Introduction

Urinary tract infections (UTIs), including asymptomatic bacteriuria, cystitis, urethritis, and pyelonephritis, are the most common bacterial infections1, affecting ~150 million people worldwide per year, with high social costs2. UTIs are commonly treated with antibiotics; however, prolonged use disrupts the normal microbiota balance and causes the development of multidrug-resistant microorganisms3. UTIs mostly affect women, with over 60% experiencing a UTI during their lifetime4. After an initial infection, approximately 25% of women experience recurrence within 6 months, with 68% of recurrent infections caused by the original strain5.

Uropathogenic Escherichia coli (UPEC) is the primary cause of UTI1. UPEC enters the urinary tract from the intestinal tract, binding to the surface of bladder epithelial cells (BECs) via type 1 fimbria and invades BECs via fusiform vesicles6. Upon invasion, vesicular UPEC utilizes the phospholipase PldA to disrupt the fusiform vesicle membrane and rapidly escape into the cytosol, evading Toll-Like Receptor 4-mediated exocytosis7,8. Within the BEC cytosol, UPEC clonally replicates to form intracellular bacterial communities (IBCs). After IBCs mature, UPEC can flux out to invade neighboring naive BECs, causing subsequent rounds of IBC formation9. IBC formation provides a protected niche whereby UPEC can evade neutrophil phagocytic activity, sustain the infection between bouts of bladder voiding, and promote infection recurrence following antibiotic treatment, as intracellular locations protect UPEC from antibiotics10,11. IBC formation enables a single bacterium invasion to rapidly increase UPEC numbers in the urinary tract, leading to UTI12,13.

Invasion of BECs is the first and most critical step in establishing a successful UTI by UPEC14. Type 1 fimbria present the tip adhesin, FimH, for UPEC to bind to BEC surfaces, which is required for UPEC invasion15,16. The deletion of type 1 fimbrial genes in UTI89 significantly attenuated bacterial colonization in mouse bladder17. The fimH mutant NU14 is non-invasive to human BECs 563716, and there was a minimal invasion in mice bladders infected with the fimH mutant NU1418. Type 1 fimbria expression significantly increased when UPEC invaded the superficial BECs in mouse bladders19. RNA sequencing (RNA-Seq) assays of isolated UPECs revealed the upregulation of type 1 fimbria genes in human UTI urine samples20 and infected CBA/J mice urine21. This suggests the activation of type 1 fimbria expression in response to host-derived cues for promoting invasion. However, host cues and the underlying signal transduction mechanisms are poorly understood.

Type 1 fimbria undergoes phase variation due to rearrangement of the invertible element (fimS), containing a 314 bp DNA segment flanked by 9 bp inverted repeats and the fim gene cluster promoter (fimAICDFGH)22,23. When the invertible element is in the phase-ON orientation, the promoter enables the transcription of the structural fim genes, whereas transcription is inhibited in the inverted phase-OFF orientation. The adjacent tyrosine site-specific recombinases FimB and FimE catalyzes fimS inversion24. Multiple DNA-binding proteins, including integration host factor (IHF), leucine-responsive protein (LRP), and histone-like nucleoid structuring (H-NS), interact directly with fimS to facilitate DNA bending to form the appropriate conformation necessary for fimS phase inversion25,26. A diverse range of regulators indirectly control type 1 fimbria expression by regulating fimB and fimE transcription27.

Two-component systems (TCSs) consist of a membrane-bound histidine kinase (HK) sensor protein that responds to environmental conditions and a homologous response regulator that acts as a DNA-binding protein to regulate downstream gene expression28. TCSs enable bacterial pathogens to adapt to different environments during their pathogenic process and regulate virulence responses to environmental signals29. A number of TCSs are implicated in the adaptation of host environments by UPEC, including PhoB/PhoR, OmpR/EnvZ, and QseB/QseC30, which also regulate type 1 fimbria expression31,32,33. PhoB/PhoR, which responds to phosphate-limiting condition34, indirectly represses type 1 fimbria expression, indicated by the increased amount of type 1 fimbrial major subunit FimA in phoB/phoR mutant compared to wild-type (WT) UPEC in vitro31. EnvZ/OmpR represses fimB transcription by binding to fimB promoter and indirectly downregulating type 1 fimbria expression under acidic and high osmolality conditions in vitro33,35. Deleting envZ/ompR attenuated UPEC virulence in a mouse bladder colonization assay at 24 h post-infection (p.i.)36. However, whether PhoB/PhoR- and EnvZ/OmpR-regulated type 1 fimbria expression affects UPEC virulence remains unclear. QseC directly dephosphorylates QseB, which in turn de-represses type 1 fimbria expression and facilitates UPEC colonization and IBC formation, as observed in bacterial titers and IBC numbers in WT and qseC mutant-infected mouse bladders at 6 h and 16 h p.i.32. However, the mechanisms by which QseC/QseB regulates type 1 fimbria remain unknown. Hence, the signaling regulatory mechanisms underlying the activation of type 1 fimbria to promote UPEC invasion in BECs remain unclear.

GrpP (encoded by c4545) and GrpQ (encoded by c4546) is a potential TCS in UPEC, as they have sequence similarity to HK and RR, respectively37. GrpP and GrpQ are required for the expression of 1,2-propanediol degradation genes, located in the same cluster as grpP and grpQ in UPEC, to produce ATP and stimulate growth under anaerobic conditions37. grpP and grpQ are present in the genome of UPEC CFT073 but not in non-pathogenic E. coli K12 MG1655 or pathogenic enterohemorrhagic E. coli (EHEC) O157, suggesting that GrpP/GrpQ may be a TCS specific to UPEC38. However, whether GrpP and GrpQ are associated with UPEC pathogenesis has not yet been reported.

This study aimed to investigate the role of the GrpP/GrpQ in UPEC and its specific involvement in the regulation of type 1 fimbria expression. In this study, autophosphorylation and transphosphorylation assays confirmed that GrpP and GrpQ constitute a TCS, with GrpP as the HK sensor and GrpQ as the RR. Next, we investigated whether GrpP/GrpQ contributes to UPEC virulence by generating grpP/grpQ mutant and performing infection assays using a BALB/c mouse model of acute UTI and a human BEC 5637 invasion model. The results showed that GrpP and GrpQ enhance UPEC virulence by promoting the initial invasion of BECs. Further investigation showed that GrpP/GrpQ positively regulates type 1 fimbria expression by directly binding to fimS. Moreover, GrpP/GrpQ is activated in response to D-serine, which is abundant in host urine, leading to increased type 1 fimbria expression and enhanced invasion. This study reveals a new host-derived signal transduction regulatory pathway that contributes to UPEC pathogenicity.

Results

GrpP and GrpQ constitute a cognate TCS

To verify whether GrpP and GrpQ constitute a TCS, we constructed recombinant plasmids expressing GrpP-GST and GrpQ-His6 proteins with pGEX-6p-2 and pET-28a (+), separately. These proteins were purified and used for autophosphorylation and transphosphorylation assays as described39. Purified GrpP-GST was incubated with ATP at 37 °C for specified durations. Phosphorylated GrpP was detected using a pIMAGO-biotin phosphoprotein detection kit. We observed that GrpP underwent autophosphorylation at 1 min, with a significant increase in autophosphorylation over time (Fig. 1A). Transphosphorylation assay involved co-incubation of autophosphorylated GrpP with GrpQ at 37 °C for the specified duration. We observed a significant increase in GrpQ phosphorylation within 5 min compared to the phosphorylation levels at 0 min (Fig. 1B). In addition, reverse transcription PCR (RT-PCR) results showed that grpP and grpQ were co-transcribed (Fig. S1). These results suggest that GrpP and GrpQ constitute a cognate TCS, with GrpP as the HK and GrpQ as the RR.

Fig. 1: GrpP and GrpQ constitute a cognate TCS.
figure 1

A Autophosphorylation of GrpP in vitro. The purified GrpP protein was mixed with ATP and incubated at 37 °C, and phosphorylated proteins were detected using the pIMAGO-biotin phosphoprotein detection kit. Coomassie gel of GrpP acted as loading control. B Phosphotransfer from GrpP to GrpQ. The autophosphorylated GrpP and purified GrpQ protein were mixed with ATP, and the reaction mixture was incubated at 37 °C and detected using the pIMAGO-biotin phosphoprotein detection kit. Coomassie gel of GrpP and GrpQ acted as loading control. Data were obtained from three independent experiments, and representative images are shown. Source data are provided as a Source data file.

Full size image

GrpP/GrpQ TCS contributes to UPEC virulence

To determine whether GrpP/GrpQ contributes to UPEC virulence, we constructed a mutant strain ΔgrpP/grpQ lacking both grpP and grpQ genes using the λ-Red recombination system, and a complemented strain, ΔgrpP/grpQ+, via transformation of ΔgrpP/grpQ with a pBluescript II KS (+) recombinant plasmid carrying the grpP and grpQ genes under its own promoter of the CFT073 WT strain. We transurethrally infected female BALB/c mice with 1 × 107 WT, ΔgrpP/grpQ or ΔgrpP/grpQ+ and monitored bladder bacterial colony-forming units (CFUs) and IBC formation at 6 h p.i., indicating the first round of IBC formation, and 24 h p.i., representing subsequent rounds of IBC formation, during acute UTI9. The results showed that both bacterial CFUs in mouse bladders and the IBC numbers per bladder were significantly lower in ΔgrpP/grpQ than in WT at either 6 h or 24 h p.i. (Fig. 2A–D). ΔgrpP/grpQ+ restored bacterial CFUs and IBC numbers of ΔgrpP/grpQ to levels similar to that of WT-infected mouse bladders (Fig. 2A–D). These results indicate that a lack of grpP/grpQ decreases UPEC colonization and IBC formation in mouse bladders. The bacterial CFUs in mouse bladders were also lower in ΔgrpP/grpQ than in WT at 48 h p.i., a later timepoint of acute UTI (Fig. 2E). Furthermore, ΔgrpP/grpQ grows as well as WT in Luria-Bertani (LB) and RPMI 1640 media, ruling out a growth defect as the cause of UPEC colonization and IBC formation in vivo (Fig. S2). Therefore, grpP and grpQ promoted UPEC colonization in mouse bladders during acute UTI, thereby contributing to UPEC virulence.

Fig. 2: GrpP/GrpQ TCS contributes to UPEC virulence by promoting its invasion into superficial BECs.
figure 2

A, B Total bacterial titers of UPEC in the bladders of BALB/c mice transurethrally infected with WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ at 6 h (A) or 24 h p.i. (B) (n = 10 mice). C, D IBC enumeration in BALB/c mouse bladders transurethrally infected with WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ at 6 h (C) or 24 h p.i. (D) determined using confocal microscopy (n = 10 mice). Representative images of the IBC were shown on the right. Bacteria and cell membranes were stained with wheat germ agglutinin (red), and nuclei were stained with DAPI (blue). Scale bars, 20 μM. E Total bacterial titers of UPEC in the bladders of BALB/c mice transurethrally infected with WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ at 48 h p.i. (n = 10 mice). F Intracellular bacterial titers of WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ in 5637 cells at 2 h p.i. (n = 3 independent experiments). G Intracellular bacterial titers of UPEC in the bladders of BALB/c mice transurethrally infected with WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ at 2 h p.i. (n = 10 mice). Data were obtained from three independent experiments and presented as mean ± SD. P values were determined using two-tailed Mann–Whitney U test (A, B, C, D, E and G) and two-tailed unpaired Student’s t-test (F). Significance was indicated by a P value. n.s., No significant difference. A P = 0.000022; C P = 0.000076; D P = 0.0000111; G P = 0.000043. Source data are provided as a Source data file.

Full size image

GrpP/GrpQ TCS contributes to UPEC virulence by promoting invasion into superficial BECs

Invasion of superficial BECs is essential for UPEC colonization of the host bladder13. We then investigated whether the decreased colonization in ΔgrpP/grpQ is caused by an invasion defect and assessed the invasion abilities of WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ into 5637 cells in vitro using a gentamicin protection assay. The results revealed a significant reduction in the invasion of ΔgrpP/grpQ into 5637 cells compared to that of WT (Fig. 2F). ΔgrpP/grpQ+ restored bacterial invasion of 5637 cells to levels similar to that of WT (Fig. 2F). We further investigated whether grpP/grpQ affects UPEC invasion of superficial BECs in vivo by infecting BALB/c mice with WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+, and determining bacterial CFUs in mouse bladders at 2 h p.i., when UPEC invaded BECs18. Consistent with in vitro results, intracellular bacterial CFUs of ΔgrpP/grpQ-infected mouse bladders were significantly lower than those of WT-infected mouse bladders (Fig. 2G). ΔgrpP/grpQ+ fully restored the ability of UPEC to invade mouse bladders (Fig. 2G). These results indicate that GrpP and GrpQ promoted UPEC invasion of BECs in vitro and in vivo.

GrpP/GrpQ TCS promotes UPEC invasion by activating type 1 fimbria genes

Type 1 fimbria are essential for UPEC invasion16. qRT-PCR analysis showed that the expression of type 1 fimbria was upregulated in CFT073 cultured in mouse urine for 2 h (Fig. 3A). We investigated whether GrpP and GrpQ promoted UPEC invasion by regulating type 1 fimbria expression. To investigate downstream genes controlled by GrpP/GrpQ, we performed NGS-based RNA sequencing (RNA-seq) on WT and ΔgrpP/grpQ cultured in LB medium. The RNA-seq results showed that grpP/grpQ mutation significantly decreased the transcription of all seven genes (fimA, fimC, fimD, fimF, fimG, fimH, and fimI) encoding type 1 fimbria (Fig. 3B). Quantitative reverse transcription PCR (qRT-PCR) analysis confirmed the positive regulation of type 1 fimbria genes by GrpP and GrpQ (Fig. 3C). These results indicate that GrpP and GrpQ positively regulate the expression of type 1 fimbria genes. Western blotting further revealed that grpP/grpQ mutation decreased the protein levels of FimH (Fig. 3D), indicating that GrpP/GrpQ promoted the production of type 1 fimbria. Hemagglutination (HA) assays were performed to verify the effect of GrpP/GrpQ on the surface production of type 1 fimbria. The result showed a significant decrease in HA titer in ΔgrpP/grpQ compared with that of WT, and ΔgrpP/grpQ+ restored the HA phenotype to WT levels (Fig. 3E). As a negative control, we constructed a mutant strain ΔfimA-H, which was previously shown to lack type 1 fimbria expression and barely hemagglutination40, and a double mutant ΔfimA-HΔgrpP/grpQ, which lacks both fimA-H and grpP/grpQ genes. The HA titer of ΔfimA-HΔgrpP/grpQ was similar with both ΔgrpP/grpQ and ΔfimA-H, indicating the contribution of GrpP/GrpQ on surface production of type 1 fimbria depends on fimA-H. We also constructed a fimS phase-locked-ON (left inverted repeat, LIR) CFT073 mutant, in which the inversion is blocked allowing the fimS promoter element to be persistently reoriented into the ON orientation41. We found that the LIR fimS mutation restored the HA titer of the ΔgrpP/grpQ mutant to WT levels, indicating GrpP/GrpQ regulated the surface production of type 1 fimbria by controlling the inversion of fimS. The difference in HA titer between WT and ΔgrpP/grpQ was eliminated in the presence of mannose, a competitive inhibitor of type 1 fimbria-mediated adhesion42 (Fig. 3E). These results confirm that GrpP/GrpQ promotes the production of type 1 fimbria on the bacterial surface.

Fig. 3: GrpP/GrpQ TCS promotes UPEC invasion by activating type 1 fimbria genes.
figure 3

A qRT-PCR analyses of the mRNA levels of fimH in CFT073 cultured in mouse urine compared to that in CFT073 cultured in M9 medium for 2 h (n = 5 mice). B Volcano plot showing differentially expressed genes in the ΔgrpP/grpQ transcriptome versus WT cultured in LB medium. Data are expressed as a log2-fold change in gene expression levels (X-axis) plotted against the −log10 p-value (Y-axis). The red color indicates upregulated genes. The green color indicates downregulated genes. fim genes are marked by a black circle. Differentially expressed genes were tested by a two-part hurdle model; P-values were adjusted for multiple comparisons by the Bonferroni’s method. C qRT-PCR analyses of the mRNA levels of fim genes in ΔgrpP/grpQ and WT strains cultured in LB medium. D Western blot analysis of FimH protein levels in WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ statically grown in LB medium at 37 °C overnight. DnaK was used as the loading control. Representative image from three independent experiments. E HA assays of the production of type 1 fimbria in WT, ΔgrpP/grpQ, ΔgrpP/grpQ+, ΔfimA-H, ΔfimA-HΔgrpP/grpQ, WT-LIR, or ΔgrpP/grpQ-LIR in the presence or absence of 3% mannose. F, G Competitive index of ΔfimA-H versus ΔfimA-HΔgrpP/grpQ (F) or ΔgrpP/grpQ-LIR versus WT-LIR (G) in BALB/c mouse bladders at 2 h p.i. (n = 10 mice). Data were obtained from three independent experiments and presented as mean ± SD. P values were determined using two-tailed Student’s t-test (A, C and D) or two-way analysis of variance (E) or two-tailed Wilcoxon signal-rank test (F, G). Significance was indicated by a P value. n.s., No significant difference. C P = 0.0000000061 (fimH), P = 0.0000000083 (fimA), P = 0.0000000034 (fimC), P = 0.0000000015 (fimI), P = 0.0000000036 (fimD), P = 0.0000000173 (fimF), P = 0.0000000078 (fimG); D P = 0.000015; E P = 0.000052 (WT vs ΔfimA-H), P = 0.000052 (WT vs ΔfimA-HΔgrpP/grpQ). Source data are provided as a Source data file.

Full size image

To verify that the impaired invasion of ΔgrpP/grpQ into BECs occurred via type 1 fimbria, we performed in vivo competition assays to measure the difference in BEC invasion between ΔfimA-H and ΔfimA-HΔgrpP/grpQ. BALB/c mice were co-infected at a 1:1 ratio of ΔfimA-H and ΔfimA-HΔgrpP/grpQ, and the colonization capabilities of these two strains were compared by calculating the competitive index (CI) at 2 h p.i. The CI averaged ~1.074 for ΔfimA-H versus ΔfimA-HΔgrpP/grpQ at 2 h p.i. (Fig. 3F), indicating that the colonization capability of ΔfimA-HΔgrpP/grpQ is similar to ΔfimA-H, suggesting that grpP/grpQ mutation did not confer additional invasion defects in the fimA-H mutant background. In vivo competition assays between WT-LIR (fimS phase-locked-on) and ΔgrpP/grpQ-LIR showed that the colonization capability of ΔgrpP/grpQ-LIR was similar to that of WT-LIR (CI ≈ 0.89) (Fig. 3G), suggesting that grpP/grpQ mutation did not confer additional invasion defects in the fimS phase-locked-on background. These results demonstrate that the contribution of GrpP/GrpQ to UPEC invasion of BECs is entirely dependent on type 1 fimbria. Further in vivo competition assays between ΔfimA-H and ΔgrpP/grpQ showed that the colonization capability of ΔfimA-H was lower than that of ΔgrpP/grpQ (CI ≈ 0.27) (Fig. S3A), suggesting that GrpP/GrpQ contributes partially to the type 1 fimbria-dependent invasion.

Collectively, these results demonstrated that GrpP/GrpQ promotes UPEC strain CFT073 invasion of BECs by activating type 1 fimbria expression.

GrpQ directly binds to fimS and facilitates its inversion to phase-ON orientation

We investigated whether GrpP/GrpQ activates type 1 fimbria by directly binding to fimS, an invertible element containing the fim gene cluster promoter23. Competitive electrophoretic mobility shift assays (EMSAs) revealed the appearance of slowly migrating bands for FAM-labeled fimS DNA with adding of phosphorylated GrpQ. Moreover, addition of unlabeled promoters effectively competed for GrpP binding to the labeled promoters (Fig. 4A), whereas no retarded bands were observed for the negative control (kanamycin cassette fragment), indicating that phosphorylated GrpQ directly binds to fimS in vitro. Chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) revealed a 6.5-fold enrichment of fimS in the GrpQ-ChIP sample compared with that in the control sample (Fig. 4B), indicating that GrpQ directly binds to fimS in bacterial cells. These results demonstrate that GrpQ directly binds to the fimS element to regulate the expression of type 1 fimbria genes. We performed a dye-based DNase I footprinting assay to identify precise GrpQ-binding sites on fimS. The result showed that GrpQ protects a 29-base pair motif (5′-TTACATATCAAACAGTTAGATGCTTTTTG-3′) in the fimS phase-ON orientation (Fig. 4C), situated from −219 bp to −191 bp relative to the proximal transcriptional start site. Mutation of this identified binding site completely abolished the binding of phosphorylated GrpQ to fimS (Fig. 4D), indicating phosphorylated GrpQ specifically binds to this site in fimS. Furthermore, qRT-PCR analysis indicated that fimB and fimE transcription in ΔgrpP/grpQ were similar to that in WT when cultured in LB medium (Fig. S3B), indicating that GrpP/GrpQ does not regulate type 1 fimbria expression via fimB and fimE.

Fig. 4: GrpQ directly binds to fimS and facilitates its inversion to the phase-ON orientation.
figure 4

A Competitive EMSAs of the binding of the fimS DNA fragment with phosphorylated GrpQ protein Kana fragment was used as the negative control. Representative images from three independent experiments. B Fold changes of the fimS DNA fragment in the GrpQ-ChIP sample relative to the mock-ChIP sample. Data were obtained from three independent experiments and presented as mean ± SD. P values were determined using a two-way analysis of variance. C DNase I footprinting assay identified precise GrpQ binding sites on fimS. The protected region shows a significantly reduced peak intensity (blue) pattern compared with that of the control (red). D Competitive EMSAs of the binding of purified phosphorylated GrpQ to the mutant fimS binding region. Representative images from three independent experiments. E Quantification of the percentage of bacteria with fimS in the phase-OFF orientation. F Phase-OFF orientation of fimS was calculated using qPCR in ΔgrpP/grpQ relative to WT. Data were obtained from three independent experiments and presented as mean ± SD (B, E, and F). P values were determined using two-way analysis of variance (B) or two-tailed Student’s t-test (E and F). Significance was indicated by a P value. n.s., No significant difference. B P = 0.00004; F P = 0.000099. Source data are provided as a Source data file.

Full size image

Next, we analyzed the effect of GrpQ binding to fimS during phase inversion. Using the classical procedure described by Stentebjerg-Olesen et al.43, we amplified a 601-bp fragment containing fimS from the genomic DNA of the indicated strains and digested this fragment using the restriction endonuclease SnaBI. The asymmetric location of the cleavage site within the invertible element allowed us to approximate, based on fragment size and band intensity, the percentage of bacteria predicted to be transcriptionally active with respect to the type 1 fimbria gene. Our results revealed that 54.8% of fimS was found in the phase-OFF orientation in ΔgrpP/grpQ compared with 30.3% in WT (Fig. 4E), indicating that GrpP/GrpQ facilitates fimS inversion to the phase-ON orientation. We also performed qPCR to confirm fimS inversion using primers Phase-OFF-F and Phase-OFF-R to amplify the OFF orientation, as described previously31. WT and ΔgrpP/grpQ were statically cultured in LB medium, and qPCR was performed with 50 ng of genomic bacterial DNA. The result showed that the number of bacteria in the OFF orientation for ΔgrpP/grpQ was 2-fold higher than that of the WT strain, by comparing the amount of OFF orientation amplification products (Fig. 4F). These results indicate that GrpQ facilitates the inversion of fimS from phase OFF to phase ON.

Collectively, these data indicate that GrpP/GrpQ activates type 1 fimbria expression via GrpQ binding directly to fimS to facilitate phase-ON inversion.

GrpP/GrpQ is activated in response to D-serine in the urine to promote UPEC invasion

Consistent with the role of GrpP and GrpQ in UPEC invasion (Fig. 2G), grpP and grpQ expression in WT-infected mouse bladders significantly increased at 2 h p.i., as measured by qRT-PCR (Fig. 5A), compared to that in LB medium, indicating the activation of grpP and grpQ by host-derived signals.

Fig. 5: GrpP/GrpQ is activated in response to D-serine.
figure 5

A Fold changes in the mRNA levels of grpP and grpQ in CFT073-infected BALB/c mouse bladders compared to that in CFT073 inoculum at 2 h p.i. (n = 5 mice). The mouse experiments were repeated thrice. B Fold changes in the mRNA levels of grpP and grpQ in CFT073 cultured in M9 medium with the indicated concentrations of D-serine compared to that without D-serine for 2 h (n = 3 independent experiments). C Phos-tag™ SDS-PAGE analysis of phosphorylation status of GrpQ in CFT073 cultured in M9 medium containing 0, 0.02, 0.1, 0.5, and 1 mM D-serine. Representative image from three independent experiments (upper). The bar graph shows the relative GrpQ-P/GrpQ. GrpQ-P phosphorylated GrpQ, GrpQ non-phosphorylated GrpQ (bottom). D qRT-PCR analysis of the mRNA levels of grpP, grpQ, fimA, or fimH in WT in the presence or absence of 1 mM D-serine (n = 3 independent experiments). E qRT-PCR analysis of the mRNA levels of fimA or fimH in WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+ in the presence of 1 mM D-serine (n = 3 independent experiments). F Regulation of type 1 fimbria expression by GrpP/GrpQ in the presence or absence of 1 mM D-serine in vitro. GFP activation visualized in WT and ΔgrpP/grpQ carried a fimS-GFP fusion reporter plasmid. Images are representative of three independent experiments. DIC, differential interference contrast. Scale bars, 5 μm. Data were obtained from three independent experiments and presented as mean ± SD (A, B, C, D, and E). P values were determined using a two-way analysis of variance (A, B, D, and E) or two-tailed Student’s t-test (C). Significance was indicated by a P value. n.s., No significant difference. A P = 0.000091; B P = 0.000003 (0 mM vs 0.1 mM); C P = 0.00000088 (0 mM vs 0.02 mM), P = 0.000077 (0 mM vs 0.1 mM); D P = 0.000093 (grpP), P = 0.000049 (fimH); E P = 0.000001 (fimA), P = 0.0000006 (fimH). Source data are provided as a Source data file.

Full size image

We also investigated the host signals responsible for activating grpP and grpQ. As UPEC transitions from the intestinal tract to the urinary tract, it encounters distinct environmental challenges, including abundant D-serine, limited iron, and low L-arginine concentration compared to the intestinal tract environment44. To identify the specific signal to which GrpP/GrpQ responds, we first determined the response of GrpP/GrpQ to abundant D-serine (1 mM), limited iron (5 μM), and low L-arginine concentration (50 μM) equivalent to that in human urine44, compared with the absence of D-serine, high iron concentration (1 mM), and high L-arginine concentration (500 μΜ) in the human intestinal tract. We assessed the mRNA levels of grpP and grpQ in the M9 medium supplemented with L-arginine, D-serine, or iron II sulfate at concentrations equivalent to those in the urine or intestinal tract. qRT-PCR analysis showed that alterations in L-arginine or iron II sulfate did not affect the mRNA levels of grpP and grpQ in the WT, whereas the presence of D-serine significantly upregulated the mRNA levels of grpP and grpQ (Fig. S4). Physiological level of D-serine is at a range from 0.02 to 1.09 mM in host urine45,46. We next determined the dose effect of D-serine (0–1 mM) on grpP and grpQ expression, and the result showed that the expression of grpP and grpQ was induced by D-serine at concentrations from 0.02 to 1 mM (Fig. 5B), indicating that D-serine at the physiological level is able to activate GrpP/GrpQ. Phosphorylation assay further showed that the phosphorylation level of GrpQ in response to D-serine was increased in the dose-dependent manner (Fig. 5C). These results suggest that D-serine in host urine is a trigger that induces the expression of grpP and grpQ.

To determine whether GrpP/GrpQ regulates type 1 fimbria expression in response to D-serine, we assessed the mRNA levels of two type 1 fimbriae genes, fimA and fimH, in the presence or absence of 1 mM D-serine47, when the bacteria were grown in M9 minimal medium. qRT-PCR analysis showed that the addition of D-serine increased the mRNA levels of fimA and fimH in the WT, consistent with grpP and grpQ results (Fig. 5D), indicating that D-serine induces the transcription of type 1 fimbria genes. However, grpP/grpQ mutation inhibited the upregulation of fimA and fimH induced by D-serine (Fig. 5E), indicating that grpP/grpQ is required to induce type 1 fimbria genes by D-serine. To verify the qRT-PCR results, we constructed a transcriptional fusion plasmid fimS-GFP containing fimS linked to a GFP reporter gene and transformed this plasmid into WT or ΔgrpP/grpQ. GFP expression in M9 minimal medium with or without D-serine was determined using confocal microscopy. In the WT strain, GFP expression was only detectable in the M9 minimal medium with D-serine (Fig. 5F). In contrast to the results obtained for WT with D-serine, GFP expression in ΔgrpP/grpQ was abolished in the presence of D-serine (Fig. 5F), confirming that D-serine acts as the signal for GrpP/GrpQ activation and, therefore, the expression of type 1 fimbria genes.

To further verify whether D-serine affects GrpP/GrpQ-dependent UPEC invasion, 5637 cells were infected with WT and ΔgrpP/grpQ in the presence or absence of D-serine. The results showed that the number of WT that invaded the cells was higher in the presence of D-serine than in its absence, whereas the number of ΔgrpP/grpQ was unaffected by the presence or absence of D-serine and lower than that of WT (Fig. 6A), indicating that GrpP/GrpQ responds to D-serine to promote UPEC invasion. ΔgrpP/grpQ displayed growth similar to WT in 1 mM D-serine M9 medium, indicating that the decreased invasion of ΔgrpP/grpQ is not due to a growth defect (Fig. S5A). The same as in the presence of 1 mM D-serine (Fig. 6A), the number of WT that invaded the cells was also higher in the presence of 0.02 mM D-serine, the lower limit of D-serine concentrations in host urine45, than in its absence (Fig. 6B). Accordingly, qRT-PCR analysis showed that both 1 mM and 0.02 mM D-serine treatment increased the expression of fimH in WT-infected 5637 cells compared to that of without D-serine treatment (Fig. 6C), while the expression of fimH in ΔgrpP/grpQ was decreased in contrast to that in WT and was not affected by D-serine treatment (Fig. 6C), indicating that D-serine activates type 1 fimbria expression during invasion, and the activation requires the presence of grpP and grpQ. These results confirmed that D-serine acts as a host signal to induce the expression of grpP and grpQ, and thus enhance the expression of type 1 fimbria to promote invasion.

Fig. 6: GrpP/GrpQ is activated in response to D-serine in the urine to promote UPEC invasion.
figure 6

A, B Bacterial titers in 5637 cells infected with WT or ΔgrpP/grpQ at 2 h p.i., in the presence or absence of 1 (A) or 0.02 mM (B) D-serine (n = 3 independent experiments). C Fold changes in the mRNA levels of fimH in WT, or ΔgrpP/grpQ that infected 5637 cells at 2 h p.i., with 0.02, 1 mM or without D-serine (n = 3 independent experiments). D Total bacterial titers in WT- or ΔgrpP/grpQ-infected BALB/c mouse bladders pretreated with siRNA targeting PHGDH or control siRNA at 24 h p.i. (n = 10 mice). E Intracellular bacterial titers in WT- or ΔgrpP/grpQ-infected BALB mouse bladders pretreated with siRNA targeting PHGDH or control siRNA at 2 h p.i. (n = 10 mice). Data were obtained from three independent experiments and presented as mean ± SD (A, B, C, D, and E). P values were determined using a two-way analysis of variance (C) or two-tailed Student’s t-test (A and B) or two-tailed Mann–Whitney U test (D and E). Significance was indicated by a P value. n.s., No significant difference. A P = 0.000077 (0 mM), P = 0.00000009 (1 mM); E P = 0.000011 (WT vs ΔgrpP/grpQ) P = 0.000076 (Control siRNA WT vs PHGDH siRNA WT). Source data are provided as a Source data file.

Full size image

To verify that D-serine acts as a signal to promote CFT073 invasion in vivo. Accell siRNA was used to deplete D-serine in mouse urine8,48 by targeting PHGDH, a rate-limiting enzyme in D-serine synthesis49. siRNA targeting PHGDH significantly reduced the production of PHGDH (85.5%), resulting in almost completely depletion of D-serine (0.0009 mM) in mouse urine (Fig. S5B, C). Mouse bladder colonization experiment was carried out using BALB/c mice pretreated with siRNA targeting PHGDH or control siRNA, by transurethrally infecting the mice with WT or ΔgrpP/grpQ. The bacterial CFUs in WT-infected mouse bladders pretreated with siRNA targeting PHGDH were lower (0.657-log-fold) than those with control siRNA at 24 h p.i. (Fig. 6D), indicating that the depletion of D-serine prevents UPEC colonization. The bacterial CFUs were higher (0.801-log-fold) in WT-infected mouse bladders pretreated with control siRNA than those in ΔgrpP/grpQ-infected, whereas the bacterial CFUs in WT-infected mouse bladders pretreated with siRNA targeting PHGDH were similar to those in ΔgrpP/grpQ-infected (Fig. 6D), indicating that the depletion of D-serine prevents UPEC colonization through the inactivation of GrpP/GrpQ. The same result in bacterial CFUs was also obtained at 2 h p.i. (Fig. 6E), indicating that the depletion of D-serine prevents UPEC invasion through the inactivation of GrpP/GrpQ. These results confirmed that D-serine in host urine acts as a signal to promote UPEC CFT073 invasion and thus virulence by activating GrpP/GrpQ in vivo.

Collectively, these findings suggest that D-serine is a host cue that induces GrpP/GrpQ activation and, therefore, type 1 fimbria production to promote UPEC invasion.

grpP and grpQ are highly associated with UPEC genomes

A previous study suggested that GrpP/GrpQ may be a strain-specific TCS in UPEC based on the presence of grpP and grpQ in the UPEC CFT073 genome, but not in EHEC O157 or non-pathogenic E. coli (K12 MG1655) genome38. We analyzed the presence of grpP/grpQ in 2263 publicly available complete genomes of E. coli, including UPEC (307 strains), avian pathogenic E. coli (APEC) (87), EHEC (279), enteropathogenic Escherichia coli (EPEC) (161) and enterotoxigenic E. coli (ETEC) (39), and commensal E. coli (45). We found that grpP/grpQ is present only in UPEC and APEC, which shares genetic similarity with UPEC50, but not in commensal E. coli and other gut pathogenic E. coli (EHEC, EPEC, and ETEC) genomes (Fig. S6). 87 out of 307 UPEC (28.1%) and 2 out of 87 APEC genomes (2.29%) contain grpP/grpQ, indicating grpP/grpQ is highly associated with UPEC and likely contributes to UPEC evolution.

We then transferred grpP/grpQ to two distinct evolutionary lineages of E. coli (K12 strain MG1655 and E. coli Nissle 1917), both are commensal E. coli that naturally lack these two genes. We compared the expression of fimH in the original MG1655 and Nissle 1917 and the strains containing grpP/grpQ cultured in M9 medium supplemented with D-serine (0.02 and 1 mM), or in human or mouse urine. The results showed that the expression of fimH was upregulated in the two commensal E. coli strains with grpP/grpQ compared with the strains without grpP/grpQ under all conditions, indicating that the presence of grpP/grpQ provides MG1655 and Nissle 1917 ability to increase type 1 fimbria expression in response to D-serine (Fig. S7A–F). This result further confirms that the acquisition of grpP/grpQ contributes to the UPEC evolution.

We further analyzed the serotypes of all 307 UPEC genomes using the method of SerotypeFinder51, and the distribution of grpP/grpQ among the serogroups. We found that grpP/grpQ is most frequently present in the isolates that belong to O1 (29 out of 30, 96.67%), which is strongly associated with cystitis UPEC clinical isolates than other serogroups, including O2, O4, O6, O7, O12, O15, O16, O18, O25, O7552. Cystitis UPEC isolates predominantly expresses type 1 fimbria16. The prevalence of grpP/grpQ in O1 clinical isolates suggests that this TCS contributes to the evolving of O1 virulence. grpP/grpQ is also present in the isolates that belong to O2 (14 out of 33, 42.42%), O101 (25 out of 64, 39.06%) and other 9 serogroups (15 out of 77, 19.48%) (Fig. S8), and expected to contribute to their virulence.

Discussion

The first and critical step in establishing a successful UTI involves invading the BECs on the bladder luminal surface by UPEC via type 1 fimbria53. By invading BECs, UPEC avoids bulk urine flow, accesses a more nutrient-rich environment, and evades both innate and adaptive host immune defenses12. The expression of type 1 fimbria is induced when UPEC invades the superficial BECs of mouse bladders16,19. However, their underlying regulatory mechanisms remain unclear. This study revealed a new virulence regulatory pathway mediated by a previously uncharacterized GrpP/GrpQ TCS, which activates type 1 fimbria expression to promote BEC invasion. A model depicting the GrpP/GrpQ-dependent virulence regulatory pathway has been proposed (Fig. 7). grpP/grpQ expression was activated in response to D-serine in host urine. Subsequently, GrpQ upregulated the expression of type 1 fimbria by directly binding to fimS element containing the fim gene cluster promoter, facilitating fimS inversion from the phase-OFF to the phase-ON orientation. This upregulated type 1 fimbria expression facilitated BEC invasion by UPEC, thereby promoting UTI. This study reported the regulatory mechanism underlying the activation of type 1 fimbria expression to promote invasion in response to a host cue and revealed a complete signaling transduction pathway.

Fig. 7: Schematic model of virulence regulation by GrpP/GrpQ.
figure 7

We proposed that once inside the urinary tracts, UPEC senses the presence of D-serine in urine through histidine kinase GrpP, which, in turn, transphosphorylates the response regulator GrpQ. Phosphorylated GrpQ directly binds to the fimS and activates the expression of type 1 fimbria, enabling UPEC to invade BECs and cause UTI.

Full size image

H-NS, LRP, and IHF in E. coli also regulate type 1 fimbria expression by directly binding to fimS25. These regulators bind to distinct sites on fimS, thereby affecting fimS phase inversion in different ways. H-NS maintains fimS in the OFF orientation by binding to sequences that straddle the left inverted repeat54. LRP binds to three distinct sites (sites 1, 2, and 3) in the fimS element, inducing sharp DNA bending55. IHF binds to two proposed fimS binding sites (sites I and II). IHF binding to site I (located outside fimS) may facilitate protein-protein or protein-DNA interactions that stimulate FimB recombination. Site II (inside fimS) plays a direct structural role in recombination via the ability of IHF to introduce sharp DNA bends55. IHF and LRP were reported to establish the phase-ON state of fimS by directly binding to fimS25. Our results revealed that the binding site of GrpQ in fimS partially overlaps with site II of IHF, suggesting that the specific mechanism of GrpQ in controlling the phase variation of type 1 fimbria may be similar to that of IHF. While GrpP/GrpQ directly regulates type 1 fimbria expression by binding to fimS, PhoB/PhoR and EnvZ/OmpR indirectly regulate type 1 fimbria expression31,33. The utilization of multiple TCSs, responsive to different environmental stimuli, to regulate type 1 fimbria phase transition could be of advantage for UPEC when adapting to different environmental challenges while colonizing various sites during infection.

During its transition from the intestine to the urinary tract, UPEC encounters distinct environmental challenges56. D-serine, which is abundant in urine, is toxic to many bacteria by inhibiting pantothenic acid synthesis57. In addition to using D-serine as a cue to promote initial BEC invasion, UPEC can also catabolize D-serine via expression of the D-serine deaminase DsdA, facilitating the growth of UPEC in the urinary tract32. D-serine has also been implicated as a regulatory signal that alters the transcription of other UPEC virulence-associated genes, including fmlA (encoding the major F9 fimbria subunit) and papA (encoding the major pap pilus subunit)58. F9 fimbria are advantageous for UPEC during chronic cystitis59, and P pili are associated with acute pyelonephritis60. Therefore, the utilization of D-serine by UPEC, both as a niche-specific signal for virulence gene activation and supporting growth within the host, significantly promotes UTIs. Our transcriptome analysis revealed a downregulation of the expression of F9 fimbria and P pili-related genes in ΔgrpP/grpQ compared with WT cultured in LB medium. grpP/grpQ is also abundantly present in O2 serogroup (14 out of 33, 42.42%), which highly expresses P-pili for kidney invasion61, as indicated by our serotyping result (Fig. S8). Whether GrpP/GrpQ regulates P pili-related genes to promote kidney invasion that can lead to pyelonephritis, and other virulence genes associated with UTI in response to D-serine and/or other host signals requires further investigation.

Concentrations of D-serine in human urine (0.02 to 1.09 mM)45,46 is in the similar range as that in mouse urine (0.03 to 1 mM)45. The result showed that the expression of grpP and grpQ was upregulated in CFT073 cultured in mouse urine for 2 h (Fig. S9A). As expected, the expression of grpP and grpQ, and thus fimH was also upregulated in CFT073 cultured in human urine for 2 h (Fig. S9B, C), and the concentration of D-serine (0.25 mM) was sufficient for GrpP/GrpQ activation (Fig. S9D and Fig. 5B). These results indicate that GrpP/GrpQ TCS is also activated in response to D-serine early in the infection to promote type 1 fimbria-dependent invasion.

UPEC has a serine protease62, and D-serine is depleted in the urine by serine protease of CFT073 cultured in human urine for 18 h63. D-serine was also almost completely depleted (0.004 mM) in human urine cultured with CFT073 for 24 h (Fig. S9D), and the expression level of grpP and grpQ in CFT073 in urine at this timepoint remained unchanged in comparison to that in M9 treatment (Fig. S9E), indicating that the depletion of D-serine turns off the GrpP/GrpQ TCS and thus GrpP/GrpQ-dependent type 1 fimbria upregulation. Previous studies reported that the expression of type 1 fimbria is downregulated in UPEC cultured in human urine for much longer period (e.g., 20 and 24 h)64,65, and the downregulation of type 1 fimbria expression leads to the inhibition on UPEC invasion65. Type 1 fimbria expression was also downregulated in CFT073 cultured in human urine for 24 h (Fig. S9F). However, other regulators also contribute to the downregulation of type 1 fimbria expression later in the infection64, as that the inactivation of GrpP/GrpQ only prevented the upregulation of type 1 fimbria (Fig. S9E, F).

Higher doses of D-serine (0.5 mM and above) have negative effect on bacterial growth66. Growth curve experiment in M9 medium supplemented with different concentrations of D-serine (0.02, 0.1, 0.5, and 1 mM) showed that the growth of CFT073 was not affected by 0.02 and 0.1 mM D-serine, but affected by 0.5 and 1 mM D-serine (Fig. S5A). The fact that D-serine at all physiological concentrations reported in host urine (0.02 to 1 mM), in regardless of whether or not affecting bacterial growth, induces grpP and grpQ expression (Fig. 5B), indicating that D-serine acts as a specific signal to activate GrpP/GrpQ. The smaller increase in the number of CFT073 that invaded the 5637 cells with 1 mM D-serine (compared to without D-serine) than the increase with 0.02 mM (Fig. 6A, B) is likely due to the negative effect of higher doses of D-serine on bacterial growth.

The expression of grpP and grpQ in UPEC CFT073-infected mouse bladders at 6 h p.i., when UPEC escapes into cytosol and forms IBC after the invasion17, was still upregulated (Fig. S9G), while the concentration of the D-serine in BECs (0.0054 mM) was very low (Fig. S9H) and insufficient to induce the expression of grpP and grpQ (Fig. 5B), indicating the presence of other factors in the cytosol of BECs that can activate the expression of grpP and grpQ. Previous study reported that type 1 fimbria is also upregulated during IBC formation to regulate biofilm surface structure formation17. The upregulation of grpP and grpQ could also contribute to IBC formation by upregulating type 1 fimbria expression during the IBC formation period, which requires further study.

A previous study showed that GrpP/GrpQ mediates the induction of the grp genes in response to 1,2-PD, as the deletion of either grpP or grpQ prevented the induction of grp genes by 1,2-PD under anaerobic conditions37. However, our RNA-seq results showed that the expression of grp gene cluster was undetected in WT and ΔgrpP/grpQ cultured in LB medium (aerobic conditions) (Fig. 3B). qRT-PCR analysis confirmed that the expression of grpR (a representative of grp gene cluster) was unchanged in ΔgrpP/grpQ compared with WT cultured with 1,2-PD under aerobic conditions (Fig. S9I), while was downregulated under anaerobic conditions (Fig. S9J), indicating that GrpP/GrpQ is activated by 1,2-PD only under anaerobic conditions. In contrast, the expression of fimH was downregulated in ΔgrpP/grpQ compared with WT cultured with D-serine under both anaerobic and aerobic conditions (Fig. S9K, L), indicating D-serine activates GrpP/GrpQ independent of oxygen condition. Urine is a condition of low oxygen67, and 1,2-PD is almost non-existent68, suggesting that 1,2-PD is unlikely to contribute to GrpP/GrpQ activation in host urine. In contrast, D-serine that is abundantly present in host urine is responsible for GrpP/GrpQ activation as shown in this study.

Notably, grpP/grpQ was mostly predominant in UPEC clinical strains and occasionally in APEC strains that share genetic similarity with UPEC50 but not in other pathogenic E. coli, including EHEC, EPEC, and ETEC. The activation of GrpP/GrpQ by D-serine, which is abundantly present in urine, implies that grpP/grpQ is specifically acquired by UPEC under host-derived selective pressure during co-evolution. In addition to GrpP/GrpQ, two other TCSs, KguS/KguR and OrhK/OrhR, were predominant in UPEC strains and contributed to UPEC pathogenicity. KguS/KguR responds to α-ketoglutarate and enables UPEC to use α-ketoglutarate as a carbon source under anaerobic conditions in vitro. Deletion of kguS/kguR in UPEC CFT073 reduces its colonization of mouse bladders and kidneys; however, the underlying mechanisms are unclear69. OrhK/OrhR responds to host-derived reactive oxygen species (ROS) and regulates a putative methionine sulfoxide reductase system to repair bacterial methionine residues damaged by ROS, thereby facilitating the survival of UPECs in macrophages. OrhK/OrhR also activates hemolysin expression in response to ROS, triggering hemolysin-induced pyroptosis in renal epithelial cells and macrophages in vitro. However, whether OrhK/OrhR contributes to UPEC colonization in vivo remains unclear70. The utilization of multiple TCSs enables UPEC to adapt to various environmental changes in diverse niches, indicating the importance of acquiring TCSs during UPEC evolution.

In conclusion, our findings reveal that UPEC uses a specific TCS, GrpP/GrpQ, to sense D-serine in host urine and activate type 1 fimbria expression, enabling UPEC to invade BECs and cause UTI. This offers new insights into UPEC pathogenesis, making GrpP/GrpQ a promising target for controlling UPEC infections.

Methods

Ethics statement

All animal experiments were conducted according to protocols approved by the Institutional Animal Care Committee of Nankai University (Tianjin, China) and performed under protocol no. 2020030501. Human urine collected under the guidelines approved by the Nankai University Institutional Review Board (NKUIRB2024021). Human urine specimens were collected and pooled from 2 healthy female donors with no history of UTI, written informed consent was acquired from each donor.

Cells and culture conditions

Human bladder epithelial cell 5637 (ATCC HTB-9) was purchased from the American Type Culture Collection (Manassas, USA) and cultured in RPMI 1640 medium (Gibco; 61870036) containing 10% fetal bovine serum (FBS, Gibco; 10100147) at 37 °C in a 5% CO2 incubator.

Bacterial strains and culture conditions

The bacterial strains and plasmids used in this study are listed in Tables S2 and S3. Primers used in this study are listed in Supplementary Data 1. The UPEC strain CFT073 was used as the WT strain in this study. Mutant strains, including ΔgrpP/grpQ, ΔfimA-H, ΔgrpQ, ΔfimA-HΔgrpP/grpQ, WT-LIR, and ΔgrpP/grpQ-LIR were generated using the λ-Red recombinase method. The complemented strain ΔgrpP/grpQ+ was generated by ligating grpP and grpQ with their promoter into the plasmid pBluescript II KS (+) and then transforming the recombinant plasmid into ΔgrpP/grpQ. To express GrpP and GrpQ proteins, DNA fragments of grpP or grpQ were ligated into pET-28a (+), and the recombinant plasmid was transformed into E. coli BL21 (DE3) cells for protein expression. For ChIP-qPCR, inducible expression vector pTRC99a carrying 3 × FLAG-tagged grpQ was constructed and transformed into ΔgrpQ. To generate the FimH-Flag strain, the primers were designed to amplify the KanR resistance cassette and the 3 × Flag tag-coding sequence, flanked by 50 bp from the targeted chromosomal region. The PCR products were transformed in the WT, ΔgrpP/grpQ, and ΔgrpP/grpQ+ strains harboring the pSim plasmid for λ red recombinase expression. Bacterial strains were generally cultured in LB medium under aerobic conditions at 37 °C. Antibiotics were supplemented in the medium when required, using the following final concentrations: ampicillin, 100 μg/mL; chloramphenicol, 25 μg/mL; gentamycin, 100 μg/mL; and kanamycin, 50 μg/mL.

Autophosphorylation and transphosphorylation assays

Autophosphorylation and transphosphorylation assays were performed as previously described39. The autophosphorylation reaction was conducted by purifying GrpP-GST, which was co-incubated with ATP in 1× kinase buffer (CST; 9802S). The reaction was performed at 37 °C for 1 min, 5 min, 15 min, or 30 min, and terminated by adding sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) protein loading buffer. After SDS-PAGE separation, the proteins were transferred to a polyvinylidene fluoride (PVDF) membrane and detected using a pIMAGO-biotin Phosphoprotein Detection Kit (TYMORA; 801) following the manufacturer’s protocol. For transphosphorylation, the autophosphorylated forms of GrpP and ATP were mixed with GrpQ-His6 protein, and the reaction mixture was incubated at 37 °C for 5 min, 10 min, 20 min, or 30 min. The transphosphorylated proteins were detected as described above.

Mouse infection assays

Bladder infections were induced as described previously71. Six-week-old female BALB/c mice were purchased from Beijing Vital River Laboratory Animal Technology (Beijing, China). All mice were group-housed in a temperature-controlled room in a specific pathogen-free facility with a 12 h light/12 h dark cycle. Mice were given water and standard chow diet ad libitum. Overnight-cultured bacteria were pelleted and resuspended in sterile PBS at a concentration of ~3 × 108 CFU/mL. Mice were infected via intraurethral catheterization with 50 μL of the bacterial suspension (~1 × 107 CFU) after being anesthetized with intraperitoneally injected 3% chloral hydrate (Sangon Biotech; A600288). For mouse colonization experiments, at 6, 24, or 48 h p.i., BALB/c mice were sacrificed, and bladders were harvested aseptically. The bladders were aseptically removed and homogenized in sterile PBS (0.5 mL). Homogenates were serially diluted and cultured on LB agar plates for total bacterial CFU determination. For mouse invasion experiment, at 2 h p.i., BALB/c mice were sacrificed, and bladders were harvested aseptically. Bladders were treated with 100 μg/mL gentamicin sulfate for 30 min to kill extracellular bacteria. The bladders were aseptically removed and homogenized in sterile PBS (0.5 mL). Homogenates were serially diluted and cultured on LB agar plates for intracellular bacterial CFU determination. For in vivo competitive assays, mice were co-challenged with the inoculum of 1 × 107 bacteria of the ΔfimA-H and ΔfimA-HΔgrpP/grpQ, WT-LIR and ΔgrpP/grpQ-LIR, or ΔgrpP/grpQ and ΔfimA-H. At 2 h p.i., the CI was calculated as the ratio of recovered intracellular bacterial CFU of ΔfimA-H to that of ΔfimA-HΔgrpP/grpQ, of ΔgrpP/grpQ-LIR to that of WT-LIR, or of ΔfimA-H to that of ΔgrpP/grpQ.

IBC counting and imaging

IBC fluorescence analysis was conducted as previously described with minor modifications72. The infected bladders of BALB/c mice were removed and placed onto a silicone bladder pinning pad covered with 100 μg/mL gentamicin sulfates for 30 min, then fixed with pre-chilled 4% paraformaldehyde for 30 min. The tissues were permeabilized with 0.1% Triton X-100 for 30 min and blocked with 5% bovine serum albumin (BSA) in PBS for 30 min. Subsequently, bladders were stained with mouse anti-E. coli LPS antibody (1:100; Abcam; ab35654) overnight at 4 °C. The following day, the bladders were washed thrice with PBS and incubated with the appropriate secondary antibody, goat anti-mouse IgG H & L (Alexa Fluor® 488, 1:200, Abcam, ab150113), for 1 h. They were subsequently stained with 1 μg/mL wheat germ agglutinin (WGA, Invitrogen; W32464) for 5 min, washed with PBS, and stained with DAPI (Abcam; ab104139) for 5 min. The IBC in the mouse bladders were counted using confocal microscopy from top to bottom and from left to right.

Bacterial invasion of 5637 cells

We seeded 5637 cells in 24-well plates at a density of 2 × 105 cells/well 24 h prior to infection. Cells were infected with the indicated strain (WT, ΔgrpP/grpQ, or ΔgrpP/grpQ+) at a multiplicity of infection (MOI) of 100 for 1 h, followed by an additional 1 h of incubation in the medium containing 100 μg/mL gentamicin sulfate to kill extracellular bacteria. After 1 h, the plate was washed thrice with PBS. Each well was lysed with 200 μL 0.1% Triton X-100. Intracellular bacteria were counted by plating serial dilutions onto LB agar plates.

RNA extraction and qRT-PCR

Total RNA was isolated using TRIzol Reagent (Invitrogen; 5596026), following the manufacturer’s protocol. The RNA was determined using a NanoDrop2000 spectrophotometer (Thermo Fisher Scientific). For qRT-PCR analysis, total RNA was reverse transcribed using the PrimeScript Reverse Reagent Kit (Thermo Fisher Scientific; A15299) to synthesize cDNA. The cDNA was used as a template mixed with SYBR Green PCR Master Mix kit and forward and reverse primers. qRT-PCR was performed using an ABI 7500 Thermocycler Sequence Detection System (Applied Biosystems). The expression levels of the indicated genes were normalized to gyrA. Fold changes in transcript levels were calculated using the 2−ΔΔCt method.

RNA sequencing

Total RNA was isolated as described above. RNA quality and quantity were assessed using a NanoDrop spectrophotometer (Thermo) after each manipulation step. The MICROBExpress Bacterial mRNA Enrichment Kit (Thermo Fisher Scientific, AM1905) was used to deplete bacterial 23S and 16S rRNAs. Illumina library preparation, sequencing, and data analysis were conducted by Novogene company. RNA Sequencing data were mapped to the genome of CFT073 (AE014075.1). Normalized gene locus expression levels were calculated as RPKM. The genes showing an RPKM R10 were chosen for subsequent statistical analysis. Differential gene expression was analyzed using the edgeR package73 with the default parameters. Genes exhibiting p values < 0.05 and a threefold or greater difference in RPKM between two conditions were defined as differentially expressed. Raw sequence read files are available in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA980889.

Western blotting

Bacterial strains were grown statically overnight in LB at 37 °C. Bacterial lysates were harvested by centrifugation at 6000 × g for 5 min and washed twice with PBS (pH 7.4). The pellet was resuspended in 4× SDS-PAGE loading buffer (Takara; 9173) and boiled at 100 °C for 15 min. Samples were subjected to 15% SDS-PAGE and transferred to a PVDF membrane (Millipore; ISEQ00010). Membranes were blocked with 5% BSA in Tris-buffered saline with Tween 20 (TBST) and incubated with the primary antibody, anti-Flag M2 (1:1000; Sigma; F1804), anti-PHGDH (1:1000; CST; 13428S), anti-β-actin (1:1000; CST; 4967) or anti-DnaK (1:1000; Abcam; ab69617) at 4 °C overnight. After washing thrice with 0.1% TBST, the membranes were incubated with the secondary antibody, HRP-conjugated goat anti-rabbit IgG (1:10,000; Abcam; ab205718), HRP-conjugated rabbit anti-mouse IgG (1:10,000; Abcam; ab6728) for 1 h at 25 °C. The protein band was detected by using the ChemiDocTM MP imaging system (Bio-red) with enhanced chemiluminescence western blotting reagents (Thermo; 34580). DnaK was used as a loading control. The bands were quantified using ImageJ densitometry and normalized to that of DnaK.

HA assay

Bacterial strains were cultured statically in LB medium for 24 h. HA assay was performed on normalized cells (OD600 = 1), and 1 mL bacteria were harvested by centrifuging 6000 × g for 5 min and then suspended in 100 μL of PBS. Bacteria (25 μL) was serially diluted and plated on HA plates containing 25 μL PBS per well. An equal volume of 1% guinea pig erythrocyte suspension was added to the microtiter plates. The plates were fully mixed and incubated at 4 °C for 12 h. HA titers were determined as the endpoint dilution ratio before erythrocyte agglutination.

EMSAs

The GrpQ-His6 fusion protein was purified from E. coli BL21 (DE3) as described previously74. Competition EMSAs were used to study the binding of GrpQ to DNA probes. DNA probes of fimS and the mutant fimS (the binding motif was mutated to 5ʹ-CCGTGCGCTGGGTGACCGAGCATCCCCCA-3ʹ) were amplified respectively with and without 6-FAM-labeled primers and purified (QIAGEN; 28006). 10 ng of FAM-labeled DNA probe and various concentrations of unlabeled DNA fragments (0 to 500 ng) were incubated with GrpP in binding buffer (100 mM Tris [pH 8.0], 100 mM KCl, 20 mM MgCl2, 0.1 mM dithiothreitol, and 20% glycerol) in the presence or absence of 30 mM acetyl-phosphate. The mixture was subjected to electrophoresis on an 8% native polyacrylamide gel in 0.5× TBE buffer (44.5 mM Tris [pH 8.0], 44.5 mM boric acid, and 1 mM EDTA) at 90 V for 2 h. FAM-Labeled fragments were visualized by Amersham Imager 600 (GE Healthcare).

Chromatin immunoprecipitation (ChIP) qPCR

ChIP assay was performed as described75. Bacteria were incubated at 37 °C to an OD600 of 0.3 and induced by 0.1 mM IPTG for 1 h. The bacteria were harvested and crosslinked by adding 1% formaldehyde for 25 min. The reactions were stopped using 0.5 M glycine at 25 °C for 5 min. Cells were harvested, lysed in lysis buffer (50 mM Tris [pH 7.5], 1 mM EDTA, 100 mM NaCl, 1 mM protease inhibitor cocktail, and 20 mg/mL lysozyme), and incubated at 37 °C for 30 min. The DNA was then fragmented into approximately 500 bp fragments using a sonicator. After centrifugation at 12,000 × g for 10 min, DNA-containing supernatants were used for IP with an anti-3× FLAG antibody (Sigma-Aldrich; F1804) and protein A magnetic beads (Invitrogen; 10002D). ChIP was performed using an aliquot without antibodies as the negative control. The sample was incubated with RNase A (Thermo; EN0531) and proteinase K solution for 2 h at 37 °C and 55 °C to degrade RNA and protein, respectively. The DNA samples were subsequently purified using a PCR purification kit (Qiagen; 28104). ChIP-qPCR was performed to measure the enrichment of fimS and Kana fragments (negative controls) using an ABI 7500 thermocycler sequence detection system. The relative fold enrichment was calculated as fold change using the 2−ΔΔCt method.

DNase I footprinting assay

The DNase I footprinting assay was performed as described76. Briefly, the fimS region was amplified using CFT073 as a template, 6-FAM-labeled forward primers (with 6-FAM modification at the 5ʹ end), and reverse primers. 6-FAM-labeled probe (40 nM) was incubated with 4.0 μM GrpQ in band-shift buffer (10 mM Tris-HCl [pH 7.5], 0.2 mM DTT, 2.5 mM MgCl2, 50 mM KCl, and 10% glycerol), after partially digesting the protein-DNA mixture with 0.05 units of DNase I (Sigma-Aldrich; AMPD1) for 10 min at 37 °C. The digestion reaction was stopped by heating at 85 °C for 10 min. Genotype samples were analyzed by MAP Biotech Co., Ltd. (Shanghai, China). The DNA sequences were analyzed using a Peak Scanner (Applied Biosystems).

Assay for fimS orientation

The orientation of fimS was assayed by PCR amplification and restriction enzyme digestion as described43. Briefly, genomic DNA was used as a template for PCR amplifying the invertible element using primers 5ʹ-CAGTAATGCTGCTCGTTTTGCCG-3ʹ and 5ʹ-CAGAGCCGACAGAACAACG-3ʹ to generate a 601-bp product. The PCR product was digested with SnaBI (Thermo Fisher Scientific; FD0404) to yield asymmetric products that differed between the ON (403 and 198 bp) and OFF (440 and 161 bp) orientations. The digested products were separated on 2% agarose electrophoretic gels, stained with Gel-Red, and visualized by ultraviolet transillumination.

Quantitative PCR (qPCR)

qPCR was performed using 50 ng of genomic DNA extracted from WT or ΔgrpP/grpQ strains grown under static culture conditions. Primers Phase-OFF-F and Phase-OFF-R were used to amplify the OFF orientation (Table S4). The threshold cycle (Ct) of the OFF orientation was normalized to that of gyrA. Fold change was calculated using the 2−ΔΔCt method.

Determination of GrpQ phosphorylation state

Bacterial strains were grown overnight in LB and diluted 1:100 in fresh M9 medium. When the cultures were grown for 6 h and reached an optical density of OD600 ~ 0.6, the indicated of D-serine was adding to bacterial culture and followed by 2 h. Then bacterial cells were harvested, washed, and resuspended in PBS at 4 °C. The resuspension solution was sonicated for 3 min and centrifuged at 12,000 × g for 10 min at 4 °C to remove cell debris.

GrpQ phosphoproteins were separated and detected using Phos-tagTM SDS-PAGE (FUJIFILM; 198-17981).

Serine measurement in urine

Measurement of D-serine were conducted using a fluorometric DL-Serine assay kit (Abcam；ab241027). Briefly, pooled, filter-sterilized human and mice urine from female donors was deproteinized using the sample cleanup mix provided in the kit, filtered through a 10-kDa-molecular-weight-cutoff spin column, and assayed in triplicates according to the manufacturer’s protocol.

In vivo RNA silencing

BALB/c mice were intraperitoneal injected with Accel siRNA with a nontargeting control sequence (Dharmacon Inc.; D-001910) or sequences simultaneously targeting four different regions of mouse PHGDH (SMARTpool, Dharmacon Inc.; E-045115). 24 h later, the treatment was repeated. At 72 h after the final treatment, the mice were transurethrally infected with 1 × 107 CFU of CFT073. At 2 h or 24 h p.i. the mice were sacrificed, and the intracellular or total bacterial CFUs in mouse bladders were determined as described above.

Statistical analyses

Statistical analyses were performed using GraphPad Prism version 9.0.1 (GraphPad Software, La Jolla, CA, USA). Data were compared using two-tailed unpaired Student’s t-test, two-way analysis of variance, or two-tailed Mann–Whitney U test, according to the test requirements. P < 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.