Multispecies bacterial invasion of human host cells

Authors: Charlotte Abell-King, Alaska Pokhrel, Scott A Rice, Iain G Duggin, Bill Söderström

Abstract

Urinary tract infection (UTI), one of the most common bacterial infections worldwide, is a typical example of an infection that is often polymicrobial in nature. While the overall infection course is known on a macroscale, bacterial behavior is not fully understood at the cellular level and bacterial pathophysiology during multispecies infection is not well characterized. Here, using clinically relevant bacteria, human epithelial bladder cells and human urine, we establish co-infection models combined with high resolution imaging to compare single- and multi-species bladder cell invasion events in three common uropathogenic Escherichia coli (UPEC), Klebsiella pneumoniae and Enterococcus faecalis. While all three species invaded the bladder cells, under flow conditions the Gram-positive E. faecalis was significantly less invasive compared to the Gram-negative UPEC and K. pneumoniae. When introduced simultaneously during an infection experiment, all three bacterial species sometimes invaded the same bladder cell, at differing frequencies suggesting complex interactions between bacterial species and bladder cells. Inside host cells, we observed encasement of E. faecalis colonies specifically by UPEC. During subsequent dispersal from the host cells, only the Gram-negative bacteria underwent infection-related filamentation (IRF). Taken together, our data suggest that bacterial multispecies invasions of single bladder cells are frequent and support earlier studies showing intraspecies cooperation on a biochemical level during UTI.

Introduction

Urinary tract infections (UTI) are amongst the most common bacterial infections globally (Antimicrobial Resistance 2022). With an estimated 150 million people experiencing an UTI annually (Flores-Mireles et al. 2015). Clinically, UTI are predominantly attributed to only one species of bacteria at a time, with Uropathogenic Escherichia coli (UPEC) being the most commonly diagnosed infectious agent at a reported prevalence between 70 and 95% (Kucheria et al. 2005, Foxman 2014, Xia et al. 2017, Shah et al. 2019, Klein and Hultgren 2020). However, it is well established that there are a multitude of different bacterial species present in the bladder environment at any given time (Gaston et al. 2021), and it is known that in acute UTI, polymicrobial infections are common, with both Gram-negative and Gram-positive pathogens present (Stickler 2008, Croxall et al. 2011). Curiously, multispecies UTI are clinically underdiagnosed, especially if they contain Gram-positive species (e.g. enterococci), and are often disregarded as sample contamination (Kline and Lewis 2016). It was found that in UTIs from 80 female patients with cystitis, that UPEC was over twenty-fold more common than Enterococcus faecalis (Gram-positive cocci) as the causative agent of the infections (Rosen et al. 2007). E. faecalis and UPEC are often associated with each other during outgrowth of cultures from UTI (Fourcade et al. 2015, de Vos et al. 2017) and E. faecalis can suppress the immune activation, promoting UPEC virulence (Tien et al. 2017). Recent studies have investigated co-colonization in UTI models (Flores et al. 2023,, Jafari and Rohn 2023), but very limited information regarding bacterial invasion behaviours under liquid flow conditions simulating the bladder environment is available at high resolution at a single cell level.

Here, in an effort to simulate the infection environment we set up infection models where we use bacteria commonly associated with UTI, human epithelial bladder cells and human urine. We use two parallel approaches (‘flow’ and ‘dish’) to investigate differences in human epithelial bladder cell invasion rates of one, two or more bacterial species in the same infection experiment. The first approach is based on a flow channel model with constant exchange of nutrients (Iosifidis and Duggin 2020), while the second approach use glass bottom petri dishes under constant orbital agitation and manual exchange of media at fixed time points. Pathogens used in this study were Gram-negative Uropathogenic Escherichia coli (UPEC) and Klebsiella pneumoniae, as well as the Gram-positive Enterococcus faecalis, as they are the three most commonly found organisms in healthcare-associated UTIs (HAUTIs) (Tandogdu et al. 2016), and are commonly co-isolated in clinical settings (Galvan et al. 2016), especially in samples from catheter-associated urinary tract infections (Ganderton et al. 1992, Desai et al. 2001, Macleod and Stickler 2007, Wang et al. 2010). Challenging the human epithelial bladder cells with various combinations of the bacterial species and examining the system using high-resolution live cell fluorescence microscopy we show that more than one species of bacteria frequently invades the same bladder cell.

Materials and methods

Human ethics

This study was approved by the UTS Human Research ethics committee with approval numbers HRCH REF No. 2014000452 and HREC ETH22-7590. All urine doners gave their approval to participate in this study by informed consent under this study's ethics approval numbers above.

Plasmid construction

pCAK1 and pAP1 plasmids were constructed by replacing the GFP sequence in pGI5 (Iosifidis and Duggin 2020) with mCerulean3 and mOrange2 sequences, respectively. The mCerulean3 and mOrange2 gene fragments were amplified from pEB1-mCerulean3 and mOrange2-pBAD plasmids, both obtained from Addgene (Addgene plasmids #103973 and #54531) (Shaner et al. 2008, Balleza et al. 2018). To construct pCAK1 and pAP1, pGI5 was digested with NcoI and BamHI, and assembled with mCerulean3 and mOrange2 PCR products containing 20-30 bp homologous regions from pGI5 on both the ends. This was performed using a in vivo DNA assembly method (Watson and Garcia-Nafria 2019). The final products were confirmed by Sanger sequencing and fluorescence microscopy. Primers used in this study are shown in the List of Primers table.

Bacterial growth

Bacterial strains were used as Uropathogenic Escherichia coli UTI89 transformed with mOrange2 (pAP1) or mCherry (pGI6) (Wright et al. 2021), Klebsiella pneumoniae TOP52 transformed with mCerulean3 (pCAK1) or sfGFP (pGI5), and Enterococcus faecalis OG1RF expressing GFP from the chromosome (Debroy et al. 2012). Strains used in this study are listed in Table 1.

All fluorescent proteins were produced as constitutively expressed freely diffusing molecules in the cytoplasms of the corresponding bacterial strain. A single colony of respective strain was grown overnight in a 20 ml Luria Broth (E. coli and K. pneumoniae) or Brain Heart Infusion (E. faecalis) culture at 37°C without shaking to induce expression of type-1 pili which facilitates adhesion to the bladder cells during infection. Antibiotics were added when appropriate (Spectinomycin 50 µg ml^−1^, Rifampicin 100 µg ml^−1^). The following morning, the cultures were pelleted and resuspended in 1 x PBS added to the infection model.

Bladder cell growth

Immortalized epithelial bladder cells (PD07i (Klumpp et al. 2001, Berry et al. 2009) were grown and maintained in EpiLife media supplemented with Human Keratinocyte Growth supplement (HKGS) in 5% CO2 at 37°C. Cells were split as required using standard Trypsinization methods upon reaching ∼ 80% confluency.

Human urine preparation

Human urine was collected from both male and female donors. Samples were collected in the mornings and stored at 4°C for at least 2 days before further processing. Samples were pelleted at 3000 rpm, the supernatant filtered through a 0.2 µm filter and aliquoted into 50 ml falcon tubes before placed in—20°C for storage until use. Urine was only used if the pH was between 5 and 6.5 and the Urine Specific Gravity (USG) was in the range between 1.024 and 1.030, values that has been shown to produce high degree of filamentation (Iosifidis and Duggin 2020, Söderström et al. 2022).

Infection models

Flow model

The in-vitro UTI flow channel model has been described previously (Andersen et al. 2012, Iosifidis and Duggin 2020). Here only slight changes were introduced as follows, µ-Silde I^0.2^ Luer (IBIDI #80166, total channel volume 50 µl) flow channels were seeded with bladder cells according to the manufacturer's recommendations. All channels and dishes were left in the incubator until a confluent layer of bladder cells had developed (note that after bacterial addition and growth some bladder cells shed).

The channels were connected to a New Era pump system for continuous flow of nutrients, this was left to run until a confluent layer of bladder cells formed (Söderström et al. 2022). Bacterial cells resuspended in PBS at an OD600 of 0.2–0.4 were introduced fully in the flow channels and flushed at a flow rate of 15 µl min^−1^ for 10–15 minutes (resulting in a total of ∼10^7^ bacterial cells run over the bladder cells depending on concentration and species, as determined by CFU counts), before fresh EpiLife media again was flowed over the cells for 7 hours (initial 10 minutes at 100 µl min^−1^ to flush out excess bacteria, then 15 µl min^−1^ for the reminder of the time) to allow for invasion. After this, the media was exchanged to EpiLife containing Gentamycin (f.c. 100 µg ml^−1^ (Iosifidis and Duggin 2020), for the dual infection of UPEC and K. pneumoniae was on additional 100 µg ml^−1^ of Ampicillin added) for 13 hours to eliminate all extracellular bacteria. To monitor formation of intracellular bacterial communities, channels were taken to the microscope for imaging at this point. Immediately prior to imaging, bladder cell membranes were stained with CellBrite405 or 650 in PBS (f.c. 0.7:100, with the addition of CellBrite Enhancer to mask intracellular fluorescence according to the manufacturer's recommendations) under flow (15 µl min^−1^) for 40 minutes, and washed twice with 1 x PBS. Channels were re-filled with EPILIFE media to sustain cell health during imaging.

To generate filaments, after subjecting samples to the EpiLife/antibiotics mixture, human urine was added to the flow system (15 µl min^−1^) for 16–20 hours. Bacterial samples were collected through the back-end of the flow channels, washed once in 1 x PBS, resuspended in LB or BHI. 4 µl of respective culture was placed on 2% agarose pads (w/w) and directly taken for imaging.

Dish model

In the semi-static infection model a similar experimental workflow was followed as for the flow model with the same time intervals in changing media, with the exception that all steps were carried out in a 35 mm Petri dish (IBIDI glass bottom dish with glass coverslip bottom (#1.5), pre-sterilized, cat number #81218–200) placed in a CO2 incubator at 37°C with 50 rpm orbital agitation. For the dish models the total amount of bacteria (∼10^7^, based on CFU measurements. This corresponds roughly to an MOI of 100) was distributed evenly over the bladder cells at one time point only. Bacteria were incubated for 15 minutes before liquid was aspirated of and fresh EpiLife media (2 ml) was added and grown for 7 hours. Following this, media was exchanged to EpiLife containing Gentamycin (f.c. 100 µg ml^−1^ (Iosifidis and Duggin 2020)), for 13 hours to eliminate all extracellular bacteria. To monitor formation of intracellular bacterial communities, channels were taken to the microscope for imaging. Immediately prior to imaging were bladder cell membranes stained with CellBrite405 or 650 in PBS (f.c. 100, with the addition of CellBrite Enhancer to mask intracellular fluorescence according to the manufacturer's recommendations) for 40 minutes, washed twice in 1 x PBS and again covered with 2 ml EpiLife to sustain cell health during imaging. To generate filaments in the Petri dish model, 1 ml urine was added and manually exchanged once an hour for the first 4–5 hours before a final resuspension of 3 ml was done and left in the incubator for 12–15 hours.

Imaging

To minimise biased imaging, regions of interest [ROIs] were chosen based on fluorescence signal from bladder cell membranes only, and Z-stacks were acquired for each ROI. Imaging was performed on a Leica Stellaris confocal microscope equipped with a 63x oil objective enclosed in an environmental chamber operated at 37°C and 5% CO2 (Oko-Lab). The fluorophores were excited by a white laser at optimized wavelengths, and emission was collected in pre-set system optimised detector intervals for AlexFluor405, mCerulean3, EGFP, mOrange2 and AlexaFluor647 depending on experiment, to minimize channel crosstalk. Z-stacks were always acquired to validate that the bacteria in fact were inside the bladder cells. Image size was either 2048 × 2048 or 4096 × 4096, with pixel size 90 and 45 nm, respectively. The pinhole was set to 1 AU and Z-stacks were collected with software optimized step length of either 125 or 250 nm (30–99 images per stack, depending on step length and thickness of the bladder cell in question). 3D reconstruction and deconvolution of Z-stacks were performed in the Leica LAS software and further visualised in FIJI (ImageJ).

Quantification and statistical analysis

Raw microscopy images, deconvoluted Z-stacks and movies were transferred to FIJI (ImageJ) for final analysis and processing. Fluorescence traces were analysed in OriginPro 2021 (V. 9.8.0.200 [Academic]). Note that all fluorescence traces were generated from raw microscopy data and not deconvoluted data. Cell lengths were extracted from MicrobeJ (rod and cocci) (Ducret et al. 2016) or by manual tracing in Fiji (filaments). Evaluations of statistical significance were performed using students T-tests in GraphPad Prism software (v.9.2). Levels of significance are indicated as ns, not significant; *, P < 0.05; , P < 0.01, *, P < 0.001, ^^, P < 0.0001

Results

Flow and semi-static agitation affect invasion of human epithelial bladder cells by Gram-positive bacteria differently from Gram-negative bacteria

Initially, we established monospecies infections simulating the invasion phase of uncomplicated UTI using PD07i immortalized human epithelial bladder cells (Klumpp et al. 2001, Berry et al. 2009), challenged with either UPEC (strain UTI89 expressing cytoplasmic mOrange2), K. pneumoniae (strain TOP52 expressing cytoplasmic mCerulean3) or E. faecalis (strain SD234 expressing cytoplasmic GFP), to ascertain whether they were capable of infecting host cells in a previously established in-vitro UTI flow-chamber model and to establish baseline infection rates (Iosifidis and Duggin 2020, Söderström et al. 2022). This model is based on a commercial flow-chamber system (IBIDI I^0.2^  µ-Slides connected to NewEra syringe pumps) where constant flow is applied throughout the infection cycle to mimic bladder flow. In parallel, we also used a semi-static infection model using 35 mm glass-bottom Petri dishes under constant orbital agitation (50 rpm). In both models, the bladder cells were initially exposed to a total of ∼ 10^7^ bacterial cells of each species. All three pathogens invaded the epithelial bladder cells both in the flow-chamber and Petri-dish conditions (Fig. 1A–D, Fig. S1).

At 20 h post inoculation, UPEC had invaded the most bladder cells of the tested Gram-negative species (23.9 ± 8.5% of the bladder cells in the semi-static petri dish model and 20.6 ± 7.4% in the flow model, Mean ± S.D., n > 432 from three independent infection experiments), while K. pneumoniae invaded approximately half of this number (8.49 ± 5.3% in the dish model and 10 ± 6.9% in the flow model, n > 321 from three independent infections of each condition) (Fig. 1D). E. faecalis showed the highest overall invasion with an average 27.9 ± 17.4% of the bladder cells invaded in the dish model (n = 335, from 3 experiments) (Fig. 1D). In contrast, E. faecalis invaded less than 1% of the bladder cells under the flow-chamber conditions (n = 462, from 3 independent experiments). For UPEC and K. pneumoniae the variations between the models were statistically nonsignificant, while for E. faecalis they were P = 0.0357 (statistically significant).

Elevated invasion rates for UPEC in dual species infection with E. faecalis

Since so few bladder cells were invaded by E. faecalis in the flow-chamber model, the culture-dish model was mainly used to follow invasion during multispecies UTI. We co-inoculated the bladder cells with equal numbers of UPEC and E. faecalis (a combined total of ∼ 10^7^ bacteria (Tien et al. 2017)). Based on images of more than 560 randomly inspected bladder cells at ∼ 20 h post inoculation (regions of interest [ROIs] were chosen based on fluorescence signal from bladder cell membranes, and Z-stacks were acquired for each ROI) from three independent experiments, almost half (∼ 45%, Fig. 2A,B) were invaded by at least one type of bacterial species. Most infected bladder cells were only invaded by one species (Fig. 2A, arrows).

In these dual-species infections, UPEC alone were internalised in 33% of cells (compared to ∼ 24% in single species dish infections (Fig. 1D)), whereas E. faecalis alone were internalised in only 4% (Fig. 2B) (compared to ∼ 28% in single species dish infections (Fig. 1D)). The percentage of bladder cell invasions of either or both bacteria in dual-species infections was almost 45% (43.6 ± 19.3% (n = 564 cells, from 3 experiments)). Surprisingly to us, a relatively high number, 44 of 564 or approximately 8% of the individual bladder cells were invaded by both bacterial species at the same time (Fig. 2A, arrow heads).

We noticed that the arrangement of the internalised bacteria was distinct for each species; UPEC formed condensed IBCs as previously established (Justice et al. 2004), while E. faecalis were often arranged in more loosely organised clusters in which single cells were readily resolved (Fig. 2C). Initial 3D reconstructions of low-resolution images suggested that UPEC clusters were often in close spatial proximity of the E. faecalis cells (Fig. 2D,E). With increased resolution, it became evident that tightly packed UPEC cells formed multicellular communities that frequently surrounded one or a few E. faecalis cells (Fig. 2F,H). This type of UPEC ‘encirculation’ of E. faecalis cells was apparent in a substantial fraction of the total observed UPEC IBCs (32%) during dual species invasion (Fig. 2I). We speculate that E. faecalis and E. coli might cooperate to increase the likelihood of prolonged infection (Tien et al. 2017).

Lower overall infection rates of bacteria in dual and triple-species infections when K. pneumoniae is present

To investigate whether other mixed bacterial cultures would co-invade PD07i cells and exhibit similar interactions as seen for UPEC/E. faecalis, we carried out infection experiments with other combinations of E. faecalis and K. pneumoniae (Gram-positive/Gram-negative pair), UPEC and K. pneumoniae (Gram-negative/Gram-negative pair) or all three species (UPEC, E. faecalis and K. pneumoniae). At 20 h post inoculation, we detected host cells containing all combinations of the bacteria investigated (Fig. 3A-C, Figs S3-S4 and Supplementary Movies SM6–SM8), but the invasion frequencies varied substantially.

In contrast to the UPEC/E. faecalis case, encirculation was not observed when E. faecalis was co-infected with K. pneumoniae (4 independent infection experiments, n = 463 cells in total), nor when the two Gram-negative species UPEC and K. pneumoniae were used in the same infection (4 independent infection experiments, n = 519 cells in total). UPEC and K. pneumoniae often formed dense IBCs adjacent to one another but did not encase or mix with each other (Fig. 3A). Overall, UPEC and K. pneumoniae invaded ∼ 12% of all bladder cells in the Dish model and ∼ 8% of all bladder cells in the flow channel model, however they coinfected only 1–2% of bladder cells both in either model (Fig. 3A, Supplementary Movie SM6).

During E. faecalis and K. pneumoniae co-infection, we noticed that while they were capable to co-invade bladder cells, we did not observe dense IBCs (Fig. 3B, Supplementary Movie SM7).

E. faecalis has previously been shown to antagonize K. pneumoniae biofilm formation during mixed growth albeit under different conditions compared to the intracellular conditions in the present study (Ballen et al. 2020). The total percentage of invaded bladder cells in a E. faecalis and K. pneumoniae dual species combination was ∼ 13%, with 3.7% of all observed of bladder cells invaded by both (Fig. 3B).

For three-species infections (UPEC, E. faecalis and K. pneumoniae), UPEC/E. faecalis again showed similar organizational patterns as they did in two species infections, while K. pneumoniae was predominately localized separate from the other two (Fig. 3C, Fig. S4, Supplementary Movie SM8). In the three species infections roughly 30% of bladder cells were invaded by at least one species, with almost 4% invaded by all three (Fig. 3C, n = 349, 3 independent infection experiments). This was less than the invasion frequency of the UPEC/E. faecalis dual infections where ∼ 45% of the bladder cells were invaded by at least one type of bacteria (Fig. 2B), suggesting that K. pneumoniae could have an antagonistic effect on invasion efficiency during multi-species infections.

Gram-negative bacteria undergo infection-related filamentation in multispecies infections, while Gram-positive do not

During murine model UTI, UPEC have been seen to undergo morphological changes during intracellular bacterial community (IBC) formation and dispersal, where UPEC grow into long filamentous cells (Justice et al. 2006). These morphologies were later also seen in female patents with cystitis (Rosen et al. 2007) and more recently also confirmed in flow-chamber models using human bladder cells and urine (Andersen et al. 2012, Klein et al. 2015, Iosifidis and Duggin 2020, Söderström et al. 2022). To determine whether single and multi-species co-infections undergo the same morphological changes under the established conditions, we next visualized cells sampled after the dispersal stage of infection by exposing infections to human urine (20 h) (Iosifidis and Duggin 2020, Söderström et al. 2022). Both UPEC and K. pneumoniae formed filaments of several hundreds of microns long (Fig. 4A,B). For K. pneumoniae, similar morphology cycles have been previously observed in murine models (Rosen et al. 2008), but not in in-vitro models using human epithelial bladder cells and urine. The average length of both UPEC and K. pneumoniae filaments from infections was ∼ 50 µm (Fig. 4D). By comparison, UPEC and K. pneumoniae rods grown in LB medium were approximately 3.5 µm (Fig. 4D). We classified cells as filaments if they were at least two times the average WT cell length, i.e. 7 µm. We observed that the number of cells that underwent filamentation (but not the average length of filaments) was lower in the dish model compared to the flow-chamber model. While the molecular reasons for this is not currently clear, we speculate that proficiency of morphology changes may be connected to flow and shear-force dynamics, or the continual exchange of constituents in the medium (Andersen et al. 2012). In contrast to the Gram-negative species, E. faecalis cells did not filament or grow significantly larger upon exposure to human urine (Fig. 4C,D). Average lengths after an infection were ∼ 1.4 µm (n = 215) from petri-dish infection and ∼ 1.6 µm (n = 102) from flow chambers which was similar to lengths for E. faecalis cells grown in BHI medium only (1.75 (n = 238)) (Fig. 4D). Similar results for filamentation were observed during triple-species infections (Fig. 4e). Taken together, these observations suggest that infection-related filamentation may be a broadly occurring phenomenon in uropathogenic Gram-negative bacteria and it also transpires during multispecies infections.

Discussion

Numerous studies have investigated infection-related microbial interactions and biofilm formation in-vitro in liquid media (e.g. rich media, synthetic or human urine) and on catheters (Galván et al. 2016, Juarez and Galván 2018, Gaston et al. 2020, Ch'ng et al. 2022), their effect on immune activation (Tien et al. 2017, Flores et al. 2023, Jafari and Rohn 2023, Kao et al. 2023), and in-vivo in animal models (Lavigne et al. 2008, Keogh et al. 2016, Tien et al. 2017) (a list on further reading (Gaston et al. 2021)). However, hardly any studies have investigated bacterial co-invasion behaviours at the single host cell level. Here, to study and to better understand bacterial invasion during a multispecies infection model on single cell levels in a systematic way, we established a set of multi-species urinary tract infection in-vitro models. We compared invasion frequencies of three common urinary tract pathogens, Uropathogenic Escherichia coli (UPEC), K. pneumoniae and E. faecalis, using human epithelial bladder cell in-vitro infection models under two different conditions (‘flow’ and ‘dish’). In the flow condition, UPEC showed highest overall invasion frequency, followed by K. pneumoniae, while E. faecalis barely invaded any bladder cells at all. Differences in invasion frequency between UPEC and K. pneumoniae has been observed previously in murine models, and was attributed differences in adhesion capabilities through lower type 1 pilus expression of K. pneumoniae (Rosen et al. 2008). The invasion frequencies of UPEC and K. pneumoniae were only marginally impacted by switching from continuous flow to circular agitation. On the other hand, the Gram-positive E. faecalis exhibited a large variation in invasion frequency between the models tested in this study. While E. faecalis is a well-established uropathogen (Garsin et al. 2014, Ch'ng et al. 2019), our observations suggest a weakened ability to adhere and invade bladder cells under the flow conditions in our model. At this point we do not fully understand the details resulting in the large variation in bladder cell invasion of E. faecalis between our flow and dish models, but similar behaviours have previously been linked to the reduced adherence capabilities of E. faecalis (Shankar et al. 2001, Kau et al. 2005, Kline and Lewis 2016, Ch'ng et al. 2019). One possible limitation of the E. faecalis invasion experiments is the lack of fibrinogen in the system (Flores-Mireles et al. 2016). Fibrinogen has previously been implicated in biofilm formation on urinary tract catheters and is important for infection in murine models which may in part explain the relatively low invasion rate of E. faecalis in our experiments (Flores-Mireles et al. 2014), however multiple additional cell surface adherence pathways have been identified (Rozdzinski et al. 2001, Toledo-Arana et al. 2001, Mohamed et al. 2006, Ch'ng et al. 2019).

In the multispecies co-infection experiments, we saw that multiple species of bacteria readily invade the same bladder cell. While molecular details governing intracellular coexistence of bacteria during infection remains elusive, our observations indicate that whenever UPEC and E. faecalis are introduced together the overall rate of infection increases. Similar cooperation and increase in pathogenicity by UPEC and E. faecalis have been suggested based on results from murine models (Tien et al. 2017). The observation that UPEC encirculate E. faecalis is perplexing and its biological significance is unclear but suggests a spatial importance of the previous observations that these two species cooperate on a biochemical level to increase pathogenesis (Montravers et al. 1994, Lavigne et al. 2008, Hughes and Winter 2016, Keogh et al. 2016).

In contrast, whenever K. pneumoniae was introduced into a multispecies infection situation the average invasion frequency was always lower than in the corresponding single species infections. Supporting this observation, E. faecalis has previously been implicated in suppressing K. pneumoniae growth during polymicrobial biofilm formation in human urine (Ballen et al. 2020). This suppression by E. faecalis seems also be extended to other Gram-negative species (i.e*. Pseudomonas aeruginosa*) (Zhu Tan et al. 2022).

Both infection models presented in this study caries its own merits, and each may be favoured depending on the research question in mind, e.g. ‘Flow’ could represent bladder release, while ‘Dish’ would possibly depict a resting bladder situation. We believe that the observations presented in this study could initiate a spark for further investigation into detailed genetics studies exploring dedicated signalling pathways and potential quorum sensing in multispecies UTIs at a single cell level. This will undoubtably play an increasingly important role in understanding competing or cooperative behaviours of bacteria during infections as antimicrobial resistance continue to rise.

List of primers

Supplementary Material