Female C57BL/6 mice were purchased from Jackson Laboratory (stock #000664). Mice were maintained under specific pathogen-free conditions and used at 6-9 weeks of age, following protocols approved by the Institutional Animal Care and Use Committee (protocol # 1902006) at the University of Texas Medical Branch (UTMB) in Galveston, TX. All mouse infection studies were performed in the ABSL3 facility in the Galveston National Laboratory located at UTMB; all tissue processing and analysis procedures were performed in the BSL3 or BSL2 facilities. UTMB operates to comply with the USDA Animal Welfare Act (Public Law 89–544), the Health Research Extension Act of 1985 (Public Law 99–158), the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and the NAS Guide for the Care and Use of Laboratory Animals (ISBN-13). UTMB is a registered Research Facility under the Animal Welfare Act and has a current assurance on file with the Office of Laboratory Animal Welfare, in compliance with NIH Policy. Mice (5/group) were inoculated intravenously (i.v.) with 6 × 10⁴ FFU (200 µl) of Ot Karp strain or PBS (day 0) and monitored daily for weight loss, signs of disease, and survival. Animals were euthanized at day 0, 2, 6 and 10 post infection for tissue collection (28–30). Brain tissues were stored in RNAlater (Qiagen) and incubated in 4°C overnight for inactivation of bacteria. This animal experiment has been conducted independently three times.

Vero cells (African green monkey kidney epithelial cells) were cultured in MEM medium (Gibco) containing 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin. Bacterial stock was prepared, as in our previous reports (29, 30). BV2 cells were cultured with DMEM medium plus 10% FBS; C20 was cultured in DMEM/F12 medium (Gibco) supplemented with 1% FBS and 1×N-2 supplement (Gibco) (31). Cells were infected with bacteria (MOI 10) and harvested at 24 and 72 hours (h). Uninfected cells were used as negative controls.

Tissues were homogenized using metal beads in a BeadBlaster 24 Microtube Homogenizer (Benchmark Scientific) with an RLT lysis buffer (Qiagen). Cultured cells were also harvested and lysed by using RTL lysis buffer. RNA was extracted by using RNeasy Mini kits (Qiagen) and used for cDNA synthesis with an iScript Reverse Transcription kit (Bio-Rad). cDNA was amplified in a 10 μL reaction mixture containing 5 μL of iTaq SYBR Green Supermix (Bio-Rad) and 5 μM each of gene-specific forward and reverse primers. The PCR assays were denatured for 30s at 95°C, followed by 40 cycles of 15s at 95°C, and 60s at 60°C, by utilizing the CFX96 Touch real-time PCR detection system (Bio-Rad). Relative quantitation of mRNA expression was calculated using the 2^−ΔΔCt method. The primers are listed in Supplementary Table 1 .

Brain tissues were fixed in 4% paraformaldehyde (PFA) and embedded in Tissue-Tek OCT, as described in our previous report (24). Briefly, cryosections (6 µm) were fixed for 10 min in ice-cold acetone and rehydrated with PBS. For immunofluorescent staining, sections were permeabilized and blocked using 5% normal goat serum, 0.2% Triton X-100 in PBS for 1 h, followed by primary antibody incubation overnight at 4°C. The anti-IBA1 Ab (ab178846) was purchased from Abcam and Biolegend, respectively. The Alexa Fluor555-conjugated or Alexa Fluor488-conjugated anti-rabbit IgG (H+L), F(ab’)2 Fragment secondary antibodies (Cell Signaling Technology, 1:500 dilution) were used to stain the sections for 1 h at room temperature, followed by three washes with PBS. The TrueVIEW Autofluorescence Quenching Kit (Vector Laboratories) was applied to diminish unwanted autofluorescence and the microscope slides were covered by using Fluoroshield Mounting Medium with DAPI (Abcam). Images were captured using a Carl Zeiss Axio Observer fluorescence microscope or a confocal microscope Zeiss LSM 880 with Airyscan. For each mouse, at least three fixed-frozen sections were included for each experiment. All images were captured with the same conditions of laser power and detector gain.

Brain RNA was extracted by using RNeasy RNA Isolation kit (Qiagen). Samples (3 mice/group) were sent to LC Sciences (Houston, TX) for RNA purity/quantity assessment via Bioanalyzer 2100 and RNA 6000 Nano LabChip Kit (Agilent), mRNA extraction, cDNA libraries construction, and sequencing by using the Illumina Novaseq™ 6000. The Mus musculus genome build mm10 was used as a reference genome. Quality control analysis was done by using FastQC (32) and RSeQC6 (33). Semi-supervised hierarchical clustering analysis and heatmap generation were performed by using the Morpheus tool (https://software.broadinstitute.org/morpheus). Principal component analysis and volcano plot were done in R using PCAtools (34) and EnhancedVolcano (35) packages. The RNA-seq data presented herein were deposited in NCBI’s Gene Expression Omnibus, accessible through GEO Series accession number GSE227525 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE227525).

To determine bacterial burdens, DNA was extracted from brain tissues via a DNeasy Blood & Tissue Kit (Qiagen) and used for qPCR assays, as previously described (28, 36, 37). The primers were OtsuF630 (5’-AACTGATTTTATTCAAACTAATGCTGCT-3’) and OtsuR747 (5’-TATGCCTGAGTAAGATACGTGAATGGAATT-3’) (Integrated DNA Technologies). Bacterial loads were normalized to total nanogram (ng) of DNA per µL for the same sample, and data are expressed as the gene copy number of 47-kDa protein per ng of DNA. The copy number for the 47-kDa gene was determined by known concentrations of a control plasmid containing single-copy insert of the gene. Gene copy numbers were determined via serial dilutions (10-fold) of the Ot Karp 47-kDa plasmid.

Pathway enrichment analysis using gene set enrichment analysis with GSEA version 3.0 was performed with a ranked gene list (38, 39). Specifically, ranked gene list is prepared from the differential gene expression analysis by using the following metrics: sign of the Log₂fold change multiplied by the inverse of the p-value. GSEAPreranked was run with 2000 permutations. Primary gene sets investigated were obtained from the MSigDB and published reports (40, 41). GSEA FDR Q <0.05 cutoff was applied to examine enriched gene sets in our dataset.

Enrichment results from the GSEA analysis were visualized using Cytoscape (42). GSEA output files were uploaded in the EnrichmentMap app of Cytoscape and FDR Q value cut off was set to 0.01. Clusters are automatically defined using AutoAnnotate Cytoscape app.

Data were presented as mean ± standard error of mean (SEM). Differences between individual treatment and control groups were determined by using unpaired Student’s t test, utilizing Welch’s correction when appropriate. One-way ANOVA was used for multiple group comparisons, with a Tukey’s Post Hoc for comparisons between groups. Two-way ANOVA was used for the statistical analysis of body weight change. Log-rank (Mantel-Cox) test was used for survival curve analysis. All data were analyzed by using GraphPad Prism software 8.0. Statistically significant values are denoted as *, p < 0.05; **; p < 0.01, ***; p < 0.001, and ****; p < 0.0001, respectively. All in vitro cell culture experiments were performed three times independently, with triplicate samples obtained each time. All qRT-PCR and qPCR assays (4-5 samples/group) were repeated twice independently.

To understand the brain pathogenesis in severe Ot infection, we intravenously infected mice with lethal dose of bacteria (6 × 10⁴ FFU) and collected brain tissues at days 2 (incubation period), 6 (disease onset), and 10 (disease peak), respectively. As in our previous reports (28, 37), infected mice started body weight loss around day 7, and approximately half of mice expired between days 11 and 14 ( Figures 1A, B ). Brain bacterial burdens were detectable at day 2, reached the peak at day 6, and reduced at day 10 ( Figure 1C ). We used RNA-seq based transcriptomics profiling of brain cortex at the indicated timepoints ( Figure 1D ). Mice with PBS injection were used as mock controls (day 0). We used Principal Component Analysis (PCA) and semi-supervised Hierarchical Clustering Analysis (HCA) to understand the pattern of gene expression changes in the brain during the infection. PCA analysis revealed a clear separation between infected samples compared to uninfected brain tissues ( Figure 1D ). Infected samples from the same timepoints were closer to each other as compared to others, suggesting a consistent difference in the brain transcriptomes at different stages of infection. HCA analysis also followed the same trends, with clear differences in gene expression at different timepoints ( Figure 1E ). Day 2 samples had a total of 1,333 differentially expressed genes, while days 6 and 10 samples contacted a total of 1,478 and 2,469 differentially expressed genes, respectively. Several genes previously reported to be involved in host responses against Ot (24, 43) were differentially expressed in infected brains, including interferon-stimulated genes (ISGs) that are critical components of immune response to microbial infections (Gbp2/4/6, Stat1, Mx1 and Irf7) at day 10 ( Figure 1F ). Consistent with our previous reports (24, 29), IFNg and Tnf were the dominate type-1 immune markers in infected brains. Together, the RNA-seq transcriptomics revealed widespread dysregulation of gene expression with strong type 1 immunity in Ot-infected brain tissues.

To define cellular processes that may cause neuropathology in Ot-infected brains during disease progression, we performed comparative pathway analyses on the transcriptomics data from brains (mocks versus days 2, 6, and 10, respectively) by using the Gene Set Enrichment Analysis (GSEA) software, as well as the Cytoscape Enrichment Map to visualize the biological processes that may play a role in host response against bacterial infection. Consistent with our previous PCR- and NanoString-based studies (4, 24, 29), day 2-vs-mock GSEA analysis of RNA-seq data revealed minor, but statistically significant, upregulation in a few biological pathways, including “complement activation pathway” and “humoral immune response mediated by circulating immunoglobulin” ( Figure 2A , Supplementary Figure 1A ). In contrast, we observed strong enrichment of several pathways related to immune signaling and inflammation for day 6- and day 10-versus-mock comparison. The notable top 10 pathways were related to “Hallmark interferon gamma response,” “defense response to bacteria,” and “Immunoglobulin mediated immunity” ( Figures 2A, B , Supplementary Figure 1B ). To further validate genes differentially induced at days 6 and 10, we analyzed their transcriptomic profiles by using mSigDB Hallmark gene data set ( Supplementary Figure 2 ). Consistently, “Hallmark interferon gamma response,” “Hallmark interferon alpha response,” “IL-6, JAK-STAT signaling,” and “TNF signaling via NF-κB” were among the top 10 pathways induced at days 6 and 10, respectively, as compared to control brains. The lack of an enrichment for these pathways at day 2, suggested to us that immune signaling in response to Ot infection peaked around day 6 and had sustained activation at day 10 (prior to host death).

It is well known that BBB tightly regulates the movement of biological substances between blood and neural cells and is required for maintaining brain homeostasis. BBB disruption is often observed in various pathological conditions, including severe bacterial infection (4, 22, 44). Our previous histological analyses of Ot-infected mouse brains have revealed prominent endothelial cell damage, with increased recruitment of CD3⁺ T cells and CD45⁺ leukocytes to the cortex and cerebellum (24, 30). Other groups have recently reported a core of BBB genes that can be used as the signature biomarkers of BBB disruption in various neurological diseases (40), as this core module includes 136 genes critical for various BBB functions ( Supplementary Table 2 ). To assess if BBB dysfunction contributed to our observed mouse mortality, we considered and examined these BBB signature genes in our analyses of transcriptomics. Strikingly, we found that many upregulated genes (FDR <0.05, log2FC = 1) in our dataset were consistent and overlapped with the core BBB signature. For day 2 samples, this overlap was minimal and only 7 upregulated genes were found ( Figures 3A, B ), which was consistent with our previous report of minimal vascular damage at initial infection (24). As expected, a considerable overlap was found between our dataset and the reported BBB dysfunction module (40), with 31 matched genes at day 6, and 60 genes at day 10, respectively ( Figures 3A, B ). GSEA analysis also showed an enrichment of BBB dysfunction module at days 6 and 10 ( Figure 3B ). We further analyzed several BBB genes at day 10 via qRT-PCR and confirmed a significant increase in Cxcl10, Casp4, Upp1, Ch25h, Selp, Lrg1, Ccl2, and Slfn9 expression as compared to mocks (day 0) ( Figure 3C ). Together, these data indicate a disruption in BBB during severe stages of lethal Ot infection.

Since we observed activation of immune signaling in the brain as the main biological process in response to Ot infection, we next examined key transcription factors that might drive this host signaling response. We performed GSEA analysis by using ENCODE ChIP-Seq dataset. The significant enrichment of transcription factors critical for immune signaling was evident at days 6 and 10 (FDR <0.05), but not detectable at day 2 (FDR >0.05; Figure 4 ). The top enriched 5 transcription factors at day 6 were STAT2, STAT1, FOXM1, TCF12, and CEBPB; at day 10, STAT2, STAT1, TCF12, BATF and TEAD4 were identified as top 5 enriched transcription factors ( Figures 4A, B ). Our qRT-PCR data further confirmed RNA-seq findings, judged by significantly increased transcript levels of Mx1, Mx2, Isg20, Isg15, Irf7, Oas3, Ifit1, and Rsad2 at day 10 as compared to mocks (day 0) ( Figure 4C ). Importantly, STAT1 and STAT2 were both identified as top enriched transcription factors at days 6 and 10 data sets, suggesting a critical role for IFN signaling in mediating neuroinflammation during Ot infection.

Microglia are brain-resident myeloid cells that are considered immune sentinels, playing key roles in brain homoeostasis during physiological and pathological conditions. To determine microglia activation following Ot infection, we stained mouse brain sections with IBA1 (a microglia marker). While the control samples contained resting microglia, characterized by long, ramified processes with comparatively small cell bodies, microglia activation was significant at day 10, as judged by their increased branching and lengthening of processes and enlarged cell bodies ( Figure 5A ). Mathys, et al. have recently used single-cell RNA-seq and identified a disease stage-specific microglia gene signature in a neurodegeneration model with Alzheimer’s disease-like phenotype, defining microglia reactive states as “early response” (a transient intermediate activation of microglia that proliferate) and “late response” (reactive microglia), respectively (41). To assess microglia activation state during Ot infection, we utilized Mathy’s reported analysis strategy. Interestingly, we found a significant enrichment of gene sets for early response (cluster 3/cluster7) and late response (cluster 6) at days 6 and 10, as compared to control brains by using GSEA ( Figures 5B, C ). Such trends were consistent with increased number of IBA1⁺ cells in infected brain tissues ( Figure 5A ). Notably, late-response-gene set had higher NES compared to early-response-gene set, suggesting a dominant response from reactive microglia during Ot infection ( Figure 5 ). Together, our data supports a role of microglia activation in driving neuroinflammation in the brain of Ot-infected mice.

Astrocytes, also known as astroglia, are characteristic star-shaped glial cells in the brain and are often thought of as “nurse” cells for neurons in the brain parenchyma. Reactive astrocytes in CNS injury or disease can be polarized to A1 and A2 types that are neurotoxic and neuroprotective, respectively (45, 46). Since A1 astrocytes are induced by classically activated neuroinflammatory microglia (45), we speculated that severe Ot infection may also promote A1-like responses in the brain. We then analyzed our RNA-seq data for gene profiles associated with astrocyte activation and found that most pan-astrocytic markers were upregulated at day 10, as compared to those of day 0. Importantly, most A1-specific genes were also upregulated, while 70% of A2-specific markers were not altered or slightly decreased at the severe stage of infection ( Supplementary Figure 3 ). Oure data suggest a strong brain A1-like responses in severe Ot infection, and these neurotoxic astrocytes may contribute to Ot-induced neuroinflammation and acute encephalitis.

To support our conclusion, we further investigated whether Ot replicated in microglia and promoted inflammation in vitro. We infected mouse and human microglia cell lines (BV2 and C20, respectively) with bacteria (MOI 10). Immunofluorescence staining clearly indicated established Ot infection in mouse microglia ( Figure 6A ). Furthermore, qRT-PCR analysis of proinflammatory genes indicated significantly increased levels of Cxcl10, Il-1b, and Tnf, especially at 72 h post-infection ( Figure 6B ). Similar results were also found in human microglia C20 ( Figures 6C, D ). The time-dependent increase in bacterial burdens of C20 cells confirmed productive intracellular replication ( Supplementary Figure 4 ). Together, our in vitro data demonstrate Ot-induced bacterial growth and a pro-inflammatory response in microglia, potentially leading to neuroinflammation and subsequent brain damage.