**About the Special Issue Editors**

**Juan Jose´ Torrado Duran´** , Pharmaceutics and Food Technology, School of Pharmacy, Complutense University of Madrid, Madrid, Spain. He is author of about 100 articles related to Pharmaceutical Technology and Drug Delivery Systems. Interests: conventional pharmaceutical dosage forms (tablets, capsules, semisolid and liquid formulations); new controlled release systems (pellets, nanoparticles, microcapsules, microspheres and liposomes) including production and quality control.

**Dolores R. Serrano**, Pharmaceutics and Food Technology, School of Pharmacy, Complutense University of Madrid, Madrid, Spain. She is author of about 60 articles related to Pharmaceutical Technology and Drug Delivery Systems. Interests: development and optimization of novel drug delivery systems of poorly water soluble drugs with the aim of increasing their bioavailability and their effectiveness with special focus on antifungals and antiparasitic drugs, as well as 3D printing of medicines.

**Javier Capilla**, Unitat de Microbiologia, Facultat de Medicina i Ciences ` de la Salut, Universitat Rovira i Virgili and Institut d'investigatio´ Sanitaria Pere Virgili (IISPV), Reus, Spain. He is author of about 70 articles related to fungal diseases and their treatments. Interests: the study and development of antifungal therapies against opportunistic fungal infections as well as virulence factors of fungi causing human infections.

### *Editorial* **Antifungal and Antiparasitic Drug Delivery**

#### **Juan José Torrado 1,\*, Dolores R. Serrano 1,\* and Javier Capilla <sup>2</sup>**


Received: 25 March 2020; Accepted: 1 April 2020; Published: 4 April 2020

**Abstract:** Fungal and parasitic diseases affect more than a billion people across the globe, one-sixth of the world's population, mostly located in developing countries. The lack of effective and safer treatments combined with a deficient diagnosis lead to serious chronic illness or even death. There is a mismatch between the rate of drug resistance and the development of new medicines. Formulation of antifungal and antiparasitic drugs adapted to different administration routes is challenging, bearing in mind their poor water solubility, which limits their bioavailability and efficacy. Hence, there is an unmet clinical need to develop vaccines and novel formulations and drug delivery strategies that can improve the bioavailability and therapeutic effect by enhancing their dissolution, increasing their chemical potency, stabilising the drug and targeting high concentration of drug to the infection sites. This Editorial regards the ten research contributions presented in the Special Issue "Antifungal and Antiparasitic Drug Delivery".

**Keywords:** liposomes; transferosomes; nanoparticles; emulsions; candidiasis; aspergillosis; azoles; amphotericin B; combined therapy; quality by design; leishmaniasis; malaria; trypanosomiasis

In order to obtain new antifungal and antiparasitic drug delivery systems, scientists of different disciplines have to collaborate in coordinated research teams. This volume includes ten papers, five of them about antifungal formulations and other five related to antiparasitic formulations. Amongst fungal infections, candidiasis has received special attention due to its world prevalence, as well as leishmaniasis as a parasitic disease. Interestingly, an old molecule, amphotericin B, is the active component studied in six out of the ten papers. Other active components also studied are butenafine, praziquantel, fluconazole, meglumine antimoniate and the enolase-base vaccine.

The administration route plays a key role in the development of novel antifungal and antiparasitic formulations. In this issue, a special focus on oral, parenteral and topical formulations is highlighted. Dosage forms are obviously related to the administration route. For example, suspensions, solutions and tablets are developed for oral administration while semisolid gels and wound patches are fabricated for topical application. Different parenteral administration routes are covered, such as subcutaneous, intravenous and at the bone cavity.

The originality of the new formulations proposed are usually based on the selection of already approved excipients along with the active components, such as Montanide™ Petgel A as vaccine adjuvant [1], poly(vinyl alcohol) [2], Poloxamer 407™ [3], a combination of Capryol 90™, Peceol™ and Labrasol™ [4], dextran and maltodextrin [5], a combination of modified chitosan nanoparticles with a standardized extract of cultured *Lentinula edodes* mycelia (AHCC™) [6], ground calcium carbonate [7], poly (d,l-lactide-*co*-glycolide) 50:50 [8] and Sepigel 305™ [9]. Only in one paper [10], authors have synthesized a new material based on copolymers of poly(ethyleneglycol) and poly(ε-caprolactone) conjugated with retinol as drug vehicle. Moreover, the characterization of the new formulations is described in detail in the ten papers with a special focus on toxicity and efficacy studies required in

order to bring to the market these formulations. This Special Issue is an update on novel drug delivery strategies of antifungal and antiparasitic drugs to treat both topical and systemic infections. A brief description of the ten research papers included in the issue is described below.

Tellez-Martínez et al. propose a new vaccine based on recombinant enolase-Montanide™ PetGel A against virulent fungus *Sporothrix schenckii*. The incorporation of Montanide™ PetGel A as adjuvant was able to induce specific Th1 response and protective immunity against the fungal in Balb/c mice [1]. Interestingly, the virulence of *S. schenckii* was enhanced by toluene exposure. Toluene is an example of environmental contaminant. In this work, authors proved that the combination of some environmental contaminants can enhance the virulence of pathogen agents. Effective vaccines are an important pharmacological tool to protect us against this type of severe infections.

The work of Alexandrino-Junior et al. is a clear example of the potential pharmacological effect of new formulations of old drugs. Amphotericin B was formulated on a poly(vinyl-alcohol) hydrogel as a new topical formulation for the treatment of cutaneous leishmaniasis [2]. Although topical treatment of cutaneous diseases seems to be an ideal approach, conventional topical formulations of amphotericin possess low activity on cutaneous leishmaniasis due to permeability issues. Nevertheless, these new hydrogels developed in this work have exhibited, in vitro, a promising antiparasitic activity against *Leishmania* parasites and also against some fungal infections.

Sosa et al. performed an interesting study whose aim was the development and evaluation of a topical formulation of amphotericin B for the treatment of dermal and vaginal candidiasis [3]. Poloxamer 407™ was selected as excipient based on its thermoreversible properties. This excipient is liquid at low temperatures (4–5 ◦C) but turns into a semisolid gel above 32 ◦C. A thermoreversible gel containing amphotericin was developed and evaluated. Ex vivo permeation studies on human skin and pig vaginal mucosa showed that no permeation was observed. In vitro, antifungal activity studies against *Candida* spp showed that this formulation was more efficient than free amphotericin. Moreover, the amount of amphotericin remaining on the skin and vaginal mucosa was high enough to obtain antifungal activity.

Bezerra-Sousa et al. described the preparation of an oral nanomedicine of butenafine for visceral leishmaniasis [4]. Butenafine is currently used as a topical antifungal drug with low oral bioavailability. In this work, the low solubility of butenafine was increased by preparation of optimized self-nanoemulsifying drug delivery systems which have proved in vitro to be effective against promastigotes and amastigotes of *Leishmania infantum*. Moreover, these promising systems were then transformed by spray-drying into a solid dosage form of butenafine. Development of solid oral nanomedicines enables the non-invasive and safe drug administration, being a cost-effective and readily scalable repurposed medicine for visceral leishmaniasis.

Serrano et al.'s work focused on the design of fast-dissolving orodispersible films of amphotericin B for oropharyngeal candidiasis [5]. Amphotericin B is a low water soluble antifungal drug. A quality-by-design study was applied to select the best combination of GRAS excipients. A fast disintegration film with quick amphotericin release in artificial saliva and high in vitro efficacy against several *Candida* spp. was obtained.

Pérez-Cantero et al. carry out an interesting study related to the increased prophylactic efficacy of parenteral and oral amphotericin B treatments against aspergillosis when combined with standardized extract of cultured *Lentinula edodes* mycelia (AHCC™) [6]. Amphotericin was encapsulated in modified chitosan-nanoparticles suitable for oral administration. The addition of AHCC™ significantly improved the efficacy of both oral and parenteral treatments in a mice model of experimental aspergillosis. Moreover, the weight loss of treated animals was lower when AHCC™ was administered, suggesting a protective effect of the extract. In relation to the control group, treated animals showed stimulation of the Th1 immune response, which can explain the improvement of its efficacy.

The work of Borrego-Sánchez et al. focused on the increase of solubility and dissolution rate of praziquantel [7]. Praziquantel is also a poorly water-soluble antiparasitic drug, highly effective against schistosomiasis. Ground calcium carbonate is a cheap, hydrophilic porous carrier that was combined with praziquantel by using two easily scalable processes: physical mixture or solid dispersions. An in vitro dissolution test proved that solid dispersions increase drug solubility and dissolution rate. In vitro cytotoxicity studies against HTC116 cells showed that the praziquantel solid dispersions are safe.

Hsu et al. studied how amphotericin B and fluconazole can be incorporated into resorbable beads [8]. These beads are made of biodegradable Poly(d,l-lactide-*co*-glycolide) (50:50) and they were fabricated using a compression-molding method. The beads were evaluated, showing that the in vitro release of the fluconazole beads was better than the one obtained from amphotericin B beads. The in vivo assay in rabbits showed a sustained antifungal activity of fuconazole for more than 49 days, and thus, was suitable for the treatment of bone infections.

Berenguer et al. developed and characterized a semi-solid gel dosage form of meglumine antimoniate for the topical treatment of cutaneous leishmaniasis [9]. The gel is easy to prepare and its main excipient is Sepigel 305™. It was stable for over 6 months. The pH and rheological characteristics were suitable for topical application. Ex vivo permeation studies in human skin show low permeation and high retention in the skin layer, so low systemic toxicity and enhanced local activity can be expected from this formulation. Low toxicity and good tolerance were observed in keratinocyte cell lines and human volunteers, respectively. In vitro anti-leishmanial activity of the gel showed a reduction of the IC50 compared to the reference solution. This new formulation could be a promising alternative for topical treatment of cutaneous leishmaniasis.

Rodriguez et al. described the development of amphotericin B micellar formulations based on copolymers of poly(ethyleneglycol) and poly(ε-caprolactone) conjugated with retinol [10]. Biodegradable and biocompatible polymers were initially synthesized and then conjugated with retinol. These micellar formulations were less haemolytic than Fungizone™. Furthermore, the antifungal activity of amphotericin incorporated in these new formulations showed a reduction of the MIC of up to eight-fold compared with reference Fungizone™. The low toxicity and high in vitro antifungal activity of these formulations make them good candidates for future in vivo experiments.

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **A Recombinant Enolase-Montanide™ PetGel A Vaccine Promotes a Protective Th1 Immune Response against a Highly Virulent** *Sporothrix schenckii* **by Toluene Exposure**

#### **Damiana Téllez-Martínez, Deivys Leandro Portuondo, Maria Luiza Loesch, Alexander Batista-Duharte \* and Iracilda Zeppone Carlos \***

Department of Clinical Analysis, School of Pharmaceutical Sciences, São Paulo State University (UNESP), Araraquara 14800-903, SP, Brazil; damianatellezm@gmail.com (D.T.-M.); deivysleandro@gmail.com (D.L.P.); ma\_luizaloesch@hotmail.com (M.L.L.)

**\*** Correspondence: batistaduhartea@gmail.com or batistaduhartea@fcfar.unesp.br (A.B.-D.); carlosiz@fcfar.unesp.br (I.Z.C.)

Received: 25 February 2019; Accepted: 21 March 2019; Published: 25 March 2019

**Abstract:** The effect of vaccination in fungal strains that suffered changes in their virulence by exposure to environmental contaminants is largely known. Growing reports of resistance to antifungal drugs and the emergence of new highly virulent strains, possibly acquired in the environment, prompt the design of new vaccines able to prevent and combat emerging mycotic diseases. In this study, we evaluated the protective capacity of an enolase-based vaccine and Montanide PetGel A (PGA) as an adjuvant against *S. schenckii* with increased virulence by exposure to toluene. The adjuvanted vaccine induced a strong specific Th1 response and protective immunity against a challenge with either wildtype or toluene-adapted *S. schenckii* in Balb/c mice. This study highlights the role of the adjuvant PGA driving the quality of the anti-sporothrix immunity and the key component in the vaccine efficacy.

**Keywords:** vaccine; adjuvants; *Sporothrix schenckii*; toluene; virulence; enolase; Montanide PetGel A

#### **1. Introduction**

Sporotrichosis is an emergent subcutaneous mycosis in tropical and subtropical regions, caused by several pathogenic species of the genus *Sporothrix*; that include *Sporothrix brasiliensis*, *S. schenckii sensu stricto*, *Sporothrix globosa*, and *Sporothrix luriei* [1]. Classically, infection is acquired after traumatic inoculation of contaminated soil, plants, and organic matter into skin or mucosa or, more rarely, by inhalation of conidia. Over the last years, cat–human zoonotic transmission of sporotrichosis caused by *S. brasiliensis* has become a health problem in Brazil [2]. The disease can manifest as fixed cutaneous and regional lymphocutaneous forms in immunocompetent individuals, and disseminated forms, mainly reported in immunocompromised patients [3].

Ecological determinants of the genus *Sporothrix* remain poorly understood [4,5]. However, experimental evidence suggests that environmental contaminants can modify the fungal virulence by reducing the host immunity [6] or modifying the fungal biology [4,7]. Previous studies showed that fungal exposure to toluene, a common soil contaminant that shares the same environmental niche of *S. schenckii*, is able to increase the *S. schenckii* virulence [7]. Ongoing studies are evaluating the role of chemical contamination and other environmental factors in sporotrichosis outbreaks.

Conventional treatment of sporotrichosis requires long periods of antifungal drug administration often accompanied by adverse effects and fungal resistance, principally during the treatment of disseminated sporotrichosis [8]. These problems have stimulated the search for new strategies for sporotrichosis management, including anti-sporothrix vaccination that has been proposed as a feasible way for both therapeutic and prophylactic purposes [9,10]. However, the development of antifungal vaccines has not been as successful as antiviral and antibacterial vaccines due to, among other things, a general under-appreciation for the impact of fungal diseases and the high cost of preclinical and clinical studies [11]. Another challenge has been the use of immunological adjuvants with an adequate safety and efficacy profile [12]. Aluminium-derived adjuvants have been used in human and veterinary vaccines for more than eight decades in licensed vaccines [13,14]. However, they have not been successful in preventing intracellular infection due to a weak capacity to induce cell-mediated immunity when used along with small immunogenic antigens [13]. Moreover, there are reports of tumors in the inoculation site in felines vaccinated with alum-based vaccines and a possible causal association between chronic inflammation induced by alum and these tumors has been suggested [15].

Current advances in the understanding of antifungal immune response support rational use of more effective adjuvants, such as pattern recognition receptors (PRR) agonists, inhibitors of regulatory T cells, and others [12,16,17], to achieve effective immune responses against specific fungi.

For several decades, biodegradable natural and synthetic polymers have been used for antigen delivery and as immunological adjuvants. Due to their biocompatibility, biodegradability, easy production and low toxicity polymers are attractive candidates for substituting conventional adjuvants [18]. MontanideTM PetGel A (PGA), is a polymer-based adjuvant composed of highly stable dispersion of microspherical particles of sodium polyacrylate in water (Figure 1A, Table 1). This polymeric technology has already been used in several vaccine models, including pet vaccines, with a promising safety and efficacy profile [19].

**Figure 1.** External aspect of (**A**) MontanideTM PetGel (PGA) and (**B**) PGA+rSsEno vaccine formulation. (**C**) SDS-PAGE showing the expression and purification of *rSsEno.* C1. Molecular pattern, C2. rSsEno band stained with Coomassie blue with the expected 47 kDa molecular mass (in duplicate). (**D**) Immunoblotting analysis of rSsEno with a pooled anti-*rSsEno* serum from Balb/c mice immunized with Freund's adjuvant/rSsEno (Either 1/1600 or 1/6400 dilutions of serum were used). A pooled serum from non-immunized mice (dilution 1/50) was used as negative control (C−) to identify non-specific binding.


#### **Table 1.** General properties of Montanide™ GEL adjuvants.

Recently, our group evaluated a cell-wall protein extracted from *S. schenckii* (SsCWPs) in an experimental vaccine formulated with Aluminum hydroxide (AH) gel [20]. Immunized mice developed a specific immune response characterized by a balanced Th1/Th2/Th17 response and the production of protective antibody response. In a more recent study, the AH-based vaccine was compared with other experimental vaccine candidate containing SsCWPs and PGA. Both formulations induced a protective immune response in mice. However, AH stimulated the development of granulomas in the inoculation site while PGA-based vaccine exhibited a Th1 protective response and better local tolerance than AH-based vaccine in vaccinated mice [21]. In other recent study, we evaluated the immunogenic and protective effect of enolase, one of the main antigens that were found in SsCWPs, against *S. brasilensis*, the most virulent species of the genus *Sporothrix* [22]. The recombinant enolase of *S. schenckii* (SsEno) was formulated with PGA and after three subcutaneous administrations in mice, a significant specific immune response was observed. Furthermore, a reduction in mortality (over 90%) was observed after 45 days of an intravenous challenge with viable yeast, compared with non-vaccinated mice.

Until now, there are no studies on the effects of vaccination on fungal strains whose virulence changed due to exposure to environmental contaminants. In the current context of growing reports of fungal resistance to antifungal drugs and the emergence of new highly virulent strains, possibly acquired in the environment, these assessments can provide important information on the ability of new vaccines to prevent and combat emerging mycotic diseases.

In this study, we evaluated the protective capacity of an enolase-based vaccine formulated with PGA as an adjuvant, against *S. schenckii* with increased virulence by experimental exposure to toluene [7].

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

#### *2.1. Animals*

Male Balb/c mice (five to seven weeks old) were purchased from "Centro Multidisciplinar para Investigação Biológica na Área da Ciência de Animais de Laboratório" (CEMIB), Universidade de Campinas (UNICAMP), São Paulo, Brasil. Mice were housed in microisolator cages in a controlled ambient and receiving water and food ad libitum. The study was carried out in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The experiments were approved by the Ethics Committee for Animal Use in Research of Araraquara's School of Pharmaceutical Sciences from UNESP (Protocol CEUA/FCF/CAr: 19/2018).

#### *2.2. Microorganisms and Preparations*

The *S. schenckii sensu stricto* strain ATCC 16345 (here named as *S. schenckii*) used in this work was kindly provided by the Oswaldo Cruz Foundation (Rio de Janeiro, Brazil). The mycelial phase was maintained at room temperature in Mycosel (BD Biosciences) agar. A piece of a well-defined colony was grown in 100 mL of Sabouraud dextrose broth (SDB) (Difco, Detroit, MI, USA) for four days in a rotary shaker (130 rpm and 30 ◦C). The conidia were separated from the hyphae by filtration with sterile gauze using a Buchner funnel. Them conidia were counted and suspended in phosphate-buffered saline (PBS) at 1 × <sup>10</sup>7/mL.

#### *2.3. S. schenckii Growth in Toluene*

Fungal cultures were performed in 125-mL Erlenmeyer flasks containing 50 mL of Sabouraud dextrose broth (SDB) and sealed with Teflon Mininert valves (SUPELCO, 24 mm, (Merck KGaA Darmstadt, Germany) to prevent evaporation of the solvent. SDB was supplemented with toluene 0.1% (*v*/*v*). An aliquot of 1 × 107 conidia was inoculated and incubated during five days on a rotary shaker (30 ◦C and 130 rpm). Control cultures without toluene were included. Fungal viability was determined at fifth day by counting colony forming units (CFU) on Sabouraud dextrose agar (SDA) plates [7].

#### *2.4. Expression and Purification of Recombinant S. schenckii Enolase (rSsEno)*

The detailed procedures were previously described [22]. Briefly, the gene that encodes *S. schenckii* enolase with molecular mass 47 kDa and 438 amino acids (access code: ERS97971.1, GenBank database) was synthesized by Epoch Life Science Inc. (Missouri, TX, USA). The enolase gen was subcloned into the pET28a plasmid and optimized for production in *Escherichia coli* (pET28a::SsEno). *E. coli* DH5α was used for the propagation of pET28a::SsEno on lysogeny broth (LB) agar medium containing 30 μg/mL of kanamycin. For recombinant protein expression, *E. coli* BL21 cells cotransformed with pET28a::SsEno were grown at 37 ◦C in LB medium with kanamycin until they reached an OD600 in the range of 0.5–0.7. The expression of rSsEno was induced by 0.2 mmol/L of isopropyl β-D-1-thiogalactopyranoside (IPTG) at 30 ◦C for 4 h. The cells were centrifuged (20 min at 8000 rpm), and the pellet was resuspended in buffer A (NaPO4 20 mM, NaCl 500 mM and imidazole 20 mM, pH 7,4) containing 5 U of DNAse (Promega, Madison, WI, USA) and 30 μg/mL lysozyme (Merck KGaA Darmstadt, Germany) for 30 min on ice. The cell homogenate was sonicated, filtrated and then centrifuged at 19,000 rpm for 20 min at 4 ◦C. The supernatant containing rSsEno was filtered (0.45 μm nitrocellulose membrane, Millipore and initially purified by Ni2+-affinity chromatography in buffer A. The rSsEno eluted in buffer B (NaPO4 20 mM, NaCl 500 mM, and imidazole 500 mM, pH 7.4) was subjected to size exclusion chromatography (SEC) with a Superdex 200 pg 16/60 column (GE Healthcare Life Sciences, Chicago, IL, USA) in Tris-HCl 25 mM, NaCl 100 mM and β-mercaptoethanol 2 mM at pH 7.5, and the eluted protein was concentrated using the Amicon® Ultra 15 mL 3k device (Millipore, Burlington, MA, USA) after being dialyzed for 24 h at 4 ◦C against phosphate buffer saline. The rSsEno concentration was measured by the Pierce BCA assay (Thermo Scientific, Waltham, MA, USA), and the efficacy of the expression and purification processes was assessed by 12% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting using anti-rSsEno serum.

#### *2.5. Adjuvants and Vaccine Formulation*

The vaccine formulation was prepared by mixing 100 μg of rSsEno with 5% PGA adjuvant kindly provided by Seppic (Paris, France) (Figure 1B). Other formulations composed by either PGA or rSsEno alone were used as control.

#### *2.6. Immunization Schedule*

Balb/c mice (*n* = 5) received subcutaneous (s.c.) vaccination (on days 0 for priming and 14 for booster) in the back of the neck, with 100 μL of one of the following formulations: PGA+rSsEno, 100 μg rSsEno or PBS alone as a negative control. One week after the booster, mice were euthanized in CO2 chamber and bled by heart puncture to obtain serum, which was aliquoted and stored at −20 ◦C until use.

#### *2.7. Quantification of the rSsEno-Antibody Response by Enzyme-Linked Immunosorbent Assay (ELISA)*

rSsEno IgG antibody titration was conducted as described previously [22]. Briefly, a 96-well ELISA plate (Merck KGaA Darmstadt, Germany) was coated with 5 μg rSsEno/mL in PBS and at 4 ◦C (overnight). The plate was washed with washing buffer (0.1% Tween 20) and then blocked 1 h at room temperature with 5% dried skim milk in washing buffer. Dilutions of the serum samples (1:500 in blocking buffer) were added to each well and incubated at room temperature for 2 h. After washing, peroxidase-conjugated anti-mouse IgG (1/500) (Merck KGaA Darmstadt, Germany) was added and incubated at 37 ◦C for 1 h. After exhaustive washing, tetramethylbenzidine was added to reveal the antigen-antibody reactions (30 min at room temperature). The reaction was stopped by the addition of 50 μL/well 1M H2SO4, and the absorbance was read with an ELISA reader (Multiskan Ascent, Labsystem, Vantaa, Finland) at 450 nm.

#### *2.8. Th1-Th17 Phenotipagem*

Spleens were aseptically removed and splenocytes were extracted. Viable splenocytes were adjusted to 1 × <sup>10</sup><sup>7</sup> cells/mL in complete RPMI-1640 culture medium (Merck KGaA Darmstadt, Germany), which was supplemented with 2 mm L-glutamine, 100 U/mL penicillin, 100 μg/mL penicillin/streptomycin, and 10% fetal calf serum (RPMI complete). For study of Th1 and Th17 lymphocytes subpopulations, the following anti-mouse mAb were used: anti-CD16/CD32, anti-CD3-FITC, anti-CD4-APC, anti-IL-17-PE, anti-IFN-G-Percp, and respective isotype controls (all purchased from BD Biosciences, (Franklin Lakes, NJ, USA). Splenocytes were assessed for the frequency of Th1(IFN-G+), Th17 (IL-17+). Briefly, viable splenocytes were stained for the extracellular markers, then fixed and permeabilized using eBiosciences' intracellular fixation (Thermo Scientific, Waltham, MA, USA) and permeabilization buffer set, and then the intracellular IFN-G and IL-17A were stained with a fluorescent respective marker. Intracellular cytokines were detected after in vitro stimulation with 10 μg/mL of rSsEno and Brefeldin A for intracellular retention of the induced cytokine. Events were acquired using a BD Accuri C6 flow cytometer (BD Biosciences) and analyzed with the flow cytometer's proprietary software.

#### *2.9. IFN-*G*, IL-4, and IL-17 Measurement in Supernatant of Splenocytes Culture*

Splenocytes from immunized and non-immunized mice were cultured as previously described and stimulated with 10 μg/mL of rSsEno for 24 h. The levels of IFN-G, IL-4, and IL-17 after rSsEno stimulation were measured in the supernatant of splenocytes culture by Cytometric Bead Array (CBA) (BD Biosciences) according to the manufacturer's instructions using a BD Accuri C6 flow cytometer (BD Biosciences).

#### *2.10. Fungal Challenge and Infection Assessment*

Either vaccinated or non-vaccinated mice were intraperitoneally inoculated with 106 conidia of either wild type (WT) or toluene-adapted (Tadap) *S. schenckii* suspended in 100 μL of PBS or with an equal volume of PBS alone as control. To confirm the fungal cell count and viability of the inoculum, appropriately diluted samples of the conidia suspension were plated onto Mycosel agar plates and after seven days of incubation growing colonies were counted. At seventh day post-infection the mice were euthanized in CO2 chamber. The liver and spleen of each animal were removed to measure the

relative organ weight and assess the systemic fungal load. The relative weight of livers and spleens was calculated by the following formula:

Related weight = organ weight (g)/body weight (kg).

To evaluate the fungal load, liver and spleen were macerated under sterile conditions and adequate dilutions of the macerate in PBS were cultured in duplicate, on Mycosel agar plates and the growing CFU were counted after three and six days. The final count was adjusted according to the dilution used.

#### *2.11. Statistical Analysis*

Statistical analysis was performed in GraphPad Prism ver. 6.01 (San Diego, CA, USA). A one-way analysis of variance (ANOVA) with Tukey comparisons test was used. The confidence interval was set at 95% for all tests. The significance level and *p*-values were shown as \* (*p* < 0.05); \*\* (*p* < 0.01); \*\*\* (*p* < 0.001); \*\*\*\* (*p* < 0.0001).

#### **3. Results**

#### *3.1. Production and Purification of rSsEno*

Figure 1C,D show that the production and purification of rSsEno was effective. SDS-PAGE showed a unique band stained with Coomassie blue with the expected molecular mass of 47 kDa previously characterized as rSsEno [13]. In addition, a pooled anti-rSsEno serum obtained from mice immunized with Freund´s Adjuvants/rSsEno recognized specifically the rSsEno band in the immunoblotting while there were not detected non-specific binding in the control strip treated with a serum from non-immunized mice.

#### *3.2. Post-Vaccination rSsEno-Antibody Response*

The rSsEno-specific IgG antibody reaction after the second immunization with or without the PGA are displayed in Figure 2 as optical density (OD) measured at 450 nm. Immunization with PGA+rSsEno markedly enhanced the IgG antibody response to rSsEno compared to the response induced by the non-adjuvanted vaccine, seven days after the second immunization (*p* < 0.0001).

**Figure 2.** Immunization with PGA+rSsEno markedly enhanced the IgG antibody response to rSsEno. Balb/c mice were immunized (s.c.) twice with rSsEno, PGA, PGA+rSsEno, or PBS as negative control. Serum collected seven days after the second immunization was used to determine rSsEno-specific IgG antibody. \*\*\*\* (*p* < 0.0001).

#### *3.3. Th1 and Th17 Response*

The response Th1 and Th17 are determinant in the immune response against *S. schenckii*. Mice vaccinated with rSsEno+PGA induced a significant response of Th1 lymphocytes after in vitro stimulation with rSsEno compared with the other groups (*p* < 0.0001). However, the response of Th17 lymphocytes was not modified in any experimental group (Figure 3). A similar response was observed when the concentration of IFN-G, IL-4, and IL-17A were measured by CBA. A high production of IFN-G, belonging to the Th1 pattern was detected in the group vaccinated with rSsEno+PGA compared with non-adjuvanted *rSsEno* vaccination (*p* < 0.001) and with the control groups (*p* < 0.0001) (Figure 4).

**Figure 3.** Response of Th1 (intracellular IFN-G+) and Th17 (intracellular IL-17A+) cells induced by immunization with the PGA+rSsEno. Balb/c mice were immunized (s.c.) twice with rSsEno, PGA, PGA+rSsEno, or PBS as negative control. Splenocytes of each animal were collected seven days after the second immunization and purified cells were stimulated in vitro with rSsEno and Brefeldin A. The frequency of Th1 and Th17 cells was detected using a BD Accuri C6 flow cytometer. Upper figure shows representative dot plots of Th1 and Th17 frequency collected from each group. \* (*p* < 0.05); \*\* (*p* < 0.01); \*\*\* (*p* < 0.001): \*\*\*\* (*p* < 0.0001).

**Figure 4.** Cytokine profile in supernatant of splenocytes stimulated in vitro with rSsEno. Balb/c mice were immunized (s.c.) twice with rSsEno, PGA, PGA+rSsEno, or PBS as negative control. Splenocytes of each animal were collected seven days after the second immunization and purified cells were stimulated in vitro with rSsEno. Cytokine were quantified by cytometric bead array (CBA). \* (*p* < 0.05); \*\* (*p* < 0.01); \*\*\* (*p* < 0.001): \*\*\*\* (*p* < 0.0001).

#### *3.4. Fungal Challenge and Infection Assessment*

The aim of this study was to investigate if the immune response induced by rSsEno adjuvanted with PGA was able to protect against a challenge with either WT or Tadap *S. schenckii.* We analyzed the spleen and the liver as representative organs to evaluate the fungal load. All the infected mice developed hepatomegaly that was observed by measuring the relative weight of the liver. However, splenomegaly was also observed in mice vaccinated and infected. Those animals vaccinated and infected with highly virulent toluene-adapted *S. schenckii* developed the greatest hepato- and splenomegaly compared with the control group (*p* < 0.0001) (Figure 5).

**Figure 5.** Relative (**A**) spleen and (**B**) liver weight. Balb/c mice were injected (s.c.) twice with PGA+rSsEno (vaccinated) or PBS (non-vaccinated). Seven days after the second dose of administration of PGA+rSsEno or PBS, mice were infected intraperitoneally with 1 <sup>×</sup> 106 conidia of either wild type (WT) or toluene adapted *S. schenckii* (Tadap). The spleen and the liver of each animal were collected seven days after the second immunization and the relative weight (organ weight/animal weight) was measured. \* (*p* < 0.05); \*\*\* (*p* < 0.001); \*\*\*\* (*p* < 0.0001).

The protective effect induced by the adjuvanted vaccine after the fungal challenge is shown in Figure 6. The immune response induced by PGA+ rSsEno was able to reduce the fungal burden in spleen and liver of mice infected with either WT or Tadap *S. schenckii* in a similar way. However, owing to the higher virulence of Tadap *S. schenckii*, this finding suggests that PGA+rSsEno can be effective against fungus with different levels of virulence.

**Figure 6.** Vaccination with PGA+rSsEno was able to reduce the fungal burden in spleen and liver of mice infected with either wild type (WT) or toluene adapted *S. schenckii* (Tadap). Balb/c mice were injected (s.c.) twice with PGA+rSsEno (vaccinated) or PBS (non-vaccinated). One week after the boost, mice were i.p. challenged with *S. schenckii* and seven days after infection the protection was assessed by the number of CFUs recovered from the spleen and liver. \* (*p* < 0.05); \*\* (*p* < 0.01); \*\*\* (*p* < 0.001); \*\*\*\* (*p* < 0.0001).

#### **4. Discussion**

Recent advances in the understanding of relevant immunological mechanisms against pathogenic fungi favor the development of prophylactic and therapeutic antifungal vaccines [23–25]. In contrast to classical antifungal medications, vaccines can be administered to large populations with low potential risks or side-effects [26]. Currently, there is a growing tendency to develop subunit vaccines, based in well-defined microbial components, in order to increase vaccine safety [27]. Unfortunately, pure antigens are poorly immunogenic, and they should be formulated with adjuvants to improve the vaccine efficacy [14,28].

Several non-toxic polymer adjuvants are being used for sustained delivery of protein subunit vaccines [29,30]. Polymeric adjuvants act through slow release of the antigen for the selective targeting to antigen presenting cells, promoting different signaling pathways including, activation of toll-like receptor(s) and inflammasome pathway or directly interacting with B cells. Polymer–antigen complex can be phagocytosed, and the antigen effectively presented to naive T cells via major histocompatibility complex (MHC) molecules [31].

Recently, we compared AH with the polymeric adjuvant PGA in a vaccine candidate against *S. schenckii* containing proteins extracted from the fungal cell wall. PGA induced a protective immune response against both *S. schenckii* and *S. brasiliensis* with lesser local toxicity than AH in vaccinated mice [21]. In other recent work, we developed recombinant enolase that was a key antigen in SsCWPs and it was evaluated in a vaccine formulated with PGA as an adjuvant. Again, protective immune response in vaccinated mice after a challenge with *S. brasiliensis* was observed with a survival rate above 90% after 45 days of intravenous fungal challenge [22].

Here, we evaluated the protective effect of PGA formulated with rSsEno against a highly virulent *S. schenckii.* The enhanced virulence was acquired after fungal exposure to 0.1% of toluene, which induced adaptive changes previously described, including enhanced melanosome formation and stronger antioxidant mechanism compared with the wild type strain [7].

Before the evaluation of the protective effect of vaccination, the immunogenicity of the vaccine after two subcutaneous doses of PGA+rSsEno was evaluated. In the aforementioned work [22], we evaluated the immunogenicity and efficacy of this vaccine candidate using three subcutaneous

administrations and PGA 10%. However, here the immunogenicity of the vaccine was evaluated after two doses and PGA 5%, as we used in another vaccine candidate with SsCWPs [21]. As expected, the formulation of PGA+rSsEno induced elevated production of specific IgG antibodies and a strong Th1 response after in vitro stimulation of splenocytes with rSsEno. However, we did not observe a significant Th17 response analyzed by both intracellular and released IL-17A.

Several studies revealed that Th1 response plays a decisive role in the defense against *S. schenckii* infection [32–35]. In this sense, activation of Th1 lymphocytes is becoming an interesting immunomodulatory strategy against sporotrichosis. Flores-García et al. (2015) reported that treatment with recombinant murine IL-12 (rmIL-12) promotes Th1 immunity and clinical improvement in an experimental sporotrichosis gerbil model [36]. In other study, Batista-Duharte et al. (2016) evaluated the therapeutic effect of adjuvant Finlay cochleates 3 (AFCo3), a cochleate containing purified and non-toxic LPS derived from *Neisseria meningitidis B* as vehicle of Amphotericin B (AmB) to evaluate the combined effect of immunomodulation induced by AFCo3 and the antifungal effect in a murine model of *S. schenckii* infection. AFCo3 stimulated a strong Th1- and Th17 response associated with the antifungal effect of AmB, which significantly improved the fungal clearance [37]. Regarding vaccination, studies suggest that Th1 response is associated with anti-sporothrix vaccine protection [20–22,38,39] and the activation of dendritic cells by fungal wall proteins seems to be important in theTh1 bias [34]. However, the role of Th17 response to prevent sporotrichosis is controversial. Some studies reveal that anti-sporothrix vaccination can induce Th17- combined with Th1 response [20,38–40]. However, two studies using PGA as an adjuvant with either SsCWPs [21] or rSsEno [22], showed that Th1 response is sufficient to achieve protection against a challenge with *S. schenckii* or *S. brasiliensis*. In the previous comparative study, using SsCWPs as antigen with either AH or PGA as adjuvant revealed that the adjuvant is the main component that drives the quality of the immune response since AH induced a balanced Th1/Th2/Th17 response while PGA induced a Th1 response [21]. Interestingly, the reduction of fungal load in both groups was very similar despite having different overall Th cell pattern.

In this work, we observed that the anti- rSsEno Th1 response induced by PGA confers protection against Tadap *S. schenckii* after an intraperitoneal fungal challenge. Previously, we showed that infected mice with Tadap *S. schenckii* produced high levels of IFN-G and, to a lesser extent, IL-17 [7]. This finding reinforces the criterion of the role of Th1-mediated response for protection against *S. schenckii* infection. Here, we observed significant production of IFN-G after in vitro stimulation with rSsEno, compared to non-vaccinated mice, although the cytokine production was low in general. However, unlike infection where multiple lymphocyte clones are activated, vaccination stimulates specific clones, but they are sufficient in the booster phase to prevent infection. In this way, the role of vaccination is the expansion of specific lymphocytes to maintain a basal level of memory cell able to rapidly respond against microbial infection [41].

These results are important as part of the studies of the efficacy of the recombinant enolase-Montanide™ PetGel A vaccine, due to the environment can exert a direct influence on fungal virulence [4,5]. Ideally, a prophylactic vaccine should be able to prevent infections caused by microorganisms with different levels of virulence since imperfect vaccination can enhance the transmission of highly virulent pathogens [42]. In this sense, the choice of potent and safe adjuvants driving effective immunological mechanisms against the target microorganism is one of the most important factors for vaccine efficacy [28]. However, keeping in mind that besides the antigen and the adjuvant mode of action, vaccine efficacy and toxicity are strongly influenced by the genetic background of the host [43,44], additional studies in different mouse strains and non-rodent species will be performed.

#### **5. Conclusions**

In this work, we provide evidence that vaccination with a recombinant enolase-based vaccine formulated with PGA as an adjuvant promotes a protective immune response in mice, against a highly virulent *S. schenckii* by toluene exposure. To the best of our knowledge, this is the first experimental approach to assess the role of vaccination in providing protective immunity against pathogenic fungus with enhanced virulence by exposure to environmental contaminants. Further studies are necessary in order to evaluate the memory response, including other mouse strains and using a model of subcutaneous infection.

**Author Contributions:** Conceptualization, D.T.-M.; D.L.P.; A.B.-D.; I.Z.C. methodology, D.T.-M.; A.B.-D. D.L.P. M.L.L. formal analysis D.T.-M.; D.L.P.; A.B.-D.; investigation, D.T.-M.; A.B.-D. D.L.P., M.L.L.; resources, D.T.-M.; D.L.P.; A.B.-D.; data curation, A.B.-D.; writing—original draft preparation, D.T.M.; D.L.P.; A.B.-D.; writing—review and editing, D.T.-M.; D.L.P.; A.B.-D.; I.Z.; supervision, A.B.-D.; I.Z.C. project administration, I.Z.C.; A.B.-D.; funding acquisition, I.Z.C.

**Funding:** This work was supported by Fundacão de Amparo à Pesquisa do Estado de São Paulo (FAPESP, grants 2017/26774-3).

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

#### **References**


Th17 and regulatory T cells responses than Sporothrix schenckii sensu stricto in mice. *Fungal Biol.* **2018**, *122*, 1163–1170. [CrossRef]


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **A Functional Wound Dressing as a Potential Treatment for Cutaneous Leishmaniasis**

**Francisco Alexandrino-Junior 1, Kattya Gyselle de Holanda e Silva 2, Marjorie Caroline Liberato Cavalcanti Freire 3, Viviane de Oliveira Freitas Lione 2, Elisama Azevedo Cardoso 2, Henrique Rodrigues Marcelino 4, Julieta Genre 5, Anselmo Gomes de Oliveira <sup>6</sup> and Eryvaldo Sócrates Tabosa do Egito 1,5,7,\***


Received: 18 February 2019; Accepted: 21 March 2019; Published: 1 May 2019

**Abstract:** Cutaneous leishmaniasis (CL) is a parasitic disease characterized by progressive skin sores. Currently, treatments for CL are limited to parenteral administration of the drug, which presents severe adverse effects and low cure rates. Therefore, this study aimed to develop poly(vinyl-alcohol) (PVA) hydrogels containing Amphotericin B (AmB) intended for topical treatment of CL. Hydrogels were evaluated in vitro for their potential to eliminate promastigote forms of *Leishmania* spp., to prevent secondary infections, to maintain appropriate healing conditions, and to offer suitable biocompatibility. AmB was incorporated into the system in its non-crystalline state, allowing it to swell more and faster than the system without the drug. Furthermore, the AmB release profile showed a continuous and controlled behavior following Higuchi´s kinetic model. AmB-loaded-PVA-hydrogels (PVA–AmB) also showed efficient antifungal and leishmanicidal activity, no cytotoxic potential for VERO cells, microbial impermeability and water vapor permeability compatible with the healthy skin's physiological needs. Indeed, these results revealed the potential of PVA–AmB to prevent secondary infections and to maintain a favorable environment for the healing process. Hence, these results suggest that PVA–AmB could be a suitable and efficient new therapeutic approach for the topical treatment of CL.

**Keywords:** Amphotericin B; cutaneous leishmaniasis; hydrogel; wound dressing; controlled release

#### **1. Introduction**

Cutaneous leishmaniasis (CL) is a disease caused by a genus of trypanosomatid protozoa called *Leishmania*, transmitted to humans by the bite of infected female phlebotomine sandflies [1]. CL commonly appears first as a localized papule, which evolves into an ulcer upon loss of the epidermis. Afterwards, the impairment of the skin barrier in the lesion leads to the formation of long-life scars and severe skin disabilities. Moreover, CL, which represents the most common form of leishmaniasis, is currently considered a serious public health problem in 98 countries over all five continents [2]. The World Health Organization (WHO) has estimated that 0.7 to 1.3 million new cases of the disease occur worldwide annually [3].

While the development of an effective vaccine against *Leishmania* spp. is still under research [4,5], the use of pentavalent antimony organic compounds (SbV) or pentamidines are the recommended treatment for all leishmaniasis forms [6]. However, their efficiency for getting rid of the parasites is only around 60% [7]. In this context, Amphotericin B [8], an antifungal polyene agent approved by the FDA for clinical use, has been successfully applied when the abovementioned treatments failed, showing in some cases only 15% failure [9]. Nevertheless, the clinical use of AmB is limited due to the severe side effects, mainly nephrotoxicity, that the micelle system formulation containing this drug presents [8]. Besides, the use of AmB requires hospitalization for its intravenous administration, which leads to non-adhesion to the treatment by infected individuals in 75% of cases [7]. On the other hand, when less toxic AmB formulations are used, such as the liposomal ones [10], their cost is prohibitively expensive for people in developing countries, and these dosage forms still require intravenous administration.

Considering the present scenario, the development of new therapeutic approaches to improve CL treatment and to promote its world accessibility is mandatory. In this context, the topical treatment of CL lesions represents an attractive alternative to reduce the systemic toxicity associated with the use of the abovementioned dosage forms administered intravenously, promoting elimination of parasites, re-epithelization of the skin, preventing secondary infections and enabling outpatient treatment.

The currently available local treatments for CL include intralesional injection of SbV [11] or a combination of SbV with physical therapies, such as cryotherapy [12] or thermotherapy [13]. Additionally, the topical administration of paromomycin-methylbenzethonium chloride (PR–MBCL) ointments is an option. However, although this last treatment presents fewer adverse effects, less pain, and easier administration, its therapeutic activity depends on the presence of MBCL in the formulation [14], a cationic surfactant which usually leads to inflammatory reactions. Moreover, Kim et al. demonstrated that PR regimen was less effective than any SbV regimen to achieve a clinical cure, with a described efficacy varying from 17% to 67% for *Leishmania major* infection [14].

Undeniably, the development of topical formulations containing AmB seems to be the ideal approach for CL treatment. However, the commercially available lipid formulations for this drug, when topically applied, are ineffective at curing CL in an animal model [15]. This observation supports the idea that the lack of efficiency of this type of treatment could be a deficiency of the drug delivery rather than a lack of drug efficacy. Therefore, this highlights the need to develop new therapeutic systems loaded with AmB for the topical treatment of CL.

In this context, the use of hydrogels seems to be a promising strategy, since it combines numerous advantages to wound management, e.g., enhancing the healing process [16,17], reducing pain [18], enabling exchange of gases (e.g., O2 and H2O) [19], possibility to tailor the mucoadhesion [20], act as a barrier to external threats like microbes [19]. Moreover, it can, simultaneously, act as a carrier for therapeutic agents [21–24].

Hydrogels consist of polymeric networks that absorb large amounts of water while remaining insoluble in aqueous solutions due to chemical or physical cross-linking of their individual polymer chains. Since the first report in 1960, by Wichterle and Lím [25], synthetic hydrogels have been widely applied to biomedical use, and special attention should be given to the ones manufactured with poly(vinyl alcohol) (PVA). Due to its excellent properties (e.g., high biocompatibility, hydrophilicity, transparency, etc.) this polymer has a historical use in biomedical applications [26], e.g., cell culturing [27], artificial cartilage [28], long-term implants [29,30], scaffold for tissue engineering and tissue mimicking [31], wound dressing material [32,33] and soft contact lenses [34]. Lastly, PVA has been used as carriers for therapeutic agents such as drugs [35–39] and as a surface modifier, for

liposomes [40,41]. The use of PVA in the pharmaceutical field, as a material to manufacture drug delivery systems, is justified because of its biodegradability [42] and low toxicity [43]. However, due to its high water solubility, PVA needs to be subjected to a cross-linking process in order to manufacture the hydrogel system. To this end, chemical cross-linking is one of the approaches used to enhance its mechanical, chemical and thermal properties [44].

Recently, our group demonstrated the feasibility of PVA hydrogels as a carrier to control AmB release [35]. Thus, the proposal of the present work was to produce an AmB-loaded poly(vinyl alcohol) hydrogel (PVA–AmB) intended to be used as a wound dressing system for CL treatment. With this aim, a 2<sup>3</sup> full factorial design was applied to evaluate the role of the pH, the degree of cross-linking and the presence of AmB in the system´s constitution. Then, the hydrogel microstructure was characterized by evaluating the AmB aggregation state into the system, surface morphology, swelling degree, drug release kinetics, and water and microbial permeability. The final goal of this proposal was to assess the leishmanicidal, antifungal, and cytotoxic activity of the AmB-loaded system.

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

#### *2.1. Materials*

Glutaraldehyde (GA, 25%*v*/*v* aqueous solution), dimethyl sulfoxide (DMSO) and monobasic anhydrous potassium phosphate were purchased from Quimiobras Indústria Química (Masssaranduba, SC, Brazil). Sodium chloride, potassium chloride, and methanol came from ISOFAR Indústria e Comércio de Produtos Químicos (Duque de Caxias, RJ, Brazil). Amphotericin B came from Indofine Chemical Company (Hillsborough, NJ, USA). Brain heart infusion medium and Sabouraud Dextrose Agar were purchased from HiMedia Laboratories, LLC (Mumbai, India). Streptomycin and penicillin were purchased from LGC Biotecnologia (Cotia, SP, Brazil). Fetal bovine serum (FBS) was purchased from Gibco™ (Gaithersburg, MD, USA). Dibasic sodium phosphate, Schneiders modified medium, Dulbecco´s Modified Eagle´s Medium–High glucose (DMEM–HG), and anhydrous poly(vinyl alcohol) (PVA, 98% hydrolyzed, Mw 13.000–23.000 g/mol) were purchased from Sigma-Aldrich (St. Louis, MO, USA).

#### *2.2. Factorial Design for Development of AmB-Loaded Hydrogels*

A 23 full factorial design was used to evaluate the influence of three main factors on the system's characteristics. In this case, the pH (5 and 2), the cross-linker concentration (Glutaraldehyde at 132.5 <sup>μ</sup>M·L−<sup>1</sup> and at 26.5 <sup>μ</sup>M·L<sup>−</sup>1), and the presence of the AmB in the system (0 and 10 <sup>μ</sup>M) were tested.

As previously reported, PVA hydrogels were prepared using the casting technique with few modifications [45]. Briefly, PVA powder was accurately weighed, introduced into an airtight container with water, and completely dissolved at 90 ◦C for 15 min to form a 10%(*w*/*v*) solution. This solution was cooled slowly at room temperature, under magnetic stirring, and the pH was adjusted to 2 with HCl 1 M. Afterward, AmB was added to the PVA solution in a 1:20 ratio (AmB:PVA *w*/*w*), followed by ultrasound bath for 10 min. The cross-linking reaction was carried out for 16 h in a polyethylene dish (diameter of 5.5 cm). Before all analyses, except the X-ray diffraction and the scanning electron microscopy, the hydrogels were immersed twice in distilled water to remove H<sup>+</sup> excess.

#### *2.3. X-ray Di*ff*raction Study*

The X-ray diffraction (XRD) spectrum of the hydrogels with (PVA–AmB) and without AmB (PVA–H), respectively, were carried out in the range from 2◦ to 80◦ (2θ) at the speed of 0.05◦/s using a diffractometer (Miniflex desktop diffractometer, Rigaku Corporation, Tokyo, Japan). The XRD system was equipped with a CuKα radiation source (λ =1.541 Å, 40 kV and 40 mA).

#### *2.4. Hydrogel Thickness and Morphology*

In order to investigate the thickness homogeneity in the PVA–AmB, the systems were virtually divided into two regions, edge and inner. The thickness was determined using a manual ABS digimatic indicator (model 543250B /ID-C112B, Mitutoyo Corporation, Kanagawa, Japan). The final results were expressed as the mean of 20 measurements on the entire film surface (10 at random edge and inner regions, respectively).

The scanning electron micrographs of PVA–H and PVA–AmB were performed using a Scanning Electron Microscope (JSM-5610LV Scanning Electron Microscope, JEOL Ltd, Tokyo, Japan). Prior to evaluation, hydrogels were placed over carbon conductive tabs and sputter-coated with gold plasma. The morphology of the system and the presence of pores were evaluated using an accelerating voltage of 20 kV.

#### *2.5. Drug Loading E*ffi*ciency (%)*

The amount of AmB presented in the system was measured from hydrogel fragments of 1 cm2. To this end, these pieces were immersed in 3 mL of DMSO and taken to the ultrasound bath for 15 min. The resulting solution was again diluted in DMSO and the AmB content was measured by spectrophotometry (Libra S32 UV/Vis Spectrophotometer, Biochrom Ltd, Cambridge, UK) at λ = 416 nm. Drug loading efficiency was calculated by the following equation [46]:

$$\text{Drug Loading } (\%) = \frac{\text{Experimental Drug Content}}{\text{Theoretical Drug Content}} \times 100\tag{1}$$

#### *2.6. Determination of Swelling Behavior*

A gravimetric approach was used in order to measure the swelling behavior of the system. Therefore, PVA–H and PVA–AmB hydrogels were cut into discs of 1 cm in diameter and immersed into phosphate buffer saline (PBS, pH 7.3 at 25 ◦C), which was prepared according to the Brazilian Pharmacopoeia [47], for a pre-determined time. Moreover, the samples were taken out of the swelling medium and weighed again after removing PBS excess. The swelling degree (*Q*) of the hydrogel at each time was calculated as follows [48]:

$$Q = \frac{M\_t - M\_0}{M\_0} \times 100\tag{2}$$

where *M*<sup>t</sup> and *M*<sup>0</sup> are the weight of the system imbibed with the swelling medium at time *t* and dry sample, respectively.

#### *2.7. In Vitro Drug Release*

Since the AmB has a very low water solubility (<1 mg·L−<sup>1</sup> at pH 6–7) [49], the addition of a solubility enhancer is a usual approach on drug delivery. Therefore, in order to balance the duality of performing changes in the release medium and, at the same time, still produce reliable results, the experiments were performed using the release medium previously reported in the literature [50].

The total immersion method was used to evaluate the in vitro AmB release from the system. To this end, hydrogels were cut into fragments of 1 cm2 (equivalent to 216 <sup>±</sup> 20 <sup>μ</sup>g of AmB) and immersed in 150 mL of release medium (PBS:Methanol (80:20 *v*/*v*)) pH 7.3 at 37 ◦C, and stirred at 100 rpm. At specified time intervals, aliquots of 3 mL were withdrawn from the solution and the amount of AmB was measured by spectrophotometry (Libra S32 UV/Vis Spectrophotometer, Biochrom Ltd, Cambridge, UK) at λ = 416 nm with a previously validated method. The amount of drug released was calculated and represented as the accumulative percentage of the drug released versus time. To maintain sink conditions, each aliquot was replaced by the same volume of fresh medium.

#### *2.8. Mathematical Analysis of the In Vitro Release Kinetics*

Different mathematical models were evaluated in order to describe the mechanism of the AmB release that best suited the hydrogels, according to the experimental data obtained from the in vitro release assays (Table 1). Calculations were performed with the support of Add-in DDsolver for Microsoft® Excel [51]. The choice of the model took into account the statistical significance of the fitting and the thermodynamic considerations of the model.

**Table 1.** Different mathematical models used for fitting experimental data and their equations with the values of the statistical parameters.


<sup>a</sup> In all mathematic models *F* represents the fraction (%) of the drug released over time *t*; <sup>b</sup> *k*<sup>0</sup> = zero-order release rate constant; <sup>c</sup> *k*<sup>1</sup> = first-order release rate constant; <sup>d</sup> *k*<sup>1</sup> = release rate constant for the Quadratic model denoting the dependence of the drug release on the time; *k*<sup>2</sup> = release rate constant for the Quadratic model denoting the dependence of the drug release on the quadratic time; <sup>e</sup> *k*<sup>H</sup> = Higuchi release constant; <sup>f</sup> *K*BL = combined constant on the Baker–Lonsdale model; <sup>g</sup> *k*KP = release constant incorporating structural and geometric characteristics of the drug-dosage form; *n* = diffusional exponent indicating the drug-release mechanism; <sup>h</sup> *k*HB = combined constant on Hopfenberg model; *n* = 1, 2, and 3 for a slab, cylinder, and sphere, respectively; \* *T*lag = lag time prior to drug release.

#### *2.9. Water Vapor Transmission*

The water vapor transmission (WVT) of hydrogels was determined using a modified ASTM E96/E96M water method [52]. Briefly, a test tube containing distilled water was sealed with the hydrogel. To avoid water transport through the edge, the test tubes were thoroughly sealed with scotch tape. Then, the assembly was placed into a chamber at 25 ± 1 ◦C with a constant relative humidity of 33% ± 1%. Finally, the change in weight of the assembly was measured and the rate of WVT was calculated using the following equation [53]:

$$WVT = \frac{W}{A \times \Delta p} \tag{3}$$

where *<sup>W</sup>* is the amount of water vapor permeating through the hydrogel (g·day−1), *<sup>A</sup>* is the area of exposed hydrogel (cm<sup>−</sup>2), and Δ*p* is the vapor pressure difference (mmHg).

#### *2.10. Microbial Permeability Assay*

In order to investigate the ability of the hydrogel to prevent microbial penetration, microbiological tests were conducted as previously mentioned [19], with modifications. Briefly, test tubes containing 3 mL of sterile brain heart infusion medium were sealed with sterile hydrogel at aseptic conditions. Positive and negative controls were an open and cap-closed test tube, respectively. The assembly was placed in an open environment and the progress of microbial permeation was observed for 7 days. The cloudiness of the medium in any test tube was recorded as microbial contamination.

#### *2.11. Leishmanicidal Activity Assay*

In order to estimate the leishmanicidal activity of the PVA–AmB, in vitro assays were performed as previously reported [54] with modifications. Briefly, PVA–H and PVA–AmB hydrogels were cut into fragments of 1 cm<sup>2</sup> and immersed into Schneiders modified medium. Afterward, late log-phase promastigotes of *Leishmania amazonensis* (IOC/L0575(IFLA/BR/1967/PH8)) and *Leishmania braziliensis* (IOC/L0566(MHOM/BR/1975/M2903) were added to the culture medium in the concentration of <sup>1</sup> <sup>×</sup> 105 parasites·mL−<sup>1</sup> and cultured in 24-well plates at 28 ◦C under 5% CO2 for 48 h. At specific time intervals, the promastigote viability was microscopically determined at 400× magnification, through flagellar motility, and counted in a Neubauer chamber. The experiments were carried out under similar conditions using non-treated cells and free AmB (equivalent to a final concentration of 50 μg/mL) as negative and positive controls, respectively.

#### *2.12. Antifungal Activity*

The disc diffusion approach was used to evaluate the pharmacological efficacy of the PVA–AmB hydrogel against *Candida albicans* (ATCC®: 10231), as previously reported [55] with modifications. Briefly, a suspension of this fungal strain, equivalent to 0.5 on the McFarland scale, was homogeneously distributed on a Petri dish with Sabouraud Dextrose Agar using a sterile swab. Afterward, PVA–H and PVA–AmB hydrogels were cut aseptically in fragments of 1 cm2 and neutralized with PBS. Then, hydrogels were put on the surface of the culture medium and the zones of growth inhibition were measured after 24 and 48 h of incubation at 28 ◦C. Control experiments were carried out under similar conditions using 10 μL of an AmB dispersion in PBS (2 mg/mL) as the standard drug with antifungal activity. Microorganism sensitivity to PVA–H and PVA–AmB hydrogels was determined by measuring the size of the inhibitory zones on the agar surface around them.

#### *2.13. Cytotoxicity Assay*

The nephrotoxicity of the hydrogels was estimated performing an in vitro assay. Firstly, African Green Monkey Kidney (VERO, ATCC® CCL-81) cells were seeded at 10<sup>4</sup> cells per well and cultured in Dulbecco´s Modified Eagle´s Medium-High glucose (DMEM–HG) supplemented with 10% FBS, streptomycin (100 mg/mL), and penicillin (100 UI/mL). When the cultures reached confluence, transwell filters (0.8 <sup>μ</sup>m; BD) containing PVA–H or PVA–AmB hydrogel fragments of 0.5 cm2 (AmB <sup>≈</sup> 113 <sup>μ</sup>g) were applied. Then, the plates were incubated at 37 ◦C under 5% CO2 for 24 h. After treatment, the cytotoxicity was evaluated by the MTT cell proliferation assay [56]. The viability of non-treated control cells was defined as 100%.

In order to better understand the main effect of AmB toxicity to VERO cell line, a kill curve was performed. To this end, a stock solution of AmB in DMSO was diluted into the culture medium to obtain a concentration of AmB in the range from 5 to 50 μg/mL. The cells were cultivated according to the above-mentioned methodology.

#### *2.14. Statistical Analysis*

Statistical analysis was performed by the GraphPad Prism 5.03 software (GraphPad Software Inc, San Diego, CA, USA) using the Shapiro–Wilk test to evaluate the normality of the data distribution, followed by the Bartlett test to assess the homogeneity of variance. Afterward, unpaired *t* tests were performed to determine the difference between the means. The results were presented as the mean of three individual experiments ± standard error of the mean, with *p*-value < 0.05 considered significant.

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

#### *3.1. Factorial Design*

The influence of the pH over the system's characteristics was considered significant once only hydrogels produced at pH 2 were able to maintain their structure during the dissolution test in the aqueous medium, while those produced at pH 5 were completely dissolved. Furthermore, the cross-linker concentration also influenced the system´s structure. A synergism during the cross-linking reaction was observed with a higher level of both AmB (at 10 <sup>μ</sup>M) and GA (at 132.5 <sup>μ</sup>M·L<sup>−</sup>1), showing excessive cross-linking bonds, and leading to significant deformations of the hydrogel. This was an unexpected finding and it was ascribed to an increase in the number of hydrophilic interactions among the polymer chains, the cross-linker, and the AmB (Figure 1A), improving the physical and chemical cross-linking occurrence [57,58].

**Figure 1.** (**A**) Schematic representation of the hydrophilic interactions among the polymer chains, the cross-linker, and the Amphotericin B (AmB). (1) Hydrogen bound, (2) acetal formation and (3) imine formation improving the physical and chemical cross-linking [59]. (**B**) AmB content per unit area (1 cm2), (**C**) thickness variation according to the hydrogel region (**D**) AmB values normalized according to the weight of fragments from each respective region.

#### *3.2. Drug Loading E*ffi*ciency (%)*

The efficiency of AmB loading into the system was evaluated, as well as a possible difference in the drug distribution between the edge and the inner regions of the hydrogel. In fact, the quantitative analysis of the distribution of the drug into hydrogel showed a significant higher concentration of AmB at the film edges (*p*-value < 0.05), when compared to its inner region (Figure 1B). This result was ascribed to the capillarity phenomenon, which occurred during the cross-linking reaction at the border of the Petri dish used to generate the system, hence increasing the AmB content in this region.

This hypothesis was corroborated by the thickness comparison and PVA–AmB ratio of both hydrogel regions, inner and edge (Figure 1C). Indeed, the edge region was approximately 2.6 times thicker than the inner region (*p*-value < 0.05), but the PVA–AmB ratio did not change. In fact, the AmB content was normalized according to the weight of each respective fragment and showed no statistical difference between these regions (*p*-value > 0.05, Figure 1D).

Taking into account that the WHO [6] recommends the collection of samples from the swollen edge of CL ulcers for diagnostics due to the highest number of parasites found in this part of the lesion, the heterogeneity observed in the AmB distribution inside the hydrogel can be exploited as a therapeutic strategy. This could assure a better management of the administrated dose, targeting the majority of the active compound to sites where more parasites are found, leading consequently to a more efficient therapy.

#### *3.3. Swelling Behavior*

The swelling behavior of the hydrogel is a dynamic process, which depends on structural factors such as geometry, chemical composition of the polymeric network, the presence of pores and polymer–solvent interactions [60–63]. As AmB release from the hydrogel requires the previous solvation of the polymeric chains, swelling kinetic experiments were performed to better understand the mechanisms involved in this process.

Although there was no statistical difference in the swelling behavior of the hydrogels, Figure 2A shows that the PVA–AmB hydrogel had the ability to absorb four times its weight in water. This property could be of great applicability in the treatment of the leishmaniotic ulcers, as it may assist in the removal of secretions, which are presented in up to 86.9% of the lesions [64], and could be extremely beneficial for the healing process.

**Figure 2.** (**A**) Swelling degree of poly(vinyl-alcohol) (PVA)–H and PVA–AmB. (**B**) Log-log plot from which were calculated the diffusional exponent (*n*) and the diffusion constant (*k*) of PVA–H and PVA–AmB in phosphate saline buffer.

In addition, Ritger and Peppas [65] showed that from the swelling kinetic assay data it is possible to determine the mechanisms involved in the release process mainly by the use of the following equation:

$$\frac{M\_{\rm t}}{M\_{\rm eq}} = kt^{\rm n} \tag{4}$$

where *n*, *k*, *M*<sup>t</sup> and *M*eq represents the diffusional exponent, the diffusion constant, and the mass of the hydrogel at the time *t* and at the equilibrium, respectively. The values of these constants may be determined by Plotting log( *<sup>M</sup>*<sup>t</sup> *<sup>M</sup>*eq ) × log(*t*) (Figure 2B), where the angular and linear coefficients correspond to *n* and *k*, respectively [63].

Although the presence of AmB did not change the swelling degree (*p* > 0.05), its presence significantly altered the sorption mechanism. The n values for PVA–H and PVA–AmB hydrogels were 0.2 and 0.52, respectively, which indicate a change in the swelling mechanism, from the Less Fickian to the Limited Relaxation of the polymeric chains one (Anomalous Process) [65]. This hindrance solvent permeation observed for the PVA–H hydrogel was attributed to the strong inter- and intra-chain interactions between PVA and GA in the absence of AmB. This fact can be evidenced by the analysis of the diffusion coefficient values of the solvent into the hydrogels, which were 11.10−<sup>2</sup> and 17.10−<sup>2</sup> cm2·s−<sup>1</sup> for PVA–H and PVA–AmB, respectively. Then, the presence of AmB in the hydrogel increased approximately 1.5 times the speed of solvent permeation into the system microstructure.

#### *3.4. In Vitro Drug Release*

The evaluation of the AmB kinetic release profile from the hydrogel showed that, despite the rapid swelling, the AmB release occurred slowly and gradually. After 97 h, approximately 74% of the total drug content was released from the system. Afterward, the data were fitted using different mathematical models previously described in the literature [51]. The statistical parameters used to compare these different models were the Adjusted Coefficient of Determination (R2\_adj), which allowed comparing the fitting of the theoretical models to the experimental data, and the Root Mean Square Error (RMSE), which evaluated the difference between the experimental obtained data and the fitted data provided by the used model. Consequently, the model that best fit the experimental data had to exhibit the largest R2\_adj and the lower RMSE (Table 1). Although from the statistical point of view the Baker–Lonsdale with Tlag was the model that best described the experimental data, from the thermodynamic point of view this model was inappropriate since this approach, adapted from the Higuchi model, is applied to systems having spherical geometry such as microspheres or microcapsules. The literature reports that when the mechanistic models are under investigation, its selection should be based not only on the parameters that fit best, but also on the mechanistic model probability [51]. Thus, the model of Higuchi with Tlag was the chosen model (Figure 3A). This model (Table 1) depicts the release mechanism as a diffusion process based on Fick's law and dependent on the square root of time.

**Figure 3.** (**A**) Kinetic release of AmB from PVA–AmB hydrogels. Key: (•) experimental data (–) fitted data according to the Higuchi model. (**B**) XRD patterns of the AmB, the PVA–H, and the PVA–AmB hydrogels.

Another result that corroborates the hypothesis of the AmB release from the hydrogel being governed primarily by diffusion is the system microstructure organization inferred by the XRD analysis (Figure 3B). The diffractogram showed an absence of crystalline peaks ascribed to the AmB, indicating that the AmB is in an amorphous state or molecularly dispersed in the hydrogel. Additionally, the crystallinity index (CI) of the polymeric network calculated through the equation below, was 16.8% and 17.5% for PVA–H and PVA–AmB hydrogels, respectively, and showed no significant changes.

$$\text{CI}\_{\text{(\%)}} = \left(1 - \frac{A\_{\text{a}}}{A\_{\text{t}}}\right) \times 100\tag{5}$$

where *A*a and *A*t are the amorphous phases and the total area, respectively.

Furthermore, the release by erosion mechanisms, as previously reported for PVA hydrogels [66], had not fitted the data, which is justified by the absence of groups subject to hydrolysis in the main chain. Therefore, the AmB release depends only on the solvation, the chain relaxation, and the diffusion.

#### *3.5. Water Vapor Permeability*

A key property that needs to be presented by modern wound dressings is the ability to absorb excessive exudates while maintaining a moist environment [67]. Thus, the water vapor transmission of the PVA–H and PVA–AmB hydrogels was evaluated. The results showed that the presence of AmB in the hydrogel network does not significantly affect the water vapor permeability (PVA–AmB = 393 ± 33 <sup>g</sup>·m−2·day<sup>−</sup>1; PVA–H <sup>=</sup> <sup>452</sup> <sup>±</sup> 10 g·m−2·day<sup>−</sup>1) (*p*-value <sup>&</sup>gt; 0.05) (Figure 4A).

**Figure 4.** (**A**) Water vapor transmission rate of PVA–H and PVA–AmB hydrogels. (**B**) Medium aspect of positive (+) and negative (−) controls, as well as PVA–H and PVA–AmB after 7 days exposure to the environment, showing the resistance of the hydrogels to microbial permeability.

The literature reports that the skin in a healthy state has a water vapor permeability of 204 <sup>g</sup>·m−2·day<sup>−</sup>1, while in an injured state it can fluctuate between 279 to 5138 g·m−2·day−<sup>1</sup> [68,69]. Thus, based on our results, it is possible to infer that the PVA–AmB would be able to control the water loss by evaporation from a wound, allowing the skin in an injured state to have a water vapor permeability compatible with that in the healthy state, avoiding dehydration.

#### *3.6. Microbial Permeability Assay*

The ulcer, the most common clinical manifestation of CL, can remain active for several months, exposing the tissue to microorganisms both from normal skin microbiota and from the environment. The role of the secondary infection in the healing evolution of the leishmaniotic ulcers is still unclear in the literature. However, their acquisition can be considered a grievance to the patient clinical condition [6]. Furthermore, prevention of secondary infection is especially important for *Leishmania*-human immunodeficiency virus (HIV) co-infection individuals, who already have a deficient immunological system [70,71]. Therefore, wound dressing formulations able to treat *Leishmania* infection and avoid secondary microbial infections seem to be of particular importance to these patients.

Thus, the resistance of the hydrogels to microbial permeability was evaluated for 7 days. During the assay, no cloudiness was observed in the culture medium (Figure 4B), which reveals the ability of the hydrogels to act as a barrier for the microorganisms. In fact, this result shows that the use of AmB-loaded hydrogels for treatment of CL may prevent and/or hinder the development of secondary opportunistic infections. As evidenced by the SEM analysis in Figure 5A, the hydrogel displays an apparent nonporous structure. However, it is important to highlight that according to its porosity, the hydrogels are often classified as nonporous/nanoporous (10–100 Å), microporous (100–1000 Å), macroporous (0.1–1 μm) and superporous systems (1–1000 μm) [38,72–74]. Since the magnification of 1500× did not provide enough resolution to draw an accurate conclusion about the porosity of the system on a scale smaller than 1 μm (Figure 5A) and, beyond that, a collapse of the hydrogel structure

was caused (Figure 5B), no analysis at higher magnification could be performed. Therefore, such classification should be confirmed by further characterization.

**Figure 5.** Scanning electron microscope image of PVA–AmB hydrogel evidencing the sample deformation, caused by the electron beam, as the magnification increment goes from 1500× (**A**) to 3000× (**B**).

Regardless of its classification, either a nonporous system or a microporous one with pores smaller than the pathogens, the system may act as a mechanical barrier. Therefore, the microbial impermeability of the hydrogels produced in this work makes it potentially more efficient in preventing secondary infection than conventional materials such as gauze. [75].

#### *3.7. Biological Activity Assays*

The leishmanicidal activity of the system against *L. amazonensis* and *L. braziliensis* promastigotes was evaluated for 48 h in a culture medium containing PVA–H and PVA–AmB hydrogels. The results indicated that AmB released from the system performed a similar cytotoxic pattern for *Leishmania* promastigotes (i.e. 100% and 99% ± 2% of mortality for *L. amazonensis* and *L. braziliensis*, respectively) when compared to the positive control, within the first 24 h, evidenced by the absence of flagellar motility of the parasites (Figure 6A,B). Based on the drug release assay (Section 3.4), this is an expected result, since it was expected that the system would release ~38% ± 4% of the initial dose. Therefore, a concentration of AmB equivalent to the positive control (54 ± 6 μg/mL) would be produced.

**Figure 6.** Leishmanicidal activity of the PVA–H and the PVA–AmB against (**A**) *Leishmania amazonensis* and (**B**) *Leishmania braziliensis.*

Zauli-Nascimento et al [56] evaluated the AmB susceptibility of both *Leishmania* species using ATCC strains and clinical strains. The authors reported an AmB effective inhibition concentration of 50% (EC50) ranging from 36 ± 4 to 92 ± 4 ng/mL for the *L. braziliensis* and from 55 ± 1 to 83 ± 6 ng/mL for the *L. amazonensis*. In addition, no significant differences were found among *Leishmania* spp. clinical isolate susceptibility, showing a range of AmB EC90 from 80 to 650 ng/mL.

Thus, considering these findings and assuming that the AmB release in our study had followed the same profile observed in the in vitro drug release assay, the amount of AmB released, within the first 24 h (54 ± 6 μg/mL), would be enough to kill more than 90% of the promastigote forms. Furthermore, considering that the susceptibility of the *Leishmania* promastigote and amastigote forms to AmB was similar, as previously reported [76], it is possible to infer that a similar response to the PVA–AmB observed in vitro for *Leishmania* promastigotes would be observed for the amastigote stage.

Regarding the leishmanicidal activity of the PVA–H, an unexpected decrease in the number of parasites was observed when compared to the negative control (Figure 6A,B). This could be explained by the presence of residual cross-linking inside the polymer matrix, decreasing the promastigote viability.

In order to better understand the AmB cytotoxicity in VERO cells an AmB kill curve was carried out. It seems that the cell viability displays a reasonable linear correlation to the concentration of AmB, as demonstrated by fitting the data through linear regression (Figure 7A). Regarding the cell viability of the hydrogels, no significant difference between PVA–H and PVA–AmB (*p* > 0.05) was observed.

**Figure 7.** AmB killing curve for VERO cell lines (**A**); evaluation of the cytotoxicity of PVA-H and PVA-AmB against VERO cell lines (**B**).

Performing the same assumption as before, which is the pattern of AmB released into the culture medium follows the one previously observed (Section 3.4), it is possible to compare the cell viability obtained from the hydrogels (Figure 7B, PVA–H = 80% ± 5%; PVA–AmB = 72% ± 7%) to the one predicted by the linear model (~102%). Even though these values of toxicity enable to classify the systems as no cytotoxic potential [77], this analysis revealed that the systems displayed a cytotoxicity 20% higher than expected, which might be related to the presence of residual cross-linking agents.

Furthermore, it is important to highlight that the system here proposed is intended for topical use. Therefore, it is expected that a high dilution factor of the molecules will perhaps reach the bloodstream and, thereafter, the kidneys. For instance, even though the whole dose used on the assay with *Leishmania* spp. (~216 μg) reaches the bloodstream, a dilution by a factor of ~6 L is expected [78]. Therefore, obtaining a concentration of AmB ~36 ng/mL. Taking into account that AmB is highly bound in plasma protein >95% [79], only ~5% will be effectively evaluable to perform the cytotoxic effect. However, no toxic effect is expected with AmB concentrations below 40 μg/mL (Figure 7A).

Due to the recognized use of AmB as a gold standard on the treatment of fungal infections, mainly the ones caused by *Candida* spp., the antifungal properties of PVA–AmB hydrogels were evaluated by the radial disc diffusion assay against *C. albicans*. The zone of inhibition was measured using free

AmB and PVA–H as positive and negative controls, respectively. The PVA–AmB hydrogel resulted in a significant reduction in the viability of *C. albicans*, with inhibitory zones on the agar surface bigger than the positive control (16.7 ± 3 mm vs. 9.8 ± 0.5 mm for 24 h and 16.3 ± 2.5 vs. 9.5 ± 1.3 mm for 48 h, respectively, *p* < 0.05). These results indicate that the AmB entrapped inside of the hydrogel structure keeps its antifungal activity, just as free AmB, while the PVA–H did not show any antifungal activity, corroborating with the idea that AmB stands with its pharmacological activity unaltered and the residual cross-linking quantities were negligible to show antifungal activity or severe cytotoxicity. It has been reported that the affinity of AmB to the ergosterol on fungi membranes is strongly dependent on its aggregation state [80,81]. This phenomenon could be the main reason behind the enhanced antifungal activity of PVA–AmB compared to the free form of the drug. In fact, it is expected that the polymeric matrix of the hydrogel will release the AmB slowly and continuously on its monomeric form. On the other hand, free AmB, which is already evaluable into the medium, should display the aggregated form, since its concentration is higher than 10−<sup>7</sup> M [82–84].

#### **4. Conclusions**

The results presented here show that the factorial design was a useful tool for the development of PVA hydrogels by the casting method. Parameters like pH, cross-linker, and AmB concentration demonstrated significant effects over the system's structure. Importantly, the AmB loaded into PVA hydrogels remained in its amorphous state and did not significantly change the crystallinity index. Besides, the PVA–AmB showed water vapor permeation properties that make this kind of system suitable for topical treatment of CL lesions. Furthermore, the applied mathematical modeling showed that the AmB presence in the system altered the swelling mechanism from less Fickian to anomalous, by increasing the hydrogel swelling capacity. However, no burst release of AmB was observed and the release kinetic profile was adequately fitted to the Higuchi model. Additionally, AmB-loaded hydrogels were resistant to microbial permeation and showed effective activity against *Leishmania* promastigotes and *Candida albicans*, as well as no cytotoxic potential for VERO cell lines. These results are indicative of an efficient pharmacological activity and suitable biocompatibility of PVA–AmB hydrogels and show the great potential application of this system for the topical treatment of CL.

**Author Contributions:** Conceptualization, F.A.Jr. and K.G.d.H.eS; methodology, F.A.Jr.; validation, F.A.Jr and M.C.L.C.F; formal analysis, F.A.Jr. and H.R.M.; investigation, F.A.Jr., M.C.L.C.F, E.A.C. and V.d.O.F.L.; resources, E.S.T.d.E.; writing—original draft preparation, F.A.Jr., H.R.M. and J.G.; writing—review and editing, K.G.d.H.eS, A.G.d.O. and E.S.T.d.E.; visualization E.S.T.d.E. and A.G.d.O.; supervision, E.S.T.d.E. and K.G.d.H.eS; project administration, E.S.T.d.E. and J.G; funding acquisition, E.S.T.d.E.

**Funding:** This research was funded by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Finance Code 001 and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).

**Acknowledgments:** The authors would like to thank Paulo Henrique de Souza Picciani from Instituto de Macromoléculas Professora Eloisa Mano (IMA, UFRJ), M.Sc. Karen Cybelle de Holanda Silva and PharmB Bartolomeu Santos de Souza from Laboratório de Sistemas Dispersos (LaSiD, UFRN) for the technical support during the development of this study. The authors are also grateful to Glenn Hawes, M Ed. (Master of English Education, University of Georgia) for editing this manuscript.

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Thermoreversible Gel-Loaded Amphotericin B for the Treatment of Dermal and Vaginal Candidiasis**

**Lilian Sosa 1, Ana Cristina Calpena 1,2, Marcelle Silva-Abreu 1,2,\*, Lupe Carolina Espinoza 1,3, María Rincón 1, Nuria Bozal 4, Oscar Domenech 1, María José Rodríguez-Lagunas 5,6 and Beatriz Clares 2,7**


Received: 22 May 2019; Accepted: 1 July 2019; Published: 3 July 2019

**Abstract:** The present study was designed to develop a thermoreversible gel of Pluronic (P407) loaded amphotericin B (AmB-gel) for the dermal and vaginal treatment of candidiasis. P407 was used as a copolymer to exploit potential advantages related to increasing drug concentration in the tissue layer in order to provide a local effect. Parameters including internal structure, swelling, porosity, and short-term stability were determined. In addition, drug release profile and ex vivo skin and vaginal permeation studies were carried out. Antifungal efficacy was evaluated against strains of *Candida* spp. and atomic force microscopy (AFM) supported the results. The tolerance of AmB-gel was studied by evaluating biomechanical properties of skin and determining the irritation level in scarified rabbit skin supported by histological analysis. Results confirmed the development of a thermoreversible AmB-gel with high porosity exhibiting Newtonian behavior at 4 ◦C and pseudoplasticity at 32 ◦C as well as optimal stability for at least 90 days. The Amb-gel provided a sustained drug release following a Boltzmann sigmoidal model. Non permeation was observed in skin and vaginal mucosa, showing a high retained amount of AmB of 960.0 and 737.3 μg/g/cm2, respectively. In vitro antifungal efficacy showed that AmB-gel was more effective than Free-AmB in inhibiting strains of *Candida* spp. and these results were corroborated by AFM. Finally, tolerance studies showed that its application did not induce skin irritation nor alter its biophysical properties. Together, these results confirmed that AmB-gel could be proposed as a promising candidate for the clinical status in the treatment of skin and vaginal candidiasis.

**Keywords:** thermoreversible gel; poloxamer 407; candidiasis; amphotericin B; skin and vaginal mucosa

#### **1. Introduction**

Cutaneous fungal infections are a significant cause of morbidity and constitute a critical health issue on a global scale. The most identified fungal pathogenesis that affects people worldwide is based on superficial skin, vaginal and nail infections, with approximately 1.7 billion individuals affected. This has led to an increase of superficial mycoses higher than 20% in recent decades, especially in patients who are immunocompromised or hospitalized [1,2] as well as in cases of topical burn wound infections [3]. Among these fungal infections, various species of Candida cause superficial complications known as candidiasis [4]. Categorically, *Candida albicans* is reported to be the fourth most frequent cause of infection [5].

The first-line drugs used for the treatment of these mycoses are the Azoles antifungal compounds such as fluconazole, ketoconazole, voriconazole, itraconazole, clotrimazole, among others [6,7]. However, mechanisms of resistance to these antifungals have been detected [8,9]. The appearance of these resistance mechanisms leads to the use of other antifungals such as amphotericin B (AmB), especially in immunosuppressed patients [10,11].

AmB is a polyene macrolide of a broad-spectrum with high activity against the most frequently occurring fungi involved in cutaneous mycoses, including Candida [12]. It is considered the gold standard in the treatment of fungal infections and has been found to be effective against azole-resistant fungi due to the creation of transmembrane channels by complexation with the membrane sterols which results in fungal cell death. However, it is not without side effects in systemic administration due to the self-assembly process into higher-order molecular forms [13]. From a physicochemical point of view, AmB is a poorly hydrosoluble, amphoteric, amphiphilic molecule and is difficult to solubilize in organic solvents. These properties block an optimal permeation of the drug through the stratum corneum (SC), which is the major barrier that limits permeation of external substances. For this reason, there is no topical formulation of AmB commercially available at the moment [14]. This fact evidences the need to develop new formulations using excipients with permeation-enhancing properties in order to facilitate the penetration of drug into SC and its distribution from SC to epidermis and dermis. Therefore, the research of new therapeutic approaches for the treatment of superficial mycoses, including dermal and vaginal administration from galenical perspective, is a challenging yet necessary undertaking.

The topical administration of antimicrobials for superficial mycoses offers several advantages, particularly for direct application at the site of infection. Among these are the prevention of systemic toxicity of the drug, effectiveness at an eminently local level and insurance that a sufficient amount of drug is retained in the skin [15].

With the aim of achieving an improved release of AmB, this drug was incorporated to Poloxamer 407 (P407), which is a copolymer of amphiphilic nature consisting of a central hydrophobic block of polypropylene oxide (PPO) flanked by hydrophilic polyethylene (PEG) blocks (PEG–PPO–PEG). The P407 prepared above >18% shows an aqueous solution (4–5 ◦C) that turns into a gel ~32 ◦C, making it an ideal candidate for thermoreversible delivery [16]. P407 is a low toxicity excipient approved by the Food and Drug Administration (FDA) [17]. Moreover, it has beneficial properties that help to promote and improve drug permeation through the skin and mucosa [18].

Therefore, an approach to deliver AmB formulated into P407 gel form specifically developed for skin and vaginal treatment against *Candida* spp. was developed. This work has accomplished detailed research of (i) development and physicochemical characterization of AmB-gel for topical administration, (ii) biopharmaceutical studies (release profile) and ex vivo permeation studies, (iii) in vitro antifungal activity and its effect on yeast cells by atomic force microscopy (AFM), and (iv) biomechanical properties and tolerance assay complemented with histological studies.

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

#### *2.1. Materials*

AmB was obtained from Acofarma (Barcelona, Spain). Pluronic®® F127 (P407) was obtained from Fagron (Barcelona, Spain). Dimethyl sulfoxide (DMSO), methanol, castor oil and *N*,*N*-dimethylformamide were obtained from Sigma-Aldrich (Darmstadt, Germany). Transcutol®® P and Propylene glycol were kindly provided by Gattefossé (Barcelona, Spain). The water used in all experiments was obtained from a Milli-Q®® Plus System (Millipore Co., Burlington, MA, USA). All the chemicals and reagents were of analytical grade.

#### *2.2. Preparation of AmB-Gel*

Firstly, a selection of solvents was tested to determine the solubility of AmB. Transcutol®® P, Propylene glycol, *N*,*N*-dimethyformamide, Castor oil and DMSO were analyzed. The DMSO was selected as the best solubilizing component of AmB. Shortly afterwards 25 g of P407 were dispersed on ultrapure water at 4 ◦C up to 100 mL. Subsequently, 60 mg of AmB were dissolved in 10 mL of DMSO, of which 1 mL was slowly added under stirring to 20 mL of P407 solution at 4 ◦C. The resulting solution was adjusted to pH 6.5 with NaOH 2N solution, thus obtaining an AmB-gel at a final concentration of 0.03%.

#### *2.3. Physiochemical Characterization*

To analyze the morphology, the AmB-gel was dried over a period of 8 days using a desiccator with provision for a vacuum. Once it was dried, a small quantity was coated with carbon as a conductor agent. The internal structure of the gel was examined by scanning electron microscopy (SEM) using a JEOL J-7100F (Peabody, Massachusetts, MA, USA).

#### 2.3.1. Swelling and Degradation Tests

The swelling ratio (SR) and degradation as percentage of weight loss (WL) were accomplished by a gravimetric method. Half a gram of dried AmB-gel or fresh AmB-gel was used to carry out swelling and degradation tests, respectively. In both experiments the sample was incubated in PBS (pH = 5.5) at 32 ◦C for 30 min. Samples (*n* = 3) were removed and weighed after blotting the surface water at predetermined time intervals of 5 min. The PBS uptake was carried out in triplicate.

The SR ratio was calculated using the following equation and expressed by kinetic modeling:

$$\text{SR} = \frac{\mathcal{W}\_{\text{s}} - \mathcal{W}\_{d}}{\mathcal{W}\_{d}} \tag{1}$$

where *Ws* is the weight of the swollen AmB-gel at 5 min intervals and *Wd* is the weight of dried gel.

WL was calculated following the equation and expressed by kinetic modeling.

$$\text{WL } (\%) = \frac{\mathcal{W}\_i - \mathcal{W}\_d}{\mathcal{W}\_i} \times 100 \tag{2}$$

where *Wi* is the initial weight of AmB-gel and *Wd* the weight of gel at different times.

#### 2.3.2. Porosity Study

The porosity percentage (P) was calculated by displacement of the solvent. The method consisted of immersing the previously dried AmB-gel in absolute ethanol for 2 min and then weighing after the excess ethanol on the surface was blotted. The porosity percentage was calculated using the following equation.

$$P = \frac{w2 - w1}{\rho \times V} \times 100\tag{3}$$

where, *w1* represents the weight of the dried AmB-gel to be immersed in ethanol, *w2* represents the weight of AmB-gel after being immersed in ethanol, ρ is the density of absolute ethanol, and *V* is the volume of the gel.

#### *2.4. Stability Study*

The AmB-gel was stored at two temperatures: 4 ◦C and 25 ◦C. The pH values of formulation were measured in a Crison 501 digital pH/mV-meter (Crison Instruments, Barcelona, Spain) for a period of 90 days (1, 30, 60, and 90 days) at both temperatures. Values were reported as the mean ± standard deviation (SD) of six replicates.

For the chemical stability studies, the formulation was suspended in *N*,*N*–dimethylformamide: methanol:water (55:15:30, *v*/*v*/*v*) and the amount of drug was quantified by a previously validated method of High Performance Liquid Chromatography (HPLC) (Waters, Milford, MA, USA) [12].

The short-term stability of AmB-gel was studied for a period of 90 days at 4 ◦C and 25 ◦C. A TurbiScan Lab®® (Formulaction Co., L'Union, France) was used to analyze the destabilization phenomena by transmission and light retrodispersion using a pulsed near-infrared light source (γ = 880 nm) at 25 ◦C over a span of 90 days.

#### *2.5. Rheological Studies*

The rheological measurements were performed using a Thermo Scientific Haake Rheostress 1 rotational rheometer (Thermo Fisher Scientific, Kalsruhe, Germany) equipped with cone plate geometry (60 mm diameter, 2◦ angle) with mobile upper cone Haake C60/2◦ Ti (0.105 mm gap). Viscosity and flow curves were rehearsed at 4 ◦C and 32 ◦C in triplicate. The shear rate ramp program included: 0→50 s−<sup>1</sup> (3 min), 50 s−<sup>1</sup> (1 min), and 50→0 s−<sup>1</sup> (3 min). Obtained data were then fitted to different mathematical models: Newton, Bingham, Ostwald-de Waele, Cross, Casson, and Herschel–Bulkley.

#### *2.6. Gelation Time and Spreadability Test*

Ten milliliters of AmB-gel was added to a transparent vial with a magnetic bar and placed in a low temperature water bath. The solution was then heated to 37 ± 0.1 ◦C while being stirred (400 rpm). The gelation time was measured in triplicate once the magnetic bar stopped moving due to gelation.

The spreadability of AmB-gel was determined in triplicate as follows; 0.5 g of formulation was placed within a 1 cm diameter circle previously marked on a glass plate, after which a second glass plate was subsequently placed without sliding. A series of weights (15, 22, 27, 29, 32, and 37 g) were successively added and allowed to rest for 2 min each at 4 ◦C. The same operation was repeated at 32 ◦C but using different weights (100, 200, and 300 g). The diameters (cm) of the circle spreads were measured and recorded as comparative values. Experimental data were then fitted to mathematical models using GraphPad Prism®® version 6.0 (GraphPad Software Inc., San Diego, CA, USA).

#### *2.7. In Vitro Release and Kinetic Evaluation*

In vitro release studies were performed in vertical diffusion Franz cells (FDC-400, Vidra-Foc, Barcelona, Spain) using nylon membranes.

DMSO solution continuously stirred at 32 ± 1 ◦C was used as the receptor medium to accomplish sink conditions. An aliquot of 1.3 mL of the AmB-gel was placed in the donor phase. Aliquots of 300 μL were collected from the receptor compartment at different times for 24 h and replaced with an equal volume of tempered DMSO kept under stirring at 600 rpm. AmB was quantified by previously validated HPLC [12]. Results are reported as the mean ± SD of six replicates.

Experimental data were fitted to five kinetic models (zero-order, first-order, Peppas–Korsmeyer, Higuchi, and Weibull function) by nonlinear least squares regression using GraphPad Prism®® version 5.01. The best fit was selected based on the Akaike's Information Criterion (AIC) and coefficient determination (*r2*).

#### *2.8. Ex Vivo Permeation Studies: Skin and Vaginal Mucosa*

Ex vivo permeation studies were performed as described in Section 2.7 using skin samples with thicknesses of 400 μm or Vaginal porcine mucosae as membranes. Transcutol P®® was used as the receptor medium which was kept at 32 ± 0.5 ◦C (Human skin) and 37 ± 1 ◦C (Vaginal porcine mucosa). The experiment was performed with a diffusion area of 0.64 cm2. The human skin was obtained from the abdominal region of a healthy woman after abdominal plastic surgery. The experimental protocol was approved by the Bioethics Committee of Barcelona SCIAS Hospital (Barcelona, Spain)

(ref: BEC/001/16) and written informed consent was provided by the volunteer. Integrity of skin was evaluated by measuring the transepidermal water loss (TEWL) with a DermaLab® module (Cortex Technology, Hadsund, Denmark), exhibiting values below 10 g/m2·h.

Vaginal porcine mucosae were obtained under veterinary supervision, from three- to four-month-old pigs after sacrifice using an overdose of sodium thiopental at the Animal Facility at Bellvitge Campus of Barcelona University (Barcelona, Spain) in accordance with protocols prescribed by the Animal Experimentation Ethics Committee of the University of Barcelona, Spain (CEEA-UB). Mucosa was placed into of Hank's balanced salt solution (HBSS) and refrigerated until use. The experiments were carried out for 36 h (Human skin) and 6 h (Vaginal porcine mucosae), respectively.

At the end of the permeation study, the human skin and porcine vaginal mucosa were removed from the Franz diffusion cell, cleaned with a dodecyl sulfate solution 0.05% and washed in distilled water. AmB retained in the skin and vaginal mucosa was extracted with DMSO for a period of 20 min under cold sonication in an ultrasound bath. The resulting samples were measured by HPLC.

#### *2.9. Antifungal E*ffi*cacy*

The minimal inhibitory concentration (MIC), defined as the lowest concentration of an antimicrobial agent that inhibits the growth of a microorganism, was calculated by the broth microdilution method against *C. albicans* ATCC 10231, *C. glabrata* ATCC 66,032, and *C. parapsilosis* ATCC 22,019 strains (American Type Culture Collection, Manassas, VA, USA) following the procedure outlined by the European Committee for Antimicrobial Susceptibility Testing Guidelines (EUCAST) [19] and the Reference method CLSI M27-A3 [20].

This Standards Method provides a valid procedure for testing the susceptibility of glucose-fermenting yeasts to antifungal agents by determining the MIC. A synthetic medium containing RPMI-1640, glutamine, pH indicator without bicarbonate, and glucose 2% *w*/*v*: RPMI-1640 2% G (Invitrogen, Madrid, Spain) was used for working cultures. pH was adjusted to 7.0 with 1 M sodium hydroxide and the resulting solution was filtered using a 0.22 μm filter. As samples were not sterile, chloramphenicol was added to the RPMI medium at a final concentration of 500 μg/mL. The yeast strain was first cultured on Sabouraud Dextrose agar (Invitrogen, Madrid, Spain) at 30 ◦C for 48 h before testing. The inoculums were prepared by suspending colonies in sterile distilled water to achieve a density equivalent to 2 McFarland standards, the counting of which took place in a Neubauer Chamber shortly afterwards (1 to 5 <sup>×</sup> 106 Colony Forming Unit, CFU/mL). Working suspension was prepared by diluting the standardized suspension in sterile distilled water (1:10) in order to prepare 1 to <sup>5</sup> <sup>×</sup> <sup>10</sup><sup>5</sup> CFU/mL. The test was performed using 96-well polystyrene sterile microdilution plates. Serial dilutions from 75 to 0.009 μg/mL of the following forms were compared. (1) Free-AmB (AmB solution first dissolved in DMSO at 300 μg/mL and afterwards in RPMI-1640 double strength at 150 μg/mL). (2) AmB-gel (AmB gel dissolved in RPMI-1640 double strength at 150 μg/mL). (3) Blank-gel (Blank-gel dissolved in RPMI-1640 double strength at 12.5–0.00075%).

Finally, 100 μL of inoculum was added to all the wells. In addition, 3 posts were reserved in each plate for both the positive control (300 μL of the inoculum) and negative control (300 μL of the culture broth). The plates were read at t0 and then at 24 h and 48 h after incubation at 30 ◦C, with a microplate reader model 680 (Bio-Rad, Madrid, Spain) at λ = 620 nm.

#### *2.10. Atomic Force Microscopy (AFM)*

#### 2.10.1. Images

*C. albicans* cells were exposed to AMB-gel and Blank-gel during a period of 4 h. The cells were collected by centrifugation and washed three times with PBS solution (pH 7.40) in order to obtain a final suspension of 1 <sup>×</sup> 106 yeast cells/mL. The suspension was filtered through SMWP01300 Millipore® filters (Merck Chemical and Life Science, S.A, Madrid, Spain) with a pore size similar to the yeast size followed by gentle rinsing with PBS in order to clean the filter from non trapped cells, then subsequently cut and attached to a steel disk using a small piece of double-sided adhesive tape. Finally, the sample was transferred to the AFM liquid cell while avoiding dewetting. Throughout the procedure, MSNL-10, V shaped silicon nitride cantilevers (Bruker AFM Probes, Camarillo, CA, USA) with a nominal spring constant of 0.03 N/m were used.

Prior to measuring the nanomechanics the spring constants of the cantilevers were determined by thermal noise method. AFM images were acquired in contact mode by minimizing the force during the scan and continuously adjusting the set point with a 0◦ scan angle at a scan rate of 1.5 Hz. Images were processed using NanoScope®® analysis software (Bruker, AXS Co., Madison, WI, USA).

#### 2.10.2. Force-Distance Curves

Mechanical properties were measured by recording arrays of 32 × 32 force curves, using a maximum force of 0.5–1 nN to avoid sample damage, a contact time of 100 ms, and approach and retract speeds of 1.0 μm/s. The Young's modulus of the yeast cells was determined as a first approximation by using the Hertz model.

#### *2.11. In Vivo Tolerance Study by Evaluating Biomechanical Properties of Human Skin*

Ten female volunteers between 25 and 35 years old with healthy skin participated in the study. The study was approved by the CEEA-UB according to the recommendations outlined in the Declaration of Helsinki (ref: IRB00003099) and all volunteers provided written informed consent [21]. All participants were requested not to use skin care cosmetics on the test areas of application for two days prior to the study. Skin temperature, transepidermal water loss (TEWL), and stratum corneum hydration (SCH) were determined using a thermometer ST500, a Tewameter TM 300, and a Corneometer CM 825 (Courage-Khazaka electronic GmbH, Cologne, Germany), respectively [22]. The measurements of these parameters were made before applying the AmB-gel (basal readings), immediately after the application of 0.5 mL/cm<sup>2</sup> (t0) and after 2 h of application on the flexor side of the left forearm.

#### *2.12. In Vivo Tolerance Study by Draize Assay and Histological Analysis*

The tolerance of the formulation was evaluated using scarified rabbit skin. The blank-gel and AmB-gel were tested on New Zealand albino male rabbits (2 kg) according to the guidelines provided by the CEEA-UB. The rabbits were acclimated over a 5-day period before the study and were classified into three groups (*n* = 3/group): Group A (Blank-gel), Group B (Skin scarified-control group), and Group C (AmB-gel). The surrounding area of the dorsal trunk was shaved with clippers where a square was drawn for scarification with a razor before beginning the assay. After 30 min, a volume of 0.5 mL of either Blank-gel or AmB-gel was topically applied on the scarified skin of each corresponding group while group B was not exposed to any treatment. This area was protected with gauze and secured with hypoallergenic sticking plaster for 48 h. The signal of edema and erythema were determined after 24 h and 48 h of exposure. Both scores were established according to the degree of severity and the primary irritation index value was calculated. The treatment was classified according the reported specifications: "nonirritant" (<0.5), "irritant" [2–5], or "highly irritant" [5–8,23]. Afterwards, the rabbits were anesthetized and euthanized with sodium pentobarbital.

For histological analysis, the samples of back skin from the rabbits were cut and set up for 24 h in 4% buffered formaldehyde at room temperature. After fixation, all samples were paraffin embedded in paraffin blocks, cut into 5 μm sections, and mounted on microscope slides. Afterwards, the samples were stained with hematoxylin and eosin and finally viewed on blind coded samples under a light microscope (Olympus BX41 and Olympus XC50 camera) with 100× magnification for the evaluation of the tissue structure.

#### *2.13. Statistical Analysis*

Obtained experimental data were analyzed by one-way analysis of variance (ANOVA). Comparisons of findings were done by multiple comparison test. A *p* value < 0.05 was established as an indicator of statistically significant differences (SSD).

#### **3. Results**

#### *3.1. Physicochemical Characterization*

Figure 1, shows SEM micrographs of AmB-gel, which exhibited a heterogeneous structure with the formation of interconnected capillary channels in the form of holes similar to a porous sponge.

**Figure 1.** Morphological characterization by SEM. (**A**) AmB-gel porous sponge structure (×1000). (**B**) AmB-gel tubular appearance interconnected (×25,000).

The swelling process of AmB-gel followed a first-order kinetic model, which was represented by the kinetic constants k = 0.26 min−<sup>1</sup> (*r*<sup>2</sup> = 0.9986) (Figure 2). The degradation process of AmB-gel was completed in 20 min and followed a hyperbola model with a kinetic constant of 0.15 min−<sup>1</sup> (*r*<sup>2</sup> = 0.9975), whereas the percentage of P of AmB-gel was ~82.01 ± 0.5%.

**Figure 2.** (**A**) Swelling ratio and (**B**) percentage of weight loss degradation of AmB-gel.

#### *3.2. Stability Study*

The pH value was suitable for application on the skin, ranging between 6.30 ± 0.05 and 6.70 ± 0.13 (*p* > 0.05) at 4 ◦C, and between 6.50 ± 0.13 and 6.67 ± 0.24 (*p* > 0.05) at 25 ◦C over a period of 90 days. The content of AmB present in the gel stored at 4 ◦C and 25 ◦C remained within the required margins of 90 to 100% during the first 90 days. With regards to the TurbiScan Lab®® analysis, Figure 3 shows the transmission profile of AmB-gel at 25 ◦C over a period of 90 days. The lower part of the vial is

represented on the left side of the graph and the upper part on the right side. The peaks in the lower part and in the upper part of the vial correspond to the meniscus produced by the sample from contact with the glass. No sedimentation, flocculation, or coalescence phenomena were observed.

**Figure 3.** Transmission profile of AmB-gel obtained by Turbiscan Lab®® over 90 days at 25 ◦C. The left side of the curve corresponds to the bottom of the vial, whereas the right side corresponds to the sample behavior on the top.

#### *3.3. Rheological Analysis*

Figure 4A reproduces the flow curves and the viscosity curves obtained from the rheological characterization of AmB-gel. At 4 ◦C, the gel was liquid and showed a Newtonian behavior (constant viscosity under shear rate increase) adjusted by the Newton model (*r*<sup>2</sup> = 1) when under the programmed variations of shear rate. The viscosity at 4 ◦C was found to be 90.62 <sup>±</sup> 7.112 <sup>×</sup> 10−<sup>2</sup> mPa·s. Since AmB-gel transition is carried out as the temperature increases, at 32 ◦C a significantly sticky gel was formed which made it difficult to generate an adequate flow within the space between the conical plate. The rheogram (Figure 4B) showed a critical shear rate of ~10 s<sup>−</sup>1. When above this value, the sample could not flow freely under the experimental conditions. As a result, the layout became quite irregular and the results were inconclusive. However, the valid curve portion showed non-Newtonian behavior with properties of shear thinning (pseudoplastic behavior), thus adjusting the Cross model (*r*<sup>2</sup> = 0.981). The punctual viscosity values at 10 s−<sup>1</sup> and 32 ◦C determined from the ramp up period was ~3.8 Pa·s. The estimated viscosity of the gel at 32 ◦C and 50 s−<sup>1</sup> was ~9.8 Pa·s.

**Figure 4.** AmB-gel rheograms after 24 h: (**A**) Shear stress (Pa) and viscosity (Pa·s) curves at 4 ◦C and (**B**) Shear stress (Pa) and viscosity (Pa·s) curves at 32 ◦C.

#### *3.4. Gelation Time and Spreadability*

Gelation times of AmB-gel stored at 4 ◦C and 32 ◦C were 3 and 1.5 min, respectively. No differences in gelation times were evidenced between samples either with or without AmB. For AmB-gels stored for 3 months, the gelation times were even shorter at less than 1.5 min for 4 ◦C and 1 min 32 ◦C.

Figure 5 shows the results of spreadability test. The Boltzmann sigmoidal model for AmB-gel at 4 ◦C and hyperbola model for AmB-gel at 32 ◦C were the models that best fit the experimental data. At 4 ◦C, spreadability was greatly extended and with lower weight due to having lower viscosity at this temperature.

**Figure 5.** Spreadability charts of AmB-gel at different temperatures: (**A**) Spreadability curve at 4 ◦C; (**B**) Spreadability curve at 32 ◦C.

#### *3.5. Release Studies*

Figure 6 shows the release profile of AmB from the gel, where a sustained release exhibiting S-shaped behavior can be observed. After 24 h, more than 90% of the drug was released. The optimal kinetic model was fitted to the Boltzmann sigmoidal according to the following equation with AIC of 30.05 and *r<sup>2</sup>* of 0.999:

$$Y = Bottom + \left(Top - Bottom\right) / \left(1 + \exp\left(\frac{V50 - x}{Slope}\right)\right) \tag{4}$$

where *Y* is the amount of drug released and *Top* and *bottom* are the initial and final values of drug release. *V50* is the time it takes to release half of the maximum amount susceptible for release and the slope of the curve indicates the steepness.

**Figure 6.** In vitro profile of AmB release from gel. The cumulative amount released was plotted against time. Data represent mean ± SD (*n* = 6).

#### *3.6. Ex Vivo Permeation Studies of Skin and Vaginal Mucosa*

No amount of AmB from gel was detected in the receptor chamber in either skin or vaginal mucosa. Thus, neither flux nor permeation parameters could be calculated. The amount of AmB retained in the skin was 960.0 μg/g/cm2 and vaginal mucosa 737.52 μg/g/cm2.

#### *3.7. Antifungal E*ffi*cacy*

The values from the susceptibility test after 48 h are reported in Table 1 as MIC values against strains of *C. albicans*, *C. glabrata*, and *C. parapsilosis*. It can be observed that AmB-gel exhibited the lowest MIC values (0.09, 0.37 and 0.19 μg/mL) compared with Free-AmB.

**Table 1.** MIC against different cultures of *Candida* spp., Free-AmB, AmB-gel, and Blank-gel after incubation at 30 ◦C for 48 h (*n* = 3).


#### *3.8. Atomic Force Microscopy*

Figure 7 shows the effect of a Blank-gel on *C. albicans* cell. Figure 7A shows the deflection image of a single yeast cell trapped in a pore which appears as the reddish region at the top-left of the image while the cell surface shows an extended region covered by filamentous structures. Figure 7B,C depicts magnified images of topographic and elasticity maps of this region prior to the addition of blank-gel to a final concentration of 6.5% (*w*:*v*) as well as Figure 7D,E after addition. The surface was quite smooth before the addition of the Blank-gel and some filamentous structures could be observed and differentiated by its different Young's modulus values with respect to the cell bulk structure (Figure 7B). After the addition of the blank-gel, the surface of the yeast cell became rougher and no filamentous structures could be observed any longer (Figure 7D). The elasticity map of the region (Figure 7E) evidences that it becomes more homogenous with the incorporation of polymer P407.

Figure 8 shows the effect of AmB-gel on *C. albicans* cells. Deflection, topographic and elasticity images (Figure 8A–C, respectively) of a single *C. albicans* yeast cell trapped in a pore are quite similar to those shown in Figure 6, although the filamentous structures are less evident in this case. The addition of AmB in the gel apparently altered the *C. albicans* cell. Figure 8D demonstrates that two different regions on cell surface can be clearly differentiated: (i) a reddish region (low region) that covers almost the 80% of the cell surface and (ii) a yellowish arched band (high region) crossing the surface from top-left to center-right of the cell surface protruding 40–80 nm over the reddish one. A close inspection of this yellowish band revealed a structure resembling a wrinkled sheet. Figure 8E,F shows the topographic and elasticity maps of the low region of Figure 8D. From these images, it is possible to recover the filamentous structures previously observed, although the differences in the Young's modulus values between the filaments and the bulk structure are not overly distinctive. Figure 8G,H shows the topographic and elasticity maps of the high region of Figure 8D. From the images, it is possible to observe two differentiated regions: (i) one diagonal from top-left corner to the bottom-right corner of Figure 8H (darker region) and (ii) a second region formed from the higher Young's values (lighter region) in Figure 8H occupying the bottom-left and top-right corners of the image with higher Young's modulus values than those of the low region. Considering that the darker region in Figure 8H is the higher region of the topographic image (Figure 8G), it is possible to identify these "wrinkled sheets" as regions with low elasticity modulus, thus suggesting that material in these structures is not as packed as those with blank-gel.

**Figure 7.** Effect of the Blank-gel on *C. albicans* cells. (**A**) Deflection image of a single yeast cell trapped in a pore. (**B**) Topographic map zoom image of the yeast cell before application of the Blank-gel. (**C**) Elasticity map zoom image of the yeast cell before application of the Blank-gel. (**D**) Topographic map zoom image of the cell yeast after application of the Blank-gel. (**E**) Elasticity map zoom image of the yeast cell after application of the blank-gel.

**Figure 8.** Effect of AmB-gel on *C. albicans* cells. (**A**) Deflection image of a single yeast cell trapped in a pore. (**B**) Topographic map zoom image of a single yeast cell trapped in a pore. (**C**) Elasticity map zoom image of a single yeast cell trapped in a pore. (**D**) A reddish region (low region) that covers almost the 80% of the cell surface and a yellowish arched band (high region) crossing the surface from top-left to center-right of the cell surface. (**E**) Topographic map zoom image of the low region. (**F**) Elasticity map zoom image of the low region. (**G**) Topographic map zoom image of the high region. (**H**) Elasticity map zoom image of the high region.

Conversely, Figure 9 shows the profiles of the elasticity values as a function of its frequency. Figure 9A represents the individual Young's modulus values as a function of its frequency before and after the addition of blank-gel. Before the addition of blank-gel, the yeast cell showed a wide distribution of Young's modulus values with a central value of 21.0 ± 0.4 MPa. After the addition of the blank-gel, the values distribution is narrower and centered at a higher value 26.29 × 0.08 MPa. Figure 9B shows a central peak around 18.48 MPa and a wide distribution function. The low region depicted in Figure 8D corresponds to a narrower peak with a Young's modulus mean value that is shifted towards smaller values (15.63 MPa). On the contrary, the protruding structure resembling "wrinkled sheets" in Figure 8D shows a bimodal distribution, one close to the values of the blank *C. albicans* cell with a mean Young's modulus value of 18.10 MPa and another shifted towards smaller values of 16.62 MPa.

**Figure 9.** Individual Young's modulus values as a function of its frequency on *C. albicans* cells. (**A**) Individual Young's modulus values before and after the addition of Blank-gel. (**B**) Individual Young's modulus values before and after the addition of the AmB-gel.

#### *3.9. In Vivo Tolerance Study by Evaluating Biomechanical Properties of Human Skin*

The results of evaluated biomechanical parameters are depicted in Figure 10. No statistically significant changes of TEWL or skin temperature were observed between Blank-gel and AmB-gel or when compared with baseline measurements (*p* > 0.05). However, a significant decrease in SCH values

was observed in the Blank-gel immediately after application (0 h) and at 2 h as well as in the AmB-gel after 2 h, with respect to the basal state (Figure 10E,F).

**Figure 10.** Biomechanical parameters evolution monitored before the application (basal), immediately after application (0 h) and 2 h after application. (**A**) TEWL values of blank-gel. (**B**) TEWL values of AmB-gel. (**C**) Skin temperature values of blank-gel. (**D**) Skin temperature values of AmB-gel. (**E**) SCH values of blank-gel. (**F**) SCH values of AmB-gel. \* Statistically significant differences (*p* < 0.05).

#### *3.10. In Vivo Tolerance Study on Scarified Rabbit Skin by Draize Assay and Histological Analysis*

The Draize test was carried out in order to evaluate the skin irritation potential of AmB-gel. After 48 h, the resulting primary irritation index value for Blank-gel and AmB-gel was 0.38 and 0.45, respectively. This result indicates that both blank-gel and AmB-gel are nonirritants.

Regarding the histological evaluation, micrographs revealed that scarification caused histological alterations and the presence of nonspecific inflammatory cells in the skin (Figure 11B) while the topical application of Blank-gel (Figure 11A) and AmB-gel (Figure 11C) notably repaired these alterations, resulting in a less pronounced inflammatory process than that of the control group (Figure 11B).

**Figure 11.** Optical microscopic images of skin. Blank-gel (**A**), skin-scarified control group (**B**), and AmB-gel (**C**). Hematoxylin and eosin stains nuclei blue/black while keratin and cytoplasm are stained red. Scale bar = 200 μm.

#### **4. Discussion**

P127 gels are used as the vanguard of drug delivery systems development due to the improved therapeutic efficacy and adherence to the recipient [24]. They have been used as vehicles for various routes of administration of drugs, including oral and topical [25], intranasal [18,26], vaginal and rectal [27], ocular [28], and parenteral [29]. In this study, AmB-gel was developed with 5% of DMSO as an optimal solubilizing agent and 25% of P127 to confer thermoreversible character. For the physicochemical characterization, SEM images were obtained, thus confirming that AmB-gel has a dense and well-oriented tubular appearance along with reticular networks. This type of structure has been reported in other studies [30,31], therefore supporting the postulation that a porous structure could help in the controlled release of drug.

The swelling of AmB-gel is a parameter dependent on the critical micelle concentration (CMC), given that P127-gels show two forms: gel form at 25 ◦C and liquid form at 4 ◦C, with liquid form showing thermoreversible behavior. Once the AmB-gel is poured into the swelling medium (PBS), a concentration of P407 decreases rapidly below CMC, thus producing swelling and solubilization in ~19 min. In the present study, 25% of P127 was used and at this concentration the SR followed a first-order kinetic model. Additionally, the degradation process was complete at 20 min, which is a typically rapid rate of degradation for these kinds of systems, as previous studies have also found [32]. On the other hand, the AmB-gel percentage of P was about 82.01 ± 0.5% corroborated with the SEM image which showed a high porosity structure.

The pH value of skin formulations is an important factor to consider in order to avoid skin irritation, particularly when under mycosis infections. AmB-gel showed pH levels to be slightly acidic, therefore biocompatibility with the natural acidity of the skin assures suitability for skin application [30]. What is more is that the AmB content in the gel was maintained between 90 and 100% over a span of 3 months while the superimposed graphs of transmission (Figure 3) showed variations below 10%, which indicates that the product will remain physically stable for a minimum of three months [33–35].

As expected, the rheological characteristics of AmB-gel were temperature-dependent (Figure 4). It is vital to evaluate rheological characterization because rheology can modulate biopharmaceutical properties such as release rates in addition to the widespread application on affected areas [36]. The viscosity at 4 ◦C was fairly low at ~90 mPa·s, which is an important value at the time of packaging because greater liquescence can lead to optimal forms of delivery, such as spray or roll-on applications, which help prevent the spread of mycotic infections. Once in contact with the affected skin at 32 ◦C, the solubility of the PPO chains decreases, subsequently forming micelles and assuming hexagonal and/or cubic structures, thus drastically increasing viscosity. The estimated viscosity of the AmB-gel at <sup>32</sup> ◦C and 50 s−<sup>1</sup> was ~9.8 Pa·s, which is nearly 100 times higher than the viscosity at 4 ◦C. This thermal gelation of P407 can form and sustain a drug depot in this area and subsequently increase the contact time, which can then produce desirable drug release characteristics and prolong pharmacological action [37]. The thermal gelation process did not seem to be affected by the presence of AMB. All the results suggested that gel state would be reached after application on the skin and also gain viscosity instantaneously to increase residence time of the formulation on the application area [13].

The data obtained from release studies demonstrated that the AmB-gel provided a sustained release for approximately 24 h. Model fitting showed that release mechanism followed a sigmoidal model. This faster release could be attributed to the highly porous structure and the rapid degradation of P407-gel. Moreover, this mathematical model has been associated with these types of gels due to the fact that they are used to simulate the transport and reaction of fluid in porous media [38].

Ex vivo permeation studies revealed that AmB did not permeate through human skin nor porcine vaginal mucosa. Similar results were reported in previous studies in which no AmB from a nanoemulsion was detected in the ex vivo permeation study using human skin or pig's ear skin [12,39], which might be explained by the great molecular weight (926 D) and high hydrophobicity of AmB, effectively limiting its ability to pass through the aqueous structure of the vaginal mucosa and the hydrophilic barrier of the dermis. However, it was observed that AmB had high retention in the skin and vaginal mucosa with values of 960 μg/g/cm<sup>2</sup> and 737 μg/g/cm2, respectively. This result confirms that AmB succeeded in crossing the SC and was distributed to both the epidermis and dermis. Furthermore, it was able to permeate into the vaginal mucosa without reaching the systemic circulation, which was confirmed with the high amount of drug retained in the tissue and is indicative of the formulation favoring a local effect on the target area with no side effects. This high drug retention capacity in both tissues could be due to the presence of P407 in the formulation, since the chemical structure of this polymer allows it to act as a surfactant that enhances the diffusion ability of drug, resulting in increased drug concentrations in the tissue layer that leads to efficiently provide a local effect [18]. These results suggest that the AmB in this formulation could be successfully implemented in order to achieve a local effect on the skin or vaginal mucosa without adverse systemic effects.

The antifungal action of AmB-gel was clearly observed (Table 1) with a MIC value lower than Free-AmB in all tested Candida species and thus was highly effective—essentially confirming the adequacy of the formulation due to its improved effectiveness against fungus and yeast infections, which is likely due to the synergistic effects reported between the formulations of poloxamers and antifungal agents [40]. On the other hand, Blank-gel did not have any effect against Candida strains and produced values consistent with those reported in previous studies [12]. These results were subsequently observed by AFM images (Figures 7 and 8) which evidenced the alterations on the surface of Candida yeast cells that were induced by AmB-gel. In an equivocal manner, the mean values of Young's modulus (Figure 9) demonstrate the destructuring effect of AmB-gel on the outer membrane of the yeast cell.

The tolerance of the formulation was studied by evaluating the biomechanical properties of skin, since parameters such as TEWL and SCH are important indicators of skin integrity. Thus, any changes in these parameters could be directly related to an alteration of the functionality of the skin barrier [41]. From the results of this noninvasive in vivo method, no statistically significant differences were observed in temperature or TEWL values of volunteers after treatment with AmB-gel or with blank-gel when compared with baseline measurements (Figure 10), whereas SCH values showed a significant reduction in both AmB-gel and blank-gel with respect to the basal state. This decrease in the hydration levels in the area of application of the final product could be due to the ability of the gel to absorb moisture produced by the sweat glands. When the formulation containing P407 dries on the surface of the skin, it forms a film that captures moisture and does so in a state of relative equilibrium with ambient humidity. This effect would be especially suitable in skin candidiasis because it could prevent the proliferation of microorganisms around the affected area [42,43]. Moreover, when comparing the values of the biomechanical parameters evaluated for AmB-gel and blank-gel, the incorporation of the AmB does not significantly modify the properties of the gel. Furthermore, none of the volunteers exhibited side effects of burning or itching after the application. Therefore, the AmB-gel did not alter the biophysical properties of the skin of volunteers throughout the course of the study.

Finally, the tolerability of the formulation was evaluated by in vivo model using scarified rabbit skin in order to simulate the damage of the skin barrier caused by the fungal infection. The irritation index values for AmB-gel and Blank-gel were less than 0.5 and can therefore be classified as nonirritant formulations. This result was consistent with histological analysis, which showed that both formulations did not cause any alteration of the skin architecture but, on the contrary, they actually repaired the ulcerated skin. According to this finding the ability of P407 to reduce wounds in burn patients while simultaneously stimulating the proliferation of collagen fibers, consequently increasing scarring and generating new tissue, have all been reported in previous studies [44]. In conjunction, these effects might favor the healing and alleviation of wounds caused by Candida. In conclusion, AmB-gel represents a promising therapeutic option to treat these types of infections via dermal application.

#### **5. Conclusions**

The focus of the present study was on the incorporation of AmB in a copolymer P407-based gel designed for application on both the skin and vaginal mucosa, specifically for the treatment of candidiasis. P407 has shown great potential as a support system in the administration of hydrophilic and hydrophobic drugs, as is the case of AmB. P407 is suitable due to its thermoreversible capacity, low toxicity, and permeation-enhancing properties. During the experimental phase, AmB-gel remained stable for at least 3 months, thereby demonstrating pH values suitable for both skin and vaginal application. Due to its low viscosity at predetermined temperatures and its porous structure which facilitates rapid drug release, AmB-gel can also be used in aerosol form in order to avoid contact with areas infected by the fungus and thus prevent the spread of infection to other anatomical sites while minimizing the risk of a subsequent superinfection. No drug was quantified in the receptor compartments of the Franz cells, which indicates nonabsorption of AmB and therefore avoids adverse systemic effects. None of volunteers exhibited burning or itching after application and the histological images of rabbit skin displayed high tolerability of the formulation. Furthermore, AmB-gel did not alter the biophysical properties of skin, thus confirming adequate safety as topical agents. According to the values of the amount of drug retained in the skin and vaginal mucosa, along with the MIC values and the images obtained by AFM, it can be concluded that AmB-gel provides a satisfactory antifungal effect across various Candida species.

**Author Contributions:** L.S. carried out the experiments and analyzed and wrote the results of the manuscript. A.C.C. conceived the idea and analyzed the results. M.S.-A. analyzed the results and wrote and edited the manuscript. M.R. and L.C.E. participated in the methodologic analysis and editing of the manuscript. M.J.R.-L. realized the histological studies. N.B. carried out the microbiological efficacy. O.D. performed the force atomic experiments and analyzed the results. B.C. conceived the idea and edited the manuscript. All authors read the manuscript and agreed with the submission.

**Funding:** This research was funded by L'Agència de Gestiód'Ajuts Universitarisi de Recerca (AGAUR) grant number [SGR-2017 1744]. Lilian Sosa also expresses gratitude for the grant awarded by the Institute of Nanoscience and Nanotechnology IN2UB number [2017.3.IN2UB.2] in the completion of her doctoral thesis. Marcelle Silva-Abreu acknowledges the support from the Coordination for the Improvement Personnel (CAPES-Brazil).

**Acknowledgments:** We appreciate the help of Jonathan Proctor for the revision of the use of the English language. The authors are grateful to Lyda Halbaut for the help with rheological studies.

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Repurposing Butenafine as An Oral Nanomedicine for Visceral Leishmaniasis**

#### **Adriana Bezerra-Souza 1, Raquel Fernandez-Garcia 2, Gabriela F. Rodrigues 1, Francisco Bolas-Fernandez 3, Marcia Dalastra Laurenti 1, Luiz Felipe Passero 1,4, Aikaterini Lalatsa <sup>5</sup> and Dolores R. Serrano 2,\***


Received: 30 June 2019; Accepted: 18 July 2019; Published: 20 July 2019

**Abstract:** Leishmaniasis is a neglected tropical disease affecting more than 12 million people worldwide, which in its visceral clinical form (VL) is characterised by the accumulation of parasites in the liver and spleen, and can lead to death if not treated. Available treatments are not well tolerated due to severe adverse effects, need for parenteral administration and patient hospitalisation, and long duration of expensive treatments. These treatment realities justify the search for new effective drugs, repurposing existing licensed drugs towards safer and non-invasive cost-effective medicines for VL. In this work, we provide proof of concept studies of butenafine and butenafine self-nanoemulsifying drug delivery systems (B-SNEDDS) against *Leishmania infantum*. Liquid B-SNEDDS were optimised using design of experiments, and then were spray-dried onto porous colloidal silica carriers to produce solid-B-SNEDDS with enhanced flow properties and drug stability. Optimal liquid B-SNEDDS consisted of Butenafine:Capryol 90:Peceol:Labrasol (3:49.5:24.2:23.3 *w*/*w*), which were then sprayed-dried with Aerosil 200 with a final 1:2 (Aerosil:liquid B-SNEDDS *w*/*w*) ratio. Spray-dried particles exhibited near-maximal drug loading, while maintaining excellent powder flow properties (angle of repose <10◦) and sustained release in acidic gastrointestinal media. Solid-B-SNEDDS demonstrated greater selectivity index against promastigotes and *L. infantum*-infected amastigotes than butenafine alone. Developed oral solid nanomedicines enable the non-invasive and safe administration of butenafine as a cost-effective and readily scalable repurposed medicine for VL.

**Keywords:** butenafine; SNEDDS; solid SNEDDS; spray drying; leishmaniasis; design of experiments

#### **1. Introduction**

Leishmaniasis is an infectious disease caused by parasites belonging to the *Leishmania* genus. The prevalence of leishmaniasis exceeds 12 million cases, and it is endemic in 98 countries in five continents. *Leishmania* parasites are transmitted by insect vectors from the genus *Lutzomyia* sp. or *Psychodopygus* sp. in the New World and *Phlebotomus* sp. in the Old World [1,2]. Leishmaniasis presents in the cutaneous (CL) and visceral (VL) leishmaniasis forms, depending on the type of host immune response and infecting parasite species [3,4]. In the New World, parasites of the subgenus *Vianna*

cause only CL and mucocutanoues leishmaniasis (MCL) while parasites of the subgenus *Leishmania* are responsible for CL and VL [5]. VL is a chronic disease caused by *L. (L.) infantum* and *L. (L.) donovani* species [6–8] residing in host macrophages, mainly from spleen, liver, bone marrow, and lymph nodes, and is characterized by prolonged fever, hepatosplenomegaly, lymphadenopathy, anemia with leukopenia, hypergammaglobulinemia and hypoalbuminemia, weight loss, edema, and a debilitating state leading to weakening and ultimately death if untreated [6].

VL treatment mainstays involve pentavalent antimonials (SbV) and amphotericin B (AmB) as the first line in the developing and developed world, respectively [9]. Although they are both highly efficacious in vivo, SbV are linked to severe and frequent side effects, limiting their use [9], combined with a high rate of clinical resistance [10]. AmB, on the other hand, has shown limited resistance, but its clinical use is limited by the high cost, especially for safer lipidic formulations, such as Ambisome®, thermal instability, and nephrotoxicity [11,12]. Miltefosine, the only oral VL licensed treatment (licensed in 2003 in India), has several limitations as monotherapy, as its activity is highly dependent on the clinical form of leishmaniasis and the parasite strain, and patients frequently experience severe gastrointestinal disorders. In addition, it is teratogenic, and thus cannot be used in women of child-bearing age [13,14]. Miltefosine has a long elimination half-life (seven days) and a narrow therapeutic index, characteristics that limit the administered dose, which can lead to subtherapeutic levels over several weeks, encouraging the emergence of resistance [12,14,15]. Indeed, resistance has been reported in India (a country that alone accounts for 50% of the VL worldwide burden) and France [12]. Thus, available medicines for VL are outdated, impractical, insufficiently efficacious, or subject to resistance and unacceptable toxicities. Ideal treatments for VL should be able to possess greater than 95% efficacy, be orally administered, be stable in a tropical environment, be affordable, and require fewer than 11 days of treatment with an excellent side-effect profile [12,16].

Repurposing drugs is a strategy that has been crowned with success, as repurposed drugs have made up a third of all new commercially and clinically used drug treatments since 2009 [17,18]. Current VL drugs are interesting examples of repurposed drugs, since SbV, AmB, and miltefosine were originally used as an emetic, antifungal, and anticancer drug, respectively [14,19]. Butenafine is an allylamine drug, commonly employed in the treatment of fungal skin infections, such as ringworm, athlete's foot, jock itch, and pityriasis. Currently, it is only commercialised as a 1% cream for topical use. We have shown recent reports on how butenafine repositioning was effective in vitro against *L. (L.) amazonensis* and *L. (V.) braziliensis*, ethiological agents of CL and MCL. Furthermore, parasite treated with butenafine showed morphological alterations that resembled programmed cell death, which is attributed to the blockage of the biosynthesis of ergosterol [20]. Butenafine has limited oral bioavailability, with 1.5–3% of the oral dose being recovered in the plasma an hour after a single oral dosing of radiolabeled butenafine (0.2 mg/kg) in rats [21]. Butenafine is highly metabolized in the liver (methylation, dealkylation, and hydroxylation) and only 0.03% of the oral dose has been recovered intact from the plasma after 4 h [21]. This concurs with levels of its major metabolite (1-napthoic acid) in the plasma, which ranged between 1%–100% of administered parent drug dose [21].

Here, we are reporting the development of an oral butenafine nanomedicine able to enhance its aqueous solubility, maintaining a solubilized state in the gastrointestinal to allow for enhanced oral absorption, in order to target the liver and spleen (i.e., organs where the *Leishmania* parasite resides in high concentration). We have shown that SNEDDS are able to enhance the oral bioavailability of poorly soluble drugs and enable therapeutic concentrations to be delivered in the liver and spleen [12]. Thus, we hypothesised that if butenafine is formulated with GRAS (Generally Regarded as Safe) excipients with known activity against different *Leishmania* strains [22], we can develop butenafine-loaded SNEDDS (B-SNEDDS) and solid SNEDDS (solid B-SNEDDS) with enhanced activity, as well as being able to maintain butenafine's oral solubilisation capacity in the gastrointestinal tract. To ensure that a stable, and ideally solid, cost-effective formulation is available, we used design of experiments (DoE) studies to prepare butenafine SNEDDS colloidal silicon dioxide spray-dried particles that can be easily compressed into cost-effective, easily scalable, solid dosage forms of a repurposed drug for VL.

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

#### *2.1. Materials*

Butenafine hydrochloride (purity ≥ 98%) was purchased from Cayman Chemical Co. (Michigan, MI, USA). SNEDDS excipients (Capryol 90 (propylene glycol monocaprylate), Labrafil M 1944 CS (oleoyl polyoxyl-6 glycerides), Labrasol (caprylocaproyl polyoxyl-8 glycerides), and Peceol (glyceryl monooleate)) were kindly donated by Gattefosse (Saint-Priest Cedex, France). Two Aerosil silicon dioxide excipients were used as inert solid carriers: Aerosil R972 from Degussa (Frankfurt, Germany) and Aerosil 200 from Evonik Industries (Darmstadt, Germany). All other chemicals were of ACS reagent grade (Sigma Aldrich, Madrid, Spain) and were used as supplied. Solvents were of HPLC grade (Fisher, Madrid, Spain).

#### *2.2. Solubility Studies of Butenafine*

An excess quantity of butenafine was added to each of the excipients used in the preparation of SNEDDS, and the mixture was shaken (300 rpm) overnight at 25 ◦C in triplicate. The mixtures were centrifuged at 3000 rpm for 5 min, and the supernatant (0.1 mL) was diluted with 10 mL of methanol. The absorbance was measured in a spectrophotometer (JASCO V-730 spectrophotometer Madrid, Spain) at 220 nm to determine the solubility of butenafine. A calibration curve was performed previously in methanol to establish the linearity between concentration and absorbance at 220 nm.

#### *2.3. Pseudo-Ternary Phase Diagrams*

Ternary phase diagrams were constructed to study the phase behaviour of oils/surfactants over the whole concentration range. The existence of a microemulsion (type II) region within this diagram was observed visually. A D-Optimal design was developed using different mixtures of Capryol 90, Labrasol, and Peceol by using Design Expert software (State Ease, Minneapolis, MN, USA). Mixtures were vortexed for 5 min. The particle size distribution (PSD) was measured after dilution in deionised water (1:1000 *v*/*v*) in a Microtrac Zetatrac (Microtrac, Montgomeryville, PA, USA). The optimal excipient combination leading to the smallest PSD upon dilution was selected, in order to solubilise the drug and perform further experiments.

#### *2.4. Preparation of Liquid B-SNEDDS Formulations*

Based on solubility and phase diagram studies, Labrasol, Capryol 90, and Peceol were selected as a high-hydrophilic-lipophilic balance (HLB) surfactant, a medium-HLB surfactant, and an oil, respectively. Optimal SNEDDS were prepared combining the three as Capryol 90:Labrasol:Peceol (51:24:25 *w*/*w*). Butenafine (30 mg/g) was solubilised in the resulting excipient mixture, which was stirred overnight in order to obtain a homogenous isotropic mixture.

#### *2.5. Preparation of Solid B-SNEDDS Formulation*

A mini-spray dryer (Büchi B-191) was employed for the preparation of solid B-SNEDDS. Aerosil 200 (hydrophilic fumed amorphous silica, 5–50 nm (Evonik industries, Darmstadt, Germany)) or Aerosil R972 (hydrophobic fumed amorphous silica, 16 nm (Degussa AG, Frankfurt, Germany)) were used as inert carriers. A carrier (1 g) was dispersed in 100 mL of ethanol by magnetic stirring, after which liquid B-SNEDDS formulation (0.5 g) was mixed. The resulting suspension was delivered to a two-fluid nozzle (0.7 mm nozzle tip and a 1.5 mm diameter nozzle screw cap) using a peristaltic pump, at a speed of 10% (5 mL/min). Compressed air (2 bars) was used as the drying gas in a co-current mode, with the aspirator capacity set to maximum (100%). The flow-meter for the standard two-fluid nozzle was set to 600 NL/h. The inlet temperature was fixed at 62 ◦C, and the outlet temperature varied

between 32 and 36 ◦C. After spray-drying, the dry powder was collected from the collector vessel, and the yield of the process was quantified using the following equation:

$$\text{Yield } (\%) = \frac{\text{Colllected mass after sprny drying (mg)}}{\text{Total mass sprny drid } (\text{mg})} \times 100\tag{1}$$

A 22 DoE was performed to understand and optimise key formulation parameters affecting the preparation of solid B-SNEDDS. The surface properties of silica used, i.e., type of Aerosil (200 or R972), as well as the weight ratio between Aerosil and liquid B-SNEDD (1:2 or 1:3 *w*/*w*) on the physicochemical properties of the spray-dried product were investigated. Five different responses were studied: (i) powder flow, (ii) yield, (iii) drug loading, (iv) particle size upon dilution (1:1000 *w*/*w*), and (v) the percentage of drug release at 60 min in simulated gastric fluid (500 mL buffer solution of pH 1.2).

#### *2.6. Characterisation of the Solid B-SNEDDS*

#### 2.6.1. Powder Flow: Angle of Repose (AoR) Measurements

The angle of repose (AoR) was determined according to the United State Pharmacopeia (<1174> Powder Flow) by using the fixed height funnel method (i.e., by measuring the cone height versus the base, formed by the powder falling through a plastic funnel placed 10 cm from the table surface until a stable cone was formed) [23]. AoR measurements were carried out by passing 500 mg of solid B-SNEDD in triplicate. After the powder was deposited on the surface, the height of the powder cone (*h*) and the radius of the base (*r*) were measured. The inclination of the cone created between the powder and the surface (α, AoR) was calculated using the following equation:

$$
\tan(a) = \frac{h}{r} \tag{2}
$$

#### 2.6.2. Particle Size Measurements

The particle size of both liquid B-SNEDDS and solid B-SNEDDS were determined at 25 ◦C after 1:1000 *w*/*w* dilution in deionised water. Solid B-SNEDDS were centrifuged (9000 rpm, 5 min) prior to measurements, in order to precipitate the insoluble carrier. The mean size (*n* = 3) by volume (nm) was measured using a Microtrac Zetatrac (Microtac Inc., Montgomeryville, PA, USA), with an internal probe ranging from 0.0008 to 6.5 μm [24].

#### 2.6.3. Release Studies

Release studies of solid B-SNEDDS (500 mg) containing 10 mg of butenafine were performed at 37 ◦C for 1 h in simulated gastric fluid (SGF; 500 mL buffer solution of pH 1.2), followed by a second consecutive hour in simulated intestinal fluid (SIF; 400 mL buffer solution of pH 6.8) in a calibrated dissolution apparatus (Erweka type DT80, Erweka, Heusenstamm, Germany). The SGF and SIF were prepared as described in the USP [25]. Based on predicted water solubility values (7.5 <sup>×</sup> 10−<sup>5</sup> mg/mL, according to the Drug Bank Database [26]), release studies were performed in non-sink conditions, with the aim of testing a relevant pharmacological drug dose (10 mg). At different time points (5, 10, 15, 30, 45, 60, 90, and 120 min), a 2 mL sample was withdrawn, filtered through a Millipore Millex PTFE membrane filter (0.45 μm), diluted 1:2 with acetonitrile, and injected in the HPLC. Butenafine concentration was quantified using an HPLC equipped with a Jasco PU-1580 pump, a Jasco AS-2050 Plus autosampler, and a Jasco UV-1575 UV-visible detector. Integration of the peaks was performed using the Borwin 1.5 software. Butenafine was separated on an Agilent Eclipse XDB-Phenyl reverse-phase column (250 mm × 4.6 mm, 5 μm). The mobile phase consisting of methanol/water (78:22 *v*/*v*) was pumped at 1.4 mL/min, and a sample injection volume of 20 μL was used. The column

temperature was kept at 25 ◦C, and the detector was set at 220 nm. Butenafine concentrations were calculated from a linear regression calibration curve between 100.0–0.1 μg/mL.

#### 2.6.4. Drug Loading

Solid B-SNEDD formulations (10 mg) were dissolved in 0.5 mL of dimethyl sulfoxide (DMSO), vortexed for 5 min, and then diluted in acetonitrile (1:40 *v*/*v*) prior to drug quantification by the above-described HPLC method.

#### 2.6.5. Morphological Analysis

The morphology of solid B-SNEDDS was examined using a scanning electron microscope (JEOL JSM 6335F Ltd., Akishima, Japan) at 5 kV. The samples were fixed on a brass stub using double-sided adhesive tape, and vacuum-coated with gold (Au) for 180 s (coater: Q150R S, Quorum, Lewes, East Sussex, United Kingdom).

#### 2.6.6. Tabletting and Hardness

In order to investigate the compression of the solid B-SNEDDS without the addition of any other excipient, solid B-SNEDDS (500 mg) were compressed using a Perkin Elmer hydraulic press (Waltham, MA, USA) and a 13 mm punch, and die set under different pressures of 0.5, 1.0, 3.0 or 5.0 tonnes for 15 s. Hardness was undertaken according to the European Pharmacopeia [27], using a Pharma Test PTB 311 instrument (Pharma Test, Hainburg, Germany). Tablets (*n* = 3) were individually evaluated, and the mean value of the force in Newtons (N) was reported.

#### *2.7. In Vitro E*ffi*cacy and Toxicity Studies*

#### 2.7.1. Parasites and Cell Lines

The parasites were kindly provided by Prof. Dr. Fernando Tobias Silveira from the cryobank of the Leishmaniasis Laboratory of Prof. Dr. Ralph Laison, Department of Parasitology, Ministry of Health, Evandro Chagas Institute (Belem, Para, Brazil). They were identified using monoclonal antibodies and isoenzyme electrophoretic profiles at the Leishmaniasis Laboratory of the Evandro Chagas Institute. The *Leishmania* species used was *L. (L.) infantum* (MHOM/BR/72/46). Parasites in a late log stage were used for all experiments. Parasites were maintained in Schneider's Medium (SigmaAldrich, Madrid, Spain), supplemented with 10% heat-inactivated fetal bovine serum, 50,000 IU/mL penicillin, and 50 μg/mL streptomycin.

BALB/c mice, 6 weeks old, were obtained from the Medical School of the University of São Paulo, Brazil, in order to collect peritoneal macrophages to perform the in vitro test against intracellular *Leishmania* amastigotes. To obtain the macrophages, the animals were anaesthetized with thiopental (1 mg/200 mL) and euthanized. This study was carried out in strict accordance with the recommendations detailed in the Guide for the Care and Use of Laboratory Animals of the Brazilian National Council of Animal Experimentation (http://www.cobea.org.br). The protocol was approved by the Ethics Committee of Animal Experiments of the Institutional Committee of Animal Care and Use at the Medical School of Sao Paulo University (CEUA-FMUSP number 098/17).

#### 2.7.2. In Vitro Promastigote Efficacy and Cytotoxicity

Promastigote forms of *L. (L.) infantum* (2 <sup>×</sup> 10<sup>7</sup> promastigotes/well) were incubated in a 96-well culture plate in RPMI 1640 medium, with pH 4.2 and drugs in a range of 0.01 to 400 μM. The negative control group was cultivated in medium and vehicle solution (PBS plus 1% DMSO). The parasites were incubated for 48 h at 25 ◦C, and washed with 200 μL of sodium chloride 0.9% (*w*/*v*) three times with centrifugation at 3000 rpm for 10 min at 4 ◦C, followed by the addition of MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (9.6 μM). Four hours later, 50 μL of 10% sodium dodecyl sulphate (SDS) was added to each well. The plates were further incubated for 18 h and read in an ELISA reader (Labsystems Uniscience Multiskan EX, Miami, FL, USA) at 595 nm. Effective concentration 50% (EC50) was estimated using Graph Pad Prism 5.0 software (GraphPad Software, San Diego, CA, USA).

Approximately 5 <sup>×</sup> 10<sup>5</sup> peritoneal macrophages from BALB/c mice were cultured in RPMI 1640 medium, with the drugs in a range of 0.01 to 400 μM, in 96-well plates. As a negative control, macrophages were cultivated with vehicle solution. After 48 h, cell viability was analysed by the MTT method. Cytotoxic concentration 50% (CC50) was estimated with Graph Pad Prism 5.0 software.

The selectivity indexes (SI) were calculated using the ratio CC50/EC50 toward promastigote (SIp) or amastigote (SIa) forms. EC50 represents the concentration of the formulation that produced a 50% reduction in parasites, while CC50, represents the concentration of the formulation that produced a 50% reduction of cell viability in treated culture cells with respect to untreated ones.

#### 2.7.3. Macrophage Infection and Treatments

Peritoneal macrophages from BALB/c mice (5 <sup>×</sup> 105 macrophages) were cultivated in round cover slips in a 24-well plate, followed by infection with *L. (L.) infantum* promastigotes at a ratio of 10 parasites per 1 peritoneal macrophage. Plates were incubated at 5% CO2 at 37 ◦C. After 24 h of the initial infection, drugs were added at 25, 50, or 100 μM. Round cover slips from each experimental time point were dried at room temperature, fixed in methanol, and stained by Giemsa 5% (two drops). The infection index (II) was then estimated according to Passero et al. [28], using the following equation:

$$
\Pi = \% \text{ infected macropphases} \times \frac{\text{Internal amastigotes}}{\text{Macrophases}} \tag{3}
$$

#### *2.8. Statistical Analysis*

Statistical analyses were performed via a one-way ANOVA test using Minitab 15 (Minitab Ltd., Coventry, UK), followed by Tukey's test. Statistical significance was set at *p* < 0.05. Data was plotted using Origin X9 (Northampton, UK).

#### **3. Results**

#### *3.1. Solubility Studies of Butenafine*

The solubility of butenafine was tested in four different excipients (Peceol, Capryol 90, Labrasol, and Labrafil M1944CS) that are commonly utilized in the development of lipid-based formulations and have been shown to possess efficacy against leishmaniasis [22] (Table 1). Butenafine was more soluble in excipients with oleic acid lipids or triglycerides, i.e., Peceol and Labrafil M 1944 CS, followed by lower HLB excipients such as Capryol 90. Due to the low miscibility between Peceol and Labrafil, and the higher butenafine solubility in Peceol, Peceol was chosen as the oil phase to be combined with Capryol and Labrasol for the pseudo-ternary diagram and identification of the optimal composition (Table 2) able to yield microemulsions (type II) upon aqueous dilutions.

**Table 1.** Solubility of butenafine in various vehicles at 25 ◦C (*n* = 3).


\* Hydrophilic-lipophilic balance of the vehicles; values obtained from Gatefosse website [29].


**Table 2.** Design of experiments (DoE) of liquid butenafine self-nanoemulsifying drug delivery systems (B-SNEDDS). Excipient quantities are expressed as a fraction, considering that the sum of all excipients was equal to 1 g. For each combination, the average particle size (*n* = 3) in numbers after dilution in de-ionised water (1:1000 *w*/*w*) was illustrated.

#### *3.2. Pseudo-Ternary Phase Diagrams and Preparation of Liquid B-SNEDDS Formulations*

A ternary phase diagram (Table 2 and Figure 1) was constructed to study the phase behaviour of oil/surfactants over the whole concentration range. Particle size expressed in numbers led to a better predictive model with a higher R2 than the one using values expressed in volume. The blue region indicates the self-nanoemulsifying region, with a lower particle size. The particle size optimisation studies suggested two optimal excipient combinations with the following composition:


**Figure 1.** Pseudo-ternary phase diagram of SNEDDS. In the white rectangles is indicated the particle size of each corresponding composition. The blue areas indicate a lower particle size, while yellow and red areas indicate larger sizes.

Validation studies were performed by preparing the suggested optimal mixtures, followed by measuring the particle size. The resulted particle size was 159.7 nm for combination A, and 88.4 nm for combination B. Based on these results, combination B was selected for further development of solid-B-SNEDD formulations.

#### *3.3. Yield of Solid B-SNEDDS*

Four different solid B-SNEDDS were manufactured utilising the vehicle mixture (combination B) above described. Two different hydrophobic, fumed silica carriers (Aerosil 200 or R972) were combined in two ratios (1:2 and 1:3 *w*/*w*), with the optimised liquid B-SNEDDS and the resulting four formulations coded as F1 to F4, respectively (Table 3). When SNEDDS were adsorbed on Aerosil 200 at a 1:2 ratio (F1), a higher yield was obtained, which was able to carry higher amounts of liquid B-SNEDDS (1:3 ratio). Hydrophobic fumed silica resulted in a poorer yield, indicating that SNEDDS adsorbed easier towards a hydrophilic surface, suggesting a core shell particle structure for liquid B-SNEDDS, with Labrasol being orientated towards the surface, and likely to interact more strongly with hydrophilic silica surfaces. However, this stronger interaction of labrasol with hydrophilic silicas has been linked previously to poorer disintegration times, which should be taken into account later when manufacturing solid dosage forms, such as tablets [30].


\* Particle size measurements of F4 were registered as 1 nm, indicating that particles were not stable upon dilution.

All solid B-SNEDDS had an AoR below 20◦, indicative of excellent flow properties [31,32] with near maximal drug loading (Table 3). Particle size measurements confirmed microemulsion regions with a size well below 300 nm. F4 yielded particles that either had not been able to re-form after adsorption on silica, or were unstable after the 1:1000 dilution, and thus were considered not appropriate for further development.

#### *3.4. Release Studies*

Hydrophilic fumed silica particles indicated a greater and faster release of butenafine nanoemulsions type II (F1 > F2 > F3~F4) (Figure 2). Increasing the pH from 1.2 to 6.8 to mimic the intestinal pH resulted in a sharp precipitation, which can be linked with a spring–parachute effect. This can explain the low absorption of butenafine orally [21], as that only allows for absorption in the upper part of the gastrointestinal tract [33,34]. Butenafine is more soluble in acidic pH as a hydrochloride salt, while its base (pKa: 9.23) has limited water solubility (<100 ng/mL) [35]. Formulation F1 allows for higher solubilised levels up to 90 min (*p* < 0.05; one-way ANOVA), which is critical for enhancing oral absorption.

**Figure 2.** Release of solid B-SNEDDS formulations in simulated gastric fluid (SGF) (pH 1.2, first 60 min) and simulated intestinal fluid (SIF) (pH 6.8) thereafter (mean % ± SD). --- F1, ---F2, -- F3, -- F4.

#### *3.5. Morphological Analysis*

The morphology of the four solid B-SNEDDS formulations was observed by scanning electron microscopy (Figure 3). At higher magnification, the large surface porosity of both silicon dioxide carriers can be observed. This explains the high amount of liquid SNEDDS that is able to be adsorbed on to the solid carriers, up to three times their own weight. However, agglomeration between carrier particles was more noticeable in those formulations containing a higher ratio of liquid SNEDDS (F2 and F4), which can explain the hindered release observed. In the F3 and F4, both containing the most hydrophobic carrier (Aerosil R972), it was observed that even at the lower ratio of liquid-SNEDDS (1:2 *w*:*w*), the aggregation between particles is still evident to some extent.

**Figure 3.** Scanning electron microscope (SEM) micrographs of solid B-SNEDDS F1, F2, F3, and F4 formulations at two different magnifications.

#### *3.6. Tableting and Hardness*

Amongst all the solid B-SNEDDS, only the F1 formulation was able to be compacted with adequate hardness without the incorporation of other excipients. We should bear in mind that the addition of other excipients will further dilute the butenafine dose per tablet size, which currently is 10 mg/500 mg tablet in powder form (Table 4). The addition of microcrystalline cellulose is likely to allow other solid SNEDDS to be formulated with appropriate hardness.


**Table 4.** Hardness of the F1 solid B-SNEDDS formulation. Hardness expressed as minimum and maximum values.

#### *3.7. In Vitro E*ffi*cacy and Cytotoxicity*

Miltefosine was active against both promastigote and amastigote forms of *L. (L.) infantum*, exhibiting an EC50 of 17.9 ± 0.9 and 13.7 ± 0.7μM, respectively, as well as a mild cytotoxicity (CC50 of 126.3 ± 3.5 μM) leading to a selectivity index towards promastigotes forms (SIp) of 7.0 and a selectivity index towards amastigotes forms (SIa) of 9.2 (Table 5). These values are similar to previous documented studies [36]. Free butenafine showed lower efficacy than miltefosine (Table 5), with an SI close to the unit indicating poor selectivity for the parasites over the macrophages. Amongst the four solid B-SNEDDS, F1 and F4 showed better efficacy against promastigotes (*p* < 0.05), while Aerosil R972 formulations (F3 and F4) demonstrated lower cytotoxicity and higher efficacy against amastigotes (*p* < 0.05), but whether this is driven by the excipient is not clear, and further studies are needed. In any case, except for F2, formulating butenafine as SNEDDS has improved its selectivity index, making it less toxic and more active than the free drug.

**Table 5.** Antileishmanial activity of solid B-SNEDDS formulations and butenafine were assayed against promastigote and amastigote forms of *L. (L.) infantum*. Cytotoxicity was analysed using peritoneal macrophages from BALB/c mice. EC50 represents the concentration of the formulation that produced a 50% reduction in parasites, while CC50 represents the concentration of the formulation that produced a 50% reduction of cell viability in treated culture cells with respect to untreated ones. SIp: selectivity index towards promastigotes forms; SIa: selectivity index towards amastigotes forms.


\* *p* ≤ 0.05 compared to Butenafine.

#### **4. Discussion**

Butenafine, a benzylamine derivative structurally similar to terbinafine, possesses antifungal activity, attributed to its ability to directly cause damage on fungal cell membranes by disrupting the early stages of ergosterol biosynthesis via inhibition of the enzyme squalene epoxidase [37]. This enzyme converts squalene to lanosterol, and leads to the accumulation of squalene [38]. Inhibition of squalene epoxidase suppresses the biosynthesis of ergosterol, an essential lipid of fungal and *Leishmania* cell membranes [39].

Therefore, butenafine is known for its antifungal effects in infections caused by *Tinea pedis, Tinea corporis,* and *Tinea cruris*[40]. We have recently demonstrated the leishmanicidal effect of butenafine against promastigote and amastigote forms of *L. (L.) amazonensis* and *L. (V.) braziliensis* [20]. Considering the difficulties related to the treatment of leishmaniasis, such as drug relapses, toxicity, hospitalization, and parenteral administration, the development of an oral, safe, multispecies, and effective medicament could revolutionise VL treatment. Butenafine is a multispecies drug, able to be effective against a range of parasite strains, such as *L. (L.) amazonensis, L. (V.) braziliensis*, and now in this work, it has

been demonstrated that is also active against *L infantum*. However, due to its poor water solubility and precipitation in the gastrointestinal tract, no oral formulations of butenafine have been licensed.

In order to pave the way for the market of oral medicines against VL, the selection of excipients that can promote oral bioavailability and safety, as well as elicit a synergistic effect against Leishmania parasites without compromising safety is critical. Peceol, Labrasol, and Capryol 90 have been selected as GRAS excipients to formulate butenafine SNEDDS, able to enhance the drug solubilisation capacity in the gastrointestinal tract and oral bioavailability [41]. Also, due to the proven efficacy of those selected excipients against several strains of Leishmania parasites, a synergistic effect with butenafine can be expected [22].

Pre-formulation studies have indicated that butenafine is more soluble in C18 lipids and amphiphiles, and these were thus selected for forming the oil component of SNEDDS (Peceol). We decided to combine these with the medium, as well as short chain fatty acids esters (Capryol 90) and triglycerides (Labrasol) that we have recently shown to possess antileishmanial activity. The high reported activity of lauric acid and labrasol against *Leishmania* parasites can be attributed to their ability to selectively permeate the cell membrane of parasites, resulting in rapid and considerable membrane damage and the loss of cellular potassium and magnesium [22]. SNEDDS were optimised towards a minimised droplet particle size (~100 nm) for several reasons. Nanoemulsions (type II microemulsions) are more stable than microemulsions, in terms of droplet flocculation and coalescence (Ostwald ripening) [42]. The smaller the particle size, the higher the permeability across the intestinal mucus brush border layer and cell membranes [43]. Additionally, small droplet sizes (~90 nm) provide a large interfacial surface area for drug release and absorption across the intestinal cells [44]. Recent reports suggest that a particle size between 100–500 nm is optimal for lymphatic uptake via the gastrointestinal lymphatic system, but at a slower rate than particles sized 50–100 nm [45,46].

In order to reduce costs, increase the chemical stability, and improve patient compliance, B-SNEDDS was transformed into a solid free-flowing powder by spray-drying, which can be directly compressed. Fumed amorphous silica (Aerosil) are known to possess enhanced surface area for adsorption (Aerosil 200: 200 <sup>±</sup> 25 m2/g and Aerosil R972: 110 <sup>±</sup> 20 m2/g). Both silicas were able to adsorb three times their own weight in SNEDDS, resulting in powders with excellent flow properties, as indicated by AoR studies (<20◦) [30]. The higher porosity and hydrophilic surface of Aerosil 200 carriers explain the higher drug release and yield, respectively. Loading was higher with Aerosil 200, which indicates a core-shell morphology of the SNEDDS droplet, with polyethylene glycol chains stabilising the droplet surface and enabling higher interactions with the hydrophilic surface of the silica [47,48]. Aerosil R972 carriers hinder the release of butenafine, possibly due to stronger hydrophobic interactions; however, this is likely to limit oral bioavailability. Additionally, particles prepared with Aerosil R972, with high SNEDDS loading when dispersed in water, are unable to yield an emulsion, indicating that droplets are not able to re-form or are unable to maintain their morphology, and become unstable after a 1:1000 dilution. The latter, combined with limited release from these solid B-SNEDDS, makes them less likely to be able to enhance oral bioavailability. Interestingly, though, these particles have a significantly impact on the in vitro efficacy against *Leishmania* promastigotes and amastigotes. Even though butenafine release from F3 and F4 formulations was low at physiological pH, these formulations possessed enhanced efficacy and selectivity, with no higher cytotoxicity on macrophages. Enhanced surface hydrophobicity can trigger macrophage uptake compared to particles with a hydrophilic surface [49]. However, when the results are taken together, formulation F1 is more likely to be a promising solid nanomedicine for VL, as it maintains characteristics for enhanced solubilisation and oral uptake, as well as presented antileishmanial activity in *L. (L.) infantum* promastigotes and amastigotes.

Limited studies are available in the literature for understanding the ability of solid carriers to load SNEDDS. Preparing a conventional solid dosage form of SNEDDS remains challenging, as tablets can be friable if a low compression force is used, or allow for SNEDDS leakage if a high compression force is used. Tablets prepared with Labrasol and silica (Neusilin US2, a synthetic, amorphous form of magnesium aluminometasilicate) have demonstrated an enhanced disintegration time and the need for a disintegrant [32]. Our solid B-SNEDDS were able to load high SNEDDS quantities, while being free-flowing and easily compacted into the appropriate hardness tablets. Smith et al. [12] have demonstrated that liquid SNEDDS can be transformed into solid tablets with good release properties when the ratio between liquid to solid was 1:3. In our work, we have shown that even a much greater ratio of liquid to solid (2:1) can still be compacted without SNEDDS leakage and maintain a homogenous matrix for compaction. However, further studies are required to investigate the manufacturing of these free-flowing powders into tablets that meet the USP Pharmacopeia requirements.

#### **5. Conclusions**

Manufacturing of cost-effective solid dosage forms of reformulated drugs is an ideal strategy to speed up the development of novel medicines for VL and other neglected diseases. We have demonstrated for the first time the efficacy of butenafine for VL, while developing an easily scalable preparation from GRAS excipients of solid dosage forms of butenafine, able to maintain its solubilisation capacity in the gastrointestinal tract. SNEDDS were optimised for drug loading (30 mg/g) and particle size, and we have demonstrated the ability of a microemulsion to be formed after release from fumed silica particles. Solid SNEDDS demonstrated excellent flow properties, and were able to be compressed in adequate hardness tablets. They also demonstrated antileishmanial activity in *L. (L.) infantum* promastigotes and amastigotes, which indicated their potential as solid nanomedicines for the treatment of VL.

**Author Contributions:** Conceptualization, A.L. and D.R.S.; methodology, A.B.-S., G.F.R., R.F.-G., L.F.P., M.D.L., and D.R.S.; software, A.B.-S.; validation, A.B.-S.; formal analysis, A.B.S, G.F.R., R.F.-G., and D.R.S.; resources, M.D.L., F.B.-F., L.F.P., and D.R.S.; data curation, A.B.-S.; writing—original draft preparation, A.B.-S.; writing—review and editing, A.B.-S., L.F.P., A.L., and D.R.S.; supervision, M.D.L, L.F.P., and D.R.S.; project administration, M.D.L. and F.B.-F.; funding acquisition, M.D.L. and F.B.-F.

**Funding:** This research was funded by the Unión Iberoamericana de Universidades (ENF03-2017) allowing student mobility exchange between UCM and USP and by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP processes: 2016/00468-0 and 2017/09405-4).

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Designing Fast-Dissolving Orodispersible Films of Amphotericin B for Oropharyngeal Candidiasis**

**Dolores R. Serrano 1,2, Raquel Fernandez-Garcia 1, Marta Mele 3, Anne Marie Healy <sup>4</sup> and Aikaterini Lalatsa 3,\***


Received: 7 June 2019; Accepted: 22 July 2019; Published: 1 August 2019

**Abstract:** Amphotericin B possesses high activity against *Candida* spp. with low risk of resistance. However, Amphotericin B's high molecular weight compared to other antifungal drugs, such as miconazole and clotrimazole, and poor water solubility hampers its efficacy at the physiological conditions of the oropharyngeal cavity (saliva pH, limited volume for dissolution) and thereby limits its clinical use in oropharyngeal candidiasis. We have prepared fast-dissolving orodispersible films with high loading (1% *w*/*w*) using solvent casting that enables amphotericin B to remain solubilised in saliva in equilibrium between the monomeric and dimeric states, and able to produce a local antifungal effect. Optimisation of the amphotericin B-loaded orodispersible films was achieved by quality by design studies combining dextran and/or maltodextrin as dextrose-derived-polymer film formers with cellulose-derived film formers (hydroxypropylmethyl/hydroxypropyl cellulose in a 1:4 weight ratio), sorbitol for taste masking, microcrystalline cellulose (Avicel 200) or microcrystalline cellulose-carboxymethylcellulose sodium (Avicel CL-611) for enhancing the mechanical strength of the film, and polyethylene glycol 400 and glycerol (1:1 *w*/*w*) as plasticizers. The optimised amphotericin B orodispersible films (containing 1% AmB, 25% dextran, 25% maltodextrin, 5% sorbitol, 10% Avicel 200, 10% polyethylene glycol 400, 10% glycerol, 3% hydroxypropylmethyl cellulose acetate succinate, 12% hydroxypropyl cellulose) possessed a fast disintegration time (60 ± 3 s), quick release in artificial saliva (>80% in 10 min), high burst strength (2190 mN mm) and high efficacy against several *Candida* spp. (*C. albicans*, *C. parapsilosis* and *C. krusei*) (>15 mm inhibition halo). Amphotericin B orodispersible films are stable for two weeks at room temperature (25 ◦C) and up to 1 year in the fridge. Although further toxicological and in vivo efficacy studies are required, this novel Amphotericin B orodispersible films is a promising, physicochemically stable formulation with potential wide application in clinical practice, especially for immunocompromised patients suffering from oropharyngeal candidiasis.

**Keywords:** orodispersible films; fast-dissolving films; micelles; amphotericin B; fungal infections

#### **1. Introduction**

Fungal infections of the oral cavity are opportunistic, usually caused by *Candida albicans*, and occur more frequently in patients that are immunocompromised (e.g., HIV, cancer patients), diabetics or having predisposing factors such as antibiotic and corticosteroid therapy, poor buccal hygiene and ill-fitted dentures. Local therapy of oral and pharyngeal candidiasis is desirable as it avoids adverse effects linked to systemic antifungal use [1]. A major challenge to effective local treatments remains the low volume for dissolution and need for rapid permeability of the oropharyngeal cavity in order to deliver adequate drug concentrations for local action. The majority of formulations rely on oral suspensions due to the poor aqueous solubility of most antifungals that are swished around in the mouth for a few seconds, gargled, and swallowed or spat out. This short contact time with the oral mucosa requires a readily available and solubilised drug to exert an antifungal effect.

Amphotericin B (AmB) is a broad spectrum antifungal effective in the nanomolar range (IC50 of 0.25–1 μg mL<sup>−</sup>1). Compared to azoles, there is a low frequency of *C. albicans* strains that are resistant to AmB [2]. The poor aqueous solubility of AmB (BCS Class IV) makes it difficult to solubilise in an adequate concentration in the small volume of saliva that is available in the oral cavity (1 mL) [3]. AmB is only commercialised as lyophilised formulations (micellar or lipidic nanoparticulate formulations) that are stable after reconstitution in aqueous media for a duration shorter than 24 h even when refrigerated [4,5] resulting in wastage, while poor patient compliance is an issue as formulations are not palatable. Taste masked orodispersable films (ODFs) can be potential solid dosage forms to deliver AmB in safe and efficient systems for the treatment of oropharyngeal fungal infections. Compared to other solid formulations, such as tablets and capsules, ODFs benefit from better patient compliance due to the ease of administration to dysphagic, paediatric and geriatric patients, without the need for water, and a rapid onset of action [6,7]. ODFs have shown better performance than semisolid formulations such as gels, because they can be easily transported, allow for accurate dosing and possess superior chemical and physical stability when packed appropriately, which can be important for unstable drugs in aqueous media such as AmB [8,9]. From an industrial and clinical point of view, ODFs would be more cost-effective formulations compared to parenteral AmB formulations, as they do not require sterilisation and lyophilisation, avoiding wastage.

The hypothesis underpinning this work is that AmB-loaded ODFs prepared using GRAS (Generally Regarded as Safe) excipients would enhance the current therapies available for the treatment of fungal infections in the oral and pharyngeal cavities. However, to the best of our knowledge, this has not been achieved to date due to the physicochemical challenges when formulating this drug. Here, we present an optimised fast disintegrating ODF of AmB with improved stability, loaded with high amounts of drug and designed to ensure AmB solubility in small volumes of saliva, while being taste masked and locally effective. To ensure AmB solubility at the physiological pH of the oropharyngeal cavity, we entrapped AmB in sodium deoxycholate micelles that were then embedded within the ODF. Sodium deoxycholate was selected as it interacts with AmB forming micelles, while it accumulates in buccal tissue after penetration without causing a loss of superficial cell layers and interacts with the intercellular or membrane lipids increasing the permeability of drugs through the epithelium [10]. Design of experiment (DoE) studies enabled us to identify the optimal drug:excipient ratio needed to ensure high drug loading (1% AmB) that is critical in achieving local concentrations well above the IC50 against *Candida albicans*. The optimised ODF were fully characterised and their *in vitro* antifungal activity evaluated.

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

#### *2.1. Materials*

AmB was purchased from Azelis (Barcelona, Spain). Hydroxypropylmethyl cellulose acetate succinate (HPMC AS 912, Affinisol™), maltodextrin (Glucidex 12D) and sorbitol were a gift from DowPharma (Dewsbury, UK) and Roquette (Valencia, Spain). Microcrystalline cellulose (Avicel 200) and microcrystalline cellulose and sodium carboxymethylcellulose (Avicel CL-611) were kindly donated by FMC (Cork, Ireland), while hydroxypropyl cellulose (HPC, Klucel HXF) was a gift from Ashland (Barcelona, Spain). Dextrose and dextran from Leuconostoc mesenteroides (16 KDa) were purchased from Sigma (Madrid, Spain). Humidity capsules and stability chambers were purchased

from Amebis Limited (Dunshaughlin, Ireland). All other chemicals and solvents were at least of ACS reagent grade and were used without further purification.

#### *2.2. Quality by Design (QbD) Optimisation of ODF*

Several critical quality attributes (CQAs) such as disintegration time in artificial saliva and the physical characteristics of the ODF (burst strength, flexibility, tackiness) were identified as key factors in order to meet the Target Product Profile (TPP) (Tables S1 and S2 in Supplementary Materials). A Taguchi design (L8 = 2ˆ7) was carried out using Design Expert software 8.04 (Stat-Ease, Minneapolis, MN, USA). Seven formulation variables (factors) and two levels of each factor affecting the film formation were investigated (Table 1). Disintegration time, burst strength and appearance were evaluated as responses (Table 2).

**Table 1.** Formulation and process variables with their respective high and low levels included in the Taguchi screening design.


#### 2.2.1. ODF Manufacture and Response Evaluation

Eight formulations of AmB-loaded ODFs (3 g each) (Table 2) were prepared as follows: dextrose-derived-polymer film former, Avicel 200 or Avicel CL-611, taste masking agent (5%) and plasticisers were weighed and mixed in a mortar and pestle. To this mixture, 3 mL of freshly prepared AmB-loaded micelles (30 mg of AmB and 24.6 mg of sodium deoxycholate [11]) was added and manually mixed until a homogenous mixture was formed. The cellulose-derived film formers were then added, if required, and mixed. Methanol was added to reduce the viscosity of the mixture to a pourable homogenous suspension that was immediately cast onto a release liner (Primeliner 36 μm 1S, Loparex BV, Apeldoorn, The Netherlands) using a coating knife (Multicator 411, Erichsen, Hemer, Germany) and film applicator (Erichsen Coatmaster 510 film applicator, Erichsen, Hemer, Germany), at a speed of 10 mm s−<sup>1</sup> under vacuum (air pressure 60 Pa), to form a wet film with a thickness of 1000 μm. The film was allowed to dry under vacuum for 4–5 h. Once the films were dried, they were carefully removed from the release liner and properties of the film were evaluated (Table 2).


*Pharmaceutics* **2019** , *11*, 369

Disintegration times of ODFs (1 × 1 cm) were measured in 3 mL of artificial saliva prepared as previously described: 14.4 mM sodium chloride, 16.1 mM mg potassium chloride, 1.31 mM calcium chloride dihydrate, 0.54 mM magnesium chloride hexahydrate, 1.96 mM dibasic potassium phosphate adjusted to pH 5.7 ± 0.01 [12] under gently shaking (30 slow 90◦ inversions of the vial per min). The mechanical properties (burst strength) of ODFs were evaluated using a texture analyser (Texture Analyser TA-XTplus, Stable Microsystems, Godalming, UK) attached to a film support rig (HDP/FSR, Stable Microsystems) [13]. For burst strength, the force required to rupture or break films was measured using a 5 mm spherical stainless-steel ball probe with probe adapter which was connected to the load cell. A film (35 × 15 mm) was placed in a film supporting rig and the moving probe reached the surface of the film with a pre-test speed of 2 mm s−1, test speed of 1 mm s−<sup>1</sup> and post-test speed of 10 mm s<sup>−</sup>1. The force applied had a trigger load of 4.9 N and the force maximum (mN), travel distance (mm) and area under the curve (mN mm) were measured. Finally, the overall appearance of the films was evaluated as the sum of folding endurance, adhesion of the dried film to the release liner and homogeneity. Folding endurance was manually measured by counting the number of times the film could be folded at a 180◦ angle to the plane without breaking. A value of 1 was assigned for films with good/flexibility able to fold 180 degrees without breaking above 50 folds, 2 as medium flexibility films (25–50 times) and 3 as poor flexibility films, which break easily when folded more than 25 times. Adhesion of the dried film to the release liner was also taken into account, assigning a number from 1 to 3, with 1 indicating intact removal of the film and 3 complete breakage of the film upon removal. The homogeneity was also visually inspected, assigning 1 for films that exhibit a smooth surface with no cracks, 2 for films with some irregular surfaces and 3 for non-homogeneous films with lumps and/or cracks on the surface.

#### 2.2.2. ODF Optimisation

Mathematical modelling was carried out by multiple linear regression analysis (MLRA). Only the statistically significant coefficients (*p* < 0.05) were considered in framing the polynomial equations, and the model was evaluated by analysing the *p*-value, coefficient of correlation (R2) and predicted residual sum of squares (PRESS) [14]. Films were optimised in order to minimise the disintegration time and improve overall appearance and mechanical strength.

#### *2.3. Full Physicochemical Characterisation of Optimised ODF*

A full evaluation was performed including particle size and zeta potential (after disintegration in artificial saliva), powder X-ray Diffraction (PXRD), Fourier Transformed Infrared (FT-IR), Dynamic Vapour Sorption (DVS), Scanning Electron Microscopy (SEM) and surface area [11,14,15]. Briefly, particle size and zeta potential were measured in a Zetatrac after dilution (1 to 100) with artificial saliva. XRD measurements (*n* = 3) from 5◦ to 40◦ (2 theta) and a step scan rate 0.05◦ per second were performed in a Miniflex II Rigaku diffractometer with Ni-filtered Cu Kα radiation (1.54 Å) using a tube voltage and tube current of 30 kV and 25 mA respectively. FT-IR spectra were scanned in the range of 650–4000 cm−<sup>1</sup> with a resolution of 4 cm−<sup>1</sup> on a PerkinElmer Spectrum 1 FT-IR Spectrometer equipped with a UATR and a diamond/ZnSe crystal accessory. Baseline correction and data normalization were performed using Spekwin32 version 1.71.6.1. Water sorption kinetic profiles were obtained using a DVS (Advantage, Surface Measurement Systems, Alperton, UK) at 25.0 ± 0.1 ◦C. Samples (10–20 mg) were dried at 0% relative humidity (RH) for 1 h followed by step changes of 10% RH up to 90% RH, and the reverse for desorption. SEM was carried out in a Zeiss Supra Variable Pressure Field Emission Scanning Electron Microscope (Oberkochen, Germany) equipped with a secondary electron detector at 15 kV. Surface area was determined by the Brunauer, Emmett, Teller (BET) isotherm method using N2 adsorption with 6 points in the relative pressure range of 0.05–0.3 in a Micromeritics Gemini VI surface area analyser (Particular Sciences Ltd., Dublin, Ireland).

#### *2.4. Content Uniformity*

ODF (1 × 1 cm) were weighed and disintegrated in 3 mL of deionised water prior to being diluted (1 to 2) with methanol to ensure full drug solubilisation. Samples were centrifuged to precipitate undissolved excipients (5000 rpm for 10 min) and AmB dissolved in the supernatant was quantified using a validated HPLC method [16]. Experiments were performed using films from ten different sections of the cast ODF (15 × 10 cm).

#### *2.5. Release Studies in Artificial Saliva*

ODF (1 × 1 cm) were dissolved in 10 mL of artificial saliva at 37 ◦C under slow magnetic stirring (50 rpm). Samples (1 mL) were obtained at 1, 2, 4, 6, 8, 10, 15 and 30 min and media were replaced with fresh artificial saliva. Withdrawn samples (0.1 mL) were mixed with methanol (0.1 mL), vortex and centrifuged (5000 rpm, 10 min) for HPLC analysis to determine the percentage of drug released. The remaining sample volume was centrifuged (500 rpm, 1 min) to precipitate undissolved excipients such as microcrystalline cellulose. The particle size of the supernatant was measured using a Malvern Zetasizer (Malvern Nano Zs, Malvern Instruments, Malvern, UK). Additionally, the supernatant was analysed by UV (300–450 nm, Multiskan GO, Thermo Scientific, Basingstoke, UK) to determine the aggregation state of the AmB [17]. The ratio of the absorbance at 332 nm corresponding to dimeric AmB versus the absorbance at 408 nm that corresponds to the monomeric state was plotted to demonstrate the prevalence of the dimeric aggregation state of AmB.

#### *2.6. In Vitro Antifungal Assays*

In vitro antifungal activity was tested based on the agar diffusion assay as described by Ruiz et al. [1] according to the National Committee for Clinical Laboratory Standards (NCCLS) Method for antifungal disk diffusion susceptibility testing of yeast, standard M44-A2 [18]. In vitro activity was tested on three different *Candida* spp. (*C. albicans* CECT 1394, *C. parapsilosis* 57744 and *C. krusei* 52009 which was kindly provided by Dr. Pérez (CAQYM, University of Alcala de Henares, Alcalá de Henares, Spain) [1]. Strains were cultured in Sabouraud dextrose agar for 72 h to ensure viability and absence of contamination at 35 ◦C (±2 ◦C). Antifungal tests were carried out in Müeller Hinton agar (MHA) supplemented with glucose (2% *w*/*v*) and methylene blue (0.5 mg/mL). Inoculum was prepared by picking a few distinct colonies, which were suspended in 3 mL of sterile saline (0.9%). The resulting suspension was vortexed and its turbidity was adjusted with a spectrophotometer by adding sufficient sterile saline or more colonies to adjust the transmittance to that produced by a 0.5 McFarland standard at 530 nm wavelength, resulting in a yeast stock suspension of 1 <sup>×</sup> 106 cells per mL. Yeast suspension was inoculated to the MHA (200 mL) and was casted in disposable sterile petri dishes (instead of spreading it on the surface of the plate as specified in M44-A2). Once solidified, AmB ODFs (circles with a 6 mm diameter) were tested. Four disks were placed in each plate. AmB ODF in vitro activity was compared to commercially available AmB Neo-Sensitabs tablets (10 μg, 6 mm tablets from Rosco diagnostic A/S, Taastrup, Denmark) and AmB impregnated on inoculation 6 mm paper disks (10 μg/20 μL of DMSO) with appropriate DMSO controls. Once the disks were placed on the surface of the agar, the plates were inverted and placed in an incubator set to 35 ◦C (±2 ◦C) within 15 min after the disks were applied. After 24 h of incubation, the inhibition halo was measured.

#### *2.7. Stability Studies*

Physicochemical stability studies were performed under accelerated conditions (40, 60 and 80 ◦C) for one week and long term at 5 ± 3 ◦C and 25 ± 3 ◦C. AmB ODF (1 × 1 cm) were placed in sealed vials into Amebis chambers (Amebis Ltd., Dublin, Ireland) at the selected temperature. A sensor cap was used to seal the test chamber and a logger cap connected to the sensor cap was used to collect and transmit the temperature and humidity test conditions wirelessly to the Amebis Control Software [14]. Disintegration time and drug content were quantified at different time points (time zero, day 1, 3 and

7). The degradation rate of AmB was calculated by fitting the percentage of drug degraded at different time points to several degradation kinetic equations (zero order, first order, second order, Avrami and diffusion) and the best fitted degradation kinetic model was selected (i.e., highest R2). Using the degradation rates at different temperatures, the Arrhenius equation was employed to calculate the activation energy (Equation (1)):

$$\mathcal{K} = \mathcal{A} \ e^{\bar{\mathcal{R}} \bar{\mathcal{H}}} \tag{1}$$

where *K* is the degradation rate (% drug degraded/day), *A* is the collision factor, *T* is the absolute temperature in Kelvin, *R* is the gas constant (1.985 cal/mol/K) and *Ea* is the activation energy in cal/mol [14]. Drug stability at room temperature was then predicted using the Arrhenius equation and compared to experimental values.

#### **3. Results**

#### *3.1. QbD Studies for Optimisation of AmB-Loaded ODFs*

The first-order mathematical model generated for each response variable was found to be statistically significant (*p* < 0.05 in each case). Co-efficients with *p* values > 0.1 were considered insignificant based on Pareto charts and ANOVA analysis. High R<sup>2</sup> values for the polynomial equations obtained for all the response variables indicate a good fit to experimental data (Table S3). The variables, type and amount of Avicel and number of plasticizers, showed a significant effect on the disintegration time of the ODFs (Figures S1 and S2). ODFs with lower disintegration time were obtained when lower amounts of Avicel and cellulose-derived film formers were used. The disintegration time was reduced when higher percentages of plasticisers were employed and when Avicel 200 was incorporated in the film compared to Avicel CL-611, as the latter acts as a viscosity enhancing agent resulting in thixotropic gels, which retard disintegration (Figure 1A,B).

The choice of dextrose-derived film former, volume of methanol added, type/amount of Avicel and amount of cellulose-derived film formers had a significant impact on burst strength (Figures S3 and S4). The use of maltodextrin and Avicel 200 resulted in films with higher burst strength. The higher the amount of Avicel, HPMC AS/HPC and methanol, the better the mechanical strength of the films (Figure 1C,D).

The appearance of the films was rated as described above (Figure S5). DoE indicated that the main variables that contributed to the appearance of the film were: the type of dextrose-derived film former and the amount of cellulose-derived film formers utilised, followed by type and amount of Avicel and taste masking agent (Figures S6 and S7), although results were not statistically significantly different. The appearance of the film was smoother when maltodextrin was used, although high amounts of maltodextrin increased film tackiness. When higher amounts of Avicel and cellulose-derived film formers were used, appearance was improved. Avicel 200 and sorbitol resulted in smoother film surfaces compared to dextrose films as dextrose can recrystallize resulting in rougher surfaces [19] (Figure 1E,F).

#### *3.2. Manufacturing of Optimised AmB ODF*

A trade-off between key CQAs was necessary to attain optimal characteristics, i.e., short disintegration time (which is critical for faster onset of action), maximal burst strength (in order to obtain robust films that can be easily manufactured and packaged without breaking), and good appearance. A closer match to ideal CQAs was obtained with sorbitol, Avicel 200 (10%), 20% of plasticisers (PEG 400:glycerol, 1:1 *w*/*w*) and 10% of cellulose-derived film formers (HMPC 912 AS:HPC, 1:4 weight ratio). Regarding the type of dextrose-derived film former, maltodextrin conferred better flexibility and faster release from the films compared to those obtained with dextran, as dextran interacts more strongly compared to microcrystalline cellulose or the modified cellulose-derived film formers. However, high amounts of maltodextrin significantly increased adhesion of the films to the release liners. Thus, we decided to optimise the ODF using a mixture of dextran and maltrodextrin (1:1 *w*/*w*) (Table 3). The films exhibited a good overall appearance with a dried thickness of 0.14 ± 0.01 mm, a weight of 28.5 <sup>±</sup> 1.5 mg/cm2, a drug content of 0.996 <sup>±</sup> 0.045 mg/g (which was uniform with a low standard deviation and a variance coefficient of 4.5%), a burst strength of 2190 mN mm and a disintegration time of ≤60 s in 3 mL of artificial saliva.

**Figure 1.** Contour plots showing the influence of the most influential factors affecting the disintegration time (**A**,**B**), the burst strength (**C**,**D**) and the overall appearance (**E**,**F**) of the AmB-loaded orodispersable films (ODFs). Key: CD, cellulose-derived.


**Table 3.** Composition and properties of the optimised AmB-loaded ODFs.

SEM micrographs revealed a smooth and porous surface of the optimised AmB-loaded ODF compared to films obtained in the DoE experiment 1 and 2, which exhibited a granular rough texture or embedded crystals respectively probably due to the use of dextran in the first film instead of maltodextrin and the presence of dextrose in DoE 2 ODF which tends to crystallise (Figure 2). Films obtained in DoE 6 experiments appeared cracked and exhibited low elasticity likely due to the low percentage of plasticisers and Avicel 200 included in the formulation.

**Figure 2.** Scanning Electron Microscopy (SEM) micrographs of optimised AmB-loaded ODFs compared to those obtained in the Taguchi DoE before optimization. Bars; DOE Experiment 1—Top: 1 mm and Bottom: 10 μm, DOE Experiment 2—Top: 500 μm and Bottom: 20 μm, DOE Experiment 6—Top: 500 μm and Bottom: 100 μm, Optimised ODF—Top: 50 μm and Bottom: 10 μm.

#### *3.3. Performance and Further Characterisation of the Optimised AmB-Loaded ODF*

The smooth and porous surface observed by SEM can be associated with the high 2.3 ± 0.5 m2/g surface area of the optimized film (Figure 2). AmB-loaded ODFs showed a 40% increase in mass at 90% relative humidity, associated with a large water uptake due to the films hydrophilicity as shown by the water sorption kinetic profile (Figure 3A). No mass loss (associated to phase transformation/crystallization) was observed during the sorption or desorption cycle. The solid state of the film was retained (as demonstrated by post DVS XRD analysis (Figure S8) indicating an overall acceptable physical stability. The FTIR spectra showed a broadening of the peak at 1691 cm−<sup>1</sup> (C=O stretch) probably attributed to hydrogen bonding between the AmB and excipients such as sodium deoxycholate, sorbitol and acetate succinate groups of the HPMC AS (Figure 3B and Figure S9).

**Figure 3.** Physicochemical characterization of optimised AmB-loaded ODFs. (**A**) Water sorption kinetic profile. (**B**) FTIR spectra: (a) hydroxypropyl cellulose (HPC), (b) maltodextrin, (c) dextran, (d) Avicel 200, (e) sorbitol, (f) sodium deoxycholate, (g) AmB, (h) AmB-loaded ODF, (i) physical mixture, (j) HPMC 912 AS. (**C**) PXRD patterns: (a) AmB-loaded ODF, (b) Physical mixture of all components, (c) Avicel 200, (d) HPMC 912 AS, (e) HPC, (f) maltodextrin, (g) dextran, (h) sorbitol, (i) sodium deoxycholate, (j) AmB.

The XRD pattern of the physical mixture of all components revealed crystalline Bragg peaks attributed to sorbitol (Figure 3B(b,h)), which are not present in the casted film. However, a characteristic halo attributed to the semi-crystalline nature of the microcrystalline cellulose (Avicel 200) was observed in the optimised ODFs. Lower intensity values at several Bragg peaks (5◦, 14.15◦, 17.35◦, 21.75◦) were observed in the diffractogram of the AmB-loaded ODF compared to the physical mixture, which can be attributed to the presence of amorphous AmB-sodium deoxycholate complexes in the films. However, bearing in mind that the AmB content in the ODF is 1%, it is likely that the results obtained from XRD measurements are not conclusive due to the XRD detection limit for crystalline AmB.

#### *3.4. Release and Aggregation State*

Aligned with the high porosity of the films, ODF presented a fast-dissolving behaviour (>80% in 10 min) in saliva (Figure 4A). Once disintegrated, particle size and zeta potential of the resulting suspension was measured. Initial particle size was bimodal mainly dominated (>70%) by large particles (>1 μm) due to insoluble excipients. After centrifugation, a white pellet and a transparent yellow supernatant was obtained. The latter demonstrated a bimodal particle size distribution of 8.7 ± 2.5 nm and 918 ± 120 nm with an anionic zeta potential of −14 ± 3 mV, indicating that after disintegration of the ODFs, AmB remained solubilised in the supernatant in equilibrium between micelles and particles close to 1 μm in size. TEM images confirmed the presence of micelles and particles of that size (Figure 4(B1)). Characteristic crystals corresponding to unprocessed AmB (Figure 4(B2)) were not observed after disintegration of the ODF and release of the AmB in the media which supports drug solubilisation within micelles.

**Figure 4.** Release profile, morphology and aggregation state of optimised AmB loaded ODFs. Key: (**A**) Release profile of AmB-loaded ODFS in artificial saliva. (**B**) Transmission electron microscopy (TEM) of AmB-loaded ODF after reconstitution in aqueous media (left image) and crystalline unprocessed AmB (right image); (**C**) Aggregation ratio during the release studies; (**D**) Transformation of AmB aggregation states over time.

Regarding the aggregation state of AmB released from the films, UV spectroscopy indicated a shift from the monomeric to dimeric state illustrated by the faster increase in the peak at 332 nm compared to the one at 408 nm corresponding to the monomeric state (Figure 4C,D). At the earlier time points (1–2 min), there is an equilibrium in solution between monomer and dimer. As time progresses and a higher percentage of AmB is released to the media, AmB self-aggregates shifting from monomer to dimer after 2 min and from dimer to polyaggregate after 8 min (peak at 420 nm) [17,20]. AmB solubility can be enhanced by the presence of PEG 400 and sodium deoxycholate in the formulation [21]. Fungizone®, a commercial AmB parenteral formulation, consists of 1:2 AmB:sodium deoxycholate molar fraction that facilitates the solubilisation of AmB in dimeric form within micelles [20]. The presence of nanometric (≈10 nm) spherical single layer particles in solution after ODF disintegration can be explained by the partial solubilisation of the drug within sodium deoxycholate micelles. However, even though initially AmB is incorporated in the solid mixture of excipients solubilised in sodium deoxycholate, the later addition of solvent to produce a pourable castable formulation could destabilise initially formed AmB sodium deoxycholate micelles. Thus, the different aggregation states of released AmB can be a consequence of the initial solubilisation of AmB in the monomer state in the presence of PEG 400 and the formation of ion pairs with sodium deoxycholate, followed by the encapsulation of AmB within micelles resulting in prevalence of the dimeric state. When higher amounts of AmB are released over time in the media, AmB self-aggregates into polyaggregates close to 1 μm in size which are stabilised by the release of dextrose-based polymers, in particular maltodextrin, which has been previously shown to be able to effectively stabilise dispersed systems such as oil-in-water emulsions [22].

#### *3.5. Antifungal Activity*

The optimised AmB-loaded ODFs showed good in vitro antifungal activity against the three *Candida* spp. (with an inhibition zone >15 mm) equivalent to that of AmB dissolved in DMSO and the commercially available disks, which indicates good drug release from the film and diffusion across the agar of the solubilised AmB (Figure 5). Bearing in mind the potency of the drug against *C. albicans, C. parapsilopsis* and *C. krusei* (MIC50 ranges from 0.25–1 μg/mL) reported in the literature [2], the dose delivered by a 1 × 1 cm film would be adequate to ensure efficacy against buccal candidiasis. Considering that the drug loading is 1% *w*/*w* and that a 1 cm2 film weights around 30 mg, each 1 cm<sup>2</sup> film contains approximately 0.3 mg of AmB. Assuming a volume of ~1 mL in the oral cavity [3,23], the concentration of AmB would be 300 μg/mL, which is well above the MIC50 reported in literature. Even if the concentration is diluted further with 10 mL or 100 mL i.e., 10 or 100-fold, the AmB concentration would be in range between 3 or 30 μg/mL, which are concentrations still above the MIC50 and, for this reason, we believe that the drug levels would be adequate to elicit a pharmacological effect.

**Figure 5.** In vitro activity against *C. albicans*, *C. krusei* and *C. parapsilopsis*. The isolates were classified as susceptible (S) to AmB when the inhibition zone was ≥15 mm, resistant (R) when it was ≤10 mm and intermediate (I) or susceptible-dose dependent when the inhibition zone was between 10- and 15-mm. Inhibition zone diameters are expressed as mean ± SD in mm. All experiments were performed in triplicate. \* *p* < 0.05 One-way ANOVA test.

#### *3.6. Physicochemical Stability*

Stability studies showed that the optimised AmB-loaded ODF was physiochemically stable over a year (>90% drug content) at 5 ◦C under desiccated conditions. Accelerated stability studies demonstrated a pronounced chemical degradation (80 ◦C >> 60 ◦C >> 40 ◦C), while a change in colour occurred at the highest temperature (Figure 6). The disintegration times of ODFs subject to accelerated stability studies was reduced over time, probably due to the evaporation of bound water within the film and the formation of micropores (Figure 6). However, in all the tested conditions, the disintegration time remained below 2 min as considered appropriate for fast-disintegrating ODFs. The Avrami kinetic model fitted the degradation of AmB from loaded ODFs. This kinetic model is commonly applied to evaluate the growth and formation of crystals [24]. Nevertheless, several authors have also used this model in the stability prediction of nanocomposites [25,26]. The breaking of a 3D network becomes more heterogeneous as the degradation progresses, and thus, the degradation proceeds faster as the combined effects of physical and chemical degradation occur simultaneously, especially at higher temperatures [26]. The activation energy of the AmB-loaded ODFs was 5.71 Kcal/mol, which correlates with the poor physicochemical drug stability observed at high temperatures (Figure 6). Experimental and predicted data from the Arrhenius equation showed that ODFs stored at 25 ◦C at desiccated conditions remained stable over 15 days (>90% drug content, disintegration time of 55 s). Thus, our AmB ODFs would remain stable for 2 weeks at room temperature in a closed pouch, which is

a significant advantage to existing AmB commercialised parenteral formulations that are unstable after 24 h of reconstitution of the lyophilised powder [4] reducing overall cost of treatment and wastage.

**Figure 6.** Physicochemical stability studies of optimised AmB-loaded ODFs at different temperatures. Key: --- 40 degrees, --- 60 degrees, -- 80 degrees. (**A**) Chemical stability expressed as AmB content. (**B**) Change in disintegration time for films stored at different temperatures. (**C**) Physical appearance. \* *p* < 0.05 indicative of statistically significant difference compared to AmB content and disintegration time of films at time 0 (One-way ANOVA with a post-hoc Tukey's test, level of significance set at 5%).

#### **4. Discussion**

To the best of our knowledge, this is the first report of AmB, a high molecular weight poorly water-soluble antifungal, being formulated as an ODF for the treatment of oropharyngeal candidiasis. The optimised ODF was achieved by combining dextran and maltodrextrin as dextrose-derived polymer film formers with sorbitol for taste masking, microcrystalline cellulose (Avicel 200) for enhancing mechanical strength, PEG 400 and glycerol as plasticizers to ensure faster disintegration and adequate plasticity and HMPC 912 AS /HPC for facilitating film formation. The combination of PEG 400 and glycerol (1:1 weight ratio) was selected based on previous results (data not shown). Glycerol improves mouthfeel, which is a necessity in an ODF formulation, as well as enhancing the viscosity resulting in higher, unacceptable disintegration times. In contrast, PEG 400 decreases disintegration time of ODF and, in combination with HPMC, gives optimal films in terms of tensile strength and flexibility [27].

Currently, nystatin (which has a similar chemical structure to AmB) formulated as a suspension is considered as the reference treatment for oral candidiasis; nevertheless, a recent meta-analysis demonstrated a limited efficacy for nystatin, lower than fluconazole, in treating oral candidiasis in infants, children and HIV patients, which is related to a poor and variable bioavailability in the oral mucosa [28]. AmB has shown 4-fold higher in vitro efficacy than nystatin [29], but its poor aqueous solubility limits its use in clinical practice, and it is only marketed as intravenous formulations. After disintegration of the ODF, AmB is maintained solubilised in the aqueous media in equilibrium between different aggregation states mainly monomer and dimer at earlier times points, which have shown to possess much greater efficacy than polyaggregates against *Candida* [30,31]. Unlike previous studies in which the antifungal activity was tested after disintegration of the films in liquid form [32], enhanced in vitro activity of the AmB-loaded ODF was shown here for the intact film placed on top of the agar, showing that the release and diffusion of the drug across the agar occurs and hence AmB is free to elicit its effect. Excipients utilised in ODFs are used within ranges that are considered GRAS (generally regarded as safe). Sodium deoxycholate, at the concentrations used in optimised the ODFs, has been shown to cause no major morphological changes [10]. Although further toxicological and

in vivo efficacy studies are required, we have shown that this technology platform can be a promising AmB formulation for the treatment of oral candidiasis.

#### **5. Conclusions**

Fast-dissolving orodispersible films containing a very poorly soluble and unstable drug, AmB, have been successfully engineered and prepared using a solvent casting film method in which the drug is solubilised at a final concentration of 1% *w*/*w* upon disintegration of the film. The optimised AmB orodispersible flms consisted of a mixture of dextrose-derived polymers (25% dextran and 25% maltodextrin) and cellulose-derived film formers (3% HMPC AS and 12% HPC) with 10% Avicel 200 added to enhance the mechanical strength of the film, 5% sorbitol for taste masking, and 10% PEG 400 and 10% glycerol as plasticizers. The optimised ODF exhibited a fast disintegration time (60 ± 3 s), quick release in artificial saliva (>80% in 10 min), high burst strength (2190 mN mm), good chemical stability (1 year at refrigerated conditions and 2 weeks at room temperature) and high efficacy against several *Candida* spp. (*C. albicans, C. parapsilosis* and *C. krusei* with an inhibition halo >15 mm). Although further toxicological and in vivo efficacy studies are required, these novel orodispersible films prepared with GRAS excipients have potential in the treatment of oro-and pharyngeal candidiasis, and hence represent a promising system with wide applications in clinical practice among immunocompromised patients suffering from this disease.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4923/11/8/369/s1, Figure S1: Pareto charts depicting the effect of (C) Type of Avicel, (D) Amount of Avicel, (E) Amount of plasticisers and (G) Amount of cellulose-derived film formers on the disintegration time. Orange colour indicates a positive effect whereas blue colour indicates a negative effect. Figure S2: Effect of the four variables (Type of Avicel, amount of Avicel, amount of plasticizers and amount of cellulose-derived film formers) on the disintegration time. Figure S3: Pareto charts depicting the effect of (A) Type of dextrose-derived film former, (C) Type of Avicel, (D) Amount of Avicel, (F) Volume of methanol and (G) Amount of cellulose-derived film formers on the burst strength of the film (expressed as AUC). Orange colour indicates a positive effect whereas blue colour indicates a negative effect. Figure S4: Effect of the significant variables (Film former, Type of Avicel, amount of Avicel and amount of cellulose-derived film formers) on the burst strength expressed as AUC of the film. Figure S5: Appearance of the eight AmB-loaded films prepared according to Taguchi matrix design. Figure S6: Pareto charts depicting the effect of (A) Type of dextrose-derived film former, (B) Taste masking, (C) Type of Avicel, (D) Amount of Avicel and (G) Amount of cellulose-derived film formers on the appearance of the film. Orange colour indicates a positive effect whereas blue colour indicates a negative effect. Figure S7: Effect of the five variables with higher impact on the final appearance of the film (Type of Avicel, amount of Avicel, taste masking agent, type of dextrose-derived film former and amount of cellulose derived-film formers). Figure S8: PXRD patterns of raw materials and AmB-loaded ODF before and after DVS analyses. Key: AmB ODF post DVS, (b) AmB ODF, (c) physical mixture, (d) Avicel 200, (e) HPMC AS; (f) HPC, (g) maltodextrin, (h) dextran, (i) sorbitol, (j) sodium deoxycholate, (k) AmB. Figure S9: FT-IR spectra. (a) HPC, (b) maltodextrin, (c) dextran, (d) Avicel 200, (e) sodium deoxycholate, (f) sorbitol, (g) HPMC 912 AS, (h) AmB, (i) physical mixture, (j) AmB-loaded. Table S1: Target product profile (TPP) elements for AmB-loaded ODFs. Table S2: Critical quality attributes (CQAs) of AmB-loaded ODFs. Table S3: Co-efficient values and statistical parameters obtained for first order equations for the studied response variables: 1-Type of dextrose-derived-polymer film former, 2-Taste masking agent, 3-Type of Avicel, 4-Amount of Avicel, 5-Amount of plasticisers, 6-Amount of methanol, 7-Amount of cellulose-derived film formers. Results were analysed using a first order equation (*Y* = *B*<sup>0</sup> + *B*1*X*<sup>1</sup> + *B*2*X*<sup>2</sup> + *B*3*X*<sup>3</sup> + *B*4*X*<sup>4</sup> + *B*5*X*<sup>5</sup> + *B*6*X*<sup>6</sup> + *B*7*X*7) generated for the response variables investigated in the DoE. Seven coefficients (B1 to B7) were calculated with B0 as the intercept. Only those coefficients which were significant were retained in the simplified equations.

**Author Contributions:** Conceptualization, D.R.S. and A.L.; methodology: D.R.S. and A.L.; formal analysis, D.R.S., R.F.-G., M.M. and A.L.; resources, A.M.H. and A.L.; software: A.M.H. and A.L.; supervision: D.R.S. and A.L.; validation: A.L.; writing—original draft preparation, D.R.S., A.M.H. and A.L.; writing—review and editing, D.R.S., A.M.H. and A.L., supervision, D.R.S. and A.L.; project administration, A.L.; funding acquisition, A.M.H. and A.L.

**Funding:** This work was supported by a Sir Haley Stewart Trust grant (127) and a University of Portsmouth, Research and Development Fund to A. Lalatsa and a Science Foundation Ireland grant co-funded under the European Regional Development Fund (SFI/12/RC/2275) provided to A. M. Healy.

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).

*Article*
