Effect of Antimicrobial and Physical Treatments on Growth of Multispecies Staphylococcal Biofilms

ABSTRACT

The prevalence and structure of Staphylococcus aureus and Staphylococcus epidermidis within multispecies biofilms were found to depend sensitively on physical environment and antibiotic dosage. Although these species commonly infect similar sites, such as orthopedic implants, little is known about their behavior in multispecies communities, particularly in response to treatment. This research establishes that S. aureus is much more prevalent than S. epidermidis when simultaneously seeded and grown under unstressed conditions (pH 7, 37°C) in both laboratory and clinical strains. In multispecies communities, S. epidermidis is capable of growing a more confluent biofilm when the addition of S. aureus is delayed 4 to 6 h during 18 h of growth. Different vancomycin dosages generate various behaviors: S. epidermidis is more prevalent at a dose of 1.0 μg/ml vancomycin, but reduced growth of both species occurs at 1.9 μg/ml vancomycin. This variability is consistent with the different MICs of S. aureus and S. epidermidis. Growth at higher temperature (45°C) results in an environment where S. aureus forms porous biofilms. This porosity allows S. epidermidis to colonize more of the surface, resulting in detectable S. epidermidis biomass. Variations in pH result in increased prevalence of S. epidermidis at low pH (pH 5 and 6), while S. aureus remains dominant at high pH (pH 8 and 9). This work establishes the structural variability of multispecies staphylococcal biofilms as they undergo physical and antimicrobial treatments. It provides a basis for understanding the structure of these communities at infection sites and how treatments disrupt their multispecies behaviors.

IMPORTANCE Staphylococcus aureus and Staphylococcus epidermidis are two species of bacteria that are commonly responsible for biofilm infections on medical devices. Biofilms are structured communities of bacteria surrounded by polysaccharides, proteins, and DNA; bacteria are more resistant to antimicrobials as part of a biofilm than as individual cells. This work investigates the structure and prevalence of these two organisms when grown together in multispecies biofilms and shows shifts in the behavior of the polymicrobial community when grown in various concentrations of vancomycin (an antibiotic commonly used to treat staphylococcal infections), in a high-temperature environment (a condition previously shown to lead to cell disruption and death), and at low and high pH (a change that has been previously shown to soften the mechanical properties of staphylococcal biofilms). These shifts in community structure demonstrate the effect such treatments may have on multispecies staphylococcal infections.

INTRODUCTION

Staphylococci are a prominent cause of acute and chronic infections. For example, 79% of orthopedic implant-associated infections are from staphylococcal species, where Staphylococcus aureus accounts for 34% of these infections and Staphylococcus epidermidis causes 32% (1). S. aureus and S. epidermidis are nonmotile Gram-positive cocci of the genus Staphylococcus. Both species are capable of forming biofilms, structured communities of cells that are encapsulated in a matrix of polysaccharides, proteins, and DNA (2). Biofilms have been shown to impact the persistence of chronic infections (2, 3). Although S. aureus is considered more virulent than S. epidermidis, treatment of biofilms formed by both species is difficult due to the resistance of their biofilm states to antimicrobials or other treatment methods (3).

Although biofilms formed from S. aureus and S. epidermidis are responsible for the majority of prosthetic joint infections (4, 5), bacteria may be introduced to the infection site at different times because they can contact the device both at initial placement via nonsterile technique or the patient's skin and postimplantation via the bloodstream. When a prosthetic joint infection occurs less than 3 months postoperatively, it is referred to as an early infection; most early infections are caused by S. aureus (4, 5). Delayed infections of prosthetic joints occur between 3 and 24 months postoperatively and are frequently caused by less virulent bacteria, such as S. epidermidis (4, 5). Late infections occur at times greater than 24 months postoperatively and are mainly caused by hematogenous seeding (where the bacteria are introduced to the infection site via the bloodstream) or reoccurrence of inadequately treated early infections (4, 5). Hematogenous seeding is especially high for patients with S. aureus bacteremia (6). The occurrence of prosthetic joint infections presents in the following way: 29% are caused by early infection, 41% are caused by delayed infection, and 30% are caused by late infection (7). The majority of prosthetic joint infections are culture positive for only one organism; however, 10 to 16% of infections are classified as polymicrobial (1, 4).

In addition to being found at the same sites of prosthetic joint infections, S. aureus and S. epidermidis both colonize the human nares (nostrils) (8, 9). S. aureus persistently or intermittently resides in the nasal cavity of 50% of the population (9), and its presence in the nasal cavity has been linked to S. aureus bacteremia (10). Compared to nasal microbiotas in a healthy population, inpatient nasal microbiotas are enriched in S. aureus and S. epidermidis (9). Within the inpatient population, the abundance of S. aureus was negatively correlated with the abundance of S. epidermidis (9). One study indicated that the excretion of the serine protease Esp from a subset of S. epidermidis species inhibits biofilm formation and nasal colonization of S. aureus (11).

Polymicrobial biofilm communities are known as multispecies biofilms. Most studies of biofilms involve a single species; however, in many natural environments, biofilms grow in structured multispecies communities (12, 13). The presence of multiple species is able to impact the development and shape of the community (14). The interactions between species within a multispecies biofilm can be mutualistic, commensal, competitive, amensal, or parasitic (15), and the spatial heterogeneity that results from the presence of multiple species can impact the ability of cells to communicate, sense, signal, and interact with one another (14, 16, 17).

Multispecies biofilms are typically organized in one of three ways (14). They can be organized into single-species microcolonies, where each species is independently clustered (18). The two species can coaggregate and form a biofilm that has cells of each species located in proximity to the other. Finally, layering can occur, where one species is found in the upper layers and another is found in the lower layers (19, 20). Structures within multispecies biofilms have been characterized by measures such as biovolume or thickness (21). Specific features that have been quantified within multispecies biofilms are the biomass volume of each bacterial population (20, 22, 23), the distribution of distances between the surfaces of microcolonies of different species (23), and the distribution of biomass at various depths within the sample (20).

Biofilms can be controlled through a variety of treatment methods. Clinically, S. aureus and S. epidermidis biofilm infections are commonly treated using the antibiotic vancomycin (3). Sublethal concentrations of vancomycin have also been shown to create S. epidermidis biofilms with open porous structures (24). High temperatures (45°C) have been reported as a potential treatment method for S. aureus and S. epidermidis infections (25), and temperature has been shown to have an effect on the mechanical properties of the biofilm (26). Temperatures as high as 45°C have safely been used in hyperthermia treatment of cancer cells (27, 28); thus, high-temperature therapy may be a viable method for controlling bacterial infections. Increased pH (>7) softens the mechanical properties of S. aureus and S. epidermidis biofilms when the pH change is applied after the biofilms have formed (29). Low pH may be an opportune growth environment for S. epidermidis since it typically resides on the skin, where the pH is 4.0 to 7.0 (30).

In this work, we evaluate the spatial structure of S. aureus and S. epidermidis in multispecies biofilms grown under a variety of physical and antibiotic treatment conditions. Due to the prevalence of both species in prosthetic joint infections and the nares, there are grounds to consider the degree to which each of these organisms impacts the growth and structure of the other when they form a multispecies community; toward this end, herein, we present a study of the growth and structural behaviors of multispecies communities of S. aureus and S. epidermidis. We pose the following questions through our work. (i) How are the structure and cellular growth kinetics of single-species biofilms of S. aureus and S. epidermidis impacted by the introduction of an additional species? (ii) How do physical (i.e., temperature and pH) and sublethal antibiotic (vancomycin) treatments affect the organization and species abundance of an S. aureus and S. epidermidis multispecies biofilm? (iii) Do multispecies biofilms of clinical strains of S. aureus and S. epidermidis behave in a manner similar to that of laboratory strains? Answering these questions to understand the general growth behavior of these multispecies communities with and without treatment creates a basis for understanding the effects, intended or not, that treatment methods may have on an infection site if more than one species of bacteria is present.

RESULTS

Eighteen-hour single-species and multispecies biofilm growth.

To provide a baseline for studying multispecies biofilms, we considered the growth of S. aureus and S. epidermidis 18-h single-species and multispecies biofilms. After 18 h, S. aureus and S. epidermidis single-species and multispecies biofilms were structurally similar, with densely packed space-spanning structures akin to those previously observed in unstressed single-species biofilms grown in flow cells (24) (Fig. 1A to C). The overall biomasses under each growth condition were comparable, where S. aureus had a biomass of 6.5 ± 0.5 μm³/μm², S. epidermidis had a biomass of 4.8 ± 1.0 μm³/μm², and the multispecies biofilm had a biomass of 5.1 ± 0.5 μm³/μm² (Fig. 1D). However, the 18-h multispecies biofilms consisted mainly of S. aureus cells (Fig. 1C and D), where the S. aureus biomass was 5.10 ± 0.52 μm³/μm² and the S. epidermidis biomass was minimally detectable (0.03 ± 0.02 μm³/μm²) (Fig. 1D).

FIG 1.

Eighteen-hour single-species and multispecies biofilm growth of S. aureus and S. epidermidis. Projection images of the average intensities from three-dimensional (3D) CLSM image volumes of an 18-h S. aureus biofilm (A), an 18-h S. epidermidis biofilm (B), and an 18-h multispecies biofilm (C). Scale bars = 20 μm. (D) Plot of average biomass in 18-h S. aureus, S. epidermidis, and multispecies biofilms.

Single-species and multispecies growth rates.

We next considered the rates of biofilm growth within the single-species and multispecies biofilms. Because the planktonic growth of S. aureus is faster than that of S. epidermidis (Fig. S2), we hypothesized that S. aureus biofilm development would be faster than S. epidermidis biofilm development. We found that S. aureus required ∼4 h to colonize the majority of the glass substrate surface (Fig. 2), while S. epidermidis needed 6 to 12 h to form a contiguous biofilm (Fig. 2). This result validates our hypothesis and aligns with previous single-species research showing that S. aureus colonizes surfaces more rapidly than S. epidermidis on three different substrate surfaces [a titanium alloy, poly(methyl methacrylate), and ultrahigh-molecular-weight polyethylene] (31).

FIG 2.

S. aureus, S. epidermidis, and multispecies biofilm growth from 1 to 18 h. Projection images of average intensities from confocal image volumes of representative S. aureus biofilms (left column), S. epidermidis biofilms (center column), and multispecies biofilms (right column) at the following time points: 1 h (A), 2 h (B), 4 h (C), 6 h (D), and 12 h (E). Scale bars = 20 μm. (F) Biomass of S. aureus, S. epidermidis, and multispecies biofilms after 1, 2, 4, 6, 12, and 18 h of growth after being inoculated simultaneously.

The kinetic growth rate or biomass accumulation of the multispecies biofilm resembled the initial stages of biofilm formation in S. aureus and S. epidermidis single-species biofilms until ∼4 h of growth. After this period, S. aureus dominated the growth and S. epidermidis growth slowed (Fig. 2). Thus, in the unstressed biofilm growth environment (tryptic soy broth with glucose [TSB_G], 37°C, pH 7), S. aureus is the dominant species, with a cellular microstructure that is densely packed, spanning the image volume, and S. epidermidis is present as small clusters sequestered within the total biofilm volume. The overall biomass is space filling and densely packed; large pores are absent within the volume.

Delayed addition of second species to a single-species biofilm.

Because S. aureus has a planktonic growth rate that is higher than that of S. epidermidis, it was not surprising that S. aureus was the dominant species within biofilms where both S. epidermidis and S. aureus were simultaneously inoculated. S. epidermidis took 6 to 12 h to achieve a contiguous single-species biofilm (Fig. 2). Thus, we hypothesized that S. epidermidis would require 6 to 12 h of lead time to successfully colonize the substrate of the biofilm and thereby allow it to be the dominant species within the multispecies biofilm. We also tested the reverse situation in which S. aureus is inoculated before the addition of S. epidermidis.

Figure 3 shows the delayed addition of S. aureus to S. epidermidis biofilms after giving S. epidermidis an inoculation lead time of 1, 2, 4, and 6 h (Fig. 3A to H) and the delayed addition of S. epidermidis to S. aureus biofilms after S. aureus lead times of 1, 2, 4, and 6 h (Fig. 3I to P). We found that S. epidermidis is able to thrive within the biofilm if given a lead time of 2 to 6 h (Fig. 3C/G, D/H, and Q), while the S. epidermidis population is suppressed both with and without a lead time for S. aureus (Fig. 3I through P and R).

FIG 3.

Delayed addition of one species in the formation of a S. aureus and S. epidermidis multispecies biofilm. The top row shows multispecies biofilms where S. epidermidis has been given a lead time of 1 h (A), 2 h (B), 4 h (C), and 6 h (D) before S. aureus is introduced. (E to H) S. epidermidis growth (the GFP channel only) at these respective times. The third row shows multispecies biofilms where S. aureus has been given a lead time of 1 h (I), 2 h (J), 4 h (K), and 6 h (L) before S. epidermidis is introduced. (M to P) S. epidermidis growth (the GFP channel only) at these respective times. The total growth is 18 h for all projection images of average intensities from 3D CLSM image volumes. Scale bars = 20 μm. (Q and R) Biomass bar graphs of multispecies biofilms after the delayed addition of S. aureus to S. epidermidis at 0, 1, 2, 4, and 6 h of growth (Q) and S. epidermidis to S. aureus at 0, 1, 2, 4, and 6 h of growth (R). Total growth time was 18 h.

Effect of sublethal vancomycin on multispecies biofilms.

S. aureus and S. epidermidis infections are commonly treated with vancomycin (3). Vancomycin is an antibiotic that inhibits the formation of the bacterial cell wall and interferes with peptidoglycan synthesis in Gram-positive bacteria (5). The MIC, defined as the lowest concentration of antibiotic required to inhibit visible bacterial growth after overnight culture, is different for S. aureus than for S. epidermidis. The MIC of vancomycin for S. aureus is typically ∼1 μg/ml (32, 33) and is typically ∼2 μg/ml for S. epidermidis (34, 35). The literature values of 1 and 2 μg/ml vancomycin were confirmed for the S. aureus and S. epidermidis strains, respectively, used in this experiment by broth microdilution. We hypothesized that S. epidermidis would be more fit at low dosages of vancomycin than S. aureus. We therefore investigated two sublethal vancomycin concentrations: 1 μg/ml vancomycin, which is near the MIC of S. aureus, and 1.9 μg/ml, which is just under the MIC of S. epidermidis. Figure 4 shows the behaviors of multispecies biofilms at each of these vancomycin concentrations.

FIG 4.

Effect of sublethal vancomycin on multispecies biofilms. (A) Projection image of average intensities from a CLSM volume of a multispecies biofilm grown in 1.0 μg/ml vancomycin (volume height, 8.5 μm). Scale bar = 20 μm. (B) Projection image of average intensities of the S. epidermidis (GFP channel only) in panel A. White arrows indicate regions with less S. epidermidis growth. (C) Image of potential eDNA within multispecies biofilm grown in 1.0 μg/ml vancomycin. The bright-pink regions that are much larger than regular cell size are likely eDNA; examples of these bright-pink regions are denoted with yellow arrows. Example regions of predominantly S. aureus growth are denoted with red arrows. Scale bar = 5 μm. (D) Additional evidence of eDNA is shown by the blue regions of the image, where eDNA has been stained with TOTO-3 (blue), total biomass is stained with Syto40 for eDNA verification experiments (pink), and S. epidermidis fluoresces due to the presence of the GFP plasmid. Scale bar = 5 μm. Projection images of average intensities from a CLSM image volume of a multispecies biofilm grown in 1.9 μg/ml vancomycin, where S. epidermidis has outgrown S. aureus (volume height, 6.0 μm) (E), S. aureus has outgrown S. epidermidis (volume height, 8.5 μm) (F), and growth is sparse in both species (volume height, 6.3 μm) (G). Scale bars = 20 μm. (H) Plot of changes in biomass with vancomycin (vanc.) concentrations of 0, 1.0, and 1.9 μg/ml in single-species and multispecies biofilms.

When the concentration of vancomycin was 1.0 μg/ml, S. epidermidis was the dominant species (Fig. 4A and H). S. epidermidis was less contiguous under this condition (white arrows in Fig. 4B indicate voids in S. epidermidis growth), which allowed for the biofilm to have regions with predominantly S. epidermidis growth and regions with predominantly S. aureus growth within the volume (Fig. 4A/B). The less contiguous structure of S. epidermidis in multispecies biofilms grown in 1.0 μg/ml vancomycin displays much differently from the more uniform distributions of S. epidermidis in S. epidermidis-rich biofilms formed by the delayed addition of S. aureus to S. epidermidis (compare Fig. 4B with 3G and H). The regions of growth containing predominantly S. aureus (example regions highlighted by red arrows in Fig. 4C) appear to have extracellular DNA (eDNA) present, as indicated by the observation that Syto59 was not solely localized in cellular interiors, as shown by the large bright-pink regions within the image that are larger and less spherical than a typical staphylococcal cell (example regions highlighted by yellow arrows in Fig. 4C). To confirm the presence of eDNA, we stained these biofilms with TOTO-3, a stain that binds to eDNA and dead cells. When confirming the presence of eDNA, Syto40 was used instead of Syto59 due to the overlap in the emission spectra of Syto59 and TOTO-3. We found evidence of eDNA (blue regions of image) in regions abundant with S. aureus (Fig. 4D). The total biomass at 1.0 μg/ml vancomycin was similar to that of multispecies biofilms without vancomycin, but in this case, S. epidermidis is the dominant species instead of S. aureus (Fig. 4A and H). This observation is consistent with our hypothesis, since the vancomycin MIC for S. aureus is ∼1 μg/ml and the vancomycin MIC for S. epidermidis is ∼2 μg/ml.

When the concentration of vancomycin was 1.9 μg/ml, a concentration just below the vancomycin MIC of S. epidermidis, the overall biomass of all organisms was reduced (Fig. 4H). However, the growth was heterogeneous at this concentration of vancomycin. In some cases, there was more S. epidermidis than S. aureus (Fig. 4E), and in others, there was more S. aureus than S. epidermidis (Fig. 4F). At 1.9 μg/ml vancomycin, the amount of growth varied, with some volumes containing extremely sparse growth (Fig. 4G) and some with larger clusters of cells (Fig. 4E and F). This heterogeneity is reflected in the large relative variability in biomass determinations reported in Fig. 4H.

Effect of temperature on multispecies growth.

Antibiotics are less effective on bacteria within their biofilm state. The reduced effectiveness of antibiotics on biofilms motivates the development of alternative preventative and therapeutic solutions (6). Recent work has shown that increased temperatures may disrupt mature S. aureus and S. epidermidis biofilms and their mechanical properties (25, 36). Given this literature for single-species biofilms, we hypothesized that temperature would have a comparable disruptive effect on multispecies biofilms.

We found that when multispecies biofilms consisting of S. aureus and S. epidermidis were grown at 45°C, there was a significant decrease in the overall biomass from 5.13 ± 0.53 μm³/μm² at 37°C to 2.43 ± 1.97 μm³/μm² at 45°C. A significant decrease in biofilm growth at 45°C also occurred in single-species S. aureus and S. epidermidis biofilms. Beyond the changes in biomass, there were structural changes in the multispecies biofilms grown at 45°C, where the multispecies staphylococcal biofilms grown at a higher temperature formed structures that were more porous than those grown at 37°C (compare Fig. 5A and B). Specifically, the multispecies biofilm porosity, as defined in Materials and Methods, increased nearly 2-fold from 0.4 to 0.7 (Fig. 5C). This increase in porosity was accompanied by a decrease in the total biomass of the biofilm (Fig. 5D). In particular, S. aureus formed a less contiguous structure at 45°C in comparison to the multispecies biofilms formed at 37°C (Fig. 5A and B). The porous nature of the S. aureus biofilm at the substrate surface appeared to allow an increase in S. epidermidis cells at the substrate. This increase in S. epidermidis at the substrate surface agrees with the observed increase in S. epidermidis biomass (Fig. 5D) and shows that at 45°C, S. epidermidis is most prevalent in the layer of biofilm closest to the substrate. Additionally, the porosity of the 45°C multispecies biofilm was most similar to that of the S. epidermidis biofilm, which also has a 2-fold increase in porosity at 45°C (Fig. 5C). Figure 5 shows that hyperthermia treatments, similar to those used to treat cancer (37), could contribute to the control of staphylococcal biofilm infections by decreasing the biomass and increasing the porosity of the biofilm.

FIG 5.

Effect of increased temperature on multispecies biofilms. Projection images of average pixel intensities from CLSM volumes of multispecies biofilms grown for 18 h at 37°C (A) and 45°C (B). Scale bars = 20 μm. Contrast has been increased by 1% to more clearly depict the variations in porosity. (C) Porosity of single-species and multispecies biofilms at 37°C and 45°C. (D) Plot of changes in biomass with temperature in single-species and multispecies biofilms.

Multispecies growth under various pH conditions.

Another potential approach for the physical disruption of biofilms is treatment by pH adjustment. In this context, we varied pH by initially inoculating biofilm-forming bacteria in growth media with nonneutral pH (5, 6, 8, and 9). Both S. aureus and S. epidermidis biofilms have been softened by increasing the pH of the biofilm (29). Adjusting the biofilm pH to greater than 7 was suggested as a potential method to disrupt staphylococcal biofilms by softening their mechanical properties (29). S. epidermidis survives under the low pH conditions of the skin, which ranges from pH 4.0 to 7.0 (30). Thus, we hypothesized that at low pH, S. epidermidis would be capable of being the most prevalent species (due to its adaptability to the low pH of the skin) and that at high pH, the biofilm would be more sparsely populated (due to the weakening of the staphylococcal biofilm mechanical properties at high pH [29]). S. aureus may be less densely populated at lower pH, since planktonic S. aureus growth has been shown to decrease as pH drops from 7 to 4.5 (38).

We found that in multispecies biofilms with S. aureus, S. epidermidis grows much more under low pH conditions (pH 5 and 6) than it does at neutral pH. At pH 5, the biofilm had a prominent and relatively evenly distributed S. epidermidis population (Fig. 6A) and a reduced biomass compared to biofilms cultured at pH 7 (Fig. 6C and F). At pH 6, there was more S. epidermidis than in the control at pH 7; however, there were still large regions containing only S. aureus (Fig. 6B and F). The dominant behavior of S. epidermidis at low pH can be understood as a consequence of the S. epidermidis fitness for low pH environments, which is the usual state of the skin.

FIG 6.

Effect of pH on multispecies biofilms. S. aureus and S. epidermidis multispecies biofilms grown at pH 5 (A), pH 6 (B), pH 7 (C), pH 8 (D), and pH 9 (E). Images are projections of average pixel intensities from all images in a CLSM image volume. Scale bars = 20 μm. (F) Plot of changes in biomass with pH in single-species and multispecies biofilms.

When multispecies biofilms are grown at higher pH values, we find that S. epidermidis does not incorporate into the biofilm very well and remains predominantly as individual cells or clusters of cells, and S. aureus is the dominant organism. Specifically, at pH 8, the multispecies biofilm structure is similar to that of biofilms grown at pH 7, where there are large regions of S. aureus with small clusters of S. epidermidis throughout the biofilm (Fig. 6D and F). At pH 9, S. aureus remains the dominant species in the biofilm (Fig. 6E and F).

Growth behaviors of clinical isolates of S. aureus and S. epidermidis.

Up to this point, all experiments were performed with common laboratory strains. To establish the generalizability of our work to clinical infections, we investigated multispecies staphylococcal biofilm growth behavior in three clinical isolates of S. aureus and three clinical isolates of S. epidermidis. We found that S. aureus consistently outgrows S. epidermidis in all six clinical isolates (Fig. 7A and B). However, all clinical S. epidermidis strains had a higher abundance than S. epidermidis 1457 (Fig. 7A). Although the relative abundance of each strain may vary in clinical settings, the trend of the behaviors identified in the laboratory strains remains the same when clinical strains are substituted.

FIG 7.

Behavior of S. aureus and S. epidermidis (S. epi.) clinical isolates in multispecies biofilms. (A) Biomass plots of S. aureus SH1000/pCM29 and three clinical strains of S. epidermidis (P18, P37, and P47). (B) Biomass plots of three clinical strains of S. aureus (P1B, P4, and P6) grown with S. epidermidis 1457/pCM29 GFP. S. aureus outgrows S. epidermidis under unstressed growth conditions in all clinical strains studied.

DISCUSSION

This work shows that the behavior of multispecies biofilms of S. aureus and S. epidermidis varies greatly from one environmental growth condition to another. Under unstressed growth conditions (37°C, pH 7), S. aureus is the dominant species in both clinical and common laboratory strains. S. aureus is also more prevalent than S. epidermidis under conditions of high pH (8 and 9) and high temperature (45°C). S. epidermidis was the dominant species when given a lead time of 4 to 6 h, as well as under environmental conditions with low pH (5) and low concentrations of vancomycin (1.0 μg/ml). At a vancomycin concentration near the MIC of S. epidermidis (1.9 μg/ml), the overall biomass decreased; however, the species dominance varied and produced behaviors where sometimes S. aureus grew more than S. epidermidis, and vice versa. These varied behaviors were correlated with the different vancomycin MIC values for these organisms. Understanding the general behavior of multispecies communities with and without treatment can potentially help identify unintended effects that treatment methods have on an infection site with more than one species present.

Because S. aureus and S. epidermidis single-species and multispecies biofilms may form on a variety of substrate surfaces, it would also be of interest to determine if substrate surfaces impact the structures of the multispecies biofilms that form from S. aureus and S. epidermidis. The present work is limited to the case of smooth surfaces of untreated glass. Additionally, other species of bacteria are capable of infecting these sites and could be added to these multispecies biofilms to consider how a third bacterial species impacts the structures formed by these species.

Beyond adding virulent pathogens to these communities, nonvirulent pathogens could be studied with single-species or multispecies staphylococcal biofilms to aid in the development of probiotic treatments for preventing biofilm infections. Probiotic treatment to eliminate the growth of virulent pathogens has gained traction as a strategy for eliminating biofilm infections, especially since probiotic treatments have successfully been used to treat altered bowel flora (39). Our work shows that the seed time and growth environment of the organism used to displace the pathogen are critical to the displacement of the organism. For example, a probiotic organism that grows at a low pH may displace S. aureus biofilm growth; however, this will not likely be as effective against preventing S. epidermidis growth. S. aureus and S. epidermidis can coexist within biofilms formed by both laboratory and clinical isolates, and this work has established a basis for further understanding the interactions between these two species under additional treatment conditions.

MATERIALS AND METHODS

Bacterial strains.

S. epidermidis 1457/pCM29 (AH2982) was used as the model S. epidermidis strain in our experiments (kindly provided by A. Horswill, University of Iowa) (40). S. epidermidis 1457 transformed with pCM29 (40), a green fluorescent protein (GFP) reporter with a sarA P1 promoter, which is maintained in tryptic soy broth (TSB) or tryptic soy agar (TSA) supplemented with 10 μg/ml chloramphenicol, was chosen due to its ability to constitutively express GFP. S. epidermidis 1457 is a commonly used biofilm-forming strain of S. epidermidis. Colonies of S. epidermidis were cultured on TSA with 10 μg/ml chloramphenicol. We chose a strain of S. epidermidis with a fluorescent reporter in order to distinguish S. epidermidis cells from S. aureus cells within the multispecies biofilms that were grown. Clinical isolates of S. epidermidis (P18, P37, and P47) obtained by Sharma et al. (41) were used as representative clinical S. epidermidis samples.

S. aureus SH1000 (BB 386) (42), a commonly used model strain of S. aureus (43), was used as the model S. aureus strain in our study. Colonies of S. aureus were cultured on TSA. S. aureus SH1000/pCM29 was used as a control to confirm that the interactions between S. aureus SH1000 and S. epidermidis 1457/pCM29 were not changed by the presence of pCM29, a GFP-producing plasmid. Isolates of S. aureus (P1B, P4, and P6) previously obtained from positive blood cultures from the University of Michigan Hospital Clinical Microbiology Laboratory and cryopreserved as glycerol stocks at −80°C (25) were used as representative S. aureus clinical isolates. All clinical isolates of S. aureus used in this study were methicillin resistant by standard clinical methods.

The vancomycin MIC for each strain used in this study was determined using the broth microdilution method, as per the Clinical and Laboratory Standards Institute (CLSI) guidelines (50). Briefly, 5 × 10⁵ cells/ml were inoculated into increasing concentrations of vancomycin in a 96-well plate. The plate was then incubated at 37°C overnight, and the MIC was recorded as the lowest concentration that resulted in no visible growth.

Single-species and multispecies biofilm growth.

Biofilms were grown in Nunc Lab-Tek II chambered coverglass dishes (Thermo Scientific, USA) for 18 h at 60 rpm and 37°C, unless otherwise noted. We grew single-species biofilms by inoculating a single colony of either S. aureus SH1000 or S. epidermidis 1457/pCM29 in 400 μl of tryptic soy broth with 1% glucose (TSB_G). For multispecies biofilms, we inoculated a single colony of S. aureus SH1000 and a single colony of S. epidermidis 1457/pCM29 in 400 μl of TSB_G. To approximate the initial seed concentration in each well, a hemacytometer was used to determine the average concentration of cells when one colony was suspended in TSB_G. Hemacytometer counts were performed for 10 different colonies of S. aureus in 400 μl of TSB_G and 10 different colonies of S. epidermidis in 400 μl of TSB_G. The average and standard error of the mean (SEM) are reported for the initial concentration of cells. The initial concentration of cells using this single-colony-inoculum method was found to be 2E7 ± 2E6 cells/ml, with no significant differences between S. aureus and S. epidermidis.

We confirmed the stability of pCM29 in TSB_G without chloramphenicol through 18-h optical density at 600 nm (OD₆₀₀) and fluorescence growth curves of S. epidermidis 1457/pCM29 and S. aureus SH1000/pCM29 with and without chloramphenicol. We found that the presence or absence of chloramphenicol did not change the stability of pCM29 within 18 h (Fig. S1). Additionally, we verified that the observed growth behaviors were not altered by the presence of pCM29 within S. epidermidis 1457 by conducting 18-h biofilm experiments with S. aureus SH1000/pCM29 and S. epidermidis 1457 wild types. We found that the constitutive expression of GFP did not alter the growth behavior of the multispecies communities, and S. aureus outgrew S. epidermidis irrespective of the presence of pCM29 in S. aureus versus S. epidermidis (Fig. S1).

To observe biofilm growth behaviors, bacteria were stained with 5 μM Syto59 (Molecular Probes, USA) for 30 min prior to imaging. Syto59 was chosen for staining bacteria because the excitation and emission are 622 nm and 645 nm, respectively. These are much higher than the excitation and emission of GFP (489 nm/508 nm [44]). These differences in excitation and emission spectra enable us to distinguish S. epidermidis cells from S. aureus cells in the multispecies biofilms. S. epidermidis is identified by the GFP, and S. aureus is located by identifying regions where Syto59 is present but GFP is not. Apart from differences in signal intensities due to differences in the localization of the fluorophores, the GFP within S. epidermidis 1457/pCM29 overlays well with Syto59 in single-species biofilms (Fig. S1A), so subtracting the GFP channel from the Syto59 channel allows for the identification of S. aureus.

Biofilm growth kinetics.

Biofilm kinetic experiments measure changes in the biomass accumulation with time. These experiments were performed by growing single (S. aureus alone, S. epidermidis alone), and multispecies (S. aureus with S. epidermidis) biofilms for 1, 2, 4, 6, 12, and 18 h in 400 μl of TSB_G and measuring the biomass at each time point. Three replicates were performed at each time point for each single-species and multispecies biofilm.

Delayed addition of second species to single-species biofilm.

Multispecies biofilms were grown in which either S. aureus or S. epidermidis was given a lead time of 1, 2, 4, or 6 h before the second species of bacteria, S. epidermidis or S. aureus, was introduced through inoculation of a single colony in the supernatant of the biofilm well. These multispecies biofilms were grown for a total of 18 h at 37°C in 400 μl of TSB_G. Each experimental condition was performed in triplicate.

Sublethal antibiotic, temperature, and pH biofilm growth conditions.

We considered single-species and multispecies growth under sublethal-vancomycin growth conditions. We grew biofilms in both 1.0 and 1.9 μg/ml vancomycin to determine the effect of various concentrations of sublethal vancomycin. Vancomycin is an antibiotic commonly used to treat both S. aureus and S. epidermidis infections (3); the MIC of vancomycin is ∼1 μg/ml for S. aureus (32, 33) and is ∼2 μg/ml for S. epidermidis (34, 35).

Single-species (S. aureus alone, S. epidermidis alone), and multispecies (S. aureus and S. epidermidis) biofilms were grown in triplicate for 18 h in TSB_G at 45°C and compared to biofilms grown at 37°C.

We grew single-species (S. aureus alone, S. epidermidis alone) and multispecies (S. aureus and S. epidermidis) biofilms for 18 h at 37°C in TSB_G with a pH adjusted to 5, 6, 8, or 9. pH was adjusted using 1 M HCl to lower the pH and 1 M NaOH to increase the pH. We probed this range of pH values because S. epidermidis biofilms have been found to grow at pH values ranging from 4.5 to 7.5 (45–47), and planktonic S. aureus growth has been shown to decrease as pH drops from 7 to 4.5 (38). Additionally, staphylococcal biofilms treated with higher pH media have been shown to soften their mechanical properties (29). Three replicates were performed under each pH growth condition.

Confocal laser scanning microscopy imaging and analysis.

We imaged samples using a Nikon A1Rsi confocal laser scanning microscope with a 100× 1.45 numerical aperture (NA) oil immersion objective lens. The excitation wavelength was 488 nm for the GFP within the S. epidermidis samples and 622 nm for the Syto59. The GFP channel was used for identifying S. epidermidis, while the Syto59 was used for identifying the total biomass, since it stains both the S. aureus and the S. epidermidis cells. Three-dimensional image volumes of size 128 by 128 by ∼10 μm³ with voxels of 250 by 250 by 250 nm³ were collected for each biofilm grown and used to assess biofilm growth. Volumes that were much less than 10 μm were obtained only when growth was sparse, i.e., at short times (<4 h) or under high-stress growth conditions (1.9 μg/ml vancomycin).

To confirm the presence of extracellular DNA (eDNA), biofilms were stained with 1 μM TOTO-3 (Molecular Probes, USA), a stain used for detecting eDNA and dead cells (48). Because TOTO-3 has an excitation of 642 nm and emission of 660 nm, S. aureus was stained using 12.5 μM Syto40 (excitation [ex.] 420 nm/emission [em.] 441 nm) (Molecular Probes) instead of Syto59 (ex. 622 nm/em. 645 nm) for experiments confirming the presence of eDNA.

Biomass within image volumes was quantified using the computer program COMSTAT (a program developed in Matlab to quantify biofilm structures [21]). The biomass is estimated using the biovolume function within COMSTAT. The biovolume function is defined as the number of biomass pixels in all images of a confocal laser scanning microscopy (CLSM) image volume multiplied by the voxel size and divided by the substratum area of the image volume (21). The estimated biomass has units of cubic micrometers/square micrometers. The biomass was computed for the first 8.5 μm (34 slices, starting at the coverslip) of each sample to standardize across all image volumes of the study. Actual biofilm heights were often thicker than 8.5 μm; however, analysis was not performed for growth above this limit, because the image quality progressively deteriorated above this height due to residual scattering from the biofilm constituents. The average biomass from three samples and the standard error of the mean are reported. Biomass was determined for the Syto59 channel and the GFP channel of each image. The total biomass is reported using the Syto59 channel. S. epidermidis biomass is reported as the GFP channel. S. aureus biomass is approximated by subtracting the GFP channel from the Syto59 channel.

Images representative of each condition are reported in the figures. For projections of representative CLSM biofilm volumes, the Z project tool is used in ImageJ to create a composite image, where the average of the pixels from the first 8.5 μm of the image volume is displayed to enable visual comparisons of the samples.

Porosity, calculated as the ratio of the void space to the bulk space of a porous medium (49), was determined for the temperature treatment studies using the percentage of bacterial coverage at each layer outputted by COMSTAT. Briefly, the percentage of bacterial coverage at each two-dimensional (2D) image was subtracted from the total area to determine the percentage of void coverage at each layer (49). This percentage of void coverage was then averaged across 34 2D images to determine the average porosity for each sample. Three sample porosities were computed for each condition, one for each image volume collected.