# POLYMERIC NANO-BIOMATERIALS FOR MEDICAL APPLICATIONS: ADVANCEMENTS IN DEVELOPING AND IMPLEMENTATION CONSIDERING SAFETY-BY-DESIGN CONCEPTS

EDITED BY : Gerrit Borchard, Olga Borges, Vasile Ostafe, Giuseppe Perale, Claudia Som, Peter Wick and Manfred Zinn 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 978-2-88966-256-2 DOI 10.3389/978-2-88966-256-2

#### 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

# POLYMERIC NANO-BIOMATERIALS FOR MEDICAL APPLICATIONS: ADVANCEMENTS IN DEVELOPING AND IMPLEMENTATION CONSIDERING SAFETY-BY-DESIGN CONCEPTS

Topic Editors: Gerrit Borchard, Université de Genève, Switzerland Olga Borges, University of Coimbra, Portugal Vasile Ostafe, West University of Timișoara, Romania Giuseppe Perale, University of Applied Sciences and Arts of Western Switzerland, Switzerland Claudia Som, Swiss Federal Laboratories for Materials Science and Technology, Switzerland Peter Wick, Swiss Federal Laboratories for Materials Science and Technology, Switzerland Manfred Zinn, HES-SO Valais-Wallis, Switzerland

Citation: Borchard, G., Borges, O., Ostafe, V., Perale, G., Som, C., Wick, P., Zinn, M., eds. (2020). Polymeric Nano-Biomaterials for Medical Applications: Advancements in Developing and Implementation Considering Safety-By-Design Concepts. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-256-2

# Table of Contents


Jessica Da Silva, Sandra Jesus, Natália Bernardi, Mariana Colaço and Olga Borges

*17 Computational Assessment of the Pharmacological Profiles of Degradation Products of Chitosan*

Diana Larisa Roman, Marin Roman, Claudia Som, Mélanie Schmutz, Edgar Hernandez, Peter Wick, Tommaso Casalini, Giuseppe Perale, Vasile Ostafe and Adriana Isvoran

*33 A Perspective on Polylactic Acid-Based Polymers Use for Nanoparticles Synthesis and Applications*

Tommaso Casalini, Filippo Rossi, Andrea Castrovinci and Giuseppe Perale


Divesha Essa, Pierre P. D. Kondiah, Yahya E. Choonara and Viness Pillay

*149 Chitosan Nanoparticles: Shedding Light on Immunotoxicity and Hemocompatibility*

Sandra Jesus, Ana Patrícia Marques, Alana Duarte, Edna Soares, João Panão Costa, Mariana Colaço, Mélanie Schmutz, Claudia Som, Gerrit Borchard, Peter Wick and Olga Borges


Cíntia Marques, Claudia Som, Mélanie Schmutz, Olga Borges and Gerrit Borchard

#### *205 Hydrogel Biomaterials for Application in Ocular Drug Delivery*

Courtney R. Lynch, Pierre P. D. Kondiah, Yahya E. Choonara, Lisa C. du Toit, Naseer Ally and Viness Pillay

#### *223 A Methodological Safe-by-Design Approach for the Development of Nanomedicines*

Mélanie Schmutz, Olga Borges, Sandra Jesus, Gerrit Borchard, Giuseppe Perale, Manfred Zinn, Ädrienne A. J. A. M Sips, Lya G. Soeteman-Hernandez, Peter Wick and Claudia Som

*230 Permeation of Biopolymers Across the Cell Membrane: A Computational Comparative Study on Polylactic Acid and Polyhydroxyalkanoate*

Tommaso Casalini, Amanda Rosolen, Carolina Yumi Hosoda Henriques and Giuseppe Perale

# Editorial: Polymeric Nano-Biomaterials for Medical Applications: Advancements in Developing and Implementation Considering Safety-by-Design Concepts

Gerrit Borchard<sup>1</sup> \*, Claudia Som<sup>2</sup> , Manfred Zinn<sup>3</sup> , Vasile Ostafe<sup>4</sup> , Olga Borges <sup>5</sup> , Giuseppe Perale<sup>6</sup> and Peter Wick <sup>2</sup>

<sup>1</sup> School of Pharmaceutical Sciences, Université de Genève, Geneva, Switzerland, <sup>2</sup> Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland, <sup>3</sup> Institute of Life Technologies, University of Applied Sciences and Arts Western Switzerland (HES-SO Valais-Wallis), Sion, Switzerland, <sup>4</sup> Advanced Environmental Research Laboratories, Department of Biology-Chemistry, Faculty of Chemistry, Biology, Geography, West University of Timisoara, Timisoara, Romania, <sup>5</sup> Centre for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal, <sup>6</sup> Department of Innovative Technologies, University of Applied Sciences and Arts of Southern Switzerland (SUPSI), Manno, Switzerland

Keywords: nano-biomaterials, safe-by-design, nanotechnology, nanomedicines, drug carriers

**Editorial on the Research Topic**

#### Edited and reviewed by:

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

> \*Correspondence: Gerrit Borchard gerrit.borchard@unige.ch

#### Specialty section:

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

Received: 28 August 2020 Accepted: 04 September 2020 Published: 19 October 2020

#### Citation:

Borchard G, Som C, Zinn M, Ostafe V, Borges O, Perale G and Wick P (2020) Editorial: Polymeric Nano-Biomaterials for Medical Applications: Advancements in Developing and Implementation Considering Safety-by-Design Concepts. Front. Bioeng. Biotechnol. 8:599950. doi: 10.3389/fbioe.2020.599950

#### **Polymeric Nano-Biomaterials for Medical Applications: Advancements in Developing and Implementation Considering Safety-by-Design Concepts**

The aging population represents an enormous social, structural, and financial burden on society. To address the potential problems of supporting an elderly population, it is important to consider ways of enabling individuals to maintain independence and quality-of-life (QoL) for as long as possible. To achieve this, we need to develop new solutions and novel concepts through technologies, and researchers have been exploring how disease and aging are mediated by molecular processes at the nanoscale. The subsequent nanotechnology-enabled approaches that have emerged in recent years include innovative nano-materials and nano-devices, which could enable interventions at the molecular length scale and are, therefore, an important cornerstone in building solutions.

As for every novel technology or material, a careful safety assessment is needed early in its development to avoid social and economic drawbacks. While guidelines and legislation have been put in place for therapeutics of low molecular weight along with much more complex biologics as well as their follow-on products (generics and biosimilars, respectively), the regulatory approach for advanced therapeutics including polymeric biomaterials is still in its infancy. The rise of so-called "nanomedicines" or complex therapeutics along with follow-on products such as "nanosimilars" or "complex generics" have provided advantageous attributes on the nanoscale, but the lack of guidance is a difficult challenge that needs to be addressed in the development of such products in future.

This issue describes some of the different nano-biomaterials that are currently being studied for the preparation of nanoscale drug carriers. These include: chitosan (see Jesus, Marques et al.; Marques et al.), a chitin derivative biopolymer obtained mostly from crustaceans; the family of poly(lactide-co-glycolide) (PLGA) polymers (Essa et al.); poly(D,L-lactic acid) (Casalini, Rossi et al.; Da Silva et al.); polyhydroxyalkanoates (Casalini, Rosolen et al.), a polyester produced by bacteria; and hydrogel systems made up of hyaluronic acid and alginate (Lynch et al.). As indicated in this issue, some studies have found that for some materials [e.g., chitosan (Marques et al.)], the description of the specifications of the nano-biomaterial is often incomplete. The preparation of nano-vectors using these materials is also complex that needs to be better understood, in particular when self-assembly processes are used to implement therapeutic drug delivery (Yadav et al.). These issues render comparative efficacy studies difficult and preclude the introduction of a "safe-by-design" concept, as developed in the framework of the European project "GoNanoBioMat" (https://gonanobiomat. eu/). As a concept, safe-by-design is currently being adapted for use in research on nano-biomaterials, based on the processes used in drug discovery and development, which adopt safety aspects early in the process (Schmutz et al.). It is intended that this concept in combination with other existing regulatory frameworks, will guide small and medium-sized companies during the development process of nanomedicines, span stages from material selection and design, characterization, assessment of human, and environmental health risks, to manufacturing and control, as well as storage and transport of the final product.

This special issue also focuses on the hazard assessment of polymeric biomaterials for medical use [see the literature study (Jesus, Schmutz et al.)], and the determination of the impact of certain properties of nanoscale drug vectors on their safety (cytotoxicity, immunotoxicity) (Jesus, Marques et al.), which discuss membrane diffusion characteristics and pharmacokinetics. Some of these parameters, including nanomaterial interactions with their biological environment (Casalini, Rosolen et al.) and the evaluation of the risks of their degradation products (Roman et al.) can be assessed by computational simulation (Casalini, Limongelli et al.) or systematic evaluation of animal studies (Hauser and Nowack), resulting in concrete suggestions for nanobiomaterial design to achieve optimized efficacy and enhanced safety. Finally, the different options of targeted anti-schistosomal therapy using nanotechnology are reviewed (Adekiya et al.).

The diversity of the different papers presented in this special issue are indicative of the significant interplay between life and the material sciences with computational approaches. We believe that this exchange of ideas is one of the best approaches to tackling the increasingly complex challenges of an aging society.

# AUTHOR CONTRIBUTIONS

GB has drafted the manuscript and submitted in final form. CS, MZ, VO, OB, GP, and PW have revised the manuscript draft. All authors contributed to the article and approved the submitted version.

# FUNDING

This work was supported by Romanian Ministry of National education UEFISCDI - grant PN3-P3-285.

# ACKNOWLEDGMENTS

This work was financed by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational Programme under project CENTRO-01-0145-FEDER-00008: BrainHealth 2020, and through the COMPETE 2020—Operational Programme for Competitiveness and Internationalization and Portuguese national funds via FCT—Fundação para a Ciência e a Tecnologia, I.P., under project PROSAFE/0001/2016, and the strategic projects POCI-01-0145-FEDER-030331 and POCI-01-0145-FEDER-007440 (UID/NEU/04539/2019). This work is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016, and from the CTI (1.1.2018 Innosuisse) under grant agreement number 19267.1 PFNM-NM.

**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 Borchard, Som, Zinn, Ostafe, Borges, Perale and Wick. 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.

# Poly(D,L-Lactic Acid) Nanoparticle Size Reduction Increases Its Immunotoxicity

#### Jessica Da Silva1,2, Sandra Jesus <sup>1</sup> , Natália Bernardi 1,2, Mariana Colaço1,2 and Olga Borges 1,2 \*

<sup>1</sup> Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal, <sup>2</sup> Faculty of Pharmacy, University of Coimbra, Azinhaga de Santa Comba, Coimbra, Portugal

#### Edited by:

Michele Iafisco, Italian National Research Council (CNR), Italy

#### Reviewed by:

Vuk Uskokovic, University of California, San Francisco, United States Daniele Catalucci, Institute for Genetic and Biomedical Research (IRGB), Italy Michele Chiappi, Imperial College London, United Kingdom

> \*Correspondence: Olga Borges olga@ci.uc.pt

#### Specialty section:

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

> Received: 27 March 2019 Accepted: 21 May 2019 Published: 06 June 2019

#### Citation:

Da Silva J, Jesus S, Bernardi N, Colaço M and Borges O (2019) Poly(D,L-Lactic Acid) Nanoparticle Size Reduction Increases Its Immunotoxicity. Front. Bioeng. Biotechnol. 7:137. doi: 10.3389/fbioe.2019.00137 Polylactic acid (PLA), a biodegradable and biocompatible polymer produced from renewable resources, has been widely used as a nanoparticulate platform for antigen and drug delivery. Despite generally regarded as safe, its immunotoxicological profile, when used as a polymeric nanoparticle (NP), is not well-documented. Thus, this study intends to address this gap, by evaluating the toxicity of two different sized PLA NPs (PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs), produced by two nanoprecipitation methods and extensively characterized regarding their physicochemical properties in in vitro experimental conditions. After production, PLA<sup>A</sup> NPs mean diameter (187.9 ± 36.9 nm) was superior to PLA<sup>B</sup> NPs (109.1 ± 10.4 nm). Interestingly, when in RPMI medium, both presented similar mean size (around 100 nm) and neutral zeta potential, possibly explaining the similarity between their cytotoxicity profile in PBMCs. On the other hand, in DMEM medium, PLA<sup>A</sup> NPs presented smaller mean diameter (75.3 ± 9.8 nm) when compared to PLA<sup>B</sup> NPs (161.9 ± 8.2 nm), which may explain its higher toxicity in RAW 264.7. Likewise, PLA<sup>A</sup> NPs induced a higher dose-dependent ROS production. Irrespective of size differences, none of the PLA NPs presented an inflammatory potential (NO production) or a hemolytic activity in human blood. The results herein presented suggest the hypothesis, to be tested in the future, that PLA NPs presenting a smaller sized population possess increased cytotoxicity. Furthermore, this study emphasizes the importance of interpreting results based on adequate physicochemical characterization of nanoformulations in biological medium. As observed, small differences in size triggered by the dispersion in cell culture medium can have repercussions on toxicity, and if not correctly evaluated can lead to misinterpretations, and subsequent ambiguous conclusions.

Keywords: polylactic acid, poly(D,L-lactic acid), polymeric nanoparticles, drug delivery systems, immunotoxicity, size-dependent cytotoxicity, hemocompatibility, cell culture medium

#### INTRODUCTION

Polylactic acid (PLA) is a Food and Drug Administration approved polymer that has proven to be a very versatile material, with interesting properties such as biocompatibility and biodegradability (Essa et al., 2012; Legaz et al., 2016). Thus, PLA has been explored regarding many therapeutic applications, including as a nanoparticulate antigen and drug delivery vehicle (Essa et al., 2012; Legaz et al., 2016).

The great interest in using nanoparticles (NPs) for biomedical applications (Jiao et al., 2014) is transversal to various polymeric materials, despite the poorly understood correlation between their physicochemical properties and their effects on the immune system. This knowledge gap partially results from the fact that NPs physicochemical properties, particularly its reduced size, hinder the application of traditional toxicity assays and further contribute to the misinterpretation of results and ambiguous conclusions among research groups (Dobrovolskaia et al., 2009). Additionally, the mandatory use of biological medium during toxicological assays can modify the NPs characteristics such as size, surface charge and morphology, through phenomenon's like protein corona formation and particle agglomeration, which will therefore influence the immunotoxicity profile of the NPs (Kendall et al., 2014). Therefore, a detailed characterization of the NPs in the experimental assay conditions is crucial to discuss the results, but is commonly absent from the scientific published reports. Biodegradable polymers such as PLA are generally regarded as safe, but their immunotoxicological profile when used as NPs, is not well documented (Singh and Ramarao, 2013). Previously, da Luz et al. has published an interesting paper assessing the toxicity and biocompatibility of PLA NPs in A549 cells (da Luz et al., 2017). Similarly, Legaz et al. conducted toxicity studies in Schneider's D. melanogaster line 2 (S2) cells (Legaz et al., 2016).

In this study, we described the production method of two different sized PLA NPs (PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs), in order to evaluate how the NP size affects their toxicological profile using cells from the immune system. In vitro immunotoxicity studies comprised hemocompatibility assays, cell viability experiments with peripheral blood mononuclear cells (PBMCs) and RAW 264.7 macrophage cell line, and nitric oxide (NO) and reactive oxygen species (ROS) production assays in RAW 264.7 cells. Furthermore, for the discussion of these results we have included the characterization of both PLA NPs regarding its size, polydispersity index (PDI) and zeta potential in the different cell culture media used in in vitro studies. In contrast to other published reports evaluating the toxicity of PLA NPs, this report aims to highlight the importance of the NPs characterization under in vitro experimental conditions for the establishment of relationships between the NPs properties and their effect in cells of the immune system. Not being an exhaustive study of immunotoxicology, it nevertheless intends to emphasize the importance of these studies in the development of nanomedicines.

#### MATERIALS AND METHODS

#### Poly(D,L-Lactide) Polymer

Poly(D,L-lactide) (PDLLA) polymer with an average molecular weight (MW) of 1,01,782 g/mol [analyzed by gel permeation chromatography/size exclusion chromatography (GPC/SEC)] and an inherent viscosity of 0.68 dL/g was obtained from Sigma-Aldrich Corporation (St. Louis, MO, USA).

#### PLA NP Production

For PLA<sup>A</sup> NPs production, PDLLA was dissolved at 2 mg/mL in acetone. NPs formed spontaneously upon dropwise addition of 4.5 mL of PDLLA solution to 13.5 mL of an aqueous solution (pyrogen-free water) with 1% of Pluronic <sup>R</sup> F68 Prill (Basf Corporation, Ludwigshafen, Germany) using a highspeed homogenizer at 13,000 rpm. The homogenization was maintained for another 2 min, after total addition of the PDLLA solution. The PLA<sup>A</sup> NPs were concentrated by centrifugation at 13,000 g for 20 min at 10◦C, resuspended in pyrogen-free water and concentrated again. This procedure was repeated 2 more times, and finally each batch was concentrated in a final volume of 2 mL. On the other hand, for the production of PLA<sup>B</sup> NPs, PDLLA was dissolved at 0.75 mg/mL in acetone. NPs formed spontaneously upon dropwise addition of 1 mL of PDLLA solution to 2.5 mL of an aqueous solution with 0.1% of Pluronic F68 using a vortex homogenizer and the agitation was maintained for another 2 min, after the total addition of the PDLLA solution. In order to concentrate and wash the NPs, 8 batches of PLA<sup>B</sup> NPs (20 mL) were centrifuged with Vivaspin 20 centrifugal concentrator (MWCO 300 KD, ThermoFisher Scientific Inc., Waltham, MA, USA) at 3,000 g at 10◦C until <1 mL was recovered in the centrifuge tube. The NPs were then resuspended in 10 mL pyrogen-free water, the centrifugation procedure was repeated, and the NPs were resuspended in a final volume of 1 mL pyrogen-free water. In vitro experiments and the respective characterization in in vitro conditions, were performed by diluting these concentrated NP suspensions in serum supplemented cell culture media as described below.

# PLA NP Characterization

Zetasizer Nano ZS (Malvern Instruments, Ltd., Worcestershire, UK) was used to measure particle size, and the respective polydispersity index (PDI), by dynamic light scattering (DLS) and particle zeta potential through electrophoretic light scattering (ELS). The samples were characterized dispersed in pyrogen-free water and in supplemented culture media (RPMI and DMEM). In the second case, the size and zeta potential assessment was done immediately after dilution in the culture medium, and after 24 h of incubation at 37◦C. The NP size when dispersed in pyrogen-free water was also confirmed by transmission electron microscopy (TEM). Samples were placed on a microscopy grid and observed under a FEI-Tecnai G2 Spirit Biotwin, a 20–120 kV TEM (FEI Company, OR, USA).

#### Immunotoxicity and Hemocompatibility Assays In vitro Studies With Human Blood

#### **Hemolysis assay**

Hemolysis assay was performed according to published protocols with minor modifications (Pattani et al., 2009; Villiers et al., 2009). Whole blood was collected from healthy donors after formal acceptance with a written informed consent. Blood was diluted with PBS to adjust total blood hemoglobin (TBH) concentration to 10 ± 2 mg/mL (TBHd). A volume of 100 µL of PLA NPs suspensions, PBS (negative control), or Triton-X-100 (positive control) were added to 700 µL PBS in different tubes. Then, 100 µL of TBHd was added to each tube, followed by incubation at 37◦C for 3 h ± 15 min. NPs were also incubated with PBS without blood to evaluate the possible NP interference. After the incubation time, the tubes were centrifuged at 800 g for 15 min. One hundred microliter of each supernatant and 100 µL cyanmethemoglobin (CMH) reagent were added to a 96 well-plate. The CMH reagent was prepared by mixing 1000 mL Drabkin's reagent and 0.5 mL of 30% Brij 35 solution (Sigma-Aldrich, St. Louis, MO, USA). The absorbance (OD) at 540 nm was determined and the percentage of hemolysis was calculated by Equation 1:

$$\text{(Hemolys is (\%)} = \frac{\text{(OD sample (540 nm) - OD PBS (540 nm))}}{\text{(OD TBH (540 nm) - OD PBS (540 nm))}} \ge 100 \text{ (1)}$$

#### In vitro Studies With PBMCs **PBMCs isolation**

Buffy coats obtained from normal donors (heparinized syringes) were kindly given by IPST, IP (Coimbra, PT). PMBCs were isolated on a density gradient with Lymphoprep (Axis-Shield, Dundee, Scotland) according to the provider's guidance protocol and as published by our group (Jesus et al., 2017). Isolated PBMCs were cultured in Roswell Park Memorial Institute medium (RPMI) with 10% heat inactivated fetal bovine serum (FBS), supplemented with 2 mM L-glutamine, 1% penicillin/streptomycin and 20 mM HEPES.

#### **Nanoparticle toxicity**

PLA NPs cytotoxicity was evaluated on human PBMCs using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Cells were plated in a 96-well plate at a density of 5 x 10<sup>5</sup> monocytes/well. Serial dilutions of NPs and controls were incubated with the cells for 24 h, at 37◦C and 5% CO2. After this period, 20 µL of MTT solution (5 mg/mL) in PBS were added to each well-followed by additional 4 h incubation. To ensure dissolution of the formazan crystals, cell culture plates were centrifuged (800 g, 25 min, 20◦C) and the culture medium was replaced by DMSO and the OD of the resultant colored solution was measured at 540 and 630 nm. Cell viability (%) was calculated by Equation 2:

$$\begin{aligned} \text{Cell viability (\%)}\\ &= \frac{\text{(OD sample (540 nm) - OD sample (630 nm))}}{\text{(OD control (540 nm) - OD control (630 nm))}} \times 100 \end{aligned} \tag{2}$$

The inhibitory concentration for 50% of cell viability (IC50) was calculated by plotting the log concentration of the NPs vs. inhibition percentage of cell viability and extrapolating the value from a non-linear regression using Prism 6.0 (GraphPad Software, San Diego, CA, USA).

Cytotoxicity results obtained with MTT assay were confirmed with propidium iodide (PI) assay. Briefly, cells incubated with 4 nanoparticle concentrations previously used in MTT assay were centrifuged (800 g, 25 min, 20◦C), resuspended in PBS and collected for analysis in a BD FACSCalibur Flow Cytometer (BD Biosciences, Bedford, MA, USA) using PI solution (0.5 µg/mL).

#### In vitro Studies With RAW 264.7 Macrophage Cell Line

RAW 264.7 (ATCC <sup>R</sup> TIB-71TM) were acquired to ATCC (Manassas, VA, USA), cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% heat inactivated FBS, 1% Penicillin/Streptomycin, 10 mM HEPES and 3.7 g/L Sodium Bicarbonate, and used until passage 18.

#### **Nanoparticle cytotoxicity**

PLA NP toxicity in RAW 264.7 was assessed as described previously for PBMCs with some modifications. Briefly, for MTT assay, macrophages were plated at a concentration of 2 × 10<sup>4</sup> cells/well and the incubation with MTT solution was performed for 1 h 30 min.

For the assay with PI, the cells were collected using the dissociation medium (PBS-EDTA 5 mM) followed by centrifugation (250 g, 10 min, 20◦C) to replace the medium with PBS.

#### **Nanoparticle effect on production of the reactive oxygen species**

The ROS production was assessed using the dichlorofluorescein diacetate probe (DCFH-DA) (Thermo Fisher Scientific Inc., Waltham, MA, USA). The RAW 264.7 cells were incubated in a black 96-well plate for 24 h at 37◦C and 5% CO2, at density of 0.5 × 10<sup>5</sup> cells/well. After that period, serial dilutions of PLA NPs were incubated with the cells, to evaluate ROS stimulation. LPS was used as a positive control (1 µg/mL).

After 24 h, cell culture medium was replaced by DCFH-DA (50µM) in serum free DMEM and the cells were incubated for another 2 h at 37◦C and 5% CO2. The resulting fluorescence was read at 485/20 nm and 528/20 nm (excitation/emission wavelengths).

To calculate the stimulation of ROS production, Equation (3) was applied:

$$\begin{aligned} \text{ROS production (mean fluorescence increase)}\\ &= \frac{\text{FluoresenceSAMPLE}}{\text{FluoresenceNEGATIVE} \, \text{CONROL}} \end{aligned} \tag{3}$$

#### **Nanoparticle effect on nitric oxide production**

The NO production by RAW 264.7 was evaluated based on nitrite quantification using the Griess reagent. RAW 264.7 cells were incubated in a 48-well-plate at a density of 2.25 × 10<sup>5</sup> cells/well for 24 h at 37◦C and 5% CO2. After that period, cell culture medium was replaced by serial dilutions of PLA NPs diluted in cell culture medium without phenol red. LPS was used as a positive control (1µg/mL). To test if the NPs were able to inhibit LPS stimulated NO production, the same NP concentrations were incubated together with cells in the presence of the LPS (1 µg/mL).

The Cell supernatants were collected 24 h after incubation, and 100 µL of each test sample was plated in a 96-well-plate and combined with 100 µL of Griess reagent. A calibration curve performed with sodium nitrite (0–80µg/mL) was also plated in duplicate. The optical density of the samples was measured at 550 nm and NO quantification was extrapolated from the calibration curve.

To calculate the inhibition of NO production upon stimulation with LPS (Equation 4) was applied:

$$\begin{aligned} \text{Inhibition of NO production (\%)}\\ = \frac{\text{NO(\mu g/mL)} \\_ \text{SAMPLE}}{\text{NO (\mu g/mL)} \\_ \text{POSTIVE CONTROL}} \times 100 \end{aligned} $$

#### Statistical Analysis

Data were analyzed using GraphPad Prim 6 (GraphPad Software, Inc., La Jolla, CA, USA), in which significant differences were obtained from one-way ANOVA, and values were considered statistically different when p < 0.05. In vitro data were expressed as means ± standard error of the mean (SEM).

### RESULTS

### PLA<sup>A</sup> NPs Are the Largest in Water but the Smallest in Culture Medium

Although PLA polymer has been approved by FDA for human use in an extensive range of applications (Tyler et al., 2016). The information about its toxicological profile when used as a NP is scarce (Singh and Ramarao, 2013). In order to give new insights on the relationship between NP physicochemical properties and their immunotoxicity, two different sized PLA NPs were produced and characterized regarding their mean size, PDI and zeta potential (**Figure 1**). PLA<sup>A</sup> NPs presented a mean diameter of 187.9 ± 36.9 nm and a zeta potential of −24.0 ± 4.7 mV in pyrogen-free water, while PLA<sup>B</sup> NPs presented a mean diameter of 109.1 ± 10.4 nm and a zeta potential of −6.6 ± 11.2 mV, both presenting a low PDI compatible with only one narrow-size population of particles (see graphics on **Figure 1**). The more negative charge of PLA<sup>A</sup> NPs could be explained by the higher concentration of Pluronic F68 used in the NP production method, since increased surface layer of surfactant may decreases the NPs zeta potential (Santander-Ortega et al., 2006). Sizes were also analyzed after dispersion in cell culture media, in order to evaluate the stability of the NPs in the experimental assay conditions. These tests were performed right after dispersion in DMEM and RPMI, and 24 h after incubation at 37◦C. Results from initial dispersion and after 24 h incubation were comparable (**Figure 1**), so the 24 h-incubation period did not altered the characteristics of the particles. However, great differences, when compared with the initial size (pyrogen-free water), were observed when the particles were suspended in RPMI, but especially in DMEM. In case of PLA<sup>A</sup> NPs, the size decreased and in case of PLA<sup>B</sup> NPs the size increase. To better understand the differences, representative graphs of differential and cumulative intensities of size distribution were obtained for both particles, suspended in pyrogen-free water and after 24 h incubation in RPMI and DMEM. When comparing the PLA<sup>A</sup> graphs from cell culture media with the ones obtained in the original medium (pyrogen-free water), we observed the appearance of 3 size-populations, compatible with a higher PDI. To highlight, the appearance of a small size population of particles explaining the decrease of the mean size diameter. The same phenomenon was not observed with PLA<sup>B</sup> NPs. On the contrary, in RPMI the size remained unaltered and in DMEM the size increased as a result of some aggregation of the particles. In order to confirm the initial differences in size between PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs, TEM images were obtained with particles dispersed in pyrogen free water (**Figure 2**). As illustrated, both NPs are round shaped and sizes confirmed the DLS measurements.

# Both PLA NPs Present a Good Hemocompatibility Profile

Hemolysis is the breakdown of red blood cells with subsequent release of intracellular contents. In vivo, this can lead to anemia or other pathological conditions (Dobrovolskaia et al., 2008). It is important to assess the NP effect on these blood elements not only when the intravenous route of administration is considered but also when addressing other administration routes, in order to establish their hemocompatibility (Dobrovolskaia et al., 2008). For that reason, PLA NP hemocompatibility was assessed in human whole blood and hemolytic values were considered above 5%, as recommended by American Society for Testing and Materials International (ASTM, 2013).

The results from **Figure 3** showed that both PLA NPs (A and B) have a good hemocompatibility profile, since none induced hemolysis above 5%, considering the concentration range tested (38–250µg/mL for PLA<sup>A</sup> NPs and 75–400µg/mL for PLA<sup>B</sup> NPs).

# PLA<sup>A</sup> NPs Show a Pronounced Cytotoxicity Profile in Comparison to PLA<sup>B</sup> NPs in RAW 264.7

The colorimetric MTT assay for measuring cell metabolic activity is based on the cellular conversion of a tetrazolium salt (MTT) into an insoluble formazan, that can be dissolved in DMSO generating a purple signal (Altmeyer et al., 2016). Therefore, through an indirect way, MTT assay was used to evaluate the cytotoxicity of PLA NPs after 24 h incubation with PBMCs and RAW 264.7.

Results presented in **Figure 4A** show that neither PLA<sup>A</sup> NPs nor PLA<sup>B</sup> NPs induced cytotoxicity in PBMCs, since the incubation with both resulted in cell viabilities above 70% under the concentration range tested (0.55–562.5µg/mL for PLA<sup>A</sup> NPs and 1.05–536µg/mL for PLA<sup>B</sup> NPs). Importantly, the similarity in the cytotoxicity profile in this primary culture could be explained by the similar mean diameter and zeta potential of PLA NPs when dispersed in RPMI medium. In fact, the differences in size previously seen in water were masked in RPMI.

Concerning cytotoxicity in RAW 264.7, we can observe in **Figure 4B** that PLA<sup>A</sup> NPs presented a higher cytotoxicity than PLA<sup>B</sup> NPs, since they presented an estimated IC<sup>50</sup> of 540.6µg/mL, while with PLA<sup>B</sup> NPs cell viabilities below 70% were never observed, and therefore the estimation of IC<sup>50</sup> was not possible under the concentration range tested (1.05–536µg/mL). These results are probably correlated with the size of the both PLA NPs in DMEM (RAW 264.7) medium. In case of the PLA<sup>A</sup>

FIGURE 1 | Characterization of PLA NPs. Particle mean size distribution (nm), polydispersity index (PDI), zeta potential (mV) and illustrative graphics of differential and cumulative intensities, after concentration and resuspension in pyrogen-free water, or after 24 h of incubation in cell culture media (DMEM medium or RPMI medium). Data are presented as mean ± standard deviation (SD), n ≥ 3 (three or more independent experiments, each in triplicate).

NPs, the presence of great NP population with a size below 25 nm might explain their higher cytotoxicity.

In these experiments, the control with the stock solution vehicle (mainly pyrogen-free water from the last NP wash), was tested in the volume correspondent to the highest NP concentration and no decrease in cell viability was verified. These controls ensured that the decrease in cell viability is from the NPs in suspension and not the vehicle of the NP suspension.

In order to avoid possible excessive assumption regarding cytotoxicity when using only a metabolic assay, these results were confirmed with PI assay, which evaluates the integrity of the cell membrane. Results for PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs in both cellular models were similar to the ones obtained with MTT (**Figure 4**) and confirmed the higher toxicity of PLA<sup>A</sup> NPs in RAW 264.7 cells.

# PLA<sup>A</sup> NPs but Not PLA<sup>B</sup> NPs Induce a Significant Concentration-Dependent ROS Production

The ROS, such as superoxide or hydrogen peroxide, are continually produced during metabolic processes (Brüne et al., 2013; Kwon et al., 2017). ROS generation is normally counterbalanced by the action of antioxidant enzymes and other redox molecules (Brüne et al., 2013; Kwon et al., 2017). However, when overproduced by activated macrophages, ROS can lead to cellular injury (Circu and Aw, 2010; Brüne et al., 2013; Kwon et al., 2017). It has been proven by Saini and co-workers that NPs may promote apoptotic cell death, through the induction of oxidative stress by accumulating ROS (Saini et al., 2016). Therefore, it is important to evaluate the potential effect of PLA NPs in ROS production. This assay was performed using

FIGURE 2 | TEM images of PLA NPs dispersed in pyrogen-free water (scale bar: 100 nm). (A,B) PLAA NPs; (C,D) PLAB NPs.

the cell-permeable fluorogenic probe DCFH-DA, which can be detected on a standard fluorometric plate reader (Zolnik et al., 2011). ROS production assay in RAW 264.7 was performed after 24 h of incubation and as demonstrated in **Figure 5A**, there was a concentration-dependent ROS production for PLA<sup>A</sup> NPs. The same effect was not observed for the PLA<sup>B</sup> NPs, even considering that a lower PLA<sup>A</sup> NP concentrations were tested, when compared with PLA<sup>B</sup> NPs (4.3–340µg/mL for PLA<sup>A</sup> NPs and 8.6–690µg/mL for PLA<sup>B</sup> NPs). We could hypothesize that this concentration-dependent ROS production is an indication of cellular toxicity, as demonstrated by the cell viability assay in the **Figure 5B**, where for the higher PLA<sup>A</sup> NP concentration the resultant cellular viability was near 70%. For PLA<sup>B</sup> NPs it was observed an increased trend of ROS production. However, the values observed were not statistically different from the unstimulated cells. Furthermore, in opposition to the results of PLA<sup>A</sup> NPs, no trend for decrease in cell viability was shown for PLA<sup>B</sup> NPs (**Figure 5B**).

# PLA NPs Do Not Have an Inflammatory Potential in RAW 264.7

The NO is a reactive nitrogen specie, produced by nitric oxide synthase enzymes (Boscá et al., 2005; Caruso et al., 2017). It is an important inflammatory mediator released by macrophages during inflammation, and is one of the main cytostatic, cytotoxic, and pro-apoptotic mechanisms of the immune response (Boscá et al., 2005; Caruso et al., 2017). In order to assess the inflammatory or anti-inflammatory properties of PLA NPs, NO production by RAW 264.7 cells was measured using the Griess reaction method after 24 h of incubation with different test samples.

The pro-inflammatory effect of PLA NPs was evaluated by measuring the NO release upon stimulation with NPs, and the anti-inflammatory effect was evaluated by measuring the ability of the NPs to inhibit NO release induced by LPS. In the first approach, none of the PLA NPs induced NO production under the concentration range tested (0.5–50µg/mL for PLA<sup>A</sup> NPs and 1–100µg/mL for PLA<sup>B</sup> NPs) (**Figure 5C**). Importantly, these concentrations were chosen because they did not induce significant cellular death under the assay conditions, and higher concentrations would result in cellular death above 30%, which could compromise NO production (**Figure 5D**).

The second approach, using the same concentration ranges, revealed that both PLA NPs did not inhibited the NO production stimulated with LPS (**Figure 5E**) and test conditions did not significantly reduce cell viability (**Figure 5F**).

# DISCUSSION

According to our results, PLA NPs did not present hemolytic activity in concentrations up to 250 and 400µg/mL for PLA<sup>A</sup> and PLA<sup>B</sup> NPs, respectively. Importantly, these are very high concentrations, far from the reality of in vivo administrations. In fact, apart from the fact that the experiment is performed with diluted blood (>10 times diluted), 250µg/mL would correspond to a intravenously injected human dose of 1400 mg of NP and 400µg/mL to a dose of 2240 mg [in a 70 kg person, with 5.6 L of blood (Dobrovolskaia and McNeil, 2013)]. Results confirm therefore the hemocompatibility of PLA NPs and are accordant with Altmeyer and co-workers, who described that no erythrocyte damage is caused by blank PLA NPs produced by an emulsion/solvent evaporation method with polyvinyl alcohol (PVA) (Altmeyer et al., 2016).

One of the most important conclusions herein presented is that even small changes in the physicochemical characteristics of similar NPs can originate different cytotoxicity profiles. In detail, results from RAW 264.7 suggested that PLA<sup>A</sup> NPs induced the higher toxicity, and data from the NP characterization in the experiment conditions revealed these NPs presented the smaller mean diameter, resultant from a higher heterogeneity of the NP population, with emphasize for a population presenting a

mean diameter of 10 nm. However, in PBMCs, both PLA NPs presented a similar cytotoxicity profile. Interestingly, in RPMI medium, used for PBMCs experiments, PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs presented a similar mean size and a more similar sizedistribution profile than in DMEM medium. Considering these results, we can hypothesize that the smaller NP population in PLA<sup>A</sup> NPs, resultant from a modification after dispersion in cell culture medium, is contributing to the increased toxicity of PLA<sup>A</sup> NPs. These results are concordant with the concept that smaller NP can induce more cellular damages, due to increased ability to enter the cells, and particularly, sizes <10 nm can even reach the cell nucleus (Sukhanova et al., 2018). For instance, in a recent study (da Luz et al., 2017) it was proposed that small sized PLA NPs were mainly internalized in A549 cells through clathrin-coated pits in detriment of other endocytic pathways. In the future, the assessment of the mechanisms involved in the uptake of PLA<sup>A</sup> NPs and PLA<sup>B</sup> NPs could help clarify the cause of increased toxicity observed in PLA<sup>A</sup> NPs.

The evaluation of the ROS production confirmed the correlation of the different toxicity profile of PLA NPs with the NP physicochemical differences and highlighted the importance of performing case-by-case evaluations. In fact, we demonstrated that PLA<sup>A</sup> NPs induced a concentration-dependent ROS production, whereas PLA<sup>B</sup> NPs did not stimulate statistically significant ROS production even with higher concentrations. A published report from Singh and co-workers (Singh and Ramarao, 2013) suggested that PLA NPs (emulsion-diffusionevaporation method using PVA) induced no effect on ROS production up to 100µg/mL concentration, whereas 300µg/mL showed 1.5- to 2-fold stimulation of ROS production. Their results are in agreement with ours for PLA<sup>A</sup> NPs, however, they are not aligned with the results from PLA<sup>B</sup> NPs. This stresses the importance of an adequate evaluation when testing distinct polymeric nanomaterials rather than excessively extrapolating conclusions.

According to literature, PLA may induce inflammatory responses, due to its hydrophobicity, lack of bioactivity, and release of acidic degradation by-products (Li and Chang, 2004; Farah et al., 2016; Yoon et al., 2017). Nevertheless, this study showed that PLA polymer properties are not fully exchangeable with nanosized PLA particles. Actually, we showed that both PLA NPs produced within this study did not present effects on NO production under the concentration range tested, suggesting it does not induce an inflammatory response in RAW 264.7.

Importantly, during the execution of these studies we also hypothesized that the accentuated toxicity profile presented by PLA<sup>A</sup> NPs could be related to the use of a higher concentration of Pluronic F68 during the production. Despite

NPs inhibitory effect on NO production, PLA NPs were incubated simultaneously with LPS. The percentage of NO inhibition was calculated considering 100% the NO production induced by LPS without PLA NPs (F). Cell viability assay (MTT) after the performance of NO inhibition assay. Data are presented as mean ± SEM, n ≥ 3 (three or more independent experiments, each in triplicate). \*\*\*p < 0.001 and \*\*\*\*p < 0.0001.

we have washed the PLA<sup>A</sup> NPs more exhaustively than PLA<sup>B</sup> NPs to remove the surfactant, the negative zeta potential in pyrogen-free water gave an indication that PLA<sup>A</sup> NPs could have more surfactant on its surface. To better understand whether PLA<sup>A</sup> NPs accentuated effect on ROS production could result from Pluronic F68, the assay was repeated using a range of surfactant dilutions in water (0.00025–0.25%) and no pro-oxidative effect was verified and also no decrease in cell viability.

Lastly, polymeric NPs application into clinical research is dependent on more accurate knowledge of the NP interactions with the human body (Hoshyar et al., 2016). To address this issue, well-executed in vitro studies are needed to establish relationships between their biological activity and their physicochemical properties, such as the NP size (Hoshyar et al., 2016). In this sense, exploiting PLA NPs properties correlation with toxicity in a rigorous way represented an interesting challenge for our research group. Accordingly, important recommendations were considered for the development of this work, such as the detailed characterization of the NP physicochemical properties in the original medium (pyrogenfree water) and in in vitro assay conditions, the inclusion of positive and negative controls, as well as the assessment of the NP interference before implementing testing protocols. To highlight, in every experiment, the NPs solvent (vehicle control) were also evaluated, in order to ensure that the observed effects were specific from PLA NPs. Also, for cytotoxicity assessment, more than one cell type was used to estimate the same endpoint, and two different methodologies (MTT and PI) were employed to confirm the results. These details shall increase the results reliability and relevance, as extensively discussed by (Drasler et al., 2017).

#### CONCLUSION

In this study, we observed that size highly influences PLA NPs toxicity profile. A new hypothesis to be confirmed in future arose in the course of this work. The smaller NPs are able to induce higher cellular toxicity, particularly mediated by ROS production. Nevertheless, the effect of size was only accurately addressed after characterization in in vitro assay conditions. Indeed, we exposed the influence of cell culture media on these polymeric NPs physicochemical characteristics and the respective repercussions on their toxicity. This report illustrates how an adequate NP characterization is crucial, in order to avoid misinterpretations, and consequent ambiguous conclusions. This remark can be further transposable to in vivo conditions, since the contact of NP with biological solutions, such as blood, saliva, nasal or gastric fluids can change the NP physicochemical properties, and those are known to be essential for the generation of a biological effect (Oh and Park, 2014).

We strongly believe that this study will help other research groups to achieve better understanding of their results and to obtain improved conclusions supporting the current scientific evidence.

#### REFERENCES


### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the supplementary files.

### AUTHOR CONTRIBUTIONS

JD carried out the experimental work. JD, SJ, and OB conceived and planned the experiments. All authors discussed the results and contributed to the final manuscript, reading, and approving the submitted version.

# FUNDING

This work was financed by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational Programme under project CENTRO-01-0145-FEDER-000008:BrainHealth 2020, and through the COMPETE 2020 - Operational Programme for Competitiveness and Internationalization and Portuguese national funds via FCT – Fundação para a Ciência e a Tecnologia, I.P., under project PROSAFE/0001/2016, and the strategic projects POCI-01-0145-FEDER-030331 and POCI-01-0145-FEDER-007440 (UID/NEU/04539/2019).

# ACKNOWLEDGMENTS

The authors thank Dr. Ana Donato for expertise and assistance in hemocompatibility studies in the Clinical Analysis Laboratory of the Faculty of Pharmacy of Coimbra University (Portugal), Prof. Dr. Manfred Zinn for the determination of the exact MW of PDLLA by gel permeation chromatography/ size exclusion chromatography in the University of Applied Sciences and Arts Western Switzerland (HES-SO//Valais – Wallis) and Dr. Mónica Zuzart for the TEM microscopy analyses that were performed at iLAB—Bioimaging Laboratory of the Faculty of Medicine of the University of Coimbra.


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

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

# Computational Assessment of the Pharmacological Profiles of Degradation Products of Chitosan

Diana Larisa Roman<sup>1</sup> , Marin Roman<sup>1</sup> , Claudia Som<sup>2</sup> , Mélanie Schmutz <sup>2</sup> , Edgar Hernandez <sup>2</sup> , Peter Wick <sup>3</sup> , Tommaso Casalini <sup>4</sup> , Giuseppe Perale<sup>4</sup> , Vasile Ostafe<sup>1</sup> and Adriana Isvoran<sup>1</sup> \*

*<sup>1</sup> Advanced Environmental Research Laboratories, Department of Biology-Chemistry, Faculty of Chemistry, Biology, Geography, West University of Timisoara, Timisoara, Romania, <sup>2</sup> Empa, Swiss Federal Laboratories for Materials Science and Technology, Technology and Society Laboratory, St. Gallen, Switzerland, <sup>3</sup> Empa, Swiss Federal Laboratories for Materials Science and Technology, Particles-Biology Interactions Laboratory, St. Gallen, Switzerland, <sup>4</sup> Department of Innovative Technologies, University of Applied Sciences and Arts of Southern Switzerland (SUPSI), Manno, Switzerland*

#### Edited by:

*Michele Iafisco, Italian National Research Council (CNR), Italy*

#### Reviewed by:

*Sinosh Skariyachan, Dayananda Sagar Institutions, India Naveed Muhammad, Nanjing Medical University, China*

> \*Correspondence: *Adriana Isvoran adriana.isvoran@e-uvt.ro*

#### Specialty section:

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

Received: *10 June 2019* Accepted: *22 August 2019* Published: *06 September 2019*

#### Citation:

*Roman DL, Roman M, Som C, Schmutz M, Hernandez E, Wick P, Casalini T, Perale G, Ostafe V and Isvoran A (2019) Computational Assessment of the Pharmacological Profiles of Degradation Products of Chitosan. Front. Bioeng. Biotechnol. 7:214. doi: 10.3389/fbioe.2019.00214* Chitosan is a natural polymer revealing an increased potential to be used in different biomedical applications, including drug delivery systems, and tissue engineering. It implies the evaluation of the organism response to the biomaterial implantation. Low-molecular degradation products, the chito-oligomers, are resulting mainly from the influence of enzymes, which are found in the organism fluids. Within this study, we have performed the computational assessment of pharmacological profiles and toxicological effects on human health of small chito-oligomers with distinct molecular weights, deacetylation degrees, and acetylation patterns. Our approach is based on the fact that regulatory agencies and researchers in the drug development field rely on the use of modeling to predict biological effects and to guide decision making. To be considered as valid for regulatory purposes, every model that is used for predictions should be associated with a defined toxicological endpoint and has appropriate robustness and predictivity. Within this context, we have used FAF-Drugs4, SwissADME, and PreADMET tools to predict the oral bioavailability of chito-oligomers and SwissADME, PreADMET, and admetSAR2.0 tools to predict their pharmacokinetic profiles. The organs and genomic toxicities have been assessed using admetSAR2.0 and PreADMET tools but specific computational facilities have been also used for predicting different toxicological endpoints: Pred-Skin for skin sensitization, CarcinoPred-EL for carcinogenicity, Pred-hERG for cardiotoxicity, ENDOCRINE DISRUPTOME for endocrine disruption potential and Toxtree for carcinogenicity and mutagenicity. Our computational assessment showed that investigated chito-oligomers reflect promising pharmacological profiles and limited toxicological effects on humans, regardless of molecular weight, deacetylation degree, and acetylation pattern. According to our results, there is a possible inhibition of the organic anion transporting peptides OATP1B1 and/or OATP1B3, a weak potential of cardiotoxicity, a minor probability of affecting the androgen receptor, and phospholipidosis. Consequently, these results may be used to guide or to complement the existing *in vitro* and *in vivo* toxicity tests, to optimize biomaterials properties and to contribute to the selection of prototypes for nanocarriers.

Keywords: chito-oligomers, ADME-Tox, pharmacokinetics, toxicity endpoints, biological effects

# INTRODUCTION

Chitin is a polysaccharide abundantly found in nature, especially in crustaceans, but also in insects and fungi. Chitosan is obtained from chitin by chemical or enzymatic deacetylation (Rinaudo, 2006). The difference between chitin and chitosan consists of the acetyl content, chitin contains mostly N-acetyl-D-glucosamine (GlcNAc, A) units and chitosan contains especially D-glucosamine (GlcN, D). Chitin and chitosan reveal biocompatibility, biodegradability, and nontoxicity for humans and the environment (Rinaudo, 2006). These characteristics, added to the anti-fungal, anti-bacterial, antimicrobial, and anti-oxidant properties of chitosan conducted to numerous applications in different fields: food industry, cosmetic and personal care products, wastewater management, pharmacological products, and implantable materials (Enescu and Olteanu, 2008; Raafat and Sahl, 2009; Cheung et al., 2015). Chitosan nanoparticles are approved by the Food and Drug Administration (FDA) for tissue engineering and drug delivery and by FDA and EU for dietary use and wound dressing applications (Mohammed et al., 2017).

There are many different ways in which humans have exposure to chito-oligomers (COs): by the degradation of implanted materials based on chitosan, the use of pharmaceutical products containing COs and/or occupational exposure. The low-molecular degradation products of chitosan, the chitooligomers, occur as a result of the influence of enzymes which are present in bodily fluids (Saikia et al., 2015). Due to the limitations concerning the applications of chitosan polymer related to its higher viscosity and insolubility in neutral and basic environments (Giri et al., 2012; Ways et al., 2018), COs with an increased solubility and lower viscosity have been obtained by chemical or enzymatic hydrolysis of chitosan and are largely used in the pharmaceutical field (Enescu and Olteanu, 2008; Patrulea et al., 2015; Ways et al., 2018). As an example, food supplements containing COs and their derivatives are used by many people for treating osteoarthritis (Jerosch, 2011) but their clinical importance is unclear (Liu et al., 2018). Occupational exposure to these compounds may also occur.

The chitosan oligomers that result from the hydrolysis processes may be classified in two types: (i) homo-chitooligosaccharides containing only GlcN or GlcNAc units and (ii) hetero-chito-oligosaccharides, containing both GlcN and GlcNAc with varying degrees of deacetylation (the percent of glucosamine units in the oligomer, DDA) and a varying position of glucosamine residues in the oligomer chain (acetylation pattern, AP). GlcN unit carries an amino group that is protonated at physiological pH (Cheung et al., 2015). Both homo- and hetero-oligomers may differ in the degree of polymerization (the number the monomeric units within an oligomer, DP). Heterochito-oligomers with DP<10 are considered as water soluble (Liaqat and Eltem, 2018).

Taking into account the possible ways of exposure of humans to COs, the concept of Safe-by-design (SbD) must be considered when producing and using chitosan and its oligomers. Safeby-design is a relatively new concept planned to be used in the research and development (R&D) field and industry, and include safe design, safe production, safe use and waste management for new materials and technologies (van de Poel and Robaey, 2017). Safe design addresses safety early in the R&D and design phases, such as in the design of safe products for professionals, consumers and the environment, and is usually based on prediction and experimental testing (in vitro and in vivo short term assays) tools. Safe production considers the use of environmentally friendly technologies and the potential risk for professionals involved in R&D and industrial processes. Safe use and waste management reflect safe handling (by both consumers and professionals) of products and wastes. Safety profiles of COs should depend on their structure and physicochemical properties (van de Poel and Robaey, 2017). Computational studies may have a valuable contribution to assess the safe designing of compounds by predicting the biological effects and toxicity profiles and by correlating them with the physicochemical and structural properties.

COs have numerous pharmaceutical properties and medical applications: anti-microbial, anti-oxidant, anti-tumoral, anti-inflammatory, immunostimulatory, anti-obesity, antihypertension, and anti-Alzheimer (Mourya et al., 2011; Cheung et al., 2015; Muanprasat and Chatsudthipong, 2017). COs having DP<20 are soluble in water and reveal antimicrobial, antioxidant, antitumoral, antiviral, antiangiogenic, and prebiotic properties (Sánchez et al., 2017; Long et al., 2018; Lu et al., 2019). COs with a molecular weight lower than 2000 Da and DDA>90% revealed a moderate neuroprotective activity and no toxicity against neurons (Santos-Moriano et al., 2018) and COs with molecular weight lower than 1000 Da showed high antioxidant activity (El-Sayed et al., 2017). Literature data reveal that COs having DP>6 possess enhanced anti-tumor, antimicrobial and immunopotentiation properties, favorable biological activities of smaller COs being also reported (Liaqat and Eltem, 2018). These activities of COs are dependent on their physicochemical properties: molecular weight (MW), the degree of deacetylation (DDA), the degree of polymerization (DP), and charge distribution (acetylation pattern, AP) (Park et al., 2011), but their structure–activity relationships are rather unknown (Santos-Moriano et al., 2018).

Specific literature contains little or no information about the biological effects of every possible variant of COs (Liaqat and Eltem, 2018). A chito-oligomer with specific physicochemical properties (MW, DDA, AP) may display all, some or none of all considered bioactivities of COs (Sánchez et al., 2017). The low reproducibility of results and sometimes the opposite reported effects concerning the COs biological activities could mainly be due to relatively poorly characterized oligomer mixtures that have been used in experimental studies and/or to inconstant reporting of the properties of COs (Mourya et al., 2011; Park et al., 2011; Li et al., 2016). Furthermore, available information concerning the effects of COs is mainly derived from in vitro and in vivo studies with animal models and there are limited human experimental data concerning the absorption, distribution, metabolism, excretion, and toxicity of these compounds (Cheung et al., 2015; Phil et al., 2018). It reflects the insufficiency of safety data concerning their use by humans.

The number of used chemicals is increasing constantly and there is a need to assess their safety for humans and environment. Considering the costs of the laboratory studies and the ethical concerns on using animals for testing, the role of bioinformatics in hazard assessment is well-recognized. The explosive growth in the magnitude and diversity of data from physics, chemistry, and biology conducted to the creation of specific databases and computational packages for data manipulation (Luechtefeld and Hartung, 2017). The easy access to these data allowed scientists to build accurate computational models for toxicology assessment. These models are used in drug discovery and development and for assessment of the effects of xenobiotics on humans and environment. Computational tools that were developed for hazard assessment include (quantitative) structure-activity relationships [(Q)SARs], read-across methods, expert rule-based (structural alerts) methods, and molecular modeling techniques (Alves et al., 2018a; Myatt et al., 2018; Yang et al., 2018). The Organization of Economic and Co-operation Development (OECD) created QSAR guidelines already in 2004 and the principles for the construction of (Q)SAR models, computational methods, and model validation methods are described in detail since 2007 (Fjodorova et al., 2008; Lo Piparo and Worth, 2010). Furthermore, REACH (Registration, Evaluation and Authorization of CHemicals) regulation mentions the QSAR techniques for studying the toxicological profile of chemicals (Kleandrova and Speck-Planche, 2013). Consequently, QSAR has been a reliable computational tool used for decades for connecting the properties and biological activity. Taking into consideration the limitations, the number of improvements has been recorded and various descriptors have been explored: molecular properties (0D-QSAR), fragment counts (1D-QSAR), topological descriptors (2D-QSAR), spatial coordinates (3D-QSAR), a combination of atomic coordinates and sampling of conformations (4D-QSAR), multiple expression of ligand topology (5D-QSAR), considering the solvation function (6D-QSAR) and receptor or target-based receptor model data (7D-QSR) (Kar and Leszczynski, 2019). Nowadays, many software and web servers are available for predicting chemical toxicity before synthesis, as it is recognized that computational techniques provide high-quality predictions for chemical hazard assessment (Yang et al., 2018), meaning 2D-QSAR and 3D-QSAR methods are frequently used.

Besides the large applicability of these modern tools for human and environment hazard assessment, there also are some limitations, mostly related to the robustness and predictability of the used models and to the fact that they do not provide a clean mechanistic interpretation of the outcomes (Luechtefeld and Hartung, 2017). The methods mentioned above are not applicable for assessing the pharmacokinetics of nanoparticles, especially due to the fact that fundamental mechanisms that support drug-handling within the human organism are not understood for nanoparticles (Nel et al., 2009; Beddoes et al., 2015). Also, as most QSAR models are based on in vivo or in vitro data from specific experimental conditions, the applicability domain of the QSAR model is more limited for nanomaterials (Choi et al., 2018). Furthermore, data concerning the effects of the oligomer components cannot be transferred to nanoscale polymers since in the case of nanoparticles, not only the dose and their elemental composition, but their shape, size, and surface characteristics determine the biological activities and therapeutic effects and it increases the difficulty of modeling the biological effects of nanomaterials (Nel et al., 2009; Beddoes et al., 2015). Consequently, within this study we focus on the chito-oligomers both as degradation products of chitosan nanoparticles and as independent food supplements.

The objectives of this study are: (1) prediction of the pharmacological profiles and toxicological endpoints (skin sensitization potential, endocrine disruption potential, cardiotoxicity, hERG channel blocking potential, carcinogenicity, and mutagenicity) of COs containing up to 8 monomeric units (water soluble chito-oligomers) and (2) assessment of the influence of the MW, DDA, and AP on the toxicological and pharmacological profiles of investigated COs by using computational approaches.

### MATERIALS AND METHODS

Among the numerous available computational tools for predicting the pharmacological properties and toxicological effects of chemical compounds on human health, we have selected those with an accuracy of a prediction usually higher than 70% and with friendly interfaces and tutorials that are available for free (online or open-source). The chitooligomers that we have considered in this study are presented in **Table 1** together with the computed values of their molecular weights using admetSAR2.0 tool (see further). We specify that each amino group of the deacetylated units is protonated. Furthermore, **Table 1** shortly reviews known information concerning medical and side/toxicological effects of small COs. Some of these compounds are approved by FDA only as food supplements and/or for use in wound dressings (Wedmore et al., 2000). In Europe, GlcN and GlcNAc are approved as drugs in the form of glucosamine sulfate (Jordan et al., 2003).

The simplified molecular-input line-entry system (SMILES) formulas of the considered COs are built using ACD/ChemSketch software (https://chemicalize.com). This software also generates structural files in mol format to be used for further predictions We have obtained 3D sdf files using OpenBabel (O'Boyle et al., 2011) on the online server http://www.cheminfo.org/Chemistry/Cheminformatics/ FormatConverter/index.html, starting from their structural files in mol format generated by ACD/ChemSketch software. Structure minimization has been done using Chimera software (Pettersen et al., 2004) using 1000 steepest descent steps, step size 0.02 Å, 10 conjugate gradient steps, conjugate gradient step size 0.02 Å.

FAF-Drugs4 (Lagorce et al., 2017) tool has been considered for assessing the oral bioavailability as a part of the pharmacokinetic profile and the overall toxicity of investigated COs. This is a rule-based tool having the accuracy of predictions higher than 70% (Lagorce et al., 2017). FAF-Drugs4 tool allows filtering against Lipinski's rule (Lipinski et al., 2001), Egan's rule (Egan et al., 2000), and Veber's rule (Veber et al., 2002) for predicting



*(Continued)*

TABLE 1 | Continued


bioavailability and of Pfizer's and GSK rules for predicting the overall toxicity (Gleeson, 2008).

SwissADME is a web tool that allows the computation of the physicochemical parameters of a chemical compound, its pharmacokinetic profile, drug likeness and medicinal chemistry, starting from the SMILES formula, the accuracy of predictions being between 72 and 94% (Daina et al., 2017).

AdmetSAR2.0 (Cheng et al., 2012; Yang et al., 2019) tool has been used to predict pharmacokinetic profiles and organ (eye, heart, liver) and genomic toxicity of investigated COs. with a predictive accuracy of 72.3–76.7% (Yang et al., 2019). Furthermore, every prediction made by this tool is quantitatively described by a probability output.

PreADMET is also a web tool having four parts: (i) molecular descriptors calculation; (ii) drug likeness prediction considering well known rules; (iii) ADME prediction; and (iv) toxicity prediction [mutagenicity by Ames test and rodent carcinogenicity; (Lee et al., 2003, 2004)].

Because occupational exposure to chitin and chitosan may also occur through dermal contact and skin sensitization, it may have a significant impact on individual working capacity and quality of life, we have assessed the skin sensitizer potential of investigated COs using Pred-Skin computational tool (Braga et al., 2017; Alves et al., 2018b). This information is also important when we take into account the fact that chitosan is approved to be used in wound healing purposes. Pred-Skin is a web-based computational facility considering QSAR models of skin sensitization potential. It performs the following predictions: (i) binary predictions of human skin sensitization potential established taking into account human data (prediction accuracy being 73–76%); (ii) binary predictions of murine skin sensitization potential taking into account animal data (LLNA, prediction accuracy being 70–84%); (iii) binary predictions based on Direct Peptide Reactivity Assay (DPRA), KeratinoSens, and the human Cell Line Activation Test (h-CLAT) data (prediction accuracy being 80–86%); (iv) a consensus model that is generated by averaging the predictions of individual models (prediction accuracy being 70–84% (Braga et al., 2017; Alves et al., 2018b).

Predictions concerning carcinogenicity and mutagenicity are also obtained using Toxtree software, the accuracy of predictions being 70% (Patlewicz et al., 2008).

CarcinoPred-EL (Carcinogenicity Prediction using Ensemble Learning methods) utility has been used for accomplishing predictions concerning the carcinogenicity of investigated chemicals (Zhang et al., 2017). It is a free prediction online server that is based on twelve different molecular fingerprints and three ensemble machine learning models (Ensemble RF, Ensemble SVM, and Ensemble XGBoost) permitting the identification of the structural features related to carcinogenic effects of chemical compounds (Zhang et al., 2017).

Pred-hERG is another free accessible web tool that builds predictive models of the ability of a chemical compound to inhibit the human ether-à-go-go related gene (hERG) K<sup>+</sup> channels. This hERG K<sup>+</sup> channel blockage may result in cardiac side effects such as heart arrhythmia and even possibly death (Braga et al., 2015). Consequently, hERG K<sup>+</sup> channel blockage is one of the most important toxicological endpoints to be considered when assessing the safety of chemical compounds. There are two outcomes when using Pred-hERG tool: a binary prediction (hERG non-blocker or blocker) and a multiclass prediction (hERG non-blocker, weak/moderate blocker, strong blocker) along with the probability of the prediction for each class. The predictions have an accuracy of up to 89% (Braga et al., 2015).

Endocrine disruption potential is evaluated using ENDOCRINE DISRUPTOME computational tool (Kolšek et al., 2014). This tool uses the molecular docking approach for predicting interactions between the explored compound with 12 distinct human nuclear receptors, those binding sites are known: androgen receptor (AR), estrogen receptors α (ERα) and β (ERβ), glucocorticoid receptor (GR), liver X receptors α (LXRα) and β (LRXβ), peroxisome proliferator-activated receptors α (PPRAα), β/δ (PPRAβ), and γ (PPRAγ), retinoid X receptor α (RXRα) and thyroid receptors α (TRα), and β (TRβ). Both agonistic and antagonistic (an) effects are predicted for AR, ERα, ERβ, and GR. Predictions are based on computation of the sensitivity (SE) parameter and compounds are categorized in four classes: (i)


TABLE 2 | Short presentation of the computational tools that were used in the current study.

compounds with SE<0.25 expose a high probability of binding to nuclear receptors; (ii) compounds with 0.25<SE<0.50 reflect a medium probability of binding to the nuclear receptors; (iii) compounds having 0.50<SE<0.75 emphasize minor probability of binding and (iv) compounds with SE>0.75 reveal a low probability of binding to the nuclear receptors (Kolšek et al., 2014).

A summary of the computational tools that we have used in this study is presented in **Table 2**.

The computational tools that are used in this study have been elaborated for assessing the pharmacological profiles and toxicological endpoints of new drugs, but they were successfully applied for other classes of chemicals: cosmetic ingredients and pesticides (Alves et al., 2018c; Roman et al., 2018a; Gridan et al., 2019), synthetic steroids found on the market as food supplements or veterinary drugs (Roman et al., 2018b), water soluble derivatives of chitosan (Isvoran et al., 2017). It illustrates their applicability for predicting pharmacological properties and toxicological endpoints for many classes of compounds.

#### RESULTS

Estimation of the oral bioavailability and overall toxicity of investigated COs is obtained using FAF-Drugs4 tool and is based on filtering the physicochemical properties of investigated compounds in accordance with the rules mentioned above. FAF-Drugs4 tool also estimates if the investigated compounds are able to produce phospholipidosis (PI). The outcomes are presented in **Table 3**.

With the exceptions of the monomeric and some of the dimeric chito-oligomers, the outcomes of FAF-Drugs4 indicate the lack of oral bioavailability of the other investigated COs because of their molecular weight and extensive hydrogen bonding potential. Similar results concerning the lack of human oral bioavailability of investigated COs containing more than 2 monomeric units have been obtained using admetSAR2.0 (**Figure 1**), PreADMET and SwissADME tools (**Supplementary Table 1**).

As expected, the oral bioavailability decreases with increasing molecular weight and increases with the deacetylation degree. PreADMET predictions concerning the percent of the human intestinal absorption (HIA, **Supplementary Table 2**) reveal a mean absorbance (20%<HIA<70%) (Aswathy et al., 2018) for the monomeric units, the highest value (60.25%) being registered for the GlcN oligomer. Chito-oligomers containing two monomeric units reflect a poor absorption (HIA<20%) and the other COs do not reflect intestinal absorption (HIA=0). Predictions obtained using SwissADME tool reveal low gastrointestinal absorption (GI) for all investigated oligomers (**Supplementary Table 3**). All of these results suggest that smaller and deacetylated COs could be better absorbed at the gastrointestinal level and it facilitates their access to systemic circulation and distribution through the human body.

Predictions concerning the distribution (expressed as the probability of plasma protein binding—PPB, being Pglycoprotein substrate and/or inhibitor, being able to penetrate the blood-brain barrier– BBB) of the investigated COs have been obtained using admetSAR2.0 tool and the outcomes are illustrated in **Figure 2**. Negative values of the probabilities illustrate that the investigated activity is absent.

Data presented in **Figure 2** illustrate that investigated COs reveal a very low probability to bind to plasma proteins, they are not able to penetrate the blood brain barrier and to affect the central nervous system, oligomers with more than 3 monomeric units reflect a small probability to inhibit the P-glycoprotein and none of the investigated COs is considered as P-glycoprotein substrate. There are small differences in the values of predicted probabilities for a given activity between oligomers with the same DP and different DDA, reflecting the influence of the deacetylation degree on the activity of chito-oligomers. Almost similar predictions


TABLE 3 | Estimation of oral bioavailability and overall toxicity of chito-oligomers: green cells correspond to respected rules (0 violations), yellow cells correspond to partially respected rules (maximum 2 violations for Lipinski's rule and 1 violation for Veber's and Eagan's rules), light red cells correspond to broken rules.

*The number of violation for every considered rule is specified. Compounds expected to not induce phospholipidosis (PI) are marked by "No" in green cells and compounds expected to induce phospholipidosis are marked by "yes" in light red cells. (MW-molecular weight, HBA – hydrogen bond acceptors, HBD- hydrogen bonds donors).*

are obtained using PreADMET (**Supplementary Table 2**) and SwissADME (**Supplementary Table 3**) tools. PreADMET reveals that investigated COs are not inhibitors of the P-glycoprotein and outcomes values for the blood brain barrier penetration that are lower than 0.1 that correspond to a low absorbance to central nervous system (Aswathy et al., 2018). The PPB binding assessment (percentage of drug bound in plasma protein) reveals that investigated COs exhibit low binding energy with plasma proteins (PPB<90%) (Kandagalla et al., 2017) with the exception of the oligomer 6D that shows a high binding energy to plasma proteins (93.384%). SwissADME predicts that investigated compounds are not able to penetrate the blood brain barrier and are considered as substrates of P-glycoprotein.

SwissADME, PreADMET, and admetSAR2.0 tools have been also used to assess the metabolism of COs by predicting the probability for every compound to be a substrate or to inhibit the human cytochromes P450 (CYP) involved in the metabolism of xenobiotics. The outcomes of SwissADME and admetSAR2.0 tools indicate that COs considered in this study are not substrates and inhibitors of CYPs (**Supplementary Tables 3, 4**).

the P-glycoprotein (P-gpS/P-gpI), being able to penetrate the blood-brain barrier (BBB). The predicted probabilities may take values between 0 and 1 when the investigated activity is present and between −1 and 0 when the activity is considered absent. Values closer to 1 correspond to effects that are highly probable and values closer to −1 correspond to highly improbable effects.

The predictions obtained using PreADMET are not similar, they indicate that investigated COs are not considered as substrates and inhibitors of CYP2C9 and CYP2C19, but the oligomers containing deacetylated units are possible dual inhibitors and substrates of CYP2D6 and CYP3A4. It may be due to complex modulation of CYP enzymes and this aspect should be further analyzed.

Predictions concerning the probability of the inhibition of the organic anion and cation transporter peptides by the investigated COs are also obtained using admetSAR2.0 tool and are illustrated in **Figure 3**. This figure suggests the inhibitory potential of all investigated COs against organic anion polypeptide transporter OATP1B3. Deacetylated oligomers also illustrate the inhibitory potential against OATP1B1 and underline the influence of the DDA on the activity of investigated COs.

Predictions concerning the organ (eye, heart, liver) and genomic toxicity of investigated COs obtained using admetSAR2.0 and PreADMET tools are revealed in **Table 4**.

None of the investigated oligomers have carcinogenic potential, does not produce eye irritations and corrosion. Excepting the monomeric and dimeric units that are not predicted to block the hERG, the other oligomers reveal moderate potentials of hERG channel blocking. Totally acetylated oligomers and small oligomers containing deacetylated units are predicted by the PreADMET tool as displaying mutagenic potential. Besides admetSAR2.0 and PreADMET tools, TABLE 4 | Predictions obtained using admetSAR and PreADMET tools concerning the probabilities of organ and genomic toxicity of investigated COs: hERG – potassium channel blocking potential (cardiotoxicity), ECeye corrosion, EI – eye irritation, HEPT – hepatotoxicity.


*Negative values of the probabilities illustrate that the investigated activity is absent. These probabilities may take values between* −*1 and 0 when the predicted activity is absent and between 0 and 1.00 when the predicted activity is present.*

Pharmacological

Roman et al. Pharmacological Profiles of Chito-Oligomers

TABLE 5 | Outcomes of the Pred-hERG computational tool concerning the blockage of the potassium channel by the investigated chito-oligomers: red cells illustrate predictions of hERG blocking potential and green cells illustrate hERG non-blocking potential.


*Number in every cell represents the probability of the prediction for each class. The values vary between 0 and 1.*

carcinogenicity and mutagenicity of investigated COs have also been assessed using CarcinoPred-EL and Toxtree and their outcomes displayed no carcinogenicity and no Ames toxicity for every considered oligomer (**Supplementary Table 5**). The consensus of the predictions made by the majority of these computational tools emphasizes that none of the investigated COs is expected to be carcinogen and mutagen. Data presented in **Table 4** reveal different values for the predicted probabilities for COs with distinct DDA and AP and it underlines the dependence of the biological activities of investigated COs on their properties, DDA, and AP.

For assessing cardiotoxicity of investigated COs we have also considered Pred-hERG prediction tool and the results are presented in **Table 5**.

Predictions made using the binary model illustrate that, except GlcN and GlcNAc monomers, that are considered as non-hERG K+ blockers, the other investigated COs are predicted as having hERG K+ blocking potential but the probabilities of these predictions are relatively small. Predictions based on the multiclass models reveal non-hERG blocking potential for all investigated COs, also with relatively small probabilities. Predictions reliability reported using Pred-hERG are ranging between 83 and 84% for the binary model and between 66 and 79% for the multiclass models (Braga et al., 2015). Consequently, we deliberate that, excepting the monomers, the other investigated COs reveal a weak hERG K+ blocking potential, this outcome being in very good correlation with predictions made by admetSAR2.0 and PreADMET tools (see **Table 4**).

The use of Pred-Skin computational tool reveals that investigated oligomers reflect no skin sensitizer potential (**Supplementary Table 6**). It is an important result as skin sensitization is known to be a common occupational health issue (Anderson and Meade, 2014). SwissADME and PreADMET tools also predicted very small values of the skin permeability parameters (**Supplementary Tables 2, 3**). The value of skin permeability parameter computed using SwissADME tool for diclofenac (an anti-inflammatory drug known to permeate the skin) is logKp = −4.96 cm/s (Daina et al., 2017). The use of PreADMET tool to compute the skin permeability coefficient for betulinic acid conducted to the value of logKp = −2.11 cm/h proving that betulinic acid is not permeable through the skin (Khan et al., 2018). The values of logKp computed for investigated COs by using SwissADME and PreADMET tools are much smaller that the two indicated values and it illustrates that these compounds are not permeable through skin.

Endocrine disruption potential is another toxicological endpoint that must be considered when using chemical compounds. Assessment of the endocrine disruption potential of investigated COs has been obtained using ENDOCRINE DISRUPTOME prediction tool and the results are presented in **Table 6**. Monomers and dimers of GlcNAc and GlcN, and chitobiose (DA) have the moderate ability (0.75>SE>0.50, yellow cells in **Table 6**) to bind to the androgen receptor and to produce antagonistic effects. The dimers AA and DA and the trimers ADA and DDD reflect a moderate potential to bind to the glucocorticoid receptor and to produce agonistic effects (0.75>SE>0.50, yellow cells in **Table 6**).

It means that smaller COs may inhibit the androgen and the glucocorticoid receptors and could be capable of deleterious effects on the male reproductive tract or to affect the immune response of the organism. COs containing more than 4 monomers are too big to accommodate in the binding site of the human nuclear receptors considered by the ENDOCRINE DISRUPTOME prediction tool and there are not outcomes, calculations being aborted. It seems that molecular weight is an important property for the COs that are able to interact with nuclear receptors. This outcome is in good agreement with literature data revealing that small organic non-steroidal molecules (MW between 430 and 600 Da) are capable to act as AR antagonists (Song et al., 2012; Tesei et al., 2013) and to interact with GR (Harcken et al., 2014; Sundahl et al., 2015).


TABLE 6 | Outcomes of the ENDOCRINE DISRUPTOME prediction tool concerning the potential binding of investigated COs to the human nuclear receptors: androgen receptor (AR), estrogen receptors α (ERα) and β (ERβ), glucocorticoid receptor (GR), liver X receptors α (LXRα), and β (LRXβ), peroxisome proliferator-activated receptors α (PPRAα), β/δ (PPRAβ), and γ (PPRAγ), retinoid X receptor α (RXRα) and thyroid receptors α (TRα) and β (TRβ), an - antagonistic effect.

*Results are color coded taking into account the values of the sensitivity parameter. Class "yellow" corresponds to 0.50*<*SE*<*0.75 and indicates compounds with moderate probability of binding, and class "green" corresponds to SE*>*0.75 and illustrates compounds with low probability of binding to the nuclear receptors.*

# DISCUSSION

The outcomes of the computational tools have been used to study the chito-oligomers pharmacological profiles and their toxicity. It has been shown that the results obtained with the different computational tools are usually in accordance with each other. The results obtained using FAF-Drugs4, admetSAR2.0, and PreADMET reveal that oligomers GlcNAc, GlcN, and GlcNAc-GlcN may have a higher oral bioavailability by comparison to the other COs, showing that oral bioavailability seems to be decreasing with increasing molecular weight. This outcome is in good correlation with published data considering that chitosan's systemic absorption and distribution is dependent on the molecular weight, oligomers could reveal some absorption whereas larger polymers are excreted (Kean and Thanou, 2010). Other studies revealed that the absorption of chitooligomers was significantly influenced by the molecular weight, the absorption decreased with increasing molecular weight (Chae et al., 2005; Naveed et al., 2019). Moreover, low molecular weight COs produced by depolymerization are usually preferred for pharmaceutical applications (Quiñones et al., 2018) as they have been reported to show remarkable biological activities (Adhikari and Yadav, 2018).

The action of chemicals depends on their interactions with plasma proteins, with unbound molecules usually reflecting better interactions with their targets and influencing the efficacy of the molecules (Kandagalla et al., 2017). All computational facilities that we have used reveal that COs considered in this study exhibit low potential to bind to plasma proteins. Consequently, these compounds exist freely being available for transport across the cell membranes, for the interaction with specific/non-specific targets and for excretion.

Predictions concerning the potential of the investigated chito-oligomers of being substrates and/or inhibitors of the Pglycoprotein (an efflux membrane transporter that is responsible for limiting cellular uptake and the distribution of xenobiotics within the human body) are not consistent between the computational tools that were used in this study. SwissADME tool predicts that all investigated COs are substrates of Pgp, but admetSAR2.0 illustrate the contrary. PreADMET tool reveals that investigated COs are not inhibitors of P-glycoprotein, but admetSAR2.0 displays that oligomers containing at least 4 monomeric units are potential inhibitors of P-glycoprotein. It illustrates that the activity of P-glycoprotein may be affected by the presence of COs and absorption and retention of COs in the cells could be impaired. This aspect must be further considered in experimental studies. Investigated COs reveal no potential to penetrate the blood brain barrier and it underlines their minimal side effects against the central nervous system. An in vitro study emphasized that COs with MW < 2,000 Da reflected neuroprotective effects (Santos-Moriano et al., 2018).

Investigated COs reveal no toxicity, but partially and totally deacetylated chitin oligomers are predicted to produce phospholipidosis, a disorder characterized by the accumulation in excess of phospholipids in tissues. This prediction is not unexpected as chitosan is a cationic polymer and it is known that cationic amphiphilic drugs may produce phospholipidosis (Anderson and Borlaka, 2006; Muehlbacher et al., 2012).

Literature data reflect that some xenobiotics (including drugs) are able to inhibit organic anion polypeptide transporters OATP1B1, OATP1B3, and OATP2B1. OATP1B1 and OATP1B3 are exclusively expressed in the liver, this organ being responsible for the hepatic uptake of some drugs, bile acids and some endogenous compounds. OATP2B1 is found in the liver and other tissues being associated with the oral absorption of chemicals. The inhibition of these transporters conducts to clinically relevant drug-drug interactions and to modified pharmacological effects and adverse reactions of drugs (Maeda, 2015; Alam et al., 2018). Another transporter that may be affected by the presence of xenobiotics is the organic cation transporter expressed in the kidney, OCT2. OCT2 transports compounds that are positively charged from the blood to the tubular epithelial cells, its inhibition also conducting to adverse effects (Motohashi and Inui, 2013). Consequently, predicting the inhibition of these transporters is important. Predictions obtained using admetSAR2.0 reflect that all investigated chito-oligomers are considered as possible inhibitors of the OATP1B3 and the deacetylated oligomers are also inhibitors of the OATP1B1. The inhibition of organic anion transporters OATP1B1 and OPTAP1B3 by the investigated oligomers is not an unexpected result because in silico models revealed the importance of lipophilicity, polarity and hydrogen bonding for OATP inhibition (Karlgren et al., 2012). Deacetylated oligomers reveal higher lipophilicity and hydrogen bonding potential that may conduct to their inhibitory effect against OATP1B1 too. Furthermore, the role of computational methods in predicting clinically relevant transporter interactions has been recognized (Türková and Zdrazil, 2019).

The outcomes of both admetSAR2.0 and Pred-hERG revealed that investigated COs, excepting the monomers, illustrate moderate potentials of hERG channel blocking. The hERG blockage potential of investigated COs increases with the molecular weight and it slightly depends on the acetylation degree and pattern. Literature data reveal the cavity of the hERG pore is large and is able to accommodate compounds with very high molecular weight (Linder et al., 2016) and that hERG inhibition positively correlates to the logP, molecular weight and rotatable bonds (Yu et al., 2016).

The hepatotoxicity of investigated COs also seems to depend on the molecular weight, the oligomers having the molecular weights between 500 and 1500 Da reflecting a weak potential of hepatotoxicity. This result is also in good correlation with published data revealing that lipophilicity and molecular weight are the most important physicochemical properties that influence the drug induced liver injury (Leeson, 2018). Furthermore, there is a slight dependence of hepatotoxicity of COs on the deacetylation degree and pattern. Literature data concerning the organ, tissue and cellular distribution COs suggest that: (i) they are usually distributed to kidney, hepatic, and splenic cells (the highest detected concentration was in hepatic cells) and lower concentrations were distributed to cardiac and lung tissues; (ii) MW and DDA influence the tissue and cellular distribution of COs and (iii) the biodegradation of COs is considered to occurring in the liver (Naveed et al., 2019).

There are some potential lacunas when predicting the pharmacokinetic profiles and toxicological endpoints of COs that might limit their effectiveness and probably affect the experimental validations. These lacunas are common to in silico predictions and in the case of the present study, they refer to: (i) the models used for predictions could not be adequate as the data regarding the biological activities of well-defined COs are limited and these data are not considered when building the models used for predictions; (ii) these predictions do not take into account the quantity of COs and the basic variables of experimental studies (temperature, humidity, pH, etc.); (iii) there are difficulties concerning the modeling of the toxicological endpoint because the lack of a complete understanding of its biology and of the complexity of processes involved. These lacunas may affect the experimental validations as data that were used for model building may originate from various experimental approaches that are different from those used for validation, as it is recognized that inconsistencies between predictions and experiments can often be attributed to the fact that they do not test the same assumptions (Gallion et al., 2017). In order to minimize the effects of the quality of the models built on data originated from various experimental approaches we have used well-established and recognized computational tools that are based on models obtained considering data for numerous chemical compounds and that have high predictive accuracies (higher than 70%). Furthermore, we have used both computational tools that are able to predict many pharmacological and toxicological properties, but also computational tools that are associated with a defined toxicological endpoint. The consensus of the predictions obtained using these tools increase the probability that the results are further validated in experimental tests. Due to these limitations of in silico predictions, the results have to be handled carefully and should not be used isolated to determine the potential hazard of COs. However, official agencies consider that the combination of computational modeling with in vitro testing is needed for a more efficient safety assessment of all types of chemicals.

Experimental validations of the biological effects of chemicals should be reliable, precise, and performed on a sufficiently large scale to be meaningful. These conditions may require repetitive measurement with a given assay or testing the same biological action with various assays. Consequently, in most practical cases, it is challenging to evaluate whether the available data are acceptably and complete, and to assess whether the experimental design affects the relationship with computational prediction (Gallion et al., 2017). An exhaustive experimental campaign can be time- and money-consuming also because of the intrinsic high number of variables related to the material. Focusing on chitosan, an appropriate experimental design should explore the influence of molecular weight, acetylation degree and acetylation pattern, to name a few aspects to be considered and that have been included in this study.

The potential merits of the computational predictions obtained within this study are that they highlighted some trends that relate material properties and possible side effects, which can be included in a suitable design of experiment algorithms in order to minimize the experimental efforts number and maximize the outcomes. Initially, in vitro basic tests through experiments that closely match the conditions used for predictions are preferably used for validation such as to avoid the complexity of the physiological pathways and possible interactions between various chemical molecules in the animal organisms. This approach could provide several advantages. First, it would lead to a structured experimental campaign, whose results can complement the fragmentary insights available in scientific literature. Secondly, experiments can be employed to assess the reliability of model predictions and thus the suitability of the chosen approach for pure predictive simulations. Thirdly, getting experimental data for well-characterized COs may provide an insight to define the applications of COs and will also drive the development of more rigorous models that also will conduct to improved predictions.

#### CONCLUSIONS

The pharmacokinetic profiles of chito-oligosaccharides are rarely experimentally studied, but taking into account their promising applications, their efficacy, and safety assessment are points to be considered. Obtaining well-defined COs in terms of length, degree acetylation, and acetylation pattern are still not straightforward and computational approaches offer an advantage in such cases. Within this study, we have used various computational tools to assess the pharmacokinetic profiles and toxicological endpoints of investigated COs. Computational predictions revealed that investigated small chito-oligomers, regardless of molecular weight, acetylation degree and acetylation pattern, reflect favorable pharmacological profiles: they are not able to penetrate the blood-brain barrier, do not produce eye irritation and corrosion, reveal no mutagenicity, no carcinogenicity, and no skin sensitization potential.

As possible harmful effects we have noticed the followings: (i) all investigated COs revealed high potential of inhibition of the OATP1B3 and COs containing only deacetylated units also reflect inhibition potential of the OATP1B1; (ii) COs containing more than 2 units reflect a moderate potential of cardiotoxicity; (iii) some of considered COs reflect small probabilities to produce hepatotoxicity; (iv) smaller oligomers (n = 1–3) reflect a weak disruption potential against AR and GR; (v) totally deacetylated oligomers are considered to produce phospholipidosis.

Predictions concerning the interactions of investigated COs with P-glycoprotein and CYPs are unclear and they must be further considered in experimental studies.

We have also examined the influence of the molecular weight, deacetylation degree, and pattern on the pharmacological profiles and toxicological endpoints of investigated COs. The oral bioavailability of investigated COs decreases with increasing MW and deacetylation degree. Taking into account that bioavailability profile could be the main factor limiting the

#### REFERENCES

Adhikari, H. S., and Yadav, P. N. (2018). Anticancer activity of chitosan, chitosan derivatives, and their mechanism of action. Int. J. Biomaterial. 2018:2952085. doi: 10.1155/2018/2952085

efficiency of a drug, this information is of interest. There is a slightly dependence of hepatotoxicity and cardiotoxicity on the molecular weight and on the deacetylation degree and pattern of COs. The cardiotoxicity of investigated COs increases poorly with the molecular weight, decreases slightly with the deacetylation degree and is not influenced by deacetylation pattern. Hepatotoxicity increases with the molecular weight, decrease with the deacetylation degree and depends on the deacetylation pattern.

#### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/**Supplementary Files**.

#### AUTHOR CONTRIBUTIONS

DR performed the computations using FAF-Drugs, admetSAR, PreADMET, Pred-Skin, and Endocrine Disruptome, to results analysis and contributed to manuscript editing. MR performed computations using, SwissADME, Pred-hERG, CarcinoPred-El, and Toxtree, to results analysis and contributed to manuscript editing. CS, MS, EH, PW, TC, and GP contributed to the analysis of results, conception, and design of the manuscript. VO furnished the SMILES and structural formulas of oligomers and contributed to the conception and design of the manuscript. AI conceived and planned the study and contributed to the conception and design of the manuscript. All authors contributed to revise the manuscript and approved the submitted version.

#### FUNDING

This study is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016; from the CTI (1.1.2018 Innosuisse), under grant agreement Number 19267.1 PFNM-NM; by the Rumanian funding agency Executive Unit for Financing Higher Education, Research, Development and Innovation (UEFISCDI) under the grant PN3- P3-285 for research work and under the grant PNIII 28 PFE BID for publication fee, and from FCT Foundation for Science and Technology under the project PROSAFE/0001/2016.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2019.00214/full#supplementary-material

Alam, K., Crowe, A., Wang, X., Zhang, P., Ding, K., Li, L., et al. (2018). Regulation of Organic Anion Transporting Polypeptides (OATP) 1B1 and OATP1B3-mediated transport: an updated review in the context of OATP-mediated drug-drug interactions. Int. J. Mol. Sci. 19:E855. doi: 10.3390/ijms19030855


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

Copyright © 2019 Roman, Roman, Som, Schmutz, Hernandez, Wick, Casalini, Perale, Ostafe and Isvoran. 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.

# A Perspective on Polylactic Acid-Based Polymers Use for Nanoparticles Synthesis and Applications

#### Tommaso Casalini <sup>1</sup> \*, Filippo Rossi <sup>2</sup> , Andrea Castrovinci <sup>1</sup> and Giuseppe Perale1,3

<sup>1</sup> Polymer Engineering Laboratory, Department of Innovative Technologies, Institute for Mechanical Engineering and Materials Technology, University of Applied Sciences of Southern Switzerland, Manno, Switzerland, <sup>2</sup> Department of Chemistry, Materials and Chemical Engineering "Giulio Natta", Politecnico di Milano, Milan, Italy, <sup>3</sup> Ludwig Boltzmann Institute for Experimental and Clinical Traumatology, Vienna, Austria

Polylactic acid (PLA)—based polymers are ubiquitous in the biomedical field thanks to their combination of attractive peculiarities: biocompatibility (degradation products do not elicit critical responses and are easily metabolized by the body), hydrolytic degradation in situ, tailorable properties, and well-established processing technologies. This led to the development of several applications, such as bone fixation screws, bioresorbable suture threads, and stent coating, just to name a few. Nanomedicine could not be unconcerned by PLA-based materials as well, where their use for the synthesis of nanocarriers for the targeted delivery of hydrophobic drugs emerged as a new promising application. The purpose of the here presented review is two-fold: on one side, it aims at providing a broad overview of PLA-based materials and their properties, which allow them gaining a leading role in the biomedical field; on the other side, it offers a specific focus on their recent use in nanomedicine, highlighting opportunities and perspectives.

Edited by: Anderson Oliveira Lobo, Federal University of Piauí, Brazil

#### Reviewed by:

Jianxun Ding, Changchun Institute of Applied Chemistry (CAS), China Edson Cavalcanti Silva-Filho, Federal University of Piauí, Brazil

#### \*Correspondence:

Tommaso Casalini tommaso.casalini@supsi.ch

#### Specialty section:

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

Received: 05 July 2019 Accepted: 26 September 2019 Published: 11 October 2019

#### Citation:

Casalini T, Rossi F, Castrovinci A and Perale G (2019) A Perspective on Polylactic Acid-Based Polymers Use for Nanoparticles Synthesis and Applications. Front. Bioeng. Biotechnol. 7:259. doi: 10.3389/fbioe.2019.00259 Keywords: polylactic acid, degradation, processing, nanomedicine, nanoparticles

# INTRODUCTION

Polylactic acid (PLA), classified as an aliphatic polyester because of the ester bonds that connect the monomer units, has gained a key role in the biomedical field for a wide range of applications: suture threads, bone fixation screws, devices for drug delivery, just to scratch the surface. PLA merges several interesting properties that make it an ideal candidate for biomedical applications.

PLA naturally degrades in situ through hydrolysis mechanism: water molecules break the ester bonds that constitute polymer backbone. This eliminates the necessity of additional surgeries in order to remove the device, improving patient recovery and optimizing health system costs.

The main phenomena involved in the degradation mechanisms and the most important factors that influence hydrolysis rate are currently well-established in scientific literature, thanks to a devoted research activity that reached the peak between the 1980s and the 1990s. Consequently, degradation kinetics and mechanical properties can be tailored by properly tuning few polymer properties (such as composition or molecular weight), thus leading to the development of biomedical devices optimized for each specific application. Degradation products (composed of lactic acid and its short oligomers) are recognized and metabolized by the body itself: this gives PLA an intrinsic biocompatibility that dampens the attainment of critical immune responses. In addition, PLA can be processed with standard and established technologies, such as injection molding, extrusion, etc.

After this brief summary, whose main points will be discussed in the following sections, it should be no more surprising why PLA attracted a lot of attention and enthusiasm in the biomedical field. These features make PLA a suitable option also for the new paradigm recently introduced by nanomedicine, where nanomaterials (whose size is similar to molecules of biological interest, such as proteins or viruses) are distributed within the human body and can be internalized by cells.

Nanomedicine offers new unprecedented chances, thanks to the synthesis of nanoparticles, which can be employed for the targeted delivery of drugs, vaccines and genes. On the other hand, nanomaterials can also give rise to new side effects due to specific interactions with the biological components (proteins, carbohydrates, lipids) present in body fluids (blood, plasma, interstitial fluids).

The first part of this review guides the interested reader through the main peculiarities of PLA, underlining the reasons why it emerged as a material of choice in the biomedical field. The second part of the manuscript is focused on the use of PLA for the synthesis and application of nanoparticles, from the synthetic routes of nanovectors to perspectives and opportunities.

#### POLYLACTIC ACID-BASED MATERIALS: GENERAL DESCRIPTION AND SYNTHESIS ROUTES

Polylactic acid is a hydrophobic polymer that belongs to the class of biomaterials commonly referred as poly-α-hydroxy acids, poly-α-esters or aliphatic polyesters. It is synthesized starting from lactic acid (LA; 2-hydroxypropanoic acid), which a watersoluble monomer that exhibits two enantiomeric forms, namely L-(+)-LA and D-(-)-LA, as shown in **Figure 1**.

Although both enantiomers are employed in industrial practice, L-(+)-LA is the isomer of interest for biomedical applications since it is involved in the cellular metabolism of the human body and reduces the risk of adverse reactions. In in vivo environment L-(+)-LA can be either incorporated into the Krebs' cycle or converted into glycogen in the liver; eventually it is eliminated as water and carbon dioxide from the lungs (Sheikh et al., 2015). PLA can be produced starting from pure L-lactic and D-lactic isomers, which leads to poly-L-lactic (PLLA) acid and poly-D-lactic acid (PDLA) homopolymers, respectively; if a racemic mixture of L- and D-monomers is employed, poly-D,Llactic acid (PDLLA) copolymer is obtained. The stereochemistry has a relevant impact on material properties: PLLA is a semicrystalline polymer, while PDLLA is an amorphous polymer with no melting point. In addition, degradation rate of PLLA is significantly slower than PDLLA, because of the presence of crystalline regions. Main advantages and disadvantages of PLA use and production are summarized in **Table 1**.

Focusing on lactic acid itself, synthesis can be performed in different ways; the most popular route is the following one (Storti and Lattuada, 2017):

CH3CHO + HCN → CH3CH (OH) CN (1) CH3CH (OH) CN + 2H2O + HCl → CH3CH (OH) COOH +NH4Cl (2) CH3CH (OH) COOH + CH3OH ↔ CH3CH (OH) COOCH<sup>3</sup> +H2O (3)

Lactonitrile, obtained from acetaldehyde and hydrogen cyanide (1), is hydrolysed at low pH in order to lactic acid (2); it is subsequently converted to methyl lactate (3) through esterification and eventually recovered and purified by distillation. Lactic acid and methanol are obtained through hydrolysis from lactate; methanol is recycled in step (3). Anyway, this kinetic pathway leads to a racemic mixture.

Bacterial fermentation of sugar solutions is currently the most employed process; this process leads to high yields and, depending on the chosen type of bacteria, it allows obtaining one given stereoisomer or the racemic mixture. It is estimated that about 90% of the total LA produced worldwide is currently obtained with this procedure.

TABLE 1 | Main advantages and disadvantages of PLA.


on material properties due to the sensitivity of the reaction to residual non-cyclic monomers. The cyclic raw material for PLA is constituted by cyclic dimer lactide, which exhibits three stereoisomeric forms, as shown in **Figure 2**: LL-, DD-, and D,L- (also referred a meso-lactide).

and chemical purity, since impurities have detrimental effects

Lactide is usually produced through backbiting kinetic mechanism is then (promoted with suitable process conditions) starting from low molecular weight prepolymer; cycles are eventually collected by distillation. Other synthesis routes are azeotropic dehydration and enzymatic polymerization. PLAbased polymers synthesis routes are summarized in **Figure 3**.

PLA is widely employed in the biomedical field because of its biocompatibility and its processability, since it can be processed with a wide range of techniques, such as extrusion, injection molding, injection stretch blow molding, film and sheet casting, extrusion blow film, thermoforming, foaming, fiber spinning, electro spinning, blending, compounding, and nanocompositing.

### PHYSICAL AND CHEMICAL PROPERTIES

PLA can be seen as a "family" of polymers, which include homopolymers PLLA and PDLA (synthesized from mixtures of pure L- or D-lactic acid) and the copolymer PDLLA (obtained from the racemic mixture). This has a remarkable impact on material properties because of the involved stereochemistry: PLLA and PDLA are semicrystalline polymers, while PDLLA is usually amorphous. The final crystallinity depends also on the thermal and mechanical history, mainly due to fabrication processes. Mechanical properties are summarized in **Table 2**; values are expressed as ranges, since they strongly depend on the characteristic of the tested material (molecular weight, crystallinity, processing, etc.) as well as testing procedure (Van De Velde and Kiekens, 2002).

Polymer crystallinity influences mechanical and physical properties such as hardness, modulus, tensile strength, stiffness, and melting points. If the amount of PLLA is higher than 90% the polymer is semicrystalline, while lower amounts (and thus a lower optical purity) lad to an amorphous polymer. The density values lie in small range depending on the composition, as shown in **Table 2**.

PLA is soluble in dioxane, acetonitrile, chloroform, methylene chloride, 1,1,2-trichloroethane and dichloroacetic acid, while it is only partially soluble in ethyl benzene, toluene, acetone and tetrahydrofuran, only when heated to boiling temperature. PLA is not soluble in water, alcohols, and linear hydrocarbons. Crystalline PLLA cannot be dissolved in acetone, ethyl acetate, or tetrahydrofuran.

It is worth mentioning that polymer properties can change after processing, because of thermal and mechanical stresses. PLA undergoes thermal degradation above 200◦C, although degradation rate and extent depend on variables like time, temperature, low molecular weight impurities, and catalyst amount (Carrasco et al., 2010).

The success of PLA passes also through its versatility, since material properties can be modified in several ways. They can

In this framework, the critical step is the subsequent LA purification, which is expensive and determines process profitability. Commonly used techniques are liquid extraction, membrane separation, ion exchange, electrodialysis, and reactive distillation.

Polymer synthesis can be carried out through step growth polymerization or ring opening polymerization. Step growth polymerization simply takes advantage of the reactivity of the two LA functional groups: indeed, the polycondensation of hydroxyl and carboxyl moieties leads to the formation of the ester bonds that constitute polymer backbone. This synthetic route has several drawbacks: long residence times are required for longer chains (leading to unwanted side reactions, like transesterification), challenging reaction conditions (temperatures up to 250◦C and vacuum up to 100 mbar) and continuous water (side product of polycondensation) removal. Chain extenders (e.g., isocyanates or epoxides) can be in principle employed, although this approach has an inevitable impact on material purity and quality.

At industrial scale ROP is the most popular process because of its advantages: mild process conditions, short residence times, absence of side products and high molecular weights. The most widely used catalyst is 2-ethylhexanoic tin(II) salt (also referred as stannous octoate [Sn(Oct)2]), approved by United Stated Food and Drug Administration (FDA) and usually employed along with an alcohol as cocatalyst. The real bottleneck of ROP is the availability of cyclic monomers as well as their optical

Frontiers in Bioengineering and Biotechnology | www.frontiersin.org

be tuned, e.g., through the addition of suitable plasticizers, widely used in order to improve processability and flexibility of polymers. Focusing on semicrystalline PLA, plasticizer addition decreases T<sup>g</sup> , as well as T<sup>m</sup> and crystallinity.

PLA can be blended with biodegradable or non-biodegradable polymers (such as polyethylene, polypropylene, chitosan, polystyrene, polyethylene terephthalate, and polycarbonates) (Saini et al., 2016) or by making composite materials (Murariu and Dubois, 2016) through the addition of carbon nanotubes, ceramic nanoparticles, natural fibers, and cellulose (Hamad et al., 2018). A relevant example is constituted by PLA/hydroxyapatite (HA) composites for devices for bone healing, where HA micro or nanoparticles are dispersed into the polymer matrix (Rodenas-Rochina et al., 2015).

Another is the formation of stereocomplexes (Tsuji, 2016), which can be obtained by blending PLLA with PDLA (that is, the homopolymer composed by D-lactide units only) or adopting PLLA/PDLA block copolymers. The strong interactions between PDLA and PLLA blocks that derive from the formation of stereocomplex crystallization improves mechanical properties



ρ, density; σ, tensile strength; E, elastic modulus; ε, ultimate strain; Tg, glass transition temperature; Tm, melting temperature. Taken from Farah et al. (2016).

and thermal stability, slows down degradation rate and increase PLA barrier properties, allowing a more prolonged drug release. PLA-based stereocomplexed materials, by virtue of their improved stability, attracted a lot of interest also for biomedical applications, such as fibers and nanoparticles for drug delivery applications (Jing et al., 2016).

PLA-based materials can be also assembled in complex molecular architectures (Corneillie and Smet, 2015), leading to branched polymer chains, star-shaped structures (Michalski et al., 2019), grafted chains (Nagahama et al., 2007), and crosslinked matrices (Tsuji, 2016). If synthesized with both PLLA and PDLA blocks, stereocomplexation can be achieved also with these complex structures (Nagahama et al., 2007; Fan et al., 2013; Sveinbjornsson et al., 2014), which found as well-potential applications in the biomedical field for the synthesis of hydrogels, nanoparticles and micelles for drug delivery purposes.

Another popular way to tune material properties is the copolymerization with glycolic acid, which leads to the well-known polylactic-co-glycolic random copolymer (PLGA). Copolymerization is also performed with caprolactone, which allows obtaining polylactic-co-caprolactone (PLCL). Another strategy to improve material hydrophilicity is the synthesis of PLA and polyethylene glycol (PEG) block copolymers.

PLA (as well as its copolymers) degrades because of hydrolysis mechanism: water breaks the ester bonds that constitute polymer backbone, according to the following mechanism:

$$P\_{n+m} + H\_2O + H^+ \leftrightarrow P\_n + P\_m + H^+ \tag{4}$$

where Pn+m, Pn, and P<sup>m</sup> are polymer chains composed by n+m, n, and m monomer units, respectively, H2O is a water molecule and H<sup>+</sup> indicates that hydrolysis is catalyzed in acidic environment. The most important phenomena that govern PLA degradation are currently rationalized and accepted in scientific literature (Casalini, 2017). Two degradation regime can be distinguished. If hydrolysis rate is higher than diffusion rate, surface, or heterogeneous degradation takes place; only polymer surface experiences degradation and erosion (i.e., mass loss), while the bulk remains intact. The shape of the device remains unchanged, but its volume decreases in time. On the other hand, if water penetration is much faster than water consumption, homogeneous, or bulk degradation occurs: degradation rate is essentially equal in every point of the matrix and the volume does not appreciably change in time. Mass loss is observed after a certain time interval, when chain scission has created oligomers that are mobile enough to diffuse through the matrix toward the environment. Another relevant aspect is autocatalysis: polymer degradation creates small fragment that lower pH-value by virtue of their dissociated carboxyl terminal group, thus enhancing hydrolysis rate. In other words, pH decreases as degradation continues and this results in an autocatalytic behavior. Notably, when mass transport resistances and/or mean diffusive paths are relevant, a transition from homogeneous to heterogeneous degradation may occur. In this case, oligomers accumulate in the core of the device, locally lowering the pH; consequently, degradation is faster in the bulk than close to the surface. In order to discriminate the degradation mechanism, Von Burkersroda et al. (2002) proposed a distinctive parameter called critical thickness Lcrit; if the characteristic size of the device (e.g., the radius of a sphere) is larger than the critical thickness surface degradation occurs, otherwise bulk degradation govern matrix hydrolysis mechanism.

A scheme is provided in **Figure 4**.

The discussed mechanisms represent asymptotic cases and the observed experimental behavior is usually one of the many shades of gray in between.

Lcrit depends on the interplay between degradation and diffusion kinetics, and at a first glance, it depends on the specific material. Von Burkersroda et al. (2002) computed the values of Lcrit for some polymers of interest; the reference value for aliphatic polyesters is 7.4 cm. The main phenomena behind PLA degradation can be summarized as follows (Casalini, 2017):


Degradation rate depends on several factors, such as (Alexis, 2005):



By virtue of a critical thickness equal to 7.4 cm, PLA-based devices usually experience homogeneous degradation, which can become heterogeneous when oligomer accumulation in the bulk occurs.

In in vivo environment, there is an additional contribution to degradation due to enzymes that cleave ester bonds, such as lipases, cutinases, serine proteases, PHB depolymerase, PCL depolymerase, elastase esterase, proteinase K, and trypsin. This enzymatic degradation is a heterogeneous process since involves only device surface: enzymes are not able to diffuse in the polymer matrix and contribute to surface erosion through ester bonds cleavage (Armentano et al., 2018).

PLA-based materials are also subjected to thermal degradation; while this is not relevant for biomedical applications themselves (at body temperature, thermal degradation is absent), this should be taken into account in the fabrication process. Indeed, for temperature values above 200◦C (Garlotta, 2001), PLA experiences not only hydrolysis but also lactide reformation, oxidative chain scission, intra- or intermolecular transesterification reactions.

# PROCESSES FOR NANOPARTICLES SYNTHESIS

PLA-based materials experienced a wide success in biomedical field for several reasons: biocompatibility, low toxicity, degradation through hydrolysis and tailored physical and chemical properties through the selection of molecular weight or

#### TABLE 3 | Overview of PLA biomedical applications.


Adapted from Tyler et al. (2016).

through copolymerization, blending, or building more complex molecular architectures and processability. A proper tuning of polymer properties allows assuring the desired performances (in terms, e.g., of tensile strength or release rate) over a suitable time span, before an appreciable onset of degradation reactions. The natural degradation of PLA-based devices due to hydrolysis avoids the need of additional surgery for device removal, improving patient care.

All these advantages led to a wide range of applications, summarized in **Table 3** for the sake of completeness.

Nanomedicine is an emerging field, focused on the development and application of engineered nanomaterials, whose size (from 1 to 1,000 nm according to the FDA draft guideline form 2017) is comparable to many molecules of biological interest, such as proteins and viruses. Devices like polymer nanoparticles, by virtue of their small size, can be internalized by cells and this opens a wide range of new opportunities for the development, e.g., of new carriers for the targeted delivery of drugs and vaccines or image contrast agents for diagnostic purposes. Because of the interesting properties of PLA-based polymers, it is not surprising that they experienced and are still experiencing a great interest as starting materials for the synthesis of nanoparticles.

The most frequently used and promising methods to formulate nanosized particles can be divided in four categories according to the fundamental physical principles, as summarized by Lee et al. (2016). The main challenges are the control of particle size and an efficient drug encapsulation.

#### Emulsion-Based Methods

The single-emulsion/solvent-extraction method is the simplest approach for the synthesis of micro- and nanoparticles, including drug-loaded carriers. The polymer and, if needed, the hydrophobic drug are dissolved in a water-immiscible organic solvent and an emulsion in water phase is subsequently realized by adding a stabilizer and stirring. For the sake of completeness, oil in water (o/w) as well as oil in oil (o/o) and water in oil (w/o) emulsions can be suitable for this process.

The removal of the organic phase is carried out through evaporation at low pressure or vacuum or by solvent extraction; polymer particles are recovered by centrifugation or filtration and washed with water or buffer solutions in order to remove possible traces of solvent, stabilizer and free drug before lyophilization.

Single-emulsion approach leads to a poor encapsulation efficiency of hydrophilic drugs (such as peptides), which are mainly dispersed in the aqueous phase rather than the organic one. Double-emulsions methods aim at overcoming this issue. A water solution containing the hydrophilic active molecule is added to an organic solvent where the polymer is dissolved under stirring, in order to form a w/o (or an o/w) emulsion that is subsequently added to a second water phase containing a stabilizer. This leads to the formation of a w/o/w (or an o/w/o) emulsion. The organic solvent is removed by means of evaporation under low pressure or vacuum and the resulting particles are washed (to safely remove traces of solvent, stabilizer, and free drug) before lyophilization.

Emulsion-based methods, despite their simplicity, need the optimization of several process parameters, such as phase volumes (oil and water), polymer, drug and stabilizer concentration, type of solvents, and stirring rate.

Examples of particles produced with emulsion-based methods are provided in **Table 4**.

#### Precipitation-Based Methods

Nanoprecipitation, also referred as solvent displacement, is a one-step process suitable for producing nanoparticles loaded with hydrophobic drugs. The underlying physical principle is the interfacial deposition of a polymer, following the displacement of a water-compatible solvent from a lipophilic solution. Polymer and drug are dissolved in a semi-polar organic solvent (acetone, methanol, ethanol, acetonitrile); the resulting organic phase is mixed drop-wise in a water solution containing a stabilizer. This technique leads a narrow particle size distribution and allows avoiding the use of large amounts of toxic solvents as well as external energy sources. On the other hand, it is limited by drug solubility in the organic phase and it is thus not suitable for hydrophilic drugs; another drawback is the removal of the residual solvent.

Salting-out method is based on the addition of a polymer and drug solution in a water-compatible solvent (acetone, acetonitrile, tetrahydrofuran) to an aqueous solution that contains the salting-out agent (electrolytes like magnesium chloride and calcium chloride or non-electrolytes like sucrose) and a stabilizer, under stirring. This allows obtaining an o/w emulsion that is subsequently diluted with large volumes of water, promoting the formation of particles by virtue of the diffusion of the water-compatible solvent toward the aqueous phase. Particles are recovered and purified by means of centrifugation or filtration. In particular, salt residues must be removed before utilization.

Salting-out is the ideal process for encapsulating heat-sensitive molecules (such as proteins, DNA, RNA) because no heating


TABLE 5 | Examples of nanoparticles synthesis by means of precipitation-based methods.


steps are required. Anyway, the process requires the optimization of parameters like salt type and concentration, type of polymer and solvent, and their relative amounts. The principal limitations are the intensive purification of the resulting nanoparticles as well as incompatibility issues concerning most of the employed salts with bioactive compounds.

Dialysis emerged as a simple process that allows obtaining small particles with a narrow size distribution. The polymer is dissolved in an organic solvent and placed in a dialysis tube of suitable pore size; dialysis is subsequently carried out in a solvent that is miscible with the organic phase but not compatible with the polymer. This leads to the formation of polymer particles due to the loss in solubility. Selected examples from literature of particles synthesized by means of precipitation based-methods are reported in **Table 5**.

#### Compositing Methods

In spray drying technique, polymer and drug are dissolved in an organic solvent and subsequently dispersed as ultra-fine droplets in a hot air flow. The solvent evaporates instantaneously and dried particles are collected under low pressure in dry air flow. Spray drying process is easy to perform and can be potentially employed at industrial scale. However, productivity can be hindered by the adhesion of the particles to the walls of the spray dryer and their agglomeration. Moreover, it is challenging to control drug distribution.

Melting technique allows avoiding the use of organic solvents but implies the dissolution of the drug in a polymer melt; therefore, it is not suitable for encapsulating active compound that are subjected to thermal degradation. Drug/polymer melt is subsequently solidified and cooled down with water or dry air. Particles are obtained through grounding or milling; in order to achieve small particles with narrower size distribution, the ground melt can be emulsified in a hot solution with a stabilizer. Despite the absence of organic solvents, this approach is limited by the thermal treatment of the drug/polymer system and the high number of steps needed to obtain smooth particles.

In situ-forming techniques aim at overcoming the most common drawbacks of the discussed processes, such as solvent removal, particles recovering, and resuspension. A drug/polymer solution (in a water-miscible solvent) is prepared and injected in the target site. When in contact with physiological fluids, polymer phase hardens and precipitates forming microparticles that entrap the active compound. The main drawback lies in a careful choice of the solvent, whose side-effects must be previously investigated.

#### Other Approaches

Supercritical fluids-based methods attracted a lot of interest because of their advantages, such as the use of environmentally friendly solvent and the possibility to obtain nanoparticles with (virtually) no traces of residual solvents. There are two main processes that involve supercritical fluids: rapid expansion of supercritical solution (RESS) and rapid expansion of a supercritical solution into a liquid solvent (RESOLV).

In RESS technique, the polymer is dissolved into the supercritical fluid; the solution is then subjected to a rapid expansion across a nozzle in ambient air. The sudden reduction in pressure leads to a substantial supersaturation, which, in turn, promotes homogeneous nucleation and the formation of well-dispersed particles. RESOLV process is based on the same principle, but the expansion does not take place in air but in a liquid solvent. The liquid phase hinders particle growth, leading to the synthesis of nanoparticles. The most important limitation of RESS and RESOLV technique is the poor solubility of the polymer in the supercritical fluid. In addition, it is difficult to control particle size and morphology.

With microfluidic techniques it is possible to obtain uniform particles with a narrow particle size distribution, which, in turn, allows a finer control of the release rate. The starting point is usually the attainment of an o/w emulsion in the microfluidic device, where monodisperse droplets can be achieved, followed by droplet solidification by means of solvent evaporation, diffusion or extraction. Particle size can be controlled by tuning the properties of oil and water phases (density, interfacial tension, and viscosity) and flow rates. Because of the inherent micron length scale, the challenge lies in the synthesis of particles at nanoscale. The underlying principle of hydrogel template method is the possibility to control sol-gel transition of physical gel by changing the environmental conditions (e.g., temperature). A warm aqueous gelling solution is distributed on a hard master template and placed at low temperature, in order to obtain a hydrogel mold. Polymer and drug are dissolved in a suitable solvent and poured on the hydrogel mold; solvent is removed through evaporation and particles are recovered by centrifugation or filtration and washed after dissolving the mold in water. Polyvinyl alcohol water-soluble molds are also employed. Similarly to microfluidic techniques, the drawback lies in the particle size, which is still limited to the micron length scale. In principle, nanoparticles can be obtained by means of nanostructured mold templates. Notably, nanoparticles can be produced not only from preformed polymers but also starting from monomers, including the polymerization process in the nanoparticle production step. This can be achieved my means of emulsion polymerization (George et al., 2019).

#### Summary

For the sake of completeness, the main advantages and disadvantages of the most common employed methods are summarized in **Table 6**.

As mentioned in the previous sections, several parameters are involved in process optimization and strongly influence the final particle size distribution. The most important degrees of freedom, as well as their influence on the final outcome are summarized in **Table 7**.

#### THE NEW PARADIGM INTRODUCED BY NANOPARTICLES

While devices at macroscale (suture treads, polymer-coated stents, bone fixation screws, etc.) remain at the implantation site, nanoparticles, because of their size, are able to spread all over the body and to penetrate into cells. This introduces a new paradigm in the engineering of polymeric nanocarriers, since they must be designed so that they remain in the systemic TABLE 6 | Advantages and disadvantages of the most common nanoparticles production methods.


TABLE 7 | Process variables and their effect on particle size.


circulation long enough to accomplish their task and they are able to target the desired objective. Particle behavior, in terms of clearance, biodistribution (i.e., distribution in the organs), cellular uptake, and toxicity are mainly influenced by particle size, shape, morphology, surface chemistry, and charge (Blanco



et al., 2015). The techniques that can be used to characterize experimentally nanoparticles are summarized in **Table 8** (Crucho and Barros, 2017).

The acquired knowledge led to the development of proper design strategies, as shown below (vide infra). Nanoparticles synthesized from PLA-based materials are mainly employed as devices for drug delivery for cancer treatment and for imaging purposes (Kim et al., 2019). Nanoparticles are potentially able to penetrate selectively within the cancer, where they can release the loaded active compound at the desired rate, so that a therapeutic effective drug concentration is maintained for a given time period. This allows minimizing the amount of administered drug, since it mainly diffuses in the tumor following nanoparticles permeation through cancer cells, dampening potential side effects and optimizing costs for health systems.

There are various administration routes for nanoparticles, such as oral, parenteral (intravenous, subcutaneous, intradermal, and intramuscular), respiratory, and transdermal routes (Kaialy and Al Shafiee, 2016). In any case, nanoparticles must be able to cross certain barriers (which can vary according to the administration route) in order to be effective (Blanco et al., 2015). Depending on the administration route, the first barrier can be constituted by endothelial or epithelial cells.

Epithelium is essentially constituted by the skin and mucosal membranes, while endothelium separates the blood flow from the surrounding tissues. The endothelium that separates blood vessel and central nervous system is the well-known blood brain barrier (BBB), which is very challenging to cross.

Another barrier is constituted by the immune system; after injection, nanoparticles experience opsonization, which involves the adsorption of plasma proteins on the surface of the device that leads to the formation of the protein corona (vide infra). After the attainment of the layer of adsorbed proteins, nanoparticles bind to a macrophage receptor and are subsequently internalized and removed from circulation. This problem can be overcome through surface modification, hindering protein adsorption, and interactions with macrophages receptors. The most popular route is PEGylation, that is, the addition of PEG brushes on nanoparticle surface that constitute an obstacle for protein adsorption (Partikel et al., 2019). Other strategies involve surface functionalization with ad hoc peptides that delay phagocytic clearance (Rodriguez et al., 2013), or coating with cell membranes from red blood cells or leukocytes (Blanco et al., 2015). In general, the objective is to prolong the persistence in the blood circulation avoiding a rapid clearance by the immune system.

Focusing on PEGylation of PLA-based particles, three main approaches can be identified (Betancourt et al., 2009). In direct conjugation, PEG chains are covalently bound to the end groups of polymer chains already assembled in nanoparticles. This approach has the advantage that PEGylation is performed after the encapsulation of an active principle with standard techniques but it is not very efficient because of the limited exposure of the end groups on particle surface. When active conjugation in solution is chosen, preformed long polymer chains are activated and conjugated with PEG chains. Despite the moderate conjugation efficiency, the attainment of high yields is hindered by a difficult recover of the copolymer and the possible formation of PEG-PEG conjugates that can affect the purity of the product. The most used technique is ring opening polymerization, where preformed OH-PEG-COOH chains (that is, with a hydroxyl and a carboxyl end groups) are polymerized with lactide (and glycolide, if PLGA is needed) cyclic dimers. In these conditions, the hydroxyl end group of PEG acts as a protic agent initiating the reaction, while carboxyl end group remains intact. This leads to the synthesis of block PLA-PEG block copolymers.

Eventually, nanoparticles can experience cellular uptake mainly through endocytosis, which can be due to different pathways (Sahay et al., 2010; Kim et al., 2019). In receptormediated endocytosis, nanoparticles can be internalized by interacting with a specific receptor expressed on cellular membrane; a key role is played by clathrin and caveolin. Clathrinmediated endocytosis is present in essentially all mammalian cells and is responsible for the uptake of essential nutrients; caveolae-mediated endocytosis exploits the presence of caveolin proteins in the caveolae (lipid rafts along cellular membranes) and attracted some interest since this pathway allows bypassing lysosomes and thus avoiding lysosomal degradation. Carriermediated endocytosis exploits the presence of carrier proteins on cellular membrane; this pathway can be exploited to pass challenging barriers like the BBB. Since the cellular membrane, at physiological pH, has a slight negative charge, electrostatic interactions with positively charged carriers can promote particle internalization through an adsorption-mediated endocytosis. Pinocytosis implies the formation of membrane-based vesicles from the cell surface, which captures solute and fluid from the environment.

From an experimental point of view, it is possible to identify the specific endocytosis pathway by suppressing some mechanisms with suitable inhibitors and assessing the cellular uptake.

In this regard, it useful to introduce the concept of targeting. In order to maximize their effect, nanoparticles should be able to selectively penetrate within the tumor, minimizing their accumulation in healthy organs. There are two different targeting approaches: passive and active targeting.

Passive targeting exploits the so called enhanced permeation and retention (EPR) effect; according to EPR, cancer exhibits an enhanced permeation due to the hyperpermeable vasculature and an enhanced retention because of the ineffective lymphatic drainage. Although EPR concept seems to be quite well-assessed in literature, its effectiveness is still debated since it is welldocumented for small animal models but human clinical data are less clear (Danhier, 2016).

Active targeting implies the functionalization of nanoparticle surface with suitable ligands (small molecules, proteins, carbohydrates, etc.), which can interact in a specific way with receptors that are overexpressed in diseased organs, tissues and cells (Bertrand et al., 2014). Since PLA has no reactive side groups, functionalization needs the synthesis of polymer chains with end groups that can be activated for further conjugation. In this regard, two strategies can be identified; pre-conjugation, where conjugated chains are obtained and subsequently assembled in nanoparticles with a suitable technique. This approach can be used for small ligands and peptides, while it is not suitable for proteins, since they can affect self-assembling process and conjugation needs organic solvents that can influence the secondary structure. Pre-conjugation allows introducing multiple ligands with one-step formulation procedure and a good control of particle properties. Post-conjugation involves the functionalization of preformed nanoparticles; this strategy is suitable for both small and big ligands (proteins, antibodies).

Notably, functionalization can be achieved also through the physical (that is, non-covalent) adsorption of targeting moieties on nanoparticle surface (Bertrand et al., 2014).

Summarizing, Dawidczyk et al. (2014) proposed general guidelines for the design of nanoparticles as carriers of active compounds, as shown in **Table 9**.

In the following paragraphs, some relevant examples from scientific literature are reported concerning PLA-based nanoparticles for drug delivery and imaging purposes. Given the extent of the topic, the following discussion does not claim TABLE 9 | Design criteria for nanoparticles for drug delivery purposes (Dawidczyk et al., 2014).


to be exhaustive, but aims at presenting some opportunities in the field.

#### Nanoparticles for Drug Delivery

Shalgunov et al. (2017) systematically investigated the effect of PEG coverage, injected dose and release kinetics on the performance of PLA-PEG nanoparticles loaded with vincristine (an anticancer active compound), determining their impact on pharmacokinetics and biodistribution in in vivo animal model.

Pavot et al. (2013) synthesized PLA nanoparticles containing Nod1 and Nod2 receptors ligand; the aim is to induce a systemic immune response to improve the efficacy of vaccine delivery applications. Experimental outcomes showed promising results.

Zhou et al. (2018) developed nanoparticles based on hydroxyehyl starch-polylactide (HES-PLA) polymer, where they loaded two active compounds: doxorubicin and the TGF-β inhibitor LY2157299. This strategy involving combined delivery aims at suppressing both tumor growth and metastasis.

Medel et al. (2017) developed PLA-PEG nanoparticles loaded with anticancer drugs curcumin and bortezomib. These active compounds are highly hydrophobic and show synergistic effects; in addition, they can form a covalent complex stable at physiological pH but labile at midly acidic pH (such as cancer microenvironment). The use of nanoparticles improved the cytotoxicity provided by curcumin-bortezomib if compared to free, not-encapsulated drugs.

Raudszus et al. (2018) synthesized PLA nanoparticles using a newly developed stabilizer, a vinyl sulphone-modified poly(vinyl alcohol) (VS-PVA) derivative. By virtue of its enhanced reactivity, VS-PVA derivative allowed an easy functionalization of particle surface with targeting moieties such as Ovalbumin, Apolipoprotein E (ApoE), and Penetratin. In particular, ApoE and Penetratin functionalized particles exhibited a higher cellular uptake, associated to a specific interactions with cellular receptors.

Zhu et al. (2016) developed nanoparticles made of Dα-tocopherol polyethylene glycol succinate-poly(lactide) (TPGS-PLA) loaded with docetaxel, an anticancer compound. Nanoparticles were coated with polydopamine and functionalized with glucosamine in order to enhance cellular uptake in the liver through ligand-mediated endocytosis.

Zhang et al. (2016) synthesized PLA-PEG nanoparticles loaded with paclitaxel and functionalized with EGFP-EGF1 covalently bound to PEG brushes. In vivo experiments showed that such particles are able to target multiple types of key cells in tumor tissues.

Turino et al. (2017) developed paclitaxed-loaded PLGA nanoparticles functionalized with ferritin. Functionalization was possible thanks to the use of PLGA-NHS polymer, where one end group is constituted by succinimidyl ester, which reacts with protein amine groups. Nanoparticles were also loaded with a guanidinium-based (Gd-DOTAMA) agent for magnetic resonance imaging (MRI). Experimental studies in vitro proved the targeting capability using breast cancer cell lines.

Gourdon et al. (2017) investigated PLA-PEG nanoparticles loaded with acyclovir (antiviral drug) and functionalized with single amino acids or short peptides in order to target PepT1 intestinal transporter. Functionalization was performed by covalently linking amino acids to PEG chains with an amine end group. Valine-functionalized nanoparticles showed the best outcomes in terms of targeting.

Cui and Zhu (2016) prepared doxorubicin-loaded PLA nanoparticles, covered with polyethylene imine (PEI) that was functionalized with Herceptin, a monoclonal antibody, which targets the human epidermal growth factor receptor 2 (HER2), overexpressed in breast cancer. Functionalization improved cellular uptake and nanoparticles proved to enhance the therapeutic effect of the drug reducing side effects.

Xiong et al. (2016) synthesized a block copolymer containing folic acid, pluronic (a polyethylene oxide-poly propylene oxide-polyethylene oxide block copolymer) and lactic acid. Resulting product was employed to produce paclitaxel-loaded nanoparticles. Experimental data proved that folate included in polymer chain could be used for active targeting with folate receptor expressed in ovarian cancer cells.

Coolen et al. (2019) synthesized PLA nanoparticles, which exhibit a negative charge on the surface due to lactic acid resulting from degradation. In order to non-covalently bind mRNA on the surface (they are both negatively charged), the authors firstly created a non-covalent complex between mRNA and cationic cell penetrating peptides (CPPs), which could be adsorbed on PLA surface.

Tang et al. (2018) obtained PLA-PEG micelles loaded with paclitaxel, an anticancer drug. They functionalized the surface with a CPP linked to PEG polymer with a pHsensitive sequence composed of histidine and glutamic acid. At physiological pH, CPP, and the linker are strongly bound through electrostatic interactions between glutamic acid (in the linker) and arginine (in the CPP). The mildly acidic pH of the tumor microenvironment leads to the protonation of some histidine residues of the linker, which interfere with the linker/CPP electrostatic interactions. The immediate consequence is an increased exposure of the CPP in the cancer and thus a more effective targeting, which could be achieved with stimuli responsive device (pH, in this case).

Song et al. (2016) developed PLA-PEG nanoparticles loaded with AZD2811, a hydrophobic anticancer active compound. The authors extensively tested the hydrophobic ion pairing (HIP) approach in order to maximize drug loading and encapsulation efficiency. They evaluated different hydrophobic counterions (such as oleic acid, cholic acid, and so on) that increase AZD2811 hydrophobicity through the formation of ion pairs. Different counterions led to different release kinetics, which allowed obtaining a library of particle formulations.

Medina et al. (2019) synthesized PLA-PEG and PLGA nanoparticles, blended with low molecular weight PLA and PCL and lipid-conjugate PEG, respectively. They observed that this blending improved the encapsulation efficiency of adapalene (a topical retinoid). Blending had a moderate impact on release kinetics.

Discussed examples are summarized in **Table 10**.

# Nanoparticles for Imaging

Banerjee et al. (2017) radiolabeled with <sup>111</sup>In-containing moieties and IRDye680RD PLA-PEG nanoparticles functionalized with prostate-specific membrane antigen (PSMA) moieties. In vivo and ex vivo imaging allowed determining the distribution of both PSMA-functionalized and not-functionalized particles in the tumor. Xiong et al. (2017) treated Fe3O<sup>4</sup> iron oxide nanoparticles with PLA-PEG chains for MRI purposes. End groups of PLA-PEG chains where functionalized with D-glucosamine as targeting agent. In vivo MRI in tumor-bearing mice confirmed the ability of the nanoparticles to target the cancer and their potential as contrast agents.

Dos Santos et al. (2017) synthesized PLA nanoparticles loaded with betamethasone and dexamethasone (antiflammatory drugs) and labeled with technetium-99m. Experiments showed that betamethasone loaded particles were able to accumulate in the inflammation site in an in vivo model of S. aureus infection.

Kerr et al. (2017) synthesized dye-PLLA-conjugates using Difluoroboron β-Diketonates as dyes, which were subsequently employed to obtain nanoparticles (average diameter 55 nm) for imaging purposes. In order to improve stability, the authors added PDLA-PEG to promote stereocomplexation. In vivo experiments proved the ability of such particles to target tumors.


#### TOXICITY OF POLYLACTIC-BASED NANOPARTICLES

Nanoparticles behavior mainly depends on size, shape, morphology, and surface charge; this holds also for their unwanted side effects. Entering into cells, nanoparticles can provide cytotoxic effects, leading to the disruption of cell membranes or cytosolic components or to programmed cell death (apoptosis). Typical adverse effects are oxidative stress [an excess of reactive oxygen species (ROS) or reactive nitrogen species (RNS) usually neutralized by cells], apoptosis, cytokine activation (due to inflammatory response), loss of mitochondrial, and lysosomal stability. They can also be a source of genotoxic effects, damaging DNA (Ganguly et al., 2018).

In addition, nanoparticles may induce haemolysis (disruption of red blood cells) or blood coagulation (causing thrombosis) (Dobrovolskaia and McNeil, 2007). PLA-based nanoparticles may provide additional side effects through their degradation products.

Toxicity assessment in vitro in usually performed by exposing cells to given dose of the potential toxic agent and measuring, e.g., cell viability and proliferation, mitochondrial activity, ROS production, cytokine activation through suitable assays.

In vivo experiments aim at assessing the pharmacokinetics of nanoparticles, their distribution in the organ and their clearance.

While in vitro and in vivo testing are usually performed when a new formulation is discussed (vide supra) up to author's best knowledge, there are few systematic studies concerning toxicity of PLA-based nanoparticles (Da Luz et al., 2017; Da Silva et al., 2019).

Singh and Ramarao (2013), e.g., observed that PLA nanoparticles in vitro did not provide detrimental effect concerning RNS, cytokine activation, mitochondrial or lysosomal integrity. At high concentration, they stimulated ROS production and inflammation; this was linked to the accumulation of polymer degradation products in the cell.

There is an additional intrinsic risk of toxic responses when nanoparticles are used and injected in the blood stream. Nanoparticles interact with the components present in the environment (proteins, carbohydrates, small molecules, etc.) through their surface. The driving force leading to the formation of this nano-bio interface are already known in scientific literature and are basically due to electrostatic and Van der Waals interactions as well as hydrophobic and depletion effects (Nel et al., 2009). One of the main outcomes from this network of interactions is the attainment of a layer of adsorbed proteins on nanoparticle surface, usually referred as protein corona (Cedervall et al., 2007). Because of the interactions with the surface, adsorbed proteins can be subjected to relevant structural changes, which can lead to aggregation and fibrillation, loss of enzymatic activity, or the exposure of new antigenic epitopes. Such side effects emerge from the specific protein-surface interactions: while, as mentioned, driving forces are known, they depend on many factors (materials, pH, ionic strength, etc.) and are challenging to be determined a priori.

In vitro experiments allow a rapid and cost-effective evaluation of toxicity if compared to in vivo experiments with animal models (and the related ethical concerns). However, the possible lack of correlation between in vitro-in vivo tests, the challenging extrapolation of animal data to human patients and the shortage of harmonized protocols are still obstacles for extensive clinical trials (Dobrovolskaia and McNeil, 2013).

#### REFERENCES


#### CONCLUSIONS

PLA—based polymers have been extensively studied in literature and are currently an established reality in the biomedical field, thanks to their interesting properties. This led to a growing interest also in the nanomedicine field for the synthesis of nanoparticles for drug delivery and imaging purposes. Nanoparticles showed a great potential as nanocarriers to deliver poorly soluble drugs, proteins, and genes targeting the tumor and releasing the active compound at the desired rate, enhancing in the therapeutic effect.

The new perspective introduced by nanoparticles also brings new sources of toxicity connected with cytotoxicity and haemolysis; also protein corona can provide undesired side effects that are not easily predictable a priori.

Nowadays, an extensive clinical application of nanoparticles is still hindered by an exhaustive assessment of potential toxic effects, which does not allow nanoparticles unleashing their full potential.

### AUTHOR CONTRIBUTIONS

TC performed literature research and wrote the first draft of the paper. All authors discussed and approved the contents of the manuscript and contributed to its final version by reading and editing.

# FUNDING

This study is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016, from the CTI (1.1.2018 Innosuisse) under grant agreement number 19267.1 PFNM-NM and from FCT Foundation for Science and Technology under the project PROSAFE/0001/2016.


administration in rats: a microdialysis study. Biopharm. Drug Dispos. 29, 431–439. doi: 10.1002/bdd.621


new carrier of protein delivery system. J. Controlled Release 107, 158–173. doi: 10.1016/j.jconrel.2005.06.010


**47**


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

Copyright © 2019 Casalini, Rossi, Castrovinci and Perale. 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.

# Molecular Modeling for Nanomaterial–Biology Interactions: Opportunities, Challenges, and Perspectives

Tommaso Casalini <sup>1</sup> \*, Vittorio Limongelli 2,3, Mélanie Schmutz <sup>4</sup> , Claudia Som<sup>4</sup> , Olivier Jordan<sup>5</sup> , Peter Wick <sup>6</sup> , Gerrit Borchard<sup>5</sup> and Giuseppe Perale1,7

<sup>1</sup> Polymer Engineering Laboratory, Department of Innovative Technologies, Institute for Mechanical Engineering and Materials Technology, University of Applied Sciences and Arts of Southern Switzerland (SUPSI), Manno, Switzerland, <sup>2</sup> Faculty of Biomedical Sciences, Center for Computational Medicine in Cardiology, Institute of Computational Science, Università della Svizzera italiana (USI), Lugano, Switzerland, <sup>3</sup> Department of Pharmacy, University of Naples "Federico II", Naples, Italy, <sup>4</sup> Technology and Society Laboratory, Swiss Federal Laboratories for Materials Science and Technology (Empa), St. Gallen, Switzerland, <sup>5</sup> School of Pharmaceutical Sciences, University of Geneva, Genève, Switzerland, <sup>6</sup> Laboratory for Particles – Biology Interactions, Swiss Federal Laboratories for Materials Science and Technology (Empa), St. Gallen, Switzerland, <sup>7</sup> Ludwig Boltzmann Institute for Experimental and Clinical Traumatology, Wien, Austria

#### Edited by:

Roberto Molinaro, University of Urbino Carlo Bo, Italy

#### Reviewed by:

Giosuè Costa, University of Catanzaro, Italy Isabella Romeo, University of Calabria, Italy

\*Correspondence: Tommaso Casalini tommaso.casalini@supsi.ch

#### Specialty section:

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

Received: 09 July 2019 Accepted: 27 September 2019 Published: 17 October 2019

#### Citation:

Casalini T, Limongelli V, Schmutz M, Som C, Jordan O, Wick P, Borchard G and Perale G (2019) Molecular Modeling for Nanomaterial–Biology Interactions: Opportunities, Challenges, and Perspectives. Front. Bioeng. Biotechnol. 7:268. doi: 10.3389/fbioe.2019.00268 Injection of nanoparticles (NP) into the bloodstream leads to the formation of a so-called "nano–bio" interface where dynamic interactions between nanoparticle surfaces and blood components take place. A common consequence is the formation of the protein corona, that is, a network of adsorbed proteins that can strongly alter the surface properties of the nanoparticle. The protein corona and the resulting structural changes experienced by adsorbed proteins can lead to substantial deviations from the expected cellular uptake as well as biological responses such as NP aggregation and NP-induced protein fibrillation, NP interference with enzymatic activity, or the exposure of new antigenic epitopes. Achieving a detailed understanding of the nano–bio interface is still challenging due to the synergistic effects of several influencing factors like pH, ionic strength, and hydrophobic effects, to name just a few. Because of the multiscale complexity of the system, modeling approaches at a molecular level represent the ideal choice for a detailed understanding of the driving forces and, in particular, the early events at the nano–bio interface. This review aims at exploring and discussing the opportunities and perspectives offered by molecular modeling in this field through selected examples from literature.

Keywords: molecular dynamics, metadynamics, molecular modeling, protein corona, coarse grain, lipid bilayer, cellular membrane

# INTRODUCTION

Nanomedicine is an emerging discipline that is providing novel impulses to the biomedical field thanks to the use of nanotechnologies and the continuous development of engineered nanomaterials such as polymer-, metal- or metal oxide-based nanoparticles. Nanomaterials, by virtue of their small size (1–1000 nm, comparable to many biological molecules like proteins and viruses) open up a wide range of new opportunities and applications, for example as devices for targeted drug delivery and diagnostic purposes and as image contrast agents. However, as with every novel technology, the potential negative side effects have to be assessed early in the development process to avoid adverse social and economic effects.

Indeed, the injection of nanomaterials into an organism leads to complex interactions between the surface of the device and the components of the medium, such as proteins, carbohydrates, fatty acids, et cetera. These interactions play a key role in determining not only the fate of the nanomaterial (in terms of clearance and in vivo biodistribution) but also the attainment of undesired side effects. The fundamental driving forces governing the formation of this nano-bio interface have already been identified and discussed (Nel et al., 2009) and include van der Waals and electrostatic interactions and hydrophobic and depletion effects. The challenge lies in the rationalization of the synergistic effects of intrinsic nanomaterial properties (chemical composition, size, surface functionalization, et cetera), the characteristics of the surrounding medium (pH, ionic strength, et cetera), and the phenomena occurring at the interface and their impact on nanomaterial behavior.

One of the most relevant consequences is the formation of the protein corona, i.e., a layer of adsorbed proteins on the NP surface (Cedervall et al., 2007a,b; Lundqvist et al., 2008; Dell'orco et al., 2010). The attainment of such a network alters the surface properties of the nanomaterial, which may cause substantial deviations from the expected behavior concerning colloidal stability, cellular uptake, clearance, distribution within the organs, and immune response.

On top of that, the formation of the protein corona can lead to changes in the protein structure and thus to undesired consequences (not easily predictable a priori), such as (Nel et al., 2009):


Experimental protocols for the investigation of the protein corona are currently well-established (Walkey and Chan, 2012; Wei et al., 2014; Pederzoli et al., 2017), although they have some intrinsic limitations concerning spatial and temporal resolution; indeed, they do not allow the characterization of the early events leading to protein corona formation and do not provide a clear overview of specific nanomaterial/protein interactions or changes in protein structure.

Computational approaches at the molecular scale, such as molecular dynamics (MD) simulations, constitute the natural complement to experimental techniques. This is due to several factors, such as the accessible time and length scales (microsecond and nanometer, respectively), the full atomistic description of the system (which allows the specific protein/nanomaterial interactions to be identified) and its dynamic behavior (thus identifying conformational changes after binding), and the inclusion of environmental effects.

This review aims at exploring and discussing the opportunities and limitations of nano-bio as well as giving some perspectives on the use of molecular modeling techniques for characterizing these interactions. After giving a brief theoretical background, relevant applications of simulations at the molecular scale are discussed through selected examples from the scientific literature.

# MOLECULAR MODELING—A BRIEF OVERVIEW

Molecular modeling can be seen as the sum of two components: a molecular model and a computational technique to properly characterize the behavior of the molecules.

Building a suitable molecular model, that is, how the system under investigation is rationalized and represented in the framework of a meaningful simulation, is the first fundamental step. In this framework, molecular models can be essentially divided into two categories; on the one side, full atomistic models provide the highest level of detail since all atoms (considered as the smallest constitutive units of the model) are explicitly accounted for. On the other side, coarse-grained models summarize the atomic detail by enclosing groups of atoms into beads that lump the main peculiarities (in terms of charge, polarity, et cetera) of the atoms that they embed. This simplification is unavoidable for complex systems whose atomistic representation would be prohibitive from a computational point of view, in terms of the system size and/or time and length scales needed to investigate the phenomena of interest. Despite the loss of detail, a coarsegrained model that retains the main features of the system is able to provide meaningful insights at a reasonable computational cost (vide infra). For the sake of completeness, there exist more detailed representations where electrons are the smallest constitutive units and are explicitly included. Such models are treated with quantum chemistry methods, which are not considered or discussed here since their application in the field of nanomedicine is hindered by their computational inefficiency.

In a broader sense, a molecular model also includes unavoidable simplifications that allow for the simulation of complex systems, either at a full atomistic or coarse-grained level of detail, which could not be treated otherwise. The simulation of protein adsorption on a microparticle surface, for example, is unfeasible because of the system size. Such a system is usually simplified by adopting a molecular model that involves the adsorption of a protein on a flat surface with a suitable thickness. This approach is reasonable since the phenomena of interest are restricted to the solvent/particle interface; in addition, since protein size is much smaller than microparticle radius, curvature effects can be reasonably neglected.

The second component of molecular modeling is constituted by suitable computational methods that allow the characterization of the dynamics, energetics, and conformational sampling of the system of interest. Full atomistic models are usually treated with molecular dynamics, while other techniques such as coarse-grained molecular dynamics and dissipative particle dynamics are employed along with coarse-grained models.

Each method has its own strengths and limitations, as well as characteristic accessible time and length scales, as discussed in the following paragraphs.

# Full Atomistic Models—Molecular Dynamics

In molecular dynamics simulations, atoms are represented as spheres that interact with each other by virtue of a potential energy function, usually called the force field (FF). Molecular coordinates and velocities as a function of simulation time can be evaluated by solving Newton's equation of motion with a suitable numerical integration scheme, as shown in Equation (1) (Frenkel and Smit, 2002):

$$m\_i \frac{d^2 r\_i}{dt^2} = F\_i = -\nabla U(r) \tag{1}$$

where m<sup>i</sup> isthe mass of the i-th atom,r<sup>i</sup> are the spatial coordinates of the i-th atom, t is time, F<sup>i</sup> is the force acting on the i-th atom, and U(r) is the potential energy (that is, the force field), which is a function of the coordinates of all atoms present in system r. Such an approach essentially implies a couple of assumptions, as follows. First, the motion of electrons can be reasonably described by the dynamics of the corresponding nuclei (Born–Oppenheimer approximation). Second, the motion of the atomic nuclei (which are heavier than electrons) can be described as point particles that follow classical mechanics; this is an acceptable approximation when quantum effects are not important (Frenkel and Smit, 2002). Generally speaking, a force field takes into account both intramolecular and intermolecular interactions, in terms of bonds, angles, dihedrals, and long-range interactions, namely van der Waals and electrostatic.

FFs contain several parameters that are computed in order to reproduce the conformational energies and minimum energy structures obtained from high-level quantum mechanics calculations and/or experimental data, such as hydration enthalpies or structural parameters from NMR experiments (Riniker, 2018). There are "general purpose" force fields, usually employed to describe small ligands, as well as FFs specifically tailored for given categories of molecules, like proteins, nucleic acids, carbohydrates, and lipids (Riniker, 2018). The choice and the quality of the force field cannot be underestimated, since they strongly affect the reliability of the simulation outcome.

MD simulations do not explicitly consider electrons, so chemical reactions and excited states cannot be investigated; however, they constitute the ideal tool for those systems that are mainly governed by non-covalent interactions, like electrostatic and Van der Waals forces. MD also allows environmental conditions to be included through the addition of explicit solvent molecules, ions, and other solute molecules into the system. The main outputs from an MD simulation are molecular trajectories, the post-processing of which can provide structural information (binding poses, protein conformation) as well as energetic information such as interaction energies.

#### Enhanced Sampling Methods

The characteristic time and length scales of MD simulations are in the tens to hundreds of nanoseconds (up to 1000 ns) and tens of nanometers (up to 20 nm), respectively. However, many phenomena of interest (e.g., molecular binding, protein unfolding) need large time scales to occur (up to minutes), and their investigation through MD would be in principle unfeasible; this is due to the presence of metastable states separated by high free energy barriers. A way to overcome this issue is to use enhanced sampling methods, which allow enhancement of the transitions between different metastable states separated by energy barriers higher than the thermal energy kBT, which would not be crossed in a standard simulation at temperature T (where k<sup>B</sup> is the Boltzmann constant and T is absolute temperature). As recently reviewed (Camilloni and Pietrucci, 2018), there are three different suitable approaches: i) increasing the temperature T; ii) changing the potential U(r), and iii) adding an external bias potential V(r). Each approach has its own methods, the discussion of which (along with their theoretical basis) is well beyond the purpose of this review; the interested reader is referred to ad hoc reviews (Miao and Mccammon, 2016; Camilloni and Pietrucci, 2018). Some of the popular enhanced sampling techniques are Replica Exchange (RE, first approach) (Miao and Mccammon, 2016) and Well-Tempered Metadynamics (WTM) (Valsson et al., 2016), which belongs to the third group. In particular, WTM and its variant forms allow the free energy of the system under investigation to be recovered by adding an external bias on a selected number of degrees of freedom, commonly referred to as collective variables (CVs). CVs are generally functions of atomic coordinates and can range from simple quantities, such as distances and dihedral angles, to more complicated variables, like the number of hydrogen bonds/hydrophobic contacts, alpha helix-content in a protein, or Debye–Hückel interaction energy. Collective variables must be chosen so that they can discriminate between metastable states and can be representative of the transition mechanism. Typical applications of WTM and WTM-based methods are the study of protein conformations (also in the presence of denaturants) (Owczarz et al., 2015), the binding poses of small ligands to target proteins (Tiwary et al., 2015), and the conformation and self-assembly of polymeric and supramolecular systems (Bochicchio and Pavan, 2018). Some phenomena, such as protein folding, require a relevant number of collective variables to perform meaningful simulations. Although conceptually feasible, running a WTM simulation with many CVs introduces some issues such as a drop in computational efficiency and a non–trivial analysis of the results obtained. In order to overcome this issue, some WTM variants have been proposed, discussed, and validated in literature (mainly for protein folding), namely Parallel Tempering Metadynamics (PTMD) (Bussi et al., 2006), Parallel Tempering Metadynamics in the Well-Tempered Ensemble (PTMD-WTE) (Deighan et al., 2012), and Bias Exchange Metadynamics (BEMD) (Piana and Laio, 2007). The discussion of the theoretical basis of these methods is beyond the purpose of this review; the interested reader is referred to the corresponding papers.

# Coarse-Grained Models—Molecular Dynamics, Dissipative Particle Dynamics

The aim of coarse-grained (CG) models is to perform meaningful simulations of systems whose analysis would be challenging or unfeasible with full atomistic MD methods by building simplified representations that allow the main physical/chemical features (like the interplay between hydrophobic and hydrophilic effects) to be retained.

In the coarse-graining procedure, groups of atoms are enclosed into "beads" or "interaction sites" that are representative of the embedded atoms in terms of charge, size, hydrophobicity/hydrophilicity, et cetera. Beads interact with each other by virtue of a potential energy function, which takes into account both bonded interactions (that is, bond, angles, and dihedrals) and non-bonded interactions and which is parameterized in order to optimally reproduce some experimental properties (like water/octanol partition) or the behavior of more detailed full atomistic simulations.

Trajectories can be computed by integrating Newton's equation of motion and also adding other components to the force such as friction due to the solvent (if implicit solvent methods are used) (vide infra).

It is worth mentioning that the coarse-graining procedure can be performed to different extents, since a bead can enclose a group of atoms (3–4 heavy atoms), a group of monomers (or amino acids), an entire protein or an entire microparticle, according to the aim of the simulation. In this review, the term "coarse-grained models" is employed for all those approaches where there is a loss of degrees of freedom with respect to a full atomistic description.

A common drawback of CG models is that parameterization is strictly tailored for the system under investigation and in principle should be repeated for every new system; in other words, parameters are not transferable. In this regard, the MARTINI force field (Marrink et al., 2007) attracted a lot of interest due to its reliability and straightforward coarse-graining procedure. Beads (which include groups of 3–4 heavy atoms) still interact with each other through a simple potential energy function, as described for MD (vide supra). MARTINI offers a library of parameterized beads, mainly divided into four categories: polar, non-polar, apolar, and charged; in addition, each group includes subgroups representative of polarity and hydrogen bond capability. Parameters for bonded interactions (bonds, angle, dihedrals) must be determined from detailed MD simulations, while non-bonded interactions are tuned in order to reproduce thermodynamic properties like free energy of hydration, free energy of vaporization, and partitioning between water and different solvents. Explicit water and ions can also be added (a MARTINI water bead is representative of four water molecules). An example of MARTINI mapping from a full atomistic to a coarse-grained system is shown in **Figure 1**.

Bead parameterization can be further refined by the user in order to improve agreement with full atomistic simulations. Even with simulations based on the MARTINI force field, some phenomena of interest can be still characterized at a time scale that is not accessible. In this framework, enhanced sampling methods like Metadynamics can be employed to alleviate this issue, as already shown in the literature (Lelimousin et al., 2016).

Another widely employed method with CG models is Dissipative Particle Dynamics (DPD). Bead trajectories are still obtained by means of Newton's equation of motion, assuming that each i-th particle is subjected to three pair-additive forces that arise from the interactions with the other j-th particles: a conservative force, a dissipative force, and a random force (Liu et al., 2015):

$$m\_i \frac{d^2 r\_i}{dt^2} = f\_i = \sum\_{j \neq i} F\_{ij}^\circ + F\_{ij}^d + F\_{ij}^r \tag{2}$$

The conservative force F c is due to the interaction potential of particles and accounts for both bonded and long-range interactions through an elastic force and a soft repulsion force, respectively. F d is a dissipative force that damps the relative motion between particles, and F r is a random force directed along the line that connects beads centers. Dissipative and random forces are momentum-conserving and represent the minimal model that takes into account viscous forces and thermal noise between particles.

### Full Atomistic vs. Coarse-Grained Models: Strengths and Weaknesses for Nanomaterial–Biology Interactions

In this framework, full atomistic models provide the highest level of detail, since all atoms are explicitly included. On the one side, they account for all those fundamental interactions that are essential for a suitable description of the nano–bio interface, such as van der Waals, electrostatic, hydrogen bonding, π-π stacking, and π – cation interactions (provided the intrinsic limits and the accuracy of the FF). On the other side, the inclusion of explicit solvent molecules, ions, and other solute molecules allows environmental effects to be taken into account; the impact of pH is accounted for by appropriately changing protonation states. Focusing on proteins, by means of molecular dynamics simulations and their resolution at atomic scale it is possible to highlight the most relevant amino acids that drive the interactions at the nano–bio interface and protein structural changes at the single amino acid level, achieving a level of detail that is usually out of reach from an experimental point of view. On top of this, the reliability of the simulation results can be assessed by comparing theoretical quantities such as circular dichroism spectra with the corresponding experimental outcomes. The importance of this aspect cannot be underestimated since it strengthens the connection between experiments and simulations. Molecular dynamics simulations are still limited by the characteristic time and length scales accessible by the method: microseconds and nanometers, respectively. The direct use of enhanced sampling methods is still prohibitive for complex and/or large systems. In this regard, switching to coarse-grained models is a forced but attractive choice due to the longer accessible time and length scales (tens of microseconds and tens of nanometers, respectively). The drawback is the loss of the atomic detail, which implies that some

interactions (strong electrostatic interactions, hydrogen bonds, solvation effects) are accounted for only in a roughly qualitative way. Anyway, if the fundamental physical/chemical peculiarities of the system (such as the balance between hydrophobic and hydrophilic groups) are well reproduced in the CG model and if the interaction potentials (that govern the interactions between beads) are accurately parameterized against experimental data or validated simulations at atomic scale, simulations at CG scale are a powerful tool to complement the insights obtained with MD simulations. CG simulations can also provide some input guess structures for, e.g., protein–protein interactions (that would be challenging to obtain with MD simulations), which can be further employed for a more accurate analysis at atomic scale. On top of that, enhanced sampling methods (in particular, Well-Tempered Metadynamics) have proved to be useful for simulations at CG scale when the time scale is still not accessible.

All these aspects are discussed in detail, along with selected examples, in the following paragraphs.

# APPLICATIONS FOR NANOMATERIAL–BIOLOGY INTERACTIONS

Molecular modeling is essentially employed for two purposes in the framework of nanomaterial–biology interactions. On the one side, it can shed light on the early events leading to the protein corona, highlighting the main mechanisms behind protein adsorption on the nanomaterial surface (hydrophobic effects, hydrogen bonds, electrostatic interactions, et cetera), the most important amino acids involved in the binding and the attainment of conformational changes. On the other side, simulations at the molecular scale allow the evaluation (in a trend-wise manner) of the impact of environmental effects, nanoparticle material, and surface functionalization on cellular uptake; some preliminary theoretical insights can also be obtained concerning the effect of protein corona formation.

# Protein Corona

Molecular modeling, thanks to its resolution at the atomic scale, represents the natural choice for the study of early events that lead to protein corona formation. Knowledge of the structural changes experienced by the protein after adsorption is essential for understanding system behavior, as discussed in the introduction (vide supra). Molecular modeling can offer an exhaustive overview of the structural transitions thanks to the resolution at a molecular level, highlighting the portion of proteins subjected to structural changes (along with the most important amino acids that cause this) and the main driving forces (electrostatic interactions, hydrophobic effects, et cetera). This allows information to be obtained that is challenging or impossible to achieve experimentally, and this is why molecular modeling has emerged as the natural and ideal complement to experiments. A typical application is constituted by detailed MD simulations of the interactions between a protein and a particle and the resulting changes in protein structure. The particle is usually modeled as a flat surface. On the one hand, there is no need to account for the entire sphere, since the interactions occur only at the interface. On the other hand, if the size of the protein is much smaller than the particle size, surface curvature effects can be safely neglected; this approximation is not valid for nanoparticles, whose size is comparable to those of proteins, and particle curvature must be accounted for by building the molecular model of the NP surface properly.

In this framework, full atomistic simulations can provide a detailed picture of the structural changes experienced by the protein after adsorption at the surface in terms of modifications of its secondary and tertiary structure (increase/decrease of alpha-helix and beta-sheet motifs and their arrangement). The specific structural changes of the protein can be directly correlated with experimental data, circular dichroism results, or NMR spectra. In addition, since protein adsorption modifies the properties of the particle surface (in terms of charge, hydrophobicity, et cetera), the insights obtained can be correlated, e.g., to differences in the colloidal stability of the particle suspension or other phenomena related to the protein corona such as protein aggregation and fibrillation.

Environmental effects can be taken into account thanks to the addition of explicit solvent molecules and ions, so that given salt concentrations (i.e., ionic strength) can be included in the simulation. The effect of pH can be included by changing the protonation state of the protein and the NP surface accordingly; anyway, protonation states in MD simulations are fixed and not dynamic since proton exchanges are not simulated. In other words, a positively charged amino acid remains protonated during the entire simulation, although the proton may be exchanged with surrounding water molecules according to the environmental pH. On top of that, the acid dissociation constant can be heavily influenced by local environmental effects such as the neighboring units and exposure to the solvent. This issue can be overcome by means of constant pH methods, which are currently available and validated only for proteins (Swails et al., 2014).

Simulations can also account for surface functionalization and its impact on the interactions with the protein. Through trajectory post-processing, it is possible to identify the main driving forces behind adsorption (hydrophobic effects, hydrogen bonds, et cetera) and to compute interaction energies in order to obtain a quantitative estimation of the strength of the binding.

Although the results of such simulations can surely contribute to increasing understanding and rationalizing experimental data, this approach also has some limitations and drawbacks.

The accuracy and reliability of the simulated protein structural changes are strongly related to the robustness of the force field; if FF parameterization leads to, e.g., an overestimation of alphahelix content, this will unavoidably affect the simulation results. Several articles where force field performances are systematically analyzed, as well as reference FF papers, address such limitations in detail, which are therefore known a priori. It is also worth mentioning that force field improvements are continuously carried out, and updated FF versions are periodically released. In principle, changes in protein secondary and tertiary structure can occur on time scales beyond those accessible to standard MD simulations (ns–µs), so the use of enhanced sampling methods often becomes an inescapable necessity to achieve meaningful results. Standard MD simulations provide an ensemble of conformations according to the given conditions (temperature, solvent, ionic strength, et cetera), but if two metastable states are separated by an energy barrier much higher than the thermal energy, kBT, some relevant protein conformations are not accounted for because this barrier would not be crossed and simulation outcomes can provide only a partial description of the event under investigation. The use of enhanced sampling methods alleviates this issue.

Simulations are usually focused on the adsorption of a single protein on a surface, which is essentially representative of particles in a very dilute protein solution; in other words, the overall protein–protein interactions are neglected since they can occur on long time scales and their description is usually challenging, even with enhanced sampling methods. Although simulations provide interesting insights, systematic and rational validation of the molecular models is still lacking. This currently hinders the extensive use of molecular simulations for practical applications, such as the engineering of nanoparticles in order to promote or discourage the adsorption of given proteins.

In this regard, the use of coarse-grained models, along with suitable techniques to study system dynamics, represents an inescapable choice, although the atomic detail is lost. CG models allow longer time and length scales to be explored than do full atomistic models coupled with MD simulations and can thus be used to investigate the impact of protein–protein interactions, overcoming the infinite dilution condition. Small nanoparticles can be explicitly included, and the surface curvature effect can be taken into account. However, the coarse-graining procedure is not painless due to its intrinsic limits: strong electrostatic interactions, solvation effects, and anisotropic interactions like hydrogen bonding are poorly described. Focusing on proteins, it is still challenging to account for changes in secondary structures. Therefore, an accurate parameterization of interaction potentials is an essential step in obtaining reliable results. Simulations at CG scale, despite the mentioned drawbacks, can still provide useful insights and can also be employed to obtain input guess structures for protein–protein interactions that can subsequently be investigated at an atomic level. The interaction potentials are usually parameterized against more accurate simulations with full atomistic models, whose validity, in turn, must be evaluated through comparison with experimental data. This further reinforces the need for systematic experimental validation.

The advantages and disadvantages (related to both MD and CG approaches) are summarized in **Table 1**.

As mentioned above, molecular models still need to be validated against comparison with experimental data. Indeed, for every property of interest, it is possible to highlight reference experimental techniques as well as computational techniques, as summarized in **Table 2**.

The literature offers several examples of MD simulations of protein adsorption on different materials, such as graphene sheets (Chong et al., 2015), carbon nanotubes (Ge et al., 2011; Gu et al., 2015), gold nanoparticles/surfaces (Wang et al., 2013; Brancolini et al., 2014; Tavanti et al., 2015; Bellucci et al., 2016; Yang et al.,


TABLE 2 | Reference experimental and computational techniques for properties of interest of the protein corona.


2017; Ma et al., 2018), hydroxyapatite surfaces (Dong et al., 2007, 2011), fullerenes (Leonis et al., 2015), titanium oxide surfaces (Utesch et al., 2011; Mudunkotuwa and Grassian, 2014), and molybdenum disulfide (Gu et al., 2018), highlighting the specific interactions behind the binding and the attainment of structural changes. Interestingly, there are no relevant computational studies of protein adsorption on polymer surfaces. To our best knowledge, this may be due to the limited availability of validated FF parameters for polymers and to intrinsic issues with the design of molecular models. Whereas inorganic nanoparticles are characterized by an ordered atomic arrangement, a model of a disordered polymer random coil can be more challenging to build.

Among many theoretical works, only a few papers combine experimental and computational components in order to achieve an all-round understanding of the mechanisms that lead to hard corona formation. Chong et al. (2015) adopted MD simulations to study the affinity of four abundant plasma proteins (bovine fibrinogen, immunoglobulin, transferrin, and bovine serum albumin) on graphene oxide and reduced graphene oxide surfaces. The affinity trend predicted by MD is in agreement with the experimental trend for all investigated proteins. Simulations also allowed determination of the most relevant residues for the binding. Gu et al. (2018) studied the interactions of MoS<sup>2</sup> nanoflakes with potassium channels proteins highlighting potential toxic effects of the binding, which can alter the biological function. The results were further corroborated by experimental data.

As mentioned, enhanced sampling methods are currently also applied for the study of protein–surface interactions with both MD simulations (where the system is described at full atomistic level) and CG simulations (where the atomic detail is lost for the sake of computational efficiency). Indeed, the accessible time scale may not be adequate for the phenomena under investigation, and the use of enhanced sampling methods is a good solution for both MD and CG simulations.

Even if standard simulations are sufficient for small peptides, the application of enhanced sampling methods improves the efficiency of the sampling and provides additional information about the system thanks to the possibility of reconstructing the free energy as a function of the degrees of freedom of interest. In this regard, Metadynamics-based methods have proved to be a promising choice. Prakash et al. (2018) systematically analyzed the use of Metadynamics-based methods for the adsorption of GGKGG peptide on a silica surface, explicitly including the influence of ionic strength and ion charge; the authors discussed the performances of each method and suggested the best collective variables to account for, thus providing useful guidelines for meaningful simulations. Deighan and Pfaendtner (2013) employed Metadynamics to study the influence of surface functionalization on the adsorption of Lkα14 and Lkβ15 peptides on self-assembled monolayers; the model outcomes were in good agreement with experimental findings. Bellucci et al. (2016) investigated the adsorption of Aβ16−<sup>22</sup> peptide on a gold surface in order to investigate the impact of the binding on fibrillation. Their simulations revealed that binding poses are mainly influenced by the affinity between gold and phenylalanine, as shown in **Figure 2A**. The model was also validated through a comparison between experimental and calculated spectra obtained through sum generation frequency (SFG) spectroscopy (**Figure 2B**).

Hildebrand et al. (2018) employed Metadynamicsbased methods to examine the conformational changes of Chymotrypsin after adsorption on silica. Simulations highlighted that the enzyme loses part of its helical content with minor perturbation of the tertiary structure; the model results were used to compute a theoretical circular dichroism spectrum that was in good agreement with the experimental spectrum.

of Chemistry.

CG models are also extensively used (Bellion et al., 2008; Vilaseca et al., 2013; Ding and Ma, 2014; Lopez and Lobaskin, 2015; Tavanti et al., 2015; Yu and Zhou, 2016; Hu et al., 2017; Wei et al., 2017), since they allow the characteristic accessible time and length scales of full atomistic simulations to be extended and the computational cost to be reduced. It thus becomes possible to simulate entire nanoparticles whose size is equal to or less than about 20 nm (at least when MARTINI is employed), fully covered by one or more kinds of proteins. The investigation of larger particles is still also challenging for CG methods because of the required computational effort.

Adopting CG models implies the loss of the atomic detail at the single amino acid level and a less accurate description of the system. While hydrophobic effects are reasonably accounted for, it is challenging to take into account properly, e.g., water structuring, cation–π interactions, strong electrostatic interactions, and hydrogen bonds, which lose their directionality because of the coarse-graining procedure (Marrink and Tieleman, 2013). Focusing on proteins, changes in tertiary structure can be reasonably described, while it is still non-trivial to account for changes in secondary structure due to the intrinsic limitations of the method (Marrink and Tieleman, 2013).

Despite such limitations, CG models can be employed for qualitative insights or to obtain guess structures for subsequent more detailed full atomistic simulations, as is commonly done, e.g., for the non-covalent protein–protein interaction and oligomerization of membrane proteins (Lelimousin et al., 2016). Anyway, a systematic use for more quantitative results must first be corroborated through comparison with more accurate and, above all, validated atomistic MD simulations.

Yu and Zhou (2016) used CG simulations with the MARTINI force field to understand the influence of nanoparticle curvature on lysozyme adsorption on silica at different values of ionic strength. They found that while salt concentration had a modest effect, surface curvature greatly influenced structural changes.

Ding and Ma (2014) used dissipative particle dynamics to characterize the adsorption of human serum albumin (HSA) on generic hydrophobic, hydrophilic, and charged nanoparticles for different size and pH values. By computing binding free energy as a function of the distance between the protein and particle centers of mass (COM), they showed that HSA could be bound only to hydrophobic and positively charged nanoparticles. They further studied the attainment of the protein corona by computing the number of adsorbed proteins as a function of particle size at neutral pH for hydrophobic and positively charged particles.

The reported studies are summarized in **Table 3**.

Although the reported examples of simulations at the CG scale provide interesting findings, they are not coupled with validation against experimental data; therefore, the results should be taken as qualitative theoretical considerations. Notably, the literature offers many examples concerning inorganic nanoparticles (gold, silica) or carbon-based materials (graphene, carbon nanotubes). To our best knowledge, polymeric systems are not widely investigated. This is due to a lack of validated parameters as well as intrinsic issues related to system modeling since, by virtue of their ordered atomic arrangement, inorganic surfaces can be more easily described than a polymer surface composed of a disordered random coil.

To summarize, at this stage, molecular modeling of the protein corona cannot replace experimental activity, and its use as a purely predictive tool is currently premature. This is due, on the one side, to the intrinsic complexity of the system under investigation and, on the other side, to the lack of systematic validation against experimental data. Many examples discussed in the literature are purely theoretical, and only a few recent studies have critically validated simulation outcomes with experiments. In addition, comparison with experimental data is only performed in vitro; the complexity of the in vivo environment still constitutes an arduous challenge because of


the wide range of proteins present in the blood flow and their mutual interactions.

It is important to take into account another limitation of the method: usually, the investigation is focused only on the proteins directly adsorbed on the nanoparticle, usually modeled as a flat surface if the particle size is much larger than that of the protein. Small nanoparticles can be entirely included in the simulations, while in intermediate cases the molecular model of the surface must account for curvature effects.

Molecular simulations must be intended as the ideal complement to experimental activity in vitro. As shown in **Table 2**, simulation outcomes can be compared with the corresponding experimental information, providing a deeper understanding thanks to the detail provided at the molecular level.

The road toward purely predictive simulations is still long and arduous, but the main points to be addressed are clear. On the one side is the development of more reliable force fields that can accurately capture the structural transitions of proteins (in terms of both secondary and tertiary structure) after adsorption. On the other side is a systematic validation of simulations with experimental data, which can clearly highlight the strong and weak points of the molecular model and the computational technique and thus where and how to improve them. The link between experiments and simulations is becoming stronger and tighter, since it is possible to compute theoretical quantities (such as circular dichroism spectra) that can be directly compared with the corresponding experimental outcomes. The validation of full atomistic models is an unavoidable condition for exploiting the main advantages of coarse-grained models, which must be properly parameterized against more accurate simulations at the molecular level in order to obtain robust and reliable results.

# Nanoparticle–Cellular Membrane Interactions

Molecular modeling can also be employed to investigate the interactions of drug molecules and nanocarriers with lipid bilayers that act as a simplified description of the complex and heterogeneous cellular membrane. Full atomistic MD simulations are the method of choice when small drug molecules are involved, while CG models are the only opportunity if the interest lies in bigger entities like polymer nanoparticles. A detailed molecular model of a cellular membrane, which includes several kinds of lipid molecules as well as transmembrane proteins, is still out of reach, although progress has recently been made in this direction (Ingolfsson et al., 2014), as recently reviewed (Ingolfsson et al., 2016; Marrink et al., 2019). This is due not only to the long time scales needed for achieving converged results but also to the lack of the experimental data for complex membranes (that is, composed of different lipid molecules) needed to parameterize and validate molecular models. For this reason, the cellular membrane is usually represented as a homogeneous bilayer (i.e., which contains only one kind of lipid molecule such as dipalmitoylphosphatidylcholine) or a simple heterogeneous membrane (with two different lipid molecules and sometimes cholesterol). In this framework, molecular modeling can be used to qualitatively understand the impact of nanocarrier formulation and the presence of adsorbed proteins on nonspecific cellular uptake (that is, not mediated by a receptor).

A typical application of MD simulations is the study of the permeation of drug molecules through lipid bilayers, which mimic cellular membranes. Because of the energy barrier related to membrane crossing, the application of enhanced sampling methods becomes unavoidable. Further post-processing by means of an inhomogeneous solubility–diffusion model allows the evaluation of a position-dependent diffusion coefficient through the lipid bilayer as well as the overall permeation coefficient (Dickson et al., 2017). In another study, Bruno et al. (2015) elucidated the binding mechanism of the multiple sclerosis biomarker CSF-114 peptide to membrane using an unbiased atomistic MD approach inspired by the binding freeenergy method, funnel metadynamics (Limongelli et al., 2013).

This approach provides very useful insights, since it allows the relation of the observed permeation of different drug molecules to the specific interactions at the atomic level, such as hydrogen bonds. On the other hand, the use of full atomistic simulations limits the applicability of this analysis to small drug/peptide molecules (up to a few hundreds of Da). The study of nanoparticle permeation with atomic detail would lead to unfeasible or extremely challenging simulations due to the size of the system and the long time scales needed to reach converged results. Because of this, coarse-grained simulations are the method of choice for the study of nanoparticle–cell membrane interactions, as widely discussed in the literature (Rossi and Monticelli, 2014, 2016; Ding and Ma, 2015; Ge and Wang, 2016). For the same reasons, there has been an increase of interest in the use of CG simulations for the study of transmembrane proteins (Lelimousin et al., 2016). In a recent study (Molinaro et al., 2018), a MARTINI model was employed to study the introduction of a membrane protein in biomimetic vesicles (leukosomes) obtained through a microfluidic-based setup. CG simulations allowed the impact of glycosylation, steric hindrance of the protein extracellular domain versus the intracellular domain, and relative to vesicle curvature on protein orientation to be highlighted.

Another limitation is shared by both full atomistic and coarse-grained methods: as has been mentioned, cellular membranes are very heterogeneous environments because of the wide range of lipids involved and the presence of several transmembrane proteins and receptors, and simplified models are needed for affordable simulations. Lipid bilayers made of dioleoylphosphatidylcholine (DOPC) and dipalmitoylphosphatidylcholine (DPPC) are commonly used as cell membrane models thanks to the availability of validated parameters for the force fields. Simulations of bilayers with heterogeneous compositions (that is, composed by many different lipid molecules), which would constitute a more realistic cellular membrane model, are hindered by the lack of experimental data for force field validation (Poger et al., 2016). Transmembrane proteins are not included unless the investigation is focused on the interactions with a specific receptor or on the behavior of such proteins.

In summary, simulations at the molecular level of nanoparticle–cellular membrane interactions are usually performed by means of CG methods and are focused on simplified systems made up of a mimicking lipid bilayer and a small nanoparticle (up to 10–20 nm). The investigation of larger particles, although of potential interest, is still limited by the computational effort required and the difficulty of achieving converged results.

The advantages and disadvantages, for both MD and CG, are summarized in **Table 4**.

In general, the comparison with experimental data is more challenging. Simulation of naked and decorated particles (i.e., with surface functionalization and/or a hard protein corona) can highlight the different interactions with the cellular membrane and can be compared with the expected and the experimental cellular uptake. In this framework, simulations are expected to give those insights at molecular resolution, which cannot be obtained experimentally; this reinforces the need to have previously validated models of protein–particle interactions and model lipid bilayers. Computational efforts are currently focused on parametric simulations, where the influence of particle hydrophilicity/hydrophobicity (including charge), coating (e.g., PEGylation), shape, and size on membrane permeation and induced stresses are qualitatively evaluated.

The examples offered by the literature involve generic nanoparticles with different shapes or functionalization (Yang and Ma, 2010; Ding and Ma, 2012, 2014; Li and Hu, 2014; Li et al., 2014), gold nanoparticles (Lin et al., 2010; Rossi and Monticelli, 2016; Salassi et al., 2017; Lunnoo et al., 2019), and polymer systems (Schulz et al., 2012) such as dendrimers (Rossi and Monticelli, 2014, 2016), polystyrene (Rossi and Monticelli, 2016), and polyelectrolytes (Rossi and Monticelli, 2016).

Ding and Ma (2014) employed dissipative particle dynamics to study the influence of human serum albumin corona (vide supra) around hydrophobic or positively charged nanoparticles on membrane permeation. They found that at physiological pH, the HSA corona promotes particle adhesion on a DPPC lipid bilayer model of a cell membrane thanks to the specific interactions with the protein coating of a 3-nm hydrophobic particle. They also investigated the impact of pH and membrane charge.

Li et al. (2014) studied through a coarse-grained model and dissipative particle dynamics the effect of PEG grafting density (0.2–1.6 chains nm−<sup>2</sup> ) and molecular weight (550– 5000 Da) on the internalization of an 8-nm particle, proposing

TABLE 4 | Advantages and disadvantages for nanoparticle–cellular membrane interactions.


a optimal choice of parameters for maximizing cellular uptake. They also characterized the uptake process in detail, identifying three stages: membrane bending (0 < t < 122 ns), membrane monolayer protruding (122 < t < 750 ns), and equilibrium (t > 750 ns).

Recently, Lunnoo et al. (2019) employed the MARTINI CG model to simulate the cellular uptake of gold nanoparticles. Notably, they employed a more complex mammalian cell model previously proposed by Ingolfsson et al. (2014), which includes 63 different lipid species asymmetrically distributed in the bilayer. This allowed the limitations of simple models to be overcome and the complexity of a more realistic cellular membrane to be accounted for; indeed, they found that neutral 10-nm nanoparticles experienced an endocytic pathway with a DSPC/DSPG model membrane, while they exhibited direct translocation across the more complex model of a mammalian membrane. They also characterized the energy barrier related to membrane crossing by changing the shape and charge density, also taking particle aggregation into account.

Similarly to protein corona simulations, in this framework, molecular modeling must still be considered as a complementary tool to experimental activity and not as an alternative. Although it provides interesting insights, the lack of systematic experimental validation hinders the application of molecular simulations as a predictive tool. It is also necessary to take into account the inherent approximations of coarse-grained models, where some kinds of interactions are poorly accounted for. In addition, there are still some limitations concerning the size of the device; according to examples in the literature, the maximum investigated nanoparticle size is about 20 nm. Simulations of larger devices not only increase the number of beads but also require very long calculations to achieve converged results: the required computational effort is not always feasible. This issue could in principle be overcome by employing, e.g., implicit solvent methods, which further improve computational efficiency by representing the solvent as a continuum (and thus reducing the number of explicit beads in the system) at the price of an additional approximation. The implicit solvent parameterization of the MARTINI force field, called Dry MARTINI, is currently validated only for lipid bilayers, although it has been shown that it can also be used for polymeric systems after a proper re-parameterization (Bochicchio and Pavan, 2017). In general, the use of implicit solvent methods requires an accurate parameterization and validation with experimental data or more detailed simulations at an atomic scale. Currently, only qualitative insights concerning more realistic systems (in terms of particle size) can be obtained through the simulation of smaller devices.

# CONCLUSIONS AND PERSPECTIVES

Simulations at the molecular level, despite the discussed limitations and drawbacks, constitute a powerful tool for improving understanding of the governing phenomena at the nano–bio interface. The intrinsic peculiarities of molecular modeling, which account for the synergistic effects of several factors (particle material, protein adsorption, environmental effects, interactions with cellular membranes, et cetera), can provide some insights that are challenging or impossible obtain experimentally, thanks to the molecular resolution. The increasing availability of computational resources, the development of improved force fields (that are more accurate), algorithm optimization, and theoretical advancements are constantly pushing molecular simulations beyond their limits, slowly overcoming the current issues.

Focusing on the protein corona, the conditio sine qua non for a meaningful simulation is a validated force field, which allows a reasonable description of secondary and tertiary structures to be obtained and a robust sampling of the most relevant conformation. Indeed, discrepancies in the description of protein structural transitions inevitably affect result reliability and the subsequent steps (e.g., the study of the interaction of a proteindecorated particle with a cellular membrane). Descriptive capabilities are known a priori, since they are addressed in detail in several papers and FF reference papers. The development and the improvement of force fields (not only for proteins) are always ongoing, and updates are periodically released and discussed in the scientific literature. This refinement process is currently taking advantage of new state-of-the-art techniques such as machine learning (Debiec et al., 2016).

Molecular dynamics simulations provide detail at an atomic level, but they are limited by the time scale of many phenomena of interest (such as protein folding/unfolding, slow binding/unbinding kinetics), which is beyond that accessible through standard simulations. The development of enhanced sampling methods allows this issue to be alleviated and allows a more comprehensive ensemble of conformations to be obtained. Currently, the extensive application of such methods is still hindered by the size of the molecules under investigation, which cannot exceed, in the case of proteins, a few tens of amino acids in order to obtain reliable and converged results. Further improvements of the method itself and optimization of computational protocols and algorithms could allow the investigation to be focused on larger and more complex proteins.

Coarse-grained models, along with suitable methods to study system dynamics, have emerged as an attractive choice when molecular dynamics simulations are unfeasible because of the time and length scales involved. Indeed, despite the loss of atomic detail, CG models have proved that the fundamental physical/chemical peculiarities lie at the molecular model. However, in order to obtain reliable results, careful parameterization and validation against experimental data still represent essential steps that are not always addressed.

Simulations are mainly focused on inorganic particles (gold, silica) or carbon-based devices (graphene, carbon nanotubes), while there are no relevant examples concerning polymer nanoparticles. This can be attributed to the fact that molecular models of inorganic particles are easier to build given the availability of reliable force field parameters together with their known and well-parametrized structural properties.

Many efforts are also being devoted to the development of more realistic models of cellular membranes, as recently reviewed (Ingolfsson et al., 2016; Marrink et al., 2019). This aspect cannot be underestimated, because the reliability of the results concerning drug or nanocarrier–cell membrane interactions of course requires a robust description of a cell membrane with a suitable level of approximation.

The available force fields provide validated parameters for small sets of lipid molecules (although the number of available compounds increases in every FF update), and it is difficult to validate simulations of heterogeneous membranes (that is, made up of several kinds of different lipid molecules) because of the lack of suitable experimental data. In this regard, a first attempt has been performed by Ingolfsson et al. (2014), who employed a CG MARTINI model to simulate an idealized mammalian plasma membrane, including more than 63 lipid species asymmetrically distributed in the bilayer. Marrink et al. (2019) recently published a comprehensive review that summarizes all the advancements in the field and clearly describes the ultimate goal for comprehensive modeling: the simulation of a membrane with hundreds of different lipids, with a large variety of transmembrane as well as peripherally bound proteins and realistic gradients of metabolites, ions, and pH. Although this "definitive" simulation is still far off, there are in the literature some interesting attempts to model complex systems, such as viral envelopes and mesoscale simulations remodeling eukaryotic cell membranes (Marrink et al., 2019).

In conclusion, simulations at the molecular scale have emerged as a fruitful tool to complement the insights provided by experimental activity and obtain a deeper understanding of the main phenomena behind the observed behavior. Despite their use becoming more and more widespread, there are still some points that need to be addressed in the near future to overcome the current limitations:


#### AUTHOR CONTRIBUTIONS

TC performed the literature research and wrote the first draft of the paper. All authors discussed and approved the contents of the manuscript and contributed to the final version by reading and editing.

### FUNDING

This study is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016, from the CTI (1.1.2018 Innosuisse) under grant agreement Number 19267.1 PFNM-NM and from the FCT Foundation for Science and Technology under the project PROSAFE/0001/2016. VL thanks the Swiss National Science Foundation (Project

# REFERENCES


N. 200021\_163281) and the COST action CA15135 (Multitarget paradigm for innovative ligand identification in the drug discovery process MuTaLig) for the support.


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

Copyright © 2019 Casalini, Limongelli, Schmutz, Som, Jordan, Wick, Borchard and Perale. 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.

# Hazard Assessment of Polymeric Nanobiomaterials for Drug Delivery: What Can We Learn From Literature So Far

#### Sandra Jesus <sup>1</sup> , Mélanie Schmutz <sup>2</sup> , Claudia Som<sup>2</sup> , Gerrit Borchard<sup>3</sup> , Peter Wick <sup>4</sup> and Olga Borges 1,5 \*

*<sup>1</sup> Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal, <sup>2</sup> Laboratory for Technology and Society, Empa Swiss Laboratories for Materials Science and Technology, St. Gallen, Switzerland, <sup>3</sup> School of Pharmaceutical Sciences, University of Geneva, Geneva, Switzerland, <sup>4</sup> Laboratory for Particles-Biology Interactions, Empa Swiss Laboratories for Materials Science and Technology, St. Gallen, Switzerland, <sup>5</sup> Faculty of Pharmacy, University of Coimbra, Coimbra, Portugal*

#### Edited by:

*Michele Iafisco, Italian National Research Council (CNR), Italy*

#### Reviewed by:

*Smilja Markovic, Institute of Technical Sciences (SASA), Serbia Ching-Yun Chen, National Health Research Institutes, Taiwan*

> \*Correspondence: *Olga Borges olga@ci.uc.pt*

#### Specialty section:

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

Received: *02 August 2019* Accepted: *26 September 2019* Published: *23 October 2019*

#### Citation:

*Jesus S, Schmutz M, Som C, Borchard G, Wick P and Borges O (2019) Hazard Assessment of Polymeric Nanobiomaterials for Drug Delivery: What Can We Learn From Literature So Far. Front. Bioeng. Biotechnol. 7:261. doi: 10.3389/fbioe.2019.00261* The physicochemical properties of nanobiomaterials, such as their small size and high surface area ratio, make them attractive, novel drug-carriers, with increased cellular interaction and increased permeation through several biological barriers. However, these same properties hinder any extrapolation of knowledge from the toxicity of their raw material. Though, as suggested by the Safe-by-Design (SbD) concept, the hazard assessment should be the starting point for the formulation development. This may enable us to select the most promising candidates of polymeric nanobiomaterials for safe drug-delivery in an early phase of innovation. Nowadays the majority of reports on polymeric nanomaterials are focused in optimizing the nanocarrier features, such as size, physical stability and drug loading efficacy, and in performing preliminary cytocompatibility testing and proving effectiveness of the drug loaded formulation, using the most diverse cell lines. Toxicological studies exploring the biological effects of the polymeric nanomaterials, particularly regarding immune system interaction are often disregarded. The objective of this review is to illustrate what is known about the biological effects of polymeric nanomaterials and to see if trends in toxicity and general links between physicochemical properties of nanobiomaterials and their effects may be derived. For that, data on chitosan, polylactic acid (PLA), polyhydroxyalkanoate (PHA), poly(lactic-co-glycolic acid) (PLGA) and policaprolactone (PCL) nanomaterials will be evaluated regarding acute and repeated dose toxicity, inflammation, oxidative stress, genotoxicity, toxicity on reproduction and hemocompatibility. We further intend to identify the analytical and biological tests described in the literature used to assess polymeric nanomaterials toxicity, to evaluate and interpret the available results and to expose the obstacles and challenges related to the nanomaterial testing. At the present time, considering all the information collected, the hazard assessment and thus also the SbD of polymeric nanomaterials is still dependent on a case-by-case evaluation. The identified obstacles prevent the identification of toxicity trends and the generation of an assertive toxicity database. In the future, *in vitro* and *in vivo* harmonized toxicity studies using unloaded polymeric nanomaterials, extensively characterized regarding their intrinsic and extrinsic properties should allow to generate such database. Such a database would enable us to apply the SbD approach more efficiently.

Keywords: hazard assessment, exposure assessment, in vivo toxicity, oxidative stress, genotoxicity, toxicity on reproduction, hemocompatibility, polymeric nanobiomaterials

#### INTRODUCTION

Over the last decades, several nanomaterials (NMs) have been developed and studied as promisor drug delivery vehicles and medical devices, including magnetic, metallic, ceramic and polymeric nanomaterials. At present, there is fragile consensus regarding the "nano" definition among different regulatory organizations. In detail, considering medical regulatory authorities, such as the European Medicines Agency (EMA) or the United States Food and Drug Administration (FDA) some considerations can be made. In a reflection paper about nanotechnology-based medicinal products for human use published in 2006, EMA defined nanotechnology as "the production and application of structures, devices and systems by controlling the shape and size of materials at nanometer scale," considering that "the nanometer scale ranges from the atomic level at around 0.2 nm (2 Å) up to around 100 nm" (European Medicines Agency, 2006). On its turn, FDA guidance for considering whether an FDA-regulated product involves the application of nanotechnology (Food Drug Aministration, 2014) refers that it should be considered "the evaluation of materials or end products engineered to exhibit properties or phenomena attributable to dimensions up to 1,000 nm, as a means to screen materials for further examination and to determine whether these materials exhibit properties or phenomena attributable to their dimension(s) and associated with the application of nanotechnology." Therefore, for the context of academic research and to the context of this review the following definition of nanomaterial applies: materials in the size range of 1 nm to 1,000 nm and a function or mode of action based on its nanotechnological properties. In addition, by "nanobiomaterial" we considered NMs intended to interact with biological systems. The application of nanobiomaterials in the medicine field present several advantages as they can (Moritz and Geszke-Moritz, 2015; Banik et al., 2016):


Considering polymeric NMs in particular, they can be assembled in different pharmaceutical nanosystems, such as nanoparticles (NPs), dendrimers, polymeric micelles and drug conjugates (Bhatia, 2016). On its turn, polymeric NPs comprise both vesicular systems (nanocapsules) and matrix systems (nanospheres) (Bhatia, 2016). The polymeric nature of these NMs provides additional advantages that are worth exploring, such as enhanced biocompatibility, biodegradability and low immunogenicity (Egusquiaguirre et al., 2016; Rana and Sharma, 2019).

All considered, most of these advantages are frequently attributed to their distinctive size which contributes to their high surface area to mass ratio, and is also responsible for the different toxicokinetic fate of the NMs (Landsiedel et al., 2012; Boyes et al., 2017). Indeed, small sizes facilitate cell uptake, penetration through endothelial and epithelial cells, interaction with tissues and accumulation in the liver, kidney and spleen (Khan and Shanker, 2015). The increased cellular interaction can have a modulatory effect on the immune system, triggering inflammation, increased susceptibility to infectious diseases, or even to autoimmune diseases or cancer (Kononenko et al., 2015).

The unique physicochemical properties of the NMs restricts the extrapolation of toxicological data from raw materials, and makes it necessary to have specific toxicological studies adequate to the nanoscale (Ge et al., 2011). Moreover, there is a need for specific and optimized methods for NMs toxicity evaluation, since interactions between NMs and current toxicity testing protocols can lead to false positive or false negative results (Khan and Shanker, 2015; Kononenko et al., 2015).

Understanding the toxicokinetics of NMs and their modulation of the immunological system is necessary to implement their Safe-by-Design based on the literature. This is an up-to-date subject, currently widely discussed among the scientific community, but most commonly for metallic NM (Gatto and Bardi, 2018; Kanwal et al., 2019).

Therefore, the objective of this review is to summarize what is known about the toxic effects of polymeric NMs, with special focus on polymeric NPs that could be correlated to human health risks. We intend to identify the analytical and biological tests described in the literature used to assess NMs toxicity and to evaluate and interpret the available results. Furthermore, we intend to understand the obstacles and challenges related to the nanomaterial testing that are still preventing a harmonized regulation on polymeric NMs for drug delivery and biomedical applications.

We started this review by discussing the pillars of human health risk assessment: exposure assessment and hazard assessment. Next, in order to analyze the state of the art about the toxic effects of polymeric NMs, peer reviewed original research articles from the last 10 years were analyzed and discussed,

addressing the following endpoints: (1) in vivo toxicity (acute and repeated-dose), (2) oxidative stress, (3) inflammation, (4) genotoxicity, (5) toxicity on reproduction and (6) hemolysis. Importantly, articles were carefully examined regarding minimal characterization parameters, such as chemical composition, particle size, surface charge and endotoxin contamination (when relevant).

### PILLARS FOR HUMAN HEALTH RISK ASSESSMENT

To perform human health risk assessment of any material is necessary to integrate the exposure assessment with hazard assessment. The first intends to determine routes of exposure and estimate exposure dosages (dose, duration and frequency) while the second intends to characterize the possible hazards (toxic effects) of polymeric NMs when in contact with the human body.

#### Exposure Assessment

Human exposure to polymeric NMs should be considered in the context of intentional nanomedicine applications, and in the context of occupational exposures of workers during the manufacturing processes, testing methods, distribution and handling/administration of polymeric NMs. Moreover, it cannot be disregarded situations where misuse and overuse are easily attained (Sayes et al., 2016). While in nanomedicine exposure scenarios, the administration route, the dose and duration of the exposure are well-defined, occupational exposure can happen through multiple and non-expected routes (**Figure 1**) and result in potentially cumulative levels of exposure and organ accumulation, whose impact in human health might be very different from the one predicted (Sayes et al., 2016). In fact, working with NMs involves challenges different from when working with bulk size materials, since they have increased ability to enter the human body, particularly through the respiratory airways, and to be translocated to the bloodstream and different organs (Yah et al., 2012). The lack of testing methods to detect and quantify the unintentional absorbed cumulative doses of these materials in the organism is currently, one of the main difficulties for designing predictive toxicological assays for occupational exposures. Therefore, exposure modeling arises as one alternative to allow occupational risk assessment. In the context of the FP7 NanoReg project a number of risk assessment tools for manufactured NMs, such as the CB NanoTool, the Nanosafer, and the Stoffenmanager-Nano have been examined and a new two-box nano specific exposure model (I-Nano) has been implemented (Jiménez et al., 2016). However, the need to rely on detailed input data (rate of particulate release from the source as well as the particle size distribution) which is not always available and its only application to inhalable exposures are some of the limitations present (Jiménez et al., 2016).

In the main, the NM routes of administration and exposure include respiratory, oral, ocular, dermal, and parenteral (injectable and implantable), each route presenting its own biodistribution pattern, resulting in different effects on human health. Indeed, the same composition, size and surface charge of the polymeric NM, might produce a different effect only by changing the exposure route (Sharma et al., 2016; Boyes et al., 2017). Importantly, it cannot be disregarded that the characteristics of the individual exposed, such as its age and health status, might also influence the NMs effect (Boyes et al., 2017). **Table 1** below summarizes the most common administration/exposure routes and the most important characteristics of NMs related to each one.

#### Hazard Assessment

The NMs toxic effects might occur in the administration site or they can result from the nano-sized materials ability to cross biological barriers (mucosal barriers, air-blood


TABLE 1 | Common routes of administration/exposure: important considerations relating nanomaterials characteristics and the various routes of exposure (Agrawal et al., 2014; Blanco et al., 2015; Date et al., 2016; Palmer and DeLouise, 2016; Boyes et al., 2017).

barrier, blood-brain barrier, placenta barrier) reaching cells and tissues that are generally protected from bulk size materials (Buzea et al., 2007; Ai et al., 2011). This improved penetration of nanoparticles may increase the toxicity, but at the same time be advantageous in order to improve current therapies.

The uncertainties about using NMs for drug delivery and other biomedical applications result mainly from particle size reduction which is linked to increased reactivity and augmented toxicity (Ai et al., 2011). Nonetheless, several other properties can contribute to the effects of these nano-sized delivery systems, such as chemical composition, hydrophobicity/hydrophilicity, surface charge or shape. In the literature, there is a significant amount of data relating physicochemical features of NMs with cellular interaction, biodistribution, cytotoxicity and immune system activation, as reviewed elsewhere (Fröhlich, 2012; Ma et al., 2013; Salatin et al., 2015; Hoshyar et al., 2016; Jindal, 2017; Zhang et al., 2017). Nevertheless, general conclusions indicating toxicity trends for a specific nanoparticle physicochemical property, are limited to cautious hypotheses, only verified in particular scenarios (i.e., depending on the administration route, dose metrics, etc.). A review published in 2014 by Gatoo et al. (2014) discusses the correlation between the physicochemical properties of NMs and its toxicity. Briefly, smaller particles are often correlated with a higher toxicity, due to their increasing ability to cross biological barriers and reach different organs without being recognized by the reticuloendothelial system (RES) (Gatoo et al., 2014). Other characteristics, such as the nonspherical shape or the positive surface charge are also believed to contribute to an increased toxicity of NMs (Gatoo et al., 2014). Importantly, most of these conclusions are based on studies using inorganic NMs. Since chemical composition is one of the variables affecting the NMs toxicity, different behaviors can derive from the polymer composition and therefore, extensive extrapolations among all classes of NMs should be avoided. Moreover, most toxicity trends consider one characteristic at a time, but it is important to consider a holistic approach of the NM: all physicochemical characteristics are interconnected and together will influence its toxicological profile.

The key aspect to test polymeric NM for human toxic effects is the simulation of realistic human exposures. Those scenarios are difficult to simulate mainly due to: (1) the difficulty on transposing accurately human effective doses to in vitro settings; and (2) the difficulty to have complex in vitro systems, based on human cells or primary cell lines, that mimic the physiological complexity of the human body and its interaction with the materials (Sharma et al., 2016). Actually, most of the results of the application of in vitro studies to polymeric NMs might not reflect the realistic exposures, since the tests are performed at much higher concentrations than those that can be achieved in in vivo experiments (Landsiedel et al., 2017). Moreover, in vitro testing commonly use mass-based exposure metrics, which is believed to be a limiting factor, as particle number, surface areas and the formed agglomerates in suspension greatly influence the effective concentration delivered to cells (Hinderliter et al., 2010; DeLoid et al., 2014).

The intrinsic and distinctive characteristics inherent to the nanoscale dimension, might interfere with reagents and detection methods of in vitro assays recommended for bulk materials (Dobrovolskaia et al., 2009). For instance, NMs may bind to the marker enzyme lactate dehydrogenase (LDH) or they may interact with dyes and dye products, such as neutral red and the tetrazolium salt (MTT) (Landsiedel et al., 2017). On the other hand, polymeric NMs also go through modifications when in contact with biological matrices, such as: bio-corona formation, aggregation/agglomeration, dissolution, generation of new nanosized particles (as a result of ionic salvation or degradation of surface coatings) (Sharma et al., 2016). These transformations of the NM can interfere with its toxicological effect, and most of the times are not considered during in vitro testing. Lastly, the selection of relevant positive and negative nano-sized controls is most of the times ignored, mainly because there is no clear knowledge-base on the toxicity (and especially immunotoxicity) of the different NMs (Dobrovolskaia and McNeil, 2013).

It is widely accepted that in vitro assays based on cell lines are an inexpensive and direct method to evaluate nanoparticle related toxicity in target tissues. However, results significantly depend on the chosen cell line (commonly immortalized cancer cells), incubation time, cell culture media or cell culture supplementation (Lorscheidt and Lamprecht, 2016). For instance, cell culture media supplementation with serum is highly likely to induce a protein corona in the surface of positively charged nanoparticles, changing its size and zeta potential, and therefore modifying the nanoparticle-cell interaction and uptake, and ultimately its biological effect (Khang et al., 2014; Lorscheidt and Lamprecht, 2016).

Overall, despite the great effort in developing high-throughput in vitro assays, there is still much variables to accurately mimic real exposure scenarios, and the results are often in disagreement with those of animal studies (DeLoid et al., 2014). Even so, nanotechnology laboratories are still searching for the best in vitro assays to replace in vivo testing and predict real exposure scenarios. This issue has been extensively discussed by Dobrovolskaia and McNeil (2013).

The urge to replace in vivo testing of toxicity, is motivated by the high costs and relatively low throughput of the assays, the inter-species variability particularly on the structure and function of the immune system, the low sensitivity of standard in vivo toxicity tests toward mild immunomodulation reactions, and most importantly, the ethical concerns about animal use (Dobrovolskaia and McNeil, 2013).

Altogether, it is widely accepted that efficient and costeffective toxicological testing is required (DeLoid et al., 2014). For that reason, international organizations including OECD and ISO have developed official papers with considering the NMs properties and their influence on testing methods (Sharma et al., 2016; Dusinska et al., 2017).

In 2006, the OECD started a nanosafety programme overseen a Working Party on Manufactured Nanomaterials (WPMN), which aims to promote international cooperation on the human health and environmental safety of manufactured NMs, and involves the safety testing and risk assessment of manufactured NMs. Over the years they have published numerous reports and some test guidelines which are published in the OECD Series on the Safety of Manufactured Nanomaterials to provide up-to-date information on the OECD activities in this area (OECD<sup>1</sup> ).

<sup>1</sup>OECD. Available online at: http://www.oecd.org/science/nanosafety/ publications-series-safety-manufactured-nanomaterials.htm (accessed June 15, 2018).

In 2005, the Technical Committee ISO/TC 229 was created. It aims at the standardization in the field of nanotechnologies. The specific tasks of this committee include developing standards for terminology and nomenclature, metrology and instrumentation, test methodologies, modeling and simulations, and sciencebased health, safety, and environmental practices (Behzadi et al., 2014). Over the years, the committee has published several standards, from which we can highlight the recent ISO/TS 19006:2016 [Nanotechnologies-5-(and 6)-Chloromethyl-2 ′ ,7′ -Dichloro-dihydrofluorescein diacetate (CM-H2DCF-DA) assay for evaluating nanoparticle-induced intracellular reactive oxygen species (ROS) production in RAW 264.7 macrophage cell line] and the ISO 19007:2018 (Nanotechnologies–in vitro MTS assay for measuring the cytotoxic effect of nanoparticles), discussed below (Bazile et al., 1995; Behzadi et al., 2017). In addition to the specific standards generated by this committee, in 2017, the part 22—Guidance on nanomaterials, was implemented in ISO 10993 (Biological evaluation of medical devices) (Barratt, 2000). Although this technical report represents the current technical knowledge related to NMs for medical devices it does not contain detailed testing protocols.

An important contribution to this field is being given by the US National Cancer Institute Nanotechnology Characterization Laboratory, whose main objective is to facilitate the development and translation of nanoscale particles and devices for clinical applications. In fact, they have described several protocols for in vitro characterization as well as for in vivo, and for the physicochemical characterization of NMs (Assay Cascade Protocols—https://ncl.cancer.gov/resources/assaycascade-protocols). In parallel, the European Nanomedicine Characterization Laboratory (EUNCL) is also developing standard operating procedures (SOPs) to allow the physical, chemical, in vitro and in vivo testing of nanobiomaterials (http:// www.euncl.eu/).

# HAZARD CHARACTERIZATION OF POLYMERIC NANOMATERIALS— LITERATURE REVIEW

NMs toxicity should be evaluated by in vivo and in vitro assays considering its effect in the host physiological and immunological integrity (Yildirimer et al., 2011). Most of in vitro assays available for testing a NM toxicological effects are focused on the molecular mechanisms underlying toxicity (i.e., oxidative stress generation and inflammation), while in vivo assays, particularly acute and repeated dose toxicity assays assess the effects on vital organ functions [i.e., biomarkers of liver function, such as aspartate aminotransferase (AST) and alanine aminotransferase (ALT)].

**Table 2** summarizes the studies collected from the literature of the last 10 years, assessing the toxicity of polymeric NMs for the endpoints studied. The polymers considered for analysis were chitosan, polylactic acid (PLA), polyhydroxyalkanoate (PHA), poly(lactic-co-glycolic acid) (PLGA) and policaprolactone (PCL). From the table systematization we can highlight three main issues: (1) chitosan based NPs are the most studied polymeric

\**Genotoxicity*

 *includes Mutagenicity*

 *and* 

*Carcinogenicity.*

NMs followed by PLGA based NPs; (2) the different colors illustrating the generation or absence of effect for each endpoint according to the different studies, reflects the inconsistency in the results found for the same type of NM; (3) No data on PHA based NMs is available regarding those endpoints. The inconsistent results must be carefully analyzed because in fact they may be complementary results, as the NM characteristics, their concentrations, the cellular and animal models used and even the experimental methodology are significantly different among authors. Therefore, in the next sub-chapters each endpoint and respective studies will be discussed in detail in an attempt to scrutiny possible toxicity trends for polymeric NMs. To note, over the following discussion, the effect of some other polymers, such as alginate, polyethylene glycol (PEG), pluronic and polyvinyl alcohol (PVA) are addressed as they are often used as surface coatings and blends in chitosan, PLGA, PLA and PCL based nanomaterials.

#### In vivo Toxicity Studies

To study the toxicity of the NMs and to identify possible risks to the human health, researchers perform in vivo tests in animals (most time non-primates) to evaluate acute and repeated-dose (subacute, sub-chronic or chronic) toxicity. These studies, although highly valuable to understand the adsorption, distribution, metabolism and excretion (ADME) of the NMs as well as the immune system interactions, should be limited to a minimum according to the 3Rs strategy (replacement, reduction and refinement) (Oostingh et al., 2011; Dusinska et al., 2017). To note, in 2018, OECD guidelines for the testing of chemicals were adapted to accommodate the testing of NMs (OECD, 2018b,c).

As illustrated in **Table 3**, the available research articles testing in vivo the toxicity of NMs are characterized by a great variability between the rodent's species (or other animals, such as carps) used in the assays, the number of days (for the repeated-dose toxicity studies) and even for the endpoints that are analyzed. Some of the most reported endpoints are the clinical appearance of the animal, clinical signs of infection, hematological parameters, serum hemoglobin levels and albumin/globulin ratio, organ weights, and enhanced histopathology evaluation different organs (Dusinska et al., 2017).

As already stated, chitosan NMs are the most studied polymeric NMs regarding toxicity. Several studies were found in the literature evaluating the toxicity of blend chitosan NPs upon repeated oral administrations. Despite the great heterogeneity among the used NPs (chitosan/alginate NPs, chitosan/glutamic acid NPs, oleoyl-carboxy methyl chitosan NPs, chitosan coated PLGA NPs and α-tocopherol succinate-g-carboxymethyl chitosan NPs), the animal models (Wistar and Sprague Dawley rats, ICR mice and Carps) and the dosing schedules (7–19 days), all revealed no in vivo toxicity (Sonaje et al., 2009; Liu et al., 2013; Jena and Sangamwar, 2016; Aluani et al., 2017; Maity et al., 2017; Radwan et al., 2017b; Sharma et al., 2017). Moreover, the conclusion of no toxicity was based on different evaluated parameters for each study, except for the histopathological analysis, which was performed in all studies (generally liver and intestine histopathology with no signs of tissue damage). Among these studies, only Sonaje et al. (2009), Maity et al. (2017), and Radwan et al. (2017b) have evaluated biochemical parameters in blood, and in common have tested serum alanine transaminase (ALT), alkaline phosphatase (ALP) and aspartate transaminase (AST) activities, and their results were in agreement (no changes in comparison to the control group). Moreover, chitosan based NPs lack of oral toxicity was also reported for single dose administrations (Mukhopadhyay et al., 2015; Leng et al., 2018). Therefore, considering these reports, we may hypothesize that chitosan NPs (as well as bulk chitosan Chang et al., 2014) do not present oral toxicity. On the other hand, although only 2 reports were found testing chitosan NPs toxicity through the injectable route (Yuan et al., 2015; Shan et al., 2017), a dose dependent toxicity was found, even though chitosan and chitosan NPs appear to be hemocompatible in some hemolysis assays (Fernandes et al., 2010; Lü et al., 2011; Wang et al., 2014; Kumar et al., 2017; Leng et al., 2018).

On its turn, PLGA NPs also exhibited no toxicity on repeated oral administration studies (Moraes Moreira Carraro et al., 2017; Sharma et al., 2017), as well as on the majority of intravenous (i.v.) administration studies (VasanthaKumar et al., 2014; Fasehee et al., 2016; Radwan et al., 2017a). Only one article described some toxicity when using danorubicin loaded PEG-PLL-PLGA NPs (Guo et al., 2015). Unfortunately, the formulations in those reports were loaded with the active drug and no information was given on blank NPs. Therefore, not only the effects might be associated with the drugs (rather than the NPs polymers or characteristics), but also no comparison on the dose of the NPs administered can be made between articles, as they only refer to the equivalent amount of drug administered. Similarly (Li et al., 2014), tested two mPEG-PLA NPs (with different copolymerization degrees) loaded with paclitaxel in beagle dogs by i.v. administration in the foreleg. Despite the results had revealed differences between the NPs, being the ones with the 50/50 ratio mPEG:PLA more toxic than the ones with the 40/60, no experiments were made with unloaded NPs, restricting the extrapolation of data.

# Oxidative Stress

Reactive oxygen species (ROS) are produced during cellular metabolism in the forms of hydrogen peroxide (H2O2), superoxide anion (O2−•) and hydroxyl (•OH) radicals (Ngo and Kim, 2014; Lorscheidt and Lamprecht, 2016). Besides its role in cell signaling and regulation, excessive oxidative stress can induce oxidative damage to cells through lipid peroxidation, DNA disruption, interference with signaling functions, gene transcription modulation and inadvertent enzyme activation, causing several health disorders, such as hypertensive, cardiovascular, inflammatory, aging, diabetes mellitus, and neurodegenerative and cancer diseases (Sharifi et al., 2012; Ngo and Kim, 2014; Lorscheidt and Lamprecht, 2016).

The most used probe to access ROS is the H2O<sup>2</sup> specific 2′ ,7′ dichlorodihydrofluorescein diacetate (H2DCFDA or DCFH-DA), which diffuses freely through the cell membrane and is hydrolyzed inside the cells into H2DCF carboxylate anion form, which is in its turn non-permeable (Kalyanaraman et al., 2012; Oparka et al., 2016). Then, H2DCF is oxidized and results in the formation of the fluorescent product (DCF), which is excited at 495 nm and emits at 520 nm (Kalyanaraman et al.,

#### TABLE 3 | Review of original articles assessing*in vivo*the toxicity of polymeric nanoparticles.


*(Continued)*

Polymeric



*(Continued)* Polymeric



TABLE 3 | Continued

*(Continued)* Polymeric


References

Yuan et al., 2015

Aluani et al., 2017

Radwan et al.,

2017b

Sonaje et al., 2009

Jena and

Sangamwar, 2016

 decreased

 of GFAP

> altered

 or

 neuron apoptosis and

 cerebellum

 response in the frontal

 weight or relative liver

 in the condition of the

 activity

 alteration in animal's

 and weight gain

 or weight loss

 changes in liver, kidney

 *study of insulin-loaded*

 changes in the liver,

 segments

 *(5.5 mg/kg)*

*pharmacokinetic*

*nanoparticles*

*The dose (100 mg/kg) was 18 times*

 No mortality

No pathological

*higher than the dose they used in the*

Normal weight gain

Normal red blood cells morphology

kidney, and intestine

 parameters

 for the 10:1

Polymeric

Nanobiomaterials: Hazard Assessment

α-tocopherol

succinate-grafted

low molecular

22 kDa

weight chitosan: 114–187 nm *In vivo* exposure

(14 days)

−20 to −22 mV

carboxymethyl

chitosan polymeric

micelles


Sprague Dawley

rats

 Model

Sprague-Dawley

Administration

Intravenous

route Dose/ concentration

range

 3, 10, and 30 Results

Body weight of rats remarkably

TABLE 3 | Continued

Nanomaterial

Tween 80 modified  Polymer characterization

Chitosan (100 251 nm

Nanomaterial

 characterization

Testing method

exposure

*In vivo*

*(Continued)*

Oral

 500 mg/kg b.w.



TABLE


*(Continued)*




Polymeric

Nanobiomaterials: Hazard Assessment

*(Continued)*


TABLE 3 | Continued


*aDD, deacetylation degree. <sup>b</sup>M/G,* β*-D-mannuronic acid/*α*-L-guluronic acid.cRatio alg:chi:ins. <sup>d</sup>Mw, molecular weight number. <sup>e</sup>na, not available. fMv, viscosity molecular weight. gRatio chi:alg.*

2012; Oparka et al., 2016). Using this probe, the intracellular signal can be monitored by several techniques, such as confocal microscopy and flow cytometry (Kalyanaraman et al., 2012). During the H2DCF oxidation, there is a formation of a superoxide radical that can stimulate the auto-amplification of the DCF signal (Oparka et al., 2016). On the other hand, DCF is cell permeable, which means it leaks out of cells over time and can induce measurement errors depending on the analysis time (Lorscheidt and Lamprecht, 2016). A variant of the DCFH-DA probe is the 5-(and 6)-chloromethyl-derivative, that leads to the formation of fluorescent CM-DCF, which displays a lower passive leakage from the cell (Oparka et al., 2016). Alternatively, the fluorescence read-out can also be performed using a fluorescence microplate reader and in this situations errors can result from nanoparticle quenching effect over the DCF fluorescence (Aranda et al., 2013).

Free radical production is the highest in macrophages (Singh and Ramarao, 2013) which is in line with the protocol suggested in ISO/TS 19006:2016-Nanotechnologies-5-(and 6)-Chloromethyl-2′ ,7′ -Dichloro-dihydrofluorescein diacetate (CM-H2DCF-DA) assay for evaluating nanoparticle-induced intracellular reactive oxygen species (ROS) production in RAW 264.7 macrophage cell line. Nonetheless, according to this ISO, other cell lines similar to RAW 264.7 (BEAS-2B, RLE-6TN, HEPA-1, HMEC and A10) can be used with due validations. In this technical specification, the protocol was validated for conducting the assay in 24 well-plates, for 6 and 24 h incubation with the NPs and controls, and 30 min incubation with the probe before flow cytometry analysis. To note, the recommendation is the use of Sin-1 as positive control (maximum ROS production due to cell death) and polystyrene NPs as negative control.

As it is possible to observe from **Table 4**, most studies reported in the literature do not use RAW 264.7 cells, neither do they employ 6 and 24 h incubation.

In detail, Grabowski et al. found a transient production of ROS with chitosan stabilized PLGA NPs in THP-1 cells (Grabowski et al., 2015), Sharma et al. verified an increased oxidative effect of oleanolic acid when delivered by chitosan coated PLGA NPs in MDAMB-231 cells (Sharma et al., 2017), Sarangapani et al. found an increase in ROS production in BCL2(AAA) Jurkat cells with chitosan NPs (Sarangapani et al., 2018) and Gao et al. found an increase in ROS production in zebrafish embryos incubated with chitosan NPs (Hu et al., 2011). In contrast, Bor et al. found a reduction in ROS production with plasmid loaded chitosan NPs and chitosan NPs in Hela, THP-1 and MDAMB-231 cells (Bor et al., 2016). These inconsistent results, obtained with different chitosan based nanomaterials, different cellular models and concentrations do not allow for a straightforward interpretation of the oxidative effect of nanoscale chitosan. Among these articles, only Sarangapani et al. compared the activity of chitosan NPs with bulk chitosan (at the same concentrations) and verified a similar but lower concentration dependent effect for the polymer (Sarangapani et al., 2018). Also, it is important to note, that the tested concentrations (10– 50µg/mL), caused increasing cell death as verified by the MTT assay, and therefore, the oxidative stress was the mechanism identified as responsible for cellular toxicity. In contrast, Bor et al. verified that chitosan NPs reduced ROS production in several cell lines (also tumor derived cells), but they used a concentration that did not cause cell death (Bor et al., 2016). Therefore, although at first sight the results are conflicting, they cannot be directly compared, but we can hypothesize that chitosan NPs might influence ROS production in a concentration dependent manner. One of the widely reported characteristics of bulk chitosan is its anti-oxidant activity, attributed to its scavenging activity against several radicals, such as hydroxyl (•OH), superoxide anion (O•− 2 ), 1,1-diphenyl-2-picryl-hydrazy (DPPH) and alkyl (Ngo and Kim, 2014). This scavenging activity, has been widely demonstrated by cell-free in vitro assays (Je et al., 2004; Yen et al., 2008; Ngo and Kim, 2014). In fact, in the article discussed before (Sarangapani et al., 2018), although reporting that chitosan and chitosan NPs increased ROS production in BCL2(AAA) Jurkat cells, they also verified that the same concentrations increased free radical scavenging activity using chemical assays. Therefore, some compounds may demonstrate chemically some antioxidant activity, which is not verified at cellular and physiological level (Lü et al., 2010).

Regarding bare PLGA NPs its effect on ROS production was documented by 3 authors Platel, Singh, and Granbowski (Singh and Ramarao, 2013; Grabowski et al., 2015; Platel et al., 2016) all using different cellular models. Nevertheless, Platel tested only one low concentration of PLGA NPs (40µg/mL) and found no effect on ROS production (Platel et al., 2016), while the other 2 authors found an increase in ROS production that was dose dependent (Singh and Ramarao, 2013; Grabowski et al., 2015). Curiously, both tested 1 mg/mL, but Singh et al. reported that this concentration quenched the fluorescence of the probe, therefore interfering with the results (Singh and Ramarao, 2013). On its turn, Grabowski et al. found that at the concentration of 1 mg/mL only a transient production of ROS was verified at 5 min after the incubation with PLGA NPs, and at longer incubation times, no significant ROS increase was verified (Grabowski et al., 2015). Although the authors do not explore this achievement, we could hypothesize that a similar interference as reported by Singh and Ramarao might be occurring.

Overall, not only PLGA NPs, but in general the polyester NPs appear to induce ROS production in a concentration dependent manner. Other studies confirm this effect for concentrations above 300µg/mL (Singh and Ramarao, 2013; Legaz et al., 2016; Da Silva et al., 2019). Nevertheless, this conclusion has reservations since for instance, Da Silva et al. tested two different PLA NPs, and only one of these induced ROS production.

#### Inflammation

Presently, inflammation is acknowledged as a mechanism of immune defense and repair, in addition to its widely accepted role in passive cell injury and cell death (Wallach et al., 2013; Khanna et al., 2015). Interestingly, several molecules are associated with inflammation and cell death. For instance TNF-α, IL-1β, IL-6, IFN-γ, IL-17, IL-8, IL-2, GM-CSF, TGF-β, and IL-12 are examples of pro-inflammatory mediators frequently evaluated in the context of cellular toxicity induced by nanomaterials (Khanna et al., 2015; Lorscheidt and Lamprecht, 2016).

#### TABLE 4 | Review of original articles assessing oxidative stress induction by polymeric nanoparticles.


*(Continued)*

Polymeric


Polymeric

References

TABLE 4 | Continued

Nanomaterial

Polymer characterization

−12.7 mV

−12.7 mV

243 nm

c (intrinsic

viscosity 0.72 g/dl)

viscosity 0.62 g/dl)

85:15

Nanomaterial

 characterization

Testing method  Cellular model  Dose/ concentration

range

Results

concentration

production

interfered with ROS assay due

A further increase in NPs

to fluorescence  to 1,000 µg/ ml

 quenching

Observations


*(Continued)*



*aDD, deacetylation degree.*

*<sup>b</sup>DMEM, Dulbecco's Modified Eagle Medium.*

*cPLGA lactic to glycolic acid.*

Regarding the methodologies, the enzyme-linked immunosorbent assay (ELISA) is widely applied as a simple mean to perform a qualitative and quantitative analysis of cytokines, chemokines, growth factors and immunoglobulins, with a spectrophotometric readout (Lorscheidt and Lamprecht, 2016). In this assay, the pro- and anti-inflammatory mediators are released into cell supernatant, which is collected and then analyzed. Therefore, the release of cytokines or other molecules by cells during the incubation with nanoparticles can be underestimated due to the nanoparticles ability to adsorb biomolecules at its surface (Lorscheidt and Lamprecht, 2016). Kroll et al. (2012) tested the potential interference of 4 types of engineered nanoparticles on IL-8 secretion, and verified that a specific pre-dispersion of TiO<sup>2</sup> nanoparticles was able to reduce the measurable levels of the cytokine, under the assay conditions. Similarly, Guadagnini et al. (2013a), tested 4 types of nanoparticles in acellular conditions and verified that TiO2, SiO2, and Fe3O<sup>4</sup> NPs decreased the cytokines levels due to surface adsorption. In the same experiment, PLGA-PEO NPs induced an apparent increase in GM-CSF levels, which the authors believe may be due to the stabilization of the peptides, their protection from proteolysis or by avoiding the interaction of this cytokine with the plastic of the culture plates (Guadagnini et al., 2013a). Although most of the reported interferences are for inorganic nanoparticles, these are good examples that can be overlooked when performing ELISA in cell supernatants previously incubated with polymeric nanoparticles. When studying pro- and anti-inflammatory molecules release due to NPs stimulation, it can be useful to previously study the adsorption or interaction of the NPs with the molecules (i.e., cytokine standards) in acellular conditions.

Alternatively, instead of measuring cell secreted pro- and antiinflammatory molecules by ELISA, the mRNA levels inside the cell can be measured with RT-qPCR (Real-Time quantitative Polymerase Chain Reaction) or the intracellular levels of the cytokines can be measured by flow cytometry analysis using specific antibodies fluorescently labeled (Lorscheidt and Lamprecht, 2016). In the first alternative, however, an increase of mRNA expression does not necessarily lead to an increase of protein secretion (Guadagnini et al., 2013a).

Lastly, besides the masking/enhancing effect of NPs, the presence of contaminants, such as endotoxins can induce itself increased levels of pro-inflammatory molecules in cells (Oostingh et al., 2011). Endotoxins, commonly referred to as lipopolysaccharide (LPS), are present in the outer cell membrane of Gram negative bacteria and are released during multiple processes, such as cell death, growth and division (Magalhaes et al., 2007; Lieder et al., 2013). Therefore, due to the bacteria ability to growth and adapt in several environments, LPS is easily found in numerous media, including poor nutrient media (water, saline and buffers) and its removal is a struggle since it is highly resistant to extreme temperatures pHs (Magalhaes et al., 2007). LPS is comprised by a O-antigen region, a hydrophilic core oligosaccharide and a hydrophobic Lipid A (LipA) (Davydova et al., 2000; Magalhaes et al., 2007; Steimle et al., 2016). The lipid A structure, highly conserved, differs among bacterial species, and determines the molecule immunogenicity (Steimle et al., 2016). On the whole, LPS is a pathogen associated molecular pattern (PAMP), which is recognized and activates the mammalian innate immune system, leading for instance to cellular release of pro-inflammatory cytokines and free radicals, particularly by monocytes and macrophages (Yermak et al., 2006; Lieder et al., 2013; Steimle et al., 2016). Consequently, in vitro testing of LPS contaminated polymeric NMs might generate misleading results and false assumptions of bioactivity or toxicity, ultimately affecting the evaluation of possible human health effects (Lieder et al., 2013).

**Table 5** summarizes the results found in the literature for polymeric NPs stimulation of cytokines.

For chitosan NPs, it is interesting to notice that one author referred chitosan NPs induced several cytokines in BMDCs (Koppolu and Zaharoff, 2013), while other did not (Han et al., 2016). Nevertheless, in both papers, no endotoxin contamination was assessed, no concentrations of NPs were given and the chitosan polymers and NPs characteristics were not the same. Furthermore, it must be considered that cytokine secretion highly depends on the cellular model under study. Indeed, Koppolu and Zaharoff, upon stimulation with chitosan NPs, reported the production of IL-1β in BMDCs and the absence of the same cytokine in RAW 264.7 (Koppolu and Zaharoff, 2013).

The fact that no endotoxin control was made in both papers can rise several questions, mainly in the results that suggest a positive stimulation of chitosan NPs. Chitosan has a cationic charge, resultant from the N-acetyl group removal during chitin deacetylation. This positive charge, mediates for instance the electrostatic interactions with cargo molecules, allowing high loading efficacies, but it also enables chitosan interactions with the negatively charged phosphate, pyrophosphate, and carboxylic groups of LPS (Davydova et al., 2000). Actually, chitosan has been used as a selective filtration membrane for endotoxin removal due to these extensive interactions (Machado et al., 2006; Lieder et al., 2013).

But not only chitosan should be evaluated regarding endotoxin contamination. For instance, Grabowski et al. have published two reports, comparing the inflammatory ability of different PLGA NPs based on the in vitro assessment of cytokines, such as IL-6, TNF-α, IL-8 and MCP-1 (Grabowski et al., 2015, 2016). The differences among PLGA NPs resulted from the inclusion of chitosan, PVA and P68 in order to obtain, positive, neutral and negatively charged particles. In one of the reports the authors do not evaluate or discuss the presence of endotoxin contamination in the formulations (Grabowski et al., 2015). Nonetheless, in the other report, using the same methods and polymers, the authors mentioned that all formulations presented 0.1 to 0.3 EU/mL of LPS depending on the concentration used (Grabowski et al., 2016). In both reports, this information was imperative, since the authors tested IL-8, IL-6 and TNFα, cytokines whose production is induced by LPS (Agarwal et al., 1995; Grabowski et al., 2016). Therefore, despite their conclusions, as illustrated in **Table 5** (Grabowski et al., 2015, 2016), and despite the authors attribute the observed effects to the nanoparticulate form of the formulations, the effect of LPS contamination might be interfering with the results. A simple control that could be adopted in this situation, was TABLE 5 |Review of original articles assessing inflammatory cytokines induced by polymeric nanoparticles in different cells.


Jesus et al.

Polymeric

Nanobiomaterials: Hazard Assessment

*(Continued)*


TABLE 5 | Continued

Nanomaterial


Nanomaterial

Testing method  Cellular model

 Polymer


 Dose/ concentration

> range

Results Endotoxin

contamination

 References

*(Continued)*

Polymeric



*aInferred results from the graphs. The authors do not show or discuss the comparison with non-treated cells.*

*<sup>b</sup>Only statistically significant increases were considered in the results.*

*cAccording to the authors, IL-6 levels were not statically different from the control but neither were LPS levels. Considering this, chitosan stabilized PLGA NPs induced IL-6 levels similar to LPS.* to use the LPS concentration the authors quantified in the formulations, incubate with the cell and assess the cytokine secretion. In these articles, the relationship between the 0.1–0.3 EU/ml of contamination and the 0.1–10µg/mL of LPS as control was not given, and therefore, no further conclusions could be drawn regarding the effect of the LPS contamination in the formulations. Another relevant aspect to highlight, is the fact that nanoparticles, particularly polymeric nanoparticles interfere with most endotoxin quantification assays. This fact was denoted by the authors of these reports, who overcame the interference, by centrifuging the formulations and measuring the contamination in the supernatant (Grabowski et al., 2016). Unfortunately, due to what was discussed previously, the polymers, and particularly the positively charged, might adsorb the LPS through electrostatic interactions, which means the quantification on the supernatant can be underestimated. Overall, in this example, the conclusions about the mild inflammatory ability of PLGA and PLGA stabilized NPs should be extrapolated with caution, since the use of endotoxin free materials, or the presence of endotoxin inhibitor (i.e., polymycin B) might generate different results.

# Genotoxicity

Genotoxicity describes the capacity of the compounds to affect the DNA structure or the cellular apparatus and topoisomerases, modifying the genome fidelity (Słoczynska et al., 2014). Genotoxic effects are not always related with mutations but they can have serious implications for risks of cancer or chronic/heritable diseases (Słoczynska et al., 2014; Lorscheidt and Lamprecht, 2016; Dusinska et al., 2017).

NMs can cause damage to cell's DNA through direct and indirect interactions (Magdolenova et al., 2013; Lorscheidt and Lamprecht, 2016; Dusinska et al., 2017). In fact, upon cellular uptake, NMs might reach the nucleus and contact with cell genetic material, leading to physical or chemical alterations (Magdolenova et al., 2013; Lorscheidt and Lamprecht, 2016; Dusinska et al., 2017). Importantly, this direct interaction is limited by the particle size. Particles ranging between 8 and 10 nm of diameter may reach the nuclear compartment through nuclear pores, whether 15– 60 nm particles will only access the nucleus during cellular division when the nuclear wall is disrupted (Barillet et al., 2010). However, indirect interactions have a greater significance for genotoxicity, since several biomolecules involved in normal gene function (i.e., DNA repair) and cell division (i.e., DNA transcription and replication) can interact with even larger NMs, altering its function and consequently leading to DNA injury or chromosome malformation (Lorscheidt and Lamprecht, 2016; Dusinska et al., 2017). For instance, oxidative stress is a key mechanism by which NMs can cause DNA injury (Dusinska et al., 2017). Therefore, data showing non-cytotoxic increase of ROS should imply genotoxicity studies to assess the degree of damage caused by the oxidative stress (Lorscheidt and Lamprecht, 2016).

Several assays are described in the literature for genotoxicity assessment and include in vitro and in vivo approaches. In vitro assays are commonly performed in cell lines, such as the mouse lymphoma L5178Y TK+/<sup>−</sup> 3.7.2C cells, the TK6 human lymphoblastoid cells and rodent fibroblastic cell lines (CHL-IU, CHO and V79 cells) (Lorge et al., 2016). Regarding in vivo studies, the bacterial reverse mutation test (AMES test) is the most commonly used initial screening performed. Also, the Allium cepa model, allows for a simple and cost-effective assay where DNA damage is assessed after the roots of the plant grow in direct contact with the substance of interest (Bosio and Laughinghouse IV, 2012). Alternatively, other in vivo studies comprise the use of Zebrafish (Danio rerio) due to their molecular and physiological similarities with humans, therefore giving a high-throughput for genotoxicity (Chakravarthy et al., 2014). Rodents and other mammals are also widely used for genotoxicity assessment. In all these models, the comet assay, the micronucleus assay and the chromosome aberrations test are the most common used tests to evaluate nanoparticles toxicity (Magdolenova et al., 2013).

Importantly, some considerations have been published by OECD regarding the protocols to assess genotoxicity of NMs, namely the "2018 Report No. 85—Evaluation of in vitro methods for human hazard assessment applied in the OECD Testing Programme for the Safety of Manufactured Nanomaterials" and "2014 Report No. 43—Genotoxicity of Manufactured Nanomaterials: Report of the OECD expert meeting" (OECD, 2014, 2018a).

Data collected from the literature assessing genotoxicity of polymeric NMs is summarized in **Table 6**. Again, most of the data collected refers to chitosan and PLGA based NPs and should be carefully analyzed. First, we must recognize we are comparing NPs comprising a particular polymer (chitosan or PLGA) but whose chemical specifications can differ and whose composition and characteristics are very diverse. Also, comparisons should ideally be performed only when the same test is applied. In detail, chitosan/poly(methacrylic acid) NPs induced a concentration dependent genotoxic effect according to the cytogenetic test using human lymphocyte culture (De Lima et al., 2010). However, the same report reported no evidence for DNA alterations using the Allium Cepa assay (De Lima et al., 2010). In another study, Eudragit <sup>R</sup> S100/alginate enclosed chitosan calcium phosphate-loaded lactoferrin nanocapsules, was considered non-genotoxic based on the Allium Cepa and the comet assay in Vero cells (Leng et al., 2018). Overall, these two studies comprising nanoparticles with chitosan in their composition, presented a different conclusion for the NM genotoxicity, but if we compare only the same assay (Allium Cepa assay), the results were similar. Another interesting fact, is the heterogeneity of results that may be achieved with different cell lines. For instance, Platel et al. used three different cell lines, and three different PLGA NPs and evaluated genotoxicity using the comet assay and the micronucleus test (Platel et al., 2016). For bare PLGA NPs, no genotoxicity effects were verified in none of the 3 cell lines with both tests (Platel et al., 2016). On the other hand, CTAB stabilized PLGA NPs induced an increase in the number of micronuclei only in one of the cell lines (micronucleous test in HBE14o- cells) (Platel et al., 2016). These examples illustrate how an extrapolation based on one single genotoxicity assay (or cellular/animal model) can be misleading.

### Toxicity on Reproduction

The extrapolation to human health of toxic effects on reproduction using in vitro and animal models presents several specific limitations, such as the differences in reproductive structures and endocrine functions or the duration of gestation or spermatogenesis period (Das et al., 2016). Also, alike other studies, the tested concentrations and doses are much higher than the clinically relevant doses in humans (Das et al., 2016). Nevertheless, the toxicity on reproduction is a valuable endpoint since it allows the prediction of health effects not only of individuals but also of the next generation (Dusinska et al., 2017).

As mentioned before, toxicity on reproduction might be evaluated using in vitro and in vivo studies. For instance, in vitro assays test the toxicity of nanoparticles in cells from reproductive organs (such as blastocysts and granulosa cells) or use ex vivo placentae or sperm from healthy donors (Ema et al., 2010; Sun et al., 2013; Brohi et al., 2017). In these examples, the authors expect to see direct toxicity of the NPs in reproductive system cells, or to evaluate the ability of the NPs to cross for instance the placental barrier (Ema et al., 2010; Brohi et al., 2017).

Regarding in vivo testing, the use of mice as a mammalian model provides analogous experimental conditions to humans. However, the investigation of early embryonic developmental effects occurring in utero are not easily detectable (Sun et al., 2013). Interestingly, the zebrafish model has been widely applied as a rapid and cost-effective whole animal model to assess reproductive toxicity (Hu et al., 2011). Characteristics like the small size, rapidity to reach sexual maturity, great number of eggs (200–300) and the possibility to examine every stage of embryonic development through its transparency, make zebrafish one of the most used animal models (Wang et al., 2016).

Results from toxicity on reproduction assays with polymeric NMs are summarized in **Table 7**. The results for chitosan NPs (blend and bare) are consistent between reports. In fact, it appears that chitosan based NPs induce embryonic malformations when directly in contact with embryos, or intravenously administered to animal models (Hu et al., 2011; Park et al., 2013; Choi et al., 2016; Wang et al., 2016; Yostawonkul et al., 2017). However, this effect is not verified in when PLGA NPs coated with chitosan are administered through the oral route in Sprague Dawley rats (Sharma et al., 2017). Though, this conclusion is only speculative. In order to have a proven conclusion, the oral route should be tested for toxicity on reproduction using the same NPs as were used for the intravenous administration and embryonic incubation experiments. Otherwise, we cannot be sure if the result is due to the administration route, or the NPs composition and characteristics. Nevertheless, other study using PLGA based NPs also tested toxicity on reproduction through the in vitro zebrafish embryonic model, and found no toxicity for those nanoparticles (Chen et al., 2017).

#### Hemocompatibility

Hemocompatibility is frequently assessed as an endpoint of biocompatibility for chemicals and particularly NMs. In fact, blood is the first target when considering intravenous injections of NMs, but it is also a surrogate target model for other routes of exposure, since its high complexity allows for an approximation the overall body response (Tulinska et al., 2015).

In particular, hemolysis which is associated to red blood cells damage is believed to have a good correlation with toxicity, since the in vitro hemolytic assays show results that greatly relate with in vivo toxicity studies (Dobrovolskaia and McNeil, 2013).

In 2008, Dobrovolskaia et al. published a report describing the validation of an in vitro assay for the analysis of nanoparticle hemolytic properties and main interferences (Dobrovolskaia et al., 2008). In 2013, ASTM International standards organization published the Standard Test Method for Analysis of Hemolytic Properties of Nanoparticles and defined a material as hemolytic if the hemolysis values are above 5% and as moderately hemolytic if they are between 2 and 5% (ASTM International, 2013; Dobrovolskaia and McNeil, 2013). Therefore, the existence of this protocol contributes to the use of standardized procedures among research groups, allowing comparisons and extrapolations of results.

From **Table 8** we can acknowledge several authors reporting the hemolytic activity of diverse polymeric NMs. An important remark is the fact that a number of papers describe the hemolytic activity of drug loaded formulations and compare it to the free drug, but not with the unloaded nanocarrier (Essa et al., 2012; Gupta et al., 2012; Altmeyer et al., 2016; Radwan et al., 2017a). These results generally demonstrate a lower hemolysis rate of the drug loaded polymeric NM in comparison to the free drug, but still a significant hemolysis (>5%) (Essa et al., 2012; Gupta et al., 2012; Radwan et al., 2017a). In these situations, no conclusion regarding the hemolytic activity of the polymeric NM itself can be drawn. On the other hand, some other authors, test the unloaded nanoparticles but make no disclosure of their concentration (Altmeyer et al., 2016; Moraes Moreira Carraro et al., 2017).

Nevertheless, polymeric NMs appear to present good hemocompability profile, as in most tested cases, hemolysis is a concentration dependent phenomenon, reaching significant values only for high NM concentrations. Also, the encapsulation of hemolytic drugs in polymeric NMs decreases their hemolytic activity.

#### DISCUSSION

Most information available on nanotoxicity is related to inorganic NMs, such as zinc oxide NPs, nanoscale silver clusters, and titanium dioxide NPs or carbon nanotubes (Yuan et al., 2015). Information related to polymeric NMs toxicity that could be correlated with their effects on human health is still scarce and poorly harmonized.

The majority of reports on polymeric NMs are focused in optimizing the nanocarrier features, such as size, physical stability and drug loading efficacy, and in performing preliminary cytocompatibility testing (mainly through MTT and LDH assays)

Jesus et al.



*(Continued)*

Polymeric


October

2019 | Volume 7 | Article 261


References

Platel et al.,

2016

Platel et al.,

2016

**89**

hexadecyltrimethylammonium

Nanomaterial

PEG stabilized PLGA NPs

bromide (CTAB) stabilized PLGA NPs

Polymer Characterization

Resomer

RG503H, acid

Resomer

RG503H, acid

Mw

and PEG 2000

terminated,

24,000–38,000

Mw

terminated,

24,000–38,000

®

 50:50,

®

 50:50,

Nanomaterial

Characterization

78 nm

−1 mV

82 nm

+15 mV Testing method

Comet assay (3 h)

> + 40 h recovery

Comet assay (3 h)

> + 40 h recovery

time)

micronucleus

and

(3

time)

Micronucleus

 test

 test

and

(3

 Model

L5178Y and

16HBE14o-,

L5178Y and

TK6 cells

TK6 cells –

–

route (if applicable)

Administration

Dose/ concentration

range

(L5178Y and TK6

cells)

50–500µg/mL

25–100µg/mL

25–100µg/mL

 cells)

(16HBE14o-

(L5178Y and TK6

cells)

Results

No primary DNA,

no chromosomal

damage and no

increase in the

number of

L5178Y and TK6

damage on

L5178Y and TK6

concentrationrelated increase in

micronuclei

 in

cells;

the number of

cells

No primary DNA

or chromosomal

micronulei on The L5178Y mouse

The L5178Y mouse

lymphoma and TK6 human

routinely used in *in vitro*

regulatory genotoxic assays

lymphoma and TK6 human

routinely used in *in vitro*

assays. The human

bronchial epithelial cells

suitable for toxicity studies

16HBE14o-,

 a cell line is

regulatory genotoxic

B-lymphoblastoid

B-lymphoblastoid

 cells, are

 cells, are

Observations


Polymeric

Nanobiomaterials: Hazard Assessment

TABLE 7 | Review of original articles assessing toxicity on reproduction


 induced by polymeric nanoparticles.

Polymeric

Nanobiomaterials: Hazard Assessment

TABLE 7 | Continued


*ab.w., body weight.*


#### TABLE 8 |Review of original articles assessing hemolysis induced by polymeric nanoparticles.

 References

Shelma and

Sharma, 2011

Nadesh et al.,

Sarangapani

2018

Kumar et al., 2017

Liu et al., 2013

Da Silva et al.,

Da Silva et al.,

2019

Radwan et al.,

2017a

Moraes Moreira

Carraro et al.,

2017

2019

 et al.,

2013

Polymeric

Nanobiomaterials: Hazard Assessment

*(Continued)*

negligible hemolysis

(unknown concentration)




*(Continued)* Polymeric

Polymeric

Nanobiomaterials: Hazard Assessment

#### TABLE 8 | Continued


and proving effectiveness of the drug loaded formulation, using the most diverse cell lines (Lorscheidt and Lamprecht, 2016). Toxicological studies exploring the biological effects of the polymeric NMs, particularly regarding immune system interaction are often disregarded. Though, as suggested by the safe-by-design concept, the toxicity study of NMs should be the starting point for the formulation development.

After our research on original peer reviewed articles, we selected the following endpoints to analyze that are crucial to understand the toxicity of nanobiomaterials for drug delivery: acute toxicity, repeated-dose toxicity, inflammation, oxidative stress, genotoxicity (including carcinogenicity and mutagenicity) toxicity on reproduction, and hemolysis. Importantly, one of the first conclusions to retain is that among different research groups, the methodologies, the animal or cellular model, the dose or concentration, the assay duration and notably, the polymeric NM properties, are not the same, making it difficult to compare and establish trends. This issue derives in part from the absence of regulatory binding and standardized methodologies and guidelines which hardens the comparison of safety/toxicity assessments in different reports (Dhawan and Sharma, 2010), and ultimately, makes it difficult to extrapolate safety profiles for human health. A similar conclusion was achieved by Park and coworkers, who discussed the status of in vitro toxicity studies for wide-ranging NMs, particularly cytotoxicity, oxidative stress, inflammation and genotoxicity and established that important limitations were preventing their use for human health risk assessment (Park et al., 2009).

Among the different polymeric NMs available, the most studied and reported are chitosan and PLGA nanoparticles. "Chitosan nanoparticles" and "PLGA nanoparticles" are general terms used for an endless number of different nanoparticles comprising multiple polymeric combinations, cross-links and surfactants, and therefore, displaying diverse physical and chemical properties as illustrated by the first 3 columns of **Tables 3**–**8**. As expected, these variables, together with the great diversity of protocols employed by different authors for the same assays, generates ambiguous results that prevent the establishment of trends between the nanocarriers characteristics and the expected toxicological endpoints.

An adequate characterization of the polymeric NMs is crucial for a comprehensive interpretation of the results but also to allow a comparison between different NMs. In 2018, in the context of EU FP-7 GUIDEnano project, it was published the development of a systematic method to assess similarity between NMs that would allow the extrapolation of results for human hazard evaluation purpose (Park et al., 2018). In that methodology they defined the following parameters for assessing similarities between NMs: chemical composition, crystalline form, impurities, primary size distribution, aggregate/agglomerate size distribution, density, and shape. Importantly, those parameters should be tested and compared in relevant media accordingly to the exposure route or toxicity test. However, in the process of developing such methodology, the authors identified several challenges that prevented the establishment of thresholds for establishing similarity. They suggest that the awareness of researchers for the relevance of characterizing NMs when performing hazard assessments is increasing which can lead to the establishment of the thresholds in the future, facilitating the extrapolation of hazard endpoints between similar NMs. Indeed, among the different research articles analyzed, the lack of broad characterization is frequent, sometimes even ignoring important parameters, such as the polymer molecular weight or the nanoparticle size.

Another aspect that should be taken into consideration when characterizing the polymeric NMs to study their biological effects is the endotoxin contamination. In fact, when discussing for instance cytokine stimulation or oxidative stress, endotoxin contamination should not be neglected. Nevertheless, endotoxin quantification (or its acknowledgment) on chitosan and other polymeric NMs is still scarce, which compromises some of the results found in the literature regarding their bioactivity and toxicity. In addition, despite testing the presence of endotoxins is a common procedure in laboratory and several commercial tests are available, they need to be validated for use with NMs, since most are based on optical assays and may be affected by the optical density of NPs (Dobrovolskaia et al., 2010).

Not only endotoxin detection assays are susceptible of interference from NMs and consequently misinterpretation of the results. Therefore, one way of trying to overcome this problem is to use different assays to evaluate the same endpoint. Additionally, experiment controls, such as the incubation of probes (without biological matrixes) and positive controls with NMs, can reveal whether these NMs might be generating false positive or negative results.

The obstacles identified in this review prevent the identification of toxicity trends and the generation of a useful database where we can rely for the Safe-by-Design. Only by performing in vitro and in vivo harmonized toxicity studies using unloaded polymeric NMs, extensively characterized regarding their intrinsic and extrinsic properties and by performing all necessary controls it is possible to generate such database. At the present time, taking everything into account, the human health risk assessment of polymeric NMs is still dependent on a case-by-case evaluation, and it should comprise the evaluation of parameters, such as the route of administration and dose, among others, to define the required tests for the hazard assessment (i.e., type of in vitro and in vivo studies).

#### AUTHOR CONTRIBUTIONS

SJ gathered the information, analyzed, and wrote the first draft of the manuscript. SJ, MS, CS, GB, PW, and OB defined the subjects for discussion. All authors contributed to manuscript revision, read, and approved the submitted version.

#### FUNDING

This work was financed by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational Programme under project CENTRO-01-0145-FEDER-00008:BrainHealth 2020, and through the COMPETE 2020—Operational Programme for Competitiveness and Internationalization and Portuguese national funds via FCT—Fundação para a Ciência e a Tecnologia, I.P., under project PROSAFE/0001/2016, and the strategic projects POCI-01-0145-FEDER-030331 and POCI-01-0145-FEDER-007440 (UID/NEU/04539/2019). This work is part of the

#### REFERENCES


GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016 and from the CTI (1.1.2018 Innosuisse), under grant agreement Number 19267.1 PFNM-NM.


due to interference with assay processes and components of classic in vitro tests. Nanotoxicology 9, 13–24. doi: 10.3109/17435390.2013.829590


PLGA-PEO nanoparticles in human blood cell model. Nanotoxicology 9(Suppl. 1), 33–43. doi: 10.3109/17435390.2013.816798


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

Copyright © 2019 Jesus, Schmutz, Som, Borchard, Wick and Borges. 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.

# Meta-Analysis of Pharmacokinetic Studies of Nanobiomaterials for the Prediction of Excretion Depending on Particle Characteristics

Marina Hauser and Bernd Nowack\*

*Empa, Swiss Federal Laboratories for Materials Science and Technology, St. Gallen, Switzerland*

The growth in development and use of nanobiomaterials (NBMs) has raised questions regarding their possible distribution in the environment. Because most NBMs are not yet available on the market and exposure monitoring is thus not possible, prospective exposure modeling is the method of choice to get information on their future environmental exposure. An important input for such models is the fraction of the NBM excreted after their application to humans. The aim of this study was to analyze the current literature on excretion of NBMs using a meta-analysis. Published pharmacokinetic data from *in vivo* animal experiments was collected and compiled in a database, including information on the material characteristics. An evaluation of the data showed that there is no correlation between the excretion (in % of injected dose, ID) and the material type, the dose, the zeta potential or the size of the particles. However, the excretion is dependent on the type of administration with orally administered NBMs being excreted to a larger extent than intravenously administered ones. A statistically significant difference was found for IV vs. oral and oral vs. inhalation. The database provided by this work can be used for future studies to parameterize the transfer of NBMs from humans to wastewater. Generic probability distributions of excretion for oral and IV-administration are provided to enable excretion modeling of NBMs without data for a specific NBM.

Keywords: nanobiomaterials, pharmacokinetic, meta-analysis, excretion, prediction

# INTRODUCTION

In the past decade, nanobiomaterials (NBMs) have been increasingly investigated for the use in pharmaceutics and biomedical engineering (Küster and Adler, 2014). A wide range of different nanomaterials are being suggested for these purposes. For example, metals or metal oxides are very common in nanomedicine. Their relatively simple generation and surface modification as well as biocompatibility make gold (Au) nanoparticles attractive for the utilization in medical imaging or cancer detection and treatment (Hirn et al., 2011; Bonakdar and Mashinchian, 2015; Rambanapasi et al., 2015). Silver (Ag) nanoparticles are applied as coatings for indwelling catheters, antibacterial agents, wound dressing, orthopedic implants, and tissue-engineered scaffolds (Lin et al., 2015). Silica nanoparticles (SiO2) are easy to synthesize, exhibit low toxicity and have an ease for surface modification. These properties make silica applicable as biomarkers, biosensors, DNA or drug delivery, and cancer therapy (Lee et al., 2014). Also organic nanomaterials are often used in medical applications, especially due to their high biological safety, good biodegradability,

Edited by:

*Olga Borges, University of Coimbra, Portugal*

#### Reviewed by:

*Domenico Cassano, Joint Research Centre (Italy), Italy Po-Chang Chiang, Genentech, Inc., United States*

> \*Correspondence: *Bernd Nowack nowack@empa.ch*

#### Specialty section:

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

Received: *08 October 2019* Accepted: *27 November 2019* Published: *17 December 2019*

#### Citation:

*Hauser M and Nowack B (2019) Meta-Analysis of Pharmacokinetic Studies of Nanobiomaterials for the Prediction of Excretion Depending on Particle Characteristics. Front. Bioeng. Biotechnol. 7:405. doi: 10.3389/fbioe.2019.00405* low environmental toxicity (Hauser et al., 2019), and easy production and modification (Han et al., 2018). Commonly used organic NBMs are chitosan, polylactic acid (PLA), or poly(lacticglycolic acid) (PLGA). They may be preferred to other types of nanoparticles due to their flexibility, biodegradability, and relatively low levels of toxicity (Navarro et al., 2017). Chitosan is a polysaccharide which is found in the exoskeleton of crustaceans and is applied in fast wound healing or as a blood clotting agent (Singh et al., 2017). PLA is used in cartilage regeneration, bone tissue engineering, and cartilage repair due to its good elastic modulus, thermal formability, and mechanical strength. PLGA is widely used in nanoparticles, microspheres, pellets, sutures, implantable scaffolds, and microcapsules (Navarro et al., 2017; Han et al., 2018). Additionally, also carbon-based nanomaterials are used in nanomedicine. Fullerenes and carbon nanotubes (CNTs) are highly promising for medical applications as carriers in drug delivery (Yamashita et al., 2012).

NBMs can be administered to the patient's body in different ways. The most commonly used routes of administration in humans are oral, intravenous and inhalation. From these, the oral route is the most convenient one as it is non-invasive and therefore widely accepted by most patients (Schleh et al., 2012). Besides, it also has the potential to be taken at home and not necessarily in a hospital or clinic setting (Navarro et al., 2017). However, the absorption into the bloodstream after oral absorption is generally very low (Park et al., 2011; Lin et al., 2015). The lungs are considered the most important entry of nanoparticles into the human body for example via occupational inhalation of airborne particles during manufacturing (Li X. et al., 2012; Laux et al., 2017). The advantage of intravenous injection is the direct access of the NBM to the blood circulation and thereby a quick distribution throughout the entire body (Hirn et al., 2011). In animal studies also intratracheal (introduction of the material directly into the trachea) or intraperitoneal (into the body cavity) administration is common.

Increasing applications and usage of NBMs leads to an increase in the potential for environmental exposure (Laux et al., 2017; Kabir et al., 2018). Depending on the material, a NBM can biodegrade, accumulate in tissues and organs or get excreted via urine or feces. From urine and feces, they enter the sewage system and are eventually discharged into surface water from where they are distributed throughout the whole biosphere. We expect NBMs to behave similarly to pharmaceuticals as they have the same mode of application and are also excreted in urine and feces from where they reach the sewage system. The German Federal Environment Agency reported the detection of 156 pharmaceuticals in environmental media such as surface water, groundwater and drinking water (Umwelt Bundesamt, 2018). Pharmaceuticals were detected in surface water at a concentration of 0.1–10.0 µg/l (Bergmann et al., 2011).

In order to be able to assess the environmental exposure, one needs knowledge of the presence of nanomaterials in different products but also about their release throughout the life cycle (Som et al., 2010; Keller et al., 2013). The release of nanomaterials into the environment has previously been modeled for a range of engineered nanomaterials (Mueller and Nowack, 2008; Gottschalk et al., 2009, 2010; Sun et al., 2014, 2016, 2017; Wang et al., 2016). However, only one modeling study has been published for NBMs, covering the environmental exposure of gold-nanoparticles from medical applications in the United States and the United Kingdom (Mahapatra et al., 2015).

In exposure modeling the whole life cycle of the material needs to be taken into consideration. For NBMs, the excretion of the NBM from the body is the starting point from where they flow to the sewage system, the waste water treatment plant and finally can be distributed throughout the biosphere to reach different environmental compartments such as soil, ground water, oceans, as well as the atmosphere. In recent years, the number of published physiologically based pharmacokinetic models (PBPK) of NBMs has increased significantly (Grass and Sinko, 2002; Li et al., 2010; Li M. et al., 2012; Li M. et al., 2016; Moss and Siccardi, 2014; Carlander et al., 2016; Li D. et al., 2016). These studies are mostly interested in the distribution of the NBMs in the body to different organs and tissues but the excretion of the material in feces or urine is in many cases also considered.

The aim of our study was to collect data from published pharmacokinetic studies of NBMs and make predictions based on this data set about the excretion of the NBM from the body. As different studies used different materials, coatings, administrations, doses, animals, and evaluation time spans, we aimed to incorporate the different materials and particle properties or study designs into the evaluation and to make general predictions about the excretion of NBMs.

## METHODS

The literature was searched for pharmacokinetic studies of NBM or nanoparticles in general that specifically quantified excretion of the nanoparticles. The time frame of the search includes all studies until the end of April 2019. Google Scholar was used with search terms such as "pharmacokinetics nanoparticles excretion," or "pharmacokinetics nanomaterials excretion," "pharmacokinetics metallic/polymeric/organic/etc. nanoparticles/nanomaterials excretion" in all variations, or just "nanoparticles excretion." For each search term, the first ten pages each containing 10 articles were looked at. Besides, the cited articles of these studies were also evaluated.

Only studies with a time frame of a least 1 day were considered. As we were only interested in the total excretion of the nanoparticles, studies with a time frame of <1 day were deemed too short to fully excrete the nanoparticles. Additionally, only studies where the excretion in feces and/or urine is mentioned in %ID (percent of injected dose) or total excretion with the amount administered mentioned in the article (so the %ID could be calculated) were considered. Within one study, only the data point at the longest time was collected per material as it was assumed that this shows the total excretion. Only one data point was collected per material per study to avoid overrepresentation of studies with many measurements. However, several data points were collected from one study if materials with different size, zeta potential, surface coating, dose, etc. were used.

FIGURE 1 | Number of data points for each type of administration (A) and for each type of material class (B). From each pharmacokinetic study of nanobiomaterials only one data point was extracted per specific material and the cumulative excretion as well as the material properties were reported. The whole database with all data points can be found in the Supporting Information. IV, Intravenously administered, QD, Quantum dots.

For each material, the material class, the particle size (TEM measurements), the test animal, the route of administration, the zeta potential of the material, the administered dose, and the cumulative excretion (in %ID) were noted. We have taken these material characteristics as they were mentioned in other articles to be of significance for the excretion of the material (Soo Choi et al., 2007; Semmler-Behnke et al., 2008; Alric et al., 2013; Xu et al., 2018). TEM measurements of the primary particle size were preferred over hydrodynamic size as TEM measurements were more widely available and as the nanoparticles get rapidly modified by protein adsorption after administration in the body (Kreyling et al., 2014).

# RESULTS AND DISCUSSION

# Presentation of the Database

In total, 192 data points were collected from 66 studies. The whole database can be found in the **Table S1**. More than 60% of the nanomaterials were administered intravenously (IV), 30% orally, 7% intratracheally, and <3% by inhalation or intraperitoneal or intrahepatic injection (see **Figure 1A**). Of all the materials investigated, 40% were metallic, 35% metal oxides, 12% organic and <4% carbon-based, Quantum Dots (QD), clays or other (see **Figure 1B**).

Not all studies reported all relevant material or study characteristics. For almost 45% of the data points, the full data set

with zeta potential, size, and administered dose was available (86 data points), see **Figure 2**. For six data points the zeta potential was only listed as positive or negative. These data points were counted as only size and dose available. For 36% of the data points only the size and the dose were mentioned but not the zeta potential, whereas for 5% of the data points only the size and the zeta potential was available but not the dose. For more than 10% of the data points only the size and for 2% of the data points only the dose could be found. For one data point, neither the size nor the dose or the zeta potential was mentioned in the article (see **Figure 2**). The particles ranged in size from 1.1 to 360 nm, the zeta potential ranged from −76 to 106.2 mV, and the administered dose ranged from 0.0032 to 2,000 mg/kg body weight.

points can be found in the Supporting Information.

The amounts excreted through urine and feces were added together to get the total excretion of the nanomaterial. In order to evaluate if there is a relationship between the size of the material, the zeta potential, or the administered dose, each of these properties were plotted against the cumulative excretion. The dots were color-coded either for the type of administration (**Figures 3B,D**) or the material class (**Figures 3A,C**) to see if there was any relationship. Only material classes or administration types with at least three data points were used. Categories with <3 data points are shown together as "All other" just for illustrative purposes. Not all graphs have the same amount of points as for some data points the specific information was missing. For example, only 96 of the 192 data points have a zeta potential mentioned in the original study, therefore there are only 96 points in the graph for zeta potential and not 192. Plotting all data points together (**Figure 3A**), it can be seen that most (94%) of the materials are below 200 nm in size, the majority (79%) even below 100 nm, which would be the currently accepted threshold for the nanoparticle definition (European Commission, 2011). Regarding the zeta potential (**Figure 3C**), the majority (67%) of the data points have a negative zeta potential, only a few (33%) have a positive zeta potential. The doses used in most studies are below 100 mg/kg or even less, only a very small amount of studies used higher doses (**Figure 3D**). The plots for cumulative excretion versus zeta potential of the nanomaterial color-coded by type of administration and cumulative excretion versus dose of the nanomaterial color-coded by material class can be found in the Supporting Information in **Figures S1**, **S2**, respectively.

### Data Evaluation

Several studies report size and surface charge of nanoparticles to be of major influence for their biodistribution and excretion. Small particles (Soo Choi et al., 2007; Semmler-Behnke et al., 2008; Li D. et al., 2016; Jasinski et al., 2018) and positivelycharged particles (Alric et al., 2013) are reported to be excreted faster than larger or negatively and neutrally-charged particles. However, looking at the graphs above, there seems to be no correlation between size or zeta potential and excretion neither for different types of administration nor for different material classes. Therefore, a multilinear regression was calculated for the 86 data points for which the size, dose, and zeta potential was available to check if there was any relationship. Size, dose and zeta potential were used as input values and the cumulative excretion of feces and urine in percent as the output. The calculations show

color-coded by type of administration (D).

that using zeta potential, size, and dose of nanomaterials, the accuracy of predicting the cumulative excretion is low with R 2 being only 0.29. The plot of observed vs. predicted values shown in the **Figure S3** reveals that the multilinear regression does not result in an acceptable fit. Taking all data together, it is therefore not possible to predict the amount excreted based on size, zeta potential and amount administered.

Regarding dose dependencies, Xu et al. (2018) have found strong dose-dependent renal clearance of glutathione-coated gold nanoparticles. At higher doses, the same can be seen in the graphs considering all types of nanoparticles. This might be explained by the fact that these doses are so high that the tissues are saturated with the material and the body cannot take up more of the nanomaterial and it is therefore excreted.

Looking at **Figure 3B** showing the size against excretion colorcoded by type of administration, there seems to be a general trend of orally administered particles (blue dots) being excreted more than intravenously administered particles (red dots). Therefore, we have plotted the cumulative excretion vs. the administration for all administration types with three or more data points. **Figure 4** shows a boxplot of the cumulative excretion distribution for the five types of administration. The data points are plotted in red circles for each type of administration and the number of data points available for each type of administration is written in brackets next to the administration type.

To test whether the cumulative excretion of the different types of administration is statistically different, we applied a one-way analysis of variance (ANOVA) followed by a post-hoc Turkey test on the data set (**Table 1**). The criterion for statistical significance was p < 0.05. We found that only IV-oral, oralintratracheal, oral-inhalation, and intratracheal-intraperitoneal were significantly different.

#### Prediction of Excretion for Environmental Risk Assessment

In environmental risk assessments, the potential hazard of a material is compared to the extent the material will come in contact with an organism (ECHA, 2016). Several environmental hazard assessments have been performed on various nanomaterials: Coll et al. (2016) for nano-Ag, CNT,

TABLE 1 | *p*-values from ANOVA for testing statistical difference between different types of administration (*p* < 0.05 in green, *p* > 0.05 in red).


nano-TiO2, and nano-ZnO in freshwater; Hauser et al. (2019) for chitosan, nano-chitosan and HAP in freshwater, and chitosan in soil; Mahapatra et al. (2018) for nano-Au in freshwater; Wang and Nowack (2018) for nano-Al2O3, nano-SiO2, nano iron oxides, nano-CeO2, and QDs in freshwater. On the other hand, only one study has been performed so far on environmental exposure to NBMs (Mahapatra et al., 2015 for nano-Au). Therefore, more research is needed on the exposure side before environmental risk assessments of NBMs can be performed. As often the NBMs in question are only in the development stage and not yet on the market, the only way to estimate the prospective environmental concentration is through mathematical models (Gottschalk et al., 2009). The amount of a nanomaterial released into a technical or environmental compartment is a central point in any release model (Gottschalk and Nowack, 2011). For NBMs the main relevant release process is the excretion from the human body. If most of the NBM is excreted, it will end up in the wastewater, if it stays in the body or is metabolized, there is no immediate release into water.

The excretion data collected in the database (**Table S1**) can be used to predict excretion for a specific NBM or be used to obtain a generic excretion rate for NBM with a specific administration. So if a specific material has its own data, then the real excretion for this material can be used in the model. If however for the material in question, no own data is available, then data from the database can be used in the form of probability distributions. Therefore, for each type of administration, a histogram was prepared to show the distribution of the data points. For IV and oral administration, there are enough data points to see the distribution (see **Figures 5A,B** below). For inhalation only four data points were available. The histogram for inhalation can be found in the **Figure S4**. As intratracheal and intraperitoneal administration are not used on humans, their data are not shown here and will not be further evaluated. The distributions shown in **Figure 5** represent the probability that a NBM is excreted to a certain extent and can be used as input value to parameterize excretion in probabilistic exposure models such as DPMFA (dynamic probabilistic material flow analysis) (Bornhöft et al., 2016).

Recently published studies have focused on evaluating small difference in particles characteristics and their influence on the biodistribution and excretion. It is generally believed that particles below 5.5 nm in size get rapidly cleared from the body through urinary excretion (Soo Choi et al., 2007). Du et al. (2017) evaluated urinary excretion of sub-nm gold particles with the same surface ligands but different sizes after IV injection. They found that a size reduction of just a few atoms resulted in a decrease in urinary clearance. As in our database, no other

material with specific material properties from one study.

materials were in the sub-nm size-range, we could not confirm this on a general basis with other materials. As mentioned before, we have not found a size dependent relationship. Cassano et al. (2019) compared the excretion of silver, gold and platinum nanoparticles and found that while gold nanoparticles are predominantly excreted in urine, silver nanoparticles were almost completely found in feces. We have only analyzed the total excretion, however, it would be interesting to evaluate the route of excretion for the different NBMs. Jasinski et al. (2018) evaluated the effect of shape of RNA nanoparticles on their biodistribution. They compared squared, triangular and pentagon-shaped RNA nanoparticles of 10 nm size. Fluorescent images showed a high fluorescence in kidneys after 12 h for nanosquared, but none for the triangle and very little for the pentagon-shaped nanoparticles. Most studies used round nanoparticles, so to study the general effect of shape, more studies using differently shaped nanoparticles would be needed in the future.

The data collected in the database are all from animal studies. No study is available in which pharmacokinetic profiles for NBMs are compared between animals and humans to get an indication on the extrapolation of animal data to humans with regard to excretion. Data on excretion for other pharmaceuticals are available for different animals and humans. Mamidi et al. (2014) performed an excretion study of orally administered canagliflozin (used for the treatment of type 2 diabetes) in mice, rats, dogs, and humans. They have found a total excretion of canagliflozin and its metabolites of 97.8 and 98.3% for male and female mice, respectively, 96.9 and 98.4% for male and female rats, 99.1% for male dogs, and 92.9% for male humans. Maurer et al. (1983) administered bromocriptine (used for the treatment of Parkinson's disease) orally to mice, rats, monkeys, and humans. They have found a total excretion 94.2% and 101.6% for mice with a dose of 3 and 50 mg/kg, respectively, 83.4% for rats, 101.7% for monkeys, and 88.0% for humans. Comparing these studies, the total excretion from humans is in a similar range as the excretion from the animals included in our database. Therefore, we can assume that the excretion of NBMs in humans would also be in a similar range to animals and thus we can use the calculated excretion profiles for further modeling of NBMs administered to humans.

#### DATA AVAILABILITY STATEMENT

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

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

MH collected, prepared, evaluated the input data, created the figures and tables for the manuscript, and wrote the manuscript. BN supervised the study, gave inputs on the data, and contributed to the writing of the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This work was supported by the BIORIMA project which received funding from the European Union Horizon 2020 framework under grant agreement #760928.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2019.00405/full#supplementary-material


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

Copyright © 2019 Hauser and Nowack. 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.

# A Review of Nanotechnology for Targeted Anti-schistosomal Therapy

Tayo Alex Adekiya, Pierre P. D. Kondiah, Yahya E. Choonara, Pradeep Kumar and Viness Pillay\*

Wits Advanced Drug Delivery Platform Research Unit, Department of Pharmacy and Pharmacology, School of Therapeutic Science, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

Schistosomiasis is one of the major parasitic diseases and second most prevalent among the group of neglected diseases. The prevalence of schistosomiasis may be due to environmental and socio-economic factors, as well as the unavailability of vaccines for schistosomiasis. To date, current treatment; mainly the drug praziquantel (PZQ), has not been effective in treating the early forms of schistosome species. The development of drug resistance has been documented in several regions globally, due to the overuse of PZQ, rate of parasitic mutation, poor treatment compliance, co-infection with different strains of schistosomes and the overall parasite load. Hence, exploring the schistosome tegument may be a potential focus for the design and development of targeted anti-schistosomal therapy, with higher bioavailability as molecular targets using nanotechnology. This review aims to provide a concise incursion on the use of various advance approaches to achieve targeted anti-schistosomal therapy, mainly through the use of nano-enabled drug delivery systems. It also assimilates the molecular structure and function of the schistosome tegument and highlights the potential molecular targets found on the tegument, for effective specific interaction with receptors for more efficacious anti-schistosomal therapy.

Keywords: schistosomiasis, nanoparticles, drug delivery, targeted agents, molecular receptors, antibody, aptamers

# INTRODUCTION

Schistosomiasis is recognized as the second most prevalent among the group of NTDs in sub-Saharan Africa, following hookworm infection (Adekiya et al., 2017). Schistosomiasis is an infectious disease caused by parasitic worms that belong to the group of trematode and genus of Schistosoma, that results in chronic and acute disease (Adekiya et al., 2017). It poses a significant challenge on agricultural productivity and the life, growth and development of pregnant women and school children in afflicted areas. The disease-causing species of Schistosoma are Schistosoma mansoni, Schistosoma haematobium,

#### Edited by:

Olga Borges, University of Coimbra, Portugal

#### Reviewed by:

Pradipta Ranjan Rauta, University of Texas MD Anderson Cancer Center, United States Stefano Leporatti, Institute of Nanotechnology (NANOTEC), Italy

#### \*Correspondence:

Viness Pillay viness.pillay@wits.ac.za

#### Specialty section:

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

Received: 03 November 2019 Accepted: 14 January 2020 Published: 31 January 2020

#### Citation:

Adekiya TA, Kondiah PPD, Choonara YE, Kumar P and Pillay V (2020) A Review of Nanotechnology for Targeted Anti-schistosomal Therapy. Front. Bioeng. Biotechnol. 8:32. doi: 10.3389/fbioe.2020.00032

**Abbreviations:** Ach, acetylcholine; AChE, acetylcholinesterase; ACS, american cancer society; AD, alzheimer's disease; AFM, atomic force microscopy; AuNP, gold nanoparticle; BBB, blood-brain barrier; CNS, central nervous system; IgGs, immunoglobulins; IVM, ivermectin; LBNPs, lipid-based nanoparticles; LGIC, ligand-gated ion channel; LNCs, lipid nanocapsules; MB, methylene blue; nAChR, nicotinic acetylcholine receptor; NK, natural killer; NLCs, nano-lipid carriers; NTDs, neglected tropical diseases; pRBCs, plasmodium-infected red blood cells; PZQ, praziquantel; RBCs, red blood cells; SEM, scanning electron microscopy; SGTP1, schistosome glucose transporter 1; SGTP4, schistosome glucose transporter 4; SLN, solid lipid nanoparticle; TEM, transmission electron microscopy; TSPs, tetraspanins; Th1, T-helper 1 cells; Th2, T-helper 2 cells; WHO, world health organisation.

Schistosoma japonicum, Schistosoma intercalatum, and Schistosoma mekongi (Adekiya et al., 2017; da Paixão Siqueira et al., 2017). For these worms to cause disease, the intermediate hosts (freshwater snails) need to be infected with the miracidia in freshwater where it develops into cercaria. Following human-water exposure, the cercaria penetrates the intact skin of humans.

Schistosomiasis affects the world's poorest countries where there is no safe water, basic sanitation and hygiene education (da Paixão Siqueira et al., 2017). Currently, over 200 million people have been affected by schistosomiasis, including 40 million women of reproductive age and approximately 600– 779 million individuals are at risk of becoming infected. The mortality rate has been estimated at 280,000 deaths annually in Sub-Saharan countries (Cioli et al., 2014).

The parasitizing of this infectious disease results in fever, malaise, abdominal pain, and skin rashes in an acute state, while intestinal, liver, urinary tract and lung diseases are the result of chronic infection. Acute and chronic disease is solely reliant on the type of species that infects an individual. Reappearance of schistosomiasis over latent periods can result in blockage of the urinary tract and pulmonary hypertension that can lead to fatal complications. In addition, schistosome infection promotes the severity of infection with additional pathogens such as; Plasmodium falciparum, Toxoplasma gondii, Leishmania spp., Mycobacteria, Staphylococcus aureus, Salmonella, and Entamoeba histolytica (Abruzzi and Fried, 2011).

The incidence of schistosomiasis is predominant in Sub-Saharan Africa, and with the increasing rate of infection, due to climate change and other socio-economic factors. To date, PZQ remains the only drug for the treatment of this debilitating disease. PZQ has the following benefits: (1) its effective against all forms of Schistosomes, (2) it is inexpensive and readily available and (3) it has a low side-effect profile, well tolerated in patients of all ages. Unfortunately, the use of PZQ is limited by the following: (1) drug resistance, (2) poor patient compliance to treatment in certain populations, (3) its ineffective against immature forms of the Schistosoma species and (4) it cannot prevent re-infection of Schistosomiasis. Furthermore, there is an increase in parasite alteration and modification, the global parasite load and coinfection with several strains of Schistosoma parasites (Caffrey, 2007; Doenhoff et al., 2008; Fenwick et al., 2009). Coupled with cases of cerebral schistosomiasis in some regions globally, there is an urgent need for an alternative anti-schistosomal drug molecule or to improve the delivery efficacy of PZQ using approaches such as nanotechnology to achieve targeted anti-schistosomal therapy, for example in the CNS.

There has not been a considerable impetus placed on developing novel and new drug treatments for schistosomiasis. However, based on the debilitating impact of the disease, researchers need to be alerted on exploring several essential target proteins found in the Schistosoma species and could play a significant role in ensuring the possibility of designing new drug molecules for schistosomiasis (Garcia-Salcedo et al., 2016). In the absence of any meaningful drug discovery programs for identifying new drug targets and molecules for schistosomiasis, pharmaceutical researchers have turned to providing more efficacious delivery systems for the gold-standard drug PZQ. Hence, nanotechnology and the use of nano-enable drug delivery systems (**Figure 1**), has been a major focus to potentially provide better treatment outcomes for schistosomiasis using PZQ (Veerasamy et al., 2011). Nano-enabled drug delivery systems can enhance the bioavailability and therapeutic efficacy of PZQ (or other drugs) and reduce the side effect profile by having more targeted drug delivery. Nanoparticulate systems currently researched involve, but are not limited to, lipid-based nanoparticles (liposomes, micelles, solid lipid nanoparticles, nanostructured lipid carriers and nanodiscs). Others include polymeric-based nanoparticles (nanospheres, nanocapsules, nanofibers/nanotubes, nanodiscs and micelles), metallic/inorganic-based nanoparticles (nanospheres, nanocapsules, nanodiscs and nanowires/nanotubes) and metal nanoparticles; fabricated by green chemistry (gold, silver, copper, platinum, palldium and zinc nanoparticles).

The emergence of smart LBNPs (**Figure 2**) has offers secure platforms for the use of nano-biomaterials in medical applications such as encapsulation of therapeutic drugs for the targeted delivery of drugs for the treatment of diseases in biomedicine. Recently, the use of LBNPs has gained much interest, particularly in treating schistosomiasis, due to a better absorbed tegument of the schistosomes, which has an affinity for the phospholipid bilayer. LBNPs amphipathic nature allows them to play a pivotal role in the solubility modification and rate at which drugs such as PZQ can be targeted, for enhancing drug absorption across biological barriers (Cheng et al., 2017). Furthermore, targeted LBNPs can improve the efficacy and specificity of drugs to cells or tissues by upregulating surface molecular receptors such as antigens, unregulated selectin and serpin enzyme complex-receptor (Cheng et al., 2017). The Schistosoma parasite consists of different molecules that are found on the surface of the parasite tegument, which are needed for the parasite survival. This is a largely unexplored approach for targeted drug delivery in anti-schistosomal therapy. To this end, nanotechnology has played a central role in the design of systems intended to target the parasite tegument. Hence, this review aims to provide a concise incursion into the molecular structure and function of the schistosome tegument and assimilate the potential targeting proteins/molecules on the tegument to identify new targets and targeting molecules in anti-schistosomiasis therapy.

# OVERVIEW OF THE PAST AND PRESENT ANTI-SCHISTOSOMIASIS THERAPIES

In 1984, the WHO Expert Committee proposed chemotherapy as the best treatment approach to eliminate schistosomiasis (Conlon, 2005). Ever since, chemotherapy continues to be the only measure for the control of schistosomiasis and depends only on a single dose treatment with PZQ. Among other antischistosomal drugs that have been explored, PZQ is the most widely used. PZQ is active against all forms of Schistosoma species that cause schistosomiasis. It reduces the parasitic load and is able to reduce the severity of symptoms. It is also the most preferred drug because of its simple administration,

Adekiya et al. Targeted Schistosomiasis Therapy

efficacy and affordability. Although, the mechanism of action in treating schistosomiasis is not well understood, a widely proposed mechanism is the immediate alteration in the worm musculature. This was reported by Pax et al. (1978) when they noticed that the alteration in the worm musculature causes contraction probably due to rapid influx of Ca2<sup>+</sup> into the schistosome. This assertion was corroborated by interesting work undertaken by Kohn et al. (2001) that drew attention to the voltage-gated calcium channels of schistosomes as the potential target for PZQ. In their study, the mechanism of action for PZQ was suggested to be consistent with the observed effects of PZQ on Ca2<sup>+</sup> homeostasis in schistosomes. It was noted that β-subunits of schistosome channels had a unique form of β-subunit structure that was different from other common β-subunits which inhibit flow of current through the α<sup>1</sup> subunit of schistosome with which they are associated. The study further hypothesized that PZQ facilitated the opening of more channels for current to flow leading to the disruption of α1/β interaction in these channels resulting in disruption of Ca2<sup>+</sup> homeostasis (Kohn et al., 2001). It has also been reported that PZQ causes morphological transitions in the schistosomes tegument. This was initially indicated by the formation of vacuoles within the tegument and blebbing at the surface (Becker et al., 1980; Mehlhorn et al., 1981; Cioli and Pica-Mattoccia, 2003). These morphological transitions cause increased exposure of antigens on the surface of the parasite (Harnett and Kusel, 1986). Harnett and Kusel suggested that the action of PZQ on the exposed antigens may be due to its lipophilicity that makes it easier to interact with hydrophobic cores of the tegument.

Due to the shortcomings of the drugs listed in **Table 1**, researchers have resorted to the use of drug delivery technologies such as nanotechnology to provide more targeted therapies to all stages of the Schistosoma parasite such that drugs can be more effective in treating the immature forms of the parasite. These novel approaches can also reduce drug resistance and avert re-infection by clearing the schistosomes in the human host.

### THE SCHISTOSOME TEGUMENT: REVISIT OF THE MOLECULAR STRUCTURE AND FUNCTION FOR TARGETED DRUG DELIVERY

The outer-surface of the schistosome is enclosed with an uncommon structure known as the tegument where some probable receptors for targeted nano-delivery system are found. It is a rare double layered membrane structure that plays a pivotal role in protecting the worm from harsh conditions in the host system. There are several organelles present in the tegument (**Figure 3**). The heptalaminate tegumental surface is enclosed by a typical plasma membrane structure that is superimposed by a secreted membranocalyx (generated by the multi-laminate vesicles found in the tegument cytoplasm) and fuses with lateral channels protruding out into the base of the surface from the cytoplasm which also host some potential proteins for nano-delivery systems. The membranocalyx can be active by


interacting with proteins and glycans via the extracellular loops of the tetraspanins protein, this depicts tetraspanins as a probable target for nano-delivery systems.

The initiation of heptalaminate membrane surface alongside dyneins protein starts from the outer membrane of the cercarial trilaminate 30 min after invasion of cercarial into the host skin, and within 3 h, the change in cercarial membrane from the trilaminate to the heptalaminate mature membrane structure is accomplished in an immature schistosome (schistosomulum) (Mansour and Mansour, 2002), with the help of several molecules which are potential target for nano-delivery systems. The surface spines of schistosomes are made-up of paracrystalline arrangements of filamentous actin, which are found outside the tegument and basal membranes with protruding tip which is above the general level of the tegument. The dorsal surface spines protrude into the endothelial cells surface of the human host blood vessels with the help of different molecular proteins such as; dyneins, SGTP4 and tetraspanins (some of the potential targets for nano-enabled drug delivery systems), where it helps the schistosomes to hold fast against the blood flow when the

schistosomes are living in the host mesenteric blood vessels (Kašný et al., 2017).

The syncytium of the tegument consists of trilaminate vesicles that comprise membranous material, and they are either elongated or spherical in shape and are known as membranous bodies and elongate bodies. The syncytium is linked to nucleated cell bodies (cytons) by cytoplasmic tubes that are coated with microtubules. Although, cytons are not considered to be part of the tegument, they are found under the circular and longitudinal muscle fibers of the tegument where some potential targeted proteins and molecules are secreted. They consist of mitochondria, nuclei, Golgi complexes, ribosomes and glycogen particles (Mansour and Mansour, 2002). The biogenesis of new membrane material including tegumental proteins, and the maintenance of the schistosomes tegument are equipped inside the cytons, under the muscular layer and syncytium vesicles (El Ridi et al., 2017). Though, the process has not been well studied, both the elongated bodies and membranous bodies transport membranous material which are found to be scattered in the cytons in conjunction with Golgi complexes where they are possibly generated. These vesicles carry membranous material and some targeted proteins produced within the cytons and move them via the tubules of the cytoplasm to the syncytium, and thereafter, migrate to the outside of the surface tegument of the membrane (El Ridi et al., 2017; Gobert et al., 2017; Kašný et al., 2017). More so, the incorporation of membrane vesicles content with other tegumental proteins (potential targeted molecules for nano-delivery systems) take place in the cytons where a fresh heptalaminate membrane is produced. This activity is not only restricted to schistosomula developments, also to adult schistosomes after the shedding or the rupture of the exterior schistosomes surface. Surface pits is another organelle found on the surface of the schistosomes tegument. The pits have an ability to increase the surface area of the worm not less than tenfold where it provides an avenue for the worm to absorb nutrients such as glucose and other molecules from the exterior milieu using its tegumental proteins and receptors.

Wendt et al. (2018) developed a new fluorescent schistosome tegument label technique. This proved that the schistosome parasite usually repairs and replaces the tegument continuously with a half-life of 5 days in order to survive harsh conditions in the host system (Wendt et al., 2018). This corroborated with the work of Wilson and Barnes (1977), where they showed that there is a possibility for the membranocalyx of the schistosome tegument to replace itself at a variable rate and was dependent on the external environmental conditions in which the worm was found. This view was supported by Perez and Terry (1973) where they observed that the surface of the schistosome was turning over more rapidly when schistosomes cultured in monkey antimouse serum were selected from the schistosome mice model.

This unique membrane structure allows the schistosome tegument to play a significant role in protecting the schistosome to survive in the host, some of which include; host response modulation that causes host immune response evasion (Elzoheiry et al., 2018). This immune evasion occurs by rendering the infected host's antibody responses ineffective, hence fails to clear the established parasites (Han et al., 2009). In addition, the tegument of the schistosomes has some other functions such as absorption of nutrients.

Trematodes have an incomplete digestive tract, and the Schistosoma species can survive under prolonged in vitro incubation in the absence of nutrient absorption within the

intestine (Isseroff and Read, 1974; Popiel and Basch, 1984). Glucose absorption in trematodes is noticed during immature stages of the trematode's life cycle, which lacks a developed intestine (Uglem and Lee, 1985). Both, immature and mature schistosomes rely mainly on plasma glucose from the host for energy. Physiological investigations demonstrated that the influx of glucose within the tegument occurred by a carrier-mediated process (Rogers and Bueding, 1975; Uglem and Read, 1975). Several enzymes function for the absorption of amino acids on the tegument (Skelly et al., 2014), and other enzymes, for instance, leucine aminopeptidase is absent in the gut. Cholesterol is also acquired by schistosomes from the host via the tegument where it is redistributed throughout the schistosomes body (Haseeb et al., 1985; Popiel and Basch, 1986).

In terms of parasite motility control, the mature Schistosoma species move based on several degrees of flow and confinement. Zhang et al. (2019) observed that movement mechanics of schistosomes may be an important factor for the specific morphological qualities of an adult male worm, some of which include tegument topography and the strength as well as the nature of its suckers. In addition, the regulation of osmotic and electrochemical gradients of the worms is also control by the tegument. Faghiri and Skelly (2009) revealed that the schistosomes tegument controlled the movement of drug molecules and water into the parasites. This highlighted the role of the tegument in the uptake of drugs and in the osmoregulatory control of the parasite. The tegument also controls the excretion of certain metabolic products like amino acids, lactate, NH<sup>4</sup> <sup>+</sup> and H<sup>+</sup> (Faghiri et al., 2010; Skelly et al., 2014).

# POTENTIAL MOLECULAR TARGETS IN THE SCHISTOSOME TEGUMENT

There are several targets which have been identified on the surface of the tegument (**Table 2**). These are essential for engineered drug-loaded nanoparticles to target SGTP1 and SGTP4 as well as AChE and a nicotinic type of acetylcholine receptor (nAChR) that are predominantly found on the surface of the male schistosomes tegument. Other major surface proteins found on the tegument that can be targeted include dynein, aquaporins and tetraspanins among others. These molecules located on the surface of the tegument can serve as major molecular targets for the design and development of novel drug molecules and vaccines against the Schistosoma parasite.

# Glucose Transporters as a Potential Molecular Target for Nano-Delivery Systems

Several studies (Skelly et al., 1994, 1998, 2014; Skelly and Shoemaker, 2000; Krautz-Peterson et al., 2010) have shown that Schistosoma parasites rely on energy (glucose) to survive. Energy (glucose) is consumed by the tegument first and not by the intestinal cecum. Uptake within the tegument is facilitated by the glucose transporters found on the tegument (Skelly et al., 1994, 1998). Skelly et al. (1994) isolated and characterized three different cDNAs with predicted protein sequences that indicate a high degree of structural and sequence similarity to that of facilitated diffusion transporters in animals, bacteria and plants. It was discovered that two cDNAs encoded for two different glucose transporters in the tegument namely SGTP1 and SGTP4. In addition, the study described that the SGTP 1 and 4 genes are expressed in adult and larval female and male schistosomes to facilitate the uptake of glucose from the host. In another related study by Cabezas-Cruz et al. (2015) four glucose transporters were encoded in the Schistosoma mansoni genome and only two out of the four facilitated glucose diffusion. Their results further proposed that Schistosoma mansoni class 1 glucose transporters failed to carry glucose and that this function developed independently in the schistosomes-specific glucose transporter.

There is a dynamic difference from the glucose transport of the platyhelminthes-specific transporters of the schistosomes when compared with humans (Cabezas-Cruz et al., 2015). It has been shown that the sequence of SGTP1 and 4 are 60% similar. Zhong et al. (1995) used electron microscopy to map the various locations of the transporters on the tegument. They observed that localization of SGTP1 was at the basal lamina and to a lesser extent under the muscle cells. This may help in transporting free glucose inside the tegument into the interstitial fluids that paddle the interior organs of the parasite. It was also demonstrated that SGTP4 was evenly distributed on the dorsal and ventral surfaces of female and male teguments with an extraordinary structure of a double lipid bilayer (Zhong et al., 1995). The distinct location of SGTP4 on the outer tegumental membrane reveals that SGTP4 facilitated glucose transport into the parasite tegument from the host bloodstream (Skelly et al., 1998; Skelly and Shoemaker, 2001). In addition, SGTP4 is involved in the development of the free-living cercariae into schistosomula. Through maturation they satisfy the needs of the parasite for high glucose uptake as soon as they enter the host (schistosomula stage) and throughout adulthood (Skelly and Shoemaker, 1996; Skelly et al., 1998).

Thus, proposing SGTP proteins as a potential target for nano-delivery systems, this postulation was supported by Krautz-Peterson et al. (2010) where RNAi was used to knock down the upregulation of SGTP4 and SGTP1 genes in schistosomula and in the life stages of adult worms. This study was undertaken to investigate the significance of these proteins to the parasite. Downregulation of either SGTP4 or SGTP1 displayed impairment in the ability of the protein to transport glucose when compared with the control. The study further showed that the simultaneous downregulation of both SGTP1 and SGTP4 reduced the ability of the parasite to transport glucose when compared with a single downregulated SGTP gene. It was also demonstrated that none of the parasites exhibited phenotypic distinction after prolonged incubation of all the suppressed parasites in enriched medium when compared to the control. Finally, it was suggested that SGTP1 and SGTP4 were important for transporting exogenous glucose from the mammalian host for normal parasite development. This was based on the observation that parasites with suppressed SGTPs showed decrease viability in vivo after infection of experimental animals (Krautz-Peterson et al., 2010). This notion was supported by a study performed by

TABLE 2 | Different potential targets found on the schistosomes tegument for conjugated nanoparticles and their functions.


McKenzie et al. (2017), where the uptake of glucose was regulated in Schistosoma mansoni by Akt/Protein kinase B signaling. It was observed that Akt can be triggered by the host L-arginine, more so, insulin was shown to be effective in the layer of adult and schistosomula teguments. The inhibition of Akt decreased the upregulation and development of SGTP4 at the exterior of the host-invading larval stage of the parasite. The suppression of the SGTP4 upregulation at the tegument in adult worms was associated with a decrease in glucose uptake.

Hence, the functionalization of nanoparticles with targeted agents (antibodies, aptamers, antibody-like ligands, peptides and small molecules) with high specificity to SGTP proteins may be a superior alternative to anti-schistosomal treatment to nano-enabled the delivery of anti-schistosomal drugs. In achieving the desired selectivity of drug delivery, nanotechnology has allowed researchers to design nanoparticulate systems and incorporate therapeutic drugs to acts as nanocarriers. This is due to the overexpression of receptor molecules (SGTP proteins) which can serve as docking/interacting sites for targeting potential therapeutic drugs. Theoretically, the therapeutic drugs can be concentrated in a specific site in organ and tissues by functionalizing drug-containing nano-delivery systems with ligands against the receptors. Thus, nano-delivery systems with ligands specific to SGTPs as a receptor can be a potential target for designing, developing and delivering of anti-schistosomal drug.

#### Acetylcholine (nAChRs), AChE and Nicotinic Receptors; Possible Targets for Nano-Delivery Systems

Acetylcholine (ACh) is an essential neurotransmitter, both in invertebrates and vertebrates. The neuromuscular consequences of ACh are normally mediated by postsynaptic nAChRs due to their high-affinity for nicotine. Based on the structure of nAChRs, they belong to the Cys-loop LGIC superfamily (Albuquerque et al., 2009; MacDonald et al., 2014). nAChRs generate hetero and homo-pentameric structures that are arranged in a barrel shape around a central ion-selective hole. nAChRs in invertebrates are anion and cation-selective (Cl2) ACh-gated channels while in vertebrates nAChRs are cation-selective (Ca2+, Na+, K+) and facilitate excitatory responses.

Both the nicotinic type of the acetylcholine receptor (nAChR) and AChE are potential target for nano-enabled drug delivery system, because they are both found on the exterior surface of the

tegument where they play an essential role in the schistosomes ion channels and nervous system (Mansour and Mansour, 2002; MacDonald et al., 2014). AChE and nAChR are predominantly found on the surface of adult male schistosomes. The adult female schistosomes usually lodges in the gynaecophoral canal of the male containing a lower number of these proteins. AChE has been shown to have glucose scavenging modulatory activity from the human host bloodstream. The uptake of glucose is controlled by the interaction of ACh with the nAChR and AChE on the surface of the tegument (Camacho and Agnew, 1995). It has also been discovered that the exposure of low concentrations of ACh to S. haematobium or S. bovis and not S. mansoni improved the uptake of the glucose by the parasites in the host blood. At higher concentrations of ACh, the uptake of glucose in the host by parasites was inhibited. This specificity between the nicotinic receptor and ACh was supported by showing the effect of α-bungarotoxin and D-tubocurarine as antagonists to ACh. Therefore, it is significant when instituting a nanotechnology approach to deliver antagonistic drug molecules to the binding sites of nAChR and AChE in order to inhibit their glucose scavenging activities from the host bloodstream.

### Dyneins as a Possible Molecular Targets for Nano-Delivery Systems

Dyneins is a protein that produces force and movement on microtubules for biological processes such as ciliary beating, intracellular transportation and cell division, it performs these functions through the help of ATP hydrolysis (Roberts et al., 2013). Dyneins could serve as a possible target for nanodelivery system in the treatment of schistosomiasis owing to its biological function in the survival of the parasite. Several studies have employed immunostaining to identify various microtubule related proteins inside the schistosomes tegument such as actin, tubulin, paramyosin, and dyneins. Studies have suggested that cytoplasmic dyneins may have a role to play in transporting of vesicles to the surface bilayers and tegument cytoplasm from the sub-tegumental cells. Dynein chains are part of a huge enzyme complex comprising heavy, intermediate and light chains. Dyneins are implicated in the assembly of spindles that are used for chromosome movement in mitosis. The upregulation of dyneins are involved in the developmental of S. mansoni. In addition, they are found in the schistosomula stage that occurs after the penetration of the intact skin of the host by the parasite and at the lung stage in adult worms. At this stage, early upregulation of the heptalaminate exterior membranes are exhibited. Meanwhile, dynein light chains are not found in the cercariae or ciliated miracidia (Hoffmann and Strand, 1996).

The dynein light chain protein discovered recently was shown to have high affinity to other proteins tegument with which they form highly complex associations. Another dynein light chain protein has been considered as a tegument antigen with the molecular weight of 20.8 kDa. Githui et al. (2009) investigated the normal motor constituents of vesicular transport present in the schistosomes tegument. The NCBI database blast search analysis recognized clones that are myosin and dynein light chains genes. After subjecting the genes of schistosome dynein to further analysis in the databases, they detected three dynein light chains families. They also observed that the Tctex family sequences of the dynein light chains are different significantly when compare to the mammalian homologs, Hence, could serve as probable drug/vaccine target against schistosomes infection. The three dynein light chains, S. japonicum dynein light chain-1, S. mansoni dynein light chain and SM10 studied via the immunolocalization of microtubule-related motor protein components show a specific and strong immunolocalization in the distal cytoplasm of the tegument (Kohlstädt et al., 1997; Yang et al., 1999). The tegument-associated protein of the S. japonicum which has 22.6 kDa displays similar localization arrangements (Li et al., 2000). In view of these aforementioned roles of dyneins in schistosomes survival, the delivery of targeted drug to localize and bind to dynein using nanotechnology approach will be a potential technique in eradicating schistosomiasis. Nanotechnology-based targeted delivery system functionalized with specific targeted molecules (antibodies, peptides, antibodylike molecule and aptamer) can recognize and selectively bind onto the dynein protein (receptors) on its active region thereby conferring targeted delivery.

### Aquaporins as a Potential Molecular Targets for Nano-Delivery Systems

Aquaporins is another promising target for the delivery of surface-engineered drug-loaded nanoparticles. They are small integral membrane proteins that are mostly upregulated in animal and plant kingdoms. Aquaporins consist of two short and six transmembrane helical segments that enclose cytoplasmic and extracellular vestibules linked by a narrow aqueous pore. They consist of several conserved motifs, and aquaporin monomers are assembled as tetramers in membranes, with every monomer working independently (Verkman, 2013). Aquaporins act as channels to selectively control the influx and efflux of water molecules within cells. Certain aquaporins allow the diffusion of metabolites in and out of the cell (Tsukaguchi et al., 1998, 1999; Gonen and Walz, 2006).

The nano-enabling drug delivery activity of aquaporins was corroborated by Braschi et al. (2006a) where a proteomic study of the schistosome tegument was described. The presence of aquaporins was revealed on the surface of the tegument which indicated that aquaporins assisted with the influx of water and solutes within the plasma membrane of the schistosomes. The tegument (S. mansoni) as an excretory organ was investigated by Faghiri et al. (2010), where they observed that aquaporins on the surface of the tegument acted as a lactate transporter. In addition, it was also shown that the aquaporin found on the tegument was competent in transporting mannitol, water, alanine and fructose, but not glucose. Their further analysis of the tegument using immunofluorescent and immune-EM suggested that the function of the tegument was far above the known ability as an organ of nutrient uptake, but rather, it also helped in excretion of waste metabolites (Faghiri et al., 2010). The study supported the notion that the tegument controlled the osmoregulation and drug uptake in parasites (Faghiri and Skelly, 2009). It was also shown that the existence of aquaporins on the tegument controlled the

movement of water following the suppression of S. mansoni aquaporins with iRNAs (Faghiri and Skelly, 2009).

It has also been shown that aquaporin-4 (a homolog of aquaporins) enhanced the granulomatous response with an increase in the accumulation of macrophages and eosinophils around the S. japonicum eggs in the liver of the mice model. Similarly, the study showed that aquaporin-4 mice enhanced Th2, but decreased Th1 and Treg cell formation in S. japonicum., This accounts for the improvements of the liver granuloma formation (Zhang et al., 2015). These findings collectively indicate that aquaporins may be a desirable target for antischistosomal therapy using high precision delivery of drugloaded nanostructures.

# Tetraspanins as a Potential Molecular Target for Nano-Delivery Systems

Tetraspanins (TSPs) is a family of integral membrane proteins expressed by schistosomes, found in the exterior surface of the membrane of the schistosomes tegument. Braschi et al. (2006a) identified five tetraspanins in the schistosomes membrane surface, and the abundant components of these proteins are found in the tegument periphery. They speculated that the schistosome tetraspanins play an important role in the structure of the schistosomes plasma membrane, based on their analogy with other organisms (Braschi et al., 2006a). They also showed that, some tetraspanins are recognized more readily than others, and the concentrations and locations of only three biotinylated are suggested to vary within the surface of schistosomes tegument. The capacity of tetraspanins homologous interaction to generate a tetraspanin web may help scaffold organization in the lipid bilayer upon which there are assemblage of other proteins within the tegument. More so, the extracellular loops of the tetraspanins may provide platforms for gylcans and proteins which interact with the membranocalyx (Braschi et al., 2006a).

The functions of tetraspanins in the tegument of S. mansoni was investigated with the inhibition of the upregulation of Smtsp-1 and Sm-tsp-2 mRNAs using RNA. The ultrastructural morphology of mature schistosomes treated with Sm-tsp-2 dsRNA, show a thinner tegument and there is a visible formation of vacuoles on the schistosomes tegument. More so, schistosomula exposed to Sm-tsp-2 dsRNA showed a drastic thinner and extensive vacuolated tegument, and this morphological observation depends on failure of tegumentary invaginations (Tran et al., 2010). In another related study by Sotillo et al. (2015), it was reported that tetraspanins were found in biotinylated and unbound tegument tissues. It was also reported that tetraspanin-2 found in S. mansoni is essential for the formation of the schistosomes tegument and is a target of protective immunity in naturally resistant human and vaccinated mice. On the other hand, S. mansoni tetraspanin-1 are detected on the apical membrane of schistosomula. Tetraspanin-2 was found only in the unbound sample, which corroborates with other findings, which shows that the localization of tetraspanin-2 within the inner compartments of the schistosomes; relates with the exterior invaginations and vesicles in the tegument (Sotillo et al., 2015). Targeted nanocarriers has three main components that is; as a targeting moiety-penetration enhancer, an apoptosis-inducing agent and also, as a carrier. Therefore, the inhibition of tetraspanin by the means of nanotechnologybased approach will stop the interaction of glycans and proteins to the schistosomes membranocalyx, because, nanomaterials can preferentially accumulate in the parasite via tetraspanin in an active targeting mechanism thereby, release the encapsulated drugs in a regulated manner. This will provide the benefits of increasing the anti-schistosomal drug concentration and its therapeutic efficacy.

# Other Potential Molecular Targets for Nano-Delivery Systems

Several studies have used proteomic in identifying constituents found within the tegument of schistosomes which potential targets for nano-delivery systems are. Braschi et al. (2006a) used proteomics to detect molecules found within the S. mansoni tegument. In their study, they identified transporters for sugars, inorganic ions, amino acids and water, which indicated that the tegument plasma membrane was crucial for schistosomes to acquire nutrients from the host and help maintain solute levels. They also identified enzymes such as esterases, phosphohydrolases and carbonic anhydrase with their catalytic domains found in the outer core of the plasma membrane, more so, annexin, five tetraspanins and dysferlin were shown to play a pivotal role in the architecture of the membrane. The study was corroborated by another proteomic analysis of S. mansoni proteins that was performed in the same year by Braschi et al. (2006b) not less than fifty-one (51) proteins were identified based on homology with known proteins in other organisms. Some of the identified proteins were enolase which involves energy metabolism; several cytoskeletal and molecular motor proteins such as severin, actin and dynein light chains. Others include molecular chaperone heat shock proteins 17, 19, and 20, calmodulin; vesicle proteins, and plasma membrane transporters; mitochondrial proteins for example ATP synthase; structural molecules and enzymes such as glucose transporter protein, calcium ATPase, annexin, alkaline phosphatase and tetraspanins A, B, and C (Braschi et al., 2006b).

As shown in **Figure 4** Sotillo et al. (2015) used the same approach to detect novel therapeutic targets for nanocarriers in S. mansoni schistosomula. Over 450 proteins were detected on the apical membrane of S. mansoni schistosomula, in which the expression of 200 have significant controlled profiles at diverse stages of schistosomula development in vitro which are potential targets for nano-delivery systems, such as glucose transporters, heat shock proteins, sterols, antioxidant enzymes and peptidases. In addition, current vaccine antigens were also detected on the apical membrane such as calpain, Sm-TSP-1 or Sm-TSP-2, Sm29 found on sub-tegumental fractions of the schistosomula showing localization patterns that differ in some instances from those found on the adult stage of the worm. Another study used S. mansoni genome project, concurrently with proteomic and lipidomic approaches, which allowed the study to characterize the lipids and proteins within the tegument plasma membrane in details. This study detected

some tegumental targeted proteins and lipids, which depicts the role of the tegument in the uptake of nutrients from the host, and in the evasion of immune response. Furthermore, the study demonstrated that the tegument of the worm is enriched in lipids which are not found in the host. Similarly, the schistosome tegument possess proteins which have no sequence similarity

with any other sequence found in databases of species excepts in schistosomes (Van Hellemond et al., 2006). Several other studies (Liu et al., 2006; Pérez-Sánchez et al., 2006; Mulvenna et al., 2010; Castro-Borges et al., 2011; Sotillo et al., 2016, 2019) have employed proteomics technique in identification of several molecular receptors which are druggable and vaccine candidates for schistosomiasis treatment.

To date, there is no effective vaccine for schistosomiasis. Although, several potential promising vaccine candidates for S. mansoni and, to a lesser extent, S. haematobium have been discovered and published in literature. There is one vaccine, namely, BILHVAX, or the 28-kDa GST from S. haematobium, which has entered clinical trials (Capron et al., 2005). Unfortunately, published data is not available on the clinical efficacy of this vaccine, but nonetheless, it is of concern that other vaccines have not progressed to this stage. More so, there are several nanotechnology approaches in developing vaccines for schistosomiasis published in literature, but have not entered clinical trials. Some include oral vaccination with chitosan nanoparticles loaded with plasmid DNA encoding a Rho1-CTPase protein of S. mansoni (Oliveira et al., 2012). Another approach includes a novel nanoparticle formulation of the Sj26GST DNA vaccine; although there was no significant reduction in worm burden, a highly significant decline in tissue egg burden and the fecundity of female adult worms was reported (Mbanefo et al., 2015).

## AN OVERVIEW OF NANO-DELIVERY SYSTEMS

Nanomedicine is the application of nanotechnology for treatment, prevention, monitoring, and control of biological diseases. In applying nanomedicine in the treatment of diseases, the precise targets (cells and/or receptors) specific to the clinical disease is identified and the suitable nanoparticles for the delivery system to minimize the side effects and improve the efficacy of the original drug is selected. One of these precise targets are macrophages, endothelial cells, proteins, dendritic cells as well as tumor cells. Some typical examples of nano-delivery systems (**Figure 5**) used over the years in the treatment of diseases includes; liposomes, micelles, dendrimer, polymeric nanoparticles, polymeric micelles, metallic nanoparticles, nanotubes and nanocrystals.

Morphological characteristics such as rigidity, size and aspect ratio play a vital role in, and affect the impact and fate of nanocarrier properties in vivo (Wang et al., 2018). The properties of nano-delivery systems are critically dependent on the morphological characteristics of the particle, and it is of importance to deliver drug into a specific site during the treatment of disease, such as the delivery of an antitumor drug into the site of a solid tumor (Wang et al., 2018). The characterization of the nanoparticles morphology and dimensions can be determine using SEM, TEM, and AFM. Although, the most appropriate technique depends on the sample type and the desired information to be measured and in some cases, researchers usually adopt techniques which are available and well-known to them in a characteristic dimension of the nanoparticles. A typical example of TEM, SEM and AFM of nanoparticle are shown in **Figure 6**.

Several studies have explored the use of nano-delivery system in improving the therapeutic efficacy of different drug molecules in the treatment of diseases. Mehrizi et al. (2018) carried out the synthesis of a novel nanosized chitosan-betulinic acid delivery system, against resistant Leishmania, with the first clinical observation of this parasite in the kidney. It was discovered that chitosan nanoparticles synthesized using phase separation; and drug loading by phase separation, improved the therapeutic dose of betulinic acid to 20 mg/kg. More so, the successful improvement in the

et al. (2019).

use of the nanosystem loaded betulinic acid in the treatment of leishmania, displayed both in in vitro and in vivo efficacy (Mehrizi et al., 2018).

The effectiveness of IVM was investigated using nanostructured lipid carriers in the treatment of hydatidosis, with some limitations and resistance associated with the drug overcome by the carriers in in vitro experimentation. It was observed that NLCs-loaded IVM induced higher mRNA caspase-3 expression which suggested a more potent apoptotic effect on the parasite (Ahmadpour et al., 2019). In another related study, nucleoside-lipid-based nanocarriers was used to encapsulate MB; a positively charged tricyclic phenothiazine molecule used in malaria treatment. This approach showed that the nanoparticles partially protected MB from oxido-reduction reactions, thereby preventing early degradation during storage, and the carrier also prolonged the pharmacokinetics in plasma.

It was concluded that this approach was an interesting technique in improving MB stability and the delivery in malaria treatment (Kowouvi et al., 2019).

Hence, the utilization of lipid nanoparticle-based drugs in the treatment of schistosomiasis will be beneficial in terms of cost, since solid lipid nanoparticles are easy to scale-up and involves lower cost production, relative non-toxic nature, biodegradable composition and stability against aggregation. More so, lipidbased formulations have the ability to enhance the bioavailability of drugs through solubility modification and the rate at which drugs can be released for the improvement and enhancement of drug absorption across biological barriers (Cheng et al., 2017), reducing side-effects associated with these drugs. This type of approach will be beneficial and effective in treating all forms of schistosomes (both mature and immature), by functionalizing the nanoparticles with targeted molecules which has ability to recognize and bind to molecular receptors present in all forms of schistosomes. Thus, preventing reinfection by specifically targeting overexpressed schistosomes antigens present in the human host, it has been reported that nanoparticles have the ability to induce heightened T cell immunity, which can prevent disease reactivation and reinfection (Tousif et al., 2017). The list of various nano-delivery systems used in improving the therapeutic efficacy of PZQ in the treatment of Schistosomal infections are reported in **Table 3**.

#### TARGETED NANO-ENABLED DRUG DELIVERY

Targeted nano-enabled therapies are able to recognize or detect molecules that are highly expressed on the surface of specific cells. This approach has gained popularity in treating various cancers due to the overexpression of specific receptors on the membrane surface of cancer cells. In the field of cancer, targeted nanotherapies inhibit particular cell surface proteins or genes which are responsible for cancer growth and metastasis. It has been hypothesized that targeted nanotherapies may be desirable over other forms of treatment (Joo et al., 2013; Camidge, 2014). According to a 2018 review published by the ACS, targeted nanotherapies have been approved for various anticancer therapies. Thus, employing a nanotherapeutic approach to target overexpressed proteins or genes on the surface of the schistosome tegument will assist in overcoming PZQ resistant, reduce the burden of immature schistosomes (schistosomula), and finally, put an end to the morbidity and mortality of schistosomiasis (**Figure 7**). More so, this approach and can be employed in the treatment of various parasitic infections. Although there are no reports of targeted nano-enabled drug delivery against Schistosoma species to date; there are few reports of this type of approach on other similar parasites such as; the preparation of the primaquine-containing liposomes functionalized with covalently bound heparin for the targeted delivery of antimalarial drugs to pRBCs.

Heparin covalently linked with targeted nano-enabled drug delivery to pRBCs was carried out to reduce anticoagulation risks. The study showed that heparin-based targeting can be designed to have a greater half-life, further relying on antibodies with exposed antigens, whose expression is constantly modified by successive generations of the parasite (Marques et al., 2017). Further to this, targeted nano-enabled drug delivery of a 19-amino peptide from the circumsporozoite protein of Plasmodium berghei which contained a conserved region, as a consensus heparin sulfate proteoglycan-binding sequence attached to the distal end of a lipid-polyethylene glycol bioconjugated, was prepared by the incorporation into phosphatidylcholine liposomes, reflecting favorable in vitro results (Longmuir et al., 2006).

Jain et al. (2014) developed a chitosan-assisted immunotherapy for the intervention of experimental leishmaniasis via amphotericin B -SLPs to activate the macrophages in order to impart a specific immune response by improving the production of TNF-α and IL-12 (Jain et al., 2014). This study also reflected a positive hypothesis for targeted nano-drug delivery for site specific targeting.

# Antibody-Functionalized Drug Delivery Systems for Targeted Therapy in Schistosomes Infection

Antibodies are mostly IgGs or their fragments and have ability to recognize and interact virtually with any molecular target with high affinity and high specificity. Antibodies have gained special interest in targeted therapies due to their nanosized. They are biological materials which are part of the specific immune system, and they are toxin or pathogens neutralizers in nature. They help in recruitments of immune elements such as; improving phagocytosis, complement, cytotoxicity antibody dependent by NK cells. They can also help in carrying several elements such as; toxins, nanoparticles, drugs and fluorochroms to where they can be used in therapy to destroy a specific target, and for several other diagnostic procedures (Arruebo et al., 2009). Antibody-functionalized nanoparticulate systems are more sitespecific, causing higher accumulation on the target region and subsequently, reduces dosage requirements.

The bioconjugation of antibodies with nanoparticles to generate a unique product which is composed of both the properties of the antibodies and nanoparticles can take place by adsorption process that is, at the isoelectrical point of the antibody through electrostatic interaction (Arruebo et al., 2009; Greene et al., 2018). More so, the conjugation can take place by direct covalent bonding between the surface of the antibody and the nanoparticles (that is, coupling of the antibodies to nanoparticles by free carboxyl or amine functionalities on aspartic/glutamic acid or lysine residues or by thio-maleimide reaction) (Arruebo et al., 2009). Another means by which the bioconjugation of the antibodies to the nanoparticles can be achieved is through the use of adapter molecules that is, bio-recognitions like streptavidin and biotin, which usually involves the formation of the complex. Greene et al. (2018), developed a novel approach for the site-specific conjugation of nanoparticulate systems which promotes the uniform and outward projection of paratopes for utmost target interaction. They demonstrated a successful re-bridging of the inter-chain



disulfide linkage with a heterobifunctional linker and successive coupling to nanoparticles bearing complementary azide moieties in TRAZ F(ab) model. In a study by Mohammad et al., the administration of the antimalarial drug (chloroquine) loaded liposomes, targeted to infected RBCs with a tagged antibody against infected erythrocytes surface antigens on the chloroquine liposomes against drug-resistant Plasmodium berghei, presented a cure rate of 75–90% on days 4–6 post infection in mice (Mohammad et al., 1995).

Secret et al. (2013) described the preparation of antibodyfunctionalized biodegradable porous silicon nanoparticles loaded with the hydrophobic anticancer drug camptothecin. Using a novel semicarbazide based bioconjugation technique in chemistry, the specific orientation of the immobilized antibody on the nanoparticles was achieved. Three antibodies mAb528 a monoclonal antibody to EGFR; MLR2 a monoclonal antibody to p75NTR and Rituximab a monoclonal antibody to CD20 were used to target glioblastoma, neuroblastoma and B lymphoma cells, respectively in an in vitro study. The successful targeting was demonstrated by means of immunocytochemistry and flow cytometry both with cell lines and primary cells. The incubation of the antibody-functionalized nanoparticles with the cell lines for cell viability, showed selective killing of cells expressing the receptor, which correspond to the antibody coupled on the porous silicon nanoparticles. Also, the incorporation of camptothecin an anticancer drug into a nanoparticle functionalized with the antibodies showed to be very effective and efficient in targeting and killing cancer cells (Secret et al., 2013). In another recent study by Li et al. (2019), they developed an antibody-functionalized gold nanoparticle (cetuximab-AuNP) to selectively target cancer cells and probe for their potential radiosensitizing impacts under proton irradiation. It was discovered that cetuximab-AuNP interacts and bound specifically and accumulate in EGRF-overexpressing A431 cells when compared with EGFR-negative MDA-MB-453 cells. It was further shown that, cetuximab-AuNP improved the influence of proton irradiation in A431 cells but not in MDA-MB453 cells (Li et al., 2019). There are several other studies (Day et al., 2010; Dai et al., 2015; Korkmaz et al., 2016) which have employed antibody-functionalized nanoparticles to selectively target some specific receptors on the cancer cells either for treatments or imaging (diagnostics).

### Aptamer-Functionalized Drug Delivery System for Targeted Therapy in Schistosomes Infection

Aptamers are short single-stranded oligonucleotide (RNA or DNA ligands) or peptides that bind to their target molecules; either small chemicals, large molecular cell-surface or transmembrane proteins with high specificity, affinity and versatility. They have been developed for over two decades against several targets and for different applications. Aptamers have emerged as promising molecules to target specific cancer antigens in therapy and clinical diagnosis (Cerchia and De Franciscis, 2010; Catuogno et al., 2016). Nucleic acid aptamers have gained attention as an attractive molecular vehicle because of their ability to bind to specific ligands with high affinity,

activities.

they have high ability to penetrate cells, tissues and organs, and they also possess high chemical flexibility (Catuogno et al., 2016). Whereas, peptide aptamers, otherwise known as affimers are small stable proteins that are selected to interact and attach with high binding affinity to specific sites (surface) on their target molecules. They contain short amino acid of about 5–20 residues long sequences which are normally embedded as a loop inside a stable protein scaffold (Reverdatto et al., 2015). Aptamer based sensing platforms for the recognition of peptides, small molecules, proteins and cells have gained a huge interest due to their high sensitivity and selectivity. In general, aptamers are molecules that can generate unique 3-dimentional structure and has the ability to bind almost any molecular targets with higher binding affinity in the nanomolar level compared to monoclonal or polyclonal antibody.

Due to aptamers properties such as; high affinity, chemical stability, small size, ease of synthesis, low-immunogenicity and controllable chemical modification, owing to these multiple attributes, aptamer conjugated nanoparticles are well qualified nanosystems for the development of biomedical devices for imaging, analytical, drug delivery and some other medical applications. The bioconjugation of the aptamers onto the nanoparticulate systems can be attained via non-covalent (affinities interaction e.g., streptavidin-biotin or metal ion coordination) and covalent (1-ethyl-3-carbodiimide or succinimdyl ester-amine chemistry and N-hydroxysuccinimide activation chemistry which cross link the carboxylic acid group on the surface of the nanoparticles and the amino group of the ligands) interactions. The covalent interaction strategies can also be achieved by maleimide-thiol chemistry that is, the cross linking of the thiol group on the targeting moiety and the maleimide functional group on the surface of the nanoparticulate systems.

Yu et al. (2011) developed a novel aptamer bioconjugated nanoparticles in order to enhance the delivery of paclitaxel anticancer drug to MUC1-positive cancer cells. The aptamer was engineered into the surface of the nanoparticles via a DNA spacer. The flow cytometry analysis shown the higher uptake of the nanoparticulate systems conjugated with MUC1 specific aptamer into the target cells via the overexpression of MUC1. The results further showed that, paclitaxel loaded aptamer functionalized nanoparticles improved the in vitro drug delivery and cytotoxicity to MUC1 cancerous cells when compared to non-targeted nanoparticulate systems which lack the MUC1

aptamer. In Yang et al. (2013) developed a DNA aptamer envelope protein for the inhibition of hepatitis C virus. In their study, it was shown that selected aptamers for E1E2 particularly recognized the recombinant E1E2 protein and E1E2 protein from hepatitis C virus-infected cells. The aptamers exert antiviral properties via the inhibition of the virus binding to the host cell (Yang et al., 2013). Several other studies (Mathieu et al., 2016; Corda et al., 2018; Dou et al., 2018; Tan et al., 2019) have employed aptamer-conjugate as a targeted delivery system for therapeutics and diagnostics.

# Other Functionalized Drug Delivery Systems for Targeted Therapy in Schistosomes Infection

Other small molecules or peptides that are highly specific for certain molecular receptors with high affinity, can also be screened or developed in order to localize and bind with the molecular receptors found on the tegument of the schistosomes. Lei et al. (2019) designed a novel alendronatemodified nanoparticle loaded with paclitaxel and coated with polydopamine for osteosarcoma-targeted therapy. In this study, it was reported that the polymerization of dopamine in a versatile modification method was not limited by the absence of functional groups on the surfaces of the compound and do not affect the chemical properties. The successful bioconjugation of the polydopamine with nanoparticles with a surface modifier which consist of a precise affinity for osteosarcoma cells was attained. They posit further that, the targeting nano-delivery systems exhibited a higher in vitro cytotoxicity on K7M2 of osteosarcoma cells when compared with the native nanosystems. Furthermore, the in vivo study showed that the targeting nano-delivery systems could accumulate within the tumor to a greater extent with remarkable decrease in the adverse effects of paclitaxel when compared with non-targeted nanosystems (Lei et al., 2019).

Säälik et al. (2019) investigated the effect of a novel penetrating peptide-guided nanoparticles that targets cell surface LinTT1, p32 for glioblastoma targeting. In this study, the coupling of LinTT1 to albumin-paclitaxel nano-delivery systems was achieved by sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate as a linker. They demonstrated that the novel p32 targeting peptide, LinTT1 promotes the targeted accumulation of nanoparticles to tumors across a panel of high-grade glioma mouse model effectively. They further showed that the treatment of mice with LinTT1 guided nanoparticles extend the survival rate of mice with the tumor; due to the ability of LinTT1-nanopaticles to recognize the upregulation of p32 on glioblastoma (Säälik et al., 2019).

Ahlschwede et al. (2019) employed targeted nano-delivery approach in treating cerebral amyloid angiopathy and detecting cerebrovascular amyloid observed in AD. A targeted nanodelivery system was developed by a cationic blood brain barrier penetrating peptide using a covalent bioconjugation technique. The results from the targeted nanosystem depicted a higher significant brain uptake due to the high binding affinity of the peptide (K16ApoE)-nano-delivery system to amyloid plaques. In another study carried out by Colombo et al. (2019) where the targeted biodegradable nano-delivery system for CD34 + endothelial precursors in the treatment of rheumatoid arthritis was achieved. The bioconjugation of the targeting molecule was activated by N-hydroxysuccinimide in order to exploit its primary amino groups. The results in this study showed that, the targeted nano-delivery system possess a greater advantage in delivery the drug to inflamed synovia via the synovium-homing peptide as a targeting molecular receptor.

Silva et al. (2015) carried out mannose-functionalized polymeric nanoparticles to target the mannose receptors on antigen-presenting cells and therapeutic anti-tumor immune responses in a melanoma model. It was discovered that mannosefunctionalized nanoparticles potentiated the Th1 immune activity, and the nanoparticulated vaccines reduced the rate of murine B16F10 melanoma tumors growth in prophylactic and therapeutic settings (Silva et al., 2015). Also, Mufamadi et al. (2013) carried out a ligand-functionalized nanoliposomes for targeted delivery of galantamine in AD. It was shown that ligand-functionalized nanoliposomes enhanced the uptake of galantamine into PC12 neuronal cells through the receptor of Serpin Enzyme Complex (Mufamadi et al., 2013). Ruff et al. (2017) investigated the effect of gold nanoparticles surface engineered with amyloidogenic β-amyloid specific peptides in a BBB in an in vitro model. This study was carried out in order to increase the BBB permeability, as well as the nanoparticle concentration in the brain by the peptides. It was discovered that, the multivalent peptides bind selectively to Aβ-amyloid fibrils, thereby posing a strongly effect on the integrity of BBB, thus, actively cause the transport of the gold nanoparticle conjugates via the BBB.

#### CONCLUSION

Incidences of schistosomiasis continue to increase globally across sub-Saharan Africa and other tropical regions. However, the development of resistance against the only drug PZQ necessitates the design of more effective drug molecules to tackle the continual increase in Schistosomiasis cases. In this review, the molecular structure and function of the schistosomal tegument was described and several molecular targets have been identified to potentially target the schistosomes tegument as a site for enhanced PZQ delivery in antischistosomal therapy. In addition, potential agents that could target the molecular receptors identified have been highlighted. In general, surface functionalization of nanoparticles with antibodies, aptamers, antibody-like ligands, peptides and small molecules to specifically target and bind to the schistosomes tegument receptor genes and proteins presents a viable option for researchers to explore. This approach will suppress the activity of receptor genes/proteins, thereby impairing the ability of schistosomes to import nutrients from the host as well as disrupt the ability of the parasite to maintain solute balancing and evasion of the host immune response. Hence, exploration of the schistosomes tegument may be a possible and potential focus for designing and developing anti-schistosomal drug which can target receptors and proteins present on the worm tegument.

#### AUTHOR CONTRIBUTIONS

fbioe-08-00032 January 30, 2020 Time: 16:53 # 17

All authors contributed to the manuscript completion and approved the final submission. TA, PK, YC, PK, and VP

#### REFERENCES


Camacho, M., and Agnew, A. (1995). Schistosoma: rate of glucose import is altered by acetylcholine interaction with tegumental acetylcholine receptors and acetylcholinesterase. Exp. Parasitol. 81, 584–591. doi: 10.1006/expr.1995.1152


contributed to this study from framework design and manuscript content to manuscript optimization.

#### FUNDING

This work was supported by the National Research Foundation (NRF) of South Africa (Grant name: SARChI).



delivery system loaded with citraland its antiproliferative effect on colorectal cells in vitro. Nanomaterials 9:1028. doi: 10.3390/nano9071028



towards the development of new tools for elimination. PLoS Neg. Trop. Dis. 13:e0007362. doi: 10.1371/journal.pntd.0007362


**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 Adekiya, Kondiah, Choonara, Kumar and Pillay. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Design of Poly(lactide-co-glycolide) Nanocarriers for Medical Applications

#### Divesha Essa, Pierre P. D. Kondiah, Yahya E. Choonara and Viness Pillay\*

Wits Advanced Drug Delivery Platform, Department of Pharmacy and Pharmacology, School of Therapeutic Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

Polymeric biomaterials have found widespread applications in nanomedicine, and poly(lactide-co-glycolide), (PLGA) in particular has been successfully implemented in numerous drug delivery formulations due to its synthetic malleability and biocompatibility. However, the need for preconception in these formulations is increasing, and this can be achieved by selection and elimination of design variables in order for these systems to be tailored for their specific applications. The starting materials and preparation methods have been shown to influence various parameters of PLGA-based nanocarriers and their implementation in drug delivery systems, while the implementation of computational simulations as a component of formulation studies can provide valuable information on their characteristics. This review provides a critical summary of the synthesis and applications of PLGA-based systems in bio-medicine and outlines experimental and computational design considerations of these systems.

Edited by:

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

#### Reviewed by:

Daniele Di Mascolo, Italian Institute of Technology (IIT), Italy Jyothi U. Menon, The University of Rhode Island, United States

\*Correspondence:

Viness Pillay viness.pillay@wits.ac.za

#### Specialty section:

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

Received: 08 October 2019 Accepted: 22 January 2020 Published: 11 February 2020

#### Citation:

Essa D, Kondiah PPD, Choonara YE and Pillay V (2020) The Design of Poly(lactide-co-glycolide) Nanocarriers for Medical Applications. Front. Bioeng. Biotechnol. 8:48. doi: 10.3389/fbioe.2020.00048 Keywords: poly(lactide-co-glycolide), drug delivery, biodegradable polymer, nanoparticle preparation, nanomedicine, computational simulation

# INTRODUCTION

The design of novel delivery systems using nanomaterials has experienced substantial growth since the application of nanotechnology to biomedical applications established the field of nanomedicine. As a result of the ongoing discovery of numerous new pharmaceutically active compounds which have shown excellent efficacy but inadequate clinical translation, there is a growing need to fill the gap between the formulations available and their successful inclusion into active treatment. This has urged scientists to investigate alternate forms of delivery to the biological target in order to overcome the hurdles associated with conventional drug delivery, such as poor drug entrapment, inadequate bioavailability and pharmacokinetics, as well as systemic toxicity and side effects. These novel delivery systems all strive for the "magic bullet" effect (Bosch and Rosich, 2008) which is a vehicle that can form favorable interactions with a lipophilic or hydrophilic drug to facilitate high drug loading (Aravind et al., 2013; Bauer et al., 2016), can shield the drug from physiological conditions, deliver it to the biological target with minimal loss, and then can release it at the site in a sustained manner and at therapeutic concentrations (Al-Jamal et al., 2016; Almoustafa et al., 2017). Moreover, the carrier is ideally biodegradable, biocompatible and nonimmunogenic, with low systemic toxicity (Alshamsan, 2014; Ananta et al., 2016). Nanomaterials are a befitting source to meet these requirements because they can be tailored to a vast range of sizes and shapes and can suit various delivery mechanisms, while the interactions between the carrier and the physiological medium can be controlled by adapting the surface properties

of the carrier (Kakkar et al., 2017). This has given rise to the widespread implementation of nanomaterials as pharmaceutical carriers for medical diagnostics and therapeutics (theranostics) (Berthet et al., 2017; Singh et al., 2019). Nanostructures can be fabricated from organic, inorganic, metallic or non-metallic sources. Examples include carbon nanotubes (Singh et al., 2013), dendrimers, liposomes, micelles, and solid lipid nanoparticles (Mishra et al., 2014; Lombardo et al., 2019).

Polymeric nanoparticles are commonly implemented as components of drug delivery systems and the use of synthetic polymers in particular can enable the design of carriers in a wellcontrolled and reproducible manner in order to suit the desired application (Lai et al., 2014). Polymer-based nanoparticles act as drug delivery vehicles by encapsulating the active agent inside its polymeric matrix, by conjugating to the agent or by adsorbing it onto the surface of the polymer (Mahapatro and Singh, 2011; Ansary et al., 2014). Polymers can be constructed to be linear, branched or globular and their size and their properties can be modulated by the choice of synthetic process (Panyam and Labhasetwar, 2012; Chen et al., 2013). Biodegradable polymers are suitable as nanocarriers as they often self-assemble and are easily sourced. Natural biodegradable polymers are chitosan (Ali and Ahmed, 2018), alginate (Jana et al., 2016) and inorganic ceramic hydroxyapatite composites (Turon et al., 2017). Synthetic polymers such as poly lactide-co-glycolide (PLGA) are an attractive alternative as they can be precisely engineered from monomers to suit the target and physiological environment they are intended for Middleton and Tipton (2000).

Poly lactide-co-glycolide has become ubiquitous in the biomedical field for many reasons. Firstly, it is a synthetic, biodegradable polymer that is easily broken down in vivo by hydrolysis into lactic acid and glycolic acid. These monomers are biocompatible and are physiologically metabolized by the tricarboxylic acid cycle for final excretion in the lungs (Semete et al., 2010; Sequeira et al., 2018), as shown by **Figure 1** (Sun et al., 2017). Hence, PLGA as a nanocarrier is considered to produce minimal systemic toxicity when used for biomedical applications (Kumari et al., 2010) and has been used in various formulations including membranes, sponges and gels (Sun et al., 2017).

The appeal of PLGA also lies in the fact that its properties can be manipulated and adapted to modify the encapsulation profile and drug release kinetics of the nanostructure to suit the desired application (Mittal et al., 2007). PLGA is overall a hydrophobic polymer and is therefore detected by the RES and if unmodified, is bound by phagocytes for elimination by the liver or spleen and eliminated before delivering its payload to the target site (Danhier et al., 2012). To circumvent this, surface modification of PLGA is necessary. One such modification is the coating of hydrophilic poly ethylene glycol (PEG) groups on the surface of PLGA to shield the hydrophobic end groups from the reticulo endothelial system (RES), resulting in an amphiphilic di-block co-polymer (Salmaso and Caliceti, 2013). Other polymers used as surface modifiers include chitosan (Lu et al., 2019), polaxamer and poloxamines (Redhead et al., 2001) which work by altering the electrostatic and hydrophobic surface properties of the PLGA block co-polymer. To increase the therapeutic efficacy, the surface of the PLGA nanocarrier can be decorated with targeting ligands such as small molecules, antibodies, and aptamers. These molecules selectively bind to receptors on the target cell and guide the vehicle to the site of action (Jahan et al., 2017). Targeting moieties such as aptamers, have been shown to increase retention time at the site of action (Dinarvand et al., 2011).

The use of PLGA in biomedicine dates back to the 1970s when it was used as a component of biodegradable sutures and implants. With the advent of nanomedicine, it has found application as nanocarrier in various areas of medical research, including chemotherapeutics, immunology, and biomechanics (Swider et al., 2018). Numerous studies have also reported successful applications in antibiotics, antiseptics, imaging, wound healing, and as nano scaffolds (Sharma et al., 2016). The suitability and adaptability of PLGA as a nanocarrier is illustrated by **Figure 2** (Mir et al., 2017).

Optimizing the synthetic procedure by changing the parameters can affect other properties of nanocarriers and therefore a great deal of forethought should go into the design of the system for the particular application (Rezvantalab et al., 2018). During synthesis, parameters such as particle size, surface behavior, degree of crystallinity, degradation rate, and molecular weight can be modified to adapt the nanocarrier for desired dosage and site specific action (Mittal et al., 2007). Bio-nano interactions are important considerations in design as they determine the suitability of the nanostructure for the intended application as well as the undesired toxicity that may result from the engineering process. Previous research has indicated accumulation of PLGA in the liver when used as nanocarriers and therefore there could be toxicity challenges caused by dose dumping (Makadia and Siegel, 2011). While there are numerous reviews on PLGA based nanodelivery systems in general, this work considers the literature from a design perspective. Schrur's nano-toxicology editorial states "few studies offer consistent results that are of value, and it is difficult to compare studies because they are often carried out using poorly characterized nanomaterials and arbitrary experimental conditions" (Schrurs and Lison, 2012). With these considerations in mind, in silico design, which is an expanding field in drug delivery, could be used to model numerous parameters, including polymer degradation, drug loading and toxicity and hereby provide insight into the structure-behavior relationships of PLGA-based nanocarriers (Ramezanpour et al., 2016). The aim of this review is to collate research on PLGA based delivery vehicles that have been studied for common medical applications, to compare the choices of starting materials and synthetic methods on the properties and functions of the final polymer-drug systems, and to explore how computational investigations can assist in the design of these systems.

#### PLGA AS A NANOCARRIER

#### Properties

Poly lactide-co-glycolide is synthesized from its constituent monomers (Sun et al., 2017) and can be obtained commercially in varying ratios of these monomers

(Sadat Tabatabaei Mirakabad et al., 2014). Each constituent has its own physical characteristics that it brings to the copolymer. PLGA retains properties of both copolymers and can be customized using these properties, which are stiff, hydrophobic and slowly degrading lactic acid vs. malleable, less hydrophobic and faster degrading glycolic acid (Engineer et al., 2011). For example, poly-DL -lactic acid has a methyl group, as shown in **Figure 1**, and is therefore more hydrophobic than poly glycolic acid. Hence, adjusting the concentration of poly-lactic acid in PLGA varies the solubility of the final polymer (Makadia and Siegel, 2011). A study investigating the rate of hydrolysis of PLGA demonstrated that increasing glycolic acid to lactic acid ratio increases the hydrophilicity of the PLGA co-polymer and hence leads to faster degradation (Keles et al., 2015), while a separate study quantified the degradation constant to be 1.3 times higher for glycolic units than for lactic units in the PLGA co-polymers investigated (Vey et al., 2011). It has been shown that PLGA co polymer ratios can be varied to adapt the degradation rate from months to years (Sun et al., 2017). In general, the higher the glycolic acid content of the PLGA polymer, the more amorphous it is and the faster it degrades due to it being more hydrophilic. An exception is PLGA 50:50 lactic: glycolic units, which has exhibited the fastest degradation rate (Lü et al., 2009). It has been shown that increasing the glycolic acid ratio increases the wettability of PLGA for thin film applications (Ayyoob and Kim, 2018) and that increasing the lactic acid content has application in designing PLGA carriers for sustained release (Li, 1999). PLGA co-polymers with lactic acid content less than 70% have been characterized as amorphous and suitable for drug delivery applications (Habraken et al., 2006). As expected, the higher the molecular weight of PLGA, the more structural integrity it exhibits and the longer it has shown to degrade in vivo (Anderson and Shive, 2012). The PLGA co-polymer can be end-capped with different functional groups which have shown to affect the degradation kinetics of the delivery system. For example ester end-capped polymers exhibit a slower degradation rate than acid end-capped polymers and are therefore suitable for slower release applications (Gentile et al., 2014). Apart from degradation rate, it is also possible to control solubility and glass transition temperature of the PLGA system by varying the molecular weight, lactic/glycolic ratios, and end-cap functional groups of the starting material (Gentile et al., 2014).

Poly lactide-co-glycolide is also soluble in a variety of organic solvents including acetone, dichloromethane, chloroform, ethyl acetate, and THF (Sharma et al., 2016) and therefore is relatively simple to work with as carriers for both hydrophobic and hydrophilic drugs (Zhang et al., 2014).

#### Surface Functionalization Shielding

In order to avoid elimination by the RES, a stealth coating around the hydrophobic PLGA nanoparticle surface has been achieved by incorporation of co-polymers with desired properties. The most frequently used co-polymer is polyethylene glycol (PEG) as it is biocompatible and easily grafted or adsorbed onto the surface of PLGA. The hydrophilic PEG shields the PLGA carrier from being taken up by opsonins (Vllasaliu et al., 2014) and it has been shown that the PEG shield dramatically increases the blood circulation half-life of the nanocarrier (Owens and Peppas, 2006). Some studies have shown that the nanoparticle in vivo residence time is dependent on the surface density of the PEG chains (Bertrand et al., 2017). PEGylation has also been associated with enhanced drug loading and tunable carrier degradation (Khalil et al., 2013). Chitosan, a natural polymer that is formed by partial deacetylation of chitin, is also commonly grafted onto the surface of PLGA based systems to increase biocompatibility. It is biodegradable and has mucoadhesive properties as it carries a positive charge and can efficiently bind to negatively charged cell membranes (Bruinsmann et al., 2019). Hence, a coating of chitosan on the PLGA nanostructure shields it from opsonins and promotes stronger cellular interaction and retention (Lima et al., 2018). Collagen is a highly hydrophilic protein that also increases cellular interaction and when blended with PLGA, can form a delivery system with superior hybrid properties such as increased biological compatibility and mechanical strength (Sadeghi-Avalshahr et al., 2017). Heparin, a biocompatible material that can be obtained both naturally and synthetically, has been used to impart specific binding properties to delivery systems when combined with polymers (Rodriguez-Torres et al., 2018). It is a sulfated glycosaminoglycan with high binding affinity for various growth factors and has been used in sustained delivery applications by immobilization on the surface of PLGA delivery systems (Chung et al., 2006).

#### Surfactants

One of the strategies employed in the nanofabrication process to increase colloidal stability is the use of surfactants. These are agents which usually have amphiphilic properties and reduce the interfacial tension between the hydrophobic and hydrophilic components, hence increasing miscibility and dispersion (Heinz et al., 2017) and preventing particle aggregation (Shkodra-Pula et al., 2019). A commonly used agent is polyvinyl alcohol (PVA), which is a hydrophilic polymeric surfactant that has been shown to decrease the size and increase the uniformity of PLGA nanocarriers, but is also associated with hypertension and central nervous system depression in animal studies (Menon et al., 2012). The use of Polysorbate 80, 60, and 20 has also shown increased residence time and enhanced permeation of the blood brain barrier (Sharma et al., 2016). However, it has shown in some cases to cause anaphylactoid reactions (Coors et al., 2005) and long-term infertility (Gajdova et al., 1993). Polaxamer is a thermo-reversible, non-toxic coating that has been used when encapsulating hydrophobic drugs and has been shown to preferentially target cancer cells. However, it has shown rapid erosion times and it is associated with hyperlipidemia and hypercholesterolemia (Miller and Drabik, 1984). Poloxamine is an amphiphilic block co-polymer and therefore has been used to stabilize hydrophobic drugs while increasing circulatory residence time (Alvarez-Lorenzo et al., 2010). Vitamin E TGPS is a water-soluble form of vitamin E and is used as a solubilizing and emulsifying agent in nano drug delivery. It is commonly used to enhance drug loading (Zhang and Feng, 2006) and nanoparticle degradation rates (Jalali et al., 2011).

#### Active Targeting

fbioe-08-00048 February 7, 2020 Time: 15:13 # 5

In active targeting, the surface of the nanoparticle is further decorated with ligands that specifically bind to receptors on the cells of interest and enables the carrier to enter the cell by receptor mediated endocytosis (Muhamad et al., 2018). These targeted delivery systems are designed to localize drug release at the disease site (Danhier et al., 2010). There are various different kinds of targeting ligands such as small molecules, peptides, antibodies, aptamers and polysaccharides, as shown in **Figure 3** (Yoo et al., 2019). These ligands can be either conjugated or adsorbed onto the surface of the nanocarrier after formation or can be linked to one of the components of the carrier before nanoparticle formation (Yoo et al., 2019). It has been shown that increasing the conjugation density of targeting ligands has an effect on the targeting ability of the nanocarrier.

Monoclonal antibodies have had a long history as targeting ligands (Friedman et al., 2013) since they have complementarity determining regions that enable them to bind to receptors on cell surfaces with high specificity and affinity (Carter et al., 2016). However, since they are large molecules, their conjugation density capacity on the nanocarrier is substantially decreased (Yoo et al., 2019) compared to other ligands, and furthermore, they raise immunogenicity concerns (Karra and Benita, 2012). Compared to antibodies, peptides have the advantage of smaller size and non-immunogenicity but they still are able to retain target specificity (Zhao et al., 2007). Aptamers are short strands of nucleic acids that can be synthetically designed to bind specific biological targets. They are non-immunogenic and nontoxic but their synthesis can be costly (Friedman et al., 2013). Despite their advantages for in vivo targeting, both peptides and aptamers are prone to enzymatic degradation (Yoo et al., 2019). Polysaccharides are advantageous because they are biocompatible and can be used as structural components of the nanocarrier (Choi et al., 2011) as well as to target carbohydrate binding receptors on cell surfaces. However, some polysaccharides could have solubility challenges and modification of the carbohydrate structure could result in unintended toxicity (Peng et al., 2018). Small molecules form a class of targeting ligands that comprise of synthetic compounds that are designed to target certain domains on cell surface receptors. They are usually chosen because of ease and control of synthesis but they often do not bind with high specificity and some target receptors can be expressed in healthy cells (Yoo et al., 2019), resulting in unintentional cell binding. Even though active targeting strategies provide an attractive avenue for site specific drug delivery, there are many challenges in this area, such as receptor accessibility and off-target binding. Particularly during different stages of tumor development, certain receptors can be up or down regulated, which provides an additional challenge for the use of targeting ligands in chemotherapeutic drug delivery (Vhora et al., 2014).

#### Toxicity

Despite biocompatibility and biodegradability of PLGA as a polymer, its toxicological profile in nanoformulations deserves to be investigated because of altered physicochemical properties, such as higher surface area to mass ratios. Furthermore, reports have suggested that particles of any material may acquire unique toxicological properties in the nanoscale

(Makadia and Siegel, 2011). Different effects such as acute toxicity, repeated dose toxicity, inflammation, oxidative stress, genotoxicity, and reproductive system toxicity of PLGA nanocarriers have been examined in order to obtain information on the possible risks of these materials in pharmaceutical preparations. A study of danorubicin loaded PEG-PLL-PLGA nanoparticles has described some toxicity in Kunming mice (Guo et al., 2015) but since no results were reported for blank nanoparticles, it is unclear whether the toxicity was due to the drug or nanocarrier (Jesus et al., 2019). Regarding oxidative stress, studies have reported an overall increase in the production of reactive oxidative species corresponding to increasing concentrations of the PLGA nanoformulations tested (Singh and Ramarao, 2013; Grabowski et al., 2015) and another study demonstrated mild inflammatory properties of different PLGA formulations (Grabowski et al., 2016). Several studies have confirmed no genotoxicity (Tulinska et al., 2015; Platel et al., 2016), no toxicity on reproduction (Chen et al., 2017; Sharma et al., 2017) and no hemolysis (Chen et al., 2017). A study comparing the toxicity of PLGA nanoparticles to silica-, iron-, and zinc-based nanoparticles showed that the PLGA system had no appreciable adverse in vitro or in vivo toxicological outcomes, and did not produce the toxicity commonly associated with the inorganic nanomaterials (Semete et al., 2010).

# SYNTHETIC METHODS OF PLGA NANOCARRIERS

Poly lactide-co-glycolide nanocarriers may be fabricated by different methods and the choice of method has shown to affect properties such as particle size, colloidal stability, drug loading/encapsulation efficiency, and release behavior of the final product (Swider et al., 2018). Depending on the process of preparation, the structural organization may also be different. The drug is either encapsulated inside the carrier or adsorbed on the surface (Danhier et al., 2012). There are several methods that can be employed for the preparation of PLGA nanocarriers, and the following provides a brief overview on both the well-established and relatively recently developed techniques. The traditional methods based on emulsions are illustrated by **Figure 4** (Ding and Zhu, 2018).

### Single and Double Emulsion

The emulsion methods have been the most frequently used methods of synthesis and they are suitable for a wide range of drugs with varying solubilities (Wang et al., 2016). The single emulsion (oil in water or O/W) method is suitable for hydrophobic drugs. PLGA and the drug is dissolved in a small volume of suitable volatile organic solvent and added dropwise to the aqueous phase containing a stabilizer, usually PVA. The mixture is sometimes sonicated and then stirred, often under sheer stress for a fixed amount of time to allow the organic solvent to evaporate. The double emulsion method is used when the active agent to be entrapped is hydrophilic, such as proteins and peptides. The active is dissolved in an aqueous phase and then added to PLGA which is dissolved in the organic phase, and this

forms a primary water in oil (W/O) emulsion. This is then added to another aqueous phase containing a stabilizer and allowed to mix under stress, allowing the organic solvent to evaporate. The nanoparticles are therefore formed by a water in oil in water (W/O/W) emulsion (Makadia and Siegel, 2011). The product is isolated by centrifugation or ultrafiltration and washed to remove unreacted products. Thereafter it is freeze dried and can be stable for several months to years. Recently, a single emulsion method was used to successfully entrap proteins for vaccine application (Ospina-Villa et al., 2019) and a PLGA-PEG nanocarrier was formulated using the double emulsion method for intraperitoneal insulin delivery (Haggag et al., 2018). The emulsion methods can be adjusted by changing the drug to PLGA ratio, the organic solvent, the stabilizer concentration in the aqueous phase and the stirring speed and can hereby be adapted to control the size range of the nanocarriers to some extent. However, there are often batch to batch variation with these methods and the carriers prepared by this method for protein-based drugs have limited stability due to degradation of proteins at the aqueous interface and the sheer stress of homogenization leading to unfolding of the protein sheets (Ding and Zhu, 2018).

# Spray Drying

This method involves the preparation of water in oil or solid in oil emulsions, which are sprayed in a thin stream of heated air. The type of drug (hydrophilic or hydrophobic) would determine the solvent used in the emulsion (Makadia and Siegel, 2011). Recently, spray drying was used in the preparation of a PLGA nanoformulation for sustained treatment of tuberculosis (TB) (Kalombo et al., 2019) and in the fabrication of a carrier for antibiotic coating of dental implants (Baghdan et al., 2019). This method is highly advantageous because it is suitable for hydrophobic and hydrophilic drugs and can be used for sensitive compounds since the conditions are mild. It is also a rapid method (Nie et al., 2008) which can be suitable for industrial scale-up due to the minimal processing parameters

involved (Ding and Zhu, 2018). The main drawback of this technique is the wastage caused by inaccessible product that adheres to the inside of the nanosprayer (Wang et al., 2016). Parameters such as orientation of jets, temperature, and solvent choice can all affect the properties of the final nanoparticles (Berkland et al., 2004).

### Coacervation

With coacervation or phase separation, the polymer and drug are prepared as O/W for hydrophobic drugs and W/O/W for hydrophilic drugs, and then a non-solvent, e.g., silicon oil is added dropwise under stirring (Verma et al., 2018). This reduces the solubility of PLGA in the organic solvent and results in the formation of a polymer-rich phase in which PLGA surrounds the drug molecules to form microdroplets (coacervates). These are rapidly quenched in a non-soluble medium to form the solid product (Wang et al., 2016). Parameters such as starting polymer, solvent choice, stirring rate and temperature can be varied to control the properties of the particles. Coacervation usually forms micrometer sized particles (Sharma et al., 2016) but has been used in protein nanoparticle preparation (Verma et al., 2018).

# Salting Out

In the salting out method, drug and PLGA are dissolved in a miscible organic solvent and added to the aqueous phase containing stabilizer and a salt under sheer mechanical force. The salt usually used is magnesium chloride hexahydrate or magnesium acetate tetrahydrate (Wang et al., 2016) and is used at a ratio of 1:3 PLGA:salt (Eley et al., 2004). Upon addition of water, the organic solvent diffuses into the aqueous phase, causing the formation of PLGA-drug nanoparticles, illustrated by **Figure 5** (Crucho and Barros, 2017). This method is not suitable for lipophilic drugs and can be time intensive since isolation of the product involves several washing steps to remove reagents. However, it would suit drugs which are very temperature sensitive since heat is not required (Nagavarma et al., 2012). This method is robust and is suitable for nanoparticles with high polymer concentrations since the size of the particle is not generally affected by the amount of polymer (Swider et al., 2018).

# Nanoprecipitation

In this method, PLGA and the drug is dissolved in a polar, water miscible solvent and added dropwise to the aqueous phase, which may contain a surfactant. The product is formed by rapid diffusion of the water miscible solvent into the aqueous phase, resulting in precipitation of the PLGA-drug nanoparticles, as shown in **Figure 6** (Crucho and Barros, 2017). The properties of the nanocarrier are controlled by PLGA content and molecular weight, PLGA to drug ratio and choice of solvent (Wang et al., 2016). Recently, an optimized nanoprecipitation method was developed for the preparation of PLGA encapsulated alendronate sodium, a drug for osteoporosis (Oz et al., 2019) and a modified procedure was reported for a PLGA hybrid nanocarrier for simvastatin (Zhang et al., 2018). Nanoprecipitation can be used to prepare particles in the 100 nm size range and is advantageous because of the absence of shear stress (Fessi et al., 1989). However, the unmodified nanoprecipitation method does not usually work well for hydrophilic drugs as they do not form favorable interactions with PLGA in a water miscible solvent (Govender et al., 1999).

#### Supercritical Fluid Technology

fbioe-08-00048 February 7, 2020 Time: 15:13 # 8

Supercritical fluid technology, illustrated by **Figure 7** (Jog and Burgess, 2017), can provide an environmentally friendly method of generating nanoparticles since it reduces, and in some cases, eliminates the use of organic solvents (Koushik and Kompella, 2004). In this method, the polymer and drug are dissolved in a supercritical fluid which is then rapidly expanded and depressurized. The resultant mixture is then passed through a fine nozzle or capillary, resulting in supersaturation and formation of nanoparticles, which are collected separately (Mishima, 2008). This is an attractive method as it is highly tunable but the kind of nanoparticle products are restricted since not all starting materials are compatible with the supercritical fluid (Soh and Lee, 2019).

# Microfluidics

The area of microfluidics deals with channels of the micrometer size range that are used to control and manipulate the movement of volumes of fluid from the nanolitre size range and below. When working at the nanoscale, the conditions of flow can be precisely controlled and constant laminar flow is maintained, which is impossible when conducting reactions at the macroscale level (Chiesa et al., 2018). Therefore this technique has lent itself to the synthesis of nanoparticles by the formation of emulsions using droplet microfluidics. In this method, the polymer and drug are combined and the emulsion is formed in the microfluidic mixer, which can have different channel architechtures (Collins et al., 2015). The most commonly used geometries for droplet based PLGA nanoparticle formation are the t-junction, flow focusing, and continuous flow microchannels, as shown in **Figure 8I** (Shembekar et al., 2016). In the t-junction geometry,

channels are perpendicular to each other. The dispersed phase (aqueous) flows through one channel while the continuous phase (oil) flows through the other and droplets are formed at the junction. In the flow focusing system, the aqueous phase flows through a square capillary where shear force is provided on either side of it by the flow of the oil phase. The emulsion then flows through a narrow capillary and droplets are formed in a collection chamber. With continuous flow geometry channels, the aqueous phase flows through a capillary that resides in another capillary through which the oil flows in the same direction. Droplets begin to form once the two phases mix (Shembekar et al., 2016). SEM images of particles formed by this method are shown in **Figures 8II-D,E** (Xu et al., 2009).

These microchannels are used for O/W emulsions, but can be adapted for double (W/O/W) emulsions by using a combination of channels. Recently, a microfluidics method was developed for the encapsulation of cell penetrating peptides (Streck et al., 2019) and targeted delivery of taxanes (Martins and Sarmento, 2020). There are numerous advantages of microfluidics for nanoparticle synthesis. With this technique, the chemical composition of the final product can be preselected according to the desired application. The synthetic parameters can be controlled to the extent that there is a much larger particle size homogeneity compared to bulk methods, and there are also smaller volumes of solvent needed. However, the scale of nanoparticle production is limited and the microchannels are also susceptible to blockage and contamination. The time and temperature of mixing, flow rate, choice of solvents and payload type determine the properties of the final nanoparticles (Kim K.T. et al., 2019).

### Membrane Extrusion Emulsification

With this technique, single or double emulsions of PLGA and drug are either initially prepared or formed when extruded through a membrane of predetermined pore size. There are two ways to do this – direct and pre-mix membrane extrusion (ME), as illustrated by **Figure 9A** (Guo et al., 2018), and a characteristic emulsion is shown in **Figure 9B**. In direct ME, the membrane emulsifies the dispersed phase into nanosized droplets, while in premix ME, the emulsion is formed via a conventional method and thereafter extruded through the membrane, which downsizes the coarse emulsion into uniform nanosized droplets. This method can be used for hydrophobic or hydrophilic drugs and is advantageous because the size of product can be controlled by varying the nanoporous membrane pore size to create particles of required dimensions, resulting in a large size homogeneity. Premix ME in particular has been shown to have a higher uniformity in dispersity of final nanoparticles as shown by **Figure 9C**, compared to direct ME. In general, it is a mild procedure with low energy requirements and can be easily scaled up. However, it is not suitable for emulsions with high viscosity (Guo et al., 2018).

# Nanoimprint Lithography and the PRINT Technique

Nanoimprint lithography is used to form nanoparticles from a nanostructure template that is placed over a layer of precursor

material which is heated to above the glass transition temperature of the polymer. Thereby, the malleable precursor material is molded into the desired size and shape, which is retained upon cooling. The template is then removed, leaving the product on the substrate base. Su et al. (2015) have successfully used this method for the nanofabrication of submicron PLGA grooves for the control of the length and direction of retraction fibers during cell division. The major drawback of this method is the residual interconnecting layer on the substrate base that prevents the formation of isolated nanostructures (Fu et al., 2018). The PRINT (particle replication in non-wetting template) technique involves the preparation of the PLGA-drug solution matrix and casting it on a delivering sheet. Thereafter a mold with nanosized cavities is placed over the delivering tray and it is passed through a nip and separated so that the polymeric material fills the mold cavities. The particles are then solidified and placed on a high energy adhesive layer and passed through the nip without separation. After the mold is removed, the nanoparticles are collected by washing with a solvent that dissolves the adhesive (Perry et al., 2011). The method is automated with a high degree of control over the individual parameters, and can be used for a wide variety of cargos including hydrophobic and hydrophilic drugs, vaccines, and proteins. However, it is a multistep process that can be labor intensive (Swider et al., 2018). The desired particle size, surface properties and composition can be preset and controlled in the initial step. The PRINT process is illustrated in **Figure 10** (Perry et al., 2011), while **Figure 11** shows SEM micrographs of the different shapes of PLGA nanoparticles that have been prepared by this method. Enlow and colleagues have reported the PRINT process whereby PLGA micro- and nanoparticles were prepared, with cylindrical, spherical, ridged, and fenestrated morphologies. These particles demonstrated > 40% drug loading and > 90% encapsulation efficiencies of docetaxel (Enlow et al., 2011).

#### MEDICAL APPLICATIONS

#### Cancer Research

#### Actively Targeted Chemotherapeutics

In the United States of America alone, 1,762,450 new cancer diagnoses and 606,880 cancer related deaths are expected to occur in 2019 Siegel et al. (2019). Despite the ubiquity of this

disease, treatment options are challenging due to the complex pathology of the different cancers. Current chemotherapy often leaves debilitating and life altering side effects since most drugs on the market that target the rapidly dividing cancer cells also inadvertently damage cells that are vital for normal life processes (Rizvi and Saleh, 2018). Actively targeting PLGA nanoparticles are able to circumvent this; Moku et al. (2019) have shown

increased drug loading and efficacy against lung cancer by using transactivator of transcription (TAT) peptide ligands to target mesenchymal stem cells while Ganipineni et al. (2019) found that magnetically targeted paclitaxel- and SPIO-loaded PLGA-based nanoparticles effected increased cellular uptake in glioblastoma cells compared to the non-targeted carriers. Another type of nanoparticle targeting is the use of 'smart' carriers that are engineered to respond to a stimulus (Kapoor et al., 2015). Recently, a pH dependent aptamer functionalized PLGA nanocarrier system was reported to increase anti-cancer activity of doxorubicin to human lung cancer cells, with reduced toxicity to healthy cells (Saravanakumar et al., 2019) and a superparamagnetic iron oxide encapsulated nanocarrier

FIGURE 11 | SEM Images of PLGA PRINT particles. (A–C) Cylinders of different dimensions. (D) Spheres; (E) ridged cubes; (F, bottom right) particles with center fenestrations, reprinted with permission from Enlow et al. (2011). Copyright (2019) American Chemical Society.

for docetaxel demonstrated favorable pharmacokinetics and a greater degree of uptake in breast cancer cells (Panda et al., 2019).

#### Immunotherapy

fbioe-08-00048 February 7, 2020 Time: 15:13 # 11

Since Allison and Honjo were awarded the 2018 Nobel prize in Physiology and Medicine "for their discovery of cancer therapy by inhibition of negative immune regulation" (Guo, 2018), the area of nanomedical research into cancer immunotherapy has received substantial attention. This approach involves the use of pharmaceutical agents to activate a patient's immune system to fight cancers as opposed to traditional chemotherapy which involves directly drugging the cancer cells (Khalil et al., 2016). Chen et al. (2016) have described PLGA nanocarriers equipped with an immunostimulant and photothermal agent, and this formulation showed increased activation of the immune system of BALB mice compared to the free agent. More recently a sustained controlled release PLGA nanosystem was developed to activate the anti-tumor immune response in mice bearing melanoma and colon cancer (Yin, 2019) and a PLGA system containing an immune adjuvant together with an enzyme that increased the efficacy of radiation therapy demonstrated the feasibility of combination immunotherapy and targeted radiotherapy in BALB mice (Chen et al., 2019).

#### Imaging and Diagnostics

Poly lactide-co-glycolide has applications for tumor diagnostics as it is able to deliver imaging agents to cancer cells with specificity and controlled biodistribution. Advances in nanotheranostics, which is the incorporation of imaging and therapeutic agents in one nanocarrier, have shown promise for real time imaging throughout a patient's treatment course (Chapman et al., 2013). A novel theranostic PLGA nanocarrier with a near infrared imaging agent, further decorated with gold nanoparticles has been synthesized and shown to have increased activity and photodynamic properties in tumor grafted BALB mice (Xi et al., 2018) and a targeted PLGA-based nanobubble was designed with an ultrasound contrast agent, and demonstrated specificity and imaging capabilities to breast cancer in BALB mice (Du et al., 2018). More recently an image guided photothermal PLGA nanocarrier for doxorubicin showed promise for real time photoacoustic imaging in tumor bearing nude mice (Shen et al., 2019) and a near infra-red dye loaded PEGylated PLGA nanocarrier was also able to provide information on the circulation and distribution of the nanoparticles in nude mice (Kumar et al., 2019).

#### HIV Treatment

The delivery of anti-retro viral drugs faces many of the general limitations of conventional drug delivery and therefore biomaterials with low toxicity such as PLGA based nanocarriers are being implemented in formulations to treat HIV. Mannosylated PLGA nanoparticle carriers have shown promise for targeted delivery of anti-retro viral drugs to the brain (Patel et al., 2018) and the use of microfluidics technology enabled the novel synthesis of efiravine loaded PLGA nanoparticles (Martins et al., 2019). A recent study reported a PLGA based nanocarrier for the combination of the anti-retro virals griffithsin and dapirivine which showed a long acting treatment profile (Yang et al., 2019) and a separate proof of concept study showed promise for a long acting bictegravir encapsulated PLGA nanocarrier (Mandal et al., 2019).

#### Inflammatory Disorders

Many current treatments have proven to be inadequate at treating or alleviating symptoms of inflammation. The specific delivery of anti-inflammatory agents to the target site could potentially increase their therapeutic concentration in the inflamed tissue with reduced side effects (Gendelman et al., 2015), and the use of PLGA is particularly suitable to this application because of its favorable biodegradability and non-immunogenicity (Lamprecht et al., 2001). Davoudi et al. (2018) described a carrier within a carrier system using intestinal organoids to transport 5-ASA encapsulated PLGA nanoparticle to treat inflammatory bowel disease, and Perreira's research involved the development of a metformin loaded nanoformulation that showed efficacy against periodontal inflammation in diabetic rats (Pereira et al., 2018). Gholizadeh et al. (2018) have formulated a dactolisib-PLGA nanoparticle that showed activity against inflamed endothelial cells and more recently, Yang (2019) group reported the synthesis of a crocetin-loaded nanoparticles that reduced the level of pro-inflammatory cytokines in renal tissue and therefore shows potential for the treatment of diabetes induced nephropathy.

# Other Applications

Poly lactide-co-glycolide has been adapted to treat conditions in many fields of biomedicine, as shown by **Figure 12** (Mir et al., 2017). A hyaluronic acid functionalized PLGA based nanocarrier for methatrexate has been developed for targeted treatment of rheumatoid arthritis (Trujillo-Nolasco et al., 2019), a PLGA nanoparticle with protease inhibitor has shown to overcome gastro-intestinal limitations of oral insulin delivery in rats (Faheem et al., 2019), a PLGA-chitosan based nanocarrier has been synthesized and shown to be selective for human antigen presenting cells (Durán et al., 2019), a potential DNA vaccine delivery system has been designed using a PLGA based nanocarrier (Besumbes et al., 2019), and a Vitamin D encapsulated PLGA based delivery system has recently shown activity against various markers for Alzheimer's disease in mice (Jeon et al., 2019). Gonzalez-Pizarro et al. (2018) have prepared an optimized fluoromethalone-PLGA nanoparticle that demonstrated increased efficacy in treating ocular inflammation compared to the commercial formula. The use of some of the available methods in PLGA nanocarrier synthesis and their applications are summarized in **Table 1**.

# Inclusion of PLGA Formulations in the Clinic

The biocompatibility, biodegradability and versatility of PLGA has made it suitable for a wide range of clinical applications. PLGA was commercialized in the 1970s as a suture material under the trade name Vicryl <sup>R</sup> (Kamaly et al., 2016). Other sutures include Dolphin Sutures <sup>R</sup> , and Polysorb <sup>R</sup> which are both currently approved. PLGA-containing chemotherapeutic

formulations approved for clinical use include Lupron Depot <sup>R</sup> , for sustained release of leuprolide, which has application in the management of prostate cancer (Swider et al., 2018), Trelstar <sup>R</sup> , a triptorelin-containing suspension for the treatment of prostate cancer and Zoladex, a goserelin-containing implant used in the treatment of breast and prostate cancer and endometriosis. Formulations approved for other applications include Risperdal <sup>R</sup> Consta <sup>R</sup> (risperidone), Vivitrol <sup>R</sup> (naltrexone) and Arestin <sup>R</sup> (minocycline) for the treatment of schizophrenia, opioid dependence and periodontal disease respectively (Jain et al., 2016). A promising direction for clinical development is the engineering of PLGA based systems with imaging agents to monitor disease progress and/or relapse patterns using magnetic resonance imaging (MRI). Studies have shown these structures to be non-invasive and cost effective, with excellent safety profiles (Strohbehn et al., 2015). Furthermore, there are a number of PLGA based systems have been used in clinical trials that are ongoing or have been recently concluded (U.S. National Library of Medicine, 2019).

#### COMPUTATIONAL MODELING

Despite the numerous formulations and methods available for PLGA synthesis, there is a large discrepancy between in vitro, in vivo and clinical results. One of the reasons for this could be the fact that it is difficult to obtain mechanistic insight into nano-formulation behavior in the various systems by evaluation of results based solely on experimental methods (Huynh et al., 2012). Because of the ubiquity of PLGA across so many biomedical fields of research, there is an abundance of data at our fingertips for computational modeling (in silico). There are various levels of detail that can be used in computational simulations. The approach used most widely for nanoparticle drug delivery systems is molecular dynamics (MDs). This technique uses the motion of the molecules in the system to predict its behavior. The parameters it uses are the bonds, bond angles and dihedrals, and here the atoms are treated as point charges. If the degree of detail of atomistic interaction is not required, a coarse grained (CG) model can be used. Here, atoms are grouped into molecular fragments and their behavior in the system is modeled (Frenkel and Smit, 2001). Density functional theory is a model based on electronic density around atoms in the system and measures these interactions within the system of interest (Geerlings et al., 2003). Computational simulations at these levels can give insight into polymer interactions, drugcarrier miscibility, drug loading, drug release, and complex stability (Ramezanpour et al., 2016). Mathematical modeling such as finite element analysis and computational flow dynamics are particular useful when studying polymeric nanoparticle formation (Lince et al., 2011), diffusion and degradation (Kojic et al., 2017). Hence, coupled with experimental methods, they can be powerful tools in the rational design of PLGA nanocarriers for biomedical applications. A study of PLGA


TABLE 1 | Properties of some prepared PLGA nanocarriers, their methods of formation and biological targets.

binding to curcumin was conducted using MDs with the GROningen MAchine for Chemical Simulations (GROMACS) program, compared to laboratory findings and collated with the experimental results of 10 other PLGA-drug formulations. This study also predicted that PLGA could entrap curcumin with a higher encapsulation efficiency than tripalmitin, a lipidbased carrier, and this prediction was confirmed experimentally (Metwally and Hathout, 2015). A study on the drug release of the anti-cancer drug oxaliplatin in a PLGA matrix was conducted using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) program (Lange et al., 2016). A detailed insight into PLGA "patchy particles," which are particles made up of PLGA and lipid-polymer groups, was obtained from computational fluid dynamics, MDs and coarse grain simulations (Salvador-Morales et al., 2016). A study involving the simulation of PLGA-PEG co-polymer with the hydrophobic drug itraconazole, as shown in **Figure 13**, provided information about the drug loading limitations of this system (Wilkosz et al., 2018). A MDs simulation of the peptide Melittin showed that it constituted a more stable formulation with PLGA than PLA (Asadzadeh and Moosavi, 2019). The level of these studies as well as the information they provide is summarized in **Table 2**.

Another area of expansion of computational modeling on PLGA nanosystems is be the study of the transport of these nanocarriers in the circulatory system. One of the limitations of the clinical translation of nanosystems in general is the poor correlation between in vitro and in vivo results. The use

of mathematical and computational methods to model the interactions between the drug, carrier, biological transport system and tumor vasculature can be used to gain insight into these complexities (Curtis et al., 2015). Finite element analysis can be employed to model the dynamics of a nanoparticles within a channel, hence simulating transport in a blood vessel while continuum models can also be used to simulate nanoparticles in a vascular network generated by physical input parameters (Liu et al., 2012).

While computational simulations can provide valuable information on the molecular interactions in various PLGA nanocarrier systems, they can be limited by computational cost and time intensive calculations (Ramezanpour et al., 2016). It has



been proposed that a minimum reporting standard be instituted where researchers are required to present their results with enough information to make it useful for in silico modeling and future work (Faria et al., 2018).

Computational modeling could be immensely useful once it reaches a level where it can be used to select or eliminate certain experimental variables before laboratory research is conducted. Currently, even though simulation time scales are appropriate for the modeling of several nano-systems, the detailed investigation of the formation of nanoparticles by new methods, for example microfluidics, is beyond the abilities of current computational technology. The majority of studies conducted thus far involve the modeling of individual systems; however, more data is needed so that we can move away from specific systems to create profiles to generalize these delivery systems for rational design (Ramezanpour et al., 2016).

Optimizing nanoformulations especially with PLGA polymers which have numerous possible combinations of lactic acid to glycolic acid ratio, molecular weight, endcaps and surface functionalization, could be very time consuming, expensive and in some cases not experimentally feasible. Since computational simulations give a molecular insight to macroscopic properties (Huynh et al., 2012), it could provide a platform to model these initial parameters in order to narrow down the possibilities in a specific study.

#### DESIGN CONSIDERATIONS

Several studies have demonstrated the increase in particle size and decrease in drug release rate with increase in molecular weight and lactide:glycolide ratios (Song et al., 2008; Dinarvand et al., 2011). Recently, Lu et al. (2019) designed a 75:25 lactide:glycolide PLGA nanocarrier for the sustained release of paclitaxel. Surface functionalization is a component that needs to factor in when designing nanocarriers. Gu et al. (2008) conducted a study that optimized the in vitro release rate of docetaxel in PLGA, and additionally found that they could reduce the size of the nanoparticles from ∼291 to ∼160 nm by shortening the length of the PEG chains that were used for surface functionalization, while Bertrand et al. (2017) found that up to a point, increasing the density of PEG surface functionalization increased the blood circulation time of their nanoformulations. Gu's group also investigated an optimum targeting ligand density in order to provide maximum targeting ability without inhibiting the shielding effect of the PEG corona (Gu et al., 2008), while Lu's group found that increasing the density of the chitosan coating in their formulation increased the particle size from ∼133 to ∼173 nm (Lu et al., 2019).

Since the choice of fabrication methods and processes can determine the physicochemical characteristics of the resulting system, it is important to select an approach that is associated with the desired nanoparticle properties for the system of interest. For example Kim S.R. et al. (2019) reported that even though the preparation of their entacavirloaded system by spray drying produced larger particle size diameters compared to emulsion techniques, the spray dried system showed a much more favorable drug loading and release profiles and hence was the better performing delivery system. Krishnamoorthy described a multi-criteria decision making approach to the synthesis of polymeric

nanoparticles which concluded that nanoprecipitation would be the best suited preparation method for a campthothecinloaded system (Krishnamoorthy and Mahalingam, 2015), and an adapted approach could be implemented in the selection of synthetic methods for specific PLGA-based systems.

### DISADVANTAGES OF PLGA AS A NANOCARRIER

Even though the versatility of PLGA makes it an attractive option as a nanocarrier, it does present several challenges in nanomedicine. PLGA co-polymers are usually readily commercially available, but to obtain it in a high purity, and the specificity required for different molecular weights, lactic/glycolic acid ratios and end capped options can make it very costly (Danhier et al., 2012). Many formulations show poor drug loading and therefore would require large doses in order to achieve therapeutic concentrations of cargo at the target site. Furthermore, these systems often exhibit burst release kinetics, which would result in off target in vivo delivery. The degradation rate of PLGA is often unpredictable and the acidic degradation products have shown to affect the activity of the encapsulated drug (Sharma et al., 2016), and despite its biodegradability, reports have shown that the use of PLGA in medical devices may produce localized reaction at the site of delivery (Makadia and Siegel, 2011). Even though targeted PLGA based carriers theoretically have more efficient site specific delivery properties, the targeting moieties in these nanosystems can induce additional immunogenocity (Danhier et al., 2010). There are various in vivo physiological barriers and up- and down-regulation of cell surface receptors and other targets can also decrease the efficacy of the targeting agents in these systems. The adaptability of PLGA as a polymer for its specific application has resulted in it being used in many delivery systems and therefore it is difficult to make

#### REFERENCES


comprehensive predictions at this stage about its general behavior and toxicity (Sharma et al., 2016).

#### CONCLUSION AND FUTURE WORK

Even though PLGA is a polymer with many desirable features, there are various areas in which research can be conducted to improve the viability of PLGA based nanocarriers for clinical translation. Since PLGA has been developed in drug delivery systems for such a wide berth of applications, it should be precisely designed in terms of cargo suitability, particle size, drug entrapment and degradation kinetics, for its specific target. This would determine the choice of starting materials and in some cases, the method of preparation and would therefore remove some of the uncertainty present in several trial-anderror attempts in previous drug delivery systems. The innovative fabrication techniques mentioned above could also be attempted to increase control over homogeneity of the products. The use of in silico modeling for PLGA nanoparticles as an element of experimental design and could have tremendous implications for the future of nanoparticle design.

# AUTHOR CONTRIBUTIONS

DE, PK, YC, and VP from designing the framework and main content of the manuscript, revisions to optimize the manuscript, approved the final submission, and the manuscript was accomplished with contributions.

#### FUNDING

This work was supported by a National Research Foundation (NRF) SARChI grant.




carrier for the delivery of drugs to the brain. Colloids Surf. A Physicochem. Eng. Asp. 392, 335–342. doi: 10.1016/j.colsurfa.2011.10.012





**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 Essa, Kondiah, Choonara and Pillay. 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.

# Chitosan Nanoparticles: Shedding Light on Immunotoxicity and Hemocompatibility

Sandra Jesus 1†, Ana Patrícia Marques 1†, Alana Duarte1,2†, Edna Soares 1,2 , João Panão Costa1,2, Mariana Colaço1,2, Mélanie Schmutz <sup>3</sup> , Claudia Som<sup>3</sup> , Gerrit Borchard<sup>4</sup> , Peter Wick <sup>5</sup> and Olga Borges 1,2 \*

<sup>1</sup> Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal, <sup>2</sup> Faculty of Pharmacy, University of Coimbra, Coimbra, Portugal, <sup>3</sup> Laboratory for Technology and Society, Empa Swiss Laboratories for Materials Science and Technology, St. Gallen, Switzerland, <sup>4</sup> Institute of Pharmaceutical Sciences of Western Switzerland, University of Geneva, Geneva, Switzerland, <sup>5</sup> Laboratory for Particles-Biology Interactions, Empa Swiss Laboratories for Materials Science and Technology, St. Gallen, Switzerland

#### Edited by:

Mustafa Culha, Yeditepe University, Turkey

#### Reviewed by:

Subhra Mandal, Creighton University, United States Nanasaheb D. Thorat, University of Limerick, Ireland

> \*Correspondence: Olga Borges olga@ci.uc.pt

†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: 31 October 2019 Accepted: 03 February 2020 Published: 21 February 2020

#### Citation:

Jesus S, Marques AP, Duarte A, Soares E, Costa JP, Colaço M, Schmutz M, Som C, Borchard G, Wick P and Borges O (2020) Chitosan Nanoparticles: Shedding Light on Immunotoxicity and Hemocompatibility. Front. Bioeng. Biotechnol. 8:100. doi: 10.3389/fbioe.2020.00100 Nanoparticles (NPs) assumed an important role in the area of drug delivery. Despite the number of studies including NPs are growing over the last years, their side effects on the immune system are often ignored or omitted. One of the most studied polymers in the nano based drug delivery system field is chitosan (Chit). In the scientific literature, although the physicochemical properties [molecular weight (MW) or deacetylation degree (DDA)] of the chitosan, endotoxin contamination and appropriate testing controls are rarely reported, they can strongly influence immunotoxicity results. The present work aimed to study the immunotoxicity of NPs produced with different DDA and MW Chit polymers and to benchmark it against the polymer itself. Chit NPs were prepared based on the ionic gelation of Chit with sodium tripolyphosphate (TPP). This method allowed the production of two different NPs: Chit 80% NPs (80% DDA) and Chit 93% NPs (93% DDA). In general, we found greater reduction in cell viability induced by Chit NPs than the respective Chit polymers when tested in vitro using human peripheral blood monocytes (PBMCs) or RAW 264.7 cell line. In addition, Chit 80% NPs were more cytotoxic for PBMCs, increased reactive oxygen species (ROS) production (above 156µg/mL) in the RAW 264.7 cell line and interfered with the intrinsic pathway of coagulation (at 1 mg/mL) when compared to Chit 93% NPs. On the other hand, only Chit 93% NPs induced platelet aggregation (at 2 mg/mL). Although Chit NPs and Chit polymers did not stimulate the nitric oxide (NO) production in RAW 264.7 cells, they induced a decrease in lipopolysaccharide (LPS)-induced NO production at all tested concentrations. None of Chit NPs and polymers caused hemolysis, nor induced PBMCs to secrete TNF-α and IL-6 cytokines. From the obtained results we concluded that the DDA of the Chit polymer and the size of Chit NPs influence the in vitro immunotoxicity results. As the NPs are more cytotoxic than the corresponding polymers, one should be careful in the extrapolation of trends from the polymer to the NPs, and in the comparisons among delivery systems prepared with different DDA chitosans.

Keywords: chitosan nanoparticle, immunotoxicity, hemocompatibility, deacetylation degree, endotoxin-free, inflammation, reactive oxygen species, PBMCs

# INTRODUCTION

Studies have shown that nanoparticles (NPs) can interact with different components of the immune system, resulting in immunosuppression and in immunostimulation (Dobrovolskaia and McNeil, 2007). Although these interactions can be purposeful and desirable in increasing the efficacy of vaccines, cancer immunotherapy or immunotherapies for autoimmune diseases, they can also be unexpected and undesirable, causing hypersensitivity reactions, anaphylaxis, coagulopathies and body defense decrease (Dobrovolskaia and McNeil, 2007).

Chitosan (Chit) is the common name given to a family of natural polysaccharide polymers obtained from the deacetylation of chitin. Chit is a cationic polymer, considered non-toxic, biodegradable and biocompatible and is therefore extensively investigated in nanobiomedical research (Ali and Ahmed, 2018). Chit has been granted FDA Generally Recognized As Safe (GRAS) designation (GRN n◦ 73, 170, 397 and 443) and is widely used in dietary supplements (U.S. FDA, 2019a) as well as in medical devices, such as wound dressings and gels (U.S. FDA, 2019b). Chit is known for its mucoadhesive properties and its ability to stimulate cells of the immune system, which supports the value of investigating Chit NPs as vaccine adjuvants (Dedloff et al., 2019). For this purpose, it has long been used by the group with various antigens, such as the hepatitis B surface antigen (HBsAg) (Borges et al., 2008; Lebre et al., 2016; Jesus et al., 2017, 2018; Soares et al., 2018a,b; Bento et al., 2019), the protective antigen (PA) from anthrax (Bento et al., 2015) or antigens from Schistosoma mansoni (Oliveira et al., 2012). Nevertheless, in the literature, Chit NPs have also been tested as drug delivery systems, without considering its immunomodulatory activity. An example of this situation is the numerous studies with the encapsulation of insulin into chitosan particles (Al Rubeaan et al., 2016). Furthermore, although there are several studies evaluating Chit NP toxicity in vitro, most of them do not assess the dysregulation of the immune system function (immunotoxicity). From the ones that do, the results are frequently contradictory. These contradictions and ambiguity may be due to differences in the used Chit polymers or in vitro methodology, namely the cellular model, NP concentration and incubation period. Moreover, it has been observed that most of the studies do not properly characterize, or at least do not report, both the polymer and the derived NPs, nor use or report adequate controls to screen NP interferences or monitor the presence of endotoxin contamination (Jesus et al., 2019). Notably, in the context of Safe-by-Design (SbD) of new polymeric NPs for drug delivery, it is necessary to rely on assertive results of immunotoxicity and hemocompatibility, obtained with properly characterized polymeric NPs.

The aim of this study is to explore the influence of the DDA of Chit polymer on immunotoxicity and hemocompatibility of Chit NPs. Therefore, murine RAW 264.7 cells, Peripheral Blood Mononuclear Cells (PBMCs) and whole blood were used as representative in vitro models for the immune system.

Nitric oxide (NO), reactive oxygen species (ROS) and cytokine production, cell viability, hemolysis, coagulation times and platelet aggregation were studied using appropriate controls under endotoxin-free conditions, and following protocols and recommendations, with slight changes, described by the European Nanomedicine Characterization Laboratory (EU-NCL) (EU-NCL, 2019).

#### MATERIALS AND METHODS

#### Chitosan Polymers

Two different low molecular weight (LMW) Chitosans (ChitoClearTM) were kindly donated by Primex BioChemicals AS (Avaldsnes, Norway). According to the supplier's specifications, one Chit had a lower deacetylation degree (DDA) and a viscosity of 13 cP (1% solutions in 1% acetic acid), while the other had higher DDA and a viscosity of 71 cP. Their exact DDA was found to be 80 and 93%, respectively, using the methodology described below.

The polymers were purified using a routine technique used in our laboratory and previously described by us (Lebre et al., 2019). Briefly, 1 g of Chit was suspended in 10 mL NaOH (1 M) solution. This suspension was heated between 40 and 50◦C under continuous magnetic stirring for 3 h. After this time, the suspension was allowed to reach room temperature and was filtered using a Buchner funnel. Insoluble Chit on the filter was washed with water and then recovered to be further dissolved in 200 mL of 1% acetic acid solution and stirred for 1 h at room temperature. The Chit solution was then filtered through a 0.45µm filter and 1 M NaOH solution was used to adjust the pH of the filtrate to pH 8.0 to precipitate Chit. The precipitate was then washed with water through three consecutives 30 min centrifugations at 4500 × g. The precipitate was recovered and freeze dried. To note that deionized water was used to obtain the purified polymer for the first experiments, optimization of the NP production method and physicochemical characterization, while LPS-free water was used to obtain LPS-free chitosan for cell in vitro studies. The purified polymers were used in all the methods described below.

Chit deacetylation degree and mean molecular weight were obtained by nuclear magnetic resonance (1H-NMR) and by size exclusion chromatography (SEC), respectively.

Deacetylation degree was determined as previously described (Lavertu et al., 2003). The DDA was calculated using the peaks of proton at the position 1 of deacetylated (H1D) and acetylated (H1A) monomer:

$$\text{DDA (\%)} = \left(\frac{\text{H1D}}{\text{H1D} + \text{H1A}}\right) \times 100 \tag{1}$$

where H1D is shifted at 5.21 ppm and H1A at 4.92 ppm.

For Chit molecular weight (MW) analysis, two types of Chit polymers (before and after purification) were dissolved in 0.1 M acetic buffer (pH 4.0) containing 0.3 M NaCl to obtain 1 mg/mL solutions. Then they were filtered through 0.22µm filters and collected in the chromatographic sample vials. For each analysis, 100 µL were injected at a flow rate of 1 mL/min at room temperature. Each sample was measured in triplicate. The interpretation of the obtained results was done using Mnova software.

Chit polymer particle size (micrometer range) was also characterized in acetate buffer and cell culture media using Beckman Coulter LS 13 320 Laser Diffraction Particle Size Analyzer (Beckman Coulter Inc., Brea, CA, USA).

# Preparation and Characterization of Chitosan Nanoparticles

To prepare both Chit NPs, each of the polymers (Chit 80% DDA and Chit 93% DDA) were dissolved at 0.1% (w/v) concentration in 1% (v/v) of acetic acid, and the pH was further adjusted to 4.6–4.8 using 10 N NaOH. Chit NPs spontaneously formed upon dropwise addition of 1.750 mL of sodium tripolyphosphate (TPP, 0.16% w/v) to 10 mL of Chit solution under high-speed homogenization. The final suspension remained in maturation during 30 min under magnetic stirring.

Chit NPs produced with Chit with 80% DDA were concentrated, washed with LPS-free water and concentrated again by centrifugation using Vivaspin 20 centrifugal concentrator (MWCO 300 kDa, 3,000 g). Chit NPs produced with Chit with 93% DDA were concentrated by centrifugation at 10,000 g (15 min) and centrifuged again at 7,000 g (15 min) with LPS-free water.

To evaluate if all the Chit polymer used for the NP production was effectively cross-linked with TPP, Cibacron Brilliant Red 3B-A dye assay (Muzzarelli, 1998) was used to quantify the free Chit that remained in solution after NPs preparation. The quantification was performed in 3 mL of the supernatants obtained by the previously described centrifugations, which were added to 100 µL of glycin/HCl buffer, 1 mL of the dye solution (0.015% dye in water, w/v) and 900 µL of ultra-pure water. The samples were left for 20 min in agitation and then the absorbance was read at 575 nm. The quantification was performed by interpolating the values with the values from a calibration curve ranging from 0.0004 to 0.0020% of Chit. The concentration of the Chit NPs was calculated subtracting to the initial mass of the chit used to prepare the NPs, the mas of free chitosan.

DelsaTM Nano C particle analyzer (Beckman Coulter, CA, USA) was used to measure NP size by dynamic light scattering (DLS) and the zeta potential through electrophoretic light scattering (ELS). Samples comprised the aqueous concentrated dispersions obtained after centrifugation, which were diluted with water before the measurements.

Concentrated samples of Chit NPs were tested for physicochemical stability when dispersed in cell culture media at 37◦C for a maximum of 24 h. The resulting particle size and zeta potential were evaluated in DelsaTM Nano C particle analyzer (Beckman Coulter, CA, USA).

Images of Chit NPs, were acquired by two microscopy techniques. Transmission Electron Microscopy (TEM) using a FEI-Tecnai G2 Spirit Biotwin, (20–120) kV microscope (FEI company, Hillsboro, OR, USA) with NPs dispersed in water and subsequently dried out in the grid and observed with no contrast. For the second microscopy technique, a High resolution Scanning Electron CryoMicroscope (CryoSEM) (JEOL JSM 6301F/ Oxford INCA Energy 350/ Gatan Alto 2500) was used. The NP suspension was rapidly cooled in slush nitrogen, fractured and sublimated for 120 s at −90◦C, before coating with Au/Pd. The sample was studied at −150◦C.

For in vitro immunotoxicity studies, Chit purification and Chit NP production were conducted under endotoxinfree conditions following a methodology already published by our group (Lebre et al., 2019). All the reagents involved in NP production were tested with an endotoxin detection kit (Pyrochrome <sup>R</sup> Endpoint Chromogenic Endotoxin Testing, maximum sensitivity of 0.001 EU/mL, Associates of Cape Cod, Inc., East Falmouth, MA, USA) according to manufacturer's instructions.

#### In vitro Studies With RAW 264.7 Cell Line

RAW 264.7 cell line (ATCC <sup>R</sup> TIB-71TM) was acquired from ATCC (Manassas, VA, USA), cultured in Dulbecco's modified eagle's medium (DMEM, supplemented with 10% heat inactivated fetal bovine serum (FBS), 1% Penicillin/Streptomycin, 10 mM HEPES and 3.7 g/L sodium bicarbonate) and used until passage 18.

#### Cell Viability

The Cell viability of Chit NP and polymers was evaluated in RAW 264.7 cells using the 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) assay, performed in 96-well plates and cells plated at a density of 2 × 10<sup>4</sup> cells per well. Serial dilutions of Chit NPs and Chit polymers ranged from 312 to 5,000µg/mL final concentration in the well were incubated with the cells for 24 h, at 37◦C and 5% CO2. Simultaneously, the NPs solvent (supernatant from the last washing centrifugation) and the polymer solvent (acetate buffer) were also tested in a dilution equivalent to the most concentrated samples. Then, 20 µL of MTT solution (5 mg/mL, in PBS) were added to each well and incubated for additional 1 h 30 min. To ensure the dissolution of the formazan crystals, cell culture medium was replaced by 200 µL of dimethyl sulfoxide (DMSO).

The resultant colored solution OD was measured at 540 nm and 630 nm. Cell viability (%) was calculated by the following equation:

$$\text{Cell viability (\%)} = \frac{\text{(OD sample (540 nm) - OD sample (630 nm))}}{\text{(OD control (540 nm) - OD control (630 nm))}} \times 100 \text{ (2)}$$

The half maximal inhibitory concentration (IC50) of NPs that cause death or inhibition of the growth of 50% of cells was calculated by using the Log (NP concentration) vs. normalized response - variable slope analysis for the non-linear fit using Prism 6.0 (GraphPad Software, San Diego, CA, USA).

Interference controls were performed to guarantee the validity of the assay with the samples as suggested by Rösslein et al. (2015). Therefore, NPs and polymers in cell culture media without cells were plated in 96-well plates and the absorbance was measured (540 and 630 nm).

#### Production of Reactive Oxygen Species

The production reactive oxygen species (ROS) was assessed using the dichlorofluorescein diacetate (DCFH-DA) probe (Molecular Probes <sup>R</sup> , Life Technologies, Eugene, OR, USA). RAW 264.7 cells were incubated in black 96-well plates for 24 h at 37◦C and 5% CO2, at a density of 0.5 × 10<sup>5</sup> cells per well.

After that, serial dilutions of Chit NPs and Chit polymers (38µg/mL to 156µg/mL) were incubated with the cells in DMEM for 24 h at 37◦C and 5% CO2, to evaluate ROS stimulation. The NPs solvent (supernatant from the last washing centrifugation) and the polymer solvent (acetate buffer) were also tested in a dilution equivalent to the most concentrated samples. Lipopolysaccharide (LPS, 1µg/mL, from Salmonella enterica serotype minnesota, Sigma-Aldrich, Saint Louis, MO, USA) was used as a positive control or in combination with the same NP and polymer concentrations to test if the NPs were able to inhibit LPS stimulated ROS production.

Then, the cell culture medium was replaced by DCFH-DA (50µM) in serum-free DMEM and the cells were incubated for additional 2 h at 37◦C and 5% CO2. The resulting fluorescence was read at 485/20 (excitation) and 528/20 nm (emission) wavelengths.

To calculate the stimulation of ROS production (fluorescence fold increase) or the inhibition of ROS production (%) upon stimulation with LPS apply the following Equation (3) and (4), respectively.

$$\text{ROS production} = \frac{\text{Fluorescence}\_{\text{SAMPL}}}{\text{Fluorescence}\_{\text{NEGATE}} \text{CONTR}} \quad \text{(3)}$$

$$\text{ROS inhibition(\%)} = \frac{\text{Fluorescence}\_{\text{SAMPL}}}{\text{Fluorescence}\_{\text{POSITVECONTROL}}} \times 100 \text{ (4)}$$

Interference controls were performed to guarantee the validity of the assay with the samples. Therefore, NPs and polymers in cell culture media without cells were plated in black 96 well plates and all procedures were followed as in the original assay described.

#### Nitric Oxide Production

Nitric oxide (NO) has a short half-life in oxygen-containing aqueous solutions, often attributed to a rapid oxidation to nitrite. Therefore, NO production by RAW 264.7 cells was estimated based on nitrite quantification using the Griess reagent [1% (w/v) sulphanilamide mixed with 0.1% (w/v) naphthylethylenediamine dihydrochloride (1:1), both solutions previously dissolved in 2.5% (v/v) phosphoric acid].

RAW 264.7 cells were incubated in 48-well plates at a density of 2.25 × 10<sup>5</sup> cells per well for 24 h at 37◦C and 5% CO2. After that, cell culture medium was replaced by serial dilutions of Chit NPs and Chit polymers (38–156µg/mL), diluted in cell culture medium without phenol red, and cells incubated for 24 h at 37◦C and 5% CO2. The NPs solvent (supernatant from the last washing centrifugation) and the polymer solvent (acetate buffer) were also tested in a dilution equivalent to the most concentrated samples. LPS was used as a positive control (1µg/mL). To test whether the NPs were able to inhibit LPS stimulated NO production, the same NP and polymer concentrations were incubated together with LPS (1 µg/mL).

After that, 100 µL of each cell supernatant were collected and plated in a 96-well plate and combined with an equal volume of the Griess Reagent. Several sodium nitrite solutions (0µM to 80µM) were also plated in duplicate to perform the calibration curve. The absorbance (Abs) of the samples was measured at 550 nm and the NO concentration (µM) was extrapolated from the calibration curve.

To calculate the inhibition of NO production upon stimulation with LPS, the Equation (5) was used.

$$\text{NO inhibition} \left(\%\right) = \frac{\text{[NO] (\mu\text{M})}\_{\text{SAMPLE}}}{\text{[NO] (\mu\text{M}) \_{\text{POSTIVE CONTROL}}} \times 100 \text{ (5)}$$

Interference control was performed to guarantee the validity of the assay with the samples containing the particles. Therefore, 100 µL of NPs and polymers in DMEM without phenol red and without cells were plated in 96-well plates. Additionally, the NO calibration curve was performed in the presence of NPs and polymers, by plating in 96-well plates 50 µL of the samples and 50 µL of the standards used in calibration curve. Then, an equal volume of the Griess Reagent was added to each well and the absorbance was read as described above. This interference control was made at least in duplicate.

#### In vitro Studies With Peripheral Blood Mononuclear Cells PBMC Isolation

Peripheral blood (buffy coat) was kindly given by IPST, IP (Coimbra, PT) and was obtained from healthy donors in heparinized syringes followed by serum depletion. PMBCs were isolated on a density gradient with Lymphoprep (Axis-Shield, Dundee, SCT) according to the provider's guidance protocol, with minor modifications. Briefly, the blood dilution performed was 1:5 (v/v) in 0.9% sodium chloride, the centrifugation step was performed at 1,190 × g for 20 min (20◦C) and the mononuclear cell dense ring was collected and washed with PBS (pH = 7.4 at 37◦C) through consecutive centrifugations (487 × g, 10 min, 20◦C) until the supernatant was clear. At the end, cells were suspended in Roswell Park Memorial Institute Medium (RPMI 1640) supplemented with 1% Penicillin/Streptomycin and 10% heat-inactivated FBS.

#### Cell Viability

Chit NP and polymer toxicity in PBMCs was assessed by MTT as described previously for RAW 264.7 cells, with some modifications. Briefly, cells were plated at a concentration of 7.5 × 10<sup>6</sup> cells/well, test samples ranged from 2.44 to 5,000µg/mL and MTT incubation was prolonged for 4 h. To ensure the dissolution of the formazan crystals, cell culture plates were centrifuged (800 × g, 25 min, 20◦C) and 180 µL/well of the culture medium were replaced by an equal volume of DMSO.

Cell viability results obtained with the MTT assay were confirmed with propidium iodide (PI) assay, using four different NP concentrations. Cells incubated with the NPs were centrifuged (800 × g, 25 min, 20◦C), resuspended in PBS and collected for flow cytometry analysis (BD FACSCalibur, BD Biosciences, Bedford, MA, USA). A volume of 2 µL of PI solution was added immediately before the analysis to achieve a final concentration of 0.5 µg/mL.

#### Cytokine Secretion

To analyze the cytokine secretion induced by Chit NPs, cells were plated in 96-well plates at a density of 2.5 × 10<sup>5</sup> cells per well. Chit NPs and polymers (100µg/mL) and positive controls (LPS 2 ng/mL, Con A 5µg/mL) were incubated with the cells for 24 h, at 37◦C and 5% CO2. Then, cell culture plates were centrifuged (800 × g, 25 min, 20◦C) and the supernatants were collected for Enzyme-Linked Immunosorbent Assay (ELISA) according to manufacturer's instructions (Human TNF-α and IL-6 Standard TMB ELISA Development Kit, Peprotech, NJ, USA).

Interference controls were performed to guarantee the validity of the assay with the samples. Therefore, NPs and polymers in RPMI were incubated for 24 h, at 37◦C and 5% CO<sup>2</sup> without cells, in the presence of several concentrations of cytokine standards (TNF-α and IL-6), as used in the ELISA calibration curve. The same concentrations of each cytokine were also incubated in RPMI in the absence of the samples. After that time, supernatants were collected and analyzed by ELISA as described for the samples with cells, according to manufacturer's instructions.

#### In vitro Studies With Human Blood

Blood was collected from healthy volunteers at the Clinical Laboratory Analysis of Faculty of Pharmacy (University of Coimbra, Portugal). A written informed consent was obtained from all participants. Anonymous blood samples were used by the researchers for the hematological in vitro assays.

#### Hemolysis Assay

To perform hemolysis assay, plasma free hemoglobin (PFH) concentration was required to be below 1.0 mg/mL. Whole blood was collected in heparinized tubes and diluted in PBS to adjust total blood hemoglobin (TBH) concentration to 10 mg/mL ± 2 mg/mL (TBHd). A volume of 100 µL of cyanmethemoglobin (CMH, blank), Chit NP suspensions, Chit polymer suspensions, PBS (negative control), Triton-X-100 (positive control) or NPs solvent (interference control) were added to 700 µL of PBS in different tubes. Then, 100 µL of TBHd was added and incubated at 37◦C for 3 h ± 15 min. NPs were also incubated with PBS without blood to evaluate the possible NP interference with the assay. Then, the mixture was centrifuged at 800 × g for 15 min. A volume of 100 µL of supernatant and 100 µL of CMH reagent were added to a 96-well plate. The CMH reagent was prepared by mixing 1,000 mL Drabkin's reagent and 0.5 mL of 30% Brij 35 solution (Sigma-Aldrich, Saint Louis, MO, USA). The absorbance (OD) was read at 540 nm. The percentage of hemolysis was calculated using the following equation:

$$\text{(Hemolyus)} \left(\% \right) = \frac{\text{(OD sample (540 nm) - OD negative control (540 nm))}}{\text{(OD TBHd (540 nm) - OD negative control (540 nm))}} \times 100 \text{ (\ $\$ )}$$

#### Coagulation Assay

The two pathways of blood coagulation, the activated partial thromboplastin time (APTT) and the prothrombin time (PT) were separately tested. Blood was collected using sodium citrate tubes and the plasma was obtained by centrifugation of the blood at 2500 × g for 10 min. Plasma (450 µL) was incubated with a volume of 50 µL of Chit NPs and Chit polymer suspensions (two final concentrations: 0.1 and 1 mg/mL), for 30 min at 37◦C. Then, samples were evaluated using Bio-TP LI (PT) and Bio-CK (APTT) kits (Biolabo S.A.S., Maizy, France) according to manufacturer's instructions, in an Option 4 plus coagulation analyzer (BioMérieux, Marcy-l'Étoile, France).

#### Platelet Aggregation Assay

Platelet-rich plasma (PRP) was obtained from blood collected in sodium citrate tubes, and centrifuged at 200 × g for 16 min. Platelet-free plasma (PFP) was obtained after blood centrifugation at 2,500 × g for 10 min, followed by plasma centrifugation at 18,000 × g for 5 min. A volume of 100 µL of PRP or 100 µL of PFP were added to 96-well plates and incubated for 5 min at 37◦C. A volume of 25 µL of Chit NPs at 2 mg/mL, saline solution (negative control) and calcium chloride 0.25 M or collagen (positive controls) were added to the wells with PRP and incubated for 30 min at 37◦C. Then, 4 µL of Giemsa dye was added to each well and incubated for 5 min. Finally, a 1:200 dilution with saline solution was applied for platelet counting (PC) using a light microscope. Chit NPs were also incubated with PFP to evaluate the NPs interference in plasma.

The percentage of platelet aggregation was calculated using the equation 7.

$$\text{Platlelet aggregation (\%)} = \frac{\text{(PC negative control -- PC sample)}}{\text{PC negative control}} \times 100 \quad \text{(7)}$$

# Statistical Analysis

Results were expressed as mean ± standard error of the mean (SEM). Prism 6.0 (GraphPad Software, San Diego, CA) was used for all statistical analysis. Statistical significance was assessed using one-way ANOVA.

# RESULTS

# Physicochemical Characterization of Polymers and Nanoparticles

#### Polymer Purification Reduces the Molecular Weight of the Lower Deacetylation Degree Chitosan

The characterization of the polymers used and the nanoparticulate delivery system developed is critical to prevent erroneous interpretations of resultant immunotoxicity findings. Different Chit characteristics can have different biological effects. Unfortunately, most studies addressing biological activity of Chit NPs lack the used polymer characterization, which also restricts comparisons among studies.

The two Chit polymers used in this study were purified under endotoxin-free conditions to eliminate possible contaminants. Since the purification process involves harsh conditions, namely heating the polymer suspension in NaOH 1 M, their DDA and MW were assessed before and after purification and the results presented in **Table 1A**. Chit deacetylation experienced no significant alterations, resulting in polymers with 80 and 93% DDA (Chit 80% and Chit 93%, respectively). In contrast, the MW before and after purification for the lower DDA Chit (Chit 80%) was altered. An important decrease from 168 to 49 kDa is compatible with the fact that lower DDA Chit has higher TABLE 1 | Physicochemical characterization of Chit polymers and NPs. (A) Polymer molecular weight (MW), deacetylation degree (DDA), and size in acetate buffer and after resuspension in DMEM and RPMI at 37◦C (Mean ± SEM). (B) Chit 80% and Chit 93% NPs size, polydispersity index and zeta potential (ζ ), in water and after resuspension DMEM and RPMI at 37◦C (Mean ± SEM). (C) Endotoxin contamination evaluated with the Pyrochrome kit for Chit 80% NPs, Chit 93% NPs, Chit 80%, and Chit 93% and TPP solution. Endotoxin contamination of pyrogen-free water was also evaluated for comparison. Mean ± SEM; n = 3 (three different batches).


enzymatic and acid hydrolysis degradation rate (Kurita et al., 2000; Vårum, 2001; Szymanska and Winnicka, 2015).

Chit issoluble in acidic conditions, which is incompatible with cell culture as it leads to cell death. Therefore, in vitro studies with Chit polymers (purified raw material) were performed with Chit suspended in acetate buffer (pH = 5.0), further diluted in cell culture medium (156.25µg/mL). Particle size in acetate buffer and cell culture media is illustrated in **Table 1A**. The mean average size of these particle suspensions was around 500µm in all situations.

#### Chitosan With Higher Molecular Weight and Deacetylation Degree Leads to Larger-Sized NPs

Chit NPs were successfully produced by ionic gelation method, using TPP as the crosslink (Chit 80% NPs and Chit 93% NPs). These NPs were isolated and concentrated in water. Importantly, the analysis of the first supernatants revealed that more than 99% of the Chit used in the production was retained in the NPs. This result was important to calculate Chit NP concentration.

After isolation and concentration, NP mean particle size, polydispersity index (PDI), and zeta potential (ζ) were determined by DLS and ELS, respectively, and are summarized in **Table 1B**. Results illustrate the effect of the different Chit on the NP characteristics. In fact, the same methodology, when applied to Chit polymers with different DDA and MW, resulted in NPs with different sizes. Lowering the DDA from 93 to 80% caused the mean particle size to fall from 292 to 127 nm. These average particle sizes were illustrated by TEM and SEM analysis. The round shape of the NPs was the second conclusion inferred by observing the images (**Figures 1A,B**) of both techniques. Concerning zeta potential, both Chit 93% and Chit 80% NPs presented a positive charge when dispersed in deionized or pyrogen-free water, although slightly more positive for Chit 80% NPs (+20 and +29 mV, respectively).

Due to the complexity of cell culture media, and the variability of their supplementation, results from NP colloidal system characterization in water are not transposable to in vitro conditions (Moore et al., 2015). Chit NPs were therefore characterized in cell culture media to understand the changes that NPs experience during in vitro studies. Chit 80% and Chit 93% NPs were added to DMEM and RPMI (containing FBS) at 37◦C at a concentration of 156.25µg/mL for further size and PDI measurement after 1 and 24 h, and zeta potential measurement after 1 h (**Table 1B**). Even though the DLS methodology for size analysis in complex media (such as cell culture medium) has limitations, it can give us some insights about changes occurring to the different Chit NPs. Most notably, the suspension

of both Chit NPs in RPMI and DMEM resulted in increased PDI, meaning an increase of the size heterogeneity. The zeta potential of the NPs decreased when measured in both cell culture media (ranging from −2 to −5 mV) (**Table 1B**). This change induced by the adsorption of negatively charged proteins from the medium, to positively charged Chit residues, forms a protein corona, decreasing the suspension stability. Under these conditions, the appearance of aggregates is inevitable which is part of the explanation for the PDI increment. To further complement the information given by the PDI and intensity average size, graphics from size distribution illustrate the different size populations of Chit 80% NPs and Chit 93% NPs in the different media (**Figure 2**). In the water, the size of both NP was distributed over a single peak (**Figures 2A,D**), while in cell culture media, there were at least three independent peaks (**Figures 2B,C,E,F**). We can hypothesize that the alterations observed in cell culture media size dispersion, including smaller and bigger size populations simultaneously, were induced not only by the presence of proteins, but also by the high ion content in comparison to water (Moore et al., 2015). Furthermore, as the media composition is different between RPMI and DMEM, the observed changes in the NP size distribution were not similar. A comparable phenomenon was described by Yang et al. (2018) for silica and silica coated nanoparticles, whose great stability in buffered saline was not kept in cell culture media, where the authors verified the erosion of surface silica by DMEM ingredients.

#### Endotoxin-Free Conditions Guarantee the Production of LPS-Free Nanoparticles

The last step of characterization was related to endotoxin contamination. As previously mentioned, Chit polymers were purified by a method published by our group (Lebre et al., 2019). The method allows the obtainment of endotoxin-free chitosan, proved by two methods: Limulus Amebocyte Lysate (LAL) test and the absence of IL-6, secreted by dendritic cells (DCs), cultured in the presence of chitosan. The chitosan does not induce IL-6 secretion by DCs and endotoxins do that stimulation. Furthermore, for in vitro immunotoxicity studies, the NP production was performed under those conditions, to avoid endotoxin contamination, as the presence of these molecules can easily lead to false positive results. To assure that Chit purification and Chit NP production were successfully achieved, both Chit polymers and NPs as well as the pyrogenfree water and the TPP solution used for NP production, were submitted to LAL test. Importantly, before establishing the methodology for endotoxin quantification with Pyrochrome <sup>R</sup> testing kit, all recommended tests to evaluate sample interference with LAL test were done to guarantee the suitability of the LAL test for Chit NPs, as described in the manufacturer's instructions. The results were presented in **Table 1C**, and show that all tested samples were not significantly different from pyrogen-free water, the negative control, and all were far below 0.25 EU/mL, which is the limit for water for injection according to main health authorities (Ph. Eur. 9.0, 2019). Thus, it was demonstrated that the process and conditions used to minimize the contamination and remove existent endotoxins during Chit purification and NP production was effective, and that Chit polymers and NPs used in immunotoxicity tests were indeed LPS-free, supporting the reliability of the results.

#### In vitro Studies With RAW 264.7 Cell Line

The monocyte/macrophage-like RAW 264.7 cells have been widely used for 40 years, as a suitable in vitro model, since they present unique phenotype and functional characteristics of macrophages (Roberts et al., 2018). Nevertheless, these cells should be used carefully since their functional stability is not maintained at high passage number. Indeed, a recent article

mentions the phenotype and functional characteristics to remain stable from passage 10 to 30 (Roberts et al., 2018), and the American Type Culture Collection (ATTC) recommends to use them until passage 18.

#### Chitosan Nanoparticles Are More Cytotoxic for RAW 264.7 Cells Than Chitosan Polymers

The evaluation of the cytotoxic profile of Chit NPs and polymers was performed using the MTT metabolic activity assay, over a wide range of concentrations as illustrated in **Figure 3**. Results showed that Chit 80% and Chit 93% polymers were not cytotoxic in the concentration range tested (purple and orange lines, respectively), while Chit 80% and Chit 93% NPs induced significant decrease of cell viability above 2,500 and 3,000µg/mL, respectively (**Figure 3A**). Based on the nonlinear regression analysis of the cell viability data of the Chit NPs, nonsignificant differences were found for the IC50 of Chit 80% NPs and Chit 93% NPs (**Figure 3B**).

The reduction of the reagent MTT by cells leads to the generation of insoluble crystals of formazan that once dissolved in DMSO generate a purple signal (van Meerloo et al., 2011). Since it is a colorimetric assay, and although the cell medium with the testing sample was aspirated before solubilizing the formazan crystals, NP interferences with the readout were tested to validate the assay (**Figure 3C**). As it is possible to observe, the measured absorbance (Abs) was not increased by the presence of the NPs or polymer suspension. Additionally, to guarantee that the cell viability results were only related with the NP and polymers, and not with the solvents, the supernatants collected from the NPs last washing step with water, as well as the acetate buffer used to disperse the polymers, were also tested using MTT assay (**Figure 3D**). Results showed that the solvents did not cause any decrease in cell viability.

#### Both Chitosan Polymers Hamper Nitric Oxide Release After LPS Stimulation and Only the Lower Deacetylation Degree Chitosan Induces Oxidative Stress

Reactive oxygen species (ROS) are unstable molecules that easily react with other molecules and may cause damage to DNA, RNA, proteins and ultimately lead to cell death, when accumulated (Schieber and Chandel, 2014).

To evaluate the effect of Chit polymers and Chit NPs on ROS production by RAW 264.7 cells, four different concentrations were used. As it is possible to see in **Figure 4A**, only Chit 80% NPs and the respective Chit polymer were able to induce ROS production, under non-cytotoxic concentrations. The increase in ROS production was concentration dependent, however, for the concentration range tested, the effect was not as high as LPS-induced ROS production. On the other hand, Chit 93% NPs and polymer had no effect on ROS production by RAW 264.7 cells. Importantly, all tested conditions did not induce cellular death as confirmed by the MTT assay performed at the end of each experiment (**Figure S1A**). In order to have a more complete picture, studies were conducted to evaluate that the polymers and NPs would not play an inhibitory role in the production of ROS by cells stimulated with LPS. Therefore, increasing concentrations of Chit polymer or Chit NPs were incubated together with cells and 1µg/ml of LPS. Results in **Figure 4B** show that no inhibitory effect was observed for any of the tested samples. Consequently, it was possible to conclude that Chit 80%

(n ≥ 3).

NPs, Chit 93% NPs and the respective polymers, when used in non-cytotoxic concentrations (cell viability results are on **Supplementary Figure S1B**), were not able to reduce the LPSinduced ROS production.

The possibility of having the nanoparticles interfering with the methods should not be ruled out, leading to false positives or false negatives. So, to evaluate the interference of Chit NPs and Chit polymers in the fluorescence readouts, the ROS production assay was performed without cells and at the highest polymer and NPs concentrations. The values obtained for test samples were similar to the medium alone (**Figure 4E**), meaning that they do not interfere with ROS measurement. Additionally, the possible interference of solvents was also assessed under the same testing conditions and as shown in **Figure 4F**, no stimulation of ROS production, as the fluorescence increase fold values were around 1.

NO is an important inflammatory mediator released by macrophages during inflammation, being one of the main cytostatic, cytotoxic, and pro-apoptotic mechanisms of the immune response (Bosca et al., 2005). NO production by RAW 264.7 cell line was measured using the Griess reaction method. Again, all test samples were sterile and endotoxin-free in order to prevent false positive results, and used in adequate concentrations that did not affect cell viability (Cell viability study in **Supplementary Figures 1C,D**).

With the aim to evaluate whether one of the polymers or Chit NPs would be able to induce cells to produce NO, samples were incubated with the RAW 264.7 cells for 24 h and the results were presented in **Figure 4C**. None of the Chit NPs or polymer concentrations tested induced NO production. Additionally, to evaluate whether the NPs and polymers had an inhibitory effect on NO production when cells were stimulated by LPS, increasing concentrations of the polymers and NPs were incubated with cells and with 1µg/ml LPS. The results shown in **Figure 4D** indicate that there was a slight but significant inhibitory effect on LPS-induced NO production, at all concentrations tested when compared to the LPS control. Since the Chit and Chit NP concentrations tested did not induce significant reduction in cell viability (**Supplementary Figure 1D**) we can exclude the hypothesis that it was a consequence of cellular death.

For all NPs, the possible interference with optical detection methods is a hypothesis that should be tested before doing

FIGURE 4 | Immunotoxicity assays in RAW 264.7 cell line. All assays were performed with non-cytotoxic concentrations of NPs, polymers and controls (evaluated by MTT assay after every experiment). (A) ROS production stimulated by Chit 80% NPs, Chit 93% NPs and the respective polymers prepared in endotoxin-free and sterile conditions. For the experiment, test samples were incubated with RAW 264.7 cell line for 24 h, as well as LPS, as a positive control. Mean ± SEM; obtained from four independent experiments, each performed in triplicate (n = 4), \*p < 0.05 compared to control. (B) Inhibition of ROS production by Chit 80% NPs, Chit 93% NPs and the respective Chit polymers prepared in endotoxin-free and sterile conditions. For the experiment, LPS and test samples were co-incubated with RAW 264.7 cell line for 24 h. Mean ± SEM, obtained from a minimum of seven independent experiments, each performed in triplicate (n ≥ 7). (C) NO production stimulated by Chit 80% NPs, Chit 93% NPs and the respective polymers prepared in endotoxin-free and sterile conditions. For the experiment, test samples were incubated with RAW 264.7 cell line for 24 h, as well as LPS, as a positive control. Mean ± SEM, obtained from a minimum of three independent experiments, each performed in triplicate (n ≥ 3), \*\*\*p < 0.001 compared to control. (D) Inhibition of NO production by Chit 80% NPs, Chit 93% NPs and the respective polymers prepared in endotoxin-free and sterile conditions. For the experiment, LPS and test samples were co-incubated with RAW 264.7 cell line for 24 h. Negative control (C–) was not co-incubated with LPS. Mean ± SEM, obtained from a minimum of three independent experiments, each performed in triplicate (n ≥ 3), \*\*p < 0.01 and \*\*\*p < 0.001 compared to LPS control. (E) Evaluation of possible NP and polymer interference with the wavelength used to read ROS assay (Ex485/20 – Em528/20) (Mean ± SEM, n = 3). (F) Evaluation of the ROS production (fluorescence fold increase) induced by the NPs solvent and polymers solvent (n = 3). Data are presented as mean ± SEM. (G) Evaluation of possible NP and polymer interference with the wavelength used to read NO assay (550 nm) (Mean ± SEM, n = 3). (H) Evaluation of the NO production (%) induced by the NPs solvent and polymers solvent (% of control) (n = 3). Data are presented as mean ± SEM. (I) Control of interferences of (A) Chit NPs and (B) Chit polymers with known concentrations of NO without cells (Mean ± SEM, n = 3).

the test itself. So, similar to ROS assay, the NO assay was performed in the presence of the test samples, without cells and the results were presented on **Figure 4G**. The solvent of the Chit NPs suspension or the chitosan polymer suspension were evaluated to understand if they also had an effect on NO production (**Figure 4H**). No interferences were observed in the readout, and the solvents were not able to induce NO production. An additional control was performed for NO production assay, to evaluate whether Chit and Chit NPs, due to their cationic charge, could be adsorbing NO at their surface, reducing the amount of NO quantified. Such phenomenon would provide an explanation for the NO production inhibition observed. To evaluate this hypothesis, we performed the NO calibration curve in the presence and absence of Chit NPs and polymers (**Figure 4I**). As shown, the NO curves are all overlapping, meaning no interferences from Chit NPs and Chit polymers were observed.

#### In vitro Studies With Human Peripheral Blood Mononuclear Cells

PBMCs are a good model to study immune responses, since they secrete regulatory and pro-inflammatory cytokines and chemokines in the human body. In vitro cell viability experiments give an indication of a particle cytotoxic profile that may be observed in vivo.

FIGURE 5 | Cell viability studies in PBMCs and assay interference evaluation. (A) Cell viability decrease induced by different concentrations of Chit 80% NPs, Chit 93% NPs, Chit 80%, and Chit 93% in human PBMCs, evaluated by MTT assay following 24 h of incubation. Dotted line represents the 70% of cell viability. Results are expressed as mean ± SEM, obtained from four independent experiments, each performed in triplicate (n = 4). (B) Confirmation of MTT results by testing four different concentrations of Chit 80% NPs and Chit 93% NPs by flow cytometry using PI. Results are expressed as mean ± SEM, obtained from 1 to 4 independent experiments, each performed in duplicate (n = 1–4). (C) Nonlinear regression analysis of the cell viability data, allowing the extrapolation of IC50 values (720µg/mL for Chit 80% NPs and 2104µg/mL for Chit 93% NPs). Significant statistical difference between Chit 80% NPs IC50 and Chit 93% NPs IC50 calculated using extra sum-of-squares F-test. (D) Evaluation of possible NP and polymer interference with the wavelength used to read MTT assay (540/630 nm) (Mean ± SEM, n = 3). (E) Evaluation of the cell viability resultant from the incubation of PBMCs with the NPs solvent and polymers solvent (% of control) (Mean ± SEM, n = 4).

FIGURE 6 | Cytokine secretion in PBMCs and interference evaluation. (A,B) Cytokine secretion induced by 100µg/mL of Chit 80% and Chit 93% polymers and NPs on human PBMCs, after 24 h incubation (A- IL-6 and B- TNF-α). Cytokine quantification was performed by ELISA. Results illustrate the increase in cytokine production (Chit 80% NPs, Chit 93% NPs, Chit 80%, Chit 93%, ConA, LPS), when compared to the basal level (–). The experiment was repeated with blood from eight different donors (n = 8). (C,D) Evaluation of the Chit 80% and Chit 93% NPs ability to interfere with cytokine quantification when compared to the cell culture media (experiment without cells). The results of the cytokine quantification for the calibration curve in the presence of Chit 80% NPs and Chit 93% NPs were compared to the cytokine quantification of the calibration curve in simple cell culture media (C- IL-6 and D- TNF-α). Dotted lines represent the original calibration curve in ELISA diluent. Data are represented as mean ± SEM (n = 3).

#### Lower Deacetylation Degree Chitosan NPs Are More Cytotoxic for PBMCs

Similar to RAW 264.7 cell line experiments, Chit NPs and polymers were incubated with cells, in this case human PBMCs, and the cell viability was evaluated using the MTT assay. The results depicted in **Figure 5A** showed that Chit NPs were more cytotoxic than the respective polymers.

Comparing the results achieved between the two NPs, Chit 80% NPs showed a tendency to be more cytotoxic than the Chit 93% NPs. This difference was further confirmed with the PI assay, where the cell membrane integrity rather than the metabolic activity was evaluated (**Figure 5B**). A nonlinear regression of the MTT assay results clearly showed that Chit 80% NPs induced a more accentuated decrease in cell viability, with the 50% inhibitory concentration (IC50) calculated at ∼720µg/mL (**Figure 5C**). Chit 93% NPs showed a statistically different IC50, calculated to be 2,104 µg/mL.

To note, Chit NP and polymer highest concentrations tested during cell viability assessment in both RAW 264.7 and PBMCs were very high and do not correlate with concentrations required for in vivo assays. Nevertheless, 5,000µg/mL is recommended in OECD guidelines for genotoxicity testing of chemicals (test guideline 487) as the maximum concentration to be tested when no cytotoxicity or precipitates are observed. In our case, these concentrations were needed to correctly calculate the IC 50. For the Chit polymers, even though the highest sample concentration was very thick, it did not induce toxicity below 70%, confirming the great biocompatibility of the Chit polymers.

As explained for the RAW 264.7 cell line, experimental controls were performed and the results are presented in **Figures 5D,E**. The absorbance readout showed no interference for formulations (equal Abs values) and the resultant cell viability following solvent incubation with PBMCs during 24 h showed comparable cell viability to the control.

#### LPS-Free Chitosan Nanoparticles Do Not Stimulate IL-6 and TNF-α Release by PBMC's

Cytokines participate in many physiological processes, mostly in the regulation of immune and inflammatory responses (Ai et al., 2013). Interleukin-6 (IL-6) is a pleiotropic cytokine (inflammatory and anti-inflammatory properties) able to modulate the activity of immune cells (Wang et al., 2017). Tumor necrosis factor-α (TNF-α) is a pro-inflammatory cytokine released from macrophages or activated T cells which plays a crucial role in many immune and inflammatory processes, such as proliferation, apoptosis, and cell survival (Cai et al., 2017).

In order to understand if Chit NPs and polymers were able to stimulate the release of these cytokines by human PBMCs, the cells were incubated with 100µg/mL Chit test samples for 24 h and the secreted cytokine results, measured by ELISA, were depicted in **Figures 6A,B**. Results showed that neither Chit NPs nor Chit polymers stimulated the production of IL-6 and TNFα, as no differences were found before and after incubation with test samples. Importantly, the use of positive controls such as LPS and Con A, give us an indication of the cell function regarding the cytokine we are analyzing. Notably, both positive controls significantly increased the cytokine secretion in PBMCs.

Additionally, since chitosan's positive charge favors cytokine adsorption, the possible interference of Chit NPs in cytokine quantification was tested. For that, Chit NPs suspended in cell culture medium were incubated with known concentrations of each cytokine (calibration curve) for 24 h, then centrifuged and supernatant cytokine content similarly quantified by ELISA. The percentage of cytokine quantification in cell culture medium incubated with the nanoparticles in comparison to cell culture medium without nanoparticles, can reveal if the cytokines adsorbed to the NPs, preventing their quantification. Interestingly, **Figures 6C,D** suggest that Chit NPs did not adsorb IL-6 nor TNF-α, since cytokine quantification was equal or above 100%. Thus, we can assume that the absence of TNF-α and for IL-6 production upon stimulation with Chit NPs and polymers was indeed due to the lack of the samples' ability to stimulate the cells, which strengthens the conclusion that they do not induce a proinflammatory cytokine response, at least when produced under endotoxin-free conditions.

# Hemocompatibility Assays

#### Chitosan Nanoparticles and Polymers Do Not Induce Hemolysis Even at High Concentrations

Hemolysis is characterized by the rupture of red blood cells (RBCs) and the release of their contents, ultimately leading to anemia, jaundice and renal failure (Dobrovolskaia et al., 2008). All materials entering the blood get in contact with RBCs and so the evaluation of the hemolytic ability of the biomaterials is of utmost importance.

Chit NPs and polymers hemolytic activity was evaluated following a 3 h incubation at 37◦C with RBCs. Results showed that none of them induced a percentage of hemolysis superior to 5%, even in Chit concentrations of 2 mg/mL (**Figure 7A**). Triton X-100 was used as the hemolytic agent whose effect is possible to observe by the red color of the supernatant after centrifugation of the experiment tube 1 and 2 (**Figure 7B**). According to the ASTM E2524-08 standard, only hemolysis superior to 5% are considered significant. Although no hemolytic activity was induced by Chit NPs and polymers, solvents were tested as well as the NP interference with the assay readout. As depicted in **Figure 7C**, the NPs had no interference in the absorbance measurements and **Figure 7D** illustrates that no hemolysis was induced by the NPs or solvents of the suspensions of the NPs or polymers.

#### The Effect of Chitosan Nanoparticles in Coagulation and Platelet Aggregation Depends on the Nanoparticle Characteristics

The plasma coagulation cascade is responsible for blood clotting and consists of a series of protein interactions (Laloy et al., 2014). To evaluate the effect of Chit NPs and polymer samples on plasma coagulation time, two concentrations (0.1 and 1 mg/mL) of test samples were incubated with blood during 30 min. In this assay, both blood coagulation pathways, the activated partial thromboplastin time (APTT) and the prothrombin time (PT) were separately tested (**Figure 8A**).

The results showed that Chit NPs and polymers at 0.1 mg/mL concentration had no effect on plasma coagulation for both pathways. However, 1 mg/mL Chit 80% NPs prolonged APTT (intrinsic pathway), while no effect was observed with Chit 93% NPs and polymers 80% and 93% at the same concentration. NPs suspension solvent and polymer suspension solvent (acetate buffer) was also tested to discard any method interference and no effect was observed in plasma coagulation (**Figure 8B**).

Platelets play an important role not only in hemostasis but also in immune and inflammatory responses (Golebiewska and Poole, 2015). Homeostatic imbalance as a result of platelet

FIGURE 7 | Hemolysis assay. (A) Hemolytic activity of Chit polymers and NPs in human blood after 3 h incubation at 37◦C. PBS and Triton-X-100 were respectively used as negative (C–) and positive control (C+). Results are expressed as mean ± SEM, obtained from at least three independent experiments, using blood from different donors, each performed in duplicate (n ≥ 3). (B) Representation of 100% hemolysis generated by the positive control (tube 1 and 2) and the absence of hemolysis induced by the negative control (tube 3 and 4). (C) Evaluation of the Chit NP and Chit polymers interferences with the absorbance readout, without blood. Results are expressed as mean ± SEM, obtained from at least three independent experiments, using blood from different donors, each performed in duplicate (n ≥ 3). (D) Evaluation of the hemolysis resultant from the incubation of NPs solvent and polymers solvent in human blood after 3 h incubation at 37◦C.

each performed in duplicate (n ≥ 3).

solvents with the coagulation times assay. Results are expressed as mean ± SEM, obtained from three independent experiments, using blood from different donors,

function alterations affect primary hemostasis and can result in thrombotic or hemorrhagic disorders (Golebiewska and Poole, 2015). Therefore, it is important to study Chit NPs interactions with platelet function.

To assess platelet aggregation, a cytometer is frequently used to count the platelets, however, by this method the interference of NPs, due to their size have to be taken into account. To evaluate the interference of Chit NPs with the platelet count, Chit NPs were incubated with platelet-free plasma (PFP) and visualized under the light microscope. Results showed that Chit NPs, most likely in the form of aggregates, were possibly counted as platelets, which invalidated the use of such method. To overcome this setback and assess platelet aggregation, the experiment was performed by counting platelets manually under a microscope, using a Neubauer chamber. Results from microscopy observation were summarized in **Figures 9A,B**.

The **Figure 9A**-1 clearly shows the absence of platelets typical from PFP, while plenty of platelets were observed in PRP, with no signs of aggregates (**Figure 9A**-2). When platelets were incubated with calcium chloride, we observed the formation of fibrins, a sign of platelet aggregation (**Figure 9A**-3). Similarly, collagen also induced platelet aggregation, but in this case no fibrins were observed (**Figure 9A**-4). When analyzing both types of Chit NPs incubated with PFP we can see their tendency to form NPs aggregates, which were hypothetically the cause of the observed interference in the cytometry technique (**Figures 9A**-5,7). Nevertheless, under microscopic observation, these aggregates were not misinterpreted as platelets. The Chit 80% NPs (**Figure 9A**-6) when incubated with PRP did not seem to induce platelet aggregation, as there was no evidence of platelet aggregates as found in the positive controls. On the other side, when Chit 93% NPs were incubated with PRP (**Figure 9A**-8) we observed that large NP agglomerates appear to have retained some platelets. Besides that, platelet aggregation was observed.

Using platelet count to calculate the percentage of platelet aggregation, positive controls induced an effect superior to 40%. The Chit 80% NPs did not induce platelet aggregation as only 4.9% of platelet aggregation was calculated for these samples. However, Chit 93% NPs resulted in 37.5% platelet aggregation similar to what was achieved with calcium chloride and collagen positive controls.

#### DISCUSSION

A hot topic in the nanomedicine field are polymeric NPs, which are engineered to either interact or not with the immune system. In the early stages of the development of a nanotechnology-based medicine, when the drug is to be encapsulated into NPs, the first question to be considered is, whether it is supposed that the new nanomedicine, in addition to its main pharmacological action, also acts on the immune system. This kind of approach is part of the SbD. Particularly, in the case of chitosan, as it is a set of polymers with different MW and DDA [quality attributes (QA)], it is important to understand if there are differences between them, regarding possible interactions with the immune system. For Chit NPs, in addition to polymer QA, NP characteristics, like size and zeta potential or shape can also be important. Therefore, physicochemical characteristics (PCC) of the polymers and NP might influence their immunological properties, and therefore a thorough characterization of both is very important to supplement the immunotoxicity studies and to draw meaningful conclusions (Crist et al., 2013). The lack of an exhaustive characterization may preclude the correct interpretation of results and may lead to misinterpretations hindering the establishment of trends regarding how Chit NP PCC influence the immune response. Additionally, one of the most important challenges encountered in in vitro immunotoxicity tests for NPs is related to their unique physicochemical properties. These can interfere with the established tests, originally developed for testing conventional chemicals (Dobrovolskaia and McNeil, 2016). Such interference depends both, on the NPs tested and the in vitro assay and can lead to false-positive or falsenegative results (Dobrovolskaia and McNeil, 2016). Lastly, in order to achieve a correct result interpretation, it is important to identify the presence of biological contaminants in the NP preparation (Dobrovolskaia and McNeil, 2007). The main biological contamination in in vitro assays, even when working under sterile conditions, are endotoxins, which may lead to inflammatory responses (Dobrovolskaia and McNeil, 2007).

The present case study intends to provide a systematic analysis of the effects of Chit NPs and respective Chit polymers on different biological outcomes commonly tested under the immunotoxicity scope, considering, as most important the effect of DDA and MW, without neglecting possible interferences and contaminants.

In detail, as literature suggests, we found that Chit NPs appear to be more cytotoxic than the respective Chit polymers from which they were derived. In fact, for polymer concentrations up to the extraordinary concentration of 5,000µg/mL, no cytotoxic effects were found neither in PBMCs, nor in RAW 264.7 cells. On the other hand, when the polymers were assembled into NPs, the same range of Chit concentrations induced a concentration dependent reduction in cell viability. Another important result we found was that PBMCs isolated from human blood were more sensitive to the NPs than RAW 264.7 cells, which is evident from the lower IC50 values extrapolated. Furthermore, this higher sensitivity of PBMCs exposed differences between the NPs produced with Chit 80% and Chit 93%. In fact, Chit 80% NPs induced a more accentuated decrease in cell viability. To discuss these results some aspects must be analyzed. To begin with, the cell culture media were different for PBMCs and RAW 264.7 cells (RPMI and DMEM, respectively). The physicochemical characterization of the NPs in water (stock suspension) is important, but their characterization when dispersed in the medium used for in vitro assays can provide further evidence. In fact, Chit 80% NPs presented a smaller size than Chit 93% NPs in water (127 nm vs. 292 nm), but these differences were not observed in cell culture media. Moreover, the NPs size analysis in cell culture media resulted in very high PDI. We realized that in RPMI (used for PBMCs) Chit 80% NPs presented an important size population around 500–1,000 nm, while Chit 93% showed a significant size population around 1,000–2,000 nm. On the other hand, Chit 80% NPs and Chit 93% NPs in DMEM (used for RAW 264.7) did not show such size distribution profile, with the most expressive populations around 300–700 nm and 400–800 nm, respectively. Therefore, the most noteworthy size

PBS, collagen (200 and 600µg/mL) and calcium chloride (CaCl2, 0.25 M) were used as negative and positive control, respectively. (A) Representative images of platelet aggregation assay stained with Giemsa dye. Untreated platelet free plasma (PFP) is represented in image 1 and untreated platelet rich plasma (PRP) is represented in image 2. For the experiment two different positive controls were used (CaCl2–3 and collagen−4). Chit 80% NPs and Chit 93% NPs were tested both with PFP (image 5 and 7) and PRP (image 6 and 8). (B) Quantification of the platelet aggregation effect. Platelet count is presented as the final average of a minimum of three donors ± SD. The percentage of aggregation was calculated using as reference the platelet count of the negative control, and is presented as the average of all assays ± SD (n ≥ 3).

differences occurred in RPMI, which could explain the different cell viability profile between the NPs in PBMCs.

Literature review showed several contradictory results regarding Chit NPs effect on cellular ROS production. One study suggested that Chit NPs had an inhibitory activity (Bor et al., 2016), two studies reported no Chit NPs effect (Omar Zaki et al., 2015; Arora et al., 2016) and three reported a stimulating effect (Hu et al., 2011; Sarangapani et al., 2018; Wang et al., 2018) on basal ROS cellular production. Concerning the polymer, same conflicting results were also found (Arora et al., 2016; Salehi et al., 2017; Sarangapani et al., 2018). From our case study, we concluded that, despite no significant differences were found in the cytotoxic profile of both NPs in RAW 264.7 cells, in ROS assay these NPs had different effects when tested at non-cytotoxic concentrations. Only Chit 80% NPs induced ROS production in a concentration-dependent manner (starting at 156µg/mL). Nevertheless, the 80% DDA polymer suspended in acetate buffer also induced ROS production. Thus, the effect was dependent on the type of Chit polymer: Chit with the lowest DDA induced ROS production. On the other hand, neither NPs nor polymers, irrespective of the DD were able to inhibit ROS production. While our results suggested an influence of the DDA of the polymer in cellular ROS stimulation, the above mentioned studies did not. In fact, all authors mentioned used similar DDA Chit (75–85%) and no pattern could be observed. Moreover, those results are also affected by other variables, such as the different cellular models and testing conditions, namely concentrations, used by each author, as previously reviewed elsewhere (Jesus et al., 2019). Furthermore, none of the studies mentioned used RAW 264.7 cells, which hinders the comparison with the results herein presented.

Concerning the ability to induce NO by cells, only one result was found in the literature that claim the ability of the chitosan NPs to induce cells to produce this inflammatory marker and it showed a concentration-dependent increase above 68.18µg/mL, in PBMCs following 24 h incubation (Pattani et al., 2009). Our case study, however, did not allow us to confirm this trend. Our results showed that none of the Chit NP tested increased NO production, in the range 39µg/mL to 156µg/mL. To note, Pattani et al. used Chit NPs cross-linked with sodium carboxymethyl cellulose that possessed a much smaller average size (37 nm), which may have been one of the causes for the increased reactivity. For Chit polymer, two studies observed no effect in basal NO production (Jeong et al., 2000; Wu and Tsai, 2007) supporting our results (Chit polymers did not induce NO production), while two others reported an increase (Peluso et al., 1994; Wu et al., 2015). In the case of Peluso et al. (1994) we can hypothesize that the conflicting results can be due to the use of a different cellular model (rat peritoneal exudate macrophages) or a possible endotoxin contamination, which was not assessed. On the other hand, Wu et al. (2015) used RAW 264.7 cells and claimed the endotoxin level in the stock solution was <0.5 EU/mL, which is a much higher value than we have for the Chit polymers tested. In opposition, the ability of Chit NPs and polymers to inhibit LPS-induced NO production was verified for all testing samples. This effect was similar among them, suggesting no effect of the DDA or particle size. In this case, although we have excluded that Chit NPs or polymers were interacting with NO, hampering its quantification, we cannot rule out the ability of Chit to bind LPS, partially inhibiting its effect. These findings of NO inhibition are in agreement with most of the results found in the literature, where Chit NPs were reported to inhibit H2O2-stimulated NO production (Wen et al., 2013) and Chit was reported to inhibit LPS-induced NO production (Hwang et al., 2000; Wu and Tsai, 2007). In contrast, one study performed by Jeong et al. showed Chit had a synergistic effect with IFN-γ to induce NO production (Jeong et al., 2000). In this case, the polymer used had a higher MW (300 KDa) than the polymer used in this study.

Regarding the ability of the Chit polymer and NPs to stimulate cell to produce pro-inflammatory cytokines, for instance the induction of TNF-α has been reported in some studies. These studies, however, must be carefully discussed regarding endotoxin contamination. We realized that when the authors do not disclose the purity of the polymer used, namely whether it is an LPS-free chitosan or not, results are not consensual. In some of these studies, IL-6 and TNF-α were reported to be induced following Chit and Chit NPs stimulation (Feng et al., 2004; Koppolu and Zaharoff, 2013; Baram et al., 2014), while in others studies they were not (Villiers et al., 2009; Han et al., 2016). On the other hand, when authors used Chit-based samples prepared under endotoxin-free conditions (Pattani et al., 2009; Lieder et al., 2013; Stopinšek et al., 2016), they were unanimous proving that "pure/clean" non-cytotoxic Chit and particularly, Chit NPs, do not induce IL-6 or TNF-α secretion. In agreement with this, our endotoxin-free formulations confirmed that Chit NPs and polymers do not induce TNF-α or IL-6 secretion in PBMCs. Consistently, previous studies from our group using different endotoxin-free Chit-based particles (different DDA and MW polymer, cross-link compound and NPs size when compared with present NPs) also showed no ability to induce these cytokines in mice spleen cells (Soares et al., 2019) and mice bone marrow derived dendritic cells (BMDCs) (Lebre et al., 2019). However, the last study (Lebre et al., 2019) proved that Chit and Chit NPs were able to stimulate BMDCs, activating the NLRP3 inflammasome. As a consequence, it was observed an increase of the IL-1β (pro-inflammatory cytokine) secretion by cells.

Regarding hemocompatibility assessment, our studies also allowed to clarify some conflicting literature results. Considering the hemolytic activity, some original articles were found supporting the non-hemolytic activity of Chit NPs (Nadesh et al., 2013; Kumar et al., 2017) and also the Chit polymer (Nadesh et al., 2013). Nevertheless, two studies reported a slight hemolytic effect for Chit NPs (Shelma and Sharma, 2011; de Lima et al., 2015). The last study, however, suggested that the hemolytic activity was due to the NPs solvent, which was diluted acetic acid and neutralized diluted acetic acid (de Lima et al., 2015). Our studies enabled us to confirm that both Chit polymers and Chit NPs do not have hemolytic activity even at high concentrations (2 mg/mL) and that the washing procedure of the NPs eliminated the acetic acid traces of the NPs solvent, which could otherwise induce an erroneous hemolysis. Concerning coagulation studies, we found that only Chit 80% NPs caused a concentration dependent effect on coagulation. At similar concentrations, the Chit 93% NPs and both Chit polymers in acetate buffer had no effect, meaning the effect was dependent on the nanoscale dimension of the NPs and on the polymer characteristics (80% DDA and 49 kDa). We can hypothesize that Chit 80% NPs prolonged activated partial thromboplastin time due to the affinity of NPs for plasma clotting factors that are involved in the intrinsic pathway (XII, XI, IX, VIII), possibly adsorbing them (Palta et al., 2014). In previous studies, Shelma and Sharma (2011) showed that Chit NPs reduced the total normal coagulation time, while Nadesh et al. verified that Chit NPs did not alter coagulation time, when resuspended in saline (Nadesh et al., 2013). However, experimental conditions were significantly different. The first used 2 mg/mL which is a concentration similar to ours, but evaluated only the blood clotting time, and the second only used 0.05 mg/mL. Lastly, only Chit 93% NPs were able to induce platelet aggregation. We can hypothesize this effect was only observed with Chit 93% NPs due to the higher amount of NH<sup>+</sup> 3 groups resulting from deacetylation, increasing the interaction with negatively charged groups of platelets. However, through microscope slide analysis we postulated that the effect may also be related to the formation of large NP aggregates when using a concentration of 2 mg/mL, that further leads to platelet aggregation at their surface. Accordingly, Shelma and Sharma (2011) also reported that platelet aggregation was induced by Chit NPs at a concentration of 2 mg/mL. However, since in the same Chit 80% NPs concentration we could not confirm this tendency, the influence of different physicochemical properties of NPs affecting the biological activity must be highlighted.

In addition to the specific immunotoxicity and hemocompatibility results presented here, this case study aims to raise awareness of the scientific community about the importance of adequate controls (experimental and sample controls). Indeed, some studies fail to report important experimental controls to validate whether a particular assay is appropriate for each NP formulation and to avoid false-positive and false-negative results. A simple control is the evaluation of NP interference in the assay readout (absorbance, luminescence or fluorescence) in the absence of the biological matrix. This is omitted most of the times even though it highly increases the reliability of the obtained results. For instance, in the platelet aggregation study, the cytometer counted NPs instead of platelets, which was the reason why we did not use this technique and we had to use a light microscope. Another desirable experimental control is the cellular viability at the end of each assay, to guarantee that the revealed effects are not only a side effect of cytotoxic concentrations. Regarding sample controls, a parameter that is generally ignored is the solvent of the NP suspension. Usually, synthesized nanoparticles are in a solvent which is not designed to be biocompatible, but to stabilize the particles and prevent their aggregation in stock suspensions. The presence of such solvent in the culture medium may be enough to induce cell death, alter osmolality, pH, cause cellular damage, and decrease metabolic activity (Oostingh et al., 2011). Therefore, solvent control test is also useful to correctly interpret the results. Ultimately, we highlight the need to avoid endotoxin contamination of polymeric NPs, as it is frequently not considered and may be the source of false bioactivity or toxicity assumptions. In fact, endotoxins are a type of bacterial cell wall toxins, responsible for inducing a state of inflammation in organisms, resulting in fever, fibril reactions and organ damage (Dobrovolskaia et al., 2009). NPs are typically contaminated with endotoxins, mostly Chit NPs as their marked positive surface charge is especially susceptible to this kind of contamination. We believe that the increasing awareness of researchers about endotoxin contamination will contribute to reduce the disparity among NP immunotoxicity results.

Once more, we confirmed that our Chit NPs are more cytotoxic than Chit polymers, which justifies why we cannot rely on the Chit polymer attested safety to extrapolate to Chit NPs. More importantly, as we proposed, the presented results enabled us to shed light on some conflicting results found in literature. Notably, neither Chit NPs tested here demonstrated intrinsic pro-inflammatory ability. However, other assays showed that Chit DDA and MW influence Chit NPs immunotoxicity and hemocompatibility. Chit NPs with the lower DDA and lower MW (Chit 80% NPs) were more toxic in terms of reducing cell viability, ROS production and coagulation times. Nevertheless, all reported effects are concentration dependent and do not refrain Chit 80% NPs from being promising drug delivery systems or vaccine adjuvants.

To conclude, the present case-study together with further studies may contribute to the development of a knowledge-based guideline that enables NP product design based on the SbD approach. Nevertheless, we cannot overlook the current need to establish a set of methods for immunotoxicological assessments of NPs that need validation and standardization to allow the generation of a reliable database of results, essential to apply SbD more efficiently.

#### DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

SJ, AM, and OB contributed conception and design of the study. AM, AD, ES, JC, and MC performed the experimental work. SJ wrote the first draft of the manuscript. SJ, AM, ES, MS, CS, GB, PW, and OB contributed to data analysis and manuscript revision. All authors read and approved the submitted version.

#### FUNDING

This work was financed by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational Programme and through the COMPETE 2020 - Operational Programme for Competitiveness and Internationalization and Portuguese national funds via FCT – Fundação para a Ciência e Tecnologia, under project PROSAFE/0001/2016 and POCI-01-0145-FEDER-030331, and the strategic project UIDB/04539/2020. This work is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016 and from the CTI (1.1.2018 Innosuisse), under grant agreement Number 19267.1 PFNM-NM.

#### ACKNOWLEDGMENTS

The authors would like to thank Dra Ana Donato from the Faculty of Pharmacy Clinical Laboratory Analysis (University of Coimbra) for the hematological studies support, Dra Mónica Zuzarte from iLAB - Bioimaging Laboratory of the Faculty of Medicine of the University of Coimbra for TEM analysis, and Primex for supplying Chitoclear <sup>R</sup> chitosan polymers. NMR data was collected at the UC-NMR facility which is supported in part by FEDER – European Regional Development Fund through the COMPETE Programme (Operational Programme for Competitiveness) and by National Funds through FCT – Fundação para a Ciência e a Tecnologia (Portuguese Foundation for Science and Technology) through grants RECI/QEQ-QFI/0168/2012, CENTRO-07-CT62-FEDER-002012, and also through support to Rede Nacional de Ressonância Magnética Nuclear (RNRMN) and to Coimbra Chemistry Centre through grant UID/QUI/00313/2019.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2020.00100/full#supplementary-material


scripts/fdcc/?set=GRASNotices&sort=GRN\_No&order=DESC&startrow=1& type=basic&search=chitosan (accessed October 24, 2019).


**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 Jesus, Marques, Duarte, Soares, Costa, Colaço, Schmutz, Som, Borchard, Wick and Borges. 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.

# Nanoscale Self-Assembly for Therapeutic Delivery

Santosh Yadav, Ashwani Kumar Sharma and Pradeep Kumar\*

Nucleic Acids Research Laboratory, CSIR Institute of Genomics and Integrative Biology, Delhi, India

Self-assembly is the process of association of individual units of a material into highly arranged/ordered structures/patterns. It imparts unique properties to both inorganic and organic structures, so generated, via non-covalent interactions. Currently, selfassembled nanomaterials are finding a wide variety of applications in the area of nanotechnology, imaging techniques, biosensors, biomedical sciences, etc., due to its simplicity, spontaneity, scalability, versatility, and inexpensiveness. Self-assembly of amphiphiles into nanostructures (micelles, vesicles, and hydrogels) happens due to various physical interactions. Recent advancements in the area of drug delivery have opened up newer avenues to develop novel drug delivery systems (DDSs) and selfassembled nanostructures have shown their tremendous potential to be used as facile and efficient materials for this purpose. The main objective of the projected review is to provide readers a concise and straightforward knowledge of basic concepts of supramolecular self-assembly process and how these highly functionalized and efficient nanomaterials can be useful in biomedical applications. Approaches for the self-assembly have been discussed for the fabrication of nanostructures. Advantages and limitations of these systems along with the parameters that are to be taken into consideration while designing a therapeutic delivery vehicle have also been outlined. In this review, various macro- and small-molecule-based systems have been elaborated. Besides, a section on DNA nanostructures as intelligent materials for future applications is also included.

Keywords: self-assembly, nanostructures, amphiphilicity, polymers, small molecules, drug delivery

#### Edited by:

Gianni Ciofani, Italian Institute of Technology, Italy

#### Reviewed by:

Veikko Linko, Aalto University, Finland Stefano Leporatti, Italian National Research Council, Italy

> \*Correspondence: 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: 27 September 2019 Accepted: 10 February 2020 Published: 25 February 2020

#### Citation:

Yadav S, Sharma AK and Kumar P (2020) Nanoscale Self-Assembly for Therapeutic Delivery. Front. Bioeng. Biotechnol. 8:127. doi: 10.3389/fbioe.2020.00127

**Abbreviations:** Boc, t-butyloxy carbonyl; CDs, cyclodextrins; CMC, critical micellar concentration; DDSs, drug delivery systems; DEPC, dierucoylphosphatidylcholine; DMPC, dimyristoyl phosphatidylcholine; DMPG, dimyristoyl phosphatidylglycerol; DNA, deoxyribonucleic acid; DOPC, dioleoylphosphatidylcholine; DOPE, dioleoly-sn-glycerophophoethanolamine; DOPS, dioleoylphosphatidylserine; DPPC, (1,2-dipalmitoyl-sn-glycero-3-phosphocholine); DPPG, dipalmitoylphosphatidylglycerol; DSPC, distearoylphosphatidylcholine; DSPE, distearoyl-sn-glycero-phosphoethanolamine; DSPE-MPEG-2000, (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-methoxypolyethyleneglycol-2000); DSPG, distearoylphosphatidylglycerol; EPC, egg phosphatidylcholine; Fmoc, 9-fluorenylmethyloxy carbonyl; HA, hyaluronic acid; HSPC, hydrogenated soy phosphatidylcholine; MPEG, methoxy polyethylene glycol; MSPC, (1-stearoyl-2-hydroxysn-glycero-3-phosphocholine); PEG, polyethylene glycol; PGA, poly(glycolic acid); PISA, polymerization-induced self-assembly; POPC, palmitoyloleoylphosphatidylcholine; PLA, poly(D,L-lactic acid); PLGA, poly(D,L-lactic-co-glycolic acid); PTX, paclitaxel; RNA, ribonucleic acid; SM, sphingomyelin; Tm, transition temperature.

# INTRODUCTION

fbioe-08-00127 February 22, 2020 Time: 15:45 # 2

Supramolecular self-assembly has recently attracted the attention of the researchers worldwide to generate nanostructures and nanomaterials bearing unique physical and chemical properties. The organization of molecules in these nanoassemblies has made it possible to design and develop new devices that can interact with the living cells and generate the response. These are not only being focused as important components in the emergence of cellular life, but also as materials that can be used in huge applications ranging from diagnostics and sensing to biomaterials, bioelectronics, energy generation, catalysis, drug delivery, and nanocomposites (Busseron et al., 2013; Du et al., 2015; Habibi et al., 2016; Xing and Zhao, 2016). Mainly, two strategies, viz., top-down and bottom-up, are being followed for the fabrication of nanostructures (**Figure 1**). The earlier one involves the carving out of the final nanostructure with a defined shape and size from a larger block of matter. As a result, the strategy does not require atomic level control. Alternatively, the later approach involves building up of the desired nanostructures from the basic components by the processes of molecular recognition and self-assembly, which is basically derived from the interactions of basic units to form well-organized structures. Therefore, the atomic or molecular level control is possible in the later approach over the formation of nanostructures by manipulating the structures of self-assembling molecular units.

# SELF-ASSEMBLY

Self-assembly is the spontaneous molecular arrangement of the disordered entities of molecules into ordered structures resulting from specific local interactions among the components themselves (Mendes et al., 2013; Mattia and Otto, 2015; Stoffelen and Huskens, 2016). Formation of the most of the biological nanostructures is the outcome of the self-assembly such as construction of cell membranes by assembly of phospholipid bilayers, helical structure of DNA, folding of polypeptide chains, etc. The interaction of a ligand with its receptor is also attributed to self-assembly (Haburcak et al., 2016; Azevedo and da Silva, 2018). It also accounts for the development of molecular crystals, self-assembled monolayers, phase separated polymers, and colloids (Busseron et al., 2013; Mendes et al., 2013; Du et al., 2015; Mattia and Otto, 2015; Habibi et al., 2016; Haburcak et al., 2016; Stoffelen and Huskens, 2016; Sun et al., 2017; Azevedo and da Silva, 2018). In fact, molecular self-assembly is a natural process which is very essential in the emergence and maintenance of life. Synthetic molecules like amino acids, oligo- and polypeptides, polymers, dendrimers, and π-conjugated compounds have been considered as the primary focus used for building up nanostructures, such as nanotubes, nanofibers, micelles, and vesicles (Buerkle and Rowan, 2012; Correa et al., 2012). Moreover, self-assembly of small molecules as building units is a useful strategy for the formation of structurecontrolled materials (Ariga et al., 2019). Likewise, DNA-based nanomaterials have shown their potential in diagnostics and therapeutic delivery.

The process of self-assembly plays a key role in the design, synthesis, and development of newer nanomaterials (Whitesides and Boncheva, 2002).


Thus, self-assembly has exhibited a profound impact in a wide range of fields, viz., physical, chemical, and biological sciences, materials and biomedical sciences, and manufacturing. Besides, the concept has provided opportunities to develop new materials and components of life through the exchange of ideas and methodologies among these fields.

# CLASSIFICATION OF SELF-ASSEMBLY

The term self-assembly was initially used by the researchers in different fields and subsequently, it was adopted by the chemists to describe the ordered arrangement of the molecules. Now, it has applied to materials of any size (from small molecules to galaxies) in the world around us (Xing and Zhao, 2016). Recently, the strategy has been shifted to synthesis of molecules which can be manipulated at the molecular level. This has become possible due to integration of chemistry, biology, and material science. Based on the size and nature of building blocks, self-assembly can be classified into three main categories, i.e. atomic, molecular, and colloidal self-assemblies (**Figure 2**). A variety of building blocks have been embraced in the term "self-assembly." The process of self-assembly not only covers bulk materials, but also it can apply to two-dimensional systems, i.e. surfaces and interfaces. Thus, on the basis of the systems and where it occurs, it can be classified as biological or interfacial. Further, based on its processing, it can be categorized as thermodynamic or kinetic self-assembly. Atomic, molecular, biological, and interfacial selfassemblies are covered under thermodynamic processes, while colloidal and some interfacial self-assemblies come under kinetic ones. Among these types of assemblies, atomic and biological self-assemblies are directional while others are random or nondirectional such as colloidal, molecular, and interfacial selfassemblies. Self-assembly involving large building units can also be responsive to one or the other external stimuli, viz., gravity, magnetic field, flow, electric field, surrounding media, etc.

Thus, as a result of self-assembly, spontaneous association can lead to generation of ordered structures in a range from angstrom to centimeter of different sizes and shapes. Historically,

the concepts of self-assembly have come from the investigation of molecular/biological processes.

# TYPES OF INTERACTIONS IN SELF-ASSEMBLY

Basically, the types of interactions that involved in the selfassembly processes occurring at colloidal, molecular, or atomic length scale are usually fragile and long range in contrast to chemical forces (Mendes et al., 2013). These are mainly non-covalently linked via van der Waals forces, hydrophobic, electrostatic, hydrogen bonding, π-π aromatic stacking, metal coordination, etc., which are normally weak (2–250 kJ/mol) individually in comparison to covalent linkages (100–400 kJ/mol) but together, if present in adequate numbers, they form very stable self-assembled structures and the shape, size, and functionality of the final assembly are administered by their fine balance (Mendes et al., 2013). Self-assembly between molecular units occurs when they interact with one another through a balance of usually weak and non-covalent intermolecular forces (Lee, 2008; Genix and Oberdisse, 2018). These interactions play a significant role in the alignment of molecular units in an ordered structure. These interactions are the main force that facilitates self-assembly of the units. Besides, the directionality and functionality of self-organized structures are determined by other functional interactions or forces (**Figure 3**). All these non-covalent interactions stabilize the self-assembled structures under different environmental conditions. Moreover, exhibition of completely new type of behavior as well as unique physical and chemical properties by self-assembled nanostructures have made them of special interest to researchers and scientists worldwide (Xing and Zhao, 2016). The distinctive intermolecular forces important in molecular self-assembly are given below (Mendes et al., 2013).

#### van der Waals Interactions

van der Waals interactions consist of attractive or repulsive forces between molecules which operate at moderate distances. These forces arise from dipole or induced dipole interactions at the atomic and molecular level (Lee, 2008). These are strong in vacuum or if there is no medium between two molecules. If a medium (such as water) comes between the two molecules, these forces are reduced because of dielectric screening from the medium. Obviously, this screening effect is particularly strong for water due to its high dielectric constant. The energy of the van der Waals interactions is around 1 kJmol−<sup>1</sup> whereas a covalent bond has a binding energy of around 150 kJmol−<sup>1</sup> or more (hydrogen bonds, for comparison, have typical energies of around 50 kJmol−<sup>1</sup> ). Overall, at atomic and molecular levels,

van der Waals interactions are predominantly attractive, while, under certain conditions, these can also be repulsive (particularly, at short range).

### Electrostatic Interactions and Electric Double Layer

Electrostatic interactions occur between two charged atoms, ions, or molecules, which can be either - attractive or repulsive forces, depending upon the sign of charges. These interactions are quite strong and act even at long range (upto ∼ 50 nm), however, decrease gradually with distance. Ionic self-assembly is straightforward and considered to be a reliable method for the organization of polyelectrolytes, charged surfactants, peptides, and lipids (Lee, 2008; Mendes et al., 2013). These forces originate from electrostatic interactions and impart a strong effect on many self-assembly processes. Further, these forces act as balancing interactions along with hydrophobic interactions, which result in the finite size and shape of selfaggregated structures. Sometimes these interactions get added onto during self-assembly process. Self-assembly processes at the atomic scale involve the electrostatic interactions in air as well as in vacuum, while in solution, molecular and colloidal/mesoscale self-assembly processes occur.

The interfacial double layers are generally quite evident in systems having large surface area to volume ratio, such as porous or colloidal bodies with pores or particles, respectively, in the range of micrometers to nanometers. Layer by layer or double

layer self-assembly plays an important part in several routinely employed materials, e.g. existence of homogenized milk, which is owing to coverage of fat droplets with a double layer that inhibits their agglomeration into butter.

#### Hydrophobic Interactions

Hydrophobic interactions play a big role in understanding the process of self-assembly. These interactions occur in water due to poor dispersibility of the hydrophobic moieties. Interaction of a hydrophobic moiety with water can be elucidated using thermodynamic effects which result in the change in free energy, entropy, enthalpy, and heat capacity. These changes can be studied by the thermodynamic principle, 1G = 1H - T1S. When a hydrophobic substance interacts with water, the structure of water around that substance varies with the size and shape of the substance. This networking around the hydrophobic substance is called iceberg cluster or iceberg formation (Lee, 2008). The iceberg formation itself is not an entropic or enthalpic effect rather it depends upon the temperature and the geometry of the hydrophobic substance (Lee, 2008). Hydrophobic substances have been shown to exhibit extraordinary stronger interactions in aqueous phase as compared to the interactions in the gaseous state primarily because of van der Waals interactions. Therefore, due to poor dispersibility of hydrophobic moieties in water, they tend to form aggregates which ultimately result in self-assembly to generate micelles and lipid bilayers.

# Hydrogen Bonding

Hydrogen bonding constitutes the most attractive type of bonding in controlling inter- or intramolecular orientations in self-assembly. It also helps in understanding the variety of events in biological systems (Lee, 2008; Mendes et al., 2013). The strength of H-bonding varies from 10–50 kJmol−<sup>1</sup> , which indicates that this bonding has capability to provide sufficient stability to the self-assembled clusters. Basically, H-bonding occurs due to dipole-dipole attraction which takes place between a H-atom attached to an electronegative atom and an electronegative atom with lone pair of electrons present in the vicinity. Generally, it happens between H and O, F, and N. Strength of H-bonding is also affected by the surrounding medium, i.e. solvent. An additional feature of H-bonding is that it imparts stability as well as directionality to self-assembly. This property facilitates self-assembled structures to gain various morphologies useful for various biomedical applications.

# Aromatic π-π Stacking

Aromatic π-π stacking refers to another type of non-covalent interactions which are quite attractive to researchers for cooperative binding during self-assembly. It occurs between aromatic residues as they contain pi bonds. These interactions have been found to be of considerable importance in DNA and RNA molecules (nucleic base stacking), folding of polypeptide/protein chains, template-directed synthesis, materials sciences, and molecular recognition (Bissantz et al., 2010). Large polarizabilities and a significant quadrupole moment, generated by a particular shape and electronic properties of the aromatic ring systems, result in a set of preferred interaction geometries. As demonstrated by various theoretical and practical investigations, it has been well established that aromatic ring systems have tendency to form ordered clusters of four different types, viz., parallel displaced, T-shaped, parallel staggered, or Herringbone (Gazit, 2002). These geometries might be possibly potential minimum configurations in the Lennard– Jones–Coulomb empirical potential calculations. For interactions between two π systems, the predominant arrangements are the T-shaped edge-to-face and the parallel-displaced stacking arrangement. In proteins, the parallel-displaced stacking arrangement is observed more frequently. Stacking is potentially more favored between electron deficient aromatic rings rather than electron rich rings. Moreover, the alignment of positive and negative partial charges and molecular dipoles significantly affects the preference among the orientation of heteroaromatic

rings. This becomes even more attractive when edge-to-face interactions are increased as a result of increased acidity of the interacting hydrogen atom. The effect is visible when a strongly electron withdrawing substituent in ortho or/and para position is introduced (Wheeler and Houk, 2009).

The steric constrains observed during the formation of the organized stacking structures have an essential role in the process of self-assembly that leads to the formation of supramolecular structures. Such π-π stacking interactions are responsible for stabilization of the tertiary structure of proteins, host–guest interactions, double-helix structure of DNA involved in core packing, and porphyrin aggregation in solution.

Gazit (2002) has also reported that π-π stacking interactions play a significant role in self-assembly of amyloid fibril formations. π-π stacking provides two important elements for the formation of these structures, (i) an energetic contribution that drives the self-assembly process thermodynamically and (ii) specific directionality and orientation that are driven by the set of stacking pattern (Gazit, 2002). This becomes more important because amyloid fibrils are well-defined supramolecular structures and a pre-determined pattern of stacking leads to formation of an organized structure. On analyzing a group of proteins with known structures having ππ stacking in them, it was noticed that a parallel displaced π-π stacking is the major organization of π-π interactions in proteins.

### FABRICATION OF SELF-ASSEMBLED AGGREGATES

Self-assembly is a process that involves balancing between attractive driving force, repulsive opposition force, and directional force (Lee, 2008; Mendes et al., 2013; Genix and Oberdisse, 2018). Particularly, a sweet balance between attractive and repulsive forces initiates the formation of self-assembled aggregates, which is a random process and also shows nonhierarchical structures (**Figure 3**). Most of the colloidal and micellar systems fit in non-hierarchical type of self-assembly. Addition of directional force to the other forces, the self-assembly processes become directional. Moreover, the self-assembled aggregates in such cases usually show hierarchical structures that include biological and bio-mimetic systems.

#### Micelles

In case of micelle formation by surfactant molecules, the attractive and repulsive forces guide surfactant molecules to come close enough to acquire an ordered structure (Lee, 2008; Xing and Zhao, 2016). The driving force that allows the formation of micellar system is the hydrophobic attraction while ionic repulsion and/or solvation force acts as the opposition force. As a result of this arrangement, at a certain position, the attractive and repulsive forces balance each other, which results in the formation of micelles. Concentration of the surfactant is the concentration that is required to form the first micelle (CMC). Addition of more amounts of surfactant molecules in bulk solution will result in the formation of additional micelles following the same force balance scheme. During this process, the size of the micelles remains invariable.

#### Vesicles

Vesicles are sphere-shaped lamellar structures having a hollow aqueous core (Xing and Zhao, 2016). The formation of vesicles can be viewed as two-step self-assembly process in which amphiphiles first form a bilayer which then closes to form a vesicle. A number of amphiphilic organic compounds, varying from natural to synthetic, exhibit vesicle formation (Lombardo et al., 2015; Xing and Zhao, 2016). Natural phospholipids, amphiphilic polymers, and polypeptides capable of forming vesicles are called liposomes, polymerosomes, and peptosomes, respectively (Xing and Zhao, 2016). Among these classes of compounds, most of them are formed of a hydrophilic head and a lipophilic tail that induces formation of vesicles. During exposure to aqueous media, the hydrophilic head interacts with water while the hydrophobic tail contracts inside to minimize exposure to water. In this process, the lipophilic part of the amphiphile buries inside the bilayer and the hydrophilic part forms the interior and exterior positions exposed to aqueous environments. Differences in the arrangement of molecules lead to unilamellar or multilamellar vesicles with diameters in range from 20 nm to several micrometers according to the number of bilayers present in the newer structures formed.

In a self-assembly of amphiphiles into a vesicle or other types of structures, the volume ratio of the hydrophilic and lipophilic parts plays a significant role and it is a dominant factor which is now being applied for designing and development of vesicular structures.

In the equation, P = v/la,

"v" and "l" symbolize the volume and length of the lipophilic part, while "a" symbolizes the volume of the hydrophilic head. "P" values can help in speculating the morphology of the nanostructures and explaining the phase transitions.

If

P < 1/3, spherical micelles 1/3 < P < 1/2, worm-like micelles 1/2 < P < 1, vesicles P = 1, planar bilayers P > 1, inverted structures.

This theory was initially applied to surfactant systems but now it is being applied in studying self-assemblies of other kinds of amphiphiles that include amphiphilic block copolymers which follow the same principles as the small-molecule-based systems (Chu et al., 2013). Having both the hydrophilic interior and hydrophobic membrane, vesicles can be used to entrap both the hydrophobic as well as hydrophilic drugs at the same time. Liposomal vesicles have been well demonstrated to carry a wide range of therapeutic molecules and some of them are currently being used in clinical applications. Some recent papers focus on manipulating the size, shape, physical properties, and biodistribution of vesicles for drug delivery applications and emphasize the need of further control of these parameters of vesicles for therapeutic delivery applications (Zhao et al., 2017).

#### Fibrillar Networks or Hydrogels

fbioe-08-00127 February 22, 2020 Time: 15:45 # 7

Hydrogels are 3-D continuous interpenetrated network of phases, the solid and the liquid phase (Lee, 2008; Mendes et al., 2013; Shang et al., 2019). The liquid phase of a hydrogel comprises of water while the solid phase is network structure in which nanofibers are formed via molecular self-assembly (i.e. molecular gelators). Fibers can be formed from self-assembled proteins, peptides, lipids, and hybrid amphiphiles. However, their formation is significantly dependent on hydrophobic– hydrophilic balance as it is essential for self-assembly. These nanofibers act as the matrices of a hydrogel. It also prevents the undesirable precipitation or dissolution of the hydrogelators (Du et al., 2015). The hydrophilic part of the molecule locates itself as the exterior portion of the nanofibers, which gets involved in hydrogen bonding with the surrounded water molecules making it certain that hydrophilic biomolecules such as drug molecules (small peptides) can be translated into hydrogelators. Such supramolecular structures interact with the target molecules/sites more efficiently than the native biomolecules thereby increasing their bioactivity. As a hydrogel contains ∼97% of water, still it behaves like a solid and can flow only when a shear force is applied. Generally, hydrogels display response to an external stimulus and undergo a phase transition upon its application because these are formed via the self-assembly of small molecules through hydrogen bonding and hydrophobic interactions which are quite weak interactions. Apart from this, the supramolecular hydrogels offer an added advantage that these are biocompatible and biodegradable, as well as resemble to extracellular matrices which help in design and synthesis of novel supramolecular hydrogelators as materials for biomedical applications. Hydrogel materials have been intended to synthesize for encapsulation and delivery of water soluble therapeutic molecules. There are many reports in literature which demonstrate the encapsulation and release of small hydrophilic molecules, proteins, and cells from the hydrogels (Narayanaswamy and Torchilin, 2019). Drug molecules can be entrapped into the networked structure during initiation of self-assembly process. Hydrogels, formed mainly by the process of self-assembly, are joined together through noncovalent crosslinking (covalent or physical hydrogels), which also determines its actual mechanical strength. The classification of the hydrogels can be made on the basis of their source (natural or synthetic), nature (degradable or non-degradable), networking (covalent or physical), and the nature of network (homopolymeric, copolymeric, interpenetrating networks, and double networks).

#### APPLICATIONS OF SELF-ASSEMBLED MATERIALS

The term nanostructure generally refers to those materials/structures which have structured components with at least one dimension less than 100 nm. The properties (both physical and chemical) of nanostructures are markedly dissimilar from their monomeric unit or the bulk material having identical chemical composition. The main reason for this unique behavior at nano-scale is due to the appearance of new quantum effects as well as enhanced surface area to volume ratio (Dahman, 2017). As the nanostructures have higher surface area to volume ratio as compared to their conventional forms, they exhibit greater chemical reactivity and strength. These emergent properties exhibited at nano-scale have the potential for greater impacts in biomedical applications. Suitable modulation of the properties and response of nanostructures may result in the creation of new desired gadgets and technologies.

The area of nanobiotechnology for therapeutic delivery is flooded with new challenges as the demand for new medical therapies is increasing exponentially. Earlier, the nanomedicines were developed by reformulating the available drugs in nanostructures. With the development of nanomedicines, which have shown the potential to treat the diseases in a much better way, the demand for personalized medicines has grown up that requires the customization of the fabrication of the nanostructures with in-built desired properties (Cui et al., 2010). For this customization, an improved control over structure, composition, as well as function of the matter at molecular level is needed. To achieve this control at molecular level, self-assembly comes into picture which can play a very crucial role by adjusting various parameters such as size, shape, and surface chemistry mimicking the 3-D structure of biomacromolecules. Thus, novel nanomaterials can be produced with greater ease and economically by employing the tools of molecular self-assembly. Furthermore, diverse nanostructures with varied functionality can be produced by this process (Genix and Oberdisse, 2018). The great advantage of this scheme for the formation of nanosized structures is the structural control over the final self-assembled nanostructures which can be achieved by varying the monomer, its composition, and chemistry, by inducing environmental changes (solvents, temperature, pH, and co-assembling molecules), and changing the rate of self-assembly process (Dallin et al., 2019). The ultimate goal of these self-assembled nanostructures is to attain their required functions; whether these structures are thermodynamically stable or not. As discussed above, selfassembly easily provides the flexibility to develop newer materials with customized morphologies and preferred functionalities and thus provides better control over bulk properties of the resulting nanostructures. Hence, it is quite simple to presume the behavior of final assembly by controlling the structural changes in the constituent molecules. In recent past, a plethora of nanostructures have been produced using different biopolymers (proteins, carbohydrates, nucleic acids, etc.) which have further refined the concepts and knowledge of this process as well as enhanced the use of these self-assembled materials in diverse medical applications such as in fabrication of molecular devices, delivery systems, or scaffolds (Panda and Chauhan, 2014; Azevedo and da Silva, 2018; Lombardo et al., 2019). These systems have shown their promising potential, however, need more attention to address some limitations in terms of their in vivo stability which has hindered their safe use in human beings.

Self-assembled nanomaterials are being used for a very broad range of applications from fundamental to applied research, with striking implementations in biomedical sciences, information

technology, and environmental sciences. Here, in this article, selfassembled nanostructures useful for biomedical applications have been the main focus, specially, drug delivery and gene delivery, so the subsequent part deals with these aspects (Busseron et al., 2013; Xing and Zhao, 2016).

# Drug Delivery

Therapeutic delivery is a very significant area to address concerns related to healthcare and medicine. Certain problems associated with the use of free drugs can be minimized by using the appropriate carriers for drugs such as stability issue of free drugs in biological system, short half-life, insolubility in aqueous environment, abnormality in biodistribution, and pharmacokinetics of the delivered drugs (Mohanraj and Chen, 2006). Controlled drug delivery has shown enhanced bioavailability of the therapeutic by avoiding their untimely degradation and improving their uptake, maintaining the therapeutic dose of the drug by controlling the kinetics of drug release, and reducing toxicity by targeting to desired sites/tissues. In this regard, nanoparticles have proved to be potential DDS due to their advantageous characteristics. Many positive aspects of nanoparticle-mediated delivery of therapeutics have been realized (Wang et al., 2018).

Nanoparticles as therapeutic delivery systems offer several advantages:


# Criteria for the Designing of New Delivery Vehicles

Criteria for the designing of a new delivery vehicle are highly dependent on the therapeutics to be delivered and intended applications. Some of the common points, that are kept in mind while designing these vectors, are given below (Yu et al., 2016; Knauer et al., 2019).

i. The delivery vehicles need to be non-toxic, biocompatible, and biodegradable, and get readily eliminated from the body.


#### Polymers

A wide range of self-assembled polymeric nanostructures have been used for drug delivery, but biodegradability is essential to overcome side effects and toxicity to healthy tissues (Sofi et al., 2018; George et al., 2019). The selfassembled nanostructures are formed from both natural and synthetic polymers. Numerous self-assembled DDSs have been developed which have successfully encapsulated drug molecules to improve bioavailability, bioactivity, and controlled delivery, with some achieving clinical testing (Felice et al., 2014) and some of them have been launched for commercial purposes (Felice et al., 2014) (**Table 1**).

#### **Natural polymers**

Natural polymers possess abundance of functional groups, amenable to modifications via chemical or biochemical routes that result in the generation of different types of biopolymerbased materials (Nitta and Numata, 2013; Abedini et al., 2018; Sofi et al., 2018). Among these natural polymers, polysaccharides constitute an important class of polymers which are being used more frequently for various biomedical applications. Polysaccharides are carbohydrate polymers of monosaccharidebased repeating units connected through glycosidic linkages. Their source of production is quite diverse; hence, these have different structures and properties, a wide variety of reactive functionalities, different chemical compositions, and molecular weights (Nitta and Numata, 2013). Based on their functional groups, these have been divided into two main categories, viz., non-ionic (dextrin, pullulan, dextran) and ionic polysaccharides (heparin, chitosan, alginate, etc.). Polysaccharides are considered to be highly stable, safe, non-toxic, biodegradable, hydrophilic, and biocompatible. By tethering lipophilic moieties on the polysaccharides, the resulting conjugates can readily selfassemble into micelles in aqueous solutions and can potentially be used for drug delivery applications. Some of the polysaccharides possess certain bioactive groups which can act as targeting

#### TABLE 1 | List of nanoengineered polymers for drug delivery applications (Felice et al., 2014; Bobo et al., 2016; Patra et al., 2018).


moieties. HA can act as ligand for targeting receptors present on the endothelial cells of liver and certain cancer cells. Self-assembled nanostructures of amphiphilic HAs have been highly investigated as active targeting agents in drug delivery (Cho et al., 2012). Self-assembled structures of modified cellulose, chitosan, and pullulan-based polysaccharides have also been used for colon targeting. These polymers promote drug absorption due to enhanced mucoadhesion in the small intestine. Similarly, amphiphilic heparin-based systems have also shown to reduce tumor size and blood vessel formation in tumor area (Niers et al., 2007). The most commonly and extensively used polysaccharides, namely, alginate, chitosan, and dextran, have been described here along with their therapeutic advantages.

Alginate. Alginate (sodium salt) is a water soluble polysaccharide which is made up of 1–4 linked α-L-glucuronic acid and α-Dmannuronic acid in alternating order. Modification of alginate produces diverse polymers which behave in different manners under physiological conditions. As biodegradability of polymer is improved by oxidation of hydroxyl group, sulfonation increases

blood compatibility (Kumar et al., 2004). Self-assembled PEG derivatives of alginate have shown significant improvement in hypocalcemia efficacy in rats by enhancing the oral delivery of calcitonin (Li et al., 2012). Recently, phenylalanine ethyl ester modified alginate self-assembled nanoparticles showed good in vitro cellular uptake efficiency and biocompatibility profile in human intestinal cell lines (Zhang P. et al., 2019). Ayub et al. (2019) synthesized cysteamine conjugated disulfide crosslinked sodium alginate nanoparticles by layer by layer self-assembly mechanism to get better delivery of an anticancer drug, PTX, for colon cancer. Further, the alginate nanoparticles have been used for antigen delivery also. Antigen-BSA encapsulated polylysinesodium alginate nanoparticles were formed by process of selfassembly using electrostatic interactions between oppositely charged polyelectrolyte complexes. These particles showed sustained release behavior of vaccine and enhanced cellular uptake without imparting cytotoxicity in vitro (Yuan et al., 2018). The self-assembled alginate-based nanoparticles have been used in treatment of multidrug resistant tumors (Kumar et al., 2019) and in combinational chemotherapy (Zhang et al., 2017) as stimuli responsive (redox and light responsive) nanoparticles. Bazban-Shotorbani et al. (2016) synthesized alginate nanogels via microfluidics with tunable pore size and these nanogels were used for protein delivery.

Chitosan. Chitosan modifications and their use in delivery applications of various therapeutic molecules have been extensively reviewed in the literature (Coviello et al., 2007; Wasiak et al., 2016; Wang et al., 2017a; Quiñones et al., 2018). Chitosan, an unbranched linear polysaccharide, is made up of β-(1-4)-linked D-glucosamine (deacetylated unit) and N-acetyl-D-glucosamine (acetylated unit). It is produced from the skeleton of shellfish, including shrimp, lobster, and crab. It is used in various medicinal formulations such as filler in tablets, controlled-release drugs, and to improve the solubility of drugs. Self-assembly of the modified chitosans into micelles in aqueous solution with hydrophobic pockets has been used to entrap various anti-tumor therapeutics such as PTX, doxorubicin, and camptothecin (Quiñones et al., 2018). Recently, Trummer et al. (2018) synthesized N-benzyl-N,O-succinyl chitosan, N-naphthyl-N,O-succinyl chitosan, and N-octyl-N,O-succinyl chitosan-based self-assembled nanocarriers and successfully coordinated to antitumor drug cisplatin and evaluated the efficacy of these nanocarriers in vitro in human carcinoma cell line HN22 and HN29. The results showed high efficacy of N-benzyl-N,Osuccinyl chitosan-mediated cisplatin delivery. They observed that the encapsulated formulation was less cytotoxic and caused lower cisplatin-induced renal cell death but it exhibited greater apoptosis in HN22 cells as compared to native cisplatin. Besides, the formulation provided long-lasting treatment with reduced nephrotoxicity. Chen et al. (2019) prepared polyelectrolyte complexes via self-assembly of opposite charged alginate-coated nanoparticles and chitosan nanoparticles and used this complex for pH sensitive controlled release of insulin.

Dextran. Dextran is a polymer formed from joining of glucose units through α-1,6-linkages with branch points at α-1,2, α-1,3, and α-1,4 linkages. It is non-toxic, highly biocompatible, and could be widely used in medicinal products including development of drug-delivery systems (Wasiak et al., 2016). It has been extensively used as a supplementary material to prevent the formation of blood clots by reducing blood viscosity and iron-dextran conjugates have been applied for fulfilling the iron deficiency. Derivatives of dextran have also been used as the source of biocompatible hydrogels for drug delivery applications to attain regulated and sustained drug release profile for longer time periods (Coviello et al., 2007). Wang et al. synthesized dextran nano-hydrogel by conjugating polyacrylic acid via disulfide crosslinking to dextran. The anticancer drug (doxorubicin) was conjugated to Dex-SS-PAA. The results showed that these nanohydrogels exhibited stimuli (pH and redox) responsive drug release behavior as well as greatly reduced the toxicity of free doxorubicin, and inhibited the growth of MDA-MB-231 tumors (Wang et al., 2017a). In another recent study, folic acid-dextran conjugates were synthesized which showed pH responsive self-assembly behavior. This conjugate self-assembled into nanoparticles at pH 7 and dissociated at pH > 9. The anticancer drug, doxorubicin, was efficiently entrapped in these particles and exhibited targeted drug delivery in vitro with enhanced antitumor activity in 4T1 subcutaneous tumor bearing mouse model (Tang et al., 2018). The modified soy-protein and dextran nanogels have been used for the delivery of riboflavin (Jin et al., 2016).

Cyclodextrins (CDs). Cyclodextrins are oligomers of glucose consisting of six, seven, and eight glucose units in α-, β-, and γ-CDs, respectively. The exterior of the cup-shaped molecule is hydrophilic while the internal part is hydrophobic, thus, they are readily soluble in aqueous environment and they can include small, hydrophobic "guest" molecules in their interior and thus forming inclusion complexes (Muankaew and Loftsson, 2018). Due to their inherent biocompatible nature, FDA approved their use in pharmaceutical formulations as solubilizing agents. CD derivatives are synthesized by replacing hydroxyl group on CDs with desired functional groups. The natural biocompatibility and self-assembling attributes of CDs have made them efficient nanocarriers for drug and gene delivery. These molecules can form diverse nanostructures such as nanoparticles, nanospheres, nanogels, nanomicelles, etc. The various modifications have been done on CDs to form amphiphilic derivatives that can self-assemble in aqueous environment and enhance interaction with cell membranes (Simoes et al., 2015). The modified CD amphiphiles can be cationic, anionic, or neutral depending on the groups attached to them. To form sustained drug release carriers, hydrophobic modifications have been done on CDs. He et al. (2013) synthesized acetalated α-CD material that showed pH modulated hydrolysis and pH triggered drug release behavior of encapsulated PTX drug in vitro. CDs have also been conjugated to various polymers to improve their physiochemical properties and enhance their drug delivery efficiency (Zhang and Ma, 2013; Zerkoune et al., 2014). Song et al. (2016) prepared β-CD conjugated poly[n-isopropylacrylamide] polymer as a temperature responsive drug carrier. This polymer self-assembled and formed inclusion complexes with PTX drug via host–guest interactions. The enhanced cellular uptake and antitumor effect were observed in cancerous cell lines (Song et al., 2016).

#### **Synthetic polymers**

fbioe-08-00127 February 22, 2020 Time: 15:45 # 11

Among the commonly used synthetic polymers, block copolymers are a special class of polymers in which two or more blocks of polymers are attached in a regular arrangement. Block copolymers containing two, three, or more blocks are named as diblocks, triblocks, or multiblocks, respectively. PLA, PGA, and their copolymer PLGA, apart from being biodegradable and biocompatible, have been explored for therapeutic delivery and these are approved by the US Food and Drug Administration (Vilar et al., 2012).

Block copolymers are macromolecules which are formed by the linear and/or radial array of two or more dissimilar blocks having different monomer composition to impart amphiphilicity to molecule. The ever-increasing interest in block copolymers has recently arisen due to combinatorial qualities attained by the combination of two different polymers which leads to generation of micellar systems useful for carrying hydrophobic therapeutics.

A variety of amphiphilic copolymers, viz., di-block (A-B) and tri-block (A-B-A) grafted polymers, are being used to form selfassembled nanostructures for different biomedical applications. Among these nanostructures, the micelles are the most common structures formed from these copolymers or block polymers. On dissolving a block copolymer in a solvent, which can be an excellent solvent for dissolving one block and a poor solvent or precipitant for the second block, the copolymer molecules quickly align themselves to attain micellar structure and this phenomenon of micelle formation is reversible also. The most frequently used hydrophilic block is PEG. Other hydrophilic polymers which are commonly used are poly(Nvinylpyrrolidone) and poly(N-isopropylacrylamide). The core forming hydrophobic blocks which are most frequently used are poly(propyleneoxide), poly ε-caprolactone, polylactic acid (PLA), poly(lactide-co-glycolic acid) (PLGA), poly(L-aspartic acid), poly(L-histidine), poly(β-amino ester), etc. Among these polymers, PLA, PLGA, and PEG are the ones which have been approved or entered the clinical trial phases (Vilar et al., 2012; Felice et al., 2014) (**Table 1**).

In the last decade, for the fabrication of polymeric nanoparticles, the process of PISA has been used extensively in which polymerization as well as self-assembly occur simultaneously in one vessel to form polymeric nanoparticles. Drug can be encapsulated during the PISA process of nanoparticle formation as well as post-PISA process (Zhang W.J. et al., 2019).

Polylactic acid (PLA). Poly(D,L-lactic acid) is biodegradable polyester used in the fabrication of stents, implants, and various other medical devices (Hoffman, 2008). On hydrolysis, it degrades into monomeric lactic acid, which is also produced during anaerobic respiration in living beings. The polymer, characterized by its inherent viscosity, is dependent on its chain length/molecular weight. A controlled release of the entrapped therapeutic is also dependent on the PLA chain length. PLA is available commercially as Lupron Depot and Risperdal Consta in the form of microparticles. Among PLA matrices, the PLA-PEG micelles have extensively been used in drug delivery applications. For instance, Genexol PLTM is PTX encapsulated PLA-PEG micelles. It is clinically approved in South Korea and Europe (Kim et al., 2004); however, in United States, it is still under phase II clinical trials. Amphotericin B was also encapsulated in PLA micelles and sustained drug release was observed. PLAbased micelles have been used in other drug delivery formulations also (Liu et al., 2008). Apart from these, PLA-based nanoparticles have also been used for entrapment of nucleic acid and their delivery (Munier et al., 2005). Several PLA-based nanostructures are under pre-clinical investigation.

Poly(lactic-co-glycolic acid) (PLGA). Poly(D,L-lactic-co-glycolic acid) is made up of two polymers, i.e. lactic acid and glycolic acid, which on hydrolysis yields biodegradable metabolite monomers, i.e. lactic acid and glycolic acid. These biodegradable metabolites are involved in several biochemical and physiological cycles in the living systems displaying minimal systemic toxicity. Degradation rate of PLGA highly depends on its molecular weight and monomer ratio (Danhier et al., 2012). Till now, PLGA-based therapeutic delivery systems have not been approved but certain PLGA-based systems are under pre-clinical and clinical trials. PLGA-based nanostructures are primarily used in the entrapment of lipophilic antitumor therapeutics, viz., PTX, vincristine sulfate, doxorubicin, curcumin, tetrandrine, etc. (Que et al., 2019). In one of the latest reviews, the industrial and scientific aspects of PLGA nanoparticles have been highlighted (Qi et al., 2019).

Polyethylene glycol (PEG). Polyethylene glycol is a polyether, a non-biodegradable hydrophilic polymer with a variety of applications in pharmaceutical and biomedical areas (Hutanu et al., 2014). PEG helps in increasing the dispersibility of the attached molecules. It has been used in the preparation of polymer-drug conjugates and provides stabilization as well as imparts stealth properties to the so formed DDS.

Polyethylene glycol polymers with reactive functionalities at their termini have been demonstrated to exhibit wide variety of applications. Bi-functional and mono-functional derivatives are also being used as crosslinkers and linkers or spacers. PEGbased carriers for drug delivery such as micelles, nanoparticles, dendrimers, and liposomes are better than PEG conjugates of the drugs (Hutanu et al., 2014).

Dendrimers. Dendrimers are the specialized macromolecules which offer regular and highly branched three-dimensional structures. Their unique structures show high density of functionalities at the periphery of the molecules. For instance, dendrimers with peripheral amines allow efficient condensation of negatively charged nucleic acids while the tertiary amines in the core remain available for playing an important role during endo/lysosomal acidification which enables more efficient endosomal release. Dendrimers consist of three major architectural components: a core, inner shell, and an outer shell. These can be synthesized in two ways to have different functionality in each of these components to modulate various properties such as solubility, thermal stability, and attachment of compounds for particular applications (Thota et al., 2015).

A dendrimer is typically symmetric around the core. Its structure provides relatively easy access to control their size, composition, and chemical reactivity very precisely. The degree of branching is expressed in the form of generation of the dendrimers. The size and surface charge on dendrimers vary with the number of "generations" during synthesis. Because of the presence of large number of tertiary amines, PAMAM dendrimers act not only as proton sponges in gene delivery applications but also along with carbon skeleton, they have been used as drug carriers simultaneously (Salzano et al., 2016; Abedi-Gaballu et al., 2018; Li et al., 2018). Haensler and Szoka (1993) reported for the first time the use of PAMAM dendrimers in gene delivery. They showed that the sixthgeneration dendrimer was almost 10-folds better than lower generation ones. Based on this study, PAMAM dendrimers have recently been used in several in vivo and in vitro gene delivery applications and found to be biocompatible (Maruyama-Tabata et al., 2000; Yang et al., 2015; Araújo et al., 2018). PAMAM dendrimers have a well-defined size and shape but offer limited flexibility. Therefore, attempts have been made to hydrolytically cleave some of the amide bonds in the inner part. Breaking of some of the branches of dendrimers in the core enhances the flexibility and the resulting molecules are known as activated dendrimers. Although activated and inactivated dendrimers were found to form complexes with DNA by electrostatic interactions and mediated transfer of bound DNA into eukaryotic cells, the overall transfection efficiency of activated dendrimers was found to be two to three times higher than the inactivated (native) dendrimers. The fractured or activated dendrimers not only showed the greater flexibility to interact with plasmid DNA but their solubility also enhanced and presented less tendency to aggregate. This enhanced flexibility of activated dendrimers showed better endosomal release of the DNA and subsequently, the transfection efficiency. SuperFect (from QIAGEN, Hilden, Germany) is an example of commercially available efficient transfection reagents based on the fractured G-5 PAMAM dendrimer. In another attempt, the peripheral amines of PAMAM-G4 were converted into guanidinium (Gn) and tetramethylguanidinium (TMG) moieties. Although these modified dendrimers did not display cytotoxicity in various mammalian cells, higher transfection efficiency was observed only in case of guanidinium-PAMAM-G4 (Yadav et al., 2014b). Somani et al. have investigated the effect of pegylation (2 and 5 kDa) on G-3 and G-4 diaminobutyric polypropylenimine dendrimers. Cytotoxicity decreased significantly in these modified dendrimers without compromising DNA condensability; however, enhanced gene expression was found in G-3 and G-4 daminobutyric polypropylenimine dendrimers conjugated to 2 kDa PEG in cell specific manner (Somani et al., 2018). Further, Gao et al. studied structure activity relationship to design efficient gene delivery vectors. They demonstrated that both hydrophobic modification and density of amines modulate the gene transfer ability of synthetic vectors (Gao et al., 2018).

Subsequently, several hydrophobic modifications have also been incorporated in dendrimers to make them amphiphilic which can self-assemble and be used as delivery vectors (Bolu et al., 2018). Han et al. (2017) developed an amphiphilic conjugate of PAMAM dendrimer by conjugating hydrophobic PLA which on self-assembly formed core-shell nanostructures in which 5-FU and doxorubicin were entrapped efficiently for combinatorial anticancer therapy. This nanomicelle system showed synergistic antitumor effect on MDA-MB 231 tumor cell line and MDA-MB 231 xenograft mice model (Han et al., 2017). In another report, an amphiphilic block micelle was synthesized by conjugating hydrophobic block of linear poly e-caprolactone polymer with hydrophilic part of methoxy terminated PEG decorated G-3 polyester dendron (WooáBae, 2011). They further explored the effect of peripheral functional group such as amine, carboxylic, and acetyl using –NH2, –COOH, –COCH<sup>3</sup> group terminated PEG chains instead of methoxy terminated PEG used in their earlier study (Pearson et al., 2012). The group used this amphiphilic micellar system to encapsulate and deliver anticancer drug, endoxifen. Dendron micelle system containing carboxy terminated PEG showed the highest potential to deliver the drug across skin layers among the tested systems (Yang et al., 2014). This research group further evaluated the effect of density of targeting moiety, folic acid, and PEG length on these dendritic micelles in terms of interaction with cells (Pearson et al., 2016). There are lots of preclinical studies on dendrimers-based drug delivery; however, the clinical ones are very few. A dendrimer drug formulation, DTXSPL8783, for advanced cancer treatment, is currently undergoing clinical phase 1 trial (Caster et al., 2017; Ventola, 2017), while another dendrimer-based antiviral product, Vivagel from Star Pharma, is in phase III trials for bacterial vaginosis (Ventola, 2017). Some of the nanoengineered polymeric systems either approved by FDA or in the advanced clinical phases are listed in **Table 1**.

#### Self-Assembling Small Molecules

The self-assembled nanostructures formed from the small peptides (Fleming and Ulijn, 2014; Panda and Chauhan, 2014), lipid-based systems (Li et al., 2015), and other small molecules (Xing and Zhao, 2016) can be used as carriers for different therapeutic molecules (Wang et al., 2017b; Liu et al., 2018; Yang et al., 2018). The small molecules can be produced easily as compared to the larger ones and act as efficient and economical alternatives to the large molecule-based systems. Moreover, different small peptides can be combined with diverse synthetic molecules thus producing tailored nanoscale engineered biomaterials that can be used as carriers of genetic materials, drug molecules, and antimicrobial agents. Now-adays, carrier free and self-assembled small molecule-based nano-DDS are also being developed with the aim to eliminate the issue of undefined metabolism and clinical safety of the carriers (Guo et al., 2017; Jiang et al., 2017; Zheng et al., 2019). Recently, Guo et al. have demonstrated the efficacy of a carrier free system by developing a theranostic nanodrug delivery formulation for NIR imaging and chemotherapy. In this system, indocyanine green, ursolic acid, and PTX formed a selfassembled system via aromatic pi-pi and electrostatic interactions (Guo et al., 2017). Similarly, another carrier free nano-DDS was developed in which anticancer drug doxorubicin and ursolic

acid were co-assembled by pi-pi stacking, hydrophobic, and electrostatic interactions, and modified with HER-2 aptamer for targeting to HER-2 receptors overexpressing cancer cells (Jiang et al., 2017).

#### **Lipids**

This group of carrier materials comprise of cholesterol and phospholipids as the key constituents. Phospholipids, the constituent of all cell membranes as well as the major component of liposomes, are mainly one or two fatty acids linked to glycerol or sphingosin with a phosphate head group which impart amphiphilicity facilitating the formation of bilayered membranes in liposomes. It is reported in the literature that the ordinary amphiphiles have critical micellization concentrations in the range of 10−2–10−4M, whereas CMC of phospholipids is four to five times smaller, thus these materials have extremely low water solubility. As a consequence of this, they have high stability after administration in comparison to micelles (Felice et al., 2014). Among the natural and anionic phospholipids, the phosphatidylcholine, which is the most studied lipid, present in both plants and animals, consists of a phosphate and quaternary ammonium group. Among the natural anionic phospholipids, phosphatidic acid, phosphatidylglycerol, phosphatidylserine, and phosphatidyl ethanolamine are the most commonly used ones.

The natural cationic lipids such as the stearylamine are mainly used for encapsulation of nucleic acids. The synthetic phospholipids such as dimyristoyl-phosphatidylcholine, dipalmitoylphosphatidylcholine, and DSPC are also used. Natural phospholipids are preferred over synthetic ones as they show greater chemical biostability against phospholipases, esterases, bile salts, and serum proteins thus imparting the greater thermodynamic stability to the vesicles against alkaline pH, high temperature, and oxidative stress conditions. On the other hand, liposomes permeability can be controlled in a better way using synthetic lipids. In biological environment, the phospholipids are degraded by lipolysis and thus result in low toxicity. The fluidity of the liposomal membrane is influenced by the viscosity of the phospholipids which is regulated either by using phospholipids possessing elevated phase shift temperatures, or through insertion of cholesterol molecules. The second most commonly used lipid in nanoengineered DDSs is cholesterol, which is responsible for reducing the permeability of hydrophilic molecules by increasing the stability of liposomal membrane (Felice et al., 2014). Kirby et al. studied the effect of cholesterol on liposomal membranes prepared using natural phospholipids and found that cholesterol containing liposomal membranes were more stable in comparison to cholesterol deficient membranes (Felice et al., 2014) in blood. Deng et al. (2018) and Massiot et al. (2019) developed X-ray radiation triggered (Deng et al., 2018) and photo-triggered (Massiot et al., 2019) liposomal systems for sustained release of the drug molecules (Pattni et al., 2015; Deng et al., 2018; Massiot et al., 2019). A great development has been achieved on liposomal technology advancement; however, their full potential is yet to be explored, as successful bench to bedside applications are very few. So, there is still the need of further development of liposomal

technology for advancement in therapeutic delivery systems (Pattni et al., 2015).

Most of the commercialized liposomes have been used to encapsulate anticancer drugs. Among them, MyocetTM and DoxilTM, which encapsulate doxorubicin, are the most efficacious formulations (Felice et al., 2014) (**Table 2**).

Doxil is the first liposomal formulation of anticancer drug, doxorubicin, permitted by the FDA (United States) in 1995 for AIDS associated with Kaposi's sarcoma (Northfelt et al., 1998). In this case, the stealth liposomal carrier consists of HSPC, cholesterol, and PEGylated phosphoethanolamine. By encapsulating doxorubicin into stealth liposome carriers, both the circulation half-life of drug and its accumulation in tumor environment were got enhanced. Doxorubicin, an anthracyclin, is currently being used against various carcinomas and exerts its effect by inhibiting DNA and RNA synthesis but it causes side effects such as severe myelo-suppression and cardiotoxicity (Felice et al., 2014). These side effects were reduced to an extent when doxorubicin was entrapped in liposomes (Fan and Zhang, 2013). Initially, doxorubicin, encapsulated in multilamellar liposomes by passive entrapment, was found to be unsuccessful because of fast release of the drug followed by rapid clearance by phagocytic system of the body. Then active drug loading technique was employed to enhance the drug encapsulation content and stability which resulted in two formulations, MyocetTM and DoxilTM. In MyocetTM, doxorubicin was encapsulated by a pH gradient, while in DoxilTM, potential gradient was used to load the drug molecules. MyocetTM comprises cholesterol and EPC while DoxilTM contains cholesterol and hydrogenated soya phosphatidylcholine. Of these two formulations, PEG coating in DoxilTM improved its pharmaco-kinetic profile over MyocetTM. Both MyocetTM and DoxilTM showed significant reduction in the toxic effects of doxorubicin (Hofheinz et al., 2005).

Other liposomal drugs clinically approved are AmbisomeTM, in which Amphotericin, an antifungal drug is loaded, DepodurTM, in which morphine, an analgesic, is loaded and VisudyneTM, in which verteporfin, a drug for treatment of macular disintegration, is loaded. Besides, there are a number of other liposomal systems which are under phase II and III clinical trials. Liposome-based formulations in clinical trials are more than other types of nanoengineered DDSs. Among these, ThermoDoxTM is a temperature-sensitive liposomal formulation encapsulating doxorubicin. ThermoDoxTM comprises DPPC, MSPC, and DSPE-MPEG-2000 (Poon and Borys, 2009), the synthetic phospholipids. ThermoDoxTM releases its doxorubicin content under the influence of temperature, i.e. above 39.5◦C as this formulation has comparatively low Tm. Tm, phase Tm, is the temperature needed to induce change in the arrangement of lipid chains. At this temperature, the aligned gel phase structure changes into the non-aligned liquid crystalline phase structure. In the living system, this temperature (Tm) can be attained by heating the tumor with radio-frequency electromagnetic waves.

#### **Small peptides**

Liposomes are associated with some technological issues such as stability, reproducibility, poor drug loading, and insufficient


TABLE 2 | List of nanoengineered liposomal therapeutic delivery systems either approved by FDA or in advanced clinical trials (Felice et al., 2014; Bulbake et al., 2017; Patra et al., 2018).

control over drug release (Torchilin, 2005). The basic idea of fabricating nanostructures from peptide-based biomaterials came from the literature survey that showed that in certain diseases such as Parkinson's, Alzheimer's, and in Prion-related ones, the self-organized tubular shaped proteins were formed from otherwise soluble amphiphilic proteins in the cells (Gazit, 2004). These findings led the way to divert attention and focus researches on investigating self-assembly of peptides to form ordered structures. Peptides possess several inherent properties such as biocompatibility and biodegradability which make them very useful building blocks for forming self-assembled nanostructures for various biomedical applications (Panda and Chauhan, 2014). Furthermore, in case of peptides, the versatility of the chemical design and their capability to assume highly ordered organized structures offers chance of manipulating the final assembly by controlling structural features at the nanometric range. The properties of peptides such as sequencebased unique self-organization and recognition provide them a function of acting as a significant signaling molecule and key

building component of living beings. A range of self-organized nanostructures with distinct shapes such as fibrous, tubular, rod, particles, and various other shapes are also formed through selfassembly of peptides (Gazit, 2004; Panda and Chauhan, 2014; Prakash Sharma et al., 2015). In the past few years, research on medium sized peptides, small peptides, ultra small peptides, and peptide-conjugates have opened up new avenues for designing and synthesis of peptide-based self-assembled nanostructures for different biological applications (Fan et al., 2017; Guyon et al., 2018; Ni and Zhuo, 2019; Tesauro et al., 2019). These selfassembled peptide-based nanomaterials can be designed with ease to act as new scaffolds to mimic various biomaterials, tissues, etc. The usefulness of di- or tri-peptide based self-assembled nanostructures has been reported by many research groups. A variety of peptides such as surfactant like peptides, amphiphilic peptides, bolaamphiphiles, peptides containing α-helix, β-sheets, and β-turns, as well as cyclic peptides have been nanotized and studied in detail the process involved in their conversion. Peptides are generally labile to enzymatic degradation which

limits their use as therapeutic delivery agents but this limitation has been circumvented, to an extent, by incorporation of noncoded residues in peptide sequences (Gupta et al., 2007).

The self-assembled peptide-based nanostructures, i.e. nanotubes were developed for the first time by Ghadiri et al. using cyclic polypeptides, cyclo-[-(L-Gln-D-Ala-L-Glu-D-Ala)2- ], containing even number of D- and L-amino acids alternatively (Hartgerink et al., 1996). These cyclic peptides self-assembled to form nanotubes via intermolecular hydrogen bonding in which side chains of the amino acids lied toward the exterior surface and formed a hollow tube type arrangement. The cyclic peptide-based nanotubes have been used as antimicrobial agents, biosensors, catalysts, etc. (Brea et al., 2010).

A large number of peptides self-organize via formation of β-sheeted secondary structures. β-Sheeted peptides selfassemble to form diverse supramolecular architectures such as nanoribbons, nanotubes, monolayers with nanoscale order, etc. (Reches and Gazit, 2003; Ashkenasy et al., 2006). The 16 amino acid residues containing peptides, RADA-16 I and RADA-16 II, formed β-sheeted structures that produced nanofibrous network followed by pH-responsive hydrogels (Holmes et al., 2000) via self-assembly. The stimuli responsive hydrogel network formation by these peptides can also be enhanced by increasing ionic strength or altering the pH of the assembling environment. These peptides are now available commercially with the name, PuraMatrix (3DM, Inc., Cambridge, MA, United States).

The surfactant-like peptides also self-assemble to form nanostructures (Vauthey et al., 2002). These seven to eight amino acids long surfactant-like peptides (A6D1, V6D1, V6D2, L6D2) have similar properties as observed in biological surfactant molecules. They also contain a negatively charged hydrophilic head group at C-terminus, i.e. aspartic acid, and the hydrophobic amino acids, i.e. alanine, valine, or leucine formed the part of lipophilic tail. The N-terminus of the peptides was acetylated to create a neutral moiety, which facilitated self-assembly via electrostatic and hydrophobic forces (Tsonchev et al., 2008).

Amphiphilic peptides are formed via hydrophilic peptide forming head group and hydrophobic alkyl tail. The hydrophobic alkyl end helps in arrangement of the hydrophilic part to selfassemble in different higher order structures. These amphiphilic peptides self organize to form diverse morphological structures with nanodimensions such as micelles, vesicles, or tubules (Panda and Chauhan, 2014).

Bolaamphiphiles (KA6K, KA4K, KA8K) are amphiphilic molecules made up of two hydrophilic groups flanked by hydrocarbon chains. In these peptides, β-sheet H-bonding interactions result in formation of a variety of structures such as nanofibers, nanorods, nanotubes, nanoribbons, nanospheres, etc. (Panda and Chauhan, 2014). However, in case of nanostructure formation by these large linear peptides, cyclic and dendritic structures, the related expense and complexity of synthesis restrict the use of these peptides practically. In addition to this, these peptides are also not stable under enzymatic exposure which hampers the use of these peptides in biological applications.

Very short peptides such as di-, tri-, tetra-, and pentapeptides also self-assemble to form diverse nanostructures (Panda and Chauhan, 2014). These short peptide fragments were carefully studied in order to find out the minimum sequence required for amyloid formation. In amyloid fibrils polypeptide, a hexapeptide fragment NFGAIL (hIAPP22–27) of the islet amyloid polypeptide (IAPP) formed the well-organized amyloid fibrils which were alike to amyloid fibrils of complete polypeptide. Further, it was found that a pentapeptide fragment FGAIL (hIAPP23–27) of the IAPP also formed a fibrillar structure. Similarly, AILSS fragment was also discovered as strong amyloidogenic region of IAPP. Another peptide part, KLVFF of the amyloid β-peptide Ab-42, self-organized in saline buffer forming hydrogel. A pentapeptide sequence in human calcitonin, i.e. DFNKF, also formed the well-ordered amyloid fibrils similar to those observed in case of full length polypeptide. All these observations revealed that peptides which form amyloid fibers have a shorter sequence of amino acids which can also self-organize to form amyloid fibrils similar to those formed by complete peptide. Furthermore, these research studies recommended the significant role of aromatic amino acids in formation of amyloid fibrils. Further research on amyloid like fibrils suggested that α-amino isobutyric acid (U) containing small peptide, i.e. Boc-AUV-OMe, Boc-AUI-OMe, and Boc-AGV-OMe form β-sheet structures which self-organize in amyloid like fibrils (Panda and Chauhan, 2014).

Dipeptides also self-assemble to form nanostructures was demonstrated for the first time by Gazit et al. using dipeptides, FF (Reches and Gazit, 2003). This dipeptide self-assembled into different nanostructures, i.e. nanotubes/microtubes, nanowires, and nanoforests (Reches and Gazit, 2003; Marchesan et al., 2015). The nanotubes formed by FF dipeptide were thermally stable. They further demonstrated that an incorporation of – SH group in FF resulted in transition from tubular to spherical structures (Reches and Gazit, 2004). There are reports in literature that hydrophobic dipeptides, LL, LF, FL, and IL selfassemble to nanotubes via head to tail (NH<sup>3</sup> <sup>±</sup>OOC) H-bond formation (Panda and Chauhan, 2014). Further reports exist in the literature, which demonstrate that VA, LS, and FF can also form nanoporous structures. The amine and carboxyl terminal modified dipeptides also form self-assembled nanostructures. Fmoc-FF dipeptide form hydrogels with a nanofibrillar structure in aqueous conditions and physical attributes of these hydrogels were found to be better than the hydrogels formed by longer polypeptides. The same research group further evaluated the effect of modification of –NH<sup>2</sup> and –COOH terminals of FF dipeptides on self-assembly behavior and found that N and C terminal modified dipeptides also form self organized structures in nano-range. Furthermore, the hydrogels were formed from Fmoc protected dipeptides using a combination of glycine, alanine, leucine, and phenylalanine. The physical and structural features of these hydrogels were different, depending on the characteristic of amino acids used in dipeptide sequences. There are numerous reports in literature which suggest that aromatic moieties such as Fmoc and pyrene protected peptides form nanofibrillar hydrogel network due to π-π stacking and hydrophobic interactions. Unsaturated amino acids such as dehydrophenylalanine have also been used to form selfassembled nanostructures (Panda and Chauhan, 2014). The

introduction of dehydrophenylalanine provides conformational constrain and proteolytic stability to peptides. The extensive studies on dehydrophenylalanine (1F) containing dipeptides have been done by Chauhan et al. where the 1F was used as C-terminal amino acid and N-terminal amino acid residue was varied with any one of the natural amino acids (Gupta et al., 2007). They found that dipeptides with aromatic amino acid at N-terminus formed nanotubes while the dipeptides having charged amino acid at N-terminus formed vesicles. They also revealed that these dipeptides having hydrophobic groups at their N-terminus give rise to self-assembled structures which can be seen from human eye while the structures formed with hydrophilic N-termini were invisible to naked human eye (Panda and Chauhan, 2014). The dipeptide, F-1F, assembled into hydrogels at pH 7.0. The hydrogel formed from F-1F dipeptides efficiently entrapped and released drugs, antibiotics, and vitamins. The kinetics of drug release was affected by change in pH and ion concentration (external stimuli). Thus, this system was used as a controlled self-regulated drug release system (Panda et al., 2008). Among the tested dipeptides having charged amino acids at N-termini, E1F, K1F, R1F, and D1F, the dipeptide, R1F formed vesicular structures and was easily functionalized with folic acid to target folic acid receptors. These nanostructures showed enhanced cellular uptake in various cancer cell lines, like MDA-MB-231 and HeLa. These folic acid functionalized R1F nanostructures also encapsulated doxorubicin efficiently. These doxorubicin-loaded nanostructures showed enhanced cytotoxicity toward cancer cells. These nanostructures further showed enhanced targeting and accumulation in tumor tissue in Ehrlich ascitic tumor bearing mice.

In yet another study, Mahato et al. (2012) prepared selfassembled nanostructures in aqueous environment from small glycolated dehydropeptides, Boc-F-1F-εAhx-GA and H-F-1F-εAhx-GA, wherein glucosamine was attached to peptides via 6-aminohexanoic acid linker. These peptides efficiently entrapped the hydrophobic dyes, eosin and N-fluoresceinyl-2 aminoethanol (FAE), in their core (TEM images in **Figures 4a,b**). At higher concentration, Boc-F-1F-εAhx-GA formed hydrogel also. Likewise, Yadav et al. synthesized Boc-F-1F-εAhx-Neomycin by conjugating Boc-F-1F-εAhx-OH with an aminoglycoside, neomycin, which on self-assembly in aqueous environment formed nanostructures having hydrophobic core and cationic hydrophilic shell. These cationic nanostructures efficiently interacted with pDNA and showed enhanced transfection efficiency in mammalian cells in vitro. Besides, these nanostructures efficiently entrapped hydrophobic molecules, eosin and curcumin, in the core of nanostructures which were characterized by electron microscopic imaging (TEM images, **Figures 4c,d**) (Yadav et al., 2014a). Later on, the same research group fabricated a tripeptide, Boc-P-F-G-OMe, which on self-assembly yielded nanostructures and acted as drug carrier by efficiently encapsulating hydrophobic drug molecules such as eosin, aspirin, and curcumin (**Figures 4e,f**) (Yadav et al., 2015). This group further synthesized a dehydropeptide, Boc-P-1F-G-OMe containing dehydrophenylalanine, an unnatural amino acid, to check the effect of dehydrophenylalanine on the formation of self-assembled nanostructures. The incorporation of dehydrophe instead of Phe improved the encapsulation efficiency of hydrophobic drug, curcumin, in these nanostructures (**Figures 4g,h**). These nanostructures were further stabilized with vitE-TPGS. These nanostructures showed that incorporation of constrained dehydro amino acid in peptides resulted in the construction of stable nanostructures for the development of nanomaterials with controlled drug release (Deka et al., 2017).

Stimuli-responsive peptide nanostructures have, recently, attracted the attention of the researchers as controlled drug delivery vehicles since these are capable of releasing the drug at the intended sites (Panda et al., 2008). The therapeutic moleculeencapsulated peptide nanostructures can be made to release the therapeutics on exposure to stimuli such as change in pH, temperature, and others. There is a report in the literature where anti-inflammatory prodrug, olsalazine, has been conjugated with a tripeptide derivative and this conjugate self-assembled in water to form supramolecular hydrogels (Li et al., 2010). Moreover, the release of 5-aminosalicylic acid from these hydrogels was made to occur in an organized way, attained by reducing the azo bond which resulted in disassembly of the hydrogels. Moreover, the nanovesicles developed from dipeptides using glutamic acid at C-terminus showed constant behavior over a range of pH. However, these vesicles were responsive to Ca2<sup>+</sup> ions. In these nanovesicles, anticancer drugs, fluorescent dyes, and various biologically active molecules were entrapped and released in response to calcium ions (Naskar et al., 2011). The peptide nanostructures formed from peptide, Acp-YE (Acp, 3-amino caproic acid), showed stimuli responsive behavior to Ca2<sup>+</sup> ions concentration and change in pH. The release of encapsulated anticancer drug (doxorubicin) from these vesicular structures was achieved on exposure to Ca2<sup>+</sup> ions concentration.

Enormous literature exists on self-assembled peptide nanostructures useful for drug delivery applications; however, most of them are under in vitro studies. Only few in vivo studies have been undertaken and some of them have been listed in **Table 3** (Leite et al., 2015; Habibi et al., 2016; Lee et al., 2019).

#### **DNA nanotechnology-based drug delivery**

DNA nanotechnology involves assembly of synthetic DNA fragments into self-assembled nanostructures of different sizes, shapes, and morphology. The basic principle behind DNA nanotechnology is the complementary base pairing. Using this principle, a large number of simple DNA nanostructures have been produced (Hu et al., 2018; Madhanagopal et al., 2018; Kim et al., 2019). Initially, the research on DNA nanotechnology was initiated with the formation of simple topological morphologies which later on advanced to production of DNA tiles, periodic lattices, and 2D and 3D structures. Using this superamolecular DNA technology, wherein interactions such as van der Waals, pi-pi stacking, H-bonding, metal-coordination etc., are involved in DNA self-assembly, different molecules have been encapsulated (Sharma et al., 2018; Ariga et al., 2019). In the last decade, the advent of DNA origami in 2006 by Rothemund's group expanded the research on DNA nanotechnology from 2D to 3D confirmation forming complex 3D nanostructures of diverse shape and design. In DNA origami, a large natural singlestranded DNA is folded with the help of several chemically


TABLE 3 | List of peptide-based self-assembled nanostructures for drug delivery applications.

synthesized short DNA strands called staple strands to form well-defined 2D and 3D nanostructures. The different types of DNA nanostructures can be fabricated by varying the number and length of staple strands as well as by changing the relative sequence of nucleotides in individual strand. These small staple strands induce the folding of larger DNA strand by annealing with it. The complementary base pairing interaction of DNA scaffold with staple strands help in self-assembling in well defined nanoarchitectures (Hu et al., 2018; Madhanagopal et al., 2018; Sharma et al., 2018).

The self-assembled DNA nanostructures have been used in various biomedical applications such as drug delivery, gene delivery, biosensing, etc. (Ariga et al., 2019). The design and the strategy used to construct these DNA nanoarchitechtonics depend on the type of application these nanostructures are required. The DNA nanostructures, formed via self-assembly approach, are mostly based on sticky end cohesion of DNA strands. The sticky ends are unpaired nucleotide overhangs at the end of DNA molecules. These overhangs are mostly palindromic sequences. The sticky ends are used to combine DNA nanostructures via hybridization of their complementary single strands. Initially, the polyhedral DNA nanostructures were formed by this approach. Subsequently, periodic lattices were formed by tile-based self-assembly approach. With the

dawn of DNA origami approach, a lots of 2D and 3D objects were created. DNA origami approach is successfully used to create large nanostructures compared to tile-based approach, as in DNA origami, thousands of nucleotide long scaffold DNA strand is employed. Another strategy has also been used for DNA nanostructures in which single-stranded DNA tiles containing four domains are used to create DNA nanostructures and adjacent tiles bind with complementary parts forming DNA lattices composed of parallel DNA helix (Madhanagopal et al., 2018).

Construction of customized DNA nanostructures is driven by the type of therapy and therapeutic molecules to be delivered. Various types of therapeutic molecules can be delivered, e.g. drug molecules, fluorescent dyes, protein molecules, siRNA, miRNA, CpG sequences, mAB, etc. Fluorescent dyes, viz., fluorescein, cyanine, and rhodamine, have been tagged with self-assembled DNA nanostructures and delivered to cells for different cellular analysis (Hu et al., 2018; Lacroix et al., 2019). Ding et al. have used DNA origami to synthesize 2D DNA triangle and 3D DNA tubes to load anticancer drug, doxorubicin. These DNA origami nanostructures showed enhanced cellular uptake in adenocarcinoma cell lines, MCF-7 and Dox-resistant MCF-7 cells (Jiang et al., 2012; Zhao et al., 2012; Zhang et al., 2014; Mei et al., 2015; Li et al., 2017).

DNA nanostructures have also been used to deliver oligonucleotides, i.e. CpG dinucleotides as vaccine adjuvants for immunotherapy of infectious diseases (Zhang et al., 2015; Wang et al., 2016; Qu et al., 2017; Hu et al., 2018). The CpG motifs are present in bacterial genome and they are recognized as foreign molecules by vertebrate immune system. So these DNA motifs have been used to trigger host immune response as these are recognized by TLR-9 located on endosomes of host membrane of immune system which activates the innate immune pathway of host immune system. Similarly, siRNA and miRNA delivery have also been carried out using DNA nanostructures for gene silencing applications (Lee et al., 2012; Fakhoury et al., 2013; Bujold et al., 2016; Roh et al., 2016; Qian et al., 2017; Nahar et al., 2018). Various types of DNA nanostructures, that have been used as delivery vehicles, are listed in **Table 4** (Hu et al., 2018; Madhanagopal et al., 2018). Some of the recent reviews on DNA nanotechnology have described in detail the applications of DNA nanotechnology in drug delivery. Despite enormous advantages of DNA-based nanostructures, their stability in vivo is an issue as they are sensitive to cellular environment as well as salt concentration. Moreover, the high cost involved in the synthesis of DNA hampers their large-scale applications in biomedical field (Linko et al., 2015). In an attempt to address this concern, Praetorius et al. (2017) have presented a method to knock down the price of DNA nanostructure synthesis using biotechnological mass production. Although this method is not currently available in every lab, it is expected, in near future, that this cost-effective protocol would overcome this obstacle expanding the scope of DNA nanotechnology in other branches of science and technology.

# CONCLUSION AND FUTURE PERSPECTIVES

In the last 20 years, tremendous developments have been made in the area of self-assembly of bioactive molecules. Post self-assembly, the nanostructure-based materials are potentially useful and have offered newer tools to revolutionize the area of biological and biomedical sciences. Nanotechnology has significantly contributed toward the realization of targeted and controlled delivery of therapeutics. For their delivery, different types of materials/systems have been developed. Barring a few, many of these materials have their own merits and demerits. Certain materials have been claimed to exhibit biocompatibility but the others that have been developed and are being used showing toxicity and hence proved inappropriate for in vivo applications. For example, cationic lipid-based nanostructures are found to activate the immune system. Besides, these are also associated with some technological issues such as stability, reproducibility, low drug loading, encapsulation, and uncontrolled drug leaching problems. Polymeric systems were then developed and evaluated but they were also associated with the similar types of limitations and hence surface functionalization was thought of to improve drug or gene-


targeting, which is usually complicated. Similarly, natural polymers elicited unwanted immune reactions and also showed a batch to batch inconsistency, thus in vivo performance of these polymers became complex and questionable. Peptides and small molecule-based nanostructures can be good alternatives as carriers for therapeutic delivery as they possess certain characteristics such as good biocompatibility, ease of synthesis, and functionalization. Their self-assembled nanostructures present numerous prospective applications in biomedical field. Beside this, easy stimuli-responsiveness (internal/external stimuli) of self-assembled small molecules makes their role vital in the advancement of therapeutics delivery systems, where the therapeutic release behavior can be better controlled according to the requirements. Thus mild and rapid synthesis conditions, easy dispersibility in aqueous medium, simple functionalization, low production cost, and non-requirement of specialized equipments are some of the advantages which have advocated their promising potential to be used as future candidates for applications such as in drug/gene delivery, diagnosis, imaging, sensors, tissue engineering, bioelectronics, production of biomaterials, healthcare-related systems, etc. Various types of structures can be generated simply by varying the conditions. Thus, this area has emerged as a newer area of research which has shown promising potential. However, there exist several challenges which still need to be addressed in order to make them materials of choice for researchers. Although self-assembly results in the generation of various types of structures such as nanotubes, nanoparticles, nanospheres, nanotapes, nanofibers, nanogels, nanorods, etc., controlling the size of these structures during processing, their behavior under aqueous environment, degree of loading/entrapment of therapeutics and stability, as well as upscaling are still the gray areas where sincere attention of the researchers is required. Besides, studies to establish the biocompatibility and immunogenicity of these nanostructures are lacking.

#### REFERENCES


Self-assembled DNA-origami nanostructure-based drug delivery offers a newer area which has shown tremendous potential in cancer treatment. These structures have been shown to possess stability in cell lysates upto 12 h, while, on prolonged exposures, degradation begins to occur. To improve their stability, several modifications have been suggested. Likewise, optimization in size and shape of these nanostructures reveals their effectiveness during drug release. Because of their multifunctional nature, easy amenability to modifications, biodegradability, as well as biocompatibility, these systems can be developed as safe and efficient drug delivery vectors. However, translation from bench to bedside applications, some crucial aspects are still required to examine in detail such as stability of these nanostructures under different conditions, their efficacy in different types of diseases, comparison of their performance with the commercially available formulations, systemic clearance, morphological parameters during their interactions with the different types of cells, effect of surface charge on their stability during circulation, etc. These investigations are required to ascertain that these systems will provide fascinating and promising solutions to improve the area of human healthcare.

### AUTHOR CONTRIBUTIONS

SY wrote the manuscript with guidance from AS and PK. PK provided the critical feedback and helped in shaping the manuscript in current form. AS and PK supervised the manuscript.

# FUNDING

SY is thankful to CSIR for providing Research Associateship (RA) vide reference File No. 31/43(349)/2017-EMR1 to carry out this work.



peptide scaffolds. Proc. Natl. Acad. Sci. U.S.A. 97, 6728–6733. doi: 10.1073/pnas. 97.12.6728


multidrug resistance in targeted anticancer drug delivery. Nano Res. 8, 3447– 3460. doi: 10.1007/s12274-015-0841-8




**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 Yadav, 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.

# How the Lack of Chitosan Characterization Precludes Implementation of the Safe-by-Design Concept

Cíntia Marques1,2, Claudia Som<sup>3</sup> , Mélanie Schmutz<sup>3</sup> , Olga Borges2,4 and Gerrit Borchard<sup>1</sup> \*

1 Institute of Pharmaceutical Sciences of Western Switzerland, University of Geneva, Geneva, Switzerland, <sup>2</sup> Faculty of Pharmacy, University of Coimbra, Coimbra, Portugal, <sup>3</sup> Empa, Swiss Federal Laboratories for Materials Science and Technology, Technology and Society Laboratory, St. Gallen, Switzerland, <sup>4</sup> Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal

#### Edited by:

Qingxin Mu, University of Washington, United States

#### Reviewed by:

Giulia Suarato, Italian Institute of Technology, Italy Gaoxing Su, Nantong University, China

\*Correspondence:

Gerrit Borchard gerrit.borchard@unige.ch

#### Specialty section:

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

Received: 03 October 2019 Accepted: 18 February 2020 Published: 10 March 2020

#### Citation:

Marques C, Som C, Schmutz M, Borges O and Borchard G (2020) How the Lack of Chitosan Characterization Precludes Implementation of the Safe-by-Design Concept. Front. Bioeng. Biotechnol. 8:165. doi: 10.3389/fbioe.2020.00165 Efficacy and safety of nanomedicines based on polymeric (bio)materials will benefit from a rational implementation of a Safe-by-Design (SbD) approach throughout their development. In order to achieve this goal, however, a standardization of preparation and characterization methods and their accurate reporting is needed. Focusing on the example of chitosan, a biopolymer derived from chitin and frequently used in drug and vaccine delivery vector preparation, this review discusses the challenges still to be met and overcome prior to a successful implementation of the SbD approach to the preparation of chitosan-based protein drug delivery systems.

Keywords: safe by design, polymeric drug carriers, chitosan, insulin, protein drug delivery

#### INTRODUCTION

Nanoparticles (NPs) have been extensively investigated as delivery systems for targeted drug delivery, controlled drug release, in vivo imaging, diagnostics, and medical devices. These systems may offer more convenient routes of administration, decrease drug toxicity, and potentially reduce healthcare costs (Vasile, 2019). However, despite numerous publications on nanoparticulate drug carrier systems ("nanomedicines"), the extent of their translation into clinical application has been unsatisfactory (Hua et al., 2018; Rosenblum et al., 2018). The first generation of these nanomedicines passed regulatory approval by meeting standards in place for "conventional" drugs of low molecular weight. However, with regard to the complexity of nanomedicines, these standards were reviewed and partially replaced by nano-specific critical quality attributes (CQAs) that need to be reported in order to confirm quality, safety, and efficacy of NPs (Gaspar, 2007; U. S. Food and Drug Administration, 2017). Quality control assays for nanomaterial characterization, the need of establishing specialized toxicology studies for nanomedicines, and the lack of suitable standards and dedicated regulatory guidelines are a few examples of the challenges to their development and effective clinical translation (Hua et al., 2018).

The research community is working to establish protocols for nanomaterial characterization (Brown et al., 2010). The Nanotechnology Regulatory Science Research Plan, established by the Food and Drug Administration (FDA), addresses five major criteria, namely, physicochemical characterization, pre-clinical models, risk characterization, risk assessment, and risk communication (Rosenblum et al., 2018). In this regard, the Nanotechnology Characterization Laboratory (US NCL) was founded, focusing on the characterization of nanomedicines for cancer therapy. In Europe, the European Nanomedicine Characterization Laboratory (EU-NCL) was created as a multi-national organization within the H2020 framework. EU-NCL focuses on the pre-clinical characterization of nanomaterials in order to accelerate their development toward the approval by the regulatory agencies (European Nanomedicine Characterization Laboratory, 2019a). Moreover, in the European Union, other projects such as NANoREG, NANoREG II, ProSafe, and NanoDefine have also explored the standardization of nanomaterial characterization, and the development of better prediction models, such as the application of the Safe-by-Design (SbD) approach to nanosystems (Kraegeloh et al., 2018).

The principle behind SbD includes the safety assessment of nanomedicines as early as possible in their innovation process and throughout their lifecycle by designing out the physicochemical properties with an adverse effect on human health and the environment (Bottero et al., 2017; Soeteman-Hernandez et al., 2019). Several concepts of SbD have arisen from the European projects mentioned above. For example, the NANoREG project describes three pillars: safe product by design, safe use of products and safe industrial production. In addition, according to NANoREG II, the SbD concept aims at the development of functional and safer nanomaterials, safer processes as well as safer products. In general, the application of this concept requires the examination of which physicochemical properties render a nanomaterial safe, means to implement this knowledge into industrial innovation processes, and information exchange between stakeholders. The SbD concept can be implemented to design nanomaterials with an optimal balance between functionality and risk, based on relevant physicochemical parameters (Kraegeloh et al., 2018).

The European project GoNanoBioMat created a SbD approach to support industries, particularly small and mediumsize enterprises (SMEs) to identify risks and uncertainties early in the research and development phase, support safe production and handling, and deliver safe products. The SbD approach is applied to polymeric nanobiomaterials for drug delivery and it focuses on safe nanobiomaterials, safe production and safe storage and transport (Som et al., 2019).

Particularly, one goal of GoNanoBioMat was to establish the characteristics of different types of chitosan nanoparticles (Chit NPs), to establish a correlation between the physicochemical properties of this biopolymer and its immunostimulatory activity and, finally, to establish a guideline to select the most suitable chitosan polymer according to its purpose, allowing an SbD approach. To address these points, an extensive literature search was initiated and will be presented in this report.

Chitosan, the deacetylated form of chitin, is a biopolymer investigated for the preparation of particles as vectors for drug delivery. Chitosan nanoparticles are under investigation for a wide variety of biomedical applications, due to the polysaccharide's exceptional versatility (Koppolu et al., 2014). One of the major applications of chitosan is the preparation of medical micro- and nano-particles. Nanoparticles of natural polymers are a promising approach for drug delivery due to their biocompatibility and biodegradability, as well as for their ability to provide a controlled drug release profile (Erel et al., 2016). Even though chitosan is one of the most studied biopolymers, there is no standardization as far as its properties and the resulting biological activity are concerned.

The goal of this review was to understand whether it is possible to identify physicochemical properties of chitosan that are correlated to its biological effects. To this end, supportive information on protocols used to prepare chitosan NPs encapsulating insulin (Chit-Ins NPs) as a model protein drug were collected. Protocol details and Chit-Ins NPs characterization data were compared. Literature was also examined for available information on the immunotoxicological response to Chit-Ins NPs administration. Finally, the report summarizes the current state of the art, identifies the challenges in applying the SbD concept to the bionanomaterial chitosan and establishes future perspectives on Chit NPs characterization.

### METHODS

A literature search was performed through PubMed and Science Direct using as Medical Subject Headings (MeSH) keywords chitosan, immune activity, gelation, insulin, encapsulation, and adjuvant. We focused on ionotropic gelation, using tripolyphosphate (TPP) as crosslinker because it is the most used process to prepare Chit NPs. Insulin was chosen as a model for protein encapsulation into these nanoparticles.

# CHITOSAN: POTENTIAL AND VERSATILITY

Chitosan is the partially deacetylated form of chitin – a poly (Dglucosamine) – and comprises a wide range of linear polymers differing in polymer length and deacetylation degree. The polymer is composed of randomly distributed β-(1→4)-linked D-glucosamine (deacetylated unit) and N-acetyl-D-glucosamine (acetylated unit) (**Figure 1**) and it appears in the market with different purity degrees (Primex, 2019). Chitin is a natural biopolymer extracted from the exoskeleton of crustaceans (shrimp, crabs, lobsters, etc.) and from the cell walls of fungi or yeast (Illum, 1998; Mukhopadhyay et al., 2013; Vasiliev, 2015; Bugnicourt and Ladavière, 2016; Jafary Omid et al., 2018; Primex, 2019).

In fact, chitosan is one of the most studied biopolymers. This polysaccharide is exceptionally versatile as it can be used in solutions, suspensions, hydrogels and/or micro- and

nanoparticles. Moreover, it is possible to proceed to its chemical functionalization through its amino and hydroxyl groups, and/or by conjugation of peptides and other molecules to the polymer backbone. This allows for the modification of physicochemical properties and/or the introduction of desirable characteristics, further broadening chitosan potential applications (Sreekumar et al., 2018).

Chitosan is well known for its inherent biological properties, namely biocompatibility (Hirano et al., 1990), non-toxicity (Hu et al., 2011; Pradines et al., 2015), antimicrobial activity (Zheng and Zhu, 2003; Qin et al., 2006; Cerchiara et al., 2015), plant strengthening (Choudhary et al., 2017), hydrating ability (Cerchiara et al., 2015), gel and film forming (Shan et al., 2010; Nie et al., 2016), mucoadhesive properties (Cerchiara et al., 2015; Patel et al., 2015), immunostimulant activity (Nishimura et al., 1985; Scherliess et al., 2013), hemocompatibility (Malette et al., 1983; Lee et al., 1995; Zhao et al., 2011), and biodegradability (Lee et al., 1995; Patel et al., 2015).

This polymer is one of the most widely used for biomedical applications. Actually, chitosan has been under investigation for drug and vaccine delivery (Borges et al., 2008; Esmaeili et al., 2010; Jafary Omid et al., 2018; Soares et al., 2018; Bento et al., 2019), gene delivery (Thanou et al., 2002), surgical sutures (Muzzarelli et al., 1993; Altinel et al., 2018), rebuilding of bone (Lee et al., 2009), corneal contact lenses (Silva et al., 2016), dental implants (Yokoyama et al., 2002), wound healing (Mizuno et al., 2003), antimicrobial applications (Dai et al., 2009), and tissue engineering (Madihally and Matthew, 1999; Kanimozhi et al., 2016). Moreover, chitosan has been used as a dietary supplement in preparations for treatment of obesity and hypercholesterolemia (Bokura and Kobayashi, 2003; Zhang et al., 2008) and also in medical devices for the treatment and control of bleeding (Millner et al., 2009). The polysaccharide is classified by FDA as Generally Recognized As Safe (GRAS) for food (Nutrition Center for Food Safety Applied, 2019a,b). The polymer description was first introduced into the European Pharmacopeia 6.0 and the 29th edition of the United States Pharmacopeia (USP) 34-NF. Monographs contain the assays and establish limits to be observed when the polymer is used as a pharmaceutical excipient (Council of Europe, 2019). Currently the efficacy of chitosan nanoparticles in the treatment of postoperative pain and antibacterial activity against Enterococcus faecalis in infected root canals is being studied in a phase 2 clinical trial (U. S National Library of Medicine, 2018).

# CHALLENGES FOR SAFE-BY-DESIGN OF CHIT-NPs

#### Characterization of Chitosan Is Not Standardized

Despite the large number of papers about chitosan, reproducibility of the reported results is often an issue (Nasti et al., 2009). As mentioned above, chitosan is a family of polymers, which differ in their degree of deacetylation (DD), molecular weight (MW) and purity. The different characteristics can be correlated with the diversity of physicochemical properties and diverse biological activities of the polysaccharide. As a matter of fact, these structural characteristics are dependent on the source of chitin, its extraction, and the deacetylation method (Bellich et al., 2016), whose correlations with chitosan biological properties has been reviewed elsewhere (Younes and Rinaudo, 2015).

As illustrated in **Table 1**, chitosan basic characterization is neglected in many papers making it difficult to critically comment on conflicting experimental results (Vasiliev, 2015; Bellich et al., 2016). Even when the MW is provided, there is often an ambiguous classification. For example, Mehrabi et al. (2018) classify chitosan into high molecular weight (HMW) at the range of 700–1,000 kDa, low molecular weight (LMW) when less than 150 kDa, and medium molecular weight (MMW) between low and high molecular weight. On the other hand, Vila et al. (2004) mention chitosan of 23 and 38 kDa as LMW and chitosan of 70 kDa as HMW.

Moreover, Vasiliev (2015) pointed out the importance of method harmonization and validation to chitosan analysis, such as size exclusion chromatography (SEC) to determine MW, capillary viscosimetry to check for viscosity, nuclear magnetic resonance (NMR) to define the degree of deacetylation (DD), and Limulus amebocyte lysate (LAL) test to verify endotoxin content.

Other authors go even deeper with respect to chitosan characterization. Even knowing that patterns of acetylation (PA) – random, alternating or blockwise – are linked to different polymer functionalities, such as polymer-solvent interactions (Bellich et al., 2016; Wattjes et al., 2019) and biological activity (enzyme recognition) (Weinhold et al., 2009), it is not usually taken into consideration in papers on chitosan characterization. In fact, studies have shown that chitosan with the same DD can have different solubility properties due to different patterns of distribution of its monomers N-glucosamine and N-acetylglucosamine (Bellich et al., 2016). Because commercially available chitosan is produced by chemical deacetylation of chitin under heterogeneous conditions (Wattjes et al., 2019), it usually results in heterogeneous products with random patterns of acetylation (Varum et al., 1991; Weinhold et al., 2009). Enzymatic deacetylation is an interesting alternative to chitosan preparation as the application of chitin deacetylases allows for a controlled process, resulting in a polysaccharide with well-defined patterns of acetylation (Tsigos et al., 2000).

Despite different opinions, the accurate determination of chitosan properties should be unavoidable (Bellich et al., 2016). MW, DD, viscosity and purity should be presented as

TABLE 1| Summary of Chit-Ins NPs production protocols by ionotropic-gelation method.


(Continued)

Safe-by-Design Chitosan Characterization

TABLE 1| Continued


Cs NPs, chitosan nanoparticles; TPP, tripolyphosphate; AE, association efficiency; CAM assay, Chick Chorioallantoic Membrane assay; XTT assay, Cell Viability Assay. Units were converted to standardization, so they can differ from the ones at the original paper. ? refers to data that could not be confirmed in the respective publication and thus remain unknown.

Safe-by-Design Chitosan Characterization

fbioe-08-00165 March 10, 2020 Time: 12:59 # 5

chitosan characterization parameters. Moreover, it is known that the properties discussed above will influence Chit NP physicochemical properties such as size and zeta potential, but also determine its biological activity. It is therefore essential to define the properties of chitosan in order to assure the reproducibility of Chit NP preparation (Hua et al., 2018) and to obtain the desired biological response. Moreover, in order to follow a SbD approach, as mentioned before, it is important to classify with accuracy the physicochemical properties that determine the safety of the nanomaterial.

# Drug Encapsulation Into Chitosan/Tripolyphosphate Nanoparticles (Chit-TPP NPs): Insulin as Case-Study

Chit NPs can be prepared through numerous methods. Among them, ionotropic gelation is based on the electrostatic interactions between charged polymers and non-toxic anionic cross-linking agent species, such as citrate, sulfate, or TPP. Ionotropic gelation is performed in aqueous media, avoiding organic solvents, high temperatures, and high shear rates. Because of that, it is a safe preparation method resulting in low-toxicity NPs (Dash et al., 2011; Bugnicourt and Ladavière, 2016). In the case of Chit NPs preparation, the convenient characteristics of ionotropic gelation along with the cationic sites available all along the polymer chain of chitosan allow the interaction and encapsulation of fragile poly-anionic molecules, such as proteins and deoxyribonucleic acids (DNA), producing stable colloidal complexes (Xu and Du, 2003; Bugnicourt and Ladavière, 2016).

Chit NP production, particularly by using TPP as a crosslinker, is a generally established method and it is by far the most mentioned in the literature. Usually, the preparation of Chit-Ins NPs by ionotropic gelation consists in dissolving the polysaccharide in an aqueous acetic acid solution, while TPP is dissolved in deionized water. Then, TPP solution is added dropwise to the chitosan solution under stirring (magnetic stirring or using a high-speed homogenizer), leading to the spontaneous formation of Chit NPs (Calvo et al., 1997).

There are many different protocols for insulin encapsulation into Chit NPs (**Figure 2**). Insulin can be pre-dissolved in diluted hydrogen chloride (HCl) solution (Abbad et al., 2015), the pH of this final solution can be adjusted with sodium hydroxide

(NaOH) (Hecq et al., 2015; Li et al., 2017), or insulin can even be directly solubilized into diluted NaOH, or directly into TPP solution (Zhao et al., 2014). Then, the insulin solution is added to the chitosan solution right before or during TPP addition (Ma et al., 2005; Azevedo et al., 2011; Makhlof et al., 2011) or added after TPP addition to chitosan (Ma et al., 2002; Erel et al., 2016). Nanoparticles form spontaneously, the system stays under stirring for a while in order to stabilize the nanoparticles.

Despite similar formulation and preparation procedures, different properties of the resulting insulin-loaded chitosan nanoparticles (Chit-Ins NPs) have been reported (Ma et al., 2002), as shown in **Table 1**. Factors such as chitosan and TPP concentrations, pH, chitosan origin and its characteristics, rotation speed, insulin concentration, among others, greatly influence the final nanoparticle properties, thus having a serious impact on batch reproducibility and bioactivity (Ma et al., 2002; Sreekumar et al., 2018).

Note that the systems listed in **Table 1** were developed mainly for the oral administration of insulin. This protein is highly susceptible to enzymatic degradation in the gastrointestinal (GI) tract, thus nanoparticles may aid to protect it from the acidic environment and enzymatic degradation, and to promote insulin absorption by using mucoadhesive polymers, such as chitosan (Makhlof et al., 2011; Al Rubeaan et al., 2016). Despite their well-known potential, Chit-TPP NPs are not stable under acidic conditions, as the protonation of the amino groups of chitosan at low pH values promotes their dissolution and successive insulin degradation, decreasing its bioavailability (Al Rubeaan et al., 2016).

In order to increase nanocarrier stability in the gastric environment recent delivery systems have been developed based on modified chitosan through conjugation, quaternization, thiolation, substitution, and grafting (Chaudhury and Das, 2011; Al Rubeaan et al., 2016). For example, permanently positively charged N-(2-hydroxy)propyl-3-trimethyl ammonium chitosan chloride (HTCC), increases Chit-Ins NPs stability (Hecq et al., 2015). Another derivative, thiomalyl chitosan, produces negatively charged NPs that, curiously, seem to enhance mucoadhesion and permeation, when compared to Chit NPs. This system is also suggested to inhibit insulin degradation due to its protease inhibitory effect (Rekha and Sharma, 2015).

Moreover, hydroxypropyl methylcellulose phthalate used as crosslinker (instead of TPP) in Chit-Ins NPs preparation also proved to increase NP stability and, additionally, to improve intestinal mucoadhesion and penetration (Makhlof et al., 2011). Finally, another interesting approach is an emulsion-based delivery system, where Chit-Ins NPs were suspended in a microemulsion, successfully protecting insulin under gastric conditions and reducing blood glucose levels for 8 h after oral administration (Erel et al., 2016).

As can be extracted from **Table 1**, depending on the preparation method, reported NP size may range between 112 and more than 400 nm. Zeta potential, when measured, was also highly variable with values ranging between 20 and 40 mV. Even more variable was the encapsulation efficiency for insulin reported, with values ranging from as little as 2% to almost 90%. Overall it can be said that generally not all relevant data

on the materials and methods used were reported, rendering the selection of the optimal preparation method from the literature difficult.

### Chitosan as an Immunostimulant: An Additional Source of Disagreement

As mentioned before, chitosan is known for its immunostimulatory activity. Because of that, the polysaccharide has been extensively studied and reviewed as an adjuvant and/or as a delivery system for vaccines (Van der Lubben et al., 2001a; Ghendon et al., 2009; Esmaeili et al., 2010; Mehrabi et al., 2018).

Establishing the physicochemical properties that are correlated with chitosan immune stimulation is important to define Chit NPs activity in view of a SbD approach. However, as for other data available for chitosan, reports on its immunomodulation activity are contradictory. Some publications claim that chitosan is not able to stimulate antibody production (de Geus et al., 2011), while other studies confirm that chitosan can only induce immunostimulation due to the synergic effect between the components of the chitosan formulation and the antigen (Seferian and Martinez, 2000; Bivas-Benita et al., 2004). In addition, many articles claim the obvious adjuvant potential of the polysaccharide (Nishimura et al., 1985; Zaharoff et al., 2007; Ghendon et al., 2009; Esmaeili et al., 2010; Dzung et al., 2011; Vasiliev, 2015; El Temsahy et al., 2016).

The adjuvant activity of chitosan was first attributed to its mucoadhesive properties, which prolong the residence time of the loaded antigen at mucosal sites. This, in turn, increases antigenic uptake (Illum, 1998; Alpar et al., 2005) and improves immunological response via transmucosal routes (Illum, 1998): nasal (Van der Lubben et al., 2001b; Esmaeili et al., 2010), pulmonary (Esmaeili et al., 2010), and oral (Van der Lubben et al., 2001b; Borges et al., 2006, 2007; Esmaeili et al., 2010). Furthermore, the physical association of chitosan with an antigen (Calvo et al., 1997; Seferian and Martinez, 2000) and its slow release are very important to the overall adjuvant activity of the biopolymer (Calvo et al., 1997).

Other authors explored the potential of chitosan immune stimulation through the parenteral route (Borges et al., 2008), based on preliminary data that attributed adjuvant activity to chitin derivatives, including chitosan. These biopolymers increased immune response in guinea pigs after immunization applied to their footpads (Nishimura et al., 1985). Zaharoff et al. (2007) vaccinated mice with β-galactosidase dissolved in a viscous chitosan solution. The adjuvant activity was attributed to the combination of an antigen depot with the stimulation of both humoral and cell-mediated immune responses (Zaharoff et al., 2007). Correspondingly, Ghendon et al. (2009) explored the properties of chitosan as an adjuvant for inactivated influenza vaccines, showing that the polysaccharide induced the production of high titers of antibodies against the antigen and increased cytotoxic activity of NK-cells. Furthermore, Chit NPs are known to induce mixed Th1/Th2 responses with a great variability of antigens. An increase of interferon-G (IFN-G) and IgG2a is characteristic for a Th1 response, while the Th2 pathway is elicited by IL-4 and IgG1 production (Zaharoff et al., 2007; Borges et al., 2008). Additionally, Chit NPs interact with antigenpresenting cells (APCs), such as macrophages, and induce CD4<sup>+</sup> T cell proliferation (Zaharoff et al., 2007). In case of mucosal administration, an increased production of sIgA has been shown (Vila et al., 2004; Borges et al., 2007).

Recently, Chit NPs prepared by ionotropic gelation have been tested as an adjuvant in several vaccine systems (Vila et al., 2004; Danesh-Bahreini et al., 2011; Dzung et al., 2011; El Temsahy et al., 2016). For example, El Temsahy et al. (2016) produced Toxoplasma lysate vaccines by encapsulating virulent RH and avirulent Me49 Toxoplasma strains into Chit NPs, while Danesh-Bahreini et al. (2011) applied the Chit-TPP system to develop a leishmaniasis vaccine. In the first example, the Toxoplasma lysate vaccines were injected by the intraperitoneal route into mice, stimulating both humoral and cellular immune responses (El Temsahy et al., 2016). Furthermore, the Chit-TPP-antigen system was shown to be as effective as Freund's incomplete adjuvant (FIA) in enhancing the efficacy of Toxoplasma vaccine (El Temsahy et al., 2016). The reported data are in agreement with other studies comparing the polysaccharide with commonly used adjuvants, FIA and aluminum hydroxide, demonstrating the biopolymer to be equipotent to those adjuvants (Zaharoff et al., 2007; Dzung et al., 2011). Chit NPs where also loaded with Leishmania superoxide dismutase (SODB1), and injected into BALB/c mice, eliciting both IgG2a1 and IgG1 production (Danesh-Bahreini et al., 2011). Therefore, chitosan is an alternative to traditional adjuvants applied in vaccine development (Zaharoff et al., 2007; El Temsahy et al., 2016).

In general, immune responses depend on the system's physicochemical characteristics, properties and dose of antigen (Amidi et al., 2010). Furthermore, polysaccharide features appear to influence the elicited response. Chitosan from different sources and suppliers, of different DD (Nishimura et al., 1985; Scherliess et al., 2013) and MW (Ghendon et al., 2009; Dzung et al., 2011; Scherliess et al., 2013) have been used to explore its immunostimulant activity. Nishimura et al. (1985) observed a correlation between the immunological activity and chitosan DD, in which 70% DD was the optimal value, whereas 30% DD resulted in lower adjuvanticity. This appears to be in agreement with data showing that positively charged particles are associated with increased immunogenicity (Foged et al., 2005). However, recent reports also showed that chitosan with 76% DD elicited higher immune responses than 81% DD chitosan (Scherliess et al., 2013).

Data is also contradictory with respect to the influence of MW on chitosan immunostimulant activity. While some authors claim that LMW chitosan (10 kDa) is more effective in immune system stimulation than HMW chitosan (300 kDa) (Ghendon et al., 2009), others show that MW around 300 kDa has a greater effect than LMW chitosan (Dzung et al., 2011). Moreover, another paper stated that MW had no significant impact on Chit NPs stimulated immune response (Vila et al., 2004). Note that the last classification of LMW and HMW was based on Ghendon et al. (2009).

The contradictory information suggests that the chitosan formulation can also affect its adjuvant action (Scherliess et al., 2013). In case of chitosan particulate systems, the preparation

technique has a direct influence on the particle size, which also influences the triggered immune pathway (Bueter et al., 2011; Scherliess et al., 2013; Soares and Borges, 2018). Note that the particle size also depends on chitosan MW and DD (Scherliess et al., 2013). Moreover, the antigen release pattern from the chitosan system and the injection site seem to affect the immune response, as well (Vila et al., 2004; Scherliess et al., 2013).

Furthermore, there is a lack of information on the biopolymer purity, such as the presence of endotoxins, LPS, proteins, nucleic acids and heavy metals, which can have an important influence on the immune response elicited. As a consequence, it has been proposed that the adjuvant activity attributed to chitosan can be related to its impurities and not to the polymer itself (Vasiliev, 2015).

In the end, it is not clear which factor is responsible for the differences in immune responses elicited by the biopolymer. There is most probably an interaction between all the properties mentioned before affecting chitosan adjuvanticity (Scherliess et al., 2013).

# Undesired Adjuvanticity of Chit: Potential Immunotoxicity of Chit-Ins NPs

The adjuvant activity of chitosan has been studied for the purpose of vaccine formulation. That means that the active pharmaceutical ingredient (API) encapsulated is already known to have immunogenic properties, whether the antigen is highly or poorly immunogenic. The great majority of Chit-TPP systems loaded with insulin are studied as an alternative to the subcutaneous administration of insulin. Thus, immunogenic studies are not usually a concern as shown in **Table 1**, which illustrates the lack of information on the immunotoxicological and immunopharmacological profile of Chit-Ins NPs.

Note that mucosal delivery routes—oral, nasal, etc.—studied for insulin administration generally imply absorption through a mucosal surface, where chitosan has also been widely applied as a vaccine adjuvant (Illum, 1998; Van der Lubben et al., 2001b). Insulin is indeed poorly immunogenic (Fineberg et al., 2007). Its formulations for subcutaneous administration have been developed and improved, indicating rare severe immunological complications. Actually, less than 0.1% of recipients experience insulin resistance due to immune reactions (Fineberg et al., 2007). However, insulin resistance due to Chit-Ins NPs administration cannot be totally excluded in the absence of in-depth studies.

Chit NPs adhere to the mucosa and transiently open intercellular tight junctions. Due to the pH variation, these NPs become less stable and disintegrate releasing the insulin, which is absorbed through the paracellular pathway into the systemic circulation (Borchard et al., 1996; Sung et al., 2012). In reality, other transport pathways can be involved after oral administration of Chit-Ins NPs (Abbad et al., 2015), such as transcytosis through enterocytes, receptor-mediated transcytosis, and transcellular absorption by M cells in the Peyer's patches. As part of the gut associated lymphoid tissue (GALT), Peyer's patches have an important role in eliciting immune responses against oral antigens, as reviewed elsewhere (Soares and Borges, 2018). However, since absorption studies do not use models that include enterocytes, goblet, and M cells simultaneously, the insulin absorption pathway is still unknown (Abbad et al., 2015). Furthermore, these studies showed NP uptake by epithelial cells, but did not prove their transport across those cells. Thus, there is a risk of intercellular degradation of the NPs (Amidi et al., 2010; Hu and Luo, 2018).

Depending on the route of administration, Chit-Ins NPs can be taken up and processed by APCs, or transported into lymphatic tissues, triggering a local and/or systemic immune response against the protein (Amidi et al., 2010; Soares and Borges, 2018). Furthermore, it should be kept in mind that the repeated administration of the formulation increases the potential risk of antibody formation against insulin (Jiskoot et al., 2009).

### The Hurdles of Protein Delivery by Chit-NPs

Even though there is plenty of information on chitosan in the literature, there is also a huge gap with regard to chitosan standardization, making it difficult to relate its characteristics with the outcomes reported (Vasiliev, 2015) and to establish guidelines for SbD implementation. Note that polymer composition is a requirement of the assay cascade for nanomedicines elaborated by both US NCL and EU-NCL (European Nanomedicine Characterization Laboratory, 2019b; Nanotechnology Characterization Lab, 2019), thus the complete characterization of chitosan is revealed to be the greatest need and challenge of all.

The FDA Department Guidance for Industry "Drug Products, Including Biological Products that Contain Nanomaterials" requires the full description of nanomaterial composition, based on their functionality and intended use (U. S. Food and Drug Administration, 2017). Furthermore, the FDA guidance states that the nanomaterial critical quality attributes (CQAs) should be determined as early as possible, considering their functions and potential impact on the final product performance (quality, safety, and efficacy). Moreover, risk assessment should be applied linking the structure-function relationship of the nanomaterial to attributes that need to be examined and controlled in case of manufacturing changes – for example, the source and supplier of chitosan for NP production (U. S. Food and Drug Administration, 2017). Scarce good laboratory practice (GLP) conditions and questions regarding the validity and reproducibility of the scientific results are obstacles to collaboration with pharmaceutical industry and approval by regulatory authorities (Rosenblum et al., 2018). For example, clinical translation relies on a consistent and reproducible product (Anselmo and Mitragotri, 2016). As far as chitosan is concerned, contradictory information available in the literature on chitosan-biological activity correlation may be a potential source of problems during the drug approval process.

The risk assessment approach should also be applied to evaluate possible adverse immune responses that may be associated with nanomaterial administration, affecting both safety and efficacy. Biological products with a nanomaterial component may have a different immunogenic profile compared

to the biological substance alone, which may apply to Chit-Ins NPs (U. S. Food and Drug Administration, 2017).

As reviewed elsewhere (Jiskoot et al., 2009), the particulate character of drug delivery systems makes them predisposed to be recognized as foreign by immune cells and the complement system. In general, the elicited immune response depends on the route and frequency of administration. Moreover, in case of Chit-Ins NPs the potential immune response will also depend on chitosan characteristics and its source, on the properties of the nanocarrier (size, surface charge, polydispersity, etc.), and on the insulin employed. Often recombinant human insulin is applied, which usually does not stimulate immune responses. However, the immunogenicity risk of frequent administration of Chit-Ins NPs is unknown, as chitosan is known to have adjuvant properties, and recombinant human therapeutic proteins are also known to trigger antibody production after chronic treatment (Hermeling et al., 2004). Chitosan systems stimulate both cellular and humoral responses. Therefore, studies should be carried out to detect anti-insulin IgG1 and IgG2a production after Chit-Ins NPs administration. Screening of cytokine production, such as IL-4 and IFN-G, and detection of IgA, in the case of mucosal administration, would also be of interest.

In the end, the potential problems regarding Chit-Ins NP administration can be analyzed from a larger scope. The application of Chit NPs to protein delivery, in general, should take into account chitosan characteristics and the potential triggering of an immune response. These must be taken into consideration when examining the human health risks of a formulation in the framework of a SbD approach, especially when it is not desirable to stimulate the immune system.

#### CONCLUSION

This review shows that the characterization of chitosan is frequently missing in scientific reports, which complicates the translation into a SbD driven approach. Since the term chitosan is applied to a large group of polymers, the biological effects can be different and dependent on the degree of deacetylation and molecular weight of the polymer used on the study. This fact may explain, at least in part, the contradictory biological effects of chitosan reported in literature. Moreover, the purity of the polymers is not always mentioned, and the observed effects may be influenced by the presence of contaminants and impurities. Additionally, a similar situation can be observed with Chit NPs. Several protocols can be found in literature for insulin encapsulation into Chit NPs, however, in view of the lack of complete information given, it is difficult to reproduce them.

#### REFERENCES


Protocols also differ, which is an additional problem for data analysis and its comparison.

Furthermore, even though the immunostimulatory effect of chitosan systems has been well reported in the vaccine delivery field, the undesirable potential immune stimulation of those nanocarriers has been given less attention.

The data presented in this report illustrate the challenges encountered when implementing the SbD concept to polymeric drugs based on chitosan. The SbD approach defined by GoNanoBioMat establishes an early risk identification through material design and characterization. However, as it is shown in this report, the correlation between chitosan's physicochemical properties and its activity is far from being established. Consequently, it is also difficult to correlate Chit NP characteristics with the efficacy of the final drug product. Moreover, the potential hazard, namely, the eliciting of an unwanted immune activity, is also difficult to predict.

The full understanding of the composition of the nanoformulation is a critical point, thus a lack of knowledge in this field may explain why the number of approved drugs with chitosan as excipient is limited. Harmonization and validation of chitosan analysis will enable comparison between future studies. By developing these studies, it will be possible to establish the characteristics of different types of chitosan nanoparticles, establish a correlation between chitosan properties and its immunostimulant activity and, finally, to establish a guideline to select the most appropriate chitosan according to its purpose, allowing a safe-by-design approach.

#### AUTHOR CONTRIBUTIONS

CM drafted the manuscript, was responsible for the acquisition, analysis, and interpretation of the data for the work. CS, MS, and OB provided critical revision and redrafted the manuscript. GB provided critical revision, redrafted the manuscript, and gave approval for publication of the content.

#### FUNDING

This study was part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016; from the CTI (1.1.2018 Innosuisse), under grant agreement Number 19267.1 PFNM-NM; and from FCT Foundation for Science and Technology under the project PROSAFE/0001/2016.

Opin. Drug Deliv. 13, 223–237. doi: 10.1517/17425247.2016.110 7543


anastomosis. Ulus. Travma Acil Cerrahi Derg. 24, 405–411. doi: 10.5505/tjtes. 2018.59280




groups in partially N-deacetylated chitins (chitosans) by high-field n.m.r. spectroscopy. Carbohydr. Res. 211, 17–23. doi: 10.1016/0008-6215(91)84142-2


**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 Marques, Som, Schmutz, Borges and Borchard. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Hydrogel Biomaterials for Application in Ocular Drug Delivery

Courtney R. Lynch<sup>1</sup> , Pierre P. D. Kondiah<sup>1</sup> , Yahya E. Choonara<sup>1</sup> , Lisa C. du Toit<sup>1</sup> , Naseer Ally<sup>2</sup> and Viness Pillay<sup>1</sup> \*

<sup>1</sup> Wits Advanced Drug Delivery Platform Research Unit, Department of Pharmacy and Pharmacology, School of Therapeutics Sciences, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa, <sup>2</sup> Division of Ophthalmology, Department of Neurosciences, School of Clinical Medicine, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

There are many challenges involved in ocular drug delivery. These are a result of the many tissue barriers and defense mechanisms that are present with the eye; such as the cornea, conjunctiva, the blinking reflex, and nasolacrimal drainage system. This leads to many of the conventional ophthalmic preparations, such as eye drops, having low bioavailability profiles, rapid removal from the administration site, and thus ineffective delivery of drugs. Hydrogels have been investigated as a delivery system which is able to overcome some of these challenges. These have been formulated as standalone systems or with the incorporation of other technologies such as nanoparticles. Hydrogels are able to be formulated in such a way that they are able to change from a liquid to gel as a response to a stimulus; known as "smart" or stimuli-responsive biotechnology platforms. Various different stimuli-responsive hydrogel systems are discussed in this article. Hydrogel drug delivery systems are able to be formulated from both synthetic and natural polymers, known as biopolymers. This review focuses on the formulations which incorporate biopolymers. These polymers have a number of benefits such as the fact that they are biodegradable, biocompatible, and non-cytotoxic. The biocompatibility of the polymers is essential for ocular drug delivery systems because the eye is an extremely sensitive organ which is known as an immune privileged site.

#### Edited by:

Gerrit Borchard, Université de Genève, Switzerland

#### Reviewed by:

Mark William Tibbitt, ETH Zürich, Switzerland Jianxun Ding, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, China

> \*Correspondence: Viness Pillay viness.pillay@wits.ac.za

#### Specialty section:

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

Received: 21 October 2019 Accepted: 05 March 2020 Published: 20 March 2020

#### Citation:

Lynch CR, Kondiah PPD, Choonara YE, du Toit LC, Ally N and Pillay V (2020) Hydrogel Biomaterials for Application in Ocular Drug Delivery. Front. Bioeng. Biotechnol. 8:228. doi: 10.3389/fbioe.2020.00228 Keywords: biopolymers, ocular drug delivery, hydrogel, nanotechnology, biomaterials, safety by design

# INTRODUCTION

There have been many recent advancements made in the delivery of drugs to the eye, a site that is challenging to treat. The eye is a relatively isolated organ within the body, with many barriers and mechanisms that limit the entry of foreign substances into the eye. These include, among others, the cornea, blinking reflex, blood-aqueous barrier, blood–retina barrier, and the nasolacrimal drainage system. Collectively, these systems make the delivery of drugs to both the anterior and posterior segment of the eye more difficult (Patel et al., 2013). Novel drug delivery systems are constantly being developed to overcome the low bioavailability observed in many conventional ophthalmic formulations; these novel systems include the development of hydrogels.

Hydrogels have been largely investigated within the medical industry for a number of purposes; including drug delivery and tissue engineering. These systems are composed of cross-linked polymers which are capable of swelling when placed in water or an aqueous environment. Hydrogels have been researched in terms of drug delivery because they are able to hold, within the cross-linked matrix, a number of different substances. These range from hydrophobic and

hydrophilic molecules to both micro- and macromolecules (Kang Derwent and Mieler, 2008). An example of the effectiveness of hydrogels in drug delivery is shown in the article by Li et al. (2018) where the delivery of antibiotics by hydrogel systems was discussed. It was highlighted how hydrogels are able to deliver antibiotics to a local site (overcoming the severity of side effects often seen with systemic administration), offer controlled release of the active ingredient, and have better biocompatibility than conventional drug delivery systems (Li et al., 2018). These benefits can be translated into the development of hydrogel systems for the delivery of drugs to the eye.

Due to the fact that hydrogels are so versatile and are able to be modified to exploit the environment and function they are being designed for; these systems are highly advantageous in the effective delivery of drugs to the eye (Kang Derwent and Mieler, 2008). **Figure 1** indicates the various potential applications for hydrogels in ocular drug delivery.

Hydrogels have been shown to alter the drug release profiles of a formulation (to a sustained drug release profile), largely due to the swelling rate and water adsorption properties of the biotechnology platform. This swelling rate of the hydrogel can be induced as a response to a change in the environment into which the hydrogel is placed; these are known as "smart" or stimuliresponsive hydrogels. The stimulus can be chemical or physical and allows for the development of drug delivery systems which are regulated by the body. In addition, these "smart" hydrogels are able to respond to external stimuli such as in the process of iontophoresis (Fathi et al., 2015).

Through the development of stimuli-responsive hydrogel systems, not only are researchers able to overcome the issues of low bioavailability and rapid removal from administration site which is currently seen with conventional formulations, they are also able to do so without comprising on patient comfort. These delivery systems are able to be administered as a liquid and then form a gel once in contact with the eye (Hamcerencu et al., 2020). This is an important factor to consider in terms of patient compliance as patients are less likely to make use of an ophthalmic formulation if it is difficult to administer which is often the case with formulations that are highly viscous such as ointments (Singh et al., 2019).

Polymers have received much attention for use in drug delivery, and more specifically ocular drug delivery, over recent years. Although there are countless polymers available, this review article focuses on those which occur naturally, also known as biopolymers. These specific polymers offer the beneficial properties of being biodegradable, biocompatible, and noncytotoxic. They also have the advantages of being readily available, renewable, and less expensive in comparison to synthetic polymers (Oh et al., 2009).

# PHYSIOLOGICAL OCULAR BARRIERS AND DEFENSE SYSTEMS WHICH IMPACT DRUG DELIVERY

There are many challenges when it comes to effective delivery of drugs to the eye. Many of these are as a result of the barriers and mechanisms present within the eye which are designed to protect it from foreign particles and substances. A brief overview of the major ocular defense mechanisms is discussed below.

The first defense mechanism found in the eye is pre-corneal factors which result in the low bioavailability of topically applied ocular formulations. These include the blinking reflex, high tear turnover rate, and the lacrimal drainage of the solution. The cul-de-sac of the eye can hold approximately 30 µl of an administered eye drop. However, majority of this is removed within 15–30 s after the drops have been administered (Gaudana et al., 2010). Considering these factors, drug delivery systems need to be developed that are able to improve the retention of the formulation at the administration site. Consequently, this will improve the penetration of the active ingredient into the eye. Both hydrogel systems and mucoadhesive biopolymers could furnish formulations with these much-needed advantages (Biro and Aigner, 2019).

One of the major barriers to foreign substance entry into the eye is the multiple layers through which substances must pass through in order to penetrate into the target tissues. These layers include the cornea and the conjunctiva, among others. The cornea is located in the anterior segment of the eye and it made up of six layers: the epithelium, Bowman's membrane, stroma, Dua's layer, Descemet's membrane, and the endothelium (Ludwig, 2005; Dua et al., 2013). It is one of the main penetration-limiting layers in terms of drug delivery. This layer is highly lipophilic which largely prevents the entry of hydrophilic molecules into the eye (Moiseev et al., 2019).

The conjunctiva is a highly vascularized membrane that covers most of the anterior aspect of the eye. This high vascularity means that, although it can be used for the delivery of hydrophilic and large molecules, a large portion of the administered drug will be removed via the conjunctiva and enter systemic circulation before penetrating into the eye. This is also one of the main reasons why topically administered drugs are not able to reach the posterior segment of the eye in effective concentrations (Willoughby et al., 2010).

The eye is composed of two segments; the anterior segment (composed of the aqueous humor, conjunctiva, cornea, iris, ciliary body, and lens) and the posterior segment (composed of the choroid, optic nerve, retina, sclera, choroid, and vitreous humor). Each segment is susceptible to a range of conditions and each poses its own challenges when it comes to drug delivery (Souto et al., 2019). There are two blood-ocular barriers; the blood-aqueous barrier and the blood-retinal barrier. These largely prevent the entry of substances into the eye from systemic circulation. Although systemic administration has been considered as a route for drugs needed in the posterior segment of the eye, the dose needed is often high which leads to unwanted side effects (Nettey et al., 2016). **Figure 2** highlights the blood-ocular barriers in addition to the tissues which comprise these barriers.

When a formulation is applied to the surface of the eye (i.e., topical administration), it is rapidly removed through the blinking reflex and nasolacrimal drainage. This drainage system removes the drug from the eye via the nasolacrimal duct. It then enters the nose and is absorbed by the nasal mucosa where it

FIGURE 1 | Highlighting the potential application for hydrogels in ocular drug delivery. These include the delivery of drugs to both the anterior and posterior segments of the eye which will aid in overcoming the physiological barriers. Possible topical formulations for delivery to the anterior segment include systems which gel upon application (in situ gelling formulations) and contact lenses. Posterior segment formulations include intravitreal injections, which are made more effective by hydrogel technology, and cell carrier systems (adapted with permission from Kirchhof et al., 2015).

enters into systemic circulation. This is another factor which furthers the low bioavailability of topical applied ophthalmic preparations (Rajasekaran et al., 2010).

Hydrogels have been shown to increase the residence time of an active ingredient, allowing more time for it to diffuse through the layers of the eye. This plays a major role by increasing the bioavailability of topically administered ophthalmic formulations (Vashist et al., 2014). Due to the increased viscosity of a hydrogel system, it is also better able to withstand the clearance of the formulation due to blinking, further improving the bioavailability (Li Z. et al., 2013).

Biopolymers have also been shown to help overcome these barriers to drug delivery. Some, such as chitosan, have inherent mucoadhesive properties which allow the formulation to remain at the administered site for a longer period of time (Fulgencio et al., 2012). Cellulose derivatives have also been used to enhance the viscosity of a formulation, thereby preventing it from being washed away from the ocular surface too rapidly (Rajasekaran et al., 2010).

# CURRENT COMMERCIAL FORMULATIONS UTILIZED FOR THE DELIVERY OF DRUGS TO THE EYE

There are many formulations currently on the market which are designed to treat ophthalmic conditions. These range from anterior segment conditions such glaucoma, bacterial conjunctivitis, and post-operative inflammation to posterior segment conditions such as neovascular age-related macular degeneration, uveitis, and macular edema (Sultana et al., 2006b; Bao et al., 2017; Kaji et al., 2018). Each of the drug delivery systems discussed below has distinctive disadvantages when it comes to the effective delivery of drugs to the eye. It has been shown that the inclusion of hydrogels into the drug delivery system has been able to overcome some of these challenges, as is highlighted by the various studies included below.

Currently, the most common dosage form used to treat ocular conditions is eye drops. These formulations can be solutions or suspensions. However, although they are the first line treatment, there are many limitations to their use. These range from low bioavailability and rapid clearance from the administration site, to poor patient compliance (Yellepeddi and Palakurthi, 2016). Active ingredients in eye drops are not able to penetrate through to the posterior segment of the eye and thus are mainly used to treat anterior segment conditions (Urtti, 2006).

Conventional, commercially available eye drops often have frequent dosing schedules (ranging from daily to multiple times a day) and, in the case of chronic conditions such as glaucoma, require the patient to use them on a long-term basis. This can lead to unwanted side effects, which, for example, has been seen with latanoprost eye drops (daily administered dose of one drop). These side effects can cause patients to stop using their medications as prescribed, or to not use them at all. This is another reason why novel drug delivery systems such as hydrogels are needed; to reduce the frequency of dosing, reduce side effects and be patient-friendly enough so that patients will use them for an extended period of time if need be (Cheng et al., 2016).

In a recent article written by Yadav et al. (2019), it was highlighted how pre-corneal factors lead to the low absorption of ocular active ingredients used to treat glaucoma, administered as eye drops. These factors, such as tear turnover rate and the drainage of the formulation from the administration site, result in a 70–80% loss of the amount of drug which is administered. It was also highlighted how the frequent dosing schedules of eye drops can cause damage of to the eye. The consideration of ointments has been made, as these formulations have a higher viscosity and are not as rapidly drained from the eye as a liquid formulation. However, ointments are known to cause blurred vision when administered which leads to poor patient compliance (Yadav et al., 2019).

Posterior segment conditions are generally treated using sub-tenon, intravitreal, or systemic administration. However, each of these routes also comes with challenges of its own. One of the main objectives in the development of new drug delivery systems for the posterior segment is to reduce the invasiveness of the formulations which are currently used. For example, anti-vascular endothelial growth factors (anti-VEGFs) are used to treat a number of posterior segment conditions, namely those affecting the retina such as myopic choroidal neovascularization and diabetic macular edema. However, anti-VEGF is currently only able to be administered via intravitreal injections as the molecules are large and hydrophilic which prevent them from penetrating through the various barriers. This highlights the need for new technologies and drug delivery systems which are able to deliver molecules such as anti-VEGF without frequent, invasive injections (Wong and Wong, 2019).

Intravitreal injections are able to deliver a high concentration of the drug directly into the vitreous of the eye but are invasive and pose risks such as retinal detachment, vitreous hemorrhage, and endophthalmitis. The chances of these happening increase with the frequency of administration (Urtti, 2006; Gaudana et al., 2009). The use of hydrogels as intravitreal injections, with their extended drug release profiles, can delay the frequency of intravitreal injections, thus lowering the chances of the aforementioned risks occurring. **Table 1** highlights the formulations which are currently used to treat ophthalmic conditions, both in the anterior and posterior segment of the eye. A brief breakdown of the disadvantages of each of the formulations is also given.

# CHARACTERIZATION BETWEEN PHYSICALLY AND CHEMICALLY CROSS-LINKED BIOTECHNOLOGY HYDROGEL SYSTEMS

As previously mentioned, hydrogels are formed from polymers through a process known as cross-linking. Cross-linking occurs when one polymer chain is linked to another chain via a bond, either through a chemical or physical process. It is these


TABLE 1 | Current ophthalmic formulations which are used to treat anterior and posterior segment conditions.

These formulations, both topical and intraocular, each have a number of disadvantages or challenges in terms of drug delivery which can be overcome by hydrogel systems.

bonds which give hydrogels their stability and multidimensional network structure. The process of cross-linking a hydrogel can have an impact on its physical properties such as elasticity, viscosity, and solubility (Maitra and Shulka, 2014).

Although chemical and physical cross-linking methods each have their own advantages and disadvantages, it is worth noting that physically cross-linked hydrogels do not employ agents containing reactive functional groups which may cause inflammatory responses in vivo. However, these hydrogels also result in limited control over how the hydrogel is degraded within the body and, if the physical bonds are not strong enough, the inevitable dilution within the body can negatively impact the mechanical integrity of the hydrogel (Patenaude et al., 2014).

# Hydrogels Which Are Cross-Linked Through Physical Bonds

Physical bonding occurs through interactions between the polymer chains such as ionic bonding, Van der Waals forces, hydrogen bonding, or hydrophobic forces. Due to these types of bonds, the hydrogels formed through physical bonds are known to be reversible and have a degree of instability (Trombino et al., 2019). The hydrogels formed through physical interactions are generally less stable than those formed through chemical interaction as these bonds are susceptible to formation and breakage when there are changes in pH, temperature, and ionic strength. However, this can be a favorable characteristic if the desired outcome is a reversible hydrogel (Kirchhof et al., 2015).

# Hydrogels Which Are Cross-Linked Through Chemical Bonds

Chemically formed hydrogels are known as "permanent" hydrogels due to the covalent bonds which form between polymer chains. These systems allow more stability and maintain their structure better than the physical hydrogels (Trombino et al., 2019). However, it is important that the cross-linking agent can be removed completely from the hydrogel or a non-toxic agent is used so as to prevent adverse tissue reactions when the hydrogel is placed into the eye (Hoare and Kohane, 2008).

The stability of a chemically cross-linked hydrogel was demonstrated by Yu et al. (2015). In this study a hydrogel comprised of hyaluronic acid and dextran was evaluated for the delivery of bevacizumab, a monoclonal antibody which is used to treat neovascular diseases (Grisanti and Ziemssen, 2007). The hydrogel system was designed so that once it had been injected intravitreally, the polymers would form a solid gel. While this delivery system design has the benefits of a chemically cross-linked hydrogel, it also does not contain any cross-linking agent (the polymers cross-link with each other in response to physiological conditions) thereby improving its biocompatibility. The hydrogel system was able to release the active ingredient via a controlled release mechanism and maintain a therapeutically relevant concentration within the vitreous over a period of 6 months during in vivo studies. This would eliminate the current monthly schedule needed for bevacizumab administration, the risks of which have been discussed above (Yu et al., 2015).

### STIMULI-RESPONSIVE AND IN SITU HYDROGEL SYSTEMS AND THEIR APPLICATIONS IN OCULAR DRUG DELIVERY

In situ forming gel preparations offer an interesting advancement in sustained drug release profiles. This can be particularly useful

in terms of the delivery of drugs to the eye as these systems provide an increased retention time at the cornea as well as prevent the rapid removal of the formulation via the nasolacrimal drainage system (Cheng et al., 2016). Both of these factors play a role in overcoming the current challenge of low bioavailability seen in many ocular drug delivery preparations.

These in situ gelling systems are a type of stimuli-responsive hydrogels that are able to be administered to the eye as a liquid drop and subsequently form a gel after administration; known as a sol–gel transition. Gelation can be brought about as a response to a change in pH, ionic content, or temperature; although not all hydrogel systems are designed as stimuli-responsive systems and are simply administered as a gel (Al Khateb et al., 2016). Along with the ease of administration and prolonged retention time, in situ gelling systems have other advantages such as accurate dosing, simple formulation processes, and easy sterilization (Agrawal et al., 2010). **Figure 3** depicts the various stimuli which can cause a hydrogel to swell or de-swell.

In situ gelling systems have also been shown to exhibit sustained drug release profiles, another beneficial factor in ophthalmic drug delivery. This has been observed in many of the studies which are discussed below.

# Temperature-Sensitive Hydrogel Systems

Temperature-sensitive, also known as thermosensitive, hydrogels undergo swelling or de-swelling in response to a change in temperature. There are three classifications of thermosensitive hydrogels; negatively thermosensitive (these contract in response to an increase in temperature), positively thermosensitive (these contract in response to a decrease in temperature), and thermally reversible gels (Masteikova et al., 2003).

Thermosensitive in situ hydrogels, which are commonly utilized for drug delivery purposes are liquid at room temperature (20–25◦C) and form viscous gels at body temperature (34–37◦C). The polymers used in these systems have a lower critical solvent temperature; the temperature at which the sol–gel transition occurs. It is important that this critical temperature is close to bodily temperatures so that the systems do not require an external heat source to form a gel (Kumar et al., 2013). The thermosensitive properties of these hydrogels have also be proven to be beneficial in recent cartilage tissue engineering research as they allow for minimally invasive administration yet form a scaffold with suitable mechanical strength. These systems are also able to mold into the irregular shaped area into which they are administered (Wang et al., 2019).

An in situ thermosensitive hydrogel was developed by Chen et al. (2012) for the delivery of a model drug, levocetirizine dihydrochloride. The hydrogel system was comprised of chitosan and disodium α-D-glucose-1-phosphate (DGP) and showed many favorable results. The formulation was a low viscosity liquid at room temperature and a gel at physiological temperature. It showed an initial rapid release of the drug, followed by a sustained drug profile. When in a gel form, the system showed that it had a prolonged residency time, in comparison to that of an aqueous solution, as well as improved cornea penetration of the drug (Chen et al., 2012). This shows that a thermosensitive hydrogel system is able to overcome some of the challenges seen in conventional ophthalmic treatments.

# pH-Sensitive Hydrogel Systems

These in situ gelling systems either swell or de-swell as a response to a change in the pH within the environment into which it is placed. The polymers used in pH-sensitive hydrogels have ionic groups which give them their responsive ability. For example, cellulose acetate phthalate latex (formulation pH of 4.4) has been shown to form a viscous gel when placed into the culde-sac of the eye. However, the development of pH-sensitive gels must take into account the delicate environment of the eye. The formulation must have a buffer capacity that can form a gel when placed into the eye but not cause damage to the eye (Kushwaha et al., 2012).

Although many of the polymers used in pH-sensitive hydrogels are synthetic polymers, such as carbopol [polyacrylic acid (PAA)] and polyethylene glycol, natural biopolymers are also used in the formulations to give them more favorable characteristics (Kushwaha et al., 2012; Wu et al., 2013). For example, in a study performed by Kumar and Himmelstein (1995), it was shown that, although PAA is able to change from a low viscosity liquid when in an acidic solution to a gel at a higher pH, the amount of PAA needed for this to occur was too high. This means that the solution could not be neutralized by the tear fluid which acts as a buffer in the eye. To overcome this, hydroxymethylcellulose, a natural polymer also able to act as a viscosity modifier was added. Both the PAA and the hydroxymethylcellulose were low viscosity liquids at pH 4.0 and transformed

into viscous gels at a pH of 7.4. This meant that the concentration of PAA could be reduced to a safe level, without compromising the gelling and rheological behavior of the system (Kumar and Himmelstein, 1995).

The ability of methylcellulose, as hydroxypropylmethylcelullose, to act as a viscosity modifier in a pH-sensitive gelling system was further demonstrated by Srividya et al. (2001). The researchers developed a pH-triggered in situ gelling system comprised of PAA and hydroxypropylmethylcellulose which was shown to be a viable system in the topical delivery of ofloxacin.

#### Ion-Sensitive Hydrogel Systems

An ion-sensitive gel transforms from a liquid to a gel as a result of a change in ion concentration within the environment it is exposed to. An example of such a gel is shown in a study by Liu et al. (2006). The researchers formulated an alginate hydrogel for the delivery of gatifloxacin, a broad-spectrum antibiotic, which underwent a sol–gel transition when exposed to divalent cations. Methylcellulose was incorporated in order to decrease the amount of alginate needed for gelation. This formulation was able to release the active ingredient over an 8-h period in vitro and formed a gel within the cul-de-sac of the eye when administered as a drop. This renders an ion-sensitive hydrogel a suitable alternative to conventional eye drops as it increased the residence time and sustained drug release profile will lead to an improved bioavailability (Liu et al., 2006).

# Ultrasound-Responsive Hydrogel Systems

Ultrasound responsive systems are able to deliver drugs to a specific site which prevents the side effects which can be seen with systemic administration of certain drugs. These systems can incorporate nanotechnology. Polymeric hydrogels or nanocarriers such as nanobubbles are loaded with the drug and, once administered, exposed to ultrasound waves. This then leads to cavitation and high temperatures at the site, causing the rupture of the polymeric chains of the nanobubble (Mura et al., 2013; Mahlumba et al., 2016).

Ultrasound-responsive systems are able to deliver a drug at a rate which is controlled from an external source which make them particularly useful in the investigation of cancer treatment. An example is the use of oxygen nanobubbles used for the delivery of mitomycin-C. The nanobubbles system was capable of lower tumor progression rates with a 50% lower drug concentration (Bhandari et al., 2018).

The application of ultrasound waves has been shown to be beneficial in the penetration of drugs through the various barriers of the eye, including the cornea. This was shown to be true in a study performed using dexamethasone where a significant increase in the permeability of the cornea was observed (Nabili et al., 2013). However, there is some concern over the increase in temperature which is induced as it may cause damage to the sensitive structures within the eye. A study was completed by Nabili et al. (2015), which showed that the ultrasound frequency which had previously been shown to increase penetration was safe for the ocular tissues tested.

# Iontophoresis: An External Stimulus for More Effective Ocular Drug Delivery

Iontophoresis is a physical force-based response technique which is used to enhance the penetration of an ocular active ingredient through the various tissue layers found in the eye. This is done by applying an electric current between two electrodes; one which is used to deliver the drug and another which is placed on the body. The ionized drug is then able to travel through the tissue as a conductor of the current. Iontophoresis has been illustrated extensively in transdermal applications but has also been investigated for use in ocular drug delivery (Eljarrat-Binstock and Domb, 2006).

There are many challenges, which have highlighted throughout this article, associated with the delivery of drugs to the anterior chamber of the eye but there are even more challenges in the delivery to the posterior segment. Most active ingredients aren't able to penetrate through to the posterior segment when they are applied topically. This has led to the investigation of alternative routes of delivery such as intravitreal, subconjunctival, or transscleral. Iontophoresis has also been considered to aid in delivering drugs to the posterior segment. This allows for the treatment of conditions such as retinitis, uveitis, diabetic retinopathy, and age-related macular degeneration (Myles et al., 2005).

There are various device designs which can be utilized for iontophoresis; one such design includes a hydrogel. A hydrogel pad is saturated with a drug and acts as the delivery probe. This system has been shown to have promising results when tested with various drug entities such as dexamethasone. Transscleral hydrogel-based iontophoresis devices have been tested in both in vivo studies and clinical trials in healthy subjects and have shown good safety profiles as well as successful delivery of drug to the retina and choroid (Huang et al., 2018).

Although there are some iontophoresis devices which have been designed for transscleral drug delivery, the process does have some disadvantages. As with any medical procedure, there are risks involved; these include epithelial edema, inflammation, and burns (depending on the current density and duration of treatment). Iontophoresis has been demonstrated to be effective in improving the penetration of steroids, antibiotics, and antivirals. However, it has been reported that it is not able to deliver macromolecules to the vitreous in rabbits at a significant concentration (Thrimawithana et al., 2011).

In a study by Eljarrat-Binstock et al. (2008), hydrogel iontophoresis was employed to deliver nanoparticles to the eyes in an in vivo rabbit model. This study also investigated whether positively or negatively charged fluorescence nanoparticles penetrated through the tissues better. The researchers noted that, while iontophoresis is effective in improving the penetration of drugs into the eye, each active ingredient needs to be evaluated separately due to the fact that the physicochemical properties of the molecule will influence its behavior during the procedure. In this study, the, respectively, charged nanoparticles

were loaded into a hydrogel sponge and were administered via an iontophoretic device at the central cornea and at the sclera. After a specified amount of time the eyes of the rabbits were enucleated and tissue samples collected. The negatively charged particles showed penetration into the inner ocular tissues after 4 h, which increased after 12 h. However, the positively charged nanoparticles showed extensive penetration into the inner tissues at just 4 h after administration, illustrating the effect of the physicochemical properties of the particles on their behavior. Both of these indicate that iontophoresis is an effective way of ensuring the penetration of nanoparticles (which are able to be loaded with an active ingredient) through the eye (Eljarrat-Binstock et al., 2008).

Iontophoresis has also been used for the delivery of drugs through the suprachoroidal space (SCS). In a study performed by Jung et al. (2018), a micro-needle device was tested for the delivery of nanoparticles in an ex vivo rabbit model. The results showed that with an injection into the SCS without iontophoresis the nanoparticles that were localized around the site of injection (less than 15% delivered to the posterior region of the SCS). However, in the eyes on which iontophoresis was performed, over 30% of the nanoparticles were found in the posterior region of the SCS; this was also found in the in vivo study. These studies show how iontophoresis is able to improve the delivery of drugs to the eye and is able to be used in place of other delivery systems such as intravitreal injections (Jung et al., 2018).

# BIOPOLYMERS EMPLOYED IN THE FORMULATION OF OCULAR HYDROGEL SYSTEMS

Natural polymers have been widely investigated in a number of medical fields, including tissue engineering and drug delivery. This is largely due to the fact that they are biodegradable within the body and do not induce an inflammatory reaction (Singh, 2011). In terms of tissue engineering, they have also been shown to be conducive to cell growth and have a structure similar to the tissue matrix (Zhang et al., 2019). This section will focus on how natural polymers are employed in drug delivery systems.

These polymers, also known as biopolymers, have long been viewed as a crucial aspect in the developments that are achieved in the field of drug delivery. Highlighted below are biopolymers commonly used in ocular drug delivery systems. Their chemical structures are shown in **Figure 4**.

# Chitosan Polymeric Bio-Platforms

Chitosan is one of the most widely used polymers in polymeric drug delivery systems due to its biocompatibility, biodegradability, and low toxicity profiles (Bhattarai et al., 2010). It is a cationic polysaccharide which is derived from chitin. One of chitosan's most beneficial qualities is its mucoadhesive properties. The mucoadhesion is due to the fact that the positively charged chitosan is able to interact with the negative charges found in mucin (Fulgencio et al., 2012). This quality allows for improved permeation of drugs through ocular tissues as well as their controlled release from the formulation; both of which are vital in improving the delivery of drugs to the eye (Duttagupta et al., 2015).

Although chitosan is a very useful biopolymer for drug delivery, it is only soluble in acidic solutions. This is not desirable, especially when it is being formulated in ophthalmic formulations. For this reason, chitosan is often modified, for example through PEGylation and carboxymethylation (Xu et al., 2013).

A thermosensitive chitosan-based hydrogel was formulated by Cheng et al. (2016). This system was designed to overcome some of the challenges seen with latanoprost eye drops such as unwanted side effects after long-term use and low bioavailability. The hydrogel was characterized using both in vitro and in vivo tests for drug release and biocompatibility. The system was shown to be well tolerated and non-cytotoxic. During the in vivo studies, using a rabbit model, latanoprost was found in the aqueous humor 7 days after a single topical administration of the system, suggesting that this system could be administered on a weekly base instead of a daily basis as the commercial product is currently (Cheng et al., 2016).

Chitosan is often used in combination with other natural or synthetic polymers. For example, a study was performed by Cao et al. (2007) where a poly(N-isopropylacrylamide)-chitosan (PNIPAAm-CS) polymer was formulated into a thermosensitive in situ gelling system for the topical delivery of timolol, an active ingredient used for the treatment of glaucoma. The PNIPAAm-CS delivery system showed a higher Cmax and area under the curve (AUC) of blood concentration against time than that of a convention eye drop containing timolol. The gel system was also able to lower the intraocular pressure more than the eye drop over a 12-h period (Cao et al., 2007).

Another example is a hydrogel system was developed by Yu et al. (2017) containing carboxymethyl chitosan and a poloxamer composed of poly (ethylene oxide)/poly (propylene oxide)/poly (ethylene oxide) (PEO–PPO–PEO). The hydrogel was chemically crosslinked using glutaraldehyde and was able to undergo a reversible sol–gel transition in response to a change in pH and/or temperature. Preliminary studies, including cell studies performed with human cornea epithelial cells, showed that the hydrogel was not cytotoxic and has sustained drug release profiles (in comparison to a sample drug solution systems). This shows that this system could be further developed for ocular drug delivery (Yu et al., 2017).

# Hyaluronic Acid Polymeric Platforms

Hyaluronic acid is an anionic biopolymer which is found naturally within the human body. It is biodegradable and does not cause an immune response when used in medical systems. Due to this, hyaluronic acid has been a major interest in the design of drug delivery systems. It is particularly useful in respect to ocular drug delivery because it is a component within the vitreous humor of the eye and also has ligands for receptors found in many types of retinal cells, such as CD-44 (Martens et al., 2015).

Hyaluronic acid is endogenous to the body, making it highly biocompatible and non-immunogenic. However, it is not able to form a gel on its own and thus hydrogels made from hyaluronic acid rely on chemical modifications

and cross-linking or gelling agents. Hyaluronic acid hydrogels have been investigated as a drug delivery system because they are able to be formulated as both static and stimuli-response (Trombino et al., 2019).

Hydrogels are able to be utilized in conjunction with other technologies in order to improve ocular drug delivery. This can be seen in a study by Widjaja et al. (2015), where a hyaluronic acid-nanocomposite hydrogel was formulated with a sample drug, latanoprost. This system, in which the modified hyaluronic acid was combined with liposomes which contained the drug before crosslinking occurred, showed longer drug release profiles than the hydrogel and liposomes each did on their own. The composite system also improved the stability of the liposomes and the viscosity of the formulation. The hyaluronic acid was modified in two ways, using either adipic dihydrazide (ADH) or methacrylic anhydride (MA). Both modifications were tested throughout the study. The drug release mechanism is shown in **Figure 5**; it was found that both liposomes with entrapped drug and free drug were released from the hydrogel matrix which is what is believed to be the reason behind the sustained drug delivery profile which was observed. Although only preliminary studies were conducted; with further research, these nanocomposite systems are a potential candidate for the delivery of drugs to the eye after a single administration (Widjaja et al., 2015). **Figure 5** shows how the drug is released from the system.

Another hyaluronic acid-based hydrogel system was developed by Wu et al. (2013). This system was designed to be a thermoresponsive microgel for the topical delivery of drugs to the eye. Hyaluronic acid was coupled with g-poly(Nisopropylacrylamide) to form HA-g-PNIPAAm which was shown to have high drug loading capabilities. The gel was tested for biocompatibility in rabbit eyes with the results showing that is was safe and did not cause any irritation. The formulated system, with a sample drug cyclosporine A (CyA), was tested against a castor oil solution of CyA and a commercial product also containing CyA. There was a significantly higher concentration of CyA in the corneas of rabbits who received the HA-g-PNIPAAm system than in those who received the other two solutions. This shows that in situ thermoresponsive gels are able to improve the bioavailability of ocular active ingredients (Wu et al., 2013).

Hyaluronic acid hydrogels have been investigated not only as a drug delivery system but also as an artificial vitreous substitute. Schramm et al. (2012) completed a study whereby hyaluronic acid hydrogels were formulated using two different cross-linking methods; the first through the use of dihydrides as a cross-linking agent and the second through photocrosslinking. Both methods resulted in three-dimensional hydrogels which had suitable optical transparency and rubber-like consistency. The results of this study showed that these hydrogels are able to replace the conventionally used silicone oils, which have disadvantages such as the formation of cataracts and a need for surgical removal of the oil, as a vitreous replacement on a long-term basis (Schramm et al., 2012).

#### Gelatin Polymeric Platforms

Gelatin is a natural polymer which is biocompatible and biodegradable. It is derived from collagen, a substance which is found naturally within the stroma of the cornea and sclera. It has been investigated for a number of ocular drug delivery systems; including nanoparticles (Vandervoort and Ludwig, 2004). Natu et al. (2007) performed a study where gelatin hydrogels were investigated as a drug delivery system for pilocarpine, an ocular active used in the treatment of glaucoma. The hydrogels were formulated through chemical crosslinking with N-hydroxysuccinimide (NHS) and N, N-(3-dimethylaminopropyl)-N 0 -ethylcarbodiimide hydrochloride (EDC). These crosslinkers were used in a variety of concentrations which altered the degree of crosslinking and subsequently the release of the drug from the hydrogel. The release of pilocarpine from the various hydrogels ranged from 29.2 to 99.2% over an 8-h period. The hydrogels also displayed good adhesion and non-cytotoxicity profiles. This shows hydrogels comprised of gelatin to be a viable option for the delivery of drugs to the eye (Natu et al., 2007).

In a study by Song et al. (2018), chitosan and gelatin were used to form a hydrogel aimed at improving the sustained delivery of drugs to the eye. The hydrogel was formed using a double crosslinking method; using both genipin and β-glycerophosphate

disodium salt hydrate as crosslinking agents. The resulting hydrogel had in situ gelling properties; showing rapid gelation at 37◦C. Timolol maleate was used as a sample drug as a comparison could be made against a commercially available product. The hydrogel delivery system was non-toxic and showed a sustained release drug release profile. During in vivo studies, in comparison to the commercial product, the hydrogel delivery system was able to show a longer lasting and more effective reduction (due to a twofold increase in duration) in the intraocular pressure. The in situ gelling property also prevented the system from being rapidly removed from the lower conjunctival sac by tears following administration (Song et al., 2018). **Figure 6** shows the double crosslinking-method which is used in this formulation.

# Alginate Polymeric Platforms

Alginate is another highly biocompatible polysaccharide that is able to undergo ion-responsive gelation (Liu et al., 2008). It is classified as a polyanionic copolymer and is extracted from brown sea algae. Alginate forms a hydrogel when it is exposed to divalent cations such as Ca2<sup>+</sup> (Lin et al., 2004). It has been used in ocular hydrogel preparations because it is non-cytotoxic and biodegradable. It was used in a formulation by Lin et al. (2004) which is discussed below under "ion-sensitive hydrogels."

The utilization of alginate can also be seen in the study reported by Mandal et al. (2012) where an in situ forming gel was prepared using sodium alginate for the sustained delivery of moxifloxacin hydrochloride, a broad spectrum antibiotic. In this formulation, although sodium alginate was used as the primary gelling polymer, hydroxypropyl methylcellulose (HPMC) was also added as a viscosity enhancer. The resultant formulation was able to lengthen the precorneal residence time of the drug (also due to sodium alginate's mucoadhesive properties) and improve its bioavailability. The polymer was able to undergo a sol–gel transition in response to an ion exchange when administered to the eye. In vivo studies were performed for biocompatibility using healthy male albino rabbits. The rabbits showed no signs of irritation after the formulation was administered to the eye and no ophthalmic damage was noted. This makes this formulation a viable alternative to conventional eye drops for the delivery of moxifloxacin with a less frequent dosage schedule (Mandal et al., 2012).

Sodium alginate hydrogels have also been used in the delivery of anti-inflammatory drugs to the eye. One such formulation is that prepared by Pandit et al. (2007). They highlighted the preference for hydrogel systems over implants as novel ocular drug delivery systems due to the fact that hydrogels are more cost effective and comfortable to the patient while still overcoming the bioavailability issues that are seen with convention drug delivery systems. The hydrogel which was produced supported these sentiments; sodium alginate was formulated into an in situ gelling system which would increase the residency time of the drug as well as exhibit sustained drug release profiles; both of which are vital in improving the bioavailability of ocular drugs (Pandit et al., 2007).

## Methylcellulose Polymeric Platforms

Methylcellulose is natural polymer which is often used as a viscosity enhancer in ocular formulations. It is capable of undergoing a reversible sol–gel transition when it is heated. This makes it useful in the development of in situ gelling hydrogel systems (Sultana et al., 2006a).

In a study by Silva et al. (2017), a HPMC hydrogel was used to aid in the delivery of chitosan-hyaluronic acid nanoparticles to the eye, giving another example in how a hydrogel can be employed in a drug delivery system. Methylcellulose was used because it is safe to sterilize within an autoclave, it has a suitable pH for the eye and has been shown to be used successfully in other ophthalmic preparations (Silva et al., 2017). This study highlights one of the derivatives of methylcellulose, among others, which are often used in preparations. This is due to the fact that these derivatives influence the temperature at which the methylcellulose is able to undergo a sol–gel transition. For example, by lowering the molar substitution of hydroxyl propyl groups, the transition temperature is reduced from between 75 and 90 to 40◦C (Gambhire et al., 2013).

Methylcellulose can also be added to a formulation to adjust its gelation behavior. This was investigated by Dewan et al. (2015) in a study where methylcellulose of varying molecular weights were added to Poloxamer 407 (PM), a polymer previously investigated for the delivery of various drugs to the eye. However, when used in these formulations, PM is diluted by the lacrimal fluid of the eye and loses its ability to form a gel. Increasing the concentration of PM is not a viable solution as it causes the gelation temperature to drop; resulting in the formulation turning into a gel at room temperature. It was found that the addition of methylcellulose resulted in a decrease in the gelation temperature of the PM formulations and facilitated extended drug release profiles of the sample drug; making it a viable option for sustained drug delivery to the eye (Dewan et al., 2015).

A further study which illustrates that methylcellulose can be utilized in ophthalmic drug delivery preparations is that

performed by Bain et al. (2009). Agents such as fructose and sodium citrate tribasic dehydrate were added to the formulation to reduce the gelation temperature. These additives have an impact on the gelation temperature by affecting the interactions between the polymer and the water molecules. The sample drug used was ketorolac tromethamine (KT). The resulting formulation was able to extend the release of the drug from 5 to 9 h, largely due to the presence of fructose which further enhances the viscosity of the formulation. Although further testing and in vivo studies are needed, the resulting formulation is a viable option for the delivery of drug to the eye in the place of conventional eye drops (Bain et al., 2009).

### Collagen Polymeric Platforms

Collagen is a natural polymer which is also available to be used in ocular drug delivery systems. Type 1 collagen is one of the primary components of the cornea and has been used in scaffolds for tissue engineering (Chen et al., 2005). Collagen shields have been formulated and are able to deliver drugs to the eye for a maximum of 72 h. This is more beneficial than soft contact lenses, which have been shown to only delivery the drug for the first 1–2 h after insertion. These shields are generally used following ophthalmic surgery for the delivery of anti-inflammatory or immunosuppressive active ingredients, among others. However, these shields are non-transparent and have to be applied by a surgeon (Liu et al., 2008).

However, there are some collagen shields available which have the potential to be self-administered. As reported by Khan and Khan (2013), these bandage contact lenses are able to facilitate the healing of the cornea following surgery or injury by protecting it from abrasion caused by blinking. They are also able to be laden with active ingredients; as the tears dissolve the contact lens, the drug is released along with a layer of collagen which is able to lubricate the eye. This provides a system which is able to increase the residency time of the drug at the cornea, allowing for increased permeability and bioavailability (Khan and Khan, 2013).

An example of a formulation where collagen, along with hydrogel technology, has been developed is that reported by Liu et al. (2006) where composite collagen hydrogels were formulated which contained alginate microspheres for the delivery of drugs to the eye. The composite hydrogels were characterized and shown to be suitable for use in ocular inserts or contact lens formulations as they were biocompatible and showed sustained drug release profiles as well as supported the attachment and growth of corneal epithelial cells (Liu et al., 2006).

Collagen has also been used in hydrogels that are intended for tissue engineering purposes. They have been investigated as an alternative to amniotic membrane which is used for clinical ocular surface reconstruction. This is due to the fact that they biodegrade at a suitable rate and offer very low immunogenicity. In a study by Mi et al. (2010), these collagen-based scaffolds were investigated. It was found that collagen gels are difficult to manipulate because of their weak structure. This was overcome through controlled unconfined plastic compression which, depending on the collage concentration and time for which the gel was compressed, produced a scaffold which closely mimiced the structure of the cornea. These hydrogel scaffolds were able to adequately support cell attachments and epithelial cell growth (Mi et al., 2010).

# SAFETY BY DESIGN OF POLYMERIC HYDROGELS THROUGH OCULAR BIOCOMPATIBILITY AND BIODEGRADATION

The eye is an organ of immune privilege, which protects its visual capability from the potentially sight-threatening sequelae of intraocular inflammation (Keino et al., 2018). Consequently, any potential formulations used in the eye, whether it be for drug delivery, tissue engineering, or any other medical procedure need to be vigorously tested for biocompatibility.

# Biocompatibility

Many studies in which new ophthalmic formulations are being investigated include biocompatibility studies. Typically, the first step in determining biocompatibility is to determine the cytocompatibility of the formulation. This is done through cytotoxicity or cell proliferation tests which are performed in vitro. The cell line most commonly used for these tests is human corneal epithelial cells (HCECs). These in vitro tests are useful in determining biocompatibility as they provide a controlled environment whereby researchers can observe the impact of the polymers used in their formulation on cell characteristics such as adhesion, proliferation, and viability. It has been noted that cell studies which are performed with multiple, different cell lines provide a more accurate representation of the cells found within tissues than studies where only a single cell line is used (Huhtala et al., 2007).

The second process in determining biocompatibility is through in vivo testing. This is usually performed using animal models. The New Zealand white rabbit model is most commonly used in ophthalmic bioavailability studies. This is because the eye of an adult rabbit is big enough to ensure the procedure is performed accurately (for example, rat eyes are sometimes used but are often too small for formulations designed for use in human eyes) and there is no pigment epithelium in the eye (Short, 2008).

Although the majority of the studies that are detailed in this review include biocompatibility studies in addition to other characterizations, either through in vitro or in vivo testing, there are those available which focus primarily on biocompatibility. One such study is that performed by Lai (2010). The authors investigated the effect of different cross-linkers [namely glutaraldehyde (GTA) and 1-ethyl-3-(3 dimethyl aminopropyl) carbodiimide (EDC)] on the ocular biocompatibility of gelatin hydrogels. Gelatin has been shown to have a rapid dissolution when it has not been cross-linked and is placed within an aqueous environment, which would limit its potential application in the delivery of drugs to the eye. The biocompatibility was tested using both cell culture

techniques and in vivo animal testing. The cell line selected was primary rat iris pigment epithelial cells; these were cultured and observed for cell proliferation, viability, and presence of pro-inflammatory genes.

The results showed that the EDC cross-linked gels were better tolerated than the GTA hydrogels. This was then corroborated in the in vivo tests whereby the gelatin hydrogels were inserted into the anterior chamber of the eye of New Zealand white rabbits and observed for 12 weeks. The rabbits who were given the GTA cross-linked hydrogels showed a significant inflammation reaction whereas the EDC cross-linked hydrogels were well tolerated, concluding that EDC is more suitable as a cross-linking agent for the formulation of ophthalmic gelatin hydrogels. This study highlights that, although gelatin itself is biocompatible, the cross-linking agents which are used in the formulation of hydrogels have the ability to change the biocompatibility of a formulation (Lai, 2010).

The results mentioned in the study above were further corroborated in another study; also focusing on the biocompatibility of GTA and EDC cross-linked hydrogels, with the exception of using hyaluronic acid as the polymer. The results of the in vivo tests, performed using rabbits, showed that the EDC crosslinked hydrogel elicited no inflammatory response whereas the GTA cross-linked hydrogels produced a severe tissue response. This further highlights the importance of biocompatibility testing, not only for the polymer, but also for the other reactants used within a formulation (Lai et al., 2010).

Other in vitro methods for testing biocompatibility have been developed. An example of this is the development of a three-dimensional, curved epithelium model which is able to mimic the cornea. This model was designed and created by Postnikoff et al. (2014) in the hopes of removing the need for the use of animal testing in the development of some ophthalmic preparations. This particular model was shown to be multi-layered and responsive to cytotoxic compounds, as a cornea would which makes it a viable option in the biocompatibility assessment of contact lenses (Postnikoff et al., 2014).

#### Biodegradability

Biodegradability is one of the aspects which makes the polymers discussed in this review beneficial for use in ocular drug delivery. This allows sustained drug release systems to be able to breakdown and be absorbed by the body, eradicating the necessity for surgical removal. The most common form of biodegradable system is that where a drug is embedded within a polymeric system and is released as the polymer degrades. The advantage of biodegradable over non-biodegradable ocular systems has been seen in implants developed for sustained drug release. Majority of ocular implants currently available on the market are non-biodegradable but research is being done into the development of biodegradable formulations (Lee et al., 2010).

The biodegradable nature of polymers, while advantageous, can sometimes hinder their ability to maintain their integrity for an extended time within the environment into which they are placed. For example, hyaluronic acid, which is broken down by hyaluronidase, does not have a sufficient residence time for longterm delivery. Hyaluronic acid is often modified to overcome this issue (du Toit et al., 2013).

# INCORPORATION OF HYDROGELS AND NANOTECHNOLOGY FOR OCULAR DRUG DELIVERY

Hydrogels can form a vital role in the development of nanotechnologies for the delivery of drugs to the eye. An example of this is the formulation of hydrogel nanoparticles. This drug delivery system combines the benefits of a hydrogel (hydrophilic and high-water content) with the minute size of a nanoparticle. These have been developed using both synthetic and natural polymers but, in this article, only those employing natural polymers are discussed (Hamidi et al., 2008).

Although hydrogels themselves offer many advantages to overcome these challenges, by combining hydrogels in colloidal drug delivery systems the effective delivery of drugs to the eye is further improved. Nanotechnology, such as nanoparticles and nanoliposomes, has been given a lot of focus in recent years for use in ocular drug delivery. These nanocarriers are able to offer advantages such as the more targeted delivery of drugs and controlled release as well as reduced toxicity and improved efficacy of formulations. These carriers, which range from 1 to 1000 nm in size, are also able to deliver drugs which are poorly water soluble (a problem that in the past has seen ocular active drugs not being made into effective preparations) as well as provide improved penetration into tissues. Colloidal drug delivery systems are also able to increase the retention time at the surface of the cornea, resulting in improved bioavailability (Ameeduzzafar et al., 2016).

In terms of ocular drug delivery, nanoparticles are useful due to their small size which allows for targeted drug delivery and improved bioavailability. The drugs in these delivery systems can be incorporated into the nanoparticle either through entrapment, encapsulation, or attachment to the surface. Nanoparticles with intrinsic hydrogel structure are able to be formulated using either physical or chemical cross-linking methods and have been prepared using a number of synthetic and natural polymers. Nanoparticles are able to be combined with hydrogel technology either in the way that they are synthesized or in the way that they are administered where the hydrogel acts as a suspending agent (Hamidi et al., 2008).

A further example of the combination of hydrogels and nanotechnology is nanogels. These nanoparticle carriers have many beneficial properties in terms of ocular drug delivery. These include sustained drug delivery profiles and improved stability of the drug in water (Jamard et al., 2016).

In a study by Jamard et al. (2016), it was noted that many nanogels require harsh conditions for formulation, such as high temperatures and the use of organic solvents. However, it was noted that by using biopolymers (such as methylcellulose) which have been modified with hydrophobic moieties [such as poly(N-tert-butylacrylamide)], self-assembling nanogels could be formulated through hydrophobic interaction within an

aqueous environment. This renders the resultant, non-cytotoxic nanogel suitable for the delivery of biological compounds with a prolonged release profile (Jamard et al., 2016).

A further study, focusing on the delivery of fluconazole to the cornea, was performed by Nishil et al. (2013) where fluconazole loaded chitin nanogels were synthesized. The system was shown to have sustained drug release drug profiles while also being cytocompatible. It was also noted that the system allowed for penetration through the cornea in ex vivo studies. The nanogel can be considered for improved bioavailability for the fluconazole in the treatment of corneal fungal infections (Nishil et al., 2013).

Solid lipid nanocarriers (SLN) are another form of nanotechnology which have been researched for the replacement of conventional ocular drug delivery systems. These SLNs are advantageous as they have low toxicity due to the fact that they are prepared from lipids natural to the body, are able to undergo autoclave sterilization, and are able to be loaded with both hydrophilic and hydrophobic drugs (Farid et al., 2017). SLNs fall under a larger group of lipid-based nanocarriers which also includes lipid-drug conjugates (Puglia et al., 2015).

Nanoparticles offer a particular benefit in that, due to the large surface area-to-volume ratio, they are able to support a vast number of surface functional groups (Jacob et al., 2018). These surface modifications are able to improve some of the disadvantages which are seen in certain nanotechnologies. An example of this can be seen in a study by Attama et al. (2008) where a phospholipid was used as a surface modifier on SLNs. The results showed that the drug release from the SLNs which were formulated without the phospholipid happened in a burst release fashion due to the fact that there was more drug present in the periphery of the nanoparticles. In addition, a large amount of drug was found in the bulk aqueous medium. Those that were formulated with the phospholipid had a sustained drug release profile. This illustrates how surface modifications are able to have an effect on not only the drug release profiles but also the encapsulation efficacy of SLNs (Attama et al., 2008).

The concept of colloidal nanoparticulate-based systems has been investigated for therapeutic contact lenses. The incorporation of nanoparticles allows for improved drug release from the contact lens as well as prevents the interaction of the drug with the polymers of which the lens is composed. An example of such system was formulated by Jung et al. (2013). Nanoparticles which contained timolol, a drug used to treat glaucoma, were loaded onto commercial contact lenses. The contact lenses were tested in preliminary drug release and in vivo studies which showed that, in addition of being biocompatible, they were able to release timolol over an extended period (5 days) resulting in a lowering of the intraocular pressure. These are promising results as an alternative to conventional timolol eye drops which must be administered multiple times a day; however, there is still further research which needs to be conducted (Jung et al., 2013). This research would include the impact of colloidal systems on the contact lens' transparency and ion and oxygen permeability (Maulvi et al., 2016).

# FUTURE PERSPECTIVES

The primary focus of the research that is being done, and that has been commented on in this article is to improve the shortfalls seen in current ophthalmic treatments. Whether that be the low bioavailability and rapid clearance from the administration site found with eye drop formulations or the frequency of invasive procedures seen with intravitreal injections, future developments made in ocular drug delivery are vital (Sapino et al., 2019).

Many of the advancements being made in this area of drug delivery include harnessing the benefits highlighted for both biopolymers and hydrogel systems. One of the main focuses of the future perspectives is the further testing of the systems that have been discussed in this paper. This testing includes in vivo animal testing of systems that have undergone cell testing, and clinical trials for the systems that have undergone animal pilot studies. It has been noted that not many of the newly developed systems have been made commercially available and these studies would further this process (Barbu et al., 2006).

Natural, biodegradable polymers have uses in other future prospects for ocular drug delivery outside of their use in hydrogel systems, both on their own and in conjunction with synthetic polymers. These include the development of polymeric ocular inserts [as an example, an insert was developed by Jain et al. (2010) with sodium carboxymethylcellulose and polyvinyl alcohol for the topical delivery of ciprofloxacin]. Majority of the ocular inserts which are commercially available are composed of synthetic polymers so the development and commercialization of biopolymer-based inserts is a definite avenue for the future prospects of biopolymer technology.

Hydrogel systems have been demonstrated in many studies to be highly beneficial in their role as ophthalmic drug delivery systems. The advances that have been made in recent years, particularly in terms of "smart" or stimuli-responsive hydrogels, have made a large impact. However, many of these formulations have not been made commercially available, mainly because many of them have yet to undergo clinical trials. This would be a vital step in improving the quality of life of patients; especially those who require eye drop administration on a daily basis. According to the research that has been done, hydrogels provide an option for far less frequent dosing schedules (in some cases weeks or months) (Chang et al., 2019).

# CONCLUSION

Although hydrogels are not as extensively investigated as some of the other developments that are being made in ocular drug delivery, they are making an impact. These systems provide two vital benefits to drug delivery; sustained drug release and increased retention time. They are able to be formulated in such a way that they are able to respond to stimuli, which has been shown to be very beneficial. This stimuli-response ability allows for ease of administration, making these formulations more favorable for patients. This takes the ease of administration of eye drops and combines it with the increased viscosity of ointments, resulting in effective topical drug delivery without

frequent dosing schedules (seen with eye drops) and blurred vision (seen with ointments).

Biopolymers are at the forefront of many studies undertaken in ocular drug delivery. These polymers, with their noncytotoxic, biodegradable profiles enable researchers to develop technologies without the risk of causing inflammation and the need for surgical removal. They also lend themselves to safetyby-design aspects for new formulations as there are many studies which illustrate their low toxicity profiles. Biopolymers provide an easily available and relatively cheaper option to some synthetic polymers.

Both hydrogels and biopolymers lend themselves to use in nanotechnology for ocular drug delivery. Whether it be in the form of the intrinsic make-up of the nanoparticles, nanoliposomes, or nanowires, or as a suspending agent, hydrogels can greatly impact the developments which are

# REFERENCES


being made in this field of drug delivery. Although there are still developments to be made, both hydrogel and biopolymer technology play a vital role in the improvements being investigated for the effective delivery of drugs to the eye.

# 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 financially supported by the National Research Foundation (NRF) of South Africa.





**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 Lynch, Kondiah, Choonara, du Toit, Ally and Pillay. 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.

# A Methodological Safe-by-Design Approach for the Development of Nanomedicines

Mélanie Schmutz<sup>1</sup> , Olga Borges<sup>2</sup> , Sandra Jesus<sup>2</sup> , Gerrit Borchard<sup>3</sup> , Giuseppe Perale<sup>4</sup> , Manfred Zinn<sup>5</sup> , Ädrienne A. J. A. M Sips<sup>6</sup> , Lya G. Soeteman-Hernandez<sup>6</sup> , Peter Wick<sup>7</sup> and Claudia Som<sup>1</sup> \*

<sup>1</sup> Technology and Society Laboratory, Empa – Swiss Federal Laboratories for Materials Science and Technology, St. Gallen, Switzerland, <sup>2</sup> Center for Neuroscience and Cell Biology, Faculty of Pharmacy, University of Coimbra, Coimbra, Portugal, <sup>3</sup> School of Pharmaceutical Sciences Geneva-Lausanne, Geneva, Switzerland, <sup>4</sup> Polymer Engineering Laboratory, Department of Innovative Technologies, Mechanical Engineering and Materials Technology Institute, University of Applied Sciences and Arts of Southern Switzerland, Manno, Switzerland, <sup>5</sup> Institute of Life Technologies, University of Applied Sciences and Arts Western Switzerland (HES-SO Valais-Wallis), Sion, Switzerland, <sup>6</sup> National Institute for Public Health and the Environment, Bilthoven, Netherlands, <sup>7</sup> Particles-Biology Interactions Lab, Empa – Swiss Federal Laboratories for Materials Science and Technology, St. Gallen, Switzerland

#### Edited by:

Michele Iafisco, Italian National Research Council, Italy

fbioe-08-00258 April 1, 2020 Time: 15:47 # 1

#### Reviewed by:

Diana Boraschi, Istituto di Biochimica delle Proteine (IBP), Italy Ivana Fenoglio, University of Turin, Italy Henriqueta Louro, Instituto Nacional de Saúde Doutor Ricardo Jorge, Portugal

> \*Correspondence: Claudia Som claudia.som@empa.ch

#### Specialty section:

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

Received: 31 October 2019 Accepted: 12 March 2020 Published: 02 April 2020

#### Citation:

Schmutz M, Borges O, Jesus S, Borchard G, Perale G, Zinn M, Sips ÄAJAM, Soeteman-Hernandez LG, Wick P and Som C (2020) A Methodological Safe-by-Design Approach for the Development of Nanomedicines. Front. Bioeng. Biotechnol. 8:258. doi: 10.3389/fbioe.2020.00258 Safe-by-Design (SbD) concepts foresee the risk identification and reduction as well as uncertainties regarding human health and environmental safety in early stages of product development. The EU's NANoREG project and further on the H2020 ProSafe initiative, NanoReg2, and CALIBRATE projects have developed a general SbD approach for nanotechnologies (e.g., paints, textiles, etc.). Based on it, the GoNanoBioMat project elaborated a methodological SbD approach (GoNanoBioMat SbD approach) for nanomedicines with a focus on polymeric nanobiomaterials (NBMs) used for drug delivery. NBMs have various advantages such as the potential to increase drug efficacy and bioavailability. However, the nanoscale brings new challenges to product design, manufacturing, and handling. Nanomedicines are costly and require the combination of knowledge from several fields. In this paper, we present the GoNanoBioMat SbD approach, which allows identifying and addressing the relevant safety aspects to address when developing polymeric NBMs during design, characterization, assessment of human health and environmental risk, manufacturing and handling, and combines the nanoscale and medicine field under one approach. Furthermore, regulatory requirements are integrated into the innovation process.

Keywords: Safe-by-Design, polymeric nanobiomaterials, nanocarriers, drug delivery, nanomedicine

# INTRODUCTION

The concept of Safe-by-Design (SbD) was addressed in the field of nanotechnology because of the continuous uncertainty about the potentially harmful effects of nanomaterials on humans and the environment. Its implementation started with the Dutch NanoNextNL program<sup>1</sup> and the European NANoREG project and was further developed by the H2020 ProSafe initiative and

<sup>1</sup>www.nanonextnl.nl

H2020 NanoReg2 project (Soeteman-Hernandez et al., 2019). Since then, an increasing number of European Union projects focused on SbD for nanomaterials (Lynch, 2017). Even though various concepts of SbD coexist, they share the purpose of assessing safety as early as possible in the innovation process of a nanomaterial or nanoproducts. They aim at reducing adverse effects on human health and the environment by altering nanoproduct design (Soeteman-Hernandez et al., 2019) and by ensuring safety along its lifecycle (Bottero et al., 2017; Kraegeloh et al., 2018). The SbD concept is therefore different from conventional risk assessment approaches, which only consider safety when the product is already fully developed (Schwarz-Plaschg et al., 2017).

Despite being a rather novel concept in the context of nanotechnology, the principle behind SbD is not new and already applied by other industries (Kraegeloh et al., 2018). The medicine field has also long expertise in ensuring safety throughout the drug discovery and development process (Hjorth et al., 2017). However, how to handle safety issues effectively at the very beginning of drug development, to allow the selection of drug candidates, and mitigate toxicity is still being investigated (Kramer et al., 2007; Loiodice et al., 2019). The concept of Quality-by-Design (QbD) is widely used by pharmaceutical industry and its implementation is foreseen by the pharmaceutical development guidelines. The SbD is a new concept for the Pharmaceutical industry and it is not yet included in ICH, EMA, or FDA guidelines. This means that even if safety is considered during the pharmaceutical development, there is no systematic SbD approach yet in place. The concept of QbD presupposes the definition of the critical quality attributes (CQA) that will lead to the achievement of a product with proven effectiveness and the SbD would establish CQA that will lead to a product with high safety.

The application of nanotechnology in the medicine field (nanomedicine) brought new barriers precluding the prediction of potential adverse effects to human health and the environment because of the complexity of nanobiomaterials (NBMs). The unpredictability of nanomedicines' interaction with biological systems makes it difficult to bring these to the market (Resnik and Tinkle, 2007; Accomasso et al., 2018) and consequently, their potential benefits in medicine are still underexploited (Tinkle et al., 2014; Troiano et al., 2016). The lack of guidelines, standards and tools adapted to nanomedicines for assessing their risks represents one of the causes for this situation (Accomasso et al., 2018).

The development of such products remains therefore challenging. In addition, nanomedicines are costly and based on an interdisciplinary approach. They are at the junction of pharma, medtech, biotech and nanotech companies, and academia, which are important economic and social players in Switzerland and Europe. These companies may have different roles in the value chain of nanomedicines' development and as they have different backgrounds, they may have various needs to overcome the complexity of nanotechnology for medical applications.

As there is no systematic SbD approach in place for nanomedicines and that not all actors (coming from different fields) are experienced in considering safety to reduce risks on human health and the environment, there is a need for a methodological approach enabling to consider all necessary aspects to evaluate the safety of nanomedicines early during product development. This would ultimately improve the efficiency of the innovation process and the collaboration of all involved interdisciplinary actors and thus ensure the development of a safe product from the beginning of the process.

In order to fill this gap, within the GoNanoBioMat project,<sup>2</sup> we aimed at elaborating a methodological SbD approach by taking up the principles of the SbD approach developed for nanotechnologies in general and by adapting it to the field of nanomedicines. The developed methodological approach has a focus on polymeric nanocarriers for drug delivery (Som et al., 2019) as they are valuable materials, widely used to prepare nanoparticles and microparticles for the purpose of encapsulating drugs (Etheridge et al., 2013), can be biodegradable, biocompatible, and can be tailored to have targeting abilities (Bennet and Kim, 2014; Moritz and Geszke-Moritz, 2015). Therefore, these materials are expected to increase drug efficacy and safety (Ariën and Stoffels, 2016).

The aim of this paper is (1) to present what we adapted from the SbD concept developed within the EU projects NANoREG and NanoReg2, and the ProSafe initiative (hereafter general SbD approach) for the field of nanomedicines and (2) present the methodological SbD approach (hereafter the GoNanoBioMat SbD approach).

#### ADAPTATION OF THE GENERAL SbD APPROACH TO NANOMEDICINES

The general SbD approach can be applied in many different fields (e.g., paints, textiles, etc.), is addressed to industries, and can be used by regulators as a reference tool (Kraegeloh et al., 2018). Its goal is to "reduce uncertainties and risks of human and environmental safety of nanotechnology, starting as early as possible during the innovation process, on the basis of mandatory and voluntary safety and efficacy compliance requirements" (Soeteman-Hernandez et al., 2019). The main elements of the general SbD approach are: (1) it uses a stage-gate innovation approach, (2) it is based on three pillars, which are Safe materials and products, Safe production, and Safe use and end-of-life; (3) it includes SbD action for maximizing safety while maintaining functionality, and (4) it is integrated into a Safe Innovation Approach (see extensive description in Kraegeloh et al., 2018; Soeteman-Hernandez et al., 2019). Below we show how we changed or adapted these elements of the general SbD approach to nanomedicines (the comparison can be seen in **Table 1**).

As can be seen in **Table 1**, the GoNanoBioMat SbD approach is not based on a stage-gate innovation approach. Instead, it is an iterative approach. This decision was made in order to better represent the reality of "drug discovery and development" field, which also uses an iterative approach (Hjorth et al., 2017). In addition, the iterations are necessary to build up knowledge, as

<sup>2</sup>The GoNanoBioMat project was initiated by the ProSafe initiative

TABLE 1 | Comparison of the general SbD approach developed by NANoREG, NanoReg2, and the ProSafe initiative with the GoNanoBioMat SbD approach.


it will be shown in the section "GoNanoBioMat SbD Approach," on physico-chemical properties and their biological effects. This is because currently, it is not possible to predict these effects only based on literature and modeling.

The GoNanoBioMat SbD approach is also based on three pillars (**Table 1**), but these were modified to match the scope of the topic at hand. The pillar Safe Nanobiomaterials corresponds to the first pillar of the general SbD approach and has the same aim. In the general and GoNanoBioMat SbD approaches, the second pillar is Safe Production. However, the focus in the GoNanoBioMat SbD approach is not only on the safety of workers but also on ensuring safety and quality of the NBMs and on applying good manufacturing practices (GMP), which are a prerequisite to produce medicines and consequently nanomedicines. On the one hand, the third pillar of the general SbD approach is about Safe use and end-of-life. Its main goal is to prevent exposure during use and to have adapted recycling and disposal routes. On the other hand, the third pillar of the GoNanoBioMat SbD approach is about Safe Storage and Transport in order to ensure the safety and quality of NBMs because they may experience transformations (Cobaleda-Siles et al., 2017), which may affect their safety and quality (USP36–NF31, 2012; European Commission, 2013). Storage, and more particularly shelf-life, is an aspect being highly connected to the logistics and costs of the final nanomedicine and therefore its viability on the market (Diven et al., 2015). As can be seen, here the pillar is a bit narrower than in the general approach. This is to better represent the needs for developing nanomedicines.

Both approaches include SbD actions (**Table 1**). The difference between the two is that the functionality is specified into efficacy in the GoNanoBioMat SbD approach. It was changed into efficacy because efficacy is a measurement of the successful pharmacological effect of a drug and therefore more representative for developing nanomedicines. The goal of these SbD actions is to maximize safety while optimizing efficacy and costs by comparing different forms of NBMs. However, it should be pointed out that sometimes it is not feasible to maximize both efficacy and safety at the same time (Soeteman-Hernandez et al., 2019). Optimization will always require iterations in order to be able to balance efficacy and safety (Hjorth et al., 2017).

Finally, in **Table 1** it is possible to see that the general SbD approach is integrated into a Safe Innovation Approach and provides a Trusted Environment (Soeteman-Hernandez et al., 2019). The Safe Innovation Approach combines the SbD concept and the Regulatory Preparedness concept. The Regulatory Preparedness concept being the improvement of anticipation of regulators to keep up with the fast growing knowledge on nanomaterials and thus facilitate the development of adaptable regulations. The Trusted Environment is a space for enabling a dialogue among stakeholders and regulators for sharing and exchanging knowledge on nanomaterials. The GoNanoBioMat SbD approach, however, is embedded into and sets the frame for a document whose title is "Guidelines for implementing a SbD approach for medicinal polymeric nanocarriers" written and published by the GoNanoBioMat project consortium (Som et al., 2019). The guidelines provide the state of scientific knowledge with meta-analyses, decision trees, methods for producing NBMs, relevant endpoints to test, and safety aspects to consider early and throughout the development of polymeric NBMs for drug delivery. The guidelines can be downloaded under this link: www.empa.ch/gonanobiomat.

# GoNanoBioMat SbD APPROACH

As mentioned, the GoNanoBioMat SbD approach is a methodological approach for developing nanomedicines with a focus on polymeric NBMs for drug delivery and is presented in **Figure 1**. It contains the following steps: Material Design, Characterization, Human Health and Environmental Risks (first pillar), Manufacturing and Control (second pillar), and Storage and Transport (third pillar). The regulatory framework for developing nanomedicines is also included within the approach starting at the end of the Material Design step. The bullet points inside the boxes correspond to methods and tools that can be used or endpoints that should be considered and tested in each step. The blue arrows represent the flow of polymeric NBMs from their design until their storage and transport. The red arrows are feedback loops (iterations) going back to the Material Design step.

It is important to note, that most of these steps also apply to other NBMs and other type of nanomedicine applications. For example, the Human Health and Environmental Risks steps

could be applied to any type of NBMs. However, in the Material Design step and in the Characterization step, specific questions (e.g., what is the type of drug and what is the release kinetics) for drug delivery and specific parameters to characterize polymers are provided, respectively. Therefore, the total of questions only applies to polymeric NBMs for drug delivery, even if many questions also apply to other NBMs or other applications.

effect concentration.

As can be seen in **Figure 1**, the GoNanoBioMat SbD approach starts with the Material Design step. This step is divided into three sub-steps, which are (a) set the context and generate ideas, (b) define material properties and screen for unwanted toxicity and efficacy, and (c) produce the prototype.

In the first sub-step, a set of questions can be used to guide the conceptual process for developing NBMs for drug delivery, and searching for the relevant literature. The questions include the type of application, type of drug (possibility of chemical interaction between drug and polymer), administration route, the biological barriers, target cells, release kinetics, and dose needed. All these aspects influence the design of nanocarriers (Elsabahy and Wooley, 2012), in other words, its physicochemical properties to be efficient as a drug delivery system and lining up for safe application. An important consideration to bear in mind is that the properties of the polymer (particles larger than 1 micron) may not be equal to the properties of the polymer when the size of its particles is reduced to the nanoscale. Once the data from literature are collected, the data can be used to screen for efficacy but also toxicity and to define the wished material properties of the nanocarriers (second sub-step) by using modeling tools (i.e., non-testing tools), such as quantitative structure–activity relationship tools (OECD, 2007). These tools have for aim to find a correlation between NBMs properties and their corresponding

effect (e.g., cell internalization, cytotoxicity) and may enable to assess whether a material is safe for medical purposes. However, it has to be noted that such methods still need to be further developed.

As aspects of safety and functionality should be taken into account at the very beginning of the project's conception (Cobaleda-Siles et al., 2017), these two sub-steps based on literature and modeling are facilitating their consideration. However, assessing the human health risks in an early stage of innovation only based on data found in the literature is currently not adequate. This may be a result of the lack of standardized assays, which lead to a high variation in reported studies (Hofmann-Amtenbrink et al., 2015). Also some studies have no proper characterization and lack appropriate controls specific to the nanoscale (Jesus et al., 2019), which makes comparisons between toxicity outcomes difficult. Therefore, experimental studies are still needed.

After these two sub-steps, comes the first SbD action. Its goal is to compare different possible NBMs for the intended use/application, which was defined in the beginning of the Material Design step, and to select the NBMs having a good balance between, safety, efficacy, and costs. After this, the selected NBMs should be produced as prototypes.

These prototypes should be then characterized in order to be able to find relationships between physicochemical properties of NBMs and their biological effects, and thus apply the concept of SbD. As can be seen in **Figure 1**, the properties attributed to the polymer itself (e.g., molecular weight) and the properties attributed to the nanosize (e.g., size) should be characterized. If the desired properties of the prototypes do not correspond to the measured properties, the prototypes should go back to the prototype production sub-step in order to optimize the production process. One criterion in SbD requires understanding the variables contributing to undesired side effects (Lin et al., 2018). Therefore, to have a thorough characterization of polymeric NBMs, the Characterization step includes specific parameter to be tested for polymers NMBs, such as molecular weight, size and surface area. This step is also essential to determine later the CQAs, which are defined as "physical, chemical, biological, or microbiological properties or characteristics that should be within an appropriate limit, range, or distribution to ensure the desired product quality" (ICH Q8 (R2), 2009).

The next two steps are experimental steps to evaluate the human health and the environmental risks of the selected NBMs. For both, the exposure and the hazard should be evaluated. For the Human Health Risks step, the route of administration/exposure, the dosage, the duration and frequency should be determined as safety of NBMs depends on the route of administration/exposure and the resulting respective pharmacokinetic profiles (Jesus et al., 2019). For the hazard, the following endpoints should be tested: immunotoxicity, biocompatibility, carcinogenicity, mutagenicity, genotoxicity, toxicity on reproduction, acute, repeated, or chronic toxicity studies. All tested endpoints should as well include appropriate controls for the nanoscale. The proposed endpoints are in line with current regulation (Jesus et al., 2019).

In parallel, the assessment of the environmental risks should be performed. To do so, the predicted environmental concentration and the predicted no effect concentration have to be calculated (Hauser et al., 2019). The former can be assessed via a material flow analysis and the latter via performing a (probabilistic) species sensitivity distribution. For this, ecotoxicity data are needed, which can be obtained either via literature or experimentally by following OECD guidelines.

After the Human Health and Environmental risks steps comes the second SbD action. As for the first one, the goal is to compare the selected NBMs and choose the one maximizing safety, while optimizing efficacy and costs. At this point, either one NBM is selected as the final candidate or if no NBMs have a good balance between benefits and risks, the developer should go back to the Material Design step. The results of these two steps can help to build up a useful database. In other words, with iterations, a database with the experimental results could be established and these data could be used for modeling. Ultimately, it would enable better predictions of NMBs' efficacy and toxicity.

If one final candidate has been selected, the developer of NBMs should go to the Manufacturing and Control step. The goal of this step is to scale-up the production by applying GMP, preventing contamination and ensuring uniformity between the batches. In this step, CQAs of NBMs must be identified as well as Critical Process Parameters. These are defined as the "process parameters that influence CQAs and therefore should be monitored or controlled to ensure the process produces the desired quality" (ICH Q8 (R2), 2009). It can be noted that this step is typically valid for any type of NBMs.

After scale-up, usually the nanocarrier and their encapsulated drug system would go to clinical trials. However, as we did not include clinical trials in the approach because it was out of the scope of the project, the next step is Storage and Transport. The (nano)medicine stability studies have to be performed (SME Office, 2016; MDR, 2017), because nanocarriers and encapsulated drug, both, or just one of them, might experience degradation process during their life cycle, which might affect the quality and safety of the nanomedicine (Cobaleda-Siles et al., 2017).

Finally, the Swiss and European regulatory frameworks for the marketing authorization of nanomedicine is embedded within the GoNanoBioMat SbD approach. More information on this aspect can be directly found in the GoNanoBioMat guidelines.<sup>3</sup>

## DISCUSSION

In case of nanomedicines, SbD approaches should be included in the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) guidelines and relevant OECD guidance and guidelines. ICH is unique in bringing together the regulatory authorities and pharmaceutical industry to discuss scientific and technical aspects of drug registration and thus to discuss what should be included within guidelines concerning the safety of the NBMs and nanomedicines.

<sup>3</sup>www.empa.ch/gonanobiomat

The GoNanoBioMat SbD approach is methodological, contains all important elements to consider in order to integrate safety early and throughout the development of polymeric NBMs for drug delivery. It can as well to a certain extent be applied to other types of NBMS and nanomedicine applications. Including safety in the design of NBMs is an important aspect, especially for nanomedicines, which are highly regulated, cost and time consuming, and complex. However, the approach should not be seen as a warranty of complete safety, because absolute safety is unreachable (Cobaleda-Siles et al., 2017; Hjorth et al., 2017; van de Poel and Robaey, 2017), and should therefore be considered as a design strategy (van de Poel and Robaey, 2017) since the past showed that each nanomedicine has to be taken as caseby-case (McNeil, 2009). As for the general SbD concept, it has no legal binding and does not replace regulatory requirements (Soeteman-Hernandez et al., 2019).

The GoNanoBioMat SbD approach was focusing only on the safety of nanocarriers (polymeric NBMs) and not on the nanocarriers and its encapsulated drug. For regulatory purpose, it is necessary to test the safety of the nanocarrier alone in addition to the nanocarrier/drug system. Therefore, this GoNanoBioMat SbD approach is a first step toward the integration of safety early in the development of such products. Efficacy, which is closely related to the drug used, could not be included in the approach and therefore must be evaluated case-by-case. In a future development of the GoNanoBioMat SbD approach adding steps for clinical trials and use will be developed.

Finally, we believe that the GoNanoBioMat SbD approach as presented here may facilitate the implementation of the general SbD concept and to find a balance between benefits

#### REFERENCES


and risks by comparing different nanocarrier candidates in terms of their respective safety, efficacy, and costs. For instance, the GoNanoBioMat SbD approach provides all relevant steps for developing polymeric NBMs; provides methodology and endpoints to test human health and environmental risks, which are in line with current regulations; is an iterative process; and combines the nanoscale and medicine field under one methodological approach. In addition, the approach may bring the different actors of the value chain on a common ground. Ultimately, the approach may enable to move toward safe and efficient NBMs, safe production, and safe storage and transport.

#### AUTHOR CONTRIBUTIONS

MS created the GoNanoBioMat SbD approach in collaboration with CS, OB, SJ, PW, GB, GP, ÄS, and LS-H. MS wrote the manuscript. CS wrote some part of the manuscript. OB, SJ, PW, GB, GP, ÄS, and LH read and reviewed the manuscript. All authors approved the submitted version.

#### FUNDING

This study is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016; CTI (1.1.2018 Innosuisse), under grant agreement Number 19267.1 PFNM-NM; and FCT Foundation for Science and Technology under the project PROSAFE/0001/2016.



agile system for dealing with innovations. Mater. Today Commun. 20:100548. doi: 10.1016/j.mtcomm.2019.100548


**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 Schmutz, Borges, Jesus, Borchard, Perale, Zinn, Sips, Soeteman-Hernandez, Wick and Som. 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.

# Permeation of Biopolymers Across the Cell Membrane: A Computational Comparative Study on Polylactic Acid and Polyhydroxyalkanoate

Tommaso Casalini<sup>1</sup> \*, Amanda Rosolen<sup>1</sup> , Carolina Yumi Hosoda Henriques<sup>1</sup> and Giuseppe Perale1,2

<sup>1</sup> Polymer Engineering Laboratory, Department of Innovative Technologies, Institute for Mechanical Engineering and Materials Technology, University of Applied Sciences and Arts of Southern Switzerland, Manno, Switzerland, <sup>2</sup> Ludwig Boltzmann Institute for Experimental and Clinical Traumatology, Vienna, Austria

Polymeric nanoparticles, which by virtue of their size (1–1000 nm) are able to penetrate even into cells, are attracting increasing interest in the emerging field of nanomedicine, as devices for, e.g., drugs or vaccines delivery. Because of the involved dimensional scale in the nanoparticle/cell membrane interactions, modeling approaches at molecular level are the natural choice in order to understand the impact of nanoparticle formulation on cellular uptake mechanisms. In this work, the passive permeation across cell membrane of oligomers made of two employed polymers in the biomedical field [poly-D,L-lactic acid (PDLA) and poly(3-hydroxydecanoate) (P3HD)] is investigated at fundamental atomic scale through molecular dynamics simulations. The free energy profile related to membrane crossing is computed adopting umbrella sampling. Passive permeation is also investigated using a coarse-grained model with MARTINI force field, adopting welltempered metadynamics. Simulation results showed that P3HD permeation is favored with respect to PDLA by virtue of its higher hydrophobicity. The free energy profiles obtained at full atomistic and coarse-grained scale are in good agreement each for P3HD, while only a qualitative agreement was obtained for PDLA. Results suggest that a reparameterization of non-bonded interactions of the adopted MARTINI beads for the oligomer is needed in order to obtain a better agreement with more accurate simulations at atomic scale.

#### Keywords: molecular dynamics, lipid bilayer, permeation, molecular modeling, biopolymers

# INTRODUCTION

The detailed knowledge of drug/membrane interactions plays a key role for the determination of the ADME (adsorption, distribution, metabolism and excretion) profile of active compounds. The efficacy of an administered drug also depends on its ability to cross cellular membranes or barriers of biological interest, such as the blood–brain barrier, to reach the desired target. Membrane permeation can occur through different mechanisms; passive diffusion (i.e., membrane crossing due to the concentration gradient) plays a key role when small uncharged molecules are involved (Smith et al., 2014) and its detailed understanding is essential for drug design. There are established experimental techniques and protocols for investigating drug permeation in model membranes, but their limited spatial resolution does not allow shedding light behind the specific interactions. Simulations at fundamental molecular level emerged as the ideal tool

#### Edited by:

Emilio Isaac Alarcon, University of Ottawa, Canada

#### Reviewed by:

Jeffrey Robert Comer, Kansas State University, United States Ariela Vergara, University of Talca, Chile

> \*Correspondence: Tommaso Casalini tommaso.casalini@supsi.ch

#### Specialty section:

This article was submitted to Biomaterials, a section of the journal Frontiers in Bioengineering and Biotechnology Received: 01 December 2019

Accepted: 08 June 2020 Published: 30 June 2020

#### Citation:

Casalini T, Rosolen A, Henriques CYH and Perale G (2020) Permeation of Biopolymers Across the Cell Membrane: A Computational Comparative Study on Polylactic Acid and Polyhydroxyalkanoate. Front. Bioeng. Biotechnol. 8:718. doi: 10.3389/fbioe.2020.00718

to improve our knowledge, thanks to the detail at atomic scale that allows highlighting the most relevant interactions behind the observed or expected permeation rate (Di Meo et al., 2016; Shinoda, 2016). A lipid bilayer is a heterogeneous environment because of the presence of polar head groups and hydrophobic chains (Nagle and Tristram-Nagle, 2000). These aspects can be accounted for, in detail, by means of simulations at molecular level, which allow developing mechanistic interpretations and models for lipophilic compounds permeation, as widely discussed by Dickson and coworkers (Dickson et al., 2017). In this regard, the growing use of computational techniques such as molecular dynamics (MD) simulations is the result of several aspects: First, the increasing availability of computational resources, coupled with software optimization, which lead to affordable and meaningful simulations. Second, the continuous development and improvement of accurate force fields tailored for lipid bilayers; indeed, the reliability of MD simulations outcomes is strongly dependent on the robustness of the chosen force field, whose importance cannot be underestimated. Third, it should be mentioned that membrane permeation usually involves an energy barrier much higher than the thermal energy kBT (where k<sup>B</sup> is Boltzmann constant and T is absolute temperature) available to molecule in standard simulation at temperature T. This implies that a membrane crossing event would rarely be observed in a standard MD simulation, while multiple events should occur in a simulation in order to obtain statistically meaningful results. In other words, the characteristic time scale of molecule diffusion is larger than the time scale accessible to MD simulations. This issue can be overcome by means of enhanced sampling methods, which enhance the transition between metastable states separated by free energy barriers higher than kBT. The most popular method for drug/membrane interactions is umbrella sampling (US) (Torrie and Valleau, 1977), which allows obtaining the potential of mean force (PMF) as a function of a relevant reaction coordinate, usually taken as the distance between the center of the membrane and the center of mass of the molecule of interest. Position-dependent diffusion coefficients and permeation coefficients can be also obtained through the inhomogeneous solubility-diffusion model (ISDM) (Marrink and Berendsen, 1994). Such protocol is still widely employed nowadays for different systems of interest (Bochicchio et al., 2015; Dickson et al., 2017, 2019; Teixeira and Arantes, 2019). Another useful technique is constituted by well-tempered metadynamics (WTMD) (Barducci et al., 2008); briefly, WTMD allows recovering the free energy landscape of the system of interest as a function of few relevant degrees of freedom [commonly referred as collective variables (CV)] by adding a time-dependent bias. WTMD attracted some interest for the study of the permeation of small molecules, because of its increased computational efficiency with respect to US and to the possibility to add easily a bias potential to other CV that can play a role in membrane permeation, such as permanent orientation or intramolecular hydrogen bonds (Minozzi et al., 2011; Jambeck and Lyubartsev, 2013; Loverde, 2014; Saeedi et al., 2017). Simulations usually consider the interaction of a single drug molecule with a model membrane, usually made of dioleoylphosphatidylcholine (DOPC) or dipalmitoylphosphatidylcholine (DPPC), thanks to the availability of validated force fields (Dickson et al., 2014; Ingolfsson et al., 2016; Frederix et al., 2018). The use of a model membrane is an accepted approximation; adopting more realistic models still suffers from the lack of experimental data needed to validate force field parameters (Poger et al., 2016) but there is an increasing number of examples of heterogeneous membranes in literature. A common solution is the addition of cholesterol or other molecules in the model membrane (Murzyn et al., 2005; Hoopes et al., 2011; Tse et al., 2018). Recently, Tse et al. (2019) proposed a full atomistic model of a mammalian cell membrane, which contains 26 different components. The same considerations can be in principle extended also to biomaterials/membrane interactions, whose simulations are attracting an increasing interest because of the new paradigms introduced by nanomedicine. Indeed, simulations at fundamental molecular level, due to the involved time and length scales, are the natural modeling tool for improve our understanding of the interactions between nanocarriers (whose size is between 1 and 1000 nm) and biological components (proteins, carbohydrates, membranes, et cetera). Focusing on biomaterials/membrane interactions, on the one side, nanocarriers such as nanoparticles can cross the cellular membrane also through passive permeation. On the other side, when bioresorbable polymers are employed, degradation products can permeate through cellular membranes and accumulate into the cells, thus leading to adverse effects. Overall, this approach matches the requirements of the "safety by design" paradigm too. Because of the involved time and length scales, MD simulations with enhanced sampling methods are not always suitable to investigate nanoparticles/membrane interactions (Schulz et al., 2012; Casalini et al., 2019a) and coarse-grained (CG) models should be employed. As recently discussed (Ingolfsson et al., 2014; Lunnoo et al., 2019), CG models also allow including heterogeneous lipid bilayers, moving toward a more realistic description of the cellular membranes. Despite the loss of the atomic detail, they provide interesting insights if accurately parameterized against experimental data or full atomistic simulations (Marrink and Tieleman, 2013). Parameterization can be performed, e.g., by reproducing with a CG model the PMF of interested obtained with MD simulations (de Jong et al., 2013). In this work, we study by means of molecular dynamics simulations the diffusion across a DOPC model membrane of small oligomers made of poly-D,L-lactic acid (PDLA) and poly(3-hydroxydecanoate) (P3HD) chosen as representative compound of the family of polyhydroxyalkanoates (PHA). PDLA and PHA gained a wide interest in the biomedical field since they merge several interesting peculiarities, such as biocompatibility, good mechanical properties and an in situ degradation due to hydrolysis mechanism (Bassas-Galia et al., 2017; Butt et al., 2018; Casalini et al., 2019b). This led to the development of a wide range of biomedical devices, from bone fixation screws to nanoparticles for targeted drug delivery. On the one side, the excessive accumulation of degradation products inside cells may lead to adverse effects (Ramot et al., 2016); on the other side, a deeper understanding of the endocytic pathway for nanoparticle uptake can support the experimental design of new and more effective formulations. This constitutes the starting

point of this work, which is structured as follows. First, the free energy landscape related to the permeation of small PDLA and P3HD oligomers (representative of degradation products from polymer hydrolysis) is obtained adopting umbrella sampling. Membrane crossing is subsequently simulated adopting a coarsegrained model and the free energy landscape is computed by means of WTMD. The assessment of the suitability of a coarsegrained model, parameterized on more accurate simulations at atomic scale, is fundamental to investigate the permeation of entire nanoparticles in model membranes, which would not be feasible with full atomistic simulations due to the involved time and length scales.

# METHODS

#### Force Field Parameterization

The second-generation general amber force field (GAFF2) (Wang et al., 2004) was employed for PDLA and P3HD. Atomic charges were computed by means of restrained electrostatic potential (RESP) method (Bayly et al., 1993; Cornell et al., 1993), consistently with force field parameterization procedure. Oligomers composed of 6 monomer units were optimized in vacuo through density functional theory (DFT) calculations at B3LYP/6-31G(d,p) level of theory. The obtained conformations were subsequently employed to compute electrostatic potentials in vacuo at HF/6-31G<sup>∗</sup> level of theory. Calculations were performed my means of Gaussian09 software (Frisch et al., 2016). Atomic charges were then fitted by means of RESP procedure, adopting a two-step protocol. First, partial atomic charges were calculated starting from the previously obtain electrostatic potential values, imposing an overall charge value equal to zero. In the second step, charge equivalence is imposed for chemically equivalent atoms. This procedure allowed obtaining a library of building blocks that can be used to build polymer chains of different length. Lipid17 force field (Dickson et al., 2014) was adopted for DOPC lipid bilayer because of its validated parameters. TIP3P water model (Jorgensen et al., 1983) was employed for explicit solvent molecules, consistently with force field parameterization. Parameters for monovalent ions, optimized for TIP3P model, were taken from Joung and Cheatham (2008, 2009). Details are reported in **Supplementary Material**.

## Creation of the Molecular Models

Polymer chains were built using tLeap module included in AmberTools; chain ends were saturated with methyl groups. The same tool was used to solvate with TIP3P water molecules and add ions to assure electroneutrality, where needed. A DOPC lipid bilayer composed of 128 DOPC molecules was assembled and solvated by means of CHARMM-GUI web server (Wu et al., 2014; Lee et al., 2019). The membrane lies on xy-plane and water molecules were placed only along z direction, so that an infinite surface can be obtained by applying periodic boundary conditions. A suitable number of Na<sup>+</sup> and Cl<sup>−</sup> was added to reach 0.15 M salt concentration, which mimics physiological conditions. The bilayer/polymer system was assembled starting from equilibrated configurations of the single components by means of AddToBox module included in AmberTools; ions were added in order to mimic physiological conditions. A coarse-grained model was built adopting MARTINI force field (Marrink et al., 2007), chosen for its validated results and its straightforward parameterization procedure. A bilayer composed of 128 DOPC molecules was built by means of CHARMM-GUI web server, similarly to the full atomistic model, with explicit water and ions beads. Parameters for bonded and nonbonded interactions of DOPC molecules are already available in MARTINI library. PDLA was coarse-grained by adopting 7 C5 beads, while P3HD was modeled using 7 Na beads for the backbone and one C1 bead and one C3 bead for each side chain. Structures are depicted in **Figures 4A,B**.

Parameters for bonded interactions were computed to best reproduce the bond, angle dihedral distributions obtained with MD simulations at full atomistic level. Parameters for nonbonded interactions were taken from MARTINI library. Details are reported in **Supplementary Material**.

## Molecular Dynamics Simulations

Molecular dynamics simulations were performed according to the following protocol. First, energy minimization step procedure was carried out by fixing the solute with a harmonic restraint (force constant equal to 500 kcal mol−<sup>1</sup> Å −2 ), in order to remove bad solvent/solvent and solute/solvent contacts due to the random placement of water molecules. Energy minimization was subsequently repeated removing the restraint on solute molecules. Temperature was raised from 0 to 310 K by means of 20 ps in NVT ensemble (constant number of particles N, volume V, and temperature T). When the lipid bilayer was present in the simulation box, temperature was slowly increased from 0 to 310 K through 10 ns in NVT ensemble adopting a linear ramp. Solute was kept fixed through a weak harmonic restraint (force constant equal to 10 kcal mol−<sup>1</sup> Å −2 ); temperature was maintained to the desired production value by means of Langevin thermostat, adopting a collision frequency equal to 1 ps−<sup>1</sup> . Finally, system equilibration was achieved by means of molecular dynamics simulations in NPT ensemble (i.e., at constant number of particles N, pressure P, and temperature T) at 310 K and 1 atm. Pressure was controlled by means of isotropic (for polymer/water systems) and anisotropic (for systems containing lipid bilayer) Berendsen barostat. Simulations were performed adopting periodic boundary conditions; the chosen cutoff value for long-range interactions was set equal to 1 nm. Particle Mesh Ewald (PME) was chosen for treating electrostatic interactions. SHAKE algorithm was employed to constrain all covalent bonds involving hydrogen atoms; this allowed propagating system dynamics through Leap-Frog algorithm using a time step equal to 2 fs. Simulations were carried out with GPU cards using the pmemd.cuda module implemented in AMBER 16 (Salomon-Ferrer et al., 2013; Case et al., 2016). A summary of performed MD simulations is reported in **Supplementary Material**.

#### Umbrella Sampling

Umbrella sampling was performed by choosing the distance between the center of mass (COM) of oligomer chain and the

center of the lipid bilayer as the relevant reaction coordinate (Di Meo et al., 2016; Shinoda, 2016). Simulations were carried out using 41 windows, covering a distance range from 0 to 40 Å with a spacing value equal to 1 Å.

Oligomers were restrained to the reference distance of each window by means of a harmonic potential with a force constant equal to 2.5 kcal mol−<sup>1</sup> Å −2 , chosen so that a good overlap between distance distributions among adjacent windows could be obtained. Only the z component of the distance was subjected to the restraint, while oligomers were free to move along x and y directions.

First, the oligomer was placed in the center of the membrane by applying a harmonic potential and 40 ns MD simulation in NPT ensemble were carried out to reach equilibration. Then, Umbrella Sampling simulations were performed so that the oligomer was pulled out from membrane center to the external environment; indeed, it has been shown in scientific literature that this procedure (rather than gradually placing a molecule in the bilayer) improves the convergence of the results (Filipe et al., 2014); 80 ns MD simulations in NPT ensemble at 1 atm and 310 K were carried out for each window, leading to 3.2 µs of total sampling time for each system.

Free energy as a function of the chosen reaction coordinate was obtained by means of weighted histogram analysis method (WHAM) (Kumar et al., 1992; Roux, 1995), using a 0–40 Å distance grid with a grid spacing equal to 0.025 Å; a further decrease of grid spacing did not lead to appreciable variations of the obtained results.

For each window, the first 50 ns were used for system equilibration and discarded; free energy was computed using the last 30 ns, using three blocks of 10 ns each. Results are expressed as average ± standard deviation. Details are reported in **Supplementary Material**.

A position-dependent diffusion coefficient can be computed by means of inhomogeneous solubility-diffusion model (Marrink and Berendsen, 1994), which was applied in literature to small solutes (water, methanol, etc.) (Bemporad et al., 2004; Orsi et al., 2009; Lee et al., 2016) as well as to drug-like molecules (Dickson et al., 2017). Diffusivity as a function of z coordinate D(z) can be computed as follows (Hummer, 2005):

$$D\left(z\right) = \frac{\nu a r(z)^2}{\int\_0^\infty \mathcal{C}\_{\text{xz}}\left(t\right) dt} \tag{1}$$

where var(z) is the variance of the z-component of the distance in a US window and Czz is the position autocorrelation function, defined as follows:

$$C\_{\rm xz}(t) = \delta z(0)\delta z(t) \tag{2}$$

$$
\delta z\left(t\right) = z\left(t\right) - z\tag{3}
$$

where <z> is the average value of the distance in the US window.

Autocorrelation function was numerically integrated by means of trapz algorithm implemented in MATLAB (which takes advantage of the trapezoidal rule) until it decayed to var(z)·10−<sup>2</sup> , in order to exclude from the integration the noise around Czz(t) = 0 (Dickson et al., 2017). The position-dependent resistance can be also computed:

$$R\left(z\right) = \frac{\exp\left(\beta \,\Lambda G\left(z\right)\right)}{D(z)}\tag{4}$$

where β is equal to (kBT) −1 , k<sup>B</sup> is Boltzmann constant, T is the absolute temperature, and 1G(z) is the free energy profile. While D(z) is evaluated for every window (41 values), 1G(z) is obtained for every grid point. Therefore, in order to compute R(z), the diffusion coefficient is evaluated along the distance grid using a shape-preserving interpolant by means of mpich algorithm implemented in MATLAB. An overall permeation coefficient can be obtained by integrating the resistance profile:

$$P\_{\text{eff}} = \frac{1}{R\_{\text{eff}}} = \frac{1}{\int\_{-z\_{\text{B}}}^{z\_{\text{B}}} R \, (z) \, dz} \tag{5}$$

where the integration boundaries are referred to water phase at either side of the lipid bilayer (i.e., z<sup>B</sup> = 40 Å and –z<sup>B</sup> = -40 Å). Binding free energy 1G 0 Bind and membrane partitioning constant Klip can be also obtained:

$$
\Delta G\_{\text{Bind}}^{0} = -k\_{\text{B}}T \ln \left( \frac{1}{z\_{\text{B}}} \int\_{0}^{z\_{\text{B}}} \int \exp \left( -\beta \Delta G \left( z \right) \right) dz \right) \tag{6}
$$

$$K\_{lip} = \exp(-\beta \Delta G\_{\text{Bind}}^0) \tag{7}$$

#### Coarse-Grained Simulations

Simulations were performed with GROMACS 2018.3 (Pall et al., 2015). Focusing on DOPC bilayer, after energy minimization the system was progressively equilibrated by running five simulations of 10 ns each in NPT ensemble at 310 K and 1 atm. A harmonic restraint was applied to lipid molecules, reducing the value of the force constant in each simulation; force constant values equal to 200, 100, 50, 20 and 10 kJ nm−<sup>2</sup> mol−<sup>1</sup> were chosen for this purpose. Temperature and pressure were controlled by means of velocity rescaling algorithm (Bussi et al., 2007) and semiisotropic Berendsen barostat, respectively, with coupling time constants equal to 1 and 12 ps. Finally, 600 ns in NPT ensemble at 310 K and 1 atm was performed, adopting velocity rescaling algorithm and semiisotropic Parrinello-Rahman barostat (Parrinello and Rahman, 1981) for temperature and pressure control, respectively. Coupling time constants were not modified. A cutoff value equal to 1.1 nm was chosen for longrange electrostatic and Van der Waals interactions, which were computed adopting a reaction field (with a dielectric constant beyond the cutoff equal to 15) and a straight cutoff. A potential modifier was applied to VdW interactions to better perform with the Verlet cutoff scheme. Periodic boundary conditions were applied along x, y, and z directions; dynamics were propagated using Leap-Frog algorithm using a time step equal to 20 fs.

WTMD simulations were carried out with GROMACS 2018.3 patched with PLUMED 2.5 (Tribello et al., 2014). PDLA and P3HD were added in the water phase in the simulation box of the equilibrated DOPC bilayer, replacing water beads if necessary, with the insert-molecule tool implemented in GROMACS. After energy minimization and a brief equilibration (20 ns) in NPT

ensemble at 310 K and 1 atm, WTMD simulations were carried out, considering the component along z direction of the distance between oligomer and bilayer centers of mass. The initial Gaussian height, sigma, and bias factor values were set equal to 0.8 kJ mol−<sup>1</sup> , 0.05 Å, and 30, respectively. Bias potential was added every 5000 steps (100 ps). Harmonic potentials were applied by means of upper\_walls and lower\_walls algorithms implemented in PLUMED (force constant equal to 50 kJ mol−<sup>1</sup> nm−<sup>2</sup> ) in order to promote membrane crossing events and to limit the CV exploration in a range of values of interest, i.e., between -45 and 45 Å.

The convergence of the free energy landscape was evaluated with two different methods, that is, checking the free energy difference as a function of simulation time and computing the error according to Bonomi et al. (2009):

$$\varepsilon^{2}\left(t\right) = \frac{1}{\nu o l(\Omega)} \int ds \left[V\left(s, t\right) - F(s)\right]^{2} \tag{8}$$

where ε is the error, t is time, s represents the chosen collective variables, is the explored CV region, V(s,t) is the external bias added to the system, and F(s) is the reference free energy profile, i.e., the one obtained at the end of the simulation. Plots are reported in **Supplementary Material**.

Free energy profiles as well as binding free energies were computed as an average of the last 2000 ns.

#### RESULTS AND DISCUSSION

#### Oligomers and Lipid Bilayer Equilibration

First, MD simulations were carried out in order to obtain equilibrated structures of both oligomers and the DOPC lipid bilayer, which mimics a cellular membrane. Each oligomer was equilibrated with 50 ns MD simulations in NPT ensemble at 1 atm and 310 K (**Figures 1A,B**). The attainment of reasonable equilibrated structures was checked by computing the root

mean square displacement (RMSD) and the solvent accessible surface area (SASA) as a function of simulation time, as shown in **Figures 1C,D**. While PDLA oligomer did not experience substantial structural variations, P3HD oligomer folded due to its increased hydrophobicity related to aliphatic side chains. Focusing on DOPC bilayer, 150 ns MD simulations were performed for equilibration and the attainment of an equilibrated structure (**Figure 1E**) was verified by computing the area per lipid (**Figure 1F**) and membrane thickness (computed from the peak-to-peak distance of electron density profiles) as a

function of simulation time. Equilibration led to an area per lipid and membrane thickness values equal to 72.04 ± 0.88 Å<sup>2</sup> lipid−<sup>1</sup> and 35.75 ± 0.45 Å, respectively; values are expressed as average ± standard deviation. The obtained values are consistent with the reported computational and experimental data provided by Dickson et al. (2014).

The equilibrated structures were thus employed for the study of oligomers permeation in the lipid bilayer.

#### Oligomers Permeation

The main outcome from Umbrella Simulations is the free energy landscape as a function of the z-component of the distance between the center of mass of the oligomer and the center of the membrane. In this regard, it is possible to identify three different zones, related to the heterogeneous environment of the bilayer: tail groups (0 < z < 13 Å), head groups (13 < z < 27 Å) and water phase (27 < z < 40 Å). Results are shown in **Figure 2**. PDLA and P3HD free energy landscapes are consistent with the results shown in literature for hydrophobic molecules (Bemporad et al., 2004; Orsi et al., 2009; Bochicchio et al., 2015), since such oligomers preferably partition inside the membrane. Indeed, 1G 0 Bind computed through equation 6 is equal to - 11.49 ± 0.69 and -23.85 ± 0.99 kcal mol−<sup>1</sup> for PDLA and P3HD, respectively. The more favorable value related to P3HD is due to the relevant interactions between polymer/bilayer aliphatic chains. The minimum of the free energy lies in the region with the tail groups, by virtue of hydrophobic effects. Free energy increases

moving toward the hydrophilic head groups, where no favorable interactions take place since no hydrogen bonds can be formed.

coarse-grained simulations for PDLA (E) and P3HD (F). Profile from full atomistic simulations was mirrored for the sake of clarity.

Hydrophobic effects behind the free energy landscape can be highlighted through SASA values, computed using the last 10 ns of each window, as shown in **Figures 3A,B**; indeed, free energy profiles for PDLA and P3HD exhibit the same trend of SASA decrease due to permeation. Notably, P3HD oligomer also experiences unfolding inside the bilayer (**Figure 3C**), when it is surrounded by the hydrophobic tails. In addition, P3HD is still unfolded at the bilayer/water interface (**Figure 3D**); aliphatic chains point toward the lipid bilayer, while the backbone is exposed to the solvent. Up to authors' best knowledge, experimental diffusion coefficients are not available, while computational studies are usually focused on smaller molecules. Comparison with literature data reveals that P3HD and PDLA diffusion coefficients are about two orders of magnitude lower if compared to low molecular weight compounds (ranging from water to benzene) or small drugs (Orsi et al., 2009; Dickson et al., 2017) and can be considered acceptable. The resistance as a function of collective coordinate reaches it minimum value in the center of the bilayer (by virtue of the favorable interactions) and it is maximum at water/bilayer interface. Indeed, polar head groups are the major obstacle to permeation, due to the not favorable interactions with the oligomers. All computed values are reported in **Supplementary Material**.

#### Coarse-Grained Simulations

The first step was evaluating the attainment of an equilibrated bilayer structure at CG level (**Figure 4C**) and its agreement with the outcomes from atomistic simulations. The average values of area per lipid and membrane thickness are equal to 68.57 ± 1.25 Å<sup>2</sup> lipid−<sup>1</sup> and 36.67 ± 0.57 Å, respectively, in

good agreement with the results obtained at full atomistic level. It should also point out that in this case membrane, thickness was computed from the distance between the beads representative of the phosphate groups. Moreover, an equilibrated structure is rapidly obtained, as shown by the time evolution of the area per lipid (**Figure 4D**).

Free energy landscapes were obtained by means of WTMD; thanks to the higher accessible time scales provided by the intrinsic computational efficiency of CG simulations with respect to full atomistic ones, the sampling was performed considering the full CV range from -40 to 40 Å, in order to observe multiple membrane crossing events. The comparison between free energy profiles from full atomistic and coarsegrained simulations is shown in **Figures 4E,F** for PDLA and P3HB, respectively.

While the agreement for P3HD is good from both a qualitative and a quantitative point of view, only a fair qualitative agreement was obtained for PDLA. This is evident also focusing on 1GBind values, which were computed also from CG simulations using the last 2000 ns. The value obtained for P3HD, equal to - 22.69 ± 0.23 kcal mol−<sup>1</sup> is in good agreement with the estimation from US, equal to -23.85 ± 0.99 kcal mol−<sup>1</sup> . On the other hand, the analogous comparison for PDLA oligomer showed an expected poor agreement, by virtue of 1GBind values equal to -8.33 ± 0.11 and -11.49 ± 0.69 kcal mol−<sup>1</sup> obtained from coarse-grained and full atomistic simulations, respectively.

The observed disagreement for PDLA results can be explained by taking into account the parameterization of nonbonded interactions of MARTINI beads. Indeed, PDLA has a backbone composed of ester bonds, which act as polar groups, and hydrophobic side chains constituted by methyl groups. The parameterization of the non-bonded interactions of the chosen MARTINI bead is thus not able to account for this balance, since the hydrophobicity in underestimated. Modeling PDLA polymer with Na MARTINI beads, representative of ester bonds only, leads to physically not consistent results: preliminary explorative simulations showed that the oligomer would preferably partition in water phase. On the other hand, more hydrophobic beads essentially take into account aliphatic backbones and would provide an overestimation of the affinity for the lipid phase.

Summarizing, while C5 MARTINI beads for PDLA polymer represent the best compromise, they do not provide a description of the polymer at CG level with an acceptable accuracy level. Therefore, while the CG model for P3HD presented here could be readily used to simulate an entire nanoparticle, a reparameterization of non-bonded interactions for PDLA oligomer is needed to improve the agreement with more accurate atomistic simulations.

#### CONCLUSION

In this study, the passive permeation of small oligomers of polymer of interest in the biomedical field was studied by means of molecular dynamics simulations, at both full atomistic and coarse-grained level.

Simulations at atomic scale allowed obtaining the free energy landscape as a function of the distance between the center of the membrane and the center of mass of PDLA and P3HD, chosen as collective coordinate. Results showed that both oligomers preferably partition into the membrane; this trend could be explained in terms of hydrophobic effects by computing the solvent accessible surface area as a function of the collective coordinate.

The obtained free energy landscape can be in principle employed to tune a coarse-grained model, which can be used to simulate the permeation of an entire nanoparticle into a lipid bilayer, by virtue of the higher accessible time and length scales. For this reason, coarse-grained simulations were performed using MARTINI force field, to check whether the free energy landscape from atomistic simulations could be reproduced without further reparameterization.

Results showed a good quantitative agreement for P3HD oligomer and only a fair qualitative agreement for PDLA, highlighting the need of a further reparameterization of non-bonded interactions in order to better account for the hydrophobicity due to the methyl side groups.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

# AUTHOR CONTRIBUTIONS

TC, AR, and CH performed simulations and post processing. TC wrote the first draft of the manuscript. GP contributed to supervision of the work. All authors discussed and approved the contents of the manuscript and contributed to its final version by reading and editing.

# FUNDING

This study is part of the GoNanoBioMat project and has received funding from the Horizon 2020 framework program of the European Union, ProSafe Joint Transnational Call 2016, from the Swiss KTI (from 1.1.2018 Swiss Innosuisse) under Grant Number 19267.1 PFNM-NM and from EU FCT Foundation for Science and Technology under the project PROSAFE/0001/2016.

# ACKNOWLEDGMENTS

TC acknowledges the contribution of Michela Castelnuovo, B. Des., for image editing.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe.2020. 00718/full#supplementary-material

#### REFERENCES

fbioe-08-00718 June 27, 2020 Time: 19:53 # 9



**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 Casalini, Rosolen, Henriques and Perale. 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.