Fibrinolytic-deficiencies predispose hosts to septicemia from a catheter-associated UTI

SUMMARY Catheter-associated urinary tract infections (CAUTIs) are amongst the most common nosocomial infections worldwide and are difficult to treat due to multi-drug resistance development among the CAUTI-related pathogens. Importantly, CAUTI often leads to secondary bloodstream infections and death. A major challenge is to predict when patients will develop CAUTIs and which populations are at-risk for bloodstream infections. Catheter-induced inflammation promotes fibrinogen (Fg) and fibrin accumulation in the bladder which are exploited as a biofilm formation platform by CAUTI pathogens. Using our established mouse model of CAUTI, we identified that host populations exhibiting either genetic or acquired fibrinolytic-deficiencies, inducing fibrin deposition in the catheterized bladder, are predisposed to severe CAUTI and septicemia by diverse uropathogens in mono- and poly-microbial infections. Furthermore, we found that E. faecalis, a prevalent CAUTI pathogen, uses the secreted protease, SprE, to induce fibrin accumulation and create a niche ideal for growth, biofilm formation, and persistence during CAUTI.


INTRODUCTION 50
Urinary catheterization is a common procedure to drain urine from patients' bladders due 51 to chronic conditions or while in healthcare facilities, intensive care units, during surgical 52 procedures and recovery 1-3 . Despite its benefits, catheter placement increases the risk of 53 developing a catheter-associated urinary tract infection (CAUTI) 4-7 . CAUTIs are one of the most 54 common nosocomial infections and often lead to septicemia with a 30% mortality 3,8 . In fact, ~25% 55 of sepsis cases come from complicated UTI, including CAUTIs 9 . Current CAUTI management 56 colonization in the bladders (~1 log) and catheters (~1.5 logs) (Fig. 3b,c). Soluble Fg presence 187 (Fg AEK ) significantly decreased kidney dissemination (Fig. 3d). Enterococcal dissemination to the 188 spleen and hearts was not observed in any of the mouse lines (Fig. 3e,f). These results further 189 confirm that Fg is critical for enterococcal CAUTI. Importantly, soluble Fg or low fibrin formation 190 also results in lower colonization, indicating that fibrin accumulation is critical for persistent 191 colonization. 192 193

Fibrin accumulation enhances enterococcal bladder and catheter persistence during CAUTI. 194
Fibrinolysis is critical for dissolving fibrin net/clots and defective fibrinolysis results in fibrin 195 accumulation. Based on this and that Pg was found on the human and mouse catheters (Fig. 2a), 196 we investigated fibrinolysis in CAUTI, focusing in Pg and its key activators. We used C57BL/6-197 background mice deficient in different fibrinolysis factors (Fig. 3a). To increase fibrin 198 accumulation, we used mouse lines deficient for: i) plasminogen (Pg -/-) 37 , ii) urokinase 199 plasminogen activator (uPA -/-) 38 , and iii) tissue plasminogen activator (tPA -/-) 38 . To decrease fibrin 200 levels, we used a deletion of plasminogen activator inhibitor (PAI -/-) 39 , which exhibits unregulated 201 fibrin degradation (Fig. 3a). 202 We found that persistent fibrin accumulation led to significant increases of enterococcal 203 bladder (~ 2 logs) and catheter (~1.5 logs) colonization in Pg -/mice when compared to WT mice 204 (Fig. 3g,h). Importantly, fibrin reduction in the bladder via PAI deficiency, resulted in significantly 205 lower bladder (~1 log) and catheter (~1.5 logs) colonization (Fig. 3g,h). 206 To further confirm that inactivation of Pg proteolytic activity correlates with enterococcal 207 colonization enhancement, we used a transgenic mouse expressing Pg with a plasmin-inactivating 208 active site mutation (PGB) 40 , finding that bacterial burden in the bladder and catheter was similar 209 to Pg -/mice (Fig. 3g,h). This suggests that fibrinolysis inhibition correlates with exacerbating E. 210 faecalis colonization. 211 Importantly, bladder and catheter bacterial colonization in uPA -/mice significantly 212 increased similarly to Pg -/mice, whereas tPA -/mice exhibit similar bacterial burden to WT mice. 213 Thus, uPA is the primary activator of Pg into plasmin in the catheterized bladder (Fig. 3g,h). Since 214 coagulation dysregulation is not uncommon in human populations (Supplementary  further assessed the effect of Fg/fibrin modulation in enterococcal dissemination, finding that 220 fibrinolytic defects (Pg -/-, PGB, and uPA -/-) significantly increased bacterial burden in kidneys, 221 spleen, and hearts compared with WT mice colonization ( Fig. 3i-k). Furthermore, tPA deficiency 222 resulted in a bimodal kidney enterococcal colonization (Fig. 3i). Interestingly, kidneys with higher 223 enterococcal colonization came from mice with higher bladder colonization, suggesting that a 224 persistent bladder colonization may result in their dissemination (Fig. 3i). These demonstrate that 225 fibrin accumulation predisposes the host to enterococcal systemic dissemination. 226 227 E. faecalis proteases degrade plasmin and plasminogen but not thrombin. We previously 228 demonstrated that during CAUTI and growth in urine conditions, E. faecalis OG1RF induced the 229 expression and activity of two enterococcal secreted proteases, a serine protease (SprE) and 230 metalloproteinase gelatinase (GelE), which are critical for enterococcal CAUTI and systemic 231 dissemination 29 . Given that E. faecalis in the catheterized bladder increases Fg/fibrin levels ( Fig. 2b) and that fibrinolysis deficiencies enhance enterococcal colonization (Fig. 3), we examined if 233 the enterococcal secreted proteases antagonizes the fibrinolytic system by assessing their ability 234 to degrade human Pg (hPg) and plasmin (hPm) in vitro. 235 We compared the protease activity of WT, mutants defective for each protease (∆sprE or 236 ∆gelE), or both proteases (∆gelE∆sprE) against hPg and hPm under urine conditions (See 237 Methods; Supplementary Fig. 1). Filtered WT supernatants degraded hPg (~95 kDa) and hPm 238 (~85 kDa), producing a major degradation fragment ~35 kDa and 3 visible degradation products 239 (DP) between 50-40 kDa (Fig. 4a,b). WT and ∆gelE supernatants produced similar hPg and hPm 240 degradation pattern while the ∆sprE supernatant was not able to fully degrade hPg and hPm, 241 showing three faint fragments between 50-40kDa and no major degradation product at ~35 kDa 242 ( Fig. 4a,b). The ∆gelE∆sprE mutant completely lost the ability to degrade hPg and hPm into 243 smaller peptides, similar to urine or PBS controls (Fig. 4a,b,d,e). 244 To understand whether these proteases are promiscuous and capable of targeting other 245 factors in the coagulation cascade, specifically the fibrin formation pathway, we incubated the 246 strains' supernatants with thrombin, finding that thrombin was not degraded in any of the 247 treatments ( Fig. 4c,f). 248 To further validate these, we determined the proteolytic activity of supernatants from E. 249 faecalis WT and various mutants against mouse Pg and thrombin in a complex catheterized 250 environment. Cell-free bacterial supernatants were incubated with bladder homogenates (1:5) from 251 PGB mice catheterized for 24 hrs (Supplementary Fig. 1). WT and ∆gelE supernatants 252 significantly degraded Pg while no degradation was observed in the urine control, the ∆sprE single 253 mutant and the double ∆gelE ∆sprE mutant supernatant (Fig. 4g-h). No cleavage of thrombin was 254 observed with any of the treatments (Fig. 4g,i). 255 To determine whether E. faecalis proteases could functionally disrupt fibrinolysis by 256 targeting plasmin, we incubated purified plasmin with WT and mutant strains' supernatants, 257 followed by a secondary incubation with thrombin-generated fibrin (Fig. 4j) or Pg -/homogenates 258 ( Fig. 4k) that were catheterized, non-infected, for 24hrs (Supplementary Fig. 1). Supernatants-259 containing SprE (WT and ∆gelE) degraded plasmin, resulting in lack of fragment E generation, 260 demonstrating plasmin inhibition ( Fig. 4j-m). These data suggests that E. faecalis' SprE protease 261 activity in urine targets the fibrinolytic system by inactivating Pg and plasmin, which results in 262 dysregulated fibrin accumulation. reduced in Fg -/mice. Predictively, defective colonization was rescued in mice that accumulate 269 fibrin due to fibrinolytic deficiencies (Pg -/and PGB) (Fig. 5a,b), exhibiting higher than the WT 270 strain colonization similar to fibrinolytic-deficient mice infected with WT strain (compare to Fig.  271 5a,b to Fig. 3). Importantly, fibrin accumulation in the catheterized bladder promoted E. faecalis 272 ∆gelE∆sprE systemic dissemination (Fig. 5c- 19 . Here, we found that bladder and catheter colonization in Fg -/mice or 280 coagulation-deficient mice (Fg AEK ; soluble Fg) was significantly reduced (Fig. 5f,g). Conversely, 281 not only was UTI89 bladder and catheter colonization significantly increased (~2 logs) in mice 282 with fibrinolytic deficiencies (Pg -/and PGB) (Fig. 5f,g), but these mice also experienced 283 significant systemic dissemination ( Fig. 5h-j). These data provide evidence that uropathogenic E.  (Table S3). TXA is 289 extensively used to treat patients with postpartum hemorrhages, traumatic injury, and surgical 290 procedures that increase the risk of bleeding 42,43 . TXA is a synthetic anti-fibrinolytic amino acid 291 that acts by competitively blocking the lysine binding sites on plasmin(ogen), inhibiting plasmin 292 interaction with fibrin 43 (Fig. 6a). As 86% of surgery patients require urinary catheterization 44 , we 293 assessed whether TXA treatment during urinary catheterization increased fibrin accumulation in 294 the bladder, enhancing colonization and systemic dissemination by the three most prevalent 295 CAUTI uropathogens: E. faecalis, E. coli, and the fungal pathogen C. albicans. Mice were treated 296 with TXA or vehicle control (PBS) intraperitoneally (Fig. 6b). Then, mice were catheterized and 297 infected with the respective pathogen and sacrificed at 1 dpi. TXA-treated mice exhibited a 298 significant increase of bladder and catheter colonization and promoted systemic dissemination of 299 all pathogens when compared with PBS-treated mice ( Fig. 6c-e). No significant differences in 300 bladder weights between treatments were observed, suggesting that TXA did not exacerbate 301 catheter-induced inflammation (Fig. 6f). To confirm fibrinolysis inhibition and fibrin 302 accumulation, Fg/fibrin concentration and fragment E generation was compared and quantified 303 between PBS-and TXA-treated bladders ( Fig. 6g-i). We found that Fg/fibrin concentration 304 increased and fragment E generation decreased in TXA-treated bladders ( Fig. 6g-i). Since Fg 305 induces IL-6 expression in immune and epithelial cells 45,46 , we measured IL-6 levels in the bladder. 306 We found that IL-6 levels significantly increased in TXA-treated mice (Fig. 6j). This is consistent 307 with increased Fg/fibrin levels in the TXA-treated bladders, suggesting a perpetual inflammatory 308 positive-feedback loop in the catheterized host. 309 We further tested whether fibrin accumulation could enhance polymicrobial CAUTI. For 310 this, TXA and PBS-treated mice were catheterized and infected with all three main uropathogens, 311 E. faecalis, E. coli, and C. albicans. We found no difference in inflammation between PBS-and 312 TXA-treated mice (Fig. 6k). However, blocking fibrinolysis with TXA significantly enhanced the 313 burden of all pathogens in the bladders and catheters and increased systemic dissemination (Fig.  314 6l-n). In conclusion, these data show that pharmacological inhibition of fibrinolysis in the host 315 enhances mono-and polymicrobial CAUTI and systemic dissemination by diverse uropathogens. 316

Discussion 317
CAUTIs remain prevalent and costly, increasing patient morbidity and mortality 47,48 . 318 CAUTI prevention and treatment faces major challenges, including the ability to predict: i) when 319 catheterized patients develop infections; ii) which uropathogen will cause the infection; and iii) 320 which patient population is more susceptible to septicemia. Previous attempts at treating CAUTIs 321 include vaccine development, antibody therapy, and compounds that target specific pathogens 6,49 . 322 However, multiple pathogens from different kingdoms can simultaneously cause CAUTIs and the 323 risk of developing an infection increases with dwelling time. Efficiently predicting when patients 324 will develop an infection, the causative organism, and which patient populations are more 325 susceptible to systemic dissemination is imperative for successful CAUTIs prevention and 326 treatment. Thus, a better understanding of CAUTI pathophysiology is critical to develop better 327 clinical practices for improving patient outcomes. 328 Here, we found that the coagulation cascade plays an important role in CAUTI outcome. 329 Importantly, we identified that host coagulopathies that result in fibrin clot accumulation promote 330 higher pathogen colonization in the catheterized bladder by the most prevalent uropathogens in 331 mono-or poly-microbial infections, enhancing persistence during CAUTI and promoting 332 bloodstream infection and systemic dissemination (Fig. 3, 5, and 6). Fibrin is a major thrombus-333 formation end product 50 and its concentration is tightly controlled by a series of cofactors, 334 inhibitors, and receptors 15 . Once healing is resolved, the fibrinolytic pathway is triggered by 335 activation of Pg into plasmin by uPA and tPA, promoting fibrin clot degradation and restoring 336 tissue homeostasis (Fig. 3g) 14,15 . In the catheterized bladder, our results show that uPA, which was 337 originally isolated from human urine, is the main Pg activator, but not tPA (Fig. 3). This could be 338 related to their action mechanisms as tPA activation of Pg requires attachment to the fibrin on the 339 clot surface to activate fibrin-bound Pg 51 . Since CAUTI pathogens form biofilms on fibrin, they 340 may occlude tPA binding. In contrast, uPA is fibrin-independent and activates Pg in solution or 341 when associated with its cellular receptor uPAR 52 , providing a cell-mediated Pg activation 53 . 342 uPAR is present on many immune cells including macrophages 53 that are recruited by urinary 343 catheterization 20,27,28 ; thus, it is possible that these immune cells are important for uPA-cell-344 mediated Pg activation. The uPA activation mechanism of Pg will require further experimentation. 345 Hypofibrinolysis and several thromboembolic diseases, including strokes, venous 346 thromboembolism, rheumatoid arthritis, and renal diseases are accompanied by fibrin accumulation 54,55 . Furthermore, there are several congenital deficiencies that result in fibrin clot 348 accumulation, including individuals with deficiencies of Pg, tPA, uPA, PAI-1, PAI-2, TAFI, and 349 Annexin A2 50 (Table S3). Consistent with our CAUTI data, fibrin accumulation from imbalance 350 between activation of coagulation and inhibition of fibrinolysis, is detrimental for patients 351 suffering from COVID-19 56 and ventilator-associated pneumonia 57 . Moreover, fibrin 352 accumulation can also result from antifibrinolytic treatments, which are commonly use in patients 353 with bleeding disorders 55,58 (Table S3) IL-6 is the most highly elevated cytokine in the bladder, bloodstream, and liver and E. 364 faecalis infection promotes its concentration (Fig. 1i). IL-6 is a potent regulator of inflammation,  Fig. 4b). Interestingly, IL-6 is repressed during uUTI 28 , further demonstrating that uUTI and 374 CAUTI have different pathophysiologies. Therefore, targeting IL-6/Fg signaling to modulate its 375 effects on pathogenesis may provide a novel approach to improving CAUTI outcomes. 376 Here, we found that SprE actively degrades Pg and plasmin, suggesting that E. faecalis 377 uses SprE to inactivate the fibrinolytic system. This strategy differs from other bacterial pathogens, 378 including Group A Streptococcus (GAS), which activates Pg into plasmin via streptokinase, 379 dissolving blood cloth to cause an invasive infection 66 . Therefore, identifying Pg/plasmin cleavage 380 sites using purified SprE will be important to determine its action mechanism. IBC formation is significantly reduced without affecting overall bladder colonization 68,69 . We found that Fg, which is highly glycosylated, is used by E. coli colonize the bladder and catheter 394 during CAUTI 19 . Fg deposition onto the urothelium could block E. coli-cell interaction, affecting 395 IBC formation. Consistent with this, we found that Fg/fibrin levels directly affected E. coli CAUTI 396 (Fig. 5f,g), decreasing colonization when Fg was soluble or absent while increasing colonization 397 when fibrin accumulated. Soluble Fg and other host glycosylated proteins, including uromodulin 69 , 398 may act as decoy receptors to prevent binding to urinary tract surfaces, resulting in the pathogen's 399 expulsion by urine flow. In contrast, fibrin accumulation on surfaces can provide a platform for E. 400 coli adherence to counteract its clearance by urine flow. 401 Given how common urinary catheterization is and the rise of multidrug resistant pathogens, 402 the CAUTI frequency is expected to keep increasing. This study not only has identified 403 mechanisms by which catheter-induced inflammation predispose patients to development of 404 CAUTIs but also identified host populations that will be at higher risk of a CAUTI-associated 405 septicemia caused by the most prevalent CAUTI pathogens (Extended Data Fig. 5)

Lead Contact 419
Further information and requests for resources and reagents should be directed to and will be 420 fulfilled by the lead contact, Ana Lidia Flores-Mireles (afloresm@nd.edu). 421

Materials availability 422
The study did not generate new unique material or reagents. the final CFUs/50 μl were ∼2 × 10 7 CFU for bacteria and ~1x10 6 CFU for fungi. Immediately 455 before sacrifice, blood was harvested via cheek poke and serum was extracted from whole blood (BD Microtainer #365967) for cytokine analysis as described below. To harvest catheters and 457 organs, mice were sacrificed at their specified time-point by cervical dislocation after anesthesia 458 inhalation; the bladder, kidneys, heart, spleen, liver, and if present, silicone catheter were 459 aseptically harvested and bladders weighed. Except livers, all organs were homogenized (1min 460 shake, 5 min rest, 1 min shake; MP Biomedical 116005500) and plated for CFU enumeration as 461 described below. A subset of bladders used for histology analysis were weighed, fixed and 462 processed as described below. Homogenized samples of bladder, liver and blood serum were sent 463 for cytokine analysis as described below. Catheters were subjected to 15min sonication (Branson 464 2800) for CFU enumeration or sent for proteomic analysis as described below using nonimplanted Mouse bladders were fixed in 10% neutralized formalin for 18hrs, embedded, sectioned, and 497 stained as previously described 72 . Briefly, bladder sections were deparaffinized, rehydrated, and 498 rinsed with water. Hematoxylin and Eosin (H&E) stain for light microscopy was done by the 499 CORE facilities at the University of Notre Dame (ND CORE). All imaging was done using a 500 Zeiss Axio Observer inverted light microscope. Zen Pro and ImageJ software were used to analyze 501 the images.

Microbial Enumeration 503
Microbe load from bladder, kidney, spleen, hearts and silicone catheters from sacrificed animals 504 was determined via serial dilution and enumeration. Organs were homogenized and silicone 505 catheters were cut in small pieces before sonicated for CFU enumeration. Pathogens were plated 506 in their corresponding media conditions (see Microbial strains and growth conditions). 507 508

Cytokine analysis 509
Bladder, liver, and blood serum samples from mice catheterized and infected with E. faecalis 510 OG1RF or catheterized and mock infected with PBS for 1h, 3h, 6h, 9h, 12h, 1d, 3d, 7d, and 14d 511 were frozen at −80 °C until time of assay. Before cytokine analysis, homogenates were thawed on 512 ice and microcentrifuge at 11,000 × g for 10 min, and supernatants transferred to a new tube. were for 1 dpi (mice) and 24-26 hrs dwell time (humans) were compared using an E Venn 532 network 73 online software (http://www.ehbio.com/test/venn/#/) to identify common proteins in the 533 catheterized bladder environment between mice and humans. The 76-shared proteins were then 534 submitted to metascape.org (Table S3) for gene ontology classification, gene ontology 535 enrichment, protein-protein interaction identification, network analysis, and prediction of 536 transcriptional regulatory relationships 32 . Generated networks were modified using Cytoscape 537 software 74 . 538 539 Bacterial protease activity assay 540 Cultures of E. faecalis strains were centrifuged for 10 min at 3000x and washed three times with 541 1xPBS. Pellets were resuspended in 10ml of filter sterilized human urine at pH 6.5. The 542 resuspended cultures were statically incubated for 18 hrs at 37 o C. Following incubation, cultures 543 were centrifuged and the supernatants were collected, filter-sterilized, and concentrated and size 544 excluded using Macrosep Advance Centrifugal Devices (PALL MAP010C37). 40 µg/mL of 545 purified protein (hPlasmin, hPlasminogen, or hThrombin) were added to supernatants and 546 incubated statically for 4 hrs at 37 o C. To determine proteolytic activity of E. faecalis strains in 547 mice, urine-cultured supernatants were also added to PGB mouse bladder homogenates (1:5) that 548 were catheterized only for 24 hrs. The mixture was incubated for 4 hrs at 37 o C. Following 549 incubation, 5x SDS sample buffer was diluted to 1x into the samples and boiled at 95 o C for 2 550 minutes before separation on 12% acrylamide gels at 100V. Plasmin that was incubated with E. 551 faecalis mutant strains' supernatants was subsequently incubated with 25 µl of thrombin-generated 552 fibrin (from 2 mg/mL of purified fibrinogen) or 25 µl of 24hpc Pg -/bladder homogenates for 4hrs 553 at 37 o C to determine functional disruption of fibrinolytic activity. Fibrin and homogenates 554 incubated with supernatant-treated plasmin was then also processed for acrylamide gel protein 555 separation. Normalization and western blots were done as described below. 556 557

Western blot 558
Western blotting was done as previously described 75 . Briefly, bladder homogenate samples were 559 diluted 1:1 into 5X sample buffer then boiled at 95°C for 2 min. Following boiling, 10uL of sample 560 in SDS was run on a 12% polyacrylamide gel for separation. Gels were then stained with 561 Coomassie Blue for 60min followed by de-staining and imaging on an Odyssey infrared imaging 562 system (LI-COR Biosciences, Lincoln, NE) to normalize protein concentrations. Following 563 normalization, samples were again run on 12% acrylamide gels at 4 o C and 50V. The samples were 564 then transferred to polyvinylidene difluoride membrane using a semi-dry transfer (Millipore Sigma 565 Cat# IPFL00005). Following transfer, membranes were blocked in 5% skim milk and probed by 566 Western immunoblot with anti-β-actin (Abcam ab8229) as a loading control, and anti-plasminogen 567 primary antibodies (Proteintech 17462-1-AP, respectively), anti-thrombin (ThermoFisher 568 Bladder homogenates from respective mouse bladder groups were analyzed to determine 576 abundances of Fg and IL-6 cytokine with respective ELISA kits following manufacture-provided 577 protocols. Briefly, bladder homogenates were centrifuged and supernatants were diluted 1:10 in 578 ELISA diluent. Dilution standard preparation was performed following manufacturer guidelines. 579 Standards/samples were then placed on to pre-coated ELISA plates overnight at 4 o C, removed, 580 plates washed three times with wash buffer, incubated with detection antibodies for 1hr at room 581 temperatures, washed three times, incubated with appropriate biotinylated-detecting enzyme for 582 30min at room temperature, washed 5 times, and incubated with appropriate chromogenic 583 substrate for 15min at room temperature, and stopped with appropriate stop solution. Absorbances 584 were measured at appropriate wavelengths with a plate reader (Molecular Devices Spectramax 585 ABS plus). 586 587

Measurement of Fragment E 588
Fragment E was probed via SDS-PAGE and western blot with an anti-Fg primary antibody and 589 anti-β-actin as a loading control. Fragment E probing was performed for the following 590 experiments: i) in vitro bacterial protease activity of plasmin incubated with fibrin or with Pg-/-591 homogenates (Fig. 4j-m), ii) in vitro inhibition of plasmin fibrinolytic activity by TXA (Fig. 7h), 592 and iii) homogenized bladders from both TXA-treated and vehicle controls (Fig. 7h). Fragment 593 E band was visible at ~41kDa. Measurements of the median fluorescent intensity from each sample

Statistical Analysis 598
Data derived from this study was entered into Graphpad Prism 9 to generate statistical results and 599 graphs. T-tests were used to determine significance between samples. When data was 600 nonparametric, and the median better represented the distribution, comparisons were made with 601 Mann-Whitney U Tests. Pearson's correlation statistical analysis was used to measure association 602 between variables.