Function and contribution of two putative Enterococcus faecalis glycosaminoglycan degrading enzymes to bacteremia and catheter-associated urinary tract infection
Department of Microbiology and Immunology, Jacobs School of Medicine and Biomedical Sciences, State University of New York at Buffalo, Buffalo, NY, USA
Department of Biological Sciences, University of Texas at Dallas, Richardson, Texas, USA
Department of Biostatistics, School of Public Health and Health Professions, State University of New York at Buffalo, Buffalo, NY, USA
Department of Urology, University of Texas Southwestern Medical Center, Dallas, Texas, USA
*Correspondence: chelsiea@buffalo.eduAbstract
Enterococcus faecalis is a common cause of healthcare acquired bloodstream infections and catheter associated urinary tract infections (CAUTI) in both adults and children. Treatment of E. faecalis infection is frequently complicated by multi-drug resistance. Based on protein homology, E. faecalis encodes two putative hyaluronidases, EF3023 (HylA) and EF0818 (HylB). In other Gram-positive pathogens, hyaluronidases have been shown to contribute to tissue damage and immune evasion, but function in E. faecalis has yet to be explored. Here, we show that both hylA and hylB contribute to E. faecalis pathogenesis. In a CAUTI model, ΔhylA exhibited defects in bladder colonization and dissemination to the bloodstream, and ΔhylB exhibited a defect in kidney colonization. Furthermore, a ΔhylAΔhylB double mutant exhibited a severe colonization defect in a model of bacteremia while the single mutants colonized to a similar level as the wild-type strain, suggesting potential functional redundancy within the bloodstream. We next examined enzymatic activity, and demonstrate that HylB is capable of digesting both HA and CS in vitro while HylA exhibits only a very modest activity against heparin. Importantly, HA degradation by HylB provided a modest increase in cell density during stationary phase and also contributed to dampening of LPS-mediated NF-Bκ activation. Overall, these data demonstrate that glycosaminoglycan degradation is important for E. faecalis pathogenesis in the urinary tract and during bloodstream infection.
Introduction
Enterococcus faecalis infections are common and impose a high burden on the healthcare system. E. faecalis ranks as the 5th most frequently encountered healthcare-associated pathogen in the United States and can cause a variety of infections including endocarditis, central line-associated bloodstream infection (CLABSI), and catheter-associated urinary tract infections (CAUTIs)(1–3). E. faecalis is the most common CLABSI pathogen in both pediatric intensive care units and adult long-term acute-care hospitals (1, 2), and is among the three most common causes of CAUTI (3). E. faecalis CAUTI can also progress to disseminated infection and secondary bacteremia, and is estimated to cause approximately 13,000 deaths per year in the United States alone (4, 5).
One host factor that may protect against invasive Enterococcal infections are glycosaminoglycans (GAGs). GAGs are acidic polysaccharides that are ubiquitous on the cell surface and extracellular matrix and they provide tissue integrity and structure in addition to mediating immune signaling and tissue repair (6–10). With respect to infection, GAGs are present at many barrier sites such as the endothelial glycocalyx and the urinary tract urothelial lining where they are thought to play a protective role against pathogen invasion (11, 12). For example, the GAG-rich glycocalyx of the bladder mucosa is believed to mask uroplakins and other surface proteins that are targeted by bacteria for adhesion and invasion (11–13). In support of this hypothesis, removal of GAGs from the bladder mucosa increases attachment of Enterococcus faecalis as well as other uropathogenic bacteria (14–16).
The most common GAGs in urothelium and endothelial glycocalyx are hyaluronic acid (HA), chondroitin sulfate (CS), and heparin (6, 17). Several pathogenic bacteria produce enzymes that digest GAGs, which can destroy barrier functions, provide a carbohydrate source for microbial growth, and alter the innate immune response either by promoting inflammation or disrupting inflammatory signaling. For example, some Streptococcus species produce hyaluronidase enzymes that depolymerize HA and have been shown to promote tissue invasion, disrupt inflammatory signaling, and facilitate bacterial growth by liberating a new carbon source (18–21). We therefore hypothesized that GAG degradation by E. faecalis could contribute to pathogenesis and invasive disease.
There are several conflicting reports pertaining to E. faecalis hyaluronidase activity. In 1964, Rosan and Williams reported hyaluronidase activity in oral Enterococcus isolates (22). However, this study was performed at a time when enterococcal taxonomy was in a state of flux, so it is unclear whether the hyaluronidase-positive strains were E. faecalis, E. faecium, or another species (23, 24). A putative hyaluronidase was identified in Enterococcus faecium that is predominantly associated with vancomycin-resistant clinical isolates (25), and transfer of the plasmid encoding the putative hyaluronidase increased virulence of a fecal isolate in a mouse peritonitis model (26). However, no hyaluronidase activity was observed for any of the E. faecium strains (27), and deletion of the putative hyaluronidase itself did not reduce fitness in the mouse peritonitis model (28).
A recent assessment of in vitro GAG digestion by several E. faecalis clinical isolates found that none of the isolates degraded HA, CS, or HS (29). However, a homology of the Streptococcus hyaluronidase gene is present in all 518 E. faecalis complete genome sequences available through the Bacterial and Viral Bioinformatics Resource Center (BV-BRC)(30) and 68% of the isolates also have a second putative hyaluronidase. E. faecalis strain V587 is one of the clinical isolates with both putative hyaluronidases. This strain is particularly notable as it is vancomycin-resistant and was isolated from the urine of a patient who later experienced sepsis with the same strain (31), making it relevant to two of the most prevalent types of E. faecalis infection. The purpose of this study was to characterize the enzymatic activity of both putative hyaluronidases and examine contribution to CAUTI and bacteremia.
Results
Enterococcus faecalis V587 encodes two putative hyaluronidases with differing predicted protein architectures
The genome of E. faecalis strain V587 in the Bacterial and Viral Bioinformatics Resource Center (BV-BRC)(30) contains two putative hyaluronidases located in different parts of the chromosome: ef3023 (hylA) is 4,119 bp at position at 2,884,596 bp (Figure 1A), and ef0818 (hylB) is 3,015 bp at position 763,339 bp (Figure 1B). The hylA gene is monocistronic, and no neighboring genes are predicted to be involved in oligosaccharide digestion or uptake. The hylB gene is similarly monocistronic but immediately downstream of an operon that encodes for a putative hyaluronate-oligosaccharide phosphotransferase system with a divergently-transcribed GntR-family transcriptional regulator (32).
Using the Bacterial and Viral Bioinformatics Resource Center (BV-BRC)(33), we searched 646 complete genome sequences of E. faecalis for the presence of hylA and hylB. A total of 582 strains (90%) had at least one putative hyaluronidase with >95% amino acid identity and >80% query coverage: 470 strains (73%) had a homolog of hylA, 459 strains (71%) had a homolog of hylB, and 289 (45%) had both. Thus, hylA and hylB are well conserved across E. faecalis strains. Both are predicted to encode “polysaccharide lyase family 8” enzymes, and both are predicted to contain Sec pathway protein export signal peptides according to SignalP 5.0 (34). However, the amino acid sequences of HylA and HylB share only 36% identity to each other and 24-30% identity with hyaluronidases from other bacterial species (Supplemental Table 1). There are also substantial differences in the domain architecture of the two proteins: in addition to the GAGase domain, HylA contains a Discoidin (DS) and Bacterial immunoglobulin-like fold (Big2) domains, a section of repeated amino acid residues annotated as a “Found In Various ARchitectures” (FIVAR) domain, and a predicted LPXTG domain for cell wall attachment (Figure 1C) (35), while HylB lacks cell wall anchoring motifs and has only two small flanking domains with no predicted function by an NCBI Conserved Domain Search (36) (Figure 1D). Structures predictions of HylA (Figure 1E) and HylB (Figure 1F) were generated using AlphaFold (37, 38), with the GAGase domain shown in pink and other domains in different colors. Importantly, an alignment of the two proteins could not be achieved, which underscores the substantial differences in structure.
E. faecalis V587 does not exhibit hyaluronidase activity under in vitro conditions
Considering the prior literature regarding a lack of hyaluronidase activity in E. faecalis isolates (29), we first sought to examine expression of hylA and hylB in strain V587. Notably, RNA sequencing data from the closely-related V583 strain indicates that hylA and hylB are not expressed during growth in BHI (39). We therefore used RT-qPCR to examine expression of hylA, hylB, and the housekeeping gene recA in V587 after 4 hours of growth in BHI, 0.5X TSB (a condition previously found to support hyaluronidase activity in other species) (29), and human urine (Figure 2). TSB and urine were also supplemented with either unfractionated high molecular weight HA or low molecular weight HA (∼5 kDa) to determine if expression is induced by hyaluronic acid. The Cq values for recA were similar across all conditions, confirming that recA is an appropriate housekeeping gene for normalizing expression (Figure 2A), and all samples were detected at least 10 cycles before the respective no template controls (Figure 2B). When growth in BHI was used as the calibrator, growth in 0.5X TSB resulted in a slight increase in expression of both hylA and hylB gene while growth in urine reduced expression (Figure 2C). To examine the impact of HA on expression, we next used unsupplemented 0.5X TSB or urine as the calibrator. In TSB, supplementation with unfractionated HA slightly induced expression of hylA but had no effect on hylB (Figure 2D). In contrast, unfractionated HA had no impact on expression in urine, but supplementation with 5 kDa HA induced expression of both hylA and hylB (Figure 2E). Thus, hylA and hylB are expressed by E. faecalis V587 under multiple growth conditions, and expression can be stimulated to some extent by HA.
We generated markerless deletion mutants of hylA and hylB, as well as a double deletion mutant (ΔhylAΔhylB). All of the mutants grew similarly to V587 in BHI, TSB, and human urine (Figure 3 A-C). To examine in vitro hyaluronidase activity, V587 and the ΔhylAΔhylB mutant were incubated on agar plates containing HA and examined for development of a zone of clearance around the colony (Figure 3D). Consistent with prior studies, no hyaluronidase activity was detected for either strain, suggesting that while hylA and hylB are expressed in vitro, active protein is not produced under these conditions. To determine if the lack of activity is specifically due to lack of expression or function under these conditions, hylB was cloned into pBAD and transformed into Lactococcus lactis, a Gram-positive species that lacks endogenous hyaluronidases. Notably, expression of the V587 hylB gene in L. lactis resulted in modest hyaluronidase activity, suggesting that the E. faecalis HylB can indeed function as a hyaluronidase and the lack of activity in V587 is likely due to regulatory control of expression and activity.
HylA and HylB contribute to E. faecalis bladder colonization and bacteremia
Prior to further probing the enzymatic activity of HylA and HylB, we first sought determine contribution to E. faecalis pathogenesis in our established mouse model of CAUTI. In brief, female CBA/J mice were inoculated transurethrally with 1 x 105 CFU of either V587, ΔhylA, ΔhylB, or ΔhylAΔhylB. During inoculation, a small piece of silicone tubing was left in the bladder to simulate a urinary catheter, as previously described (40). At 48 hours post infection (HPI) the mice were sacrificed and bladders, kidneys and spleens were collected for quantitation of CFUs, with spleen CFU used as an indicator of bacteremia (Figure 4A). Inoculation with the ΔhylA mutant significantly reduced bladder and spleen colonization compared to infection with wild-type V587, and also reduced the overall incidence of bacteremia. In contrast, inoculation with the ΔhylB mutant resulted in similar bladder colonization as wild-type V587 but a significant defect in the kidneys. Therefore, HylA and HylB both contribute to E. faecalis CAUTI but may be differentially expressed in different host niches or have differing roles in pathogenesis. Interestingly, infection with the ΔhylAΔhylB double mutant phenocopied the mean bladder colonization of infection with ΔhylA and the men kidney colonization as ΔhylB, but the only statistically significant defect compared to wild-type V587 was a reduction in spleen colonization and incidence of bacteremia.
As we observed a reduced incidence of secondary bacteremia in the CAUTI model and E. faecalis is a common cause of CLABSI, we next sought to determine whether HylA and HylB contribute to primary bacteremia. CBA/J mice were inoculated via tail vein injection with 1 x 108 CFU of V587, ΔhylA, ΔhylB, or ΔhylAΔhylB, mice were sacrificed 24 HPI, and CFUs were quantified in the liver, spleen, and kidneys (Figure 4B). While the ΔhylA and ΔhylB mutants exhibited similar overall colonization to V587 in this model, the ΔhylAΔhylB double mutant had a significant colonization defect in the kidneys and the spleen. Thus, HylA and HylB may be functionally redundant during primary bacteremia.
Constitutive expression of hylB confers hyaluronidase and chondroitinase activity
With a contribution to pathogenesis confirmed, we next sought to define the enzymatic properties of HylA and HylB in E. faecalis V587. To bypass any issues due to endogenous gene expression, we opted to constitutively express hylA or hylB in the ΔhylAΔhylB double mutant. HylB is predicted to be fully secreted and lacks complex domain architecture, so the full-length sequence of hylB was cloned under control of the ermB promoter in pAOJ84, and exported protein was detected as the expected molecular weight of 109 kDa (Figure 5A). HylA is predicted to be anchored to the cell wall, which presents a potential challenge for assays requiring unattached secreted enzyme and could also impact cell wall integrity when over-expressed. We therefore designed vector pAOJ86 to contain the full-length hylA sequence lacking the LPXTG anchor, and exported protein was again detected at the expected molecular weight of 146 kDa (Figure 5A).
Using an established semi-quantitative method for measuring GAG digestion with live bacteria (29), we incubated E. faecalis WT, ΔhylAΔhylB, and ΔhylAΔhylB containing the constitutive expression vectors for 24 hours at 37°C under low oxygen conditions in 0.5X TSB containing 2.5 mg/mL of hyaluronic acid (Figure 5B). In agreement with the literature and our prior results, V587 did not digest any of the GAGs and had an identical profile as the negative control strain ΔhylAΔhylB (Figure 5B-D), confirming that E. faecalis does not exhibit GAGase activity under in vitro conditions (29). Unexpectedly, ΔhylAΔhylB expressing hylA (pAOJ86) did not exhibit any HA degradation, but ΔhylAΔhylB expressing hylB (pAOJ84) fully degraded HA, confirming that HylB is indeed a hyalouronidase.
Four additional variants of HylA were generated for constitutive expression to account for potential issues in protein folding, including 1) full-length HylA with the LPXTG anchor intact (pAOJ83), 2) HylA lacking both the LPXTG and FIVAR/repeat domains (pAOJ87), 3) HylA lacking the non-enzymatic conserved domains at the N-terminus (pAOJ88), and 4) a HylA chimera in which the LPXPTG anchor was removed and the N terminal region upstream of the predicted GAGase domain was replaced with the N terminal region upstream of the GAGase domain of HylB (pAOJ90). All resulted in secreted protein at the correct molecular weight except for pAOJ87 (Supplemental Figure 1), which may indicate a requirement for the FIVAR/repeat domain for proper protein expression, stability, or secretion. However, none of the HylA expression vectors conferred HA degradation, indicating that either none of the HylA constructs produce catalytically-active enzyme or that HylA acts on a different substrate.
To determine if either HylA or HylB act on other GAGs, we repeated the assay using chondroitin sulfate (CS, Figure 5C) and heparin sodium salt (HS, Figure 5D). The strain expressing HylB degraded the majority of the CS, which was not entirely unexpected as hyaluronidases from other bacteria have been shown to exhibit activity against CS (41). In contrast, the only phenotype observed for the strain expression HylA was a slight though statistically significant decrease in HS levels.
An important consideration when studying GAG degradation is the final product size. Host hyaluronidases and other factors such as oxidative damage (42) can cleave HA into intermediate sized fragments (< 500 kDa) or oligosaccharides that act as damage-associated molecular patterns and stimulate inflammation via TLR-2 and/or TLR-4 (10, 43, 44), while hyaluronidases from some pathogenic bacteria can cleave terminal disaccharides from HA which are anti-inflammatory (18, 45). Since the semi-quantitative GAG degradation assay is based on precipitation of high molecular weight GAGs, it would not detect subtle decreases in GAG fragment size. We therefore incubated HA directly with live E. faecalis V587, ΔhylAΔhylB or ΔhylAΔhylB harboring each of the constitutive hyaluronidase constructs or a vector control (pAOJ55) for 24 hours and used agarose gel electrophoresis to examine HA degradation product size (Figure 5E). An uninoculated control was incubated under the same conditions. Since E. faecalis acidifies BHI to a pH of ∼5.7 after 24 hours, an additional BHI control was adjusted to pH 5.7 with lactic acid to account for any HA degradation that might occur due to acidification alone. Incubation of HA with commercially available hyaluronidases from Streptomyces hyalurolyticus (HylSh) and Bovine Testicular Hyaluronidase Type I-S (BovHyl), that both digest HA down to short oligosaccharides (46), were utilized as positive controls.
Incubation with HylSh and BovHyl fully degraded the HA to a small band at ∼5 kDa, while no degradation was observed for either BHI control (Figure 5E). Importantly, HylSh is highly specific for HA (46), confirming the identity of the bands stained by this method. The strain containing pAOJ84 (hylB) displayed obvious digestion of HA and left only a low molecular weight smear of degradation products visible on the gel, albeit larger than the ∼5 kDa products of HylSh or BovHyl (Figure 5E). No degradation was observed for any of the HylA constructs. We hypothesized that HylA may only act on lower molecular weight HA fragments produced through the action of HylB, and further digest them to even smaller oligo- or disaccharides. To test this hypothesis, we incubated HA with a combination of ΔhylAΔhylB carrying pAOJ83 (full length hylA, in case the LPXTG cell wall anchor was required for full activity) and ΔhylAΔhylB carrying pAOJ84 (hylB). The combination still only resulted in a low molecular weight smear of HA similar to that of ΔhylAΔhylB carrying pAOJ84 alone (Figure 5E), indicating that HylA does not provide any additional depolymerization of HA digested by HylB.
A puzzling observation was that all E. faecalis strains including ΔhylAΔhylB partially depolymerized HA, resulting in a large smear that ran lower on the agarose gel than the BHI negative controls at either neutral pH or pH 5.7 (Figure 5E). This observation suggests that E. faecalis V587 has an additional mechanism to partially degrade HA that would not have been apparent in the semi-quantitative assay (Figure 5B) as large fragments would still precipitate and appear as undigested HA. To determine if E. faecalis exported protein was responsible for the observed basal level of HA degradation, HA was incubated with cell-free supernatants of E. faecalis ΔhylAΔhylB carrying pAOJ84 (hylB), pAOJ86 (hylA), or pAOJ55 (vector control) that had been filter sterilized and concentrated using a 10 kDa molecular weight cutoff filter (Figure 5F). Concentrated supernatants from the strain constitutively expressing hylB showed partial digestion of high molecular weight HA, possibly suggesting that continual production of HylB is required for reaction completion. However, no HA degradation was observed for any of the other strains, indicating that the partial HA degradation that occurs during incubation with live bacteria is not mediated by a soluble, secreted product ≥10 kDa.
Intrigued by this result, we searched the genome of V587 and the related V583 for other potential GAGases and noted the presence of a putative heparinase, EF2268, that could be responsible for the weak digestion. No signal peptide was predicted in the EF2268 amino acid sequence by SignalP 5.0, which would align with our observation that HylB was the only large, secreted product that depolymerized HA (Figure 5E). We therefore deleted ef2268 from the ΔhylAΔhylB strain background to generate a triple GAGase mutant. Surprisingly, partial depolymerization was still observed during incubation with the triple mutant, demonstrating that EF2268 is not responsible for the basal HA degradation (Supplemental Figure 2).
HylB purified from E. coli retains hyaluronidase activity
To characterize the activity of purified HylB enzyme, we cloned hylB into the E. coli expression plasmid pBAD/myc his A, transformed the plasmid into E. coli BL21(DE3) pLysS, and purified the recombinant HylB-Myc-6x His protein. The extracted protein was of high purity (Supplemental Figure 3). While E. coli BL21(DE3) pLysS is not suspected to encode any enzymes that could degrade HA, we used lysate from a culture of E. coli expressing unrelated proteins in a pBAD/myc his A background to serve as a negative control. Hyaluronidase activity was measured in a turbidimetric assay designed for use with purified protein rather than live bacteria. In this assay, HylB-Myc-6x His achieved complete HA digestion in less than 15 minutes, compared with ∼40% HA remaining with bovine hyaluronidase at equimolar protein concentration (Figure 6A). Digestion product size was also examined by agarose gel electrophoresis. Incubation of a 601 kDa HA fragment with Streptomyces hyaluronidase (HylSh) completely digested all HA, such that no remaining product was visible on the gel (Figure 6B). Incubation with recombinant HylB-Myc-6x His resulted in near complete HA digestion with only a low molecular weight smear around ∼30 kDa remaining, mirroring the results obtained with live E. faecalis cells (Figure 6B). Importantly, incubating HA with the E. coli protein lysate showed no digestion in either the gel or turbidometric assay (Figure 6A and B), confirming that all activity was derived from HylB-Myc-6x His. Combined, these data confirm HylB as a hyaluronidase, and demonstrate that purified HylB enzyme is sufficient for complete digestion of HA. While a similar approach was tried for expression and purification of HylA, all attempts were unsuccessful. Thus, the function of HylA remains to be defined.
E. faecalis does not use HA as a nutrient source during growth in vitro
Some bacteria can use HA as a nutrient source after breaking down the repeating polymers into N-acetyl-glucosamine and glucuronic acid monomers. For example, the hyaluronate lyase activity of S. pneumoniae allows this species to use HA as a carbon source in minimal media (21). We therefore sought to determine whether HA degradation by HylB could support growth of E. faecalis in a simplified defined media (SDM) (47) in which E. faecalis is unable to grow without a supplied carbohydrate source (47). E. faecalis V587 exhibited robust growth in SDM supplemented with glucose, but failed to grow in SDM with HA as the sole carbohydrate (Figure 7A). However, HylB may not be active in WT E. faecalis under these conditions, even when HA is supplied as the sole carbohydrate source. Since HylB-mediated degradation of HA was clearly observed in the semi-quantitative GAG degradation assay in 0.5X TSB, we next asked whether HA supplementation enhanced growth when HylB was constitutively expressed. V587, ΔhylAΔhylB, and ΔhylAΔhylB with pAOJ84 (constitutively expressing hylB) all grew similarly in 0.5X TSB (Figure 7B) and they all achieved stationary phase at a similar rate and density when supplemented with HA (Figure 7C). The addition of unfractionated HA caused a slight reduction in the cell density at which stationary phase was achieved for all strains. However, constitutive expression of hylB resulted in an increase in ΔhylAΔhylB culture density from 8-24 hours that was not observed for the other strains (Figure 7C), indicating that HylB activity either provides E. faecalis with an additional nutrient source or alleviates any inhibition imposed by the presence of high molecular weight HA.
HylA and HylB contribute to suppression of LPS-induced inflammation but do not dampen inflammation during CAUTI
Hyaluronidases from other Gram-positive bacteria have been shown to mediate inflammatory regulation (18). For example, S. agalactiae digests HA to disaccharides that facilitates pathogenesis by interfering with TLR-2/TLR-4 signaling (18) and subverting neutrophil and macrophage killing (19, 48, 49). Since E. faecalis is well-known for the ability to modulate the immune response, including suppression of lipopolysaccharide (LPS) stimulated inflammation (50, 51), we sought to determine whether HylA and HylB contribute to modulation of the innate inflammatory response using a well-established NF-κB macrophage-like reporter cell line, RAW-Blue. This cell line expresses a secreted embryonic alkaline phosphatase (SEAP) under the control of an NF-κB inducible promoter, allowing for a readout of NF-κB induction via an alkaline phosphatase assay.
To determine if HA disaccharides dampen NF-κB activation in this cell line, RAW-Blue cells were incubated for six hours with increasing concentrations of HA disaccharides, LPS, and each combination thereof (Figure 8A). Interesting, the HA disaccharides appears to be pro-inflammatory at 0 and 0.1 ng/ml of LPS, but dampened NF-κB induction at higher concentrations of LPS. While these effects were all modest, the data confirm that HA disaccharides can dampen LPS-stimulated activation of NF-κB in RAW-Blue cells.
We next examined NF-κB activation in response to infection. In preliminary experiments, we observed that infection with MOIs greater than 10 resulted in substantial acidification of the culture media before assay completion. Since low pH has been shown to interfere with the NF-κB pathway (52–54), an MOI of 10 was chosen for all experiments. RAW-Blue cells were incubated for six hours with wild-type E. faecalis V587, ΔhylA, ΔhylB, ΔhylAΔhylB, ΔhylAΔhylB with pAOJ83 (constitutively expressing the full length HylA protein), or ΔhylAΔhylB with pAOJ84 (constitutively expressing the full length HylB protein) with or without 10 ng/mL LPS and 100 ug/ml of high molecular weight HA (∼700 kDa). In uninfected cells, LPS supplementation resulted in a ∼5-fold increase in NF-κB activation as expected while the addition of HA had no impact under any conditions (Figure 8B), confirming that 700 kDa HA does not alter stimulation in the absence of hyaluronidase activity.
Infection with any of the E. faecalis strains resulted in a >6-fold increase in NF-κB activation over DMEM alone (Figure 8C-H), clearly demonstrating that strain V587 is capable of inducing inflammation at an MOI of 10. The addition of HA during infection with V587 resulted in a slight though statistically-significant decrease in NF-κB activation, which was also observed during infection with ΔhylB or ΔhylAΔhylB constitutively expressing the full length HylA protein but not during infection with any strain lacking hylA. These data suggest that HylA may provide a small degree of immune suppression in the absence of other stimuli.
Supplementation with LPS during infection resulted in much greater NF-κB activity for all strains compared to infection in DMEM alone (∼10-fold induction vs ∼6-fold), demonstrating that E. faecalis V587 does not suppress LPS-stimulated NF-κB activation under these experimental conditions. This is in contrast to prior reports and likely due to differences in MOI and subsequent media acidification. The addition of HA slightly mitigated NF-κB activation during infection, as HA+LPS resulted in a modest though significant decrease in NF-κB activation for all strains except the ΔhylAΔhylB double mutant. Since loss of either gene alone or complementation with either gene alone allowed for decreased in NF-κB activation, these data suggest that digestion of HA by either HylA or HylB is sufficient to modestly suppress LPS-stimulated inflammation.
We next sought to determine whether HylA or HylB contribute to immune modulation during CAUTI. Since individual mutants exhibited colonization defects 48 HPI in the CAUTI model, we conducted three additional infection studies with V587 and the ΔhylAΔhylB mutant to examine colonization at 6 HPI and two studies at 24 HPI (Figure 9A and B). No significant differences in colonization were observed at these earlier time points, allowing us to pinpoint the potential contribution of HylA and HylB to immune modulation.
For initial assessment of the innate immune response, bladder homogenates were pooled by experiment date and infection group, resulting in three biological samples per inoculum at 6 HPI and two biological samples at 24 HPI. Homogenates from mock-infected (PBS) catheterized mice were also included for each time point. Samples were then analyzed by Luminex to quantify the levels of 32 cytokines, chemokines, and growth factors, of which 14 were above the limit of detection (Figure 9C and D, Supplemental Figure 4). Unexpectedly, the global profiles of the catheterized mock-infected PBS mice were not significantly different from either infection group at either time point, suggesting that the majority of the inflammation was driven by the presence of the catheter. This, E. faecalis V587 does not appear to substantially dampen catheter-driven inflammation, and also does not further stimulate the immune response in this model.
To determine the potential contribution of HylA and HylB to inflammation, the profile of bladders from mice infected with V587 were compared to those infected with. Interestingly, the only analyte that was significantly different at any time point was a reduction in IL-6 at 6 HPI for mice infected with the ΔhylAΔhylB double mutant (P = 0.0004). This observation was unexpected, as we hypothesized that HylA and HylB would most likely dampen the immune response, resulting in an increase in pro-inflammatory cytokines during infection with the mutant strain. To confirm whether HylA and HylB might contribute to IL-6 induction, an additional set of mice were inoculated as above along with an extra control group lacking the catheter segment, and IL-6 was quantified by ELISA (Figure 9E). As expected, the presence of the sterile catheter segment alone resulted in ∼8-fold induction of IL-6 in the absence of infection, but no further increase in IL-6 was observed during infection with V587. These data confirm the hypothesis that the catheter segment drives the majority of inflammation in this model at 6 HPI. Further, IL-6 levels were significantly lower in mice inoculated with the ΔhylAΔhylB double mutant compared to V587, confirming that HylA or HylB contribute to stimulation of IL-6. It is important to note that the effect is modest at 6 HPI and appears to be resolved by 24 HPI, and likely does not substantially contribute to the differences in dissemination that are observed in this model. In summary, HylA and HylB may contribute to immune modulation but do not appear to be the primary mediators of immune suppression during CAUTI.
Discussion
Enterococcus faecalis infections are very common and impose a high burden on the healthcare system, ranking as the 5th most commonly encountered healthcare acquired pathogen in the United States overall, both in pediatric patients and adults (1, 2). E. faecalis can cause a variety of infections, including CAUTI and CLABSI (2, 3). In this study, we revealed HylA and HylB as two new virulence factors for CAUTI and bacteremia caused by vancomycin-resistant E. faecalis.
Even though we detected hylA and hylB mRNA transcripts in WT E. faecalis V587 during growth in vitro, we did not detect any hyaluronidase activity in the WT strain. These data confirm prior work indicating a lack of HA degradation by E. faecalis. However, by cloning hylB under a constitutive promoter in E. faecalis strain V857 we conclusively demonstrated that HylB (EF0818) is a bona fide hyaluronidase with substantial chondroitinase activity. This enzyme appears to act specifically on HA and CS, as we did not detect any evidence of HS digestion. We further demonstrated that expression and purification of HylB in E. coli also provides robust hyaluronidase activity, showing that HylB is both necessary and sufficient to digest HA. As the amino acid sequencing of HylB in E. faecalis V587 shows low identity with previously characterized Gram-positive bacterial hyaluronidases (55–59), structural studies of this enzyme could provide critical information for GAGases in general.
In contrast to HylB, the only in vitro evidence of enzymatic activity for HylA (EF3023) was a very subtle degradation of HS and a potential role in dampening NF-κB activation in the presence of HA. However, hylA contributed to bladder colonization and dissemination to the bloodstream in the mouse model of CAUTI and appeared to provide functional redundancy to hylB in the bacteremia model. There are many potential explanations for the discrepancy between the in vitro and in vivo phenotypes of HylA. One possibility is that none of the HylA expression constructs resulted in properly folded or active enzyme. Alternatively, HylA may require processing by additional factors not present under our in vitro conditions to become a catalytically-active enzyme. For example, the HylB enzyme of S. agalactiae requires post-translational proteolytic cleavage to be active (60). Another possibility is that the substrate of HylA may be a specific glycosaminoglycan, glycoprotein, or sulfonation pattern that was not directly tested in this study but is present in the urinary tract and circulatory system. This would not be the first time that a predicted bacterial hyaluronidase turned out to have a different target substrate; Streptococcus pyogenes Spy1600 was initially labeled as a possible hyaluronidase but experimentally confirmed to act as a β-N-acetylglucosaminidase (61). Experimentation with additional host polysaccharides and expression constructs will be needed to either confirm or disprove HylA as a glycosaminoglycan degrading enzyme.
While some bacteria such as S. pneumoniae can use HA as a carbon source (21), this was not the case for E. faecalis in vitro. However, constitutive expression of hylB did allow the ΔhylAΔhylB double mutant to achieve higher cell density during HA supplementation than even the WT strain. Considering the proximity of hylB to a putative oligosaccharide PTS system, there is likely additional regulation that must occur for E. faecalis to potentially use HA degradation products as a nutrient. Our data also do not rule out a possible role for HA catabolism to fuel growth during infection.
The contribution of hyaluronidases to microbial pathogenesis has been studied in S. agalactiae in the context of pre-term birth (48, 49) and systemic infection (19). During systemic infection in mice, disruption of hyaluronidase by mutation of hylB decreased tissue colonization and lethality (19). In a nonhuman primate model, hyaluronidase activity decreases neutrophil bactericidal activity by interfering with TLR2/TLR4 signaling, and a hylB mutant deficient in hyaluronidase activity was deficient in establishing fetal invasion and bacteremia (48). In mice, the hylB mutant colonized the vaginal tract to a similar level as the parental isolate but exhibited defects in ascending infection and in the immune-subversion (19). Vaginal colonization has also been examined in E. faecalis strain OG1RF using a genome-wide transposon mutant screen for factors that allow E. faecalis to persist within the vaginal tract (62). Intriguingly, the homolog of hylB in this strain (OG1RF_10550) exhibited reduced colonization at all time points (62).
However, OG1RF_10550 was just one of many genes identified as important for vaginal colonization in this study, and we are unaware of any follow-up studies on this gene. Considering that our studies demonstrate that E. faecalis HylA primarily contribute to bladder colonization and dissemination to the bloodstream while HylB contributes to kidney colonization and almost 50% of E. faecalis isolates possess homologs of both genes, expression and activity are likely highly regulated. This is supported by the recent observation that HylA expression is regulated by a BglG/SacY antiterminator homolog in E. faecalis strain V19, a plasmid-cured derivative of V583 (63). Further exploration into the conditions and signaling pathways required for each to be expressed and active is expected to provide further insight into how E. faecalis is able to adapt to many different infection niches.
Hyaluronic acid also modulates inflammatory signaling, with the outcome being largely dependent on HA fragment size. Hyaluronidases from S. agalactiae and other Gram-positive pathogens were shown to digest HA down to disaccharides, which block TLR-2 and TLR-4 signaling (18). In contrast, Streptomyces hyaluronidase produces HA oligosaccharides that stimulate inflammation, as do mammalian hyaluronidases (18). Our data indicate that infection of RAW-Blue cells with E. faecalis V587 activates NF-κB regardless of condition or infecting strain, but constitutive expression of HylB can modestly counteract LPS-stimulated activation of NF-κB. Thus, HA degradation by E. faecalis V587 could contribute to immune modulation during infection.
We also examined a broad array of cytokines, chemokines and growth factors in the bladders and the kidneys of mice infected via our CAUTI model. Strikingly, we did not observe drastic differences between infected and mock infected catheterized mice, which led us to confirm that the majority of inflammation during CAUTI with E. faecalis V587is due to the insertion of the catheter itself. We also did not observe dramatic difference in immune response between infection with wild-type V587 and the ΔhylAΔhylB strain, other than a slight reduction in IL-6 in the bladders of mice infected with ΔhylAΔhylB at 6 HPI. Infection with E. faecalis as well as catheter implantation are both known to stimulate IL-6 (64–66), so this observation may suggest a role for HylA or HylB in tissue invasion and damage.
In summary, this study highlights the importance of in vivo validation of in silico functional predictions. We have confirmed HylB as a hyaluronidase/chondroitinase enzyme in vancomycin-resistant E. faecalis, despite having low amino acid identity to other bacterial hyaluronidases. We have further demonstrated that active HylB can be produced in vitro, both by E. faecalis and by E. coli, and that hylA and hylB both contribute to E. faecalis pathogenesis. Further examination of the substrate and functions of HylA and the distribution of its target in mammalian hosts may reveal novel host-pathogen interactions or new information about GAG distribution in mammalian tissues.
Materials and Methods
Animal model ethics statement
Animal protocols were approved by the Institutional Animal Care and Use Committee of the University at Buffalo (MIC31107Y), in accordance with the Office of Laboratory Animal Welfare, the U.S. Department of Agriculture, and the Association for Assessment and Accreditation of Laboratory Animal Care. Mice were anesthetized with a weight-appropriate dose (0.1 ml for a mouse weighing 20 g) of ketamine/xylazine (80 to 120 mg/kg ketamine and 5 to 10 mg/kg xylazine) by intraperitoneal injection and euthanized by CO2 with vital organ removal.
Bacterial strains and culture conditions
Vancomycin resistant Enterococcus faecalis strain V587 (NR-31979) (31) was obtained through BEI Resources, NIAID, NIH. This strain was chosen for its clinical history that mirrors our mouse infection models, and for containing a large number of known E. faecalis virulence genes (67), including Enterococcal Surface Protein (esp) previously implicated in urinary tract colonization (68). E. faecalis V587 and isogenic mutants were routinely cultured from frozen glycerol stocks in 5 ml of Brain Heart Infusion (BHI) broth (Dot Scientific) or Tryptic Soy Broth (TSB) with 500 µg/mL gentamicin sulfate (ACROS Organics, catalog # 455310250) in tightly capped 14-ml polypropylene culture tubes at 37°C with shaking at 300 RPM. E. faecalis harboring plasmids were cultured with 20 µg/mL chloramphenicol (Research Products International, catalog # C61000) instead of gentamicin.
RNA Extraction and qRT-PCR
E. faecalis V587 was grown overnight in BHI plus gentamicin 500 µg/mL, 0.6 mL aliquots of the overnight culture were centrifuged and washed once with PBS (pH 7.4), the cell pellet was resuspended in 0.6 mL of PBS, then diluted 1:100 in one of the following media: BHI, 0.5X TSB (TSB diluted to 50% concentration with sterile water), 0.5X TSB with 2.5 mg/mL ∼5 kDa size fractionated HA (Lifecore Biomedical, catalog # HA5K), 0.5X TSB with 2.5 mg/mL unfractionated HA sodium salt from Streptococcus equi (Sigma-Aldrich), filter sterilized human urine (Lee BioSolutions catalog # 991-03-P-FTD, lot # 01J5567), human urine with 2.5 mg/mL ∼5 kDa size fractionated HA, or human urine with 2.5 mg/mL unfractionated HA. 3 mL cultures were used for samples with rich media as the base, and 6 mL cultures were used for the samples grown in urine to account for lower growth density. Cultures were incubated with shaking for 4 hours at 37°C, pelleted by centrifugation, washed once with 1 mL of TE buffer (pH 8), then suspended in 50 µl RLT Buffer with BME (Qiagen RNeasy Mini Kit). 1.5 mL safe-lock microfuge tubes (Eppendorf) were filled halfway with 0.5mm glass disruption beads (RPI research Products International), the suspended pellet was added to the bead tube, and samples were homogenized using a Bullet Blender Gold (Next Advance) at max speed for 5 minutes.
After bead beating, 250 µl of RLT Buffer with BME was added and samples were vortexed to mix. A small hole was punched in the bottom of the tube with an 18G needle (BD Precision Glide), and the tube was immediately placed inside a 1.5 mL microfuge tube to collect the homogenate while leaving the beads behind. The collected homogenate was centrifuged at 21,000 rcf for 10 minutes at 4°C, supernatant was removed, and 1 volume of 70% ethanol was added and vortexed to mix. The sample was then transferred to an RNeasy Mini spin column for on-column DNase digestion kit (Qiagen), and RNA was eluted with 40 µl of RNase free water. A second DNase digestion was then performed off-column by adding 0.1 volume of 10X DNase buffer and 1 µl of DNase (Invitrogen) and incubating at 37°C for 30 minutes. The digestion was terminated by adding 0.1 volumes of DNase inactivation reagent, incubating at room temperature for 2 minutes, and centrifuging for 90 seconds at 8000 rcf. cDNA was synthesized using the iScript cDNA Synthesis Kit (BioRad) and qRT-PCR was performed using PCRBio SyGreen Blue Mix Lo-Rox (PCR Biosystems) and a BioRad CFX-Connect Real Time system. Data were normalized to recA as the reference gene and analyzed via the ΔΔCT method.
Generation of E. faecalis deletion mutants
Markerless deletion mutants of E. faecalis V587 were constructed by allelic exchange using the plasmids designated in the Supplemental primer table. Roughly 1,000 base pairs upstream and downstream of the desired deletion location were amplified via PCR and assembled into allelic exchange vector pIMAY (Addgene plasmid # 68939, gifted by Tim Foster) (69) via either restriction digest and ligation or NEBuilder HiFi DNA Assembly Mastermix (NEB), as indicated in the Supplemental vector table. Assembled vectors were propagated in E. coli Top10 and verified a combination of restriction digest and sequencing (Plasmidsaurus).
Plasmids were electroporated into E. faecalis, recovered and plated at 30°C, and isolated colonies were struck onto pre-warmed BHI-chloramphenicol plates and incubated overnight at 37°C to induced plasmid integration (70). Chloramphenicol-resistant colonies were then struck on BHI agar containing 1 µg/mL anhydrotetracycline hydrochloride (Cayman Chemical catalog # 10009542) to induce expression of the counter selection marker, and incubated overnight at 30°C. Large, isolated colonies from these plates were then re-struck on anhydrotetracycline BHI plate and incubated at 37°C overnight to ensure loss of the plasmid backbone. Mutants were PCR verified by checking for the presence of the desired mutation, the absence of the original gene, the absence of the pIMAY backbone, and retention of natural V587 plasmids using the primers described in the Supplemental primer table. The esp gene was also amplified using esp11 and esp12 from a previous publication (71). Verified colonies were cultured in BHI-gentamicin broth overnight to generate glycerol stocks and PCR-verified a second time. The ΔhylAΔhylB double mutant was generated by sequential deletion of hylB from the ΔhylA strain using the method described above, and was verified by PCR as described above and by sequencing of ∼1200 base pairs up and downstream of each gene (Eton Biosciences).
Hyaluronidase and Chondroitinase plate assay
HA plates were made using a published recipe (18). A hyaluronic acid solution was first generated by suspending 300 mg unfractionated HA sodium salt and 8 g of bovine serum albumin fraction V at pH 5.2 (Sigma-Aldrich) in 200 mL of water and slowly adjusting to pH 7.5 using 0.1 M sodium hydroxide. The HA solution was then filter sterilized and warmed to 37°C. The agar base (10 g Noble agar, 10 g yeast extract, 30 g Todd-Hewitt broth powder, and 800 mL of RO water) was autoclaved, cooled to ∼56°C, supplemented with the warmed hyaluronic acid solution, and 16 mL was distributed into petri dishes. Once cooled, plates were inoculated with 10 µL of overnights cultures of E. faecalis or L. lactis and incubated overnight at 37°C. After overnight incubation, plates were cooled to ambient temperature and flooded with 2 M acetic acid to precipitate high molecular weight GAGs, and incubated for at least 10 minutes. The acetic acid was then aspirated and the plates were imaged on an Axygen Gel Documentation System (Corning) with blue light illumination.
Mouse infection models
For establishment of CAUTI, 6-8 week old female CBA/J mice (The Jackson Laboratory, strain # 000656) were inoculated transurethrally with 50 µl of a suspension of 2 x 106 colony forming units (CFU)/mL in PBS of either wild-type V587, ΔhylAΔhylB, ΔhylA, ΔhylB, or mock infected using PBS. A 4 mm segment of silicone tubing was placed in the bladder during inoculation to simulate a urinary catheter, as previously described (40, 72). At 6, 24 or 48 hours post infection, mice were sacrificed, and the bladders, kidneys and spleens were homogenized in 5 mL tubes (Eppendorf) containing 500uL of 3.2mm stainless steel beads (Next Advance) and 1 mL of sterile PBS in a Next Advance 5 E Gold Bullet Blender. Organ homogenates were plated for CFU counts and an aliquot of each homogenate was flash frozen and stored at -80°C for cytokine and chemokine measurements as described below.
For establishment of bacteremia, 7 week old female CBA/J mice were inoculated via tail vein injection of 100 µl of a suspension of 5 x 108 CFU/mL of either wild-type V587, ΔhylA, ΔhylB or ΔhylAΔhylB. 24 hours post infection, the mice were sacrificed and livers, spleens and kidneys were harvested, homogenized and plated for CFU counts.
Construction of constitutive expression vectors for E. faecalis
pAOJ20, an E. coli-Gram-positive shuttle vector described previously (73), was utilized as the backbone for plasmids pAOJ83 (full length HylA), pAOJ84 (full length HylB), pAOJ86 (HylA[codons 1-1334], minus LPXTG domain), and pAOJ55 (vector control with a non-enzymatic 5’ fragment of unrelated protein AtlA (74, 75)). All plasmids have hylA or hylB expression driven by the constitutive ermB promoter (76). To construct plasmid pAOJ88 (HylA[codons Δ29-246, Δ1335- 1372], HylA minus the Discoidin/Big2 domains and the LPXTG domain), pAOJ86 was amplified via inverse PCR using phosphorylated primers AOJ_726 and AOJ_727, and circularized by ligation. Plasmid pAOJ90 (fusion construct with the HylB N-terminus [codons 1-215] and HylA C-terminus [codons 247-1334, Δ1335-1372]) was constructed by performing inverse PCR on pAOJ84 using primers AOJ_729 and AOJ_730 to obtain the plasmid backbone and first 216 codons of hylB, which was fused to codons 247-1334 of hylA amplified via AOJ_731 and AOJ_732 by digesting both DNA fragments with BspQI (NEB) and ligation using T4 DNA ligase.
Due to technical issues encountered while propagating the constitutive expression constructs in E. coli, all ligation products were transformed into Lactococcus lactis NZ9000 (MoBiTec catalog # VS-ELS09000-01) for plasmid amplification. L. lactis competent cells were made and transformed using the previously described Lithium acetate-Dithiothreitol (DTT) method (77). All constructs were sequence-verified via Plasmidsaurus before transformation into E. faecalis V587 ΔhylAΔhylB.
Secreted protein profiles of E. faecalis V587, ΔhylAΔhylB, and constitutive expression strains
E. faecalis overnight cultures were diluted 1:200 into 50 mL conical tubes containing 20 mL BHI. Cultures were further supplemented with 20 µg/mL chloramphenicol for ΔhylAΔhylB with pAOJ83, pAOJ84, pAOJ86, pAOJ87, pAOJ88, and pAOJ90, or 500 µg/mL gentamicin sulfate for V587 WT and V587 ΔhylAΔhylB. The tubes were tightly capped and incubated with shaking for 4 hours, at which point the cells were pelleted and the supernatant filter sterilized using a 0.22 µm PES filter. The 4 hour time point was chosen to balance bacterial cell density while limiting contact time with the broad-spectrum E. faecalis proteases GelE and SprE, which exhibit peak expression in late exponential phase (78). 20 ml of filter-sterilized supernatant was placed on ice to cool, mixed with 5 mL of cold 6.1 N trichloroacetic acid (TCA), and incubated at 4°C overnight for protein precipitation. Tubes were then centrifuged at 3000 RCF for 30 minutes, supernatants discarded, and protein pellets washed twice with 90% acetone saturated with Tris base and once with 100% acetone. Pellets were air-dried, resuspended in 0.5 mL of 2X Laemmli Sample Buffer (Bio Rad), and incubated at 95°C for 10 minutes. Samples were then cooled to room temperature and centrifuged at 21,130 RCF for 10 minutes to remove remaining insoluble material for better gel resolution. 10 µL of each sample was loaded into the lanes of a Bio Rad Any kD SDS-PAGE gel. Gels were washed and stained with Coomassie Brilliant Blue R-250 and destained with water/methanol/acetic acid 50:40:10 until protein bands were clearly distinguishable from background. Gels was photographed with a Nikon Z30 on white background. Photographs were adjusted using Adobe Photoshop 2022 for exposure, levels, hue, saturation and/or contrast until fainter bands, easily visible by eye, were also distinguishable from background in the captured image. These adjustments were applied to the entire photograph equally.
Semi-quantitative In vitro measurement of E. faecalis GAGase
A recently-published semi-quantitative turbidimetric assay of GAGase activity was used to examine the profiles of wild-type E. faecalis V587, ΔhylAΔhylB, and ΔhylAΔhylB with each constitutive expression construct (29). E. faecalis strains for GAG degradation assays were cultured overnight at 37°C in 3 mL of BHI with 20 µg/mL of chloramphenicol for plasmid containing strains. Cultures were then diluted 1:4 in BHI, adjusted to an OD600 nm of 0.04, pelleted at 6010 RCF for 8 minutes, and the supernatant was removed. Cell pellets were resuspended in 1 mL of 0.5X TSB containing 2.5 mg/mL of either hyaluronic acid (Sigma-Aldrich, catalog # 53747), chondroitin sulfate A (Sigma-Aldrich, catalog # C9819), or heparin sodium salt (Fisher BioReagents, catalog # BP2425), and 300 µL aliquoted in triplicate into 96-well plates. Plates were incubated in a hypoxia chamber for 24-hours at 35°C with 10% CO2 and 5% O2, followed by a semi-quantitative GAG degradation measurement as previously described (29). Briefly, the OD600 nm of all replicates was measured, bacteria were pelleted by centrifugation at 3214 RCF for 10 minutes, 180 µL of supernatant was transferred from each well to a new 96-well plate, and supernatants were then serially diluted in 1X PBS. 10 µL of 10% bovine serum albumin (BSA) was then added to all wells and OD600 was measured. Finally, 40 µL of 2M acetic acid was added to all wells, mixed by gentle stirring with the pipette tip, and OD600 was measured again. The percentage of GAG remaining was calculated by comparing the post-acetic acid OD600 values of all experimental conditions to bacteria-free controls (29). Percentage GAG remaining was calculated using the following formula using the OD600 reading of the post-acetic acid serial dilution falling in the center of the linear range of the assay for each condition.
Agarose gel electrophoresis for determination of GAG degradation product size
100 µL of indicated E. faecalis overnights were diluted into 20 mL of BHI containing appropriate antibiotics and 1 mg/mL HA (Sigma-Aldrich), incubated for 24 hours at 37°C with shaking, pelleted, and supernatants were filter sterilized. Two uninoculated HA BHI tubes with 20 µg/mL chloramphenicol were incubated the same way was included as a negative control, one of which had no pH adjustment, and one was acidified to pH 5.7 using 85-90% lactic acid (BeanTown Chemical) to simulate E. faecalis fermentation. Hyaluronidase positive controls were made by incubating HA BHI with either 12.5 µg/mL Bovine Hyaluronidase (Sigma-Aldrich) or Streptomyces hyalurolyticus hyaluronidase (Sigma-Aldrich) at 3 units/mL. Agarose gel electrophoresis was followed using a modified protocol from Echelon Biosciences (79) to determine HA size ranges. 750 µL of each supernatant was added to 250 µL of loading buffer (0.02% Bromophenol blue, 2 M sucrose in 1X Tris Borate EDTA (TBE) buffer), 10 µL of which was added to each well of a 2% (w/v) agarose gel and electrophoresis was performed at 100 V in TBE buffer. Gels were washed with 30% (v/v) ethanol for ∼15 minutes to fix the HA fragments within the gel prior to staining, and then washed in 50% (v/v) ethanol for ∼1 hour to remove the aqueous buffer as Stains-All dye has low water solubility. Gels were transferred to 0.01% Stains-All in 50% ethanol (Alfa Aesar) overnight, then destained with 50% ethanol until bands were clearly distinguishable from background staining. Gels were imaged on a ChemiDoc MP (BioRad) using the DyLight 680 setting.
The gel assay was also utilized to determine what HA size ranges were produced by recombinant HylB. For these experiments, 1 mg/ml of 601 kDa HA (Echelon Biosciences, catalog # HYA-601KEF-1) was incubated with either 27.5 µg/mL of recombinant HylB, 3 units/mL of Streptomyces hyaluronidase, or 50 µg/mL E. coli protein overnight at 37°C, and gels were prepared as above.
Production of E. faecalis concentrated cell free supernatants for determination of GAG degradation by secreted products
To collect E. faecalis protein supernatants, 50 ml conical tubes containing 20 mL of BHI were inoculated with 100 µL of ΔhylAΔhylB with pAOJ86, pAOJ84, or pAOJ55 and incubated for 4 hours with shaking at 37°C. After incubation, cultures were centrifuged and supernatants were filter sterilized and concentrated using a 10 kDa molecular weight cutoff filter. Retentates were then mixed with an equal proportion of 2 mg/mL HA in 20 mM sodium phosphate (pH 7) with 77 mM sodium chloride. A blank containing only the HA in phosphate buffer mixed 1:1 with PBS was also used. All samples were incubated for 24 hours statically at 37°C then subjected to agarose gel electrophoresis as described above.
Cloning, expression, and purification of HylB Myc-6x His
The HylB expression vector pAOJ66 was generated by amplifying the ef0818 gene with primers AOJ_651 and AOJ_652 and cloning into NcoI-HF/ApaI digested pBAD/myc his A (Invitrogen). The vector was propagated in E. coli TOP10, then transformed into E. coli BL21(DE3) pLysS (80) and cultured overnight in 50 mL LB with 100 µg/mL carbenicillin (Research Products International catalog # C46000) and 20 µg/mL chloramphenicol. The following day, 10 mL aliquots were sub-cultured into four 2 L flasks containing 900 mL pre-warmed Terrific Broth (81) supplemented with 10 mM of MgCl2 and 100 µg/mL carbenicillin. Flasks were incubated with shaking at 37°C until the OD600 reached 0.4-0.5 (∼3 hours post inoculation), after which 100 mL of 200 mM arabinose was added to each flask (20 mM final concentration) to induce protein expression (82). Cultures were incubated for 4 hours at 37°C with shaking, centrifuged at 8000 RCF for 10 minutes, and pellets were stored at -80°C overnight.
Cell pellets were thawed on ice, suspended in 200 mL total volume of lysis buffer (PBS pH 7.4 with 35 µM polymyxin B, 2% Triton X-100, 2 SIGMAFAST EDTA-free protease inhibitor tablets (Sigma-Aldrich, catalog # S8830-20TAB), 20 µL Benzonase nuclease (Sigma-Aldrich), and 50 mg of egg white lysozyme (Gold Biotechnology)), then incubated at 37°C for 30 with gentle agitation to lyse the cells. Lysates were centrifuged at 8000 RCF at 4°C for 30 minutes to remove insoluble material. The supernatant was filtered through a 0.22 µm PES membrane, and mixed with cold ammonium sulfate solution in PBS (pH 6) to a final concentration of 20% (v/v) ammonium sulfate. Samples were then centrifuged at 14500 RCF for 45 minutes at 4°C to remove higher molecular weight contaminants, supernatants were mixed with ammonium sulfate solution to a final concentration of 75% (v/v), and protein was precipitated overnight at 4°C. The next day, supernatants were centrifuged for 30 minutes at 14500 RCF at 4°C, the resulting protein pellet was dissolved in 100 mL of cold PBS pH 7.4 with 10 mM imidazole, and passed through a 3 mL His-Pur Ni-NTA column (Thermo Scientific, catalog # 88226) according to manufacturer instruction. The column-bound protein was eluted with 10 mL of 250 mM of imidazole in PBS pH 7.4, washed with PBS on a 10 kDa molecular weight cutoff column to remove imidazole, further washed with PBS + 5% glycerol, and the retentate was snap frozen and stored at -80°C.
Concentration and purity of recombinant protein was estimated using the Blue Dry Western technique (83) via a Trans-Blot Turbo RTA Transfer Kit, PVDF (Bio Rad, catalog # 1704272), and Trans-Blot Turbo transfer apparatus (Bio Rad) on the “large protein” pre-programmed setting. The blot was stained with the primary mouse α-c-Myc Monoclonal 9E10 antibody (Invitrogen, catalog # 13-2500) and Goat anti-mouse IgG1 Cross-Adsorbed Horseradish Peroxidase secondary antibody (Invitrogen, catalog # A10551), imaged colorimetrically (for Coomassie Brilliant Blue R-250), treated with Clarity Western ECL Substrate (Bio Rad, catalog # 170-5060), and imaged for chemiluminescence on a Bio Rad Chemi Doc MP imager. The chemiluminescence and colorimetric images were overlaid with the multi-image setting on the Chemi Doc MP to find bands contained the c-Myc tag of the recombinant protein. Densitometry was performed using Adobe Photoshop on raw colorimetric images to determine protein concentration by comparison to a Bovine Serum Albumin (BSA) dilution series of known concentrations loaded on the same gel.
E. faecalis growth curves
Growth in rich medium of WT V597, ΔhylA, ΔhylB, ΔhylAΔhylB, or ΔhylAΔhylB (pAOJ84) was measured by diluting overnight cultures 1:100 in BHI, TSB, or 0.5X TSB, aliquoting into a 96 well flat bottom plate, and incubating for 18-24 hours at 37°C with double-orbital shaking in a Synergy H1 plate reader (BioTek). Where indicated, media were supplemented with 2.5 mg/mL unfractionated HA or 2.5 mg/mL ∼5 kDa HA. Cell density was measured at 600 nm every 15 minutes.
To determine if E. faecalis V587 could use HA as a sole source of carbon, the Simplified Defined Media (SDM) recipe was used (47). Overnight cultures of E. faecalis V587 were centrifuged, washed once with an equal volume of PBS to remove traces of BHI, and diluted 1:100 in SDM with one of the following carbohydrate sources: 12 mM glucose (21), a two fold dilution series of HA (Sigma-Aldrich) starting at 2.5 mg/mL, or water (no carbon source). Growth was assessed as above.
Growth in human urine was assessed as previously described (40). Briefly, overnight cultures of E. faecalis were washed with PBS and diluted to an approximate cell density of 2 x 107 CFU/mL in sterile, pooled urine from healthy female donors (Lee BioSolutions catalog # 991-03-P-FTD, lot # 01J5567) and incubated at 37°C with shaking. CFU/mL was determined by dilution and plating every hour for 7 hours.
RAW-Blue NF-κB reporter assay
NF-κB activation by E. faecalis was assessed using RAW-Blue cells as previously described (50). RAW-Blue cells were obtained from InvivoGen (catalog # raw-sp) and routinely cultured at 37°C in a 5% CO2 in pre-warmed Dulbecco’s Modified Eagle’s Medium (DMEM, Corning catalog # 10-017-CM) with 4.5 g/L glucose, 0.584 g/L L-glutamine, 10% fetal bovine serum (Sigma-Aldrich), and no pyruvate. For each experiment, low-passage vials of cells were thawed and incubated in Corning Falcon 75 cm2 vented culture flasks at a density of 2 x 106 / 15 mL of DMEM with 100 µg/mL Normocin (InvivoGen catalog # ant-nr) for approximately 72 hours. Culture media was then removed, cells were dislodged with a cell scraper and suspended in 10 mL of DMEM with 200 µg/mL Zeocin (InvivoGen catalog # ant-zn), counted on a Beckman Coulter Z2 Particle Counter and Size Analyzer set to a 10-20 µm size range, diluted to 5 x 105 cells/mL in DMEM with 200 µg/mL Zeocin and 200 µL (1 x 105 cells), and 100 µl were seeded into a Corning Falcon 96 well flat bottom polystyrene tissue culture treated plate and incubated overnight. The following day, the media was removed, the RAW-Blue cells were washed once with Dulbecco’s Phosphate Buffered Saline (DPBS) and then incubated with either LPS and/or HA disaccharide sodium salt (Biosynth catalog # OH11537) or infected with E. faecalis at a multiplicity of infection (MOI) of 10 (1 x 106 CFU/well) in either DMEM or DMEM plus additives. MOI 10 was chosen as values higher than 10 were observed to acidify the media in a dose-dependent manner, low extracellular pH has been shown by to interfere with the NF-κB pathway (52–54). For infection experiments, media conditions used were: DMEM, DMEM with 100 µg/mL HA sodium salt at a molecular weight range of 500-749 kilodaltons (Lifecore Biomedical, catalog # HA700K), DMEM with 10 ng/mL ultrapure lipopolysaccharide (LPS) from E. coli O111:B4 (InvivoGen, catalog # tlrl-3pelps), or DMEM with both HA and LPS.
For all experiments, 200 µL of the treatment media or inoculum was added to the washed RAW-Blue cells and incubated for 6 hours. The plate was then centrifuged at 3000 RCF for 10 minutes, and 20 µL of supernatant was added to 180 µL of QUANTI-Blue Alkaline Phosphatase Detection Medium (InvivoGen catalog # rep-qbs), incubated at 37°C with shaking, and OD640 was read every 15 minutes for up to 16 hours to identify the optimal linear range. Under our experimental conditions, a 2 hour incubation was found to produce reads near the end of the linear range of the reporter assay, so this time point was used for reporting NF-κB activation.
Analysis of cytokines, chemokines and growth factors from mouse bladders and kidneys
Flash frozen tissue homogenates were thawed on ice and pooled by infection date and inoculation group (WT, ΔhylAΔhylB, or mock infected with PBS). Pooled samples were extracted using the Thermo Scientific Tissue Protein Extraction (T-PER) reagent bundle (Fisher Scientific catalog # B90002). 600 µL of total pooled sample volume was lysed with 700 µL T-PER lysis buffer with 1 Pierce Protease Inhibitor Mini Tablet per 7 mL of T-PER lysis buffer. Vortexed samples were centrifuged for 5 minutes at 10,000 RCF at 4°C and the supernatant removed. The protein concentration of 20 µL of supernatant was determined with Pierce Rapid Gold BCA Kit. Individual pooled samples were diluted to obtain 10 µg total protein in 50% T-PER buffer, 50% PBS.
Quantification of 32 different analytes in the adjusted bladder and kidney pools was provided by the Flow and Immune Analysis Shared Resource at the Roswell Park Comprehensive Cancer Center. The following analytes were measured: Eotaxin, G-CSF, GM-CSF, IFNγ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13, IL-15, IL-17, IP-10, KC, LIF, LIX, MCP-1, M-CSF, MIG, MIP-1α, MIP-1β, MIP-2, RANTES, TNFα, and VEGF, using a commercially available kit (MCYTMAG-70K-PX32, Millipore Sigma, Burlington, Massachusetts). Data was acquired on a Luminex 200 instrument (Luminex Corporation, Austin, Texas). The experiment and instrument set-up were performed based on the manufacturer’s kit instructions. In brief, serially diluted standards were analyzed in duplicate wells, while the experimental samples were tested in single wells. The plate was incubated overnight with the multiplex beads on a plate shaker, in the dark, at 4°C and processed with the reporter reagents the next day, as per manufacturer’s instructions. Analyte concentrations were determined by extrapolating individual experimental fluorescence intensity values against each analyte’s standard curve using the BeadView multiplex data analysis software, version 1.0, (Upstate Cell Signaling Solutions, Lake Placid, New York). The analytical performance was checked using high concentration and low concentration quality controls provided with the kit of which the determined concentration for each analyte needs to be within the manufacturer-determined concentration range.
Radar charts were developed for select Luminex analytes to visualize the multivariate and multi-scale data simultaneously on a single plot. Separate plots were made for 6 hours and 24 hours using the same grid. The upper bound of the range (outer ring) for each variable was set as the log of the maximum average pg/ml values across replicates for V587, ΔhylA ΔhylB and PBS. The inner rings represent 25%, 50%, and 75% of the maximum, respectively Radar charts were developed in the R programming language using the “fmsb” packagev (85, 86).
IL-6 ELISA
A colorimetric Mouse IL6 ELISA kit (Novus Biologicals NBP1-92668) was used to measure IL-6 from mouse tissue homogenates following the manufacturer instructions. Briefly, the ELISA plate was washed, bladder homogenates were thawed on ice and gently vortexed, and 50 µl of homogenate was mixed with 50 µl of diluent. 50 µl of Biotin conjugate was added to all wells (samples, blanks, and standards), the plate was incubated at room temperature for two hours with shaking at 400 RPM, then washed 5 time with a multichannel pipette. Strep-HRP was added and incubated for 1 hour at room temperature with shaking at 400 RPM. The plate was then washed, developer was added, and OD620 was monitored every minute on a BioTeck Synergy H1 until Standard 1 reached OD 0.9. Stop solution was then added and absorbance was read at 450 nm.
Statistical analysis
Statistical analysis was performed using GraphPad Prism Software version 9. Significance was determined using a P < 0.05. All P values are two tailed at a 95% confidence interval. For colonization data in both the CAUTI and bacteremia model, we performed two types of statistical analysis to assess fitness of wild-type versus mutant strains. Data were first analyzed by Mann Whitney U-Test (PMW < 0.05). Bacterial infection data does not necessarily follow distributions that are simple to perform standard statistical analysis on (87). To mitigate this issue, in addition to these tests, contingency tables of values above and below certain thresholds were constructed and analyzed for statistical significance by Fisher’s exact test (PFe < 0.05). In the CAUTI model a threshold of 1 x 105 CFU per gram of tissue was chosen for the bladders / kidneys because it represented a middle point between the limit of detection (102 CFU) and the upper range of recovered CFU (108). The exception to this was at 6 HPI, in which the threshold was set to 1 x 104 CFU due to lower overall colonization at this time point. For incidence of bacterial dissemination to the blood from CAUTI, the threshold was set at the limit of detection in the spleen (200 CFUs). For the bacteremia model in which mice were inoculated via tail-vein injection, statistical analysis was performed as in the CAUTI samples, with the exception that a cutoff of 1 x 105 CFU per gram of tissue was used for all organs.
Statistical analysis of relative expression was performed with Wilcoxon signed rank test against a hypothetical value of 1. Analysis of growth curves, the NF-κB reporter assay with diHA, and the Luminex assay was performed using a two-way ANOVA for overall trends, with Dunnett’s Multiple Comparisons used to compare specific strains/conditions. The semi-quantitative GAG assays were analyzed with one-way ANOVA with Dunnett’s Multiple Comparisons.
Acknowledgements
We would like to thank Dr. Thomas Russo and members of his laboratory for helpful comments and critiques, as well as Dr. Elsa Bou Ghanem for her useful suggestions. This work was funded by the National Institute of Diabetes and Digestive and Kidney Diseases under award R01 DK123158 to CEA, by university start-up funds to CEA, and by the Welch Foundation, award number AT-2030-20200401 to ND. Cytokine and chemokine analysis was performed by the Flow and Immune Analysis Shared Resource at the Roswell Park Comprehensive Cancer Center (Supported in part by NCI Cancer Center Support Grant 5P30 CA016056 and NCI R50 R50CA211108). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.