**Supramolecular Systems for Gene and Drug Delivery**

Editors

**Francisco Jos´e Ostos Jos ´e Antonio Lebr ´on Pilar L ´opez-Cornejo**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Francisco Jose Ostos ´ University of Seville Spain

Jose Antonio Lebr ´ on´ University of Seville Spain

Pilar Lopez-Cornejo ´ University of Seville Spain

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Pharmaceutics* (ISSN 1999-4923) (available at: https://www.mdpi.com/journal/pharmaceutics/ special issues/supramolecular gene drug delivery).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-3377-3 (Hbk) ISBN 978-3-0365-3378-0 (PDF)**

© 2022 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**


#### **Fasih Bintang Ilhami, Enyew Alemayehu Bayle and Chih-Chia Cheng**

Complementary Nucleobase Interactions Drive Co-Assembly of Drugs and Nanocarriers for Selective Cancer Chemotherapy Reprinted from: *Pharmaceutics* **2021**, *13*, 1929, doi:10.3390/pharmaceutics13111929 ........ **175**

## **Anna Jagusiak, Katarzyna Chłopa´s, Grzegorz Zemanek, Izabela Ko´scik and Irena Roterman**

Interaction of Supramolecular Congo Red and Congo Red-Doxorubicin Complexes with Proteins for Drug Carrier Design

Reprinted from: *Pharmaceutics* **2021**, *13*, 2027, doi:10.3390/pharmaceutics13122027 ........ **193**

## **About the Editors**

**Francisco Jos´e Ostos**, Ph.D., is a postdoctoral fellowship at the Clinical Unit of Infectious Diseases, Microbiology, and Preventive Medicine from the Institute of Biomedicine of Seville (IBiS) and School of Medicine from the University of Seville (since 2021). His research focuses on developing an innovative multidisciplinary approach at the interface of nanotechnology, supramolecular chemistry, cell biology, and immunology in order to eradicate viral infections. His research period resulted in the publication of impressive works in journals with a high impact factor, including 18 research articles, 2 book chapters, and 2 books. Furthermore, he presented inventions in 30 international/national meetings, and participated in 7 research projects. He received some awards (One secondary award from the Spanish Ministry of Education, Culture and Sport (XIV Certamen Universitario Arqu´ımedes), and three awards as best publication from the University of Seville.

**Jos´e Antonio Lebr ´on** has a research contract funded by the "Fundacion ONCE" in its ´ "Oportunidad al Talento" program at the "Professor Rodr´ıguez Velasco Kinetics Group (FQM 206)" of the University of Seville. His research has led him to publish papers in high-impact-factor journals, including 19 research articles. He has also participated in six research projects and two book chapters, and has published two books of his own. He has contributed 33 communications to national/international congresses. His research focuses on the synthesis and development of new drug nanocarriers (drugs or genetic material) for the treatment of diseases. The candidate has also participated in the organization of two congresses, one national 2017 "J2IFAM2017" and another international 2013 "(8th ICCK 2013)".

**Pilar L ´opez-Cornejo**, Professor of Physical Chemistry at the University of Seville. Her research is related to restricted geometry systems, macromolecules, colloids, and ligand–receptor interactions. Her research focuses on the development of biocompatible nanocarriers based on the use of calixarenes, metallomicelles, liposomes, CNTs, MOFs, and polymers for encapsulating nucleic acids, antineoplastic drugs, peptides, proteins, antiretrovirals, and toll-like receptors. Publications include 2 books, 8 book chapters, 95 research articles, and >50 meetings. She has participated in more than 20 projects, leading some of them. She is also the supervisor of three doctoral theses (+1 under progress) and is the reviewer of several scientific journals of high impact. She is also responsible for bilateral agreements, including Sevilla-Pisa and Sevilla-Hamburgo (Erasmus Program), and Seville-Tunisia (Erasmus +). As a member of the Colloids and Interface Group and the RSEQ, she has collaborated with different national and international research groups.

## *Editorial* **Supramolecular Systems for Gene and Drug Delivery**

**José A. Lebrón 1, Pilar López-Cornejo 1,\* and Francisco J. Ostos 2,3,\***


Several biomaterial-based supramolecular systems (cyclodextrins [1], calixarenes [2,3], polymers [4], carbon nanotubes [5], nanoparticles [6,7], liposomes [3,8], nanogels [9], and nanocomplexes [10], among others) have been widely used for biomedical applications, such as gene and drug delivery. Numerous researchers have developed novel supramolecular systems for enhancing their biocompatibility and pharmacological activity, thus increasing their therapeutic properties. These nanosystems are considered to be promising platforms in gene therapy and drug delivery due to their higher transfection (or encapsulation) efficiency and low cytotoxicity.

This Special Issue, "Supramolecular Systems for Gene and Drug Delivery", brings together the latest research articles, published in *Pharmaceutics*. Noticeably, 10 original research articles were published by authors from 12 different countries on what is a hot topic in this research field.

I. Asela et al. [1] prepared nanosponges based on β-cyclodextrin (βCDNS), which were loaded with the drugs phenylethylamine (PhEA) and 2-amino-4-(4-chlorophenyl)-thiazole (AT). Subsequently, the supramolecular βCDNS drug complexes were functionalized with gold nanoparticles (AuNPs), forming the βCDNS-PhEA-AuNP and βCDNS-AT-AuNP systems. The drug-loading capacity was higher for the βCDNS and βCDNS-drug-AuNP systems than with native βCD.

B. Gómez-González et al. [2] studied the formation of inclusion complexes between alkyl sulfonate guests and a cationic pillar [5] arene receptor in water using NMR and ITC measurements. The results demonstrated the formation of host–guest complexes stabilized by electrostatic interactions and hydrophobic effects.

J. A. Lebrón et al. [3] studied the formation of calixarene-based liposomes. Four amphiphilic calixarenes were used. The lipid bilayer was formed with one calixarene and with the phospholipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE). The liposomes containing the least cytotoxic calixarene (TEAC12)4 were used as nanocarriers of both nucleic acids and the antineoplastic drug doxorubicin (DOX). The results showed that (TEAC12)4/DOPE/p-EGFP-C1 lipoplexes, of a given composition, can transfect the genetic material, although the transfection efficiency substantially increases in the presence of an additional amount of DOPE as coadjuvant. On the other hand, the (TEAC12)4/DOPE liposomes showed a high doxorubicin encapsulation efficiency and a slow controlled release, which could diminish the side effects of the drug.

V. Karava et al. [4] prepared microparticles (MPs) based on newly synthesized poly(llactic acid)-co-poly(butylene adipate) (PLA/PBAd) block copolymers for the preparation of aripiprazole (ARI)-loaded long-acting injectable (LAI) formulations. In terms of in vitro dissolution profile, results suggested that the newly synthesized PLA/PBAd block copolymers can successfully control the release rate and extent of the API's release from the prepared MPs, indicating that, probably, under in vivo conditions, their use may lead

**Citation:** Lebrón, J.A.; López-Cornejo, P.; Ostos, F.J. Supramolecular Systems for Gene and Drug Delivery. *Pharmaceutics* **2022**, *14*, 471. https://doi.org/ 10.3390/pharmaceutics14030471

Received: 18 February 2022 Accepted: 21 February 2022 Published: 22 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

to new formulations that will be able to maintain a continuous therapeutic level for an extended period of time, with reduced lag time compared to the currently marketed ARI LAI product.

L. Tang et al. [5] successfully prepared a multi-walled carbon nanotube (MWNT) based drug delivery system with the synergistic effect of PTT photothermal therapy and chemotherapy for efficient tumor removal. The integration of photothermal agents ICG-NH2 to MWNT was achieved by linking hyaluronic acid (HA). To realize the synergistic therapeutic effect of chemotherapy and phototherapy, DOX was attached on the wall of MWNT via a π–π interaction to obtain the final MWNT-HA-ICG/DOX nanocomplexes. Both in vitro and in vivo experiments verified the great therapeutic efficacy of MWNT-HA-ICG/DOX nanocomplexes.

L. S. Mbatha et al. [6] formulated folic acid (FA)-modified, poly-amidoamine-generation-5 (PAMAM G5D)-grafted gold nanoparticles (AuNPs) and evaluated their cytotoxicity and transfection efficiency using the luciferase reporter gene (FLuc-mRNA) in vitro. These nanosystems showed low cytotoxicity and good transfection efficiency.

S. Yin et al. [7] prepared NPs based on the insertion of two types of functional peptides, half-life extension peptide PAS and tumor-targeting peptide RGDK (Arg-Gly-Asp-Lys), into human heavy-chain ferritin (HFn) at the C-terminal through flexible linkers with two distinct enzyme-cleavable sites. RGDK peptide enhanced the internalization efficiency of HFn and showed a significant increase in growth inhibition. Pharmacokinetic study in vivo demonstrated that PAS peptides extended ferritin half-life. RGDK peptides greatly enhanced drug accumulation in the tumor site, rather than in other organs, in a biodistribution analysis. Drug-loaded, PAS-RGDK-functionalized HFns curbed tumor growth with significantly greater efficacies in comparison with drug-loaded HFn.

C. E. Torres et al. [8] prepared magnetoliposomes (MLP), which are liposomes that contain magnetite nanoparticles (MNP) inside. This study presents a low-cost microfluidic approach for the synthesis and purification of MLPs to improve their biocompatibility, with functional testing via hemolysis, platelet aggregation, cytocompatibility, internalization, and endosomal escape assays to determine their potential application in gastrointestinal delivery. In addition, the authors achieved encapsulation efficiencies between 20% and 90% by varying the total flow rates (TFRs), flow rate ratios (FRRs), and MNP concentrations.

F. Bintang Ilhami et al. [9] developed a new concept in cooperative adenine–uracil (A-U) hydrogen bonding interactions between anticancer drugs and nanocarrier complexes, which was successfully demonstrated by invoking the co-assembly of water-soluble, uracil end-capped polyethylene glycol polymer (BU-PEG) upon association with the hydrophobic drug adenine-modified rhodamine (A-R6G). This concept holds promise as a smart and versatile drug delivery system, which leads to the formation of self-assembled A-R6G/BU-PEG nanogels in aqueous solution, for the achievement of targeted, more efficient cancer chemotherapy.

A. Jagusiak et al. [10] described the Congo red–doxorubicin (CR-DOX) complexes, analyzed their interaction with some proteins, and explained the mechanism of this interaction. This kind of interaction between CR-DOX and the described proteins may in future become an important therapeutic system, with the possibility of targeted drug transport and delivery. Supramolecular ribbon-like CR complexed with doxorubicin is a promising system in the treatment of cancers and may open new avenues for novel treatment strategies.

We would like to thank all the authors and reviewers of this Special Issue. We also acknowledge the Assistant Editor, Ms. Daisy Tu, for her tremendous efforts in ensuring its implementation. In addition, authors are encouraged to submit original research articles and reviews in the next Special Issue, "Supramolecular Systems for Gene and Drug Delivery (Volume II)", led by us.

**Funding:** F. J. Ostos thanks the Junta de Andalucía for the postdoctoral grant (PAIDI-DOCTOR, DOC\_00963). J. A. Lebrón also thanks the Fundación ONCE funded by the Fondo Social Europeo.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* β**-Cyclodextrin-Based Nanosponges Functionalized with Drugs and Gold Nanoparticles**

**Isabel Asela 1,†, Orlando Donoso-González 1,2,†, Nicolás Yutronic 1,\* and Rodrigo Sierpe 1,2,3,\***


**Abstract:** Drugs are widely used as therapeutic agents; however, they may present some limitations. To overcome some of the therapeutic disadvantages of drugs, the use of β-cyclodextrin-based nanosponges (βCDNS) constitutes a promising strategy. βCDNS are matrices that contain multiple hydrophobic cavities, increasing the loading capacity, association, and stability of the included drugs. On the other hand, gold nanoparticles (AuNPs) are also used as therapeutic and diagnostic agents due to their unique properties and high chemical reactivity. In this work, we developed a new nanomaterial based on βCDNS and two therapeutic agents, drugs and AuNPs. First, the drugs phenylethylamine (PhEA) and 2-amino-4-(4-chlorophenyl)-thiazole (AT) were loaded on βCDNS. Later, the βCDNS–drug supramolecular complexes were functionalized with AuNPs, forming the βCDNS–PhEA–AuNP and βCDNS–AT–AuNP systems. The success of the formation of βCDNS and the loading of PhEA, AT, and AuNPs was demonstrated using different characterization techniques. The loading capacities of PhEA and AT in βCDNS were 90% and 150%, respectively, which is eight times higher than that with native βCD. The functional groups SH and NH2 of the drugs remained exposed and allowed the stabilization of the AuNPs, 85% of which were immobilized. These unique systems can be versatile materials with an efficient loading capacity for potential applications in the transport of therapeutic agents.

**Keywords:** β-cyclodextrin-based nanosponge; phenylethylamine; 2-amino-4-(4-chlorophenyl)thiazole (AT); gold nanoparticles; carrier of therapeutic agents

#### **1. Introduction**

β-cyclodextrin (βCD) is a cyclic oligosaccharide approved by the FDA (Food and Drug Administration) that has been widely used as a pharmaceutical excipient in food products, textiles, cosmetics, and medical products [1]. In modern drug delivery investigations, βCD has been used as a host molecule for the preparation of drug carrier systems in diverse forms, such as vesicles, hydrogels, micelles, metal–organic systems, and nanoparticles [2–6]. Structural modifications of native βCD have been shown to increase its inclusion capacity and solubility and have allowed bioapplications of a large number of guest biomolecules [7–9]. An innovative modification to βCD recently studied was the synthesis of a polymeric crosslinked network, forming a highly porous and branched matrix of nanometric dimensions called the β-cyclodextrin-based nanosponge (βCDNS) [10,11]. This nanostructure contains multiple lipophilic cavities and carbonate bridges, leading to a network of hydrophilic

O.; Yutronic, N.; Sierpe, R. β-Cyclodextrin-Based Nanosponges Functionalized with Drugs and Gold Nanoparticles. *Pharmaceutics* **2021**, *13*, 513. https://doi.org/10.3390/ pharmaceutics13040513

**Citation:** Asela, I.; Donoso-González,

Academic Editors: Francisco José Ostos, José Antonio Lebrón and Pilar López-Cornejo

Received: 17 February 2021 Accepted: 4 April 2021 Published: 8 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

channels [12], which allows βCDNS to serve as a polymeric conjugate, increasing the loading capacity, association, and stability of the included drugs [7,13–19]. Notably, a high loading capacity is a characteristic feature of βCDNS since they can interact with different molecules of suitable dimensions, using either the cavities of βCD or the multiple pores generated in the crosslinking [7,11,12]. Due to the above, studies on βCDNS applied to drug administration have been reported.

Drugs are some of the most widely used therapeutic agents; however, they may present some limitations, such as early instability, poor aqueous solubility, and low bioavailability. Therefore, strategies for the inclusion of drugs in matrices of native or modified βCD have been an excellent alternative for solving these disadvantages. In this work, the loading of the drugs phenylethylamine (PhEA) and 2-amino-4-(4-chlorophenyl)-thiazole (AT) on βCDNS was studied, which led to formation of two new systems: βCDNS–PhEA and βCDNS–AT. PhEA is a psychoactive stimulant that is used as an antidepressant without inducing tolerance; however, it is rapidly metabolized in organisms by the MAO-B enzyme, making it difficult to reach the site of action [20,21]. AT is a thiazole derivative that is currently used as an antimicrobial and anti-inflammatory agent but is rapidly degraded and has a poor aqueous solubility [22–24]. Our group previously studied complex formation between native βCD and these drugs. An increase was reported in the aqueous solubility and stability of PhEA over time due to its inclusion; in addition, the drug was released from the βCD–PhEA complex using laser irradiation and gold nanoparticles (AuNPs) present in the medium [25]. The inclusion of AT in βCD increased its aqueous solubility, allowing the application of higher doses in in vitro studies of permeability and antibacterial activity. Finally, it was demonstrated that the βCD–AT complex maintained its antibacterial activity against six strains of clinical relevance [26]. In this sense, the incorporation of βCDNS could show novel results, increasing the loading capacity or solubility, among other advantages [7,10,11,16]. Notably, AuNPs could also be added as a remarkable second therapeutic agent.

AuNPs have been widely employed in nanobiotechnology due to their unique properties, which allow them to be incorporated into new nanomaterials [27]. The main characteristics of AuNPs include their optoelectronic properties, as surface plasmon resonance (SPR), which are related to their shape, size, and large surface-to-volume ratio; besides its excellent biocompatibility and low toxicity [28]. The chemical reactivity of the surface atoms of AuNPs allows their functionalization and assembly with various chemical species, enabling their application in chemical and biological sensing, imaging, therapeutics, detection and diagnostics, biolabeling, and drug delivery [29–35]. Notably, AuNPs have been used as therapeutic and diagnostic agents, even in hard-to-reach places, such as the brain, since they can cross the blood–brain barrier [36–40]. Due to their photothermal properties, AuNPs can release compounds that are attached or close to its surface, such as drugs, at specific sites of action in a controlled manner due to the generation of thermal energy when excited by a laser specifically tuned to the SPR frequency [25,41,42]. Furthermore, AuNPs can accumulate passively in sites with an immature vasculature and with extensive fenestrations, such as tumor tissues, or in injured sites where an immune response develops. This is called the enhanced permeability and retention effect (EPR effect) [43–45]. It has been shown that the EPR effect combined with a longer blood circulation time of some types of nanoparticles can increase drug concentrations in tumors by 10 to 100 times compared to the use of free drugs [46]. In recent years, a series of complexes based on βCD have allowed the stabilization of AuNPs, building systems with promising applications in biological and chemical areas [25,26,47–53]. Therefore, if properly designed, βCDNS loaded with drugs and AuNPs could be used as new systems with biomedical applications, acting synergistically in nanotherapy.

In this work, we propose the development of a new system based on βCDNS for the transport of two therapeutic agents, drugs and AuNPs. For this, inclusions of PhEA and AT were studied separately. Later, the complexes were functionalized with AuNPs, forming the βCDNS–PhEA–AuNP and βCDNS–AT–AuNP systems. We believe that these unique systems, based on βCDNS, drugs, and AuNPs, can be versatile materials with potential applications in the therapy and diagnosis of diseases.

#### **2. Materials and Methods**

#### *2.1. Material*

Anhydrous βCD (C42H70O35) ≥ 97%, 1134.98 g/mol; diphenylcarbonate (DPC, C6H5O)2CO) 99%, 214.22 g/mol; PhEA (C8H11N) ≥ 99.5%, 121.18 g/mol, density (δ): 0.962 g/mL; AT (C9H7ClN2S) ≥ 98%, 210.68 g/mol; sodium hydroxide (NaOH) ≥ 97%, 40.00 g/mol; tetrachloroauric acid (HAuCl4) ≥ 99.9%, 393.83 g/mol; and sodium citrate (Na3C6H5O7) ≥ 99%, 294.10 g/mol were provided by Merck (Merck, Darmstadt, Germany). Ultrapure water (18 MWcm−1) was obtained from a Milli-Q water system (Synergy UV equipment, Merck, Darmstadt, Germany).

#### *2.2. Synthesis of β-Cyclodextrin Nanosponges*

For βCDNS synthesis, anhydrous βCD and DPC were used as precursors. Synthesis was carried out by adapting Patel's protocol [54]. βCD (0.189 g) and 0.143 g of DPC were mixed in solid state at a 1:4 molar ratio βCD:DPC. A round-bottom flask with the mixture was heated inside an oil bath on a heating plate, with constant stirring for 5 h at 100 ◦C, observing its melting. The solid mixture obtained was ground in a mortar, washed with distilled water, and filtered under vacuum. The product was washed in a Soxhlet apparatus with acetone for 24 h, to remove phenol by-product. Later, it was moistened with water and dried for 2 h in a vacuum system using a Buchner funnel connected to a Kitasato flask to remove trace βCD. Finally, the product was dried for 72 h at 65 ◦C and stored.

#### *2.3. β-Cyclodextrin Nanosponges Loading with Drugs*

To load βCDNS with PhEA and AT, the saturated solutions method [55] was used with minor modifications. βCDNS were dispersed in a NaOH 0.1 M solution at room temperature, while the drugs were dissolved in ethanol. The solutions were mixed under constant agitation for 15 min and then left without agitation for 24 h. The resulting solution was centrifuged, and the supernatant was lyophilized and reserved [56,57]. The loading capacity of the βCDNS–PhEA and βCDNS–AT systems was calculated from the weights of βCD and drugs obtained using Equation (1) [58].

$$\text{Loading capacity} = \frac{\text{Weight of drug in \\$CDNS}}{\text{Weight of \\$CD in \\$CDNS}} \times 100\tag{1}$$

#### *2.4. Association Constant, Ka*

For both drugs, studies were performed following the Higuchi and Connors method [59]. First, known concentrations (C) of each drug were measured by UV-Vis. From the Amax vs. C graph, the slope corresponded to the ε of each drug. Then, the βCDNS concentration versus the loaded drug concentration (calculated by Beer–Lambert law) was plotted. The value of the slope of the graphs related the amount of βCDNS added to the amount of solubilized drug, indicating the degree of solubilization. Degree of solubilization was used to calculate the association constant (Ka) and complexation efficiency of each system using Equations (2) and (3), respectively.

$$\mathcal{K}\_{\mathbf{a}(1:1)} = \frac{\text{Degree of solubilitization}}{[\mathcal{C}\_{\mathbf{o}}](1 - \text{Degree of solubilitization})} \tag{2}$$

$$\text{Complexation efficiency} = \text{K}\_{\text{4(1:1)}}[\text{C}\_{\text{o}}] \quad \frac{\text{Degree of solubilitization}}{(1 - \text{Degree of solubilitization})} \tag{3}$$

[Co] corresponds to the concentration of the free drug in the absence of βCDNS.

#### *2.5. Synthesis of Gold Nanoparticles and Their Immobilization on β-Cyclodextrin Nanosponges–Drug Systems*

Synthesis of AuNPs was performed using the Turkevich method [60]. A reflux system on a round-bottom flask (with three necks) was mounted by placing a thermometer, a condenser, and a rubber stopper on each neck. Here, 0.474 mL of HAuCl4 was added with 18 mL of water. Sodium citrate (22.8 mg) was dissolved in 2.0 mL of water and heated at 60 ◦C for approximately 5 min. When aqueous solution of HAuCl4 was refluxed and the gas–liquid equilibrium stabilized at a temperature of 186 ◦C, the citrate solution (at 60 ◦C) was added through the neck with the stopper. The reflux was continued under constant agitation (6× *g*) for 30 min until a deep red solution was obtained. Later, the solution was cooled slowly to room temperature. The obtained AuNPs were filtered, diluted, set to pH 8.8 using an NaOH solution, and stored at 4 ◦C.

Immobilization was carried out via solubilization of the βCDNS–drug supramolecular complexes in an alkaline environment of AuNPs, setting the pH to 8.8 using NaOH. These mixtures formed homogeneous colloidal solutions that were centrifuged to decant only the βCDNS–drug systems interacting with AuNPs. Once the systems βCDNS–drug–AuNP were separated from the supernatant, they were resuspended in a new aqueous solution, forming the systems βCDNS–AT–AuNP and βCDNS–PhEA–AuNP. The concentration of AuNPs was calculated using UV-Vis spectroscopy. The molar extinction coefficient was obtained from the literature [61,62], and it was applied together with the Beer–Lambert equation.

#### *2.6. Analysis by Nuclear Magnetic Resonance of Protons, 1H-NMR*

All the samples were dissolved in deuterated dimethylsulfoxide (DMSO)-d6.

#### *2.7. Preparation of Samples for Studies by Scanning and Transmission Electron Microscopy, SEM and TEM*

For SEM studies, the βCD and βCDNS samples were prepared directly depositing the solid material onto carbon tape, then a gold coating was applied using magneton sputtering (pressure 0.5 mbar, Ar atmosphere, current 25 mA over 15 s). βCDNS–drug–AuNP samples were prepared by dropping aliquots on carbon tape, allowing them to dry overnight. The AuNPs immobilized on βCDNS–drug systems allowed the conductivity of these samples.

For TEM studies, the βCD and βCDNS samples were dissolved in ethanol (20% *v*/*v*), then mixed, sonicated, and dripped onto a copper grid with a continuous Formvar film. The βCDNS–drug samples were dissolved in ethanol (20% *v*/*v*), then mixed, sonicated, and dripped onto a holey carbon grid. Finally, all these samples were stained with phosphotungstic acid. The AuNPs samples were deposited directly on the grid with a continuous Formvar film.

#### *2.8. Preparation of the Samples for Studies by Dynamic Light Scattering (DLS) and ζ Potential*

βCDNS and βCDNS–drugs were redispersed to measurements. To determine the size distribution of the samples, the results were retrieved from the intensity distribution values using the cumulant method. The measurement conditions were set for organic βCD-based samples (refraction index: 1.49 and k: 0).

AuNPs and AuNPs with βCDNS–drug were diluted 10× for measurements. Sonication and filtration were performed through a 0.45 μm filter. To determine the size distribution of AuNPs on the samples, the results were retrieved from the intensity distribution values using the cumulant method. On the other hand, the Smoluchowski approximation was used to calculate the ζ potentials from the measured electrophoretic mobility. The measurement conditions were set for colloidal gold samples (refraction index: 1.33 and k: 0.20).

#### *2.9. Equipment Used for Characterization of the Samples*

#### 2.9.1. Nuclear Magnetic Resonance of Protons, 1H-NMR

1H-NMR characterizations of the βCDNS, PhEA, AT, and βCDNS–drug samples were performed in a Bruker Advance 400 MHz instrument (Bruker, Billerica, MA, USA) at 30 ◦C using TMS as an internal reference. The MestreNova program was used for data processing.

#### 2.9.2. Infrared Spectroscopy, IR

The analyses were performed on a Jasco FT/IR-4600 instrument (Jasco, Easton, PA, USA). Spectral resolution: 1 cm−1, number of scans: 4. CO2 and H2O correction through the software of the equipment was made. Baseline correction of KBr was performed.

#### 2.9.3. Thermogravimetric Analysis, TGA

Analyses were performed on Perkin-Elmer model 4000 equipment (Perkin-Elmer, Waltham, MA, USA) over a temperature range from 0 ◦C to 800 ◦C with a rate of 10 ◦C/min under an air atmosphere with a flow of 20 mL/min.

#### 2.9.4. Scanning and Transmission Electron Microscopy, SEM and TEM

For both characterizations, Inspect F50 HR-SEM instrument (Fei Company, Hillsboro, OR, USA) was used. For the scanning electron microscopy (SEM) images, an Everhart-Thornley detector was used, while for the transmission electron microscopy (TEM) images, the detector was scanning transmission electron microscope (STEM). An acceleration voltage of 10.0 kV, pressure of 9.71 × <sup>10</sup>−<sup>8</sup> Pa, and observation magnitudes of 16,000× and 100,000× were used.

#### 2.9.5. UV-Visible Spectrophotometry

A Shimadzu UV-2450 instrument (Shimadzu, Kyoto, Japan) was employed to obtain the absorbance spectra. Measurements were made in 1.0 cm diameter quartz cuvettes between 200 and 800 nm using water at pH 8.8 as the reference. The UVProve program, version 1.10, was used for data processing.

#### 2.9.6. Dynamic Light Scattering (DLS) and ζ Potential

The samples were measured on a Malvern Zetasizer Nano ZS instrument (Malvern, Malvern, UK).

#### 2.9.7. Lyophilization of the Samples

BenchTop Pro, Omnitronic team equipment, SP Scientific (Omnitronic team, Gardiner, NY, USA) was used.

For data processing and graphic design, OriginPro 8.0 software (OriginLab, Northampton, MA, USA) was used.

#### 2.9.8. Metallization of the Samples

PELCO SC-6 magnetron sputtering (PELCO, Fresno, CA, USA) was used. A gold foil was placed inside the vacuum chamber at 0.5 mbar, under inert atmosphere of argon. To begin the process, a current of 25 mA was used over 15 s to ionize the gas, hitting the metal foil and releasing Au atoms. These Au atoms were deposited over the βCD and βCDNS systems.

#### **3. Results and Discussion**

#### *3.1. Synthesis and Characterization of β-Cyclodextrin-Based Nanosponges*

Different synthesis routes have been reported for βCDNS formation, and they use ultrasonic baths; heating plates; solvents, such as ethanol or acetone for the washing stages; and even different molar ratios of βCD and DPC [7,10]. For this reason, different methodologies were evaluated to optimize the synthesis of βCDNS, eliminate byproducts, and increase yield. For the ultrasonic bath (A) and heating plate (B) methods, the use of acetone and a 1:4 molar ratio showed yields greater than 60%, as shown in Figure A1 (Appendix A). Considering the reproducibility of the synthesis and the lower amount of generated byproducts exhibited by method B relative to method A, method B with a heating plate was selected.

βCDNS formation was confirmed using 1H-NMR. The technique allowed us to compare the chemical shifts of the signals for βCD protons in βCDNS and in native βCD. Figure 1 shows the spectra of (A) βCD, (B) DPC, and (C) βCDNS with a scheme showing the proton assignments for βCD and DPC. Table 1 shows the proton assignment for βCD and their respective chemical shifts and integrations in the 1H-NMR spectra. The shifts of the signals are due to the change in the chemical environment of the βCD matrices when they are linked to form βCDNS. Notably, the greatest changes were observed in the integration delta (Δ - ) of the hydroxyl groups, because they react with DPC to form linkers between βCD matrices, strongly suggesting βCDNS formation.

**Figure 1.** 1H-NMR spectra of (**A**) native β-cyclodextrin (βCD), (**B**) diphenylcarbonate (DPC), and (**C**) β-cyclodextrin-based nanosponge (βCDNS) synthesized (**left**) together with the molecular structures and the assignments of the protons of βCD and DPC (**right**).

**Table 1.** Proton assignments, 1H-NMR chemical shifts, and integrations of the β-cyclodextrin (βCD) and β-cyclodextrinbased nanosponge (βCDNS) signals.


βCDNS formation was also characterized using IR vibrational spectroscopy. Commonly, this study focuses on comparing the signals of native βCD and βCD forming

nanosponges and recognizing the vibration signal of the carbonyl group, which is an indicator of βCD crosslinking. Figure 2A shows the IR spectra of (A) βCD, (B) DPC, and (C) βCDNS. Characteristic peaks of βCD are observed at 3363 cm−<sup>1</sup> (O-H alcohol stretching), 2924 cm−<sup>1</sup> (C-H stretching), 1417 cm<sup>−</sup>1, 1368 cm−1, 1157 cm−<sup>1</sup> (O-H bending), 1080 cm<sup>−</sup>1, and 1029 cm−<sup>1</sup> (C-O stretching). These data are consistent with literature data [63–65]. For βCDNS, the characteristic peaks are located mostly in the same regions observed for βCD, but with shifts or variations in intensity due to changes in the chemical environment. These were observed at 3366 cm−<sup>1</sup> (O-H alcohol stretching), 2928 cm−<sup>1</sup> (C-H stretching), 1645 cm−<sup>1</sup> (C=O stretching), 1367, 1234, and 1155 cm−<sup>1</sup> (O-H bending), and 1079 cm−<sup>1</sup> and 1030 cm−<sup>1</sup> (C-O stretching). Notably, the appearance of peaks at 1783, 1715, and 1235 cm−<sup>1</sup> derived from signals present in DPC confirm the crosslinking of βCD forming nanosponges. The peak at 1760 cm−<sup>1</sup> (C=O stretching) of DPC is masked by a peak in the βCDNS spectrum.

Thermogravimetry was performed to analyze and confirm the formation of βCDNS, differentiating it from its precursors through changes in their thermal decomposition, as is typically observed in the synthesis of polymeric materials [66]. Figure 2B shows thermograms of (A) βCD, (B) DPC, and (C) βCDNS. The loss of hydration water was observed in the first decomposition at temperatures up to 100 ◦C, with the percentage of mass loss being 11.5% for βCD and 2.7% for βCDNS of the total mass samples. Decomposition of 100% of the mass of DPC was observed in the range 130 to 250 ◦C. A second range of decomposition in βCD was observed between 300 and 350 ◦C, corresponding to a loss of 71% of the sample mass. For βCDNS, this second range was between 210 and 350 ◦C, consuming 70% of the total mass. The decrease in the temperature at the beginning of thermal degradation suggests that DPC, a crosslinker molecule, binds to the primary OH groups of βCDs, forming the nanopolymer through carbonyl groups. Changes in the peaks of the TGA curves (see Figure A2 in Appendix A) from 337 ◦C (βCD) to 327 (βCDNS) are typically observed in the formation of polymeric materials due to changes in chemical structure [67–69]. Finally, the oxidation interval for βCD ranged from 350 to 700 ◦C, encompassing 17.5% of the mass. However, βCDNS oxidation ranges from 350 to 580 ◦C, encompassing 27.3% of the mass. This also suggests modifications in the reactive structure of the polymer relative to native βCD.

To explain the change in the beginning of the range of thermal degradation for βCDNS, the average between the beginning temperatures for βCD and DPC, which were 300 and 130 ◦C, respectively, was evaluated. The calculated average temperature was 215 ◦C, which coincided with the value of the beginning of thermal degradation observed in the βCDNS thermogram, fulfilling the "eutectic mixture" criterion [70]. In addition, the high value of the degradation interval for βCDNS supports its thermal stability.

To obtain information on the morphology and size of βCDNS, the material was characterized using electron microscopy techniques and DLS. Figure 3 shows micrographs obtained by FE-SEM of native βCD (Figure 3A) and βCDNS (Figure 3B), directly revealing the morphological differences between both. βCD has irregular crystalline structures, while βCDNS has a characteristic porous appearance. TEM images were obtained to determine the average diameter of βCDNS, which were previously dispersed by sonication. Figure 3C,D shows the βCDNS and the resulting histogram, respectively. The average diameter, obtained from the count of more than 450 nanoparticles seen in various TEM images, was 146 ± 54 nm (see more images in Figure A3 in Appendix A). The staining of the βCDNS sample revealed some βCD crystals, which was verified by obtaining TEM images of native βCD with the same dispersion and staining protocol described for βCDNS (see Figure A3 in Appendix A). In addition, a hydrodynamic diameter of 133.9 ± 66.9 nm was found for βCDNS using DLS. These size data are concordant and strongly suggest the nanometric dimensions of the system studied (see more details in Appendix C).

**Figure 2.** (**A**) FT-IR spectra of (A) βCD, (B) DPC, and (C) βCDNS; (**B**) normalized thermograms of (A) βCD, (B) DPC, and (C) βCDNS.

**Figure 3.** SEM micrographs of (**A**) βCD and (**B**) βCDNS. (**C**) transmission electron microscopy (TEM) micrograph of βCDNS and (**D**) the size distribution observed in TEM micrographs of βCDNS. Scale bar for figure (**A**) and (**B**) is 200 μm; scale bar for figure (**C**) is 1000 nm.

#### *3.2. Loading of β-Cyclodextrin-Based Nanosponges with Drugs*

The βCDNS obtained was loaded with two drugs separately, forming the βCDNS– PhEA and βCDNS–AT systems. Once each supramolecular complex was formed in the

solubilized phase of the aqueous solution, the effective inclusion of the drugs and the stoichiometric relationship of both systems were analyzed using 1H-NMR.

Figure 4 shows the spectra of βCDNS–PhEA (A) and βCDNS–AT (B) with their molecular structures and proton assignments for the respective drug. The loading of PhEA to form the βCDNS–PhEA system (A) and the loading of AT to form the βCDNS–AT system (B) were confirmed with the respective assignments of protons in the molecular structures of PhEA and AT (see full spectra, Figures A4 and A5, in Appendix B). Tables 2 and 3 show the chemical shifts and integrals recorded for the protons of βCDNS and of the PhEA and AT drugs resulting from the inclusion process.

For the βCDNS–PhEA system, Table 2, the largest chemical shifts for βCDNS were observed for the internal protons H3 and H5 and the hydroxyl groups OH2 and OH3, probably due to preferential inclusion in the widest zone of the βCD cavity. In addition, chemical shifts for all the βCDNS protons were observed, mainly towards lower fields, which demonstrates the effective loading of PhEA within βCD cavities and in the multiple interstitial spaces of the interstitial βCDNS produced by crosslinking. Analyzing the chemical shifts of the PhEA protons, a change in the chemical environment due to inclusion was also evidenced, consistent with that reported in the literature [25,71].

For the βCDNS–AT system, Table 3, chemical shifts were observed in all the βCDNS protons oriented towards the interior and exterior of the cavity due to the change in the chemical environment of βCDNS resulting from AT loading. This finding shows that the inclusion of the drug occurs in βCD cavities and between the formed interstitial spaces. Chemical shifts towards higher fields were observed in the protons NH2b, Hb'/f', and Hc'/e' of AT, which demonstrates the electronic shielding effect of the drug due to its inclusion in the nanosponges, in accordance with that reported in the literature [26].

**Figure 4.** 1H-NMR spectra of (**A**) phenylethylamine (PhEA) loaded in βCDNS (βCDNS–PhEA) and (**B**) 2-amino-4-(4 chlorophenyl)-thiazole (AT) loaded in βCDNS (βCDNS–AT) (**left**) together with the molecular structures and the assigning protons with respect to PhEA and AT (**right**).


**Table 2.** Proton assignments, 1H-NMR chemical shifts, and integrations of the βCDNS, phenylethylamine (PhEA), and PhEA loaded in βCDNS (βCDNS–PhEA) signals.

**Table 3.** Proton assignments, 1H-NMR chemical shifts, and integrations of the βCDNS, 2-amino-4-(4-chlorophenyl)-thiazole (AT) and AT loaded in βCDNS (βCDNS–AT) signals.


Notably, the integration of the βCDNS protons and the protons of each drug in their respective 1H-NMR spectra, Tables 2 and 3, showed a stoichiometric βCD:drug ratio of 1:8 in both systems, which is an amount of drug eight times greater than those reported for βCD–PhEA [25] and βCD–AT [26], each of which exhibits a 1:1 stoichiometry. This amount is equivalent to 0.9 mg of PhEA loading per 1 mg of βCD unit in βCDNS, and on the other hand, to 1.5 mg of AT loading per 1 mg of βCD unit in βCDNS. Applying Equation (1) [58] (Section 2, Material and Methods), the loading capacity in βCDNS is 90% for PhEA and 150% for AT, which is higher than the loading capacity of 11% for PhEA and 19% for AT in βCD native, according to reported data [25,26]. These results show that the drug loading of the βCDNS formed is higher than that of native βCD and that βCDNS could be used as a more efficient drug carrier than native βCD (see the details in the Appendix B).

The loading of drugs into βCDNS was also analyzed by FT-IR spectroscopy by comparing peaks for vibrations before and after the inclusion process. Figure 5 shows the spectra of (A) PhEA, (B) βCDNS-PhEA, (C) AT, and (D) βCDNS–AT.

**Figure 5.** IR spectra of (**A**) phenylethylamine (PhEA), (**B**) PhEA loaded in βCDNS (βCDNS–PhEA), (**C**) 2-amino-4-(4-chlorophenyl)-thiazole (AT), and (**D**) AT loaded in βCDNS–AT.

In the vibrational analysis of the βCDNS–PhEA system, the βCDNS peaks at 3570 cm−<sup>1</sup> and 3170 cm−<sup>1</sup> corresponding to O-H alcohol stretching and N-H primary amine asymmetric and symmetric stretching, respectively, were identified. The peaks at 2926 cm−<sup>1</sup> corresponding to C-H stretching, at 1642 cm−<sup>1</sup> corresponding to C=O stretching, at 1333 cm−<sup>1</sup> and 1157 cm−<sup>1</sup> corresponding to O-H bending, and at 1081 cm−<sup>1</sup> and 1029 cm−<sup>1</sup> corresponding to C-O stretching were also identified. These vibrations remain unchanged in comparison to those of the βCDNS spectrum without loaded drugs. The peak from PhEA found for the βCDNS–PhEA system corresponding to N-H symmetric stretching was observed at 2950 cm−1, while the peak at 745 cm−<sup>1</sup> corresponding to C-H aromatics was masked due to the inclusion process.

In the case of the βCDNS–AT system, decreases in the intensity of some peaks with respect to those of βCDNS were observed. However, the characteristic peaks were located in the same regions of the spectra. O-H alcohol stretching, and N-H primary amine asymmetric and symmetric stretching vibrations were observed at 3170 cm−<sup>1</sup> and 3570 cm−1, respectively. C-H stretching appeared at 2924 cm−1, C=O stretching at 1637 cm−1, O-H group bending at 1384 cm−<sup>1</sup> and 1157 cm−1, and finally, C-O stretching appeared at 1079 cm−<sup>1</sup> and 1029 cm−1. The characteristic peaks of AT at 1476 cm−1, corresponding to C=C aromatics, and at 3438 cm<sup>−</sup>1, corresponding to N-H aromatic stretching, were masked in βCDNS–AT due to the inclusion in βCDNS.

The changes in the intensity and definition of the βCDNS peaks observed in the IR spectra suggested a change in their conformations due to drug loading, which was also corroborated by DLS and TEM. The hydrodynamic diameters of βCDNS–PhEA and βCDNS–AT were 270.5 ± 48.0 nm and 335.5 ± 150.5 nm, respectively, observing an increase in the size of both systems with respect to βCDNS (see more details in Appendix C). Figure 6 shows TEM images of βCDNS loaded with PhEA (A–E) and AT (F–I). Changes in the shapes of the systems with respect to that of βCDNS were also observed; in addition, the average diameter calculated using TEM images of these systems increased to 252 ± 39 nm with respect to βCDNS. The loading of the drugs PhEA and AT could promote a process of

association and intermolecular interactions between different βCDNS. This would explain the increase in size observed using TEM and DLS.

**Figure 6.** TEM micrographs of (**A**–**E**) βCDNS–PhEA and (**F**–**I**) βCDNS–AT. Scale bar for all images is 1000 nm (Red arrows highlight the nanosystems in the micrographs).

The degree of solubilization indicates the tendency to increase the aqueous solubility of the drugs due to the action of βCDNS, while the complexation efficiency corresponds to the concentration of drug included versus the concentration of drug initially used in the process. This is directly related to the effectiveness of βCDNS and intermolecular interactions to keep drugs entrapped in the complex. The degree of solubilization of the drugs, the *Ka*, and the complexation efficiency for the βCDNS–PhEA and βCDNS–AT systems were calculated using phase solubility studies (Equations (2) and (3), Section 2, Material and Methods) [59] and are shown in Table 4. Additionally, they were compared with the results obtained for the complexation of PhEA and AT using native βCD [25,26].

An increase in the aqueous solubility of PhEA and AT using βCDNS was observed, when they were compared to the solubility of free drugs (see Figures A7 and A8, Appendix B). Notably, the degree of solubilization achieved by the presence of βCDNS was more than 1.3 times higher for PhEA and 5 times higher for AT than with native βCD. This is especially relevant in therapy since drugs to be pharmacologically active must be soluble in water. The *Ka* values are 1318 M−<sup>1</sup> and 484 M−<sup>1</sup> for the βCDNS–PhEA and βCDNS–AT systems, respectively. These results indicate that the interactions that allow inclusion are strong, forming two highly stable systems over time due to the incorporation of βCDNS. The complexation efficiency values obtained for both systems show that the complexation using βCDNS is optimal, being the same for PhEA in native βCD and seven times greater for AT in native βCD. The above findings are in accordance with the previous discussion given by stoichiometry studies and loading capacity calculated using NMR (more details in the Appendix B).


**Table 4.** Comparative data on the degree of solubilization, association constants, and complexation efficiency of the drugs PhEA and AT included in βCDNS versus native βCD.

\* Reference values obtained from the literature [25,26].

In general, the Ka values of the βCD complexes vary between 50 and 2000 M<sup>−</sup>1. Lower values at 50 M−<sup>1</sup> indicate a limitation in the pharmaceutical formulation since they have low stability and do not release the drug at its site of action [25,72–74]. On the other hand, Ka values greater than 2000 M−<sup>1</sup> also present limitations, such as poor pharmacokinetics, since the drug release rates can be affected [72,73]. This is why the use of a strategy for the controlled release of the drugs included in βCDNS becomes relevant. AuNPs can release absorbed energy in the form of heat and can release molecules near their surface as a result of the photothermal effect [28,75,76]. This was demonstrated for a drug in AuNPand βCD-based systems using laser irradiation [25,47]. In this sense, the incorporation of AuNPs into the two systems could, in addition to acting as a therapeutic agent, promote the controlled release of the drugs.

#### *3.3. Synthesis and Immobilization of Gold Nanoparticles on Drug-Loaded β-Cyclodextrin-Based Nanosponges*

Once the βCDNS–drug systems were obtained, the interactions with colloidal AuNPs were studied to load another therapeutic agent and form the βCDNS–PhEA–AuNP and βCDNS–AT–AuNP systems. AuNPs were synthesized following the Turkevich method at pH 5.5. These AuNPs were then stabilized at pH 8.8 to facilitate their immobilization on drug-loaded βCDNS. Figure 7A shows the absorbance spectra of AuNPs at pH 5.5 and 8.8, and Figure 7B shows a representative TEM micrograph of spherical AuNPs with an average diameter of 18 ± 4 nm (see histogram in Figure A9, Appendix C) AuNPs with diameters between 4 and 100 nm do not present cytotoxic effects [77], which would allow possible drug delivery applications.

Figure 7C,D shows the UV-Vis spectra of the βCDNS–PhEA–AuNP and βCD–AT– AuNP systems, respectively, in addition to those of the initial AuNP solution and the supernatant resulting from the functionalization of each mixture. The recorded plasmon bands demonstrate a preferential interaction of AuNPs with βCDNS–drug, with an immobilization of 85%, maintaining the main characteristics of the plasmon band and indicating that AuNPs remain stable in both systems.

Table 5 shows the intensities and the maximum wavelengths from the absorbance spectra. In addition, the hydrodynamic diameter and surface charge of βCDNS–PhEA– AuNP and βCDNS–AT–AuNP in aqueous solution are shown. These analyses represent the behavior of AuNPs in the different systems, because Au is highly efficient to absorb and scatter light, being superior to the organic material present.

A shift in the wavelength of the maximum absorbances with respect to those for the as-synthesized AuNPs occurred for both systems due to the interparticle coupling caused by the increased proximity between these nanostructures when immobilized; in turn, the permanence of the plasmon bands was evidence of the stability achieved and that the aggregation of AuNPs did not occur. In turn, increases in hydrodynamic diameters from 33.9 ± 13.2 nm for AuNPs with citrate to 51.2 ± 24.7 nm for AuNPs in the βCDNS– PhEA–AuNP system and up to 114.0 ± 42.2 nm for AuNPs in the βCDNS–AT system were observed due to the proximity between the immobilized AuNPs and the presence of βCDNS–drug complexes. Furthermore, this behavior was consistent with the increase in size of the βCDNS when they were loaded with the drugs. The reported partial and dynamic inclusion of AT in βCD could explain the greater hydrodynamic diameter of the

AuNPs on βCDNS–AT with respect to βCDNS–PhEA. The two functional groups, NH2 and SH, of AT are exposed [26], facilitating its interaction with AuNPs, while PhEA only has one NH2 group that is completely included within βCD [25,71].

**Figure 7.** (**A**) UV-Vis spectra of AuNPs at pH 5.5 and pH 8.8; (**B**) TEM micrograph of AuNPs with their size histogram inserted (scale bar of 1000 nm); (**C**) UV-Vis spectra of AuNPs with citrate and with βCDNS–PhEA, including supernatant of the functionalization; and (**D**) UV-Vis spectra of AuNPs with citrate and with βCDNS–AT, including supernatant of the functionalization.

The registered surface charge of the AuNPs was −51.4 ± 7.9 mV due to the stabilizing citrate ions, which changed to −33.0 ± 5.3 mV and −38.4 ± 6.9 mV for AuNPs in the βCDNS–PhEA–AuNP and in the βCDNS–AT–AuNP systems, respectively, due to the replacement of a fraction of citrate molecules by neutral supramolecular complexes. As a control, a drug-free βCDNS solution was subjected to the same mixing protocol with colloidal AuNPs, confirming through different characterization techniques that the interaction between βCDNS and AuNPs does not occur (see the details in the Appendix C).


**Table 5.** Data obtained from the UV-VIS spectra, dynamic light scattering (DLS), and ζ potentials of the as-synthesized gold nanoparticles (AuNPs) and AuNPs immobilized on the βCDNS–PhEA and βCDNS–AT supramolecular systems.

Figure 8A,B shows SEM micrographs of the βCDNS–PhEA–AuNP (A) and βCDNS– AT–AuNP (B) systems, respectively. The images clearly show the AuNPs immobilized on the βCDNS–drug supramolecular complexes. In addition, an irregular morphology was observed, probably due to the process of functionalization of βCDNS, as suggested by the TEM images (Figure 6).

Various characterization techniques and direct observation using electron microscopy confirmed the simultaneous loading of βCDNS with two therapeutic agents, drugs and AuNPs, forming the βCDNS–PhEA–AuNP and βCDNS–AT–AuNP systems. If properly designed, that is, by establishing parameters for the colloidal stability, concentration, surface charge, and size, among others, βCDNS and AuNPs could be considered nontoxic and used in therapy without generating adverse effects in the organism. In this sense, in the design and formation of these two new systems, the established parameters were realized.

**Figure 8.** SEM micrographs of (**A**) βCDNS–PhEA with gold nanoparticles (AuNPs) immobilized on the surface and a zoomed view, with bar scales of 2000 nm and 500 nm; (**B**) βCDNS–AT with AuNPs immobilized on its surface and a zoomed view, with bar scales of 1000 nm and 500 nm.

#### **4. Conclusions**

The formation of βCDNS was confirmed by different techniques that indicated its polymeric characteristics and nanometric dimensions. Therapeutic agents PhEA and AT were successfully included in the multiple cavities of the nanostructures, forming the βCDNS–PhEA and βCDNS–AT systems. The loading capacity of βCDNS was 90% for PhEA and 150% for AT, being eight times higher than with native βCD. An increase in the aqueous solubility of PhEA and AT when complexed with βCDNS was demonstrated. In addition, a higher degree of solubilization and complexation efficiency of both drugs was obtained with βCDNS than with native βCD. The synthesized AuNPs were also loaded into each system, reaching an immobilization percentage of 85%. The hydrodynamic diameter and surface charge of AuNPs were 51 nm and −33 mV in the βCDNS–PhEA–AuNP system and 114 nm and −38 mV in the βCDNS–AT–AuNP system, respectively, which are relevant parameters for biological studies. βCDNS loaded with the two therapeutic agents (drug and AuNP) were observed directly by SEM images, showing the porous morphologies of the nanosponges and the nanoparticles immobilized on their surfaces due to the SH and NH2 functional groups of the drugs. We believe that these unique systems, based on βCDNS, drugs, and AuNPs, can be versatile materials with an efficient loading capacity for potential applications in the transport of therapeutic agents. Finally, to continue researching in the field of drug delivery, studies that demonstrate the controlled release of PhEA and AT from βCDNS–drug–AuNP using laser irradiation are required and this, together with studies of cell permeability, toxicity, and pharmacological activity, has been considered in a future perspective.

**Author Contributions:** Conceptualization, I.A., N.Y. and R.S.; methodology, I.A., N.Y. and R.S.; validation, I.A. and R.S.; formal analysis, I.A., O.D.-G. and R.S.; investigation, I.A. and R.S.; resources, I.A., O.D.-G., N.Y. and R.S.; data curation, I.A. and R.S.; writing—original draft preparation, I.A., O.D.-G. and R.S.; writing—review and editing, I.A., O.D.-G., N.Y. and R.S.; visualization, I.A. and R.S.; supervision, I.A. and R.S.; project administration, I.A., N.Y. and R.S.; funding acquisition, I.A., O.D.-G., N.Y. and R.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** Orlando Donoso-González gives thanks for financing of ANID doctoral scholarship No. 21180548. Rodrigo Sierpe gives thanks for financing to ANID-FONDECYT for postdoctoral research grant No. 3180706. Orlando Donoso and Rodrigo Sierpe acknowledge the financing of ANID-FONDAP No. 15130011.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

*Appendix A.1.* β*-cyclodextrin Nanosponge Synthesis* The methods for the synthesis of βCDNS were:

#### *Appendix A.1.1. Method A. (Patel, 2014)*

In a round-bottomed flask, βCD was mixed with DPC, and two different molar ratios were studied: 1:4 and 1:8. The flask was then placed in an ultrasonic bath at 90 ◦C for 5 h. The solid product obtained was then repeatedly washed with distilled water and vacuum filtered for 2 h using a Kitasato flask with a Büchner funnel with filter paper. The product was washed for 24 h in a Soxhlet system, and two different solvents, ethanol and acetone, were studied for this process; the product was finally dried and stored in an amber flask with a Teflon seal.

#### *Appendix A.1.2. Method B. (Modified from Patel)*

In a round-bottomed flask, βCD was mixed with DPC, and two different molar ratios were studied: 1:4 and 1:8. The flask was heated in an oil bath on a heating plate with constant agitation for 5 h at 100 ◦C. The mouth of the flask was covered with a septum, and a syringe was introduced into it to let the phenol gas released from the reaction pass through. The solid mixture obtained was extracted from the flask, a small amount of water was added and an ultrasonic bath was used to release residues from the walls. Once all the solid was extracted, it was ground in a mortar, washed with distilled water, and then vacuum filtered for 2 h. The product was washed for 24 h, and two different solvents, ethanol and acetone, were studied for this process: ethanol and acetone. Finally, the product was dried for between 48 and 72 h at 65 ◦C and stored.

Figure A1 shows a bar graph with the mass yields of different types of βCDNS syntheses with changes in the heating method, the solvents used in the washing steps and the βCD:DPC molar ratio.

**Figure A1.** Mass yield (in percentage) of βCDNS versus the washing solvent, heating method and βCD:DPC molar ratio.

#### *Appendix A.2. Derivative Curves from TGA Characterization of* β*-Cyclodextrin Nanosponge*

Figure A2 shows the derivative curves from TGA (Figure 2B), analysing the differences in the peaks of the thermogravimetric decomposition curves.

**Figure A2.** TGA derivative curves of (A) βCD, (B) DPC and (C) βCD.

*Appendix A.3. TEM Characterization of Native* β*-Cyclodextrin*

Figure A3 shows a TEM image of Figure A3A–C βCDNS to obtain its mean size; and Figure A3D native βCD with the same dispersion and staining protocol described for βCDNS.

**Figure A3.** TEM micrographs of: (**A**,**B**) βCDNS with bar scales of 2000 nm; (**C**) βCDNS with bar scale of 400 nm; and (**D**) native βCD with bar scale of 4000 nm. Acceleration voltage used: 10.0 kV.

#### **Appendix B**

*Appendix B.1. 1H-NMR Full Spectra of Loading Drugs Process*

Figures A4 and A5 show the full 1H-NMR spectra of βCDNS, PhEA, AT, βCDNS-PhEA and βCDNS-AT, summarized in Figure 4.

**Figure A4.** 1H-NMR spectra of (**A**) βCDNS, (**B**) PhEA and (**C**) βCDNS-PhEA.

**Figure A5.** 1H-NMR spectra of (**A**) βCDNS (**B**) AT and (**C**) βCDNS-AT.

#### *Appendix B.2. Calculation of the Stoichiometric Ratio of the Drug Loading Process*

The stoichiometric ratios were calculated in the 1H-NMR spectra by comparing the integrals of the PhEA and AT protons with the integrals of the βCDNS protons from the βCDNS-PhEA and βCDNS-AT systems. First, the integrals of the PhEA signals (protons He/g, Hd/h, and Hf) were analysed using the H1 signal of βCDNS as reference, which integrated 7. In turn, the integrals of the AT signals (He, Hb'/f', Hc'/e' and NH2b) were analysed using the H1 signal of βCDNS as reference, which integrated 7 (see the data in Tables A1 and A2). Finally, the stoichiometric ratios calculated are summarized in Table A3.

**Table A1.** Values of the integrated PhEA and βCD proton signals in the 1H-NMR spectra of the βCDNS-PhEA system, with the integrated H1 proton signals of βCDNS as a reference.


**Table A2.** Values of the integrated AT and βCD proton signals in the 1H-NMR spectra of the βCDNS-AT system, with the integrated H1 proton signals of βCDNS as a reference.


**Table A3.** Summary of the molar ratios of the drugs in the βCDNS for βCDNS-drug systems calculated from Tables A1 and A2.


#### *Appendix B.3. Extinction Coefficient Determination of Drugs*

For each drug, a calibration curve was made with stocks of aqueous solutions of known concentrations to obtain the value of ε. Table A4 presents the data used for this determination for PhEA and AT.

**Table A4.** Data on the concentrations of the drugs, PhEA and AT, and their absorbance maxima at 310 and 290 nm, respectively.


By plotting the PhEA concentration versus the maximum absorbance at 310 nm, the line shown in Figure A6A was obtained, and the value of <sup>ε</sup> was 4.7497 ± 0.2110 mM<sup>−</sup>1cm−1. By plotting the AT concentration versus the maximum absorbance at 290 nm, the line presented in Figure A6B was obtained; the slope corresponds to the value of ε, which was 0.8031 ± 0.0692 mM<sup>−</sup>1cm−1.

**Figure A6.** Linear plots of the (**A**) PhEA and (**B**) AT absorbance maxima at 310 and 290 nm, respectively, vs. concentrations.

#### *Appendix B.4. Determinations of Association Constants of Drug Loading Systems*

For the determination of *Ka*, a stock solution was prepared with 200 mg of βCDNS and water in a 25 mL measuring flask. Volumes of 0 to 2 mL of the stock were taken and diluted with water to produce a total volume of 2 mL with a fixed amount of each drug, 0.5 mL for PhEA and 5.0 mg for AT, added. All the data obtained are presented in Table A5 for PhEA and Table A6 for AT. By applying the extinction coefficient value ε to the Lambert-Beer equation, it was possible to determine the PhEA and AT concentrations in the different assays using the Higuchi-Connors method.

**Table A5.** Values of the different tests carried out to calculate the Ka and complexation efficiency of the βCDNS-PhEA system in water.


**Table A6.** Values of the different tests carried out to calculate the Ka and complexation efficiency of the βCDNS-AT system in water.


The linear relationship obtained from a plot of the solubilized PhEA concentration versus the added βCDNS concentration is shown in Figure A7. The value of the slope was 0.03534 (±0.00115). Using Equation (1), the association constant Ka was calculated, resulting in a value of 1318 M−1. Finally, using Equation (2), the value of complexation efficiency was calculated, resulting in a value of 0.03663 for the βCDNS-PhEA system.

**Figure A7.** Graph of the concentration of solubilized PhEA versus the concentration of added βCDNS and the linear fit.

Figure A8 shows the linear relationship obtained from a plot of the solubilized AT concentrations versus the added βCDNS concentration. The value of the slope was 0.297 (±0.024). The Ka for the <sup>β</sup>CDNS-AT system was 484 M−<sup>1</sup> and the complexation efficiency value was 0.422.

**Figure A8.** Graph of the concentration of solubilized AT versus the concentration of added βCDNS and the linear fit.

#### **Appendix C**

#### *Appendix C.1. Size Histogram of Gold Nanoparticles*

Figure A9 shows the size distribution histogram for synthesised AuNPs from representative TEM images. The observed diameter was 18 (±4) nm.

**Figure A9.** Histogram of size distribution of synthesised AuNPs.

*Appendix C.2. Dynamic Light Scattering and* ζ *Potential Studies of Loading Systems and Gold Nanoparticles Interacting with Supramolecular Systems*

The studies using DLS and the ζ potential are summarized in Table A7. βCDNS was dispersed in water (pH 8.8) and measured, while βCDNS-drug systems were sonicated and measured. AuNPs stabilized with citrate were filtered and then characterized. AuNPs with βCDNS and drugs were centrifuged and resuspended in water (pH 8.8) prior to characterization.

**Table A7.** Data obtained using dynamic light scattering (DLS) and the ζ potential of βCDNS and βCDNS-drugs, and AuNPs with citrate, βCDNS, βCDNS-PhEA and βCDNS-AT.


#### **References**


## *Article* **Molecular Recognition by Pillar[5]arenes: Evidence for Simultaneous Electrostatic and Hydrophobic Interactions**

**Borja Gómez-González 1, Luis García-Río 1,\*, Nuno Basílio 2, Juan C. Mejuto 3,\* and Jesus Simal-Gandara <sup>4</sup>**


**Abstract:** The formation of inclusion complexes between alkylsulfonate guests and a cationic pillar[5]arene receptor in water was investigated by NMR and ITC techniques. The results show the formation of host-guest complexes stabilized by electrostatic interactions and hydrophobic effects with binding constants of up to 107 M−<sup>1</sup> for the guest with higher hydrophobic character. Structurally, the alkyl chain of the guest is included in the hydrophobic aromatic cavity of the macrocycle while the sulfonate groups are held in the multicationic portal by ionic interactions.

**Keywords:** pillararene; host:guest; supramolecular; hydrophobic; ITC; NMR

#### **1. Introduction**

Supramolecular chemistry is a topic of great interest to the scientific community that wants to take advantage of non-covalent interactions, such as van der Waals forces, hydrogen bonds, π-π stacking interaction, electrostatic interactions, or hydrophobic/hydrophilic interactions, with the aim of implementing and explaining increasing complexity systems (bottom-up approach) [1–3]. During the last decades, numerous supramolecular systems have been successfully developed and in the literature, there are numerous investigations regarding their applications as functional materials, in catalytic processes, electronic devices, sensors, or drug carriers, etc., [4–6]. Among these applications, nanomedicine presents a promising potential for modernizing traditional biomedical practices, and in this context, the design of new supramolecular systems in the nanometric range is one of the new frontiers that will offer new diagnostic and therapeutic applications in the field of nanomedicine (drug delivery, gene delivery, drug/gene co-delivery, bioimaging or photodynamic therapy) [7,8].

Noncovalent interactions present several advantages in comparison to covalent ones:


**Citation:** Gómez-González, B.; García-Río, L.; Basílio, N.; Mejuto, J.C.; Simal-Gandara, J. Molecular Recognition by Pillar[5]arenes: Evidence for Simultaneous Electrostatic and Hydrophobic Interactions. *Pharmaceutics* **2022**, *14*, 60. https://doi.org/10.3390/ pharmaceutics14010060

Academic Editors: Francisco José Ostos, José Antonio Lebrón and Pilar López-Cornejo

Received: 16 November 2021 Accepted: 22 December 2021 Published: 28 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

(vi) In addition, it allows the manipulation of supramolecular molecules or building blocks at the molecular level, modulating sizes and morphologies using the "bottom-up" method, providing a variety of novel diagnostic and therapeutic platforms toward applications in nanomedicine.

Within the different non-covalent interactions considered as supramolecular phenomena, the host:guest between different substrates and macrocycles have been studied extensively in recent decades. By including host:guest, two or more molecules can be integrated in a simple and reversible way. This offers us multiple possibilities for new supramolecular structure design. Molecular recognition that involves host:guest interactions play a vital role in life-sustaining biological processes [17,18]. Macrocyclic compounds have been extensively used and intensively investigated as prime host receptors with high affinity and selectivity for complementary small guest molecules or ions. Examples of macrocycles include cryptands [19], crown ethers [20–22], cyclophanes [23], cyclopeptides [24–26], cyclodextrins [27–29], resorcinarenes [30], cucurbit[n]urils [31–34], calix[n]arenes [35–38], and pillar[n]arenes [39]. These macrocycles (hosts) have cavities that allow the encapsulation of substrates of interest (guest). The external properties exhibited by the host molecules favor interaction with the surrounding solvent. On the other hand, the characteristics of the cavities allow the inclusion of the guest. This inclusion occurs through different causes (hydrophobic and/or electrostatic interactions, formation of hydrogen bonds, suitable molecular shape and/or size, etc. In fact, encapsulation in an aqueous solution of hydrophobic guest molecules in macrocyclic hydrophobic cavities is one of the most common cases. The host:guest complex will exhibit high stabilities, providing robust and reliable structures for obtaining supramolecular systems in aqueous media.

Pillar[n]arenes are one of the most recent families of macrocyclic hosts used in supramolecular chemistry [40]. Pillararenes bring together some interesting characteristics of other host systems in a single molecular structure, such as a highly symmetrical pillar-shaped structure which is similar in many respects to that of highly symmetrical cucurbiturils, a π-rich aromatic cavity, also found in calixarenes, and several hydroxyl moieties on both rims, a feature shared with the highly functionalized cyclodextrins. Substituents on both rims of pillararenes affect their physical properties, such as solubility, conformational and host:guest properties. Pillar[n]arenes are very useful structures useful for the design of different supramolecular systems [41–51]. In particular, these systems have interrelated applications in the biomedical and pharmacological fields as drug carriers [46], transmembrane channels [47], or cellular glue [48].

Another aspect to consider is that the presence of charged groups on the pillars[n]arenes convert them into water-soluble substrates (see Scheme 1). Furthermore, its ability to incorporate a wide variety of hosts into its cavity [52,53], its applications as catalysts [54–56], detection [57,58] and gene nanocarriers [59] in aqueous media, has caused that numerous investigations focus on them. When oppositely charged molecules were evaluated as hosts, electrostatic interactions contribute significantly to stabilizing the resulting supramolecular system.

Pillararenes-based host-guest systems comprising amphiphilic guests offer interesting strategies for the development of novel stimuli-responsive drug-delivery systems which improve precision and efficiency in drug delivery [60–66].

In this sense, knowledge of the driving forces behind the host:guest complex formation process is important. We know about the electrostatic interactions between the charged groups of the pillar[5]arene and the ionic substrates, however, the role played by hydrophobic/hydrophilic interactions must be explored in detail. In this context, fundamental studies on the interaction between charged pillararene receptors and model amphiphilic compounds are of utmost importance for the intended pharmaceutical applications in relation to these macrocycles.

**Scheme 1.** Cationic pillar[5]arene and alkylsulfonates.

In this article, we present a structural and thermodynamic study on the host-host complexes between a cationic pillar[5]arene and charged amphiphilic compounds, which by keeping the head group constant, its hydrophobicity can be modulated by modifying the length of the chain hydrocarbon. The family of amphiphiles chosen was the alkylsufonates (see Scheme 1).

#### **2. Materials and Methods**

#### *2.1. Materials*

The highest purity commercially available reagents were supplied by Sigma-Aldrich (Madrid, Spain) and were used without further purification. The water-soluble cationic pillar[5]arene was obtained by a synthetic procedure described elsewhere [67]. Br- exchange by BF4- was carried out as follows: To a solution with Br- (1.17 g, 0.514 mmol) in Milli-Q water at room temperature and with stirring, AgBF4 was added slowly in little portions. A grayish precipitate was obtained. The suspension was centrifuged, and supernatant was collected and filtered (0.45 μm). A yellowish solid was obtained after removing the solvent (1.15 g, 96%). 1H NMR (D2O, 300 MHz): δ = 6.89 (s, 10H); δ = 4.36 (s, 20H); δ = 3.91 (s, 10H); δ = 3.72 (s, 20H); δ = 3.19 (s, 90H); 13C NMR (D2O, 75 MHz): δ = 149.2 (C, 10C); δ = 129.8 (C, 10C); δ = 115.9 (CH, 10C); δ = 64.8 (CH2, 10C); δ = 62.3 (CH2, 10C); δ = 53.7 (CH3, 30C); δ = 29.3 (CH2, 5C); MS (ESI): m/z calcd for [TMAP510+.9BF4 −] 2+ 2253.4; found 2253.2; calcd for [TMAP510+.8BF4 −] 2+ 1083.3; found 1083.1. The final product was analyzed by thermal gravimetric analysis to assess volatile content.

#### *2.2. Microcalorimetry*

An isothermal titration microcalorimeter (VP-ITC) supplied by Microcal Co. (Northamptoh, MA, USA) at 1 atm and 25 ◦C to carry out the microcalorimetric titrations. The procedure used for each titration consisted in sequentially injecting a guest solution in a syringe (0.270 mL) with shaking (459 rpm) into a host solution in the sample cell (1.459 mL). Before each titration, the samples were degassed and thermostatted using an accessory supplied by ThermoVac (Leybold Hispánica, Barcelona, SPAIN). For the reference cell, the same sample was used as in the sample cell. The first injection was discarded in all the experiments carried out in order to suppress the diffusion effects in the calorimetric cell of the syringe material. The number of injections, their volume and the spacing time between each one were varied according to the experiment. Binding constants were calculated from the titration curve by using the AFFINImeter software (S4SD, Santiago de Comostela, SPAIN).

#### *2.3. NMR Spectrometry*

NMR experiments were conducted at 25 ◦C on a spectrometer supplied by Bruker (Bruker NEO 17.6 T) (Billerica, MA, USA) with 750 MHz proton resonance, equipped with a 1H/13C/15N triple resonance PA-TXI probe with deuterium lock channel and shielded PFG z-gradient. The control software was TopSpin 4.0. Chemical shifts were referenced to the lock deuterium solvent. Spectra have been processed and analyzed using Mestrenova software v14.0 supplied by Mestrelab Inc (Santiago de Compostela, SPAIN). The 1D 1H spectrum has been measured with 128 scans, d1 2s relaxation delay and 2.75 s FID acquisition time (aq). The FID has been acquired with 64k complex data points. It has been processed using Fourier Transformation (FT) and zero-filling. 131 k data points spectra have been obtained. The total measurement time was ~10 min.

A two-dimensional 2D COSY spectrum magnitude mode was measured (pulse sequence "cosygpppqf" of Bruker library). The relaxation delay (d1) and the FID acquisition time (at) were 2 and 0.172 s, respectively. The spectrum was measured with eight scans. The number of points in the direct and indirect dimensions was 4 k and 160, respectively. The spectrum was processed with apodization with a sine-bell function in both dimensions and represented in the magnitude mode. The total measurement time was ~48 min.

A two-dimensional 2D HSQC multiplicity edited 1H-13C spectrum was measured (pulse sequence "hsqcedetgpsisp 2.4" of the Bruker library). The spectrum includes adiabatic inversion pulses in 13C and suppression of COSY type artifacts. The INEPTs transfers were optimized for a nominal value of 1JCH of 145 Hz. The delay for multiplicity selection was set to 1/(2 · 1JCH) to detect with the same sign signals of CH3 and CH groups and with opposite phase CH2 groups. The relaxation delay (d1) and the FID acquisition time (at) were 1.6 and 0.112s, respectively; 2048 and 160 complex points in the t2 and t1 dimensions spectrum were acquired. Scans number per t1 increment was 8. The total measurement time was ~1 h 15 min.

#### **3. Results**

δ

The hydrophobic cavity of the pillar[5]arene, together with the presence of five positive charges in each rim makes this macrocycle an excellent receptor for amphiphilic anionic guests. The complexation of the different alkylsulfonates, whose hydrophilic head exhibits a negative charge, (G) by the pillar[5]arene (H) was studied by different experimental techniques.

#### *3.1. NMR Evidence of N-Octylsulfonate Complexation by Pillararene*

NMR spectroscopy has been used to determine the structures of macrocycles complexes. The 1H NMR spectra of octylsulfonate upon mixing in different proportions with pillararene can be observed in Figure 1.

**Figure 1.** 1H NMR spectra in D2O at 25 ◦C for pillararene (1.5 mM); octylsulfonate (1.5 mM) and mixtures of both with a constant concentration of pillararene (1.5 mM) and different concentrations of octylsulfonate.

All protons of octylsulfonate appear upfield-shifted with respect to the free guest upon addition of pillararene, indicating that an inclusion complex was formed. These results indicate that octylsulfonate is incorporated into the magnetic shielding region of the pillararene aromatic cavity with the sulfonate group pointing towards the trimethylammonium groups of the host. Moreover, the host proton signals are also affected by complexation due to the asymmetric structure of the guest and the manner in which it is inserted into the host cavity [52]. To determine the binding stoichiometry of the host:guest complex, considering that fast exchange on the NMR chemical shift timescale was observed for this complex, an NMR titration at constant host concentration, was carried out. Figure 2 shows that the magnitude of the upfield shift for guest hydrogen atoms increases upon a gradual increase of the [host]/[guest] ratio, reaching a plateau for values higher than 1, indicating a 1:1 stoichiometry for the inclusion complex.

**Figure 2.** Chemical shifts for hydrogen atoms in alpha position to the sulfonate group in octylsulfonate in the presence of increasing concentrations of host.

Detailed analysis of spectrum for [Pillararene] = 1.5 mM and [octylsulfonate] = 0.50 mM (Figure 1) reveals that the signal corresponding to the methylene groups in positions C4–C7 of octylsulfonate splits into different signals, allowing a clear characterization of the inclusion complex. Figures 3 and 4 show the HSQC and COSY spectra respectively allowing the assignment of all signals in the NMR spectrum.

**Figure 3.** HSQC spectra for a mixture of 1.5 mM of Pillararene and 0.5 mM of octylsulfonate in D2O a 25 ◦C. Labels for octylsulfonate hydrogen atoms are according to the picture shown in the figure.

**Figure 4.** COSY spectra for a mixture of 1.5 mM of Pillararene and 0.5 mM of octylsulfonate in D2O a 25 ◦C. Labels for octylsulfonate hydrogen atoms are according to the picture shown in the figure.

Assignment of NMR signals allows us to quantify the magnitude of the complexationinduced upfield effect for each hydrogen atom in octylsulfonate (results showed in Table 1). We refer to a complexation-induced chemical shift as the difference between the chemical shift observed for the guest free and complexed, Δδ = δfree − δbound. The magnitude of Δδ is dependent on the hydrogen atom position along the alkyl chain of octylsulfonate. It is remarkable the very large magnitude of the upfield effects with values larger than Δδ = 3 ppm for some central chain nuclei. Hydrogen atoms Hc and Hd show the large Δδ values allowing to propose a structure for the host:guest complex as shown in Figure 5. Hydrogen atoms in positions c and d are located inside the aromatic region of the pillararene allowing the large Δδ values, Δδ > 3 ppm. Hydrogens at position e should be just below this region but close to the aromatic groups (Δδ = 2.3 ppm). It is remarkable that hydrogen atoms at positions g and h (Δδ < 1 ppm), as well as in the alpha position to the sulfonate group, are clearly located outside the aromatic region.

**Table 1.** Magnitude of the complexation induced chemical shifts (ppm) for host:guest complexes between pillararene and different alkylsulfonates.


Δδ

Δ

**Figure 5.** (**Left**) Plot of the magnitude of complexation induced chemical shift, Δδ, as a function of the hydrogen atom position (starting at the sulfonate group). (**Right**) Schematic picture of the host:guest complex showing hydrogens Hc and Hd fully incorporated into the aromatic region of the host. For simplicity only two trimethylammonium groups of pillararene are shown.

Similar experiments were conducted for shorter chain alkylsulfonates with three to six carbon atoms (see Table 1) revealing that hydrogen atoms in positions Hc and Hd show the large upfield effects confirming that these atoms are clearly included inside the pillararene cavity. The complexation picture shows the sulfonate aligned with the trimethylammonium head groups of the receptor in such a way that electrostatic interaction should be the major driving force for complexation. It is remarkable that Δδ values are also dependent on the nature of the alkylsulfonate (see Figure 6). In fact, Ha hydrogen atoms show the large upfield effect for alkylsulfonates with four and five carbon atoms, meanwhile, alkyl sulfonates with three and eight carbon atoms present smaller values. On the other hand, hydrogens Hc show the large upfield effect for C5SO3 − and C6SO3 −, and hydrogens Hb show the large Δδ for C4SO3 − and C5SO3 −. More clearly, Figure 6-left shows that the magnitude of Δδ is strongly dependent on the number of carbon atoms in the alkylsulfonate for hydrogens Hc > Hb > Ha, being indicative of a different degree of penetration into the pillararene cavity. Figure 6-right represents normalized Δδcorr by subtracting the values corresponding to hydrogens Ha. The normalized values are directly comparable and indicate that Hc hydrogens are much closer to the cavity than Hb and that an optimal degree of penetration is reached for five atoms of carbon. Alkylsulfonates with three and four carbon atoms can form external complexes where the carbon atoms do not fit neatly together. This causes that the magnitude of Δδcorr does not reach an optimal value. Likewise, it is observed that for octylsulfonate, the Hc hydrogens present a lower inclusion than for the 5 carbon atom homolog. This behavior may be due to a hydrophobic pushup effect that compels the sulfonate group towards a plane superior to the portal of the pillararene in order to accommodate more methylene groups inside the cavity. At the same time, the possibility that the hydrophobic effect induces a greater degree of folding of the alkyl chain in order to maximize the number of carbon atoms that can be included in the cavity should be considered.

These results indicate that the location of the sulfonate group should be dependent on the number of carbon atoms, being closer to the positive portal of the pillararene for C5SO3 − and C4SO3 −. This behavior can be observed for hydrogen atoms in positions Hb and Hc, being clear evidence of a different degree of guest penetration into the host cavity and, consequently, ruling out the electrostatic attraction as the only interaction stabilizing the host:guest complex.

**Figure 6.** (**Left**) Influence of the number of carbon atoms in the alkyl chain of alkylsulfonate on the chemical induced upfield effect, Δδ: (•) Ha; (•) Hb and (•) Hc. (**Right**) Values of upfield effect for hydrogen atoms in positions (•) Hb and (•) Hc after correction by upfield effect of hydrogens Ha.

#### *3.2. Calorimetric Titrations for Alikylsulfonate Recognition by Pillararene*

In order to quantitatively evaluate the complexation of pillar[5]arene with each guest and the stoichiometry of the complex formed, an isothermal calorimetry titration was carried out at 25 ◦C under neutral conditions. Each titration was done by consecutively adding the guest to the host in the sample cell. As an example, each butylsulfonate titration in the sample cell containing the pillar[5]arene is shown in Figure 7 (see Supplementary Materials for other alkylsulfonates). The experimental data were satisfactorily fitted to a model of "a set of binding sites", obtaining the binding constant (K) and the thermodynamic parameters (Table 2).

**Figure 7.** Microcalorimetric titration of butylsulfonate (G) with pillar[5]arene (H) in water at 25 ◦C. (**Top**): Raw data for the 28 sequential injections (10 μL per injection) of a solution of G (0.5 mM) into a solution of H (0.04 mM). (**Bottom**): "Net" heat effects fitted using the "one set of sites" binding model.

ΔΔ


**Table 2.** Thermodynamic parameters obtained for host:guest complexes between pillararene and different alkylsulfonates.

The results indicate that complexation is mainly enthalpy-driven (ΔH<sup>0</sup> <sup>=</sup> −(6.60 ± 0.01) kcal/mol) accompanied by favorable entropic changes (TΔS<sup>0</sup> = 2.12 kcal/mol), this balance is more favorable to the enthalpic term with the other alkylsufonates.

From the results of the experiments obtained for guests and other macrocyclic compounds, it has been shown that non-covalent interactions contribute to enthalpic changes, while entropy changes can be attributed to conformational changes and/or effects associated with desolvation processes [68]. Thus, hydrophobic or electrostatic interactions, together with dehydration processes, have a positive contribution to entropy. The negative contribution would be produced by the loss of conformational freedom degrees (both on the guest and on the host). Thus, the values obtained for the thermodynamic parameters would indicate that the electrostatic interactions, π-π, and C-H··· π interactions between the aromatic ring and the methyl group of the alkylsulfonate and the electron-rich pillararene cavity would give rise to a favorable contribution on enthalpy. At the same time, the solvent molecules (water) that surround both the host and the guest are released into the bulk water and would be the cause of the entropic increase. The binding constant obtained, K = (2.63 ± 0.01) × 106 <sup>M</sup>−1, is comparable with those reported for negatively charged pillararenos [52,53,68,69] or calixarenes [70–72].

Experimental results reported in Table 2 show alkylsulfonate binding constants to be very sensitive to alkylsulfonate chain length with an increase of almost 10<sup>4</sup> fold ongoing from propane to octanesulfonate. Quantitative analysis of these binding constants requires correction of binding constant for propanesulfonate. Because of its smaller value, experimental results were obtained in the presence of [Pillararene] = 0.25 mM instead of [Pillararene] = 0.04 mM used for other alkylsulfonates. Previous results from our group have shown that toluenesulfonate binding constant to pillararene decreases from 1.37 × 106M−<sup>1</sup> to 3.18 × 104M−<sup>1</sup> by increasing the host concentration from 0.01 to 0.1 mM [53]. This behavior is due to BF4 − complexation by the pillararene, which difficult the entrance of the guest. Extrapolation to alkylsulfonates implies that propanesulfonate binding constant of 1.86 × <sup>10</sup><sup>5</sup> <sup>M</sup>−<sup>1</sup> should be used for comparative proposes.

Figure 8 plots the dependence of the binding constant with the alkyl chain length and includes similar results using β-cyclodextrin as a receptor [73]. Quantitative analysis of the thermodynamic parameters involved in the complex formation between surfactant molecules and cyclodextrin can be simplified by considering the process divided into three stages:

(i) Dehydration of surfactants and cyclodextrin This process is entropically favored due to a strong water structuring that hydrates the exposed hydrophobic residue of the surfactant and to geometric constraints within the CD cavity. Water is structured around the surfactant hydrophobic chain, giving rise to a strong network of hydrogen bonds. The amount of water molecules involved in hydration scales linearly with the alkyl chain length, therefore, the linear relationship between the number of carbons present in the surfactant hydrocarbon chain and the micellization free energy, and similar phenomena involving removal of the surfactant chain from the aqueous medium.


**Figure 8.** (•)Influence of alkyl chain length of alkylsulfonates on their binding constants to pillar[5]arene using a [pillararene] = 0.04 mM at 25 ◦C. Value for propanesulfonate was extrapolated from [host] = 0.25 mM (see text). (•) Binding constants for alkylsulfonates to β-cyclodextrin taken from ref. [73].

Alkylsulfonate binding constants to β-CD increase with the number of methylene groups into the alkyl chain in a non-linear way. The binding constants found for short and very large alkyl chains present lower values than expected due to the fact that the cavity occupation is not complete. This implies that a small amount of water molecules is expelled into the bulk. On the other hand, in the case of large chains, the fact that the binding constants present values lower than those expected would be due to the tolerance of the cyclodextrin cavity to accommodate 6–8 methylene groups.

Figure 8 shows that pillararene is a much more effective receptor for alkylsulfonates than β-CD by a factor of 106. This effect should be ascribed to electrostatic interactions between the negative charge of the guest and the positive ones on the upper and lower rim of pillararene. Note that this interaction is not possible in the case of β-CD as a receptor. The influence of the alkyl chain length on the binding constants to pillararene parallels that observed with β-CD indicating that hydrophobic interactions are playing an important role in the recognition ability of pillararene.

Hydrophobic effects in pillararene recognition are responsible for the different locations of the sulfonate group with respect to the positive upper or lower rim of the host. This different location is reflected by the complexation-induced upfield effect observed in Figure 6-left for hydrogens in alpha position (Ha) to the sulfonate group. Electrostatic interaction in the host:guest complex will compel the sulfonate group close to the trimethylammonium ones in such a way that the distance between the hydrogens Ha of the guest and the aromatic ring of the host keeps constant. However, experimental results indicate that this distance decrease for the following alkylsulfonates: C8SO3 − > C3SO3 − > C6SO3 − > C5SO3 <sup>−</sup> ≈ C4SO3 −. X-ray crystal structure of 1,4-dipropoxypillar[5]arene confirmed that it is a pentagon from the upper view and a pillar structure from the side view. The diameter of the internal cavity was 4.7 Å, which is similar to that of cyclodextrin, allowing the perfect inclusion of methylene chain [75]. The height of pillararene cavity, taken as the distance between the oxygen atoms in the upper and lower rims, is 5.5 Å, allowing accommodation of 4–5 methylene groups. It means that C5SO3 − and C4SO3 − are deeply included in the

pillararene cavity in comparison to C8SO3 − and C3SO3 −. The smaller alkylsulfonate does not displace a large amount of water from the host cavity resulting in a small hydrophobic effect. On the other hand, three methylene groups of C8SO3 − will be outside the cavity. Their hydration in the host:guest complex will contribute unfavorably to its stability.

#### **4. Conclusions**

To sum up, we have demonstrated that alkylsulfonates with different chain lengths are effectively bound by a decacationic pillar[5]arene receptor in an aqueous solution with binding constants in the micro/submicromolar range. The formation of the complexes is enthalpy and entropy driven suggesting that ionic, C-H··· π, van der Waals interaction along with hydrophobic effects contribute to the binding stability. The observed increase in the binding constants as the guest alkyl chain length increases provides strong evidence for the contribution of the hydrophobic effect for the recognition process. This view is supported by the structural NMR studies showing that hydrophobic alkyl chains are deeply included in the aromatic cavity of the macrocyclic receptor. The results obtained herein suggest that cationic pillararene receptors are potentially strong binders for anionic and eventually zwitterionic lipids, and therefore, further studies addressing this class of natural molecules as a guest should be considered due to the potential pharmaceutical applications of these macrocycles.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10.339 0/pharmaceutics14010060/s1, Figures S1–S4 show each titration of alkylsulfonates fitted by the "one set of binding sites" model.

**Author Contributions:** Conceptualization, L.G.-R. and N.B.; methodology, L.G.-R.; software, J.C.M.; validation, L.G.-R. and N.B.; formal analysis, N.B.; investigation, B.G.-G.; resources, B.G.-G.; data curation, L.G.-R. and N.B.; writing—original draft preparation, L.G.-R.; writing—review and editing, N.B., J.C.M. and J.S.-G.; visualization, J.C.M.; supervision, L.G.-R.; project administration, L.G.-R. and J.S.-G.; funding acquisition, L.G.-R. and J.S.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** Financial support from the Ministerio de Economia y Competitividad of Spain (project CTQ2017-84354-P), Xunta de Galicia (GR 2007/085 and ED431C2018/42-GRC) and the European Regional Development Fund (ERDF) is gratefully acknowledged. This work was also supported by the Associate Laboratory for Green Chemistry-LAQV which is financed by national funds from FCT/MCTES (UIDB/50006/2020). N.B. acknowledges the FCT/MCTES for the research contract CEECIND/00466/2017.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Multivalent Calixarene-Based Liposomes as Platforms for Gene and Drug Delivery**

**José Antonio Lebrón 1, Manuel López-López 2, Clara B. García-Calderón 3, Ivan V. Rosado 3, Fernando R. Balestra 4,5, Pablo Huertas 4,5, Roman V. Rodik 6, Vitaly I. Kalchenko 6, Eva Bernal 1, María Luisa Moyá 1,\*, Pilar López-Cornejo 1,\* and Francisco J. Ostos 1,\***


**Abstract:** The formation of calixarene-based liposomes was investigated, and the characterization of these nanostructures was carried out using several techniques. Four amphiphilic calixarenes were used. The length of the hydrophobic chains attached to the lower rim as well as the nature of the polar group present in the upper rim of the calixarenes were varied. The lipid bilayer was formed with one calixarene and with the phospholipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine, DOPE. The cytotoxicity of the liposomes for various cell lines was also studied. From the results obtained, the liposomes formed with the least cytotoxic calixarene, (TEAC12)4, were used as nanocarriers of both nucleic acids and the antineoplastic drug doxorubicin, DOX. Results showed that (TEAC12)4/DOPE/p-EGFP-C1 lipoplexes, of a given composition, can transfect the genetic material, although the transfection efficiency substantially increases in the presence of an additional amount of DOPE as coadjuvant. On the other hand, the (TEAC12)4/DOPE liposomes present a high doxorubicin encapsulation efficiency, and a slow controlled release, which could diminish the side effects of the drug.

**Keywords:** cationic calix[4]arenes; liposomes; nucleic acids; transfection efficiency; doxorubicin; encapsulation

#### **1. Introduction**

Liposomes are spherical structures, similar to vesicles, which have an inner aqueous polar region and a hydrophobic lipid bilayer [1]. Their spontaneous formation in aqueous solutions is due to interactions among water molecules, hydrophilic head groups and hydrophobic chains of the amphiphilic molecules forming the lipid bilayers [2–5]. Liposomes have been prepared by several techniques such as thin lipid film hydration, solvent injection, detergent dialysis or reverse phase evaporation [2,4,5]. They are usually characterized according to their size and number of bilayers, and their charge can be positive, negative or neutral [6,7].

Since the pioneering work by Bangham [8], liposomes have been broadly used as delivery systems for several diagnostic and therapeutic compounds including drugs, genes,

**Citation:** Lebrón, J.A.; López-López, M.; García-Calderón, C.B.; V. Rosado, I.; Balestra, F.R.; Huertas, P.; Rodik, R.V.; Kalchenko, V.I.; Bernal, E.; Moyá, M.L.; et al. Multivalent Calixarene-Based Liposomes as Platforms for Gene and Drug Delivery. *Pharmaceutics* **2021**, *13*, 1250. https://doi.org/10.3390/ pharmaceutics13081250

Academic Editors: Franco Dosio and Giovanna Della Porta

Received: 20 July 2021 Accepted: 8 August 2021 Published: 12 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

imaging agents, proteins, or vaccines, among others [9–13]. Liposomes draw great interest for many researchers working on biomedical applications due to the numerous advantages they offer. It is easy to prepare biocompatible and biodegradable liposomes, with low toxicity and immunogenicity, high stability, and with the ability to host hydrophilic and hydrophobic compounds, whose release can be controlled [2,13,14]. A great variety of amphiphilic molecules have been used to prepare liposomes, making the formation of vesicles in aqueous solution favorable [15–17]. Among them, calixarenes have been frequently used [18–20].

Calixarenes are macrocycles prepared by the base-catalyzed condensation of formaldehyde and p-substituted phenols [21]. They are composed of phenolic units disposed in cyclic arrays and linked by methylene spacers. Their name comes from the word *calix crater* because their tridimensional structure resembles that of an ancient Greek vase. The most common calixarenes are calix[4]arenes, calix[6]arenes, and calix[8]arenes, where [n] indicates the number of phenolic units. The synthetic, easy-to-obtain multivalent ligands by introducing substituents of different nature at the upper and lower rim is one of their main advantages. The structural variety of calixarenes permits to have selective receptors for the inclusion of several neutral molecules, metal and ammonium anions, and cations [22]. Besides, calixarenes can be used in the self-assembly of nanoparticles [23], in the building of molecular machines and rotaxanes [24], for molecular encapsulation [25], and many other applications [22,26,27].

The ability of calixarenes to bind, condense, and transport DNA across cell membranes has been previously investigated [28]. In particular, calix[4]arenes in cone conformation have been found to facilitate cell transfection effectively. Among them, those with long alkyl chains usually lead to small aggregates with low polydispersity, promoting more efficiently gene transfection [29]. On the other hand, it was previously shown that calixarenes interact with the antineoplastic drug doxorubicin, DOX (Scheme 1) [30]. The interaction between calixarenes and DOX is mainly mediated by host–guest and π–π interactions. Works in the literature have shown that calixarenes can be used for the treatment of different types of cancer [31], the results obtained for distinct cancer lines showing that a higher therapeutic effect of the drug is achieved as well as a decrease in side effects.

**Scheme 1.** Structure of doxorubicin.

The authors have been interested in the interaction of calixarenes with nucleic acids and antineoplastic drugs such as DOX [30,32–34]. In this work, the authors wanted to investigate if the use of calixarene-based liposomes improves the results of their use as nanodelivery systems as compared to the naked calixarenes. It is worth noting that thanks to their unique structural properties, calixarene-based liposomes could provide hybrid systems that will synergistically lead to non-viral vectors with enhanced cell transfection properties [35,36]. Moreover, the hydrophobic cavity of calixarenes provide the posibility that, within the calixarene-based liposomes, host–guest phenomena with different drugs can occur [37]. The more appropriate type of calixarenes for preparing

calixarene-based liposomes are the amphiphilic ones. They can be obtained by introducing polar groups at one rim and hydrophobic chains at the other rim [30]. With the goal of preparing calixarene-based liposomes for delivering of genetic material and doxorubicin, in this work the cationic calix[4]arenes 5,11,17,23-tetratriethylammoniummethylene-25,26,27,28- tetradodecyloxycalix[4]arene tetrachloride, (TEAC12)4; 5,11,17,23 tetra(3-methylimidazolium)-methylene-25,26,27,28-tetradodecyloxycalix[4]arene tetrachloride, (Im12)4; 5,11,17,23-tetra(3-methylimidazolium)-methylene-25,27-dihexadecyloxy-26, 28-dipropoxycalix[4]arene tetrachloride, (Im16Im3)2; and 5,11,17,23-tetra(3-methylimidazolium)-methylene-25,26,27,28-tetrahexadecyloxycalix[4]arene tetrachloride, (Im16)4, have been used to prepared liposomes (see Scheme 2). The abbreviations used tried to inform the reader about the structure of the calixarenes. TEAC stands for tetraethylammonium chloride and Im for imidazolinium, in order to distinguish between the two different charged heads present in the upper rim of the calixarenes. The subscripts 3, 12, and 16 inside the parentheses indicate the length of the hydrophobic chains attached to the lower rim. The subscripts 2 and 4 outside the parentheses indicate the number of units inside the parentheses present in the calixarene molecules. Throughout this work the abbreviation CAL means calixarene.

**Scheme 2.** Structure of the calix[4]arenes investigated in this work. (**A**) Calix[4]arene ammonium derivate; and (**B**) Calix[4]arene imidazolinium derivates.

The results obtained will be of interest for researchers working on the use of calixarenes in several biotechnological applications.

#### **2. Materials and Methods**

#### *2.1. Materials*

The lipid 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) was purchased from Avanti Polar Lipids (Alabaster, AL, USA). Red Safe was from iNtRON (Biotechnologiy, Chicago, IL, USA). The rest of the materials, including doxorubicin, were from Sigma-Aldrich (Darmstadt, Germany) and used without further purification.

ctDNA concentration was estimated by UV–visible spectroscopy measuring at 260 nm (molar absorptivity 6600 mol−<sup>1</sup> dm<sup>3</sup> cm−<sup>1</sup> [38]). The average number of base pairs was estimated by agarose gel electrophoresis, using ethidium bromide, EB. The results indicate that there are above 10,000 bp [39]. Throughout the manuscript, the ctDNA concentration will be expressed per base-pairs. The pEGFP-C1 plasmid (Clontech, Biocientífica S.A., Buenos Aires, Argentina), pDNA, was extracted from competent *E*. *coli* bacteria previously transformed with pEGFP-C1; the extraction was done using a GenElute HP Select Plasmid Gigaprep kit (Sigma Aldrich, Darmstadt, Germany). A protocol previously described was used [40]. FuGENE 6 was from Promega Corporation (Madison, WI, USA).

The syntheses of the cationic calixarenes (TEAC12)4 and (Im16Im3)2 were previously described [30], and (Im12)4 and (Im16)4 were purchased from Life Chemicals Inc. (Niagaraon-the-Lake, ON, Canada). Their purity (≥99%) was checked by 1H and 13C NMR, elemental analysis, and mass spectra.

Solutions were prepared with MilliQ water (resistivity > 18 MΩ × cm). The pH was kept constant at 7.4 by using 10 mM HEPES (4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid sodium salt) buffer.

#### *2.2. Preparation of Liposomes*

Liposomes were prepared using the lipid thin-film hydration method [41]. Briefly, adequate quantities of calixarenes, CAL, and DOPE were dissolved in chloroform. Different volumes of these solutions were mixed in order to obtain the desired cationic calixarene molar fraction, α, given by:

$$\alpha = \frac{\text{nCAL}}{\text{nCAL} + \text{nDPE}} \tag{1}$$

where nCAL and nDOPE are the mole number of the cationic calixarene and the zwitterionic DOPE, respectively, in the total volume of the organic solution.

A rotary evaporator was used to evaporate the organic solvent, at 303 K for 50 min. The resultant dry lipid film was stored at 193 K for at least 24 h. In this way, degradation is avoided [42]. Afterwards, 2 mL of HEPES 10 mM, pH = 7.4, was added for hydrating the lipid film, and the mixture was submitted to 10 cycles of vortex (3 min/1200 rpm) and sonication (2 min, JP Selecta Ultrasons system 200 W, 50 kHz, Abrera, Barcelona, Spain). In the final step the solution was vortexed for 2 h at room temperature. The liposome solution had a high polydispersity, with multilamellar liposomes. In order to obtain a homogeneous size distribution solution with unilamellar liposomes, 1 mL of liposome solution was extruded 10 times with a manual mini extruder from Avanti Polar Lipids (Alabaster, AL, USA), using polycarbonate membranes of 100 and 200 nm (Whatman, Maidstone, UK). After extrusion, the solutions were maintained in the dark at 277 K for 24 h for a complete stabilization. In this work the calixarene/DOPE liposomes will be named CAL/DOPE liposomes.

Only in the case of (Im16)4 liposomes was a mixture of ethyl acetate:ethanol 1:1 used, instead of chloroform, because of solubility problems. Nonetheless, by using this mixture in the thin film hydration method, the characteristics of the liposomes containing (TEAC12)4, (Im12)4, and (Im3Im16)2 were similar to those observed using chloroform.

The composition (mole ratio) of the liposomes prepared is summarized in Table 1.


**Table 1.** Composition (mole ratio) of the CAL/DOPE liposomes prepared.

Mole ratio is expressed with respect to the nCAL.

#### *2.3. Preparation of Lipoplexes*

The lipoplexes were prepared by mixing appropriate volumes of the liposome solution and of the aqueous ctDNA (or p-EFGP-C1) HEPES 10 mM solutions in order to obtain the desired L/D ratio. For each α value, the mass ratio L/D is given by the expression:

$$\frac{\text{L}}{\text{D}} = \frac{\text{m}\_{\text{CAL}} + \text{m}\_{\text{DOPE}}}{\text{m}\_{\text{DNA}}} \tag{2}$$

where mDOPE, mCAL, and mDNA are the masses of the zwitterionic phospholipid, of the calixarene, and of the DNA, respectively, in the solution. In all the liposome solutions investigated, the mass of DNA was kept constant at 10−<sup>4</sup> g (the concentration was 1.0 mg/mL or 8.1 × <sup>10</sup>−<sup>5</sup> mol L−<sup>1</sup> given in base-pairs). The calixarene/DOPE/ctDNA lipoplexes will be named CAL/DOPE/DNA lipoplexes.

The stability of the lipoplexes was followed by changes in their size and polydispersity with time. The size remained unchanged for more than 48 h. The authors also checked the stability of the lipoplexes, of different compositions, after dilution with buffer HEPES 10 mM. No variations in their size were observed.

#### *2.4. Zeta Potential Measurements*

Zeta-potential, ζ, values were calculated measuring the electrophoretic mobility of the liposomes and of the lipoplexes from the velocity of the particles, using a laser Doppler velocimeter (LDV). A Zetasizer Nano ZS Malvern Instrument Ltd. (Malvern, Worcestershire, UK) was used. Temperature was kept at 303.0 ± 0.1 K, and DTS1060 polycarbonatecapillary cells were utilized. ctDNA concentration in the buffered solutions of liposomes was 8.1 × <sup>10</sup>−<sup>5</sup> M. Data are expressed as mean ± SD from at least three separate experiments, *n* = 9.

#### *2.5. Dynamic Light Scattering, DLS, Measurements*

A Zetasizer Nano ZS Malvern Instrument Ltd. (Worcestershire, UK) was used to estimate the hydrodynamic diameter, dH (Z average), and the polydispersity index, PDI, of the lipoplexes using DLS measurements. A scattering angle of 90◦ was used. A fixed concentration of 8.1 × <sup>10</sup>−<sup>5</sup> M of ctDNA was present in all the liposome solutions investigated. Data are expressed as mean ± SD from at least three separate experiments, *n* = 9. Temperature was maintained at 303.0 ± 0.1 K.

#### *2.6. Agarose Gel Electrophoresis*

Agarose gel (1%) was prepared in a TAE buffer (40 mM Tris-acetate, 1 mM EDTA) in a total volume of 180 μL and stained with the dye Red Safe (10 μL) for the visualization of the nucleic acid bands. The ctDNA concentration was kept constant at 8.1 × <sup>10</sup>−<sup>5</sup> M. The method was as follows: (i) 20 μL of the buffered liposome solution was mixed with 5 μL of 5 × DNA loading buffer. (ii) After homogenization, the resulting solution was added in each well. Electrophoresis was performed at 90 V for 90 min. A detector Ultima 16si (Hoefer Inc., Holliston, MA, USA) was used for visualizing the nucleic acid bands by irradiation with UV light (254 nm).

#### *2.7. Circular Dichroism, CD, Spectra*

A Biologic Mos-450 spectropolarimeter (Cambridge, UK) was used to register the CD spectra. Scans were taken from 220 to 310 nm with a standard quartz cell of 10 mm path length. Three independent experiments were done. Each spectrum was obtained from an average of 10 runs, with a 5 min equilibration before each scan, at 303.0 ± 0.1 K. The ctDNA concentration was kept constant at 8.1 × <sup>10</sup>−<sup>5</sup> M in the lipoplex solutions. All solutions were prepared in 10 mM HEPES buffer, pH = 7.4.

#### *2.8. Atomic Force Microscopy, AFM*

Atomic force microscopy was used to study the structures of the lipoplexes. A resonance frequency of around 240 KHz and a nominal force constant of 42 N/m were the working conditions in a Molecular Imaging PicoPlus 2500 AFM (Agilent Technologies, Santa Clara, CA, USA). Silicon cantilevers (Model Pointprobe, Nanoworld, Neufchâtel, Switzerland) were used. The images were recorded in air and in tapping mode. Data collection (256 × 256 pixels) was registered with scan speeds about 0.5 Hz. The ctDNA concentration in the buffered HEPES 10 mM liposome solutions, pH = 7.4, was 0.6 μM.

Images of the buffered liposome solutions were obtained using the following method: (a) In order to prepare a modified mica surface, 0.1% (*v*/*v*) APTES aqueous solution was dropped onto a freshly cleaved mica surface. It was washed with ultra-pure water after 20 min and air dried. (b) A 30 μL droplet of the lipoplex solution was deposited on the modified mica surface and incubated for 30 min. (c) Subsequently, the mica surface was washed with pure water and air dried for AFM imaging.

#### *2.9. Electron Transmission Microscopy, TEM*

TEM images of the lipoplexes were obtained in a Zeiss Libra 120 scanning electron microscope (Carl Zeiss AG, Oberkochen, Germany), at 80 kV. Samples were prepared by impregnation, using a 300 mesh copper grid coated collodion that, subsequently, was stained with a solution of uranyl acetate (2.0%). Images were processed with a bottom-mounted TEM CCD camera and recorded with a resolution of 2048 × 2048 pixels. ImageJ (National Institutes of Health (NIH), Bethesda, MD, USA) bundled with 64-bit Java 1.8.0\_172 was used to analyze TEM images from independent experiments, for each of the lipoplex solutions investigated.

#### *2.10. In Vitro Cytotoxicity Assays*

The cytotoxicity of the CAL/DOPE liposomes with α = 0.5 for (TEAC12)4, (Im12)4, and (Im16Im3)2, and with α = 0.3 for (Im16)4, at different L/D values, was estimated in vitro using the MTT assay [43]. These are the maximum α values that could be prepared for the different calixarenes, and they were chosen to carry out the cell viability assays because they correspond to the highest content of the cationic calixarene within the liposomes. The cell lines used were RPE-1 (normal cell line), A549 (adenocarcinomic human alveolar basal epithelial cell line), HepG2 (human liver cancer cell line), LS180 (adenocarcinomic human colonic epithelial cell line), and MCF7 (breast cancer cell line). They were a gift from different research groups from the IBIS (the Institute of Biomedicine of Seville). In any case, all cell lines used were from commercial suppliers. Cell lines were plated out into 96-well plates at a density of 3000 cells per plate. The next day, the liposome solutions were added to the wells, and the plate wa s returned to the incubator for 4 days more. Later, they were pulsed with MTS (ROCHE, Basilea, Switzerland). According to the manufacturer instructions, cell viability was measured by luminometry in a Varioskan Flash (Thermo Fisher Scientific, Waltham, MA, USA). Each liposome concentration was measured in triplicate.

#### *2.11. Transfection Assays*

The cell line chosen to carry out these experiments was the U2OS, from human osteosarcoma, because these cells are suitable for transfection experiments. The non-viral vectors investigated were (TEAC12)4/DOPE liposomes, containing the plasmid pEGFP-C1. Liposomes of (TEAC12)4 were selected because they are the least toxic of all the CAL/DOPE liposomes investigated. On the other hand, pEGFP-C1 is a plasmid carrying an enhanced GFP coding sequence with the required regulatory elements for constitutive expression of the gene in human cells. The method used to carry out the transfection experiments was as follows: 3 μg of pEGFP-C1 was added to a solution containing 180 μL of Opti-MEM (Gibco, Thermo Scientific, Waltham, MA, USA), and the amount of liposome buffered solution (HEPES 10 mM) necessary to obtain the L/D ratio for each α value was investigated. The resulting mixture was incubated at room temperature for 20 min, and afterwards it was added to a 50% confluent 6 cm plate with 3 mL of DMEM medium (Sigma Aldrich, Darmstadt, Germany).

The cells were transfected with a mixture of transfection reagent and Opti-MEM (not pEGFP-C1 included) as negative control. As positive control, FuGENE 6 transfection reagent (E2311, from Promega Corporation, Madison, WI, USA) was used, according to the manufacturer's protocol (i.e., 3 μg of pEGFP-C1 in 200 μL Opti-MEM plus 9 μL of FuGENE 6). Transfection efficiency was evaluated by flow cytometry with a FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA) 24 h after transfection.

#### *2.12. UV–Visible Spectroscopy*

The doxorubicin concentration was determined by UV–visible spectroscopy, using a Hitachi UV-visible 3900 (Chiyoda, Tokyo, Japan) by measuring absorbance at 490 nm. Temperature was kept using a Lauda (Stuttgart, Baden-Würtenberg, Germany) flow cryostat connected to the cell compartment.

#### *2.13. Encapsulation Efficiency Measurements*

A dialysis method was used to estimate the doxorubicin, DOX, encapsulation efficiency of the (TEAC12)4/DOPE liposomes. A total of 900 μL of drug-loaded liposomes was added to a Spectra/Por® 3 (MWCO 3.5 kDa) from Spectrum Laboratories, Inc. (Rancho Dominguez, CA, USA) The final concentration of antibiotic was 2 × <sup>10</sup>−<sup>4</sup> M. The dialysis membrane was plunged into a beaker containing 30 mL of 10 mM HEPES buffer (pH = 7.4), the same used for liposome hydration. The liposomes' stabilization was ensured by keeping the temperature at 4 ◦C throughout the process in order to avoid doxorubicin degradation [32]. An aliquot of 1 mL from the beaker was taken at different time intervals. These aliquots were replaced each time by an equal volume of buffer HEPES 10 mM in order to keep constant the total volume of buffer in the beaker. Dialysis was followed for at least 24 h. The encapsulation efficiency, EE, was calculated by using Equation (3):

$$\text{EE}(\%) = \frac{[\text{DOX}]\_{\text{Liposomes}}}{[\text{DOX}]\_{\text{T}}} \times 100 = \frac{[\text{DOX}]\_{\text{T}} - [\text{DOX}]\_{\text{buffer}}}{[\text{DOX}]\_{\text{T}}} \tag{3}$$

where DOXLiposomes, [DOX]T, and [DOX]buffer are the DOX concentration encapsulated into the calixarene-based liposomes, the total DOX concentration present in the system, and the DOX concentration in the buffer solution, respectively. All the concentrations are referred to the total volume of solutions.

The loading capacity, LC, can be calculated using the following expression:

$$\text{LC} = \frac{\text{п}\_{\text{DOX.enc.}}}{\text{п}\_{\text{DOX,total}} + \text{п}\_{\text{CAL} + \text{DOPE}}} \times 100\tag{4}$$

where nDOX.enc., nDOX,total, and nCAL+DOPE represent the moles of the encapsulated drug, total drug, and total lipids, respectively.

Quantification of doxorubicin concentration was done by UV–visible spectroscopy (λ = 490 nm). Each experiment was performed in triplicate.

#### *2.14. In Vitro Drug Release*

After applying the dyalisis method, the doxorubicin loaded (TEAC12)4/DOPE liposomes were suspended in buffer HEPES 10 mM, pH = 7.4, in a glass vial under continuous magnetic stirring, 200 rpm, at 37.4 ◦C (the human body temperature). A sample was removed at determined time intervals and, subsequently, replaced with an equal amount of buffer. This is a way to simulate the in vivo removal of a drug into a systemic circulation. The concentration of the antibiotic was estimated by UV–visible spectroscopy, measuring the absorbance at 490 nm. The absorbance data were corrected from the dilution effect. Three separate experiments were done. The precision was close to 7%.

#### *2.15. Statistical Analysis*

Values are expressed as the mean ± standard errors of independent experiments. Statistical analysis was performed with Student's t-test and one-way analysis of variance (ANOVA). When *p* < 0.05 (95% confidence) the differences were considered as significant.

#### **3. Results and Discussion**

#### *3.1. Calixarene-Based Liposomes*

Calixarene-based liposomes were prepared at several cationic CAL molar fractions α (α = nCAL/(nCAL+ nDOPE)). In particular, this magnitude was varied within the interval from 0.1 to 0.5. Molar ratios higher than 0.5 were not investigated because these liposome solutions showed a high polydispersity index, PDI (PDI > 0.8). Figure 1 shows the dependence of the hydrodynamic diameter of the calixarene-based liposomes on the calixarene molar ratio, α, in these nanostructures. The interval of α studied for each calixarene was limited by solubility problems. In the case of (Im16)4, it was particularly narrow. Figure 1 shows that the size of the CAL/DOPE liposomes initially decreased upon increasing α, but a subsequent increase in the molar ratio led to an increase in the liposome sizes. This trend was observed for all CAL/DOPE liposomes with the exception of (Im16)4, for which no clear trend was observed due to the narrow α interval studied.

**Figure 1.** Dependence of the hydrodynamic diameter, dH, on the cationic calixarene molar fraction, α. T = 303.1 ± 0.1 K.

The experimental observations could be explained by considering the main interactions controlling the liposome sizes. On one hand the hydrophobic interactions are the driving force for the formation of liposomes, and they favor a diminution in the liposome size [44]. On the other hand, the electrostatic repulsions between the positively charged head groups of the calixarenes in the bilayer make an increase in the liposome size more favorable. At low α values the hydrophobic interactions mainly control the liposome sizes, and an increment in α results in a diminution of the hydrodynamic diameter, dH. A further increment in α means an increase in the amount of cationic CAL within the liposomes, and, as a consequence, the electrostatic repulsions will augment. At a given α value, depending on the calixarene nature, the electrostatic repulsions overcome the hydrophobic interactions, and an increase in α is followed by an increase in dH since the cationic head groups tend to separate. The hydrophobic interactions are expected to be stronger in the case of (TEAC12)4 and (Im12)4 than for (Im3Im16)2. The latter has alternate chains, which makes the interaction between the two 16 C tails more difficult. Besides, two of the chains are short (three C atoms), this resulting in weak hydrophobic interactions when compared to those with 12 C atoms. Therefore, the minimum in the plots of dH vs. α is expected to be observed for higher α values in the case of (TEAC12)4 and (Im12)4 than in the case of (Im3Im16)2, as in fact is observed. Following the same reasoning, (Im16))4 should present the minimum at the highest α value. However, this is not observed.

Apart from the solubility problems, a possible explanation would be that the long hexadecyl chains could fold towards themselves, resulting in steric hindrance [45], which would make smaller liposomes more favorable.

The cytotoxicity of the CAL/DOPE liposomes was estimated by using the MTT assay. Since DOPE is considered a biocompatible lipid, the cell viability in several cell lines was

determined for the highest α value, corresponding to the liposomes with the largest amount of cationic calixarenes. Figure 2 shows the results obtained, from which some conclusions can be reached. (TEAC12)4/DOPE liposomes are the less toxic of all. From the comparison of the data corresponding to the (TEAC12)4/DOPE liposomes to those of (Im12)4, one can say that the substitution of quaternary ammonium groups by imidazolinium ones results in an increase in the cytotoxicity of the liposomes. A similar result was observed by Rodik et al. in a previous work [46]. On the other hand, an increase in the length of the hydrophobic chains attached to the lower rim of the calixarenes caused a diminution in the cell viability. This was particularly evident for the (Im16)4/DOPE liposomes. This behavior can be attributed to a higher lipophilicity, which results in a higher ability to penetrate and disrupt the cell membrane. It was particularly evident for the (Im16)4/DOPE liposomes. A similar trend has been previously observed by other authors [47]. Besides, Figure 2 shows that (Im3Im16)2/DOPE liposomes present a specific selectivity towards tumor cells, while (Im12)4/DOPE liposomes act particularly on MCF7 mammary cells and, at high concentrations, on HepG2 and A549 cells.

**Figure 2.** Cell viability values in the presence of different liposome concentrations at α = 0.5 for (TEAC12)4, (Im12)4, and (Im3Im16)2, and at α = 0.3 for (Im16))4. The results are the average of three independent experiments: (**A**) (TEAC12)4; (**B**) (Im12)4; (**C**) (Im3Im16)2; (**D**) (Im16)4.

#### *3.2. CAL/DOPE/DNA Lipoplexes*

In order to study the characteristics of the CAL/DOPE/DNA lipoplexes, the zeta potential, ζ, was measured for a given molar ratio, α, at different mass ratios L/D. Figure 3 shows the results obtained for the different calixarenes studied in this work. In all cases when L/D increases, the charge of the lipoplex goes from negative to positive. The solutions were stable, and no turbidity was observed in any system. This is an expected result since, for a given α, an increase in L/D means an increment in the amount of cationic lipid, CAL, in the liposomes. The charge inversion observed indicates that the polynucleotide interacts with the cationic calixarene within the lipoplex. From the data in Figure 3 it is possible to calculate (L/D)Φ, the value of the mass ratio corresponding to a zeta potential equal to zero; that is, when the lipoplexes are neutral. (L/D)<sup>Φ</sup> values can also be calculated theoretically using Equation (5). The deduction of this equation is described in the Supplementary Information.

$$
\begin{split}
\begin{pmatrix} \frac{1}{\mathsf{L}}\\ \mathsf{D} \end{pmatrix}\_{\mathsf{\Phi}} &= \frac{\mathsf{q}^{-}\_{\mathsf{ADN}}}{\mathsf{q}^{+}\_{\mathsf{CAL}}} \times \frac{\mathsf{M}\_{\mathsf{CAL}}}{\mathsf{M}\_{\mathsf{pp}}} \times \frac{(\mathsf{n}\_{\mathsf{CAL}}\mathsf{M}\_{\mathsf{CAL}} + \mathsf{n}\_{\mathsf{DOPE}}\mathsf{M}\_{\mathsf{DOFE}})}{(\mathsf{n}\_{\mathsf{CAL}} + \mathsf{n}\_{\mathsf{DOPE}})} \times \frac{1}{\frac{\mathsf{n}\_{\mathsf{CAL}}}{(\mathsf{n}\_{\mathsf{CAL}} + \mathsf{n}\_{\mathsf{DOPE}})}}
\end{split}
$$

$$
&= \frac{\mathsf{q}^{-}\_{\mathsf{ADN}}}{\mathsf{q}^{+}\_{\mathsf{CAL}}} \times \frac{\mathsf{M}\_{\mathsf{CAL}}}{\mathsf{M}\_{\mathsf{bp}}} \times \frac{\alpha \mathsf{M}\_{\mathsf{CAL}} + (1 - \alpha)\mathsf{M}\_{\mathsf{DOPE}}}{\alpha}
\end{split}
\tag{5}
$$

where q− ADN and <sup>q</sup><sup>+</sup> CAL are the charges of the polynucleotide and of the CAL, respectively. The charge of calf thymus DNA is considered to be−2 per base-pairs [48]. MCAL and MDOPE are the molecular weights of the cationic calixarene and non-ionic lipid, respectively, Mbp being the polynucleotide molecular weight per base-pair. nCAL and nDOPE are the number of moles the CAL and DOPE, respectively. The rest of the symbols have been previously defined. Table 2 summarizes the experimental and the theoretical (L/D)<sup>Φ</sup> values. One can see that, within experimental errors, the theoretical and the experimental values agree quite well. It is important to know the charge of the different lipoplexes prepared because one of the requirements for an efficient cellular uptake is that the charge of the nanocarrier (the lipoplexes in this work) has to be positive in order to cross the negatively charged cellular membrane [49].


**Table 2.** Theoretical and experimental (L/D)Φ values.

<sup>a</sup> Theoretical values; <sup>b</sup> Experimental values.

Charge inversion of DNA in the lipoplexes can also be investigated by gel electrophoresis. Figure S1 (Supplementary Information) shows the results obtained by using this technique. In this figure a migration to the anode is observed for L/D values lower than (L/D)Φ, for the systems investigated, although a diminution in the mobility of the band is found when L/D approximated to (L/D)Φ. Once this value is reached, the mobility is hindered, this pointing out that the charge of the polynucleotide has been inverted from negative to positive. These results are in agreement to those found by zeta potential measurements (see Figure 3).

**Figure 3.** Dependence of the relative zeta potential, −(ζ/ζo), of the CAL/DOPE/DNA lipoplexes on L/D for different molar ratios α. T = 303.0.1 ± 0.1 K. (**A**) (TEAC12)4; (**B**) (Im12)4; (**C**) (Im3Im16)2; (**D**) (Im16)4.

Another important magnitude is the size of the lipoplexes. Recent studies indicated that there is a particular size range adequate for cellular uptake [18,50]. Although cellular uptake depends on the type of cells and on the different barriers making this process difficult [49], the appropriate nanocarrier size is usually considered to be a few hundred nanometers. Figure 4 shows the dependence of the hydrodynamic diameter, dH, on the mass ratio L/D, for a given α value, for the calixarenes investigated. In all cases a Gaussian dependence of dH on L/D was observed. This dependence can be explained as follows. At low as well as at high L/D values, the lipoplexes are negatively and positively charged, respectively (see Figure 3). That is, there are repulsive forces among them that kept them apart and maintained a stable size distribution. When L/D is approaching (L/D)Φ, the charge of the lipoplexes is moving closer to zero. As a consequence, the lipoplexes do not repel each other and an aggregation process occurs, a steep increment in dH being observed. This explanation is supported by Figure S2 (Supplementary Material), which shows the dependence of the relative zeta potential, (ζ/ζo), and of the hydrodynamic diameter, dH, of CAL/DOPE/DNA lipoplexes on L/D for α = 0.2. Apart from the L/D values close to (L/D)Φ, the lipoplex sizes observed for the different molar ratios α, and for all the CAL studied, are within the hundred nanometers size range.

A way of getting information about the conformational changes of the DNA in the lipoplexes, when the mass ratio L/D varies, is by circular dichroism, CD. Figure 5 shows the CD spectra of the CAL/DOPE/DNA lipoplexes for different α and L/D values. First, it was checked that the liposomes in the absence of DNA did not contribute to the spectra. All CD spectra were run taking as reference an aqueous buffer solution HEPES 10 mM, at pH = 7.4. Figure 5 shows the CD spectrum corresponding to pure DNA in aqueous buffered solution of HEPES 10 mM. This spectrum presents a negative band, at about 247 nm, due to the right-handed helicities of the polynucleotide, and a positive band, close to 280 nm, coming from the π–π stacking interactions between the bases. This spectrum is in agreement with that expected for the right-handed B form of the double-stranded ctDNA [51]. For all the CAL/DOPE/DNA lipoplexes investigated, a diminution in the positive band intensity upon increasing L/D was observed. The (Im16)4/DOPE/DNA lipoplexes could not be studied at higher L/D values because of solubility problems. This makes the comparison of the results obtained for this calixarene with those corresponding to the rest of the macrocycles more difficult.

The dependence of the positive band intensity on L/D could be explained considering the attractive electrostatic interactions between the DNA phosphate groups and the positively charged calixarenes within the lipoplexes. These interactions could cause the opening of the DNA double strand and conformational changes in the polynucleotide. An increment in L/D is accompanied by an increase in the amount of cationic calixarene in the lipoplexes. Therefore, it would be expected that the diminution in the positive band intensity was larger, for a given α value, the higher L/D is, as is observed.

The displacement of the inflection point (observed at 260 nm for pure DNA) towards higher wavelengths as well as the increase in the negative band intensity are usually related to the DNA denaturation and to DNA conformational changes [52,53]. Bombelli et al. found a similar dependence of the DNA spectrum in gemini surfactants/DOPE/DNA lipoplexes on L/D [54]. These authors proposed that the gemini surfactant/DNA interactions in the liposomes causes a conformational DNA change from a B form to a more condensed Ψ phase, where the polynucleotide molecules are partially inserted within an inverted hexagonal lipid rearrangement, which gives the DNA a certain spatial organization and a fixed directionality.

**Figure 4.** Dependence of the hydrodynamic diameter, dH, of the CAL/DOPE/DNA lipoplexes on L/D for different molar ratios α. T = 303 ± 0.1 K. (**A**) (TEAC12)4; (**B**) (Im12)4; (**C**) (Im3Im16)2; (**D**) (Im16)4.

**Figure 5.** Dependence of the circular dichroism spectra of the CAL/DOPE/DNA lipoplexes on L/D for different molar ratios α. T = 303 ± 0.1 K.; (**A**) α= 0.20 for (TEAC12)4; (**B**) α = 0.50 for (TEAC12)4; (**C**) α = 0.20 for (Im12)4; (**D**) α = 0.50 for (Im12)4; (**E**) α = 0.20 for (Im3Im16)2; (**F**) α= 0.50 for (Im3Im16)2; (**G**) α = 0.20 for (Im16)4; and (**H**) α = 0.30 for (Im16)4.

The displacement of the inflection point wavelength diminishes when α increases. This could be due to the effects of the presence of DOPE on the DNA conformation, as was pointed out by Marty et al. [55] in different liposome formulations. It is also observed in Figure 4 that the wavelength displacement is lower for (Im12)4 than for (TEAC12)4 (14 nm vs. 6 nm, respectively). Both calixarenes have the same hydrophobic tail length. This observation was explained considering that the imidazolinium groups were intercalated, at least partially, between the DNA base pairs, this stabilizing the B form of the DNA [56]. At this point it is worth noting that there are not enough experimental results for (Im16)4 that permit the comparison with the other CAL.

The circular dichroism results seem to indicate that the lipoplex formation results in DNA conformational changes. In order to support this hypothesis, atomic force microscopy measurements, AFM, were carried out. Figure 6 shows the AFM images corresponding to pure DNA (Figure 6A), together with those of CAL/DOPE/DNA lipoplexes of different compositions, for two of the calixarenes investigated (Figure 6B–E). In the absence of liposomes, the pure DNA presents an elongated form (see Figure 6A). For (Im3Im16)2/DOPE/DNA and (TEAC12)4/DOPE/DNA lipoplexes at α = 0.2 for L/D = 1 (Figure 6B,D), which is lower than (L/D)Φ, some globular structures are observed, and the length of the DNA seems to be somewhat shorter, although no substantial conformational variations are observed. Yan et al. [57] explained the formation of the globular structures, linked across DNA strands, on the basis of an increased bending of the DNA double helix, this leading to the formation and stabilization of intramolecular loops. The separation of the double DNA strand into single strands, due to electrostatic attractions between the DNA and the cationic calixarenes, favors this process. For L/D = 7, (L/D) > (L/D)Φ, the number of globular structures present in the images increases. However, a full condensation of the polynucleotide is not reached (Figure 6C,E) since some elongated fragments of DNA are still observed. These experimental observations are in agreement with the CD spectra.

**Figure 6.** AFM topographic images of CAL/DOPE/DNA lipoplexes in buffered solutions, 10 mM HEPES (pH = 7.4), adsorbed on APTES modified mica surface. (**A**) Pure DNA; (**B**) (Im3Im16)2 α = 0.2 and L/D = 1; (**C**) (Im3Im16)2 α = 0.2 and L/D = 7; (**D**) (TEAC12)4 α = 0.2 and L/D = 1; (**E**) (TEAC12)4 α = 0.2 and L/D = 7.

With the goal of visualizing the morphology of the CAL/DOPE liposomes and CAL/DOPE/DNA lipoplexes, transmission electron microscopy measurements were carried out. Figure 7 shows the TEM images obtained for some of the systems investigated. One can see that a spherical morphology is observed for both the liposomes and the lipoplexes studied. The molar ratio α = 0.3 was chosen because Figure 1 shows that, for the two calixarenes studied, the minimum size is found close to this molar ratio value. It is interesting to indicate that (TEAC12)4/DOPE and (Im12)4/DOPE liposomes are the less cytotoxic among the calixarene-based liposomes investigated. For the mass ratio L/D = 5, the charge inversion of the polynucleotide is complete for (TEAC12)4/DOPE/DNA and (Im12)4/DOPE/DNA lipoplexes, at α = 0.3 (see Figure 3). Figure 7 shows that the two lipolexes not only are spherical, but their size is in the order of a few hundred nanometers. This is important in relation with the use of lipoplexes for gene delivery since it has been shown that a spherical morphology and a small size are two characteristics of the genetic material nanocarriers that favor transfection efficiency [49]. Besides, the liposome and lipoplex sizes measured using DLS (Figure 4) and TEM are in agreement, as one can see in Table 3.

**Figure 7.** TEM images of the following systems: (**A**) (TEAC12)4/DOPE liposomes with α = 0.3; (**B**) (Im12)4/DOPE liposomes with α = 0.3; (**C**) (TEAC12)4/DOPE/DNA lipoplexes with α = 0.3 and L/D = 5; (**D**) (Im12)4/DOPE/DNA lipoplexes with α = 0.3 and L/D = 5.

**Table 3.** Sizes of various liposomes and lipoplexes, with a molar ratio α = 0.3, measured by dynamic light scattering, DLS, and electronic transmission microscopy.


#### *3.3. Transfection Efficiency of CAL/DOPE/pDNA Lipoplexes*

The transfection experiments were carried out for the lipoplexes containing the least cytotoxic calixarene: (TEAC12)4. Before carrying out these measurements, the cell viability of the liposomes for the human bone osteosarcoma epithelial cells U2OS using the MTT assay was carried out. This is the cell line used in the transfection experiments because it is considered an easy-to-transfer cell line. Figure 8 shows the results obtained. In this figure, the cytotoxicity of the (TEAC12)4/DOPE liposomes in the presence of additional DOPE (+1/4 of the DOPE amount present in the liposomes) is also presented. These systems were investigated because in the transfection experiments the addition of DOPE could make the delivery of genetic material more efficient [58]. Therefore, given that the transfection efficiency, TE, of the lipoplexes was studied in the presence of this phospholipid, the cell viability experiments in the presence of different cell lines for (TEAC12)4/DOPE liposomes + DOPE was also carried out.

**Figure 8.** Dependence of the cell viability (%) of (TEAC12)4/DOPE liposomes in U2OS cancer cell line, at 48 h, on the concentration of (TEAC12)4 within the liposomes, for a constant cationic lipid molar ratio α = 0.3. The experiments were done in triplicate. (**A**) In the absence of additional DOPE; (**B**) in the presence of additional DOPE (+1/4 of the DOPE amount present in the liposomes), which is not forming part of the (TEAC12)4/DOPE liposomes.

One can see that, for α = 0.3, the presence of additional DOPE, which is not present in the liposomes, substantially diminishes the cell viability of (TEAC12)4/DOPE liposomes. In the absence of additional DOPE, the cell viability of the (TEAC12)4/DOPE liposomes is lower than 60% for concentrations of the cationic calixarene [(TEAC12)4] ≥ <sup>20</sup> <sup>μ</sup>g mL<sup>−</sup>1. However, the addition of DOPE to the system caused the cell viability of the (TEAC12)4/DOPE liposomes to be lower than 60% for cationic calixarene concentrations [(TEAC12)4]>5 μg mL−1. That is, keeping the cationic calixarene molar ratio α equal to

0.3, the cell viability substantially decreases when the [(TEAC12)4] within the liposomes increases if additional DOPE is present.

The transfection process of the plasmid pEGFP-C1 was carried out on the U2OS cells. The TE of the (TEAC12)4/DOPE/pEGFP-C1 lipoplexes, with α = 0.3, within a L/D range between 9 and 90, in the presence as well as in the absence of additional DOPE, was estimated. Expression of GFP is frequently used to follow transfection, and it does not require any additional manipulation of the sample since GFP is an intrinsically fluorescent protein. Therefore, its fluorescence can be readily measured directly. The lowest L/D value studied was 9, with the idea of assuring that the lipoplexes have a positive charge, a requirement to cross the cell membrane. On the other hand, L/D values higher than 90 were not investigated because this would mean a high amount of cationic calixarene present in the liposomes, this increasing cytotoxicity (see Figure 2). In regard to the introduction of additional DOPE in order to improve the TE, amounts of DOPE up to 1/4 of that present in the lipoplexes were investigated. Higher additional phospholipid amounts were not studied to avoid a further increase in cytotoxicity (see Figure 8).

Figure 9 shows that for the mass ratio L/D = 9 no transfection was observed in the absence as well as in the presence of additional DOPE. Negligible TE values were also found for L/D values lower than 90. For this reason, they are not shown in Figure 9. However, for L/D = 90 a low TE was observed, close to 3%. This TE is much lower than that of the FuGENE 6 reagent. When DOPE was added (1/4 of the amount of phospholipid present in the lipoplexes), the TE for L/D = 90 increased up to 16%. For additional DOPE amounts lower than 1/4 no changes in the TE for L/D=9 were observed. The increment in the TE caused by the addition of the helper lipid DOPE could be due to an increase in the stabilisation of the interactions cationic lipid/p-EGFP-C1 [59,60]. Besides, the addition of DOPE could also make the transfer of the genetic material in the context of endosomal escape more favorable, due to its fusogenic character [61].

**Figure 9.** Percentage of GFP-positive cells after transfection with 3 μg of p-EGFP-C1, 24 h post transfection; 10,000 cells were analyzed per condition by flow cytometry (FACS). (a) (TEAC12)4/ DOPE/pEGFP-C1 liposomes at α= 0.30; and (b) (TEAC12)4/DOPE/pEGFP-C1 liposomes + DOPE at α = 0.30.

Figure 10 shows the FACS analyses of cells transfected with 3 μg of p-EGFP-C1 with the indicated reagents. Representative images of GFP-positive cells after transfection are shown in Figure 11.

**Figure 10.** FACS analysis of cells transfected with 3 μg of p-EGFP-C1 for the different reagents. The charts show green fluorescent emission (X axis) vs. red fluorescent emission (Y axis) of 10,000 live cells analyzed per condition 24 h post-transfection. The GFP gate defines the area where cells with a clear increase in their green fluorescent emission are observed without a parallel increase in their red fluorescent emission. (**A**) Control; (**B**) FuGENE 6; (**C**) (TEAC12)4/DOPE/pEGFP-C1 liposomes at α = 0.30 for L/D = 90; and (**D**) (TEAC12)4/DOPE/pEGFP-C1 liposomes + DOPE at α = 0.30 for L/D = 90.

**Figure 11.** Representative images of GFP positive cells after transfection with 3 μg of pEPFG-C1. (**A**) Control; (**B**) FuGENE 6; (**C**) (TEAC12)4/DOPE/pEGFP-C1 lipoplexes at α = 0.3 and L/D = 90; (**D**) (TEAC12)4/DOPE/pEGFP-C1 lipoplexes at α = 0.3 and L/D = 90 in the presence of additional DOPE. Scale bar: 10 μm.

#### *3.4. Encapsulation of Doxorubicin*

Doxorubicin (see Scheme 2), DOX, is an antineoplastic drug displaying a strong antitumoral activity against a wide spectrum of human cancers. It is used in the treatment of various lung, breast, or ovarian cancers. It is also used in chemotherapy for leukemia and lymphomas [62–64]. DOX intercalates between the base-pairs of the DNA, this resulting in the inhibition of the synthesis and transcription of the genetic material. The result is the blocking of the enzyme topoisomerase II, which hindered the division and growing of cells. Besides, the interactions DOX/DNA cause variations in the chromatin structure, which triggers apoptosis in cells [65]. In spite of the beneficial DOX activity, its clinical use is limited by its side effects. Gastrointestinal toxicity, stomatitis, myelosuppression, or cardiotoxicity are some of the most frequent side effects caused by the treatment with DOX [66–68].

The study of the encapsulation of drugs within nanocarriers is of great interest because it could permit the transportation of the drug towards its therapeutic target but, simultaneously, diminishing the drug side effects. One of the most frequently used nanocarriers are liposomes, particularly in the case of doxorubicin [69–71]. In fact, there is a commercial liposome preparation for DOX administration in chemotherapy called Doxil® [72]. Other types of nanovehicles have also been used to administer doxorubicin [73–76]. In this work, the (TEAC12)4/DOPE liposomes were used to study the encapsulation and the release of doxorubicin. These liposomes were chosen because of the low cytotoxicity they present, which is one of the requirements of nanocarriers for biomedical applications.

Before studying the encapsulation of doxorubicin within the calixarene-based liposomes, the stability of these nanostructures was investigated. Stability was followed by DLS measurements, through the dependence of the hydrodynamic diameter, dH, and the polydispersity, PDI, on time, at 310 K (simulating the human body temperature). Figure 12 shows the results. One can see that the liposomes were stable during approximately 6 days. After that time the size as well as the PDI increased, this indicating that the system was not stable for longer times. The results observed in Figure 12 could be explained by the fragmentation of the lipid membrane of the liposomes due to the hydrolytic decomposition of the phospholipid molecules. As a consequence, their structure will vary, and an increment in the surface of the liposome membranes can occur, this causing an increase in both the hydrodynamic diameter and the polydispersity.

The encapsulation of doxorubicin within the (TEAC12)4/DOPE liposomes was done during the hydration process of the lipid bilayer (thin lipid film method). A DOX buffered solution (HEPES 10 mM, pH = 7.4) was added to the dry lipid bilayer. Afterwards, the vortex-sonication cycles were carried out, followed by the extrusion process. The final DOX concentration was 2 × <sup>10</sup>−<sup>4</sup> mol L−<sup>1</sup> in all cases. In order to estimate the amount of DOX encapsulated, the drug-loaded liposomes were dialyzed (see Section 2.12). The doxorubicin concentration was determined by UV–visible spectroscopy measuring absorbance at 490 nm. The temperature was kept at 277 K in order to avoid doxorubicin degradation. The results obtained are summarized in Table 4.

**Table 4.** Encapsulation efficiency, EE%, and loading capacity, LC, of doxorubicin in the (TEAC12)4/ DOPE liposomes. T = 277 K. Three independent experiments were carried out for each system studied.


Table 4 shows that for the two molar fractions α investigated, the encapsulation efficiency was high. EE% increased when α augmented. This experimental observation could be explained by considering the interactions between (TEAC12)4 and doxorubicin, which were investigated in a previous study [30]. An increase in the amount of the antineoplastic drug in the liposome, an increase in α, will favor these interactions, this

leading to an increment in the encapsulation efficiency. The loading capacity also followed the same trend as EE%, and the explanation is similar to that given above.

**Figure 12.** Variation of the hydrodynamic diameter, dH, and of the polydispersity, PDI, with time for (TEAC12)4/DOPE liposomes at 310 K. (**A**) α = 0.2; (**B**) α = 0.4.

The size and the polydispersity, PDI, of the (TEAC12)4/DOPE liposomes with and without loaded DOX were compared. The measurements were done by DLS, and the data are listed in Table 5. One can see that the hydrodynamic diameter as well as the PDI of the liposomes were similar in the absence and in the presence of doxorubicin.

**Table 5.** Hydrodynamic diameter, dH, and polydispersity, PDI, of (TEAC12)4/DOPE liposomes and doxorubicin loaded (TEAC12)4/DOPE liposomes. T = 310 K. The results are the average of three independent experiments.


<sup>a</sup> (TEAC12)4/DOPE liposomes. <sup>b</sup> DOX loaded (TEAC12)4/DOPE liposomes.

When a drug is loaded in a nanocarrier, it is important to study the release time of the drug. The method used to investigate the release was described in the Experimental section, and it was carried out also at 310 K, in order to mimic the human body temperature. Figure 13 shows the variations of EE% against time for doxorubicin loaded (TEAC12)4/DOPE liposomes with molar fractions 0.2 and 0.4. The concentration of doxorubicin was estimated measuring the absorbance at 490 nm. From the variations of EE% with time, it is possible to deduce that the release follows a pseudo-first-order kinetics.

**Figure 13.** Release of doxorubicin from DOX loaded (TEAC12)4/DOPE liposomes at 310 K. (**A**) α = 0.2; (**B**) α = 0.4. The values are the average of three independent experiments.

The following kinetic rate constants were estimated: 1.9 <sup>×</sup> <sup>10</sup>−<sup>4</sup> min−<sup>1</sup> and 1.6 <sup>×</sup> <sup>10</sup>−<sup>5</sup> min−<sup>1</sup> for α = 0.2 and α = 0.4, respectively. Figure 13 shows that not all the doxorubicin was released from the liposomes. This could be explained by considering that, once the liposomes are fragmented, CAL molecules free from the lipid bilayer will be present in the solution. (TEAC12)4 can form different aggregates at low CAL concentrations, and DOX could be bound to them [30]. This hypothesis is in agreement with the results shown in Figure 13, since less doxorubicin was released in the case of α = 0.4 than of α = 0.2. An increase in the molar ratio α means an increment in the amount of calixarene present in the liposomes. Therefore, once the liposomes are broken, there will be a larger number of free CAL molecules to associate with the doxorubicin and, consequently, the amount of antineoplastic drug released from the liposomes will be lower.

The results obtained indicate that (TEAC12)4/DOPE liposomes can be used as noncytotoxic nanocarriers for the antineoplastic drug doxorubicin. Besides, even when liposomes are fragmented, the DOPE as well as the (TEAC12)4 molecules are non-cytotoxic [30]. The half-lives of the release were 2.5 and 3 days for α = 0.2 and α = 0.4, respectively, this pointing out that the side effects of the doxorubicin could be diminished due to a controlled release of the drug encapsulated as compared to the use of the naked drug.

#### **4. Conclusions**

In this work the formation of calixarene-based liposomes, CAL/DOPE, was investigated. Calixarenes with hydrophobic chains of different length attached to their lower rim were considered. The nature of the hydrophilic head present in their upper rim was also changed. The phospholipid DOPE was used, together with the cationic calixarene, for forming the lipid bilayer of the liposomes. The liposomes were characterized using several techniques. TEM images showed their spherical morphology. Cell viability experiments

permitted the estimation of the cytotoxicity of the CAL/DOPE liposomes of different compositions, the results showing that the (TEAC12)4 liposomes are the least cytotoxic.

The formation and characterization of CAL/DOPE/DNA lipoplexes of different compositions was investigated and the nanostructures characterized. They have a spherical geometry and a size on the order of a few hundred nanometers. Subsequently, transfection experiments were carried out only for the least cytotoxic calixarene, (TEAC12)4. Results showed that some of these (TEAC12)4/DOPE/p-EFGP-C1 lipoplexes can transfect, although the transfection efficiency, TE, is low. However, the presence of an additional amount of DOPE substantially increases the TE. This could be explained considering that the presence of DOPE can stabilize the interactions between the cationic lipid and the plasmidic DNA. Besides, the addition of DOPE could also make the transfer of the genetic material in the context of endosomal escape more favorable, due to its fusogenic character.

The antineoplastic agent doxorubicin was encapsulated in the (TEAC12)4/DOPE liposomes with high encapsulation efficiencies. The liposomes were stable for close to 6 days, at 310 K (the human body temperature). The drug release was studied, and the results showed that the liposomes can be utilized for a controlled release of the drug. Their use could suppose a diminution in its side effects.

The results obtained also show that the calixarene-based liposomes seem to be better nanocarriers, for both nucleic acids and doxorubicin, than their aggregates (micelles and vesicles), formed by the naked calixarenes.

Future investigations will be oriented to the design and preparation of new noncytotoxic amphiphilic calixarenes with the goal of using the CAL/DOPE/DNA lipoplexes as nanocarriers for the delivery of genetic materials, with a high transfection efficiency. The CAL/DOPE liposomes can also be checked as nanovehicles for different drugs.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/pharmaceutics13081250/s1, Deduction of Equation (5); Figure S1: Electrophoretic mobility shift assay on an agarose gel (1%) for CAL/DOPE/DNA lipoplexes; Figure S2: Dependence of the relative zeta potential, (ζ/ζo), and of the hydrodynamic diameter, dH, of CAL/DOPE/DNA lipoplexes on L/D for α = 0.2. T = 303.0.1 ± 0.1 K.

**Author Contributions:** Conceptualization, M.L.-L., P.L.-C. and M.L.M.; methodology, F.J.O., J.A.L., C.B.G.-C., I.V.R., M.L.-L., F.R.B., P.H., P.L.-C. and M.L.M.; software, J.A.L., F.J.O., E.B. and P.H.; validation, M.L.-L., P.L.-C., P.H. and M.L.M.; formal analysis, F.J.O., J.A.L., E.B., M.L.-L., P.L.-C. and M.L.M.; investigation, F.J.O., J.A.L., C.B.G.-C., I.V.R., R.V.R., V.I.K., F.R.B., E.B., M.L.-L., P.L.-C., P.H. and M.L.M.; resources, P.L.-C., P.H. and M.L.M.; data curation, F.J.O., J.A.L. and M.L.M.; writing original draft preparation, F.J.O., M.L.-L., P.L.-C., P.H. and M.L.M.; writing—review and editing, F.J.O., M.L.-L., P.L.-C., P.H., R.V.R., V.I.K. and M.L.M.; visualization, F.J.O., M.L.-L., P.L.-C., P.H. and M.L.M.; supervision, F.J.O., M.L.-L., P.L.-C. and M.L.M.; project administration, M.L.-L., P.L.-C. and M.L.M.; funding acquisition, P.L.-C., P.H., V.I.K. and M.L.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was financed by the Consejería de Conocimiento, Innovación y Universidades de la Junta de Andalucía (FQM-206, FQM-274, and PY20-01234), the VI Plan Propio Universidad de Sevilla (PP2019/00000748), RTI2018-100692-B-100; P18-RT-1271; PI18-0005-2018; VI-PP AY.SUPLEM-2019; RYC-2015-18670, The R+D+I grant PID2019-104195G from the Spanish Ministry of Science and Innovation-Agencia Estatal de Investigación/10.13039/501100011033 (P.H.) and the European Union (Feder Funds). The authors thank the University of Seville for the grant VPPI-US. J.A.L. also thanks the Fundación ONCE funded by the Fondo Social Europeo.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Poly(L-Lactic Acid)-co-poly(Butylene Adipate) New Block Copolymers for the Preparation of Drug-Loaded Long Acting Injectable Microparticles**

**Vasiliki Karava 1, Aggeliki Siamidi 1, Marilena Vlachou 1,\*, Evi Christodoulou 2, Nikolaos D. Bikiaris 2, Alexandra Zamboulis 2, Margaritis Kostoglou 3, Eleni Gounari 4,5 and Panagiotis Barmpalexis <sup>6</sup>**


**Abstract:** The present study evaluates the use of newly synthesized poly(L-lactic acid)-copoly(butylene adipate) (PLA/PBAd) block copolymers as microcarriers for the preparation of aripiprazole (ARI)-loaded long acting injectable (LAI) formulations. The effect of various PLA to PBAd ratios (95/5, 90/10, 75/25 and 50/50 *w*/*w*) on the enzymatic hydrolysis of the copolymers showed increasing erosion rates by increasing the PBAd content, while cytotoxicity studies revealed non-toxicity for all prepared biomaterials. SEM images showed the formation of well-shaped, spherical MPs with a smooth exterior surface and no particle's agglomeration, while DSC and pXRD data revealed that the presence of PBAd in the copolymers favors the amorphization of ARI. FTIR spectroscopy showed the formation of new ester bonds between the PLA and PBAd parts, while analysis of the MP formulations showed no molecular drug–polyester matrix interactions. In vitro dissolution studies suggested a highly tunable biphasic extended release, for up to 30 days, indicating the potential of the synthesized copolymers to act as promising LAI formulations, which will maintain a continuous therapeutic level for an extended time period. Lastly, several empirical and mechanistic models were also tested, with respect to their ability to fit the experimental release data.

**Keywords:** long acting injectables; poly(L-lactic acid); poly(butylene adipate); block copolymers; aripiprazole; microparticles; sustained release

#### **1. Introduction**

In the past few decades synthetic polymers that degrade under physiological conditions (i.e., biodegradable polymers) have become increasingly common in medical and pharmaceutical applications [1–3]. Especially, in the case of particulate drug formulations (such as nano- or microparticles) an increasing number of polymeric materials, and especially polyesters, have been introduced and implemented for drug delivery [4–6]. Amongst them, the preparation of long-acting injectable (LAI) formulations is probably the most intensively studied application for such polyester-based systems. In general, LAIs' are being utilized to reduce drug administration frequency, resulting to higher patient compliance.

**Citation:** Karava, V.; Siamidi, A.; Vlachou, M.; Christodoulou, E.; Bikiaris, N.D.; Zamboulis, A.; Kostoglou, M.; Gounari, E.; Barmpalexis, P. Poly(L-Lactic Acid)-co-poly(Butylene Adipate) New Block Copolymers for the Preparation of Drug-Loaded Long Acting Injectable Microparticles. *Pharmaceutics* **2021**, *13*, 930. https://doi.org/10.3390/ pharmaceutics13070930

Academic Editors: Francisco José Ostos, José Antonio Lebrón and Pilar López-Cornejo

Received: 19 May 2021 Accepted: 21 June 2021 Published: 23 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Compared to other formulations, LAIs can provide a prolonged and constant therapeutic effect, enhance the biological half-life of drugs, improve bioavailability and protect the active pharmaceutical ingredients (APIs) against harsh environmental conditions [7–9]. In addition, compared to other materials, such as lipids, polyesters can be customized more easily in order to adapt to any specific type of drug [4]. However, despite the numerous scientific reports, the fact that only about 20 different LAI products are available in the market, suggests that the design of such drug formulations is a rather difficult task [10].

In this context, perhaps the most widely explored polyester used is poly(L-lactide) (PLA) and its copolymer with glycolic acid (i.e., poly (lactic-co-glycolic acid), PLGA) [11,12]. In general, PLA or PLGA LAI formulations have been investigated as suitable matrix/carriers to deliver a variety of APIs, including small molecules, peptides and proteins for periods ranging from one week to several months [13–19]. In the last two decades several PLA or PLGA-based products have been brought into the market [11,20,21]. The characteristics of the polymer, such as molecular weight (MW), copolymer composition, terminal groups functionality and glass-transition temperature (Tg) are the key factors affecting its biodegradability and, hence, release kinetics. However, despite their inherent flexibility, PLA and PLGA-based LAIs face a number of challenges, including initial burst release, enhanced lag-time, incomplete drug dissolution and poor drug stability during both production and storage [10,11]. In an attempt to overcome these limitations, a wide range of suitable biodegradable polymers, including poly-ε-caprolactone, polyorthoesters, polydioxanones, polyphosphazenes, polyanhydrides, poly(acyanoacrylates), polyiminocarbonates, polyoxalates and polyurethanes, have been proposed as alternatives. Among them, poly(alkylene adipate) derivatives were only recently introduced showing promising results [22,23].

In general, poly(alkylene adipate)s, derived from dicarboxylic acids and different aliphatic diols, such as poly(ethylene adipate) (PEAd)), poly(propylene adipate) (PPAd) and poly(butylene adipate) (PBAd), seem to be a promising PLA or PLGA substitute in terms of ecological and economic (balance of cost–benefit) factors [24–27]. In a recently published attempt, the use of poly(alkylene adipate)s, as sole matrix/carriers for the preparation of drug LAI microparticle formulations showed promising results, in terms of efficacy, although incomplete drug dissolution was recorded in addition to rather short sustained action (a plateau was reached in dissolution at 3 days) [22]. These results indicate that a certain amount of tunning is needed in order for this type of polyesters to be suitable for LAI formulations. Similar results were also obtained from another study, where the combination of poly(butylene adipate) (PBAd) with PLA in the form of a physical blend was utilized in order to prepare novel electrospun nanofibrous matrices for the sustained delivery of the immunomodulatory drug, teriflunomide [23]. In this study, a controlled release pattern of the drug was achieved and varied analogous to the proportion of the PBAd and the drug content.

In view of these findings, we recently published a study on the synthesis of a new block copolymer with enhanced physicochemical and mechanical performance, based on butylene adipate segments [28]. Specifically, block copolymers of PBAd combined with PLA (Figure 1a) were synthesized via a two stage polycondensation and analyzed in regard to thermal and mechanical properties. The results showed that the continuity of the two polymers throughout the copolymer volume and the semicrystalline morphology were both easily tuned by either the preparation method conditions and the ratio of PBAd to PLA to the drug. Based on these features it can be assumed that the new prepared block copolymer may be a promising candidate for the preparation of drug-loaded LAI microparticles (MPs).

**Figure 1.** Chemical structures of (**a**) PLA/PBAd copolymers and (**b**) aripiprazole (ARI).

In the present study the use of the recently synthesized PLA/PBAd block copolymer was evaluated as a suitable LAI carrier. Aripiprazole (ARI, Figure 1b), a second-generation antipsychotic, is used as a model drug, which according to the FDA, is one of the several antipsychotic drugs marketed as a LAI formulation (please see Abilify Maintena®), which, however, presents a long initial lag-time (this is why oral ARI is simultaneously given for 14 consecutive days after the initial LAI injection) and is administrated rather frequently (i.e., once per month) [29–31]. Hence, within the set of the present study, after the initial evaluation of the biodegradation and the cytotoxicity profile of the neat PLA/PBAd block copolymers, ARI-loaded PLA/PBAd MPs were prepared via the emulsification/solvent evaporation method and thoroughly evaluated in terms of physicochemical and pharmacotechnical properties.

#### **2. Materials and Methods**

#### *2.1. Materials*

ARI (7-(4-[4-(2,3-Dichlorophenyl) piperazin-1-yl]butoxy)-3,4-dihydroquinolin-2(1*H*) one) form III crystals were kindly donated by Pharmathen S.A. (Athens, Greece). Adipic acid (ACS reagent, ≥99.0%), 1,4-butanediol (99%), tetrabutyl titanate (TBT) (97%) and the Tin(II) 2-ethylhexanoate (TEH) (96%) catalysts were obtained from the Sigma-Aldrich (Saint Louis, MO, USA). L-Lactide (98%) and (*S*,*S*)-3,6-dimethyl-1,4-dioxane-2,5-dione were purchased from Alfa Aesar Chemicals (Kandel, Germany). *Rhizopus delemar* and *Pseudomonas cepacia* lipases were purchased from Fluka BioChemika, Steinheim, Germany. All other solvents and reagents used were of analytical or pharmaceutical grade and were used as received.

#### *2.2. Synthesis of PBAd and PLA/PBAd Block Copolymers*

PBAd and PLA/PBAd copolymers were prepared by the method we previously published [28]. Briefly, PBAd was prepared via a two-stage esterification and polycondensation. During the first stage (esterification), accurately weighed amounts of adipic acid and 1,4-butanediol, in a 1/1.1 molar ratio, were placed in a round-bottom flask and the polymerization mixture was degassed and purged with nitrogen several times, before heating to 180 ◦C under constant stirring and then gradually heating up to 220 ◦C over a period of three hours. After removal of the water formed, the nitrogen flow was stopped and 400 ppm of TBT (0.05 g mL−<sup>1</sup> in toluene) was added to the mixture under high vacuum (5.0 Pa), in order to avoid excessive foaming. The temperature was then increased to 240 ◦C and the polycondensation reaction was carried out for another two hours.

After the preparation of the neat PBAd the PLA/PBAd copolymers were prepared via ring opening polymerization of L-lactide. Briefly, proper amounts of L-lactide and PBAd (corresponding to a final copolymer weight ratio of 95/5, 90/10, 75/25 and 50/50 *w*/*w* PLA to PBAd) were placed in round bottom flasks along with the THE (used as a catalyst at 400 ppm based on the L-lactide concentration). After nitrogen purging, the mixture was heated up to 200 ◦C and the reaction was initiated and carried out under constant mechanical stirring for one hour. The MW of the prepared copolymers was increased by heating up to 220 ◦C, under high vacuum (5.0 Pa) for 15 min. Then, the flasks were cooled to room temperature and the copolymers were purified by dissolving them in chloroform and precipitating in cold methanol twice, prior to using them. The precipitates were filtered and dried in a vacuum oven at 50 ◦C for 24 h. All samples were collected, placed in hermetically sealed vials, after purging with N2, and stored at 5 ◦C before further use.

#### *2.3. Characterization of PLA/PBAd Block Copolymers*

Following their preparation, the newly synthesized block copolymers were characterized in terms of biodegradation and cytotoxicity profiles, characteristics that are extremely significant when preparing drug LAI formulations. Results, in terms of structural characterization, MW, physical state, thermal properties and molecular mobility, are given in a previous study of ours (Table 1) [28].

**Table 1.** Values of interest for the synthesized copolymers: molecular weight values, weight averaged (Mw) and number averaged (Mn), and polydispersity index (PDI), estimated by SEC, and crystallization, melting and glass transition temperatures (Tc, Tm and Tg accordingly), estimated by DSC.


#### 2.3.1. Enzymic Hydrolysis

PLA/PBAd enzymatic hydrolysis was performed based on a previously employed method [32]. Briefly, the neat PBAd and PLA and the PBAd/PLA copolymers were prepared in the form of films, using an OttoWeber Type PW 30 hydraulic press (Paul-Otto Weber GmbH, Remshalden, Germany). The films were placed in petri dishes and 5 mL of phosphate buffer solution (0.2 M, pH 7.4) was added, containing 0.09 mg/mL of *Rhizopus delemar* lipase and 0.01 mg/mL of *Pseudomonas cepacia* lipase. The petri dishes were kept at 37.0 ± 1.0 ◦C in an oven for twenty days, while the media were replaced every 24 h. After predetermined time intervals, the films were removed from the lipase solution, washed thoroughly with distilled water and dried at 40 ◦C in vacuo, until constant weight. Every measurement was repeated three times. The degree of enzymatic hydrolysis was estimated from the weight loss, as compared to the initial weight of the samples.

#### 2.3.2. Size-Exclusion Chromatography

The molecular weights of all samples after enzymatic hydrolysis were estimated by size-exclusion chromatography (SEC). The analysis was performed by means of SEC equipment consisting of a Waters 600 high pressure liquid chromatographic pump (Waters, Milford, MA, USA), Waters Ultrastyragel columns (HR-1, HR-2, HR-4 and HR-5) and a Shimadzu RID-10A refractive index detector (Shimadzu Corporation, Kyoto, Japan). Col-

umn calibration was performed using polystyrene standards (1–300 kg/mol in molecular weight). The concentration of the prepared solutions was 20 mg/1000 mL, the injection volume was 150 mL and the flow rate was 1 mL/min, operating at 60 ◦C.

#### 2.3.3. Cytotoxicity Studies

Human adipose-derived mesenchymal stem cells culture (hAMSCs): For the evaluation of neat polymer and copolymers cytotoxicity, hAMSCs were provided from Biohellenika S.A. (Thessaloniki, Greece) after adipose tissue isolation from healthy volunteer donors. Experimentally, after liposuction adipose tissue washed twice in PBS (phosphate buffered saline) (1X, pH 7.4) (BIOWEST, Nuaillé, France). Overnight digestion was performed with 5 mg collagenase type I (Sigma-Aldrich, Saint Louis, MO, USA) per 10 g of adipose tissue after overnight incubation. The mixture was filtered using a 70 μm cell strainer (CORNING, Glendale, AZ, USA) and centrifuged at 850× *g* for 10 min. The pellet was resuspended in Dulbecco's modified Eagle's medium (DMEM)(BIOWEST, Nuaillé, France) supplemented with 10% fetal bovine serum (FBS)(BIOWEST, Nuaillé, France) and 2% penicillin/streptomycin and plated in culture flasks for 72 h until cells' adherence to the plastic surface (37 ◦C incubation with 5% CO2). The cell culture medium was replaced every 2–3 days until 80–90% confluence was reached. Cells were used in the experiments between passage 4 and 5. Every cells' detachment was performed with 0.05% trypsin–EDTA (BIOWEST, Nuaillé, France).

Sterilization of the materials and cell seeding: All the materials were sterilized in gradually reduced ethanol concentrations (100%, 70% and 50% in ddH2O) and, after washing twice with ddH2O, were left to air dry for 5 h under sterile conditions. Fibrin glue was prepared after the blood sampling of a healthy volunteer donor. A total of 10 μL of fibrin glue per film were placed in the bottom of a 24-well plate and the materials were seeded using a sterile pincher from above by applying minimal manual pressure and were left to air dry overnight under sterile conditions.

hAMSCs were detached using trypsin–EDTA 1x in PBS. A total of 3.5 × <sup>10</sup><sup>5</sup> cells were resuspended in the DMEM full medium and were subsequently placed above the films of each condition. A total of 3.5 × 105 cells were also plated in a plastic surface without any material and used as a control group. Upon air drying for 4 h in the incubator 1 mL of the DMEM full medium was added per well for the culture initiation. After 48 h, the cytotoxic effect of the materials was determined with an MTT (3-[4,5-dimethylthiazol-2-yl]- 2,5 diphenyl tetrazolium bromide) assay.

In vitro cytotoxicity assays: The MTT cell proliferation assay, which employs the reduction of tetrazolium salts by metabolically active cells for examining cellular viability, was used for in vitro cytotoxicity assessment (Trevigen, Gaithersburg, MD, USA 4890- 025-K). After 48 h of coincubation with the formulations, the medium was removed and cells were washed once with PBS before adding fresh medium including the 1/10 MTT reagent (Sigma-Aldrich, Saint Louis, MI, USA). Upon the removal of the MTT, 1 mL/well of DMSO was introduced for one additional hour of incubation. The optical density of MTT formazan deposits was quantified by a spectrophotometer at a 570 nm and 630 nm wavelength (PerkinElmer, Boston, MA, USA). All experiments were conducted in triplicate.

#### *2.4. Preparation of ARI MPs*

ARI MPs were prepared using PLA and PBAd polymers and their copolymers using an emulsification/solvent evaporation method. Briefly, 250 mg of polymer (pure PLA, pure PBAd and copolymers) were initially dissolved in 5 mL of dichloromethane and stirred with a magnetic stirrer. Then 50 mg of ARI were added to the solution and sonicated for 1 min until complete dispersion. The aqueous phase (50 mL of deionized H2O and 50 mL of 1% *w/v* PVA solution) was then added to the dispersion phase, homogenized and left under stirring (1200 rpm), at room temperature, until the solvent was completely evaporated. When the microspheres were formed, they were separated from the rest of the solution by centrifugation at 4500 rpm for 10 min. Possible solvent or emulsifier residue was removed by three consecutive washes with deionized water. The microspheres were then freeze-dried in order to remove any water residue. For the preparation of non-drug loaded MPs, the same procedure as the one described above was followed without the addition of the API. All final samples were subsequently stored at 4 ◦C using hermetically sealed amber glass vails before further use.

#### *2.5. Characterization of MPs*

#### 2.5.1. Differential Scanning Calorimetry (DSC)

DSC studies were conducted, using a Perkin–Elmer, Pyris Diamond DSC. In brief, accurately weighed samples (5.0 ± 0.1 mg) of the raw materials (i.e., neat PLA, neat PBAd and neat ARI) and the PLA-ARI, PBAd-ARI and PLA/PBAd-ARI MPs were hermetically sealed in aluminum pans and placed in the DSC sample holder. Then the samples were heated from 25 to 180 ◦C with a heating rate of 20 ◦C/min and the various thermal events were recorded using the Pyris Diamond software. The melting points (Tmelt) were determined as the peak temperature and the glass-transition temperature (Tg) was determined as the inflection point temperature, while the enthalpy of fusion (ΔHf) was determined as the integrated area of the heat flow curve in all cases. Nitrogen flow (50 mL/min) was applied in order to provide a constant thermal blanket within the DSC cell. The instrument was calibrated for temperature using high purity benzophenone, indium and tin, while the enthalpic response was calibrated using indium. All measurements were conducted in triplicate. The standard deviations of temperatures and enthalpies determined, in this work, were not higher than 1.0 ◦C and 3.0 J/g, respectively.

#### 2.5.2. Wide Angle Powder X-ray Diffractometry (pXRD)

pXRD patterns of the raw materials (i.e., neat PLA, neat PBAd and neat ARI) and the PLA-ARI, PBAd-ARI and PLA/PBAd-ARI MPs were recorded using an XRD-diffractometer (Rigaku-Miniflex II, Chalgrove, Oxford, UK) with a CuKα radiation for crystalline phase identification (λ = 0.15405 nm for CuKα). All samples were scanned from 5 to 50◦ with a scanning rate of 1 ◦/min.

#### 2.5.3. Scanning Electron Microscopy (SEM)

The morphology of the neat PLA, PLBAd and PLA/PBAd copolymers in the form of film before and after the enzymatic hydrolysis study and the ARI-loaded PLA, PBAD and PBAd MPs before and after the completion of the dissolution study was examined in a SEM system (JEOL JMS-840, JEOL USA Inc., Peabody, MA, USAmanufacturer, city, country). All samples (either in the form of thin films or MPs) were covered with carbon in order to provide good conductivity of the electron beam. All SEM images were collected with the following operating conditions: (1) accelerating voltage 20 kV, (2) probe current 45 nA and (3) counting time 60 s.

#### 2.5.4. Fourier-Transformed Infrared Spectroscopy (FTIR)

The chemical structure and the formation of molecular interactions in PLA/PBAd copolymers and the PLA-ARI, PBAd-ARI and PLA/PBAd-ARI MPs was elucidated by FTIR spectroscopy. FTIR spectra of the samples were received with an FTIR spectrophotometer (model FTIR-2000, Perkin Elmer, Dresden, Germany) using KBr discs (thickness of 500 μm). The spectra were collected in the range from 4000 to 400 cm−<sup>1</sup> at a resolution of 2 cm−<sup>1</sup> (total of 64 coadded scans) and were baseline corrected and converted into the absorbance mode.

#### 2.5.5. Yield, Encapsulation Efficiency and Drug Loading

MPs' yield, drug loading and encapsulation efficiency (EE) were determined by applying the following equations:

Yield (%) = [weight of MPs]/[initial weight of polymers and ARI] × 100 (1)

Drug loading (%) = [weight of ARI in MPs]/[total weight of MPs] × 100 (2)

## EE (%) = [weight of ARI in MPs]/[initial weight of ARI] × 100 (3)

Microspheres equivalent to 10 mg of aripiprazole were dissolved in the minimum quantity of dichloromethane and then diluted with the mobile phase: H2O pH 3.5: acetonitrile 60:40 (*v*/*v*). The resulting solution was filtered through 0.45 μm filter paper and the filtrate was assayed for ARI using a Shimadzu Prominence HPLC system (Shimadzu Corporation, Kyoto, Japan), consisting of a degasser (Model DGU-20A5), a pump (Model LC-20AD), an automatic sampler (Model SIL-20AC), an ultraviolet–visible variable detector (Model SPD-20A) (λmax = 254 nm) and a thermostatic oven (Model CTO-20AC). A reverse phase C18 column (250 mm × 4.6 mm I.D., 5 μm particle size) was used for chromatographic analysis. The flow rate was adjusted to 1 mL/min and the infusion volume was 20 μL. The chromatograms obtained were processed with the LC Solution software (v1.2, Shimadzu Corporation, Kyoto, Japan). All measurements were conducted in triplicate.

#### 2.5.6. In Vitro Dissolution Test

The MPs (having 10 mg of ARI) were suspended in 2 mL of PBS and inserted in a dialysis tubing cellulose membrane bag (D9402-100FT; Sigma-Aldrich, Steinheim, Germany) with a molecular weight cut-off of 12,000–14,000, which was then sealed and placed into the dissolution basket (Distek Inc., North Brunswick Township, NJ, USA, model 2100C Dissolution Test System), equipped with an automatic sampler (Evolution 4300 Dissolution Sampler). The dissolution studies were performed under sink-conditions, using 400 mL of a phosphate-buffered saline (PBS) solution (pH 7.4), at 50 rpm/37 ± 0.5 ◦C. The solubility of ARI in PBS was 0.3 mg/mL (measured at 37 ◦C with the shaking flask method). Samples (2 mL) were withdrawn at predetermined time intervals, filtered and the concentration of ARI was determined using the above described validated HPLC method. Additionally, after the completion of the dissolution experiments the remaining MPs were withdrawn from the dialysis tubes, dried and analyzed for ARI content, using the HPLC method described in Section 2.5.5. All experiments were conducted in triplicate.

In order to evaluate the drug release mechanism, in vitro dissolution results were fitted to the following release kinetics models [33]:

$$\text{Zero order model: } \mathbf{D}\_{\mathbf{t}} = \mathbf{D}\_0 + \mathbf{k}\_0 \mathbf{t} \tag{4}$$

First order model: logDt = logD0 + k1t/2.303 (5)

$$\text{Higgsuchi square root model: } \mathbf{D}\_t = \mathbf{D}\_0 + \mathbf{k}\_{\rm Pl} \mathbf{t}^{1/2} \tag{6}$$

$$\text{Hixon-Crowell model: } \text{D}\_{\text{t}}^{1/3} = \text{D}\_{\text{0}}^{1/3} - \text{k}\_{\text{HCl}} \text{t} \tag{7}$$

$$\text{Korsmeyer-Pepps model: } \mathrm{D\_l/D\_{\infty} = D\_0 + k\_{\mathrm{P}}t^{\mathrm{R}}} \tag{8}$$

where, Dt is the amount of drug released at time t, D0 is the initial amount of drug released, Dt/D<sup>∞</sup> is the fraction of drug released at time t, k0 is the zero-order release constant, k1 is the first-order release constant, kH is the Higuchi release constant, kHC is the Hixson–Crowell release rate constant, kp is the Peppas release constant and n is the release exponent respectively.

#### *2.6. Statistical Analysis*

Statistical significance in the differences of the means was evaluated by using Student's *t*-test or Dunnett's test for the single or multiple comparisons of experimental groups, respectively. A difference with a *p*-value (*p*\*) < 0.05 was considered statistically significant.

#### **3. Results and Discussion**

#### *3.1. Evaluation of neat PLA/PBAd Block Copolymers*

As stated in the Introduction, the present study attempts to build upon the previously published promising results regarding the thermal and mechanical properties of the newly synthesized PLA/PBAd block copolymers and to evaluate their use as matrix/carriers for the preparation of ARI loaded LAI MPs. In this context, cytotoxicity and enzymatic hydrolysis of the neat copolymers are initially evaluated, since these two features are extremely important before proceeding with the preparation and evaluation of the LAI MPs.

#### 3.1.1. Cytotoxicity Results

In general, polymers or copolymers that will be used to prepare such drug delivery systems should possess low cytotoxicity. Polyesters, based on PLA, poly (glycolic acid) (PGA) and polycaprolactone (PCL) and their copolymers, have been widely used as such biomaterials with a low cytotoxicity profile [34–40]. However, the cytotoxicity arising from the biodegradation of the newly prepared PLA/PBAd is unknown, and, hence, systematic evaluation is needed in order to verify their safety. The MW of the newly synthesized copolymers (measured by size-exclusion chromatography) varied from 98k–95k, while the polydispersity index (PDI) was below 2.0 in all cases (1.60–1.85) [28].

Figure 2 illustrates the cytotoxicity effect of the prepared block copolymers on hAM-SCs, where the *y*-axis shows the reduction of yellow 3-(4,5-dimethythiazol2-yl)-2,5-diphenyl tetrazolium bromide (MTT) by mitochondrial succinate dehydrogenase.

**Figure 2.** Cytotoxic effect on hAMSCs after incubation with neat PBAd, neat PLA and the newly prepared PLA/PBAd block copolymers.

During this study, the MTT, which enters hAMSCs, passes through the mitochondria where it is reduced to formazan. Subsequently, the cells are solubilized and the formazan content is measured spectrophotometrically. Since MTT reduction can only happen in metabolically active cells, the degree of activity is a measure of the cells' viability. Generally, in order for a material to be classified as toxic a reduction in the measured absorbance should be more than 50% as compared to the control sample. Hence, based on the obtained results, all studied materials can be considered as non-toxic since the max reduction in the measured absorbance was 40%. Specifically, in the case of PLA/PBAd block copolymers the obtained results showed a similar (for PLA/PBAd 50/50 and 75/23 *w*/*w*) or a significantly better (for PLA/PBAd 90/10 and 95/5 *w*/*w*) cytotoxicity profile as compared to the neat PLA, which is considered to be a non-cytotoxic biopolymer. Furthermore, results showed that the metabolic activity of the cancer cell line was dependent on the PLA to PBAd ratio within the copolymer, with samples higher in PLA showing a remarkably lower toxicity. Therefore, based on the MTT assay results it can be said that all prepared copolymers are non-toxic and, hence, are suitable candidates for the preparation drug LAI formulation.

#### 3.1.2. Enzymatic Hydrolysis

In addition to non-toxicity, evaluation of the enzymatic hydrolysis profile of a polymer (or copolymer) is needed in order to clarify whether this material can be used as a LAI matrix/carrier, including, of course, MP based formulations. Generally, enzymatic hydrolysis (i.e., the path to polymer's degradation) is controlled by various factors related to the structure, the solid and thermal properties, etc. Among them, the mobility of polymer (or copolymer) segments, the crystalline morphology (including spherulite size), the ratio and the balances of hydrophilic/hydrophobic segments, the molecular weight, the Tmelt and Tg are all factors that significantly affect the hydrolysis rate and extent [41–45].

In the present study, the enzymatic hydrolysis of the raw materials (i.e., the neat PBAd and PLA) and the newly synthesized PLA/PBAd block copolymers were evaluated in solutions containing a mixture of *R. delemar* and *Pseudomonas cepacia* lipases, at 37 ◦C and pH 7.4. Figure 3 shows the calculated hydrolysis in terms of weight loss vs. time profiles.

**Figure 3.** The enzymatic hydrolysis profile measured as % weight loss vs. time plots for the neat PBAd, the neat PLA and the various PLA/PBAd block copolymers.

In the case of the neat PLA, the results showed an extremely slow enzymatic hydrolysis rate reaching 3% within the first six days of testing. This slow degradation for PLA may be attributed to the polymer's high hydrophobic nature and to its high degree of crystallinity and its rather high Tmelt and Tg (i.e., 150 ◦C and 55 ◦C, respectively [28]). In contrast to PLA, PBAd showed a substantially higher degree of enzymatic hydrolysis. This higher hydrolytic rate is in agreement with previous results [23] and can be attributed to the polymer's low Tg (approximately −55 ◦C) and Tmelt onset (40 ◦C), which allow the polymer's segments to move around more freely, thus, enabling water to penetrate and hydrolyze the PBAd ester bonds more easily. In the case of PLA/PBAd samples, results showed an increase in the copolymer's hydrolysis, which was proportional to the PBAd content. Specifically, as the content of PBAd increased, the copolymer's degradation (measured in terms of weight loss) also increased. Hence, based on the obtained results, it should be noted that the prepared block copolymers also show highly tunable enzymatic hydrolysis characteristics. This is extremely important since, depending on the pharmacological properties of the API, the specific disease features and patients' individual characteristics, the proposed new biomaterials may be received as a universal solution for tailored drug or patient treatment.

However, despite the above presented significant findings, regarding the hydrolysis rate and extent of the prepared copolymers, in depth analysis of the degradation process is also needed in order to gain a true insight into the enzymatic biodegradation phenomena. In this context, the morphology of the prepared samples, before and after enzymatic hydrolysis, was evaluated via SEM. The results, presented in Figure 4, showed that the neat PLA remained almost unaffected after six days of testing, while neat PBAd showed an extensive mass degradation, which was dispersed uniformly along the whole surface of the sample. Similarly, the PLA/PBAd copolymers showed increased mass loss, as the

content of PBAd increased, while, according to all collected images, it was obvious that the degradation mechanism of the copolymers, during their enzymatic hydrolysis at 37 ◦C, was related initially to surface erosion. This was also confirmed by SEC measurements after the first six days of study, which showed that molecular weight values remained practically unchanged compared to the initial samples, while weight loss was taking place (Table 2). However, even in this case, hydrolysis is a dynamic procedure. It has been found that the hydrolytic chain cleavage proceeds preferentially in the amorphous regions of polyesters, leading initially to the increase in polymer crystallinity [46]. Due to the interconnections of amorphous fractions, hydrolysis becomes also a bulk erosion process after a period of time.

**Figure 4.** SEM micrographs of the neat PBAd and PLA and the various PLA/PBAd block copolymers during enzymatic hydrolysis at zero time (initial) and after six days.


**Table 2.** SEC estimated molecular weight values, Mw and Mn, and % weight loss, after six days of enzymatic hydrolysis.

#### *3.2. Evaluation of ARI-loaded MPs*

Based on the previously obtained results, the newly synthesized PLA/PBAd block copolymers show a good cytotoxicity profile and highly tunable enzymatic hydrolysis characteristics, features that makes them good candidates as LAI matrix/carriers. Hence, in the following section the preparation of such drug-loaded formulations (in the form of MPs) will be thoroughly evaluated.

#### 3.2.1. MPs Morphology Evaluation Via SEM

As stated previously, in the present study the newly prepared PLA/PBAd block copolymers (at several PLA to PBAd ratios) were tested as suitable biopolymers for the preparation of ARI LAI MPs. In this set framework, the effect of the PLA and PBAd content on the size and the morphological characteristics of the drug-loaded MPs was initially investigated via SEM. Results in Figure 5 showed the formation of spherical MPs with a smooth exterior surface, while in all cases no particle agglomeration was observed.

Specifically, regarding the MPs prepared with the initial polymeric raw materials (i.e., PLA and PBAd), results showed the formation of significantly larger particles in the case of PLA with more spherical shape and uniform size distribution, while some defects were also observed on the surface of the said MPs. These differences can be attributed to the more hydrophobic nature of PLA (as compared to PBAd), which leads to a better homogenization and consequently more controlled solvent removal processes. Additionally, a significant role also plays the notable differences in the thermal properties of the two tested biopolymers, with PBAd's lower Tmelt and Tg values enabling the 'softening' of the just formed MPs during the solvent removal phase, leading in this way to the formation of smaller drug-loaded spherical MPs (as compared to PLA). In the case of PLA/PBAd, results showed that as the PBAd content increased within the block copolymer the particle size of the obtained MPs decreased. Specifically, the average particle size (measured as d50) of the prepared ARI-loaded MPs, measured from at least ten SEM images, was estimated as 58.2 ± 15 μm, 43.3 ± 10 μm, 30.15 ± 10 μm and 18.8 ± 5 μm, for the MPs prepared with PLA/PBAd 95/5, 90/10, 75/25 and 50/50, respectively. Considering the previously published results on the thermal properties of the prepared neat block copolymers [28], where it was found that the melting properties of the two monocomponents (i.e., PLA and PBAd) are retained in the newly prepared biomaterial, the obtained results indicate that the addition of PBAd in the block copolymer chain (and its more hydrophilic nature and lower melting temperature) is responsible for the reduction of the resultant ARI-loaded MP's size.

#### 3.2.2. MPs Yield, Drug Loading and EE

Table 3 summarizes the yield, drug loading and EE values for the prepared ARIloaded MPs.

**Figure 5.** SEM images for the ARI-loaded MPs prepared using PLA, PBAd and the newly synthesized PLA/PBAd block copolymers as the matrix/carriers.


**Table 3.** Yield, drug loading and EE of the prepared ARI MPs.

Based on the obtained results, the yield in all MPs ranged from 60.35 to 98.6% with the PLA-PBAd copolymers showing much more improved yields, as compared to the neat PLA and PBAd MPs. Additionally, a closer look at the obtained results revealed that the higher yield values were recorded in the case of MPs containing higher amounts of PBAd (i.e., PLA/PBAd 75/25 and 50/50), indicating that the presence of PBAd in the backchain of the prepared copolymer results in a more efficient (at least in terms of yield productivity) MP's preparation process. However, in contrary to the previous findings, results regarding MP's drug loading and EE showed the opposite effect. Specifically, as the content of PBAd increased in the PLA/PBAd copolymer, the resultant ARI-loaded MPs showed lower drug loadings and EE values. This indicates that, contrary to MP's productivity, the presence of PBAd results in droplets that were harder to solidify, and hence there was much more time for the API molecules to diffuse away from the droplet into the aqueous phase medium, resulting in the preparation of MPs with a lower drug content.

#### 3.2.3. MPs' Thermal Properties and Physical State Evaluation Via DSC

The thermal properties and the physical state of the drug-loaded MPs as compared to the neat raw materials were evaluated with the aid of DSC (Figure 6a). In the case of the neat ARI, results showed an initial endothermic peak at 140.6 ◦C, corresponding to the melting of the ARI form III crystals, followed by a small recrystallization exotherm (at 144 ◦C) and a second endothermic peak at 151. 9 ◦C corresponding to the ARI form I crystals melting. These results indicate that the initially used ARI was in the form of polymorph III crystals, while during its melting a phase transition from polymorph III to polymorph I was recorded. This behavior is in agreement with previous studies evaluating the phase transition phenomena occurring during ARI's DSC heating [47]. Regarding the neat initial polymers, the results in the case of PBAd showed a broad DSC endotherm with a peak at 63.3 ◦C, corresponding to its melting, while PLA showed a Tg transition point at 69.9 ◦C and a melting endotherm at 153.4 ◦C, both indicative of its semicrystalline nature.

Looking at the DSC thermograms of the drug-loaded PBAd- and PLA-MPs, similar thermal events were detected, as is in the case of pure (neat) polymers. Specifically, in the case of ARI-PBAd MPs, a broad endothermic peak was recorded at 58.9 ◦C corresponding to the melting of the crystalline part of the polymer, while for ARI-PLA MPs a Tg (with an endothermic overshoot due to the molecules' relaxation) was recorded at 67.5 ◦C followed by an endothermic melting peak at 143.9 ◦C, corresponding to the melting of the PLA. It is important to note that in all thermograms a small drop in the obtained thermal events was recorded (as compared to the neat polymeric raw materials), which is attributed to the presence of the API and the remaining solvents (used for the preparation of the MPs) that act as plasticizers to the whole system.

Additionally, it should be pointed out that in the case of ARI-loaded PBAd MPs, no thermal events were recorded in respect to the API, indicating that probably the drug was amorphously dispersed within the polymeric matrix, although in situ solubilization of the ARI crystals during the DCS heating scan cannot be excluded. On the contrary, results from the DSC thermograms of the drug-loaded PLA MPs, showed a small melting endotherm at 137 ◦C, which is probably attributed to some of the remaining ARI form III crystals. Similar results were also obtained for the drug-loaded MPs prepared with the newly synthesized PLA/PBAd block copolymers, where the DSC endotherm corresponding to the API melting was decreasing as the PBAd content increased, while at the higher PBAd content used (i.e., PLA/PBAd 75/25 and 50/50) no such API melting peaks were recorded. Hence, based on the DSC results it seems that the presence of PBAd in the PLA/PBAd block copolymers favors the amorphization of the API leading to its complete amorphization in ratios higher that 75/25 *w*/*w* PLA to PBAd.

**Figure 6.** DSC thermograms (**a**) and XRD diffractograms (**b**) of the neat raw materials (i.e., ARI, PLA and PBAd) and the prepared drug-loaded MPs.

#### 3.2.4. Physical State Verification Via pXRD

The physical state of the API after the preparation of the drug-loaded MPs was also evaluated via pXRD in order to verify the results suggested by DSC. Figure 6b shows the pXRD diffractograms of the raw materials, the recrystallized neat PLA (after solubilizing in dichloromethane, i.e., the solvent used for the preparation of the MPs) and the respective MPs. In the case of the neat ARI, results showed several sharp pXRD diffractogram peaks at 2θ of 10.9, 16.6, 19.3, 20.3 and 22.0◦, which were all characteristic of the ARI form III crystals [48]. In the case of neat PBAd, two characteristic pXRD peaks were recorded at 2θ of 21.1 and 24.2◦, which were both located over a broad amorphous halo, indicating that the neat copolymer was semicrystalline in nature. Regarding the neat PLA, two different pXRD patterns were recorded before and after its recrystallization. Specifically, the polymer as received showed a characteristic amorphous halo, indictive of its highly amorphous nature, while upon its recrystallization new crystals were recorded at 2θ positions of 14.8, 16.9, 19.1 and 22.5◦, respectively, all of which were also seen in the case of PLA's melt recrystallization [28]. In regard to the drug-loaded MPs using only PBAd, the recorded pXRD diffractograms showed only the characteristic peaks of the neat copolymer, indicating that the API was amorphously dispersed within the MPs' matrix/carrier. In contrast, in the

case of PLA drug-loaded MPs, in addition to the polymer's characteristic pXRD pattern, two diffractogram peaks corresponding to the ARI's form III crystals (i.e., 2θ of 20.4 and 22.0◦, respectively) were also recorded, indicating that the API was recrystallized during the formation of the said MPs. Similarly, in the case of MPs prepared with the newly synthesized PLA/PBAd block copolymers having high PBAd content (i.e., 75/25 and 50/50 PLA to PBAd ratio), no ARI characteristic pXRD peaks were recorded, indicating that the API was amorphously dispersed within the said matrix-carriers. On the contrary, in the rest of the samples, i.e., those using PLA/PBAd copolymers with high PLA content, two characteristic ARI form III peaks (although of low intensity) were recorded, indicating that a small portion of the API was recrystallized in these cases. Hence, based on the obtained results, the pXRD analysis verifies the previously presented DSC findings, since the increase in PLA's content within the newly synthesized PLA/PBAd block copolymers, leads, indeed, to ARI's recrystallization.

#### 3.2.5. Evaluation of Molecular Interactions

In a further step, FTIR spectroscopy was used in order to identify the formation of molecular interactions during the preparation of the drug-loaded MPs. Initially, before proceeding with the FTIR analysis of the MPs, the spectra of the newly prepared PLA/PBAd block copolymers were evaluated (Figure 7a) in an attempt to identify the molecular interactions evolving between the two polymeric components (i.e., PLA and PBAd) during the copolymerization process. Specifically, in the case of PLA the asymmetric and symmetric vibrations of the methylene groups were recorded at 2995 cm−<sup>1</sup> and 2945 cm<sup>−</sup>1, respectively, while the vibrations of the carbonyl C=O and the C-O-C ester groups were recorded at 1757–1710 cm−<sup>1</sup> and 1188 cm<sup>−</sup>1. In the case of PBAd, the characteristic absorption peaks of the ester -COO- and the C-O-C appeared at 1735 cm−<sup>1</sup> and 1100–1300 cm<sup>−</sup>1, respectively, while the peaks located at 1450–1465 cm−<sup>1</sup> were attributed to the C-H bending vibrations of the methylene and methyl groups. In all spectra the low intensity peaks recorded at 3300–3550 cm−<sup>1</sup> can be attributed to the presence of -OH end groups. Regarding the newly synthesized copolymers, results showed increased similarities among the recorded spectra. Specifically, in all cases a strong absorption peak at 1730 cm−<sup>1</sup> was recorded, due to the formation of a new ester bond between the PLA and the PBAd (responsible for the formation of the new block copolymer). Additionally, there were also several peaks in the range of 750–1100 cm−<sup>1</sup> and 1100–1400 cm−1, corresponding to the C-C and C-O vibrations, respectively. Finally, the presence of the methylene groups within the newly synthesized copolymers was also confirmed by the specific FTIR absorption peaks recorded in the region of 2700–3000 cm<sup>−</sup>1.

Moving along with the evaluation of molecular interactions, evolving within the prepared drug-loaded MPs, Figure 7b shows the recorded FTIR spectra of all systems along with the spectrum of the neat API. In regard to ARI, results showed the presence of several characteristic FTIR absorption peaks at 3195 cm−<sup>1</sup> (corresponding to the NH vibrations), 2949 and 2840 cm−<sup>1</sup> (attributed to the CH), 1679 cm<sup>−</sup>1, due to C=O, and 1628 cm<sup>−</sup>1, due to C=C, vibrations), while the peaks at 1160 and 1123 cm−<sup>1</sup> are attributed to the stretching vibration of the single C-O and C-C bonds, respectively. Before proceeding with the analysis of the MPs' FTIR spectra, it is important to note that in general, polyesters, such as those evaluated in the present study, consist mainly of ester bonds and terminal carboxylic and hydroxyl groups, which can interact, via hydrogen bonding (HB), with the ester groups or the amino groups of ARI and its two chlorine atoms located in the dichlorophenyl part of the molecule. Hence, in order to determine if such interactions exist in the prepared systems, we will focus our analysis on the characteristic peaks recorded in the region of the hydroxyl and carbonyl groups of the FTIR spectrum. Looking at the obtained MPs' spectra, the hydroxyls of the polyesters in the MPs were recorded at 3480 cm−<sup>1</sup> and no obvious shifts were apparent amongst the examined systems. Additionally, in the region of carbonyls' absorption bands, all polyesters show a similar wide peak at 1730 cm<sup>−</sup>1, while next to it (at 1679 cm−1) the carbonyl vibration of the pure drug is recorded indicating

that the API was successfully encapsulated in the prepared MPs. Finally, since, there are no differences (or shifts/displacements) in the FTIR absorption peaks, it seems that no molecular interactions are taking place between the API and the copolymer.

**Figure 7.** FTIR spectra of: (**a**) the neat PLA, PBAd and the PBA/PBAd copolymers and (**b**) the API and the prepared API-loaded MPs.

3.2.6. In Vitro Dissolution Profile

In the final step of the present work, the effect of the newly synthesized block copolymers on the in vitro dissolution characteristics of ARI were evaluated. Figure 8a depicts

the dissolution profiles of the prepared drug-loaded MPs. The maximum ARI released from the PLA-PBAd MPs ranged from 37.70% (with PLA-PBAd 95/5) to 60.38% (with PLA-PBAd 50/50) indicating a wide distribution, in terms of drug release extent. This can be partially attributed to the amorphization of the drug within the MPs, induced by the presence of the PBAd, and its fine dispersion within the polymeric matrix (confirmed previously by the XRD and DSC results). Additionally, drug assay analysis of the remaining MPs after the completion of the dissolution trials (Table 4) revealed the presence of un-dissolved API still "trapped" within the polymeric structure. This may explain the incomplete delivery of the API observed in all MPs formulations. Lastly, regarding the ARI that was neither recovered from the microparticles nor released, we can postulate that this was probably lost during the withdrawal of the MPs from the dialysis tubes and the drying process conducted before the ARI content analysis. Additionally, we may assume that a small portion of the API may be "lost" due to the drug's degradation during dissolution, although in order to support/verify this hypothesis ARI solution stability at 37 ◦C for 30 days has to be performed.

**Figure 8.** In vitro drug % release vs. time of the encapsulated ARI in PLA/PBAd MPs, during the 30 (**a**) and the 0.5 (**b**) days of the experiment.


**Table 4.** % Aripiprazole drug remained in the MPs at the end of the in vitro release study.

The MPs prepared, using PLA, showed the lowest drug release rate, on contrary to the MPs prepared with PBAd where the maximum release rate was achieved. The rest of the formulations using the newly synthesized PLA-PBAd block copolymers showed increasing API release rate (and extend) as the PBAd content increased. Therefore, it seems that the addition of PBAd to the polymeric matrix significantly improves the hydrolysis rate of PLA and, consequently, the dissolution rate of the encapsulated API. Based on these findings it must be said that the use of the newly synthesized PLA/PBAd block copolymers as matrix/carriers for the preparation of ARI-loaded MPs, results in highly tunable extended release profiles for the API, which may be controlled for up to 30 days. Therefore, under in vivo conditions this could possibly lead to new formulations able to maintain continuous therapeutic levels for an extended time period (>30 days), with no lagtime, and hence, emerge as an alternative long-acting treatment option for the management of chronic diseases.

Looking again back to the obtained dissolution results, it is obvious that ARI's dissolution from the prepared MPs followed a biphasic release profile in all cases. Specifically, after an initial burst phase (Figure 8b) attributed to the active substance present on the surface of the MPs, a fast release phase was observed for up to approximately five (5) days, followed by a slower release phase for the remaining twenty-five (25) days. Keeping in mind that drug release from such MPs is mainly controlled by the interplay between API's diffusion from the polymeric matrices and polyester's erosion/degradation behavior, it can be assumed that in both phases (i.e., the fast and the slow) these two different mechanisms have a different impact. Therefore, in an attempt to identify the differences prevailing in each release phase, the obtained dissolution data were fitted in the various kinetic models described in Section 2.5.6. The goodness of fit (expressed by the correlation coefficient, R2) and the k-constants for each model are summarized in Table 5.

**Table 5.** Dissolution data model fitting results for the employed drug release kinetic models.


Looking at the obtained results, in the case of the initial release phase (i.e., up to five days) the higher R<sup>2</sup> values for all samples were obtained for the Korsmeyer–Peppas equation, indicating that the said model is more suitable to describe the obtained dis-

solution data. In general, the Korsmeyer–Peppas model is able to describe the several mechanisms that simultaneously control the dissolution behavior in such systems, by the use of the exponent *n*. Specifically, *n* values below 0.5 suggest that the drug diffuses through the matrix and is released with a quasi-Fickian diffusion mechanism, while values between 0.5 and 1 indicate an anomalous, non-Fickian, drug diffusion and values above 1 suggest a non-Fickian, Case II, release kinetics mechanism [33]. Based on the obtained Korsmeyer–Peppas fitting results, the exponent *n* in the initial fast release phase was below 0.5 in all cases (i.e., *n*(PLA) = 0.088, *n*(PBAd) = 0.337, *n*(PLA/PBAd 95/5) = 0.268, *n*(PLA/PBAd 90/10) = 0.256, *n*(PLA/PBAd 75/25) = 0.366 and *n*(PLA/PBAd 50/50) = 0.348) indicating that the drug released from all prepared MPs in the first five days was diffusion controlled. Interestingly, in the case of the slow-release phase (i.e., starting from the 6th day and lasting up to 30 days) the fitting results in Table 5 showed that the release of the API from all prepared MPs followed a zero-order release mechanism. This is also verified by the *n* exponent of the Korsmeyer–Peppas model fitting, where, in all cases, was approximately one (i.e., *n*(PLA) = 0.917, *n*(PBAd) = 1.057, n(PLA/PBAd 95/5) = 0.920, *n*(PLA/PBAd 90/10) = 0.989, *n*(PLA/PBAd 75/25) = 1.032 and *n*(PLA/PBAd 50/50) = 1.066). Hence, it seems that in the latter stage of API's dissolution the initially diffusion-controlled phase is compensated by the simultaneous matrix swelling (due to the polyester's wetting) and a small portion of matrix erosion (due to the polyester's hydrolytic degradation) leading in this way to a 'balanced' zero-order release profile, which is essential in achieving stable in vivo pharmacokinetic behavior.

#### Morphology Evaluation after Dissolution Studies

In a further step, in order to examine the process of polyester degradation/erosion during dissolution and to correlate this with the enzymatic hydrolysis results presented in Section 3.1.2, SEM images were taken after the completion of the test (Figure 9). As evidenced, in the case of neat PLA and the two polyesters containing only a small amount of PBAd, namely PLA/PBAd 95/5 and 90/10, the surface and shape of the prepared microspheres remained practically unchanged (Figure 9). On the other hand, neat PBAd and the polymeric matrices containing high PBAd load (i.e., PLA/PBAd 75/25 and PLA/PBAd 50/50) demonstrated some clear evidence of surface erosion, presumably due to polyester hydrolysis or drug dissolution. From these images, we can thus conclude that the amount of PBAd bares a crucial role to the extent of polyester degradation and consequently the drug release rates, which is in accordance to previously discussed results from neat polymer enzymatic hydrolysis studies.

#### A Mechanistic Release Model

Finally, since the so-called "standard" dissolution release models used in the literature and herein (see Equations (4)–(8)) present several limitations related to the assumptions made for their implementation (for details please see Reference [33]), new, more sophisticated models were also tested for modeling the obtained results.

In general, there are two types of models to describe a physicochemical process such as the drug's dissolution. The first kind is the so-called empirical models. The physical content of these models is limited. Some of them are just equations used to describe appropriately a large amount of experimental data and some of them have a kind of qualitative information of the physical mechanism that is responsible for the process evolution. The second type of models are the so-called mechanistic models. These models include information of the underlying mechanism and in addition they can consider several mechanisms acting simultaneously. After considering exhaustively the whole toolbox of existing empirical models to describe the present data, an attempt to construct a mechanistic model of the present release process was made.

Looking closely at the form of the release data, the performance of the "standard" release models and the data for hydrolysis evolution of the polymer matrices, the following scenario appears: There is an initial fast release phase that can be partially attributed to the presence of API probably in the form of a thin film layer located on, or near, the surface of the MPs. It is not clear if the mechanism of this layer release is diffusion or matrix erosion since both are equally probable. The second release phase (which is slower) is mostly controlled by Fickian diffusion, although a small erosion contribution is also there (verified by the SEM images presented in Figure 9). Assuming that a fraction of the drug in the polymer is free to move and its motion occurs through the diffusion mechanism and that the shape of the particles is approximately spherical (verified by SEM that is imaged in Figure 5), the transient partial differential equation of diffusion is probably the best model to describe the dissolution behavior of the API [49]. However, in this case a very

simple exponential form, called the linear driving force approximation, can be also used to model the obtained results [50]. This same approximation was also used in the present study for modeling the dissolution kinetics of the API located on the surface layer, despite its unknown release mechanism. Finally, there is a fraction of drug immobilized in the polymer matrix. This fraction can be released only through matrix erosion (due to polyester hydrolysis). In the absence of any other information a zero order release dynamics model will be assessed for this fraction. It should be pointed out that for the limited extent of hydrolysis observed here, this approximation is quite realistic. By summarizing the above arguments, the released drug fraction evolution can be approximated by the following (uniformly valid in time) expression:

$$\mathbf{C\_{r}} = \varphi\_{1}(1 - \exp(-\mathbf{k\_{1}t})) + \varphi\_{2}(1 - \exp(-\mathbf{k\_{2}t})) + \mathbf{k\_{3}t} \tag{9}$$

where Cr is the cumulative API released (%), ϕ<sup>1</sup> and k1 are the percentage of drug in the excess layer and the corresponding kinetic constant respectively, ϕ<sup>2</sup> and k2 are the percentage of mobile drug and the corresponding kinetic constant respectively and k3 is the kinetic constant of the erosion process. It is noted that the linear superposition of diffusionand erosion-induced release is allowed only because the erosion extent is small.

Figure 10 shows the comparison between the predicted and the experimentally derived points after fitting to Equation (9).

**Figure 10.** Comparison of experimental release data (symbols) to the mechanistic model-based Equation (9) (continuous lines). The presentation is made in two-time scales just for clarity: (**a**,**b**) 0–30 days and (**c**,**d**) 0–0.5 days.

The R2 factor was larger than 0.99 in all cases except for composites 75/25 and 50/50 for which it was 0.98 and 0.99, respectively (probably due to a more complicated release

×

scenario than the one described by Equation (9)). Nevertheless, and despite this small pitfall, the mechanistic model proposed herein is still more efficient compared to the "standard" empirical model tested previously, since it is able to model the dissolution profile of the API within the whole-time domain of the test (i.e., both release phases simultaneously). The values of the fitting parameter according to Equation (9) are presented in Table 6.


**Table 6.** Parameters derived by fitting Equation (9) to the experimental drug release data.

According to the obtained results, the percentage of drug in the excess layer was 20% and the corresponding parameters ϕ<sup>1</sup> and k2 did not show any systematic correlation to the copolymer matrix composition. This is expected to be the case for a rather random procedure of accumulation of drug in the surface layer. The fraction of the mobile drug appears to increase consistently from 13% (for PLA) to 32% (for PBAd), while the corresponding kinetic constant k2 appeared also to increase in the same order (with the exception of the 95/5 composite). Finally, the erosion constant k3 increased as the content of PBAd increased, which is in agreement with the hydrolysis rates evaluation presented previously.

The diffusion coefficient, D, of the mobile drug presented also in Table 6 was calculated based on the following equation [50]:

$$\mathbf{D} = \mathbf{k}\_2 \mathbf{r}^2 / 15 \tag{10}$$

where r is the radius of the particles (presented in Table 6). Based on the obtained results, the range of values corresponding to D consisted of the drug diffusion within the polymer matrix and is in agreement with the results describing the second (and slower) release phase (depicted by k2 constant). However, no such relation was proven in the case of the k1 constant, since the characteristic length of diffusion in the first fast release phase is unknown. So, it can be said that the release of the drug's initial fraction (i.e., ϕ1) may be either from the fast erosion of the very thin API layer located on the surface of the MPs or due to the fast initial API diffusion from this surface layer.

#### **4. Conclusions**

In the present study PLA/PBAd-based ARI-loaded LAI MPs were successfully prepared for the first time. Results regarding the highly tunable enzymatic hydrolysis profile and the low cytotoxicity of the new copolymers, amplified the previously made suggestions that these new copolymers can be considered as a quite promising candidate for the preparation of drug sustained release formulations. Evaluation in terms of morphological characteristics (via SEM), productivity (in terms of MPs' yield) and drug loading also showed extremely promising results. Physicochemical analysis of the prepared formulations revealed the amorphous API dispersion with increasing PBAd content, while no specific molecular interactions between the drug and the polyesters were recorded, based on FTIR spectroscopy. Lastly, in terms of the in vitro dissolution profile, results suggested that the newly synthesized PLA/PBAd block copolymers can successfully control the release rate and extent of the API's release from the prepared MPs, indicating that, probably, under in vivo conditions their use may lead to new formulations that will be able

to maintain a continuous therapeutic level for an extended time period (>30 days), with reduced lag-time, as compared to the currently marketed ARI LAI product.

**Author Contributions:** Methodology, investigation, V.K.; A.S.; N.D.B.; investigation, formal analysis, writing, E.C.; methodology, formal analysis, A.Z.; release modeling, writing, M.K.; cytotoxicity studies, writing, E.G.; writing—original draft preparation, review and editing, P.B.; conceptualization, supervision, writing—review and editing, M.V. All authors have read and agreed to the published version of the manuscript.

**Funding:** The APC was funded by Pharmathen SA (Hellas).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **NIR Light-Triggered Chemo-Phototherapy by ICG Functionalized MWNTs for Synergistic Tumor-Targeted Delivery**

**Lu Tang 1,2, Aining Zhang 1,2, Yijun Mei 1,2, Qiaqia Xiao 1,2, Xiangting Xu 1,2 and Wei Wang 1,2,\***


**Abstract:** The combinational application of photothermal therapy (PTT), chemotherapy, and nanotechnology is a booming therapeutic strategy for cancer treatment. Multi-walled carbon nanotube (MWNT) is often utilized as drug carrier in biomedical fields with excellent photothermal properties, and indocyanine green (ICG) is a near-infrared (NIR) dye approved by FDA. In addition, ICG is also a photothermal agent that can strongly absorb light energy for tumor ablation. Herein, we explored a synergistic strategy by connecting MWNT and a kind of ICG derivate ICG-NH2 through hyaluronic acid (HA) that possesses CD44 receptor targeting ability, which largely enhanced the PTT effect of both MWNT and ICG-NH2. To realize the synergistic therapeutic effect of chemotherapy and phototherapy, doxorubicin (DOX) was attached on the wall of MWNT via π–π interaction to obtain the final MWNT-HA-ICG/DOX nanocomplexes. Both in vitro and in vivo experiments verified the great therapeutic efficacy of MWNT-HA-ICG/DOX nanocomplexes, which was characterized by improved photothermal performance, strengthened cytotoxicity, and elevated tumor growth inhibition based on MCF-7 tumor models. Therefore, this synergistic strategy we report here might offer a new idea with promising application prospect for cancer treatment.

**Keywords:** multi-walled carbon nanotube; photothermal therapy; indocyanine green; synergistic strategy; cancer treatment; targeted drug delivery

#### **1. Introduction**

Cancer remains one of the deadly diseases that seriously threatens human health. Despite the encouraging progress of medical advancement, effective therapeutic methods against cancer are still insufficient. Conventional treatment modalities such as chemotherapy, radiotherapy, and surgery often undergo many drawbacks such as unavoidable side effects, severe pain, potential development of drug resistance, and inadequate effectiveness due to the instability and rapid clearance of drugs, which all cause unsatisfactory outcomes of anticancer therapy [1–3]. To overcome the limitations mentioned above, more and more attempts based on targeted drug delivery have been developed to improve cancer treatment efficacy. In this context, nanomaterial-mediated platforms have been widely explored in anticancer drug delivery, which serve as effective carriers of both therapeutic agents and diagnostic agents due to their distinctive properties and unique advantages, such as increased drug stability, reduced systemic toxicity, improved pharmacokinetics, elevated bioavailability, precise drug transportation capability, and controlled drug release ability [4–7].

From all the nanomaterials, carbon nanotubes (CNTs) have captured many researchers' attention due to their multiple application possibilities in cancer theranostics [8]. CNTs

**Citation:** Tang, L.; Zhang, A.; Mei, Y.; Xiao, Q.; Xu, X.; Wang, W. NIR Light-Triggered Chemo-Phototherapy by ICG Functionalized MWNTs for Synergistic Tumor-Targeted Delivery. *Pharmaceutics* **2021**, *13*, 2145. https:// doi.org/10.3390/pharmaceutics13122145

Academic Editors: Francisco José Ostos, Pilar López-Cornejo and José Antonio Lebrón

Received: 22 November 2021 Accepted: 10 December 2021 Published: 13 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

possess tiny tubular shapes composed of carbon atoms that are ordered to form a honeycomb nanostructure with many unique physicochemical characteristics [9]. Generally, CNTs can be sorted into either single-wall carbon nanotubes (SWNTs) or multi-wall carbon nanotubes (MWNTs) according to the sheet number of carbon atoms, both playing a significant role in anticancer therapy [10]. Due to their multifunctionality, CNTs are widely investigated in cancer treatment through various therapeutic modalities. For instance, the strong absorption of CNTs in near-infrared (NIR) regions make them ideal candidates in phototherapy. In addition, CNTs can transform the laser energy to acoustic signals and display excellent resonant Raman scattering and photoluminescence in NIR region, which are all advantageous to their application in cancer imaging [11]. Moreover, many studies have reported that CNTs can be taken up by various cell types due to their needle-like architecture, which enhances their deep tumor penetration to act as ideal drug delivery platforms in anticancer therapy [12]. Additionally, the ultra-high surface area of CNTs for drug loading also benefits their utilization in anticancer drug delivery. However, there are still some obstacles that limit their broad application in biomedical fields. For example, the unique nanostructure of CNTs promotes their hydrophobicity and causes cytotoxicity [13]. Therefore, functionalizing CNTs to increase their hydrophilicity as well as attenuate their inherent cytotoxicity is a key point to improve their biocompatibility for safe application in cancer therapy [14].

Phototherapy is an emerging therapeutic modality that takes advantage of laser energy to eliminate the target tumor due to its high selectivity [15,16]. Photothermal therapy (PTT) and photodynamic therapy (PDT) are two typical phototherapeutic approaches, and NIR light is always used as the light source for phototherapy due to its deep tissue penetration capability [17]. The conversion of absorbed photon energy into thermal energy to cause hyperthermia for tumor ablation is known as PTT, while the absorbance of specific light energy to produce cytotoxic reactive oxygen species (ROS) is termed as PDT, both of which play critical roles in cancer phototherapy [18]. Moreover, in recent decades, light-triggered therapies have been developed as safe treatment modalities to ablate numerous tumors with great effectiveness [18,19]. PTT, one of the representative phototherapy methods, has gained considerable attention in cancer treatment due to its various merits such as minimal trauma, easy implementation, and fewer side effects, which can effectively inhibit solid tumor growth through localized thermal destruction [20]. Indocyanine green (ICG) is a NIR dye approved by FDA for application in phototherapy due to its strong light absorbance in the NIR window [21,22]. However, some intrinsic limitations such as poor solubility, instability, concentration-dependent aggregation, and rapid clearance largely impede the effectiveness of ICG in phototherapy [23]. With the aim of overcoming these aforementioned drawbacks, construction of an appropriate drug carrier to load ICG is of great necessity [24].

In this work, a synergistic chemo-phototherapy was achieved by adopting MWNT to carry photothermal agent ICG-NH2 and anticancer drug doxorubicin (DOX) (Scheme 1). Briefly, ICG-NH2 was conjugated with hyaluronic acid (HA) by an amide bond to form the HA-ICG conjugate, which improved the water solubility of ICG-NH2. Then, MWNT-HA-ICG was synthesized via an ester bond between the carboxyl groups on acidified MWNT and hydroxyl groups on HA, which significantly enhanced the photothermal performance compared to MWNT or ICG-NH2 alone. In addition, the connection through HA could not only elevate the targetability of MWNT-HA-ICG due to its affinity with CD44 receptors that are over-expressed on the membrane of many tumor cells, but also endow the whole drug delivery system with good dispersity and biocompatibility [25–27]. Furthermore, to realize synergistic chemo-photothermal therapy, DOX was attached on the surface of MWNT by a non-covalent π–π bond to obtain the final MWNT-HA-ICG/DOX nanocomplexes (MWNT-HA-ICG/DOX). The targeted delivery of DOX through MWNT-HA-ICG greatly enhanced the therapeutic efficacy of DOX compared to free administration, while causing reduced side effects due to its non-specificity [28,29]. The novelty of this work is that we combined the drug carrier MWNT with good optical property and photothermal agent ICG-NH2

in one platform to achieve synergistic photothermal therapeutic effect. Meanwhile, HA was used a targeting ligand in this nanosystem, which not only increased the targetability of the whole delivery platform, but also elevated the overall biocompability. To further improve the therapeutic efficacy, the classical anticancer drug DOX was employed in the constructed delivery system to take chemotherapy effect. Altogether, this integrated nanoplatform using combined chemo-phototherapy provides a promising strategy for precise cancer therapy.

**Scheme 1.** Schematic illustration of the construction of MWNT-HA-ICG/DOX nanocomplexes for synergistic cancer chemo-phototherapy.

#### **2. Materials and Methods**

#### *2.1. Materials, Cell Lines and Animals*

MWNTs (diameter: 10–20 nm, length: 5–15 μm) were purchased from Shenzhen Nanotechnologies Port Co., Ltd. (Shenzhen, China). DOX·HCl was purchased from Nanjing Chemlin Chemical Industry Co., Ltd. (Nanjing, China). ICG-NH2 was synthesized by Prof. Dun Wang in Shenyang Pharmaceutical University. Sodium hyaluronic acid (MW = 5 kDa), N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide hydrochloride (EDC·HCl), and N-hydro xysuccinimide (NHS) were obtained from Aladdin Reagent Database Inc. (Shanghai, China). Concentrated sulfuric acid (98%), concentrated nitric acid (65%), 4-dimethylaminop yridine (DMAP), and N,N'-Carbonyldiimidazole (CDI) were purchased from Aladdin Reagent Database Inc. (Shanghai, China). 4 , 6-diamidino-2-phenylindole (DAPI) were obtained from Beyotime Institute of Biotechnology (Shanghai, China). All other reagents were of analytical grade and were commercially available.

Human breast cancer cell line MCF-7 were purchased from the Cell Bank of Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China) and cultured in DMEM medium (Hyclone) supplemented with 10% FBS (Hyclone), 100 U/mL penicillin, and 100 μg/mL streptomycin in a humidified atmosphere of 5% CO2 at 37 ◦C.

Female BALB/c nude mice (4–6 weeks, 18–20 g) were purchased from Qinglongshan Animal Farm (Nanjing, China) and were given free access to food and water. All mice were cared for in compliance with the National Institute of Health (NIH) Guidelines for

the Care and Use of Laboratory Animals and was approved by the Ethics Committee of China Pharmaceutical University (Ethics Code: 2021-12-002). The tumor xenograft models were established by injecting MCF-7 cell suspensions (1 × 106 cells) subcutaneously into the right flank region of nude mice. A caliper was used to measure tumor sizes, and the tumor volume was calculated as follows: (tumor length) × (tumor width)2/2.

#### *2.2. Preparation of MWNT-HA-ICG/DOX Nanocomplexes*

#### 2.2.1. Carboxylation of MWNTs

Pristine MWNTs (300 mg) were suspended in mixed acid (H2SO4/HNO3, *v*/*v* = 3:1, 140 mL) by ultrasonication for 18 h. Then, the suspension was diluted with deionized water and filtered through a 0.22 μm micro-porous membrane, followed by repeated washing through deionized water until the pH of filtrate was neutral. The solid product on the membrane was then redispersed in deionized water and lyophilized to obtain the final MWNT-COOH, which was stored at room temperature for further use.

#### 2.2.2. Synthesis of HA-ICG

Firstly, HA (60 mg) and NHS (43.79 mg) were dissolved in the mixture of water and DMF (1:3), after which EDC solution was added into the HA solution and stirred for 1 h to activate the carboxyl groups of HA. Then, ICG-NH2 (44.4 mg) was dissolved in the mixture of water and DMF (1:3) and added to the HA solution dropwise, followed by stirring for 24 h at room temperature. The resultant solution was dialyzed (MWCO: 3500) using 50% DMF for 24 h to remove excess ICG-NH2 and then dialyzed using deionized water for another 48 h to remove DMF. Finally, the solution was lyophilized to obtain the HA-ICG and stored at −20 ◦C for further use.

#### 2.2.3. Synthesis of MWNT-HA-ICG

MWNT-COOH (10 mg) and CDI (12 mg) were dissolved in 6 mL formamide and stirred for 1 h to activate the carboxyl groups of MWNT-COOH. Afterwards, the activated MWNT-COOH was dropped into HA-ICG solution in formamide with DMAP (15 mg) and stirred for 24 h under the protection of nitrogen. Next, the mixed solution was dialyzed (MWCO: 8000–14000) using 50% DMF for 24 h to remove the excess CDI and DMAP and then dialyzed using deionized water for another 48 h to remove the organic solvent. The final MWNT-HA-ICG was lyophilized and stored at −20 ◦C for further use.

#### 2.2.4. Synthesis of MWNT-HA-ICG/DOX

First, 10 mg MWNT-HA-ICG and 10 mg DOX·HCl were dissolved in deionized water, respectively. Then, DOX solution was added into MWNT-HA-ICG solution and stirred for 24 h at room temperature. Afterwards, the solution was centrifuged (12,000 rpm, 10 min), and the precipitate was washed with PBS (pH 7.4) to remove free DOX by repeated centrifugation. Finally, the resultant precipitate was MWNT-HA-ICG/DOX and was lyophilized and stored at −20 ◦C for further use.

#### *2.3. Characterization of MWNT-Based Formulations*

Solid samples of raw MWNT and MWNT-COOH were prepared to obtain their Raman spectra using confocal micro-Raman spectroscopy (LabRam HR800, Paris, France). The successful synthesis of HA-ICG and the grafting percentage of ICG on HA were confirmed by 1H-NMR spectroscopy (Avance™ 600, Bruker, Germany, 300 MHz), UV, and fluorescence spectroscopy. For the measurement of 1H-NMR spectra, HA and HA-ICG were dissolved in D2O, and ICG-NH2 was dissolved in DMSO-*d6*. Then, the measurement was carried out using 300 MHz under 20 ◦C. UV spectra were collected by dissolving ICG-NH2, DOX, MWNT-COOH, MWNT-HA-ICG, and MWNT-HA-ICG/DOX into deionized water, following by the scanning of their wavelength from 200–900 nm. The fluorescence spectra of samples were obtained by dissolving ICG-NH2, DOX, and MWNT-HA-ICG/DOX into deionized water, which was then irradiated by 760 nm and 490 nm excitation light to obtain

their fluorescence spectra. In addition, thermogravimetric analysis (TGA) was performed to analyze raw MWNT, MWNT-COOH, and MWNT-HA-ICG (heat from 30 ◦C to 700 ◦C, nitrogen, heating rate at 10 ◦C/min). The particle size and zeta potential of various MWNTbased formulations were measured by dynamic light scattering (DLS) using a Malvern Zetasizer (Nano ZS-90, Malvern Instruments Ltd., Malvern, UK). The morphology of raw MWNT, MWNT-COOH, and MWNT-HA-ICG/DOX was verified by transmission electron microscope system (TEM, Hitachi, Japan, 80 kV).

#### *2.4. In Vitro Release of DOX from MWNT-HA-ICG/DOX*

In vitro drug release profiles of DOX from MWNT-HA-ICG/DOX were performed at 37 ◦C in three different pH phosphate-buffered solutions (PBS). Briefly, 1 mg MWNT-HA-ICG/DOX was dispersed in 2 mL release medium with pH of 7.4, 6.5, or 5.5 in dialysis bag (MWCO: 3500). Then, the dialysis bag was placed in 40 mL corresponding pH release medium. The resultant dispersions were gently shaken at 37 ◦C in water bath at 100 rpm. At predetermined time intervals, 4 mL of release medium was taken out and the amount of released DOX was determined by a fluorospectrophotometer (HORIBA, Fluoromax-4, Palaiseau, France). In the meantime, 4 mL of same fresh release medium was replenished.

#### *2.5. In Vitro Photothermal Effect of MWNT-Based Formulations*

First, 1 mL of different concentrations free ICG-NH2, MWNT-COOH, and MWNT-HA-ICG/DOX (containing 20 μg/mL ICG-NH2 and 100 μg/mL MWNT-COOH) were dispersed in PBS and placed in 4 mL tubes. Then, different prepared solutions were irradiated by 808 nm laser (1 W/cm2) for 5 min. The temperatures were recorded every 25 s, and thermographic maps of the dispersion in tubes were taken by thermo imager (FLIR, E64501, Goleta, CA, USA) at the time point of 5 min.

#### *2.6. In Vitro Cytotoxicity Assay*

The cytotoxicity of various prepared formulations was evaluated through MTT assay. Briefly, MCF-7 cells were seeded in 96-well plates at a density of 1 × <sup>10</sup><sup>4</sup> cells/well and incubated for 24 h. For treatment groups without laser irradiation, the culture medium was replaced with 200 μL medium containing ICG-NH2, DOX, MWNT-HA, MWNT-HA/DOX, MWNT-HA-ICG, and MWNT-HA-ICG/DOX at different concentrations and incubated for 24 h. For treatment groups with laser irradiation, the culture medium was replaced with 200 μL medium containing ICG-NH2, MWNT-HA/DOX, MWNT-HA-ICG, MWNT-HA-ICG/DOX, and MWNT-HA and incubated for 4 h, followed by laser irradiation (1 W/cm2, 5 min); then, laser treatment groups were further incubated for 20 h. After the total 24 h incubation of treatment groups with/without laser irradiation, 20 μL MTT solution (5 mg/mL) was added to each well, and the cells were incubated for 4 h. Afterwards, the medium was removed, and 150 μL DMSO was added. The absorbance measured by microplate reader (Thermo, Multiskan FC) at 570 nm. The cell viability was calculated using the following equation: Cell viability (%) = (Asample − Ablank)/(Acontrol − Ablank) × 100%.

#### *2.7. Cellular Uptake and Intracellular Trafficking*

MCF-7 cells were seeded in 24-well plates and cultured until 80% cell confluence. The medium was discarded, and cells were washed twice with PBS before adding serum-free medium with ICG-NH2, DOX, MWNT-HA-ICG, MWNT-HA/DOX, and MWNT-HA-ICG/DOX at a final concentration of 10 μg/mL ICG-NH2 and/or 10 μg/mL DOX for cellular uptake for 1, 2, and 4 h. For competition assay, cells were pretreated with free HA (5 mg/mL) for 4 h before adding MWNT-HA-ICG/DOX. The cellular uptake of different formulations was determined by flow cytometry (BD FACS Calibur, San Jose, CA, USA) quantitatively according to the fluorescent property of ICG and DOX. The intracellular fluorescence of the above formulations was observed under a confocal laser scanning microscope (CLSM, Leica TCS SP5, Wetzlar, Germany) after staining cells with DAPI.

For intracellular trafficking study, cells were cultured, incubated with different formulations, washed, and fixed by 4% paraformaldehyde for 15 min. After discarding the paraformaldehyde, DAPI was added to stain the cell nucleus for 10 min. Finally, distribution of different formulations was observed by CLSM.

#### *2.8. Cell Apoptosis Assessment*

The in vitro antitumor efficacy was carried out by an apoptosis experiment. MCF-7 cells were seeded in 24-well plates at a density of 5 × 104 cells/well and incubated for 24 h. Cells were co-incubated with control (PBS), free DOX, and MWNT-HA-ICG/DOX (containing 10 μg/mL ICG and 10 μg/mL DOX) for 24 h to evaluate the chemotherapeutic efficacy of constructed nanocomplexes. To study the photothermal therapeutic efficacy, cells were co-incubated with control (PBS), ICG-NH2, MWNT-HA-ICG, MWNT-HA/DOX, and MWNT-HA-ICG/DOX (containing 10 μg/mL ICG and 10 μg/mL DOX) for 4 h and then irradiated for 5 min (1 W/cm2), followed by another 20 h incubation. Afterwards, cells were collected and washed with PBS. Next, cells were suspended in binding buffer and stained with Annexin V-FITC/PI apoptosis detection kit in dark. Finally, the apoptotic cells were detected by flow cytometry.

#### *2.9. In Vivo Imaging Study*

Mice were randomly divided into three groups and injected with ICG-NH2, MWNT-HA-ICG, and MWNT-HA-ICG/DOX at a dose of 10 μg ICG-NH2/mouse via tail vein, respectively. Due to the fluorescent characteristic of ICG, the in vivo distribution and targeting efficiency could be evaluated using an in vivo imaging system (FX PRO, Kodak, Rochester, NY, USA) at predetermined time points. After 12 h post-injection, mice were sacrificed to obtain the tumor and major organs for ex vivo imaging quantitative analysis using the same imaging system. The temperature changes of tumors during the irradiation (5 min, 1 W/cm2) were monitored by a thermo-imager (FLIR, E64501) after 6 h post-injection at the time points of 0 min and 5 min.

#### *2.10. In Vivo Antitumor Efficacy*

The in vivo antitumor efficacy was evaluated using MCF-7 tumor bearing xenograft nude mice with average tumor volume around 100 mm3. All the mice were weighed and randomly divided into nine groups. Of the nine groups, four groups were administrated with PBS, DOX, ICG-NH2, and MWNT-HA/DOX intravenously without irradiation, while the rest groups were treated with PBS, ICG-NH2, MWNT-HA-ICG, MWNT-HA/DOX, and MWNT-HA-ICG/DOX with laser irradiation. All the formulations were administrated at a dose of 10 μg ICG-NH2/mouse and/or 0.5 mg/kg DOX per mouse every 2 days. For laser groups, tumors on mice were irradiated by 808 nm laser (1 W/cm2) for 5 min at 6 h after injection with different formulations. In the meantime, the tumor volumes and body weights of mice were recorded every 2 days until day 14.

#### *2.11. Statistical Analysis*

The data were expressed as mean ± S.D. from triplicate experiments conducted parallelly, unless otherwise noted. Statistical analysis was carried out through one-way analysis of variance (ANOVA) test for comparison of multiple groups. Statistical significance was regarded as \* *p* < 0.05, \*\* *p* < 0.01, or \*\*\* *p* < 0.001.

#### **3. Results**

#### *3.1. Synthesis of MWNT-HA-ICG/DOX*

Pristine MWNT was long and covered by impurities, which hindered the direct application of MWNT as a drug delivery vehicle. Therefore, mixed acid (H2SO4/HNO3 *v*/*v* = 3:1) was used to remove the impurities, shorten the length of MWNT, and enable it to be modified with carboxyl group for further reaction. Moreover, MWNT is highly hydrophobic, which restricts its application in vivo. Thus, hydrophilic polymers are essential to be conjugated on the surface of MWNT to improve its dispersity. Hyaluronic acid (HA) is a negative polysaccharide with high molecular weight and consists of repeated D-glucuronic acid and *N*-acetyl-D-glucosamine disaccharide units [30]. HA has great water solubility and active hydroxyl and carboxyl groups, making it a prospective connection between MWNT-COOH and ICG-NH2 [31]. As the synthesis route shown in Figure 1, ICG-NH2 and HA were firstly conjugated through amide reaction, the resultant HA-ICG improved the water solubility of ICG-NH2 and was used as the reactant in the second step. Next, hydroxyl group on HA-ICG and carboxyl group on MWNT-COOH were connected through ester bond. By virtue of excellent water solubility of HA, the dispersity of MWNT-HA-ICG complex was greatly improved. Moreover, both the photosensitizer ICG and the drug vector MWNT have excellent optical features, which enabled ICG and MWNT to realize synergistic PTT effect. To further improve the antitumor efficacy, chemotherapy drug DOX was conjugated onto the wall of MWNT via a π–π bond; thus, the final MWNT-HA-ICG/DOX nanocomplexes could be obtained.

**Figure 1.** Synthesis route of MWNT-HA-ICG/DOX.

#### *3.2. Characterization of MWNT-HA-ICG/DOX*

The successful synthesis of MWNT-HA-ICG/DOX was characterized by Raman, 1H-NMR, UV, fluorescence spectra, and thermogravimetric analysis (TGA). As shown in Figure 2A, both Raman spectra of MWNT and MWNT-COOH displayed two characteristic peaks. Tangential G band (~1590 cm−1) represented the in-plane vibration of C–C bond, and the D band (~1350 cm−1) reflected the disorder in the carbon system. ID/IG value of raw MWNT was 1.13, and the value of MWNT-COOH increased to 1.31. The increased value of ID/IG could reflect the elevated degree of deficiency of MWNT, probably due to the carboxylation of MWNT. The conjugation between carboxyl groups on HA and amine groups of ICG-NH2 was verified by 1H-NMR. The solvent for HA and HA-ICG was D2O, and the solvent for ICG-NH2 was DMSO-*d*6. As shown in Figure 2B, the typical chemical

shift of -CH- in HA was identified at 4.4~4.6 ppm, and the characteristic peaks of N-acetyl group and hydroxyl groups in HA were identified at 2.0 ppm and 4.0 ppm, respectively [32]. In addition, the evident peak of Ar-H due to proton resonance and the characteristic peak of -CH=CH- in ICG-NH2 were identified at 7.5–8.3 ppm and 6.3–6.7 ppm, respectively. In the 1H-NMR spectrum of HA-ICG, the characteristic peaks of HA and ICG-NH2 were both identified, manifesting the successful conjugation of ICG-NH2 onto HA. Moreover, as the UV spectrum was illustrated in Figure 2C, free ICG-NH2, DOX, and MWNT-COOH had absorbance peaks at 780 nm, 496 nm, and 253 nm, respectively, while modified MWNT-HA-ICG showed absorbance at wavelengths of 785 nm and 253 nm, and the final MWNT-HA-ICG/DOX displayed absorbance peaks at 785 nm, 496 nm, and 253 nm, indicating that HA-ICG was connected onto the wall of MWNT-COOH, and MWNT-HA-ICG/DOX was successfully synthesized. Similarly, as the fluorescence spectra shown in Figure 2D, free ICG-NH2 and DOX could be excited by wavelengths of 760 nm and 490 nm, which showed emission peaks at 803 nm and 590 nm, respectively. Meanwhile, MWNT-HA-ICG/DOX also had emission peaks at the wavelength of 792 nm and 590 nm, confirming the successful synthesis of MWNT-HA-ICG/DOX.

**Figure 2.** (**A**) Raman spectra of MWNT and MWNT-COOH. (**B**) 1H-NMR spectra of HA, ICG-NH2, and HA-ICG. (**C**) UV spectra of ICG-NH2, DOX, MWNT-COOH, MWNT-HA-ICG, and MWNT-HA-ICG/DOX. (**D**) Fluorescence spectra of ICG-NH2, DOX, and MWNT-HA-ICG/DOX. (**E**) TGA curves of MWNT, MWNT-COOH, and MWNT-HA-ICG.

TGA is a common way to determine the contents of grafting substance on MWNT by measuring the weight loss from 30 ◦C to 700 ◦C. As shown in Figure 2E, pristine MWNT was quite stable when the temperature was below 500 ◦C. In the range of 500 ◦C to 700 ◦C, there was around 9.84% weight loss of MWNT, mainly due to the impurities on pristine MWNT. With the temperature rising, the carboxyl groups on MWNT-COOH decomposed. Additionally, at a temperature of 700 ◦C, the weight loss of MWNT-COOH was approximately 22.37%, which implied that the content of carboxyl groups in MWNT-COOH was about 12.53%. For MWNT-HA-ICG, its weight loss accompanied with temperature increment also reflected the amount of HA-ICG grafted on MWNT-COOH, which was approximately 43.25%, demonstrating that HA-ICG could be effectively conjugated on the wall of MWNT-COOH.

#### *3.3. Particle Size, Zeta Potential, and Morphology*

The particle size, zeta potential, and polydispersity index (PDI) were measured to investigate the variation of MWNT after modified with different molecules. As is shown in Figure 3A, the particle size of MWNT-HA-ICG/DOX was around 190 nm and was larger than that of MWNT-COOH, which was due to the functionalization of ICG-conjugated HA and the non-covalent loading of DOX. In addition, the negative-charged nanocomplexes tended to be more stable in vivo due to their avoidance of the interaction with the negativecharged components in plasma [33,34]. The morphology of MWNT, MWNT-COOH, and MWNT-HA-ICG/DOX were observed by transmission electron microscopy (TEM). As shown in Figure 3B, raw MWNT was long, twined, and aggregated together with impurities on its surface. After acidification, MWNT-COOH was much shorter than raw MWNT, with a particle size of approximately 160 nm, which was in accordance with the result measured by dynamic light scattering (DLS). Moreover, the surface of MWNT-COOH was very smooth and no longer woven together, demonstrating that the mixed acid could efficiently shorten the raw MWNT and remove the impurities. After conjugation with HA-ICG and loading with DOX, the thickness of the MWNT was obviously increased and the length of functionalized MWNT was a little longer, demonstrating that HA-ICG and DOX were successfully attached to the surface of MWNT-COOH.

#### *3.4. In Vitro Drug Release and Photothermal Effect of MWNT-Based Formulations*

To investigate the in vitro drug release profile of DOX from MWNT-HA-ICG/DOX, three different pH phosphate buffers were used to simulate physiological pH, tumor pH, and lysosomal pH, respectively. As illustrated in Figure 3C, the cumulative released amount of DOX was pH-dependent. At pH 7.4, approximately 20% DOX was released from MWNT-HA-ICG/DOX after 48 h, implying that DOX would not release too much under normal physiological environment because the π–π stacking interaction between DOX and MWNT was stable. In contrast, approximately 30% and 50% of the loaded DOX was released from MWNT-HA-ICG/DOX at pH 6.5 and pH 5.5 after 48 h, respectively, suggesting the accelerated release profile of DOX in acidic tumor sites after internalization inside the tumor cells through receptor-mediated endocytosis. Altogether, the data above demonstrated that DOX could be released from MWNT-HA-ICG/DOX in a sustained manner in tumor sites and the constructed nanocomplexes were stable under normal physiological condition, which was attributed to attenuated π–π stacking interaction between DOX and MWNT due to the amino protonation of DOX under acidic PH conditions. To determine the in vitro PTT efficiency of ICG-NH2, MWNT-COOH, and the synergistic effect of ICG-NH2 and MWNT-COOH, a series of concentrations of the aforementioned three solutions were irradiated for 5 min (808 nm laser, 1.0 W/cm2) and the corresponding temperatures were recorded at the designed time points. As shown in Figure 3D,E, the photothermal performance of both free ICG-NH2 and MWNT-COOH exhibited a concentration-dependent and time-dependent profile. The temperature of MWNT-COOH was almost proportional to time, while the temperature of ICG-NH2 was not that case. When the concentration of ICG-NH2 was relatively low, the temperature increased signifi-

cantly with the increment of its concentration and time. However, when the concentration of ICG-NH2 was over 50 μg/mL, the temperature did not show an obvious increase with the increment of concentration, which was probably due to the aggregation of ICG-NH2 that blunted its photothermal properties. As illustrated in Figure 3F, the temperature of MWNT-HA-ICG/DOX that contained 20 μg/mL ICG-NH2 and 100 μg/mL MWNT-COOH was higher than that of free ICG-NH2 and MWNT-COOH alone, demonstrating the synergistic photothermal effect of ICG-NH2 and MWNT-COOH. Moreover, as shown in Figure 3G, the thermographic maps taken at 5 min after laser irradiation displayed an increased temperature in MWNT-HA-ICG/DOX group compared with the others, which was in accordance with the results above.

**Figure 3.** (**A**) Hydrodynamic diameter, zeta potential, and PDI of different MWNT-based formulations. (**B**) TEM images of raw MWNT, MWNT-COOH and MWNT-HA-ICG/DOX. (**C**) Cumulative release curves of DOX from MWNT-HA-ICG/DOX in three phosphate buffers with different pH. In vitro temperature curves of (**D**) free ICG-NH2, (**E**) MWNT-COOH at various concentrations, and (**F**) free ICG-NH2 (20 μg/mL), MWNT-COOH (100 μg/mL) and MWNT-HA-ICG/DOX after laser irradiation for 5 min (808 nm, 1.0 W/cm2). (**G**) Infrared thermographic maps of free ICG-NH2, MWNT-COOH, and MWNT-HA-ICG/DOX determined at 5 min after continuous laser irradiation.

#### *3.5. In Vitro Cytotoxicity Studies*

In vitro cell viability of free ICG-NH2, free DOX, and different MWNT-based formulations with or without laser irradiation was evaluated using MCF-7 cells by MTT assay. Since the amount of ICG-NH2 and DOX in constructed nanocomplexes was the same, and their concentrations were consistent with the concentration increment of MWNT-COOH, cell viability assays were carried out according to the concentrations of therapeutic agents (ICG-NH2/DOX and MWNT-COOH). Figure 4A showed the cell viability of different treatment groups without laser irradiation, which illustrated that the cell viability of free ICG-NH2 group was relatively high, implying the hypotoxicity of ICG-NH2. Meanwhile, the cell viability after MWNT-HA and MWNT-HA-ICG treatment decreased as the concentration increased, indicating that the cytotoxicity of MWNT-HA and MWNT-HA-ICG was concentration-dependent. However, the overall cytotoxicity of this nanocarrier was relatively low, even at the highest concentration, which was attributed to the improved biocompatibility and reduced cytotoxicity of MWNT through HA modification. Furthermore, the cytotoxicity of MWNT-HA/DOX and MWNT-HA-ICG/DOX was significantly higher than that of MWNT-HA and MWNT-HA-ICG, which was mainly due to the cytotoxic effect of DOX. For the laser treatment groups shown in Figure 4B, MCF-7 cell viabilities upon laser irradiation (808 nm, 1.0 W/cm2) for 5 min declined than those of unirradiated groups in Figure 4A. The higher the concentration of ICG-NH2 and MWNT-COOH, the better the photothermal therapeutic effect. In addition, from all the treatment groups in Figure 4B, cells incubated with MWNT-HA-ICG/DOX plus laser irradiation showed the lowest viability, indicating the synergistic therapeutic effect through chemo-phototherapy of the constructed nanocomplexes.

#### *3.6. In Vitro Cellular Uptake Studies*

The evaluation of in vitro cellular uptake of ICG-NH2, DOX, and different MWNTbased formulations was carried out using flow cytometry. ICG-NH2 and DOX were used as fluorescent probes to quantitatively indicate the amount of MWNT-HA-ICG/DOX internalized by MCF-7 cells. In this experiment, ICG-NH2, DOX, and their corresponding MWNT-based formulations were divided into two groups to detect their fluorescent intensity in two channels according to the emission wavelength. As shown in Figure 4C,E, the fluorescent intensity of free DOX was the lowest compared to other groups, while the fluorescent intensities of MWNT-HA/DOX and MWNT-HA-ICG/DOX were significantly stronger than free DOX. Moreover, after pretreatment with HA, the cellular uptake of MWNT-HA-ICG/DOX was significantly reduced, implying that HA-modified MWNTbased formulations could enter the cells through CD44 receptor-mediated endocytosis [32]. All the results above demonstrated that MWNT-based formulations were more likely to enter the cells due to the targeting property of HA. Moreover, the results of cellular uptake by ICG-NH2 fluorescence shown in Figure 4D,F also confirmed the same cellular uptake mechanism of MWNT-HA-ICG/DOX, which were in accordance with the results obtained in Figure 4C,E.

**Figure 4.** Cell viabilities of MCF-7 cells treated with different formulations at a series of concentrations of ICG-NH2/DOX and MWNT-COOH for 48 h (**A**) without laser irradiation and (**B**) with laser irradiation (ICG-NH2/DOX means ICG-NH2 and/or DOX). Flow cytometric profiles and fluorescence intensities of different formulations of (**C**,**E**) 10 μg/mL DOX, (**D**,**F**) 10 μg/mL ICG-NH2 at 4 h after treatment. Data were expressed as mean ± S.D. (*n* = 3). *\* p* < 0.05, *\*\* p* < 0.01, and *\*\*\* p* < 0.001.

#### *3.7. Intracellular Distribution and Cell Apoptosis Studies*

CLSM was utilized to evaluate the intracellular distribution of MWNT-based formulations. DAPI was employed to dye the cell nucleus (blue). As shown in Figure 5A, the fluorescence of ICG-NH2 (green) and DOX (red) appeared in cytoplasm and nucleus, respectively. The merged fluorescence appeared light purple due to the fluorescent overlay

of DOX (red) and DAPI (blue). More fluorescence of ICG-NH2 and DOX could be observed in MWNT-HA/DOX, MWNT-HA-ICG, and MWNT-HA-ICG/DOX groups, indicating that the functionalization of MWNT through HA enabled more ICG-NH2 and DOX to enter into tumor cells due to the targetability of HA. After pretreatment with HA, the cellular uptake of MWNT-HA-ICG/DOX obviously decreased, which was verified by the decreased ICG-NH2 and DOX fluorescence. In summary, the observation of CLSM and the results of flow cytometry were consistent. Cell apoptosis assay was carried out using flow cytometry to quantitatively evaluate the apoptosis-inducing efficacy of different formulations. The double staining of FITC labeled Annexin V and propidium iodide (PI) was used to discriminate live/early apoptotic cells and dead/late apoptotic cells [16]. As demonstrated in Figure 5B, only about 2% and 3% apoptotic cells were examined in the groups of control and control with laser irradiation, respectively, indicating that treatment using 808 nm laser was safe and would not cause damage to normal cells. In contrast, all the treatment groups using MWNT-based formulations exhibited obvious cell apoptosis. Notably, the combination therapy through MWNT-HA-ICG/DOX upon laser irradiation displayed the highest cell apoptotic rate of 98.18%, illustrating the synergistic chemo-photothermal therapeutic effect through the integration of DOX, ICG-NH2, and MWNT.

#### *3.8. In Vivo Targeting Study*

A drug carrier that is able to target tumor sites plays a significant role in achieving elevated therapeutic outcome and reducing systemic side effects. Therefore, in vivo tumor targeting performance of constructed nanocomplexes were studied on MCF-7 xenograft tumor in nude mice using an in vivo fluorescence imaging system. As is shown in Figure 6A, mice were administered with ICG-NH2, MWNT-HA-ICG, and MWNT-HA-ICG/DOX and then photographed at the time points of 1 h, 4 h, 6 h, and 12 h. Mice injected with ICG-NH2 exhibited a primary liver accumulation with weak fluorescence at the tumor site. At 4 h post-injection, the fluorescence intensity of ICG-NH2 in the liver became weak, partly due to the quenching aggregation and instability of free ICG [35,36]. In contrast, mice administered with MWNT-based formulations exhibited strong fluorescent signal at the tumor sites, even at 12 h after injection, implying that nanocomplexes conjugated with HA were able to produce selective accumulation and retention at tumor sites due to the specific affinity of HA to overexpressed CD44 receptors on MCF-7 cells [37]. As illustrated in Figure 6C, after 12 h injection, tumors and major organs were isolated for ex vivo imaging, which demonstrated that ICG-NH2 was non-specifically accumulated in liver, spleen, and lung with no accumulation in tumor sites. In contrast, fluorescent images of mice treated with MWNT-HA-ICG and MWNT-HA-ICG/DOX showed favorable accumulation at tumor sites, which was attributed to the EPR effect and CD44-guided tumor targetability [38]. In the quantitative analysis shown in Figure 6D, the accumulation of nanocomplexes in tumor and major organs was consistent with that of ex vivo imaging above, which confirmed the excellent tumor targeting ability of the constructed nanocomplexes. In order to assess the PTT efficacy of the constructed formulations, the local temperatures of tumors upon laser irradiation (808 nm, 1.0 W/cm2) after 6 h treatment with PBS, free ICG-NH2, MWNT-HA, and MWNT-HA-ICG/DOX were recorded. As is illustrated in Figure 6B, the temperature of mice injected with ICG-NH2 increased from 30.9 ◦C to 37.6 ◦C after 5 min laser irradiation, which did not show apparent temperature enhancement compared to that in the mice injected with PBS solution, indicating that free ICG-NH2 could not specifically target tumor sites with a slow temperature increment. In contrast, due to the photothermal property of MWNT and targetability of HA, mice administered MWNT-HA showed an improved PTT performance over free ICG-NH2. Notably, the temperature of mice injected with MWNT-HA-ICG/DOX could rise to 55.6 ◦C after irradiation for 5 min, which was high enough for tumor ablation [39]. All the data above confirmed that the constructed nanocomplexes could contribute to a synergistic PTT effect due to the combination of MWNT and ICG-NH2 as well as the enhanced tumor targetability mediated by HA.

**Figure 5.** (**A**) Representative confocal images of intracellular trafficking of ICG-NH2, DOX, MWNT-HA/DOX, MWNT-HA-ICG, MWNT-HA-ICG/DOX, and MWNT-HA-ICG/DOX pretreated with HA in MCF-7 cells. Nucleus stained by DAPI showed blue fluorescence, ICG showed green fluorescence, and DOX showed red fluorescence. (**B**) Cell apoptosis induced by different formulations with and without laser irradiation using flow cytometry analysis. Cells treated with PBS were used as control.

**Figure 6.** (**A**) Representative time-lapse in vivo imaging and biodistribution of MCF-7-tumor-bearing nude mice intravenously injected with ICG-NH2, MWNT-HA-ICG, and MWNT-HA-ICG/DOX. (**B**) Infrared thermographic maps of mice upon laser irradiation (808 nm, 1.0 W/cm2) at 6 h after IV injection with PBS, ICG-NH2, MWNT-HA, and MWNT-HA-ICG/DOX. (**C**) Representative ex vivo NIR imaging of tumors and major organs excised from mice at 12 h post-injection. (**D**) Quantitative analysis of the fluorescence intensity in tumors and major organs at 12 h post-injection. (**E**) Tumor growth and (**F**) body weight curves of mice after IV administered with different formulations. Data were expressed as mean ± S.D. (*n* = 3). *\* p* < 0.05, *\*\* p* < 0.01 and *\*\*\* p* < 0.001.

#### *3.9. In Vivo Antitumor Efficacy Study*

MCF-7 xenograft tumor model was established to investigate the synergistic antitumor efficacy of chemo-photothermal therapy during 14-day treatment. Mice were *i.v.* administered with different formulations with comparable amounts of therapeutics. As illustrated in Figure 6E, tumor volumes of mice treated with PBS, PBS plus laser, free ICG-NH2, and ICG-NH2 plus laser rapidly increased with the time, indicating that single treatment through PTT could not result in obvious inhibitory effect on tumor growth and ICG-NH2 plus laser treatment could not cause considerable antitumor effect either, which was mainly

due to the lack of targetability towards tumor sites [40]. Meanwhile, mice injected with free DOX, MWNT-HA-ICG plus laser, MWNT-HA/DOX, and MWNT-HA/DOX plus laser showed a significantly slight tumor volume increment. Moreover, mice treated with MWNT-based formulations exhibited remarkably slower tumor growth compared to the free DOX group, which indicated that formulations based on MWNT-HA exhibited better tumor targetability, thus causing enhanced antitumor efficacy. In contrast, mice injected with the final MWNT-HA-ICG/DOX plus laser showed a significant tumor volume reduction compared to the other groups above, confirming the synergistic therapeutic efficacy of the chemo (DOX)-photothermal (MWNT and ICG-NH2) strategy in MWNT-HA-ICG/DOX nanocomplexes. Meanwhile, as Figure 6F illustrated, the body weight of mice treated with PBS and PBS plus laser increased in the first 8 days due to the growth of tumor and then decreased in the following days, which might be the result of the enlarged tumor that influenced the health of mice. In addition, the body weight of mice treated with DOX continuously reduced due to the cytotoxicity effect of DOX. Moreover, due to the non-specificity of DOX, treatment with free DOX also caused toxic effects on normal cells, which affected the life quality of mice and contributed to their weight reduction [41]. In contrast, the body weight of mice treated with MWNT-based formulations showed an increased tendency, indicating the reduced toxicity and increased therapeutic efficacy of constructed nanocomplexes that improved the life quality of mice, which was due to the elevated tumor-targeting ability and good biocompatibility of MWNT-based formulations after the modification of HA on MWNT.

#### **4. Conclusions**

In summary, we successfully fabricated a nano-based drug delivery system with synergistic PTT and chemotherapy effect for efficient tumor elimination. The integration of two photothermal agents ICG-NH2 and MWNT through the connection of HA could not only elevate the photothermal performance compared to the single treatment modality, but also improve the targetability of the whole nanocomplexes due to the specific binding of HA and CD44 receptors overexpressed in tumor cells. Moreover, a simultaneous therapeutic effect could be achieved after involving DOX in this drug delivery system. In vitro results showed that MWNT-HA-ICG/DOX plus laser irradiation could lead to significant cytotoxic effects towards MCF-7 cells. In vivo experiments demonstrated that the combinational treatment strategy through PTT and chemotherapy could result in a favorable inhibitory effect on MCF-7 tumor growth. Therefore, MWNT-HA-ICG/DOX provided a promising therapeutic opportunity based on synergistic strategies for cancer treatment.

**Author Contributions:** Conceptualization, W.W.; Methodology, L.T., A.Z., Y.M., X.X. and Q.X.; Software and Data Analysis, L.T., X.X. and A.Z.; Original draft preparation, L.T., A.Z., Y.M., Q.X. and X.X.; Review and editing, L.T. and W.W.; Supervision, W.W.; Project administration, W.W.; Funding acquisition, W.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by National Nature Science Foundation of China (Nos. 31872756 and 32071387), National Major Scientific and Technological Special Project for 'Significant New Drugs Development' (No. 2016ZX09101031), Six Talent Peaks Project in Jiangsu Province (JY-079).

**Institutional Review Board Statement:** The animal study in this work was conducted with the approval of the Ethics Committee of China Pharmaceutical University (Ethics Code: 2021-12-002).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Dendrimer-Coated Gold Nanoparticles for Efficient Folate-Targeted mRNA Delivery In Vitro**

**Londiwe Simphiwe Mbatha, Fiona Maiyo, Aliscia Daniels and Moganavelli Singh \***

Nano-Gene and Drug Delivery Group, Discipline of Biochemistry, School of Life Sciences, University of KwaZulu-Natal, Private Bag X54001, Durban 4000, South Africa; londiwem3@dut.ac.za (L.S.M.); fcmaiyo@kabarak.ac.ke (F.M.); DanielsA@ukzn.ac.za (A.D.) **\*** Correspondence: singhm1@ukzn.ac.za; Tel.: +27-31-2607170

**Abstract:** Messenger RNA (mRNA) is not an attractive candidate for gene therapy due to its instability and has therefore received little attention. Recent studies show the advantage of mRNA over DNA, especially in cancer immunotherapy and vaccine development. This study aimed to formulate folic-acid-(FA)-modified, poly-amidoamine-generation-5 (PAMAM G5D)-grafted gold nanoparticles (AuNPs) and to evaluate their cytotoxicity and transgene expression using the luciferase reporter gene (F*Luc*-mRNA) in vitro. Nanocomplexes were spherical and of favorable size. Nanocomplexes at optimum nanoparticle:mRNA (*w*/*w*) binding ratios showed good protection of the bound mRNA against nucleases and were well tolerated in all cell lines. Transgene expression was significantly (*p* < 0.0001) higher with FA-targeted, dendrimer-grafted AuNPs (Au:G5D:FA) in FA receptors overexpressing MCF-7 and KB cells compared to the G5D and G5D:FA NPs, decreasing significantly (*p* < 0.01) in the presence of excess competing FA ligand, which confirmed nanocomplex uptake via receptor mediation. Overall, transgene expression of the Au:G5D and Au:G5D:FA nanocomplexes exceeded that of G5D and G5D:FA nanocomplexes, indicating the pivotal role played by the inclusion of the AuNP delivery system. The favorable properties imparted by the AuNPs potentiated an increased level of luciferase gene expression.

**Keywords:** gold nanoparticles; PAMAM dendrimers; folic acid; mRNA; gene expression

#### **1. Introduction**

Over the years, non-viral gene delivery modalities based on plasmid DNA (pDNA) were extensively evaluated in vitro as potential treatments of inherited diseases [1]. However, their failure to demonstrate potency at a clinical level due to their inability to bypass hurdles posed by the nuclear membrane of non-dividing cells and immunogenic responses of cytosine-phosphate-guanine (CpG) motifs contained by unmethylated DNA has aroused interest in using mRNA instead of pDNA [2,3].

Since an early study conducted by Malone and co-workers, the use of mRNA in gene therapy was limited by the belief that mRNA is too unstable when transfected into cells [4,5]. Recently, researchers have disproved that notion by successfully demonstrating the feasibility of mRNA-based modalities in several therapeutic applications, including tumor vaccination [6] and cancer immunotherapy. The feasibility and non-toxicity of naked mRNA and mRNA complexed with protamine were demonstrated in human patients via intradermal injections, resulting in promising immunological responses [7,8].

The recent interest in mRNA-based systems is due to the pharmaceutical safety advantages demonstrated over their pDNA-based counterparts. These include, first, the ease of mRNA to be formulated into an efficient therapeutic agent since it does not require the incorporation of promoters and terminators such as pDNA. It lacks immunogenic CpG motifs, which are present in pDNA, and does not need to traverse the nuclear membrane to elicit expression, as it is delivered into the cytoplasm, resulting in early and improved transfection activities [9]. Lastly, mRNA can transfect non-dividing cells, and its inability

**Citation:** Mbatha, L.S.; Maiyo, F.; Daniels, A.; Singh, M. Dendrimer-Coated Gold Nanoparticles for Efficient Folate-Targeted mRNA Delivery In Vitro. *Pharmaceutics* **2021**, *13*, 900. https://doi.org/10.3390/ pharmaceutics13060900

Academic Editors: Francisco José Ostos, José Antonio Lebrón and Pilar López-Cornejo

Received: 16 April 2021 Accepted: 20 May 2021 Published: 17 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

to integrate into the host genome eliminates insertional mutagenesis, making it safer to deliver than pDNA [10]. However, few studies have explored mRNA transfection over the years, and consequently, knowledge regarding mRNA transfection is limited, as the application of mRNA is still restricted by the need for improved delivery systems [11]. Thus far, the general consensus is that the use of cationic non-viral mRNA-based delivery systems, particularly cationic polymers (e.g., dendrimers), results in significantly improved transgene activity compared to that elicited by pDNA-based delivery systems [5], with some researchers recently using lipid nanoparticles (LNPs) for mRNA delivery [12]. Dendrimers, particularly PAMAM, are shown to elicit high transfection activities in vitro due to their hyperbranched, well-defined, three-dimensional (3D) structure with multiple surface functionalities, extreme buffering capacity, and ability to be protonated at physiological pH for efficient nucleic acid binding [13–16]. However, their high cytotoxic profiles induced by an excess of the surface amines (tertiary, 3◦ internal and peripheral primary, 1◦) amines, especially at higher generations (>5), have tarnished their use in drug/gene delivery in the past [17]. Many reports, however, have shown that modifying these surface amines via pegylation, methylation, alkylation, acetylation, and conjugation with vitamins or amino acids significantly reduced this cytotoxicity [18–20].

Recently, several studies have exploited the remarkable properties of dendrimers as stabilizers of metal nanoparticles (NPs) [14–16,21–23]. This strategy combines the unique properties of metal NPs with those of cationic dendrimers to produce safe and highly efficient non-viral gene delivery systems. Gold nanoparticles are among the most commonly used metallic NPs to date due to their facile synthesis, biocompatibility, favorable surface-to-volume ratio, ability to be modified, and low cytotoxicity [24,25]. To the best of our knowledge, the transfection of mRNA using PAMAM dendrimer-grafted gold nanoparticles (AuNPs) was never explored. For that reason, this proof of principle study focused on designing FA-modified PAMAM-grafted AuNPs and PAMAM-grafted AuNPs and evaluating their cytotoxicity profiles and capacity to efficiently deliver F*Luc*-mRNA in vitro. FA-modified PAMAM nano-conjugates and PAMAM nano-conjugates were also evaluated for comparison purposes.

#### **2. Materials and Methods**

#### *2.1. Materials*

Starburst PAMAM dendrimer, generation five (PAMAM G5D), (*Mw* of 28,826, 128 surface amino groups), bicinchoninic acid (BCA), folic acid, 1-(3-dimethylaminopropyl)-3-ethyl carbodiimide (EDC), dimethylformamide (DMF), sodium dodecyl sulfate (SDS), dialysis tubing (*MWCO*, 12,000 Daltons), and ribonuclease A (RNase A) were supplied by Sigma-Aldrich (St. Louis, MO, USA). Ultra-pure DNA-grade agarose was acquired from Bio-Rad Laboratories (Richmond, VA, USA). Tris (hydroxymethyl)-aminomethane hydrochloride (Tris-HCl), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), 2- [4-(2-hydroxyethyl)-1-piperazinyl] ethane sulphonic acid (HEPES), Dimethyl sulphoxide (DMSO), ethidium bromide (ETB), and gold (III) chloride trihydrate 99% (HAuCl4) were purchased from Merck (Darmstadt, Germany). F*Luc*-mRNA (5-methylcytidine and pseudouridine modified) was purchased from TriLink BioTechnologies, Inc (San Diego, CA, USA). Minimum essential medium (EMEM) containing Earle's salts and L-glutamine, penicillin (500 units/mL)/streptomycin (5000 μg/mL), and trypsin-versene were purchased from Lonza-BioWhittaker (Walkersville, MD, USA). Fetal bovine serum (FBS) was purchased from Highveld Biological (Lyndhurst, South Africa). Human embryonic kidney (HEK293), hepatocellular carcinoma (HepG2), breast adenocarcinoma (MCF-7), cervical adenocarcinoma cells (KB), and colorectal adenocarcinoma (Caco-2) cells were originally obtained from the American Type Culture Collection (Manassas, VA, USA).

#### *2.2. Synthesis of Gold Nanoparticles (AuNPs)*

An adaptation of the Turkevich method was followed to synthesize the AuNPs [26]. Briefly, HAuCl4 (0.03 M, 0.1 mL) was dissolved in 25 mL of 18 MOhm water, stirred

vigorously, and heated for 15 min until boiling. This was followed by the slow addition of 1 mL of 1% trisodium citrate (Na3C6H5O7) with stirring until a red color change was produced. The mixture was then removed from the heat and stirred until it cooled to room temperature.

#### *2.3. Modification of PAMAM G5D with Folic Acid (FA)*

PAMAM G5D (dried under nitrogen) was dissolved in 18 MOhm water and conjugated to folic acid (FA) via carbodiimide chemistry as described previously by the authors [15,16]. FA (2.8 μmol in 3 mL of DMF) was reacted with 38.2 μmol EDC for 45 min with constant stirring under nitrogen. The activated FA was then added slowly with stirring into the dendrimer (3 μmol, 100 μL) solution, and the pH maintained at 9.5. The solution was stirred for 3 days under nitrogen, followed by the removal of unreacted by-products by dialysis (*MWCO* 12 000 Da) against 18 MOhm water for 24 h.

#### *2.4. Formulation of Dendrimer-Coated AuNPs (Au:G5D NPs, and Folic-Acid-Targeted, Dendrimer-Coated AuNPs (Au:G5D:FA NPs)*

The G5D and previously synthesized G5D:FA (Section 2.3) were conjugated to the citrate-reduced AuNP solution as previously described by the authors [15,16] to produce Au:G5D and Au:G5D:FA NPs in a 25:1 gold/dendrimer molar ratio. NPs were dialyzed as in Section 2.3.

#### *2.5. Ultra-Violet (UV) and Proton Nuclear Magnetic Resonance (1H NMR) Spectroscopy*

Successful functionalization of the G5D and AuNPs was monitored by UV-vis spectroscopy (UV-1650PC, Shimadzu, Japan) using a wavelength range of 200–800 nm. Further confirmation of NP synthesis was achieved using 1H NMR spectroscopy (Bruker DRX 400) with deuterated (D2O) water as a solvent.

#### *2.6. Transmission Electron Microscopy (TEM) and Nanoparticle Tracking Analysis (NTA)*

The ultrastructural morphology of the NPs and their mRNA nanocomplexes at optimum binding ratios (*w*/*w*) were determined by cryo-TEM, using a Jeol JEM-1010 transmission electron microscope containing a Soft Imaging System (SIS) fitted with a MegaView III digital camera with iTEM UIP software, operating at an acceleration voltage of 200 kV (Tokyo, Japan). The z-average hydrodynamic diameters and zeta (ζ) potentials were determined by nanoparticle tracking analysis (NTA, NanoSight NS500; Malvern Instruments, Worcestershire, UK) at 25 ◦C. NPs (1 mL) were diluted 1:100 in 18 MOhm and sonicated before analysis. Although the characterization of these NPs was reported previously by the authors [15,16], the mRNA-based nanocomplexes are reported here for the first time.

#### *2.7. Nanocomplex Preparation and Binding Studies*

Nanocomplexes for mRNA binding, cell viability, and transfection studies contained a constant amount of F*Luc*-mRNA (0.05 μg) together with increasing amounts of G5D, Au:G5D, G5D:FA, and Au:G5D:FA NPs. Nanocomplexes were briefly mixed and incubated at room temperature for 60 min.

#### 2.7.1. Band Shift Assay

Band shift assays [27] were utilized to determine the binding of mRNA to the NPs. Nanocomplexes prepared as in Section 2.7 were subjected to electrophoresis on 1% (*w*/*v*) agarose gels containing ethidium bromide (ETB) (1 μg/mL) in a Bio-Rad mini-sub electrophoresis apparatus containing 1× electrophoresis buffer (36 mM Tris-HCl, 30 mM, sodium phosphate (NaH2PO4), 10 mM ethylenediamine tetra-acetic acid (EDTA), pH 7.5), for 45 min at 50 Volts. Gels were viewed and images captured using a Vacutec Syngene G: Box BioImaging system (Syngene, Cambridge, UK).

#### 2.7.2. Ethidium Bromide Displacement Assay

The compaction of the nanocomplexes was assessed using a dye displacement assay [27]. ETB solution (24 μL, 100 μg/mL) and HBS (100 μL) were initially added to a 96-well FluorTrac flat-bottom black plate, and fluorescence read in a Glomax®-Multi + detection system (Promega, Sunnyvale, CA, USA) at an excitation wavelength of 520 nm and an emission wavelength of 600 nm. This measurement was set as 0% relative fluorescence (RF). The 100% RF was obtained after the addition of 0.05 μg F*Luc*-mRNA. Thereafter, 1 μL aliquots of the respective NPs were added, and fluorescence was measured until a plateau in fluorescence was achieved.

#### 2.7.3. RNase A Protection Assay

The stability of the nanocomplexes and the protection afforded to the mRNA in the presence of degrading enzymes were evaluated by an RNase protection assay adapted from [27]. NP:mRNA nanocomplexes prepared at the sub-optimum, optimum, and supraoptimum ratios (obtained from Section 2.7.1) were exposed to 10% RNase A for 2 h at 37 ◦C. This was followed by the addition of 10 mM EDTA to halt the reaction and 0.5% SDS to release the nucleic acid from the nanocomplex. Samples were subsequently incubated at 55 ◦C for 20 min, followed by electrophoresis as described previously ( Section 2.7.1).

#### *2.8. Cell Culture-Based Assays*

All cells were maintained and propagated at 37 ◦C and 5% CO2 in 25 cm<sup>2</sup> flasks containing sterile EMEM, FBS (10%, *v*/*v*), penicillin G (100 U/mL), and streptomycin sulfate (100 μg/mL). The cells were split upon confluency into desired ratios when necessary and the medium changed routinely.

#### 2.8.1. MTT Cell Viability Assay

The MTT assay was used to determine the viability of the cells after treatment with the respective nanocomplexes as described previously [28,29]. All cells were seeded into 48-well plates at densities of 2.5 × <sup>10</sup><sup>5</sup> cells/well, and incubated for 24 h at 37 ◦C. Thereafter, nanocomplexes at selected ratios were added in triplicate, and cells were incubated for 48 h at 37 ◦C. Cells containing no nanocomplexes were used as the positive control (100% cell viability). Following the 48 h incubation, a fresh medium containing the 10% MTT reagent (5 mg/mL in PBS) was added, followed by a 4 h incubation at 37 ◦C. The medium MTT mixture was then aspirated, cells washed with PBS (2 × 0.3 mL), and 0.3 mL of DMSO was added to solubilize the resulting formazan crystals. Absorbance was then measured at 570 nm in a Mindray MR-96A microplate reader (Vacutec, Hamburg, Germany) using DMSO as the blank. The percentage cell viability was calculated against the positive control (100%).

#### 2.8.2. Apoptosis Assay

To determine if apoptosis was instrumental in the cell death recorded, a fluorescent dualstain apoptosis assay was conducted as previously described [30]. Cells (2.9 × 105 cells/ well) were plated into 12-well plates and incubated for 24 h at 37 ◦C. Following the addition of nanocomplexes at optimum binding ratios, the cells were incubated for 48 h at 37 ◦C. Thereafter, cells were washed with PBS, and 10 μL of AO/ETB (AO/ETB, 1:1 *v*/*v*, 100 μg/mL) was added. Cells were viewed for structural and morphological changes under an Olympus fluorescent microscope (×200 magnification), fitted with a CC12 fluorescent camera (Olympus Co., Tokyo, Japan). Apoptosis was quantified by calculating the apoptotic index (AI) as below:

Apoptotic Index = Number of apoptotic cells/Total number of cells

#### 2.8.3. Transfection and Competition Assays

The transfection and competition assays were conducted as previously described [15,16,28,29]. Cells with densities of 2.5 × 105 cells/well were seeded into 48-well plates and incubated for 24 h at 37 ◦C. The nanocomplexes (ratios as used for the MTT assay) were then added, and the cells were incubated for 48 h at 37 ◦C. Thereafter, the cells were washed with PBS (2 × 0.5 mL) and lysed with 80 μL/well cell lysis buffer (Promega) for 15 min with shaking at 30 rpm in a Scientific STR 6 platform rocker (Stuart Scientific, Staffordshire, UK). Cell suspensions were then centrifuged at 12,000× *g* for 1 min. The cell-free extract (20 μL) was added to 100 μL luciferase assay reagent, mixed, and luminescence recorded in relative light units (RLU) in a Glomax®-Multi+Detection System (Promega Biosystem, Sunnyvale, CA, USA). The standard BCA assay was used to determine the protein concentrations of the cell-free extracts as described previously [29,31]. The luminescence recorded was normalized against the protein concentration, and luciferase activity was expressed as RLU/mg protein.

For the competition assay, cells were seeded and treated as for the normal transfection, but FA (250 μg) was incubated with folate receptor-positive cells (MCF-7 and KB cells) for 20 min at 37 ◦C before the addition of the targeted nanocomplexes. Luciferase activity was then determined as described above.

#### *2.9. Statistical Analysis*

Cell viability and transfection studies were performed in triplicate and results expressed as means ± standard deviation (SD). The experimental data was analyzed by a two-way ANOVA and *t*-test using GraphPad Prism 6.0 and statistically significant values are indicated by \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001, and \*\*\*\* *p* < 0.0001, # *p* > 0.05.

#### **3. Results**

#### *3.1. UV-Visible and 1H NMR Spectroscopy*

The attachment of G5D and FA on the AuNPs was first confirmed by UV-vis spectroscopy (Figure 1).

**Figure 1.** (**A**) UV spectra of (**a**) AuNPs, (**b**) Au:G5D NPs, (**c**) Au:G5D:FA NPs; and (**B**) UV-spectra of (**d**) G5D, (**e**) G5D:FA NPs, and (**f**) FA.

The absorption band at 536 nm confirmed the formation of AuNPs, since the known absorption band of AuNPs range between 520 and 550 nm [32]. The band shift from 536 nm to 566 nm confirmed the attachment of G5D on the surface of the AuNPs [33]. Furthermore, the covalent attachment of the FA onto the surface of NPs is known by its absorption maxima at 280 nm with a saddle point at 360 nm [34,35] (Figure 1A), corresponding to the absorption peak of Au:G5D:FA observed at 287 nm. Figure 1B shows the λ max for G5D and FA which caused the changes in the surface plasmon resonance of the AuNPs upon functionalization. The UV-vis absorbances were further utilized to estimate the amount of bound G5D and FA, which were 53.8% and 60.6%, respectively.

The formation of Au:G5D and Au:G5D:FA NPs was also verified by 1H NMR spectroscopy (Figure 2). Significant differences in the chemical shift of protons related to Au:G5D (D), Au:G5D:FA (B), G5D:FA (A) were observed when compared to G5D(C). The 1H NMR of the G5D shows six broad peaks (Figure 2C, peaks 1–6) as indicated by a chemical shift ranging from 2.25 to 3.34 ppm, representing the protons of the amino (NH2) and methylene groups (CH2). These findings correlated with those reported [36,37]. Moreover, the three peaks between 6.50 and 8.63 ppm observed in Figure 2A,B indicate the attachment of FA protons (H-Ar (7 and 13), NH (18)). The formation of Au:G5D nanocomplexes resulted in the downfield shift of protons 4, 5, and 6 of G5D, which indicated the interaction of the surface of the AuNPs with the internal amines of the dendrimer. These findings correlate to that in literature [38].

**Figure 2.** The 1H NMR spectra of PAMAM dendrimer (G5D) and folic acid-functionalized gold nanoparticles in D2O. (**A**) G5D:FA, (**B**) Au:G5D:FA, (**C**) G5D, (**D**) Au:G5D.

#### *3.2. Morphology, Size, and Zeta Potential of Nanoparticles and Nanocomplexes*

The NPs appeared spherical (Figure 3A,B,D) with a uniform distribution and hydrodynamic diameters from NTA ranging from 65 nm to 128 nm (Table 1). Nanocomplexes prepared at optimum binding ratios (Figure 3C,E), presented as clusters of smaller particles with hydrodynamic diameters ranging from 101 nm to 265 nm (Table 1). There was no significant size difference (# *p* > 0.05) between the Au:G5D/Au:G5D:FA and G5D/G5D:FA nanocomplexes (Table 1).

Overall, ζ potentials ranged from 20.9 mV to 87.2 mV for the NPs and from −21.0 mV to −65 mV for the nanocomplexes, indicating good colloidal stability (Table 1). Au:G5D and Au:G5D:FA nanocomplexes had the highest ζ potentials of −37.3 mV and −65.7 mV, respectively. The polydispersity indices (PDI) revealed that all the NPs and nanocomplexes are highly monodisperse and uniform in size with PDI values below 0.2 (Table 1), suggesting that these NPs and nanocomplexes have a lower tendency to agglomerate [39].

**Figure 3.** TEM micrograph of (**A**) AuNPs, (**B**) Au:G5D, (**C**) Au:G5D-mRNA nanocomplex, (**D**) Au:G5D:FA, and (**E**) Au:G5D:FA-mRNA nanocomplex. Nanocomplexes were prepared at optimum binding ratios of 3:1 (*w*/*w*) for Au:G5D-mRNA and 4:1 (*w*/*w*) for Au:G5D:FA-mRNA, respectively.

**Table 1.** Hydrodynamic size, ζ potential measurements, and polydispersity indices of nanoparticles and nanocomplexes. Data are presented as mean diameter ± standard deviation (SD) (*n* = 3).

