Abstract

Introduction

Biofilms are a common and critical problem in orthopedic infections, such as septic synovitis and implant-associated infections. Infection-associated complications and resistance of biofilms to treatment contribute to increased treatment costs and treatment failure in human [1–3] and veterinary patients [4–7]. A biofilm is a community of bacteria adhered to a surface or to each other and encased in a self-secreted extracellular matrix of carbohydrates, proteins, and DNA [8–14]. Biofilms contribute to antimicrobial resistance by shielding indwelling bacteria from immune cells and antimicrobials [12,15–18], facilitating bacterial metabolic downregulation [19,20], and promoting exchange of antimicrobial resistance genes [7]. Resistance of biofilm-bound bacteria to antimicrobials and immune cells contributes to persistent infection in the face of parenteral and regional antimicrobial administration [6,21–23]. Multi-stage surgical removal of infected implants and debridement of affected tissue are often required in conjunction with prolonged antimicrobial therapy for infection resolution [4,5,24].

Despite aggressive and prolonged medical and surgical therapy, sequalae of orthopedic infections involving biofilms include irreversible tissue damage due to chronic inflammation, such as osteolysis with subsequent implant failure [4,25,26] and chondromalacia in septic synovitis [6,27]. Additionally, persistent biofilms serve as a nidus for dissemination of infection to the bloodstream and vital organs, which can be life-threatening [2,28–31]. The complications related to biofilm infections are dire; 26–55% of people with orthopedic implant infections die within 5 years of initial diagnosis [29,31]. Persistent orthopedic biofilm infections are an equally significant problem in horses and are often caused by Staphylococcus aureus and Escherichia coli, which readily form biofilms on metallic orthopedic implants [7,32–34] and in synovial fluid as floating aggregates [12,35–38]. The resulting chronic lameness, loss of athletic potential, and reduced quality of life due to chronic, unresolved orthopedic infections [4,6,22,24] make euthanasia ultimately necessary in up to 54% of horses [4,5,22,39,40]. A treatment that disrupts established orthopedic biofilms and improves success of antimicrobials and surgical intervention is desperately needed.

Mesenchymal stromal cells (MSC) derived from equine bone marrow are immunomodulatory and used for the treatment of musculoskeletal injury and osteoarthritis in horses [41–47]. Recently, bone marrow-derived MSC from humans and horses have been demonstrated to secrete antimicrobial peptides in vitro [48–52], ex vivo [53]. Production of antimicrobial factors by MSC have also been demonstrated in murine and canine models of infection [54–56]. Equine bone marrow-derived MSC exert antimicrobial effects in a paracrine manner, via secretion of antimicrobial peptides. Cathelicidin is one such antimicrobial peptide that causes microbial growth inhibition bacterial cell membrane depolarization [49,51] and interferes with bacterial metabolism of key nutrients [50]. Equine bone marrow-derived MSC also enhance the host antibacterial response through stimulation of phagocytosis by neutrophils [52] and secretion of antimicrobial peptides from keratinocytes [53]. MSC from the adipose tissue of rats and mice or from human menses have been used successfully with antimicrobials to significantly reduce bacterial numbers [55–57]. TLR-activation of equine MSC used in combination with vancomycin successfully reduced bacterial counts in an equine model of multi-drug resistant S. aureus [52]. This concept of synergy between MSC and antimicrobials has high therapeutic potential.

To date, studies have primarily evaluated the ability of MSC to prevent establishment of infection, including planktonic (free-floating) bacterial reduction and inhibition of biofilm formation [49–52]. Harman et al. demonstrated that conditioned medium from equine bone marrow-derived MSC prevented biofilm formation by planktonic cultures of S. aureus and E. coli [49]. However, it was unclear if the anti-biofilm properties were due to bacterial killing prior to biofilm establishment, or direct disruption of biofilm formation. The biofilm matrix physically protects indwelling bacteria from being killed by immune cells and can utilize the host immune response to its advantage by incorporating secreted proteins [12,37,58] and DNA [59] into its matrix, thereby enhancing the barrier function of the biofilm. Therefore, it is important to investigate the ability of equine MSC to reduce established orthopedic biofilms containing a mature extracellular matrix. It is also unknown whether equine MSC can synergize with antimicrobials used clinically in equine orthopedic infections [21,60,61] to reduce established orthopedic biofilms to a greater extent than MSC alone.

The objectives of this study were to evaluate 1) the ability of equine bone marrow-derived MSC to reduce biomass (total cells and matrix), biofilm size, and live bacterial counts of established S. aureus or E. coli biofilms in vitro, and 2) whether the combination of MSC and the antimicrobial amikacin sulfate would be more effective than either alone. We hypothesized that 1) MSC would reduce biomass, size, and live bacterial counts of established S. aureus and E. coli biofilms compared to untreated controls, and 2) the combination of MSC with amikacin would reduce biomass, size, and live bacteria of S. aureus and E. coli biofilms more than MSC or amikacin alone.

Materials and methods

Overview

MSC were isolated from bone marrow from 5 horses and co-cultured at passage 3, alone or with a concentration of amikacin sulfate achieved in vivo [21,62], with established 24-hour S. aureus or E. coli biofilms in a transwell plate system (Fig 1). This system allowed MSC and biofilms to interact in a paracrine manner, without direct contact, during co-culture and facilitated easy biofilm separation after co-culture for imaging and quantification of biomass and live bacterial counts. Biofilm biomass, live bacterial counts, and biofilm size were quantified following 24 or 48 hours of co-culture.

Open in a separate window

Fig 1

Experimental workflow to quantify biofilm reduction.

MSC and biofilms of S. aureus or E. coli were prepared separately. Biofilms were established on 24-well tissue culture-treated plates for 24 hours and were co-cultured with 1 x 10⁶ passage 3 MSC seeded in transwell inserts alone or with amikacin sulfate for an additional 24 or 48 hours. Biofilms that remained untreated for the co-culture period were the negative control. Open circles indicate preparatory steps and solid circles indicate steps to establish co-cultures and perform biofilm quantification assays.

Treatment groups

Six treatment groups were established in triplicate for 24 or 48 hours Table 1. Untreated biofilm (untreated/ negative control); biofilm + 30 μg/mL amikacin sulfate (Avet Pharmaceuticals, Inc^®) [21,62,69]; biofilm + MSC; biofilm + MSC + 30 μg/mL amikacin sulfate; biofilm + 500 μg/mL amikacin sulfate with 2% sodium dodecyl sulfate (positive control) [70–72]. Wells containing no biofilm served as contamination controls. All treatment groups were cultured in antibiotic-free MSC medium with 5% FBS. Biofilm and co-culture treatment groups were established as described below.

Table 1

Treatment groups for Biofilm-MSC co-culture.

		Biofilm (S. aureus or E. coli)	MSC in transwell insert (1.0 x 106)	Amikacin (30 μg/mL)	Other
Treatment Groups	Untreated/ negative control a
	30 μg/mL amikacin a
	MSC a
	MSC + 30 μg/mL amikacin a
	Positive control a				500 μg/mL amikacin sulfate with 2% sodium dodecyl sulfate
	Contamination control a

Open in a separate window

^a All treatment groups were cultured in antibiotic-free MSC medium with 5% FBS. Groups were established in triplicate for 24 and 48 hour time-points.

Results

MSC reduced biofilm organization and gross biofilm area

MSC maintained 75% of baseline viability following co-culture with S. aureus biofilms and 70% following co-culture with E. coli biofilms for 48 hours. Biofilms were successfully established as a central aggregate that protruded up from the bottom of the well toward the meniscus. Established biofilms of each bacterial species that were then co-cultured with MSC were visibly less organized and had smaller, less-defined biofilms compared to untreated biofilms at both time points (Fig 2A and 2B). E. coli biofilms were less organized, with a more expansive central region, than S. aureus biofilms across treatment groups. When placed under mechanical stress on the same rotary shaker used for enzymatic biofilm digestion at 200 rpm, MSC-treated biofilms subjectively dispersed more readily than untreated or amikacin-treated biofilms (S1 and S2 Videos).

Open in a separate window

Fig 2

Effect of MSC ± amikacin on biofilm size.

In situ digital images of biofilms from one representative horse for S. aureus (A) and E. coli (B) following 24 or 48 hours of treatment. The negative control was untreated biofilms. (C, D) Dot-plots of normalized biofilm area (mm²) from n = 5 horses. Horizontal black lines indicate mean biofilm area across horses, and circles indicate mean biofilm area for each individual horse at 24 (open circles) or 48 hours (closed circles) as measured from photographic analysis. Brackets and p-values indicate groups significantly different (p < 0.05) from each other.

Correlation between observers for biofilm area using pixel thresholding was high, with a Pearson’s r coefficient of 0.911 (95% CI 0.865–0.940, p < 0.001). S. aureus biofilm area was significantly affected by treatment only (F = 12.42, df = 3, p < 0.001), whereas E. coli biofilm area was affected by treatment (F = 16.46, df = 3, p < 0.001) and time (F = 7.81, df = 3, p = 0.013), with a significant interaction between treatment and time (F = 4.65, df = 1, p = 0.016). MSC reduced S. aureus biofilm area compared to untreated biofilms at 24 (p = 0.005) and 48 hours (p = 0.001), while MSC + amikacin reduced S. aureus biofilm area compared to amikacin alone at 48 hours (p = 0.037), with no difference at 24 hours (p = 0.154) or compared to MSC alone at 48 hours (p = 0.998) (Fig 2C). MSC reduced E. coli biofilm area compared to untreated biofilms at 24 hours (p = 0.007) with no changes at 48 hours (p = 0.606) (Fig 2D). MSC + amikacin reduced E. coli biofilm area compared to MSC alone (p < 0.001) at 48 hours, with no changes at 24 hours (p = 0.202) or compared to amikacin at either timepoint (p = 0.980 at 24 hours; p = 0.843 at 48 hours).

Co-culture with MSC reduced total biomass of S. aureus biofilms

Co-culture with MSC reduced S. aureus biomass compared to untreated biofilms at 24 and 48 hours (p < 0.001 for both) (Fig 3A). MSC with amikacin reduced S. aureus biomass compared to amikacin alone at 24 (p = 0.014) and 48 hours (p = 0.053). MSC + amikacin reduced biomass of E. coli biofilms compared to untreated controls at 24 (p = 0.032) and 48 hours (p = 0.023) (Fig 3B). No other differences were detected between treatment groups.

Open in a separate window

Fig 3

Biofilm biomass quantified by crystal violet staining for S. aureus and E. coli.

Bars represent mean ± SD from biofilms co-cultured with MSC from n = 5 horses at 24 (grey) or 48 hours (black). The negative control was untreated biofilms. Brackets and p-values indicate groups significantly different from each other.

Co-culture with MSC reduced S. aureus CFU at 48 hours, with no improvement in bacterial killing of either species by MSC + amikacin versus amikacin alone

CFU/biofilm of MSC-treated S. aureus biofilms was reduced by 0.5 log₁₀-fold compared to untreated biofilms at 48 hours (p = 0.036), with no difference at 24 hours (p = 0.997) (Fig 4A). MSC + amikacin reduced CFU of S. aureus biofilms compared to MSC-treated biofilms by 2 log₁₀-fold at both time points (p < 0.001). As for S. aureus, MSC + amikacin reduced E. coli CFU by 2 log₁₀-fold compared to MSC alone at both timepoints (p < 0.001) (Fig 4B). However, reductions in CFU of both bacteria by MSC + amikacin were equivalent to reductions achieved by amikacin alone. Additionally, MSC + amikacin treatment reduced CFU of both bacteria by 2 log₁₀-fold at both timepoints compared to untreated controls (p < 0.001). No other differences were detected between treatment groups.

Open in a separate window

Fig 4

Live bacteria quantified by total colony forming units (CFU) per biofilm for S. aureus and E. coli.

Bars represent mean ± SD from biofilms co-cultured with MSC from n = 5 horses at 24 or 48 hours. The negative control was untreated biofilms. Brackets indicate groups significantly different from each other.

Discussion

Our study demonstrates that equine bone marrow-derived MSC reduce biomass and the biofilm area of established S. aureus biofilms, but that MSC + amikacin did not reduce biomass further compared to MSC alone. However, MSC + amikacin was effective at reducing S. aureus biofilm biomass compared to amikacin alone at 24 hours and at reducing biofilm area compared to amikacin alone at 48 hours. Taken together, our results are supportive of a potential synergy between amikacin and MSC in combatting S. aureus biofilms. In contrast, while E. coli biofilm area and visible biofilm organization were reduced by MSC at 24 hours and by MSC + amikacin at 48 hours compared to no treatment, neither MSC nor MSC + amikacin decreased E. coli biofilm biomass. With the exception of S. aureus following 48 hours of MSC treatment, MSC did not otherwise reduce live bacterial counts in our biofilms. Furthermore, MSC + amikacin reduced live counts of both bacteria equivalently to amikacin alone.

MSC response is paracrine in nature based on co-culture of planktonic bacteria with either equine bone-marrow derived MSC or MSC-conditioned medium [49–51]. Our transwell system was chosen over using conditioned medium to enable a MSC-mediated antibacterial response through real-time paracrine feedback without allowing physical contact with the bacteria. MSC from horses and humans express pattern recognition receptors (PRR) [52,78–80] that can be stimulated by bacterial pathogen-associated molecular patterns or synthetic ligands to enhance planktonic bacterial killing via antimicrobial peptide synthesis [48,50,52]. Our co-culture model will enable future investigations using pre-stimulated MSC. A survey of equine veterinary practitioners found that 72.1% use bone marrow-derived MSC at least once a year for orthopedic conditions [81]. Therefore, use of MSC versus conditioned medium also more closely aligns with current clinical practice of MSC injection and more closely aligns with the use of horses as a naturally occurring model in future studies.

Although our co-culture design allowed for stimulation of MSC PRR stimulation by bacterial products, it is unclear why a more uniform anti-biofilm MSC response was not observed. The S. aureus and E. coli strains used in our study were selected because of their ability to form robust biofilms, their clinical relevance [82,83], and their susceptibility to amikacin sulfate [84]. However, evaluation of biofilm-forming strains that generate a strong inflammatory response, such as methicillin-resistant USA300 [25,51,53], may stimulate a greater anti-biofilm MSC response than that seen in our study. Pre-treatment of MSC to increase their immunomodulatory properties prior to bacterial co-culture may also improve biofilm reduction. Possible candidates include PRR agonists [50,52,85], hypoxic exposure [85], or three-dimensional culture [86,87]. Inclusion of neutrophils in this model may have improved biofilm reduction and more closely captured the MSC antibiofilm potential that would occur in vivo [52,55,73]. MSC may indirectly exert an antimicrobial effects through enhancement of the antibacterial responses of immune cells, such as macrophages and neutrophils, present in infected tissues [55,88].

Our study demonstrated a difference in the ability of MSC to combat established biofilms of S. aureus versus E. coli. These findings are consistent with greater reduction of live bacterial counts for Gram-positive versus Gram-negative organisms by equine adipose-derived MSC [89] and platelet lysate [35]. There may be differences in the response of MSC to S. aureus versus E. coli biofilms related to differences in biofilm matrix composition as S. aureus and E. coli biofilms differ in their specific matrix carbohydrates [11,90] and adhesion proteins [8,10,91]. Activation of the MSC antimicrobial response is driven by PRR which may differ between biofilms versus planktonic bacteria, or between bacterial species. Activation of equine MSC with a TLR3 agonist did not increase the anti-biofilm effect against S. aureus, despite successfully reducing planktonic S. aureus in vitro [52]. When equine allogeneic MSC were pre-stimulated with the same TLR3 agonist and used in an equine model of multi-drug resistant S. aureus synovial infection, bacterial numbers increased at 24 hours, but then were significantly reduced by 96 hours [73]. These findings indicate that a longer incubation time may be needed to demonstrate anti-biofilm effects, or optimization using specific or combination PRR activation is needed. The importance of specific PRR activation has been demonstrated using human umbilical cord blood derived MSC that had enhanced activity against E. coli via the TLR4 pathway but not the TLR2 pathway [92].

Donor-dependent variation in MSC anti-biofilm activities was observed with both bacterial strains but was more pronounced for E. coli and may have impacted our ability to detect differences between treatment groups. In particular, there was greater variation in E. coli untreated (negative control) and amikacin-treated biofilm biomass. Bone marrow-derived MSC from different horses vary in their immunomodulatory properties [86,93], proliferation [94], differentiation [94], and cell yield on isolation from tissues [95]. The potential for inter-donor variation in overall MSC anti-biofilm function and in the ability of E. coli versus S. aureus to stimulate anti-biofilm responses requires further investigation. Donor screening [96,97] or pooling of active factors from multiple horses [35,36] may be required to minimize the impact of inter-donor variation.

The primary mechanism of MSC against established S. aureus and E. coli biofilms in this study was dispersion of the biofilm matrix based on observed reductions in S. aureus and E. coli biofilm size and S. aureus biofilm matrix components. Equine bone marrow-derived MSC kill planktonic bacteria and prevent biofilm formation by disrupting bacterial cell membranes through secretion of amphipathic antimicrobial peptides [48,49] and cysteine proteases [51]. MSC antimicrobial proteins may disrupt the matrix of established biofilms in a similar fashion, causing reductions in biomass and biofilm area.

With the exception of S. aureus following a 48 hour co-culture, MSC did not reduce live bacteria of our biofilms compared to untreated biofilms. The reasons for this lack of correlation between live bacterial numbers and reduced biomass and biofilm area are uncertain. Biofilm area measured only the biofilm whereas CFU measures the sum of viable bacterial from both the biofilm and the suspension. The lack of reduction in CFU after biofilm reduction indicates that either (a) the bacteria removed from the biofilm remained viable or (b) the surviving bacteria that were removed were able to reproduce more quickly in the suspension than in the biofilm. However, ongoing multiplication of dispersed bacteria prior to quantification may have masked these immediate reductions, as planktonic bacteria divide more rapidly than biofilm-bound bacteria [16]. Three-dimensional imaging techniques, including confocal microscopy [12] or scanning electron microscopy [26], may have captured MSC-specific biofilm reductions not seen with our techniques. While MSC-treated biofilms were subjectively easier to disperse under mechanical stress in our study, biomechanical testing could be used to quantify changes in biofilm matrix physical properties, such as stiffness [14], following MSC treatment. The potential for MSC to disperse biofilm matrix may prove useful in a therapeutic context by increasing antimicrobial and immune cell access to live bacteria in biofilms, resulting in improved infection clearance. Biofilm matrix dispersion by MSC may also facilitate physical removal of biofilms intraoperatively by pulsatile lavage or ultrasonic debridement [98]. Staged treatment of biofilms with MSC to initially disperse the matrix, followed by surgical debridement to remove residual matrix and antimicrobials to kill liberated live bacteria, may be a useful approach to improve the therapeutic utility of biofilm matrix dispersal.

Use of established biofilms with a mature matrix likely limited the ability of MSC and/or amikacin to penetrate the biofilm matrix compared to planktonic bacteria. Our MSC:bacteria ratio or the physiologic dose of amikacin sulfate chosen, may have resulted in lower than expected live bacterial reductions [21,62]. Ratios of MSC to live bacteria used in this study were similar to those reported for a 6-hour co-incubation of human MSC with planktonic S. aureus or E. coli in a transwell plate system [48] and were chosen to maximize the MSC dose while maintaining MSC survival during co-culture. The concentration of amikacin used in our study was chosen as the minimum concentration to give a C_max:MIC ratio of 8:1 for our S. aureus and E. coli strains (MIC₉₀ ≤ 4 μg/mL) [21,99–101], as has been historically described to maximize the antibiotic effect of amikacin. Amikacin is rapidly cytotoxic to equine bone marrow-derived MSC in a concentration dependent manner [102]. The 30 μg/mL dose selected reflected a concentration in the middle of the non-cytotoxic range for 24 hours of exposure and maintained MSC viability after 48 hours in our study conditions [103]. Concentrations of amikacin in synovial fluid vary widely following regional limb perfusion in horses [104]. The selected concentration is within the expected range of concentration and fold-MIC for our isolates [21,62,69,104]. However, because the presence of a biofilm matrix increases the MIC of antimicrobials 100–2,000 times [12,15] compared to planktonic bacteria, our amikacin concentration may have been insufficient in the face of established biofilms. Further elucidation of synergy between amikacin and MSC will require testing varying doses of amikacin and MSC.

Amikacin was preferentially selected over other aminoglycosides or antibiotic classes as it is commonly used during regional limb perfusion in horses [104] and 78.7% of veterinarians who use antimicrobials during intra-articular injection use amikacin [105]. Adaptive resistance to aminoglycoside antimicrobials in vitro can develop within 24 hours of constant antimicrobial exposure [106,107] and could have limited live bacterial count reductions in groups treated with amikacin and our ability to detect synergy between MSC and amikacin. Antimicrobial peptides secreted by equine bone marrow-derived MSC inhibit the growth of bacteria commonly found in skin wounds [49,51] but their effect on antimicrobial resistant microbes remains unknown. Antimicrobial resistant infections arise in clinical patients for reasons including chronic infections containing mature biofilms [6,108], following prolonged antimicrobial pre-treatment in which adaptive resistance develops [27], or polymicrobial infections [22,28,109]. Areas that require future investigation include the ability of MSC to disrupt biofilms with varying degrees of potential treatment resistance, testing a range of amikacin concentrations above fold-MIC and using other antimicrobial classes in conjunction with MSC.

Conclusions

Our study documents MSC-mediated reductions in biomass for S. aureus biofilms and biofilm area for S. aureus and E. coli biofilms. The transwell co-culture model facilitated culture of MSC and biofilms within the same fluid space while allowing physical separation to enable downstream assays for established biofilms. We demonstrated that the primary effect of MSC for established biofilms of S. aureus and E. coli was reduction of biofilm matrix. The dose and conditions of MSC tested did not show an appreciable effect on live bacterial counts of either S. aureus or E. coli. The lower-than-expected live bacterial reduction by MSC does not correlate with biofilm matrix reduction and may reflect limitations in quantification techniques, culture conditions, or dispersal of live bacteria that were not subsequently killed by MSC. MSC affected E. coli biofilms to a lesser extent than S. aureus biofilms, which warrants further investigation into differences in resistance of E. coli biofilm matrix to MSC-mediated disruption, or in stimulation of an MSC anti-biofilm response. Our results support further investigation of the mechanisms by which MSC disrupt biofilm matrix and highlights the importance of targeting specific bacterial strains for optimization of MSC anti-biofilm therapy.

Supporting information

Acknowledgments

Funding Statement

This work was supported by the Morris Animal Foundation Training Fellowship D21EQ-401. The URL for Morris Animal Foundation is https://www.morrisanimalfoundation.org/. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

References