Gellan gum spongy‐like hydrogel‐based dual antibiotic therapy for infected diabetic wounds

Abstract Diabetic foot infection (DFI) is an important cause of morbidity and mortality. Antibiotics are fundamental for treating DFI, although bacterial biofilm formation and associated pathophysiology can reduce their effectiveness. Additionally, antibiotics are often associated with adverse reactions. Hence, improved antibiotic therapies are required for safer and effective DFI management. On this regard, drug delivery systems (DDSs) constitute a promising strategy. We propose a gellan gum (GG)‐based spongy‐like hydrogel as a topical and controlled DDS of vancomycin and clindamycin, for an improved dual antibiotic therapy against methicillin‐resistant Staphylococcus aureus (MRSA) in DFI. The developed DDS presents suitable features for topical application, while promoting the controlled release of both antibiotics, resulting in a significant reduction of in vitro antibiotic‐associated cytotoxicity without compromising antibacterial activity. The therapeutic potential of this DDS was further corroborated in vivo, in a diabetic mouse model of MRSA‐infected wounds. A single DDS administration allowed a significant bacterial burden reduction in a short period of time, without exacerbating host inflammatory response. Taken together, these results suggest that the proposed DDS represents a promising strategy for the topical treatment of DFI, potentially overcoming limitations associated with systemic antibiotic administration and minimizing the frequency of administration.

Multiple bacterial species are often found in DFI, from which Staphylococcus aureus predominates as the most frequently isolated and virulent pathogen. 5,7,9,10 Methicillin-resistant Staphylococcus aureus (MRSA), in particular, represents a significant health threat in hospital and community settings, with a prevalence ranging 15%-40%. 11,12 A wide range of antibiotics of different classes has been used for DFI treatment, isolated or combined. 12,13 For MRSA infections, vancomycin (VAN) and clindamycin (CLD) represent two frequently used antibiotics. [13][14][15] VAN is a glycopeptide administrated parenterally (500-1000 mg IV every 12 h) due to its poor bioavailability. Although some isolates of VAN resistant S. aureus have recently emerged, this antibiotic remains one of the first choices for severe MRSA infections. 16 CLD belongs to the class of lincosamides and is one of the first line antibiotics for the treatment of mild and moderate DFI, either oral (PO) or intravenous (IV) route (600-900 mg PO every 6 h, or 600-2400 mg IV once daily). 12,13 CLD can also be used to treat severe DFI, although in these cases it should be combined with other antibiotics, such as VAN. 13 Although crucial to treat DFI, antibiotic therapy faces some challenges and drawbacks. In the context of DFI, one of the most important factors limiting antibiotic therapy is the peripheral vascular disease that hinders antibiotic delivery and penetration into infected tissues at effective concentrations. Consequently, pharmacological activity of antibiotics can be compromised, resulting in failure to treat DFI, as well as an increased risk of emergence of antibiotic resistance. 7 Another limitation of antibiotic therapy is the occurrence of adverse effects. [17][18][19][20] Therefore, new and more effective therapeutic strategies are needed to improve antibiotic therapy in eradicating complicated DFI caused by MRSA, ensuring therapeutic concentrations at the target site, while minimizing potential systemic adverse effects and the risk of bacterial resistance. In this regard, drug delivery systems (DDSs) have evolved as an alternative strategy for circumventing the constraints of conventional therapy. Furthermore, different structures of DDS have been proposed in the literature, for example, particulate systems, 21,22 nanofibers, 23 hydrogels, [24][25][26][27][28][29] polymer-drug conjugates, 30 according to the physicochemical properties of the drug and to grant the benefits of the intended therapeutic administration route. Here, we propose an alternative therapeutic approach for MRSA-infected diabetic wounds, based on a topical administration of VAN and CLD using gellan gum (GG)-based spongy-like hydrogels. GG is an extracellular bacterial polysaccharide that naturally forms thermoreversible physical hydrogels and is achieving promising results for drug delivery [31][32][33][34][35][36][37] and wound healing purposes. [38][39][40][41] In particular, GG-based spongy-like hydrogels possess unique features that can benefit the wound healing process, including (i) improved mechanical properties which grant wound adaptability and manipulation without break; (ii) water retention capability allowing wound exudate absorption and moisture retention for wound hydration; (iii) act as a regenerative template. 38,42,43 These features combined with an off-the-shelf availability and a simple method of production, turns GG-based spongy-like hydrogels an appealing DDS in the context of local antibiotic therapy.
Presently, we developed GG-based formulations loading VAN and CLD that possess suitable physicochemical properties for a controlled and topical delivery of both antibiotics. The antibacterial activity of loaded GG-based spongy-like hydrogels against MRSA, as well as their potential to reduce antibiotic toxicity were firstly assessed in vitro and further corroborated in vivo using a diabetic mouse model of MRSAinfected wounds. Proposed formulation showed a good antibacterial activity, while significantly reduced the intrinsic antibiotic cytotoxicity without aggravating the host inflammatory response.

| In vitro antibacterial effect of interaction between VAN and CLD
We first conducted a checkerboard assay to confirm the susceptibility of MRSA strain to VAN

| Physicochemical properties of loaded GGbased formulations
Given the additive effect of VAN and CLD against MRSA, we next characterized the structure of loaded GG-based structures to assess F I G U R E 1 Representative result of the checkerboard assay of VAN and CLD against MRSA. Resazurin reduction to resofurin by metabolically active bacteria is indicated by color transition from blue to pink. Close circles indicate the MIC of VAN and CLD when tested alone and open circle shows the MIC of their combination. C+ corresponds to positive control wells (bacteria) and CÀ corresponds to negative control wells (medium).
the impact of antibiotic incorporation on the physical properties of these formulations and to identify any potential polymer-antibiotics interactions. Loaded GG-based structures revealed a porous and crosslinked internal network, similar to unloaded structures ( Figure 2a) that, upon rehydration, granted a water uptake capacity superior to 2000%, within the first 30 minutes of incubation. This uptake capacity remained stable through the following incubation time (Figure 2b), without significant differences when compared to unloaded GG-based spongy-like hydrogels ( p > 0.05).
Fourier transform infrared spectroscopy-attenuated total reflectance (FTIR-ATR) analysis of loaded GG-based structures (Figure 2c) revealed spectral peaks characteristic of antibiotics and polymer, with no alterations in the characteristic peaks or in the appearance of new functional groups. Specifically, the infrared (IR) spectrum of loaded GG-based structures showed small peaks in the range of 1700-1600 cm À1 related to the C O stretching of amide carbonyl group of VAN 45 and CLD, 46,47 [48][49][50] The release profile of VAN and CLD from GG-based spongy-like hydrogels was characterized by a burst release within the first 8 h of incubation, followed by a sustained release (Figure 2d

| In vitro antibacterial activity and cytocompatibility of loaded GG-based spongy-like hydrogels
We then evaluated the suitability of the attained release profile over the in vitro antibacterial activity of loaded GG-based spongy-like hydrogels against MRSA. Different concentrations of antibiotics were tested using the broth dilution assay, to assess a possible doseresponse effect (Table 1). Figure 3a shows that all tested concentrations of loaded antibiotics resulted in a complete inhibition of metabolic viability of MRSA, in addition to a drastic reduction of bacterial replication capacity. In this regard, loaded GG-based spongy-like hydrogels-1 to -3 led to approximately half decay of Log 10 colony forming units (CFU) in comparison to control, while no bacterial growth was observed for loaded GG-based spongy-like hydrogel-4.
Moreover, in comparison to the corresponding free VAN/CLD solutions, loaded GG-based spongy-like hydrogels presented a similar antibacterial activity, except for loaded GG-based spongy-like hydrogel-3, in which a higher bacterial growth inhibition was observed for free VAN/CLD solution ( p < 0.01). Moreover, unloaded GG-based spongy-like hydrogels did not present any antibacterial effect (p > 0.05 compared to control). The antibacterial effect of loaded GG spongy-like hydrogels-4 was further confirmed in an agar diffusion assay. A mean inhibition zone of 34.5 mm was observed (Figure 3b), in opposition to its absence for unloaded GG-based spongy-like hydrogels (0.0 mm), in agreement with the lack of antibacterial effect found in the broth antibacterial assay.
Furthermore, in vitro cytocompatibility of loaded GG-based hydrogels was assessed in L929 fibroblasts, to determine the potential of developed DDS in reducing free antibiotic-associated toxicity. Viability of cells treated with loaded GG-based spongy-like hydrogels-1 to -3 was superior to 80%, while loaded GG-based spongy-like hydrogel-4 reduced cell viability in about 50% ( Figure 3c). All tested free VAN/CLD solutions reduced cell viability by more than 40%, and this effect increased as higher concentrations were tested. Treatment with unloaded GG-based spongy-like hydrogels resulted in a cell viability similar to untreated cells (control). Immunological analysis was also performed to evaluate a possible effect of the treatment on the immunomodulation of the host defense. Several cytokines and chemokines were quantified in the wound tissue as well as in the serum. Overall, tissue analysis revealed an increment of most of the tested immune mediators from 3 to 7 dpt and no significant differences between treatments were observed, except for the pro-inflammatory cytokine IL-23 ( Figure 6).

| Biological effect of loaded GG-based spongylike hydrogels in a diabetic mouse model of MRSAinfected wounds
In this case, a significant decrease was observed for the group treated with loaded GG-based spongy-like hydrogel in relation to the unloaded GG-based spongy-like hydrogel and control groups a similar profile was observed between the different experimental groups, noticing only a significant difference for G-CSF at 3 dpt between loaded GG-based spongy-like hydrogel and control ( p < 0.01). Levels of lL-10 were undetectable in the serum at both timepoints.

| DISCUSSION
One of the most common complications arising from diabetes is DFI, in which S. aureus is frequently the causative pathogen. Treating S. aureus infections has been increasingly challenging, owing the ability of this pathogen to develop resistance to several antibiotics of choice. 51 Indeed, the emergence of antibiotic resistance is an important public health issue that has been threatening the efficacy of currently available antibiotics. Moreover, in the past years, new classes of antibiotics have been hard to find in the market, given the drastic reduction of investment of the pharmaceutical industry on antibiotic research due to scientific, regulatory, and financial constraints. 52,53 Under these circumstances, combination therapy using two or more antibiotics has been an attractive approach for difficult-to-treat bacterial infections, minimizing the risk of antibacterial resistance and potentially contributing to an increased drug efficacy. 54   Immunological analysis confirmed that inflammation was local, rather than systemic, and persisted from day 3 to day 7 post-treatment.
These data also suggested that treatment did not trigger an exacerba-

| Preparation of GG-based spongy-like hydrogels
GG-based spongy-like hydrogels were obtained as previously described. 38 Briefly, Gelzan™ CM powder was dissolved at 0.75% in distilled water at

| Morphological analysis
Morphology of loaded and unloaded GG-based structures was analyzed using a high-resolution field emission scanning electron microscope (SEM, AURIGA Compact, Zeiss, Germany) equipped with energy dispersive electron X-ray (EDX) spectroscopy (Bruker QUAN-TAX ESPIRIT 2.0 EDS system, X-flash detector, Germany). Previously, samples were placed onto metal holders using carbon double-sided tape and platinum sputter coated (Leica EM ACE600, Leica Microsystems, Germany).

| Water uptake and retention capacity
where W t is the weight of the spongy-like hydrogel at a predetermined time (t) and W i is the initial weight of dried GG structure.

| FTIR-ATR
Possible antibiotic-polymer molecular interactions were analyzed by FTIR-ATR using an IR Prestige-21 spectrometer (Schimadzu, Japan) with an ATR sampling mode. Transmittance spectra of loaded GGbased structures and raw materials (VAN, CLD, and GG) were recorded at 40 scans in the spectral range of 4400-700 cm À1 , with a resolution of 4 cm À1 .

| In vitro antibiotic release
The release of antibiotics from loaded GG-based spongy-like hydro- Release data were analyzed by fitting to mathematical models, including zero-order release kinetics, first-order release kinetics, T A B L E 2 LC-MS ion acquisition parameters in dynamic MRM mode for VAN and CLD detection. VAN  Higuchi model and Korsmeyer-Peppas model, to estimate the mechanism(s) governing antibiotics release. The adjusted coefficient of determination (R 2 adjusted ) was used as the indicator of the best fitting model.

| Antibacterial activity
Broth microdilution and agar diffusion assays were used to assess the antibacterial activity of loaded GG-based spongy-like hydrogels. An inoculum was prepared as described for checkerboard assay.
For broth microdilution, inoculum was first diluted in MH broth to 10 6 CFU/mL and then transferred to respective wells of a 96-well plate (100 μL/well). For testing loaded GG-based spongy-like hydrogels, MH broth (100 μL) was added to respective wells, followed by loaded GG-based structures. Free VAN/CLD solutions were also tested by dilution with bacterial suspension. Unloaded GG-based spongy-like hydrogels were used as material control. Positive (bacteria only) and negative (medium only) growth controls were also used.
Plates were then incubated at 37 C for 24 h. Bacterial metabolic viability was assessed using resazurin assay as described for checkerboard assay and measuring optical density at 575 and 610 nm wavelengths (Infinite ® 200 microplate reader, Tecan, Switzerland). In parallel, viable colonies were quantified by performing serial 10-fold dilutions of test wells and further plating on MH agar. After incubation at 37 C for 24 h, CFU were counted and presented as Log 10 CFU.
Regarding agar diffusion assay, 10 8 CFU of freshly prepared inoculum were seeded on MH agar. Loaded GG-based structures-4 and unloaded GG-based hydrogels (control) were placed on top and hydrated with PBS (5 μL). After incubation at 37 C for 24 h, the zones of bacterial growth inhibition (mm) were measured using a ruler.

| Cytocompatibility
Cytocompatibility of loaded GG-based spongy-like hydrogels was compared with free VAN/CLD solutions and analyzed by the metabolic activity of mouse fibroblasts L929 (85011425, European Collection of Authenticated Cell Cultures). Cells were cultured in DMEM supplemented with 10% FBS/1% L-glutamine, in a humidified atmosphere at 37 ± 1 C and 5% CO 2 .
Briefly, cellular suspensions were prepared at 10 5 cells/mL, seeded in 48-well plates (100 μL/well) and incubated for 24 h for adhesion. Afterwards, 0.2 mL of culture medium was added and loaded GG-based structures were placed onto the wells. For free VAN/CLD wells, 0.2 mL of antibiotics solution was added directly.
Unloaded GG-based spongy-like hydrogels were included as material control, as wells as cells only (positive control) and culture medium only (negative control). Cells were incubated for 3 days at 37 ± 1 C and 5% CO 2 , and then tested for metabolic viability using the resazurin assay as described for broth microdilution assay. Cell viability reduction higher than 30% was considered for cytotoxicity, according

| Diabetes induction
Type 1 diabetes mellitus was induced by intraperitoneal injection of STZ (50 mg/kg) for 5 consecutive days, as previously. 73 During the injection period, mice were given 10% sucrose water to prevent fatal hypoglycemia. Nine days after treatment, mice were fasted for 6 h and a blood sample was collected from the tail vein for glucose measurement (FreeStyle Precision Neo, Abbott Laboratories, USA). Glucose levels higher than 150 mg/dL were considered for hyperglycemia.

| MRSA inoculation of polycarbonate membranes
Polycarbonate membranes, 0.2 μm pore size (Merck KGaA, Germany) were used for MRSA growth and further wound infection, as previously. 46 First, membranes were cut in 5-mm diameter discs and sterilized on both sides by UV light for 30 min. Membranes were then placed on MSA and inoculated with 10 2 CFU of MRSA, previously diluted from a 10 8 bacterial suspension prepared as described for checkerboard assay, and incubated overnight at 37 C.

| Mouse model of MRSA-infected wounds
Mouse model of MRSA-infected wounds was created following our previously optimized protocol. 74 On the day before surgery, dorsal hair was removed by using a hair clipper and then applying a depilatory cream (Veet). For surgery, mice were anesthetized (75 mg/kg ketamine and 1 mg/kg medetomidine), placed on their side, and dorsal skin was pulled and punched through the folded skin using a 5-mm biopsy. A silicone splint ring (15-mm external diameter and 6-mm internal diameter) was secured around each wound with cyanoacrylate glue and four interrupted sutures of 5/0 nylon to prevent wound contraction.
Wounds were infected with MRSA-inoculated polycarbonate membranes (approximately 10 9 CFU), by placing the biofilm in direct contact with the wound bed. Wounds were covered with Durapore™ self-adhering bandage (3M, USA) and mice were allowed to fully recover from anesthesia under a warming lamp. From the day of surgery and during the following 2 days, analgesia and vitaminic supplementation was subcutaneously administered for postoperative pain relief and hydration of animals. After 2 days, mice received a lower dose of anesthesia, and polycarbonate membranes were removed before treatment application. Finally, a sterile transparent semiocclusive dressing Tegaderm (3M, USA) was applied covering the wounds and splints, followed by an Omnifix elastic bandage (Hartmann).

| Immunological assays
Quantification of cytokine and chemokine in serum and in wound tissue was performed using the LEGENDplex™ multi-analyte flow assay kit, according with manufacturer's instructions.
For serum sampling, whole blood was collected from the orbital sinus, centrifuged (2000 g/10 min) and serum was aliquoted and stored at À80 C until usage.
For wound tissue collection, the right wound of each mouse was cut across the midline and one of the half-portions was immediately frozen in liquid nitrogen and stored at À80 C until further processing for protein extraction. For this purpose, tissue samples were minced by manual grinding in ice-cold PBS containing 1% EDTA and protease and phosphatase inhibitors. Homogenates were vortexed and centrifuged (13,000 g/15 min/4 C). The supernatant containing soluble protein was collected, aliquoted and stored at À80 C until analysis.

| Wound bacterial burden quantification
Left wound was harvested, minced in a petri dish, resuspended in PBS and vortexed with 2-mm glass beads. Ten-fold serial dilutions of wound homogenates were plated onto MSA and incubated at 37 C for 24 h before CFU quantification. Bacterial burden was expressed as Log 10 CFU/wound.

| Histological analysis
Half-portion of right wound tissue was fixed in 10% neutral buffered formalin (at room temperature and mild stirring, 24 h) and further embedded in paraffin, sectioned (4-μm thickness) and stained with Hematoxylin and Eosin (H&E) using standard protocols.

| Statistical analysis
Data was expressed as mean ± standard deviation (SD) of at least three independent experiments. Normality was tested using Shapiro-Walk test and, if observed normal distribution and variance homogeneity, One-way analysis of variance (ANOVA) with Tukey's multiple comparisons test was used. Otherwise, data was analyzed using Kruskal-Wallis test with Dunn's multiple comparisons test. In vivo data was analyzed using two-way ANOVA with Sidak's multiple comparison test. Significance levels were set as *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. Analyses were performed using GraphPad Prism 7.0a (GraphPad Software Inc., USA).

| CONCLUSION
We proposed an antibiotic topical delivery system using a GG-based spongy-like hydrogel as a therapeutic approach targeting MRSA-DFI,

CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest to declare.