NsrP is an sRNA that negatively regulates NMEC virulence

By analyzing our previous RNA sequencing (RNA-seq) data of RS218 derived from the blood of mice that were injected with NMEC RS218 via the tail vein¹⁸, we identified a potential NMEC-specific sRNA, which we named NsrP. NsrP is encoded by RDI 13, which is reportedly involved in the invasion of human brain microvascular endothelial cells (HBMECs), an in vitro BBB model^19,26. This finding suggests that NsrP may play a role in NMEC virulence. Utilizing an NsrP-specific probe, we performed Northern blotting to determine whether NsrP is an sRNA transcribed in RS218. The findings showed an RNA band for the RS218 wild-type strain (WT) and the ΔNsrP complemented strain (cNsrP, which was generated by introducing a plasmid containing NsrP with its native promoter into ΔNsrP), but not for the ΔNsrP strain, indicating that NsrP is an sRNA transcribed from the reverse strand of the NMEC RS218 genome (Fig. 1a).

a Northern blotting was performed with a specific probe directed against NsrP sRNA in the NMEC WT, ΔNsrP, and ΔNsrP complemented (cNsrP) strains. 5S rRNA was used as a loading control. b The genomic position of the NsrP sequence in the NMEC RS218 genome. c Northern blotting for NsrP in RNA preparations from NMEC WT, and Δhfq. d, e Bacterial counts in the blood (CFU/mL, d) and meningitis development (e) were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the WT, ΔNsrP, or cNsrP strain (n = 8 for each group in d). CSF culture positivity was defined as meningitis (e). f H&E staining of brain sections was performed 4 h after intravenous injection of 1 × 10⁶ CFU of the WT or ΔNsrP strain. The right panels are higher-magnification images of the boxed regions, showing meningeal thickening and neutrophil infiltration (arrows). Scale bar, 100 μm. The images shown are representative of three independent experiments. g Survival curves of 2- to 5-day-old rats subcutaneously injected with 1 × 10⁵ CFU of the WT or ΔNsrP strain (n = 16 for each group). The number of living animals was recorded every 8 h. ns, nonsignificant. The two-tailed Mann–Whitney U-test (d), and the log-rank (Mantel–Cox) test (g) were applied.

The genomic location of the NsrP is shown in Fig. 1b. The 5′ and 3′ rapid amplification of cDNA ends (RACE) technique was used to determine the transcription start and termination sites of NsrP. The results showed that NsrP was exactly 244 nucleotides in length, spanning coordinates 2,920,694 to 2,920,937 in the genome (Supplementary Fig. 1a). Since Hfq is an indispensable chaperone for both sRNAs and their target mRNAs to facilitate base pairing and to protect sRNAs from degradation by cellular nucleosidases¹⁴, we therefore examined whether NsrP is hfq dependent. Northern blotting showed that the expression of NsrP exhibited no significant difference in the Δhfq strain compared with that in the WT strain (Fig. 1c). Moreover, when bacterial transcription was blocked by rifampicin treatment, the half-life of NsrP showed no significant difference in the Δhfq strain compared with that in the WT strain, indicating that NsrP is an Hfq-independent sRNA (Supplementary Fig. 1b).

To investigate the potential influence of NsrP on the pathogenicity of NMEC, we initially performed HBMEC invasion assays. Our findings indicated that the deletion of NsrP did not have any discernible impact on the capacity of the bacteria to invade the HBMECs (Supplementary Fig. 1c). To determine the potential role of NsrP in the development of bacteremia and meningitis in vivo, we used a mouse model of experimental hematogenous meningitis¹⁸. The results showed that the quantity of ΔNsrP in the blood was 7.26 ± 0.30 log CFU/ml, which was 10.42-fold higher than that of the WT strain (6.29 ± 0.22 log CFU/ml) and 11.10-fold higher than that of the cNsrP strain (6.25 ± 0.24 log CFU/ml), suggesting that NsrP hinders the survival and replication of NMEC within the murine bloodstream (Fig. 1d). In addition, CSF was collected from the mice to assess bacterial penetration across the BBB. The results showed a significant increase in the incidence of E. coli meningitis (defined on the basis of positive CSF cultures) in ΔNsrP-infected mice compared with that in mice infected with the WT or cNsrP strain (Fig. 1e). This finding is consistent with the fact that achieving a high level of bacteremia is a prerequisite for meningitis development for NMEC⁶. Since NsrP overlaps with a hypothetical open reading frame (orf14550), we mutated the putative start codon of orf14550 in the chromosome of WT (from ATG to CTG), generating strain orf14550^mut, and found that this site mutation did not influence the virulence of NMEC (Fig. 1b and Supplementary Fig. 1d, e), suggesting the influence of NsrP on the virulence of NMEC is not related with orf14550.

Additionally, brain sections of the infected mice were examined by hematoxylin and eosin (H&E) staining. Compared with those in WT strain-infected mice, thickened meninges with neutrophil infiltration were observed in the mice infected with ΔNsrP (Fig. 1f), demonstrating that NsrP deletion leads to aggravated inflammation in the cerebral meninges. We next investigated the influence of NsrP on NMEC virulence by monitoring the survival of 2- to 5-day-old rats infected subcutaneously with WT or ΔNsrP NMEC²⁷. Survival curves showed that all the neonatal rats infected with the ΔNsrP strain died at 48 h post-infection, while only 50% of the neonatal rats infected with the WT strain died at 48 h post-infection (Fig. 1g), suggesting that the mortality induced by the ΔNsrP strain was accelerated compared with that induced by the WT strain. These results demonstrated that NsrP inhibits bacterial survival in the blood and penetration of the BBB, resulting in a decrease in bacterial virulence.

NsrP inhibits NMEC virulence by repressing purD expression

In total, 387 putative target genes of NsrP were predicted in the genome of RS218 using TargetRNA3 (Supplementary Data 1). Among these potential target genes, purD may be a potential target of NsrP because the predicted base-pairing domain was included in the purD mRNA near the initiation codon (Fig. 2a). The purD gene encodes a phosphoribosylamine-glycine ligase that has been identified by transposon insertion sequencing as an NMEC gene required for survival in human serum^20,28. qRT-PCR assays revealed that purD expression was significantly greater in the ΔNsrP strain than in the WT and cNsrP strains, suggesting that NsrP negatively regulates purD expression (Fig. 2b). To determine the expression of PurD at the protein level, a chromosomally 3×FLAG C-terminal fusion construct of purD was generated. Western blotting was performed to quantify the levels of FLAG-tagged PurD in LB medium for the WT, ΔNsrP and cNsrP strains. The results showed that, compared with the WT and cNsrP strains, the ΔNsrP strain exhibited increased PurD-FLAG production (Fig. 2c). These results demonstrate that NsrP inhibited PurD production.

a The regions of base pairing between NsrP and purD as predicted by the TargetRNA3 with the probability threshold of 0.25 and p-value of 0.05. Point mutations to generate the disrupted alleles. b qRT-PCR analysis of purD in the NMEC WT, ΔNsrP, and cNsrP. c Western blot analysis and quantification of PurD-FLAG in the NMEC WT, ΔNsrP, and cNsrP. DnaK was used as the loading control. d, e Bacterial counts in the blood (CFU/mL, d) and meningitis development (e) were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the WT, ΔNsrP, ΔpurD, ΔNsrPΔpurD or ΔpurD complemented (cpurD) strain (n = 8 for each group in d). CSF culture positivity was defined as meningitis. f H&E staining of the brain sections was performed 4 h after intravenous injection of 1 × 10⁶ CFU of the WT or ΔpurD. The right panels are higher-magnification images of the boxed regions, showing meningeal thickening and neutrophil infiltration (arrows). Scale bar, 100 μm. The images shown are representative of three independent experiments. g Survival curves of 2- to 5-day-old rats subcutaneously injected with 1 × 10⁵ CFU of the WT or ΔpurD (n = 16 for each group). The number of living animals was recorded every 8 h. h Growth curves of the WT and ΔpurD in M9 medium with or without 40 mM IMP. i, j Replication of the WT and ΔpurD in 60% NMS with (j) or without (i) 40 mM IMP. k, l Bacterial counts in the blood (CFU/mL, k) and meningitis development (l) were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the WT, ΔpurD, or cpurD with or without IMP (n = 8 for each group in k). CSF culture positivity was defined as meningitis (l). ns nonsignificant. In b, c, h–j, data were presented as the means ± SDs (n = 3 independent experiments). One-way ANOVA (b, c, i, j), two-tailed Mann–Whitney U-test (d, k), log-rank (Mantel–Cox) test (g), and two-way ANOVA (h) were applied.

To determine the effect of purD on NMEC pathogenicity, we performed HBMEC invasion assays and found that there was no significant difference in the bacterial capacity to invade HBMECs among the WT, ΔpurD and ΔpurD complemented (cpurD) strains (Supplementary Fig. 2a). A previous investigation established the limited availability of nucleotide precursors in blood^23,29. In addition, the inactivation of nucleotide biosynthesis genes in Salmonella enterica and Bacillus anthracis has been demonstrated to impede their growth in human serum²⁴. Therefore, we examined the impact of purD on the capacity of NMEC to elicit bacteremia and meningitis in a mouse model of experimental hematogenous meningitis. Compared with the WT strain, the ΔpurD strain elicited a 12.4-fold lower level of bacteremia in mice, with bacterial loads of 6.46 ± 0.32 log CFU/ml for the WT strain and 5.38 ± 0.28 log CFU/ml of blood for the ΔpurD strain) (Fig. 2d). Similarly, the development of E. coli meningitis was markedly suppressed in ΔpurD-infected mice compared to that in mice infected with the WT or cpurD strain (Fig. 2e). H&E staining of brain sections revealed thickened meninges with neutrophil infiltration in mice infected with WT, whereas no discernible histopathology was observed in the meninges of the animals infected with ΔpurD (Fig. 2f). Survival curves showed that 50% of the neonatal rats infected with the WT strain died 48 h post-infection, whereas those infected with the ΔpurD strain remained viable at the end of the observation period (96 h post-infection, Fig. 2g). Collectively, these results provide evidence that purD facilitates the survival and replication of NMEC within the murine bloodstream, which enhances bacterial penetration across the BBB and meningitis development.

Furthermore, we investigated whether NsrP regulates the virulence of NMEC via purD. Animal experiments showed that the double-mutant ΔNsrPΔpurD and ΔpurD strains induced comparable levels of bacteremia in mice, with no significant difference in their capacity to cause meningitis development (Fig. 2d, e). Both strains exhibited significantly reduced virulence compared to the WT and ΔNsrP strains in infected mice (Fig. 2d, e). Additionally, the ΔNsrP strain induced significantly higher levels of bacteremia and an increased incidence of meningitis in mice compared to the WT strain (Fig. 2d, e), consistent with the results presented above (Fig. 1d, e). These findings suggest that the deletion of NsrP had no effect on NMEC virulence in the ΔpurD background and that the regulatory effect of NsrP on the virulence of NMEC is mediated by purD.

Bacterial growth curves demonstrated that, compared with the WT strain, the ΔpurD strain exhibited a growth deficiency when cultured in the M9 medium (Fig. 2h), which is a minimal medium devoid of purines and their derivatives. However, when M9 medium was supplemented with 40 mM IMP, the growth of the ΔpurD strain was similar to that of the WT strain (Fig. 2h). In addition, we showed that compared with the WT strain, the ΔpurD strain exhibited a reduced capacity for growth in both normal mouse serum (NMS) and heat-inactivated normal mouse serum (HI-NMS) after 2 or 4 h of incubation in vitro (Fig. 2i and Supplementary Fig. 2b). Conversely, the WT and ΔpurD strains exhibited comparable growth rates in NMS and HI-NMS when supplemented with IMP (Fig. 2j and Supplementary Fig. 2c). Furthermore, animal experiments demonstrated that the intravenous injection of IMP enhanced the ability of the ΔpurD but not the WT and cpurD strains to induce bacteremia and meningitis in mice (Fig. 2k, l). These findings suggest that the survival impairment observed for the ΔpurD strain is attributable to the deficiency of purines in the bloodstream and the impairment of purine synthesis.

NsrP directly binds to purD mRNA

To test for a direct interaction between NsrP and purD mRNA in vitro, we performed an RNA electrophoretic mobility shift assay (REMSA) using in vitro–transcribed and purified NsrP and purD mRNA or the NsrP complementary strand as a positive control (NsrP+) (Supplementary Fig. 3a, b). NsrP-purD mRNA complexes were observed, and the band intensity increased as the NsrP concentration increased (Supplementary Fig. 3a), confirming the direct interaction between NsrP and purD mRNA in vitro.

IntaRNA analysis showed that NsrP may interact with the purD mRNA through nine-nucleotide base (CUAAUUACU) pairing beginning at nucleotide +12 in the purD mRNA (based on the ATG site) (Fig. 2a). The prediction of the secondary structure of NsrP sRNA via RNAfold revealed that this binding motif of NsrP is located in a loop structure (Supplementary Fig. 3c). To confirm that this motif is the key sequence of NsrP required for binding to purD mRNA, we performed REMSA using a mutated NsrP (NsrP^mut) (the motif AAUUACU was complementarily mutated to UUAAUGA, which did not affect the secondary structure of NsrP), and a mutated purD^mut mRNA (the sequence AGUGAUU was complementarily mutated to UCACUAA) (Fig. 2a). No interaction was observed between the NsrP^mut sRNA and purD mRNA or between the NsrP sRNA and purD^mut mRNA (Supplementary Fig. 3d, e). However, REMSA showed a direct interaction between the NsrP^mut sRNA and purD^mut mRNA (Supplementary Fig. 3f). Moreover, we constructed a plasmid containing purDH–lux fusion (purD and purH are located in an operon³⁰) (Fig. 3a and Supplementary Fig. 3g), and introduced this plasmid into ΔNsrP, generating strain ΔNsrP-purDH–lux. We found a significant reduction in the luminescence intensity of the ΔNsrP-purDH–lux strain when NsrP was overexpressed (Fig. 3b). In contrast, overexpressing NsrP^mut with mutated binding motif did not influence the luminescence intensity (Figs. 2a, 3b). Furthermore, there was no significant difference in the luminescence intensity of the ΔNsrP-purD^mutH–lux strain (the binding motif AGUGAUU of purDH–lux on the plasmid was complementarily mutated to UCACUAA) when NsrP was overexpressed, whereas a significant reduction in the luminescence intensity was observed when NsrP^mut was overexpressed in ΔNsrP-purD^mutH–lux strain (Fig. 3a, c). In addition, northern blotting showed that there is no significant difference between the expression of NsrP and NsrP^mut (Supplementary Fig. 3h, i). Collectively, these data indicate that NsrP binds directly and specifically to the purD mRNA through the identified motif (AAUUACU).

a Schematic of the structure of purDH–lux. The region of the purDH promoter and the expected full-length purDH without termination codon were cloned and inserted into pMS402 to construct a purDH–lux fusion. Point mutations to generate the disrupted alleles in the purD^mutH transcript are indicated. b, c Quantification of the PurD (b) or PurD^mut (c) protein expression level in the ΔNsrP-purDH–lux strain (b) or ΔNsrP-purD^mutH–lux strain (c) by measurement of luminescence. The ΔNsrP-purDH–lux and ΔNsrP-purD^mutH–lux strains were transformed with the expression construct pNM12 (Control), or pNM12 encoding NsrP or NsrP^mut. d qRT-PCR analysis of purD expression in rifampicin-treated the WT and ΔNsrP strains. e Graphical presentation of proposed interaction of NsrP sRNA with the purD mRNA, and of base-pair changes to generate NsrP^bind-mut and PurD^syn-mut. f–i Bacterial counts in the blood (CFU/mL, f, h) and meningitis development (g, i) were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the WT, NsrP^bind-mut, and ΔNsrP strains (f, g) or WT, PurD^syn-mut, ΔNsrP, and PurD^syn-mutΔNsrP strains (h, i) (n = 8 for each group in f, h). CSF culture positivity was defined as meningitis. In b–d, data were presented as the means ± SDs (n = 3 independent experiments). ns nonsignificant. One-way ANOVA (b–d), and two-tailed Mann–Whitney U-test (f, h) were applied.

To investigate whether NsrP influences the stability of purD mRNA, we performed a rifampicin transcription inhibition assay. The results showed that the purD mRNA was more stable in the ΔNsrP strain than in the WT strain (Fig. 3d), indicating that NsrP was detrimental to purD mRNA stability. Furthermore, we found that the expression of purD in the Δhfq strain exhibited no significant difference compared with that in the WT strain in rifampicin transcription inhibition assay (Supplementary Fig. 3j). In contrast, the expression of purD was significantly increased in the ΔhfqΔNsrP strain compared with that in the Δhfq strain (Supplementary Fig. 3k). These results suggest that NsrP-purD base pairing promotes purD mRNA degradation in an Hfq-independent way and thus decreases PurD expression.

To further identify the contribution of the NsrP binding motif to NMEC virulence in vivo, we mutated the NsrP binding motif (AAUUACU to UUAAUGA) on the chromosome of WT, generating strain NsrP^bind-mut. We also mutated the binding motif on purD by synonymous mutation (AGUGAUU to GGUCAUA) on the chromosome of WT, generating strain PurD^syn-mut (Fig. 3e). We found that the levels of bacteremia and the incidence of meningitis elicited by the NsrP^bind-mut and PurD^syn-mut strains in mice were significantly higher than that in WT-infected mice, which exhibited no significant difference from that in ΔNsrP-infected mice (Fig. 3f–i). Moreover, we further deleted NsrP in strain PurD^syn-mut (generating strain PurD^syn-mutΔNsrP), and found that the levels of bacteremia and the incidence of meningitis in mice infected by PurD^syn-mutΔNsrP exhibited no significant difference from that in mice infected by PurD^syn-mut (Fig. 3h, i). This finding suggests that the regulatory effect of NsrP on the virulence of NMEC is mediated by its binding to purD mRNA.

NMEC senses low leucine levels in the blood to suppress NsrP expression

qRT-PCR analysis of WT NMEC recovered from the blood of mice infected via tail vein injection showed that NsrP expression decreased compared with that of WT NMEC in LB medium (Fig. 4a). To investigate the potential factors regulating NsrP expression, an analysis of the NsrP promoter region was conducted using the online promoter prediction tool BProm. This analysis confirmed the presence of an Lrp-binding site within the NsrP promoter region (Fig. 4b). qRT-PCR analysis revealed a significant decrease in NsrP expression in the Δlrp strain compared with that in the WT strain when cultured in LB medium (Fig. 4c), indicating that Lrp positively regulates NsrP expression.

a qRT-PCR analysis of NsrP expression in the NMEC WT strain in LB medium and mouse blood. b The NsrP promoter region and the potential binding site of Lrp were predicted by the BProm program (SoftBerry). c qRT-PCR analysis of the NsrP expression in WT and Δlrp. d Biacore SPR kinetic analyses of Lrp binding to the promoter of NsrP. e Fold enrichment of the NsrP promoter in the Lrp-ChIP samples compared with that in the mock-ChIP samples. f qRT-PCR analysis of the NsrP expression in WT NMEC in M9 medium supplemented with 0.1 mM leucine (Leu-L) or 5 mM leucine (Leu-H). g Northern blotting of NsrP sRNA in WT NMEC in M9 medium supplemented with 0.1 or 5 mM leucine. 5S rRNA was used as the loading control. h, i qRT-PCR analysis of the NsrP expression in the Δlrp (h) and NsrP^pro-mut (i) strain in M9 medium supplemented with 0.1 or 5 mM leucine. In a, c, e–i, data were presented as the means ± SDs (n = 3 independent experiments). ns nonsignificant. Two-tailed unpaired Student’s t-test (a, c, f–i) and two-way ANOVA (e) were applied.

To investigate whether Lrp directly binds to the NsrP promoter, we performed a surface plasmon resonance (SPR) assay to monitor the real-time binding of Lrp to the NsrP promoter. The result showed that the purified Lrp (but not BSA) binds to the promoter of NsrP in a dose-dependent manner (Fig. 4d and Supplementary Fig. 4a). Consistently, chromatin immunoprecipitation-qPCR (ChIP-qPCR) analysis revealed 9.51-fold greater enrichment of the NsrP promoter region in Lrp-ChIP samples than in the mock-ChIP samples (Fig. 4e). In contrast, the fold enrichment of rpoS did not significantly differ between these two samples (Fig. 4e). These results indicate that Lrp specifically binds to the NsrP promoter region both in vitro and in vivo.

E. coli Lrp was initially identified as a leucine-responsive regulatory protein, but subsequent research has also indicated its responsiveness to various other small effector molecules, including alanine, methionine, threonine, lysine, histidine, and isoleucine³¹. Thus, we carried out qRT-PCR to analyze NsrP expression in response to each of these amino acids. Mid-logarithmic phase WT strain incubated in LB medium was collected and reincubated in M9 minimal medium supplemented with different concentrations of these amino acids for 30 min. The qRT-PCR results revealed that NsrP expression decreased at a low leucine concentration (0.1 mM), which mimics conditions found in human blood³², compared to its expression at a high leucine concentration (5 mM). Conversely, the expression of NsrP did not significantly change in response to the other amino acids (Fig. 4f and Supplementary Fig. 4b). Similarly, northern blot analysis also revealed a decrease in NsrP expression in the WT strain under conditions of low leucine levels, suggesting that NsrP expression is repressed by low leucine levels (Fig. 4g). Furthermore, we showed that leucine levels no longer regulate NsrP expression in the strains Δlrp and NsrP^pro-mut (with Lrp binding site on the promoter of NsrP was deleted from the genome) (Fig. 4b, h, i). These findings provide evidence that NMEC senses low leucine levels through Lrp to suppress NsrP expression by binding to its promoter directly.

NMEC promotes purD expression in mouse blood by sensing low leucine levels in an NsrP-dependent manner

As NsrP expression was repressed at a low level of leucine and NsrP served as a negative regulator of purD expression, our subsequent investigation focused on the impact of leucine on purD expression. qRT-PCR was performed on mid-logarithmic phase WT strains reincubated in M9 minimal medium supplemented with different concentrations of leucine (0.1 or 5 mM) for 30 min. The results revealed a significant increase in purD expression under low-leucine conditions (Fig. 5a). This finding is consistent with the observed upregulation of purD expression in mouse blood, which is a low-leucine environment compared with LB medium (Fig. 5b). To determine the impact of leucine on PurD expression at the protein level, a 3×FLAG-PurD WT strain was incubated in the same conditions. Western blotting results showed that PurD-FLAG production was significantly induced by low leucine levels (Fig. 5c). In contrast, in the Δlrp, ΔNsrP, and NsrP^pro-mut strains, the regulation of purD expression by leucine was no longer observed (Fig. 5d–f). These results demonstrated that NMEC senses low leucine levels in the blood to increase purD expression in an Lrp- and NsrP-dependent manner.

a qRT-PCR analysis of the purD expression in the NMEC WT strain in M9 medium supplemented with 0.1 or 5 mM leucine. b qRT-PCR analysis of the purD expression in WT NMEC in LB medium or mouse blood. c Western blot analysis and quantification of PurD-FLAG in WT NMEC in M9 medium supplemented with 0.1 or 5 mM leucine. DnaK was used as the loading control. d–f qRT-PCR analysis of the purD expression in the Δlrp (d), ΔNsrP (e), and NsrP^pro-mut (f) strain in M9 medium supplemented with 0.1 or 5 mM leucine. In a–f, data were presented as the means ± SDs (n = 3 independent experiments). ns, nonsignificant. Two-tailed unpaired Student’s t-test was applied.

In summary, our study demonstrated that NMEC can sense low leucine levels in the bloodstream, leading to a decrease in NsrP expression, which in turn increases purD expression in an Lrp-dependent manner. This phenomenon increases NMEC survival in the host bloodstream. These findings suggest that a low leucine level is beneficial for NMEC survival in the bloodstream, which is consistent with the finding that NsrP mutation promotes NMEC pathogenicity.

purD did not affect the intestinal colonization of NMEC

NMEC infections arise due to the colonization of the neonatal gastrointestinal tract by maternally derived NMEC at or soon after birth, followed by transcytosis through enterocytes into the bloodstream and ultimately the development of meningitis^6,7. In addition, the gut can serve as a reservoir for the recurrence of NMEC infection³³. To investigate the impact of purD on the colonization of NMEC in the intestine, we performed qRT-PCR to determine the expression of NsrP and purD in the rat intestine. The results showed that NsrP and purD expression in the WT strain that colonized the small intestine did not significantly differ from that in the LB medium (Supplementary Fig. 5a). Moreover, bacterial colonization in the small intestine also did not significantly differ among the NMEC WT, ΔpurD, and cpurD strains (Supplementary Fig. 5b), suggesting that purD was not necessary for NMEC intestinal colonization. Previous research has demonstrated that the small intestine of infants is abundant in purines and amino acids, including leucine^34,35,36,37. Therefore, the small intestine serves as a niche for NMEC infection that does not require upregulation of the de novo purine synthesis pathway.

Intravenous administration of leucine reduced NMEC virulence in mice by blocking the Lrp-NsrP-PurD regulatory pathway

The above results showed that NMEC senses low levels of leucine in the blood to increase the expression of purD by repressing NsrP expression, which results in significantly increased NMEC pathogenicity. Therefore, the leucine level may be a potential factor for the treatment of NMEC infection. Thus, mice infected by NMEC WT were treated with leucine through tail vein injection at 0.5 h post-infection to determine the effect of leucine on NMEC pathogenicity. The results showed that leucine treatment resulted in significantly increased or decreased bacterial expression of NsrP and purD in vivo, respectively, comparted to control mice (WT-infected mice receiving PBS) (Fig. 6a, b). Moreover, leucine-treated mice exhibited significantly decreased levels of bacteremia and inhibited development of meningitis compared to control mice (Fig. 6c, d). Additionally, the leucine-treated mice exhibited less thickened meninges with neutrophil infiltration than did the control mice (Fig. 6e). Conversely, there was no significant difference in bacteremia levels or the development of meningitis between leucine-treated mice and control mice when they are infected with strain ΔNsrP, ΔpurD, NsrP^pro-mut, NsrP^bind-mut– or PurD^syn-mut (Fig. 6f–k). Collectively, these data indicate leucine is able to reduce the pathogenicity of NMEC by blocking the Lrp-NsrP-PurD regulatory pathway.

a, b qRT-PCR analysis of the NsrP (a) and purD (b) expression in mouse blood were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the NMEC with the administration of leucine or PBS. (c, d, f–k) Bacterial counts in the blood (CFU/mL, c, f–j) and meningitis development (d, k) were determined 4 h after intravenous injection of 1 × 10⁶ CFU of the WT (c), ΔpurD (f), ΔNsrP (g), NsrP^pro-mut (h), NsrP^bind-mut (i), and PurD^syn-mut (j) strains with the administration of leucine or PBS at 0.5 h post-infection (n = 8 for each group). CSF culture positivity was defined as meningitis (d, k). e H&E staining of the brain sections was performed 4 h after intravenous injection of 1 × 10⁶ CFU of the WT strain and of leucine or PBS at 0.5 h post-infection. The right panels are higher-magnification images of the boxed regions, showing meningeal thickening and neutrophil infiltration (arrows). Scale bar, 100 μm. The images shown are representative of three independent experiments. l The distribution of NsrP in 930 publicly available complete E. coli genomes. In a, b, data were presented as the means ± SDs (n = 3 independent experiments). ns nonsignificant. Two-tailed unpaired Student’s t-test (a, b) and two-tailed Mann–Whitney U-test (c, f–j) were applied.

To investigate whether the administration of leucine, which targets the Lrp-NsrP-PurD regulatory pathway, could be a potential strategy for the treatment of E. coli bacteremia and meningitis, we analyzed the distribution of NsrP in E. coli by performing a comparative genomics analysis using 930 publicly available complete E. coli genomes. The genomes of the strains included 46 NMEC strains, 77 uropathogenic E. coli (UPEC) strains, 85 enteropathogenic E. coli (EPEC) strains, 366 EHEC strains, and 356 nonpathogenic E. coli strains. The results showed that NsrP was present in the majority of the NMEC strains (36/46, 78.3%) and in a few UPEC strains (12/77, 15.6%) whose serotypes were O1 and O18 (Fig. 6l, Supplementary Data 2). In contrast, NsrP was absent in the EHEC, EPEC, and nonpathogenic E. coli strains (Fig. 6l). Furthermore, among the 34 NMEC strains that possessed a K1 capsule (34/46, 73.9%, defined as E. coli K1⁺), 29 harbored NsrP (29/34, 85.3% in E. coli K1⁺), suggesting that NsrP is widely present in E. coli K1⁺ strains. These results indicate that NsrP is widely distributed in NMEC but not in nonpathogenic E. coli or other pathotypes of E. coli. Thus, the administration of leucine to target the Lrp-NsrP-PurD regulatory pathway is a potential strategy for the prevention and treatment of E. coli bacteremia and meningitis.