Vaughan and coworkers (Vaughan et al., 2021) used published EQUC procedures (Hilt et al., 2014) to isolate urinary bacteria from catheterized urine specimens. Streak-isolated bacteria were typed by MALDI-TOF and then transferred to a set of liquid nutrient media: tryptic soy broth (TSB), Man Rogosa and Sharpe broth (MRS), and brain heard infusion broth (BHI). Inoculated media was statically incubated at 35°C for 24-72 hours with ambient or anaerobic atmospheric conditions. The isolates that did not produce enough biomass were inoculated in TSB supplemented with 5% sheep blood, cysteine (0.5 g/L), or other additives (hemin 5 mg/L, vitamin K 10 mg/L), or NYCIII medium ( Table S1 ) and preserved as above if visible biomass was obtained. The qualitative densities of the resulting cultures were recorded and cell cultures were saved in 14% (w/v) glycerol at -80°C in duplicate freezer stocks. Later experiments with preserved strains showed that growth at 37°C worked similarly to 35°C and was used from then on.

To confirm species of urinary lactobacilli, we isolated genomic DNA of each used strain from overnight MRS culture using the Qiagen PowerSoil kit. Then we amplified most of the 16S rRNA gene using classic primers 27F (5’-AGAGTTTGATCCTGGCTCAG-3’) and 1492R (5’-GGTTACCTTGTTACGACTT-3’) and Q5 polymerase mastermix (NewEngland Biolabs). The resulted PCR product was purified and submitted for Sanger sequencing (Genewiz).

Inhibition of a number of substances was measured via a well-diffusion assay using solid MRS agar plates. Lactobacilli strains were grown over two nights in 10 mL of MRS broth and uropathogenic strains were grown overnight in 4 mL of LB broth, both stationary at 37°C. Surface of the MRS agar plates was then inoculated with 100 μL of the uropathogenic culture. Wells were punched in the surface (Ø6 mm) and removed with sterile tweezers. 50 μL of the bacterial culture or filtrates to be tested were added to the wells. The plates were then grown overnight at 37°C, unless mentioned otherwise. To measure the size of the zones of inhibition, all plates were first imaged, and the zones were quantified by the graphical program GIMP (www.gimp.org) and analyzed in MS Excel.

As in the solid agar inhibition assays, lactobacilli strains were grown for two nights in 10 mL of MRS broth and uropathogenic strains were grown overnight in 4 mL of LB broth. The optical density as measured at 600 nm (OD₆₀₀) of the strains was then taken, and 20 μL of the uropathogen culture was placed into each of the tubes needed for the experiment. The OD₆₀₀ was then used to calculate the amount of lactobacilli culture needed to make specific ratios of uropathogen: lactobacilli and that amount of lactobacilli culture was added to their respective tubes. These tubes were then spun in a centrifuge at 3230 rcf for 10 minutes to form a cell pellet and the supernatant was poured off. The pellets were then resuspended in 10 mL of MRS broth and incubated at 37°C. Measurements of uropathogen density were taken via serial dilutions and colony counting and samples were taken at four time points after beginning incubation: 0 hours, 1 hour, 6 hours, and 24 hours. At each of these time points, 100 μL was taken from each tube and used to make serial dilutions ranging from undiluted to a dilution of 10^-7. From each of these, 2 μL was taken and plated in rows on LB agar plates which were then incubated for several hours at 37°C. The number of colonies in least diluted but still countable samples was then used to calculate the CFU/mL of E. coli in each tube.

In several experiments, cells were removed from the culture to isolate the effects of the cells from the effects of cell secretions in the media. This filtrate was made by first spinning a few milliliters of the culture in a centrifuge at 3230 rcf for 10 minutes. The supernatant was then poured into a syringe and pushed through a 0.22 μm PES filter into a sterile tube. This filtrate was then ready for use or further modification as needed for the experiment.

All experiments were done at least in triplicate, with well-diffusion assays being run with technical triplicates on three separate days to represent biological replicates. All p-values were determined by a two-tailed Student’s T-test to determine statistical differences between different groups. We used a significance level of 0.05 and p-values of 0.01 or less are marked as well.

A new urinary bacteria repository was created following the recent study by Vaughan and colleagues (Vaughan et al., 2021). In that study, asymptomatic postmenopausal women provided catheterized urine that was then processed with 16S rRNA amplicon-based gene sequencing and EQUC. Three cohorts of women using vaginal estrogen therapy were tested: those without rUTI [noUTI]; those with rUTI [rUTI]; and those with rUTI and also taking daily prophylactic antibiotics as part of their treatment regimen [rUTI+ab]. Full details on these cohorts including exclusion criteria and demographic information are published elsewhere (see Tables 1 , 2 of the prior report) (Vaughan et al., 2021). Bacteria isolated using EQUC were typed using mass spectrometry (MALDI-TOF) showing that the collection consists of 64 microbial strains from 24 species of bacteria and two species of fungi ( Tables 1 and S1 ). A subset of lactobacilli was confirmed using Sanger sequencing ( Table S2 ). While in 2020 the Lactobacillus genus was reclassified into 25 separate genera (Zheng et al., 2020). We have representatives of three of them and will refer to them collectively as lactobacilli throughout the manuscript for simplicity.

Consistent with many urinary microbiome studies, almost half of this collection contains lactobacilli from eight species: Lactobacillus gasseri, Lactobacillus delbrueckii, Lacticaseibacillus rhamnosus, Ligilactobacillus animalis, Lactobacillus jensenii, Lactobacillus jonsonii, L. iners, and L. crispatus. Due to the nature of the EQUC procedure, all these strains are aerotolerant and our tests showed that the majority of them grow sufficiently well in the presence of oxygen but without active mixing. The exceptions were L. iners, some L. crispatus, and a few other bacterial isolates which required specialized media or were particularly sensitive to oxygen changes ( Table S1 ). Predictably, the rest grew well in MRS broth designed for lactobacilli growth.

It is worth noting that the repository also contains 5 uropathogen species derived in the same fashion from the same research participants: E. coli, K. pneumoniae, K. oxytoca, Proteus mirabilis, and E. faecalis ( Tables 1 and S1 ). As all samples were derived in the absence of acute UTI and at least 4 weeks after treatment for UTI (Vaughan et al., 2021), it is unclear if these strains were present in the setting of different host immune tolerance mechanisms and if these strains had adapted to survive within the normal urinary microbiome. It is also possible, as recently reported, that similar to the vaginal tract, low level presence of pathogenic bacteria in the bladder is common (Thomas-White et al., 2018; Garretto et al., 2020; Price et al., 2020b).

At the beginning of the EQUC collection, we established the whole genome sequences of two initial isolates of urinary lactobacilli – L. gasseri 5006-2 (GenBank Accession code JAGEKM000000000.1) and L. delbrueckii 5010-2. These two strains come from healthy, postmenopausal women and one with a history of rUTI, respectively. Preliminary genome analysis showed the presence of several bacteriocin genes. In light of these preliminary findings, we sought to assess whether major urinary pathogens can be inhibited by these urinary lactobacilli. We used model UTI strains of three major uropathogens: E. coli, K. pneumoniae, and E. faecalis ( Table 2 ) and tested the two sequenced lactobacilli strains for inhibition of pathogens using well-diffusion assays on MRS agar plates ( Figure 1A ).

Lactobacilli strains were first competed against three different uropathogenic strains of E. coli, CFT073, DS17, and UTI89, in a well-diffusion assay ( Figure 1A ) (Tullus et al., 1984; Mobley et al., 1990; Mulvey et al., 2001). Both lactobacilli strains inhibited the growth of all E. coli strains, though L. gasseri generally showed more inhibition than L. delbrueckii. One notable difference between the two strains can be seen in their inhibition when grown for 48 hours, rather than only 24 ( Figures 1B, C ). An L. gasseri culture grown for 24 hours seems to have slightly better inhibition of E. coli strains than those grown for 48 hours. Cultures of L. delbrueckii grown for 24 hours inhibited E. coli growth at comparable, albeit slightly lower, levels than the other lactobacilli strain; however, this inhibition dropped massively when cultures were grown for 48 hours, and in some cases nearly to zero. The timing of inhibition appeared to be sensitive to properties of the initial inoculum of L. delbrueckii strain.

The inhibition was next tested against other common UTI pathogens – K. pnemoniae and E. faecalis. For these tests, one strain of K. pneumoniae (TOP52) and four strains of E. faecalis (OG1RF, JH2-2, V583, and MHH59) were used ( Figures 1B, C ). Significantly, we observed some level of inhibition against all strains tested, even the Gram-positive E. faecalis strains ( Figure 1A ). The K. pneumoniae strain was sucessfully inhibited by both lactobacilli strains and the amount of inhibition observed was similar to assays performed using E. coli UTI model strains. L. gasseri cultures grown for both 24 and 48 hours were able to inhibit TOP52. Cultures of L. delbrueckii were able to inhibit TOP52 when grown for 24 hours but not when grown for 48 hours. The E. faecalis strains were not inhibited nearly as much as the gram-negative strains; however, they were all inhibited to some degree. L. gasseri cultures, in particular, were able to inhibit all E. faecalis strains to a limited degree after growth for 48 hours and all but OG1RF was inhibited after 24 hours of growth. The L. delbrueckii strain, however, was only able to inhibit two of the E. faecalis strains at a very low level when grown for 24 hours.

Such inhibition tests were then repeated with antibiotic resistant strains of E. coli (ESBL41, ESBL 146, ESBL146, and ESBL193 (Kanamori et al., 2017; Bethke et al., 2020). We determined that lactobacilli inhibition extended to antibiotic resistant E. coli as well ( Figure S2 ). Notably, tested model E. faecalis V583 and MHH95 are also multidrug resistant and can be inhibited by L. gasseri 5006-2 ( Figure 1 ).

Overall, these tests show that the two selected lactobacillus strains were able to consistently inhibit several model uropathogens, potentially indicating a general inhibitory mechanism. These preliminary observations allowed us to set up feasible methods to screen the majority of urinary lactobacilli from our repository under the same growth conditions.

Our initial tests showed that urinary L. gasseri 5006-2 and L. delbrueckii 5010-2 strains can inhibit uropathogens in vitro. Moreover, the strength of the inhibition was coincidental with the rUTI status of the patients from whom the strains originated ( Table S3 ). We therefore set to test other strains from the same species and five other lactobacilli species in our repository that can be grown under the same conditions. This selection of the lactobacilli species, that were grown under similar conditions, left out several strains of L. crispatus and one L. iners strain. Overall, we screened 19 lactobacilli strains grown for 24 and 48 hours in static MRS broth. The Streptococcus anginosus 5008-3 strain was used as a control for unrelated lactic acid bacteria that are also frequently found in the lower urinary tract ( Table 1 , S1 ). This was done by using a well-diffusion assay against 3 model uropathogens: E. coli CFT073, K. pneumoniae TOP52, and E. faecalis OG1RF ( Figure 2 ).

Our inhibition data show that the amount of inhibition varied more among different species of lactobacilli than it did among strains within the species. Unfortunately, with only one to five representatives from each species, it is impossible to generalize, apart from recording that the differences between strains and species in our subset are notable. For instance, the L. rhamnosus strains all have similarly sized, large zones of inhibition, especially towards the two gram-negative uropathogens. The L. jensenii strains did not show a strong inhibition under conditions tested with no detectable inhibition zone on the E. faecalis plate. The L. crispatus, L. gasseri, L. delbrueckii, L. jonsonii, and L. animalis strains all showed intermediate inhibition levels. It is noteworthy that testing 24 and 48 hr. old lactobacilli cultures ( Figure S3 ) showed significant differences in inhibition strengths, with L. crispatus, L. delbruecki and L. jensenii being particularly sensitive to lactobacilli culture growth phase.

As the isolates in our repository came from three different cohorts of patients ( Table S2 ), we compared the strength of lactobacilli inhibition and cohort status with the hypothesis that there would be greater inhibition in lactobacilli derived from healthy women without rUTI. The presented screen shows that there is no correlation between the patient cohort and the ability of EQUC isolated lactobacilli to inhibit uropathogens in vitro ( Figure 2 and Figure S4 ). No statistically significant differences among cohorts are observed for such interspecies or averaged inhibition against either of the three uropathogens ( Figure S4 ).

In summary, results show that inhibition is widespread across urinary lactobacilli species, but show heterogeneity among different uropathogens and urinary lactobacilli.

In some of the collected urines, the EQUC method isolated multiple species co-existing in the same sample, including both lactobacilli and uropathogens ( Table S1 and S3 ). As noted, none of the women who provided urine had acute symptomatic UTI and all underwent screening with dipstick urinalysis confirming the absence of pyuria prior to providing research specimens. Nevertheless, the in vivo coexistence of uropathogens with lactobacilli species seem to counter the simple hypothesis that presence of uropathogen-inhibiting lactobacilli should eliminate survival of uropathogens in the same niche. For this reason, we sought to test the inhibition of these pathobiont isolates with their cohabitating lactobacilli species to see if lactobacilli that are cohabitating with uropathogens are less antagonistic. To do this, we set up a broad screen to test each of the lactobacilli species isolated from these women ( Table S3 ) for inhibition of each of the pathobiont species isolated from that woman’s same sample ( Figure S5 ). Unfortunately, the K. pneumoniae isolates failed to form robust lawns on the test MRS agar plates and therefore, were not included in this cohabitant’s screen. Otherwise, all combinations of E. coli, E. faecalis, and K. oxytoca ( Figure S5 ) with lactobacilli were measured. To our surprise, most of the measured pairs showed some levels of inhibition, including strain combinations isolated from the same individual at the same time ( Figure 3 and Table S3 ).

We then compared the ability of cohabitating lactobacilli strains to inhibit model uropathogen E. coli or E. faecalis, strains of the same uropathogen species that coexisted with the given lactobacilli strain, or were isolated from other individuals. From this small subset it appears that the cohabiting lactobacillus strain did not inhibit uropathogen isolates as strongly as the inhibition noted previously with model uropathogens (OG1RF and CFT073, Figure 3A ). With no further genetic or pathogenicity information available for these new isolates, it is not clear if this is due to properties of the lactobacilli or uropathogen strain.

While our data on cross-competing novel clinical isolates against each other, cohabiting or not, it does not support the hypothesis that in vitro inhibition can predict in vivo behavior – it again shows how the inhibitory interaction is specific and sensitive to the identity of the lactobacilli used at a strain-level. Nevertheless, we do observe that some species tend to be better inhibitors on a relative scale, when compared to others. Specifically, L. delbrueckii 5010-2, L. delbrueckii 5028-1, and L. jensenii 5024-7 did not inhibit well in a screen with model uropathogens and they consistently exhibit less inhibition with uropathogen isolates derived clinically from patients (compare Figures 1 - 3 ).

The results of testing urinary lactobacilli and pathobionts found in the same urine sample indicate that antagonistic species might coexist in vivo.

While informative, simple inhibition on agar plates is not be reflective of the physiological conditions of the lower urinary tract. Therefore, we tested the interactions of urinary lactobacilli and UPEC in liquid MRS media, which might provide a better, though still not perfect, representation of bacterial suspension in urine within the lower urinary tract. We therefore co-cultured E. coli CFT073 with increasing amounts of L. gasseri 5006-2 or L. delbrueckii 5010-2 in liquid growth media.

The presence of either of these lactobacilli strains dramatically decreased the growth of the E. coli CFT073 in media ( Figure 4 ). This can be seen in the significantly reduced amount of E. coli after only 6 hours of growth together with either strain of lactobacilli. After 24 hours, both L. delbrueckii 5010-2 and L. gasseri 5006-2 were able to reduce CFT073 densities by two to five orders of magnitude. Cultures containing the L. gasseri 5006-2 strain, similar to results in the well-diffusion assays, were able to inhibit to a greater degree than L. delbrueckii 5010-2 cultures.

These experiments showed that the dynamics and strength of inhibition in liquid is different between the lactobacilli strains.

To determine how the inhibition took place, lactobacilli L. gasseri 5006-2 and L. delbrueckii 5010-2 cultures were split into two parts: a cell-free filtrate and a suspension of cells in fresh broth. The filtrate of both lactobacilli cultures could inhibit the growth of the E. coli CFT073 strain starting after about 12 hours of incubation together on a plate and lasting until about 24 hours of incubation together ( Figure 5 ). However, this inhibition seems to be somewhat transient as when plates were incubated for longer than 24 hours, the E. coli would begin overgrowing the zone of inhibition, causing the clearance zone to shrink until it vanished in ~ 4 days ( Figure S6 ). Zones from the resuspended cells, in contrast, differed markedly between the two lactobacilli cultures. Resuspended L. delbrueckii 5010-2 cells exhibited some inhibition, but usually not as strong as that of cell-free filtrate or L. gasseri 5006-2 strain ( Figure 5 ). Resuspended L. gasseri 5006-2 cells showed inhibition comparable to that of the respective whole culture. In addition, the zones of inhibition from resuspended cells of both strains did not change if the plates were incubated longer than the standard 24 hours while filtrate-induced inhibition zones were eventually overgrown by E. coli lawn ( Figure S6 ). This seems to indicate an additional mechanism of inhibition present in the L. gasseri cultures; one that is dependent on the presence of cells and which does not seem to be resulting from acidification. This seems to show that both lactobacilli species inhibit uropathogens by decreasing the pH of their environment and the L. gasseri strain may have a strong additional cell-based mechanism of action that is independent of a low pH.

The cell-free filtrates from both L. delbrueckii and L. gasseri cultures were then heated or adjusted to a neutral pH before use in a well-diffusion assay. The filtrate showed the same amount of inhibition when heated as it did when unmodified, indicating that the mechanism of inhibition is resistant to heat. The filtrate from both cultures that was adjusted to a neutral pH, conversely, showed no inhibition. In addition to neutralizing the filtrate, another experiment increased the buffering capacity of the medium used in the experiment by adding PBS to the MRS agar formulation. These plates showed significantly reduced inhibition by filtrates from both lactobacilli cultures. Finally, pH readings of the filtrate before use in well-diffusion assays seem to indicate that a low pH (~3.9-4.0) is needed for inhibition to occur. Taken together, this seems to indicate that the mechanism of inhibition in the cell-free filtrate is highly dependent on pH and is likely a result of the acidification of their environment by lactobacilli species.

One possible explanation for cell-dependent inhibition is the presence of biosurfactants. Several reports showed that phosphate-buffers can elute such surfactants into solution (Rodrigues et al., 2006; Gudiña et al., 2011; Morais et al., 2017). To test for this hypothesis, cells were incubated in PBS for varying lengths of time (1-24 hr) to isolate any biosurfactant molecules, then the PBS solution was filter-sterilized, and used in well-diffusion assays. However, the PBS incubated with either of the strains’ cultures was not able to inhibit CFT073, indicating a lack of inhibitory biosurfactants that can be eluted this way (data not shown).