Enterococcus faecalis exploits the human fibrinolytic system to drive excess collagenolysis: implications in gut healing and identification of druggable targets

Abstract

Perforations, anastomotic leak, and subsequent intra-abdominal sepsis are among the most common and feared complications of invasive interventions in the colon and remaining intestinal tract. During physiological healing, tissue protease activity is finely orchestrated to maintain the strength and integrity of the submucosa collagen layer in the wound. We (Shogan, BD et al. Sci Trans Med 7: 286ra68, 2015.) have previously demonstrated in both mice and humans that the commensal microbe Enterococcus faecalis selectively colonizes wounded colonic tissues and disrupts the healing process by amplifying collagenolytic matrix-metalloprotease activity toward excessive degradation. Here, we demonstrate for the first time, to our knowledge, a novel collagenolytic virulence mechanism by which E. faecalis is able to bind and locally activate the human fibrinolytic protease plasminogen (PLG), a protein present in high concentrations in healing colonic tissue. E. faecalis-mediated PLG activation leads to supraphysiological collagen degradation; in this study, we demonstrate this concept both in vitro and in vivo. This pathoadaptive response can be mitigated with the PLG inhibitor tranexamic acid (TXA) in a fashion that prevents clinically significant complications in validated murine models of both E. faecalis- and Pseudomonas aeruginosa-mediated colonic perforation. TXA has a proven clinical safety record and is Food and Drug Administration approved for topical application in invasive procedures, albeit for the prevention of bleeding rather than infection. As such, the novel pharmacological effect described in this study may be translatable to clinical trials for the prevention of infectious complications in colonic healing.

NEW & NOTEWORTHY This paper presents a novel mechanism for virulence in a commensal gut microbe that exploits the human fibrinolytic system and its principle protease, plasminogen. This mechanism is targetable by safe and effective nonantibiotic small molecules for the prevention of infectious complications in the healing gut.

INTRODUCTION

Millions of invasive procedures ranging from mucosal biopsies to full-thickness resections of the colon are performed each year by interventional gastroenterologists and surgeons (25, 36). Healing complications that lead to perforation and intra-abdominal sepsis remain a real and present danger to patients (17). The unique challenge of intestinal healing, in contrast to other epithelial surfaces, is the high bioburden of potentially pathogenic bacteria that can disorder essential physiological processes (24).

Critical among the processes that govern intestinal healing is the balanced deposition and remodeling of collagen in the submucosa (29). Plasminogen (PLG), the principal protease of the fibrinolytic system once thought to act mainly in the vasculature, is central to extravascular collagen remodeling in the skin (33), bone (11), and liver (27). Its active form, plasmin, both directly cleaves collagen and activates collagen-degrading matrix-metalloproteases (9). Plasmin activity is thus tightly regulated to prevent excessive degradation of the extracellular matrix components (12). In addition, genetic overactivation of PLG has been demonstrated to cause excessive collagen degradation and postoperative perforations in mice (38).

Intriguingly, certain pathogenic bacteria have evolved mechanisms to exploit the host PLG system for tissue invasion and destruction (19). Unlike physiological host-mediated PLG activation, bacteria-mediated PLG activation leads to supraphysiological activity, generally via evasion of host regulatory mechanisms (6). Although bacteria-mediated PLG activation is well described in some species, it has not been previously described in the commensal microbe Enterococcus faecalis, which is known to rapidly expand its colonic population following antibiotic exposure, selectively colonize wounded intestinal tissues, and shift to a virulent, collagenolytic phenotype via expression of the proteases gelatinase E (GelE) and serine protease (SprE) (31, 35). The ability of the colonic immune system and commensal microbes to eliminate resistant E. faecalis is impaired after gross defaunation of the gut with broad spectrum antibiotics (8, 13). Intraperitoneal Enterococcal infections have been modeled extensively; however, the molecular mechanisms underlying such infections are incompletely understood (14). To our knowledge, direct activation of PLG by E. faecalis has not been previously described. PLG activation in Enterococcal infection has been studied indirectly in a model of valvular endocarditis, in which PLG activation by immune cells was induced indirectly by the presence of E. faecalis (4).

E. faecalis and Pseudomonas aeruginosa are the pathogens most commonly isolated from postsurgical colonic perforations in humans and can cause phenotypically similar perforations in mice (16, 18, 34). Previous work in our laboratory demonstrated that E. faecalis disrupt intestinal healing in rodents via excessive degradation of types I and IV collagen in the colonic submucosa (34). While activation of human PLG is a well-described means of virulence in P. aeruginosa, interactions between E. faecalis and the PLG system are as yet undescribed. As colonic plasmin has collagenolytic action similar to matrix metalloproteases, we hypothesized that E. faecalis is capable of utilizing human PLG for collagen degradation. We demonstrate for the first time, to our knowledge, a mechanism for PLG activation by E. faecalis. We elucidated the degree to which this process enhances collagen degradation in vitro. In addition, we demonstrate that this process is both clinically significant and targetable using murine models of both E. faecalis and P. aeruginosa-mediated postsurgical perforation. Finally, we demonstrated that the PLG inhibitor tranexamic acid (TXA) can act as a nonantibiotic anti-infective agent when applied locally to a colonic wound by preventing microbial PLG binding, activation, and subsequent intestinal wound perforation.

METHODS

Plasminogen/urokinase-type plasminogen activator activity.

Plasmin activity assays were performed as previously described, with minor modifications (3). E. faecalis were grown overnight in tryptone yeast media. Optical density at 600 nanometer wavelength was normalized by dilution to 0.1, and samples were diluted 10 times in the final reaction. Bacteria were incubated with 250 nm human glu-plasminogen (Haematologic Technologies) with or without 10 mm TXA (Fisher) for 2 h at 37°C. A final concentration of 4 nm urokinase-type plasminogen activator (uPA) or pro-uPA (Biovision) were added and incubation proceeded for 20 min. A final concentration of 6 μM plasmin-specific fluorogenic substrate (H-D-Val-Leu-Lys-AFC, AnaSpec) was added immediately before a 30-min kinetic fluorescent read at 380/500 nm. For assays of uPA activity, PLG was omitted and a substrate specific to uPA (Z-Gly-Gly-Arg-AMC, Bachem) was utilized. When assays were performed in human plasma, fluorescence was read for 120 min total. Plasmin or uPA activity is expressed as initial reaction velocity calculated from change in fluorescence over time during the initial phase of the reaction, when pseudo-first order kinetics determine the rate.

Collagen degradation.

Assays were performed per the manufacturer’s (ThermoFisher) instructions as previously described, with slight modifications (1). Bacteria were incubated with fluorescently labeled type I or type IV collagen, with human PLG at 250 nm and TXA at 10 mm. Total reaction volume was 200 μL. The incubation proceeded for 5 h to allow for bacterial attachment to collagen and PLG binding. uPA was added and the incubation proceeded for an additional hour. Change in fluorescence over time at 480/520 nm was determined kinetically over the initial 30 min of the reaction.

Flow cytometric evaluation of PLG binding.

PLG binding to the bacterial surface was measured as previously described, with minor modifications (7). Bacteria were diluted to a final density of 8 × 10⁶ colony forming units/mL. These cells were incubated with 250 nm FITC-labeled PLG (Oxford). Cells were pelleted, washed three times with PBS, and resuspended. Fluorescence was analyzed using an Imagestream ISX flow cytometer. Enterococci were detected using log-forward and log-side scatter, and gating was set to exclude debris and aggregates of bacteria. FITC-range fluorescent signal was confirmed by light and fluorescent microscopy of each event.

Mouse model of E. faecalis-induced perforation.

This study utilized our validated murine model of E. faecalis-mediated colonic perforation (16). C57BL/6 mice (12 wk old, Charles River) were provided standard chow and tap water ad libidum. Mice received oral clindamycin (100 mg/kg gavage) and subcutaneous cefoxitin (40 mg/kg) the day before and the day of surgery. Mice underwent general anesthesia with ketamine and xylocaine and underwent laparotomy, followed by transection of the colon and a sewn anastomosis utilizing seven interrupted polypropylene sutures. A leak test was performed before the conclusion of the operation to rule out stricture or mechanically inadequate anastomoses. Rectal enemas containing E. faecalis at an optical density of 0.1 (representing 8 × 10⁷ colony forming units/mL) in 10% glycerol were administered in 100 μL doses on postoperative days 1, 2, and 3. Rectal enemas of P. aeruginosa at the same concentration were delivered only on postoperative day 1. All mice received separate 100 μL enemas containing 50 mm TXA (corresponding to 0.03 mg/kg) or vehicle control on postoperative days 1, 2, and 3. The dose of 0.03 mg/kg was selected based on prior human and murine studies of topical application (5). Both the surgeon and the investigator performing analysis of healing were blind to the treatment group. All experiments utilizing animal models were approved by the Institutional Animal Care and Use Committee at the University of Chicago.

RESULTS

PLG and uPA enhance collagen degradation by E. faecalis.

Collagenolytic E. faecalis strain V583 were incubated with labeled type I or type IV collagen in a purified system complemented with human PLG, uPA, and pro-matrix metallopeptidase 9 (MMP9) at the concentrations present in colonic tissue. The presence of V583 increased type I and IV collagen degradation after a 6-h incubation. Pro-MMP9 alone did not significantly increase bacterial collagen degradation. The presence of PLG with uPA increased degradation of both types I and IV collagen threefold. Collagen degradation in the presence of E. faecalis, PLG, and uPA was significantly greater than the additive activities of sterile PLG with uPA and that of E. faecalis alone. Additive collagen degradation between PLG and pro-MMP9 activation was observed in the case of type IV but not type I collagen (Fig. 1, A and B).

Fig. 1.

Enterococcus faecalis-mediated plasmin activation enhances collagenolysis and is inhibited by tranexamic acid (TXA) to a degree that prevents clinical perforation in mice. A: type 1 collagen degradation assay demonstrating significantly enhanced degradation by E. faecalis with the addition of plasminogen (PLG) and urokinase-type plasminogen activator (uPA) [*P < 0.01, analysis of covariance (ANCOVA)]. B: type IV collagen degradation assay. Again, E. faecalis alone demonstrated collagenolytic activity, which was enhanced in the presence of PLG with uPA. In this case, the addition of pro-matrix metallopeptidase 9 (MMP9) induced further degradation in the presence of PLG, uPA, and E. faecalis (*P < 0.01, ANCOVA). Error bars indicate means ± SD. All data are representative of three separate experiments, each was run in triplicate. C: in models of E. faecalis- and Pseudomonas aeruginosa-induced perforation following colonic surgery, TXA enema resulted in a complete rescue of the perforated phenotype observed in 60% of vehicle-treated mice in the case of E. faecalis (n = 10 mice per group, *P < 0.05, Fisher’s exact test). In P. aeruginosa-inoculated mice, 80% of vehicle-treated mice perforated, whereas 0 mice receiving TXA did so (n = 5 mice per group, *P < 0.05, Fisher’s exact test, error bars represent SE). D and E: trichrome stains of murine colonic anastomoses on postoperative day 8 following performance of an E. faecalis-induced perforation model. Mouse D and its cohort received a postoperative TXA enema, did not suffer perforation, and demonstrated organized trichrome staining in the submucosa. Mouse E and its cohort received a vehicle control enema and suffered perforation, demonstrating diffuse, disorganized trichrome positivity and significant inflammatory infiltrate. F and G: no discernible difference in healing or inflammatory infiltrate is evident on hematoxylin and eosin stains of postsurgery, nonleaking mice without contamination with E. faecalis that received postoperative TXA vs. vehicle enema. Yellow arrows indicate the anastomotic suture line. Blue circular structures are sutures. Images are representative of 3 mice per group. RFU, relative fluorescence unit.

PLG activation inhibitor TXA rescues supraphysiological collagen degradation and colonic perforation in postoperative mice.

Twelve-week-old mice underwent a validated model of surgical injury and E. faecalis or P. aeruginosa-induced perforation, followed by a postoperative enema with TXA or vehicle control. In vehicle-treated mice, 60% suffered a clinically significant perforation, indicated by abscess or frank feculent peritonitis on necropsy. In contrast, none of the 10 mice that received TXA enema suffered a clinically apparent perforation. A complete rescue of P. aeruginosa-induced perforation was observed as well in repeat experiments (Fig. 1C, n = 5 mice/group, P < 0.05, Fisher’s exact test). In the same model, standard histological analyses were performed on postoperative day 8 surgical tissue in mice that received TXA versus vehicle enemas (n = 3 mice/group). Trichrome staining demonstrated increased quantity and organization of fibroblastic products in the submucosa of mice that received TXA (Fig. 1D) compared with those that received vehicle and suffered perforations (Fig. 1E). Hematoxylin and eosin staining revealed no discernible difference in mucosal healing or inflammatory infiltrate to the anastomotic tissue of nonperforated mice that received either TXA or vehicle control (Fig. 1, F and G).

E. faecalis throughout the murine colon bind and activate human PLG in a manner inhibitable by TXA.

We next investigated the presence of E. faecalis capable of activating PLG in the mouse colon at the site of surgical injury and remote uninjured cecal tissue. E. faecalis cultured from throughout the colon demonstrated the ability to activate plasmin, which was in all cases inhibited by 10 mm TXA (Fig. 2A). After colorectal anastomotic surgery and introduction of E. faecalis to the surgical site via enema, mice (n = 3 mice per group) again received enemas of TXA or vehicle control on postoperative days 1, 2, and 3. FITC-labeled PLG was administered systemically before euthanasia. Fluorescent microscopy for PLG and E. faecalis DNA demonstrated colocalization of E. faecalis and PLG at the surgical site in control mice and inhibition of this process, along with diminished mucosal penetrance of E. faecalis, in TXA-treated mice (Fig. 2B).

Fig. 2.

Tranexamic acid (TXA) inhibits plasminogen (PLG) activation by Enterococcus faecalis ex vivo and prevents colocalization in vivo. A: E. faecalis collected from sites along the murine colon demonstrated the capacity to activate PLG, and this was inhibited by 10 mm TXA in all cases (*P < 0.05, analysis of covariance). Error bars indicate means ± SD. All data are representative of three separate experiments, each was run in triplicate. B: mice undergoing E. faecalis-contaminated anastomoses were administered FITC-labeled PLG (green) systemically. Anastomotic tissues are shown at ×40 magnification after staining for E. faecalis DNA (red) and colonocyte nuclei (blue). Top row: mice treated with vehicle enema had significant penetration of E. faecalis (red spheroids) into the colonic mucosa along with increased PLG at the surgical site. Merged images demonstrate colocalization of E. faecalis and FITC-PLG (yellow spheroids) at the surgical site. Bottom row: mice treated with TXA enema demonstrated diminished penetrance of E. faecalis into the mucosa and no colocalization of E. faecalis with FITC-PLG at the anastomotic site. Images are representative of one experiment with three animals per treatment group. RFU, relative fluorescence unit.

E. faecalis bind PLG and enhance its activation with urokinase.

We tested the ability of V583 to bind FITC-labeled PLG using flow cytometry. Negative controls incubated with unlabeled PLG demonstrated negligible autofluorescence (Fig. 3A). When V583 was incubated with FITC-PLG, a right shift in fluorescence was observed (Fig. 3B). Concomitant single-cell microscopy confirmed colocalization of the FITC-range fluorescent signal with light microscopic images of V583 (Fig. 3C).

Fig. 3.

Enterococcus faecalis bind plasminogen (PLG) and enhance urokinase-based activation via secreted gelatinase E (GelE) and serine protease (SprE). A and B: incubation with FITC-labeled PLG induced a right shift in fluorescence, indicating surface binding of PLG by E. faecalis (P < 0.05 vs. A. K-S analysis). C: single cell microscopy performed concomitantly with A and B, illustrating colocalization of the FITC-PLG fluorescent signal with E. faecalis. Scale bars = 10 μm. D: plasmin activity assay indicating that PLG activation was enhanced in the presence of E. faecalis and that active urokinase is required for detectable activity [*P < 0.05, analysis of covariance (ANCOVA)]. E: in solution with static amounts of urokinase and PLG, plasmin activity was directly related to the concentration of E. faecalis. F: E. faecalis increased PLG activation in human plasma in the presence of a full complement of its activators and inhibitors (*P < 0.05, ANCOVA). G: antibodies directed at the PLG receptor α-enolase diminished PLG activation by E. faecalis (*P < 0.05, ANCOVA). H: E. faecalis activate the PLG activator pro-urokinase in a fashion dependent on the virulence factors GelE and SprE. I: in the presence of E. faecalis and pro-urokinase, downstream PLG activation depends on GelE and SprE (*P < 0.05 for deficient mutants vs. parent strain; **P < 0.05 for complemented mutants vs. deficient strain, ANCOVA). J: plasmin activity assay in acellular conditioned media demonstrating that secreted GelE and SprE enhance PLG activation in the presence of pro-urokinase (*P < 0.05, ANCOVA). K: activation of urokinase by E. faecalis is fully inhibited by the physiological regulator plasminogen activator inhibitor-1 (PAI-1, *P < 0.05, ANCOVA). Error bars indicate 95% confidence intervals. All data are representative of three separate experiments, each performed in triplicate. CFU, colony forming unit; ENOL, α-enolase; RFU, relative fluorescence unit; uPA, urokinase-type plasminogen; EF, E. faecalis.

We next tested the ability of V583 to activate PLG using fluorogenic kinetic assays. Plasmin activity was observed in the presence of the tissue-based PLG activator urokinase and was enhanced by the presence of low concentrations of V583 (Fig. 3D). Total plasmin activity had a direct relationship with the concentration of bacteria present (Fig. 3E). To determine whether E. faecalis induces plasmin activity in the presence of a full complement of native PLG activators and inhibitors, we performed plasma-based plasmin generation assays. Low concentrations of V583 in human platelet-poor plasma induced significantly more plasmin activity than controls did (Fig. 3F).

Prior work by others demonstrated binding of PLG to polylysine motifs at the COOH-terminus of surface-exposed α-enolase on bacterial and eukaryotic cells (22). We therefore attempted to inhibit PLG activation using rabbit anti-human enolase IgG specific to the COOH-terminus or rabbit IgG raised against a 16 amino acid peptide from the COOH-terminus of V583. Anti-human enolase significantly decreased plasmin activity compared with isotype controls, and antienterococcal enolase decreased plasmin activity compared with the absence of antibody but did not reach a significant difference from isotype controls (Fig. 3G).

E. faecalis activate pro-urokinase via secreted gelatinase E and serine protease.

The ability of E. faecalis to disrupt colonic healing depends on expression of the proteolytic virulence factors GelE and SprE (34). The PLG activator uPA is secreted by inflammatory cells as a zymogen, pro-uPA, and activated by local proteases. We incubated the parent strain V583 and its isogenic mutants deficient in GelE and SprE with human pro-uPA. Incubation with V583 increased uPA activity in a fluorogenic assay. This activity was diminished in strains that did not express GelE (ΔgelE), SprE (ΔsprE), or both (ΔΔgelEsprE). The ability to activate pro-uPA was restored when these factors were reintroduced to deficient strains in plasmid-based complemented mutants (ΔgelE/gelE, ΔsprE/sprE, and ΔΔgelEsprE/gelEsprE, Fig. 3H). Downstream PLG activation in the presence of pro-uPA depended on expression of both GelE and SprE (Fig. 3I). To confirm this effect was mediated by secreted factors, the results were replicated in conditioned media (CM) from both V583 and ΔΔgelEsprE. The secretome of ΔΔgelEsprE, compared with V583, had a reduced ability to activate PLG in the presence of pro-uPA (Fig. 3J). Activation of pro-uPA by V583 was fully inhibited by the human inhibitor plasminogen activator inhibitor 1 provided in equimolar quantities to uPA (Fig. 3K).

E. faecalis increase PLG activation and suPAR release by macrophages.

Macrophages are the most abundant inflammatory cells in the colon, and circulating monocytes aggregate at sites of injury, such as an anastomosis (15). The macrophage phenotype, including regulation of protease activity and extracellular matrix degradation, can change alongside shifts in the local bacteria population, as is known to occur when the intestinal mucosa is injured. Therefore, we investigated the impact of E. faecalis on PLG activation in murine monocyte/macrophage cell lines.

CM from RAW 267.4 monocyte/macrophages cocultured with V583 induced more plasmin activity than CM from either cell line cultured separately (Fig. 4A). We next exposed RAW 267.4 macrophages to CM from V583 and observed significantly more plasmin activity in CM-treated cells than in controls exposed to sterile media (Fig. 4B). In coculture, live RAW 267.4 macrophages and V583 demonstrated increased activation of PLG at a relatively low multiplicity of infection (MOI) of 20; however, this effect was not observed at a lower MOI of 2 (Fig. 4C). As macrophages constitutively produce and activate pro-uPA, it was notable that when grown with V583 at an MOI of 20 and incubated with PLG, the elimination of exogenous uPA from the reaction did not decrease PLG activation (Fig. 4D).

Fig. 4.

Enterococcus faecalis induce plasminogen (PLG) activation and soluble urokinase-type plasminogen activator receptor (suPAR) release by macrophages. A: conditioned media (CM) from coculture of RAW 267.4 murine monocyte/macrophages with E. faecalis activates PLG to a greater extent than CM from either cell line cultured alone or their additive values [*P < 0.01, analysis of covariance (ANCOVA)]. B: bacterial CM added to the culture medium of live macrophages induced significantly more PLG activation than controls treated with sterile nonconditioned media (*P < 0.01, ANCOVA). C: macrophages cocultured with live E. faecalis activate significantly more PLG at a multiplicity of infection (MOI) of 20 times that of macrophages cultured alone (*P < 0.01, ANCOVA). D: incubated with E. faecalis at an MOI of 20, plasmin activation by RAW 267.4 macrophages was not significantly enhanced by supplemental urokinase-type plasminogen activator (uPA). E: incubation of RAW 267.4 cells with E. faecalis and PLG induced release of suPAR as measured by ELISA of CM (*P < 0.05, Student’s t test). F: incubation with antibodies directed at the COOH-terminal PLG binding site significantly diminished PLG activation by E. faecalis-stimulated macrophages (*P < 0.01, ANCOVA). G: schematic of a proposed mechanism for increased plasmin generation by macrophages after exposure to E. faecalis. This process is independent from and supplemental to activation of PLG on the bacterial surface, as described in Fig. 2. Error bars indicate 95% confidence intervals. All data are representative of three separate experiments, each was run in triplicate. GelE, gelatinase E; PLN, active plasmin; SprE, serine protease; RFU, relative fluorescence unit; EF, E. faecalis; ENOL-N, IgG raised against the NH₂-terminus of α-enolase; ENOL-C, IgG raised against the COOH-terminus of α-enolase.

The uPA receptor (uPAR) binds pro-uPA and facilitates cleavage to its active form and colocalization with PLG. It is expressed by stimulated macrophages and cleaved from the membrane as a soluble form (suPAR) by active plasmin (41). Thus, we performed ELISA for suPAR in the secretome from macrophages incubated overnight with PLG and live V583 and observed increased levels of suPAR antigen compared with sterile controls (Fig. 4E). α-enolase is a well-described PLG receptor on macrophages, binding PLG at a poly-lysine residue on its exposed COOH-terminus (37). Incubation with an antibody raised against the COOH-terminus of α-enolase diminished plasmin activation in V583-stimulated macrophages, in which control antibody specific to the NH₂-terminus of α-enolase had no effect (Fig. 4F).

DISCUSSION

Previous work in our laboratory demonstrated that E. faecalis disrupt intestinal healing in rodents via excessive degradation of types I and IV collagen in the colonic submucosa (34). We describe for the first time, to our knowledge, activation of human PLG by E. faecalis and demonstrate its dependence on the virulence factors GelE and SprE. As plasmin has direct collagenolytic action similar to matrix metalloproteases, we investigated a distinct virulence paradigm of E. faecalis-mediated collagen degradation dependent on PLG activation. In an intestinal wound, E. faecalis expands its population, colocalizes with native PLG, and upregulates expression of GelE and SprE, which are necessary to cause perforation and which we demonstrate are major contributors to PLG activation (28, 34). TXA prevented both PLG activation and healing complications induced by E. faecalis, which is associated with the failure of colonic healing in both rodents and humans. This is significant given that E. faecalis is difficult, if not impossible, to eradicate from the gastrointestinal tract with standard antibiotics used for prophylaxis (23, 39).

To our knowledge, microbial PLG activation has not been targeted to prevent intestinal healing complications. While activation of human PLG has been described in multiple bacterial species, to date it has not been described in E. faecalis. Although prior work suggested a role for E. faecalis in inflammatory-mediated PLG activity in the context of cardiac vegetations, noncollagenolytic strains were used, and the mechanism of direct activation of the fibrinolytic system was not investigated (4). The novel proposed mechanism of PLG activation by E. faecalis, its relationship to collagen degradation and wound disruption, and its mitigation by TXA is illustrated in Fig. 5. Activation is dependent on the binding of PLG to α-enolase on the surface of E. faecalis and likely other as yet unidentified receptors. PLG activation also depended on the presence of its tissue activator pro-urokinase and expression of the virulence factors GelE and SprE by E. faecalis. Findings from the present study and prior work suggest a bacterial MMP9-PLG integrated process in the pathobiology of failed colonic healing that can be effectively dampened by TXA, providing an exciting opportunity to potentially prevent anastomotic leak in response to multiple collagenolytic pathogens.

Fig. 5.

Molecular paradigm of pathogen-mediated anastomotic leak pathogenesis and pharmacological rescue with tranexamic acid (TXA). A: pathogens are a disordering agent to a cycle of protease activation normally regulated in a highly defined temporospatial context. Activation of human plasminogen (PLG) by Enterococcus faecalis is a previously unrecognized virulence mechanism that can cause supraphysiological degradation of collagen and failure of normal healing. Similar mechanisms have been described in other leak-causing pathogens, such as Pseudomonas aeruginosa. B: TXA applied locally at the anastomotic site temporarily and partially inhibits plasmin activation by various microbes, attenuating the loop of excessive PLG activation while allowing anastomotic healing to occur. ENOL, α-enolase; GelE: gelatinase E; MMP9, matrix metallopeptidase 9; PLGR, plasminogen receptor; PLN, active plasmin; SprE, serine protease; uPA, urokinase-type plasminogen activator.

Limitations of our study include the nondiscriminant inhibition of PLG activation by TXA; both host and bacterial processes are presumably affected. Further work is needed to quantitatively determine the full temporospatial course of total PLG activation and collagen degradation in healing intestinal wounds and to determine the minimum-necessary dosing of TXA enema to prevent perforation. The dose of TXA utilized in this study corresponds doses previously given to mice (26) and humans (2) via rectal enema. TXA is poorly absorbed in the rectum and acts locally, which aligns with the current findings. The bioavailability of rectal TXA at this dose was found to be well below the levels required for systemic effects (2). Prior studies of pharmacological sepsis mitigation in animals have not translated to clinically applicable treatment paradigms in humans (30). In one mouse model of bacteremia, systemic administration of TXA worsened the severity of already-present sepsis (32). However, the goal of local, rather than systemic, TXA therapy, as we describe it, is to prevent the inciting event (the perforation) that leads to systemic infection rather than mitigating an already-disseminated infection, which we believe accounts for the significant clinical improvement observed in our experiments. Although antimicrobial resistance is most often described in Enterococcus faecium, more recent papers utilizing nonculture-based microbial analysis demonstrate that E. faecalis is present in the colon in nearly all humans and is a common cause of postsurgical complications (10, 21, 34, 40). E faecalis is an extremely low-abundance commensal organism that, although present in all mammals, can be challenging to detect under baseline conditions given that it only proliferates under conditions of physiological stress and antibiotic use when the normal microbiota become disrupted (20). It is the most common organism to be cultured from sites of anastomotic leak, and, therefore, its clinical relevance in the current anastomotic leak model is highly translatable to the human condition (18, 21).

Based on the results of the current study, the efficacy of TXA as an anti-infective agent in the context of anastomotic leak pathogenesis can now be invoked. As this drug has a proven safety record and is Food and Drug Administration approved for topical use in elective and emergency procedures for the prevention of blood loss, the findings in this paper may be rapidly translatable to clinical trials investigating TXA as an anti-infective healing adjunct in the colon.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.A.J., K.W., H.v.G., A.Z., B.D.S., O.Z., and J.C.A. conceived and designed research; R.A.J., K.W., A.J.W., S.H., and A.Z. performed experiments; R.A.J., K.W., A.J.W., S.H., O.Z., and J.C.A. analyzed data; R.A.J., K.W., A.J.W., S.G., S.H., H.v.G., A.Z., B.D.S., O.Z., and J.C.A. interpreted results of experiments; R.A.J. and O.Z. prepared figures; R.A.J. drafted manuscript; R.A.J., K.W., B.D.S., O.Z., and J.C.A. edited and revised manuscript; R.A.J., K.W., A.J.W., S.G., S.H., H.v.G., A.Z., B.D.S., O.Z., and J.C.A. approved final version of manuscript.