organs will be damaged owing to the undesirable side effects caused by non-specific distribution of chemotherapeutic drugs and lack of specificity

for tumor cell recognition (Jia et al., 2012; Shen et al., 2017). Besides, the repeated treatment with one single drug can cause multidrug resistance in tumor cells, compromising the anticancer effects of chemotherapeutic drugs (Yang et al., 2018). Currently, the rise of nanotechnology have provided an versatile drug delivery systems (DDSs) for the efficient treatment of cancer, which offers a solution to the problem of body tissue damage caused by non-specific distribution of traditional chemotherapeutic drugs (Shi et al., 2010). The developed DDSs with a diameter between 10 and 200 nm exhibits favorable pharmacokinetic property, prolonged systemic circulation time, sustained drug release profile, and enhanced intratumoral accumulation when compared to the free drugs (Hu and Zhang, 2012). With the continuous improvement of the performance of intelligent DDSs, it is possible to achieve (1) improved stability of hydrophobic drugs and the possibility of hydrophobic drugs for systemic administration; (2) targeted drug delivery; (3) combined delivery of two or more drugs or other therapeutic agents for combination therapy; (4) combined delivery of the therapeutic agents and imaging agents for the visualization of drug delivery; (5) favorable biodistribution and pharmacokinetic property of drugs and finally realizing the enhanced therapeutic effects and reduced side effects (Farokhzad and Langer, 2009).

Since liposomes were proposed as protein and drug delivery vehicles for the treatment of diseases in the 1960s, nanotechnology has had a huge impact on the development of novel DDSs for controlled drug delivery and combination therapy (Shi et al., 2010). A variety of inorganic or organic materials have been utilized to prepare novel DDSs including inorganic nanoparticles (Li et al., 2016), liposomes (Zhao et al., 2015), polymer micelles (Shi et al., 2014), and polymer nanoparticles (Yang et al., 2016; Shen et al., 2019) for the effective cancer treatment. Among all these nanocarriers, one kind of polymer nanoparticles based on poly(D,L-lactic-co-glycolic acid) (PLGA) have attracted considerable attention due to their unique physical and chemical properties, such as tunable particle size, regular morphology, large surface area, favorable pharmacokinetic property, excellent biocompatibility, and biodegradability (Deng et al., 2014; Yang et al., 2015, 2018; Shen et al., 2017, 2019). Polymer nanoparticles are colloidal particles ranging in size from 10 to 1,000 nm, which can be fabricated by using synthetic polymers, such as poly(butylcyanoacrylates) (PBCA), poly(lactic acid) (PLA), poly(D,L-lactic-co-glycolic acid) (PLGA), or natural polymers, such as chitosan, gelatin, and albumin (Bamrungsap et al., 2012; Mc Carthy et al., 2015). A majority of polymer nanoparticles are biodegradable and biocompatible and their reactive functional groups allows easy conjugation with other ligands or polymers. PBCA nanoparticles are commonly used to deliver some conventional drugs such as chemotherapeutic drugs or nucleic acids such as plasmids (Alyautdin et al., 1995; Schneider et al., 2008). However, polycyanoacrylic acid and alcohol as the hydrolysis products of PBCA are cytotoxic, which limits the application of PBCA as a nanocarrier in the field of biomedicine (Mc Carthy et al., 2015). Contrarily, PLGA and PLA have been preferentially used for the preparation of nanocarriers since they can be degraded into lactic acid and glycolic acid, which are natural and non-toxic and able to be eventually degraded into water and carbon dioxide (Graf et al., 2012). Moreover, PLGA has been approved for the medical applications by the United States Food and Drug Administration (FDA) (Hamori et al., 2015). The various targeting moieties can be introduced to the surface of PLGA-based DDSs, providing them with tumor-targeting capability. And the appropriate outer surface engineering (such as PEGylation) of PLGA-based DDSs may prolong the blood circulation time of the DDSs (Graf et al., 2012). Chemotherapeutic drugs, photosensitizers, photothermal agents, therapeutic gene/siRNA, small-molecule inhibitors, and other therapeutic agents can be easily loaded in the PLGA-based DDSs for tumor treatment (Lee et al., 2012; Zhao et al., 2015; Shen et al., 2017, 2019). Furthermore, the imaging agents can also be integrated into PLGA-based DDSs to acquire the imaging property for tumor diagnosis (Chen et al., 2016). Co-loading the therapeutic agents and imaging agents into PLGA-based DDSs can make them the potential candidates for cancer therapeutics and diagnostics (known as cancer "theranostics") (Jia et al., 2012).

Several previous literatures have summarized the application of DDSs based on PLGA in cancer therapy and imaging from the aspects of preparation methods of PLGA-based nano- and microparticles, uptake of PLGA particles into cancer cells, controlled drug release property, and combination treatments (Rezvantalab et al., 2018; Swider et al., 2018; Kim et al., 2019). In this review, we introduce the recent status of research on the application of PLGA-based DDSs in remotely triggered cancer therapy and the strategies for tumor imaging provided by PLGA-based DDSs. We firstly emphasize the external stimuli-triggered cancer therapy approach and then discuss the strategies for tumor imaging provided by PLGAbased DDSs, the cancer theranostics in biomedical applications as well as the targeting strategies which endow the PLGAbased DDSs with passive, active or magnetic tumor-targeting abilities are further discussed. We introduced various types of PLGA-based DDSs in this review, some of them are nanoscaled, while the others are micro-scaled. Hao et al. (2015) have synthesized the tumor-targeted and gold nanoshell-surrounded PLGA-based nanoparticles (denoted as ANG/GS/PLGA/DTX NPs) for cancer chemotherapy and PTT, which were spherical with a size of about 200 nm. The nano-sized particles can more easily accumulate in tumor regions owing to the leaky vasculature and poor lymphatic drainage and their smaller size than the cell gaps of vascular endothelial cells of tumors (Torchilin, 2011; Bamrungsap et al., 2012; Shen et al., 2019). Nanoparticles are appropriate for various administration routes including intravenous administration. ANG/GS/PLGA/DTX NPs with exposure to an 808 nm laser irradiation displayed considerable tumor inhibition efficiency after the intravenous administration. Their pharmacokinetic parameters were further analyzed. The AUC of ANG/GS/PLGA/DTX NPs was higher than that of DTX solution with a 1.42-fold increase, and the clearance of ANG/GS/PLGA/DTX NPs was much lower than that of DTX solution. These results indicated that the nanosized ANG/GS/PLGA/DTX NPs exhibited the prolonged blood circulation time and could improve the bioavailability of free DTX. Fang et al. (2015) have designed and synthesized the

magnetic responsive PLGA microspheres contained DOX (DOX-MMS) for combined chemotherapy and hyperthermia, of which the average size was measured to be 2.4 µm. Benefiting from their hollow structure and large size, the quantities of loaded DOX in DOX-MMS were 6.2%, and the encapsulation efficiency of DOX was up to 85%. The micro-sized DOX-MMS were intratumorally administrated, and showed an effective tumor inhibition effect in 4T1 tumor-bearing nude mice with exposure to alternating current magnetic field (ACMF). Niu et al. (2013) have designed and constructed the gas-filled multifunctional polymer microbubbles (MPMBs) to co-encapsulate iron oxide nanoparticles and DOX for tumor lymph node detection and therapy. Although these gas-filled microbubbles are not suitable for intravenous injection, the gas-filled microbubbles can be used as the drug/gene delivery systems and they have been proven to be capable of enhancing the ultrasound signals when used as the ultrasound agent (Hernot and Klibanov, 2008; Eisenbrey et al., 2010).

### PLGA-BASED DDSs FOR REMOTELY TRIGGERED CANCER THERAPY

Cancer is one of the most fatal diseases and has long been a threat to human health. Their heterogeneity and complexity make tumors grow aggressively, leading to a significant increase in patient mortality (Das et al., 2009; Mohanty et al., 2011; Parhi et al., 2012). The nanocarriers based on PLGA has been demonstrated to be the most promising DDSs due to their unique physical and chemical properties, such as tunable particle size, favorable stability, excellent biocompatibility, and biodegradability (Deng et al., 2014; Yang et al., 2015, 2018; Shen et al., 2017, 2019). PLGA-based DDSs can load hydrophobic and hydrophilic chemotherapeutic drugs, photosensitizers, photothermal agents, therapeutic gene/siRNA, and other therapeutic agents and display the prolonged blood circulation time and tumor-targeting capability after the appropriate surface modification, resulting in the improved antitumor efficacy (Lee et al., 2012; Schleich et al., 2013; Topete et al., 2014; Zhao et al., 2015; Luo et al., 2018). Remotely triggered cancer therapy can allow for selective and precise eradication of the tumors and the controlled release of chemotherapeutics by putting the external stimuli (photo, ultrasound, magnetic field, and radiofrequency) on targeted regions (Rai et al., 2010; Shenoi et al., 2013). Moreover, remotely triggered cancer therapy can determine when the treatment starts and how long it lasts, allowing for precise treatment and reduced systemic toxicity. The external stimuli-triggered cancer therapy approach [photodynamic therapy (PDT), photothermal therapy (PTT), photo-triggered chemotherapeutics release, high intensity focused ultrasound (HIFU) treatment, ultrasound-triggered chemotherapeutics release, ultrasound-enhanced efficiency of gene transfection, magnetic field-triggered cancer therapy, and radiofrequency-triggered cancer therapy] provided by PLGAbased DDSs will be briefly summarized as follows. Representative applications of PLGA-based DDSs for remotely triggered cancer therapy are listed in **Table 1**.

### Photo-Triggered Cancer Therapy

PLGA-based DDSs have been developed for the photo-triggered therapeutic application including the PDT, PTT, and phototriggered chemotherapeutics release by employing an external light as a trigger. It has been demonstrated that photo-triggered cancer therapy could selectively and precisely eradicate the solid tumors by putting the laser probe on targeted regions, realizing the satisfactory antitumor efficacy (Su et al., 2015; Yan et al., 2016b).

Photodynamic therapy (PDT) can be used as a crucial approach for the treatment of various types of cancer by irradiating the photosensitizer-enriched tumor sites with light, the generating reactive oxygen species (ROS) can directly kill the cancer cells (Jang et al., 2011; Master et al., 2013; Chen et al., 2014). The photosensitizer-loaded PLGA-based DDSs have been widely concerned because of their good targeting capability and biocompatibility (Ricci-Junior and Marchetti, 2006). Importantly, the hydrophobic photosensitizers can be integrated into the PLGA-based DDSs, making it possible for the hydrophobic anticancer drugs to achieve the tumor-targeted delivery (Lee et al., 2012). Rojnik et al. (2012) designed the PEGylated PLGA nanoparticles for delivery of temoporfin, a effective second generation photosensitizer. In this study, the temoporfin delivered by PEGylated PLGA nanoparticles exhibited less dark cytotoxicity than the free temoporfin, while the phototoxicity of temoporfin-loaded PEGylated PLGA nanoparticles was not reduced when compared to the free temoporfin. The zinc(II) phthalocyanine (ZnPc)-loaded PLGA nanoparticles synthesized by Ricci-Junior and Marchetti (2006) showed the spherical morphology with a narrow size distribution and exhibited excellent biocompatibility due to their low dark toxicity assessed by the MTT assay. The significant photocytotoxicity of ZnPc-loaded PLGA nanoparticles showed them the great potential as the photosensitizer-loaded nanocarriers for PDT. Lee et al. (2012) have fabricated the PEGylated PLGA nanoparticles to co-load chlorin e6 (Ce6, a photosensitizer that can generate ROS upon the laser irradiation) and iron oxide (T<sup>2</sup> contrast agent) for tumor diagnosis and PDT. Firstly, Ce6 and mPEG were respectively coupled to the terminal hydroxyl group of PLGA, and then the obtained PLGA-mPEG and PLGA-Ce6 were used to synthesize the multifunctional PLGA nanoparticles through the double emulsion method, and the feeding ratio of PLGA-Ce6 and PLGA-mPEG was 75%: 25% (w/w) (**Figure 1A**). The whole body fluorescence imaging of KB tumor-bearing nude mice after intravenous administration of Ce6/iron oxide co-loaded PLGA nanoparticles (NP1) showed a remarkable fluorescence signal of Ce6 in the tumor region, while the mice after intravenous administration of free Ce6 exhibited a weak fluorescence signal of Ce6 in the tumor region, demonstrating the efficient tumor targeting capability of Ce6/iron oxide co-loaded PLGA nanoparticles (NP1) (**Figure 1B**). The subcutaneous tumor volume of NP1 (equivalent 0.1 mg Ce6/kg body) treated KB tumor-bearing nude mice was about 1.5 or 3 times smaller than those of free Ce6 (2.5 mg/kg) or PBS treated nude mice (**Figure 1C**). The low-dose administration of NP1 (equivalent 0.1 mg Ce6/kg body) displayed better tumor inhibition effect than high-dose administration



of free Ce6 (2.5 mg/kg), suggesting that the Ce6-loaded PLGA nanoparticles showed enhanced PDT for tumor.

Photothermal therapy (PTT) has been proposed to be an attractive method for solid tumor elimination, which utilizes the light-absorbing agents to convert light energy into heat energy, thus the generated local hyperthermia can destroy the cancer cells irreversibly without causing damage to the healthy tissues (Dong et al., 2016; Yan et al., 2016a; Wang et al., 2017). Compared to the radiotherapy, chemotherapy and surgery, PTT has been supposed to be a less invasive, controllable and efficient cancer treatment approach (Shen et al., 2015). A large number of nanomaterials have been reported to act as the lightabsorbing agents for PTT, such as gold nanomaterials (Wu et al., 2017), carbon nanotubes (Robinson et al., 2010), and graphene (Markovic et al., 2011), which have strong absorption in the near-infrared region. Indocyanine green (ICG), as an organic molecule, is another kind of near-infrared light-absorbing agent (Li et al., 2017). Notably, the near-infrared light with a wavelength range of 650–950 nm has low phototoxicity to skin and tissues due to the minimal light absorption of skin and tissues in near-infrared region (Yu et al., 2016). Topete et al. (2014) have designed and synthesized a multifunctional nanoplatform for tumor diagnosis and therapy. As shown in **Figure 1D**, the primarily synthesized DOXO-loaded PLGA nanoparticles were subsequently modified with the chitosan biopolymer, then the Au seeds were deposited onto the surface of chitosanmodified DOXO-loaded PLGA nanoparticles, next the DOXOloaded branched gold nanoshells (BGNSHs) were obtained in the presence of HAuCl4/K2CO<sup>3</sup> and ascorbic acid through a seeded-growth surfactant-less method, and finally the human serum albumin (HSA)-ICG-FA conjugated and DOXO-loaded branched gold nanoshells (DOXO-loaded BGNSH-HSA-ICG-FA) were obtained by adsorbing the prefabricated HSA-ICG-FA complex to the DOXO-loaded BGNSHs. The photothermal efficiency of the nanoplatforms were further evaluated. As shown in **Figure 1E**, the temperature of BGNSH-HSA-ICG-FA was rapidly increased and the 1T of BGNSH-HSA-ICG-FA was ∼19◦C after 5 min of irradiation (808 nm, 2 W/cm<sup>2</sup> ), as compared to those of BGNSHs (1T = 15◦C), free ICG(1T = 6◦C), and buffer solutions (1T = 1◦C). The enhanced photothermal efficiency of BGNSH-HSA-ICG-FA was mainly due to the strong absorption of gold nanoshells and ICG molecules in the NIR region, implying the great potential of BGNSH-HSA-ICG-FA for PTT of cancer. The cell viability of HeLa cells treated with BGNSH-HSA-ICG-FA in the presence of NIR laser irradiation was much lower than that of HeLa cells treated with BGNSH-HSA-ICG-FA in the absence of NIR laser irradiation, indicating the remarkable photocytotoxicity of BGNSH-HSA-ICG-FA as a consequence of the hyperthermia generated from gold nanoshells and ICG molecules. And the cell viability of HeLa cells treated with DOXO-loaded BGNSH-HSA-ICG-FA in the presence of NIR laser irradiation was the lowest among all groups, suggesting the significant phototoxicity of DOXO-BGNSH-HSA-ICG-FA and their latent capability for combined chemotherapy and PTT of cancer (**Figure 1F**).

The light as an external stimuli has also been used for ondemand drug release from the PLGA-based DDSs at the suitable position (e.g., tumor region). The DOX and ICG co-loaded PLGA-based nanoparticles (DINPs) fabricated by Zheng et al. (2013) exhibited the faster DOX release property and enhanced cellular uptake of DOX and ICG in MCF-7 and MCF-7/ADR cells under NIR laser irradiation. Hao et al. (2015) synthesized

FIGURE 1 | The photo-triggered cancer therapy based on PLGA-based DDSs. (A) Schematic illustration of Ce6/iron oxide co-delivered PEGylated PLGA nanoparticles (NP1). (B) The whole body fluorescence imaging of KB tumor-bearing nude mice after intravenous administration of Ce6/iron oxide co-loaded PEGylated PLGA nanoparticles (NP1) or free Ce6. (C) Tumor volume changes in KB tumor-bearing nude mice after intravenous administration of NP1 (equivalent Ce6 0.1 mg/kg body), free Ce6 (2.5 mg/kg) or PBS (control). Adapted with permission from Lee et al. (2012). Copyright 2012, Elsevier. (D) Schematic illustration of the synthetic process of DOXO-loaded BGNSH-HSA-ICG-FA. (E) The temperature profiles of BGNSH-HSA-ICG-FA, BGNSHs, free ICG and PBS buffer under continuous NIR laser irradiation (808 nm, 2 W/cm<sup>2</sup> ). (F) Cell viability of the various nanoplatforms treated HeLa cells after 24 h of incubation in the presence and absence of NIR laser irradiation (808 nm, 2 W/cm<sup>2</sup> ). Adapted with permission from Topete et al. (2014). Copyright 2014, American Chemical Society. (G) Transmission electron microscopy (TEM) image of ANG/GS/PLGA/DTX NPs after 808 nm laser irradiation for 5 min. (H) The in vitro drug release profiles of DTX and ANG/GS/PLGA/DTX NPs with or without laser irradiation treatment. (I) Relative tumor volume of various treatment groups. Adapted with permission from Hao et al. (2015). Copyright 2015, Elsevier.

the docetaxel (DTX)-loaded PLGA@Au nanoparticles, and then the angiopep-2, one kind of brain tumor-targeted peptide, was conjugated onto the gold nanoshell of DTX-loaded PLGA@Au nanoparticles via Au-S bond to form the tumor-targeted and gold nanoshell-surrounded PLGA-based nanoparticles (denoted as ANG/GS/PLGA/DTX NPs) for cancer chemotherapy and PTT. ANG/GS/PLGA/DTX NPs showed the excellent photothermal response and their structure observed by TEM was collapsed, the

core-shell structure of the nanoparticles was also destroyed due to the local hyperthermia (**Figure 1G**). The in vitro drug release profile demonstrated the photo-triggered chemotherapeutics release property of ANG/GS/PLGA/DTX NPs. As shown in **Figure 1H**, the drug release of ANG/GS/PLGA/DTX NPs treated with 808 nm laser irradiation was fast and their final cumulative chemotherapeutics release percentage has increased by ∼17.0% compared to that of ANG/GS/PLGA/DTX NPs without 808 nm laser irradiation. The in vivo anti-glioma efficiency of drug-loaded NPs was further assessed, as shown in **Figure 1I**, ANG/GS/PLGA/DTX NPs with exposure to an 808 nm laser irradiation displayed considerable tumor inhibition efficiency, when compared to the GS/PLGA/DTX NPs and ANG/GS/PLGA/DTX NPs without laser irradiation, implying the promising application of ANG/GS/PLGA/DTX NPs in gliomatargeted chemotherapy and PTT.

#### Ultrasound-Triggered Cancer Therapy

Ultrasound has been widely used in medicine for multifarious diagnostic and therapeutic purposes due to their advantages of feasibility and non-invasiveness (Ferrara, 2008; Zhang X. et al., 2014). This section will focus on ultrasound-triggered cancer therapy based on PLGA nano-/micro-particles which involved HIFU treatment, ultrasound-triggered chemotherapeutics release, and ultrasound-enhanced efficiency of gene transfection.

High intensity focused ultrasound (HIFU) is an advanced technology that was proposed for the first time in the 1940s (Manthe et al., 2010; Zhang K. et al., 2014). HIFU ablation has been demonstrated to be a feasible, non-invasive, and effective procedure for solid tumor treatment (Wu et al., 2004; Kennedy, 2005; Orsi et al., 2010). Sun et al. (2012) in their 2012 Biomaterials article, used PLGA to load the hydrophobic Fe3O<sup>4</sup> nanoparticles for improving the therapeutic efficiency of HIFU ablation of breast cancer and realizing the ultrasound/magnetic resonance dual-modality imaging of tumors (**Figure 2A**). The obtained Fe3O4/PLGA microcapsules were characterized by uniform spherical morphology and an average diameter of 885.6 nm. The therapeutic effect of HIFU ablation for the breast cancer-bearing rabbits was further assessed. The tumor was treated with HIFU (150 W of acoustic power for 5 s) after percutaneous injection of Fe3O4/PLGA (saline and pure PLGA microcapsules was used as the control). After the treatment, the coagulative necrosis volume of the excised tumor tissues from the rabbits was calculated and the results showed a larger coagulative necrosis volume and lower positive index of proliferating cell nuclear antigen (PCNA) in tumor tissues of Fe3O4/PLGA microcapsules treated group when compared to the other groups treated with saline and the pure PLGA microcapsules (**Figures 2B,C**). These results were mainly attributed to the most obvious acoustic signal enhancement in the breast tumor region induced by the administration of Fe3O4/PLGA microcapsules and exposure to HIFU.

Ultrasound (US) can be used as another promising external trigger for on-demand drug release from the PLGA-based DDSs at the tumor region. Multifunctional polymer microbubbles (MPMBs) were designed by Niu et al. (2013) to co-encapsulate iron oxide nanoparticles and DOX for tumor lymph node detection and therapy. **Figure 2D** shows the process of MPMBs delivery into tumor lymph node and the controllable release of DOX from MPMBs triggered by low frequency US sonication. The workers recorded the in vitro DOX release profiles of MPMBs with or without low intensity sonication. As illustrated in **Figure 2E**, the cumulative drug release of MPMBs with low frequency US sonication showed an ∼90% of the released DOX from MPMBs, compared to that of MPMBs without low frequency US sonication showed <75% of the released DOX after 48 h. This result proved that the DOX-loaded MPMBs can be remotely triggered by US and then promote the DOX release from the microbubbles, showing the great potential for controllable drug release and accumulation of drugs into the targeted sites under US imaging guidance. The therapeutic efficacy of the MPMBs against tumor lymph nodes was further assessed. And the cell proliferation of the various treatment groups was evaluated in tumor lymph nodes model by using immunohistochemical staining methods. The results showed that the expression of PCNA could be observed in all groups and the proliferative index (PI) of tumor lymph nodes in the group treated with MPMBs and low frequency US sonication is substantially lower than those of any other groups (**Figure 2F**). However, no obvious difference for the PI was observed between the saline treated group and the pure MB (PLGA microbubble without loading iron oxide nanoparticles and DOX) combined with low frequency US sonication treated group. These results revealed that the developed MBMPs could remarkably improve the therapeutic efficacy against tumor lymph nodes when exposed to low frequency US sonication, which might be attributed to the controllable drug release behavior upon US irradiation and the cavitation effect.

Ultrasound has been utilized for gene delivery in recent years (Huber and Pfisterer, 2000; Pitt et al., 2004; Zarnitsyn and Prausnitz, 2004; Larina et al., 2005b). Larina et al. (2005a) has confirmed that combining the gene delivery systems based on nanoparticles with ultrasound may enhance the gene and drug delivery in targeted areas. Chumakova et al. (2008) fabricated the PLGA/PEI/DNA nanoparticles and combined them with ultrasound to improve the gene transfection efficiency in a nude mouse model. They firstly used PEI to absorb the β-galactosidase plasmids, and then the obtained PEI/DNA nanoparticles were loaded on the PLGA nanoparticles to form the PLGA/PEI/DNA nanoparticles. Polyethylenimine (PEI) as one type of cationic polymers has been used for gene delivery for the first time in the 1990s (Boussif et al., 1995). In recent years, PEI molecules with high molecular weight have been considered as one of the most efficient non-viral gene vectors for gene delivery (Gary et al., 2007). The PLGA/PEI/DNA nanoparticles were intravenously injected into the nude mice with DU145 human prostate tumors and then transfection efficiency was assessed by western blot analysis. As shown in **Figures 2G,H**, the expression of β-galactosidase showed an ∼8-fold enhancement in ultrasound irradiated tumors when compared to the control tumors. The enhanced transfection efficiency might be attributed to the changed structure of tumor cell membrane and tumor vasculature induced by ultrasound. These results indicated that the combination of PLGA/PEI/DNA nanoparticles with ultrasound was an efficient approach forin vivo gene transfection.

FIGURE 2 | The ultrasound-triggered cancer therapy based on PLGA-based DDSs. (A) Schematic illustration of the developed Fe3O4/PLGA microcapsules used for HIFU ablation of breast cancer in rabbits. (B) The volume of coagulative necrosis in excised tumor tissues from different groups exposed to HIFU (\*P < 0.05). (C) Positive index (PI) of PCNA in tumor tissues from different groups after HIFU ablation (\*P < 0.05). Adapted with permission from Sun et al. (2012). Copyright 2012, Elsevier. (D) Schematic illustration of the process of MPMBs delivery into tumor lymph node and the controllable release of DOX from MPMBs triggered by low frequency US sonication. (E) The in vitro DOX release profiles of MPMBs with or without low intensity sonication. (F) The proliferative index (PI) of tumor lymph nodes in various groups. \*P < 0.05 vs. the other groups. Adapted with permission from Niu et al. (2013). Copyright 2013, Elsevier. (G) The expression of β-galactosidase (β-gal) in control and ultrasound irradiated tumors assessed by western blot analysis. (H) Densitometry analysis of the β-gal expression from Figure 3G. Adapted with permission from Chumakova et al. (2008). Copyright 2008, Elsevier.

## Magnetic Field-Triggered Cancer Therapy

Magnetic hyperthermia induced by ACMF has been demonstrated to be a promising antitumor approach due to their relative non-invasive property (Jaber and Mohsen, 2013; Ramimoghadam et al., 2015). It has been proved that the magnetic nanoparticles (MNPs) could be used to produce magnetic hyperthermia via dipole relaxation under an external ACMF placed in the tumor regions, which can be employed to induce apoptosis of the tumor cells and make the tumor cells more sensitive to chemotherapy (Johannsen et al., 2007; Issels, 2008; Torres-Lugo and Rinaldi, 2013; Shen et al., 2015).

Fang et al. (2015) have designed and synthesized the magnetic responsive microspheres based on PLGA for combined chemotherapy and hyperthermia. **Figure 3A** shows the preparation procedure of magnetic PLGA microspheres contained DOX (DOX-MMS). Firstly, the workers in this study used the modified double emulsion solvent evaporation method to develop the DOX-loaded PLGA microspheres (DOX-MS). And then the positively charged polyethylenimine (PEI) was decorated on the surface of DOX-MS by electrostatic deposition to form an interlayer. Ultimately, the DOX-MMS were obtained by adsorbing the γ-Fe2O<sup>3</sup> nanoparticles (IOs) on the surface of DOX-MS through the electrostatic incorporation. SEM images of the DOX-MMS after ACMF treated for 30 min showed that the DOX-MMS were broken and the pores could be observed on the surface of DOX-MMS (**Figure 3C**), while the SEM images of DOX-MMS without exposure to ACMF showed no obvious changes in morphology and almost all the microspheres could be observed to maintain the integrated structure (**Figure 3B**). The destroyed structure of DOX-MMS after exposure to ACMF was induced by the heat effect generating from the IOs on the PLGA shell triggered by ACMF, which can lead to an accelerated DOX release behavior of DOX-MMS. **Figure 3D** shows the cell viability of 4T1 cells treated with DOX-MMS at different DOX concentrations with or without exposure to ACMF for 30 min. There was an apparent DOX dose-dependent decrease of cell viability in the ACMF treated group. However, the cell viability of 4T1 cells without exposure to ACMF exhibited a slight decline with the increase of DOX concentration. The in vivo tumor inhibition assay revealed that the DOX-MMS treated with ACMF showed an effective tumor inhibition effect in 4T1 tumor-bearing nude mice models. The severe necrotic tumor tissues with dark gray color can be observed and the relative tumor volume in DOX-MMS treated mice with exposure to ACMF was the lowest among all experimental groups (**Figures 3E,F**). And there was no significant body weight loss observed in all groups (**Figure 3G**). The hematoxylin and eosin (H&E) staining, TUNEL staining, and immunohistochemical anti-CD31 staining of the tumor tissues from mice in all groups were further performed. As shown in **Figure 3H**, the visible necrotic areas, necrotic or apoptotic cells stained brown, and the low microvessel density of tumor tissues can be observed from the DOX-MMS treated mice with exposure to ACMF. These results indicated

that the local hyperthermia triggered by the ACMF and the accelerated drug release behavior could enhance the antitumor effect of DOX-MMS.

### Radiofrequency-Triggered Cancer Therapy

Besides the ultrasound- and ACMF-triggered cancer therapy approach, the anti-cancer strategy triggered by radiofrequency (RF) has been extensively applied in various types of tumors (Miao et al., 2000; Dromain et al., 2002; Heynick et al., 2003). Although radiofrequency is a minimally invasive tool to inhibit tumor growth, the necessary high output power and long irradiation time of RF is needed, which can inevitably cause the damages to normal organs and tissues (Ryan et al., 2000). Magnetic metal nanoparticle have been validated to improve the ablated volume of tumors and in consequence promote the RF ablation of tumors, however, the high output power of RF is still needed to get high oscillating magnetic field (Kruse et al., 2011; Xu et al., 2012; Yun et al., 2014). Recently, the bubblesinduced cavitation has been proposed to be a promising strategy to enhance RF ablation of tumors.

Zhang K. et al. (2016) introduced the biocompatible DL-menthol (DLM) (melting point: 32–36◦C) to a PLGA nanocapsule to construct a DLM encapsulated PLGA-based nanocapsule (abbreviated as DLM@PLGA). The solid DLM possesses the property of continuous solid-liquid-gas (SLG) triphase transformation, which is perfectly suitable for RF ablation (Zhang K. et al., 2014). The continuous cavitation triggered by radiofrequency solidoid vaporization (RSV) can result in the continuous enhancement of RF ablation in a reduced RF power output and shorten irradiation time manner. And the external RF-mediated local heat in the RSV process can trigger the vaporization of encapsulated solid DLM and resulting in DLM bubbles continuously generating from DLM@PLGA (**Figure 4A**). SEM image of as-prepared DLM@PLGA exhibited the uniform spherical morphology with an average particle size of 450 nm (**Figure 4B**). A plenty of DLM bubbles with different particle sizes can be observed in the confocal laser scanning microscope (CLSM) image of DLM@PLGA suspension after exposure to RF heating (60◦C) (**Figure 4C**). The efficiency of RF ablation based on continuous cavitation was further evaluated in HeLa tumor-bearing nude mice models after intratumoral injection of PBS, PLGA, free DLM, and DLM@PLGA with exposure to RF (1 W of the output power and 30 s of the irradiation time). The obtained excised tumors from each group were used to calculate the ablated volume. As shown in **Figures 4D,E**, the DLM@PLGA treated HeLa tumors showed the largest ablated volume. And the three isolated ablation regions observed in the free DLM treated HeLa tumors were mainly resulting from the non-uniform distribution and agglomeration of free DLM induced by their hydrophobic property. Furthermore, tunnel staining and PCNA assay were further carried out to evaluate the molecular mechanism of the enhanced RF ablation. As shown in **Figure 4F**, the most apoptotic cells (brown color areas) but the least proliferating cells (brown color areas) were both observed in the DLM@PLGA treated

FIGURE 4 | The radiofrequency-triggered cancer therapy based on PLGA-based DDSs. (A) Schematic illustration of an external RF-triggered RSV process mediated by DLM@PLGA. (B) SEM image of as-prepared DLM@PLGA (indicated by red arrows). (C) Confocal laser scanning microscope (CLSM) image of DLM@PLGA suspension after exposure to RF heating. (D) Optical images of the HeLa tumors treated with PBS (c1), PLGA (c2), free DLM (c3), and DLM@PLGA (c4) upon exposure to RF (1 W of the output power and 30 s of the irradiation time). (E) The corresponding ablated volumes of HeLa tumors calculated from Figure 5D. (F) Tunnel staining and PCNA immunochemistry analysis of tumor tissues, wherein the ablated and un-ablated regions were separated by the red curves (scale bar is 100 µm). Adapted with permission from Zhang K. et al. (2016). Copyright 2016, American Chemical Society.

HeLa tumors. These results demonstrated that the solid DLM encapsulated into the PLGA nanocapsules could strengthen RF ablation through the continuous cavitation which mediated by the RSV process triggered by RF field.

### DESIGN OF PLGA-BASED DDSs WITH IMAGING PROPERTY

In recent years, contrast agents have opened up a new development direction for tumor imaging, gradually become a new hotspot for cancer diagnosis and treatment (Chen et al., 2014). Applying the contrast agents into tumor diagnosis can realize imaging of tumors with a high spatial resolution and sensitivity (Shen et al., 2017; Yang et al., 2017). At present, the commonly used diagnostic methods in clinic mainly includes ultrasound imaging, X-ray computed tomography imaging (CT), magnetic resonance imaging (MRI), which can provide certain information for the preoperative staging and prognosis of tumors (James and Gambhir, 2012). However, because of some drawbacks of contrast agents including poor stability, rapid elimination from the body, and lack of targeting, the location of the lesions cannot be accurately located during diagnosis, which may lead to the compromised therapeutic effect of tumors (Kirchherr et al., 2009; Niu et al., 2013). It has been demonstrated that loading imaging agents into PLGA-based DDSs could improve the targeting ability and biocompatibility of imaging agents (Shen et al., 2017, 2019). The imaging agents can be encapsulated into PLGA-based DDSs or conjugated on to the outer surface of PLGA-based DDSs through the covalently linking with functional groups (such as carboxylic acid and hydroxyl groups) of PLGA (Martins et al., 2018). For example, the imaging agents which possess the amine groups can be covalently linked to the terminal carboxylic acid groups of PLGA by forming an amide linkage (Malinovskaya et al., 2017). Additionally, the imaging agents can also be introduced into PLGA by an appropriate linker such bifunctional PEG linker (El-Gogary et al., 2014; Park et al., 2014). T2-contrast agents based on superparamagnetic Fe3O<sup>4</sup> nanoparticles can be encapsulated into the PLGAbased DDSs aiming for magnetic resonance (MR) imaging of tumors, which is a powerful and non-invasive imaging technique with relatively high temporal and spatial resolution (Li et al., 2013; Shen et al., 2017). A series of fluorescent dyes, nearinfrared fluorescent imaging agents, and quantum dots, etc. can also be introduced into the PLGA-based DDSs aiming for fluorescence imaging of tumors, which plays an important role in the research of tumorigenesis and development due to its high sensitivity and low cost property (Edgington et al., 2009; Rotman et al., 2011). ICG, as a near-infrared (NIR) fluorescent imaging agent with low toxicity, has been used for cardiac function monitoring, liver output, and retinal angiography (Owens, 1996). Moreover, ICG can be utilized as a photoacoustic imaging contrast agent to produce the enhanced photoacoustic signal due to its strong NIR light absorbance, realizing the photoacoustic (PA) imaging of tumors in a non-invasive manner (Chen et al., 2016). It is noted that the gold nanomaterials could be encapsulated into the PLGA-based DDSs or coated onto the surface of the PLGA-based DDSs for PA imaging or X-ray CT (Hao et al., 2015; Song et al., 2015). The highresolution tissue structure image of tumors will be obtained when employing the gold nanomaterials as the contrast agents for X-ray CT, which possesses the advantages of fast acquisition time, simple operation, and high availability (Li et al., 2010; Chen et al., 2014).

### PLGA-BASED CANCER THERANOSTICS

Nowadays, clinical diagnosis and treatment of tumors are two separate processes, patients usually need to be diagnosed before treatment, and the two separate medical procedures are likely to delay the best time to treat diseases (Wang et al., 2012). With the rapid development of molecular imaging technology, various imaging modes have been studied to improve diagnostic imaging. Therefore, the development of a multifunctional nanoplatform with diagnostic and therapeutic functions has been the future development trend of nanomedicine (Song et al., 2017). By applying the nanoplatform which integrated diagnostic and therapeutic agents to tumor treatment, the entire process of chemotherapy can be monitored in real time, and whether the chemotherapeutic drugs are effectively delivered to the tumor sites can be monitored. The position, size change, and metastasis of the tumors can also be monitored to determine whether the chemotherapeutic drugs are effective in killing tumors. Many researchers have developed various types of PLGA-based DDSs that can be used for MRI, CT, fluorescence imaging, ultrasound imaging, and photoacoustic (PA) imaging of tumors to achieve the accurate tumor detection, meanwhile these PLGAbased DDSs can also be used to load the therapeutic drugs such as chemotherapeutic drugs, small molecule inhibitors, photosensitizers, photothermal agents and siRNA (**Figure 5**; Menon et al., 2013). Yang et al. (2018) have developed the chargereversal PLGA-based ultrasound nanobubbles to co-load Dox and P-gp shRNA for reversal of drug resistance and enhancing the antitumor effect of chemotherapeutics. The in vitro and in vivo data substantiated that the drug and gene co-delivered PLGAbased nanobubbles could be used as an available theranostic agent for ultrasound imaging-guided chemotherapy and gene therapy of multiple drug resistance (MDR) tumors. Another novel PLGA nanoparticles coated with cancer cell membrane for dual-modal imaging-guided photothermal cancer therapy was prepared by Chen et al. (2016). The near-infrared light absorbing agent, ICG, was firstly encapsulated into the PLGA nanoparticles to obtain the ICG loaded PLGA nanoparticles which were employed as the core, then cancer cell membrane shell was coated onto the surface of ICG loaded PLGA cores by co-extruding the membrane vesicles and ICG loaded PLGA cores through a 220 nm polycarbonate membrane. The ultimately obtained cancer cell membrane-coated PLGA nanoparticles (ICNPs) with ICG loaded PLGA cores and cancer cell membrane shell were validated to have a favorable photothermal response and homologous targeting effect both at the cellular level and animal level. The results demonstrated that ICNPs could substantially

accumulate into subcutaneous breast cancer tissues in MCF-7 tumor-bearing nude mice through homologous targeting and EPR effect, and can be used as the fluorescence imaging and photoacoustic (PA) imaging agents to clearly identify tumor locations and boundaries. The in vivo results also corroborated that ICNPs could eradicate the tumors upon exposure to NIR laser and prevent the recurrence of tumors, showing the great potential of the developed cancer cell membranecoated PLGA nanoparticles as the versatile nanoplatform for homologous-targeting and fluorescence/photoacoustic imagingguided PTT. Gu et al. (2016) have reported a facile approach to construct a multifunctional nanocapsule based on PLGA to co-load BSA capped gold nanoclusters (AuNCs) and ICG. Because of the carboxyl groups existed on the surface of AuNCs and ICG co-loaded mPEG-PLGA (AuIP) nanocapsules, the amino groups of RGD peptides can be coupled to AuIP nanocapsules by forming the amide linkage. The obtained AuIP-RGD nanocapsules were demonstrated to be able to specifically target the U87-MG cancer cells that overexpress integrin αvβ<sup>3</sup> by CLSM analysis. The results further validated the satisfactory performance of AuIP-RGD in both one-photon and two-photon fluorescence imaging of tumors as well as the PTT of tumors, showing the great potential of AuIP-RGD nanocapsules as the theranostic nanoplatform for tumor diagnosis and treatment applications.

#### TARGETING STRATEGIES

The PLGA-based DDSs should be developed to achieve the effective drug delivery. In brief, the drugs loaded in the PLGAbased DDSs should be able to reach the targeted tumor site after administration, and the loss of drug activity and dose in the blood circulation before reaching the targeted site should be minimal. More importantly, the drug should only kill tumor cells without harmful effects on normal tissues and organs. Various modifications can be performed to fabricate the PLGA-based DDSs, such as changing their size, shape, structure, chemical and physical properties, etc. to make them accumulate into the tumor regions through passive targeting or active targeting approach (Torchilin, 2000; Farokhzad and Langer, 2009).

DDSs are inclined to passively extravasate through the leaky vasculature, which is the unique pathophysiological characteristics of solid tumors, and preferentially accumulate into tumor tissues through the passive targeting (Shen et al., 2019). Generally, the abundant tumor vasculature are leaky, and the pore size of the leaky vascular endothelial cell gap ranges from 100 to 800 nm (Torchilin, 2011). Meanwhile, the tumor tissues lack effective lymphatic drainage (Bamrungsap et al., 2012), which can lead to the decreasing diffusion process, ultimately resulting in the prolonged retention time of DDSs.

Therefore, the DDSs with a particle size smaller than the pore diameter can easily penetrate the interstitium and finally become accumulated in tumor sites. The phenomenon of passive accumulation of DDSs into tumor tissues is referred to the enhanced permeability and retention (EPR) effect (Danhier et al., 2010; Greish, 2010).

PLGA-based DDSs has been demonstrated to be the multifunctional DDSs that can target tumor sites owing to their characteristics such as small and tunable particle size, high stability, excellent biocompatibility, and simple surface modification (Deng et al., 2014; Yang et al., 2015, 2018; Shen et al., 2017, 2019). Appropriate surface modification (such as PEGylation, poloxamers and Tween 80 conjugation) of the PLGA-based DDSs can evade the phagocytic uptake by the reticuloendothelial system (RES) (Cai et al., 2016), leading to the prolonged circulation time in blood, which provide more opportunities for these surface-modified PLGAbased DDSs to accumulate into the tumor regions through passive targeting approach (**Figure 6A**). In our previously published literature, we have designed the PLGA-based nanoparticles modified with bovine serum albumin (BSA) to co-load near-infrared dye, indocyanine green (ICG) and chemotherapeutic drug, doxorubicin (DOX) for passive tumor-targeted combination cancer therapy (Shen et al., 2019). BSA modification can not only act as a surface stabilizer but also a biocompatible shell of the PLGAbased nanoparticles to evade the non-specific adsorption of plasma protein and the recognition of macrophage. The enhanced fluorescence signals of PLGA-based theranostic nanoplatform (denoted as IDPNs) were detected in tumor region after 24 h intravenous injection when compared to free ICG molecules, suggesting that the IDPNs possess the capability of passive accumulation into tumor sites via the EPR effect (**Figure 6C**).

To improve the targeting efficiency of PLGA-based DDSs to the targeted sites, affinity ligands such as antibodies (Venugopal et al., 2018), peptides (Graf et al., 2012), aptamers (Wu et al., 2010), or small molecules (Shen et al., 2012) can be introduced onto the surface of PLGA-based DDSs, which can be recognized by the specific receptors overexpresses on tumor cells and then bind tumor cells through ligand-receptor interactions, achieving the active targeting and accumulation of PLGA-based DDSs into tumor sites via the receptormediated cell uptake, as illustrated in **Figure 6B**. Folic acid (FA) is a kind of small molecule nutrient, as an nonimmunogenic targeting ligand, it possesses high binding affinity to the folate receptor, which overexpresses on the surface of various types of human cancer cell membranes (Shen et al., 2012). Folate receptor (FR) has been extensively utilized for active tumor-targeting of nanocarriers via the receptormediated endocytosis (El-Gogary et al., 2014). Besides the passive and active targeting, magnetic targeting is another targeting strategy, which can be achieved by loading the magnetic nanoparticles into PLGA-based DDSs before exposing to an external magnetic field (Yu et al., 2016; Shen et al., 2017). Our group has developed the folic acid-conjugated PEGylated PLGA nanoparticles co-encapsulated with CdSe/ZnS

quantum dots, doxorubicin and Fe3O<sup>4</sup> nanoparticles followed by the adsorption of vascular endothelial growth factor (VEGF)-targeted small hairpin RNA (abbreviated as LDM-PLGA/PPF/VEGF shRNA) for tumor targeted drug delivery and cancer imaging (Shen et al., 2017). The folic acid modified on the surface of nanocomposites and the Fe3O<sup>4</sup> nanoparticles encapsulated into the nanocomposites endow LDM-PLGA/PPF/VEGF shRNA with folate and magnetic dual targeting functions by folate receptor-mediated endocytosis and magnetic guidance. HeLa cells treated with LDM-PLGA/PPF nanocomposites exhibited the increase of intracellular DOX fluorescence signal along with the decreasing distance from the magnet, while an increased fluorescence signal of DOX in HeLa cells without free folate pretreatment was observed compared to the cells with free folate pretreatment, demonstrating the enhanced cellular uptake of LDM-PLGA/PPF by HeLa cells via magnetic guidance and folate receptor-mediated endocytosis (**Figure 6D**).

Currently, applying the intrinsic homologous adhesion property of the cancer cells to tumor targeting is another promising active tumor-targeting approach. Surface adhesion molecules such as galectin-3 or N-cadherin expressed on cancer cells have been proven to be the leading cause of multicellular aggregation formation. Therefore, the cancer cell membrane can be utilized for the surface functionalization of DDSs in order to obtain the homologous targeting capability. Chen et al. (2016) reported a cancer cell membrane-coated PLGA nanoparticles with encapsulation of indocyanine green (ICG) molecules for specific homologous tumortargeted and photoacoustic/fluorescence imaging-guided photothermal cancer therapy. The results of the in vivo biodistribution and photoacoustic/fluorescence imaging showed a remarkably enhanced tumor accumulation of MCF-7 cell membrane-coated PLGA nanoparticles at 24 h post-injection when compared to the non-cancer cell membrane-coated PLGA nanoparticles in human breast cancer cells (MCF-7) bearing nude mice models. The illustrated examples validate that the PLGA-based DDSs with appropriate surface functionalization can provide a promising approach for passive tumor-targeting via the EPR effect or active tumortargeting through ligand-receptor interactions, resulting in the enhanced delivery and accumulation of PLGA-based DDSs into tumors.

### CONCLUSION AND PERSPECTIVES

In this review article, we summarize the recent status of research on the application of PLGA-based DDSs in remotely triggered cancer therapy and the strategies for tumor imaging provided by PLGA-based DDSs, specifically focusing on employment of PLGA-based drug DDSs for external stimuli-triggered cancer therapy including phototriggered, ultrasound-triggered, magnetic field-triggered, and radiofrequency-triggered cancer therapy. These drug delivery systems based on PLGA were shown to possess excellent biocompatibility and biodegradability, uniform particle size,

nanocomposites for 8 h under exposure to an external magnet with or without free folic acid (FA) pretreatment. Adapted with permission from Shen et al. (2017).

Copyright 2017, Dove Medical Press.

on-demand drug release behavior and external stimuli-triggered cancer therapy approach, with the overall goal to enhance the antitumor efficacy by the enhancement of drug accumulation in tumor region and reduced side-effects of PLGA-based DDSs.

Although a variety of inorganic or organic materials have been utilized to fabricate the smart DDSs including inorganic nanoparticles, liposomes, polymer micelles, and polymer nanoparticles which can be used to co-deliver therapeutic agents and imaging agents for cancer theranostic applications, several problems are still needed to be solved, such as the complicated synthesis process, non-biodegradability, and uncontrollable drug release behavior of some DDSs. For instance, the inorganic nanomaterials such as gold and silica which have poor biodegradability are most likely to cause the long-term toxicity and raise the safety concerns. And other metal nanomaterials such as copper and silver also exhibit some cytotoxicity. Moreover, the uncontrollable drug release behavior of some DDSs is highly likely to cause the compromised anticancer effect. However, polymer nanoparticles based on poly(D,L-lactic-co-glycolic acid) (PLGA) have attracted considerable attention due to their unique physical and chemical properties, such as tunable particle size, uniform and regular morphology, large surface area, favorable pharmacokinetic property, excellent biocompatibility and biodegradability. More importantly, PLGA has been approved for the medical applications by the United States FDA, making them a favorable material to construct the nano- or micro-platforms for medical applications. Unlike the polymer nanoparticles based on poly(butylcyanoacrylates) (PBCA) of which the hydrolysis products are cytotoxic polycyanoacrylic acid and alcohol, the PLGA-based polymer nanoparticles can be degraded into the non-toxic lactic acid and glycolic acid, which can be eventually degraded into water and carbon dioxide. The biocompatible and biodegradable nature of PLGA makes them preferentially used to prepare the nano- or micro-platform. And the appropriate outer surface engineering (such as PEGylation, BSA modification and conjugation of targeting ligands) of PLGA-based DDSs can endow them with a prolonged blood circulation and enhanced accumulation in tumor sites.

Besides the hydrophilic therapeutic drugs and imaging agents, the hydrophobic therapeutic drugs and imaging agents can be encapsulated into the PLGA nanoparticles by using the water-in-oil-in-water (W/O/W) double emulsion method, which can improve the targeting and bioavailability of hydrophobic drugs and imaging agents, and eventually achieving the goal of cancer theranostics. However, it is not appropriate to design the extremely complicated nano- or micro-platform based on PLGA, since their unstability or potential uncertainty in physiological environment may lead to the unfavorable tumor inhibition effect. How to design and build a versatile but simple drug delivery systems with the effective antitumor activity remains a challenge for the new research. Employing external stimuli to trigger the cancer therapy is a promising antitumor approach, which can selectively and precisely eradicate the solid tumors and remotely control the drug release. However, the applications of remotely triggered cancer therapy remains limited. For example, the PDT efficacy will be weakened with the oxygen consumption during PDT (Wang et al., 2017). Developing a strategy that can afford sufficient oxygen for photosensitizers to consume and continuously generate ROS, has been proved to be a promising approach for enhancing PDT efficacy (Tang et al., 2018). In addition, the tissue penetration depth of the irradiation light is an essential issue for the photo-triggered therapeutic application. Employing the light in the NIR region may address this issue due to their advantages of real-time dosage adjustment, high spatiotemporal precision, and deep tissue penetration (Lin et al., 2016; Zhang P. et al., 2016; Shen et al., 2019).

Although the PLGA-based DDSs have been extensively studied, which can provide a safe and reliable approach for the tumor diagnosis and therapy in a minimally invasive manner, most of them are just preclinical studies, and still face many challenges. The property of low drug loading, high production costs, and inability to mass-produce of PLGA-based DDSs may limit their clinical applications. In order to apply the PLGA-based DDSs load with diagnostic and therapeutic agents to the clinic, the researchers should firstly overcome the obstacles of large-scale production of the nanoparticles. And it will provide a solid foundation for the application of PLGA-based DDSs in tumor therapy if more methods are provided to improve the drug loading capacity of PLGA-based DDSs. In addition, the cytotoxicity and immune response of all components of the PLGA-based DDSs in the body should be evaluated. The clinical transformation of PLGA-based DDSs can be further promoted with the continuous development of materials science and nanotechnology. And the great efforts have been made to improve their biocompatibility, stability, safety, drug loading capacity, and targeted delivery, which will endow PLGA-based DDSs with great potential in cancer diagnosis and treatment.

### AUTHOR CONTRIBUTIONS

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

## FUNDING

This work was supported, in part or in whole, by the National Natural Science Foundation of China (81671821, U19A2006, 11772088, 31700811, 11802056, 31800780, 11972111, and 31900940), the China Postdoctoral Science Foundation (2018M640904, 2019T120831), the Sichuan Science and Technology Program (2019YJ0183 and 2019YJ0184), and the Fundamental Research Funds for the Central Universities (ZYGX2019J117).

#### REFERENCES

fbioe-08-00381 May 2, 2020 Time: 20:32 # 16


nanostructures for cancer imaging and therapy. Pharmaceutics 11:E280, doi: 10.3390/pharmaceutics11060280


composites for tumor-targeted drug delivery. Bioconjug. Chem. 23, 1010–1021. doi: 10.1021/bc300008k



**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 © 2020 Shen, Li, Xie, Feng, Chen, Yang, Wu, Deng and Liu. 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.

# Tagged Halloysite Nanotubes as a Carrier for Intercellular Delivery in Brain Microvascular Endothelium

Mahdi Yar Saleh† , Neela Prajapati† , Mark A. DeCoster\* and Yuri Lvov\*

Institute for Micromanufacturing and Biomedical Engineering Program, Louisiana Tech University, Ruston, LA, United States

Neurological disorders that are characterized by unpredictable seizures affect people of all ages. We proposed the use of nanocarriers such as halloysite nanotubes to penetrate the blood–brain barrier and effectively deliver the payload over an extended time period. These 50-nm diameter tubes are a natural biocompatible nanomaterial available in large quantities. We proved a prolonged gradual drug delivery mechanism by the nanotube encapsulating rhodamine isothiocyanate and then ionomycin into brain microvascular endothelial cells (BMVECs). Through delayed diffusion, the nanotubes effectively delivered the drug to the primary BMVECs without killing them, by binding and penetration in time periods of 1 to 24 h.

#### Edited by:

Stefano Leporatti, Institute of Nanotechnology, Italian National Research Council, Italy

#### Reviewed by:

Marina Massaro, University of Palermo, Italy Mel De Villiers, University of Wisconsin–Madison, United States Xiangyang Shi, Donghua University, China

#### \*Correspondence:

Mark A. DeCoster decoster@latech.edu Yuri Lvov ylvov@latech.edu †These authors have contributed

equally to this work

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology

> Received: 13 March 2020 Accepted: 20 April 2020 Published: 14 May 2020

#### Citation:

Saleh MY, Prajapati N, DeCoster MA and Lvov Y (2020) Tagged Halloysite Nanotubes as a Carrier for Intercellular Delivery in Brain Microvascular Endothelium. Front. Bioeng. Biotechnol. 8:451. doi: 10.3389/fbioe.2020.00451 Keywords: halloysite (HNT), brain delivery, endothelia cell, nanomaterial application, drug–drug interactions

## INTRODUCTION

Brain diseases, such as central nervous system disorders, are of the most poorly treated diseases in today's world. Occurring frequently in people of all ages, impacting their way of life, and increasing the chances of premature death. Epilepsy is a common example of a neurological disorder that is characterized by unpredictable seizures that results in unusual behavior, such as involuntary movements. Around the globe today, an estimated 50 million people are diagnosed with this disorder causing it to be one of the most common brain diseases. Thus, if the disease could be treated properly through various methods, such as nanotechnology, up to 70% of diagnosed patients could be cured (Silva, 2008; Bennewitz and Saltzman, 2009; Pehlivan, 2013; World Health Organization [WHO], 2020). Current anti-seizure medication has a wide range of side effects which include dizziness, tiredness, and trouble speaking. Utilizing nanosized carriers is an effective route to determine targeted and slow release drug delivery techniques to minimize these side effects and provide an overall more effective treatment. Using less than 50 nm capsules promises better penetration into cells, especially if one uses tubule drug carriers which have improved efficiency in passing cell membrane as defined by the smallest cross-section size. Gamma-amino-n-butyric acid (GABA), glutamic acid, and dopamine are important brain neurotransmitters for epilepsy research, and their nano-formulation for intracellular sustained delivery is promising for higher efficiency.

Brain microvascular endothelial cells (BMVECs) are the key cells in the blood–brain barrier, which is the multicellular membrane between the brain's blood vessels and a brain tissue. BMVECs constitutes the tight junction proteins which prevent the entry of pathogens and other toxic substances into the brain, but it also prevents most potential drugs against neurological and mental disorders to cross the barrier and readily reach into the brain tissue (Mahringer et al., 2013; Lvov et al., 2016). We exploit halloysite clay nanotubes (HNT) to penetrate the endothelial cells and effectively deliver the payload over an extended time. Our hypothesis is based on the conception of "nano-torpedo" when rod-like tubule hollow clay capsules bind or penetrate through

cell membrane exploiting its very small, 50-nm diameter crosssection and deliver the drug load in the interior.

Halloysite nanotubes are formed by 10–15 revolutions of 0.7 nm thick kaolin aluminosilicate sheets and have diameters ranging between 50 and 60 nm, lumen diameters of 12–15 nm, and lengths within 500–900 nm (**Figure 1**) (Liu et al., 2014; Lvov et al., 2016). It is an environmentally friendly, natural, and cheap tubule nanomaterial available in large quantities. Halloysites outer surface is composed of SiO2, and the tube's interior is composed of Al2O3, which are oppositely (negative/positive) charged in the pH range of 3–9. The structure of halloysite in **Figure 1** shows how the payload can be loaded inside the positively charged lumen of halloysite, which is especially efficient for spontaneous adsorption of negatively charged drug molecules. Based on geometrical sizes of halloysite, one may conclude the maximal volume load inside the tubes of 10–12 vol.%, which may reach for organic drugs ca 15 wt.%. This is a typical drug load given in many publications, as summarized in Santos et al. (2019). Higher drug loading means that drug molecules are adsorbed on the tube external surface, which may change the formulation properties that are observed through zeta-potential and colloidal stability. Zeta potential is an important though indirect indication of the innermost drug loading. Thus, inner adsorption of negative molecules usually increases the electrical potential magnitude from ca. −30 mV in pristine to minus 45–50 mV in loaded nanotubes, simultaneously improving the sample colloidal stability (Liu et al., 2019). Drugs that were used are khellin, oxytetracycline, gentamicin, ciprofloxacin, vancomycin, atorvastatin, metronidazole, dexamethasone, doxorubicin, furosemide, nifedipine, curcumin, resveratrol, povidone iodine, amoxicillin, brilliant green, chlorhexidine, and DNA and viral genes were also successfully loaded in halloysite (Santos et al., 2019).

Halloysite is a biocompatible material with low toxicity assessments (Vergaro et al., 2010; Dzamukova et al., 2015b; Kruchkova et al., 2016; Hu et al., 2017; Kamalieva et al., 2018; Mehdia et al., 2018; Wang et al., 2018; Fakhrullina et al., 2019; Zhao et al., 2019). Many researchers reached a consensus that these clay nanotubes are safe up to a 10 mg/mL formation, which is less toxic than common table salt (Vergaro et al., 2010). This was tested on several in vitro and in vivo systems: cells lines, microworms, infusoria, fishes, mice, and rats (Santos et al., 2019). The only minor toxic effect was found with high oral halloysite consumption. When acidic clay decomposes, the

FIGURE 1 | A TEM and SEM image of halloysite nanotubes (a,b). A scheme of rolled clay (c).

stomach increased Al3+accumulation (Kamalieva et al., 2018). The mice that were orally fed with low nanoclay doses (5 mg/kg mice weight, which corresponds to 3 g of halloysite daily consumption for adult human for 1 month) have shown no oxidative stress or other toxicity signs, and even demonstrated higher growth rates.

Clay nanotube loaded with drugs will penetrate cells more efficiently than spherical particles of the same mass (Dzamukova et al., 2015a; Wang et al., 2018). This approach allowed for an effective halloysite delivery of doxorubicin and other anticancer drugs (Yang et al., 2016; Zhang et al., 2019). HNT/brilliant green formulations with intracellular delivery allowed for preferable elimination of human lung carcinoma cells (À549) as compared with hepatoma cells (Hep3b) due to different -intracellular penetration (Dzamukova et al., 2015a).

In this work, we studied halloysite nanotube penetration into primary rat BMVECs, which are the main cell type that prevent entry of drugs through the blood–brain barrier. We demonstrated with fluorescent rhodamine B isothiocyanate dye that halloysite binds on the cell, penetrates the cell interior, concentrates around the nuclei, and may deliver a drug load. Next, we developed a model for brain cell stimulation using halloysite loaded with ionomycin, a widely used Ca2<sup>+</sup> ionophore (Morgan and Jacob, 1994; Kaushik et al., 2018; Long et al., 2018; Wu et al., 2018) to monitor calcium transport across membrane and to stimulate a response from brain cells. Ionomycin-HNT formulations resulted in intracellular delivery, which was monitored by Ca2<sup>+</sup> changes in the cells, showing gradual and prolonged delivery of ionomycin into the brain endothelial cells. An outline in **Scheme 1** displays the progressive change in calcium signals throughout the stimulation, proving halloysite formulations have the potential as an efficient carrier for drugs or selected neurotransmitters (glutamate) targeted for brain disorders treatments.

#### MATERIALS AND METHODS

#### Halloysite Loading With Rhodamine Isothiocyanate

Halloysite samples were prepared by loading rhodamine B isothiocyanate **(**RITC) and ionomycin through stirring, centrifugation, and sonication at various ratios including 10 mg halloysite/1 mL DI water per 0.5, 1, and 2 mg of RITC. The solutions were then sonicated and vortexed for 1 min, then mixed on a stir plate for 24 h at room temperature. The mixture was washed once by centrifugation at 2500 RPM for 2.5 min and then dried at 70◦C for 24 h. When loading ionomycin, 20 mg of pristine halloysite clay nanotubes were stirred with 1 mM ionomycin for 24 h. The mixture was then washed with sterile water by centrifugation 2500 RPM for 3 min one time. The solution was then freeze dried for 20 min and placed in a vacuum for 24 h to remove the excess solution. Samples were then characterized by the zeta potential analyzer, which displays the surface charge and thermogravimetric analysis, allowing us to calculate the loading percentage.

## Neurotransmitter Loading

This procedure is accomplished by taking 50 mg of etched or pristine halloysite with 50 mg of glutamic acid separately in 5 mL of DI water, creating a super saturated solution through sonication for 5 min and stirring for 24 h at room temperature. The last step is for the suspensions to dry in the oven for 24 h at 70◦C. After loading the halloysite samples were characterized through zeta potential analyzer and thermogravimetric analysis to determine a surface charge and measure the percentage of material loaded.

## Cell Culture

Primary brain microvascular endothelial cells (BMVECs) were obtained directly from the rat brain cortex. Rat pups, 1 or 2 days old were euthanized to obtain the rat brain cortex. Cortex obtained by dissection of the pup's skull is cleared out of meninges under the microscope. The cortical tissue thus obtained were then treated with trypsin and triturated, with the process repeated at least three times in order to break the brain tissue into cells. After each trituration, cells were incubated at room temperature under sterile conditions for about 10 min and supernatant was collected in a 15 mL tube. The collected supernatant was then centrifuged to obtain a pellet of mixed culture of brain cells. BMVECs were isolated from this primary culture by treating them with 5.5 µM of puromycin dihydrochloride to kill all other cell types except the endothelial cells as shown in **Figure 2A**. Endothelial cells are encoded with a puromycin N-acetyl transferase gene (PAC gene), which confer resistance to the action of puromycin (Thiel and Audus, 2001; Wang et al., 2012). Thus, isolated BMVECs were then cultured in vitro at 5% CO<sup>2</sup> and 37◦C in rat endothelial growth medium (Sigma Aldrich). The cells were characterized by staining them against Von Willebrand Factor (VWF), **Figure 2B**, an essential blood clotting protein specific to endothelial cells (Mbagwu and Filgueira, 2020).

### Treatment With Rhodamine Isothiocyanate (RITC) Loaded Halloysites

Brain microvascular endothelial cells between primary passages of 3 and 7 were used for the experiments. Cells were plated in 48 well cell culture plates at 10K per well density and treated with RITC loaded halloysite, halloysite alone (negative control) and RITC alone (positive control) at 50–70% confluency. Cells were treated with 10 µg/mL of halloysite alone (negative control) and 10 µg/mL of halloysite loaded with RITC [for all nanoclay samples loaded with different concentration of RITC (1:5 and 1:10)]. The concentration of RITC alone was chosen to correspond to the loaded halloysite formulations.

### Brain Microvascular Endothelial Cells (BMVECs) Stimulation Using Ionomycin Loaded Halloysite

Primary BMVECs were plated at the density of 10K per well in 48 well plates. At the confluency of 70 to 80% each well-containing the cells were loaded with 500 µL of Fluo-3 AM loading solution, i.e., Lockes' solution with Fluo-3 AM dye (1:500) and pluronic

acid (1:1000) (Yamamoto et al., 2011). The cells were treated with this solution for 1 h at 37◦C, in 5% CO<sup>2</sup> incubator and then – with 500 µL of recovery solution for 1 h. The recovered cells were then stimulated with halloysite loaded with ionomycin (10 and 50 µg/mL), halloysite alone (50 µg/mL), ATP 100 µM, and ionomycin 1 µM at different instances depending on the experiment. Real time intracellular calcium (Ca2+) change due to ionomycin transport across the system was recorded by capturing images every 4 s with InCyt Im1 software on the imaging system. The images were then used for Ca2<sup>+</sup> signal analysis, which utilizes change in fluorescence intensity as the function of time as a measure for quantifying calcium changes in cells.

## RESULTS

### Intracellular Rhodamine Isothiocyanate (RITC) Delivery With Clay Nanotubes

An observation of halloysite binding and penetration into the endothelial cells was visualized as red fluorescence concentrated within the cells, **Figure 3**. The addition of only RITC dye did not color the cells interior within 30 min of treatment; while the dye loaded halloysite bound or penetrated the cells and start releasing the dye inside the cells, coloring them red. Within a 4-h time frame RITC and RITC-HNT both showed cells with more fluorescence compared to the same samples in the 30 min exposure time. Furthermore, RITC-HNT formulations displayed greater fluorescence compared to RITC only, proving the dye delivery into the cells.

The timeframe of these images is 30 min after delivery of both samples (RITC and HNT + RITC) to the cells, with a concentration of 10 µg/mL for HNT + RITC and 2 µg/mL of RITC alone. The loading of RITC in HNTs was 20 wt.%, thus making the same amount of RITC added to the samples. The small bright dots found in the bottom images of **Figure 3E** are the aggregated halloysite tubes that contain RITC. HNT-RITC aggregation is displayed mostly along the cellular membrane and inside the endothelial cells as indicated by nuclear exclusion.

Therefore, halloysite nanotubes are highly capable of encapsulating, transporting, and slowly releasing the dye (or drugs as we will show with an example of ionomycin) over a few hours. It is important to note that we did not modify the surface of the halloysite with any type of polymer or silane coating. **Figure 3B** displays RITC alone added at the same concentration showing a much dimmer visualization of the cell's responsiveness to this non-encapsulated dye.

Results for the 24-h treatment were similar with more profound fluorescence in both conditions but RITC loaded halloysite delivered more dye into the cells at every time period as compared to just the dye alone, **Figure 4**. Images at a time point of 24 h displayed the nanotubes distribution more evenly over the cell interior, still contained within the cell body **Figure 4**. One could see that RITC-HNT concentrated in some smaller spots of ca 1 µm dimeter, which may be the nuclear surrounding, as it was found for MCF-7 cells treated with halloysite (Vergaro et al., 2010). Throughout the trials, we detected a nuclear exclusion, extended length of fluorescence, and that the tubes did not stress or kill the cells. The clay nanotubes are displayed as small dots in **Figure 4E** and are brighter than the dye spread inside the cells. The merged image (**Figure 4F**) of phase and fluorescence settings (**Figures 4D,E**) for cells treated with HNT-RITC gives a clear picture of the localized nanotubes with red dye in the cell cytoplasm with distinct nuclear exclusion, which displays greater binding and aggregation of the tubes on the cell surface in 24 h compared to the 30 min treatment. After cell fixation (4 days), we see that cells treated with dye only was washed away with only minimal fluorescence remaining on the cells. While the HNT-RITC treated cells still showed significant fluorescence indicating a prolonged delivery of the dye from the clay nanotubes. **Figures 4G,H** show that the cells treated with halloysite nanotubes alone (negative controls) do not show any fluorescence by themselves.

The fluorescence images (middle) highlights nanotubes or dye localization in the outer region of the cell networks, while

the merged images (left) illustrate a clear picture of materials (nanotubes and dye) localization inside or on the cell surface.

#### Halloysite Loading With Glutamic Acid and Ionomycin

Glutamic acid has a negative charge and being loaded into the tube's lumens demonstrated weight percent change of 3.4 ± 0.2 wt l%, corresponding to the halloysite loading. Loading of ionomycin was 16.2 ± 0.2 wt% (**Figure 5**). The percentages of glutamic acid and ionomycin encapsulation were estimated based on the weight change at different temperatures using thermogravimetric analysis. There is a consistency in the drugs loading results, probably, based on the similar loading mechanism enhanced by an attraction of negative drug molecules into the positive lumen of nanotubes. The used pristine halloysite had a zeta-potential value of −30 ± 2 mV and after the drugs loading, it became −45 ± 1 and −48 ± 2 mV correspondingly for ionomycin and glutamic acid. These formulations provided an enhanced colloidal stability that took 3–4 h for settling for pure halloysite and 8–10 h in the loaded samples. The precipitation time for aqueous unloaded and loaded halloysite ranged from minutes to days, in correspondence to the respective zetapotential that were measured. Therefore, the ability to load these neurotransmitters into clay nanotubes is achievable but further testing on the surface area and inner lumen with the transmitters need to be performed to know their exact location along with in vitro testing.

These results demonstrate that one can sufficiently load selected brain drugs into halloysite nanotubes and deliver them into the cells in a manner similar to the procedure of loading RITC. Furthermore, we will concentrate on the analysis of ionomycin delivery because we have a well-elaborated method to characterize the drug release kinetics with Ca2<sup>+</sup> analysis.

### Delivery of Ionomycin – Halloysite Formulations Into Endothelial Cells (Ca2+-Analysis)

In real time calcium imaging, we observed that the cells response to ionomycin had a spiked increase in Ca2<sup>+</sup> which decayed quickly due to clearance by cells as shown in baseline (**Figure 6A**). When we used halloysite alone, there was only a small response (**Figure 6B**). When ionomycin was encapsulated with halloysite, we achieved a higher response of Ca2<sup>+</sup> for the same concentration (1 µM) loaded that was also used for the control resulting in a gradual rise in Ca2<sup>+</sup> which remained higher for a longer time period until it was diluted by the addition of any other stimuli. This increase of Ca2<sup>+</sup> indicates the gradual and prolonged transport of ionomycin through halloysite across the cell membrane (**Figures 6B,C**). Each experiment was ended with ionomycin stimulation to ensure that the stimulus did not kill the cell under observation. To ensure that the cells were healthy and responding normally to other physiological stimuli, cells were tested with ATP, a well-known stimulator for BMVECs. Ionomycin is a well-known antibiotic and has been also known to induce cancer cell death and proliferation. Results showing sustained delivery of ionomycin might suggest a potential application of these formulations for cancer treatment (Park et al., 2005; MacLean and Yuste, 2009; Han et al., 2013).

All cells were stimulated with ionomycin at the end of experiment (B-E), except (A) which was stimulated by

HNT-ionomycin of 50 µg/mL. This demonstrated that the cells were still responsive to Ca2<sup>+</sup> changes, ensuring no cell death, and avoiding occurrence of false signals during the experiments.

Ca2<sup>+</sup> peak analysis presented in **Figure 7** allows us to come to following conclusions: stimulation by ionomycin (positive control) shows an instant peak of Ca2<sup>+</sup> that decays quickly compared to HNT-ionomycin nanocapsules (50 and 10 µg/mL) which shows a gradual influx of Ca2<sup>+</sup> and higher delivery of ionomycin in the cells for both concentration (**Figures 7A,B**). Cells that responded to ATP showed a significant peak that is normally observed for physiological conditions (**Figure 7C**). Stimulus by halloysite alone produces a slight Ca2<sup>+</sup> response, which was much less compared to HNT-ionomycin formulations with the cells showing no visible toxicity as well as remaining active afterward.

### Statistical Analysis of Calcium Response to Different Stimuli

The graphs in **Figure 7** show signals obtained for only 5 cells for each condition for simplification and clear representation of the data. The number of cells captured per frame for a condition being tested in an experiment act as a region of interest, and it ranged from 41 to 140 cells. After obtaining the calcium signals for different stimulations, the Ca2<sup>+</sup> fluorescence intensity data as a function of time are used to extract the percentage of peak Ca2<sup>+</sup> response above the baseline (**Figure 8B**) and the cells that responded to the stimuli (**Figure 8C**). The time taken for the stimuli to reach the peak response (**Figure 8D**) was also extracted. The bar graphs are obtained by averaging the values over the region of interest from each condition used for the experiment. At least five conditions were tested here, three times. Overall, the number of cells analyzed was 1,232 for all wells and conditions reported for calcium imaging.

One can see that the calcium peak response above baseline gave us an idea of how high the calcium response to a stimulus is. The results (**Figure 8B**) shows that the increase in peak calcium response above the baseline was 120% for HNT-ionomycin (50 µg/mL), and it was greater than for ionomycin alone (positive control) at 82 ± 2%.

The halloysite (negative controls) showed a slight Ca2<sup>+</sup> response, which was insignificant at 16% compared to the high

response of 94% for the same concentration of the HNTionomycin samples. The results that the nanotubes alone can also produce much smaller 16% peak Ca2<sup>+</sup> response suggest us that there is some form of advantageous cellular interaction and communication between the nanoclay and endothelial cell networks which is served by the signaling molecule, calcium. For ATP stimulation the signal was 40%. The peak response to ATP showed that the cells were healthy and at normal physiological condition.

(C) peak stimulation by 50 µg/mL nanotubes loaded with ionomycin. Magnification = 200X.

Each experiment had different numbers of cells giving rise to different regions of interest. **Figure 8C** represents the percentage of cells that responded for the given stimulus in an experiment. One can see that a higher number of cells responded to the HNT – ionomycin formulation with 97% for concentration of 10 µg/mL and at 92% for 50 µg/mL, while it was 65% for 50 µg/mL halloysite alone, 72% for 1 µM ionomycin, and 76% for 100 µM ATP. This result indicates that all the parameters discussed here are supported by a high response of cells, but the data specifically displays a higher value in the loaded samples compared to the other conditions tested.

**Figure 8D** represents the time taken for the cells to produce the peak calcium response. It tells us how delayed or instant the peak response was, indicating the delivery properties of halloysite clay nanotubes. The duration of release until peak calcium response after the stimulation was found to be 80 s for the nanoclay loaded with ionomycin at 10 µg/mL, 54 s for halloysite loaded with ionomycin at 50 µg/mL and 66 s for 50 µg/mL halloysite alone. This result along with results of **Figures 8B,C** explains that using a low concentration 10 µg/mL HNT-ionomycin formulation is enough to get enhanced delivery of ionomycin for a prolonged time as compared to using just ionomycin itself. Using a higher concentration (50 µg/mL) can increase this response further but delay is better achieved with the use of lower concentrations.

(Iono) 1 µM (positive control); (B) stimulated by 50 µg/mL HNT- ionomycin (sample tested); (C) stimulated by ATP, a well-known Ca2<sup>+</sup> stimulator (positive control indicating healthy cells), followed by HNT-ionomycin formulation (10 µg/mL) showing the comparison between the Ca2<sup>+</sup> responses of physiological stimulation (ATP) and HNT-ionomycin formulation for the same cells. (D) Cells stimulated by 10 and 50 µg/mL of HNT-ionomycin showing comparative results for the cells when stimulated by lower and higher dose of HNT-ionomycin; (E) stimulated by 50 µg/mL of empty HNTs (negative control).

In contrast, the duration was very short for ionomycin 1 µM at 13 s and ATP at 16 s. The duration was found higher for the halloysite loaded and unloaded samples as compared to ionomycin and ATP in their soluble forms. This indicates that there is delayed diffusion from the nanotubes due to the soluble ionomycin and ATP release being much faster.

All the results discussed in **Figures 7**, **8** indicate that the halloysite loaded with ionomycin showed delayed and gradual release of ionomycin into the cells, that once it reaches the peak response it continues to diffuse for a prolonged time which is expected to be up to 24 h as was observed for the nanotubes loaded with RITC **Figures 3**, **4**. 24 h treatment of BMVECs with ionomycin concentration of 1 µM used throughout the experiment didn't show any cytotoxicity to the cells.

### CONCLUSION

Halloysite nanotubes have a great potential in delivering drugs effectively to the brain because they were not toxic to the endothelial cells, they were capable of slowly releasing drugs over various time spans ranging from minutes to hours, and are attracted to the cells that reside in the blood–brain barrier. The ability of halloysite to enhance the calcium response in BMVECs by loading it with ionomycin drastically extended the delivery time of the compound compared to the use of the ionophore alone (non-encapsulated). With this new information, we can extend this approach for treating brain cancer cells through drug delivery based upon the data of inhibitory effects of ionomycin on the cells and its use as a chemosensitizer. We confirmed a delayed and prolonged diffusion of the drug delivery mechanism due to the halloysite nanotubes loading cell probes (ionomycin + RITC). This provides a sustained delivery strategy for drug penetration across the blood–brain barrier.

## DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

## ETHICS STATEMENT

The animal study was reviewed and approved for all cell studies carried out in this work, which used primary tissues derived from laboratory animals as approved by Louisiana Tech University IACUC (Louisiana Tech University IACUC, Center for Biomedical Engineering and Rehabilitation Science, Ruston, LA, United States).

## AUTHOR CONTRIBUTIONS

MS: nanoformulation experiments under YL supervision. NP: cell culture treatments under MD: supervision. YL and MD writing the manuscript.

### FUNDING

We thank Louisiana Board of Regents grant LaSpace-GSRA#19- 287 for support of this work. Support by NSF – 1632891 grant was acknowledged. Any opinions, findings, and conclusions, or recommendations expressed in this report are those of authors and do not necessarily reflect the view of National Science Foundation.

### REFERENCES

fbioe-08-00451 May 12, 2020 Time: 20:53 # 10


**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 © 2020 Saleh, Prajapati, DeCoster and Lvov. 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.

# Smart Nanotheranostics Responsive to Pathological Stimuli

Alessandro Parodi <sup>1</sup> \*, Magdalena Rudzinska<sup>1</sup> , Stefano Leporatti <sup>2</sup> , Yuri Anissimov 1,3 and Andrey A. Zamyatnin Jr. 1,4 \*

*1 Institute of Molecular Medicine, Sechenov First Moscow State Medical University, Moscow, Russia, <sup>2</sup> CNR NANOTEC - Istituto di Nanotecnologia, Polo di Nanotecnologia, Lecce, Italy, <sup>3</sup> School of Environment and Sciences, Griffith University, Gold Coast, QLD, Australia, <sup>4</sup> Belozersky Institute of Physico-Chemical Biology, Lomonosov Moscow State University, Moscow, Russia*

The development of nanotheranostics represents one of the most dynamic technological frontiers in the treatment of different pathological conditions. With the goal in mind to generate nanocarriers with both therapeutic and diagnostic properties, current research aims at implementing these technologies with multiple functions, including targeting, multimodal imaging, and synergistic therapies. The working mechanism of some nanotheranostics relies on physical, chemical, and biological triggers allowing for the activation of the therapeutic and/or the diagnostic properties only at the diseased site. In this review, we explored new advances in the development of smart nanotheranostics responsive to pathological stimuli, including altered pH, oxidative stress, enzymatic expression, and reactive biological molecules with a deep focus on the material used in the field to generate the particles in the context of the analyzed disease.

Keywords: nanotheranostics, smart nanoparticles, pH-responsive theranostics, ROS-responsive theranostics, Enzyme-responsive theranostics

#### INTRODUCTION

Nanotheranostic development perhaps represents the highest level of technological advance in the nanomedicine field, and it aspires to combine in the same delivery platform therapeutic and diagnostic properties (Wong et al., 2020). Nanotheranostics are nanoparticles designed to provide real-time information about drug biodistribution, release, and targeted treatment in vivo, representing one of the last frontiers in personalized medicine (Jo et al., 2016). Nanotheranostics are usually generated through complex synthetic protocols (Silva et al., 2019) necessary to transfer multiple functions to the same delivery platform. For this reason, in several cases, nanotheranostics' targeting simply relies on particle passive accumulation in the diseased tissue via enhanced permeability and retention effect (EPR) usually achieved with biological coating (i.e., albumin, peptides) or polyethylene glycol (PEG) surface modification.

However, beyond exploiting a specific targeting, nanotheranostic selectivity for the pathological area can derive from a specific "responsiveness" of the carriers to an external stimulus (Sneider et al., 2017). This "activation trigger" (e.g., near-infrared light) is usually remotely applied directly in the area of interest (Wang et al., 2017b) and has benefits of non-invasiveness. Typical diagnostic or therapeutic carriers are generally in the "on" state (Zhao et al., 2017), and their detection and payload release occur from the moment of their administration. On the other hand, the designing of carriers with "off-on" theranostic properties (Yu et al., 2018) can favor a personalized assessment of the amount of drug that effectively reaches the pathological site. For this reason, responsive nanotheranostics have the potentialities to open new avenues for optimizing and controlling the

#### Edited by:

*Gabriele Candiani, Politecnico di Milano, Italy*

#### Reviewed by:

*Enzo Terreno, University of Turin, Italy Melani Solomon, University of Maryland, United States Jing Lin, Shenzhen University, China*

#### \*Correspondence:

*Alessandro Parodi aparodi.sechenovuniversity@ gmail.com Andrey A. Zamyatnin Jr. zamyat@belozersky.msu.ru*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

> Received: *04 March 2020* Accepted: *29 April 2020* Published: *25 May 2020*

#### Citation:

*Parodi A, Rudzinska M, Leporatti S, Anissimov Y and Zamyatnin AA Jr (2020) Smart Nanotheranostics Responsive to Pathological Stimuli. Front. Bioeng. Biotechnol. 8:503. doi: 10.3389/fbioe.2020.00503* treatment dose and repetition (Fan et al., 2016; Yang et al., 2016b). In this scenario, it seems almost impossible to increase further the level of complexity of these technologies. To date, many theranostics can permit multimodal imaging and therapy for enhancing diagnostic accuracy and treatment efficacy (Dong et al., 2016). However, new trends in nanomedicine aim at imparting the carriers with responsiveness to the biological environment conditions. In other words, the physiological alterations that differentiate diseased from healthy tissue can serve as "triggers" to turn in vivo the nanotheranostics "on" (Ma et al., 2016). Also, these concepts showed the potential to improve current experimental protocols. For example, nanotheranostics that can enhance their detection signal in response to precise biological stimuli could mitigate the background noise issues related to the imaging of fluorescently modified carriers (Wu et al., 2014).

The working mechanism of the stimuli-responsive carriers, often referred to as "smart" theranostics, usually depends on functional molecules or chemical linkers used to assemble the nanoparticles (Karimi et al., 2016). However, in some cases, it is the intimate structure of the particles that changes or responds to the environmental conditions, like in the case of polymers (Li et al., 2019a). In this review, we explored recent advances in the development of nanotheranostics that can respond to biological stimuli. These carriers were designed to provide their curative and/or diagnostic properties only in the pathological site, exploiting altered chemical and biological features of the diseased tissue. These concepts, as well as the field of nanomedicine, are traditionally applied to cancer disease because tumor tissue undergoes profound changes in cell metabolism, generating significant variations in local pH and oxidative stress. Sometimes the nanoparticles were designed to respond to more than one biological stimuli as well as in combination with physical stimuli that can be remotely administered. Also, tumor growth very often depends on the differential expression of enzymes that can be exploited as biological triggers as well (Xiao et al., 2018). However, recent evidence demonstrated that the concepts of biologically responsive nanotheranostic could be beneficial also for other pathological conditions, expanding the application of these technologies to a new whole portfolio of clinical conditions.

#### PH-RESPONSIVE THERANOSTICS

In the area of cancer therapy, pH represents an essential cue of differentiation compared to healthy tissue. Even though not all the regions of the tumor become significantly acidified, changes in cancer metabolism [i.e., the Warburg effect (Tekade and Sun, 2017; Shamsi et al., 2018)], can induce an overall average decrease of 0.2–0.4 points, although values <6 were registered as well (Liu et al., 2014a). In this scenario, the tumor microenvironment pH can eventually represent a targetable characteristic. Also, after cell internalization, the nanoparticles are usually sequestered in the endolysosomal compartment, where the pH drops significantly below 5 (Wang et al., 2017a), making these organelles optimal targets for pH-responsive technologies.

pH-responsive properties can be coupled with the activation of different therapeutic mechanisms, including reactive oxygen species (ROS)-based cytostatic therapeutics, as well as different imaging modalities. In this context, sono-, photo-, and chemodynamic therapies are treatments in which the cytostatic properties rely on the overproduction of ROS. In sono- and photodynamic therapy (PDT), ROS can be generated upon an external stimulus. In contrast, chemodynamic therapy (CDT) depends just on the chemical properties of the carriers or the payload responsible for ROS generation, usually when interacting with the cellular H2O<sup>2</sup> (Lin et al., 2018).

The pH-responsiveness can derive from pH-sensitive linkers used to stabilize the particles (Kanamala et al., 2016) or its payload in the carrier structure. An extensive overview of this topic can be found in Cao et al. (2019). Acotinyl linkers were intensely investigated in the field to impart pH responsiveness to the carriers. For example, Zhu et al. generated pH-responsive theranostics incorporating gold nanoparticles in poly(amidoamine) dendrimers (Zhu et al., 2018). The gold nanoparticles represented an optimal contrast agent in computed tomography (CT) imaging. The dendrimers were modified on their surface with folic acid via EDC chemistry to provide the particles with high tumor targeting. Cis-aconitic anhydride pHsensitive linkers were used to conjugate Doxorubicin (DOX) to the particles. Drug release was triggered by the acidic conditions (typical of the tumor microenvironment and the endosomal compartment) even though in vivo proof of the efficacy of this technology was not reported. Similarly, nanotheranostics were generated with bovine serum albumin (BSA) linked to the porphyrin photosensitizer pheophorbide-a via cis-aconityl pH-sensitive linkers. Stable nanoparticles were generated by complexing this structure with graphene oxide via π-π stacking and hydrophobic interactions (Battogtokh and Ko, 2016). The carriers were further modified with pegylated folate to impart the system with extended circulation properties and tumor targeting. Porphyrins are known for their therapeutic potential (PDT) and fluorescent properties, but their administration requires encapsulation since they are strongly hydrophobic (Yan et al., 2015). This complex facilitated dual therapy through PDT and photothermal therapy (PTT) due to the porphyrin payload and the graphene oxide, respectively, upon irradiation at 670 nm. In vivo, compared with the free administered photosensitizer, the particle highly accumulated in the tumor tissue through EPR effect. Here they could release their payload due to the pHsensitive linkers both in the acidic tumor microenvironment and well as in the cell cytoplasm after internalization favored by the folate functionalization. The pH-sensitive mechanism was fundamental to activate the theranostic properties of the porphyrin since this molecule is affected by aggregation-caused quenching (ACQ) effect when encapsulated.

On the other hand, different materials can be manipulated in the nanoscale and dissolve (releasing a payload) in acidic pH. Calcium carbonate (CaCO3) nanoparticles represent a wellinvestigated delivery platform (Idris et al., 2019; Vidallon et al., 2020) in this field. At physiological pH, CaCO<sup>3</sup> nanoparticles are stable, but under acid conditions, they degrade into Ca2<sup>+</sup> and CO2, releasing whatever payload was previously loaded. Pegylated CaCO<sup>3</sup> nanoparticles were synthesized by gas diffusion-reaction in combination with the photodynamic theranostic agent chlorin e6 (Ce6) (Dong et al., 2016). Like porphyrins, Ce6 is per se a theranostic because it can serve as a therapeutic tool for PDT while emitting a detectable fluorescent signal upon near-infrared light (NIR) irradiation (Liu et al., 2014b). The particles were also doped with Mn2+, to favor the photosensitizer precipitation with the CaCO3. Mn2<sup>+</sup> also allowed for detecting particle degradation and payload release via magnetic resonance imaging (MRI), and it provided a mean for CDT increasing ROS levels via Fentonlike reaction with tissue H2O2. In this chemical reaction a transition metal (i.e., Fe, Al, Mn, Cu, Zn) interacting with H2O<sup>2</sup> is oxidized while catalyzing the formation of a hydroxyl radical and a hydroxide ion (Das et al., 2015). The system was characterized by a mesoporous structure compatible with the loading of conventional chemotherapeutics like DOX that, in acidic conditions, was released as well. The particles were also modified with PEG allowing for extravasation in the tumor microenvironment via EPR. Under the acidic conditions of cancer tissue, they could exert their theranostic properties demonstrating high cytostatic properties both in vitro and in vivo against a model of breast cancer, The authors demonstrated that chemotherapy (DOX), CDT (Mn2+) and PDT (Ce6) could work synergistically against cancer cell growth (Dong et al., 2016) accomplishing multimodal therapy.

Another example of pH-responsive structure was proposed by Xiao et al. (2019) that developed a theranostic platform named MCDION-Se able to provide at the same time chemodynamic and limotherapy while representing an optimal contrast agent for MRI application. The system was composed of a core of manganese carbonate-deposited iron oxide nanoparticles coated with negatively charged selenium nanoparticles coordinated to the core via polyethylenimine (PEI) (**Figure 1**). The manganese carbonate in the system could easily dissolve in slightly acidic conditions releasing Mn2<sup>+</sup> for CDT and MRI while inhibiting ATP generation. On the other hand, the presence of the selenium enhanced the formation of H2O2, fueling the Fenton-like reaction catalyzed by the iron oxide nanoparticles to form ROS that further inhibited ATP synthesis. The combination of these elements was designed to breakdown in the tumor microenvironment and tumor cells after internalization, triggering different cascades of events affecting the energetic metabolism (limotherapy) while accelerating cell apoptosis via CDT. In vivo, the particles showed a higher circulation time and tumor accumulation than free Mn2+, probably via the EPR effect even though specific modifications to favor this phenomenon were not described. Manganese was also exploited by Liu et al. that designed PEG-coated MnO<sup>2</sup> nanoparticles stabilized with BSA. The particles were loaded with the radiosensitizer hafnium and a prodrug form of cisplatin. The core of MnO<sup>2</sup> could catalyze the conversion of tumor H2O<sup>2</sup> to O<sup>2</sup> to revert hypoxic tumor conditions while the hafnium synergistically increased radiotherapy efficiency. After particle extravasation via EPR, the particles dissolved in acidic conditions offering an optimal contrast agent for MRI. Besides, the prodrug could be internalized by tumor cells and converted to cisplatin via cellular glutathione. The system provided a dual therapy mechanism (chemo and radiotherapy) effective against a model of breast cancer in vivo (Liu et al., 2017).

Finally, it is worth reporting the work of Li et al. (2020a) that exploited the acidic properties of the tumor microenvironment to increase the generation of ROS in hypoxic tissue, where this strategy is usually not achievable due to the lack of oxygen (Chen et al., 2015). The system was composed of biodegradable magnetic mesoporous nanocubes that efficiently induced hyperthermia when exposed to an external high frequency alternating magnetic field. The particles were loaded with Vitamin C to selectively kill cancer cells through the formation of the ascorbate radical and H2O<sup>2</sup> (Yun et al., 2015; Lv et al., 2018). Its release was induced by the material phase-change that turned from the solid to the liquid state when the surrounding temperature was higher than 38◦C. More importantly, in acidic conditions, the metallic nature of these nanoparticles induced the transformation of H2O<sup>2</sup> into O<sup>2</sup> as well as its further conversion to hydroxyl radical via Fenton reaction (Cao et al., 2018) while serving as an optimal T<sup>2</sup> MRI contrast agent. The system showed theranostic properties both in vitro and in vivo, where the particle biodistribution could be tracked via MRI. Upon application of the external magnetic field, the generated hyperthermia induced the release of vitamin C with its consequent tumor-killing properties.

As shown in this section, pH responsiveness can be imparted through different chemical linkers and materials sensitive to pH to trigger the therapeutic and/or the imaging properties. The responsiveness of most of these technologies depends on the sensitivity of their ultrastructure to the pH. More importantly, the pH-responsiveness could be exploited for increasing tumor oxygen levels for improving radiotherapy effectiveness or CDT in hypoxic tumor conditions. Finally, pH-sensitivity can be coupled with other mechanisms of responsiveness that rely on external stimuli to increase the carrier therapeutic and diagnostic properties.

#### ROS-RESPONSIVE THERANOSTICS

Due to their high metabolism and accelerated growth, cancer cells are characterized by an increased generation of ROS (Trachootham et al., 2009). On the other hand, this enhanced oxidative stress is compensated by a higher average content of glutathione (Desideri et al., 2019) that in the field of nanotheranostics is exploited as a biological trigger as well, and will be discussed in the last section of this review. It is worth mentioning that all ROS-responsive mechanisms eventually depend on a biological trigger because their working mechanism depends on the oxygen content in the tissue, and they are ineffective in hypoxic regions (Chen et al., 2015). In some cases, the ROS generation is catalyzed directly by the nanoparticles, that due to the properties of their synthesis material, can activate other features of the carriers like payload release (Sun et al., 2018).

In the field of PDT, new evidence are indicating that the best treatment efficacy occurs when the photosensitizer is coupled with a chemotherapeutic agent generating a synergistic effect.

positively charged polymer and the negatively charged Se. Polyvinylpyrrolidone (PVP) was used to stabilize the system. The scheme also illustrates the synergistic action of Mn and Se in inducing the formation of ROS and consequent cell apoptosis. In particular, the selenium nanoparticles induced the formation of superoxide radicals and the activation of the enzyme superoxide dismutase (SOD) to generate H2O2. Manganese and iron oxide nanoparticles catalyzed further conversion of H2O<sup>2</sup> to hydroxyl radical. Mn and Se also negatively impacted on ATP synthesis. (B) After IV administration, the particles extravasated in the tumor microenvironment *via* EPR. Here they can be internalized by cancer cells and induce a cascade of reactions that increase cell apoptosis *via* ROS production and inhibition of ATP synthesis. Reproduced with permission from Xiao et al. (2019).

To achieve this goal, it is mandatory designing particles in which the ROS can induce a burst release of the therapeutic (Yang et al., 2016a; Zhou et al., 2017). For this reason, the carriers needed to generate ROS as well as being sensitive to these chemical species. To this goal, Sun et al. (2018) developed nanoparticles of pegylated polyphosphate crosslinked with a thioketal linker via (A2 + B3) type polycondensation. The carriers were loaded with the photosensitizer Ce6 and DOX. Under NIR light irradiation, the photosensitizer induced the generation of ROS, while favoring the degradation of the thioketal linker resulting in a burst release of DOX. This strategy was effective in overcoming DOX drug resistance in vitro in a cell line of breast cancer (MCF-7/ADR) overexpressing Pglycoprotein. In vivo, the particle demonstrated good tumor accumulation via the EPR effect as well as high tumor-killing properties. The system showed bimodal imaging capabilities achieved both in vitro and in vivo, exploiting Ce6 photoacoustic (PA) properties and gadolinium (loaded in the particles via its natural affinity for Ce6) that allowed for efficient MRI. Supported by dual-modal imaging, the tumor sites could be precisely irradiated, sparing healthy organs and reducing kidney and liver toxicity.

Within the cell cytoplasm, mitochondria are the most active organelles in terms of ROS generation (Dunn et al., 2015) and, compared to healthy cells, cancer cells showed a higher mitochondrial membrane potential favoring their targeting via internalized cationic molecules (Modica-Napolitano and Aprille, 2001). Inspired by this evidence, Yue et al. developed a nanotheranostic composed of triphenylphosphine, a positively charged molecule with a high affinity for the mitochondria, condensed via amphiphilic block polymerization with camptothecin (CPT) and the photosensitizer zinc phthalocyanine (ZnPC) for PDT generation. The drug conjugation occurred through pegylated thioketal linkers sensitive to ROS (Yue et al., 2016). The particles were designed to extravasate in the tumor microenvironment via EPR, while tumor cell internalization and mitochondrial targeting were achieved through the membrane penetrating properties of the positively charged triphenylphosphine. The ROS generated by the mitochondria and via PDT favored the release of the CPT (topoisomeraseinhibitor) and a dual therapy mode. The system could be detected via fluorescence imaging in vitro and in vivo (when ZnPC was replaced by Ce6) in a subcutaneous model of lung cancer. However, for a more comprehensive description of remotely responsive activated technologies, we suggest the readers look elsewhere (Kim et al., 2013; Zhang et al., 2016) since this review focuses mostly on the nanoplatforms that are activated by tissueor cell-generated ROS.

In this context, the probe IR790s was recently used to generate a platform named perylene diimide –IR790s–Fe/Pt NPs to detect and trace ROS generation through ratiometric photoacoustic (PA) imaging. Perylene diimide showed a strong NIR light absorption at 680 nm, while IR790 absorbed light at 790 nm. The structure of IR790 could be cleaved by ROS, with a consequent decrease of its adsorption at 790 nm. Ratiometric PA imaging was achieved by irradiating the carriers at 680 and 790 nm. The increase in the Ab680/Ab<sup>790</sup> value was directly proportional to the ROS concentration. ROS generation was facilited by cisplatin conjugated on the surface of the particles via PEG and ferric anions chelated by the perylene diimide. The PEG modification also increased particle biocompatibility and permitted EPR extravasation in vivo. After cell internalization, the cell reductive environment (i.e., glutathione) favored the cisplatin release. Besides its cytostatic effect, it activated the nicotinamide adenine dinucleotide phosphate oxidase (NOX) enzyme that transformed the molecular oxygen in O.<sup>−</sup> furtherly transformed in H2O<sup>2</sup> via superoxide dismutase. The hydrogen peroxide could be further transformed in hydroxyl radical through a reaction catalyzed by the ferric ions inducing effective CDT (Yang et al., 2018b) (**Figure 2**). The synergistic effect of chemotherapy and CDT was confirmed by an improved in vitro and in vivo tumor cells killing. Qiao et al. (2018) developed a nanotheranostic platform to enhance the temozolomide effect in glioblastoma by reverting the tumor microenvironment immunosuppressive properties through the inhibition of TGF-β expression. The carriers were designed to (1) overcome the blood-brain barrier; (2) target to glioma cells; (3) escape from the endolysosomal compartment; (4) co-deliver temozolomide and the siRNA against TGF-β; (5) be tracked via MRI. The system consisted of superparamagnetic iron nanocubes (an excellent MRI contrast agent) encapsulated in poly[(2-acryloyl)ethyl(p-boronic acid benzyl)diethylammonium bromide] polymer loaded with the TGF-β siRNA and coated with zwitterionic lipids coordinating the drug. The particles were additionally modified on their surface with the peptide angiopep-2 targeting low-density lipoprotein receptor-related protein to favor particle translocation through the blood-brain barrier and cancer cell internalization. The system could escape from the endolysosomal compartment through the zwitterionic coating that, in acidic conditions, could acquire a net positive charge destabilizing the membrane of these organelles. Upon interaction with cellular ROS, the benzylboronic acid oxidized reversing the charge of the carriers from positive to negative. This phenomenon induced the release of the therapeutic payloads, and the iron nanoparticles exploited as a contrast agent for MRI.

The application of ROS-responsive theranostics was applied to different conditions. To decrease the generation of H2O<sup>2</sup> in peripheral artery disease that negatively affects the neoangiogenesis process in this condition, Jung et al. generated a new concept of nanotheranostic to detect the diseased tissue through ultrasound and PA imaging (Jung et al., 2019). The system consisted of boronated maltodextrin that, in the presence of tissue H2O2, released 4-hydroxybenzyl alcohol with proven antioxidant and anti-inflammatory properties (Luo et al., 2017; Tan et al., 2018). Also, upon ROS interaction with the boronate, the system generated CO<sup>2</sup> bubbles with echogenic properties for ultrasound imaging. The nanoparticles were loaded with indocyanine green (ICG), allowing for multimodal fluorescent and PA imaging. The theranostic properties of these particles were successfully tested in vitro and in vivo in a model of hindlimb ischemia via intramuscular administration. H2O<sup>2</sup> is considered a mild but very common ROS in hepatic ischemia/reperfusion injury, which is a potentially fatal condition for many conditions, including liver transplantation, liver surgical resection, and hemorrhagic shock (Ushitora et al., 2010). Kang et al. (2016) designed polymeric nanotheranostic based on poly(vanillin oxalate) that can serve as a scavenger for H2O<sup>2</sup> while showing anti-inflammatory and antiapoptotic properties. These particles incorporated a prodrug form of vanillin, a compound known for its anti-inflammatory effect, but not extensively investigated in the clinic due to its short half-life. The molecule was loaded in the particles through H2O2-sensitive peroxalate ester linkers. More importantly, once activated by H2O2, the peroxalate esters decomposed in CO<sup>2</sup> bubbles that could be tracked via ultrasound imaging in a model of murine hepatic ischemia/reperfusion injury. In this case, the particles were not targeted to exploit their natural tropism toward the liver.

The development of ROS-responsive molecules showed promising results in preclinical testing. Despite their theranostic properties, they can also serve as in vivo nanosensors to measure ROS generation. The responsiveness of some material to ROS can be exploited to generate CO<sup>2</sup> for improving current ultrasound diagnostic methods. Besides, these technologies can be applied to different diseases, since the generation of ROS is a characteristic of various pathological conditions.

FIGURE 2 | (A) Self-assembly of perylene diimide cisplatin prodrug and the infrared dye IR790 in the presence of ferric ions. One of the amide groups of the PDI was modified with polyphenols coordinating the ferric ions necessary for catalyzing H2O<sup>2</sup> in hydroxyl radical in acidic conditions. The second amide of the PDI was conjugated with PEG that increased nanoparticle solubility and allowed for further modification with the cisplatin prodrug. (B) The working mechanism of the system. After cancer cell internalization, the cisplatin induced the activation of the nicotinamide adenine dinucleotide phosphate oxidase (NOX) transforming molecular oxygen in O.<sup>−</sup> with consequent generation of H2O<sup>2</sup> *via* superoxide dismutase. The hydrogen peroxide is further transformed into hydroxyl radicals by the ferric ions inducing cell apoptosis. ROS formation degraded IR790. The measurement of the perylene diimide/IR790 absorption ratio could be used for ratiometric PA imaging of the ROS formation. Reproduced with permission from Yang et al. (2018b).

### ENZYME RESPONSIVE THERANOSTICS

Tissue remodeling and the overexpression of the lytic enzymes that govern this process characterize many pathological conditions. These enzymes can be exploited as triggers since they can favor the degradation of the carriers and the consequent release of the theranostic payloads in the diseased area.

In the case of enzyme-responsive theranostics, are not rare examples of nanoparticles with multiple responsive properties. For instance, Chen et al. developed ferritin nanocages sensitive to pH and matrix metalloproteinase (MMP)-13 activity. The particles, named CMFn@HCQ, were conceived to deliver hydroxychloroquine in the cartilage tissue to ameliorate osteoarthritis conditions (Chen et al., 2019b). In osteoarthritis, the joint microenvironment is characterized by an acidic pH (close to 6) (Li et al., 2017) and MMP-13 overexpression. Ferritin was genetically modified to increase its targeting for collagen II through the addition of a specific peptide in its structure. These carriers were further modified with an MMP-13 cleavable peptide conjugated with the near-infrared (NIR) dye cy5.5 and a quencher to provide the system with enzyme-sensitive diagnostic properties. Because of this chemical modification, only after MMP-13 activity separating the dye and the quencher, the diagnostic signal could be registered. The particles were loaded with the anti-inflammatory hydroxychloroquine, and their structure degraded under acidic conditions inducing the release of the payload. The resulting nanocages had a size of 20 nm, favoring their diffusion in the dense protein matrix of the joint. These carriers demonstrated high therapeutic properties as well as the ability to detect MMP-13 overexpression in vitro and in vivo when administered via intra-articular injections. Ren et al. (2019) engineered a carrier named HA-Ce6 DOX composed of a hyaluronic acid ultrastructure that, despite the ability to offer a natural targeting for the CD44 receptor, overexpressed on cancer cells, also granted the presence of multiple modification sites. The system was conjugated with Ce6 (for PDT and fluorescent detection) and DOX through a pH-sensitive hydrazine bond (**Figure 3**). The release of the photosensitizer and the chemotherapeutic was further accelerated by the enzyme hyaluronidase, overexpressed in the tumor microenvironment. The enzymatic degradation of the particles was fundamental for the activation of the diagnostic properties of the system since, when encapsulated, Ce6 was affected by the aggregation-caused quenching (ACQ) effect (Li et al., 2020b). CPT-loaded mesoporous silica nanoparticles were functionalized with a cyclo-RGD peptide, and another peptide conjugated with a fluorescent dye and a quencher. Both the peptides were sensitive to the proteolytic action of MMP2. Cyclo-RGD had the function of improving cancer cell targeting and stabilizing the drug in the carriers' pores. When the particles were internalized in the cells, both the peptides were digested inducing the quencher and the dye separation with consequent fluorescent signal detection and drug release (Hu et al., 2016). More investigation will be necessary to test the efficacy of this theranostic platform in vivo. Gold nanorods were used to generate a dual stimuli theranostic nanocarrier to provide efficient PTT as well as a detectable diagnostic signal in response to tumor pH and MMPs (Zhao et al., 2017). The gold nanorods were modified with an asymmetric cyanine via an MMP sensible linker. This dye could emit near-infrared fluorescence in a pH-dependent manner (Zhao et al., 2015), presenting a reversible "off/on" signal emission as a function of this parameter. Despite its sensitivity for the pH, the gold nanorods represented a significant FÖrster resonance energy transfer (FRET) quencher for the asymmetric cyanine that could emit the fluorescent signal only when in free form. Both the gold nanorods and the asymmetric cyanine allowed for PTT upon irradiation at 808 nm. Finally, drug delivery was achieved through an additional functionalization with glycosyl groups that significantly increased particle biocompatibility, EPR effect, and tumor targeting via GLUT-1 receptor. Liu et al. (2016b) designed a nanotheranostic system based on fluorescent quantum dots embedded in a nanoporous silica matrix with pH-activatable targeting properties and a protease-sensitive drug delivery mechanism. The system was surface modified with a zwitterionic anti-biofouling layer composed of chemical groups with positive and negative charges [COO <sup>−</sup> and –HN <sup>+</sup> (Me), respectively]. In physiological pH conditions, the particles showed high circulation time, reduced sequestration in the organ of the mononuclear phagocytic system, and reduced protein corona formation due to their neutral surface charge. To this purpose, this strategy was previously demonstrated to be more effective and stable than PEG surface modification (Holmlin et al., 2001; Gui et al., 2013). When exposed to pH below 6.8, like in the tumor microenvironment, they acquired a positive charge favoring their uptake into the tumor cells. The pores of the silica could be loaded with a drug (DOX) and coated with the polymer polycaprolactone that was exploited to stabilize the drug in the particle structure. This polymer was sensitive to the enzyme esterase that is overexpressed and secreted by cancer cells, making tumor microenvironment and cells favorable sites for drug release. Also, it is important to highlight that the enzymatic degradation was further favored at acidic pH, where the particles acquired a positive charge and were more accessible to the enzyme and prone to cell internalization.

The enzyme-responsive theranostic working mechanism usually depends on hydrolytic enzymes favoring carrier or peptide linkers degradation. This phenomenon can also be exploited through coatings used to stabilized the therapeutics or the diagnostic molecules in the particles. The efficient application of enzyme-responsive theranostics strictly depends on the over-expression of some enzymes that can characterize different pathological conditions. Their synthesis relies on the engineering of biological substrates, usually functionalized with other molecules (including targeting modifications). Some of these modifications aim at providing higher biocompatibility and biological interaction, opening new avenues of research in the field of bioinspired nanomedicine.

#### THERANOSTICS RESPONSIVE TO OTHER BIOLOGICAL STIMULI

Glutathione (GSH) is a reactive molecule controlling the cellular redox balance. Cancer cells significantly overexpress this molecule, pointing out its potential role as a targetable trigger for the development of smart nanotheranostics (Liu et al., 2016a). To this purpose, theranostic platforms based on GSH reactivity are usually composed of a targeted particle modified with a disulfide bond coordinating a chemotherapeutic and/or a fluorescent probe (Han et al., 2017). A typical example of GSH-responsive theranostic based on a facile synthesis is represented by nanoparticles generated through hydrophilic polymers conjugated with hydrophobic drugs. The amphiphilic monomers can favor nanoparticle self-assembly, leaving room for further modification with a diagnostic probe. Exploiting these principals, Li et al. (2019b) generated multimodal imaging and therapeutic nanotheranostics named DHP. These carriers consisted of a disulfide-bond-linked hydroxyethyl starch conjugated with the chemotherapeutic paclitaxel (PXT). The addition of dioctadecyl-3,3,3,3-tetramethylindotricarbocyanine iodide (DiR) in the synthesis solution resulted in the entrapment of the dye in the hydrophobic core of the particles through one single step of dialysis. DiR allowed for both fluorescent and PA imaging, and when irradiated with at 808 nm, PTT providing synergistic effects with the chemotherapeutic. When packaged within the nanoparticle structure, the dye was affected by ACQ. However, after particle internalization, the cellular GSH induced the release of the theranostic payloads (Li et al., 2019b) (**Figure 4**). These properties were proven in vitro and in vivo

in a model of breast cancer. These particles could efficiently accumulate in the tumor via EPR due to the high circulation properties offered by the hydroxyethyl starch. Similarly, a GSHsensitive platform was synthesized through low molecular weight heparin polymerization that, in the presence of cystamine, polymerized with Ce6 in biocompatible carriers (Yang et al., 2018a). In addition, heparin was previously shown to possess antimetastatic and anti-angiogenic properties (Yang et al., 2015; Mei et al., 2017). The system offered sensitivity to a reductive environment since the cystamine linker contained a disulfide bond in its structure. GSH induced particle degradation and Ce6 release activating the theranostic properties of the dye that, when encapsulated, was affected by the ACQ effect. Finally, the system was also loaded with PXT in combination with alpha-tocopherol succinate (Yang et al., 2018a) to provide therapeutic synergism between chemotherapy and PDT. When injected intravenously, the particles accumulated via EPR in a model of 4T1 breast cancer. After cell internalization, the particles were degraded by the abundant levels of GSH releasing PXT and Ce6 that in free form was easily detectable through NIR imaging and inducible for PDT. Song et al. (2017) engineered iron oxide nanoparticles to be responsive to ATP, which is highly generated in tumor cells when compared to healthy tissue as well as to a decrease in tissue pH. The system was composed of small clusters of superparamagnetic iron nanoparticles and tannic acid that self-assembled in water-soluble aggregates smaller than 100 nm due to hydrophobic interactions. The particles were modified with ICG and DSPE-PEG via non-covalent hydrophobic interactions and hydrogen bonds, respectively. The presence of PEG increased particle circulation and tumor accumulation via EPR. The core of iron oxide permitted strong MRI properties while its high affinity for ATP induced particle disassembly in the presence of this molecule (Yu et al., 2013). On the other hand, the ICG permitted particle fluorescence detection after degradation (since the dye was affected by the ACQ effect) as well as PTT applications when irradiated at 808 nm. Particle degradation was further boosted in acidic pH, perhaps due to the protonation of the hydroxyl groups of the tannic acid that lowered the affinity of this molecule for the iron oxide nanoparticle structure. The disassembly of the system was pivotal to activate its imaging properties as well as its renal clearance. The system was successfully tested in vitro and in vivo in a subcutaneous tumor model for its theranostic and responsiveness properties obtained through facile synthesis and loading protocols.

Biological responsiveness was also used to ameliorate the conditions of diseases different than cancer. In Alzheimer's disease, the aberrant generation of plaques of soluble β-Amyloid proteins in the central nervous system occurs (Benilova et al., 2012). This phenomenon is favored by the accumulation of

microenvironment and be internalized inside the cells. In the cell cytoplasm, the disulfide bond degradation occurred due to the cellular GSH that favored the release of PXT (chemotherapy) and DiR. When encapsulated in the particles, DiR was affected by ACQ, but upon its release, it could exert its therapeutic (photothermal therapy) and diagnostic (photoacoustic and fluorescent imaging) properties. Reproduced with permission from Li et al. (2019b).

metal ions in the tissue like copper, which via ROS generation can support protein aggregation (Atwood et al., 2018). Cui et al. designed upconversion NaYF4:Yb/Er/Tm nanocrystals conjugated with 8-hydroxyquinoline-2-carboxylic (HQC) and DSPE-PEG to provide the carriers with diagnostic, therapeutic and biocompatibility properties, respectively (Cui et al., 2016). The particles were designed to (1) detect Cu2<sup>+</sup> via luminescence resonance energy transfer from the particle to the copper, (2) targeting and imaging of the β-Amyloid- Cu2<sup>+</sup> complexes by registering the upconverted light signal emitted further NIR light irradiation (demonstrated in vitro and ex-vivo) (3) chelate the copper via HCQ. The theranostic properties of this system were successfully proven in a zebrafish model and ex-vivo.

These examples demonstrated that any reactive molecules could be exploited as a tool to generate responsive nanoparticles. GSH, in particular, was extensively investigated in literature for this purpose since cancer cells are characterized by a very high content of this molecule. On the other hand, they are also characterized by the high generation of ROS. For this reason, further advances in the field need to take into consideration the


*AI, anti inflammatory; Ch.T., Chemotherapy; CDT, Chemodynamic therapy; CPT, Camptothecin; DOX, Doxorubicin; EPR, Enhanced permeability and retention effect; FA, Folic acid; Far, Folic acid receptor; FL, Fluorescence; GO, Graphen Oxide; GSH, Glutathione; HA, Hyaluronic acid; HT, Hyperthermal therapy; IV, intravenous; LRP1, low-density lipoprotein receptor related protein 1; MMP, Matrix metallo proteinases; MRI, Magnetic resonance Imaging; NIR, Nearinfrared light; NP, Nanoparticles; PA, Photoacoustic imaging; PDT, Photodynamic tehrapy; Pep, peptide; PTT, Photheral therapy; PXT, Paclitaxel; ROS, Reactive oxygen species; RT, Radiotherapy; Sc, scavenger; TU, Tumor; US, Ultrasound imaging.*

efficient alteration of this balance, both when generating GSH- as well as ROS-responsive carriers.

#### CONCLUSIONS

The fine-tuning of the therapeutic regimens and diagnostic methods represents an emerging need in all the fields of experimental medicine. Current efforts of the scientific community are paving the way to achieve this goal through the generation of materials that are sensitive and can respond to chemical, physical, and biological triggers. Therapeutic and diagnostic properties can reside in the same molecule embedded in the nanocarrier structure that can be activated at the injury site, or they can independently derive from different chemicals loaded into and/or conjugated onto the surface of the nanoparticles. Also, the very same structure of the particles can alternatively provide therapeutic or diagnostic properties. In this work, we presented some examples of pathological stimuli-responsive theranostics with multiple therapeutic and diagnostic properties, sometimes combined with external stimuli-responsive materials (**Table 1**). The goal of this work was to present the current strategies and materials used for imparting the theranostics with biological-responsiveness.

To our knowledge, no nanotheranostic was FDA approved to date (Anselmo and Mitragotri, 2019). However, clinical trials are currently active to evaluate their efficacy and safety. In this effort, most of the theranostic under clinical trials base their working mechanisms on remotely applied stimuli. In particular, iron (Nardecchia et al., 2019) and gold (Pedrosa et al., 2015) nanoparticles are extensively tested in humans for their multimodal imaging modalities, PDT and PTT properties as well as for their relatively easy synthesis and functionalization. External stimuli-responsive carriers activation strongly depends on the penetration and the precision of the external signal, and most importantly, they can be activated only in the areas that current diagnostic tools detect as disease sites. On the other hand, the responsiveness of theranostics to pathological stimuli theoretically depends solely on their ability to target the environmental conditions of the diseased tissue or cell, with no further interventions of external procedures. In this context, they could reveal and be activated in sick areas that were not detected by other imaging tools. Precise targeting, therefore, represents their main limitation. Due to the complex nature of these technologies, targeting is usually imparted through the pegylation of the carriers and depends on their passive accumulation in the diseased tissue through the EPR effect. However, this kind of targeting could

#### REFERENCES


jeopardize their responsiveness. For example, in some cases, the carriers need to be internalized in the cells to perform their theranostic functions while PEG functionalization could affect this process (Moros et al., 2012). The complexity of their synthesis protocols, as well as their low biocompatibility and inherent toxicity, represent other significant limitations in the clinical translation of these technologies (Kunjachan et al., 2015). However, more investigations on the development of theranostic molecules like porphyrins, Ce6, or Mn2<sup>+</sup> could streamline their synthetic routes while combining the advantages of external- and biological-responsive theranostics. Finally, relying on biological stimuli, they could suffer from unspecific activation in other tissues (i.e., the organs of the mononuclear phagocytic system). For this reason, more research to evaluate their adverse effect, metabolization, and excretion is necessary for future developments.

#### AUTHOR CONTRIBUTIONS

AP and YA: conceptualization. AP and MR writing (original draft preparation). SL review and editing. AZ supervision and funding acquisition.

#### FUNDING

This research was supported by the Russian Science Foundation (grant # 16-15-10410) and by the Russian Academic Excellence Project 5-100.


imaging and drug delivery to tumor. Adv. Healthc. Mater. 5, 1401–1407. doi: 10.1002/adhm.201600002


**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 © 2020 Parodi, Rudzinska, Leporatti, Anissimov and Zamyatnin. 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.

# Selective Cytotoxic Activity of Prodigiosin@halloysite Nanoformulation

Ivan Guryanov<sup>1</sup> \*, Ekaterina Naumenko<sup>1</sup> , Farida Akhatova<sup>1</sup> , Giuseppe Lazzara2,3 \*, Giuseppe Cavallaro2,3, Läysän Nigamatzyanova<sup>1</sup> and Rawil Fakhrullin<sup>1</sup> \*

1 Institute of Fundamental Medicine and Biology, Kazan Federal University, Kazan, Russia, <sup>2</sup> Dipartimento di Fisica e Chimica, Università degli Studi di Palermo, Palermo, Italy, <sup>3</sup> Consorzio Interuniversitario Nazionale per la Scienza e Tecnologia dei Materiali, INSTM, Florence, Italy

#### Edited by:

Salvador Pané Vidal, ETH Zürich, Switzerland

#### Reviewed by:

Alessandra Quarta, CNR NANOTEC, Italy Paolo Bigini, Istituto Di Ricerche Farmacologiche Mario Negri, Italy

#### \*Correspondence:

Ivan Guryanov ivan.guryanov@gmail.com Giuseppe Lazzara giuseppe.lazzara@unipa.it Rawil Fakhrullin kazanbio@gmail.com

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology

Received: 27 January 2020 Accepted: 14 April 2020 Published: 26 May 2020

#### Citation:

Guryanov I, Naumenko E, Akhatova F, Lazzara G, Cavallaro G, Nigamatzyanova L and Fakhrullin R (2020) Selective Cytotoxic Activity of Prodigiosin@halloysite Nanoformulation. Front. Bioeng. Biotechnol. 8:424. doi: 10.3389/fbioe.2020.00424 Prodigiosin, a bioactive secondary metabolite produced by Serratia marcescens, is an effective proapoptotic agent against various cancer cell lines, with little or no toxicity toward normal cells. The hydrophobicity of prodigiosin limits its use for medical and biotechnological applications, these limitations, however, can be overcome by using nanoscale drug carriers, resulting in promising formulations for target delivery systems with great potential for anticancer therapy. Here we report on prodigiosinloaded halloysite-based nanoformulation and its effects on viability of malignant and non-malignant cells. We have found that prodigiosin-loaded halloysite nanotubes inhibit human epithelial colorectal adenocarcinoma (Caco-2) and human colon carcinoma (HCT116) cells proliferative activity. After treatment of Caco-2 cells with prodigiosinloaded halloysite nanotubes, we have observed a disorganization of the F-actin structure. Comparison of this effects on malignant (Caco-2, HCT116) and nonmalignant (MSC, HSF) cells suggests the selective cytotoxic and genotoxic activity of prodigiosin-HNTs nanoformulation.

Keywords: cancer, anti-cancer drugs, comet assay, drug delivery, genotoxic effect, halloysite nanotubes, malignant cells, prodigiosin

## INTRODUCTION

Currently, the number of effective chemotherapeutic agents for the therapy of neoplasms is very limited (Patel et al., 2010). Existing drugs often have toxic side effects on adjacent nonmalignant cells and tissues. The problem of safety of the medications used in chemotherapy is of pivotal importance, one of the ways to solve it is the search for non-toxic substances and fabrication of new drug formulations. Natural drugs that can suppress the proliferation of cancer cells and metastasis formation are becoming increasingly popular (Huryn and Wipf, 2014). Prodigiosin, a bioactive secondary metabolite produced by Serratia marcescens and certain other bacteria is of particular interest (Williamson et al., 2006). The antitumor properties of prodigiosin have already been confirmed in previous studies (Montaner and Perez-Tomas, 2003; Perez-Tomas et al., 2003; Francisco et al., 2007; Lins et al., 2015). Prodigiosin can induce apoptosis in hematopoietic, colorectal, gastric cancer cells (Montaner and Perez-Tomas, 2003), human breast carcinoma cell lines (Lu et al., 2012), choriocarcinoma (Zhao et al., 2019) and prostate cancer cell lines (PC3) in vitro and JEG3 and PC3 tumor-bearing nude mice in vivo (Li et al., 2018)

with metastases suppression (Zhang et al., 2005). The cytotoxic activity of prodigiosin against various human cancer cell lines and relatively lower toxicity toward non-malignant cells has been demonstrated previously (Francisco et al., 2007; Stankovic et al., 2014; Zhao et al., 2019). In addition, prodigiosin can be used to replace synthetic colorants in food industry and sunscreen cosmetics (Darshan and Manonmani, 2015).

Hydrophobic nature of prodigiosin is an obvious disadvantage for medical and biotechnology applications. Limited aqueous solubility of prodigiosin result in poor absorption and low bioavailability (Tran et al., 2019), as well as it may disturb regular distribution of prodigiosin in biological fluids. Bioavailability of prodigiosin can be enhanced similarly, as reported previously for anticancer drug doxorubicin (Li et al., 2017) and curcumin (Ni et al., 2019), employing fabrication of nanoscale drug formulations to overcome the limitations caused by the intrinsic hydrophobicity prodigiosin. Recently, a technique of prodigiosin encapsulation was developed and anticancer effect of targeted nanoformulations of prodigiosin was investigated (Zhao et al., 2019).

Targeted delivery and controlled release of antitumor drugs, antibiotics, enzymes, and nucleic acids are currently among of the most significant challenges in biomedicine (Martín del Valle et al., 2009; Tiwari et al., 2012; Yendluri et al., 2017). The pharmacokinetics and pharmacodynamics of a number of drugs require special procedures for their administration. Using nanoscale drug delivery vehicles is one of the most promising approaches for targeted drug delivery systems (Miyazaki and Islam, 2007). Nanocarrier-based drugs allow preventing possible side effects of drugs and to overcome physiological barriers of the body (for example, blood–brain barrier) (De Jong and Borm, 2008). Nanoscale anticancer formulations can be designed using natural substances or derivatives, such as chitosan, dextran, gelatin, alginate, liposomes (De Jong and Borm, 2008), gold (Kohout et al., 2018; Singh et al., 2018) and magnetic iron oxide nanoparticles (Dulinska-Litewka et al., 2019 ´ ; Rozhina et al., 2019), mesoporous silica nanoparticles (Li et al., 2019), carbon nanotubes (Cirillo et al., 2019) and clay nanotubes (Naumenko and Fakhrullin, 2017, 2019; Yendluri et al., 2017). Natural aluminosilicate halloysite, due to its tubular structure and surface chemistry, is a potent platform to fabricate nanocontainers for drug-delivery systems. Halloysite has a hollow tubular structure, with the length of up to 1 µm, external diameter 70 nm and an inner lumen 15 nm (Shchukin et al., 2005). Halloysite nanotubes are widely used for the fabrication of polymeric nanocomposites to enhance their tensile strength and stability (Naumenko et al., 2016; Suner et al., 2019). The tubular structure of halloysite allows the internal cavity to be loaded with various macromolecules including drugs, proteins, and nucleic acids, followed by the release of the loaded compounds in the delivery region (Joussein et al., 2005). Such features as very low toxicity (Lai et al., 2013; Fakhrullina et al., 2015) and directed modification of the surface and internal cavity (Abdullayev et al., 2012; Tarasova et al., 2019; Rozhina et al., 2020) make halloysite nanotubes promising candidates for the fabrication of nanocontainers for theranostic targeted drug delivery (Hu et al., 2017). Halloysite nanotubes can be efficiently filled with hydrophobic drug via physical entrapping in the internal cavity (Naumenko and Fakhrullin, 2017, 2019; Fakhrullina et al., 2019). As a result, halloysite-based drug formulation demonstrate lower drug side effects, render the protection of drug molecules from possible degradation in aggressive conditions (low/high pH, enzymatic activity), increase the aqueous solubility of hydrophobic insoluble drugs, accumulate in pathological sites in the body, and help controlling drug release rates (De Jong and Borm, 2008; Naumenko and Fakhrullin, 2017, 2019).

In this paper we report for the first time fabrication of prodigiosin-based nanoformulation (p-HNTs) and its effects on viability of malignant and non-malignant cells.

#### MATERIALS AND METHODS

#### Prodigiosin-HNTs Fabrication and Characteristics

The red pigment prodigiosin was obtained by cultivation of the producer strain S. marcescens ATCC 9986 on agarized peptone–glycerol medium. Pigment purification was performed as described previously (Guryanov et al., 2013). Ethanol solution (96% vol. 300 µl of purified prodigiosin (4.4 µg) was mixed with glycerol (70 µL), dry HNTs (30 mg) in centrifuge tube and placed into desiccator for loading by vacuum displacement (**Supplementary Figure S1**). Prodigiosin loading procedure was performed for 24 h. Subsequently, the loading efficiency was evaluated by thermogravimetric analysis (TGA) while Fourier transform infrared spectroscopy (FT-IR) highlighted the interaction mechanism and involved functional groups. Optical absorption spectra of purified prodigiosin in ethanol and extracts of glycerol-HNTs and prodigiosin-HNTs after 30 min and 2 h exposure in PBS were obtained and compared for estimation of pigment release from loaded halloysite nanotubes. Absorption spectra were analyzed using a Lambda 35 spectrometer (PerkinElmer).

### Dark-Field Imaging and Hyperspectral Microscopy

Dark-field images and reflected light spectra were obtained using an Olympus BX51 (Olympus) upright microscope equipped with a CytoViva <sup>R</sup> enhanced dark-field condenser with a halogen light source (150 W) Fibre-Lite DC-950 (Dolan-Jener) and control module ProScan III (JH Technologies). Images were obtained using acquisition software for visualization Exponent 7 (Dage-MTI). Spectra were registered using a Specim V10E spectrometer and CCD camera in the range between 400 and 1000 nm with a spectral resolution –2 nm. Hyperspectral data were collected with ENVI software, version 4.8 (Harris Geospatial Solutions) (Akhatova et al., 2018). The presence of prodigiosin inside the treated cells was visualized using a CytoViva dual fluorescence module. Images were processed using ImageJ freeware (NIH). The cells were fixed on coverslips, nuclei of the cells were stained with DAPI. An X-cite 120Q wide-field fluorescence microscope excitation light source (Excelitas Technologies) and CytoViva <sup>R</sup> Dual Mode Fluorescence system equipped with the

visualized using fluorescence module of hyperspectral microscope CytoViva <sup>R</sup>

cell membrane, forming large clusters. (E) prodigiosin exhibits autofluorescence, therefore its leakage from the nanotubes and the presence in the cytoplasm can be

. More details see in Supplementary Figure S4.

Triple Pass Filter were used to image DAPI nuclear staining with transmitted fluorescence illumination imaging, exposure time was 100 µs. Fluorescence nuclei images were rendered with artificial red color to enhance local contrast using GIMP software, version 2.10.8. The resulting dark-field images were merged with transmission fluorescence images using the freely available image processing GIMP.

#### Fourier Transform Infrared Spectroscopy

Fourier transform infrared spectra were registered using a Frontier FTIR spectrometer (PerkinElmer). The measurements were conducted on KBr pellets at room temperature in the range between 500 and 4000 cm−<sup>1</sup> with a spectral resolution of 2 cm−<sup>1</sup> .

#### Thermogravimetry

Thermogravimetry (TG) experiments were carried out by means of a Q5000 IR apparatus (TA Instruments) under the nitrogen flows of 25 cm<sup>3</sup> min−<sup>1</sup> for the sample and 10 cm<sup>3</sup> min−<sup>1</sup> for the balance. The mass of each sample was ca. 5 mg. TG measurements were conducted between 25 and 600◦C using a constant heating rate of 20◦C min−<sup>1</sup> . The temperature calibration was carried out by means of the Curie temperatures of standards (nickel, cobalt, and their alloys) (Blanco et al., 2014). The encapsulation efficiency into HNTs was determined by considering the rule of mixtures for the residual mass at 600◦C. Details are provided in literature (Lisuzzo et al., 2019).

### Atomic Force Microscopy

Atomic force microscopy (AFM) images of HNTs and cells were made using a Dimension Icon microscope (Bruker) operating in PeakForce Tapping mode. ScanAsyst-air (Bruker) probes were used to obtain images (nominal length 115 µm, tip with a radius of 2 nm, spring stiffness 0.4 N m−<sup>1</sup> ). Images were obtained at 512–1024 scan lines at a scanning speed of 0.8–0.9 Hz. The adhesion of the nanoparticles was analyzed using an atomic force microscope and calculated from 30 × 30 nm sites on the nanoparticles surface. The obtained data were processed using Nanoscope Analysis software version 1.7 (Bruker).

#### Cell Culture

A frozen stock of HCT116, Caco-2, rat adipose-derived MSCs and human skin fibroblasts (HSF) was grown for 7 days in α-MEM (Sigma-Aldrich) supplemented with 10% of fetal bovine serum (Thermo Fisher Scientific), 100 IU/ml penicillin, 100 µg/ml streptomycin and 2 mM L-glutamine on 25 cm<sup>2</sup> tissue culture flasks in humidified atmosphere with 5% CO<sup>2</sup> at 37◦C. After cultivation, cells (2.5 × 10<sup>5</sup> ) were seeded on 24 well cell culture plates (BD Biosciences) and cultured for 48 h in CO<sup>2</sup> incubator at 37◦C. Glycerol-HNTs or prodigiosin-HNTs in PBS were added to each well to final concentration 100 µg per ml and cells were incubated for 48 h.

#### Comet Assay (Alkaline Comet Assay)

Live trypsinized cells were mixed with low-melting agarose (1.5%) and added to fully frosted slides precoated with 1% normal melting point agarose. After solidification, the slides were lysed in TABLE 1 | Thermogravimetric parameters for HNTs, glycerol-HNTs and prodigiosin-HNTs.


buffer and alkaline solutions (lysis buffer pH 10, alkaline solution pH 13, Tris-acetate-EDTA buffer 2 h at 4◦C. After lysis, the slides were immersed in the neutralizing solution (0.4 M Tris, pH 7.5) for 15 min. Then the slides were placed in the alkaline solution (300 mM NaOH, 1 mM EDTA-Na2, pH 13) for 20 min to allow DNA unwinding and subsequently electrophoresed for 30 min at 20 V, 300 mA. Upon completion of the electrophoresis the slides were placed in 70% ethyl alcohol for 5 min at room temperature, for DNA fixation. After drying at room temperature for 1 h the slides were stored in a dry and dark place until further analysis. Finally, the slides were stained with ethidium bromide and then visualized using confocal microscopy (Carl Zeiss LSM 780) equipped with diode laser (405nm), argon laser (488 nm) and He-Ne laser (633nm). Hundred randomly captured nuclei were examined from each slide in two independent experiments. The analysis does not include comet DNA of apoptotic cells detected on microscopic preparations as fluorescent comets with a broad diffuse tail and a very small head, so-called hedgehogs. Data was processed using CometScore software (v. 2.0).

### Cytoskeleton Visualization

F-actin was stained with Alexa Fluor 488 <sup>R</sup> conjugate of phalloidin according to the protocol, provided by Life Technologies, nuclei were stained with DAPI. Samples were visualized using confocal microscopy. Images were processed using ZEN software.

### Live/Dead Staining of Cell Cultures

Viability of cells was tested using Cell Viability Imaging Kit (Blue/Green) (Life Technologies) according to the protocol, provided by producer. Samples were using confocal microscopy. Living cells were visualized as blue, and dead cells as green.

## RESULTS AND DISCUSSION

We have isolated prodigiosin by acidic ethanol extraction from S. marcescens bacteria and purified as described previously (Guryanov et al., 2013). Prodigiosin purity was confirmed using UV–vis spectroscopy (**Supplementary Figure S1**). Next, prodigiosin was loaded (dissolved in ethanol/glycerol solvent) into the lumens of halloysite nanotubes via vacuum-facilitated loading (Dzamukova et al., 2015), as schematically shown in **Supplementary Figure S2**. To determine the stability of prodigiosin loaded halloysite nanotubes we investigated prodigiosin leakage in phosphate buffered saline (PBS) at 37◦C (**Supplementary Figure S1**). Absorption spectra of p-HNTs suspension in PBS confirm the prodigiosin binding to halloysite. Spectrophotometry data demonstrate non-specific prodigiosin

aggregate on the cell surface which cannot be visualized on fluorescent image.

absorption, the increase of absorption value over time in this experiment can be explained by the following dissociation of hydrophilic HNTs in PBS solution. Spectral signatures of p-HNTs obtained in reflected light mode demonstrate the effective loading of prodigiosin due to the presence of characteristic peaks of pure HNTs and prodigiosin in complex spectra (**Supplementary Figure S3**).

The detailed characterization of HNTs results are shown in **Figure 1**. Prodigiosin, as a hydrophobic compound (de Araújo et al., 2010), after the loading into halloysite can located within halloysite lumen and on the surface of the tubes (**Figure 1A**). This, in turn, increases the adhesion of prodigiosin-loaded halloysite nanotubes. AFM images of halloysite nanotubes loaded with glycerol and prodigiosin (**Figure 1A**) confirm the increase of non-specific adhesion activity after prodigiosin loading (10.5 ± 0.7 nN) compared with the control pristine halloysite nanotubes (4.5 ± 0.6 nN). We also observe a slight increase in surface adhesion of nanotubes with glycerol (7.3 ± 0.5 nN). We suggest that an increase in the adhesion of the surface of the p-HNTs may contribute to their aggregation.

The thermal behavior of modified HNTs (glycerol-HNTs and prodigiosin-HNTs) was investigated by thermogravimetric method as described previously (Blanco et al., 2014). As evidenced in **Figure 1B** both glycerol-HNTs and p-HNTs exhibit a mass loss in the temperature interval between 450◦C and 550◦C. These results highlighted that the presence of the interlayer

water molecules of halloysite is preserved in the functionalized HNTs. Moreover, p-HNTs showed an additional degradation step in the temperature range between 180 and 320◦C that can be attributed due to the destabilization of bonds in the prodigiosin molecule (Sumathi et al., 2014). The corresponding mass loss was estimated at 8.73 wt%. According to literature,

the mass loss between 25◦C and 120◦C (ML120) can be ascribed to the moisture content of the investigated material (Cavallaro et al., 2018). **Table 1** evidences that the addition of glycerol generated an enhancement of the water amount physically adsorbed on HNTs, while the opposite effect was detected after the introduction of prodigiosin in the halloysite nanostructure.

(MSC/prodigiosin-HNTs and HSF/prodigiosin-HNTs inserts).

cell types. The presence of prodigiosin-HNTs in the culture medium of non-malignant cells increased the cytoplasm vacuolization of some cells

non-specific staining of the cells cytoplasm (prodigiosin-HNTs column) is the result of autofluorescence of prodigiosin released from prodigiosin-HNTs into the cytoplasm. Changes in nuclear morphology of Caco-2 and HCT 116 cells treated with p-HNTs shown in the far right column.

Further insights on the functionalization of HNTs surfaces were obtained by comparing the residual masses at 600◦C (MR600). We estimated similar MR<sup>600</sup> values for HNTs and glycerol-HNTs. By contrast, the presence of prodigiosin induced a significant MR<sup>600</sup> reduction indicating the successful modification of halloysite.

**Figure 1** compares the FT-IR spectra of modified HNTs (glycerol-HNTs and prodigiosin-HNTs) with those of pristine HNTs and prodigiosin (**Figure 1C**). The wavelength range between 1350 cm−<sup>1</sup> and 1800 cm−<sup>1</sup> can be considered as a fingerprint region for prodigiosin, which presents numerous characteristic signals that were not detected neither for pure HNTs nor for glycerol-HNTs. On the other hand, the typical FT-IR peaks of the drug at 1467 cm−<sup>1</sup> (bending of C–H) and 1548 cm−<sup>1</sup> (aromatic C=C, NO<sup>2</sup> stretch) (Suryawanshi et al., 2014) were observed in the p-HNTs confirming the successful loading. Interestingly, the band at 1734 cm−<sup>1</sup> (C=O stretching vibration) (Arivizhivendhan et al., 2015) of prodigiosin was split in two peaks (1734 and 1717 cm−<sup>1</sup> ) in the p-HNTs. According to the literature (Cavallaro et al., 2012), this result could indicate that the delocalization of the negative charge along the carboxylate group of the drug is no longer present in the loaded HNTs. On this basis, the loading of prodigiosin in HNTs might be partly related to electrostatic interactions between the two components.

The changes of nanotubes adhesiveness after loading of prodigiosin affects their distribution over the cell surface (**Figure 1D**, red arrows). One can clearly distinguish relatively large amorphous aggregates of p-HNTs on cells (**Figure 1D**). HNTs per se distributed more evenly and less aggregate that was demonstrated in **Supplementary Figure S4**. The overall

morphology of both HSF and MSC cells remains unaffected when cells interact with HNTs or p-HNTs, while some morphology changes observed in malignant cells (**Figures 1D**, **3** and **Supplementary Figure S4**). Internalization of p-HNTs and prodigiosin release from nanotubes started immediately after adding of nanotubes and increased in course of incubation time (**Figure 2B**). We observed the accumulation of prodigiosin in the perinuclear space as red fluorescent vesicles (**Figures 2A,B**). Prodigiosin leakage from loaded nanotubes into cytoplasm during cell cultivation may cause disorganization of the nucleus structure and reduce the volume of the cytoplasm in Caco-2 and HCT116 cells (**Figures 1E**, **4**). In Caco-2 cells exposed to p-HNTs extracellular vesicles filled with prodigiosin can be observed (**Figure 1E**, yellow arrows). Noteworthy, we detected small vacuoles in the cytoplasm of p-HNTs treated fibroblasts (**Figure 1E**, green arrows). Vacuolization of HCT116 cells also

Living cells appear blue and dead cells appearing green on the color-coded confocal images.

was demonstrated after incubation of cells with p-HNTs for 3 h (**Figure 2A**). In **Figure 2C** the arrows indicate the absence of extracellular release of prodigiosin that can be seen in simultaneously taken dark-field and fluorescence images (red colored p-HNTs aggregate on the cell surface and cannot be visualized on fluorescence images).

The addition of p-HNTs during the cultivation of cancer cells led to aberration of their morphology and subsequent detachment from the bottom of the dish (**Figure 3**). Glycerolloaded HNTs used as a control did not affect the morphology and viability of all types of cells. The lack of long-term toxicity were observed for different type of test-organisms including yeast cells, Protista and worms (Konnova et al., 2013; Fakhrullina et al., 2015; Kryuchkova et al., 2016). The presence of p-HNTs in the culture medium of nonmalignant cells led to increasing of vacuolization in some

of them (**Figure 3**, inserts). However, fibroblasts and MSCs maintained normal morphology and viability as demonstrated using Live/Dead staining (**Figure 5**) where blue fluorescence indicates the nuclei of all cells while green stain is located exclusively in the nuclei of dead cells with compromised plasma membranes and cytoskeleton visualization (**Figure 6**). We summarized in **Figure 5** the numerical data demonstrating the sensitivity of cells to HNTs and p-HNTs. Caco-2 cells were less resistant to HNTs in culture medium. The interaction with p-HNTs in concentration of 10µg per 10<sup>5</sup> cells resulted to rapid 100% cell death. In opposite, the viability of nonmalignant cells (HSF) did not change in all experimental variants. Considering the fact that prodigiosin can suppress cell proliferation, it is expected that the sensitivity of cancer cells to the inhibitory activity of prodigiosin depends on cell proliferation rates (Liu et al., 2018; Sam and Ghoreishi, 2018; Ji et al., 2019).

Cytoskeleton components are a known target for certain anticancer drugs (Yvon et al., 1999; Lin et al., 2016), which suppress microtubule dynamics, inhibiting, as a result, cell

with characteristic undamaged nucleoids and comets in each variant of experiment (specified in figure). The histogram shows the ratio of the DNA percentage in the

tail and head of the comets. The semi-automated image analysis system Comet Score was used to evaluate 100 comets per sample.

proliferation rates. Interestingly, treatment with prodigiosin demonstrates the same effects (**Figure 6**), apparently due to binding to cytoskeleton components followed by their structure disruption that in turn disrupts cytoskeleton functions. Thus, in this study we have found for the first time the changes in the morphology and viability of cancer cells and absence of such effect in non-malignant cells after exposure to prodigiosin-loaded halloysite nanotubes.

To assess DNA damage caused by p-HNTs we used the Comet Assay or single-cell gel electrophoresis (Lorenzo et al., 2013) (**Figure 7**), which allows detecting DNA degradation in individual cells. Briefly, cells were embedded in agarose gel and distributed on adhesive microscope slide, then cells were lysed, leaving nucleoids (DNA structures without nuclear membrane), and then electrophoresed in alkali conditions. DNA with strand breaks are relaxed and extend toward the anode during the electrophoresis, forming a comet-like tail viewed by fluorescence microscopy with ethidium bromide staining. DNA damage (strand breaks frequency) is related to the percentage of DNA in the tail. Undamaged DNA remains in the place of initial cell localization and represent the head of comet-like structure. Anticancer drug doxorubicin was used as a positive control. Prodigiosin can bind with DNA by intercalation and acts mainly as inhibitors of topoisomerases I and II (Lins et al., 2015). This effect can cause DNA damage, which is directly correlated with the level of cytotoxicity and genotoxicity of p-HNTs in our study. We found that genotoxic effect of p-HNTs studied by Comet Assay was more pronounced in the case of malignant cells (Caco-2, HCT116). The formation of strand breaks of DNA was observed with much less frequency in human skin fibroblasts treated with p-HNTs. HNTs without prodigiosin caused only background level of DNA damage in all cell types, therefore we assume that the possible reason for p-HNTs selective antitumor action might be the destruction of malignant cells DNA.

#### CONCLUSION

In summary, prodigiosin was successfully adsorbed by HNTs surfaces and encapsulated within halloysite lumen. Then, we found that prodigiosin could release in the cytoplasm of cells, while no release occurs extracellularly. In vitro anticancer

#### REFERENCES


effects of p-HNTs were manifested in the suppression of Caco-2 and HCT116 cells proliferation, followed by alteration of cell morphology and F-actin structure disorganization. Comet assay response where fragmented chromatin were observed, indicating a high therapeutic effect of halloysite formulated prodigiosin. Comparison of the effects of p-HNTs on malignant (Caco-2, HCT116) and non-malignant (MSC, HSF) cells allows to conclude that the p-HNTs demonstrate the selective cytotoxic and genotoxic activity. We hypothesize that prodigiosin entrapped into halloysite may have significant advantages for treatment of living tissues in vivo, due to higher bioavailability and extended intracellular release.

### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

#### AUTHOR CONTRIBUTIONS

IG, EN, and RF designed the research. IG and EN performed the cell culture studies. FA performed the AFM experiments. GL and GC performed the thermoanalysis. LN and EN performed the dark-field microscopy experiments. All the authors read and approved the final version of the manuscript.

### FUNDING

The work was performed according to the Russian Government Program of Competitive Growth of Kazan Federal University. This study was funded by the Russian Foundation for Basic Research grant #20-015-00353 À and RFBR #18-29-11031 mk. GL and GC thank the University of Palermo for financial support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2020.00424/full#supplementary-material

(POSSs)/polystyrene (PS) nanocomposites: the influence of the bridge length on the resistance to thermal degradation. Polym. Degrad. Stab. 102, 132–137. doi: 10.1016/j.polymdegradstab.2014.01.029




**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 © 2020 Guryanov, Naumenko, Akhatova, Lazzara, Cavallaro, Nigamatzyanova and Fakhrullin. 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.

# Ultrashort Peptide Self-Assembly: Front-Runners to Transport Drug and Gene Cargos

#### Seema Gupta<sup>1</sup> \*, Indu Singh1,2, Ashwani K. Sharma<sup>2</sup> \* and Pradeep Kumar <sup>2</sup> \*

*<sup>1</sup> Chemistry Department, Acharya Narendra Dev College, University of Delhi, New Delhi, India, <sup>2</sup> Nucleic Acids Research Laboratory, CSIR-Institute of Genomics and Integrative Biology, New Delhi, India*

The translational therapies to promote interaction between cell and signal come with stringent eligibility criteria. The chemically defined, hierarchically organized, and simpler yet blessed with robust intermolecular association, the peptides, are privileged to make the cut-off for sensing the cell-signal for biologics delivery and tissue engineering. The signature service and insoluble network formation of the peptide self-assemblies as hydrogels have drawn a spell of research activity among the scientists all around the globe in the past decades. The therapeutic peptide market players are anticipating promising growth opportunities due to the ample technological advancements in this field. The presence of the other organic moieties, enzyme substrates and well-established protecting groups like Fmoc and Boc etc., bring the best of both worlds. Since the large sequences of peptides severely limit the purification and their isolation, this article reviews the account of last 5 years' efforts on novel approaches for formulation and development of single molecule amino acids, ultra-short peptide self-assemblies (di- and tri- peptides only) and their derivatives as drug/gene carriers and tissue-engineering systems.

#### Keywords: amphiphilicity, peptide, self-assembly, drug delivery, tissue engineering

### INTRODUCTION

The global trend is growing toward precise medicines and diagnoses through multi-centered approaches of drug delivery technology. The poor systemic bioavailability, solubility, absorption, and stability of large sized materials pose major challenges in the area of drug delivery. Novel natural biomaterials which can qualify to be biodegradable, biocompatible, non-toxic, renewable, and readily available to deliver therapeutic agents to precise targeted sites in a controlled manner is one of the most sort after research-field. Presently, nanotechnology (Lombardo et al., 2020) has provided a solution by opening up of newer avenues in terms of developing advanced controlled drug delivery and release systems that have met with huge success (Webber et al., 2016; Webber and Langer, 2017; Patra et al., 2018). The customizable nanoparticles with the manipulation in size, surface characteristics and materials used enhance the efficacy of drug delivery in a paramount manner along with the advantage of safer treatment (Eskandari et al., 2017; Rizvi and Saleh, 2018). A relatively newer area to deliver drugs across biological barriers for improved site-specific absorption is also being explored (Kou et al., 2018) i.e., transporter-targeted nanoparticles. A specific application based self-assembled materials, injectable biomaterials, are also being investigated to improve the advancing practices in healthcare (Sahoo et al., 2018b).

#### Edited by:

*Stefano Leporatti, Italian National Research Council, Italy*

#### Reviewed by:

*Remigiusz Marcin Bachor, University of Wrocław, Poland Hakan Ceylan, Max Planck Institute for Intelligent Systems, Germany*

#### \*Correspondence:

*Seema Gupta seemagupta@andc.du.ac.in Ashwani K. Sharma ashwani@igib.in Pradeep Kumar pkumar@igib.res.in*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

Received: *29 January 2020* Accepted: *29 April 2020* Published: *29 May 2020*

#### Citation:

*Gupta S, Singh I, Sharma AK and Kumar P (2020) Ultrashort Peptide Self-Assembly: Front-Runners to Transport Drug and Gene Cargos. Front. Bioeng. Biotechnol. 8:504. doi: 10.3389/fbioe.2020.00504*

Uncomplicated to design and synthesize, biocompatible, embedded with appropriate opportunities for chemical alterations, demonstrating molecular selectivity and specific interaction with diverse types of biological systems- all these characteristics make peptides ideal and flexible candidates for constructing tuneable nanostructures with normal end functionalization (Sun L. et al., 2016; Yu et al., 2016; Galdiero and Gomes, 2017) (**Figure 1**). It is hypothesized that simple amino acids may be the first catalysts for formation of peptide bonds (Luisi, 2015). The evolutionary model put forward by Carny and Gazit on the mechanism of origin of life links the aptitude of short peptides to the creation of the present living systems (Carny and Gazit, 2005). According to this model, the properties like encapsulation, catalytic potential in chemical reactions and a highly ordered template for the assembly of nucleotides, which might be the early events, could lead to create the biological systems. Self-assembling peptides, a category of peptides, assemble spontaneously into ordered nanostructures. Peptides, even as short as dipeptides, are blessed to hold all the desirable molecular information to form well-ordered structures, when worked for nano-scale. The self-assembly of short peptides turning to β-sheet amyloid conformers brings extraordinary structural stability and multi-functionality like; self-replication, catalytic activities and information transfer, which is impossible

for the corresponding non-aggregated peptides. The emergence and evolvement of such mutualistic networks may eventually lead to origin of life (Maury, 2018).

Peptide self-assemblies are the gait of 20 amino acids which can be manipulated in terms of number, type, sequence, and side chain groups. These nanostructures can be customized by incorporating modified amino acids in the peptide design to have superior assembling properties and enzymatic stability. This bottom-up science, inspired by the wonders of nature operating on the nanoscale, generates many biological nanostructures such as proteins and DNA/RNA, enabling functioning of life. Even the top-down method, with an advantage over the bottom-up approach supports to discover new peptide sequences aimed to a specific binding site on bio-macromolecules based on their structural properties. It is compelling the scientists to reach after the chemistry of these macromolecules and synthesize newer molecular self-assemblies. The peptide synthesis is a simple and affordable synthetic chemistry via conventional procedures, solution or solid phase. The cost of making peptide is found to be associated with the motif length, purity, chirality of amino acids as well as the expertise of the fabricator. The ability of natural and synthetic amino acid building blocks, both, to form ordered assemblies through self-association process with defined architectures and prominent physical properties has

been reviewed by Chakraborty and Gazit (2018). The review by Lee et al., has summarized the building blocks of peptide selfassemblies categorized by their constituting amino acids, the bound chains/motifs, characteristics and regulatory factors (Lee et al., 2019). The study of ordered nanofibres from protein selfassembly in natural systems has been recognized as reductionist approach, wherein, core peptide building blocks could be derived from the parent protein (Guterman et al., 2016).

#### Regulators of Self-Assembly

The fundamental mechanisms for the self-assembled nanostructures of different types and structures of peptides are explored regularly (Mandal et al., 2014; Chen et al., 2016; Liao et al., 2016; Fan et al., 2017; Pandit et al., 2018; Mason and Buell, 2019). The constituent amino acid residues govern the supramolecular nanostructures and peptide secondary structures to adopt the conformations led by intra-/intermolecular interactions during peptide self-assembly. The multifunctional materials with highly ordered structures could be achieved by hierarchical architectures with anisotropy and suggested to be facilitated by rational self-control of weak intermolecular interactions regulating the self-assembly process at different steps (Yuan et al., 2019).

### Inter/Intramolecular Non-Covalent Interactions

The design and nature of these ordered nanostructures is the outcome of the synergistic effect of various intermolecular non-covalent interactions, comprising H-bonding, π-π bonding, electrostatic forces, hydrophobic, and van der Waals' interactions. Molecular self-assembly, a more stable structure, is formed under thermodynamic equilibrium conditions by arrangements through various non-covalent weak interactions which can generate assemblies of excipient patterns (Mateescu et al., 2015). This alliance of non-covalent interactions strategizes the self-assembly process and determines the thermodynamic stability, though, kinetic parameters also play a critical factor in concluding the dynamic material (Wang et al., 2016a; Liu et al., 2017).

The hydrogen bonds have the competence to assist the growth of biomolecules in one direction with a long-range order to manifest one-dimensional (1D) nanostructures. The hydrogen bond stacking in case of cyclic peptides has been reported to provide the primary structure and comprehensive cylindrical morphology to the self-assembly (Rho et al., 2019). Besides the contribution from individual amino acids, the peptide backbone itself also provides significant stability through hydrogen bonds (Leite et al., 2015).

Electrostatic interactions portray remarkable performance in the self-assembly of peptides and stabilize the nanostructures. Electrostatic attraction between the positively charged peptide fibril and negatively charged small molecule drugs for controlled drug delivery (Mauri et al., 2017).

Amino acids based on amino acid residues can be categorized into hydrophobic and hydrophilic ones. Peptides can form highly ordered self-assembled superstructures due to their hydrophobic property. These hydrophobic interactions are observed to progress from open networks of secondary structures toward closed cylindrical nanostructures (β-sheets or random coils) (Fu et al., 2014).

van der Waals forces, contemplating entire intermolecular forces and relatively weaker to covalent bonds, play an integral role in supramolecular organization in nanotechnology. van der Waals interactions in plane are observed to control not only the molecular self-assembly structure but their phase transition as well (Gao et al., 2015). The dipole-dipole interactions also support in the self-assembly process and are reported to enhance the mechanical properties as dual non-covalent bonding strategy (Cao et al., 2019).

#### Ionic Interactions

The arrays of ionic interactions are also part of the driving energy of the self-assemblies in water along with the hydrophobic interactions and peptide-backbone hydrogen bonds. Selfassembling, ionic-complementary peptides are also being studied by researchers (Chen, 2005). The synergistic effect of both non-covalent and ionic interactions has been reported to provide better stability to the studied hydrogel (Xie et al., 2016).

### Aromatic Interactions

π-π stacking between large π-conjugated surfaces provides an overall stability to supramolecular polymers bound together by non-covalent interactions (Cockroft et al., 2005). Various collagen like peptides mimic the fibril formation and assemble into higher order hierarchical structures through π-π stacking interactions (Chen and Zou, 2019). The hypothesis put forward by Gazit et al. regarding the lead role played by aromatic interactions in the self-assembly of peptide nanotubes/amyloidlike structures has been established time and again (Reches and Gazit, 2005). The aromatic-aromatic interactions have been reported to transform into β-sheet conformation from α-helix on being connected to an aromatic motif at C-terminal (Li J. et al., 2017). Though the importance of aromatic interactions in amyloid formation has been challenged (Lakshmanan et al., 2013) but, are significant in amyloid β-peptide (Genji et al., 2017).

In peptide self-assembly, nonpolar amino acids (aromatic and aliphatic amino acids) aggregate through π-π stacking and hydrophobic interactions, while the polar amino acids, depending on whether they have uncharged or charged residues, stabilize through either electrostatic interactions or hydrogen bonding. The weak bond-based injectable hydrogels based on hydrogen bonding, ionic, hydrophobic and π-π stacking interactions and host-guest chemistry have been reviewed by Ding and Wang (2017). These non-covalent interactions act as driving force in designing the gels (Dou and Feng, 2017). In a characteristic assembly, electrostatic repulsions control the nanofiber length. Addition of a covalent bond forming unit, which could conjugate to the peptide sequence by amide bond condensation altered the balance between hydrogen bond formation and compensated the repulsive electrostatic interactions. Thus formation of covalent bond reinforced hydrogen bonds between peptides enabling the fiber elongation, which otherwise energetically is not possible (Sato et al., 2017). The study on reversible covalent chemistry displayed that the difference in self-assembly modes at the non-covalent level could also be reflected at the covalent level (Komáromy et al., 2017).

It is difficult to predict the combo of all these molecular forces after the peptide self-assembly. Extra facts about the interaction of these forces are essential to plan much efficient, chemically stable peptide self-assemblies.

#### Secondary Structural Conformations

Most of the self-assembling peptides are supposed to be readily soluble in water due to the presence of amino acid molecules containing charged residues (alternating hydrophilic and hydrophobic regions), periodically repeated and discrete polar and non-polar surfaces. α-amino acids comprising the peptides have the inclination to adopt various secondary structural conformations like α-helices (Boyle, 2018), β-sheets (Leite et al., 2015), β-hairpins (Nagarkar et al., 2008) and even the folds (Yoo and Lee, 2017; Kulkarni et al., 2019) and this dynamic behavior of the self-assembling peptides at the molecular structural level influences the peptide-based selfassembly processes in water (Gopalan et al., 2015; Bera and Gazit, 2019; Kulkarni et al., 2019). The process continues and these structures assemble further spontaneously to form nanofibers which consequently aggregate into supramolecular scaffolds and can entrap large volumes of water. The hierarchical self-assembly process can potentially stabilize diverse β-sheet hydrogen bonded architectures. Among the secondary structures, hydrogen bonding between carbonyl oxygen and the amino group of every third residue in the helical turn stabilizes an α-helix (each helical turn consisting of 3.6 amino acid residues) while βsheets originate with hydrogen bonding between two or more β-strands (along which, the backbone of the peptide stretches) (Kulkarni et al., 2019). The self-association phenomenon of aromatic side-chains in β-peptide oligomers supports the helical secondary structure formed by intramolecular backbone-side chain CH-π interactions and yields large vesicles due to the gain in the hydrophobic area (Mándity et al., 2014). In a study, Sarkar et al. (2015) have also demonstrated the effect of solvent interactions on the folding pattern resulting in a change in initial helical conformation and structural diversity of short aromatic γpeptides. Another architecture, coiled-coil peptide self-assembly, though suffering from the drawback of having longer amino acid sequences compared to other self-assembling peptide systems such as β-sheet fibrillizing peptides or peptide amphiphiles has been reported to offer unique advantages as reviewed by Wu and Collier (2017). All these functional architectures are the result of molecular recognition process and self-assembly led by non-covalent interactions. Zhou et al. enlightened on the adoption of amino acid conformations in the complex interactive process in relation to peptide sequence. The study evidenced that peptides with high sequence similarity could self-assemble into diverse nanostructures though could acquire similar secondary structures while completely different sequences assembled into one type of nanostructures (Zhou et al., 2019).

Various molecular forces make peptides self-assemble in different supramolecular peptides. As per the observations, a peptide with an electrostatically charged/hydrophilic head and a hydrophobic tail would self-assemble in spherical micelles or vesicles which on elongation could lead the way into fibers or tubes, respectively. Peptides with a β-sheet show inclination to assemble into flat structures suchlike tapes or ribbons. Nonetheless, on increasing the concentration of the peptides, these tapes and ribbons could stack on one another and turn into more firmly packed fibers.

A relatively new innovation, the use of co-assembly, to produce nanostructures is also being explored. When an individual component is incompetent to have the basic properties required for the self-assembly, a co-assembly option provides the necessary support. The cooperativity movement is directed by non-covalent interactions, in particular, electrostatic. The combination of experimental justification and computational simulations provides a basic support to identify structural and functional components (Raymond and Nilsson, 2018).

### Amphiphilic Peptides

It is an aqueous peptide self-assembly, characteristically spurred by the presence of amphiphilic character in the monomer units, contains hydrophilic and hydrophobic domains, which impulsively arrange to shield hydrophobic groups and minimize contact with bulk water. These molecules contain one or more alkyl chain tails along with a terminal peptidic head group. The fine tuning of balance among the hydrophilic block with polar amino acids and hydrophobic blocks could stabilize various supramolecular structures by hydrophobic, electrostatic, β-sheet hydrogen bonds and π-π stacking interactions (Mikhalevich et al., 2017; Qiu et al., 2018). A naïve investigation by Accardo et al. (2013) wherein use of an intrinsically disordered peptide as a polar head connected to alkyl chain led to some disorderto-order transition upon their self-assembly in supramolecular aggregates. The group recommended this kind of ordered core and a "disordered" surface as a potential scaffold for future biomaterials. Furthermore, Tesauro et al. (2019) discussed the versatility of arrangements in side chains, option to load charges/functional groups and select physical and chemical patterns to acquire desired biostructures during the design of peptide amphiphile (PA) and suggested the option of using conformational preferences in structured and/or disordered peptides in the review.

The review by Cui et al. has emphasized on controlling the environment of PA self-assembly manufacturing and details of their applications (Cui et al., 2010). Addition of an ionizable/charged amino acid to the PA structure is suggestive to increase the number of charges per aggregate. A selfassembly model of PAs produced by a C<sup>16</sup> alkyl tail linked to a chain of two lysines (C16K2) or three lysines (C16K3) stand reasonably accurate on the predicted behavior in terms of morphology, size and the state of protonation of the aggregates. The work by Zaldivar et al. revealed that the system followed a charge regulation mechanism and found to decrease electrostatic repulsions between charged lysines (Zaldivar et al., 2019). An atomic level study by Rad-Malekshahi et al. (2015a) analyzed the vesicle surface structure and dynamics of self-assembled nanovesicles along with the intermolecular forces between amphiphilic peptides to improve and tune the biophysical properties of the nanocarrier.

#### Impact of Chemical Modifications

Different chemical reactions are employed to construct selfassembled nanostructures (Rasale and Das, 2015). A general approach followed by peptide chemists is to incorporate modifications in the form of easy and reversible cleavable appropriate moieties/groups at the carbonyl–nitrogen bond (readily cleavable urethanes containing appropriate alkyl groups such as benzyl and tert-butyl liberating the amino groups), incorporating alkyl spacers of various lengths to modulate the chirality (Panda et al., 2019), π-clamping to tune the reactivity (Zhang C. et al., 2016), backbone amide modifications for peptidomimetic hydrogelators with enhanced stability and mechanical properties compared to peptide hydrogelators (Basavalingappa et al., 2019), enforcing a conformational constraint to prevent β-sheet structure (Bowerman and Nilsson, 2010), formulated sequence patterns to optimize charge distribution for conjugating bioactive cargo (Zhang H. et al., 2017), alternating d/l-chirality for 1D- to 2D-selfassembly (Insua and Montenegro, 2019), a racemic mixture of the mirror-image peptides to provide more rigidity to the gel (Nagy-Smith et al., 2017), alternative hydrophobic and hydrophilic residues for catalytic activity (Song et al., 2018), coordination with metal ions to inhibit amyloid-like structure (Ji et al., 2019), fine-tune assembly for hydrogel (Loic, 2017) and so on. N-Acetylation (compound attached to the amino; N-terminus) and C-amidation (compound attached to the carboxyl; C-terminus) is the most usual policy to stabilize nearly all categories of peptides. The peptides, without modification, perhaps may not be toxic, but the impact after the modification, though also unsettled, is being explored (Soleymani-Goloujeh et al., 2018). Amino acids and short peptides bearing moieties like 9-fluorenylmethoxycarbonyl (Fmoc) (Tao et al., 2016), (Chakraborty and Gazit, 2018), aromatic naphthalene-2-methoxycarbonyl (Nmoc) (Rasale et al., 2015), 9-anthracenemethoxycarbonyl (Amoc) (Gavel et al., 2018a), at N-terminal provide extra advantage to the fabrication of self-assemblies due to inherent hydrophobicity and aromaticity and are also, significant in gel formation (Orbach et al., 2012; Fleming et al., 2013; Singh et al., 2015). tert-butyloxycarbonyl (Boc) is also used to protect α-amino group in peptide synthesis (Ragnarsson and Grehn, 2013). Introduction of non-natural d-amino acid at the N-terminus, has shown unexpected effects on peptide secondary conformation and even the biological performance and so is another useful strategy to confer self-assembling properties (Melchionna et al., 2016). Small chemical modifications in peptides at the N- or Cterminus, intrinsic to self-assembly into ordered supramolecular architectures and their biomedical applications, have been reviewed by Rad-Malekshahi et al. (2015b). N- and C-terminal of the aromatic components as well as linker segment and peptide sequence have also been demonstrated to control the self-assembly of aromatic peptide amphiphiles (Fleming and Ulijn, 2014).

Recently, guiding principles to customize the kinetics and morphological changes in supramolecular peptide nanostructures have been introduced by Son et al. (2019) upon exposure of matrix metalloproteinase. These guiding principles enumerate systematic customization of enzyme-responsive peptide nanostructures by exchange of just a few amino acids, for general use in performance optimization of enzymeresponsive materials (Son et al., 2019). The advantages of enzyme-instructed self-assembly (EISA) in triggering the molecular self-assembly in situ, by overexpression, to prepare supramolecular biofunctional materials and hydrogels has been reviewed by Gao et al. (2019).

The focus has also turned to design self-assemblies of peptidebased conjugates. The multidisciplinary studies involving conjugated short peptides and single amino acids highlighted the role of conjugated material and broadened the horizon of self-assembling materials (Acar et al., 2017; Edwards-Gayle and Hamley, 2017). The reformed attempts to enhance in vivo half-life time and widespread applications of peptides are being made by conjugating them with nanoparticles (Jeong et al., 2018; Spicer et al., 2018; Wang et al., 2018; Jiang et al., 2019). A critical comparison of peptide materials with non-peptide materials has been attempted by Santis and Readnov to count on the contribution of peptide self-assemblies in real-life applications i.e., commercial products (De Santis and Ryadnov, 2015). The use of sequence-specific peptides as biological recognition elements has nicely been reviewed by Slocik and Naik (Slocik and Naik, 2017). The insertion of suitable spacers (charged or neutral) between the hydrophobic region and the peptide are reported to uphold flexibility, mobility, and sometimes increase the solubility of the molecule. The linker is supportive for creating functionalized nanofibrils and expand the modules of chemoselective bio-conjugation approaches in site-specific titivation of self-assembling peptides (Biscaglia et al., 2016; Scelsi et al., 2019). Cui and coworkers (Cui and Chen, 2017) worked on a themed issue envisioned to bring leading researchers working on peptides and peptide conjugates to assess the recent progress in utilizing peptide-based constructs and describe the challenges to interface with biology for specific biomedical applications.

The peptide-templated noble metal catalysts also play an important role in chemical biology (Wang W. et al., 2017). Metal coordination to natural and non-natural binding sites of different peptides has been reported to stimulate the peptide self-assembly (Zou et al., 2015). This knowledge of the forces to obtain an ordered organization can assist innovative peptide based materials for more assorted applications.

#### Applications of Peptide Self-Assemblies Peptide Self- Assemblies as Drug Carriers

The rich chemistry of various non-covalent interactions has led to swift development of self-assemblies as drug carriers, particularly, in short peptides (Huang et al., 2013; Panda and Chauhan, 2014; Iglesias and Marchesan, 2017; Amit et al., 2018; Raza et al., 2018; Mishra and Jyoti Panda, 2019).

The impact of finite peptide nanostructures for the development of systemic therapeutic delivery vehicles is, in particular, of interest, as the length of the assembly plays important roles during cell uptake and tissue penetration (Mendes et al., 2013) (**Figure 2**). A strategy has been reported, where the length of charged peptide-amphiphile supramolecular assemblies could be controlled through covalent bond formation (Sato et al., 2017).

The extensive non-covalent interactions provide several advantages in developing self-assembled materials for drugdelivery (Leite et al., 2015). These are mainly at the three working stages, namely, drug loading, self-assembly–drug conjugate transport, and finally the cellular drug delivery (Doane and Burda, 2012; Habibi et al., 2016; Fan et al., 2017). Devadasu et al. have suggested that understanding of the drug and disease is of utmost importance before designing a delivery system (Devadasu et al., 2012). The self-assembled-ordered structures of peptides with extensive π-π and hydrogen bonding, also a precondition for semiconductor properties, has aspired the scientists to assemble such structures for biological semiconductors alongwith biocompatible and drug release materials (Tao et al., 2017). A review by Sis and Webber has discussed about the basic designs in peptide self-assemblies and ways to mend the efficacy of drug delivery (Sis and Webber, 2019). Self-assembled photosensitizers resulting from amphiphilic dipeptide- or amino-acid-tuned for photodynamic therapy (PDT) have been reported (Liu et al., 2016). The tuneable size, surface charge and multi-responsiveness toward pH, detergents, and enzymes suggest the simple and efficient self-assembled method to deliver photosensitizers (Li L.L. et al., 2018).

Amphiphilic peptides have brought a paradigm shift toward self-assembled PA in drug delivery. The amphiphilic peptides provide a lot of option in both linear and cyclic peptide sequences, side chains loaded with charges, self–assembled as vesicles, micelles, nanofibers and nanotubes for delivery systems and for other biotechnological applications (Goel et al., 2015). An account on the versatility of cyclic peptides and the safety measures in terms of size control, length and bundle width of nanotubes during the successful delivery of active pharmaceutical ingredients has been presented in a mini-review by Hsieh and Liaw (2019). The impact of amino-acid side-chains or covalently linked hydrophobic chain in PA on their stimuli-responsive drug delivery applications has been reviewed by Song et al. (2017).

Another strategy is to use the self-assembly of amphiphilic drug molecules to do the drug-loading and then deliver the cargo as well on its own. In case of low water-solubility of the drug, hydrophilic segments may be conjugated to bestow amphiphilic behavior. The conjugation pushes the peptide sequences for one-dimensional elongation through β-sheet formation. Their architecture fundamentals administer the self-assembly of PA's into supramolecular systems to be applied in drug delivery (Lock et al., 2013).

Ample attention is also being directed to cell-penetrating peptides (CPPs), as carriers for intracellular transport cargoes such as siRNA, nucleic acids, proteins, various nano-particulate pharmaceutical carriers (e.g., liposomes, micelles), small molecule therapeutic agents as well as quantum dots and MRI contrast agents (Brasseur and Divita, 2010; Bechara and Sagan, 2013; Choi and David, 2014; Copolovici et al., 2014; Wang et al., 2014; Huang et al., 2015; Skotland et al., 2015; Dinca et al., 2016; Guo Z. et al., 2016; Kurrikoff et al., 2016; Lehto et al., 2016; Guidotti et al., 2017; Hoffmann et al., 2018; Panigrahi et al., 2018; Ramaker et al., 2018; Vánová et al., 2019), since these are internalized by cells in an exceedingly effective manner. The study by Ramaker et al. observed a statistically substantial dependence of CPPs' uptake efficiency on both net charge and peptide length; longer CPP possibly due to more ordered α-helical structure with high charge could ferry conjugated cargo across membranes more efficiently (Ramaker et al., 2018). The non-covalent approach for complexing CPPs to nucleic acids or viruses has shown better gene delivery both in vitro and in vivo (Alhakamy et al., 2013). Gallo et al. have provided a comprehensive list of recognized CPPs along with their reported applications (Gallo et al., 2019). Another review by Borrelli et al. has discussed biological properties of CPP upon conjugation with specific molecules with special emphasis on uses in cancer therapy (Borrelli et al., 2018).

Nanomedicines directly assembled from pharmaceutical ingredients, termed as small molecule nanomedicines (SMNs), have the potential to improve the drug delivery efficiency, biosafety and largely reduce the research and development cost. Xue et al. (2020) highlighted the recent advances in a section on drugs and photosensitizers with peptides and exhibited advantages of SMNs in the review article.

Tailored drug delivery vehicles are continuously gathering attention. Peptide–drug conjugates (PDCs) enable selective delivery of cytotoxic cargoes to target cells (Ma et al., 2017; Wang et al., 2017a; Wang W. et al., 2017; Vrettos et al., 2018). PDCs have exclusive and precise features to build onecomponent nanomedicines (OCNs) containing only one type of chemical substance. These OCNs do not require additional carriers. In fact, these are equipped with desired physicochemical features to involuntarily aggregate as well as accumulate at target sites (Su et al., 2015). Since peptides can be manufactured effortlessly in large quantities and need simple purification, their range of selection of peptide sequences as per the requisite physicochemical properties like stability, solubility, overall charge and availability of the characteristic groups for the conjugation with the therapeutic payload, these are considered as sought-after prodrugs (He et al., 2019).

Peptide-based hydrogels is another class of drug delivery vehicles programmed via drug encapsulation or conjugated covalently with therapeutics. Small peptide molecules, in general, have the desired state to form specific secondary structures in solution and then self-assemble into fibrillary network under various physical conditions (Fu et al., 2013; Tomasini and Castellucci, 2013; Zhang L. et al., 2015). An analysis of the structural/molecular features of different β-sheet peptide hydrogels and their relation to mechanical properties to design effective hydrogels has been reviewed by Rodriguez et al. (2016). The sticky-ended fibrillation designs, applied in DNA and coiled fibers, have inspired Sarkar et al. (2014) to develop a strategy to form staggered triple helical species assisted by interchain charged pairs. The comparison between the two classes of collagen mimetic peptides having same composition but different domain arrangements showed that the larger nucleation domains resulted in rapid fiber formation and gelation while short nucleation domains left the peptide soluble for longer period (Sarkar et al., 2014). The thixotropic supramolecular hydrogels

formed in response to external environmental stimuli have wide ranges of potential biological applications (Levin et al., 2014; Seow and Hauser, 2014; Loic, 2017; Zanna and Tomasini, 2017; Mondal et al., 2020). The stability of peptide-based hydrogels, specially to enzymatic degradation has been reviewed by Yadav et al. (2020). The composition of hydrogel material with tuneable properties controls the release of sensitive drugs (Du et al., 2015; Raza et al., 2018). Various methods to design and control the self-assembly mechanism in hydrogels for drug-delivery have been elaborated in the literature (Altunbas and Pochan, 2011; Briuglia et al., 2014; Yu et al., 2015; Li et al., 2019b). A review by Zhang et al. has highlighted on tuning the unique features/morphology of the nanostructures functions formed during the gelation process by controlling the morphologydependent variations (Zhang L. et al., 2015). The loaded drugs not only get physically trapped, but also bring morphological modifications in the peptide hydrogel (Kurbasic et al., 2017; Parisi et al., 2019), thus converting drug–peptide co-assembly into a supramolecular hydrogel. The attempts have also been made to design peptide-based supramolecular hydrogels for protein drug delivery and gene therapy (Li Y. et al., 2016; Youngblood et al., 2018). The incorporation of non-viral vectors within hydrogels to promote tissue regeneration is another challenging area (Rivas et al., 2019). A photo-cross-linking strategy, based on the ruthenium-complex-catalyzed conversion of tyrosine to dityrosine, to enhance the mechanical stability of nanofibers by 104-fold with a storage modulus of ∼100 kPa (perhaps, one of the highest reported so far among the small peptide hydrogels), with potential to be used in tissue engineering and controlled drug release has been reported by Ding et al. (2013). The low molecular mass organic gelators (LMOGs) have gained much interest in recent years with impending applications in drug delivery and tissue engineering (Sagiri et al., 2014; Skilling et al., 2014; Li Z. et al., 2016). The work on nanostructures containing D-amino acids, for their role in biologics delivery is also catching attention, but has a long way to go (Wang H. et al., 2016). The covalent conjugation between a drug and a D-peptide impacts the gelation properties of a hydrogel including its biostability (Li et al., 2012). Insertion of D-amino acids is reported to twist heterochiral self-assembled peptide hydrogels. Its impact on the drug delivery along with other biological performances makes them versatile tools for therapy in future (Fichman and Gazit, 2014; Melchionna et al., 2016). The use of low molecular weight compounds in hydrogels for drug delivery is turning distinctive due to easy injectability, responsiveness to various stimuli and comfort of synthesis (Raeburn et al., 2013; Mayr et al., 2018). Even the presence of a single amino acid can influence the donor–acceptor charge-transfer interaction in a two-component co-assembled nanofibrous hydrogel (Nelli et al., 2017).

β-Hairpin hydrogels, a subgroup of hydrogels, are another exciting candidates as drug delivery vehicle. These are formed through a molecular self-assembly mechanism which occurs only after desired triggering of intramolecular peptide folding. A review by Worthington et al. (Worthington et al., 2017) has discussed the physical properties of this kind of hydrogel network and material properties which can be used for drug delivery.

Another class of self-assembling peptides, multi-domain peptides (MDPs), allows a wide range of modifications in βsheet motif without disruption so that the materials for delivery can be trapped within the hydrophobic core of the nanofiber depending on the MDP design and cargo. (Kumar et al., 2015; Li I.C. et al., 2016; Moore and Hartgerink, 2017; Lopez-Silva et al., 2018; Chen and Zou, 2019) In a two-component system, in which a porphyrin cap is combined with a cyclic peptide the combination of various binding forces, e.g., hydrogen bonding, metal coordination, and dynamic covalent bonds, allows the delivery of encapsulated ligand (Ozores et al., 2017). These supramolecular injectable biomaterials, that can mimic the natural extracellular matrix nanostructure and show marked cellular infiltration, are ideal scaffolds for tissue engineering strategies. The self-assemble process of MDPs to a nanofibrous hydrogel requires the peptide sequence containing a core of alternating hydrophilic and hydrophobic amino acids and flanked by presence of charged amino acids which further modifies to nanofibers with bilayered β-sheets. These are further modified to a viscoelastic hydrogel processed via nanofiber elongation and cross-linking. The flexible short β-structure and the governing strong forces allow modifications to incorporate functionality, and so are the attractive choices for research (Li and Hartgerink, 2017; Carrejo et al., 2018).

As mentioned earlier, the peptide self-assemblies may have various morphologies and accordingly possess impressive range of applications though sometimes these might be undervalued. To overcome these limitations, assembly of multiple peptidic components can result in a broader range of applications as compared to the self-assemblies of either component. This epitome expands the conformational space of peptide selfassemblies in terms of structural and functional complexities (Makam and Gazit, 2018; Diaferia et al., 2019). For example, mixed dipeptide gelators are assumed to co-assemble to form fibers containing both the gelators randomly. But the pH triggered methodology, determined by the pKa of the gelator, a chemically programmed method, could alter the rate at which self-sorting occurs and forms self-assembled networks (Morris et al., 2013). These sophisticated multicomponent peptide assemblies are being seen as the next-generation bio-inspired materials (Draper and Adams, 2018; Raymond and Nilsson, 2018).

Tissue engineering is a special branch among self-assemblies whose perseverance is the replacement of damaged tissues with newly engineered tissues to restore the normal activity of the target organ, tissue or system. The rudimentary requirement of tissue engineering is a scaffold to sustain the cells and escort the regeneration of the new tissue (**Figure 3**). The selfassembling peptides are able to serve this resolute as scaffolds. The special focus in the review articles of hydrogels and porous scaffolds remains in controlled release of drugs from the tissue engineering platforms (Boekhoven and Stupp, 2014; Loo et al., 2015; Rambhia and Ma, 2015; Koutsopoulos, 2016; Banerjee et al., 2018; Lee, 2018; Inaba and Matsuura, 2019). One of the very important areas in tissue engineering is to explore the ways, to modify the properties of the materials, including mechanical and chemical functionality through changes at the sequence level to influence the cell behavior and modulate the scaffold stability (Zhou et al., 2014). Abbas et al. has reviewed the role of peptide and protein self-assembly in photodynamic and photothermal therapy (Abbas et al., 2017). There is a growing interest in the self-assemblies led by non-covalent hydrogels for three-dimensional cell scaffolding applications. The role of ultra-short molecules, namely, dipeptides and amino acids, holds special position in this area (Ryan and Nilsson, 2012). The collaboration between the molecular structure of the assembled materials and the mechanism of self-assembly grips the nerve on evolving biochemical and viscoelastic properties in this network (Maude et al., 2013). In one of the literature reports, Sarkar et al. (2018) have included an elaborated discussion on possible approaches to assimilate the functionality of peptide scaffolds that may be implanted in vivo at the site of ischemia for application in functional tissue regeneration. A review article by Rubert Pérez et al. (2015) and highlights the selfassembled peptides through solid-phase peptide methodologies, wherein the accurate amino acid sequence can be selected, for constructing bioactive matrices for regenerative medicine. Major advantages, promising applications and current limitations of peptidic materials has been reviewed by Pugliese and Gelain (2017).

Beyond the biological functions, designing peptides, with desired structures and functions by mimicking natural supramolecular systems, are in the air which can provide innovative designing principles to intricate "molecular robots" as next-generation peptide nanomaterials (Inaba and Matsuura, 2019). Handelman et al. (2016b) studied the reconstructive phase transition exhibited by some of the self-assembled ultrashort di- and tripeptide nanostructures to modify optoelectronic properties followed by the appearance of visible photoluminescence. The physics of light propagation (functional properties of linear and non-linear light propagation) in nanostructures of biological origin has been exploited to develop novel integrated nanophotonic devices (Handelman et al., 2016a).

Though the role of self-assembled peptides in drug and gene delivery is blossoming and there is a substantial flow in the research and review articles as mentioned above, but the short peptides (di- and tri-peptides) appear more promising and game changers, due to their simple structure, cost-effectiveness, nontoxic/non-antigenic nature and superior biocompatibility with enhanced bioactivity (Panda and Chauhan, 2014; Goel et al., 2015; Alam et al., 2016; Guo C. et al., 2016; Habibi et al., 2016; Hamley, 2017; Mishra and Jyoti Panda, 2019; Ni and Zhuo, 2019). Furthermore, the low molecular weight of short peptides allows purification via simple HPLC techniques. This review article has been conceptualized to express the potential of nanostructures resulting from single molecule of amino acids, di- and tri-peptides in the field of targeted biologics delivery and tissue engineering.

#### Single or Modified Single Amino Acids

Preparation of nano-shaped aggregates by self-assembly of simple building blocks is fascinating. It is quite startling that a selfassembly of a single amino acid, H-Phe-OH (Adler-Abramovich et al., 2012), was reported in 2012 much after the self-assembly of a dipeptide, H-Phe-Phe-OH (Reches and Gazit, 2003), in 2003. Single amino acid in nanodomain with a stable structure is rare. The chemical modifications of single amino acids have also been investigated for their potential in self-assembly (Chakraborty and Gazit, 2018). The aromatic α-amino acids; H-Phe-OH, H-His-OH, H-Tyr-OH, and H-Trp-OH, have been found to generate ordered self-assembled architects such as fibrils, ribbons, rods, and twisted nanosheets on varying the solvent systems (Singh et al., 2017).

The nanoparticles formed by a single amino acid derivative, a tryptophan derivative, Fmoc-Trp(Boc)-OH, with the molecular weight of 526.6 Da, showed biocompatibility and ability to release the encapsulated bioactive molecules to various cells by Dube et al. (2017). Protection to N-terminus with the Fmoc group and to the side chain with Boc group impacts the nature of peptide self-assembly and overall morphology. The spherical nanoparticles of Fmoc-Trp(Boc)-OH, with hollow interior at pH 6, showed no significant change in structure/morphology with change in pH and retained stability toward thermal perturbations upto 3 months. This asserted the overall impact of Fmoc and Boc groups. Efficient loading and release capabilities along with enhanced toxicity against cancer cells emphasize the role of self-assembly of a simple amino acid as a drug carrier (Tao et al., 2015; Zanna et al., 2015; Dube et al., 2017).

The study to unravel the stimulatory effect of H-Trp-OH on insulin absorption established that H-Trp-OH also possesses bio-enhancing effect as compared to other hydrophobic amino acids, viz., phenylalanine (Phe), proline (Pro) and isoleucine (Ile). The hypoglycemic reaction and surface plasmon resonance (SPR)-based assay showed enhancement in the oral absorption of insulin but without intermolecular interaction. Further, it showed the ability to enhance intestinal absorption of fluorescently labeled hydrophilic dextrans as well as GLP-1 and Exendin-4 peptide drugs (Kamei et al., 2018). H-Arg-OH, as single amino acid, has also shown the potential as insulin absorption enhancer and can be developed as oral delivery systems for insulin (Kamei et al., 2017). Likewise, the hydrophobic and π-π interactions between Fmoc-Lys-OH and pyrrole groups of Chlorin e6 (Ce6), a hydrophobic photosensitive drug, led to the formation of coassembly, which showed an improved cellular uptake. The results advocate the promising potential of a non-toxic photosensitizer delivery system (Liu et al., 2016).

The pyrene conjugated H-Phe-OH derivative, Pyrene-Phe-OH, displayed its ability to gelify in aqueous solutions over a wide range of pH. The change in pH brought about significant alterations in its properties, namely, (i) a distinct change in morphology (nanoscale) from a chiral (left-handed helical) nanofibers to achiral (non-helical) tape like nanofibers with increase in pH, and (ii) change in thixotropic (macroscale) property. The thixotropic behavior facilitated the encapsulation of vitamin B<sup>12</sup> and an anticancer drug, doxorubicin (Dox), within the hydrogel and sustained release paving way to its drug-delivery applicability. The worth mentioning is that Pyrene-Val-OH was unable to form gel owing to absence of π-π interactions for self-association and the gelation process (Nanda et al., 2013).

The hydrophobic interactions hold good position in selfassembly to gel formation. Interestingly, in a research article, it has been shown that phenylketonuria formed due to defective phenylalanine hydroxylase, the H-Phe-OH level in brain increases and the self-assembly forms toxic amyloid fibrils by hydrophobic interactions. The administration of H-D-Phe-OH converts the fibrous formation route of H-Phe-OH to flakes formation and the flakes further restrict H-Phe-OH to form fiber. This slows down the toxic fibril formation (Singh V. et al., 2014), though doxycycline is observed to counteract these toxic effects (De Luigi et al., 2015).

The role of aromatic moieties is quite important in the self-assembly of ultrashort peptides, exclusively Fmoc-peptides, to form hydrogels (Orbach et al., 2012). Small angle neutron scattering (SANS) technique has been used to have a view on gel properties and cross-links of the fiber structures of low molecular weight gels formed by dipeptide gelators (Fleming et al., 2014; Mears et al., 2017). The presence of Fmoc protecting group enhances the hydrophobicity of H-Phe-OH, a hydrophobic amino acid, and lowers the solubility in water. The hydrophobicity of phenyl ring has been reported to contribute toward hydrogel formation ring while its aromaticity brings thermal stability to the supramolecular hydrogel system (Murali and Shanmugam, 2019). The study of the recipe of phenylalanine and Fmoc system for gel formation concluded that the covalent linkage between the two is highly imperative to provide accurate configuration and interactions. The methylene side chain serves in stacking process (in buffer conditions) to facilitate gel formation along with the non-covalent ionic interactions and hydrophobic stacking interactions during close proximity of Phe ring and Fmoc moiety in space. The gel showed enhanced dye diffusion and faster gel erosion at higher temperature suggesting further exploration of the molecule in drug delivery (Singh et al., 2015). The unidirectional hydrogen bonding of the carbamate group within Fmoc-Phe-OH assemblies directed the assembly into 1D fibrils while Fmoc-peptoid analogs displayed 2D/3D morphology due to the alteration in H-bonding (Rajbhandary et al., 2018).

Fmoc-Phe-OH derivatives, modified at the carboxylic acid position with diaminopropane (DAP), (non-fluorinated Fmoc-Phe-DAP, monofluorinated Fmoc-3F-Phe-DAP, Fmoc-F 5 -Phe-DAP, and co-assembly of Fmoc-Phe-DAP and Fmoc-F<sup>5</sup> -Phe-DAP) twisted to hydrogels on addition of physiologically relevant sodium chloride concentrations. The hydrogels exhibited encapsulation of non-steroidal antiinflammatory drug, diclofenac, during the self-assembly process and in vivo drug release profile showed their usefulness as injectable materials. The release study of diclofenac, which decreased from Fmoc-F<sup>5</sup> -Phe-DAP to monofluorinated to the non-fluorinated gelator, demonstrated the role of the molecular structure of the gelators advocating the importance of aromatic benzyl side chain for π-π interactions between the gelator and cargo (**Figure 4**) (Raymond et al., 2019).

An amino acid derivative, N,N′ -dibenzoyl-L-cystine (DBC), has been turned into a supramolecular hydrogel, as the carrier of salicylic acid (SA), by adjusting the pH of the solution. The DBC molecule with two amide groups and two carboxyl

groups was able to oblige as hydrogen-bond donors and acceptors, respectively. The self-assembly of DBC molecules was facilitated by strong intermolecular hydrogen bonds linking neighboring amides and carboxylic acid as well as π-π stacking interactions amid aromatic rings, and hydrogen bonds among water molecules. The pH-responsive hydrogel along with thermo-reversibility and suitable mechanical properties is another addition to the controlled drug release system (Zhong et al., 2019).

Not much has been reported about fluorescent peptide nanotubes. Babar and Sarkar (Babar and Sarkar, 2017) synthesized stable, homogeneous nanotube type aggregates from individual amino acids, H-Trp-OH and H-Tyr-OH, retaining the characteristic fluorescence, which can be used as drug carriers. A simple approach, sonication followed by deposition onto brass stub, was used in the synthesis which could be extended to various nanotubes from single or mixture of amino acids. An inexpensive derivative of tyrosine, H-Tyr(t-Bu)-OH, is perhaps one of the lightest reported low molecular weight organogelator and only natural amino acid-derived gelator, decked with free -NH<sup>2</sup> and -COOH groups. The analysis outcomes made this molecule more special as the gelation could be carried out in all classes of solvents; be it polar, protic, apolar, etc. The gelation in sunflower oil and diesel makes it an appropriate candidate for drug delivery (Aykent et al., 2019). Further investigations revealed the same capability of other enantiomer, H-D-Tyr(tBu)-OH to form the organogel while racemic H-DL-Tyr(tBu)-OH was unsuccessful to do so. The new work also exposed the role of tert-butyl moiety as H-Tyr-OH, H-Phe-OH and H-Tyr(Me)-OH could not form gel while H-Tyr(tBu)-OH could do so as mentioned above (Aykent et al., 2019).

Injectable self-assemblies of derivatives of H-Ala-OH, turned into hydrogels, were used to attach Dox, through an imine bond and the resulting Dox-Gel showed regression in tumor. The presence of rotationally-flexible, an aromatic N-protecting group on the H-Ala-OH accentuated the nanofiber creation. The intermolecular hydrogen-bonding in the derivatives, formed with the addition of carboxamide and hydrazide units at the carboxylic end of H-Ala-OH, H-Ala-CAM and H-Ala-HYD, respectively, along with the protonated H-Ala-HYD+, aided in the formation of gel. The imine bond between the hydrogel and drug was reversible which could cleave in the surrounding areas of the tumors and release the drug. The proposed work highlighted the use of self-assembly of molecules with molecular weight <300 Da in anticancer therapy (Singh M. et al., 2014).

A metallo-hydrogel was fabricated using H-Val-OH based ligand, L-3-methyl-2-(pyridine-4-ylmethylamino)-butanoic acid and Zn(II). The results revealed reversible gel-to-sol and xerogelto-gel phase transitions and demonstrated a release of a polar drug stimulated by change in pH (Saha et al., 2014).

A two-component assembly of Fmoc-Tyr-OH and Fmoc-DOPA-OH (DOPA: 3,4-dihydroxyphenylalanine) resulted in a macroscopic structure with different characteristics from formed by modified amino acids hydrogels individually (Fichman et al., 2015). The amino acid H-Phe-OH in its zwitterionic state, under fibrillization conditions, is reported to be stabilized by hydrogen bonds and aromatic interactions network while organized into an ordered β-sheet-like layered assembly (Mossou et al., 2014). All coded amino acids are suggested to display a layer-like assembly (resembling supramolecular β-sheet structures) which is stabilized by α-amine to α-carboxyl H-bonds regardless of the presence of different side-chains that remarkably differ in their chemical properties though these side-chains govern the higher order organization of the layers. Overall, it recommends that the generic inclination of peptides and the proteins backbones to assemble as layered organizations might be ensued from their basic building block, the amino acid (Bera et al., 2018). Coassembly, though an efficient strategy to form supramolecular structures, suffers in its tailor made functionality due to the lack of knowledge on structural correlation amid different amino acids. This makes the prediction on the resultant co-assemblies a difficult job. Bera et al. have demonstrated that the co-assembly of naturally occurring amino acids with similar chirality is strongly steered by their interlayer separation distances. a basic rule to predict the supramolecular co-assembly relationship with structure (Bera et al., 2019).

The role of disubstituted 1,2,3-triazoles, with the ability to mimic a trans- or a cis- configuration of the amide bond, prompted the researchers to use it as a co-assembly of isosteric amino acid-based hydrogelators. An amphiphilic N-stearoyl-L-glutamic acid (C18-Glu) and its analog with amide moiety replaced by 1,4-disubstituted 1,2,3-triazole unit (click-Glu) displayed distinctive nanostructures due to different hydrogen-bonding configurations. In C18-Glu, the polar protic environment (with intermolecular hydrogen-bonding between the amide NH-bond and the CO-group of the acid moiety next to the chiral center) favored nano-almond crunch-like structures while the non-polar environment preferred (with equal contribution from both intermolecular and intramolecular hydrogen-bonding) formation of nanofibers. The reverse was true for its analog click-Glu. The co-assembly, triggered by the formation of anticipated nanostructures in specific solvents, was applied for the release of an antibiotic, vancomycin (Bachl et al., 2015).

#### Dipeptides for Biologics Delivery

Since the beginning of the journey of the self-assembly, H-Phe-Phe-OH, the dipeptide, reported for the first time by Gazit and his group (Reches and Gazit, 2003), has taken the center-stage responding to its various roles. The core-recognition motif of the Alzheimer's disease associated β-amyloid polypeptide, has potential hydrophobic and hydrophilic moieties which are key parameters for molecular assembly. Other than self-assembly by itself, H-Phe-Phe-OH also functions as a co-assembling peptide for adjusting the self-assembly of various functional molecules, providing an easy and efficient method to modulate the morphology and property of these functional molecules. The potential applications of self-assembled dipeptides are catching up in the field of drug delivery and tissue engineering, as these moieties frame into variety of nanostructures, such as, spherical vesicles, nanotubes, and nanowires. The modes of interaction in general are π-π stackings, hydrophobic associations, electrostatic interactions, and hydrogen-bonding. Various morphologies based on free and protected H-Phe-Phe-OH molecule (Ribeiro et al., 2019) and its analogs/derivatives forming fibrils (Reches and Gazit, 2005), vesicles (Guo et al., 2012), tubes (Reches and Gazit, 2006), wires (Huang et al., 2014; Marchesan et al., 2015b), plates (Tamamis et al., 2009), sheets, flakes (Singh V. et al., 2014), necklaces (Yuran et al., 2012) could be demonstrated by manipulating the stimuli and physical/chemical conditions (Adler-Abramovich and Gazit, 2014; Datta et al., 2018, 2019). Concentration dependent ordered nanoarchitectures such as planar bilayers and other diverse shapes of vesicles, namely, discoid, toroid, ellipsoid, and pot-shaped, have been reported by Guo et al. (2012). Further, by chiral control at different stages, two elementary forms of the peptide self-assembly for the helical twisting of the β sheets were developed. These could turn into wide variety of hierarchical chiral nanostructures, viz., tubelike helical ribbons, twisted ribbons, big twists, nanoscrews and nanosprings (Wang et al., 2015). Even the rarely observed toroid nanostructures could be seen in H-Phe-Phe-OH and H-Phe-Phe-Phe-OH co-assembly (Guo C. et al., 2016) and triaromatic system, Z-Phe-Phe-OH (Z= benzyloxycarbonyl) (Brown et al., 2018).

In one of the studies by Wang et al. (2016b) involving H-Phe-Phe-OH dipeptide molecules, it has also been reported that the non-covalent interactions between trace solvents and peptides performing as solvent-bridged hydrogen bonding lead to directional hydrogen bonding between C-O and N–H in these molecules without inducing π-π stacking. It can promote long-range-ordered arrangement to form nanofibers /nanobelts preferentially along one dimension. The study on H-Gly-Pro-OH dipeptide has supported the ability of n→ π ∗ interactions between carbonyl groups for the stability of the structure (León et al., 2019). The nanotubes formation in self-assembly of a dipeptide, (S,S)-3-amino-2-(2-fluorophenyl)-3-phenylpropanoic acid and H-Ala-OH, has been observed to be supported by intermolecular N–H···O hydrogen bonds, Cπ-H···O, Cπ-H···F, and van der Waals interactions but no evidence of the presence of stacking interactions between the phenyl moieties has been observed by Bonetti et al. (2015).

The ability of peptides, even the ultrashort peptides as small as dipeptide, to form nano- and microstructures born with unique physical properties is the product of reductionist approach. It proposes the creation of well-ordered, amyloid-like β-sheet-rich assemblies comparable to supramolecular structures made of much larger proteins. This approach is exploited for the design and synthesis of peptide structures of technological utilization while establishing simple in vitro model systems to study the parent architecture. This has also resulted in the developing new bio-inspired configurations that could mimic the naturallyoccurring architectures while keeping their functional properties intact, or even sometimes new functionalities are born. The similarity in the assembly mechanism between H-Phe-Phe-OH nanostructures and the aromatic amino acid containing amyloid fibrils inspired Brahmachari et al. to use the reductionist approach. Herein, a screening model was set up to identify molecules possibly capable of interfering with the aggregation process and their mode of action on the modulation of both the assembly and disassembly processes of H-Phe-Phe-OH assemblies (Brahmachari et al., 2017). Gazit is pioneer to demonstrate ultrashort peptides forming ordered assemblies by reductionist approach. His review on the reductionist approach educates on future of minimalistic peptide structures and the latitude of bioinspired self-assembly by the final reduction from very short peptides to extremely small metabolites (Gazit, 2018).

The analysis on temperature dependence of the kinetics of H-Phe-Phe-OH assembly, within the skeleton of crystallization theories, exposed that the transition state from solution to crystalline aggregates is enthalpically unfavorable-entropically favorable, qualitatively similar to longer sequences (Mason et al., 2017). Hydrophobic dipeptides, based on crystallization mode, could be categorized as self-assembled crystals composed of (i) elongated helical tubes; narrow hydrophobic channels and (ii) compact helical tubes; wide hydrophilic channels. The molecular mechanism build on density functional theory, advised by González-Díaz et al. recommends the position of the side chain, where it branches, during crystallization, to be the determining factor to drive the dipeptides into either hydrophobic or hydrophilic channels (González-Díaz et al., 2019). The structural transformation of a dipeptide self-assembly just by changing the types/ratios of the metal ion or the dipeptide to inhibit amyloid-like structure has been presented by Ji et al. (2019).

In a pair of dipeptide enantiomers, it is natural for both to exhibit opposite circular dichroism and handedness in selfassembly. In case, one of the amino acids is chiral, this chiral one exploits handedness. The situation is ambiguous, if both the amino acids are chiral. Fu et al. selected four dipeptides derived from L- and D-alanines and found that the chirality of the alanines at the terminals controls the handedness of their selfassemblies (Fu et al., 2013). The molecular hydrogelators, made of D-amino acids, have been shown to improve the selectivity of non-steroidal anti-inflammatory drugs (NSAIDs) (Li et al., 2012). It suggests that the chirality of the C-terminal amino acid commands the chiral orientation of the supramolecular helical nanostructures. In the study of dipeptides, viz., H-Phe-Phe-OH/H-Ala-Ala-OH, with NSAIDs, self-assembled as hydrogels, the peptides made of D-amino acids provided assistance to preserve the actions of NSAIDs (Li et al., 2013). Erdogan et al. (2015) observed morphological differences in H-Ala-Val-OH and H-Val-Ala-OH dipeptide molecules in the same solvent medium, though particular solvent property having impact on morphology difference could not be fixed. The two peptides differ in terms of the positional disorder of methyl group side chains in L-Val residues and torsion angles. The symmetrical intra- and intermolecular H-bonds, missing in H-Ala-Val-OH, but, present in H-Val-Ala-OH dipeptides were suggested to be accountable for long-range-ordered structures. The position and as well as number of methyl groups at the α carbons on N-terminus and C-terminus Fmoc-Phe(CH3)-Phe-OH or Fmoc-Phe-Phe(CH3)- OH has been observed to impact supramolecular nanostructure and the ability to form hydrogel (Arakawa et al., 2020).

An article dedicated to the role of dipeptides in diverse fields has been published by Panda and Mishra (2016). The bioinspired dipeptides in innovative emerging field of photoelectronics have been highlighted by Chen C. et al. (2015). Marchesan et al. discussed the preparation, characterization and pointed out the potential of H-Phe-Phe-OH motif in nanomedicine (Marchesan et al., 2015b). Multicomponent metallo-nanodrugs as coordination self-assemblies of Fmoc-His-OH and Z-His-Phe-OH in presence of Zn2<sup>+</sup> ions, formed by the combined efforts of coordination and multiple non-covalent interactions were studied for cooperative coordination of photosensitizer Ce6. These multifunctional nanodrugs with enhanced tumor-specific delivery showed the potential for clinical translation (Li S. et al., 2018).

A controllable self-assembly of H-Phe-Phe-OH in an evaporative dewetting solution has been reported by Chen J. et al. (2015). The all-atom simulations of the small H-Phe-Phe-OH oligomers could accomplish the role of several driving forces and structural motifs that command initial assembly (Jeon et al., 2013). H-Phe-Phe-OH oligomers are associated by hydrophobic interactions between side chains while the H-Phe-Phe-OH zwitterionic peptides having charged termini form ordered, clustered, and compact shapes (electrostatic interactions steer their backbones into a more ordered state as compared to those of uncharged ones). From here on, the hydrophobic interactions of the side chains further lead to higher order oligomers just like in amphiphilic peptides. Therefore, the projected study concludes that the initial precursors of H-Phe-Phe-OH might first assemble into structures that form portions of the packed hydrophobic regions in the nanotube walls (Jeon et al., 2013; Chronopoulou et al., 2014). The aptness of the peptide prompted to explore the mechanical parameters of the H-Phe-Phe-OH self-assembled tubes. Atomic force microscope studies revealed these to be the stiffest (averaged point stiffness of 160 N/m) and with much higher Young's modulus (≈ 9 GPa) as compared to other biological nanostructures, significantly enough to provide mechanical strength to cytoskeleton (Kol et al., 2005). In peptides nanotubes, since the control over length of nanotubes in solution, posed a challenge, a slower co-assembly process alongwith the adjustment in the molecular ratio of the H-Phe-Phe-OH assembly unit and its end-capped analog has been suggested (Adler-Abramovich et al., 2016).

The potential of H-Phe-Phe-OH microtubes as intracellular vehicle to release therapeutic compounds has been explored by Silva et al. (2013). The biological marker, rhodamine B, incorporated at the time of self-assembly, had the possibility to intercalate and showed its presence in the inner core to be accepted as intracellular drug delivery for hydrophilic molecules (Silva et al., 2013). These H-Phe-Phe-OH microtubes were further explored for delivering anti-cancer therapeutics, viz., 5-fluorouracil (5-FU) and anti-inflammatory cargo, flufenamic acid (FFA), by Emtiazi et al. (2017) post-conjugation with folicacid and magnetic nanoparticles. These were found to show the potential as molecular carriers (Emtiazi et al., 2017).

The hydrophobic forces in these well-ordered tubular structures further facilitated to be soluble in a suitable solvent and self-assemble into well-organized films on various substrates. This technique was tried by Zohrabi et al. (2016) wherein FFA loaded inside the H-Phe-Phe-OH nanotubes were coated onto Au surfaces functionalized with 3-mercaptopropionic acid (MPA). The biocompatibility and in vitro release studies confirmed the potential of H-Phe-Phe-OH nanotubes as an alternate system for polymer coating in drugs eluting stents. Further, these H-Phe-Phe-OH nanotubes were also shown to possess anti-biofilm activity (Porter et al., 2018). Among the three H-Phe-Phe-OH variants at terminals, such as H-Phe-Phe-OH (l-enantiomer; carboxylic acid terminus), H-D-Phe-D-Phe-OH (d-enantiomer; carboxylic acid terminus), and H-Phe-Phe-NH2, investigated for this purpose, H-Phe-Phe-OH peptide nanotubes were found to be proficient to degrade the biofilm matrix, disrupt cell membranes and hold the potential as efficient drug carrier, though the precise link between these short self-assemblies and biofilm activity needs to be explored.

The peptide-porphyrin macrocycle conjugation has extended its use in biological systems, where the photophysical properties of these molecules and their ability to coordinate metals could be exploited. A review by Biscaglia and Gobbo has presented a survey on biomedical applications of these hybrid compounds, majorly in photodynamic therapy (Biscaglia and Gobbo, 2018). The reversible, biocatalytic and co-assembled nanofibers of Fmoc protected di-peptide (Fmoc-Thr-Leu-NH2) and its porphyrin derivative [TCPP, tetrakis(4-carboxyphenyl)porphyrin] advocate the role of Fmoc-moiety in biocatalysts (Wijerathne et al., 2019). The biocompatible photothermal nanodots formed by self-assembly of peptide-porphyrin conjugate; TPP-Phe-Phe-OH (TPP= tetraphenylporphyrin) have been found suitable for tumor ablation and stand good potential for biomedical photoactive applications. The peptide moieties could provide aqueous stability (through hydrophilic interactions) as well as a spatial barrier (through the strong π-π-stacking interactions between porphyrin groups) to inhibit the further growth of lightto-heat converted nanodots with totally inhibited fluorescence emission and singlet oxygen production (Zou et al., 2017).

Self-assembled cationic dipeptides (CDP), H-Phe-Phe-OH, nanocarriers have displayed the ability to encapsulate and transport drug molecules in vitro at physiological pH condition by the action of enzymes. A covalent bond formed via Schiff base between oligomeric glutaraldehyde and amino groups of CDP followed by aging yielded CDP nanocarriers (CDPNCs). The π-π interactions of aromatic rings have been proposed to be the driving force for the assembled nanocarriers. Remarkably, the auto-fluorescence due to n-π ∗ transitions of C = N bonds offers visually traceable property in living cells. These highly biocompatible carriers with the ability to encapsulate small guest molecules and enzyme-sensitive nature exhibited a high cytotoxicity against tumor cell proliferation leading to be even used as in vivo applications in future (Zhang H. et al., 2015). Though the thermo-induced morphology changes modifying the optical properties in H-Phe-Phe-OH microtubes have been explored by Li et al. (2015), Semin et al. (2015), Nikitin et al. (2016), and Vasilev et al. (2016) the suitability of these H-Phe-Phe-OH based nanocarriers in the intracellular environment of tumor infected tissues was substantiated by Li Q. et al. (2016). Self-assembled hybrid nanospheres, pH- and glutathione (GSH)-responsive, based on H-Phe-Phe-OH and natural alginate dialdehyde as cross linker were used to deliver hydrophobic chemotherapeutic drugs. These nanospheres also contained Au3+, reduced to Au in situ by ADA, and formed H-Phe-Phe-ADA-Au hybrid nanospheres to facilitate ligand exchange reaction of GSH (Li Q. et al., 2016).

The impact of different degrees of hydrogen bonding and nitrogen substitution on mechanical properties of H-Phe-Phe-OH based peptides, indole-diphenylalanine, N-methyl indole-diphenylalanine, benzimidazolone-diphenylalanine and benzimidazole-diphenylalanine, capped at the Nterminus with heterocycles was executed by Martin et al. (2016). In conjugation with NSAIDs, this change altered the gelator properties. H-Phe-Phe-OH covalently conjugated to NSAIDs was found to form hydrogels and exhibit improved selectivity as compared to the native drug displaying its reputation in drug-delivery (Li et al., 2013). Moreover, the inclusion of ferrocene (Fc) moiety in H-Phe-Phe-OH hydrogel, a redox-active site, could change the morphology from nanosphere to nanofiber with the additional feature to reversibly control the self-assembly process of Fc– Phe-Phe-OH by altering the redox state of the Fc group (Wang et al., 2013b).

The study on impact of different N-terminal capping group to vary the aromaticity on hydrogel formation and properties with glyoxylamide mimics as self-assembly hydrogel by Aldilla et al. (2018)showed that the aromatic caps having the bulky indole side chain of tryptophan and another with electronegative substituent on the phenylalanine ring failed to form a hydrogel; might be because of hindrance in the intramolecular stacking. On the other hand, hydrogels with N-naphthalene sulfonyl cap, exhibited a β-sheet secondary structure with viscoelastic properties and topical delivery of ciprofloxacin and recommended as a drug delivery vehicle. Since H-Phe-Phe-OH forms nanofibers with remarkable optical and electrical properties, optical microscopy methods, fluorescence microscopy and super resolution single molecule localization microscopy have been suggested for the study of H-Phe-Phe-OH –based nanostructures (Pujals et al., 2017).

Kuang et al. made an attempt to establish the difference in properties between supramolecular nanofiber assemblies of small molecules and the individual molecules on interaction with the cells. The study of nanofibers formed by the selfassembly of two phenylalanine residues and a naphthyl group, with β-sheet-like-structure revealed the threshold concentration for the formation of nanofibers and suggested hydrophobicity to be responsible for higher cytotoxicity as compared to unassembled monomers (Kuang and Xu, 2013). A photoresponsive hydrogelator, wherein, H-Phe-Phe-OH was protected by 6-nitroveratryloxycarbonyl moiety; Nvoc-Phe-Phe-OH, has been designed and synthesized by Roth-Konforti et al. (2018). The photo-labile trigger, 4,5-dimethoxy-2-nitrobenzyl alcohol, served as a π-π stacking element in place of the aromatic Fmoc group. This stiff 3D-hydrogel displayed responsiveness to UV light irradiation and it completely disassembled subsequently on irradiation at room temperature to release the entrapped drug at the site of administration. The patterning of the photo-responsive peptide based hydrogel along with gradual release of insulin-FITC conferred linear correlation to the stimulus duration. Fmoc-Phe-Phe-OH gel remained unaffected by UV irradiation and showed insignificant FITC-insulin release post-UV application (Roth-Konforti et al., 2018).

The self-supporting gel, Fmoc-Phe-Phe-OH, an ultra-short peptide, is one of the most studied hydrogel. Fmoc-Phe-Phe-OH peptide has been reported by Aviv et al. to deliberate the mechanical rigidity and stability to hyaluronic acid (HA), a major component of the extracellular matrix, without the use of molecules to initiate chemical cross-linking. The Fmoc-Phe-Phe-OH/HA hydrogel, composed of two components, could allow fine-tuning of the hydrogel parameters, adjusted for injection and malleable and facilitating its use in drug-delivery and tissue engineering applications (Aviv et al., 2018). A review by Diaferia et al. describes the immense potential of the hydrogel specially in the field of drug delivery, tissue-engineering and catalytic behavior (Diaferia et al., 2019). Fmoc-Phe-Phe-OH based hydrogel nanoparticles have been explored as nano-carriers to encapsulate and deliver drugs/bioactive molecules. A scalable process for the assembly of Fmoc-Phe-Phe-OH peptide into hydrogel nanoparticles as ensuing drug delivery carriers has been outlined by Ischakov et al. (2013) The results of encapsulated Dox and 5-FU with different structural characteristics endorsed the assumption that the encapsulation ability of nanoparticles could be dependent upon the physicochemical nature of the molecules, thus affecting the release kinetics. Erdogan et al. have used freeze-quenching technique to prepare plasmonic nanoparticle-embedded organogels from Fmoc-Phe-Phe-OH and gold nanorods for efficient drug delivery. The controlled and enhanced release of the Dox was manipulated using laser illumination (Erdogan et al., 2016). The study by Truong et al. (2015) on Fmoc-Phe-Phe-OH self-assembled hydrogels with 5- FU and paclitaxel cautioned that it is the leaching time, not the exposure time which affects the overall cell viability and thus the cytotoxic effects could be observed only after the gel is completely dissolved. Since more stable gels are prone to leach their monomers slower into the neighboring biological environment, the overall concentration of these monomers and any potential adverse effects get reduced. Therefore, the factors controlling the stability of Fmoc-Phe-Phe-OH self-assembled hydrogels are more critical to continue this field (Truong et al., 2015). Argudo et al. have discussed the rules for chemical design through selection of the amino-acid sequence of Fmoc-dipeptides with the self assembling capability for 2D self-assembly at the air/water interface to form supramolecular structures (Argudo et al., 2018). The values of log P and -log S parameters have been proposed based on experimental results which can predict the possibility for a particular Fmoc-dipeptide to self-assemble at fluid interface. A cellular biosensing system, based on 3D culture model using Fmoc-Phe-Phe-OH dipeptide hydrogel employed as both, a 3D cell culture scaffold to provide a confinement in environment and an immobilized enzyme matrix through simple one-pot selfassembly is suggested for the detection of superoxide anion (O s-) by Lian et al. (2017).

A non-centro-symmetric β-sheet structure in Fmoc-Phe-Phe-OH nanofibrils, branded as biomimetic materials with mechanical properties equivalent to the biological gels, also displayed piezoelectric behavior which could enable these assemblies as scaffolds for tissue-engineering (Ryan et al., 2015). Fmoc-Phe-Phe-OH, hydrogel matrix, has been used by Lian et al. to construct a smart biointerface, followed by enzymebased electrochemical biosensing and cell monitoring. The encapsulation of horseradish peroxidase (HRP) was achieved during self-assembly, which was subsequently used to detect the release of H2O<sup>2</sup> from living cells (Lian et al., 2016).

The synergistic effect of the co-assembly of Fmoc-F<sup>5</sup> -Phe and Fmoc-Phe-Phe-OH has been reported to result in a ultrarigid hydrogel with controllable mechanical properties optimum for tissue engineering (**Figure 5**) (Halperin-Sternfeld et al., 2017). The morphological transition from hybrid nanospheres to visible macroscopic films in the co-assembly of the cationic dipeptide (CDP), H-Phe-Phe-NH<sup>2</sup> s HCl and a Keggin-type polyoxometalate (POM), phosphotungstic acid (PTA), upon photothermal treatment was observed which could be suitable for tissue engineering (Xing et al., 2015).

The in vivo studies of self-assembled, injectable, fibrous hydrogels obtained by the combination of dipeptide Fmoc-Phe-Phe-OH and poly-L-lysine have been reported to deliver the photosensitive drug Ce6 in a controlled manner at the tumor site and are suggested to work well under the strategy "once injection, multiple-treatments" by Abbas et al. (2018). Taking in and away plan of sheer forces has been observed to work well for the transmission from gel to sol state supporting the self-healing behavior.

Lately, the study by micro-second molecular dynamics simulations on the co-assembly of H-Phe-Phe-OH with different types of non-H-Phe-Phe-OH dipeptides manifested regularshaped vesicles, single-/multi-cavity assemblies, and planar sheets, which otherwise, are hardly observed in self-assemblies of non-H-Phe-Phe-OH dipeptides. The balancing act between electrostatic repulsion, hydrophobic and aromatic stacking interactions attributes toward the formation of varied structures (Tang et al., 2020). Another novel, multi- component, organicinorganic peptide-based hydrogel, with two building blocks, viz., Fmoc-Phe-Phe-OH and Fmoc-Arg-OH, and an inorganic material, hydroxyapatite (HAP) as 3D scaffolds was designed for bone tissue regeneration. Fmoc-Phe-Phe-OH served as rigid hydrogel that could mimic the ECM, arginine aided in tissue development and might also be helpful to fine tune the bioactivity of HAP, while HAP itself promoted the mechanical properties of the composite material. An optimum composition of the three resulted in a rigid and biocompatible hybrid suitable for cell adhesion and a bone tissue engineering scaffold (Ghosh et al., 2017).

A multicomponent dipeptide hydrogel from Fmoc-3F-Phe-Arg-NH<sup>2</sup> and Fmoc-3F-Phe-Asp-OH, without covalent connection between Arg and Asp, co-assembled by aromatic, hydrophobic, and coulombic interactions displayed the feasibility in the design of innovative materials for tissue engineering (Liyanage et al., 2015).

It had been established that removal of the N-terminus charge via addition of a Boc group could not stop the formation of H-Phe-Phe-OH like structures (Levin et al., 2014). Boc-Phe-Phe-OH in different solvent conditions was observed to display tuneable structures of two different morphologies: tubular or spherical under different solvent conditions and the presence/absence of superparamagnetic iron oxide (SPIONs) core and studied for dual use (drug delivery and heat generation) (Majid et al., 2017). Also, molecular dynamics simulations had suggested that the N-terminus stabilizes the initial stage aggregates during the self-assembly formation while in the final tube structure, it does not rely on the charged termini regions. So when a mixture of H-Phe-Phe-OH and Boc-Phe-Phe-OH monomers, where the uncharged Boc group was added to the N-terminus of the dipeptide, leaving the aromatic residues free to interact, the fibers with varying morphologies and mechanical properties were grown depending on the ratio of the two monomeric dipeptides. Atomic force microscopy (AFM), fluorescence and FTIR spectroscopy showed a reduced Young's modulus and rigidity as well as reduction in hydrogen bonding along with an alteration in the electronic state

of the aromatic residues in co-assembled fibers. So it was hypothesized that this reduction in the hydrogen bonding and a modification in π-π stacking between monomers altered the packing co-assembly. So, the study gives clue to develop various materials with different surface groups resulting in different interactions impacting the mechanical properties of scaffolds to be used in tissue-engineering (Creasey et al., 2016).

The co-assembly of H-Phe-Phe-OH and Boc-Phe-Phe-OH formed the fibers in aqueous conditions with reduced stiffness and curvier morphology. Creasey et al. proposed the disruption of the hydrogen bonding among neighboring N-termini and a variation of π-π stacking between monomers altered the packing of co-assembly and thus provided an option of postassembly chemical functionality and supramolecular sites for drug delivery vehicles (Creasey et al., 2016). Just to briefly mention, a new prototype, an orthogonal strategy, for the cargo release from micro-compartments by modulating the intrinsic self-assembly state and opening the prospects to use the triggered self-assembly to finally control the transport of various molecular species across interfaces has been demonstrated with Boc-Phe-Phe-OH, which could trigger jet-like release within seconds (Levin et al., 2018). Among the short hybrid dipeptides composed of β(O)-δ5 -amino acids synthesized by Reja et al., the injectable hydrogel, Boc-β(O)-δ5-Phe-β(O)-δ5-Phe showed its utility in 2D-cell culture and could encapsulate proflavine solution in the gel matrix and release slowly (Reja et al., 2019).

Rhein (4,5-dihydroxyanthraquinone-2-carboxylic acid), a traditional Chinese anti-inflammatory compound, exhibiting planar aromatic structure, is branded to enhance intramolecular π-π stacking in dipeptide-based materials. The rhein-Phe-Phe-OH peptide gets muddled into uniform spherical nanoassemblies spontaneously, suggesting its competence as carrier for drug delivery. The virtual screening of rhein-Phe-Phe-OH from Traditional Chinese Medicine (TCM) database (based on MD simulations), showed its significant binding energy toward camptothecin (CPT), known for strong antitumor efficacy but limited in use due to poor solubility and stability (perhaps the first compound to be designed cooperatively by both experimental studies and computation simulations). In vitro and in vivo experimental evaluations validated the computation simulations (Sun et al., 2019).

Amoc-Leu-Phe-OH; an Amoc-capped dipeptide, wherein, Amoc moiety is supposed to provide optimal hydrophobicity, which in turn, could enhance self-assembling propensity demonstrated anti-inflammatory action using the rat air pouch model and antibacterial efficiency against Gram-positive and Gram-negative bacteria. The non-cytotoxic hydrogel has been suggested to be used in targeted drug/cell delivery and for wound healing applications by Gavel et al. (2018b).

Non-proteinogenic α,β-didehydro-α-amino acids, with peculiar conformational properties- planar conformation due to a double bond, between Cα and Cβ carbon atoms and fixed values of t φ and ψ torsion angles, have shown their ability to form hydrogels (Vilaça et al., 2015b) and applicable in drug delivery. A procedure for cheaper, less laborious and less time taken than other methods of dehydropeptide synthesis (α,β-didehydro-α-amino acid) through oxidation of the C-terminal residue in N-Boc-protected and N-Fmocprotected peptides, both in solution and on the solid support for peptide synthesis has been reported by Wołczanski and Lisowski (Wołczanski and Lisowski, 2018). The ability to form secondary and super secondary structures and highly stable nanostructures has made α,β-didehydrophenylalanine a center of attraction for the researchers (Gupta and Chauhan, 2011). Self-assembled nanoparticles of dipeptide, methioninedehydrophenylalanine (Met-1Phe), having an unsaturated analog of naturally occurring phenylalanine to entrap curcumin, an anticancer agent, have been synthesized and used by Alam et al. (2012). In this study, the role of 1Phe in self-assembly of H-Met-1Phe-OH was compared with the self-assembly of H-Met-Phe-OH having natural L-phenylalanine. The comparison signified that the presence of 1Phe in H-Met-1Phe-OH promoted the formation and stability of self-assembled dipeptide NPs remarkably, and enhanced entrapment efficiency of curcumin. Besides, the combination of dipeptide assembly with curcumin offered several additional advantages such as high biocompatibility of the former, inhibition of the tumor growth and increased longevity of melanoma tumor-bearing animals of the later. Self-assembly of the dipeptide not only demonstrated the entrapment of curcumin, which could not be used up to its potential due to its hydrophobic nature, leading to poor aqueous solubility, but also improved its bioavailability. Therefore, such dipeptide-NPs could also be used to improve the delivery of other effective hydrophobic drug molecules. Further work on the self-assembled nanotubes using two dipeptides with non-natural amino acids, one containing a flexible H-βPhe-Phe-OH amino acid (βPhe–Phe) and the other containing both H-βPhe-Phe-OH as well as a backbone constrained α,β-didehydrophenylalanine amino acid, H-βPhe–1Phe-OH, (βPhe–1Phe), was reported to have different properties than the native H-Phe–Phe-OH nanotubes. The study also addressed the effect of backbone length, introduction of conformational flexibility (caused by β-Phe) and constraint (produced by 1Phe) on the self-assembly process and morphology. The dipeptide crystal, H-βPhe–Phe-OH, showed two conformations, viz., a gauche conformer forming the channel and a trans conformer trapped inside the channel formed by the gauche conformer, probably the first example presenting such self-entrapment. The nanotubes could efficiently encapsulate small hydrophobic drug molecules. The efficient delivery of anti-cancerous drug, mitoxantrone, entrapped in nanotubes as compared to free mitoxantrone can lead to development of more biocompatible and proteolytically stable drug delivery vehicles (Parween et al., 2014). Similarly, H-Phe–1Phe-OH based nanotubular system as sustained release nanocarriers for intravitreal delivery of the angiogenic drug was investigated. The sustained delivery potential, as pre-loaded and post-loaded options during nanotube formation assembly, was compared to the plain drug. It was speculated that in post-loading approach, there was a possibility of hydrophobic and aromatic π- π stacking interactions between

the drug and the dipeptide. Additionally, certain amount of dipeptide might have released from the nanotube surface which could further entrap the drug within the nanotubes. The hydrophobic nature of the drug speculates its participation in the development of ordered nanotubes. Phenylalanineα,β-didehydrophenylalanine (Phe-1Phe) dipeptide nanotubes were also explored by Khatri et al. to deliver cancer-testis antigens (CTAs). H-Phe-1Phe-CTA nanocomplex showed an inhibition in tumor growth in animal studies, opening the door for this dipeptide as in vivo delivery vehicle for CTAs (Khatri et al., 2019).

An arginine containing dipeptide, arginine-α,βdidehydrophenylalanine (Arg-1Phe), showed the capability to condense the plasmid DNA into positively charged nanoparticles of spherical morphology and deliver it inside the cells and distribute in cytosol and endosomes. The methodology offers simple synthesis route and characterization, viz., stability, biocompatibility and significant transfection efficiency, and made H-Arg-1Phe-OH a suitable contender for in vivo gene delivery applications (Khatri et al., 2017). The introduction of non-protein amino acid, namely N-methylated amino acid, Damino acid and α,β,-didehydroamino acid, in peptides provides enhanced stability with regard to enzymatic degradation. Selfassembled nanoparticles of H-Arg-1Phe-OH were synthesized, characterized and derivatized with folic acid which showed high affinity for its cell surface receptors. The results exhibited significant cellular uptake and tumor regression toward different cancer cells compared to free drug, Dox /drugloaded underivatized NPs both in vitro and in vivo. Iron oxide nanoparticles (IONP) conjugated H-Arg-1Phe-OH also showed heat-induced cytotoxicity on lung cancer cells which resulted in the death of these cells (Baskar et al., 2017). Further, it was demonstrated that the presence of arginine in the cationic dipeptide nanoparticles endorsed pH responsive behavior. FTIR and XRD analysis results for compatibility assessment ensured the retention of Dox activity and efficiency within the developed H-Arg-1Phe-Dox NPs. pH dependent release studies showed signs of maximum release at acidic pH and so, an excellent choice for site-specific drug delivery in tumors and tissues harboring acidic microenvironment like that of stomach (Singh et al., 2018). The H-Arg-1Phe-OH NPs, when conjugated with lactobionic acid (LA), a ligand for the asialoglycoprotein receptors, overexpressed in hepatocellular carcinoma (HCC) cell lines, achieved targeted delivery. H-Arg-1Phe-LA/miR NPs demonstrated the selective delivery of miR-199a-3p and so, the formulation could be used in HCC therapy (Varshney et al., 2018).

Spontaneous self-assembly of a dipeptide, leucine-α,βdidehydrophenylalanine (Leu-1Phe), into a stable hydrogel, was observed to enhance the antitumor activity of the drug, mitoxantrone. The amphipathic nature of the hydrogel, due to unprotected N- and C-terminals in H-Leu-1Phe-OH, could lend polar character. The side chains, bearing a strong hydrophobic character, offer the recipe for the entrapment of both hydrophilic and hydrophobic drugs in the gel matrix and release in a controlled manner (Thota et al., 2016).

Discrete nanoparticles synthesized by combination of different plasmid DNAs with cationic dipeptides. H-Lys-1Phe-OH and H-Arg-1Phe-OH, showed the potential of these dipeptides to protect DNA from enzymatic degradation and carried them to cellular cytoplasm and nucleus. The results showed that the DNA was packaged inside peptide NPs with no visible cytotoxic effect and exhibited enhanced cellular uptake. Therefore, these could be influential in non-viral mediated gene delivery applications (Panda et al., 2013).

Vilaça et al. synthesized naproxen N-conjugated dehydrodipeptide hydrogelators, having at least one aromatic amino acid (Vilaça et al., 2015b). Also, the evaluation of tryptophan N-capped naproxen and C-terminal dehydrophenylalanine (1Phe), dehydroaminobutyric acid (1Abu), and dehydroalanine (1Ala), as protease resistant hydrogelators, demonstrated that Npx-Trp-Z-1Phe-OH (with C-terminal 1Phe) was found to be the most suitable nanocarrier among the lot for the delivery of encapsulated low molecular weight drugs (Vilaça et al., 2015a). These results supported the development of a peptide construct, Npx-Ala-Z-1Phe-Gly-Arg-Gly-Asp-Gly-OH, which was also demonstrated to have properties of an efficient drug carrier (Vilaça et al., 2017). Further studies with tyrosine and aspartic acid residues, tyrosyldehydrophenylalanine [Npx-Tyr-Z-1Phe-OH] and aspartyldehydrophenyl-alanine [Npx-Asp-Z-1Phe-OH], showed that the incorporation of superparamagnetic iron oxide nanoparticles (SPIONs) could keep the magnetic behavior alive in spite of the strong diamagnetic contribution from the involved organic matrix. In vitro heat generation on magnetic excitation with AMF could convert the gel to a solution, acting as a remote trigger to release SPIONs and encapsulated drugs displaying an image-guided drug delivery through self-reporting mechanism (Carvalho et al., 2019). Recently, a newer dehydropeptide hydrogelator; Npx-Met-Z-1Phe-OH (containing naproxen and a thioether), was combined with two different architectures, core/shell manganese ferrite/gold nanoparticle and gold-decorated manganese ferrite nanoparticles (acclaiming the affinity of sulfur toward gold) and tested for the delivery of antitumor drug, curcumin by the group (Veloso et al., 2020). 1Phe is known to endow proteolytic stability and add conformational restraints in the peptide backbone. The naproxen group is added to boost selectivity toward cyclooxygenase. The drug delivery was higher in both irradiated gels as compared to the non-irradiated gels, suggestive for multimodal cancer therapy by combining both the photothermia and controlled drug delivery.

Different metal ions as well as anions modulating the selfassembly of the peptide amphiphiles have been investigated by Sharma et al. advocating an ion responsive behavior to the hydrogels. The group (Sharma et al., 2019) has reported various metal salts to influence self-assembly of histidine based dipeptide amphiphile accompanying a gel to sol transition. In another study, Fan et al. have shown the capability of a dipeptide, H-Trp-Phe-OH, nanoparticles conjugated with an aptamer to bind Dox, due to the π-π stacking interaction and Zn(II) chelation. The formulation exhibited the possibility to act as optical probe to monitor the drug uptake and release (Fan et al., 2016). As recommended by Gazit, the study on fluorophores derived from green fluorescent protein and dipeptide conjugates sets the stage for new frontiers in peptide optics and to load various aromatic drug molecules into these nanoparticles via aromatic stacking interactions (Gazit, 2016).

The nanotubes of H-Lys-Lys-OH functionalized with camptothecin (CPT) chromophore at the ε-amino position, Ac-Lys-Lys(CPT)-NH<sup>2</sup> and NH2-Lys-Lys-(CPT)-NH2, were observed to protect the drug from lactone hydrolysis by sequestering the CPT segment within the hydrophobic nanotube walls, thereby increasing its stability. Herein, CPT conjugated dipeptide played the dual function, both as drug and precursor to the nanostructured carrier (Kim et al., 2015). Similar to this scheme, the research group Sun et al. worked for the delivery of 5-FU and synthesized di-lysine conjugated with 5-fluorouracil (5-FU), wherein hydrophobic π-π association of both uracil and Fmoc could drive β-sheet self-assembly (Sun Y. et al., 2016).

Fluorescein isothiocyanate (FITC) is known as sensitive fluorescent probe which can be detected easily in living cells. The attachment of fluorescein to N-terminal of H-Leu-Leu-OH dipeptide carried out by Kirkham et al. has found to be more cytocompatible as compared to the molecule itself. The strong ππ stacking interactions between the aromatic units in conjugate molecule, Fl-Leu-Leu-OH, formed could be responsible in the induction of self-assembly process. Fl-Leu-Leu-OH adopted a β-sheet structure as shown by FTIR spectroscopy, cryogenic transmission electron microscopy (cryo-TEM) and small-angle X-ray scattering (SAXS) studies. The conjugation improved the capacity of Fl-Leu-Leu-OH to be uptaken by cells without compromising cell viability as compared to FITC alone (Kirkham et al., 2016).

Use of a dipeptide, H-Asp-Phe-OH, to assist the uptake of a therapeutic polypeptide drug, calcitonin, broadly used to treat bone diseases, has been demonstrated by Cao et al. (2017) as co-assembly of the peptide-drug nanoparticles. The dipeptide (Asp-Phe) simply interacted with the Salmon calcitonin (sCT) through hydrophobic and electrostatic interactions leading to the supramolecular nanoparticles of sCT-Asp-Phe-OH. This selfassembly established long-lasting therapeutic effect and could control in vivo release of sCT, thereby, shun the need for multiple injections in patients (Cao et al., 2017).

The formation of NDI-Tyr-Phe-NH<sup>2</sup> by thermodynamically driven enzymatic condensation of naphthalenediimide (NDI) functionalized tyrosine (NDI-Tyr) and phenylalanine-amide (Phe-NH2) has been reported by Nalluri et al. (2014). The presence of dihydroxy/alkoxy naphthalene donors could produce proficient charge-transfer complexes. The fully reversible selfassemblies driven on minimized free-energy and fewer defects holds potential to develop self-healing materials.

Microvesicles, formed from a water-soluble, synthetic, amphiphilic dipeptide, containing a glutamic acid residue at the C-terminus and hydrophobic residue (Val/Leu/Nva) at N-terminal have been established to deliver and release the encapsulated anticancer drug and a fluorescent dye in response to a stimulus of the presence of calcium ions. The dye/drug-loaded vesicles have been found to be non-toxic using 2-(4,5-dimethyl-1,3-thiazol-2-yl)-3,5-diphenyl-2,3-dihydro-1H-tetrazol-4-ium bromide (MTT) based cytotoxicity assay. These biocompatible microvesicles were reported to have the capability to carry cyclic adenosine monophosphate (cAMP) at the target site suggesting the potential of these delivery vectors to carry drugs and other bioactive molecules (Naskar et al., 2011).

A dipeptide hydrogel, containing double Fmoc-group, selfassembled at very low concentration, exhibiting two-step selfassembly process, cytotoxic, suitable for 2D/3D cell scaffolding has been reported recently. The self-assembly of Fmoc-Lys(Fmoc)-Asp-H is reported to have gained advantage from (i) additional H-bonding from the carbonyl group, (ii) aromatic and hydrophobic interactions from the fluorenyl ring, and (iii) steric optimization from the methoxycarbonyl moiety. The balance between the hydrophobic and hydrophilic functionalities is essential for the self-assembly. So the hydrophobicity due to the additional Fmoc group is suggested to be balanced by the two hydrophilic carboxylic acid moieties present at the C-terminus as well as the Asp side chain (Chakraborty et al., 2020).

As linkers play an important role in speedy and effective delivery of the cytotoxic drugs, a lysosomal protease-cleavable dipeptide H-Val-Cit-OH, has commonly been used in the designing and synthesis of antibody-drug conjugates (ADCs) for delivering cytotoxic drugs to antigen bearing cells. Besides, H-Val-Ala-OH, another important dipeptide, has also been employed for synthesizing next-generation ADCs useful for drug loading and releasing purposes. The comparison studies of H-Val-Cit-OH and H-Val-Ala-OH based ADCs conjugated to the monomethyl auristatin E(MMAE) payload using xenograft model test showed their ability as patent payload for the antitumor activity (Wang et al., 2017b; Wang W. et al., 2017). It has also been reported that adding a hydrophilic group to the N-terminus of the valine residue in the H-Val-Cit-OH linker, viz., glutamic acid, prevents the premature cleavage of the linker in mouse plasma which otherwise is susceptible to extracellular carboxylesterase. This enhancement in the polarity of H-Val-Cit-OH linker offers higher in vivo stability to the antibody–drug conjugate in mouse plasma and allows the designing of ADC linkers (Anami et al., 2018).

An amphiphilic dendron, N-octadecanoyl-1,5-bis(L-glutamic acid)-L-glutamic diamide with potential to entrap Vitamin B<sup>1</sup> (VB1) was found to shrink and turn into supergel triggered by monovalent to tetravalent metal ions. The mechanism of continuous, pH-responsive and thermally reversible shrinkage process has been proposed to involve protonation of carboxylic acid groups in presence of metal ions transforming morphology of the hydrogel from nanotubes to nanobelts followed by crosslinking between protonated carboxylic acid groups and metal ions again transforming morphology to nanofibres. During this transition process, the encapsulated water was released but immobilized due to cross-linked 3D structures and VB<sup>1</sup> was released in water phase. The release rate was found to be tuneable upon the shrinking rate of OGAc/metal hydrogel arrangement (Qin et al., 2013). Since the best shrinkage capability was exhibited by Mg2<sup>+</sup> ions, this research group used the proposed phase regulation property with change in pH in a further study. It was also observed that the addition of positively charged species accelerated the shrinkage and negatively charged additives suppressed the shrinkage of the supramolecular hydrogel. These pH change and ionic separation properties were exploited for the step-wise release of two-component drugs, pralidoxime iodide and phenol red, first the anionic drug, phenol red, was released through gel shrinkage and then on increasing pH, released the second cationic drug, pralidoxime iodide, through gel collapse, which suggests that the system still demands more sophisticated work to be carried out on the concept (Qin et al., 2014). Another pH- responsive hydrogel from the class of N-acetylglucosamine derivatives, especially, the heptylurea derivative with a p-methoxybenzylideneacetal protective group, showed release of trapped naproxen on addition of acid (Goyal et al., 2014).

Cyclic peptides have advantage to linear peptides in terms of structure rigidness and stability against proteolytic enzymes. This structure rigidity can reduce the freedom of possible structural conformations and so enhance the binding affinity of the ligands toward receptors. The cyclo-D-Trp-Tyr peptide nanotubes have been synthesized and assessed as carriers of caspase 3 silence shRNA. This could penetrate the intact cornea and deliver the CAP3 pRFP-C-RS DNA reducing the apoptosis in the wounded cornea triggered by corneal epithelial debridement (Lee et al., 2015). The injectable, in situ ambidextrous supergelator, cyclo-Phe-Glu(O-tert-But) with cytocompatibility has been reported to hold potential in drug delivery and regenerative medicine (Manchineella et al., 2017). A cationic, cyclic dipeptide, photoresponsive hydrogelator, with a diketopiperazine (DKP) derivative containing an unnatural amino acid bearing an azobenzene photoswitch (PAP–DKP); PAP-DKP-Lys favorably encapsulated oligonucleotides and released upon irradiation, thereby showing the potential for tuning the amino acid residue for the particular cargoes (Pianowski et al., 2016).

#### Tripeptides as Biologics Carriers

Amino acid chirality, considered a key tool to drive peptide self-assembly, exploited at the molecular and supramolecular levels for D- and L-amino acids positions in case of gelling tripeptides has been reported by Marchesan et al. (2015a), to establish a higher or lower supramolecular order. So, a suitable design of the chirality in sequence of self-assemblies has been suggested to assign and fine-tune the properties of reported tripeptide gel biomaterials. The amphiphilic tripeptides have been reported to generate diverse nanostructures with variable aspect-ratio and geometry as the amino acid is altered by Matson and Stupp. Hydrogen bonding between amide proton donors and adjacent carbonyl proton acceptors leading to βsheet-type association in self-assembling oligopeptides has been found to favor the formation of high aspect-ratio nanostructures (Matson and Stupp, 2012). Sahoo et al. carried the gelation of tripeptides, by connecting a hydrophobic phenylalanine residue at the N-terminus, a variable amino acid to Ac-Phe-X-Asp-NH<sup>2</sup> to alter the tendency for intermolecular hydrogen bonding [X= G/A/V/L/I residue], and a charged C-terminal aspartic acid residue. They found that even without a strong peptidespecific secondary structure, the tripeptides with last three of the mentioned amino acids formed hydrogels (Sahoo et al., 2017). Furthermore, the work on selected Ac-Phe-X-Asp-NH<sup>2</sup> sequence with high aspect ratio; Ac-Phe-Ile-Asp-NH2, suggested the requirement of the non-aromatic and uncharged residue in this tripeptide, which in this case was isoleucine–to be at least hydrophobic as well to facilitated high aspect-ratio nanostructure and aid hydrogel formation (Sahoo et al., 2018a). The study by Yang et al., on structural transition in Fmoc-Phe-Phe-Cys-OH from parallel-aligned worm-like micelles to entangled, coiled micelles persuaded by the cross-linking of the disulfide bonding, elucidated the relationship between micelle chirality and gel properties and suggested a strategy to fabricate materials with controlled chirality by means of synchronized non-covalent and covalent self-assembly (Yang et al., 2019). A series of Fmoc- protected tripeptide containing β-amino acid from H-βAla-His-OH [L-carnosine (Car)] joined by covalent link to different Fmoc-L-amino acids with a wide range of sidechains having different hydrophobicities were observed by Das et al. The tripeptides were found to be proteolytically stable under physiological condition and produce helical structures and the gelation capability to depend on the hydrophobicity of the side chain moiety where the additional π-π stacking interaction by the phenyl group supported the process. This study also demonstrated that the Fmoc-Phe-βAla-His-OH containing Fmoc-tripeptide exhibited the highest gelation ability among the examined amino acid residues, may be the phenyl group provided π-π stacking interaction. The tyrosine residue, expected for similar associative interaction, showed less gelation ability due to intermolecular H-bonding through the phenolic –OH group (Das Mahapatra et al., 2017).

Two tripeptides, viz., H-Gly-His-Lys-OH and H-Phe-Phe-Asp-OH, co-assembled through non-covalent assimilation, easily formed complex selectively with copper ions and converted into a hydrogel formed of nanofibers from clear solution of nanotapes in response to complexation (Abul-Haija et al., 2017).

A tripeptide, H-Lys-Phe-Gly-OH, accomplished the morphology change in secondary structure with switching of nanostructures from nanospheres (vesicles) to nanotubes reversibly depending on the concentration. A transition in the secondary structure could change the morphology of nanostructures. At the low concentration, the tripeptide formed spherical structures with diameters of 50±10 nm, while, at the higher concentration, the tubular aggregates with diameters of 190±10 nm were observed by Moitra et al. (2014, 2017). Even a temperature-induced change from tubular to vesicular has been observed. The nanovesicles, formed at physiological pH and disrupted at pH 6, a fit condition for the drugdelivery, were loaded with a chemotherapeutic anticancer drug, Dox, which resulted in the intracellular release of the drug.

The self-assembled nanoparticles of Fmoc-Leu-Leu-Leu-OMe have been reported by Li et al. as carriers of hydrophobic porphyrin derivative, meso-tetra(p-hydroxyphenyl) porphine (m-THPP) (Li J. et al., 2018). The doping of peptide nanoparticles facilitated the solubility of hydrophobic porphyrin in aqueous media. Additionally, the doping preserved the intrinsic fluorescent property and reduced the risk of side effects due to overdose. The doping could enhance the twophoton fluorescence absorption ability of m-THPP. In vitro reactive oxygen species detection and the cytotoxicity test suggested high anticancer efficacy of the peptide nanoparticles by the two-photon irradiation (Li J. et al., 2018; Li et al., 2019a).

Fmoc-halogenated phenylalanine hydrogels have also been prepared and reported by Wang et al. (2013a). Out of various fluorinated analogs, Fmoc-4F-Phe-OH was found to be most suitable since the fluorenyl groups stacked more extensively. However, none of these hydrogels were found suitable to serve as a bioactive scaffold for NIH 3T3 cell culture in 2D environments. Thereafter, H-Arg-Gly-Asp-OH peptide was included in Fmoc-4F-Phe-OH to aid the cells' survival. This compound showed 3T3 cell adhesion and competent cell division. The results were suggestive of using similar bioactive scaffolds for drug delivery, tissue engineering and cell culture etc (Wang et al., 2013a). A self-assembled protected tripeptide, containing hydrophobic core, Boc-Pro-Phe-Gly-OMe, showed the ability to encapsulate hydrophobic drugs—eosin, aspirin and curcumin. The stabilization of the encapsulated nanostructures with Vitamin E-TPGS illustrated the controlled release of drugs endorsing their role as drug carriers (Yadav et al., 2015). Later, replacement of phenylalanine with dehydrophenylalanine, a constrained dehydroamino acid, self-assembled as hydrophobic matrix dehydrotripeptide, Boc-Pro-1Phe-Gly-OMe. The enhanced encapsulation competence of hydrophobic molecules showed soundness of constrained dehydroamino acids in drug-delivery applications (**Figure 6**) (Deka et al., 2017).

H-Phe-Phe-OH could promote self-assembly and gelation through π-π and hydrophobic interactions to construct gel materials. Also, an amino acid, 11-aminoundecanoic acid (AUDA), non-proteinaceous with long aliphatic chain, could add thixotropic property to the hydrogel. A combination of AUDA and H-Phe-Phe-OH residues as Boc-AUDA-Phe-Phe-COOH could promote both hydrogelation and thixotropic nature. The resultant hydrogel exhibited encapsulation and sustained release of an antibiotic (vancomycin) and vitamin B<sup>12</sup> at physiological pH and temperature and endorsed itself for drug-delivery applications (Baral et al., 2014).

High-resolution scanning electron microscopic analysis revealed the architectural differences in the synthesized tripeptides, Boc-Phe-Gly-Glu-OH (L1) and Boc-Phe-Val-Glu-OH (L2), differing in hydrophobic groups and their corresponding Cu(II) conjugates (Das et al., 2018). The tripeptide with more hydrophobic glycine generated flower-like decorated branches and the less hydrophobic formed wellordered spherical assemblies with nano-meric dimensions. The corresponding metallic conjugates self-assembled as nano-belt like structures in the former and nano-flake like in the later,

which could be due to the metal–peptide interactions that controlled the self-assembly process, entailing that the metal– peptide interactions settle on the morphology of the assemblies. The observation on these non-cytotoxic metallo-peptides; Boc-Phe-Gly-Glu-Cu (L1M) and Boc-Phe-Val-Glu-Cu (L2M), could establish the role of metal ions- peptides coordination in removal of strong non-covalent interaction and the formation of supramolecular architectures with various morphologies (Das et al., 2018).

Fmoc-tripeptide hydrogels from Fmoc-Phe–Phe–Phe-OH with different chirality were synthesized by Chronopoulou et al. by a lipase-supported reaction in an aqueous phase using genipin as a crosslinker. The proposed crosslinker could modulate the physicochemical features of such hydrogels. Hydrogels displayed release of almost same amount of the entrapped drug, Dexamethasone (DXM), though the release kinetics varied. The D-amino acid based hydrogels were observed to be "dense" materials, so could detain the entrapped drugs for longer period than the corresponding L-counterparts and could also resist protease degradation. The results could establish the importance of chirality to induce the specific secondary structural features and also modulate physicochemical properties of the hydrogel. These biomaterials hold bright future in drug delivery and scaffolding uses in tissue engineering (Chronopoulou et al., 2014, 2015, 2017).

The prevalent challenges among cell-scaffolding materials are to upkeep with the vital physiognomies of the otherwise complex natural extracellular matrix (ECM) and incorporate the essential characteristics with minimal complexity into a scaffold. A biomimetic, nanofibrous hydrogel, co-assembled from two aromatic PAs, a structural unit, Fmoc-Phe-Phe-OH, and a functional unit, Fmoc-Arg-Gly-Asp-OH showed its supremacy to reconstruct a normal dermal tissue analog with enhanced mechanical strength. Fmoc-Phe-Phe-OH/Fmoc-Arg-Gly-Asp-OH hydrogel with elongated spindle-like morphology resembling myofibroblasts organized the nanofibrous 3D-networks and contracted the gel besides cell secretion of fibronectin and collagen I. As per in vitro results, fibronectin deposited ahead of denser collagen I, indicated construction of a normal dermal tissue within the scaffold. This simplified design by Zhou et al. (2014) supplies a pathway to create such versatile cell-scaffolds in skin engineering and similar bioactivities.

The policy to integrate D-amino acid residues indicates their potential use in drug delivery. The incorporation of D-amino acid residues in three chiral centers of homochiral tripeptide, Boc-Phe-Phe-Phe-COOH, could change the orientation of the corresponding aromatic ring of Phe residue(s) as well as the direction of π-π stacking interaction, which, in turn, could alter the mechanism of drug-delivery. The various combinations tried by Basu et al., hinted that the mechanical strength and proteolytic stability of a particular gelator could be modulated by changing the molecular chirality and placing the D-residue instead of Lresidues in the appropriate position (at or toward the C terminus) (Basu et al., 2016).

A heterochiral supramolecular H-D-Leu-Phe-Phe-OH tripeptide hydrogel, loaded with an anti-metabolite and pyrimidine analog, hydrophobic antineoplastic drug, 5 fluorouracil (5-FU), formed hydrogel under physiological conditions and demonstrated a fast release. Interacting through π-π stacking between the H-Phe-Phe-OH and 5-FU rings as well as hydrogen bonding between the carboxylic acid group and N-H group of the 5-FU, this work examined the potential of a drug capable to interact non-covalently with a self-assembling tripeptide (Parisi et al., 2019). The outcome of the loading of

the anti-inflammatory drugs, naproxen and ketoprofen, in the tripeptide hydrogel in earlier works could establish that the naphthalene unit engages more efficiently through non-covalent interactions than the benzene rings with the aromatic units of the peptide (Mayr et al., 2018). Though the release of 5-FU was found to be faster than that of naproxen, the role of π-π interaction and hydrogen bonding is concurrent with the hypothesized one. H-Phe-DAla-Phe-OH is an exception to the design rule, where the tripeptides of syndiotactic L-D-L stereochemistry adopt an amphipathic conformation, and the side chains in anisotactic configuration enabling the long-range self-organization turn into gel-forming fibers, as the molecule does not possess the required hydrophobicity and steric hindrance and so, it does not turn to hydrogel (Garcia et al., 2018). The attachment of p-aminobenzoyl moiety at the N-terminus of the tripeptide, (p-aminobenzoyl)- Phe-D-Ala- Phe-NH2, could address both the issues to yield a new gelator (Kieffer et al., 2019) that could be explored further.

The self-assembly of nanodrugs, as self-delivery of drugs, involving a photosensitizer (chlorine e6, Ce6) and Dox as carrierfree nanoparticles which combined photodynamic therapy with chemotherapy to inhibit tumor recurrence, has been tried (Zhang R. et al., 2016). Following the same working model, the supramolecular drug-drug delivery system (SDDDS) comprising of a tripeptide, H-Tyr-Ser-Val-OH (YSV) and Gefitinib (GEF), nanoparticles was studied. The low toxicity of the former and molecular targeting ability of the later led to the co-assembly, that required no chemical modification and performed better in drug efficacy both in vitro and in vivo as compared to individual drugs (Zhang Z. et al., 2017).

A phosphorylated and succinylated tripeptide, H-Gly-Tyr-Lys-OH, on conjugation with folic acid and Taxol, FA–Gly-Tyr-Lys–Taxol, self-assembled into a molecular hydrogel. The much higher weight percentage of Taxol in this gel compared to other similar gels, proposes it as injectable material to deliver Taxol for long period of time during chemotherapy (Wang and Yang, 2012).

A cytocompatible, photoresponsive hydrogelator with tetrazole-containing moiety, Tet(I)-Gly-Phe-Phe-OH, with fast responses to mild light irradiation, has shown the ability to modulate cellular microenvironment using a bio-orthogonal photoclick reaction. The mechanically strong hydrogel could entrap horse serum (HS) and human mesenchymal stem cells (hMSCs) for longer duration of time and its release could be modulated through photodegradation of the gel matrix (phototriggered tetrazole-to-pyrazoline transformation). Further investigation holds bright chances to control the biological behavior of the live cells in spatially defined channels (He et al., 2013). Nap-Phe-Phe-Gly-OH, another tripeptide hydrogelator, was found to create surface-induced self-assembly that could inhibit human platelet aggregations. The comparison with Nap-Phe-Phe-OH, which could not inhibit the platelet aggregation, evidenced the status of Gly residue (Zheng et al., 2012). Further exploration on naphthaleneimide (NI) conjugates; NI-Gly-Phe-Phe-OH and NI-Phe-Phe-Gly-OH supramolecular hydrogels, articulated the good biocompatibility of NI-Gly-Phe-Phe-OH with cell lines. The observations could suggest that these supramolecular materials are promising candidates for their use in biomedical applications (Yeh et al., 2016).

Among the hydrogel biomaterials, a tripeptide, Ac-Phe-Phe-Ala-NH2, developed by Pospišil et al., self-assembled into non-covalent, transparent, stable and biocompatible nanofiber hydrogel scaffolds at physiological pH without use of any organic solvent. The intramolecular hydrogen bonding of cross-β-type structure aggregates and non-covalent interactions between molecules of hydrogelators were proposed to lead the creation of the 3D gel network. The physicochemical and biological properties (in vitro) of the hydrogel suggest it as a hopeful scaffold for tissue engineering applications (Pospišil et al., 2016).

Acknowledging the importance of various factors, viz., morphology, rigidity, periodicity and epitope spacing, which impacted the performance of nanostructures and in turn vividly sway cell adhesion and signaling, a peptide-based amphiphile with a rigid amphiphilic structure of the derivative was obtained. A novel compound, N-3β-(4-tert-butylbenzoylamine)- (7α,12α-dihydroxy-5β-cholan-24-oyl)-(S)-Arg-(S)-Gly-(S)-Asp (tert-but-PhC-RGD), containing the H-Arg-Gly-Asp-OH epitope, a recognition motif involved in cell adhesion processes, was synthesized. This unusual self-assembly, compared to conventional H-Arg-Gly-Asp-OH containing amphiphiles, formed fibers with a circular cross-section and tested positive as scaffolds for tissue regeneration. The circular dichorism results showed weak interactions among H-Arg-Gly-Asp-OH groups as compared to the usual twisted ribbons superstructure which stipulate specific interactions among the chiral groups of the amphiphiles. So, relatively free H-Arg-Gly-Asp-OH groups provided opportunities to interact with the environment and, thereby, were found vital to the applications (Travaglini et al., 2017).

A tri-β-peptide, Ac-βAla-Z-βLys-βAla-OH, self-assembled to form non-toxic, biocompatible and mechanically stable hydrogel, probably the first peptide hydrogel comprising exclusively of βamino acids, with stiffness similar to brain tissue. This tripeptide contained a β-homo-lysine (K) residue (to enhance solubility of peptide in aqueous buffer) and a C14 acylated alkyl chain (a hydrophobic acyl tail to produce a peptide amphiphile to direct self-assembly into nano-cylinders) and formed stable hydrogels. The peptide monomers, driven by a 3-point hydrogen-bonding motif associated with the 14-helical structure of N-acetyl-β3– peptides, self-assembled into helices in a unique head to-tail fashion. The hydrogel could rapidly recover stiffness since high strain could break just the 3D fibrous network by upsetting the non-covalent interactions and transform to sol, without destroying the fibers, but reorganized to 3D network again on reducing the strain in a relatively short time. This superior selfhealing property of the hydrogel with minimal post-operative damage could fasten the recovery (Del Borgo et al., 2013). Following the recipe, Motamed et al. (2016) had developed a new technique to synthesize peptide based self-assembly wherein helical N-acetyl-β3–peptides could spontaneously form fibers, ranging from nano- to macroscale. The β3-tripeptides were observed by Del Borgo et al. to devise distinct self-assembly units segregated by a linker and turn to fibrous assemblies. The linkers within the peptide sequence, whether rigid or flexible, could

be composed of a bioactive α-peptide, could demonstrate fiber formation and create a hybrid α/β peptide scaffold regardless of amino acid sequence, or interruption of the sequence in the presented work (Del Borgo et al., 2018). A structural model developed for self-assembled N-acetyl-β3-peptides as stable three-stranded helical coiled coil could be used to explore the fibril structure for other materials (Christofferson et al., 2018) and be tailored for different applications (Kulkarni et al., 2019). A summary of this content is provided in **Table 1**.


*(Continued)*

acid

#### TABLE 1 | Continued


*(Continued)*

#### TABLE 1 | Continued


### CONCLUSION AND FUTURE PERSPECTIVES

Self-assembly by bioactive molecules is the talent to associate into ordered 3D structures via noncovalent interactions, through a bottom-up approach, without guidance by an external source (Dehsorkhi et al., 2014). In the last 20 years, remarkable expansion has revolutionized, the aptitude has navigated to the potentially useful nanostructure-based materials and has offered newer tools in the field of biological and biomedical sciences. Nanotechnology, the driving force behind this evolution and the revolutionary modifications, has contributed immensely toward the realization of targeted and controlled delivery of available therapeutics. Different materials/systems have been developed with this vision. Except for few, these materials hold their own merits and demerits. Biocompatibility, the nature of the interaction with surrounding tissues, is crucial. Certain materials have claimed to demonstrate biocompatibility, while other developed materials showing toxicity, are proved unsuitable for in vivo applications. Cationic lipid-based nanostructures, found to activate immune system, are such an example (Campani et al., 2018). Besides, these are also linked with certain technological issues, such as, reproducibility, stability, low drug-loading, uncontrolled leaching and encapsulation problems. Further polymeric systems have been developed, but the evaluation studies showed these materials also facing similar limitations. The surface functionalization was considered to improve drug or gene-targeting which in general, is complicated (Abd Ellah and Abouelmagd, 2017). Likewise, natural polymers, elicited undesirable immune reactions, illustrated batch to batch inconsistency (Ginjupalli et al., 2017) and so in vivo performance of these polymers became debatable.

Peptides and small molecule-based nanostructures can be virtuous alternatives for therapeutic delivery due to characteristics, like, good biocompatibility, easy on design/synthesis and also the functionalization. These distinctive and superior qualities make their self-assembled nanostructures smart tools in biomedical field. The masterstroke lies in the ability of these self-assembled small molecules to exhibit stimuli-responsiveness to the environment (internal/external stimuli), thus enabled to control and sustain the therapeutic release as per the requirement, an exciting prospect for multiple biomedical and bionanotechnology applications in drug delivery (Mart et al., 2006). Thus, virtues like, mild and rapid synthesis setting, non-requirement of specialized equipments, low production expenditure, easy dispersibility in aqueous

#### REFERENCES


medium and simple functionalization promote their use as future candidates for diverse applications such as, production of biomaterials, healthcare systems drug/gene delivery, tissue engineering, imaging, sensors, diagnosis, bioelectronics, and so on. Diverse types of structures/architectures can be generated simply by altering the physical or chemical conditions. Thus, this newer area of research is growing at an accelerating pace. Self-assemblies do result in the a variety of type of structures; nanoparticles, nanospheres, nanotubes, nanorods, nanotapes, nanofibers, nanogels, etc., yet there still exist several challenges to be addressed to make these self- assemblies- the materials of choice for research. The limited information on the influence of nanostructures on the organisms, control over size and composition during processing, tenability, behavior in aqueous environment, stability, up-scaling and degree of loading/entrapment of therapeutics are some obstacles which still need sincere attention of the researchers. Besides, follow-on these studies to establish biocompatibility and immunogenicity of these nanostructures are lacking. Nonetheless, it is viable to fine-tune the physicochemical properties by assimilating chemical modifications and so optimize the peptide functionality to minimize the toxicity without threatening their therapeutic activity (Lombardi et al., 2019). Hydrogel based drug delivery systems are of special mention as these are customized for modified release to the target site, reduce toxicity and side effects (Du et al., 2015; Li Y. et al., 2016).

Peptides offer a striking platform as non-viral gene delivery vectors which can minimize the disadvantages associated with viral vectors. Several groups have deliberated on nonviral gene delivery vehicles coupled with targeting peptides (Levine et al., 2013; Sharma et al., 2014; Kang et al., 2019; Wang et al., 2019). Though several functional peptide classes possess essential characteristics to overcome extracellular and intracellular barriers and make gene delivery feasible, still none of these functional peptides have been observed to contain all the essential characteristics required to overcome all of the barriers. Attempts are being made to develop novel multifunctional peptide vectors by combinatorial strategies and high-throughput screening (Jia et al., 2013; Raad et al., 2014; Riley and Vermerris, 2017).

### AUTHOR CONTRIBUTIONS

PK and AS conceived the idea to write a review article on peptide self-assembly. SG compiled the article and IS helped in the graphics and compilation. SG, AS, and PK also did editing.


minimalistic and high drug loading approach to increased efficacy. Chem. A Eur. J. 21, 101–105. doi: 10.1002/chem.201404520


eradicate bacterial biofilm infection. Acta Biomaterialia 77, 96–105. doi: 10.1016/j.actbio.2018.07.033


**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 © 2020 Gupta, Singh, Sharma and Kumar. 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.

# Polylactic Acid-Based Patterned Matrixes for Site-Specific Delivery of Neuropeptides On-Demand: Functional NGF Effects on Human Neuronal Cells

Olga A. Sindeeva1,2†, Olga Kopach3†, Maxim A. Kurochkin<sup>4</sup> , Andrei Sapelkin<sup>5</sup> , David J. Gould<sup>6</sup> , Dmitri A. Rusakov <sup>3</sup> \* and Gleb B. Sukhorukov 1,7,8 \*

#### Edited by:

*Stefano Leporatti, Institute of Nanotechnology (CNR), Italy*

#### Reviewed by:

*Hakan Ceylan, Max Planck Institute for Intelligent Systems, Germany Giulia Suarato, Italian Institute of Technology (IIT), Italy*

#### \*Correspondence:

*Dmitri A. Rusakov d.rusakov@ucl.ac.uk Gleb B. Sukhorukov g.sukhorukov@qmul.ac.uk*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

Received: *02 February 2020* Accepted: *28 April 2020* Published: *12 June 2020*

#### Citation:

*Sindeeva OA, Kopach O, Kurochkin MA, Sapelkin A, Gould DJ, Rusakov DA and Sukhorukov GB (2020) Polylactic Acid-Based Patterned Matrixes for Site-Specific Delivery of Neuropeptides On-Demand: Functional NGF Effects on Human Neuronal Cells. Front. Bioeng. Biotechnol. 8:497. doi: 10.3389/fbioe.2020.00497* *<sup>1</sup> Center for Neurobiology and Brain Restoration, Skolkovo Institute of Science and Technology, Moscow, Russia, <sup>2</sup> Remote Controlled Theranostic Systems Lab, Department of Nanotechnology, Educational and Research Institute of Nanostructures and Biosystems, Saratov State University, Saratov, Russia, <sup>3</sup> UCL Queen Square Institute of Neurology, University College London, London, United Kingdom, <sup>4</sup> Center for Photonics and Quantum Materials, Skolkovo Institute of Science and Technology, Moscow, Russia, <sup>5</sup> School of Physics and Astronomy, Queen Mary University of London, London, United Kingdom, <sup>6</sup> Biochemical Pharmacology, William Harvey Research Institute, Queen Mary University of London, London, United Kingdom, <sup>7</sup> School of Engineering and Material Science, Queen Mary University of London, London, United Kingdom, <sup>8</sup> Center of Biomedical Engineering, I.M. Sechenov First Moscow State Medical University, Moscow, Russia*

#### The patterned microchamber arrays based on biocompatible polymers are a versatile cargo delivery system for drug storage and site-/time-specific drug release on demand. However, functional evidence of their action on nerve cells, in particular their potential for enabling patterned neuronal morphogenesis, remains unclear. Recently, we have established that the polylactic acid (PLA)-based microchamber arrays are biocompatible with human cells of neuronal phenotype and provide safe loading for hydrophilic substances of low molecular weight, with successive site-specific cargo release on-demand to trigger local cell responses. Here, we load the nerve growth factor (NGF) inside microchambers and grow N2A cells on the surface of patterned microchamber arrays. We find that the neurite outgrowth in local N2A cells can be preferentially directed towards opened microchambers (upon-specific NGF release). These observations suggest the PLA-microchambers can be an efficient drug delivery system for the sitespecific delivery of neuropeptides on-demand, potentially suitable for the migratory or axonal guidance of human nerve cells.

Keywords: polylactic acid (PLA), patterned microchamber array, drug delivery system, NGF, human N2A cells

### INTRODUCTION

Micro- and nanostructured matrices have prompted new lines of study focusing on cell behavior (adhesion, proliferation, morphology, alignment, migration, gene expression, and even differentiation) and tissue engineering (Sousa et al., 2019). Photolithography and electroplating techniques enable creation of templates with the choice of different geometries. Such templates can be used either as independent matrices for growing cells, or as templates for the reusable synthesis of patterned films composed of polymers, proteins, and colloids, with nanoscale fidelity.

The current patterning techniques to control nerve cell morphogenesis and function include the NeuroArray device designed for the patterned growing of nerve cells (Li et al., 2014), graphene oxide-based hybrid patterns for guiding axonal growth (Min et al., 2017), microcontact-printed polymeric substrates for directed neuronal regeneration (Schmalenberg and Uhrich, 2005), micropatterned polymer brushes for cell directionality (Pardo-Figuerez et al., 2018), and others. The ultimate goal of these techniques has been, however, to modify neuronal growth and cell proliferation, with no attempts to stimulating activity of individual cells. A recently developed patterned microchamber array (MCA) (Kiryukhin et al., 2018) is, therefore, of particular importance: it provides a reservoir matrix consisting of microchambers (microcontainers). Such a matrix can be made of various synthetic and biocompatible polymers using the layerby-layer method (Ermakov et al., 2019). MCA could be composed of a variety of hydrophobic polymers, for instance, polylactic acid (PLA) (Zykova et al., 2019), poly(lactic-co-glycolic acid) (Sindeeva et al., 2018a), etc., with various inclusions inserted into the shell of microchambers, such as gold nanoparticles or the aggregates of carbon dots (Sindeeva et al., 2018b; 2019; Kurochkin et al., 2020). Having biologically active substances inside microchambers allows spatially and temporally modulated control of cell function, not only due to the periodic structure of the material (Norman and Desai, 2006; Bettinger et al., 2009; Ge et al., 2015; Sousa et al., 2019) but also due to the encapsulated cargo release (Kopach et al., 2019). A wide range of biocompatible polymers for the MCA synthesis enables control of the cargo release rate. For example, it was shown that the PLGA-based microchambers provided relatively slow, continuous release of adrenaline hydrochloride from the first day after entering the aqueous environment (Sindeeva et al., 2018a), which has a clear advantage in many clinical applications. The release of a significant quantity of cargo can be induced by ultrasound as a result of the simultaneous opening of many microchambers (Sindeeva et al., 2018a), or otherwise individual chambers can be opened by optical laser targeting (Sindeeva et al., 2018b; Kopach et al., 2019; Kurochkin et al., 2020).

Notwithstanding the advantages of MCA as a system for targeted delivery of drugs and biologically active substances, its applications in human cells remain poorly understood; this precludes perspective clinical use of these systems. This is mainly because the methods of encapsulation, the duration of storage, the release rate of the substance depend not only on the shell material (Lee and Yeo, 2015) and container geometry (Macha et al., 2019), but also on the cargo's chemical and physical properties (Albinali et al., 2019), which vary widely. Earlier, we have demonstrated that PLA-based MCA are fully biocompatible with human cells of neuronal phenotype (Kopach et al., 2019). Furthermore, we showed a site-specific cellular response to the release of compounds of low molecular weight from individual microchambers: doxycycline for enhancing biosynthesis of green fluorescent protein in individual C2C12 cells (Gai et al., 2018) and the excitatory neurotransmitter glutamate for activation of N2A cells (Kopach et al., 2019). These observations have validated MCA as an effective delivery system for modifying cell activity on demand.

The site- and time-specific delivery of neuropeptides is another important step in managing nerve cell growth, including directed neurite outgrowth that has strong potential in neuroregenerative medicine. Here, we demonstrate the feasibility of modulating neuronal cell function through loading and site-specific release of the nerve growth factor (NGF), a neuropeptide which is primarily involved in the regulation of neurotrophic activities, growth, and proliferation of nerve cells (Levi-Montalcini, 1987). Since NGF is a neuropeptide, it is particularly sensitive to the small temperature fluctuations characteristic of laser exposure. The laser-triggered NGF release from MCA to the targeted N2A cells growing on the MCA surface was therefore performed using focused near-infrared (NIR) laser light. The advantage of using NIR laser rests with the minimal light absorption of biological tissues within the 650–975 nm spectral range, which minimizes photothermal effects on cells. The site-specific localization of the laser-triggered photothermal influence on the chamber wall was enabled by the inclusion of gold nanoparticles (GNPs) in the shells of microchambers as highly photo-absorbing agents (Wijaya et al., 2009; Agarwal et al., 2011). GNPs is a safe (Sperling et al., 2008; Boisselier and Astruc, 2009), well-established, and widely used thermosensitive material for polymer carrier opening in vitro and in vivo (Radt et al., 2004; Skirtach et al., 2005; Boisselier and Astruc, 2009; Singh, 2010; Kunzmann et al., 2011).

### MATERIALS AND METHODS

#### Materials

For MCA synthesis, biopolymer PLA (3 mm granules, molecular weight 60,000), chloroform, and NGF-β (molecular weight 13,5 kDa) were purchased from Sigma-Aldrich (UK). The Poly(dimethylsiloxane) (PDMS) kit (Sylgard 184) was purchased from Dow-Corning (Midland, USA).

For gold nanorods (GNRs) synthesis, cetyltrimethylammonium bromide (CTAB, >98.0%), hydrochloric acid (HCl, 37 wt% in water), L-ascorbic acid (>99.9%), and sodium borohydride (NaBH4, 99%) were purchased from Sigma-Aldrich (UK). Hydrogen tetrachloroaurate trihydrate (HAuCl4·3H2O) and silver nitrate (AgNO3, >99%) were purchased from Alfa Aesar.

#### Synthesis of GNRs

GNRs were fabricated by the modified seed-mediated method (Nikoobakht and El-Sayed, 2003; Khlebtsov et al., 2011). At the first step, the seed solution was obtained by mixing 250 µL of 10 mM HAuCl<sup>4</sup> and 10 mL of 0,1 M CTAB. The ice-cold 10 mM NaBH<sup>4</sup> was added to the mixture in the volume of 1 mL. At the second step, 10 mL of the seed solution were mixed with 900 µL of 0.1 M CTAB, 20 mL of 4 mM AgNO3, 50 mL of 10 mM HAuCl4, 10 mL of 1 M HCl, and 10 mL of 0.1 M ascorbic acid for preparing GNRs. Then nanorods were centrifuged at 12,000 g for 60 min. The pellet was re-suspended in deionized water. The final solution was containing about 1012 GNRs per mL; their average width was 11 ± 3 nm and length was 40 ± 6 nm. The axial ratio was ∼3.8, according the longitudinal resonance was ∼790 nm.

FIGURE 1 | A diagram of the MCA, with a drug cargo and GNRs (A). Typical SEM image of PLA-based MCA with GNRs (B). SEM images of PLA patterned film, without (C) and with GNR aggregates (D). The arrangement of GNR aggregates (E, TEM image).

#### Fabrication of PLA-Based MCA

The silicon master was previously made at Shenzhen Semiconductor (Shenzhen, China) using photolithography for MCA synthesis. The pattern on silicon master was represented by a plate with 185,000 cylinders equidistant from each other (diameter 10µm, height 4µm, and distance from center to center 20µm). For the synthesis of patterned films, the PDMS stamp was made as a reverse impression from a silicone master from a mixture of the prepolymer and curing agent (10:1 ratio). The mixture was degassed for 30 min in a vacuum and consolidated (at 70◦C for 3 h). After this, the PDMS master was cut out and separated from the silicon master. The shell of the PLA-based MCA was made by sealing (printing) of two films: the patterned and the flat ones (2 kg cm−<sup>2</sup> , 15 s, at 55◦C) (**Figure 1A**). For the synthesis of the patterned film, the PDMS stamp with microwells was dip-coated for 5 s into the 1 wt% PLA chloroform solution; for obtaining the flat PLA microfilm, the same procedure was made with a cover glass. After printing, the PDMS stamp was removed and the MCA was located on cover glass.

The patterned film was covered with GNRs before printing by scattering GNRs on the inner surface of the PLA film by sedimentation (Sindeeva et al., 2018b). For precipitation and sedimentation of GNRs, 200 µL of 0.5 M NaCl solution was added to 200 µL of the nanoparticles solution to enhance aggregation (Madzharova et al., 2018). The resulting solution was centrifuged at 10,000 rpm, and supernatant was removed. Next, GNRs were resuspended in 200 µL of deionized water. After that procedure, aggregates of GNRs started to adsorb on a hydrophobic PLA surface. Aggregates in comparison with nonaggregated particles have a larger size and mass which led to amplification of sedimentation rate (Midelet et al., 2017). As a result, aggregates of gold nanoparticles were visualized with an optical microscope, as well as with scanning electron microscopy (SEM) and transmission electron microscopy (TEM).

### NGF Loading Into Microchambers; NGF Leak Test

NGF loading was carried out by applying 10 µL of an aqueous solution (10µg/mL) on the inner surface of the patterned film after the deposition of GNRs. For homogeneous loading, the solution was evenly distributed over the entire film surface and allowed to completely dry (**Figures 2A–C** and **Supplementary Movie 1**). To confirm homogeneous loading of

FIGURE 2 | Optical images of the patterned PLA film (A), showing the NGF crystallization process, and an empty patterned film (B), bright-field microscopy in phase contrast mode. The border of NGF solution drop is marked with a blue line. Schematic illustration of the NGF crystallization process on the patterned PLA film (C).

the microwells, NGF crystals were visualized inside the fabricated microchambers with SEM (**Figure 3**).

To test whether spontaneous NGF release from the microchambers could occur after sealing, the fabricated MCA were placed in Dulbecco's phosphate-buffered saline (DPBS) at 37◦C (95% O<sup>2</sup> and 5% CO2) to represent a relatively physiological environment. The amount of NGF was tested in DPBS at different time-points for up to 3 days using the Brilliant Blue G reagent and spectrophotometry.

### Laser-Induced Opening of Individual PLA-Based Microchambers

The NIR lasers are widely used for the opening of targeted drug delivery systems as they are associated with good penetration ability in tissue without damaging living cells. Laser-induced opening of individual PLA microchambers with the N2A cells growing on the MCA surface was performed using a home-made system. The in-house-made system was based on an inverted microscope (Olympus ix71, Japan), into the optical path of which

we integrated a continuous-wave NIR laser module (LD830- MA1W, 830 nm, maximum optical power 1W, Thorlabs Inc., USA) with adjustable output power, to enable photo-thermal activation of selected microchambers (**Figure 4A**). First, NIR laser light was collimated by an aspheric lens and was 3x expanded by an anamorphic prism pair. Next, laser light was directed into the microscope infinity port by a two-mirror periscope. Then, the laser light was directed by an infrared short-pass dichroic mirror (DMSP805, 805 nm cutoff wavelength, Thorlabs Inc., USA) into an exit pupil of an infinity-corrected objective lens LCAch 20x/0.4 PhC (Olympus, Japan) and focused by the objective into a 1µm spot on the surface of a selected microchamber, at a power of 15 mW over 0.5 s. The laser light irradiation exposure time was controlled by a mechanical shutter. The optical system has been tuned for the confocality between the transmitted light and the NIR channel to control the NIR channel focus by the visible-light focus. Thus, we routinely focused the NIR laser on the microchamber base to minimize damage, if any, to NGF crystals which tend to accumulate toward the top (**Figure 3**). The N2A cell reaction to triggered NGF release was recorded using a monochrome CMOS sensor (DCC3260M, Thorlabs Inc., USA) with an infrared filter.

before (B) and after (C) laser exposure (bright-field microscopy). SEM image of an opened microchamber (D).

#### SEM and TEM Techniques

To visualize MCA morphology at different steps through the fabrication procedure (payload, sealing) and after opening microchamber(s), SEM was used to ensure appropriate samples (FEI Quanta ESEM, electron microscope, FEI, Hillsboro, USA). SEM was carried out using an accelerating voltage of 10 kV, a spot size of 3.5, and a working distance of ∼10 mm.

TEM images of the MCA with GNRs were obtained using a Jeol 2100 microscope (Tokyo, Japan). GNRs diameters and lengths were evaluated from digitized TEM images (Grapher 8, Golden Software, Inc.) of about 500 GNRs.

#### Human N2A Cell Culture

To test functional effects of laser-triggered release of NGF from PLA-microchambers, we used human N2A cells. The cell line was maintained as we have recently described in detail (Kopach et al., 2019). Briefly, N2A cells were cultured in Dulbecco's modified Eagle medium (DMEM, Invitrogen, Carlsbad, CA, USA), supplemented with 2 mM L-glutamine and 10% fetal bovine serum, 2% penicillin-streptomycin, and 1% non-essential amino acids at 37 ◦C (5% CO2). After harvesting, cells were washed out and plated on a surface of PLA-based MCA, pre-treated with UV light for at least 2 h in advance. For cell differentiation to neuronal phenotype, the culturing medium was low serum (2%) DMEM. We used N2A cells at earlier passages only (before cells pass passage 20). Microscopic images of differentiating N2A cells on the fabricated MCA were collected before microchamber opening and afterward at various time-points. For the time-lapse imaging, a MCA with differentiating N2A cells on its surface was placed in a microscope cage incubator (5 cm<sup>2</sup> Petri dish) to maintain experimental conditions favorable for live-cell imaging (37◦C, 5% CO2). Images were acquired every 10 min for up to 60 h total.

In separate experiments, N2A cells were plated on glass coverslips placed into a 8 × 8 wells-plate. Experimental groups consisted of the cells of the same passage grown on glass in culture medium without NGF or NGF supplemented at the

FIGURE 5 | Dose-dependent effect of NGF on the neurite outgrowth in N2A cells on glass coverslips. A snapshot of differentiating N2A cells after 1 day of cell growth on glass (A). Statistical summary of the neurite length in N2A cells grown without or with NGF supplemented to culture medium at the concentration of 10 or 100 ng/mL. Numbers of neurites measured for their length are indicated; at least four independent samples (coverslips) were tested for each group. \*\*\**P* < 0.001 (two-tailed, unpaired *t*-test). (B) Representative images of differentiating N2A cells after 1 day of cell growth with NGF at different concentrations: 10 ng/mL (C) or 100 ng/mL (D).

concentration of 10 or 100 ng/mL. There were at least four independent samples tested for each experimental group.

#### Assessment of Neurite Length; Cell Density Analysis

Neurite outgrowth by N2A cells was assessed by measuring the neurite length in different experimental conditions, using a NeuronJ, a plugin of ImageJ software (NIH, Bethesda, USA). Neurites were traced in individual cells manually, using variable digital zooming. Analyses were performed in the cell culture field of view, across multiple areas selected in a pseudorandom manner.

The N2A cell density was analyzed by counting cell bodies on the surface of the fabricated array, within the area of interest (close to the opened microchambers). Cell density was estimated as the number of viable cells per mm<sup>2</sup> over the selected period of time-lapse recording, as indicated.

#### Statistical Analysis

Data are presented as mean ± standard error of the mean, with n referring to the number of neurites measured for their length, for each experimental group. To determine the statistical difference between experimental groups, two-tailed unpaired Student's ttest was used. A p < 0.05 was considered as an indicator of the statistically significant difference.

#### RESULTS AND DISCUSSION

#### PLA-Based MCA With Gold Nanoparticles: Fabrication and Characterization

We created MCA by printing both flat and patterned films (**Figure 1A**). The thickness of the finished film was 0.8–1.0µm. GNPs (as a classical method) were included in the MCA composition before printing, to

enable controlled opening of microchambers with laser light (Singh, 2010; Kunzmann et al., 2011).

from their original positions toward the opened microchambers. Data are representative of images on (A).

For surface modification, an in-advance concentrated water solution with GNRs was prepared (200 mg/mL). Two hundred microliter of this solution were placed on the inner surface of the patterned microfilm with microwells, for 3 h. During this time, the patterned film was horizontally oriented, after which the drop was removed using a micropipette. The entire surface of the patterned film was covered with GNR aggregates, which were clearly visible under an optical microscope. **Figure 1** shows an SEM image (D) and TEM image (E) of the GNR aggregates location. GNRs content in the patterned film was 0.47 pg/µm<sup>2</sup> , as estimated from the absorption spectrum change in the solution, before and after deposition of aggregates.

#### NGF Loading

Microchambers were filled by applying 10 µL of the NGF solution (concentration 10µg/mL) on the patterned PLA film surface (8.5 × 8.5 mm, 185000 microwells), before printing it on a flat film. Although the PLA film has hydrophobic properties (Alakrach et al., 2018), the NGF solution uniformly wetted the patterned surface due to the low surface tension. The drying of the NGF solution occurred evenly over the entire film surface, with a gradual decrease of the solution drop thickness (**Figure 2A**). When the water layer thickness reached a critical point, the rapid formation of crystals in the wells began over the entire surface (**Figures 2A,C**). The crystallization process could be clearly observed in a light microscope in real time (**Supplementary Movie 1**).

FIGURE 7 | Monitoring morphogenesis of differentiating N2A cells upon triggered, site-specific NGF release from PLA-microchambers. (A) Representative snapshots of differentiating N2A cells on the surface of MCA with NGF-loaded microchambers before and following laser-triggered microchamber opening at various time-points. Images taken from the same area of interest; red dotted lines and blue marks, trajectories for optical targeting (the sequence consists of varied trajectory for opening 3 times, ∼20 h apart). Scale bars, 40µm. (B) Time-course of cell density changes within the targeted area during the time-lapse imaging (∼60 h total) before and following triggered NGF release from PLA-microchambers, shown on (A). Red arrows, time of laser-triggered opening.

The images obtained using SEM confirmed the uniformity of filling the microwells with crystals, and the absence of NGF between them (**Figure 3**).

In general, the amount of NGF was 100 ng per sample (8.5 × 8.5 mm) and 0.54 pg per microchamber. This amount was calculated theoretically, by taking into account the total amount of substance deposited on the patterned film surface, and the number of microwells. This amount of NGF was highest possible to achieve reliable loading into the microwells of pre-designed configuration (diameter of 10µm, height 4µm). Further concentration increases led to the formation of crystals across the entire film surface (outside of the microwells) that prevented tight and reliable sealing of microchambers by the two films.

In order to increase the capability of NGF payload, larger microchambers can be used. At the same time, the use of larger (especially taller) microchambers may cause difficulties for cells to freely move, since the surface topography was found to have substantial effects on the cells behavior in different cell types (Norman and Desai, 2006; Bettinger et al., 2009; Ge et al., 2015; Sousa et al., 2019).

Controllable microchamber permeability is an important issue, which we addressed in some detail in our previous work for the PLA-based MCA loaded with low molecular weight compounds (Kopach et al., 2019). To examine potential leakage of NGF from PLA-microchambers, we next carried out testing of the fabricated MCA loaded with NGF in DPBS (at 37◦C, 95% O<sup>2</sup> and 5% CO2) over time and analyzed the amount of NGF in DPBS at different time-points. The amount of NGF that could leak out did not exceed 10–12% after 3 days of incubation in a mimicked physiological microenvironment (**Supplementary Figure 1**). Notably, that level does not exceed sensitivity of the assay. It should be also noted that we detected no leak of cargo payload in our previous study in which the PLA-microchambers contained the excitatory neurotransmitter glutamate over at least 1 week (Kopach et al., 2019).

Clearly, the release and dissolution kinetics of signaling molecules in aqueous physiological solutions (the point of interest here) could vary widely across molecular species. Furthermore, control over this process could be an important issue in the experimental design. In our case, it has been difficult to detect small amounts of NGF in the medium, so that is why we focused on documenting its neurophysiological effects.

#### Individual Microchamber Opening Using NIR Laser

The NIR laser light (15 mW for 0.5 s) was focused only onto a ∼1µm spot over an individual microchamber (i.e., on the microchamber base) to ensure the most local effect on the GNR aggregates in the microchamber wall. The exposure to laser light at 830 nm (**Figure 4B**) was accompanied by the appearance of a small gas bubble (**Figure 4C**) and by structural changes of the microchamber surface (**Figure 4D**). The microscopic bubble formation is associated with the liquid boiling on GNRs surface as a result of energy absorption and fast plasma formation occurring after liquid evaporation and subsequent vapor expansion, which are accompanied by a shock wave (Lauterborn and Ebeling, 1977; Baghdassarian et al., 1999; Link et al., 2000; Link and El-Sayed, 2001).

NIR lasers are used for heating up GNRs in the polymer shell because GNRs efficiently absorb laser energy (Gordel et al., 2014). The heating of light absorber agents such as GNRs by laser irradiation leads to the rapid melting of the carrier walls and subsequent cargo release (Radt et al., 2004; Skirtach et al., 2005, 2007; Singh, 2010). Although we did not measure the dynamics of local temperature during the laser-induced opening, it is known that laser irradiation of metallic plasmonic nanoparticles causes a very local heat increase, especially for isolated GNRs. In such cases, the temperature drops exponentially around plasmonic particles having a negligible impact on the environment (Govorov and Richardson, 2007), which was demonstrated in multilayer capsules with embedded GNR where the elevation of heat is within a one-micron spot (Skirtach et al., 2008). This has been presently confirmed to have no impact on live cells over a micron distance away from the laser exposure zone (Gai et al., 2018; Kopach et al., 2019). Again, we note that light absorption of an aqueous solution peaks at ∼970 nm. At 830 nm used here, it drops ∼20 times: in our case, the bulk of the energy is absorbed by nanoparticles and released as mechanical decomposition rather than heat. Besides, the structural PLA changes were found to take place from above 50◦C (Zhou et al., 2015), with the reported PLA glass-melting transition point near 55–60◦C (Marek and Veney, 2016). We ensured that no neighboring cells were affected by the NIR laser light during the microchamber opening as evidenced by the images shown.

#### Directed Neurite Outgrowth by Local N2A Cells Towards the Opened Microchambers With NGF Payload Inside

Next, we sought to test functional effects of NGF following the laser-triggered opening of microchambers. We utilized the human N2A cell line, a cell type providing rapid cell growth and differentiation to neuronal phenotype of human origin, as shown previously (Kopach et al., 2019). As expected, differentiating N2A cells developed typical axon-like processes and numerous neurites 1 d post-plating (**Figure 5A**), which could extend up to 50µm in length, with morphogenesis progressing during cell growth.

First, we evaluated the NGF-induced effects on differentiating N2A cells grown on glass (control group). Since NGF is a highly potent neuropeptide acting in the ng/mL concentration range, we supplemented NGF to culture medium at the concentration of 10 and 100 ng/mL. There was a clear, dose-dependent effect of NGF on the neurite outgrowth by N2A cells observed after 1 day of cell differentiation with NGF (**Figures 5C,D**). The neurite length was on average ∼31.1µm in control (0 NGF, n = 666 neurites), but ∼40.3µm in the presence of 10 ng/mL NGF (n = 541 neurites; p < 0.0001) and ∼46.8µm with 100 ng/mL NGF (n = 544 neurites, p < 0.0001; **Figure 5B**) for N2A cells of the same passage.

Next, we placed N2A cells on the surface of the fabricated MCA with NGF payload inside microchambers, and grew the cells on MCA. The cells showed no signs of toxic damage during their growth on the top of the fabricated MCA, thus confirming biocompatibility for this type of carriers. We monitored uninterrupted growth of differentiating N2A cells growing on the top of arrays, and away from the arrays, over at least 3 days, with no detectable location-specific deterioration of any kind. This is consistent with the previously reported biocompatibility of PLA as the constituent material in our recent study involving the same cell type and a variety of microchamber cargo loads (Kopach et al., 2019). We monitored N2A cells before and after laser-triggered opening of microchambers, throughout the area of interest for several days. We could observe that, after microchamber opening, the cell density increased within the targeted area (**Figures 6A,B**) and that local cells extended their neurites towards the opened microchambers (**Figure 6A**, **Supplementary Figure 2**). These effects were observed across 6 independent experiments (fabricated MCA/cell preparations), at the day 1 or 2 after opening. The effect was observed regardless of the trajectory applied for microchamber opening: a line segment (**Figure 6A**, **Supplementary Figure 2**), or a sequence of line followed by rectangular or square shape (**Figure 7A**, **Supplementary Movies 2, 3**). These physiological effects further demonstrate that the heat-induced PLA melting required to open individual microchambers for site-specific NGF release was highly localized, leaving the integrity of most NGF molecules intact.

Finally, we carried out time-lapse recording of differentiating N2A cells before and after opening microchambers with NGF payload, by collecting images from the area of interest every 10 min, starting soon after plating N2A cells on the PLA arrays, for up to 3 days in total (**Figure 7**, **Supplementary Movie 2**). We detected an increased cell density within the targeted area, with a sharp rise in response to each sequence of microchamber opening (3 times, ∼20 h apart) (**Figure 7B**). On a finer scale, cells growing in close proximity to the opened microchambers directed their neurites towards the sites of NGF release from opened microchambers (**Figure 7**, images on an expanded scale; **Supplementary Movie 3**). These results demonstrate a directed neurite outgrowth triggered by the site-targeted cargo release from PLA-microchambers on demand. Moreover, the physiological effects observed following the light-triggered NGF release after ∼20 h confirm robust preservation of functional NGF inside microchambers over an extended time. This enables prolonged load storage and selective microchamber opening at a required time point, at a selected microscopic location.

The effects of NGF observed here in human N2A cells of neuronal phenotype are similar to those attributed to NGF across the literature when tested in cell lines and primary neuronal cultures (Craig and Banker, 1994; Jareb and Banker, 1997; Brann et al., 1999; Secondo et al., 2015; Selvaraj et al., 2015). This suggests a potential, in using this type of carrier matrixes, for functional modulation of individual cell activity in response to triggered release of neuropeptides.

#### CONCLUSION

The patterned PLA-based MCA are a versatile drug delivery system for site-specific, geometrically constrained cargo release on demand. Here, we confirm that the PLA-based matrix is fully biocompatible with human-derived cells, which is particularly

#### REFERENCES

Agarwal, A., Mackey, M. A., El-Sayed, M. A., and Bellamkonda, R. V. (2011). Remote triggered release of doxorubicin in tumors by synergistic application important for highly sensitive cells of neuronal phenotype. Microchambers appear to provide safe loading for hydrophilic peptides and, because of the presence of light-absorbing gold nanoparticles in the container shell, enable laser-sensitive, site-specific cargo release on demand. Optical targeting of microchambers for drug release has triggered functional cell responses locally. Importantly, N2A cells demonstrate enhanced neurite outgrowth toward individual microchambers releasing NGF. The PLA-based MCA are therefore a potentially suitable platform for site-specific targeting of neuronal cells of human origin.

### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

#### AUTHOR CONTRIBUTIONS

OS, OK, MK, AS, DR, and GS: contributed conception and design of the study. GS, DR, OS, OK, and MK: experiment design and manuscript writing. OS, OK, and MK: conducting experiments. OK: cells state statistical analysis. MK and AS: design and development of optical system for chambers activation. DG: discussions. DR and GS: project supervision. All authors contributed to manuscript writing and revisions, they have approved the submitted version.

### FUNDING

This work was supported by the Biological Sciences Research Council grant 315 BB/J001473/1 (DR, AS, and GS); Wellcome Trust Principal Fellowship (212251\_Z\_18\_Z), ERC Advanced Grant (323113), and European Commission NEUROTWIN grant (857562) (DR); Russian Science Foundation - project number 19-75-10043 (OS) (development of NGF encapsulation approach), RFBR project number 19-32-60058 (MK) (optical setup for chambers activation).

#### ACKNOWLEDGMENTS

We thank Nadezda V. Tarakina for providing help with the gold nanorods TEM visualization.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2020.00497/full#supplementary-material

Alakrach, A. M., Noriman, N. Z., Dahham, O. S., Hamzah, R., Alsaadi, M. A., Shayfull, Z., et al. (2018). Chemical and hydrophobic properties

of thermosensitive liposomes and gold nanorods. ACS Nano 5, 4919–4926. doi: 10.1021/nn201010q

of PLA/HNTs-ZrO 2 bionanocomposites. J. Phys. Conf. Ser. 1019:012065. doi: 10.1088/1742-6596/1019/1/012065


polymers: aggregates versus nonaggregates. J. Phys. Chem. C 111, 555–564. doi: 10.1021/jp065635k


transition temperature. CrystEngComm 17, 5651–5663. doi: 10.1039/C5CE 00669D

Zykova, Y., Kudryavtseva, V., Gai, M., Kozelskaya, A., Frueh, J., Sukhorukov, G., et al. (2019). Free-standing microchamber arrays as a biodegradable drug depot system for implant coatings. Eur. Polym. J. 114, 72–80. doi: 10.1016/j.eurpolymj.2019.02.029

**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 © 2020 Sindeeva, Kopach, Kurochkin, Sapelkin, Gould, Rusakov and Sukhorukov. 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.

# Lipid-Based Nanovesicles for Simultaneous Intracellular Delivery of Hydrophobic, Hydrophilic, and Amphiphilic Species

Antonella Zacheo<sup>1</sup> , Luca Bizzarro<sup>2</sup> , Laura Blasi <sup>3</sup> , Clara Piccirillo<sup>1</sup> , Antonio Cardone<sup>4</sup> , Giuseppe Gigli 1,5, Andrea Ragusa1,6 \* and Alessandra Quarta<sup>1</sup> \*

*<sup>1</sup> CNR NANOTEC—Institute of Nanotechnology, c/o Campus Ecotekne, Lecce, Italy, <sup>2</sup> Dipartimento di Scienze Biomolecolari (DISB), University of Urbino Carlo Bo, Urbino, Italy, <sup>3</sup> CNR, Institute for Microelectronics and Microsystems, Lecce, Italy, 4 Institute of Chemistry of OrganoMetallic Compounds—ICCOM, Italian National Council of Research—CNR, Bari, Italy, <sup>5</sup> Department of Mathematics and Physics E. de Giorgi, University of Salento, Campus Ecotekne, Lecce, Italy, <sup>6</sup> Department of Biological and Environmental Sciences and Technologies, University of Salento, Lecce, Italy*

#### Edited by:

*Angela Tino, National Research Council (CNR), Italy*

#### Reviewed by:

*Federica Sodano, University of Turin, Italy Alfredo Ambrosone, University of Salerno, Italy Monica Terracciano, University of Naples Federico II, Italy*

#### \*Correspondence:

*Andrea Ragusa andrea.ragusa@unisalento.it Alessandra Quarta alessandra.quarta@nanotec.cnr.it*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

> Received: *31 March 2020* Accepted: *02 June 2020* Published: *03 July 2020*

#### Citation:

*Zacheo A, Bizzarro L, Blasi L, Piccirillo C, Cardone A, Gigli G, Ragusa A and Quarta A (2020) Lipid-Based Nanovesicles for Simultaneous Intracellular Delivery of Hydrophobic, Hydrophilic, and Amphiphilic Species. Front. Bioeng. Biotechnol. 8:690. doi: 10.3389/fbioe.2020.00690* Lipid nanovesicles (NVs) are the first nanoformulation that entered the clinical use in oncology for the treatment of solid tumors. They are indeed versatile systems which can be loaded with either hydrophobic or hydrophilic molecules, for both imaging and drug delivery, and with high biocompatibility, and limited immunogenicity. In the present work, NVs with a lipid composition resembling that of natural vesicles were prepared using the ultrasonication method. The NVs were successfully loaded with fluorophores molecules (DOP-F-DS and a fluorescent protein), inorganic nanoparticles (quantum dots and magnetic nanoparticles), and anti-cancer drugs (SN-38 and doxorubicin). The encapsulation of such different molecules showed the versatility of the developed systems. The size of the vesicles varied from 100 up to 300 nm depending on the type of loaded species, which were accommodated either into the lipid bilayer or into the aqueous core according to their hydrophobic or hydrophilic nature. Viability assays were performed on cellular models of breast cancer (MCF-7 and MDA-MB-231). Results showed that NVs with encapsulated both drugs simultaneously led to a significant reduction of the cellular activity (up to 22%) compared to the free drugs or to the NVs encapsulated with only one drug. Lipidomic analysis suggested that the mechanism of action of the drugs is the same, whether they are free or encapsulated, but administration of the drugs by means of nanovesicles is more efficient in inducing cellular damage, likely because of a quicker internalization and a sustained release. This study confirms the versatility and the potential of lipid NVs for cancer treatment, as well as the validity of the ultrasound preparation method for their preparation.

Keywords: nanovesicle, nanoparticle, doxorubicin, SN-38, breast cancer, lipidomic analysis

## INTRODUCTION

Lipid-based vesicles are the most commonly used carriers for drug delivery in nanomedicine thanks to their ease of preparation, excellent biocompatibility, and structural plasticity. They are conventionally defined as spherical vesicles characterized by an outer bilayer of lipids with an internal aqueous cavity. Since the first description of liposomal formulation, almost 60 years ago, many types of lipid vesicles have been developed for applications in various areas, including drug and gene delivery (Grimaldi et al., 2016). Some of them have been already approved for clinical use in cancer chemotherapy as carriers of pharmaceuticals, such as doxorubicin, irinotecan, cisplatin, and placlitaxel (Bulbake et al., 2017). Other formulations are currently under clinical evaluation for the delivery of drugs in cancer and other diseases.

The considerable research efforts in this area are motivated by the numerous benefits provided by the use of liposomes, including their ability to host both lipophilic and hydrophilic molecules, the presence of an aqueous core that can accommodate large compounds, the high affinity with the cell membrane that facilitates their internalization, the biodegradability and the negligible toxicity of the vesicle components. Indeed, they are generally composed of natural lipids; among them, phospholipids, such as phosphatidylethanolamine and phosphatidylcholine, and sterols, such as cholesterol, have been used as components of the bilayer (Li et al., 2019).

The amount of cholesterol and the length and saturation of the hydrocarbon chains of the phospholipids affect the rigidity and the stability of the bilayer, and in turn the capability of the NVs to host and release drugs/biomolecules (Monteiro et al., 2014). On the other hand, functionalization of the hydrophilic heads of the lipids with polymers or biomolecules, provides additional features to the vesicle surface, thus shaping their interaction with blood components, tissues, and the immune system in vivo (Riaz et al., 2018). For instance, surface coating with polyethyleneglycol (PEG) polymer chains has been demonstrated to ameliorate the colloidal stability of the liposomes and hence their capability to elude the immune system and prolong the circulation time (Nkanga et al., 2019).

Depending on the preparation method, it is possible to tune both size and lamellarity of lipid-based vesicles. Conventional synthetic approaches, which include film hydration, reversephase evaporation, and detergent dialysis, can produce multilamellar vesicles; however, they suffer from high size polydispersity and low reproducibility (Kraft et al., 2014). On the other hand, methods based on the application of mechanical forces, such as sonication and homogenization, allow for a finer control of the vesicle's size and uniformity. More recently, microfluidic technologies have been applied to the synthesis of uniform lipid vesicles (Carugo et al., 2016).

In a previous work, we also designed a microfluidic system for the preparation of lipid nanovesicles with highly homogeneous size distribution and good reproducibility (Zacheo et al., 2015). However, this method suffers from low particle yield per volume and long preparation times. In the present work, NVs with a similar lipid composition were prepared using the more versatile ultrasonication method (Klingler et al., 2015; Salvi and Pawar, 2019). NVs were synthesized combining two phospholipids [1-lauroyl-2-hydroxy-sn-glycero-3-phosphocholine (LPC) and 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC)] with a ceramide [N-lauroyl-D-erythro-sphingosine (CER)] and cholesterol (CHOL). The molar ratio of the lipids was set in order to resemble the lipid composition of vesicles naturally released by cells, that are typically rich in cholesterol, sphingolipids, and phospholipids (Antimisiaris et al., 2018; Skotland et al., 2019). To demonstrate the high versatility of this system, several compounds with different physico-chemical properties (i.e., hydrophilic, hydrophobic, and amphiphilic molecules) and dimensions (with molecular weight ranging from few hundreds Da to hundreds kDa) were encapsulated into the NVs. More specifically, inorganic nanocrystals, such as quantum dots (QDs) and magnetic nanoparticles (MNPs), an organic amphiphilic fluorophore (i.e., DOP-F-DS), the transferrin protein, and two anticancer drugs (i.e., doxorubicin and SN-38) were loaded either into the bilayer or into the core.

Doxorubicin (DOXO) is commonly employed in the chemotherapy of several solid tumors and leukemia, especially in its lipid-based formulation, the first one to be approved for clinical use (Barenholz, 2012). On the other hand, SN-38, namely 7-ethyl-10-hydroxy camptothecin, is a topoisomerase I inhibitor used in cancer therapy against solid tumors such as colon carcinoma, breast, ovarian, and pancreatic cancers (Wallin et al., 2008). It is characterized by low water solubility and poor stability in the blood stream. Therefore, high doses are needed in clinical therapy, also causing serious toxic side-effects in patients. As such, a liposomal formulation of this drug would provide its beneficial protection until reaching the target site and a prolonged circulation time of the active ingredient (Fang et al., 2018).

The prepared NVs were fully characterized and the thermal stability, size, morphology, surface charge, and colloidal stability over time assessed. The efficiency of the drug encapsulation and the effectiveness of the system as anti-cancer carrier were tested on selected human cancer cellular lines**—**MCF-7 and MDA-MB-231. Furthermore, the effect of the two drugs, either free or loaded into the NVs, on the cancer cells was studied via lipidomic analysis. This type of study is gaining increasing attention as it can provide valuable information about the changes that occur upon a particular stimulus, such as drug administration (Zhao et al., 2015; Perrotti et al., 2016). Lipids are the mayor constituents of the cellular membrane and they are also involved in many biological processes strongly related to carcinogenic pathways, such as transformation, progression, and metastasis, and their composition is altered in many neoplastic diseases (Giudetti et al., 2019). Nuclear magnetic resonance (NMR) analysis of the variation in the lipid composition upon drug treatment can thus provide valuable information about efficacy and progression of the treatment.

### MATERIALS AND METHODS

1,2-Dilauroyl-sn-glycero-3-phosphocholine (DLPC, 12:0), 1 lauroyl-2-hydroxy-sn-glycero-3-phosphocholine (LPC, 12:0), cholesterol (ovine wool >98%, CHOL), and N-lauroyl-Derythro-sphingosine (saturated ceramide, d18: 1/12: 0, CER) were purchased from Avanti Polar Lipids. The molecular formulas of the lipids are reported in **Figure 1**. Doxorubicin and transferrin-TRITC were purchased from Sigma-Aldrich and ThermoFisher, respectively. SN-38 was a kind gift from Ospedale Oncologico Giovanni Paolo II in Bari, Italy. The

organic fluorophore DOP-F-DS (**Figure 1**) was prepared as previously reported in the literature (Cardone et al., 2012). CdSe/ZnS QDs and iron oxide MNPs were prepared according to the procedures reported in the literature (Dabbousi et al., 1997; Hyeon et al., 2001). The two cell lines, namely MCF-7 and MDA-MB-231, were obtained from ATCC.

### Synthesis and Characterization of the Nanovesicles

A mixture of lipids consisting of DLPC, CHOL, LPC, and CER with respective ratios of 4.5:4:1:0.5 was used for the synthesis of the vesicles (final concentration 5 mM in 5 mL of chloroform). The phospholipid solution was stirred for 1 h with the cap closed to prevent any loss of solvent. Subsequently, the cap was removed and the solution left stirring overnight at room temperature to evaporate the solvent. Phosphate saline buffer (PBS) was then added and the solution was sonicated for 4 min at a power of 25 W and at 20 kHz frequency. During the sonication the vial was kept into an ice bath to prevent overheating of the solution.

To prepare the vesicles containing transferrin-TRITC and the blue-emitting DOP-F-DS organic fluorophore, 150 µL of DOP-F-DS (1 mg/mL in DMSO), and 50 µL of transferrin-TRICT (5 mg/mL in PBS) were respectively employed. To load SN-38 into the liposomes, 200 µL of the drug dissolved in DMSO (200µM) were added to the phospholipids in chloroform and the mixture was left under vigorous stirring for 60 min. Subsequently, the vesicles were prepared following the protocol already described. The hydrophobicity of the SN-38 molecule facilitates its intercalation within the double phospholipidic layer of the liposomes. In the case of doxorubicin, the drug dissolved in PBS (85.8µM final concentration) was added to the dry phospholipid film. The NVs were then prepared as already

TABLE 1 | Conditions employed for the preparation of the nanovesicles (type of loaded species, volume, concentration, and solvent used) and their application in this study.


described. To load both drugs into the vesicles, solutions at the same concentration as before were used. The NVs were then prepared according to the standard conditions but sonicated for longer time (6 min). To encapsulate the inorganic nanoparticles within the vesicles, few microliters of a chloroform solution of either QDs or MNPs were added to the starting lipid mixture (see **Table 1** for details).

The solution containing the formed vesicles was then transferred into a dialysis tube (50 kDa MWCO), placed into 2 L of PBS, and kept under stirring at 4◦C for 48 h. After purification, the nanovesicles were fully characterized.

Size and morphology of the lipid vesicles were monitored through transmission electron microscopy (TEM) and dynamic light scattering (DLS) measures. Low-magnification TEM images of the nanovesicles were recorded on a JEOL Jem1011 microscope operating at an accelerating voltage of 100 kV. DLS and zeta potential measurements were performed in PBS at 25◦C using a Zetasizer Nano ZS90 (Malvern Instruments Ldt) equipped with a 4.0 mW He–Ne laser operating at 633 nm and with an avalanche photodiode detector.

Thermal stability was assessed by thermogravimetric analysis (TGA) using SDT Q600 equipment (TA Instruments) with a heat ramp of 5◦C/min and an air flow rate of 100 mL/min. After synthesis of the NVs, the solvent was evaporated and the analysis was performed on the residual dry matter. Control analyses were also done on the free lipids, which were mixed in chloroform at the same molar ratio before removing the solvent.

Encapsulation efficiency (EE) and release profile of the encapsulated drugs were determined spectrophotometrically with a fluorescence spectrometer (Cary Eclipse) measuring the fluorescence intensity of the drugs. A calibration curve at known concentrations of the molecules was first prepared. The amount of free drug dissolved in the dialysis medium (PBS), after purification, was measured and the EE, expressed as percentage of encapsulated molecules over the total, was calculated according to the formula:

$$EE\text{ (\%)} = \frac{(Initial\text{ }drug\text{ }amount\text{ } -free\text{ }drug\text{ } in\text{ }the\text{ }daily\text{ } medium)}{Initial\text{ }drug\text{ } amount} \times 100\text{ }$$

To determine the drug release profile, samples were kept at 37◦C at two pH (4.5 and 7.4) for different time lengths (6, 24, 48, 96, and 120 h). The vesicles were then pelleted and the amount of drug in the supernatant quantified. The released amount at each time point, expressed as percentage over the total encapsulated, was estimated according to the formula:

Released amount (%) = released drug concentration encapsulated drug concentration×<sup>100</sup>

#### Cellular Studies Viability Assays

Two human cell lines of mammary carcinoma, namely MCF-7 and MBA-MB-231, were used. The cells were grown in DMEM medium supplemented with 10% of fetal bovine serum (FBS), 2 mM glutamine, 100 IU/mL of penicillin, and 100µg/mL of streptomycin, and were cultured in an incubator at 37◦C in a humidified atmosphere with 5% CO2.

Two viability assays were performed, namely MTT and Trypan blue assays. In detail, 5 × 10<sup>4</sup> cells suspended in 200 µL of culture medium were seeded into each well of 96 multiwell plates. After 24 h incubation at 37◦C, the nanovesicles were added to the wells at a defined concentration (each point triplicated) and the cells were kept under incubation for three different times, 24, 48, and 120 h.

In the case of the MTT assay, at the end of the incubation time, the medium was removed, the cells were washed twice with PBS, and 200 µL of fresh serum-free medium containing 1 mg/mL MTT were added to each well. After 3 h of incubation at 37◦C, the medium was discarded from the wells and 200 µL of DMSO were added to dissolve the formazan salts. The plate was stirred for a few minutes at ambient temperature and the absorbance of the solution at 570 nm was measured on a microplate reader. To determine the percentage of cell viability, the treated samples were compared to the control samples according to the equation:

$$\text{Cell viability } (\%) = \frac{\text{Absorbance of the NVs} - \text{containing sample}}{\text{Absorbance of the control sample}} \times 100\%$$

To check the presence of a synergic effect by the combined delivery of the two drugs, the combination index (CI) was also determined according to the following formula (Chou and Talalay, 1984):

$$CI\_{50} = \frac{C\_A}{IC\_{50A}} + \frac{C\_B}{IC\_{50B}}$$

where C<sup>A</sup> and C<sup>B</sup> are the concentrations of drugs A and B coloaded into the NVs that were used to achieve the IC<sup>50</sup> effect. IC50A and IC50B are the concentrations of the single drugs used to achieve the same effect on cell viability.

In the case of the Trypan blue assay, at the end of the incubation time, the medium was removed, the cells were washed twice with PBS, and 100 µL of PBS containing Trypan blue diluted 1/10 were added to each well. Soon after, the plates were analyzed under the microscope and the number of viable and dead cells was counted. The cell viability was determined as follows:

$$\text{Cell viability } (\%) = \frac{Number \text{ of } \text{variable cells}}{Number \text{ of total cells}} \times 1000$$

#### Lipidomic Analysis

MDA-MB-231 cells (1 × 10<sup>8</sup> cells/well, three replicates each) were treated with the empty nanovesicles, with the free drugs alone or in combination, and with the nanovesicles loaded with either one or both drugs. After 24 h incubation, cells were quenched and the lipid fraction extracted according to previously published procedures (Sündermann et al., 2016). The extracts were dried under a gentle nitrogen flow, dissolved in 600 µL of deuterated chloroform containing 0.03% of tetramethylsilane (TMS), used as the reference, and transferred into 5 mm NMR tubes.

Spectra were acquired on a Varian INOVA spectrometer (Varian Inc., CA) operating at 499.792 MHz and equipped with a OneNMR Probe-PT (Agilent). Acquisition parameters were set as follows: s2pul pulse sequence, 128 scans, 25◦C, pulse angle 45◦ , sw 8012.8 Hz, relaxation delay 1 s, acquisition time 2.045 s, np 32,768, complex points 16,384, observed pulse 4.85, calibration pw90 9.70 µs. The resulting spectra were Fouriertransformed, phase corrected, and calibrated to the signal of TMS. The region between 0.5 and 5.5 ppm was binned into 0.02 ppm-wide buckets (equal to 238 variables) and exported to a csv file containing the data for all samples. The obtained matrix was normalized to the total sum and Pareto scaled before multivariate statistical analysis.

#### Confocal Microscopy Imaging

For cell imaging analysis, 10<sup>5</sup> cells were seeded onto a coverslip placed into each well of a six-well plate. After 24 h, the fluorescent nanovesicles were added to each well and the cells were incubated for the settled time (6, 24, and 48 h). Then, they were washed with PBS and fixed with 4% paraformaldehyde prior to be imaged by means of a Leica confocal microscope (TCS-SP5, Leica, Mannheim, Germany) equipped with an Argon laser source (excitation at 488 nm for green emitting QDs, at 545 nm for Transferrin-TRITC and Phalloidin-TRITC) and a pulsed laser (excitation at 405 nm for DOP-F-DS and DAPI).

For the colocalization study, MDA-MB-231 cells were seeded as already described and then incubated with the NVs loaded with transferrin-TRITC for the settled time (6, 24, and 48 h). Then, cells were washed with PBS and stained with Lysotracker green (Molecular probes) according to the manufacturer protocol. The samples were imaged by means of the confocal laser scanning microscopy (CLSM).

#### Statistical Analysis

All data represent the average value of at least three independent experiments, unless otherwise specified. Normally distributed data was compared with a two-tailed Student's t-test using GraphPad Prism software (version 6.0). The bars in graphs represent mean ± S.D. values. Differences were considered significant when the p-value was <0.05. Multivariate statistical analysis in the lipidomic study was performed with SIMCA 14.1 software (MKS Umetrics, Malmö, Sweden) using pairs of classes in each analysis (control vs. free drugs; control vs. encapsulated drugs; free drugs vs. encapsulated drugs; control vs. empty NVs). Orthogonal partial least-squares discriminant analysis (OPLS-DA) was performed using the NMR buckets as variables, the samples as observations, and the formulations (drugs-loaded NVs, empty NVs, free drugs, no drugs) as classes. Cumulative R 2 X and R 2 <sup>Y</sup> were used as parameters for describing the goodness of the fit.

### RESULTS AND DISCUSSION

### Synthesis and Characterization of Lipid Nanovesicles

NVs were prepared using the above-mentioned compounds according to the molar ratio DLPC/CHOL/LPC/CER 4.5:4.0:1.0:0.5. The lipid composition of these vesicles was designed to mimic the structure of the extracellular vesicles released physiologically by the cells having important roles in intercellular communication and in the onset and transmission of diseases (Trajkovic et al., 2008; De Toro et al., 2015; Zhang et al., 2015; Sarko and McKinney, 2017; Skotland et al., 2017; Steinbichler et al., 2017). This affinity should in turn facilitate the interaction with the cell membrane and promote cellular uptake.

Simultaneous encapsulation of hydrophobic, hydrophilic, and amphiphilic species inside the NVs was achieved, with the hydrophobic molecules being located into the lipid bilayer while the hydrophilic ones into the aqueous core. An intermediate behavior would be expected in the case of amphiphilic molecules. Several parameters, such as the molar ratios of the three components, sonication time and power together with the volumes of PBS were modulated to obtain particles with an average size around 100 nm and a good size distribution. Particles with the desired characteristics were obtained by sonicating at 25 W for 4 min.

To study the thermal stability of the developed nanovesicles, TGA analysis was performed and the results are reported in **Figure 2**. As a comparison, the test was also performed on the free lipid components. The weight loss percentage (**Figure 2A**) showed a completely different behavior between the nanovesicles and the free lipids; for the lipids, in fact, complete degradation was observed for 200 < T < 600◦C. Such degradation took place in two main steps, the first corresponding to a weight loss of about 70% (T < 350◦C) while the second of about 25% (T > 350◦C). The curve of the first derivative (**Figure 2B**) highlighted additional smaller steps at lower temperatures.

On the other hand, the data showed a much higher stability of the nanovesicles, as very small weight loss (about 5%) was observed for T < 700◦C, while their degradation took place between 700 and 900◦C. This behavior was confirmed by the first derivative curve.

Significative differences between the nanovesicles and the free lipids could be also observed in the DTA curves (**Figure 2C**). For the free lipids, in fact, an exothermic peak was observed for 200 < T < 400◦C; this can be due to the combustion of some organic fragments/molecules, in agreement with the

literature (López-González et al., 2015). This peak is not present in the corresponding curve of the nanovesicles, indicating that the formation of the vesicle structure makes the whole system more thermally stable. The endothermic peaks, corresponding to the full chemical degradation of the molecules/nanovesicles, can be observed in both curves, but at different temperatures, (i.e., between 500 < T < 800◦C and 770 < T < 900◦C for free lipids and nanovesicles, respectively). Literature data already report of increased thermal stability of the vesicle structure in comparison to the free compounds (Pinilla et al., 2019). This is likely due to the different special arrangements of the molecules; indeed, stronger interactions can take place when lipids are confined in a layer, leading to a more thermally stable system.

Different molecules were loaded into the vesicles, as listed in **Table 1**, either in the lipid bilayer or the aqueous core, according to their physico-chemical characteristics. The same sonication power was used for all experiments, although a longer duration (6 min compared to the four used for all the other cargos) was necessary only when both drugs were encapsulated simultaneously.

The procedure was slightly modified depending on the hydrophobic/hydrophilic/amphiphilic nature of the species to be introduced into the vesicles. In the case of the amphiphilic DOP-F-DS fluorophore or the hydrophobic antineoplastic SN-38 drug, the species were added to the lipid suspension. On the other hand, when hydrophilic molecules, such as the fluorescent transferrin or doxorubicin, had to be encapsulated, they were dissolved in PBS and then added to the lipid dry film before sonication.

The vesicles were analyzed by different techniques to determine their morphology, size, charge and thermal stability. **Figure 3** shows the TEM images of the empty NVs and of the NVs loaded with different species, such as hydrophobic QDs, MNPs, the amphiphilic fluorophore DOP-F-DS, transferrin-TRITC, both DOP-F-DS and transferrin-TRITC, doxorubicin, SN-38, and both drugs.

The empty vesicles displayed an average size of about 65 nm but did not present regular contours. On the other hand, when the vesicles were loaded with inorganic nanoparticles (i.e., QDs and MNPs), their structure became more regular with an evident turgidity and three-dimensionality. This effect could be caused by the evaporation of the solvent during TEM sample preparation, with a different outcome depending on the inner lipid support structure. In fact, while empty vesicles are free to shrink while drying, the presence of inorganic nanoparticles inside the core provides a physical resistance to the contraction. It is likely that both QDs and MNPs accommodate inside the lipid bilayer because of their hydrophobic nature (thanks to the alkyl chains of the surfactants covering the inorganic core). However, the thickness of such layer should be smaller than 4 nm considering that the phospholipids have an alkyl chain of 12 carbon atoms, while the average core diameter of the QDs and of the MNPs is around 4 and 6 nm, respectively, as determined by TEM analysis (**Figure S1**). Moreover, both inorganic NPs are coated by surfactant molecules that would further increase their size of about 1 nm. As a result, the dimension of the inorganic particles should exceed significantly that calculated for the double layer. Despite this, both QDs and MNPs were successfully incorporated in the NVs, probably because of the flexibility of the lipid bilayer. Indeed, other studies already reported the encapsulation of inorganic nanoparticles with size comparable or even larger than that of the bilayer thickness. QDs, for instance, were incorporated into the lipid shell because of such flexibility, leading in some cases to the formation of small protrusions (Al-Jamal et al., 2008; Bothun et al., 2009; Kethineedi et al., 2013).

TEM images of the vesicles encapsulating the nanocrystals showed regular and uniform morphology with an average diameter of about 104 nm in the case of QD-loaded NVs, and of about 125 nm for the magnetic ones (**Table 2**; see also **Figure S4** for the histograms of the size distribution). **Figure 3D** shows the morphology of the lipid vesicles loaded with the organic DOP-F-DS fluorophore, whose average size is about 67 nm, comparable to that of empty liposomes: it is highly plausible that its alkyl chain is situated into the lipid bilayer (Cardone et al., 2012). The image of panel E refers to the vesicles loaded with transferrin-TRICT (a protein with 80 kDa molecular weight): the hydrophilic nature of the protein suggests an accommodation into the aqueous core, leading to an enlargement of the vesicle diameter up to 78 nm. The same sample was imaged after 2 weeks of storage in PBS at 4◦C (**Figure S2**): size and the of the vesicles were retained but it can be denoted the presence of debris, material depots on the grid bed that may be due to an initial degradation of the lipid texture. Panel F refers to the vesicles loaded with

FIGURE 3 | TEM images of nanovesicles either empty (A) or loaded with (B) QDs, (C) MNPs, (D) DOP-F-DS, (E) transferrin-TRITC, (F) DOP-F-DS, and transferrin-TRITC, (G) doxorubicin, (H) SN-38, and (I) both drugs.



*The fourth column reports the polydispersity index (PDI), while the last ones shows the values of surface charge for each type of vesicles.*

both molecules yielding slightly enlarged NVs compared to those encapsulating either DOP-F-DS or transferrin.

Next, two chemotherapeutic drugs, namely doxorubicin and SN-38, being the former dispersible in water while the latter poorly soluble and instable at physiological pH, were encapsulated into the vesicles. **Figures 3G–I** report the TEM images of the liposomes loaded with doxorubicin, SN-38, and both drugs, respectively. The size of the vesicles loaded with the individual drugs was around 70 nm in both cases, while the particles simultaneously encapsulating the two molecules displayed an average size of around 100 nm (see **Table 2**). As already mentioned, it was necessary to prolong the sonication time to 6 min when preparing vesicles simultaneously loaded with doxorubicin and SN-38. Indeed, after 4 min (that is the time needed for preparing all the other formulations) the vesicles had not formed yet, while closely attached lipid droplets of various sizes could be observed by TEM (**Figure S3**), evidencing the incomplete

formation of the spherical vesicles. The need for prolonged sonication time might be attributed to the interaction between the two drugs that might influence bending and sealing of the bilayer.

The analysis of the hydrodynamic diameter of the vesicles (third column of **Table 2** and **Figure S4**) evidenced a broadening of the size when the particles were hydrated and dispersed in PBS. As expected, the size of the empty vesicles (137 ± 11 nm) was smaller compared to that of the loaded vesicles, as already observed by the TEM analysis. The largest particles were those loaded with both drugs (245 ± 9 nm), QDs (278 ± 21 nm), DOP-F-DS, and transferrin-TRITC (298 ± 19 nm), and the magnetic nanoparticles (319 ± 24 nm). The polydispersity index (PDI) values ranged between 0.2 and 0.3 and the suspensions resulted to be very stable over time, as confirmed by DLS measurements after 7 and 14 days (samples were stored at 4◦C but they were kept for 1 h at room temperature before measurement). As shown in **Table S1**, the size of the empty vesicles and of those loaded either with the single drugs or with both doxorubicin and SN-38 remained constant and the suspension did not look altered even after 2 weeks.

The average surface charge of the vesicles was slightly negative (about −6 mV) for all the different formulations prepared, independently of the loading or the dimension of the NVs (**Table 2**).

Incorporation of the MNPs into the NVs conferred a magnetic behavior to the loaded NVs. As a proof of concept, a magnetic field (0.3 T) was applied through a magnet close to the wall of a vial containing a solution with the NVs. As expected, the nanovesicles adhered to the wall adjacent to the magnet within 2 h (**Figure S5**). The process was shown to be reversible and repeatable.

To confirm whether the different species were encapsulated in the bilayer or within the core, the fluorescence spectra of the loaded vesicles were collected. **Figure 4A**) shows the emission spectra of the organic DOP-F-DS fluorophore dissolved in DMSO (black curve), in PBS (red curve), and in the vesicles suspended in PBS (dark blue curve). The fluorescence curve of the encapsulated dye is quite similar to that of the free dye in DMSO, showing the same two peaks (at 390 and 410 nm) although slightly broadened when encapsulated. On the other hand, the DOP-F-DS fluorescence quenches completely in PBS, indicating that the dye most likely settles into the double layer when encapsulated. As shown in **Figure 4B**, the fluorescence spectrum of transferrin-TRITC loaded into the vesicles is comparable to that of the free protein dissolved in PBS.

SN-38 is poorly soluble in aqueous media; indeed, the fluorescence curve of the molecule dissolved in PBS was considerably different from that in DMSO (**Figure 4C**). In this solvent (dark green curve), the spectrum displayed two peaks at 405 and 425 nm, respectively; while in PBS (gray curve) the fluorescence shifted to 555 nm, likely due to spatial arrangement and π-π interactions among the molecules. On the other hand, the fluorescence curve (light green) was slightly shifted toward higher energies when it was encapsulated into the vesicle, with the two peaks detected at 387 and 417 nm, respectively. Therefore, it is reasonable to suppose that, similarly to the DOP-F-DS dye, SN-38 is also hosted into the lipid bilayer of the vesicles.

The main emission peak at 595 nm of free doxorubicin in PBS (dark red curve) was retained when encapsulated into the vesicles (orange curve), as shown in **Figure 4D**. The fluorescence spectrum in PBS of the nanovesicles loaded with the green-emitting QDs was also detected (**Figure S6**) and the obtained curve was similar to that of the free surfactant-coated nanocrystals. However, a slight broadening of the maximum peak was noted, again suggesting encapsulation of the QDs into the lipid bilayer, as also expected since they are not soluble in aqueous medium and their fluorescence would be quenched in this solvent.

### Drug Loading and Release From the Lipid Nanovesicles

The optical spectra of doxorubicin and SN-38 were used to monitor the amount of drug incorporated into the NVs and to calculate the encapsulation efficiency (EE). As expected, EE for doxorubicin was dependent on the concentration of the feeding solution, reaching a maximum of about 6% when using an 85.8µM initial solution. EE dropped to 3.5 and 2.7% when using lower initial concentrations (42.9 and 21.5µM, respectively). On the other hand, higher EE values were generally obtained with SN-38, although a decreased efficiency was observed (40, 30, 20, and 15%) by increasing the concentration of the feeding solution (0.5, 1, 2, and 3µM, respectively).

Two parameters must be considered to explain the different behavior of the two drugs, i.e. the different solubility and the preparation conditions of the vesicles. SN-38 has a poor solubility in PBS and tends to intercalate spontaneously within the double layer given the greater affinity for the lipophilic environment. On the other hand, doxorubicin is perfectly soluble in PBS and the encapsulated amount mainly depends on the maximum capacity of the vesicle core.

The in vitro release kinetics of the two drugs based on the NVs with the highest EE were also studied. As reported in **Figure 5**, the release was monitored up to 5 days keeping the samples at 37◦C and at two pH values, 7.4 and 4.5, the latter being resembling the condition in the intracellular endolysosomes. As expected, the different solubility of the two drugs also led to completely different patterns. In the case of SN-38 the highest release was observed at physiological pH, reaching 35% after 24 h and 57% after 120 h. The release resulted considerably reduced at lower pH, reaching a maximum of about 30% after 5 days.

On the other hand, there was a quicker and more significant release of doxorubicin at acidic conditions as compared to physiological ones. Indeed, after 24 h incubation at 37◦C the release was around 35% at pH 7.4 and almost doubled (68%) at pH 4.5. The highest release was detected after 5 days in acidic solution, reaching 83% of the total encapsulated amount.

## Cellular Studies

#### Viability Assays

Cellular viability studies were performed to assess the effectiveness of the NVs loaded with the chemotherapeutic agents and to evaluate the effects of the empty ones. Two human cell lines derived from breast cancer, MDA-MB-231, and MCF-7, were used.

Two viability assays, namely the metabolic MTT and the Trypan blue exclusion assay, were performed administering the cells with the vesicles encapsulating the drugs, either alone or simultaneously. Additionally, the viability of the cells incubated with the free drugs was also evaluated. Based on our previous studies, the drug amount administered to the cells was fixed

to 0.5µM for doxorubicin and 0.1µM for SN-38 (Deka et al., 2011; Elbialy and Mady, 2015; Fang et al., 2018). These values are consistent with average IC<sup>50</sup> values from the literature for MDA-MB-231 and MCF-7 cells, being between 1.7 and 0.2µM that for doxorubicin and between 0.3 and 0.02µM that for SN-38 (https://www.cancerrxgene.org/; Jandu et al., 2016; Lin et al., 2019).

As evidenced in **Figure 6**, the empty NVs displayed negligible toxicity according to the MTT assay, reaching a viability close to 80% after 5 days. At the administered drug concentration, the nanovesicles loaded with doxorubicin led to a reduction of viability up to 60% in both cell lines after 120 h, while the toxic activity of the free drug resulted to be very limited even after 5 days, with viability values close to 70%. In the case of SN-38, after 24 h the viability of both cell lines dropped to 70% when encapsulated, while it settled at 80% when administered as free molecule.

Encapsulation of the drugs into the vesicles increased significantly their effectiveness, likely thanks to a facilitated uptake from the cells. This effect was enhanced when both drugs were encapsulated within the vesicles, and it became more pronounced with increasing incubation times. Indeed, the combined toxic effect led to a reduction in cell viability to values below 50% after 48 h in both cell lines, and down to 26 and 22% after 5 days in MDA-MB-231 and MCF-7, respectively. On the other hand, when doxorubicin and SN-38 were administered simultaneously to the cells as free formulation (at the same concentrations loaded into the NVs), viability reached values close to 75 and 45% after 24 and 120 h, respectively. The increased toxicity of the vesicles loaded with both drugs became statistically significant (p < 0.005) in both cell lines only after long incubation time (5 days). These findings suggest that when the encapsulated drugs are internalized by the cells, they are not immediately available to exert their activity but they are released gradually, as confirmed by the drug release kinetic showed in **Figure 5**. The limited activity shown by the combination of the free drugs probably depend on the partial degradation of the molecules, that is prevented when they are encapsulated, as well as on the existence of several drug resistance mechanisms in breast cancer cells, as already reported (Jandu et al., 2016). These effects, (i.e., the protective role of the NVs, the facilitated internalization, and the sustained release, show the advantage of using the vesicles for delivering the drugs). However, from **Figure 6** it is evident that the pharmacological activity of the drug loaded NVs is poor in the first 48 h post administration. Several factors, such as surface chemistry of the NVs, physico-chemical properties of encapsulated drugs, and cells used, can contribute to this effect thus affecting internalization of the carrier, drug release kinetic, and activation of drug efflux pumps. To determine whether the NVs loaded with both drugs induced a synergic cytotoxic effect, the CI was calculated according to the Chou and Talalay's formula (Chou and Talalay, 1984). A value of 0.97 was obtained, suggesting a slight synergic effect.

The data obtained by the vital Trypan blue assay (**Figure S7**) are in agreement with those obtained by the MTT assay, confirming a progressive reduction of the number of viable cells upon incubation with the nanovesicles loaded with the drugs, reaching the highest mortality when both drugs were hosted into the lipid vesicles. Indeed, the toxicity of the NVs loaded with both drugs is statistically significant (p < 0.005) in both cell lines after 120 h. This assay also confirmed the biocompatibility of the empty vesicles, that did not impact the number of vital cells; their number, in fact, remained close to that of the control samples after 24 and 48 h, and it still maintained values above 80% after 5 days in both cell lines.

It is worth reminding that the cellular target of doxorubicin and SN-38 is the nucleus. Although their mechanism of action is not fully understood yet, they both inhibit DNA replication and translation, doxorubicin by intercalating into DNA, and inhibiting topoisomerase II, while SN-38, the active form of

irinotecan, binds to the topoisomerase I-DNA complex (Kawato et al., 1991; Yang et al., 2014). The results obtained by the cellular assays presented in this section show that the drug-loaded NVs affect cell viability of the two cell lines, but they do not shed light on the cellular pathways and the molecular mechanisms involved in this process. Deeper investigation of these effects deserves further studies that, however, are beyond the aims of this work.

#### Cellular Internalization

To elucidate the internalization process of the nanovesicles and their intracellular localization, cells doped with NVs loaded with fluorescent molecules were analyzed by optical imaging.

In detail, MDA-MB-231 cells were incubated with NVs hosting the DOP-F-DS fluorophore, contained in the lipid bilayer, and the transferrin-TRITC protein, hosted into the vesicle core. After 6, 24, and 48 h the cells were fixed and imaged by CLSM (**Figure 7**, upper panels). The vesicles were already internalized and located into the perinuclear endosomes after 6 h. Complete overlap between the DOP-F-DS signal (blue channel) and the transferrin-TRITC signal (red channel) could be observed, indicating that the vesicles did not yet undergo enzymatic degradation. However, a significant weakening of the fluorescence in the blue channel was observed after 24 h, likely due to initial partial degradation of the bilayer structure, with consequent release of DOP-F-DS and quenching of its fluorescence in aqueous medium. This effect was enhanced after 48 h, when the blue fluorescence almost disappeared. On the other hand, the TRITC-derived fluorescence, which was initially confined into the endosomal vesicles with the typical point-like fluorescence (after 6 and 24 h), seemed to diffuse into the cell cytoplasm after 48 h.

The release of molecules and drugs from the endosomes because of a temporary permeability of the organelle membrane is an already described phenomenon (Singh et al., 2004; Muripiti et al., 2018; Quarta et al., 2019). It can depend on many factors, such as size, molecular weight, charge, and solubility of the molecules that diffuse out and it is triggered by environmental parameters, such as pH and the presence of charged species.

The same CLSM experiment was also performed on MCF-7 cells, confirming the same internalization pathway (**Figure 7**, lower panels). However, despite the similar behavior, internalization seemed to be improved with the MBA-MB-231 cells. After few hours, the two fluorescence signals colocalized into vesicular structures, while the red signal of transferrin started to spread into the cytosol and the blue fluorescence tended to disappear after 24 h.

To confirm the intracellular compartmentalization of the NVs, a colocalization study of the NVs loaded with Transferrin-TRITC was performed with LysoTracker Green in MDA-MB-231 cells (**Figure S8**). Noteworthy, the distribution of the red fluorescence signal of transferrin changed over time: after 6 h some orange spots (due to the overlap of the two fluorescent signals) were visible in the perinuclear region of neighboring cells. After 24 h incubation, the endolysosomes that entrapped the NVs seemed enlarged. Then, after 48 h, the red fluorescence of transferrin started to spread into the cytoplasm (a red background was visible inside some cells) due to its release from the endolysosomal compartment.

On the other hand, when the cells were incubated with the QD-loaded nanovesicles, the fluorescence spot was localized into the endosomes at all the analyzed time points. As shown in **Figure S9**, the nucleus of the cells was labeled with DAPI while the cytoskeleton with phalloidin-TRITC, and the fluorescence signal of the QDs could be detected into perinuclear endosomes even after 48 h. The detection of fluorescence spots after 2 days might be due to continuous uptake of new nanovesicles inside the cells. In fact, once internalized into the endolysosome compartments, the NVs are probably slowly digested leading to fluorescence quenching. Indeed, the QDs, coated only by the lipophilic surfactant layer, are neither soluble nor stable in aqueous environment and their fluorescence is detectable until the vesicles remain intact, while it is quenched when the lipid bilayer is degraded. Moreover, it should be observed that the fluorescence intensity of the QDs signal did not increase over time inside the cells, as instead observed in the case of transferrin-TRITC (due to the progressive internalization of the NVs).

#### Lipidomic Analysis

To better understand the behavior of the drugs and the influence of the vesicular formulation, a metabolomic study was performed through <sup>1</sup>H NMR on the lipid extracts obtained from the MDA-MB-231 cells. Analysis were performed on the cells treated for 24 h with doxorubicin and SN-38 either free or administered through the nanovesicles. MDA-MB-231 extracts of the untreated cells were taken as control reference. The study was performed after 24 h of incubation, as confocal microscopy imaging and MTT assay already showed noteworthy internalization and physiological effect of the drugs after this time; while MDA-MB-231 cells were considered, as NVs internalization was shown to be greater with this cell line compared to the MCF-7 (see **Figure 7**). Multivariate statistical analysis (MVA) was then performed on the registered NMR spectra to check if any differences could be observed between the two types of drug administration. This approach has already proved to be a valuable tool for understanding the mechanism of action of drugs and how they influence normal physiological patterns, as well as for discriminating healthy cells from cancerous ones, both in vitro and ex vivo/in vivo, thus paving the way toward personalized medicine (Armitage and Southam, 2016; Del Coco et al., 2019; Giudetti et al., 2019).

The OPLS-DA model was able to discriminate between control and cells treated with either free or encapsulated drugs with a R 2 X of 0.766 and 0.813, respectively (**Figures 8A,B**). Moreover, some differences were also observed between the two types of administration (**Figure 8C**). Treatment of the MDA-MB-231 cells with the two types of drug administration induced similar metabolic changes in the control cells, as observed in the profiles of the corresponding S-line plots (**Figures 9A,B**). The drugs caused a significant reduction of the total fatty acids (peaks at about 0.8 and 1.2 ppm) and a concomitant increase of the unsaturated fatty acids (peak at about 1.6 ppm). The peak at about 2.8 ppm, corresponding to the bis-allylic protons

of polyunsaturated fatty acyl chains, and that one at 4.1 ppm, relative to vinylic protons in β to the carboxylic acid of fatty acids, also showed increased intensity in treated cells, although to a minor extent compared to the previous signals. A similar behavior was already observed in cells treated with antitumoral drugs and it has been related to an increased mobility of the cellular membrane, as a consequence of the reduced tight packing of the fatty acids. Eventually, this destabilization increases with time and finally ends up with the membrane disruption, resulting in cellular death and formation of lipid droplets (Hakumäki et al., 1999; Griffin et al., 2003).

The profile of the two curves (**Figures 9A,B**) looks very similar, indicating that the mechanism of action of the drugs does not change significantly depending on the mode of administration. It has to be observed, however, that phosphatidylcholine derivatives (peak at about 3.4 ppm) increased more significantly in cells treated with the free drug, but not in those treated with the loaded NVs. This

effect might be due to a selective release operated by the nanocarrier in particular regions of the cells and it deserves further investigation.

To better appreciate any differences between the two administration methods, the MVA was also performed directly on these two classes (i.e., free drugs and NVs, **Figure 9C**). This analysis confirmed what already observed by the independent comparison vs. the untreated cells, (i.e., that they induce similar effects, but it also evidenced how the amount of unsaturated fatty acids is much more abundant in the cells treated with the drug-loaded nanovesicles). This is probably due to the higher concentration of drugs able to enter the cells when administered through the nanovesicles (as already confirmed by the confocal microscopy study showing that liposomes were already internalized into vesicles after just 6 h), thus reaching more efficiently the site of action.

As a further control, untreated cells were compared to cells administered with empty NVs, thus allowing to confirm the observed cytotoxic effect is due exclusively to the drugs (**Figures 8D**, **9D**). Being just composed of cholesterol, ceramide, and phospholipids it is not expected that this type of vectors would induce any cytotoxicity, at least at the administered concentration. In fact, cells doped with the empty NVs showed to have higher amounts of cholesterol and total fatty acids (peaks at 0.8 and 1.2 ppm), but lower quantities of unsaturated ones (peak at 1.6 ppm), as observed in the corresponding S-line plot (**Figure 9D**) which is quite specular to those obtained in cells after drug administration (**Figures 9B,C**).

The use of NVs thus leads to a higher efficacy of the treatment, and the efficiency of the system is even more remarkable if we consider the protection of the drugs against degradation and the cumulative therapeutic effect of the sustained release over 5 days.

#### CONCLUSIONS

As witnessed by the number and relevance of publications in the field, lipid-based nanostructures still gain great attention thanks to their biocompatible and degradable nature, and to their similarity with the vesicles released by the cells in physiological and pathological conditions. Despite the high production cost and the evidence of interactions with the immune system upon in vivo injection, several liposomal formulations, such as Onyvide and Doxil, were approved for therapeutic use in cancer.

In this study, lipid nanovesicles composed of phospholipids, cholesterol, and ceramide with an average size of 100 nm were prepared via ultrasonication method. Several species with completely different characteristics were encapsulated into the vesicles, including a small amphiphilic organic fluorophore (DOP-F-DS), a 60 kDa protein, lipophilic (SN-38), and hydrophilic (DOXO) drugs, and also fluorescent, and magnetic inorganic nanoparticles. The morpho-structural features and

#### REFERENCES


the thermal properties were studied, showing that this system presents high flexibility and good stability. Viability studies and lipidomic analysis demonstrated that the two drugs induced similar metabolic changes independently of the administration method (either free or encapsulated) but they were significantly more cytotoxic into the nanoformulation, probably because of facilitated entrance, protection against degradation, and sustained release. Imaging of the nanovesicles loaded with two fluorescent organic probes showed that, once engulfed by the cells and internalized into endosomes, the vesicles undergo gradual degradation over time with consequent release of the cargo in the cytosol, first of the lipophilic compounds (loaded into the bilayer) and then of the hydrophilic ones (loaded into the core). The data collected so far proved the excellent versatility of this system as imaging and delivery nanoformulation with potential theranostic applications. Nevertheless, further investigation is needed to optimize drug encapsulation capability (i.e., by using longer fatty acid chains as well as different lipid ratios, and to study their efficacy in vivo).

### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

#### AUTHOR CONTRIBUTIONS

AZ and AQ designed and carried out the experiments. LBi, AC, and LBl methodology and investigation. CP and AR: data analysis, writing, and editing. All authors contributed to discussion of the results.

#### FUNDING

This study was supported by the following grants: Progetto Tecnopolo per la Medicina di precisione, Deliberazione della Giunta Regionale n. 2117 del 21/11/2018, Progetto regionale Lab on a Swab (cod. OTHZY54), and project HApECOrk, funded by Fondazione Con il Sud (project 2015-0243).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2020.00690/full#supplementary-material

Armitage, E. G., and Southam, A. D. (2016). Monitoring cancer prognosis, diagnosis and treatment efficacy using metabolomics and lipidomics. Metabolomics 12:146. doi: 10.1007/s11306-016-1093-7


dysregulation-associated disease. Chem. Biol. Interact. 240, 220–238. doi: 10.1016/j.cbi.2015.09.005

**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 © 2020 Zacheo, Bizzarro, Blasi, Piccirillo, Cardone, Gigli, Ragusa and Quarta. 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.

# Melanoma Peptide MHC Specific TCR Expressing T-Cell Membrane Camouflaged PLGA Nanoparticles for Treatment of Melanoma Skin Cancer

#### Edited by:

Stefano Leporatti, Institute of Nanotechnology (CNR), Italy

fbioe-08-00943 August 9, 2020 Time: 12:2 # 1

#### Reviewed by:

Paolo Bigini, Mario Negri Institute for Pharmacological Research (IRCCS), Italy Pradipta Ranjan Rauta, The University of Texas MD Anderson Cancer Center, United States

\*Correspondence:

Jon A. Weidanz weidanz@uta.edu Kytai T. Nguyen knguyen@uta.edu †These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology

Received: 07 May 2020 Accepted: 21 July 2020 Published: 11 August 2020

#### Citation:

Yaman S, Ramachandramoorthy H, Oter G, Zhukova D, Nguyen T, Sabnani MK, Weidanz JA and Nguyen KT (2020) Melanoma Peptide MHC Specific TCR Expressing T-Cell Membrane Camouflaged PLGA Nanoparticles for Treatment of Melanoma Skin Cancer. Front. Bioeng. Biotechnol. 8:943. doi: 10.3389/fbioe.2020.00943 Serkan Yaman<sup>1</sup>† , Harish Ramachandramoorthy1,2† , Gizem Oter<sup>1</sup> , Daria Zhukova<sup>1</sup> , Tam Nguyen<sup>1</sup> , Manoj K. Sabnani<sup>3</sup> , Jon A. Weidanz<sup>3</sup> \* and Kytai T. Nguyen1,2 \*

<sup>1</sup> Department of Bioengineering, The University of Texas at Arlington, Arlington, TX, United States, <sup>2</sup> Joint Bioengineering Program, The University of Texas Southwestern Medical Center, Dallas, TX, United States, <sup>3</sup> Department of Biology, University of Texas at Arlington, Arlington, TX, United States

Melanoma is one of the most aggressive skin cancers, and the American Cancer Society reports that every hour, one person dies from melanoma. While there are a number of treatments currently available for melanoma (e.g., surgery, chemotherapy, immunotherapy, and radiation therapy), they face several problems including inadequate response rates, high toxicity, severe side effects due to non-specific targeting of anticancer drugs, and the development of multidrug resistance during prolonged treatment. To improve chemo-drug therapeutic efficiency and overcome these mentioned limitations, a multifunctional nanoparticle has been developed to effectively target and treat melanoma. Specifically, poly (lactic-co-glycolic acid) (PLGA) nanoparticles (NPs) were coated with a cellular membrane derived from the T cell hybridoma, 19LF6 endowed with a melanoma-specific anti-gp100/HLA-A2 T-cell receptor (TCR) and loaded with an FDA-approved melanoma chemotherapeutic drug Trametinib. T-cell membrane camouflaged Trametinib loaded PLGA NPs displayed high stability, hemo- and cyto-compatibility. They also demonstrated membrane coating dependent drug release profiles with the most sustained release from the NPs proportional with the highest amount of membrane used. 19LF6 membrane-coated NPs produced a threefold increase in cellular uptake toward the melanoma cell line in vitro compared to that of the bare nanoparticle. Moreover, the binding kinetics and cellular uptake of these particles were shown to be membrane/TCR concentration-dependent. The in vitro cancer killing efficiencies of these NPs were significantly higher compared to other NP groups and aligned with binding and uptake characteristics. Particles with the higher membrane content (greater anti-gp100 TCR content) were shown to be more effective when compared to the free drug and negative controls. In vivo biodistribution studies displayed the theragnostic capabilities of these NPs with more than a twofold increase

in the tumor retention compared to the uncoated and non-specific membrane coated groups. Based on these studies, these T-cell membrane coated NPs emerge as a potential theragnostic carrier for imaging and therapy applications associated with melanoma.

#### INTRODUCTION

Melanoma is one of the most common cancer types in the United States. It is the most aggressive type of skin cancer and influences people of all ages. Heavy exposure to ultraviolet (UV) radiation, from sunlight or the use of indoor tanning devices (classified as carcinogenic by the International Agency for Research on Cancer) are some of the major causes of melanoma. Risk is also increased for people who are sun-sensitive and those with a weakened immune system (Hayes et al., 2016; Wheatley et al., 2016; Street, 2019). Although many advances have been made for the treatment of aggressive melanoma stages, including improved chemotherapy, targeted therapy, immunotherapy, radiotherapy, and the combinations of the two (Hauschild et al., 2003), most advanced melanoma cases are incurable. Chemotherapy and targeted therapy involve the administration of drugs that often stop DNA transcription or block intracellular signaling pathways in the hope of inhibiting cell growth and replication. In chemotherapy, drugs such as dacarbazine, cisplatin, vinblastine, and temozolomide are widely used. A novel mitogen-activated extracellular signal-regulated kinase (MEK) inhibitor, Trametinib, was approved by the United States Food and Drug Administration (FDA) in May 2013 as a single agent for the treatment of BRAF V600E/K mutant metastatic melanoma (Lugowska et al., 2015). While there are a wide number of therapies and drug combinations available for the treatment of melanoma, a key limitation in the efficacy of chemotherapy is the severe side effects and the development of multidrug resistance during prolonged treatment. Chemotherapy and the listed traditional treatments, however, are often accompanied with insufficient response rates and numerous severe side effects owing to the low efficacy and non-specific targeting mechanisms of drug delivery (Lugowska et al., 2015).

One particular focus of interest in immunotherapy encompasses technologies such as cell therapy. Immune cells are isolated from immunosuppressed cancer patients and reprogrammed in vitro against cancer antigens (specific to the cancer present in the patient). Current research has mostly focused on the modification of dendritic cells and T-cells, with T-cell based cancer immunotherapy having been used effectively in cancer treatment (Tran et al., 2014). T-cells are collected from patient's own body and engineered to generate specific receptors on their surface. These receptors are referred to as chimeric antigen receptors (CARs), which are a specific type of protein located on tumor cells and provide the T-cells with a way to recognize and kill cancer cells (Zhao and Cao, 2019). Although these methods have been proven to be effective, they are accompanied by several limitations. Along the high costs and complexities of isolation, the long-term presence of re-programmed cells in the body poses some major concerns. The possible risks of long-term survival of re-programmed cells have included the development of autoimmune disease and can subsequently lead to death (Sharpe and Mount, 2015; Charrot and Hallam, 2019; Stoiber et al., 2019).

Nanoparticle-based drug delivery systems have recently gained extensive attention for cancer detection and therapy. Nanoparticles possess several advantages, including increased stability, enhanced carrier capacity, varied feasible methods of administration and the ability to incorporate both hydrophilic and hydrophobic types of drugs (Pinto-Alphandary et al., 2000). Nano-vehicles often serve to protect the drug and can be customized to release the drug in a sustained fashion. Sustained

kinetics could lead to enhanced drug bioavailability at the cancer site and reduced toxicity to healthy tissues. Several currently studied, nano-based drug delivery platforms for melanoma include polymeric carriers, liposomes, polymersomes, carbonbased nanoparticles, and protein-based nanoparticles (Cai et al., 2014). Polymer based nanocarriers are widely studied due to their long-term stability and ability to provide a sustained drug release over prolonged periods of time. Current FDA approved examples of polymers include PLGA (polylactic-co-glycolic acid) and PLA (polylactic acid). Even though such polymers are biocompatible, they are still far from complete immune evasion (Danhier et al., 2012). To overcome the limitations of current drug delivery and immunotherapy methods, numerous novel modifications have been studied over the last few decades in the hopes of giving polymeric nano-carriers the ability to escape the immune surveillance. Several of these modifications include PEGylation, plasma-surface modification and lipid-based or cell membrane masking. In addition, with the emergence of nanotechnology, the use of bio/nano-carriers is widely expected to alter the landscape of cancer treatment. For example, cell membrane-based drug delivery applications have raised an interest due to the ability of providing better "stealth" properties for foreign biomaterial-based vehicles to remain unnoticed in the face of immune cells. Several of the stealth functionalities of synthetic and biopolymers are used to enable prolonged pharmacokinetics and improve biodistribution of the particles.

In this study, we have hypothesized that membranecoated nanoparticles present a comprehensive evasion strategy against the multi-faceted nature of immune clearance and a cancer cell specificity approach. Cell membrane coated biomimetic nanoparticles have made an impressive contribution to the improvement of cancer therapy (Fang et al., 2014) due to the cell membrane structure and retained cellular antigens. Biomimetic NPs carry special advantages, such as long blood circulation, ligand recognition, immune escape, homotypic targeting, and the ability for a sustained drug delivery (Tan et al., 2015). Recently, cell membrane coated nano-systems with unique features and functions have been tested and proved to have higher circulation time and immune evasion (Parodi et al., 2013; Hu et al., 2015). In this study, we have used a hybridoma T-cell with specific targeting against the melanoma cancer. The hybridoma 19LF6 expressed gp100 antigen-MHC molecule recognizing moiety. This cell membrane would be extracted and coated on PLGA nanoparticles encapsulating dyes (coumarin-6 for in vitro studies/lipophilic DiD for in vivo studies)/Drug (Trametinib) wherein serving as a highly melanoma specific carrier for diagnosis/treatment, respectively. The presence of anti-gp100 TCR on T-MNPs would specifically target gp100 presenting melanoma cells and enhance cellular uptake of T-MNPs. The chemo-drug, Trametinib, was selected due to its effectiveness against cell lines carrying a V600E BRAF oncogenic mutation acquired by DM-6 and 1520 cancer cell lines. Trametinib is an FDA approved MEK (MEK 1 and 2) that can also be used in combination with other approved anti-melanoma drugs. PLGA polymer was chosen for its excellent properties, including biocompatibility, biodegradability, FDA and European Medicine Agency approval, adaptability to both hydrophobic and hydrophilic drugs, and sustained drug release kinetics. Properties of T-MNPs, including binding affinity with skin cancer cells, cellular uptake, therapeutic efficacy, particle retention, and biodistribution, were evaluated using cell cultures and animal models. Their results were also compared to nonspecific membrane coated (A-MNPs and D-MNPs) and naked PLGA NPs (NNPs).

### MATERIALS AND METHODS

#### Materials

Poly (D, L lactide-co-glycolide) (PLGA) (inherent viscosity 50:50 with carboxyl end groups) was purchased from Akina (PolySciTech; West Lafayette, IN, United States). Poly (vinyl alcohol) (PVA, MW 30,000–70,000) and DCM (Dichloromethane) were obtained from Sigma Aldrich (St. Louis, MO, United States). Trametinib, coumarin-6, Tris-HCL, D-glucose, B-Mercaptoethanol, Phenylmethylsulfonyl fluoride (PMSF), Protease inhibitor cocktail, Triton <sup>R</sup> X-100, Dimethyl sulfoxide (DMSO) and RPMI-1640 were also received from Sigma-Aldrich. The Avanti Polar Lipids Mini Extruder Kit was ordered from Avanti Polar Lipids (Alabaster, AL, United States). RIPA buffer was purchased from Alfa Aesar (Haverhill, MA, United States). SDS-PAGE gel, and Mini PVDF transfer pack were obtained from Bio-Rad. TCR β chain (Armenia Hamster IgG) – antibody and its isotype antibody were bought from BD Biosciences. Nuetravidin biotin binding protein, Superblock solution, BCA assay kit and DNAse were obtained from Thermofisher Scientific. Formvar coated copper TEM grids were purchased from Electron Microscopy Sciences. Fetal bovine serum (FBS), 1X trypsin EDTA, Dulbecco's Modified Eagle's Medium (DMEM) and penicillin-streptomycin were ordered from Invitrogen. Biotinylated gp100-HLA-A2/B2M (gp100 refold) and biotinylated GIGL peptide-HLA-A<sup>∗</sup> 201/B2M complexes were synthesized from Dr. Weidanz's laboratory. gp100 refold heavy chain and Beta-2-Microglobulin chain were expressed in and harvested from Escherichia coli in the form of inclusion bodies. The complex was biotinylated using Avidin Biotinylation kit [Bulk BirA: BirA biotin-protein ligase bulk reaction kit]. gp100-2M (209–217) peptide was received from Genscript. bGIGL complex was synthesized in a similar fashion. DM-6, and 1520 cell lines were obtained from Dr. Weidanz's lab. T-cell hybridomas 19LF6 and DO11.10 were kind gifts from Dr. Devin Lowes (Texas Tech University Health Sciences Center) and Dr. Philippa Marrack (National Jewish Health, Denver, CO, United States), respectively.

### Synthesis of PLGA Nanoparticles

Poly (lactic-co-glycolic acid) nanoparticles (naked NPs; NNPs) were synthesized via a modified single emulsion (O/W) technique as described previously by Menon et al. (2014). Briefly, 90 mg of PLGA 50:50 and 4.5 mg of Trametinib were dissolved in 3 mL of DCM and sonicated at 30 W for 2 min to allow dispersion

of PLGA and Trametinib in the solvent. The resulting solution was added in dropwise to 15 ml of filtered 5% (w/v) PVA solution under stirring conditions. The suspension was then sonicated at 30 W for 10 min and then allowed to stir overnight to evaporate the organic solvent. The obtained nanoparticle suspension was centrifuged at 15,000 rpm for 30 min. The supernatant was saved for the drug loading evaluation, and the PLGA NP pellet was re-suspended in 3 ml of DI water followed by freeze-drying for 24 h. Nanoparticles for imaging/diagnostic techniques were synthesized by a similar procedure with coumarin-6 instead of the Trametinib.

#### Cell Lines and Culture Conditions

The cell lines used for the experiments, DM-6, 1520, A549 and 19LF6, were maintained in RPMI-1640, supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin streptomycin. DO11.10 and HDF cell lines were maintained in high glucose DMEM, supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin streptomycin. All cells were incubated at 37◦C, 5% CO2.

### Synthesis of Cell Membrane-Coated NPs (MNPs)

To harvest the cell membrane, cells were grown to confluency in T-225 cell culture flasks. Cells were isolated by trypsinization and centrifuged at 1,000 rpm for 5 min. To remove any remaining media contents, the cells were washed with cold 1X PBS and centrifuged at 1,000 g for 10 min. The resulting pellet was re-suspended in hypotonic lysis buffer (10 mm Tris-HCL, pH = 7.5) and supplemented with the ready-to-use 1× protease inhibitor cocktail. The solution was kept on ice for 20 min and then centrifuged at 1,000 g for 10 min. The pellet was resuspended in cold 0.25 × PBS and kept on ice for 20 min followed by centrifugation at 800 g for 5 min. The final pellet was collected in cold 1× PBS. The cell membrane mix was analyzed for the DNA and protein content using Nano-Drop 1000 Spectrophotometer (Thermo Fisher Scientific)/PicoGreen DNA assays and BCA assay kits, respectively. To eliminate any remaining DNA contents, DNAse reaction was performed on the cell extract by incubation with DNAase (the amount of DNAse added varied with the DNA content in the cell extract sample) for least 1 h at 37◦C.

Poly (lactic-co-glycolic acid) (PLGA) nanoparticles (naked NPs; NNPs) were coated with different cell membranes (MNPs) including 19LF6, DO11.10 and A549 by adapting a previously used technique by Hu et al. (2011). These particles were loaded with an FDA-approved chemotherapeutic drug, Trametinib, suitable for treatment of melanoma cell lines containing V600E BRAF mutation. Briefly, NNPs were re-suspended in a cell membrane solution at different NNP weight to membrane protein weight ratios (w/w): 1:0.5, 1:1, 1:2 and 1:3. The mixture was then extruded 15 times using a pre-heated Avanti Polar Lipids Mini Extruder (37◦C). The extrusion was performed using a 200-nm polycarbonate membrane. The resulting membrane coated NPs (MNPs) were dialyzed with a 100 kDa MWCO dialysis membrane for 2 h. Prior to freeze-drying, dialyzed MNP solution was supplemented with D-glucose at a final concentration of 1 mg/ml. Three types of membrane-coated NPs were created namely: (1) 19LF6 cell line (T-cells specific to melanoma) coated NPs (T-MNPs) – our treatment, (2) DO11.10 cell line (non-specific T-cells) coated NPs (D-MNPs) – control for T-cells, and (3) A549 cell line (lung cancer) coated NPs (A-MNPs) – control for other cell types.

#### Physiochemical Characterizations

Nanoparticle size, polydispersity index, and zeta potential were investigated using Dynamic Light Scattering (DLS) technique. To measure the size of the nanoparticles, NP suspension (10 µl of 500 µg/ml) was added to 3 ml of deionized water and inserted into the DLS in a compatible cuvette. To generate TEM images of MNPs, 10 µl of 250 µg/ml MNP suspension was added to plasma treated Formvar Square Mesh Copper Grids and airdried. An H-7500 TEM (Hitachi) transmission electron microscope was used to visualize the particle morphologies. Confirmation of T-cell receptor membrane protein (TCR) on the 19LF6 cell membranes coated nanoparticles was performed via flow cytometry. The particles were also stained with TCR- β chain to evaluate the presence of any TCR or TCR components on synthesized T-MNPs. Briefly, 1mg/ml of T-MNP solution was prepared in a staining buffer (0.5% BSA, 1mm EDTA in 1× PBS). 200 µl of the solutions was added with antibodies Armenian Hamster anti-TCR β and Armenian Hamster IgG antibodies at 1:100 dilution, for test group and isotype control, respectively. The solutions were then incubated for 30 min and then washed 3× times with the staining solution. The cells were collected by centrifuging at 14,000 rpm for 20 min and resuspending in 200 µl of staining buffer. Each of the groups was then analyzed using a BD Biosciences LSR II Flow cytometer without an FSC threshold to be able to detect nanoparticles.

Stability of T-MNPs with varying NP weight to membrane protein weight ratios (w/w) was evaluated by monitoring the particle size at pre-determined time-points using DLS. To observe the stability of T-MNPs, particles of different membrane ratios were incubated in 0.9% saline over 48 h. T-MNP suspensions were prepared as described above, and the size of particles was measured at different time points (0, 0.5, 1, 3, 6, 12, 24, and 48 h).

### Binding Kinetics Assay

The binding characteristics of T-MNPs were studied using ResoSens label-free optical detection system. Biotinylated pMHC complexes HLA-A<sup>∗</sup> 02:01–IMDQVPFSV (gp100209−217) and HLA-A<sup>∗</sup> 02:01–GILGFVFTL (Influenza-M, negative control) were synthesized by our group previously. In this study, T-MNPs with the highest and lowest NP weight to membrane weight ratios were tested at varying concentrations. D-MNPs were tested at the highest NP to membrane protein ratio (w/w) (1:2), and at a concentration 2× higher than the highest concentration used for T-MNPs. All samples were tested against a specific gp100-b monomer and non-specific b monomer GILG.

### Drug Loading and Drug Release Kinetics of T-MNPs

The drug/dye loading efficiency was calculated by an indirect method where the drug/dye present in the supernatant collected from the nanoparticle synthesis process was measured, and the following formula was used for loading efficiency calculation:

% loading efficiency = Amount of drug used− Amount of drug in supernatant Amount of drug used <sup>×</sup> <sup>100</sup>

The drug (Trametinib) release study was carried out for a period of 28 days. Briefly, 1 mg of NNPs and T-MNPs [NP: membrane protein weight (w/w) ratios of 1:0.5, 1:1 and 1:2] were taken at a concentration of 1 mg/ml and were incubated at 37◦C. At each pre-determined time-point, the samples were centrifuged at 14,000 rpm for 30 min, and supernatants were collected and stored at −20◦C for later analysis. The pellets were re-suspended in fresh 1× PBS and incubated for further time points. Each of the drug release aliquots was analyzed using a UVvis Spectrophotometer at 330 nm. The amount of drug released was determined against a standard curve for Trametinib.

#### In vitro Studies of T-MNPs

#### Assessment of Glycoprotein Expression (gp100) in Tumor Cells by Western Blot and RT PCR Analysis

The expression of glycoprotein (gp100) expression in melanoma cell lines, DM6 and 1520, and lung carcinoma cells A549 were evaluated by western blotting and RT (reverse transcriptase)- PCR. Cellular proteins were extracted from cancer cells using RIPA lysis buffer containing protease cocktail, and protein concentration was measured using the Pierce BCA protein assay kit (#23227, Thermo Fisher, United States). Cellular proteins (12 and 6 µg from each cell line) were separated by 10% SDS-PAGE and electrophoretically transferred onto PVDF membranes. The membranes were blocked with 5% BSA in TBST for 1 h at room temperature, followed by incubation with primary antibodies overnight at 4◦C. The membranes were washed with TBST 3× for 5 min and then incubated with HRP conjugated secondary antibodies for 1 h at room temperature. Protein bands were developed using an ECL detection system and imaged using the Chemidoc TM Imaging system.

Total RNA was extracted from DM6, 1520, and A549 cells by using RNeasy plus Mini kits (Qiagen) according to the manufacturer's instructions. Approximately 3 µg of total RNA was reverse transcribed into cDNA (High-Capacity cDNA RT Kit) and subjected to PCR in Bio-Rad Thermocycler (T100TM) with the following conditions, pre-denaturation at 95◦C for 1 min and 30 cycles of denaturation at 95◦C for 15 s, primer annealing at 57◦C for 30 s, and extension at 72◦C for 60 s. The gp100 gene was amplified using the following primers: 5 <sup>0</sup> GCTTGGTGTCTCAAGGCAACT 3<sup>0</sup> (gp100 for) and 5<sup>0</sup> CTCCAGGTAAGTATGAGTGAC 3<sup>0</sup> (gp100 rev). β-actin gene was amplified using 5<sup>0</sup> GGCACCACACCTTCTACAAT 3<sup>0</sup> and 5 <sup>0</sup> GCCTGGATAGCAACGTACAT 3<sup>0</sup> . The relative amount of each gene was normalized to the amount of β-actin and the RT-PCR result for each gene was expressed as fold change over the basal level.

#### Cellular Uptake and Therapeutic in vitro Studies

To determine the cell specific targeting function of the nanoparticle in vitro, cellular uptake studies were conducted. Briefly, 1520 (gp100 positive), DM-6 (gp100 positive) and A549 (gp100 negative) cell lines were seeded in a 96 well plate at a density of 20,000 cells/well. Coumarin-6 (C-6; fluorescent dye) loaded T-MNPs (gp100 refold specific) and D-MNPs (nonspecific to gp100 refold) at varying NP to membrane weight ratios: 1:0.5, 1:1 and 1:2, were tested. Serially diluted MNP concentrations (100, 250, 500, and 1000 µg/ml) at different NP to membrane ratios were prepared. The cells were exposed to different NP groups for approximately 2 h, subsequently washed with 1× PBS and lysed with 250 µl/well of 1% Triton <sup>R</sup> X-100. Cell lysis extracts were then analyzed for the protein content using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL, United States) and C-6 fluorescent intensity using a UV-vis Spectrophotometer (458/540 ex/em). Total protein concentration in each lysate was determined using a BSA standard curve. The uptake of the nanoparticles was calculated by normalizing the particle concentration (determined from fluorescence intensity in a lysate) in each sample with total cell protein, which correlated to the number of cells in the sample. Untreated cells were used as a negative control.

For the therapeutic efficiency of Trametinib loaded T-MNPs, melanoma cell lines, DM-6 and 1520, were used. Firstly, the effective concentration (IC50) of the Trametinib on the cell lines were determined by exposing cells seeded in a 96 well plate at density of 5,000 cells/well to a free drug of known concentrations (1.2, 2.5, 5, 10, 15, 20, 25, 50, 75, and 100 µg/mL) for a period of 72 h. After 72 h of incubation, the cell death was analyzed using MTS assays (CellTiter 96 <sup>R</sup> AQueous Non-Radioactive Cell Proliferation Assay, Promega) according to manufacturer instructions. Therapeutic efficiency of different nanoparticle groups including 1:2 NP weight to membrane weight ratio T-MNPs (melanoma-specific), A-MNPs (non-specific), and D-MNPs (non-specific) as well as free drug and NNPs (bare/naked NPs) was determined. Three different NP concentrations (0.83, 1.66, and 2.5 µg/ml for DM-6; 15, 29, and 44 µg/ml for 1520) were used based on the calculation of the IC50 studies. Cells seeded in a 96 well plate at a density of 20,000 cells/well were exposed to the described groups for 72 h, and the cell viability was determined using MTS assays.

### Cyto-Compatibility and Hemo-Compatibility

A cyto-compatibility study was performed on human dermal fibroblast cells (HDFs). T-MNPs and NNPs were re-suspended in media and added to cells at various concentrations (50, 100, 250, 500, and 1,000 µg/ml) followed by 24 h of incubation at 37◦C. After incubation, cell viability was evaluated using MTS assays. Cells without treatment and cells treated with 1% Triton <sup>R</sup> X-100 served as positive and negative controls, respectively.

Blood clotting and hemolysis property of T-MNPs were performed on fresh human blood. Briefly, for hemolysis, 200 µl of blood was added with 10 µl of the T-MNPs prepared in 0.9% saline at varying concentrations (50, 100, 250, 500, and 1,000 µg/ml). 10 µl of 0.9% saline and 10 µl of DI water was used as positive and negative controls. The tubes were incubated at 37◦C for 2 h under gentle agitation and then were centrifuged at 1,000 g for 10 min. The absorbance of sample supernatants was monitored at 545 nm using a UV-vis Spectrophotometer. For hemo-compatibility, 10 µl of T-MNPs at the various concentrations (50, 100, 250, 500, and 1,000 µg/ml) in 0.9% saline was mixed in 50 µl activated blood (0.1 M CaCI<sup>2</sup> added blood). At pre-determined time-points (10, 20, 30, and 60 min), 1.5 ml of DI water was added to all samples to inactivate blood clotting, and the samples were incubated for 5 min at room temperature. Supernatants were collected and monitored at 540 nm using a UV-vis Spectrophotometer. 0.9% saline and DI water were used as positive and negative controls.

### In vivo Biodistribution Studies of T-MNPs via Near-Infrared Fluorescence Imaging

The tumor specific targeting and accumulation of T-MNPs were assessed using a subcutaneous xenograft model of melanoma. It was established by injecting 0.5 × 10<sup>6</sup> DM-6 cells in 150 µl of Matrigel (Corning, United States) subcutaneously in athymic nude mice. Tumor volumes were checked at regular intervals and treatment started when the volume reached ∼150 mm<sup>3</sup> . For imaging of NP biodistribution in vivo, lipophilic DiD (Thermo, United States) dye was incubated with 19LF6 T-MNPs to stain membranes for 30 min and unbound dye was washed away. All nanoparticle groups were injected intravenously through the tail vein. Following, anesthetized mice were scanned at different time points using a KodakTM in vivo imaging system. After 24 h time point, the mice were euthanized under anesthesia. Tumors and organs were excised and prepared for ex vivo imaging and fluorescent analysis. Excised organs and tumors then were analyzed by their weights and fluorescent responses.

### Statistical Analysis

All results were expressed as mean ± SD performed with n = 3 for most of the experiments if not specified. Results were analyzed using either one-way or two-way ANOVA depending on experiments with p < 0.05, and the student's t-test was used to identify differences between groups. P < 0.05 was considered to be statistically significant.

## RESULTS

### Particle Size, Zeta Potential, and Morphology

The DLS results showed that the T-MNPs with 1:2 NP to membrane ratio had a particle size of about 193 ± 56 nm with a polydispersity of 0.265. The particle size of naked NPs (NNPs) was around 171.7 ± 76 nm with a polydispersity of 0.109. The particle sizes were confirmed with Transmission Electron Microscopy (**Figure 1A**). Furthermore, TEM images of T-MNPs showed that the particles were smooth and spherical with a clear core-shell structure. The T-MNPs also had a stable ZETA potential value of −36 mV compared to a −20 mV ZETA potential of the NNPs. The stability studies also showed that the T-MNPs were stable for the whole incubation period, which was denoted by no significant change in the size of the T-MNPs during the course of experiment and, therefore, there was no sign of significant aggregation (**Figures 1B,C**).

Drug loading efficiency of T-MNPs was found to be 61%. To analyze how different membrane coating ratios would affect Trametinib release from T-MNPs, the drug release kinetics of T-MNPs with different NP to membrane weight ratios (1:0.5, 1:1, and 1:2) and naked NPs were evaluated over 28 days. The rate of release kinetics differed across the three ratios of NP to membrane protein (w/w) T-MNPs. The Trametinib release rate was significantly slower and sustained for all ratios of T-MNPs when compared to MNPs, with the lowest rate of release for 1:2 T-MNPs and the highest one for 1:0.5 (**Figure 1D**). The presence of TCR β chain on 19LF6 cells and T-MNPs were confirmed by flow cytometry analysis as seen by a shift in the absorbance curve (**Figures 2A,B**).

### Binding Kinetics of T-MNPs

As shown in **Figure 2C**, T-MNPs bound onto immobilized gp100b in a dose-dependent manner with a higher binding strength observed in samples with higher concentration and higher ratio. In addition, T-MNPs at concentrations 500 and 250 µg/ml were significantly higher when compared to the binding kinetics of 1:2 D-MNPs at 1,000 µg/ml concentration.

### Western Blot and Reverse Transcription PCR

DM6 and 1520 cell lines exhibited significant expression of glycoprotein gp100. The expression level in the 1520 cell line was found to be approximately 2× higher than DM-6, whereas the lung cancer cells A549 showed no apparent gp100 protein expression (**Figure 3A**). RT-PCR identified gp100 gene amplicon (751 bp) in all 3 cell lines (**Figure 3B**). The melanoma cancer cell 1520 has demonstrated its higher expression of gp100/HLA-A2 complex protein compared to that of the DM-6 cell line previously (Weidanz et al., 2006).

### In vitro Properties of T-MNPs

The cellular uptake of NPs was evaluated on DM-6 (gp100 containing melanoma), 1520 (gp100-containing melanoma), A549 (lung cancer, no gp100 antigen) cell lines. For uptake studies, fluorescent dye (Coumarin 6) loaded NPs were used. To analyze how different NP to membrane weight ratios would affect NP cellular uptake, NPs with 1:0.5, 1:1, and 1:2 ratios were utilized. Such ratios were prepared for D-MNPs and T-MNPs. The results demonstrated that T-MNPs displayed significantly higher uptake compared to the negative control, D-MNPs (**Figures 3C–E**) in gp100-presenting melanoma cell lines. As expected, A549 did not show any selectivity/cellular uptake toward T-MNPs or D-MNPs. Such result was expected

due to the absence of the gp100 antigen on the surface. The IC50 values of Trametinib were calculated to be approximately 29.3 and 1.66 µg/ml for 1520 and DM-6 cell lines, respectively. The therapeutic potential of the Trametinib-loaded T-MNPs was evaluated on both 1520 and DM-6 cell lines. It was shown that T-MNPs had a significantly higher therapeutic efficiency on both melanoma cell lines when compared to the negative controls, specifically at IC50 and IC75 drug concentrations (**Figures 4A,B**).

#### Cyto- and Hemo-Compatibility Studies

The viability of HDF cells was evaluated after interaction with the T-MNPs and PLGA NPs (naked NPs, NNPs) at various concentrations for 24 h. Since PLGA is an FDA approved and biocompatible polymer, the cytocompatibility of both NNPs and T-MNPs were also observed in our studies. The result illustrated that T-MNPs did not show any toxicity to the cell line up to a concentration of 1,000 µg/ml, similar to PLGA NPs (**Figure 5A**). Blood clotting assay was also performed to elucidate potential blood clotting interferences by T-MNPs. The coagulation time of blood in the presence of T-MNPs was examined at different time-points: 10, 20, 30, and 60 min. Blood coagulation initiates an activation of a cascade of coagulation factors and surface mediated reactions (Smith et al., 2015). At all the tested time points, T-MNPs did not display a significantly different blood clotting pattern when compared to that of the saline control (**Figure 5B**). Furthermore, hemolysis study was performed to test T-MNPs against potential negative effects on red blood cells. T-MNPs showed hemolysis properties lower than 5% up to 1,000 µg/ml concentration.

#### In vivo Tumor Targeting and Imaging

The in vivo tumor targeting and imaging abilities of the T-MNPs were examined to demonstrate whether injected T-MNPs accumulated at the tumor site during the treatment period. In vivo imaging results showed that the T-MNPs (melanomaspecific) had better tumor accumulation than those of the control D-MNPs (non-specific) and NNPs or bare PLGA NPs (**Figure 6A**). The T-MNPs mainly accumulated in the tumor within 6 h. Furthermore, the tumor signal from the T-MNPs remained relatively constant throughout 24 h, while the overall signal of the NNP and D-MNP groups was decreased with time. In addition, ex vivo organ images revealed that

T-MNPs could efficiently accumulate at tumor tissues and avoid liver accumulation compared to those of D-MNPs and NNPs in a xenograft mouse model as shown in **Figure 6B**. The accumulation of T-MNPs was more than twice the amount of nanoparticle accumulation of D-MNPs or NNPs. This result is also confirmed with fluorescent intensity measurements from homogenized tumor tissues (**Figure 6C**). Therefore, in vivo biodistribution results showed that the T-MNP's targeting ability was confirmed via more efficient accumulation than the NNPs and D-MNPs at the tumor sites.

#### DISCUSSION

We have thus developed a theragnostic system for effective diagnosis and treatment of melanoma. The developed nanoparticle system offers great advantages compared to the conventional drug delivery systems including enhanced retention in the cancer cells and tumor sites compared to nontargeted systems. These NPs also provide controlled/sustained drug release for up to 28 days and the least/null cytotoxic behavior. The NP drug delivery system helps to reduce the cytotoxic effects of chemotherapy drugs and is predicted to minimize immune reactions as the nanoparticles are coated in endogenous T-cell membranes. The dye loaded nanoparticles will serve as an effective tool to accurately visualize the tumor site, whereas the drug loaded nanoparticles will provide the therapeutic effectiveness to treat melanoma. Thus, our results indicate the potential of our designed T-MNPs for theragnostic application to detect and treat melanoma.

In the last decade, it has been proven that cells can be used as effective drug carriers via either themselves or their membranes that can facilitate payload delivery to desired regions (Stephan et al., 2010; Tan et al., 2015). Previous works have also shown that healthy immune cells (i.e., leukocytes) are able to circulate around the body, migrate into tissues, transport through inflamed tissues and adhere to inflamed vessel walls (Muller, 2013). In this manner, cell membrane surface proteins such as ICAM's, CCLs, CXCLs, TCR's, and/or specific CD macromolecules become important to provide an effective drug delivery platform for biomimetic delivery (Thanuja et al., 2018). Recently, studies have shown that cytotoxic T lymphocyte membranes can deliver

their payloads to tumor regions; for example, it has been shown that payload delivery can be facilitated through the natural lymphocyte surface proteins such as LFA-1 (Zhang et al., 2017; Huang et al., 2018). Therefore, this study explored the role of target specific cytotoxic T lymphocyte membranes as a drug delivery platform.

The observed shortcomings of several drug/dye-loaded polymeric nanoparticles are their poor stability/increased aggregation, low control over the payload release rate, and rapid clearance from the body. Our results of T-MNPs showed that coating the nanoparticle with cell membranes could overcome these limitations. Physiochemical characterization of synthesized T-MNPs showed excellent colloidal stability in physiological conditions. The increase in stability due to the membrane coating onto the nanoparticles was evident by the stable Zeta potential of the T-MNPs compared to an unstable potential of the NNPs as described by the −30 mV Zeta rule (Kumar and Dixit, 2017). The negative surface charge (−36 mV) of the T-MNPs repel negative charged albumin molecules in the serum, preventing potential aggregations. The stability of T-MNPs in 0.90% saline solution was found to be similar across different NP weight to membrane protein weight (w/w) ratios for up to 2 days (**Figure 1C**). Fang et al. (2014) reported a similar observation where they showed a higher stability of the coated nanoparticle for up to 15 days across a wide range of NP to membrane ratios (1:0.5 to 1:4). The loading efficiency of T-MNPs was observed to be approximately 61%, and the drug release kinetics displayed an initial burst release followed by a sustained release up to 28 days. When comparing T-MNPs and NNPs, it can be clearly seen that cloaking of drug loaded PLGA nanoparticles with T-cell membranes reduced the drug release rates, especially those coated with higher amounts of the cell membrane. This reduction has also been observed in nanoparticles cloaked with membranes from erythrocytes (Hu et al., 2011) where erythrocyte membrane-coated nanoparticles showed slow drug release compared to uncoated nanoparticles. Hu et al. (2011) attributed this effect to the membrane's ability to

act as a diffusion barrier to provide better sustained drug release, as compared with PEG-based nanoparticles, thereby enhancing the therapeutic efficacy of the drug in acute myeloid leukemia cells. Furthermore, T-MNPs hold the unique characteristics of natural 19LF6 cell membranes thus avoiding clearance and enhancing circulation. Zhong et al. (2013) previously defined the 19LF6 cell line to contain a metastatic melanoma antigen, gp100 (209–217) – specific T-cell receptor. Our flow cytometric analysis and the binding kinetic results not only confirmed the higher presence of the specific TCR on the nanoparticle coating, but also its ability to bind to its specific target.

As part of the toxicological analysis, the T-MNPs, cyto- and hemo-compatibility were analyzed. The T-MNPs were found to be cyto-compatible up to 1,000 µg/ml concentration, which was shown to be at least as cyto-compatible as NNPs or bare PLGA NPs. This was similar to the results observed by Guo et al. (2015), who demonstrated that erythrocytemembrane coated PLGA nanoparticles possessed a similar cyto-compatibility compared to bare PLGA nanoparticles. This attributes to the cell mimicking characteristic and the slow drug release characteristics of T-MNPs. Blood clotting characteristics of T-MNPs were compared to the saline control and found to have no significant effect on the blood clotting cascade up to 1,000 µg/ml. According to the criterion in the ASTM E2524-08 standard, percent hemolysis >5% is considered toxic to red blood cells (ASTM E2524-08, 2013). T-MNPs-induced blood hemolysis was observed to be <5% up to 500 µg/ml. Although T-MNPs were found to be toxic to red blood cells at a concentration of 1,000 µg/ml, such high concentrations of the particles might not be used for later studies.

We also continued to investigate the in vitro characteristics of these membrane coated nanoparticles and their ability as a highly specific vehicle for cancer targeted delivery. In light of past studies that showed bare PLGA NPs have limitations owing to non-specific targeting and result in uncontrolled tissue distribution of the drug (Danhier et al., 2012) and that cell membrane coating improves immune evasion, target specificity and drug efficacy (Hu et al., 2011), we have devised the proposed design of T-MNPs. Fang et al. (2014), devised surfaceengineered PLGA NPs with platelet-membrane-derived vesicles since platelet cells have a natural ability to adhere to injured blood vessels as well as circulating pathogens. Such membrane coating provided the particles with natural platelet-like targeting functions. However, there is no data found in the literature for natural specific targeting (i.e., via TCR receptors) abilities of membrane coated drug carriers so far. The data obtained from our in vitro cell studies sheds light to the undiscovered potential of cytotoxic T-cell membrane coated nanoparticles as a biomimicking drug delivery vehicle. In our study, antigp100 TCR influence on T-MNP cellular uptake was observed in DM-6 and 1520 melanoma cell lines. Since A549 lung cell lines did not show any apparent gp100 presentation, A549 was used as a negative control cell line. Data collected from binding and cellular uptake studies showed that T-MNPs illustrated superior binding and uptake kinetics attributed to their antigp100 TCR. The uptake of T-MNPs at all ratios was significantly higher than that of D-MNPs, the negative control nanoparticles (**Figure 3**). These particles showed selective and effective binding to gp100 carrying melanoma cells when compared to that of D-MNPs made from the negative T-cell control. The uptake of T-MNPs increased with increasing concentrations and was enhanced in the 1520 cell line even more than that of DM-6 cell line. The enhanced 1520 uptake of the T-MNPs might have been due to its ability to better present gp100/HLA-A2

complex protein compared to that of the DM-6 cell line (Weidanz et al., 2006).

absorbances were quantified using a UV-vis Spectrophotometer (n = 9).

Furthermore, therapeutic efficiency of T-MNPs was evaluated. The IC50 values of Trametinib for DM-6 and 1520 were found to be approximately 29 and 1.6 µg/ml, respectively, which were very much similar to the values observed in a previous study by Roller et al. (2016). However, the IC values often depend on numerous human and in-house factors, therefore, they must be performed for each study. T-MNPs were found to be significantly more efficient in killing melanoma cancer cells than any of other groups compared, even much more than free Trametinib at the same concentration, which would attribute to the binding and uptake characteristics of T-MNPs. Particles with a higher membrane content (greater anti-gp100 TCR content) showed to be more effective when compared to that of the lower NP to membrane ratio. These results indicate that proposed membrane coated natural targeting nanoparticles could potentially be used to improve chemotherapeutic efficacy to effectively treat melanoma.

Following the in vitro characterizations, the preliminary in vivo investigation by the biodistribution study examined whether intravenously injected T-MNPs targeted and recruited/retained at the subcutaneous tumors in the tumor implanted mice. In biodistribution studies, comparing between T-MNPs and other study groups, expected tumor accumulation difference was observed. NPs from almost all groups accumulated mainly at the liver and spleen in 2 h. The T-MNP group showed distinctive accumulation in the tumor region at almost a threefold higher accumulation on the tumor site than D-MNPs and about a twofold increase in accumulation than NNPs. D-MNPs and NNPs, on the other hand, were captured more in the liver. Similarly, Zhang et al. (2017) used non-specific hCTL membranes to target gastric cancer combined with low dose irradiation and observed very similar results in in vivo targeting capabilities of hCTL membrane coated nanoparticles. In their study, membrane coated nanoparticles showed a gradual increase in the tumor sites after low dose irradiation exposure.

However, they reported that low dose irradiation was helped for these NPs to accumulate on target tissues at later time points of the study. Surprisingly, they were still able to confirm that non-specific hCTL membrane coated nanoparticles accumulated on tumor tissues without LDI and targeting molecule (Zhang et al., 2017). This effect might be due to the natural adhesion molecules like LFA-1 or integrin of the T-cell membrane surface. In our in vivo biodistribution study, anti-gp100 TCR decorated T-cell membranes were able to accumulate on target tissues. What is more to the above-mentioned literature results is tumor tissue accumulation of TCR decorated T-MNPs was superior compared to all control groups from the very early time points. All in all, our proposed T-cell membrane coated drug carrier system showed superior in vitro and in vivo targeting and uptake capabilities. Therefore, such natural biomaterials as engineered cells, bacteria membranes, biocompatible proteins, viral capsids

\*Statistically significant with P < 0.05.

and others in our drug carrier design can take the theragnostic field further than its current capabilities without compromising the effective drug delivery requirements.

Although cell membrane coated nanocarriers have great potential to deliver drugs to the desired location and could be a promising carrier to improve the theragnostic outcomes in treating melanoma, there are a few associated limitations and challenges. For instance, the cell membrane is comprised of a lot of different protein or peptide types, some of them are required for targeting, evading immune response while the other abundant proteins have unknown interactions in the host environment (Klammt and Lillemeier, 2012). Thus, further in-depth immune response and toxicity profiles must be performed for individual membrane proteins. Cell membrane isolation procedures are not robust and are limited to particular laboratory settings which can be a challenge in clinical translation such as isolation and

culture of abundant T-cells in a short duration. Therefore, quality control such as maintaining the functional and structural aspects of cell membranes for longer periods needs to be investigated. Future work should include the required studies to address the above-mentioned limitations of the proposed research. Yet a new approach, utilization of T-MNPs that have been formulated using donor cells and investigating their drug delivery potential as a translational medicine application might be beneficial toward personalized cancer therapies in the future.

#### CONCLUSION

Overall, we successfully developed T-cell coated nanoparticle carriers that displayed superior targeting capabilities toward skin cancer cells and that could serve as a potential tool as a theragnostic system to image and treat melanoma. T-MNPs maintained excellent in vitro targeting ability and had a biomimicking shell that minimizes toxicity and systemic clearance concerns in the conventional drug carrier designs. The natural targeting molecule TCR receptor on the surface of the T-MNPs was preserved after membrane isolation and synthesis of T-MNPs. In vitro assessments of T-MNPs also showed their therapeutic ability as a drug carrier platform. Finally, biodistribution studies showed the in vivo targeting abilities of T-MNPs. The cyto-compatibility and natural targeting T-MNPs recruited to the tumor regions and displayed a distinctive accumulation signal. We showed the therapeutic capabilities of T-MNPs including intrinsic targeting, prolonged drug release and therapeutic potential making T-MNPs an excellent biomimicking theragnostic carrier platform for future cancer therapy.

#### REFERENCES


#### DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

#### ETHICS STATEMENT

The animal study was reviewed and approved by Julia Kissling, IACUC/IBC Specialist, Office of Regulatory Services, The University of Texas at Arlington, Texas-76010.

#### AUTHOR CONTRIBUTIONS

SY conceived the original idea whereas JW and KN guided in developing the final nanoparticle design and the planned experiments. GO and DZ carried out the synthesis and characterization of designed nanoparticles and performed some in vitro experiments. HR cultured and isolated the cell membranes required for coating of nanoparticles used for some in vitro experiments and animal studies. SY, HR, TN, and MS designed and performed the in vivo (animal) experiments. SY and HR wrote the manuscript. All authors discussed the results and contributed to the final manuscript. JW and KN supervised the work and helped with the editing of the manuscript.

#### ACKNOWLEDGMENTS

The graphical abstract was created with Biorender.com.

cutaneous melanomas: long-term follow-up of survival in a randomised trial. Lancet Oncol. 17, 184–192. doi: 10.1016/s1470-2045(15)00482-9


biomimetic leukocyte membranes possess cell-like functions. Nat. Nanotechnol. 8:61. doi: 10.1038/nnano.2012.212


**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.

The reviewer PR declared a shared affiliation, with no collaboration, with the authors to the handling Editor at the time of review.

Copyright © 2020 Yaman, Ramachandramoorthy, Oter, Zhukova, Nguyen, Sabnani, Weidanz and Nguyen. 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.

# The New Frontiers in Neurodegenerative Diseases Treatment: Liposomal-Based Strategies

Mariafrancesca Cascione<sup>1</sup> \*, Valeria De Matteis <sup>1</sup> , Stefano Leporatti <sup>2</sup> \* and Rosaria Rinaldi <sup>1</sup>

*<sup>1</sup> Department of Mathematics and Physics "Ennio De Giorgi," University of Salento, Lecce, Italy, <sup>2</sup> National Research Council Nanotec Institute of Nanotechnology, Lecce, Italy*

#### Edited by:

*Christian Celia, University of Studies G. d'Annunzio Chieti and Pescara, Italy*

#### Reviewed by:

*Giovanni Tosi, University of Modena and Reggio Emilia, Italy Maria Chiara Cristiano, Magna Græcia University of Catanzaro, Italy Gert Fricker, Heidelberg University, Germany*

#### \*Correspondence:

*Mariafrancesca Cascione mariafrancesca.cascione@unisalento.it Stefano Leporatti stefano.leporatti@nanotec.cnr.it*

#### Specialty section:

*This article was submitted to Nanobiotechnology, a section of the journal Frontiers in Bioengineering and Biotechnology*

Received: *28 May 2020* Accepted: *14 September 2020* Published: *26 October 2020*

#### Citation:

*Cascione M, De Matteis V, Leporatti S and Rinaldi R (2020) The New Frontiers in Neurodegenerative Diseases Treatment: Liposomal-Based Strategies. Front. Bioeng. Biotechnol. 8:566767. doi: 10.3389/fbioe.2020.566767* In the last decade, the onset of neurodegenerative (ND) diseases is strongly widespread due to the age increase of the world population. Despite the intensive investigations boosted by the scientific community, an efficacious therapy has not been outlined yet. The drugs commonly used are only able to relieve symptom severity; following their oral or intravenous administration routes, their effectiveness is strictly limited due to their low ability to reach the Central Nervous System (CNS) overcoming the Blood Brain Barrier (BBB). Starting from these assumptions, the engineered-nanocarriers, such as lipid-nanocarriers, are suitable agents to enhance the delivery of drugs into the CNS due to their high solubility, bioavailability, and stability. Liposomal delivery systems are considered to be the ideal carriers, not only for conventional drugs but also for neuroprotective small molecules and green-extracted compounds. In the current work, the LP-based drug delivery improvements in *in vivo* applications against ND disorders were carefully assessed.

Keywords: neurodegenarative diseases, Blood Brain Barrier (BBB), liposome, solid lipid nanoparticles, nanostrured lipid particles, ethosomes, drug delivery

### INTRODUCTION

Pharmacological research is oriented to develop new effective therapeutic strategies in order to treat diseases and, at same time, to improve the life quality of patients. Many studies aim to discover new drugs, taking in consideration that their effectiveness is strictly dependent on the administration route and also on their intrinsic ability to access the organs and tissues in suitable amounts and times. The uptake of drugs is particularly difficult in specific districts of body, such as the Central Nervous System (CNS), in which the presence of a Blood Brain Barrier (BBB) constitutes the major obstacle. The BBB often inhibits neuroreparative and neuroprotective therapies that should carry out their pharmacological action directly on target site. According to the latest available data, it is estimated that 10 % of the world population will be affected by neurological disorders, without geographical and socioeconomic distinctions (Organization, 2006). For this reason, the World Health Organization recommends great efforts to overcome the difficulties in the brain drugs administration. Despite the neuropharmaceutical field constituting the largest sector of drug industry, its development is limited by the complication regarding the BBB crossing. It has been estimated that ca. 98% of the drugs available today for the treatment of neurological diseases, including recombinant proteins, monoclonal antibodies, and genes are not able to effectively overcome the BBB. This is true due to their deficiency to exploit specific transport mechanisms and their high molecular weight together with polarity (Pardridge and Boado, 2012). In light of this, the regulatory mechanisms of the BBB are fundamental for the development of new therapies dedicated to large groups of brain injuries such as neurodegenerative diseases.

#### NEURODEGENERATIVE DISEASES

The age-dependent disorders are becoming increasingly prevalent, in part because the elderly population has increased in recent years (Heemels, 2016).

Examples of neurodegenerative diseases are amyotrophic lateral sclerosis, Parkinson's disease, Alzheimer's disease, Huntington's disease, frontotemporal dementia, and the spinocerebellar ataxias. All these pathologies are different in pathophysiologic properties but, at the same time, are associated with memory loss, cognitive impairments, and other adverse effects including a person's inability to move, breathe, and speak (Abeliovich and Gitler, 2016; Wyss-Coray, 2016; Gitler et al., 2017). These pathologies show selective neuronal vulnerability combined with the degeneration of specific brain regions; in addition, an abnormal protein deposit (extracellular or intracellular) occurs in neurons or other types of brain cells (Ross and Poirier, 2004).

Alzheimer's disease (AD) is characterized by neuron degeneration; in particular, in the basal forebrain and hippocampus. In addition, the alteration of synaptic and neuronal connections is involved in the pathogenesis (Selkoe, 2002). The accumulation of amyloid-β (Aβ) peptides and tau protein boosts the development of the pathology progression (Combs et al., 2016). In detail, the Amyloid beta, a 39–43-amino acid residue peptide, is a physiological presence in human brain and is derived from the proteolysis process of the amyloid precursor protein (APP). The extracellular accumulation of amyloid-β (Aβ) peptides induces the formation of amyloid plaques, also called senile plaques, which form the synaptic dysfunction and neuronal loss (Citron, 2002). The tau proteins (T proteins) play a key role in microtubule stabilization and they are abundant in hippocampal areas, cortical areas, and the entorhinal cortex. In AD disease, this protein undergoes a hyperphosphorylation phenomenon that induces tau protein clumping with the consequence of developing intracellular neurofibrillary tangles (NFT): these formations lead to axon degeneration and microtubule disaggregation (Goedert, 2004). The onset of AD may be due to rare genetic mutations, which were found in the APP and in some transmembrane proteins (presenilins) that cleave the APP. Mutations of tau28 protein have been found in non-Parkinson frontotemporal dementia.

Parkinson's disease (PD) is provoked by degeneration of dopaminergic neurons in the substantia nigra of the midbrain and other monoaminergic neurons in the brain stem (Vernier et al., 2004). Despite several progresses, the main cause of PD remains unknown. It is probably born by complicate connections between environmental and genetic factors that are strong effects on cells that trigger mitochondrial dysfunction, oxidative stress, impairment of the ubiquitin proteosome process, as well as defective autophagy process (Dawson et al., 2010; Angeles et al., 2014). However, the exact mechanisms leading to neuronal death are unknown. As a consequence, the loss of several neurotransmitters and neuromodulators in the extranigral neurons, such as noradrenaline (NA), acetylcholine (ACh), glutamate, and serotonin (5-hydroxytryptaminc [5-HT]) occur. These phenomena induce the slowing of movement, muscular rigidity, and tremor (Alexander, 2004).

Numerous gene mutations could induce the early onset of PD, in particular, point mutations or increased gene dosage of the αsynuclein gene. These boost an autosomal dominant PD. Instead, the mutations of genes encoding for DJ-1 or PINK132 and parkin could trigger a recessive early-onset PD. In general, the adult PD onset regards an inclusion body near the nucleus of neurons, immersed in cytoplasm, called Lewy body, in the form of fibrillar aggregates. Several investigations demonstrated the presence of Lewy bodies in the substantial nigra in monoaminergic, cerebral, cortical, and other neurons principally constituted of α-synuclein protein (Stefanis, 2012).

### THE BLOOD BRAIN BARRIER (BBB)

The human brain contains about 100 billion of capillaries with a total length of 600 km. This dense vascular network covers an area of about 20 m<sup>2</sup> and represents an effective interface between the blood and the brain, supplying brain cells with oxygen and essential metabolites that support brain functions (Risau and Wolburg, 1990). In addition, the brain consumes 20% of the body's total glucose, increasing the blood flow and oxygen in order to quickly reach the different body's districts (Begley and Brightman, 2003). Capillaries are the major site of blood barrier brain (BBB) that constitutes a real barrier between the bloodstream and the central nervous system acting as selective filter, allowing or preventing substances (ions, glucose, proteins etc.) to move from the blood to the brain parenchyma and to cerebrospinal fluid (CSF). Thanks to this selectivity, the BBB protects the sickly chemical–physical homeostasis of the cerebral fluid environment carrying out a protective role toward CSF and the nervous tissue (Abbott and Romero, 1996).

This can be satisfied by keeping the ionic environment stable and preserving the low amino-acid gradient of the excitatory neurotransmitters (glutamic acids, aspartic acid, and glycine) that is characteristic of the brain extracellular fluid. This phenomenon is critical for reliable synaptic transmission and efficient neuroregulatory activity. BBB prevents the toxic compound uptake in cells, such as metabolites and neurotoxins both endogenous and xenobiotic, which are potentially fatal. In addition, the BBB promotes the longevity of the CNS and prevents premature cell death and neurodegeneration.

The BBB is formed by the endothelium of the brain capillaries, the processes perivascular of the astrocytes surrounding the endothelial cells and from the pericytes which are contractile connective cells that partially surround them. The anatomical structure of the BBB is responsible for its functional peculiarities, such as limited permeability to most substances and limited paracellular and transcellular transport (Zlokovic, 2008; Kisler et al., 2017). The brain capillaries are anatomically different from peripheral systemic ones since the cells that compose them form a continuous non-fenestrated endothelium, with a reduced number of pinocytic cells (Engelhardt and Liebner, 2014). Brain endothelial cells are connected by tight and adherent junctions. The first prevents the free diffusion of solutes from the blood sector (peripheral or systemic) to the CSF and brain (intrathecal), both at the level of the cerebral capillaries and the chorioid epithelium. The tight junctions involve occluding, claudin I, claudin 3, claudin 5, claudin 12, and the membrane-associated guanylate kinases, proteins ZO1, ZO2, and ZO3. The adherent junctions include cadherins, platelet endothelial cell adhesion molecules (PECAM1), and the junctional adhesions molecules (JAMs) JAMA, JAMB, and JAMC. The gas exchange (oxygen and carbon dioxide) occurs across the endothelium, whereas the pinocytosis deficit is responsible for the limitation of solutes passage (Abbott et al., 2010; Zhao et al., 2015).

Then, in order to target a specific drug to the brain, it is necessary to consider the structural and functional characteristics of the BBB and, at the same time, evaluate the chemical-physical properties of the drug (pKa, molecular weight, lipophilicity, etc.). In addition, it is important to consider the intrinsic ability of BBB to form bonds with the plasma proteins that prevent the CNS crossing, the degree of ionization (pH), and the lipid/water partition coefficient of the drug (Warren, 2018). In particular, it has been demonstrated that small lipid-soluble molecules (high partition coefficient) with a molecular weight <400 Da or containing <8 hydrogen bonds can cross BBB by simple transmembrane diffusion. Contrary, the drugs with low passive diffusion coefficient can be internalized in the brain only by active transports systems (Pardridge, 2015).

## TRANSPORT THROUGH THE BBB

Oxygen, carbon dioxide, glucose, nucleosides, vitamins, and some of fat-soluble drugs can cross the BBB by passive diffusion mechanisms or by specific transport mechanisms. The transport systems (**Figure 1**) can be located on the luminal or abluminal side of the BBB and can be classified into three categories (Abbott and Romero, 1996; Begley, 2004): (i) CMT (Carrier Mediated Transport) consists of a transport mediated by specific transport carrier proteins; they are able to move carbohydrates, fatty acids, amino acids, nucleotides, hormones, vitamins, organic cations and anions through BBB; (ii) AET (Active Efflux Transport) consists of an active transport mechanism capable of expelling a large variety of molecules from the brain compartment to the bloodstream; and (iii) RMT (Receptor Mediated Transport) which consists of receptors mediated transport system, able to drive large compounds (peptides and proteins) through an intracellular process in both directions: insulin and transferrin from blood to brain, as well as apolipoproteins (Demeule et al., 2008; Sagare et al., 2012). The transport of native plasma proteins or peptides is limited, but the cationization phenomenon can increase their uptake by adsorptive-mediated transport (AMT).

The CMT and AET systems are responsible for small molecule transport between the blood and the brain, while RMT allows large molecule delivery through the BBB. Among these, the transmembrane transporters called ABC transporters, expressed at the luminal side of BBB, play an important role due to the presence of two cytoplasmic domains useful to bind the ATP boosting transport against the unidirectional gradient (from cytoplasm to extracellular space) (Abbott et al., 2010; Mokgokong et al., 2014).

Therefore, ABC prevents the brain accumulation of xenobiotic agents, and drugs exert an important role in the body detoxification. The first ABC transporter identified is the P-glycoprotein (P-gp or ABCB1) coded by the MDR1 gene. This gene is known for its several polymorphisms (ca. 50) at the level of a single nucleotide, and for this reason, it is responsible for a strong individual variability in absorption and tolerance to drugs (Yan-Hong et al., 2006; Bartels, 2011). In addition to the ABC, OATP (Organic Anion Transporting-Polypeptide) and the OAT (Organic Anion Transporter) members are expressed in the endothelial cells of the BBB. The main function of OATP 1A2 and 2B1 carriers is to uptake several endogenous and xenobiotic compounds (10.1021/acs.molpharmaceut.0c00159), but, unlike the members of the ABC family, the OATP does not hydrolyze ATP, and consequently, they fail to uptake drugs against the concentration gradient (de Boer et al., 2003). Their presence allows the ions exchange according to the ionic gradient. In addition to the possibilities of exploiting or inhibiting these and other physiologically present transport mechanisms on the BBB (Pardridge and Boado, 2012), other approaches that can guarantee are being studied for a more effective delivery of drugs to the CNS. A spread class of carriers, abundant in mammalian neurocytes and brain capillary epithelium membrane, is also represented by glucose transporter proteins (GLUTs), GLUT1, and GLUT3 (McEwen and Reagan, 2004).

The approaches to achieve effective CNS drug concentrations can be invasive (temporary break of tight junctions, intracerebral injection, or use of intracerebral implants such as catheters, microchips, or erodible polymeric systems). It still comes cost-effective and potentially dangerous for patients, since the direct delivery of the drug exposes patients to the risk of developing serious brain infections with a consequent decrease in their compliance. Starting from these assumptions, the non-invasive (chemical, biological, or technological) approach is the hope to treat the brain pathologies (Scherrmann, 2002; Tosi et al., 2011).

In the last years, the use of nanoparticles (NPs) represents a powerful approach to improve the BBB penetration of bioactive molecules; NPs have small size (3 ÷ 200 nm) and good stability, and they can be easy functionalized by surface modifications (attaching active ligands or changing the superficial charge), offering good advantages over the free drug. The NPs can reduce the side effects on healthy cells, improving the pharmacokinetic and pharmacodynamics drug profile and increase drug concentration at the specific sites allowing their application in the neurodegenerative disease treatment.

FIGURE 1 | Transport mechanisms through BBB. Reproduced with permission from Cavaco et al. (2020).

### LIPID NANOCARRIERS FOR DRUG DELIVERY ACROSS THE BBB

The effectiveness of drugs acting on the CNS has strong limits, due to their inability to overcome the BBB in adequate amounts (de Boer and Gaillard, 2007). The issue of the limited drugs crossing through the BBB is of great interest to the scientific and medical community, which is focused on the execution and optimization of innovative therapeutic strategies against neurodegenerative diseases. In this perspective, the carriers with dimensions at nanoscale level represent a promising drug delivery platform solution for enhancing the drug crossing through the BBB (Chen and Liu, 2012; Pardridge, 2012; Upadhyay, 2014). Theoretically, the tuneable physicochemical properties of nanocarriers permit optimizing the delivery of different kinds of drugs at the specific therapeutic target; nevertheless, a careful design of drug nanocarriers needs to consider many factors, such as biocompatibility, reduction of drug toxic effects, enhancement in drug stability, and optimization of the specific target functionalization. Furthermore, the engineering of such nanocarriers become harder when the therapeutic target is the CNS, as listed by Lockman and colleagues (Lockman et al., 2002).

The growing interest in the suitable drug delivery systems research has encouraged the production of a wide range of nano-scaled carriers (Mishra et al., 2010). Among these, lipidbased carriers are recognized as promising strategies for drug delivery, due to their ability to mimic natural lipid environment constituting the biomembranes, including BBB (Pujara, 2012). In addition, lipid vectors protect the drug from degradation during transport to the target sites, reduce the toxicity, and increase the drug biocompatibility with respect to free administration (Chen et al., 2010; Nunes et al., 2017).

Among different lipid carriers, liposomes (LPs) have been extensively studied for nanomedicine applications (Fenske and Cullis, 2008; Fenske et al., 2008). The mechanism transport across the BBB is still unclear (Agrawal et al., 2017). Initially, it was thought that the liposomes spontaneously crossed several biological membranes including BBB by electrostatic interactions. However, this theory was replaced by the possible mechanism through active transport carriers by transcytosis or a specific receptor (Spuch and Navarro, 2011). The latter hypothesis is more consolidated due to the binding of specific molecules, such as glucose and glutathione, on liposome surfaces that are able to overcome the BBB promoting the liposome translocation (Noble et al., 2014).

Their biochemical architecture (**Figure 2**) is similar to the human cellular membrane; in fact, it is formed by phospholipid double layers that enclose an aqueous core. In these artificial vesicles, with spherical shape, lipophilic and hydrophilic drugs can be encapsulated: instead of lipophilic molecules only confined into phospholipidic bilayers, hydrophilic molecules can be loaded both inside the aqueous cores of LPs and in the external water phase. The encapsulation efficiency is almost equal to the loading rate in the case of lipophilic molecules, whereas it depends on composition and the LP synthetic approach method in the case of hydrophilic drug (Johnsson and Edwards, 2003). However, in any case it was found that drug toxicity is strongly reduced following internalization in the LP.

Conventionally, a first classification of LPs is based on the number of bilayers (lamellarity) and on the size of the structure (Woodle and Papahadjopoulos, 1989) because these two parameters deeply influence both the circulation half-life and the drug encapsulation efficiency. Small unilamellar vesicles (SUVs) and large unilamellar vesicles (LUVs) indicate LPs formed from a single layer with a dimension of (25 ÷ 50) nm or (200 ÷ 800) nm, respectively, while multilamellar vesicles (MLV) are referred to LPs formed by concentric multiple lipid layers, separated from each other by an aqueous solution. Most specific classifications regard the chemical composition, lipid organization, surface charge, or preparation methods (Euliss et al., 2006; Bozzuto and Molinari, 2015).

For the drug delivery aim, the LPs appear a promising system, not only for their ability to protect their cargo to enzymes degradation phenomenon, but also for their unique properties, including low toxicity, flexibility, biocompatibility, biodegradability, and non-immunogenicity. Unfortunately, LPs suffer of some limitations for specific therapeutic applications, namely shelf life and the drug-loading efficiency. In the last decades, several efforts have been made toward the ingegnerization of optimal LPs, and new categories of liposomal carriers were obtained.

Solid lipid nanoparticles (SLN) have established themselves as an alternative system to conventional LPs, due to significant advantages: controlled and targeted drug release, high stability, and effective encapsulation efficiency of drugs. The SLNs structures consist of a solid lipid matrix (tri-, di-, and monoglycerides, fatty acids, steroids, or waxes) that outlines an aqueous volume in which lipophilic and hydrophilic molecules can be internalized (Uner and Yener, 2007). However, some of the drugs could flow out during the transport following the natural recrystallization process of the SLN shell, formed by a single lipid. For that reason, the delivery efficacy results are jeopardized by the SLN architecture (Blasi et al., 2007).

With the aim to overcome these drawbacks, the nanostructured lipid carriers (NLC) were conceived. The matrix of the NLC structure is composed by a combination of solid and liquid lipids. This latter induces the nanostructuring process into the inner lipid shell. These results allow the increase of the volume available for drug loading and the boost of a more stable lipid configuration (Iqbal et al., 2012; Beloqui et al., 2016).

Several synthetic routes have been proposed to obtain SLN and NLCs; among these, high-pressure homogenization (HPH) and microemulsion are the most efficient. The microemulsion technique requires the use of hazardous toxic organic solvents; for this reason, the lipidic systems obtained through this procedure are unsuitable for medical purposes. In this perspective, the use of HPH procedure is preferred; however, this technique permits the internalization only of thermostable drugs (Das and Chaudhury, 2011).

The SLN/NLCs systems have been largely employed in vivo for the delivery of several kinds of drugs, by intranasal, intragastric, oral, and intravenous administrations. In order to optimize the carriers for dermal delivery, a new class of LPs has been proposed by Touitou and co-workers (Touitou et al., 2000). They modified LPs, adding ethanol, in order to obtain high efficiency in the delivery of different drugs through the skin.

These modified LPs (**Figure 3**), called ethosomes, are lipid vesicular systems composed primarily of phospholipids, water, and ethanol. This latter, added in high concentration (20 ÷ 45 %), interacts with the polar heads of the phospholipids; finally, it is inserted in the lipid stratum corneum, lowering the meltingpoint of lipidic bilayer. This last phenomenon makes them to be more fluid and permeable for the cell membrane. In addition, in comparison with LPs, ethosomes have small dimensions due to the negative superficial charge induced by ethanol inclusion.

The improved transport mechanism of bioactive molecules through the skin can be applied to the drug delivery to the CNS; in fact, the characteristics of the ethosomes could be useful to promote the BBB penetration.

Therefore, the improvement of the liposomal delivery systems has required the optimization of the physiochemical properties of the liposomal surface in relation to the chemical nature of entrapped drug molecules.

### CONVENTIONAL THERAPEUTIC APPROACH IN LIPOSOMAL FORMULATION AGAINST ND PROGRESSION

Currently, the treatment of ND pathologies is still not available; however, many molecules and compounds can mitigate the symptoms and slow down the ND progression (Duraes et al., 2018). The first therapeutic approaches were based on the administration of small hydrophilic molecules (levodopa, rivastigmine, tacrine, donepezil), which can penetrate the BBB through the lipid-mediated diffusion process, exerting neuroprotective effects in several ND models. In recent years, the attention of medical and scientific communities has been focused on natural compounds, to overcome the side effects induced by chemical drug treatments against ND disorders. Due to anticholinesterase and the antioxidative and antiinflammatory outcomes of different phytochemicals, these

FIGURE 3 | Schematic representation of main structure of LPs, ethosome, solid lipid nanoparticles (SLNs), and nanostructured lipid carriers (NLCs).

could be promising therapeutic agents against ND progression. In detail, the neuroprotective and anti-protein aggregation effects induced by curcumin, quercetin, and resveratrol were extensively investigated.

Despite numerous therapeutic agents showing promising results for ND treatments, the experimental evidences obtained in vivo fell short of expectations; in fact, the efficacy of all these molecules is strongly limited due to the low bioavailability and low penetration capability in BBB. In addition, the phytochemicals compounds are poorly water and blood soluble and highlighted a short biological half-life as well as low uptake in the brain (Darvesh et al., 2012; Yang et al., 2014; Kulkarni and Canto, 2015; Tellone et al., 2015; Ay et al., 2017).

The advances in the nanotechnological field have enabled the development of engineered nanocarriers to deliver drugs and to release them in a controlled manner in specific body compartments. In particular, the LPs have become the best theranostic system for anti-ND molecules, since they allow them to overcome their inherent limitations (**Table 1**).

In this section, we describe how LP-based formulations improve the delivery to the brain of conventional therapeutic drugs.

#### SOLUBILITY, BIOAVAILABILITY, AND STABILITY

The bioavailability of given administered drug depends by several factors, such as the aqueous or blood solubility, BBB permeability and metabolic stability (Helen Chan and Stewart, 1996). Rivastigmine and Donepezil are small hydrophobic molecules largely used in Alzheimer's treatment, because they act as acetylcholinesterase inhibitor, preventing the acetylcholine degradation and increasing its availability at the synapses (Di Stefano et al., 2011).

In order to improve the bioavailability of these drugs: Nageeb El-Helaly et al. (2017) fabricated electrosteric stealth (ESS) LPs in which rivastigmine was encapsulated; meanwhile, Al Asmari et al. (2016) prepared a new liposomal formulation using 1,2-distearyl-sn-glycero-3-phosphocholine, cholesterol, and polyethylene glycol to entrap Donepezil. In details, Nageeb El-Helaly and co-workers (Nageeb El-Helaly et al., 2017) obtained an augmented stability by steric hindrance due to addition of didecyldimethyl ammonium bromide and 1,2 distearoyl-sn-glycero-3-phosphoethanolamine-N-(methoxy (polyethyleneglycol)-2000) to l-α-lecithin from soybean; consequently, the in vivo studies performed on 36 rabbits showed that the bioavailability of rivastigmine strongly increased in plasma and brain compartments after intranasal injection. Analogously, the Donepezil bioavailability resulted two times higher in the brain, after intranasal administration, compared to free drug in plasma (Al Asmari et al., 2016).

Among phytochemical compounds, curcumin induces neuroprotective and anti-protein aggregation effects. The low water solubility and bioavailability can be enhanced by internalization into liposomal carriers: Cheng et al. (2017) optimized a pH-driven method to prepare curcumin-loaded LPs (**Figure 4**), testing their bioavailability by the reproduction of gastrointestinal tissue. They found an increased bioavailability (more than twofold) with respect to curcumin solution.

Bulboaca et al. (2018) investigated the therapeutic potential of curcumin in addition to the traditional sumatriptan treatment for migraine disease using an in vivo rat model, demonstrating that, after intravenous administration, the shelf life of curcumin largely increased when it was encapsulated in liposomal vectors.

The polyphenolic compound resveratrol (3,4,5 trihydroxystilbene) gained increased attention as a therapeutic agent against ND diseases, due to their ability to promote the non-amyloidogenic cleavage of the amyloid precursor proteins, to improve β amyloid-peptide clearance and to limit neuronal injury (Li et al., 2012). As well as other natural compounds, the encapsulation in liposomal systems augments the drug effectiveness.

The enhancement in resveratrol bioavailability was demonstrated by Neves and collaborators by in vitro (Neves et al., 2013) and in vivo experiments (Jose et al., 2014), loading it in SLNs and NLCs. The in vivo biodistribution study using Wistar rats demonstrated that SLN could significantly increase the brain concentration of resveratrol (17.28 ± 0.6344µg/g) as compared to non-liposomal formulation (3.45 ± 0.3961µg/g). In pharmacokinetic studies conducted in male Wistar rats, Pandita et al. (2014) showed that the oral bioavailability of resveratrol was increased 8.035-fold when loaded in stearic acid-based SLNs stabilized by mixture of surfactants PHOSPHOLIPON <sup>R</sup> 90G/poloxamer 188.

In light of these experimental evidences, it is undeniable that encapsulation in liposomal carriers may offer interesting opportunities to enhance the delivery of anti-ND drugs.

### THERAPEUTIC EFFICACY AND DRUG TOXICITY REDUCTION

The encapsulation of anti-NDs drugs in liposomal-based carriers permits to minimize their adverse effects and simultaneously to enhance their therapeutic action.

Among small hydrophobic molecules, Levodopa (L-dopa, L-3, 4-dihydroxyphenylalanine) is the most used molecules in ND treatment; in particular, in the last 4 decades Levodopa (L-DOPA)-based therapy become the golden standard therapy against Parkinson disease (Poewe et al., 2010). After oral administration, <10% of L-DOPA dose reaches the CNS, while the remaining 90% tends to accumulate into the liver and muscles, inducing dyskinesia (Dodel et al., 2001). These adverse effects can be reduced loading L-DOPA in chitosan-coated nano-LPs. Chitosan is a natural hydrogel cationic polysaccharide of co-polymers glucosamine and N-acetyl glucosamine obtained by alkaline deacetylation of chitin that is constitutive of crustacean shells. In addition, it shows a positive charge that can be exploited to form polyelectrolyte vesicular complexes able to act as delivery systems (Elieh-Ali-Komi and Hamblin, 2016). Cao and colleagues performed in vivo studies on rats, demonstrating that the abnormal involuntary movement rate significantly decreased TABLE 1 | The table summarizes the main molecules currently used against neurodegenerative diseases.


*(Continued)* Liposome-Based

Strategies for NDs Treatment

#### TABLE 1 | Continued


*The limits of the treatments are reported as well as the advantages of their administration in liposomal formulation.* Liposome-Based

Strategies for NDs Treatment

in rats treated with chitosan-coated L-DOPA NLs than those who received free L-DOPA (Cao et al., 2016).

In the case of phytochemical compounds, the ND synthoms became more attenuated after the administration of curcumin and quercetin encapsulated in liposomal carriers with respect to free formulations. In details, in vivo experiments conducted on a rat model by Sadegh Malvajerd et al. (2019) showed a reduction in inflammatory state due to increased anti-oxidant activity after intravenous administrations of encapsulated curcumin into SLN and NLC carriers with respect to free drug; in addition, the pharmacokinetic studies demonstrated that the NLC option ensures higher accumulation into rat brain. Another interesting approach was suggested by Kuo et al. (2017) proposing the combined use of curcumin (CRM) and nerve growth factor (NGF) conjugated with cardiolipin (CL) to treat Alzheimer disorders. In details, the human neuroblastoma (SK-N-MC) cell line insulted with β-amyloid peptide (Aβ) shows how treatment with curcumin CRM-CL LPs prevented the neurodegeneration process through the inhibition of the p38, JNK, and tau protein phosphorylation. Meanwhile, the exposure to NGF-CL/LIP up-regulated the expression of the p-TrkA and p-ERK5, responsible to ensure neuronal viability. The effects of the combined use of these LPs were assessed in Wistar rat models, after intraperitoneal injections. To improve the permeation rate across the BBB, CRM-CL and CRM-CL LPs were implemented with germ agglutinin (WGA). The combined action of WGA-grafted CRM-CL and CRM-CL strongly affected the Aβ plaque formation, increasing the percentage of healthy neurons and promoting the cholinergic activity in the rat hippocampus.

Sokolik et al. compared the efficacy of intranasal administrated curcumin against Alzheimer's disorder in soluble form with liposomal formulations. After neuroinflammatory quantification analysis, they performed cognitive tests on rats to evaluate the impact of curcumin exposure on mnestic functions. They evaluated the impact on the angiotensin converting enzyme activity and cytokine productions in the damaged rat brain. Treatment with CLs reduced cytokines (interleukin-1β, interleukin-6, interleukin-10, tumor necrosis factor α) and angiotensin expression in cerebral cortex and hippocampus area. Finally, the cognitive tests show a reduction in the latent period of conditioned reflex reactions (Sokolik and Shulga, 2016; Sokolik et al., 2016, 2017).

Sarkar and Das have reported that, conversely to free formulation, quercetin loaded in mannosylated LPs and administrated via carotid injection in male Sprague-Dawley rats, in which the ischemia event was induced, meaningfully sustained antioxidant enzyme activity, preventing edema formation (Sarkar and Das, 2006). In addition, Das et al. demonstrated the oral administration of LP-encapsulated quercetin enhanced the antioxidative enzyme activity in adult female Swiss albino rat brains neurodamaged after arsenic exposure (Ghosh et al., 2009). Phachonpai and colleagues (Phachonpai et al., 2010) treated male Wistar rats, used as Alzheimer's disease model, with quercetin in LPs via intranasal administration. After treatment (0.5 mg/daily for 2 weeks), the neuroprotective quercetin action was recorded: the activities of superoxide dismutase, catalase, and glutathione peroxidase increased while the malondialdehyde levels were decreased. The reduction of oxidative stress confirmed the effectiveness of quercetin against Alzheimer's disease. The extension of the time treatment until 4 weeks increased the cholinergic neuron viability in hippocampus and ameliorated cognitive performances (Terdthai et al., 2010).

In the last years, the combination of resveratrol in LPs was explored as potential strategy for ND treatment. Ethemoglu et al. (2017) demonstrated how the resveratrol administration influenced the incidence of epileptic events in rats, in which brain ND disorder was induced by penicillin administration. The treatment with resveratrol reduced the epileptic activity, probably due to the reduction of oxidative stress at the cellular level. This rate further decreased when resveratrol was encapsulated in liposomal carriers.

Wang et al. (2013) assessed the impact of resveratrol derived from Polygonum cuspidatum on rats in which Parkinson's outcome was induced. After 14 days of oral treatment, the apoptosis rate of nigral tissue cells and ROS level was decreased while the antioxidant capability increased. These effects were strongly augmented when resveratrol was administered in liposomal vectors, probably due to the enhanced bioavailability.

#### PERMEABILITY

The main limitation of anti-ND drug efficacy is their poor BBB penetration capability, in fact <98% of the small molecules and nearly 100% of large molecules are able to reach the brain. The encapsulation in liposomal-based carriers can enhance the bioavailability in brain of therapeutic agents; for example Aramugam et al. (2008) treated Wistar rat with rivastigmine encapsulated in conventional multilamellar LPs: after intranasal injection, the rivastigmine level reveled in brain tissue resulted in higher respect to free drug. However, in order to improve the uptake of anti-ND drugs, active targeting strategies have been developed. The chemical modification of LP surfaces essentially consists in attaching ligands which interact with proteins expressed by BBB.

Yang et al. (2013) produced CCP LPs using cell-penetrating peptide (CPP), with the aim to enhance the brain distribution of rivastigmine. In detail, they proceeded by an intranasal administration in order to minimize the collateral effects (including hepatic first pass metabolism and gastrointestinal adverse outcomes), enhancing the permeability of the CPP-LP complex across both the nasal olfactory pathway and the murine BBB. Relative to the extended release and absorption, an intense inhibition of acetylcholinesterase and butyrylcholinesterase activities was detected.

Among receptor-mediated transcytosis, the transferrin (Tf) receptors are the one of the most explored as targets of the drug delivery system (Sharma et al., 2016). In particular, several in vitro and in vivo studies were performed in order to improve the targeting efficacy of therapeutic molecules across BBB (Pulgar, 2018).

Mourtas et al. (2014) produced a curcumin derivative in nanosized LPs functionalized with the anti-Tf antibody to direction them against Aβ protein in amyloid plaques. The potential efficacy of curcumin-derivative LPs and curcuminderivative anti-Tf LPs was assessed in ex vivo experiments performed on post-mortem extracted brain samples from patients affected by Alzheimer disease. Both LP types exhibited high binding affinity with Aβ proteins; indeed, the thioflavin assay confirmed their ability to slow down the Aβ1-42 peptide aggregation. Although the efficacy of examined carriers was comparable, the anti-Tf receptor-decorated LPs strongly enhanced the brain permeation.

With the aim to improve the efficacy of the neuroprotectant ZL006 against ischemic stroke, Wang and colleagues (Wang et al., 2015) encapsulated it in modified LPs with HAIYPRH (T7) peptide, which directly interacts with TfR. The in vivo experiments performed on rat models (Sprague-Dawley and ICR mice) revealed that the T7 functionalization considerably improved the penetration across the BBB. This implied a reduced infarct volume and enhanced neurological deficit with respect to the untargeted LPs or SKfree ZL006.

The Tf functionalization is currently used to improve BBB targeting and transcytosis. In order to enhance the neuronal transfection in gene therapy, Dos Santos Rodrigues et al. (Dos Santos Rodrigues et al., 2018) conceived LPs with double functionalization: Tf and penetratin (Pen); the latter enabled overcoming receptor saturation and increasing the number of internalized LPs. Hydrophilic plasmid for β-galactosidase (pβgal), used as a gene model, were loaded in Pen-Tf LPs in order to explore the efficacy in gene therapy applications. After injection in C57BL/6J mice, the plasmid biodistribution was quantified in different tissues: the liposomal formulation involved a significant increase in βgal amount in brain, liver, and kidneys with respect to the endogenous levels. In addition, the Pen-Tf functionalization improved the transfection capability only in the brain.

α-Mangostin is a polyphenolic xanthone, largely used in medicine; recently, it was demonstrated its capability to slow down AD progression (Wang et al., 2012); nevertheless, its efficacy is limited by the low penetration capability across the BBB. Moving from these evidences, Chen and colleagues (Chen et al., 2016) modified LPs with Tf in order to enhance the penetration rate of α-Mangostin in brain. α-M was intravenously administrated (5 mg/kg) into the Sprague Dawley rats in different formulations: α-M solution, α-M LP, and Tf(α-M) LP. The pharmacokinetic experiments revealed that encapsulation in Tf LPs augmented the α-M bioavailability in plasma and drug accumulation rate in brain tissue. The authors proposed Tf-LPs as a promising drug delivery system, demonstrating their ability to improve the brain permeation of α-M; however, α-M distribution quantification in different body compartments showed that Tf(α-M) LPs tended to accumulate also in liver tissue; this is due to the expression of TfR on hepatocytes.

The fact that Tf receptors were expressed on several cellular types, including hepatocytes, monocytes, erythrocytes, intestinal cells, epithelial cell of choroid plexus, and neuron besides BBB, represents an important issue in targeting of delivery systems. In addition, the efficiency of the Tf targeting could be dramatically compromised by the physiological presence of high levels of endogenous Tf in the plasma, which saturated the TfR endothelial cells (Sharma et al., 2016).

Like TfR, low-density lipoprotein (LDL) receptors, involved in the specific BBB transport mechanisms, could be exploited in the increase of the drug rate permeation in brain. In particular, human apolipoprotein E (apoE) plays a key role in lipid transport in the CNS (Fan et al., 2001; Hatters et al., 2006). In addition, biological membranes contain a great variety of phospholipids; one among them is phosphatidic acid (PA), which has a special attraction for Aβ and may help rescue neurons from Aβ-induced toxicity. Drug carriers associated with PA and ApoE may improve the affinity to binding to Aβ in vitro (Re et al., 2011; Salvati et al., 2013).

Starting from these evidences, Bana et al. (2014) used a Balb/c mice model to test the uptake in brain of modified LPs mApoE-PA-LPs, obtained by functionalizing PA and a synthetic peptide (mApoE; CWGLRKLRKRLLR) that includes the apolipoprotein-E receptor-binding domain. The biodistribution experiments, performed after intravenous administration, demonstrated that mApoE-PA-LPs were more capable of penetrating the BBB respect to PA-LPs.

The apolipoprotein E (ApoE) conjugated with PA was also used by Kuo and co-worker (Kuo et al., 2018) to functionalize LPs (ApoE-PA-LP) able to deliver quercetin and rosmarinic acid to brain, in order to inhibit β-amyloid (1–42) (Aβ1-42) aggregation. The permeation capacity strongly increased in functionalization case respect to free drugs and drugs encapsulated in undecorated LPs. The administration of quercetin and rosmarinic acid in ApoE-PA-LP formulation induced in Aβ-insulted SD rats the diminishment of the acetylcholinesterase activity and the reduction in Aβ plaque formation.

Recently, other potential targeting molecules have been explored to improve the drug internalization exploiting the receptor-mediated transcytosis process. In detail, the nutrient transporters can be used for brain delivery. Among these, glutathione (GSH) is an endogenous tripeptide able to penetrate the BBB as sodium-dependent transporter; this latter is preferentially expressed in CNS and BBB of all mammalian species (Smeyne and Smeyne, 2013; Kuo et al., 2018).

Rip and co-workers fabricated PEGylated LPs with GSH, and they investigated the potential improvement in brain uptake of drug encapsulated in this carrier; for this purpose the authors loaded a fluorescent tracer (carboxifluorescein) in the PEG-GSH modified LPs (Rip et al., 2014). Quantification of carboxifluorescein in brain was obtained by microdialysis experiment on Wistar rats, in which this fluorescent molecule was intravenously administrated in three different forms: free, non-targeted PEG LPs, and GSH-PEG LPs. The results showed that the carboxifluorescein level in brain increased in liposomal formulation; in particular, the GSH-PEG LP option made this level fourfold higher with respect to undecorated PEG LPs. The efficiency of GSH-PEG LP delivery in the brain was also tested for ribavirin drug (Rip et al., 2010; Maussang et al., 2016). The concentration of drug in brain tissue increased in modified LP formulation in Wistar rats after administration by intravenous route; the permeation rate beyond the BBB became proportional to the amount of GSH coating (ranging from 0 to 2%).

In addition, in literature several works reported how the GSH-PEG LP formulation improves in brain delivery of the opioid peptide DAMGO (H-Tyr-d-Ala-Gly-MePhe-Gly-ol) (Lindqvist et al., 2013), the anti-amyloid single-domain antibody fragment (Rotman et al., 2015), flucytosine (Salem et al., 2015), methylprednisolone hemisuccinate (Kanhai et al., 2018), etc.

Another promising strategy to improve penetration capability of anti-ND drug is based on the use of cationic polyelectrolyte in LP preparation. Cationic LPs (CLs) can penetrate the BBB through adsorptive-mediated endocytosis process (Herve et al., 2008). This occurs because the surface of CLs exhibits a positive charge at physiological pH; thus, it electrostatically interacts with the polyanions present in the BBB (Miller et al., 1998).

This event was examined in vivo on Sprague–Dawley rats, in which cationic, anionic, and charge-neutral LPs were directly injected into carotid arteries. Joshi et al. reported that superficial charged LPs more actively interact with BBB endothelial cell respect to neutral LPs; in particular, the authors recorded the highest cationic concentration with respect to the anionic LPs into the brain tissue (Joshi et al., 2014a,b; Joshi et al., 2015).

Migliore et al. (2010) stated that the use of CLs as a delivery system enhanced the brain uptake of drug administered by intranasal route in Sprague–Dawley rats.

Recently, another work (Dhaliwal et al., 2020) reported how it is possible to take advantage of cationic LPs to improve the mRNA delivery. After optimization of the liposomal formulation, mRNA was administered by intranasal route to CD-1 mice. Through the GFP and luciferase reporter systems, the authors have quantified the mRNA biodistribution in vivo. The findings demonstrated the cationic LPs improved the penetration rate of mRNA and its accumulation into specific cerebral areas (cortex, striatum, and midbrain).

Although it was demonstrated that the CL formulation enhanced the anti-ND drug penetration rate in the brain, it must be admitted that this rate is limited by non-specific uptake phenomena in peripheral tissues. In addition, the superficial charge of LPs could be reduced by interaction with serum protein before reaching the BBB (Tagalakis et al., 2014).

These issues could be overcome by increasing the amounts of use of CLs, but keeping in consideration their cytotoxic potential.

#### ADMINISTRATION ROUTES

The main challenge in the treatment of CNS diseases is represented by the presence of the BBB, which makes complicated the delivery of therapeutic agents into the brain. At first, invasive approaches have performed in order to overcome the BBB, including the neurosurgery-based cerebral infusions or implants and the physical or chemical disruption of the BBB to permit the penetration of drug via osmotic shift (Chen et al., 2004). However, these techniques are highly dangerous carry risks of infection and brain tissue damage, and are able to induce chronic neuropathological effects in treated patients (Chen et al., 2004). In recent decades, several non-invasive strategies, based on physiological brain transport mechanisms, have been developed.

As already described in this review, the drug loading in LPs increases its own uptake in the brain; later on, once the importance of LP usage was proven, various studies have tried several routes of administration for finding out if they are more or less adequate.

The intravenous routes are certainly the most common, preferred to the oral route, because in this latter case there is a loss in terms of drug percentage in the gastrointestinal ambient: drug is degraded by the stomach fluids and poorly absorbed by the intestine, while increasing the hepatic and renal load; although in different reported works, it has been proven that the drug loading in LPs prevent the drug degradation. Although the intravenous injection is the most applied, the interest in alternative parenteral routes increased, for finding new direct administrations to the BBB. Among these, the intranasal way gained a noticeable increase by the scientific community, due to its ability to let the drug reach the brain via different pathways: direct ones, through olfactive epithelium, thanks to the presence of olfactory nerves; or indirect ones, through the respiratory epithelium first and the circulatory system after (Hong et al., 2019).

In the last decade, there is a growing interest of administering them via intradermal routes. In this perspective, ethosomes represent the better choice among liposomal systems, due to their ability to increase the drug permeation across the derma. For instance, Ropinirole hydrochloride is commonly used in Parkinson disease. With the aim to evaluate the possibility to administrate Ropinirole hydrochloride via transdermal route, Mishra et al. (2013) synthetized different ethosomal carriers, varying ethanol and lecithin concentrations. The experimental in vivo study showed that RHC encapsulated in ethosomes (30 and 4% w/v of ethanol and lecithin, respectively) increased their blood circulation time, becoming comparable to the oral administration case.

Flurbiprofen is largely used in ND treatments for its antiinflammatory effect. The permeation of ethosomal–flurbiprofen (EF) formulations was assessed in the albino rat model (Paliwal et al., 2019). The reported results showed a high encapsulation drug efficiency (95 %); in addition, the permeation efficacy was equal to (82.56 ± 2.11) g/cm<sup>2</sup> in 24 h, and transdermal flux was found as 226.1 µg/cm<sup>2</sup> /h.

Ligustrazine phosphate (LP) is an antioxidant used for the Alzheimer disorder. Shi and co-workers (Shi et al., 2012) optimized the ethosome-based carrier for LP transdermal delivery; in addition, they evaluated the therapeutic impact on rats, in which amnesia disease was induced by scopolamine. Experimental evidence indicated that the penetration ability of the LP ethosomal system was strongly higher than the LP in aqueous solution. Moreover, the LP ethosomal treatment normalized the levels both of antioxidant enzymes and the lipid peroxidation in a healthy rat model.

These in vivo evidences suggested the goodness of ethosome-based strategy; nevertheless, for ND disease, only a few studies explored the potential of dermal administration ethosome-mediated.

#### CONCLUSIONS AND PERSPECTIVE

In the last decades, many efforts have been made to develop an optimal therapeutic strategy against neurodegenerative disorders due to their dramatic incidence worldwide. In this perspective, several new molecules including phytochemical extracts were investigated along with the conventional drugs. Unfortunately, their therapeutic potential was strongly compromised owing to the presence of high selective barrier (BBB) that protects the brain from treatment (CNS).

The effectiveness of drugs against ND diseases increases when they are encapsulated into liposomal-based vectors.

Generally, LPs represent a promising platform for theranostic medicine purposes, thanks to their size, biocompatibility, high biological affinity, immunogenicity, biodegradability, and low toxicity. In addition, LPs have a good ability to deliver both hydrophobic and hydrophilic molecules through BBB.

In this review article, the recent state of the art about the use of liposomal-based carriers for anti-neurodegenerative drug delivery and the in vivo applications are extensively examined. In the reported researches, the enhanced efficacy of drug following liposomal encapsulation has been demonstrated, in terms of improved bioavailability, stability, and permeation capability across BBB.

Albeit a great deal of progress has been made in liposomal formulations and their efficiency was demonstrated in several in vivo studies, some issues are still present.

The scientific community is called to perform further clinical trials in order to evaluate the LP accumulation effects induced by long-term exposure and, simultaneously, to optimize the ingegnerization of these carriers in the human body.

### AUTHOR CONTRIBUTIONS

MC and VDM: conceptualization and writing (original draft preparation). MC, VDM, SL, and RR: review and editing manuscript. SL and RR: funding acquisition.

#### REFERENCES


#### ACKNOWLEDGMENTS

MC and VDM kindly acknowledge Programma Operativo Nazionale (PON) Ricerca e Innovazione 2014-2020 Asse I Capitale Umano, Azione I.2, Avviso 'A.I.M: Attraction and International Mobility [F88D18000070001] [CUP F88D1800060001]. SL has been partially supported by Progetto Fondo Integrativo Speciale per la Ricerca (FISR-CNR) Tecnopolo di Nanotecnologia e Fotonica per la Medicina di Precisione [CUP B83B17000010001].


tumors, including diffuse intrinsic pontine glioma. Front. Oncol. 8:239. doi: 10.3389/fonc.2018.00239


**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 © 2020 Cascione, De Matteis, Leporatti and Rinaldi. 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.