Comparative Analysis of Single-Species and Polybacterial Wound Biofilms Using a Quantitative, In Vivo, Rabbit Ear Model

Introduction The recent literature suggests that chronic wound biofilms often consist of multiple bacterial species. However, without appropriate in vivo, polybacterial biofilm models, our understanding of these complex infections remains limited. We evaluate and compare the effect of single- and mixed-species biofilm infections on host wound healing dynamics using a quantitative, in vivo, rabbit ear model. Methods Six-mm dermal punch wounds in New Zealand rabbit ears were inoculated with Staphylococcus aureus strain UAMS-1, Pseudomonas aeruginosa strain PAO1, or both, totaling 10∧6 colony-forming units/wound. Bacterial proliferation and maintenance in vivo were done using procedures from our previously published model. Wounds were harvested for histological measurement of wound healing, viable bacterial counts using selective media, or inflammatory cytokine (IL-1β, TNF-α) expression via quantitative reverse-transcription PCR. Biofilm structure was studied using scanning electron microscopy (SEM). For comparison, biofilm deficient mutant UAMS-929 replaced strain UAMS-1 in some mixed-species infections. Results Bacterial counts verified the presence of both strains UAMS-1 and PAO1 in polybacterial wounds. Over time, strain PAO1 became predominant (p<0.001). SEM showed colocalization of both species within an extracellular matrix at multiple time-points. Compared to each monospecies infection, polybacterial biofilms impaired all wound healing parameters (p<0.01), and increased expression of IL-1β and TNF-α (p<0.05). In contrast, mixed-species infections using biofilm-deficient mutant UAMS-929 instead of wild-type strain UAMS-1 showed less wound impairment (p<0.01) with decreased host cytokine expression (p<0.01), despite a bacterial burden and distribution comparable to that of mixed-wild-type wounds. Conclusions This study reveals that mixed-species biofilms have a greater impact on wound healing dynamics than their monospecies counterparts. The increased virulence of polybacterial biofilm appears dependent on the combined pathogenicity of each species, verified using a mutant strain. These data suggest that individual bacterial species can interact synergistically within a single biofilm structure.


Introduction
The management and treatment of chronic wounds continues to be a significant burden on the healthcare system [1][2][3][4][5][6]. The importance of bacterial biofilms to the pathogenesis, and subsequent impaired healing, of these wounds has now been validated through a series of in vitro and in vivo studies [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. Defined as a surface-adhered, complex community of aggregated bacteria within an extracellular polymeric substance (EPS), biofilms demonstrate a number of inherent virulence, defense, and survival mechanisms within a host wound environment that differentiate them from traditionally studied, free-floating, 'planktonic' bacteria. The biofilm EPS acts a physical barrier, defending against inflammatory cell phagocytosis, while also potentially inhibiting the activation of complement and the penetration of antibiotics [23][24][25][26][27]. Other in vitro work has revealed that biofilms can shed planktonic bacteria to act as 'seeds' for the development of new, remote biofilm populations [8,9]. Biofilm survival is also enhanced by the maintenance of phenotypically distinct 'persister' cells, which provide sustainability and durability in the face of a host immune response [8,9]. Individual, species-specific virulence mechanisms, such as intricate cell-cell signaling (quorum-sensing) among P. aeruginosa cells and the presence biofilm regulatory molecules sarA, agr, and cidA in S. aureus, have also been characterized using several in vitro and in vivo models [28][29][30][31][32][33][34].
Integral to our growing understanding of chronic wound biofilms have been recent improvements in imaging and molecular sampling techniques over traditional culture-based methods [12,35,36]. In particular, these studies have indicated that the bacterial burden of chronic wounds is often underestimated [11,[37][38][39][40], with the majority of biofilms within these wounds consisting of mixed bacterial species [8,10,13]. Predominant bacteria identified have included various anaerobes, Staphylococcus aureus, and Pseudomonas aeruginosa, with one study finding an average of 5.4 species per wound [41], while others have reported upwards of 106 different possible bacterial genera within human chronic wounds [12]. Consequently, an understanding of biofilm virulence in the context of multiple, interacting bacterial species is critical for placing biofilm research within a clinical framework. This includes delineating the impact of each individual species within a polybacterial setting, and the potential for synergy between different microbes within one wound, which has been previously reported in other settings [42,43]. Although in vitro studies have provided some insight into these polybacterial interactions [44][45][46][47][48][49][50], the importance of in vivo models for understanding bacteriabacteria interactions in the face of host defenses cannot be overstated. However, with only a limited number of in vivo models available, continued research is needed.
Having previously established an in vivo model of single-species wound biofilm in the rabbit ear [51][52][53], the goal of this study was to develop and validate a quantitative, in vivo model of polybacterial biofilm in order to understand its impact on wound healing dynamics. Using two of the most commonly encountered wound pathogens, S. aureus and P. aeruginosa, we verified the presence of both species at multiple time-points, resulting in a greater impact on the host inflammatory response and wound healing in comparison to single-species biofilms. We further investigated the contribution of each species to the virulence of a polybacterial wound biofilm, and the potential for synergy between them, using a biofilm-deficient mutant strain of S. aureus within the same polybacterial environment, with an expectation that the presence of a mutant would reduce the biofilm's overall impact on the host. With these findings, we hoped to lay the foundation for identifying mechanisms critical to the establishment, and maintenance, of a polybacterial biofilm phenotype and its impact on wound healing.

Animals
Under an approved protocol by the Animal Care and Use Committee at Northwestern University, adult female New Zealand white rabbits (3-6 months, ,3 kg) were acclimated to standard housing and fed ad libitum. All animals were housed in individual cages under constant temperature and humidity with a 12-hour light-dark cycle. A total of 45 animals were used for this study.

Bacterial Strains and Culture
Wild-type and mutant strains of S. aureus, and wild-type strain of P. aeruginosa, were utilized for wound infection. These included P. aeruginosa wild-type strain PAO1, S. aureus wild-type strain UAMS-1 and its biofilm-deficient mutant strain UAMS-929. This mutant is deficient in the accessory regulator protein sarA, which is known to modulate the expression of enzymes responsible for polysaccharide intercellular adhesin formation. As one of the critical mediators of biofilm formation, the lack of polysaccharide intercellular adhesin seen with UAMS-929 has been shown to reduce its capacity to form biofilm [31], with a resultant increased susceptibility to topical antibiotics in vitro [32] and in vivo [54]. S. aureus and P. aeruginosa strains were grown overnight at 37uC on Staphylococcus and Pseudomonas Isolation Agar (Hardy Diagnostics, Santa Maria, CA), and cultured in tryptic soy (TSB) and Luria (LB) broth, respectively, at 37uC until log-phase was achieved. Bacteria were harvested and washed in phosphate-buffered saline (PBS) three times by centrifugation at 5,000 rpm for 5-minutes at 20uC. An optical density at the 600-nm wavelength (OD 600 ) was measured. An OD 600 equivalent to 10 6 colony-forming units (CFU)/mL was predetermined empirically for each strain of bacteria used.

Wound Protocol and Infection Model
Wounding, bacterial infection, and biofilm formation were adapted from principles established in our previously published in vivo, wound biofilm model [51]. Rabbits were anesthetized with intramuscular injection of ketamine (22.5 mg/kg) and xylazine (3.5 mg/kg) mixture prior to surgery. Ears were shaved, sterilized with 70% ethanol, and injected intradermally with 1% lidocaine/ 1:100,000 epinephrine at the planned wound sites. Six, 6-mm diameter, full-thickness dermal wounds were created on the ventral ear down to perichondrium and dressed with Tegaderm (3 M Health Care, St. Paul, MN), a semi-occlusive transparent film. Individual wounds were inoculated with different combinations of bacteria on postoperative day (POD) 3 as dictated by the experiment being performed. Bacterial solutions were diluted such that each wound was inoculated with a total of 10 6 CFU of bacteria at a volume of 10-mL. Polybacterial wounds were inoculated with 5-mL of each bacterial species, followed by 'mixing' of the two species-solutions in vivo with pipette tip. Bacteria were allowed to proliferate in vivo under the Tegaderm dressing. Topical antibiotics (S. aureus wounds: Mupirocin (2%) ointment [Teva Pharmaceuticals, Sellersville, PA], P. aeruginosa wounds: Ciloxan ointment [Ciprofloxacin 0.3%, Alcon, Fort Worth, TX], polybacterial wounds: combination of both antibiotics) were applied POD4 to eliminate free-floating, planktonicphase bacteria, leaving a predominately biofilm-phase phenotype. To prevent seroma formation and re-growth of planktonic bacteria, thus maintaining a biofilm-dominant infection, an antimicrobial, absorbent dressing containing polyhexamethylene biguanide (Telfa AMD, Tyco Healthcare Group, Mansfield, MA) was applied to biofilm wounds on PODs 5, 6, and then every other day until harvest. All dressings were checked daily throughout the protocol.

Harvesting of Wounds
After euthanizing the animals by intracardiac euthasol injection, wounds were harvested for various analyses. For viable bacterial count measurements, polybacterial wild-type wounds were harvested at 48-hour intervals from POD6 to POD12, while bacterial counts were measured at POD12 for all other wounds. Scanning electron microscopy (SEM) to visualize the presence, and structure, of polybacterial wild-type biofilm were performed at POD6, the beginning of the biofilm-predominant 'steady-state' [51], and on POD12. Histological analyses and reverse transcription quantitative PCR (RT-qPCR) analysis, to measure mRNA levels of inflammatory cytokines, were all performed on wounds harvested at POD12. All wounds were excised using a 10-mm (histology, SEM, viable bacterial counts) or 7-mm (RT-qPCR) biopsy punch (Acuderm inc., Fort Lauderdale, FL).

Viable Bacterial Count Measurements
The dorsal side of wounds used for bacterial counts were removed to eliminate the inclusion of bacteria outside of the infected wound surface. To recover bacteria, infected wound samples harvested at different time points were placed in tubes pre-filled with homogenizer beads (Roche, Indianapolis, IN). One-mL of PBS was added to the tube and was homogenized for 90seconds at 5,000 rpm in a MagNA Lyser homogenizer (Roche Diagnostics, Indianapolis, IN), followed by sonication (Microson Ultrasonic Cell Disrupter, Heat Systems-Ultrasonics, Inc, Farmingdale, NY) for 2 minutes at 6-8 watts to disrupt any biofilm present. The resulting solutions were serially diluted and plated on S. aureus Isolation Agar (UAMS-1 wounds), P. aeruginosa Isolation Agar plates (PAO-1 wounds), or both (polybacterial wounds) and incubated overnight at 37uC. CFUs were determined by standard colony counting method.

In vitro Bacterial Growth Curves
Growth curves for each bacterial species (UAMS-1 and PAO1) and the biofilm-deficient mutant UAMS-929 were created in vitro. A single colony of target bacteria was removed from the agar plate containing overnight culture and inoculated into 10-ml of TSB and incubated overnight at 37uC. The following morning, 100-mL were transferred from the overnight growth into 50-ml of fresh TSB. This represented time point 'zero'. Beginning at time point 'zero', a small aliquot was removed from the sample and measured spectrophotometrically at 600-nm. A second small aliquot was taken, serially diluted, and then plated on agar plates in order to obtain bacterial counts. The agar plates were incubated overnight at 37uC and counted the next morning to obtain CFU/ml. This step was repeated every 60-minutes for the duration of the experiment (10-hours for wild-type and mutant S. aureus and 18hours for wild-type P. aeruginosa). Upon completion of the experiment, the results were graphed and data analyzed. The lag, log and stationary phases of growth were identified using the graph. The log phase of growth was used to obtain the mean generation time for the target organism. The formula for obtaining the mean generation time was as follows: where g = number of generations in h number of hours, N t is number of bacteria at end of log phase, and N 0 is number of bacteria at beginning of log phase. Therefore, doubling time in hours was calculated h/g, and in minutes as (h/g) 660.

Histological Analysis
Wounds excised for histological analysis were bisected at their largest diameter for H&E staining. Tissues were fixed in formalin, embedded in paraffin, and cut into 4-mm sections. Paraffin was removed with a xylene wash, followed by a standard hematoxylin and eosin staining protocol to prepare samples for analysis under a light microscope. Slides were examined for quantification of epithelial and granulation gaps, and total epithelial and granulation areas, using a digital analysis system (NIS-Elements Basic Research, Nikon Instech Co, Kanagawa, Japan), as previously described [51]. Two blinded, independent observers evaluated all histological sections. The results of both examiners were averaged.

Scanning Electron Microscopy
To determine biofilm structure, wound samples were fixed in 2.5% glutaraldehyde in 0.1 M phosphate-buffered solution (PBS) (pH 7.2), washed 36 in PBS, and dehydrated through an ethanol series and hexamethyldisilazane. Samples were mounted by double-sided tape to specimen stubs, followed by gold-platinum (50:50) ion coating (108 Auto Sputter Coater, TedPella, Inc).
Wounds for SEM had their dorsal sides removed prior to preparation to allow for better mounting for visualization. Samples were visualized using a Carl Zeiss EVO-40 scanning electron microscope operated at the scanning voltage of 10-kV.

Total mRNA Extraction and Reverse-Transcription qPCR
Wounds were harvested for mRNA extraction and subsequent cDNA conversion as part of RT-qPCR. The dermal layer on the dorsal side of the ear was removed and the wound bed was punched out and immediately snap-frozen in liquid nitrogen. Wound samples were homogenized using a Mini-bead beater-8 equipment (Biospec Products Inc, Bartlesville, OK) using Zirconia beads (2.0 mm diameter, Biospec Products Inc) in the presence of Trizol Reagent (Sigma-Aldrich, St. Louis, MO). Total RNA was isolated according to the manufacturer's protocol. Contaminating genomic DNA during RNA preparation was removed using the Turbo DNA-free kit (Ambion, Austin, TX). Five-mg of total RNA was used to prepare cDNA using superscript II (Invitrogen) with 100-ng of random primers (Invitrogen). For quantitative analysis of the level of mRNAs, RT-qPCR analyses using SYBR green 1 were performed utilizing an ABI prism 7000 sequence detection system (Applied Biosystems, Foster City, CA). PCR primers were designed using the Primer 3 program (http://frodo.wi.mit.edu/). Expression of each gene was normalized to the level of glyceraldehyde 3-phosphate dehydrogenase (Gapdh), the house keeping gene, to get DCt.

Statistical Analysis
Data are presented in graphical form as mean 6 standard errors when applicable. Statistical analyses were performed using the Student's t-test (two-tailed, unpaired) when comparing two study groups, and the Kruskal-Wallis, one-way, analysis of variance (ANOVA) when comparing the means of multiple groups. The level of significance was set at p,0.05.

Results
The validation of our in vivo, polybacterial model required the verification of viable bacteria, both S. aureus and P. aeruginosa, within our wounds. Bacterial counts were measured in wounds inoculated with wild-type strains (UAMS-1 and PAO1) of each species every 2-days from POD6 to POD12, revealing the dynamics of the established polybacterial biofilm over time (Figure 1). Following inoculation of equal concentrations of bacteria, UAMS-1 was slightly more predominant than PAO1 at POD6. However, by POD8 and beyond, the number of PAO1 within each wound was significantly higher than UAMS-1 at each measured time point (p,0.001). Demonstrated visually, SEM of wounds at POD6 and POD12 revealed a predominance of S. aureus and P. aeruginosa, respectively (Figures 2A and 2C). However, at both time points, wounds were also found to have areas of colocalization of both species surrounded by an extracellular matrix. (Figures 2B and 2D).
Having established the simultaneous presence of both UAMS-1 and PAO1 within a single wound, we utilized our model to understand the impact of polybacterial biofilms on the host inflammatory response, relative to their corresponding singlespecies biofilm wounds (Figure 3). The mRNA levels of inflammatory cytokines IL-1b and TNF-a were measured through RT-qPCR for each species, followed by normalization to the level seen in non-wounded skin. For both cytokines, polybacterial biofilm wounds showed higher levels of mRNA expression relative to wounds with only UAMS-1 (IL-1b: p,0.01, TNF-a: p,0.001) or PAO1 (IL-1b: p,0.05, TNF-a: p,0.01), indicating that polybacterial biofilm triggered a significantly heightened inflammatory response. Histological analysis of single-and dual-species wounds revealed similar trends among the three groups (Figures 4  and 5). Visual inspection showed that polybacterial biofilms severely limited the formation of new epithelial and granulation tissue as compared to their single-species counterparts (Figure 4). Quantifying these findings across several wounds, polybacterial  Given that polybacterial wound biofilm demonstrated a distinct impact on wound healing dynamics relative to single-species biofilms, we aimed to clarify the underlying factors responsible for these differences. With the presence of two different bacteria within a single biofilm, understanding the individual contribution of each species to its overall virulence against the host was of particular importance. Using a biofilm-deficient mutant strain of S. aureus, UAMS-929, polybacterial wounds were generated through the combination of wild-type P. aeruginosa and UAMS-929, followed by comparison to wounds containing both wild-type strains. Viable bacterial count measurements at POD12 revealed the predominance of P. aeruginosa regardless of which type of S. aureus was present, wild-type (UAMS-1) or mutant (UAMS-929) (p,0.0001) ( Figure 6). Furthermore, both polybacterial wounds had a similar distribution of S. aureus to P. aeruginosa at POD12, regardless of the presence of a mutant strain within the wound. This implied that the sarA mutation, responsible for the biofilmdeficiency seen in UAMS-929, did not specifically affect its growth rate, which was confirmed through in vitro growth curves generated for each bacterial strain (data not shown).
Despite having equivalent growth rates between S. aureus mutant and wild-type strains, further analysis of S. aureus mutantcontaining polybacterial wounds revealed differences in the host inflammatory response that was triggered (Figure 7). The introduction of a UAMS-929 into polybacterial wounds caused a significant reduction in the levels of mRNA for both IL-1b and TNF-a when compared to dual wild-type (UAMS-1+ PAO1) wounds (p,0.01). The resultant level of cytokine mRNA expression in the mixed wild-type + mutant wounds was similar to the levels seen in single-species S. aureus or P. aeruginosa wounds. Correlating with these differences in cytokine expression, similar relationships were found on quantitative histological analysis. Introduction of the S. aureus mutant into a polybacterial biofilm reduced its overall impact on wound healing impairment to the level of a single-species bacterial biofilm for all measured histological parameters (Figures 8). Given the known reduction in biofilm formation associated with this S. aureus mutant [31,32,54], these findings confirmed the importance of each species' intact biofilm to the polybacterial biofilm's overall virulence.

Discussion
The importance of bacterial biofilm to the development, and maintenance, of chronic wounds is now well established within the literature [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. With continued improvements in molecular sampling techniques, an increasing amount of literature has reported on the presence of several different bacterial species within these chronic wounds [12,41]. Therefore, with the majority of research to date focused on biofilms formed by single-species, an in vivo understanding of polybacterial biofilm is critical for translating the results of future research to the wounds of human patients. To address this need, we present a polybacterial adaptation of our previously published, in vivo wound biofilm model, with the goal of clarifying the dynamics and underlying mechanisms behind polybacterial biofilm.
Previous work with polybacterial biofilm systems has primarily involved in vitro systems. Pihl et al [44] studied the interactions between Staphylococcus epidermidis and P. aeruginosa, demonstrating the P. aeruginosa may counteract the colonization of S. epidermidis potentially through the extracellular polysaccharides that it produces. Woods et al [50] recently published on their in vitro biofilm system, which combined three pathogenic bacteria into one system, finding that the final biofilm structure consisted of distinct layers dominated by one particular species, unlike the colocalization that we noted within our in vivo wounds. However, only Dalton et al [55] have previously reported an in vivo, polymicrobial biofilm model, but with significant limitations for clinical applicability. The biofilms were established in vitro, and then transferred to in vivo mouse wounds, which as noted above, would result in a different biofilm organization than the colocalization that we observed. The in vivo inoculation of wounds with free-floating, planktonic bacteria in our model allows for the simulation of clinical, wound biofilm pathogenesis. In addition, the mouse model heals almost entirely by contraction, in contrast to human wounds that heal largely by formation of new tissue. They also fail to rigorously evaluate the host response to a polymicrobial biofilm challenge, which we address through an analysis of inflammatory markers. With a superior translatability to clinical wound healing [51], and the sensitivity to quantitatively and qualitatively evaluate multiple endpoints across different bacterial species, we believe our model is distinct in its ability to evaluate the impact of polybacterial biofilm on its host. Furthermore, the model's flexibility has allowed for the testing of different clinical treatment modalities [52], the introduction of comorbidities (e.g. ischemia [53]) into the wound biofilm system, and the comparison of species-specific virulence between common wound biofilms (submitted for publication). However, with the ultimate goal of better understanding biofilms within human chronic wounds, adapting our in vivo model to consistently generate polybacterial biofilm was critical to advancing our work.  Using two bacterial species commonly found within chronic wounds, S. aureus and P. aeruginosa, we have demonstrated that polybacterial biofilm significantly impairs wound healing relative to its single-species biofilm counterparts, while simultaneously triggering a greater host inflammatory response. This dramatic impairment in wound healing seen with polybacterial biofilm is indicative of a potential synergy or enhancement that occurs between both species within a single wound. The presence of each species within the same wound may lead to molecular and cellular communication, and competition, between each species, leading to an in vivo equilibrium based on the individual properties of each bacteria and the wound bed. This is supported by the spatial colocalization, and thus potential interaction, between species seen on SEM. Furthermore, the predominance of P. aeruginosa relative to S. aureus over time, followed by a plateau in the growth of each, is also representative of this equilibrium. However, as suggested by Dalton et al [55], the end-outcome of this mixed-species milieu may be a joint effect on the dynamics of the wound-healing cascade, resulting in a much lower amount of epithelial and granulation tissue than what is achievable by either single species  alone. Similarly, the presence of both species also triggers a substantial increase in the expression of host inflammatory cytokines, indicating that polybacterial wounds are recognized as being phenotypically different by the host. Consequently, given that most chronic wounds contain multiple bacterial species, the overall increase in biofilm virulence seen in our model further underscores the difficulty of eradicating chronic wound biofilm in a clinical setting.
The use of S. aureus mutant UAMS-929 within our in vivo model has allowed us to speculate on some the mechanisms underlying our findings. In particular, it was important to determine whether our perceived increase in virulence was specifically due to collaboration between multiple, biofilm-forming, bacterial species or simply from an increased total bacterial burden within polybacterial wounds. Interestingly, the biofilm-deficient mutant that we used showed a similar growth rate to its wild-type counterpart in vitro, while equilibrating to similar level of bacteria within a polybacterial biofilm in vivo. However, despite no change in the distribution and total burden of S. aureus and P. aeruginosa, polybacterial wounds that contained mutant S. aureus demonstrated a distinct reduction in wound impairment and host inflammatory response. As previously mentioned, the sarA mutation ultimately modulates the production of polysaccharide intercellular adhesin, critical to many bacterial processes including biofilm extracellular matrix production [31]. This decrease in biofilm durability may potentially reduce the virulence of UAMS-929 relative to wild-type S. aureus, as seen through previous work by others [31,32] and within our in vivo model (unpublished data). When used within our polybacterial model, the presence of wildtype P. aeruginosa was not able to 'rescue' the virulence of a biofilm containing mutant S. aureus to the level of a biofilm containing both wild-type strains. These findings suggested that polybacterial biofilm virulence was not dependent on the number of bacteria present, but rather on the ability of each species to effectively maintain, and interact within, a complex biofilm while also expressing virulence factors in the face of host defenses. Without an intact, biofilm-forming mechanism in one of the two species present, the potential synergy between species may be significantly reduced, despite the continued presence of both bacteria within the wound.
Although our findings supplement the available literature on polybacterial biofilm, we acknowledge the limitations of our work. As this study was an initial exploration of in vivo, polybacterial biofilm, we did not address the specific cellular and molecular mechanisms responsible for the proposed synergy between species. Future work, using additional mutant strains and sophisticated molecular techniques, will further delineate the importance of different biofilm constituents to the complex interactions within a polybacterial biofilm, but are beyond the scope of these initial observations. Therefore, we believe that our work with UAMS-929 was important in developing a preliminary understanding of the mechanistic principles behind polybacterial biofilm. Our findings represent a growing foundation from which continued research can move forward. Furthermore, our adapted in vivo, polybacterial biofilm model continues to provide the distinct advantages associated with our single-species model [51], including multiple, quantitative endpoints with both flexibility in experimental study and reproducibility of results. By now modeling more clinically relevant, polybacterial biofilms, the translatability of our results is further improved, allowing for future mechanistic research and testing of pre-clinical therapeutics.