Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 1.
Published in final edited form as: Ann Surg. 2021 Dec 1;274(6):e1038–e1046. doi: 10.1097/SLA.0000000000003733

Prevention of anastomotic leak via local application of tranexamic acid to target bacterial-mediated plasminogen activation: a practical solution to a complex problem.

Richard A Jacobson 1,2,3, Ashley Williamson 2, Kiedo Wienholts 4, Sara Gaines 2, Sanjiv Hyoju 2, Harry van Goor 4, Alexander Zaborin 2, Benjamin D Shogan 2, Olga Zaborina 2,*, John C Alverdy 2,*,#
PMCID: PMC7875208  NIHMSID: NIHMS1663691  PMID: 31851007

Abstract

Objective:

To investigate the role of bacterial- mediated plasminogen activation in the pathogenesis of anastomotic leak (AL) and its mitigation by tranexamic acid.

Summary background data:

AL is the most feared complication of colorectal resections. The pathobiology of AL in the setting of a technically optimal procedure involves excessive submucosal collagen degradation by resident microbes. We hypothesized that activation of the host plasminogen (PLG) system by pathogens is a central and targetable pathway in AL.

Methods:

We employed kinetic analysis of binding and activation of human PLG by microbes known to cause AL, and collagen degradation assays to test the impact of PLG on bacterial collagenolysis. Further, we measured the ability of the antifibrinolytic drug tranexamic acid (TXA) to inhibit this process. Finally, using mouse models of pathogen-induced AL, we locally applied TXA via enema and measured its ability to prevent a clinically relevant AL.

Results:

PLG is deposited rapidly and specifically at the site of colorectal anastomoses. TXA inhibited PLG activation and downstream collagenolysis by pathogens known to have a causal role in AL. TXA enema reduced collagenolytic bacteria counts and PLG deposition at anastomotic sites. Postoperative PLG inhibition with TXA enema prevented clinically and pathologically apparent pathogen-mediated AL in mice.

Conclusions:

Bacterial activation of host PLG is central to collagenolysis and pathogen-mediated AL. TXA inhibits this process both in vitro and in vivo. TXA enema represents a promising method to prevent AL in high risk sites such as the colorectal anastomoses.

Mini-abstract:

Pathogens that cause anastomotic leak utilize human plasminogen for virulent collagen degradation. Plasminogen is rapidly and persistently deposited in anastomotic tissue. Tranexamic acid can prevent these processes in vitro and rescues pathogen-mediated AL in vivo, and is a promising paradigm for prevention of AL in high risk anastomoses in humans.

INTRODUCTION

Despite improvements in operative technique and application of ever more powerful oral and parenteral antibiotics, anastomotic leaks (AL) remain a clear and present danger to patients especially in high risk sites such as the esophagus and colorectal area. For over 50 years it has been clearly established, both experimentally and clinically, that bacteria play a key and causative role in the pathogenesis of anastomotic leak. While oral antibiotics are now being widely applied to high risk anastomotic surgery in the gastrointestinal tract to address this complication, the recent finding that nearly 50% of pathogens causing serious postoperative infections are resistant to the antibiotics used for prophylaxis begs an alternative approach1.

Our lab has elucidated the molecular mechanisms by which commensal microbes that survive antibiotic decontamination protocols, such as Enteroccocus faecalis and Pseudomonas aeruginosa cause anastomotic leak. These two organisms are the most common pathobiota cultured from anastomotic leak sites in patients2,3. Our studies demonstrate that both E. faecalis and P. aeruginosa can colonize anastomotic tissues and become “in vivo expressed” by local “cues” at the healing site. These cues stimulate them to produce collagenolytic enzymes that pathologically amplify the process of collagen breakdown leading to loss of integrity of the anastomotic wound4,5. Further elucidation of the molecular details of this process has demonstrated that, in addition to its direct collagenolytic activity on the anastomotic wound, E. faecalis can convert matrix metalloprotease 9 (MMP9) from its proform to its active form resulting in an excess of tissue protease activity and, thus failure to heal the anastomotic wound2. Given that excess protease activity is the hallmark of abnormal healing across all wounds in the body, here we hypothesized that an additional tissue protease, plasminogen (PLG), might be also induced by pathogens that contribute to the pathobiology of AL.

We focused on PLG given that prior work has shown that genetic overactivation of the human fibrinolytic system leads to a non-healing anastomotic phenotype6. PLG, once thought to act mainly in the vasculature, is central to inflammatory cell migration and extravascular collagen remodeling in skin, liver and bone7–9. The impact of PLG on intestinal healing is incompletely understood, however it exists in colonic tissue at high (micromolar) concentrations and its active form, plasmin, both stimulates the MMP system and has direct collagenolytic activity10. PLG activation is a necessary component of tissue repair; its complete ablation causes severely impaired healing8. Yet under physiologic conditions, plasmin activity is finely orchestrated to prevent excessive degradation of extracellular matrix components11. Intriguingly, various pathogens have evolved mechanisms to exploit the host PLG system for tissue invasion12. However, unlike the fine-tuned activation that occurs in sterilely wounded tissues, when bacteria engage PLG, its activity is driven toward a supraphysiologic level known to interfere with healing13. PLG activation is well described in P. aeruginosa. A novel description of PLG binding and local activation by E. faecalis in a colonic anastomosis is detailed in a separately submitted manuscript (Jacobson et al 2019, under review).

Tranexamic acid (TXA) is an antifibrinolytic lysine analogue that prevents PLG from binding poly-lysine residues on fibrin clot and cellular receptors, where it would normally be cleaved and activated14. TXA has been administered systemically in surgical patients to mitigate blood loss15 and is FDA approved for this application in elective and emergency surgery. Therefore, in the present study we hypothesize that supraphysiologic activation of the host PLG system is a critical step in the pathogenesis of pathogen-mediated AL and that this pathoadaptive process could be mitigated by application of TXA at the anastomotic site. Therefore the aims of this study were to determine the role of topical TXA administration to a surgically created anastomotic wound as a method to prevent bacterial mediated anastomotic leak in a well-established mouse model involving the two most common pathogens that survive antibiotic decontamination in humans, E. faecalis and P. aeruginosa.

METHODS

Bacterial strains

Commercially available collagenolytic E. faecalis strain V583 (ATCC 700802) was utilized in all experiments. Collagenolytic P. aeruginosa strain P2 was isolated from the colon of a rat following anastomotic surgery.

Collagen degradation assays

Assays were performed per the manufacturer’s (ThermoFisher) instructions as previously described with slight modifications16. Bacteria were incubated with fluorescently labeled type I or type IV collagen, human PLG at 250nM and TXA at 10mM. Total reaction volume was 200μL. The incubation proceeded for 5 hours to allow for bacterial attachment to collagen and PLG binding. uPA at 4nM was added and the incubation proceeded for an additional hour. Change in fluorescence over time at 480/520nm was determined kinetically over the initial 30 minutes of the reaction.

Flow cytometric evaluation of bacterial PLG binding

PLG binding to the bacterial surface was measured as previously described, with minor modifications17. Bacteria were diluted to a final density of 8×106 CFU/mL. These cells were incubated with 250nM 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.

Plasmin/uPA activity

Plasmin activity assays were performed as previously described with minor modifications18. E. faecalis were grown overnight in tryptone yeast media. OD600 was normalized by dilution to 0.1, and samples were diluted 10x in the final reaction. Bacteria were incubated with 250nM human glu-plasminogen (Haematologic technologies) with or without 10mM TXA (Fisher) for two hours at 37°C. A final concentration of 4nM uPA or pro-uPA (Biovision) were added and incubation proceeded for 20 minutes. A final concentration of 6μM plasmin-specific fluorogenic substrate (H-D-Val-Leu-Lys-AFC, AnaSpec) was added immediately prior to a 30-minute kinetic fluorescent read at 380/500nm. 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 minutes 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.

Murine model of pathogen-induced AL

Pathogen-mediated anastomotic leak was performed as previously validated4. All mice received clindamycin (100mg/kg oral gavage) and cefoxitin (40mg/kg subcutaneous injection) the day prior to and the day of surgery. Mice underwent general anesthesia with intraperitoneal ketamine and xylocaine and laparotomy with transection of the colo-rectal junction followed by a primary anastomosis created with interrupted 8–0 prolene suture. A leak check was performed prior to abdominal closure with 4–0 vicryl suture in two layers.

Postoperatively, bacterial suspensions were introduced via enema to contaminate the anastomosis. E. faecalis suspensions were delivered on postoperative day (POD) 1,2 and 3, while P. aeruginosa suspensions were delivered on POD 1 only. All mice received separate 100μL enemas containing 50mM TXA (corresponding to 0.03mg/kg) or vehicle control (sterile deionized water) on postoperative days 1, 2 and 3. The dose of 0.03mg/kg was selected based on prior human and murine studies of topical application19. Both the surgeon and the investigator performing analysis of healing were blind to the treatment group.

Deposition of PLG at the anastomotic site

Mice underwent anastomotic surgery as described above followed by TXA or vehicle enema. Mice received a intraperitoneal injection of 100μL 4μM FITC-labeled PLG one hour prior to sacrifice. Anastomotic tissue was collected in Optimal Cutting Temperature compound (Tissue-Tek), immediately frozen and cryosectioned on-edge. Slides were stained with 4’,6-diamidino-2-phenylindole (DAPI) for colonic mucosa and imaged.

PLG binding by injured colon tissue

Anastomotic tissue was collected and incised longitudinally to create a sheet of tissue. Each sheet was incubated with 2μM FITC-labeled PLG (Oxford Biomedical Research) in PBS for an hour and gross tissue was used create light microscopic images with FITC-range fluorescent overlay.

Imaging

Confocal microscopy was performed on a Leica SP5 II AOBS tandem scanner spectral confocal system on a DMI6000 microscope and controlled by LASAF software (version 2.8.3). Four channels were collected at each location using sequential excitation (excitation: 405, 488, and 633; emission: 412–452, 495–537, and 654–755 nm pass bands) on either photomultiplier or HyD hybrid detectors. Objectives used were ×20, NA 0.7 dry, ×10, NA 0.7 dry, and ×40, NA 1.40 oil (Leica).

Quantitative microbiologic analysis

Anastomotic tissue and luminal contents were collected on necropsy in a sterile fashion and placed in 10% glycerol. Each sample was weighed and processed in a bead homogenizer. Samples underwent serial dilutions; 50μL of each were spread onto skim milk agar plates with enterococcal selective media and allowed to grow for 48 hours at 37°C. Collagenolytic colonies were identified by clearing of skim milk and counted by hand to calculate CFU/mL normalized to sample weight.

Statistical analyses

Statistical analyses were performed using Graphpad Prism 8 software. Unpaired Student’s t-tests were used for comparisons between two means for continuous variables. ANCOVA analysis was applied to compare slopes of linear regression lines in enzyme activity assays. Cellular populations were determined different in flow cytometric experiments through calculation of geometric mean fluorescence intensity and automated Kolmogorov-Smirnov analysis on FlowJo software. P<0.05 was considered significant for differences between mean values.

RESULTS

Plasminogen binds specifically to anastomotic tissue in a manner inhibited by TXA enema

PLG is a secreted protease readily bound by fibrin clot, inflammatory cells or pathogens in the vasculature and colonic submucosa; it circulates at high (micromolar) concentrations and is ubiquitous in tissue20,21. The role of PLG in gastrointestinal tissue repair, including its temporospatial distribution at sites of injury, is largely unknown. To confirm that PLG is specifically bound surgically injured tissue, we measured binding of intraperitoneally injected FITC-labeled PLG early in the postoperative course, and the feasibility of attenuating such deposition with postoperative TXA enema.

We first assessed whether anastomotic tissue has the capacity to bind PLG, and incubated fresh postoperative tissue ex vivo with FITC-PLG. Tissue collected remote from the anastomosis demonstrated no specific binding of FITC-PLG (Fig 1A). Specific binding of FITC-PLG to anastomotic tissue was observed as early as 10 minutes post-surgery (Fig 1B), and qualitatively increased in tissue collected 48 hours post-surgery (Fig 1C). At both timepoints, binding was centered on the anastomotic suture line.

Figure 1: PLG is bound rapidly and specifically by wounded colonic tissue in a fashion inhibited by TXA enema.

Figure 1:

Open in a new tab

Three mice per timepoint underwent surgical colorectal anastomosis and demonstrated no healing complications at the time of sacrifice. Representative images from each group are shown, yellow arrows indicate the anastomotic suture line A-C) Colon tissue incubated ex vivo in PBS with FITC-labeled PLG (green). Images are of gross tissue at 40x magnification, Scale bars = 100μm. A) Nonwounded tissue demonstrates minimal FITC-PLG binding. B) Tissue collected 10 minutes following injury demonstrates concentration of FITC-PLG at the surgical wound. C) FITC-PLG binding was qualitatively increased in tissue collected 48 hours post-injury and was spatially concentrated at the surgical site. D-J) Mice were administered intraperitoneal FITC- PLG an hour prior to sacrifice. Blue staining represents colonocyte nuclei. D) 20x magnification, scale corresponds to (E). 24 hours after surgery, a mucosal defect is seen with deposition of FITC-PLG at the surgical site. E) Spatial quantification of staining intensity in (D) indicates that FITC-PLG is bound specifically at the anastomotic site. F) FITC-PLG deposition at the anastomotic site 48 hours after surgery. G) Inhibition of PLG deposition by topical application of TXA via enema one hour prior to sacrifice. H,I) Redemonstration of PLG deposition and inhibition with topical TXA at 72 hours post-surgery. J) Quantitation of staining intensity in the groups represented in (F-I), (n=3 per group) demonstrates that TXA enema compared to vehicle control significantly decreased deposition of FITC-PLG at the surgical wound (*p<0.05, Student’s t-test). Error bars indicate 95% confidence intervals.

Next, FITC-PLG was administered intraperitoneally to postoperative mice, one hour prior to sacrifice. FITC-PLG was bound specifically at the anastomotic site at 24 hours postoperative as illustrated in Fig 1D and spatially quantified in Fig 1E. FITC-PLG deposition persisted at 48 hours postoperative (Fig 1F) and was mitigated by TXA enema (Fig 1G); a similar pattern was observed at 72 hours postoperative (Fig 1 H,I). Quantitative analysis demonstrated that TXA enema diminished deposition of FITC-PLG at the anastomotic site by roughly a third at both timepoints (Fig 1J).

TXA inhibits binding and activation of PLG by E. faecalis

Previous studies in mice and humans indicate that a 30% drop in plasminogen activity is both experimentally and clinically significant22,23. TXA prevents binding of PLG to cell surface receptors and binding sites on fibrin clot; we next investigated whether binding on the surface of E. faecalis could be inhibited with TXA. Flow cytometric analysis demonstrated that TXA significantly and concentration-dependently inhibited microbial surface binding of labeled PLG as quantified by geometric mean fluorescence intensity of 1768 (no TXA) vs 936 (2mM TXA) vs 77 (10mM TXA) with coefficients of variation 249, 252 and 363 respectively (Fig 2A). The ability of E. faecalis to activate PLG in the presence of its activator urokinase (uPA) was also significantly inhibited by TXA in a concentration-dependent manner. This was determined in vitro in a purified system. The highest dose of TXA used in these experiments, 10mM induced a 34% decrease in observed plasmin activity (Fig 2B).

Figure 2: TXA inhibits PLG-dependent collagen degradation by E. faecalis in vitro and rescues anastomotic leak in vivo.

Figure 2:

Open in a new tab

A) E. faecalis in the presence of 0 (red), 2mM (orange) and 10mM (blue) TXA and FITC-labeled PLG. Flow cytometry demonstrated that TXA concentration-dependently inhibited binding of PLG to E. faecalis (* p<0.05 compared to 0mM TXA, K-S analysis). B) TXA concentration-dependently and significantly, by over 30% at the maximum dose tested - 10mM, inhibits PLG activation by E. faecalis in the presence of uPA (* p<0.05, ANCOVA). C) TXA concentration-dependently inhibits type IV collagen degradation by E. faecalis in the presence of PLG and uPA (* p<0.05, ANCOVA). 10mM TXA almost completely eliminated bacterial plasmin-dependent collagen degradation D) Treatment with TXA had no impact on 6 hour population growth of E. faecalis. E) Treatment with TXA had no impact on collagen degradation by E. faecalis in the absence of PLG. H) In a model of E. faecalis-induced anastomotic leak, mice treated with TXA enema exhibited improved anastomotic healing scores compared to mice receiving vehicle (water) control (1.5 ± 0.5 vs 2.7 ± 0.6, *p<0.05 Student’s t-test). Each point represents one animal (n=10 per group). In this model, clinically relevant anastomotic leak is defined as AHS 3 or above. Thus, 60% of mice receiving control treatment suffered clinically relevant leak while 0% of TXA-treated mice demonstrated this outcome (p<0.01 Fisher’s exact test). Error bars indicate means ± SEM. All in vitro data are representative of three separate experiments, each run in triplicate.

TXA inhibits plasmin-dependent collagen degradation by E. faecalis

As active PLG has direct collagenolytic action, we next investigated whether TXA could inhibit PLG-dependent collagen degradation by E. faecalis in the presence of PLG and uPA. TXA inhibited collagen degradation by E. faecalis in a concentration-dependent fashion. 10mM TXA induced a 92% decrease in collagen degradation compared to control. (Fig 2C). To confirm whether this effect was truly secondary to plasmin inhibition, we measured the impact of TXA on both bacterial growth and intrinsic (plasmin-independent) collagen degradation. TXA added to culture medium had no impact on bacterial growth as measured by change in optical density over six hours (Fig 2D), nor intrinsic bacterial collagen degradation activity in the absence of PLG (Fig 2E).

TXA enema rescues E. faecalis-mediated anastomotic leak in mice.

Twelve-week-old C57BL/6 mice underwent an established model of E. faecalis-induced anastomotic leak4. On POD 1, 2 and 3 mice received a 100μL rectal enema containing 50mM TXA in sterile water or vehicle control (water alone), with 10 mice per group. We used a validated Anastomotic Healing Score (AHS) to quantify failure of healing (0 – pristine 1 – flimsy adhesions 2 – dense adhesions 3 – abscess formation 4 – gross anastomotic disruption). A score of 3 or 4 represents clinically relevant AL and correlates with sepsis in the mouse. On POD 8 mice receiving vehicle had a mean score of 2.7 ± 0.6 while mice treated with TXA had a mean score of 1.5 ± 0.5 (p<0.05 Student’s t-test). We next compared groups for the presence of a clinically relevant leak (i.e. score ≥3). By these criteria, 6/10 mice in the vehicle group developed AL, while none of the mice (0/10) in the group receiving TXA demonstrated AL (p<0.01 Fisher’s exact test). (Fig 2F).

TXA enema mitigates invasion of anastomotic tissue by E. faecalis

To assess the impact of TXA enema on the tissue penetrance of collagenolytic E. faecalis local to the anastomosis, we performed separate experiments using quantitative culture techniques. Mice underwent a model of E. faecalis-mediated anastomotic leak and treatment with TXA or vehicle (water) enema on postoperative days 1,2 and 3. Endpoint colony counts of collagenolytic E. faecalis were performed at sacrifice on postoperative days 3 and 8. At postoperative day 3, treatment with TXA did not alter the density (CFU/mg) of collagenolytic E. faecalis in anastomotic tissue (Fig 3A). However, by postoperative day 8 a significant difference in the density of collagenolytic E. faecalis was observed in the treatment group, possibly indicative of decreased tissue penetrance by bacteria or recovery of the normal microbiome in the setting of a healed anastomosis (Fig 3B). Mice treated with TXA enema had significantly diminished collagenolytic E. faecalis in anastomotic luminal contents (i.e. CFU/mg tissue) by POD3 (Fig 3C), and this trended toward but did not reach significance by POD8 (Fig 3D).

Figure 3: TXA enema reduces tissue penetrance by collagenolytic E. faecalis at the anastomotic site.

Figure 3:

Open in a new tab

Mice underwent a model of E. faecalis-mediated anastomotic leak and treatment with TXA or vehicle (water) enema. Endpoint colony counts of collagenolytic E. faecalis were performed at sacrifice on postoperative days 3 and 8. A) At postoperative day 3, treatment with TXA did not alter the concentration of collagenolytic E. faecalis in anastomotic tissue. B) By postoperative day 8 a significant difference in total CFU/mg of collagenolytic E. faecalis was observed in the TXA-treated group, possibly indicative of decreased tissue penetrance by bacteria or recovery of the normal microbiome in the setting of a healed anastomosis. C) Mice treated with TXA had significantly fewer CFU/mg collagenolytic E. faecalis in anastomotic luminal contents at postoperative day 3. D) Treatment group animals had persistently low CFU/mg, however this failed to reach statistical significance by postoperative day 8. Each data point represents 1 animal. * p<0.05 Student’s t-test. Error bars indicate 95% confidence intervals.

Rescue of pathogen-mediated AL with PLG inhibition extends to P. aeruginosa.

To generalize the rescue of AL via PLG inhibition to gram negative pathogens relevant to the human condition, reiterative experiments were performed in which P. aeruginosa was administered via enema on POD1 to contaminate the anastomosis as previously published4. In contrast to E. faecalis, activation of the host PLG system by P. aeruginosa is well-described. Prior work from our laboratory demonstrated that local phosphate abundance, at concentrations similar to the 10mM found in PBS, suppresses virulence expression in P. aeruginosa via its well described phosphosenory and phosphoregulatory system24. Enema treatments with TXA dissolved in sterile water, PBS, or sterile water as a vehicle controls were administered on POD 1, 2 and 3. Five mice per group were randomly assigned to TXA, PBS or water. The group receiving water demonstrated significantly impaired postoperative healing as measured by AHS, 3.8 ± 0.4, with 5/5 mice suffering a clinically obvious leak. No mice treated with TXA developed a clinical leak, and AHS was 1.6 ± 0.5 (p<0.01 vs water, one-way ANOVA). In the group that received PBS one mouse developed a clinical leak, and AHS was significantly lower than the water-treated group at 2.0 ± 1.2 (p<0.05 vs water, one-way ANOVA) but not significantly different than the TXA-treated group (p=0.73 vs TXA) (Fig 4A).

Figure 4: PLG inhibition is a generalizable means of preventing pathogen-mediated AL.

Figure 4:

Open in a new tab

A) In a model of P. aeruginosa-induced anastomotic leak, mice treated with either TXA (in water) or PBS (10mM phosphate, without TXA) exhibited significantly improved anastomotic healing compared to mice treated with deionized water vehicle control (1.6 +/− 0.6 (TXA) vs 2.0 +/− 1.2 (PBS) vs 3.8 +/−0.4 (water), p<0.05 Student’s t-test). Each data point represents one animal. B) Flow cytometry of P. aeruginosa cultured with added water (red) vs PBS (blue) demonstrating significantly decreased PLG binding in the presence of phosphate (*p<0.05 K-S analysis). C) P. aeruginosa cultured with added PBS activated significantly less PLG than the same strain with added deionized water. An even more significant decrease in PLG activity was observed when TXA in phosphate-free water was added to P. aeruginosa. (*p<0.05 ANCOVA). Error bars indicate means ± SD. All in vitro experiments were run in triplicate with representative data shown.

To investigate the mechanism of the rescue of P. aeruginosa-induced anastomotic leak by PBS compared to water, we examined the impact of phosphate on the ability of P. aeruginosa to activate PLG, given our previous work showing that local phosphate concentration dramatically suppresses P. aeruginosa virulence expression. These experiments were designed to control for potential effects of TXA enema besides inhibition of PLG activation, thus generalizing PLG inhibition as a means of decreasing bacterial virulence. P. aeruginosa exposed to PBS compared to water bound less FITC-PLG as measured by flow cytometry with a decrease in geometric mean fluorescence intensity from 1294 to 916 with coefficient of variation 185% and 189%, respectively (Fig 4B). Accordingly, the ability of P. aeruginosa to activate PLG was significantly attenuated in both bacteria grown in the presence of PBS and the same strain grown without PBS but treated with 10mM TXA, compared to either strain grown with added water as a vehicle control (Fig 4C). Phosphate did not diminish the ability of E. faecalis to activate PLG (data not shown).

PLG-activating microbes populate the human colon

We next examined whether human colon anastomotic tissue harbored bacteria capable of activating the PLG system. Isolates from cultures of previously published work were re-examined in this context2. Ten consecutive patients undergoing colon resection at the University of Chicago had the mucosal ends of the resected colon segment swabbed intraoperatively for microbiological analysis. Swabs were cultured and speciated. All patients had received mechanical bowel preparation (purgative cleansing and oral antibiotics) and intravenous cefoxitin preoperatively. Swabs from 2 patients were culture-negative and 8 were positive. Of the culture- positive patients, all 8 harbored bacterial species with well described activation of the PLG system - 33 of 64 total cultured strains were identified as PLG activators (Table 1).

Table 1: PLG activating and collagenolytic bacteria are prevalent in human anastomotic tissue.

Ten patients had the mucosal surface of resected colon specimens swabbed for microbiological analysis. Of the ten samples, two were culture negative. Of the remaining 8, all samples harbored species with well described PLG activation in their colonic mucosa.

Species Patient(s) PLG activation Collagenolytic
S. angiosus 1,4,8 X 37
S. salivarius 2,3,6,7,8 X 31
S. parasanguinis 3,4,7,8 X 31
S. gallolyticus 5 X 38
S. mitis 6 X 31
S. gordonii 6 X 31
S. intermedius 7* X 31
P. aeruginosa 7* X 39 X 2
E. faecalis 8 X *** X 4
S. marcescens 8 X 40 X 2
Open in a new tab
*

Of note, the one patient that harbored bacteria both with described PLG activating and collagenolytic activity (P. aeruginosa) developed an abscess requiring percutaneous drainage, presumed secondary to AL.

***

The mechanism of PLG activation by E. faecalis is described in a separately submitted manuscript, currently under review (Jacobson et al. 2019, under review).

DISCUSSION

Across all wounds, protease activity represents the hallmark of healing as a critical balance between collagen synthesis and breakdown, remodeling, strength and integrity of an injured or operated tissue. Since bacteria can disorder the process of healing, sterility, as much as is technically possible, is maintained during surgery with the goal of preventing complications. A wound in the gastrointestinal tract, particularly in the colon, represents a unique challenge, given the high density of pathogenic bacteria present before, during and after anastomotic surgery3,25. The current practice of intestinal antisepsis prior to colon surgery involves administration of both oral and intravenous antibiotics, targeting both the abundant health promoting microbiota and low abundance pathobiota26. This approach, the formulation of which has not significantly changed since its inception over 50 years ago, remains controversial, highly debated and poorly understood in terms of its precise mechanisms of action27–29. Indiscriminate elimination of the normal microbiota carries the unintended consequence of allowing resilient pathobiota to bloom and may account for the persistence of the incidence of AL following major surgery3,30.

The most common bacteria cultured from a leaking anastomosis are E. faecalis and P. aeruginosa, pathobionts that persist in the gastrointestinal tract even when powerful antibiotics are used31. Our laboratory has provided compelling evidence that collagenase-producing pathogens, including but not limited to E. faecalis, P. aeruginosa, and S. marcescens, play a causal role in AL in rodents via their ability to activate host proteases such as MMP9 in addition to their direct effect on wound collagen breakdown2,4. Data from the present study identify an additional and complementary mechanism by which collagenolytic bacteria contribute to the pathogenesis of AL - the PLG system, known to be present in all healing wounds, although poorly described in intestinal wounds. The fact that AL persists even in the face of current best practices, suggests a pressing need to develop alternative approaches to this devastating, disabling and occasionally lethal complication of anastomotic surgery. Use of TXA in combination with collagenase-suppressing agents such as phosphate active compounds (e.g. phosphorylated polyethylene glycol), could potentially decrease the incidence of AL and halt the practice of adding multiple antibiotics, which is not an evolutionarily stable strategy.

In the present study, pharmacologic targeting of PLG activation by two distinct pathogens with TXA successfully prevented AL, as might be clinically manifested in a patient (e.g. abscess formation or peritonitis). The additional finding that phosphate exposure decreased PLG activation by P. aeruginosa and rescued AL induced by that same strain supports the hypothesis that supraphysiologic PLG activation is central to pathogen-mediated AL. While phosphate-based suppression of PLG activity was not observed for both pathogens, TXA successfully ablated PLG activation in vitro and rescued AL in vivo in all experiments included in this study. Taken together these findings suggest that an advantage of PLG inhibition may be that it is pathogen-agnostic. That TXA treatment impacted plasmin-dependent collagen degradation but had no impact on plasmin-independent collagen degradation in vitro, and diminished PLG availability at the anastomosis in vivo suggests that local PLG inhibition is its specific mechanism of action in preventing AL. Further elucidation of the molecular events that underlie the TXA effect have been recently elucidated in a separate manuscript (Jacobson et al 2019, in press).

Given that multiple pathogens express collagenolytic enzymes that can potentially disrupt anastomotic healing, targeting PLG activation with TXA and phosphates may be a broadly applied method to prevent AL across a wide spectrum of pathogens and regions of the gastrointestinal tract. Local administration of TXA by enema will likely decrease the risk of thrombotic complications observed in clinical studies of systemic administration for prevention of blood loss32. It would, in theory, also decrease the risk of anastomotic bleeding. Administration of high dose TXA via the transrectal route appears to be both safe and efficacious. Its bioavailability through the colon and rectum is roughly 15% of the same dose given systemically, even in patients with active colonic inflammation33. The dosage of TXA used in this study was calibrated to previous animal and human studies, at which the risk for systemic toxicity was deemed negligible33. The in vitro decline in kinetic activity and in vivo rescue of the models represent clinically significant inhibition of PLG activation by both types of bacteria investigated. Yet another advantage of local TXA application is its temporal effect. PLG inhibition with alternative agents, such as α2-antiplasmin or PAI-1 is feasible, however these represent non-reversible inhibition of plasmin and uPA respectively, requiring de novo production of the proteases for the recovery of normal activity, whereas TXA temporarily prevents PLG binding and its effect ends after clearance from the surgical site in a matter of hours34,35. Limitations of our study include the non-discriminant inhibition of PLG activation by TXA – both host and bacterial processes are affected. Further work is needed to quantitatively determine the temporo-spatial course of total PLG activation and collagen degradation in both healing and leaking anastomoses. Based on this, the minimum-necessary dosing regimen of TXA enema to dampen these processes to physiologic levels can be estimated.

Further investigation is required to determine the source of PLG identified at the anastomotic site. PLG is secreted from the liver, circulates at high concentrations and is and ubiquitous in extracellular matrix36. While we illustrate the partial temporospatial course of PLG binding in a colonic anastomosis and its ablation with rectal TXA for the first time, our observation that PLG is specifically bound does not inform its immediate source, whether that be the vasculature or colonic submucosa. Taken together, we believe the findings enclosed in this study and our prior work represent a novel paradigm for the prevention of pathogen-mediated AL. Local application of TXA and polyphosphate compounds, either as a mucoadhesive gel or liquid enema could then be investigated as method to prevent AL in human trials of these low cost and FDA approved agents.

Acknowledgements:

The authors would like to thank Vytas Bindokas PhD for his assistance in producing microscopic images. They would also like to thank Jonathan Schoenecker MD, PhD for assistance in the conceptualization of the enclosed experiments. This article is dedicated to the memory of Sam Klonoski, MD.

Funding: The enclosed work was supported in part by NIH R01-GM062344-18 (JCA)

Footnotes

Data and materials availability: All materials are commercially available. All data associated with this study are available in the main text or the supplementary materials.

REFERENCES

  • 1.Teillant A, Gandra S, Barter D, Morgan DJ & Laxminarayan R Potential burden of antibiotic resistance on surgery and cancer chemotherapy antibiotic prophylaxis in the USA: a literature review and modelling study. Lancet Infect Dis 15, 1429–1437, doi: 10.1016/S1473-3099(15)00270-4 (2015). [DOI] [PubMed] [Google Scholar]
  • 2.Shogan BD et al. Collagen degradation and MMP9 activation by Enterococcus faecalis contribute to intestinal anastomotic leak. Sci Transl Med 7, 286ra268, doi: 10.1126/scitranslmed.3010658 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Shogan BD et al. Intestinal anastomotic injury alters spatially defined microbiome composition and function. Microbiome 2, 35, doi: 10.1186/2049-2618-2-35 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Hyoju SK et al. Oral Polyphosphate Suppresses Bacterial Collagenase Production and Prevents Anastomotic Leak Due to Serratia marcescens and Pseudomonas aeruginosa. Ann Surg, doi: 10.1097/SLA.0000000000002167 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Shakhsheer BA et al. Morphine Promotes Colonization of Anastomotic Tissues with Collagenase - Producing Enterococcus faecalis and Causes Leak. J Gastrointest Surg 20, 1744–1751, doi: 10.1007/s11605-016-3237-5 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.te Velde EA et al. Impaired healing of cutaneous wounds and colonic anastomoses in mice lacking thrombin-activatable fibrinolysis inhibitor. J Thromb Haemost 1, 2087–2096 (2003). [DOI] [PubMed] [Google Scholar]
  • 7.Yuasa M et al. Fibrinolysis is essential for fracture repair and prevention of heterotopic ossification. J Clin Invest 125, 3117–3131, doi: 10.1172/JCI80313 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Romer J et al. Impaired wound healing in mice with a disrupted plasminogen gene. Nat Med 2, 287–292 (1996). [DOI] [PubMed] [Google Scholar]
  • 9.Gong Y, Hart E, Shchurin A & Hoover-Plow J Inflammatory macrophage migration requires MMP-9 activation by plasminogen in mice. J Clin Invest 118, 3012–3024, doi: 10.1172/JCI32750 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Lijnen HR, Silence J, Lemmens G, Frederix L & Collen D Regulation of gelatinase activity in mice with targeted inactivation of components of the plasminogen/plasmin system. Thromb Haemost 79, 1171–1176 (1998). [PubMed] [Google Scholar]
  • 11.Arumugam S, Jang YC, Chen-Jensen C, Gibran NS & Isik FF Temporal activity of plasminogen activators and matrix metalloproteinases during cutaneous wound repair. Surgery 125, 587–593 (1999). [PubMed] [Google Scholar]
  • 12.Lahteenmaki K, Edelman S & Korhonen TK Bacterial metastasis: the host plasminogen system in bacterial invasion. Trends Microbiol 13, 79–85, doi: 10.1016/j.tim.2004.12.003 (2005). [DOI] [PubMed] [Google Scholar]
  • 13.Cederholm-Williams SA, De Cock F, Lijnen HR & Collen D Kinetics of the reactions between streptokinase, plasmin and alpha 2-antiplasmin. Eur J Biochem 100, 125–132 (1979). [DOI] [PubMed] [Google Scholar]
  • 14.Miles LA & Plow EF Binding and activation of plasminogen on the platelet surface. J Biol Chem 260, 4303–4311 (1985). [PubMed] [Google Scholar]
  • 15.collaborators C. t. et al. Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. Lancet 376, 23–32, doi: 10.1016/S0140-6736(10)60835-5 (2010). [DOI] [PubMed] [Google Scholar]
  • 16.Aimes RT & Quigley JP Matrix metalloproteinase-2 is an interstitial collagenase. Inhibitor-free enzyme catalyzes the cleavage of collagen fibrils and soluble native type I collagen generating the specific 3/4- and 1/4-length fragments. J Biol Chem 270, 5872–5876 (1995). [DOI] [PubMed] [Google Scholar]
  • 17.Bower K, Djordjevic SP, Andronicos NM & Ranson M Cell surface antigens of Mycoplasma species bovine group 7 bind to and activate plasminogen. Infect Immun 71, 4823–4827 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Ambadapadi S et al. Reactive Center Loop (RCL) Peptides Derived from Serpins Display Independent Coagulation and Immune Modulating Activities. J Biol Chem 291, 2874–2887, doi: 10.1074/jbc.M115.704841 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Ausen K et al. Serum Concentrations and Pharmacokinetics of Tranexamic Acid after Two Means of Topical Administration in Massive Weight Loss Skin-Reducing Surgery. Plast Reconstr Surg 143, 1169e–1178e, doi: 10.1097/PRS.0000000000005620 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Ploplis VA, French EL, Carmeliet P, Collen D & Plow EF Plasminogen deficiency differentially affects recruitment of inflammatory cell populations in mice. Blood 91, 2005–2009 (1998). [PubMed] [Google Scholar]
  • 21.Poller L Fibrinolysis and gastrointestinal haemorrhage. J Clin Pathol Suppl (R Coll Pathol) 14, 63–67 (1980). [PMC free article] [PubMed] [Google Scholar]
  • 22.Einersen PM et al. Rapid thrombelastography thresholds for goal-directed resuscitation of patients at risk for massive transfusion. J Trauma Acute Care Surg 82, 114–119, doi: 10.1097/TA.0000000000001270 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Mignemi NA et al. Plasmin Prevents Dystrophic Calcification After Muscle Injury. J Bone Miner Res 32, 294–308, doi: 10.1002/jbmr.2973 (2017). [DOI] [PubMed] [Google Scholar]
  • 24.Zaborin A et al. Red death in Caenorhabditis elegans caused by Pseudomonas aeruginosa PAO1. Proc Natl Acad Sci U S A 106, 6327–6332, doi: 10.1073/pnas.0813199106 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Weir TL et al. Stool microbiome and metabolome differences between colorectal cancer patients and healthy adults. PLoS One 8, e70803, doi: 10.1371/journal.pone.0070803 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Hayakawa M et al. Dramatic changes of the gut flora immediately after severe and sudden insults. Dig Dis Sci 56, 2361–2365, doi: 10.1007/s10620-011-1649-3 (2011). [DOI] [PubMed] [Google Scholar]
  • 27.Clarke JS et al. Preoperative oral antibiotics reduce septic complications of colon operations: results of prospective, randomized, double-blind clinical study. Ann Surg 186, 251–259 (1977). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Oliveira L et al. Mechanical bowel preparation for elective colorectal surgery. A prospective, randomized, surgeon-blinded trial comparing sodium phosphate and polyethylene glycol-based oral lavage solutions. Dis Colon Rectum 40, 585–591 (1997). [DOI] [PubMed] [Google Scholar]
  • 29.Atkinson SJ et al. In the Absence of a Mechanical Bowel Prep, Does the Addition of Pre-Operative Oral Antibiotics to Parental Antibiotics Decrease the Incidence of Surgical Site Infection after Elective Segmental Colectomy? Surg Infect (Larchmt) 16, 728–732, doi: 10.1089/sur.2014.215 (2015). [DOI] [PubMed] [Google Scholar]
  • 30.Ojima M et al. Metagenomic Analysis Reveals Dynamic Changes of Whole Gut Microbiota in the Acute Phase of Intensive Care Unit Patients. Dig Dis Sci 61, 1628–1634, doi: 10.1007/s10620-015-4011-3 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Ohigashi S et al. Significant changes in the intestinal environment after surgery in patients with colorectal cancer. J Gastrointest Surg 17, 1657–1664, doi: 10.1007/s11605-013-2270-x (2013). [DOI] [PubMed] [Google Scholar]
  • 32.Johnston LR, Rodriguez CJ, Elster EA & Bradley MJ Evaluation of Military Use of Tranexamic Acid and Associated Thromboembolic Events. JAMA Surg 153, 169–175, doi: 10.1001/jamasurg.2017.3821 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Almer S, Andersson T & Strom M Pharmacokinetics of tranexamic acid in patients with ulcerative colitis and in healthy volunteers after the single instillation of 2 g rectally. J Clin Pharmacol 32, 49–54, doi: 10.1002/j.1552-4604.1992.tb03787.x (1992). [DOI] [PubMed] [Google Scholar]
  • 34.Cubellis MV, Wun TC & Blasi F Receptor-mediated internalization and degradation of urokinase is caused by its specific inhibitor PAI-1. EMBO J 9, 1079–1085 (1990). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Wiman B & Collen D On the mechanism of the reaction between human alpha 2-antiplasmin and plasmin. J Biol Chem 254, 9291–9297 (1979). [PubMed] [Google Scholar]
  • 36.Cederholm-Williams SA Concentration of plasminogen and antiplasmin in plasma and serum. J Clin Pathol 34, 979–981, doi: 10.1136/jcp.34.9.979 (1981). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Itzek A et al. Contribution of plasminogen activation towards the pathogenic potential of oral streptococci. PLoS One 5, e13826, doi: 10.1371/journal.pone.0013826 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Isenring J et al. Streptococcus gallolyticus subsp. gallolyticus endocarditis isolate interferes with coagulation and activates the contact system. Virulence 9, 248–261, doi: 10.1080/21505594.2017.1393600 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Beaufort N et al. Activation of human pro-urokinase by unrelated proteases secreted by Pseudomonas aeruginosa. Biochem J 428, 473–482, doi: 10.1042/BJ20091806 (2010). [DOI] [PubMed] [Google Scholar]
  • 40.Molla A, Akaike T & Maeda H Inactivation of various proteinase inhibitors and the complement system in human plasma by the 56-kilodalton proteinase from Serratia marcescens. Infect Immun 57, 1868–1871 (1989). [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES