Synergistic efficacy of colistin and silver nanoparticles impregnated human amniotic membrane in a burn wound infected rat model

Antimicrobials used to treat burn wound infections have become multidrug-resistant, thus delaying wound healing. When combined with silver nanoparticles, antibiotics create a multifaceted antibacterial mechanism of action to which bacteria are incapable of developing resistance. Similarly, the amniotic membrane has been found to lower the bacterial number. The purpose of the current study was to observe the antibacterial activity of combined topical colistin with silver nanoparticles and decellularized human amniotic membrane as a dressing in burn wounds infected with bacteria with the goal of promoting faster healing. Bacteria commonly isolated from burn wounds and the most sensitive topical antibiotic were identified. Colistin, silver nanoparticles and combined colistin with silver nanoparticles were impregnated into decellularized human amniotic membranes. These wound dressings were evaluated in third-degree multidrug-resistant bacterial infected thermal burns induced in rats. Out of a total of 708 pus samples from burn wounds, Pseudomonas aeruginosa was the most prevalent pathogen 308 (43.5%), followed by Klebsiella pneumoniae 300 (42.4%). Topical colistin was 100% sensitive for both bacteria. Overall, maximum wound contraction (p < 0.05), and increased collagen deposition (+++) with no isolation of bacteria from wound swabs were noted on day 21 for the combined colistin with silver nanoparticle-loaded human amniotic membrane dressing group. Our study concluded that the increased antimicrobial activity of the novel combination of colistin and silver nanoparticle-loaded decellularized human amniotic membrane manifested its potential as an effective burn wound dressing.

Infection is the main reason for mortality after severe burns despite developments in burn care management. Due to the lack of an epidermal barrier, bacteria can colonize and rapidly multiply after burn injuries 1,2 . Pseudomonas aeruginosa (P. aeruginosa), Klebsiella pneumoniae (K. pneumoniae) and Staphylococcus aureus (S. aureus) are among the commonly isolated pathogenic bacteria from infected burn victims. However, clinicians are faced with an additional arduous task of managing these antibiotic-resistant pathogenic infections 3 . Colistin has resurfaced as a last resort antibiotic for gram-negative bacterial infections resistant to other antibiotics 4 . It operates by binding to anionic lipid A molecules of gram-negative bacteria and alters the permeability of the cell wall, ultimately leading to cell leakage and death of bacteria 5 . However, the nephrotoxicity and neurotoxicity of colistin as a side effect are of serious concern. Similarly, antibiotic therapeutic choices have been severely constrained due to the steady rise in polymyxin resistant species 6,7 .
To make antibiotics that are already in use more effective, antibiotics have been combined with silver nanoparticles (AgNPs) to maximize the antibacterial effect 8 . AgNPs have adequate antibacterial action, making them excellent candidates for exploration as alternative agents to treat multidrug-resistant (MDR) infections 9 . Their particle size in nanometers is observed to have an increased surface area which can destroy the membrane and Identification and antibiotic sensitivity testing of commonly isolated bacteria from burn wound samples of patients. A total of 708 pure pus and pus swabs were collected from hospitalized burn ward patients, applied to Blood agar and MacConkey agar and incubated at 35 ± 2 °C for 16-18 h. Further standard microbiology protocols were used to identify the bacterial isolates. Antibiotic sensitivity testing was performed by the Kirby Bauer disc diffusion method by mixing one to two colonies of commonly isolated bacteria into a tube containing 5 mL of sterile normal saline (0.5 McFarland solution) and comparing it with 0.5 McFarland standard. Lawns of each bacterial suspension were made on MHA using sterile cotton swabs, and commercially available standard antimicrobial discs for gram-negative bacteria were applied on it. The discs employed were 10 µg of ampicillin, colistin, gentamicin, imipenem and meropenem; 5 µg of ciprofloxacin and levofloxacin; 20/10 µg of amoxicillin-clavulanate, cefepime, ceftazidime, aztreonam and amikacin; 10/10 µg of ampicillin-sulbactam; and 100/10 µg of piperacillin-tazobactam. The sensitivity plates were incubated at 35 ± 2 °C for 16-18 h, and zone sizes were interpreted according to Clinical and Laboratory Standards Institute (CLSI) 2020 criteria 16,17 .
Bacteria commonly isolated from burn wounds were considered for study. Three MDR bacteria (resistance to one agent in three or more antibiotic categories) from these commonly isolated bacteria with minimum inhibitory concentrations < 2 μg/mL (considered sensitive for colistin) by the VITEK 2 system were selected www.nature.com/scientificreports/ for further study 17,18 . Similarly, the maximum sensitivity exhibited by topical antibiotic against the commonly isolated bacteria was selected for further experimentation.

Concentration determination of colistin and AgNPs in-vitro. The stock solutions of colistin and
AgNPs were prepared by adding 150 mg of colistin sulfate powder and AgNP powder separately into 100 mL of de-ionized water (20 min of sonication of AgNPs suspension with vortexing repeatedly for 5 days) in screwcapped glass bottles. A volume of 15.6 mL from the respective stock solutions was picked up and then added individually into 50 mL of distilled water in other screw-capped bottles to make 0.468 mg/mL dilutions of colistin and AgNPs for final use 17,19,20 . The formula used to calculate dilutions of colistin and AgNPs can be found in Supplementary Fig. 1 online.
Colistin and AgNPs impregnation into dHAM. All scheduled cesarean sections were included to take the placenta and were prescreened for transmissible diseases by screening kits. The amnion was isolated from the chorion, and the decellularization process was initiated by snap freezing, employing liquid nitrogen for 1 min to detach the monolayer of amnions' epithelium. Confirmation was performed through H&E staining of the samples and histological observation by light microscopy 21 .
A total of four (2 × 2 cm) lyophilized membranes were placed in 4 wells of a total 6-well (cell culture) plate. In the 1st well, 2 mL of distilled water was added as a negative control. Colistin (2 mL) was added to the 2nd well, the 3rd well with 2 mL of AgNPs and the 4 th well with 2 mL of AgNPs (1 mL) and colistin (1 mL), where in all of these cases, a colistin and AgNP dilution of 0.468 mg/mL was used. The plate was then kept overnight on the shaker.

In-vivo efficacy of colistin and AgNPsimpregnated into dHAM in rats. Thirty Sprague Dawley
rats were purchased from the animal unit of The University of Lahore, Lahore. The rats weighed approximately 200 g-250 g and were kept in a regulated room temperature (22 ± 2 °C) with unrestricted food and water for a week. Ketamine and xylazine were injected intraperitoneally to anesthetize these healthy rats. A solid iron bar of 2 × 2 cm was dipped in boiling water (at 100 °C) and thermal injury was induced by keeping this hot bar in contact for 30 s on the dorsum of the rat 22 . Evaluation of the wound was performed, and surgical debridement of the wound under general anesthesia was performed to remove necrosis the next day 23,24 .
Three MDR strains each for P. aeruginosa and K. pneumoniae (10 6 CFU/mL) were inoculated through the intradermal route. After 24 h of contamination of rat wounds, the animals were randomly divided into five dressing groups: the positive control (without any treatment) group, negative control (distilled water-dHAM) group, colistin-dHAM group, AgNPs-dHAM group and colistin + AgNPs-dHAM group. All wounds were secured with bandage afterwards. The experiment was performed in triplicate.
The bandages were opened on days 7, 14 and 21, and wound contraction was noted. The wound area was retraced on millimeter scale graph paper from the tracing paper, and wound closure was expressed as a reduction in the percentage of original wound size.
% wound contraction on day X = [(area on day 0 -open area on day X)/area on day 0] × 100. The American Veterinary Medical Association (AVMA) Guidelines for Euthanasia of Animals (2020) were followed. Tissue samples of these healed rat wounds were removed on day 21 and immediately fixed by dipping in 10% neutral buffered formalin, followed by routine histological processing. Tissues were subjected to microtomy with 4 μm cuts, and finally, H&E staining was performed. The histological analysis of wound healing was performed by a trained histopathologist. Re-epithelialization and other histological healing parameters were noted according to the following criteria: (1) Inflammatory response, featured by the presence of polymorphonuclear neutrophils (2) Granular tissue indicated by the presence of fibroblasts, myofibroblasts, and neovascularization (3) Fibrosis, indicated by the density of collagen fibers. A score was made for all parameters: − = absent, + = mild presence, ++ = moderate presence, and +++ = strong presence 25 .
Determination of antimicrobial activity. The swabs were taken from all of the wounds on day 21 and were applied separately to the Blood and MacConkey media to observe the growth of any bacteria. Bacterial morphology, staining and biochemical testing were performed to identify the bacteria 26 .
Statistical analysis. The analysis was carried out with the help of the Statistical Package for Social Sciences (SPSS) version 24 (IBM, USA). Percentages, frequencies, means, medians and standard deviations were calculated for variables wherever applicable. Data was observed to be normally distributed by applying Shapiro-Wilk statistical test. To compare the efficacy of five groups, one-way ANOVA was performed, and Dunnett's test was used for pairwise comparisons to observe the effect of time and treatment groupswith p < 0.05 was considered statistically significant. www.nature.com/scientificreports/

Frequency of bacteria isolated from burn wound samples of patients.
Of the 708 total isolated bacteria from pure pus and pus swabs from burn wounds, P. aeruginosa was isolated a predominant isolate, 308 (43.5%), followed by K. pneumoniae, 300 (42.4%), as summarized in Table 1. However, there was minimal difference between their number of isolations and other gram-negative bacteria and gram-positive cocci were isolated less frequently.

Wound healing of burn wounds in rats against P. aeruginosa and K. pneumoniae. The wound
contractions of the different dressing groups used against P. aeruginosa and K. pneumoniae and the entire findings of the statistical analysis are reported in Table 4. The most significant changes were observed on day 7, where the colistin-dHAM dressing group initially accelerated wound contraction compared to the other dressing groups but finally achieved less contraction than the combined colistin + AgNPs-dHAM dressing group on day 21 for both bacteria. Furthermore, the different dressing groups under study were highly significant with regard to their performance (p < 0.05). The administration of combined colistin + AgNPs-dHAM dressing group in the burn wound rat model significantly (p < 0.05) increased the wound contraction rate compared to the positive www.nature.com/scientificreports/  www.nature.com/scientificreports/  Histological features of rat wound healing against P. aeruginosa and K. pneumoniae. Figures 3a-e and 4a-e show the re-epithelialization of rat wound healing against P. aeruginosa and K. pneumoniae, respectively, on day 21. The re-epithelialization was complete with some differences. The animals in all treated groups reformed epidermis except the AgNPs-dHAM dressing group of P. aeruginosa and positive control group of K. pneumoniae. However, the negative control dressing of treated animals against K. pneumoniae displayed the highest thickness of epithelium in comparison to other dressing groups. Figures 3A-E and 4A-E show the histological features of rat wound healing against P. aeruginosa and K. pneumoniae, respectively, on day 21 after inducing thermal burns. The animals in the combined colistin + AgNPs-dHAM dressing groups against P. aeruginosa and K. pneumoniae reported higher collagen scores (3 ± 0; and 3 ± 1, respectively), with mild to absent inflammatory (2 ± 1; and 0 ± 1, respectively) and granulation tissue (2 ± 1; and 0 ± 1, respectively) responses on day 21. The AgNPs-dHAM of both bacteria and colistin-dHAM of K. pneumoniae revealed a strong presence of fibroblasts, myofibroblasts and new vessels on day 21, as shown in Table 5.
Wound swabbing results on day 21 against P. aeruginosa and K. pneumoniae. Culture swabs from bacterial wounds on the 21st day displayed variable results in negative controls of P. aeruginosa and K. pneumoniae (Fig. 5). Growth of S. aureus was obtained in the negative control of K. pneumoniae (Fig. 5E). However, no growth was observed in the negative control of P. aeruginosa (Fig. 5B). Furthermore, there was growth of P. aeruginosa and K. pneumoniae in their respective positive controls, as shown in Fig. 5C and F, respectively. In contrast, rest of the dressing-treated wounds remained negative for bacterial growth.

Discussion
The present study was aimed to determine the effects of newly designed wound dressings by assessing wound contraction, histopathology and the presence of bacteria.
The results showed that combination treatment with colistin and the AgNPs-dHAM dressing considerably reduced the wound area effectively for P. aeruginosa and K. pneumoniae. Surprisingly, in the case of K. pneumoniae, colistin combined with the AgNPs-dHAM dressing groups showed better wound contraction results than P. aeruginosa. The literature proves that using nanoparticles with a smaller diameter (< 30 nm) increases the antibacterial activity of AgNPs against K. pneumoniae and S. aureus 27 . Marsit et al., (2019) reported excellent antibacterial activity of dried amniotic membranes loaded with polymyxin B against P. aeruginosa 20 . There could be several possible explanations for our results. hAM is an appropriate biological scaffold for the seeding and multiplication of cells with an improved capacity to take up antibiotics. Lyophilization has additional, improved adhesion properties 12,28 . Similarly, AgNPs show remarkable antibacterial synergistic effects when combined with colistin and other antibiotics 29,30 . Moreover, AgNPs can easily control bacterial drug resistance. For this purpose, AgNPs employ many bactericidal processes working simultaneously, which could explain the reason for the uncommon resistance of bacteria to silver 31 .
Rapid re-epithelialization is a symbol of healthy wound healing 32 . Regarding histopathological analysis on day 21, complete re-epithelialization and dermal findings were encouraging in the case of K. pneumoniae. However, contradictory findings were observed in the case of P. aeruginosa, where unexpectedly, the AgNPs-dHAM Table 5. Histological features of rat wound healing against P. aeruginosa and K. pneumoniae on day 21. 0 = Absent, + = 1, + + = 2, + + + = 3.

Groups
Polymorphonuclear neutrophils www.nature.com/scientificreports/ dressing group failed to re-epithelilalize completely. Similarly, mild to moderate dermal damage was evident in all dressing groups used to treat P. aeruginosa-infected burn wounds. These findings were almost consistent with a previous study where moderate epidermal necrosis was observed on the 28th day with a moderately disordered arrangement of the dermis 11 . Further histopathological analysis of these tissues revealed that the strongest immune cell response was recorded in the negative control group of P. aeruginosa. However, the fibrosis was maximum in the combined colistin with AgNPs-dHAM dressing group on day 21 for both bacteria. Complete re-epithelialization with moderate angiogenesis evidence and fibrous tissue was observed in the dermis of AgNP-treated mice on the 28th day 10  www.nature.com/scientificreports/ Combining colistin with the AgNPs-dHAM dressing accelerated wound healing, probably by decreasing bacterial colonization, as evidenced by the absence of bacteria on day 21 of wound healing. Wound healing is a multistage process that includes three phases. The inflammatory phase is critical for cell debris removal, cytokine release and a comprehensive counter to pathogenic infection. Decreased bacterial counts in tissues minimize the inflammatory phase. The starting point for granulation tissue production and a speedy wound healing is dependent on shortening of the inflammatory step at an early beginning of the proliferative phase 35 . Earlier observations also confirmed the role of AgNPs when combined with other agents, can decrease the duration of wound healing phases 36 .

Conclusion
Our study concluded pronounced synergistic effects of combined colistin and AgNP-loaded dHAM dressing compared to positive control, negative control, colistin-dHAM and AgNPs-dHAM dressings in rats infected with P. aeruginosa and K. pneumoniae. This was confirmed by the achievement of faster wound reduction, the presence of considerable fibrosis, complete epithelial reorganization and the absence of bacteria on day 21. The friendly biological synthesis of AgNPs with easy procurement and simple processing of membranes makes them a cost-effective option to be used as a burn dressing commercially due to their improved antibacterial and healing properties. These findings may provide a gateway for other researchers to counteract MDR pathogens in burn wounds and aid clinicians in effectively managing such patients. Immunohistochemistry and molecular aspects were not addressed in our study and this work is not applicable to other species of rats. Further long term in-vitro and in-vivo studies should be carried out to confirm the functionality of the designed wound dressing.