Characterizing interactions of Staphylococcus aureus and Escherichia coli in dual-species implant-associated biofilms

Abstract

While Staphylococcus aureus is the predominant pathogen in periprosthetic joint infections (PJI), polymicrobial infections involving Gram-negative organisms, such as Escherichia coli, complicate clinical outcomes. Little is known regarding implant-associated polymicrobial interactions; consequently, current PJI treatments are not optimized for their treatment. This study explored the dynamics of S. aureus-E. coli dual-species biofilms, focusing on biofilm properties, antibiotic susceptibility, and molecular interactions. Co-culture experiments revealed that E. coli significantly suppressed S. aureus biofilm viability, observed for methicillin-susceptible S. aureus (MSSA) and methicillin-resistant S. aureus (MRSA). Microscopic analyses demonstrated enhanced E. coli attachment facilitated by S. aureus matrix proteins; however, over time, E. coli dominated the biofilm composition. In the presence of E. coli, MSSA biofilm exhibited improved gentamicin susceptibility while MRSA showed limited change, underscoring strain-specific interactions. Notably, E. coli biofilms exhibited enhanced resistance to gentamicin in dual-species settings. Gene expression profiling revealed molecular adaptation in S. aureus and E. coli, triggered by the differential regulation of stress, adhesion, virulence, and biofilm-associated genes within a dual-species implant-associated biofilm. The suppression of S. aureus by E. coli presents potential therapeutic avenues, and in vivo studies and mechanistic investigations are crucial for optimizing treatment strategies targeting polymicrobial PJIs.

Introduction

Total joint arthroplasties (TJA), prescribed for end-stage arthritis, offer an improved quality of life for >1 million patients annually, who undergo this procedure¹. Periprosthetic joint infection (PJI), which affects approximately 2% of patients, is the most severe complication associated with TJA, greatly impacting patients’ well-being due to its high morbidity and mortality rates². Revision surgery to remove infected tissues and replace infected prosthesis components, followed by local and systemic antibiotic administration, remains the recommended treatment strategy to manage PJI^3,4. However, infection remains one of the primary causes of revision failure rates, resulting in poor outcomes⁵.

Approximately 6-40% of PJI are diagnosed as polymicrobial, with more than one causative organism^6,7. In PJI, polymicrobial infections are frequent in early postoperative PJI ( ~ 30%) and patients with polymicrobial PJI are at higher risk for complications, with >70% of cases resulting in unsuccessful outcomes^8,9. In clinical settings, two or more infecting microorganisms have been frequently isolated, harboring varying antibiotic resistance profiles, which pose a critical treatment challenge^10,11. The cohabitation within a polymicrobial biofilm further increases the probability of antibiotic resistance gene transfer, resulting in a high-risk multidrug-resistant infection^12–14. PJI is diagnosed definitively only by using the culture of tissue debrided in open surgery¹⁵. While this is the standard, negative culture incidence is approximately 20%, highlighting the limitations of ex vivo culture due to prior antibiotic usage and improper diagnostic workup^15,16. For polymicrobial infections, competitive inhibition exhibited by the involved microorganisms is a crucial determinant for accurate identification, which results in further underestimation of their incidence^17,18. The high prevalence of polymicrobial PJI warrants the study of multispecies infection dynamics and interactions.

S. aureus and other coagulase-negative staphylococci (CoNS) are the predominant organisms that account for >60–70% of PJI¹⁹. Acute infections, that occur postoperatively <3 months are majorly caused by S. aureus (20–40%) spreading from surgical site contamination²⁰. The high incidence could be attributed to their capability to thrive in hypoxic environments such as that of joint space^21,22. Staphylococci can exist as aggregates in vivo²³, and joint synovial fluid further promotes the formation of a proteinaceous matrix surrounding them^24,25. These microbial aggregates utilize adhesins (MSCRAMMs, elastin, and fibronectin-binding proteins) and polysaccharide matrix (polysaccharide intercellular adhesins) to attach to implant materials and surrounding tissue, embedding microbial communities extensively over time, thereby establishing a biofilm^26,27. S. aureus biofilms produce specific virulence factors such as leukocidins and other small molecule inhibitors to interfere with host responses and modulate their responses to evade detection and promote survival^28,29. In PJI, the avascular surfaces of implantable materials make the host more prone to infections³⁰, but their role in the evolution of infectious communities is not well understood. To date, S. aureus has been reported in three distinct biofilm reservoirs within the implant, joint, and bone environments³¹. The complex biofilm dynamics in the presence of implant materials further impart physical and chemical barriers against antibiotic treatments (antibiotic tolerance) rendering them incapable of eradicating PJI. Besides, there is a compounded risk associated with the increasing prevalence of methicillin and vancomycin-resistant S. aureus (MRSA/VRSA) in PJI^32,33. Consequently, to combat antibiotic resistance and biofilm-associated antibiotic tolerance, a cocktail of antibiotics at high local concentrations (>100× MIC) are often necessary in efforts to tackle PJI^3,34.

Current treatment strategies for suspected polymicrobial PJI involve systemic and local dosing of broad-spectrum beta-lactam antibiotics and aminoglycosides, which are not optimized to eradicate polymicrobial biofilms^3,35,36. Gentamicin, specifically, remains integral to current local approaches such as bone cement and spacers in the clinic, bolstering systemic antibiotic strategies to prevent the recurrence of PJI³⁷. To improve the translational value of drug-eluting implant materials for polymicrobial PJI prophylaxis, an in-depth understanding of implant-associated polymicrobial dynamics is warranted to effectively prevent PJI while curbing the poor outcomes associated with multidrug resistance.

S. aureus is also the most isolated organism in polymicrobial PJI (~56%)³⁸, highly capable of co-forming multi-species biofilms. The presence of more than one organism creates a unique competitive microenvironment that regulates the growth, antibiotic susceptibility, and physiology of the bacteria³⁹. Among microorganisms, Gram-negative bacteria commonly harbor multi-drug resistance, and their combination with MRSA/VRSA in a polymicrobial setting tremendously increases the risk and severity of PJI^13,40. Studies have shown S. aureus frequently isolated with Gram-negative E. coli and Pseudomonas aeruginosa, wherein they often have a negative or synergistic impact on the growth and antibiotic susceptibility of S. aureus^41–44. Recent reports have identified a significant impact of polymicrobial interactions on microbial viability, colonization, and immune evasion properties of S. aureus in catheter-associated infections^45,46. These reports strongly suggest that polymicrobial interactions regulate physiological processes, promote genetic exchange conferring antibiotic resistance, and form complex biofilm matrices. These should be specifically studied in the context of PJI.

The international consensus meeting on musculoskeletal infection^47–50 has identified polymicrobial infections as a vital understudied area. Our study aims to address this knowledge gap by focusing on the behavior of S. aureus in a polymicrobial infection with E. coli due to its oversized impact on the prevalence and outcomes of PJI.

Results

E. coli affects the implant-associated biofilm growth dynamics of S. aureus

The biofilm dynamics of S. aureus (MSSA, MRSA) and E. coli (EC) during their coexistence in a dual-species implant-associated setting were determined over 48 h. Our previous study determined 6-h and 24-hour-grown biofilms as “nascent” and “established” biofilms, respectively³⁴. In dual-species biofilms, the implant-adhered viable bacteria for MSSA were significantly reduced (>3 log) at 6 and 24 h in the presence of EC when compared to its mono-species control (pvalue = 0.03120 (6 h); 0.0004491 (24 h)). By 48 h, the adherent biofilm viability for MSSA was below the detection limit in dual-species implant-adhered biofilm culture (pvalue = 0.0075) in contrast to that of the control, which demonstrated a highly viable implant-associated biofilm. For EC, no change in biofilm viability was observed in a dual-species biofilm compared to that of the control EC over the same period.

In MRSA and EC dual-species biofilms, MRSA viability at 6 h was increased compared to the control MRSA biofilm (pvalue = 0.0126). By 24 h, MRSA’s adherent biofilm viability was significantly reduced and below the detection limit in the presence of EC compared to that of the control MRSA (pvalue = 0.00076). At 48 h, the viability of MRSA was above the detection limit but significantly reduced compared to MRSA biofilm control (pvalue = 0.0217). The adherent biofilm viability of EC was not altered compared to control EC over the same timeframe (Fig. 1a, b).

Fig. 1. E. coli affects the implant-associated biofilm growth dynamics of S. aureus.

Bacterial viability quantified for (a) MSSA and EC, b MRSA and EC in an implant-associated polymicrobial biofilm cultured for 6-, 24-, and 48-h using the plate count method (n = 6) and fluorescent Gram staining. The error bars in bar charts represent the standard deviation, and an unpaired, one-tailed, and unequal variance student t test was performed. The percentage of syto9/hexidium iodide stain indicating SA and EC viability, respectively, was quantified and represented as a pie chart (n = 10 fields/condition). c Representative images of Mannitol salt and MacConkey agar plates depicting the colony morphology of SA and EC colonies recovered from polymicrobial biofilm. The images demonstrate the small colony variants (SCVs) occurrence and changes in colony color for SA in Mannitol salt and EC in MacConkey agar, respectively. The colony morphology of viable SA and EC from mono-microbial biofilms served as control.

Semiquantitative fluorescent Gram staining demonstrated the percentage of viable bacteria recovered from a dual-species implant-associated biofilm for MSSA, MRSA, and EC (Fig. 1a, b). The viability of MSSA was decreased from 78% to 14% by 48 h, and the viability of MRSA was reduced from 29% to 2% over the same time in the presence of EC. The viable EC recovered increased from 22 to 86% and from 71 to 98% over 48 h in the presence of MSSA and MRSA, respectively. In the dual-species setting, the recovered viable MSSA and MRSA from 6, 24, and 48 h dual-species biofilms, plated on selective agar plates, demonstrated white SCVs, (<0.4 mm) in mannitol salt agar compared to yellow normal-sized SA colonies (1-4 mm) from control mono-species biofilms. On the other hand, the viable EC showed more colorless colonies with corrugated morphology in MacConkey agar compared to mauve, entire colonies from the control mono-species biofilm (Fig. 1c).

Dual microbial interactions influence the implant-associated biofilm properties

The viable SA and EC within the dual-species implant-associated biofilm were confirmed by fluorescent microscopy (Fig. 2a). At 6 h, increased adhesion and colonization of SA were observed compared to EC. Cohabitation of viable SA and EC was observed after 24 h of biofilm growth, with clusters of SA and EC forming adherent aggregates. Evidently, within 48 h, viable EC biofilms dominantly adhered to the SS plates compared to SA.

Fig. 2. Dual microbial interactions influence the implant-associated biofilm properties.

a Fluorescent micrographs of syto9-stained polymicrobial biofilms (6, 24, and 48 h) on stainless-steel implant material. Scale bars,10 µm. White arrows indicate the cohabitation of SA cocci and EC rods. b Scanning electron micrographs of polymicrobial and monomicrobial biofilms (6, 24, and 48 h) grown on stainless-steel implant material (n = 3). Imaging parameters: 1.5–3 kV; magnification, 2, 5, and 10 KX; scale bars, 5, 2, and 1 µm; working distance (WD) ~ 6.5 mm. Red arrows indicate intact SA aggregates; pink arrows indicate deflated SA aggregates; dark blue arrows indicate EC and light blue arrows indicate furrowed EC in the presence of SA.

MSSA/MRSA and EC cohabitation on the implant material was visualized in detail using scanning electron microscopy at 6, 24, and 48 h (Fig. 2b). For the biofilms containing MSSA and EC, SEM revealed the substantial presence of both species at 6 h. The number of attached EC was visibly less at 6 h than the monomicrobial EC biofilm. The adhered MSSA and EC in dual-species biofilm were noticeably viable due to the intact cell walls observed for the spherical SA aggregates and interconnected EC rods, respectively. MSSA and EC mono-species controls demonstrated significant adhesion and propagation of viable biofilm on the implant materials at 6 h. For the biofilms of MRSA and EC, the number of attached MRSA was visibly less than that of EC at 6 h. The pockets of intact spherical aggregates of MRSA were found surrounded by many adhered EC.

After 24 h, there was a somewhat equal proportion of MSSA and EC adhered, with noticeably deflated spheres for MSSA. The proportion of viable MRSA was significantly reduced (as indicated by their sparsity) in the presence of EC. On the other hand, EC demonstrated biofilm formation in the presence of both MSSA and MRSA, comparable to its mono-species biofilm at 24 h. EC developed extensive chain-like growth features in the dual-species biofilm with SA cells attached to the EC biofilm matrix and/or cell surface (Fig. 2b). EC appeared to have undergone morphological changes (less pronounced biofilm sheath) in the presence of SA compared to their mono-species biofilms (prominent biofilm sheath).

At 48 h, only a few MSSA and MRSA aggregates were observed with EC compared to their respective mono-species biofilms, which showed significant colonization. Shortened rods of EC were observed in the presence of MSSA. With MRSA, EC appeared furrowed despite increased adhesion. Overall, more biofilm formation was observed for EC in the presence of SA compared to that of its monomicrobial biofilm (porous cell morphology and sparse adhesion).

E. coli alters the gentamicin susceptibility of S. aureus in implant-associated biofilms

The MIC and MBC of gentamicin for MSSA (0.5 and 2 µg/mL), MRSA (>128 and >256 µg/mL), EC (0.25 and 4 µg/mL) were determined using the micro broth dilution assay. The gentamicin concentration that resulted in >3-log reduction in biofilm viability (MBEC) for MSSA and MRSA in the presence of EC was determined for nascent (6 h) and established (24 and 48 h) biofilms (Fig. 3). In the presence of EC, the MBEC of gentamicin for MSSA was significantly reduced for nascent and established biofilms (<10–25 µg/mL) when compared to the MBEC for MSSA in a mono-species biofilm (100–500 µg/mL) (pvalue = 0.0076 (6 h); 0.0067 (24 h)). The MBEC of gentamicin for EC increased in dual-species biofilms with MSSA for nascent (<100 µg/mL) and 24-h established biofilms (<500 µg/mL) when compared to their monomicrobial biofilms (<25–50 µg/mL) (Fig. 3a).

Fig. 3. E. coli alters the gentamicin susceptibility of S. aureus in implant-associated biofilms.

Heatmaps depicting viable bacterial count (CFU/mL) for (a) MSSA and EC, b MRSA and EC recovered from implant-associated polymicrobial and monomicrobial biofilms exposed to indicated gentamicin concentrations. The data are representative of three independent experiments (n = 2/drug exposure per experiment).

For MRSA in a dual-species biofilm, no significant difference in MBEC was observed for nascent (6 h) and established (24 h) biofilms compared to the mono-species control. For the 48-h biofilms, the MBEC required for >3-log reduction of MRSA was reduced from >500 µg/mL to 200–500 µg/mL in the presence of EC. The MBEC of gentamicin for EC in dual-species biofilm was consistent (<75 µg/mL) at all indicated time points when compared to the monomicrobial biofilm that showed biofilm-maturity-associated increase in the MBEC (>500 µg/mL for 48-h established biofilms) (Fig. 3b).

S. aureus and E. coli mutually regulate molecular events in an implant-associated polymicrobial setting

Biofilm gene expression of MSSA, MRSA, and EC in nascent (6 h) and established (24 h) dual and mono-species implant-associated biofilms was determined. For SA, vraR (peptidoglycan biosynthesis), sigB (sigma factor, stress response regulator), yjbH (virulence and oxidative stress response regulator), icaA and icaD (intercellular adhesin proteins), agrA (quorum sensing, virulence regulator), spa (Protein A), atl (autolysin), clfA (clumping factor A) and sarA (transcriptional regulator) gene expression were determined. For EC, lacZ (lactose metabolism), tomB (biofilm formation, hemolysin toxicity regulator), sdiA (quorum sensing regulator), gyrA (DNA metabolism, transcription regulator), clbA and clbB (Colibactin toxin), and rpoS (stress response regulator) gene expression were determined. In dual-species nascent (6-h) biofilms, MSSA demonstrated increased expression for all genes except sarA compared to the control. On the other hand, MRSA demonstrated significant downregulation for all genes except sarA in the presence of EC compared to its control biofilm. The expression of vraR, clfA, sarA, yjbH, sigB, icaA, icaD, atl, and agrA genes was significantly downregulated in MRSA when compared to that of MSSA in the presence of EC. In an established (24-h) dual-species biofilm, the gene expression was downregulated overall for all MSSA genes, except for sarA, demonstrating increased expression. The spa gene was found to be somewhat elevated in MSSA. For MRSA, yjbH, atl, and clfA were significantly upregulated, and the genes vraR and spa were downregulated in the presence of EC compared to their respective controls. The expression of vraR, clfA, sarA, yjbH, sigB, icaA, icaD, atl, and agrA genes was significantly upregulated in MRSA when compared to that of MSSA in the presence of EC (Fig. 4a).

Fig. 4. SA and EC mutually regulate molecular events in an implant-associated polymicrobial setting.

Gene expression profiles of (a) MSSA and MRSA, b EC in dual and mono-species implant-associated biofilms. c Gene expression profile of MSSA and MRSA exposed to sub-MBEC of gentamicin in dual and mono-species biofilms. The expression of SA genes (vraR, sigB, yjbH, agrA, icaA, icaD, atl, spa, sarA, and clfA) and EC genes (clbA, clbB, lacZ, tomB, rpoS, gyrA and sdiA) was normalized to their respective 16srRNA and fold change 2^(-ΔCq) determined. Error bars indicate standard deviation (n = 3); * indicates pvalue < 0.05. (Welch’s ttest).

In dual-species, nascent biofilms (6-h), lacZ, sdiA, tomB, gyrA, and colibactin-toxin-associated genes (clbA, clbB) gene expression in EC were significantly upregulated by MSSA. In the presence of MRSA, all the genes tested were upregulated in EC compared to the respective control biofilms. The expression of all genes was significantly higher in the presence of MRSA than in MSSA. In dual-species established biofilms (24-h), all genes were downregulated in the presence of MSSA except for tomB, which stayed upregulated compared to the mono-species EC control biofilm. In the presence of MRSA, EC gene expression remained unaltered except for tomB, which was significantly downregulated (Fig. 4b). Compared to MSSA, all genes’ expression was markedly higher in the presence of MRSA, except for tomB, which was downregulated in EC.

The effects of sub-MBEC gentamicin exposure and the presence of EC on the microbial physiology of S. aureus in dual-species biofilms were determined. For the dual-species nascent biofilms, MSSA in the presence of EC and gentamicin demonstrated a notable upregulation for the genes vraR, yjbH, sigB, agrA, icaA, and icaD and downregulation for the gene spa compared to the MSSA mono-species control. On the other hand, for MRSA, the expression for all genes except atl was significantly downregulated, compared to the MRSA mono-species control. The expression of spa, clfA, sarA, yjbH, sigB, icaA, and icaD genes was significantly higher in MRSA when compared to that of MSSA in the presence of EC and gentamicin. In dual-species established biofilms, vraR, yjbH, sigB, agrA, icaA, icaD, and atl gene expression were significantly upregulated in MSSA, while clfA and sarA were downregulated. In MRSA, all genes were upregulated in the presence of EC and gentamicin except spa, which was significantly downregulated. The expression of yjbH, sigB, icaA, and icaD was upregulated considerably in MRSA compared to MSSA, while atl, clfA, and sarA genes were downregulated (Fig. 4c).

Discussion

PJI is a morbid condition associated with total joint arthroplasty that endangers the joint function and longevity of millions of patients⁵¹. Polymicrobial PJI (involving more than one organism) presents a substantial fraction of PJI cases⁷. Although clinical studies have pointed to the causal relationship between polymicrobial PJI and poor treatment outcomes, the implant-associated polymicrobial interactions within biofilms and their physiological properties has not been studied. S. aureus is the most common microorganism in mono- and polymicrobial PJI; most PJI cases would benefit from describing its infection dynamics and focusing treatment strategies targeting it. Our study focuses on polymicrobial interactions in the context of PJI, focusing on how S. aureus pathogenesis for implant-associated infection is affected by the presence of E. coli.

The definition and characterization of biofilm associated with medical devices are still in their infancy, with challenges in harmonizing the state of infections in vitro and in vivo^52,53. In our recently published work on understanding the ‘prophylactic period’ for S. aureus growth and the evolution of susceptibility relevant to PJI, we used subcutaneous implantation and the contamination of material coupons. Using longitudinal susceptibility testing, bacterial gene expression, and immune profiling of the host (in addition to bacterial viability), we identified the acute period as the first 7 days after inoculation, after which the bacteria and the immune response equilibrated^34,54. The bacterial evolution of antibiotic tolerance (for susceptible and inherently resistant bacteria) in vivo corresponded roughly to the first 6 h of in vitro culture. Thus, we identified 6 h of culture as a ‘nascent’ biofilm and 24 h of culture as an ‘established’ biofilm⁵⁵ and focused on the in-vitro study of the S. aureus-E. coli polymicrobial biofilm at 6 and 24 h as a period of potential intervention from a therapeutic and translational point of view.

For the growth dynamics, the coexistence of E. coli with S. aureus (MSSA and MRSA) on implant surfaces led to a dramatic reduction in S. aureus biofilm viability, with significant reduction observed by 24 h and 48 h for MRSA and MSSA, respectively. At 48 h, persisting populations of MRSA were observed, indicating stronger competition from MRSA, potentially changing the dynamics of the polymicrobial biofilm beyond 48 h. E. coli biofilm viability remained unaffected throughout the study period. The data demonstrated that E. coli is the dominant viable species in the presence of S. aureus within a polymicrobial biofilm setting. This observation aligns with previous studies showing that E. coli can inhibit other biofilm-forming organisms through competitive and antagonistic mechanisms, including nutrient competition, secretion of biofilm-disrupting molecules, and quorum sensing regulation^56–58. However, the conventional culture methods further limit our understanding of the exact viability status of S. aureus in the presence of E. coli.

Colony characteristics revealed a wealth of information on the stress adaptation and metabolic activity of S. aureus and E. coli in a polymicrobial biofilm setting. SCVs, usually characterized by reduced metabolic activity and enhanced antibiotic tolerance^59,60, were observed for S. aureus in the presence of E. coli. The mannitol fermentation capabilities of S. aureus, which is a determinant of its metabolic status⁶¹, were compromised by E. coli. The observations indicated a reduced mannitol metabolism by S. aureus within a polymicrobial environment that could facilitate their persistence under hostile conditions^60,62,63. For E. coli, the colonies on MacConkey agar demonstrated altered colony color and shape, indicating a bacterial response to S. aureus-induced stress, which triggered molecular regulation of metabolic pathways in E. coli. The SCV observation is novel and central to this study, which necessitates quantification and in-depth characterization utilizing a large number of samples to conclude the significance of the occurrence of S. aureus persisters in the presence of E. coli.

Fluorescent imaging revealed cohabitation of live S. aureus cocci and E. coli rods on the implant material. Differential staining was challenging on unfixed samples, and alternative approaches (fluorescent bacterial strains) could aid in visualizing live polymicrobial biofilm. Under electron microscopy, the E. coli adhesion was notably less in the presence of S. aureus at 6 h, while the corresponding monomicrobial biofilm exhibited holes that are characteristic of intra- colony channels that facilitate nutrient and oxygen availability within the biofilm⁶⁴. A striking observation was that the presence of S. aureus improved E. coli attachment to the implant material starting at 24 h when compared to its monomicrobial biofilm. S. aureus, being adept at biofilm formation due to its putative adhesins and secreted matrix proteins, could provide a favorable environment for E. coli to adhere to surfaces and establish a biofilm⁴¹. At 6 h, E. coli and MSSA adhered to implant materials similarly, forming discrete aggregates. In contrast, E. coli showed increased colonization in the presence of MRSA at 6 h. By 24 h, E. coli dominated the biofilm structure in dual-species implant-associated biofilms, with chain-like formations enveloping S. aureus. Contrary to the findings on mixed-species biofilms of E. coli with Salmonella and Enterococcus, our study suggest that in the presence of S. aureus, E. coli forms stable biofilms, by developing a robust biofilm sheath and utilizing metabolic byproducts of S. aureus^65–67. The deflation of S. aureus cells, observed for MSSA in the presence of E. coli, indicates a loss of cellular integrity, possibly mediated by direct interactions or secreted antimicrobials such as colicins or other bacteriocins^68,69. For MRSA, although viable cells were observed in the presence of E. coli at 24 h, the viability was not captured in agar plates, which suggests a non-culturable persister status⁷⁰. E. coli biofilms also underwent morphological changes (shortened rods, furrowed cell morphology, reduced curli fibers, and biofilm sheath), highlighting adaptive responses to coexistence, which may include altered matrix production or surface adhesion strategies^71,72.

Gentamicin, being an integral part of local drug delivery approaches for PJI prophylaxis, was chosen for evaluating the MBEC of implant-associated polymicrobial cultures. The presence of E. coli significantly reduced the MBEC of gentamicin for MSSA, from 100–500 µg/mL in mono-species biofilms to <10 µg/mL in dual-species biofilms. This enhanced susceptibility is likely due to the inhibition of S. aureus biofilm formation by E. coli, leading to reduced matrix-mediated protection^73,74. Conversely, the MBEC for E. coli increased in dual-species biofilms, potentially due to enhanced biofilm maturation or changes in metabolic states⁷⁵. For MRSA, E. coli had a limited impact on gentamicin susceptibility at early timepoints but reduced the MBEC for established biofilms at 48 h. This selective effect underscores the complex interactions between the two species, which may involve differential gene expression, biofilm architecture, or metabolic compatibility.

Gene expression analyses revealed significant changes in dual-species biofilms. In 6-ht biofilms, MSSA demonstrated upregulation of stress response (vraR, yjbH, sigB, agrA), adhesion (atl, spa, clfA), and biofilm-associated genes (icaA, icaD, agrA) in the presence of E. coli (Fig. 4a), suggesting an initial attempt to establish a biofilm despite competition from E. coli⁷⁶. However, MRSA showed substantial downregulation of these critical genes in nascent biofilms, indicating more potent competition from E. coli during the initial stages of biofilm formation. In 24-h biofilms, the sarA and spa expression was triggered in the presence of EC, while all other genes were deregulated, suggesting activation of critical stress adaptation by MSSA⁷⁷. For MRSA, the expression of pathogenicity-associated genes was triggered (yjbH, atl, clfA) together with the negative regulation of peptidoglycan biosynthesis (vraR) and protein A antigen (spa), indicating a potential switch to persister mode (SCVs) by MRSA, which could impact the outcome of PJI in a clinical setting⁶².

Both MSSA and MRSA, in 6-h biofilms, triggered the upregulation of quorum sensing (sdiA, lacZ), biofilm formation (tomB), and stress-response (gyrA, rpoS) genes in E. coli and colibactin-associated genes (clbA, clbB), consistent with enhanced biofilm formation, virulence, and stress adaptation^78,79. Notably, the responses were significantly augmented in the presence of MRSA compared to MSSA. In 24-h biofilms, however, the E. coli responses were muted except for further activation of tomB in the presence of MSSA, suggesting further strengthening of the biofilm⁸⁰. On the contrary, the presence of MRSA deregulated E. coli biofilm formation. The data demonstrated a strain-specific, biofilm maturity-dependent interaction dynamics by S. aureus within a dual-species biofilm with E. coli .

In 6-h biofilms exposed to sub-inhibitory concentrations of gentamicin, the stress responses were activated in MSSA and subdued in MRSA compared to their respective controls. Notably, MSSA protein A (spa) was downregulated, likely in an antibiotic-mediated manner⁸¹. The presence of E. coli and gentamicin highly impacted the physiological responses of MRSA compared to MSSA, indicating a synergistic action that could aid in MRSA eradication in its nascent biofilm stages. As biofilm maturity increased at 24-h, stress responses and biofilm-associated gene expression remained elevated in MSSA and MRSA in the presence of E. coli and gentamicin. Meanwhile, adhesion and virulence genes (clfA, sarA, and spa) were negatively regulated, suggesting the sustained synergistic action of E. coli and gentamicin against MSSA and MRSA. Taken together, E. coli is potentially the dominant species in an implant-associated polymicrobial biofilm with MSSA and MRSA and is capable of anti- S. aureus responses with and without the presence of broad-spectrum antibiotic, gentamicin.

The observed suppression of S. aureus biofilm viability by E. coli suggests potential therapeutic strategies against S. aureus, leveraging competitive microbial interactions to manage biofilm-associated infections. For example, E. coli-derived metabolites could disrupt S. aureus biofilms on implants. Nevertheless, the persistence of E. coli biofilms on implants and their enhanced antibiotic resistance in dual-species settings pose challenges, as they may serve as reservoirs for chronic infection⁸². Furthermore, the reduction in gentamicin MBEC for MSSA and the limited effect on MRSA highlight the need for tailored antibiotic regimens in polymicrobial infections. These findings also emphasize the importance of early intervention, as biofilm maturity significantly affects antibiotic susceptibility.

Since S. aureus is the primary causative microorganism cultured from tissues extracted in PJIs, and E. coli is rarely encountered as a single causative microorganism, a limitation of the study is the inability to fully capture the in vivo environment of the polymicrobial interaction. Our previous work observed that the growth dynamics and MBEC evolution for gentamicin were different for S. aureus when implanted in the joint compared to subcutaneous implantation³⁴. A significant difference between these implant locations is the hypoxic environment in the joint, which can presumably favor S. aureus growth²². Future studies of polymicrobial infection will focus on the effect of hypoxia on interspecies interactions. Critical factors include suitable culture media to grow and maintain “realistic” implant-associated biofilm in vitro and dynamic culture conditions to capture acute and chronic infection periods without media depletion (for in vitro culture >24 h). Synovial fluid or its substitutes in a dynamic setting could provide a clinically relevant environment^83,84. Although standard laboratory techniques of bacterial culture use 10⁵ CFU for inoculation, it has been shown that as few as 100 bacteria can cause an established infection in the presence of implant material⁸⁵. Further study of total inoculum and the ratio of contaminating organisms may also be appropriate. An inherent limitation of all standard microbiology techniques is the dependency of viability results on the compatibility of the specific microorganism with the plate culture, which is included after bacterial recovery from the original environment. Future studies should investigate the exact mechanisms by which E. coli inhibits S. aureus biofilm formation, including the role of secreted factors and nutrient competition. Additionally, exploring the impact of different antibiotics and implant materials on polymicrobial biofilms could inform the development of optimized treatment protocols and dosing requirements for antibiotic-eluting implant materials⁸⁶. Finally, in vivo studies are essential to validate these findings and assess their clinical relevance.

Our long-term goal is to develop effective therapeutic strategies to prevent and treat PJI, and this study is the first in-depth investigation of implant-associated polymicrobial interactions to the best of our knowledge. Understanding and manipulating the behavior of S. aureus in mono- and polymicrobial cultures will help create clinically relevant information on bacterial characteristics and have the highest potential for decreasing infection rates.

Methods

Bacterial strains and culture conditions

Methicillin and gentamicin susceptible S. aureus control strain ATCC 12600 (MSSA) and gentamicin susceptible E. coli control strain ATCC 25922 (EC) were commercially obtained (MicroBiologics®, MN, USA). Methicillin and gentamicin-resistant S. aureus clinical strain L1101 (Mu50; ATCC 700699 MRSA) was kindly provided by Dr Kerry Laplante (University of Rhode Island). The antibiotic susceptibility profiles of the bacterial strains are provided in Table S1 (Supplementary file). The lyophilized pellet or frozen glycerol stocks of bacteria were revived in Tryptic Soy Agar (TSA) plates statically incubated at 35 °C for 16–18 h. The colonies were grown overnight in Tryptic Soy Broth (TSB) before all experiments. The turbidity was spectrophotometrically determined at 600 nm, and the bacterial load was enumerated using strain-specific standard growth curves (MSSA, $y=7\times {10}^8\left(x\right)-3\times {10}^7$; MRSA, $y$ = $3\times {10}^8\left(x\right)-3\times {10}^7$; EC, $y=8\times {10}^8\left(x\right)$).

Biofilm growth dynamics

To establish dual-species implant material-associated biofilms in vitro, 1 × 10⁵ CFU/mL of S. aureus and E. coli cultures (1:1) in 1mL TSB were seeded on 316 L stainless steel plates (SS plates, 10 × 3 × 1 mm³) within 24 well plates and statically incubated at 35 °C for a period of 6, 24, and 48 h. At each time point, the spent media was aspirated, and the SS plates were washed 3× for 2 min with sterile phosphate-buffered saline (PBS) to remove non-adherent bacteria. The plates were sonicated for 40 min (sonication time optimized for maximum bacteria recovery with minimum cell death) in 1 mL PBS (Fisher Scientific, USA), and the sonicate fluid was diluted and plated onto Mannitol salt agar and MacConkey agar to differentially count viable S. aureus and E. coli, respectively. The agar plates were statically incubated at 35 °C for 16–18 h, and the viable adherent bacteria count was determined. Mono-species biofilms of S. aureus and E. coli grown on SS plates were controls for the experiment.

Fluorescent biofilm viability assay

The implant material-adhered dual-species biofilms from 6, 24, and 48 h were washed 3× and sonicated in PBS for 40 min. The sonicate fluid was stained using LIVE BacLight Bacterial Gram Stain Kit (Invitrogen, USA) containing Syto9 and Hexidium iodide fluorescent stains. 5 µL of sonicate fluid was mounted on glass slides, and differentially stained bacteria from 10 fields were counted using GFP and Texas Red filters in a Nikon Ti2 Eclipse microscope at 400X. The counts were analyzed using the Biofilm viability checker tool in Fiji image processing software, and the percentage viability for each species was quantified⁸⁷.

Biofilm visualization

Dual-species biofilms grown on SS plates for 6, 24, and 48 h were washed 3x with PBS to remove non-adherent bacteria. The adherent biofilms on SS plates were stained by applying 200 µL of diluted Syto9 dye on the surface to visualize live dual-species bacteria within the biofilm. The stained SS plates were mounted on coverslips and visualized at 600x using a Nikon Ti2 Eclipse microscope with a GFP filter. The adherent biofilms (dual and mono-species) were also visualized using scanning electron microscopy (SEM, n = 3; six fields/sample)⁵⁵. Briefly, the SS plate-adhered biofilms were washed 3x with PBS to remove non-adherent bacteria. The plates were fixed using 2.5% glutaraldehyde for a minimum of 48 h. The plates were subjected to 2% OsO4: 0.2% Ruthenium Red (1:1) staining for 1 h and washed thoroughly with deionized water (2×,10 min each) before incubating with 1% tannic acid for 1 h. The SS plate-adhered biofilms were washed with deionized water (2×, 10 min each) and air-dried completely before preparing the samples for SEM. The samples were visualized using Zeiss Gemini 560 FESEM at 1.5–3 kV and at 2 K, 5 K, and 10 K magnification. Mono-species biofilms on SS plates served as controls.

Minimum biofilm eradication concentration (MBEC) measurement

Biofilms grown on SS plates for 6, 24, and 48 h were washed 3× with sterile PBS to remove non-adherent bacteria. The biofilms were exposed to a range of gentamicin concentrations (1–500 µg/mL) prepared in 10% TSB supplemented PBS for 24 h at 35 °C. The gentamicin-exposed biofilms were rinsed further with 3× PBS for 2 min, and the viable adherent bacteria were determined using the differential plate count method described previously. The MBEC was the lowest concentration of gentamicin required to cause > 3-log reduction in biofilm viability. Mono-species biofilms grown on SS plates for the indicated time points served as controls for this experiment.

Bacterial RNA extraction

Dual and mono-species biofilms were grown on 22 × 22 mm SS plates for 6 and 24 h. At indicated time points, the plates were washed with PBS, and the material surface was swabbed thoroughly and resuspended in PBS. To determine the effect of gentamicin on the gene expression of SS-plate adhered dual-species biofilms, the biofilms grown for 6 and 24 h were exposed to sub-MBEC for gentamicin (Table S2, Supplementary file) for a period of 24 h and processed as described above. The bacterial suspensions were centrifuged at 10,000 × g for 10 min, and the total bacterial RNA was extracted from the pellet using the RNeasy PowerBiofilm kit according to the manufacturer’s instructions (Qiagen, USA). The RNA was spectrophotometrically assessed using the NanoDrop™ One (Thermo Fisher Scientific, USA).

Quantitative PCR

The bacterial RNA was converted to cDNA using the iScript™ Reverse Transcription Supermix protocol (Bio-Rad, USA). The cDNA template was used to quantify the expression of S. aureus and E. coli genes using the iTaq™ Universal SYBR Supermix (Bio-Rad, USA) and specific primers. The genes and their primers used in this study are provided in Table 1. Bacterial 16srRNA served as housekeeping gene to normalize the test gene expression. Mono-species bacterial gene expression served as a control for the experiment. The gene expression analysis was performed using the 2^-∆Cq method, and the expression fold change was determined for both dual-species and mono-species biofilms.

Table 1.

List of primers

S. aureus	vraR	FP 5ʹ- AACTCTGCGCGCTTTTTCAT-3ʹ
		RP 5ʹ- ATATCGCCGATGCAGTTCGT-3ʹ
	icaA	FP 5ʹ-TTGTCGACGTTGGCTACTGG -3ʹ
		RP 5ʹ- GCGTTGCTTCCAAAGACCTC -3ʹ
	icaD	FP 5ʹ- CGCTATATCGTGTGTCTTTTGGA -3ʹ
		RP 5ʹ- TCGCGAAAATGCCCATAGTT -3ʹ
	16srRNA	FP 5ʹ- AGACCAGAAAGTCGCCTTCG -3ʹ
		RP 5ʹ-TCAACCGTGGAGGGTCATTG -3ʹ
	agrA	FP 5ʹ- TGCCCTCGCAACTGATAATCC -3ʹ
		RP 5ʹ- TACCAACTGGGTCATGCTTACG -3ʹ
	yjbH	FP 5ʹ- CGGCACGTACACGACCTTG -3ʹ
		RP 5ʹ- TCCATTTAGCTCCGATTGCTTCA -3ʹ
	sigB	FP 5ʹ- AACCGATACGCTCACCTGTCT-3ʹ
		RP 5ʹ- TGGGGCAACAAGATGACCA -3ʹ
	atl	FP 5ʹ- GTGTTGGTGTAGGCGTAGGT -3ʹ
		FP 5ʹ- GTGTCTGGCTCTGGAAACCA -3ʹ
	clfA	FP 5ʹ- GCGGTTCAGATTCGGGTAGT -3ʹ
		FP 5ʹ- CGCTCGCTGAGTCGGA -3ʹ
	spa	FP 5ʹ- AACGCTGCACCTAAGGCTAA -3ʹ
		FP 5ʹ- CAGCAAACCATGCAGATGCT -3ʹ
	sarA	FP 5ʹ- CGTTGTTTGCTTCAGTGATTCG -3ʹ
		FP 5ʹ- ACAACCACAAGTTGTTAAAGCAGT -3ʹ
E. coli	tomB	FP 5ʹ- CGATTACCTGACTTCCGCCA -3ʹ
		RP 5ʹ- CGGCGATCAACCTCCAGTT -3ʹ
	lacZ	FP 5ʹ- TGCTGCTGGTGTTTTGCTTC -3ʹ
		RP 5ʹ- GGCGGTGATTTTGGCGATAC -3ʹ
	clbA	FP 5ʹ- TACGTGCAAATATGGCAAAC -3ʹ
		RP 5ʹ- CTAATAGCAACGGCTACTGT -3ʹ
	clbB	FP 5ʹ- TCAAGGTAGCCAATATCACG -3ʹ
		RP 5ʹ- ATCCTCATCGTCAAACAACA -3ʹ
	sdiA	FP 5ʹ- GCTGATGTTTTCCCTTCCGC -3ʹ
		FP 5ʹ- TTACTGGTGCGCGAAAGTCT -3ʹ
	gyrA	FP 5ʹ- GCTTAACAACCTCTACTCCC -3ʹ
		FP 5ʹ- CTTCAAGGATATGAGCACGA -3ʹ
	rpoS	FP 5ʹ- AATCGCCCGTTCAATCGTCT -3ʹ
		FP 5ʹ- GGTAAAAATTGCCCGCCGTT -3ʹ
	16srRNA	FP 5ʹ- CTTGCTGCTTTGCTGACGAG -3ʹ
		RP 5ʹ- GGTCCCCCTCTTTGGTCTTG -3ʹ

Statistical analysis

Welch’s t test was performed for the growth dynamics, MBEC data, and gene expression datasets to compare dual-species vs. mono-species biofilms and analyze MRSA vs MSSA groups. P value < 0.05 was considered statistically significant. The p value corresponding to each data point is included in the figures and the supplementary file (Tables S3, S4).

Author contributions

A.S. and E.O. conceptualized the study and wrote the paper. A.S., F.B.M., S.U.D., performed the bacterial growth characterization and antibiotic susceptibility studies. A.S., F.B.M., S.U.D., and N.I. performed the imaging. A.S., F.B.M., and D.M., performed the qPCR analyses. E.O. acquired the funding and E.O. and O.K.M. supervised the study. All authors read and approved the final manuscript.

Data availability

All data generated or analyzed during this study are provided within the manuscript and supplementary file.

Competing interests

Authors A.S., F.B.M., S.U.D., N.I. and D.M. declare no financial and non-financial competing interests. O.K.M. receives royalties from Corin, Mako, Iconacy, Renovis, Arthrex, ConforMIS, Meril Healthcare, Exactech, Cambridge Polymer Group; holds stake/equity in Cambridge Polymer Group, Orthopedic Technology Group, Alchimist and declares no non-financial competing interests. E.O. receives royalties from Corin, Iconacy, Renovis, Arthrex, ConforMIS, Meril Healthcare, and Exactech and is a paid consultant for WL Gore & Assoc. EO declares the following non-financial competing interests; serves as Editor for The Journal of Biomedical Materials Research and as Officer/ in Committee of Society For Biomaterials (SFB) and International Society for Technology in Arthroplasty (ISTA).

Supplementary information

The online version contains supplementary material available at 10.1038/s41522-025-00754-2.