The Medical Ethics Committee of Southern Medical University approved all of the animal experiments (Protocol number: L2018018). All the experiments on mice were done according to the corresponding guidelines. Every attempt was taken to minimize the number and suffering of mice used. We purchased neonatal C57BL/6 mice (8 days old) from the Animal Experimental Center of Southern Medical University. The α7 nAChR heterozygous (A7R^+/−) mice with C57BL/6J background were obtained from the Jackson Laboratory (B6.129S7-Chrna7tm1Bay/J, Stock No: 003232, Bar Harbor, ME). Littermate A7R ^−/− and A7R^+/+ (wild-type) mice were generated from the heterozygous for the experiment. All animals were specific pathogen free and were kept on a 12-h light/dark cycle and free to get food and water.

Two transcriptional data of E. coli infection patients were retrieved from Gene Expression Omnibus database (GSE33341, GSE65088). The metadata and SLURP1 transcriptional levels, as measured by fragments per kilobase of exon per million reads mapped (FPKM), were directly extracted from the data sets.

The chemicals and reagents used in this study were obtained as follows: Sigma-Aldrich (St. Louis, MO, USA) for bull serum albumin (BSA), Evans blue, Triton X-100, Tween-20, 4′,6-diamidino-2-phenylindole (DAPI), isopropyl-β-d-thiogalactoside (IPTG), Coomassie brilliant blue G 250, rifampin, kanamycin, gentamicin, and methyllycaconitine citrate (MLA); Thermo Fisher Scientific (Waltham, MA, USA) for fluorescent α-bungarotoxin conjugates; Gibco (Thermo Fisher Scientific, Inc., Waltham, MA, USA) for penicillin G, streptomycin, glutamine, and pyruvate; Abcam (Cambridge, UK) for antibodies against SLURP1, α7 nAChR or β-actin; Proteintech (Proteintech Group, Chicago, IL, USA) for enzyme-linked immunosorbent assay (ELISA) kits. The rest reagents were purchased from Beyotime Institute of Biotechnology, Shanghai, China.

In this study, the amino acid sequence of SLURP1 we used to construct recombinant was described by previously (23), which is secreted by N-terminal signal cleavage (23-103 aa, as showed in Supplementary Figure S1A ). Total RNA was extracted and used to amplify the SLURP1 cDNA, using primers containing 5′ BamHI and 3′Not I restriction sites at their termini. The primers used in the study were: sense, 5′-CGGGATCCCTCAAGTGCTACACCTGCAA-3′, and antisense, 5′ TTGCGGCCGCTCAGAGTTCCGAGTTGCAGA-3′. The cDNA was ligated into BamHI- and NotI- digested pET-28a, and transformed into E. coli BL21(DE3). Bacteria was added into the LB broth (1:100, containing 50 μg/ml kanamycin) and incubated at 37°C for 3–4 h (OD≈1). Afterward, IPTG was added at the final concentration of 0.1 mM and incubated at 30°C for 8 h for induction of protein expression. The protein was expressed as inclusion body form and purified using His-Tagged Protein Purification Kit (KangWeiShiJi Inc, Beijing, China) at denatured condition according to the manufacturer’s instructions. Purified SLURP1 was refolded in a series of gradient solutions of urea containing 50 mM Tris-HCl (pH 7.0), 0.5 M l-arginine, 4 mM glutathione, 1 mM glutathione disulfide, and 20% glycerin. The potential endotoxin was removed by passing through a Detoxi-Gel Endotoxin Removing Gel (Pierce Biotechnology, Rockford, IL, USA).

E. coli K1 strain RS218 (O18:K1:H7) was isolated from the cerebrospinal fluid (CSF) of a meningitis neonate and showed rifampicin-resistant property (8, 24). The brain heart infusion broth was used to culture E. coli K1 at 37°C for 14 h, with supplementation of rifampin (100 μg/ml). The immortalized human brain microvascular endothelial cells (HBMEC) were isolated and cultured as described previously (8, 24, 25). RPMI 1640 medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) was used as basic medium, with the following supplementations: 10% fetal bovine serum from Gibco (Thermo Fisher Scientific, Inc., Waltham, MA, USA), 50 U/ml penicillin G, 50 μg/ml streptomycin, 2 mM glutamine, and 1 mM pyruvate according to previous studies (8, 25).

For invasion assays, HBMEC were cultured in 24-well plates and incubated with SLUPR1 (0.1–2 μg/ml) for 2 h, followed by infected with E. coli K1 [1 × 10⁷ colony-forming unit (CFU)] for another 2 h. To kill the extracellular bacteria, the HBMEC was washed twice with sterile PBS and incubated with RPMI 1640 medium containing gentamicin (100 μg/ml) for 1 h. Then, the HBMEC were washed again and lysed using sterile water. Internalized bacteria were counted by plating the cell lysates on Luria-Bertani broth agar (containing 100 μg/ml rifampicin).

Polymorphonuclear leukocytes (PMN) transmigration experiments were carried out as previously (8, 14, 25). HBMEC monolayers on transwell filters (3 μm pore size, 6.5 mm diameter, Corning, product number 3415) were monitored by measuring trans-endothelial electrical resistance (TEER) changes across the endothelial cell monolayer using an End Ohm epithelial voltohmeter (World Precision Instruments, Sarasota, FL, USA). To exclude the possibility that the PMN migration elicited was due to destruction of HBMEC monolayer, the integrity of the monolayer was inspected by TEER and microscopy before the start of the PMN transmigration assay. Transwell filters with or without supplementation of SLUPR1 (0.1–2 μg/ml) were employed to culture fully confluent HBMEC monolayers for 2 h. Then E. coli K1 (1 × 10⁵ CFU/ml) was added to the bottom of the Transwell filters and infected for another 2 h. Then PMNs were applied to the upper compartment at a concentration of 1 × 10⁶ cells. The Transwell filters system was kept at 37°C, 5% CO₂. After incubated for 4 h, the Transwell filters were removed and migrated PMNs in the bottom of 24-well plates were harvested and counted in a blinded manner.

SLURP1 expression was knockdown using RNA interference. In brief, predesigned siRNA specific for SLURP1 and nontargeting scrambled siRNA (control) were obtained from Santa Cruz Biotechnology (CA, USA). The Lipofectamine™3000 transfection reagents (Invitrogen, USA) were mixed with the siRNA solutions, applied to the HBMEC monolayers, and maintained at 37°C, 5% CO₂ for 24 h. The transfected HBMECs were used for invasion and PMN transmigration assays as described above.

For overexpression of SLURP1, the full-length cDNA of SLURP1 was cloned into a pcDNA3.1(+) vector to construct the pcDNA3.1-SLURP1 expression plasmid. The control vector or pcDNA3.1-SLURP1 plasmid was transfected into HBMEC for 24 h using Lipofectamine™3000. The transfected HBMECs were used for invasion and PMN transmigration experiments as described above.

For the transfection in the Transwell filters, HBMECs were seeded onto Transwell filters and grown to confluency, then transfected with siRNA or SLURP1 expression vector as mentioned above. In order to ensure the barrier function remains comparable between different groups, HBMEC transfected with nontargeting scrambled siRNA or control vector were served as the scrambled control.

From postnatal days 8 to 10, neonatal C57BL/6 mice were intraperitoneally injected with SLURP1 or BSA daily at a dose of 0–100 mg/kg body weight. To establish an E. coli K1 meningitis model, mice were injected with E. coli K1 (1 × 10⁶ CFU, in 20 μl PBS) intraperitoneally at day 10. Control mice were given 20 μl PBS using the same route of injection. After infection for 18 h, the blood samples were collected and plated on Luria-Bertani agar (containing 100 μg/ml rifampicin) plates. Puncture through cisterna magna were carried out to collect CSF samples, followed by inoculating into the Luria-Bertani agar plates (containing 100 μg/ml rifampicin). Mice were perfused with 30 ml sterile PBS by the intracardiac route. Then brain tissues were harvested under aseptic conditions and homogenized in saline. Serial tenfold dilutions of brain homogenates were carried out and plated on Luria-Bertani agar (containing 100 µg/ml rifampicin) for counting. CSF samples were stained with a FITC-Ly-6G (Gr-1) (ProteinTech Group, Chicago, IL, USA) antibody and counted under fluorescence microscopy for PMN counting. For Evans blue assay, mice were injected intraperitoneally with Evans blue at a concentration of 40 mg/kg body weight 3 h before sacrificing. After intracardiac perfused with 30 ml PBS, the brain tissues were harvested and immersed in formamide. The OD620 of the supernatant was measured using a spectrophotometry.

Brain tissues were kept in formalin and transported to histological examination. After cutting into 3 µm sections, the tissues were hematoxylin-eosin (H&E) stained to assess tissue inflammation and damage. For the immunohistochemical staining, xylene was first used to dewax paraffin sections for 10 min, followed by gradient alcohol to dehydrated and rinsing in distilled water. Then the sections were heated in citrate buffer solution at 100°C for 40 min to retrieve the antigen. Hydrogen peroxide/methanol (30%) was used to stop endogenous peroxidase activity (45 min at 25°C). One percent BSA was used to block tissue sections and then incubated with an antibody specific for rabbit anti-SLURP1 (1:200, Abcam) at 4°C for 12 h. After repetitive washing by PBS, the sections were incubated with peroxidase-conjugated antirabbit antibodies, followed by visualization using 3,3-diaminobenzidine with hematoxylin counterstain. Immunostaining was quantified by ImageJ software.

For the immunofluorescence staining, the deparaffinization and antigen retrieval of sections were done as described above. Three percent normal goat serum and 0.1% Triton X-100 were used to block sections. After washing, sections were incubated with antibody specific for SLURP1 (1:200, Abcam) and a7 nAChR (1:200, Abcam) at 4°C for 16 h. After washing and incubating with appropriate secondary antibodies and DAPI, the sections were observed using fluorescence microscopy. NIH image analysis software (ImageJ) was employed to quantify the results of immunofluorescence staining.

HBMEC cells cultured in 24-well tissue were treated with BSA (2 μg/ml), SLURP1 (2 μg/ml), E. coli K1 (1 × 10⁵ CFU) or E. coli K1 (1 × 10⁵ CFU) + anti-SLURP1 antibody (1 μg) at 37°C, 5% CO₂ for 2 h. The SLURP1 antibody was added simultaneously with E. coli K1. After general washing, 4% paraformaldehyde were used to fix cells for 10 min and 5% BSA was employed to block cells for 30 min. Alexa Fluor 488-conjugated α-bungarotoxin was then added at a concentration of 1 μg/ml and kept for 6 h at room temperature. Followed by washing, the sections were observed using fluorescence microscopy. The relative fluorescence intensity was determined using the ImageJ software (NIH). Briefly, the Spot Enhancing Filter 2D plugin was used to amplify signals from the cells, and then threshold settings were used to specifically select the fluorescent regions. The selected regions were overlaid on the original images and analyzed for mean fluorescence intensity of the area.

The culture supernatants of HBMEC infected with or without E. coli K1 were concentrated using ultrafiltration for immunoblot analysis. Cell lysates were prepared in RIPA buffer. SDS-polyacrylamide gel was used to separated protein (20–30 μg), followed by transferring onto polyvinylidene difluoride membranes (Millipore). Five percent of skim milk was used to block membranes for 1 h. Membranes were then incubated with rabbit-anti-SLURP1 antibody (1:1,000, Abcam) at 4°C overnight. β-Actin (1:20,000) or Coomassie staining of total proteins was employed as an internal control. SLURP1 expression was detected using goat antirabbit IgG antibody conjugated with horseradish peroxidase and enhanced chemiluminescence reagent kit.

The proinflammatory cytokines TNF-α (ab208348, Abcam), MMP-9 (ab253227, Abcam), and ICAM-1 (ab100688, Abcam) from homogenized brain extracts were evaluated using ELISA kits according to the manufacturer’s instructions.

Data are shown in mean ± standard error. All the analyses in this study were done by SPSS (v25.0). Group differences between two groups were analyzed using the Student’s t-test. Group differences between three or more groups were analyzed using the one-way ANOVA followed by Bonferroni post-hoc test. Survival rates comparations were analyzed with log-rank test. Two-side p-value less than 0.05 was considered significant and is represented as *p < 0.05, **p < 0.01, and **p < 0.001.

Secretion of SLURP1 in the culture supernatants of E. coli K1-infected HBMEC were analyzed using immunoblot assay and ELISA. We found E. coli K1 infection enhanced the SLURP1 secretion in both time- and dose-dependent manner compared with uninfected HBMECs ( Figures 1A, B ). To confirm these results in vivo, we assessed the SLURP1 expression in brain sections of neonatal C57BL/6 mice infected with or without E. coli K1 by immunohistochemical detection. As shown in Figures 1C, D , SLURP1 protein expression from hippocampus areas was significantly increased in mice infected with E. coli K1 compared with that of control. Notably, we also found a lot of SLURP1 was specially gathered around the blood vessels in the cortex sections of mice infected with E. coli K1 ( Figures 1E, F ). ELISA showed that mice infected with E. coli K1 showed a higher concentration of SLURP1 both in the serum and CSF than that of control ( Figures 1G, H ). Pearson correlation analysis indicated that the concentrations of SLURP1 were positively correlated with E. coli K1 counts in the CSF ( Figure 1I , r = 0.7635, p = 0.0167).

To identify if E. coli infection can upregulate SLURP1 transcription in humans, we analyzed one public human data set encompassed sepsis caused by E. coli or Staphylococcus aureus (S. aureus), and found that SLURP1 transcription level was higher in E. coli-infected patients, as compared with healthy controls or S. aureus-infected patients ( Figure 1J ). Furthermore, SLURP1 transcription levels get higher at 8 h postinfection than 4 h postinfection of E. coli ( Figure 1K ). Together, these results revealed that E. coli infection could induce SLURP1 secretion.

The pathogenesis of E. coli K1 meningitis required two key events: invasion of the brain microvascular endothelial cells by the bacteria and PMN transmigration across the BBB (8), we therefore next employed immortalized HBMEC monolayers to determine whether exogenous supplement of SLURP1 could promote E. coli K1 penetrating the endothelial cells, as well as support PMN transmigration across the BBB in vitro. SLURP1 was obtained recombinantly with a His-tag as described in the Material and Methods section. The amino acid sequence is shown in Supplementary Figure S1A . The results of double-digestion analysis, protein purity, and immunoblot detection, DNA sequencing are shown in Supplementary Figures S1B, C and Supplementary Material 1 , respectively. As shown in Figures 2A, B , supplement with SLURP1 promoted E. coli K1 penetrating the endothelial cells, accompanied by enhancing E. coli K1-induced PMN transmigration across the HBMEC monolayers in a dose- and time-dependent manner. To explore the possibility that SLURP1 may promote E. coli K1 growth, we compared its growth on brain heart infusion broth in presence or absence of SLURP1. The result showed that SLURP1 has no obvious influence on E. coli K1 growth ( Figure 2C ), suggesting that the promotive effects of SLURP1 on E. coli K1 infection is not through promoting the growth of the pathogen. In order to further confirm the promotive role of SLURP1, we generated SLURP1 overexpression or knockdown HBMEC. Figures 2D, G showed the effects of SLUPR1 overexpression or knockdown, respectively. We found overexpression/knockdown of SLUPR1 significantly increased/decreased E. coli K1 penetrating the endothelial cells and PMN transmigration across the HBMEC monolayers, respectively ( Figures 2E, F, H, I ). Taken together, these results indicate that SLURP1 promotes E. coli K1 invasion of the endothelial cells, as well as enhances PMN transmigration across the BBB.

In order to confirm the biological significance of the in vitro findings described above, we further tested the effect of SLURP1 on E. coli K1 meningitis pathogenesis in the murine model. Neonatal mice were intraperitoneally administered with BSA or SLURP1 two consecutive days prior to E. coli K1 challenge. The survival rate, pathogen counts in CSF, PMN transmigration, and brain damage were detected as described in Materials and Methods . The results showed that only 20% of the SLURP1-pretreated mice infected with E. coli K1 survived within 60 h postinfection ( Figure 3A ), while the survival rate for E. coli K1-infected mice without SLURP1 supplementation reached almost 50%. Furthermore, we found administration with SLURP1 was able to markedly increase pathogen and PMN counts in the CSF ( Figures 3B, C ). Evans blue assay showed that SLURP1-pretreated mice have more severe BBB damage than the control group ( Figure 3D ). Notably, we found SLURP1 has no influence on the BBB integrity of uninfected mice. H&E staining of brain sections indicated that supplement with SLURP1 dramatically promotes neutrophil infiltration into the meninges and meningeal inflammation ( Figures 3E, F ). Additionally, we found exogenous supplement of SLURP1 could robustly enhance the levels of proinflammatory cytokines in brain homogenates ( Figures 3G–I ). Above all, these results suggested that SLURP1 promotes the pathogenesis process of E. coli K1 meningitis.

We further confirmed if E. coli K1-induced SLURP1 activates α7 nAChR. Firstly, a fluorescent α-bungarotoxin (α-bgtx) binding assay was performed to detect the activity of α7 nAChR. HBMEC were cultured in 24-well plates, followed by incubating with SLURP1 or E. coli K1 for 2 h. Afterward, fluorescently labeled α-bgtx incubation was carried out to test α7 nAChR activity. As shown in Figures 4A, B , SLURP1 or E. coli K1-treated HBMEC showed brighter green fluorescence than control, indicating an increase in α7 nAChR activity. Notably, when added with SLURP1 antibody, the promotive effect of E. coli K1 on α7 nAChR activity was blocked, indicating the E. coli K1-induced SLURP1 was responsible for the α7 nAChR activation. To further confirm whether SLURP1 is directly linked to α7 nAChR activation, we analyzed the colocalization of fluorescently labeled SLURP1 with α7 nAChRs by immunofluorescence staining. The cortex sections of E. coli K1-treated mice showed colocalization of SLURP1 (green) and α7 nAChRs (red) ( Figure 4C ). What is more, with increasing doses of E. coli K1 challenge, the Pearson’s coefficient and overlap coefficient of colocalization also increased ( Figure 4D ). These results suggest that E. coli K1-induced SLURP1 is responsible for α7 nAChR activation.

Finally, we determined whether α7 nAChR is necessary for SLURP1-enhanced E. coli K1 meningitis. We first used the MLA, an α7 nAChR inhibitor, to explore the role of α7 nAChR on the function of SLURP1 in vitro. As shown in Figures 5A, B , MLA inhibited the promotive effects of SLURP1 in a dose-dependent manner, including attenuating E. coli K1 invasion and PMN transmigration. Furthermore, we used α7 nAChR knockout (A7R^−/−) mice to confirm these findings in vitro. Wild-type (A7R^+/+) and A7R^−/− mice were intraperitoneally injected with BSA or SLURP1 for 2 days, followed by challenge with E. coli K1 for 18 h. As shown in Figures 5C – F , SLURP1 treatment has enhanced the pathogen load, PMN transmigration and the BBB damage in the A7R^+/+ mice, while all these promotive effects of SLURP1 were blocked in the A7R^−/− mice. H&E staining of the brain sections showed that neutrophil recruitment and meningeal inflammation of SLURP1-treated A7R^−/− mice was not significantly increased when compared with untreated A7R^−/− mice ( Figures 5G, H ). Above all, these results indicated that SLURP1 acts through α7 nAChR to enhance the pathogenesis process of E. coli K1 meningitis.