All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Florida (UF) and performed at UF Animal Care Facilities (IACUC Protocol #202200000637). Colonies of germ-free (GF) mice were bred and maintained in isolators by the UF Animal Care Services Germfree Division. Mixed gender, 6-10 week old GF Il10^–/– 129/SvEv mice were transferred from breeding isolators and placed into the Techniplast ISOcage P Bioexclusion system to allow for microbial manipulation.^14,15 Cages were supplied with autoclaved food and water. Mice were randomly assigned to either GF group, thirteen-member consortium (C13) group or thirteen-member consortium with C. jejuni 81-176 (C13 + C. jejuni) group. Freshly passed stools were collected on day 6 using autoclaved forceps into sterile Eppendorf tubes and flash-frozen in liquid nitrogen and stored at –80 °C. Colons were harvested and flushed with 1X PBS. One half of the proximal and distal colon snips were collected and flash-frozen in liquid nitrogen and stored at –80 °C. The other half of the colon was swiss-rolled and fixed in 10% formalin overnight at room temperature (RT) and the next day transferred to 70% ethanol for processing and paraffin embedding.

The sections were stained with hematoxylin and eosin and the slides scanned using an Olympus VS200 microscope (UF molecular pathology core). Two investigators blinded to the treatment allocations scored inflammation based on the histology of the tissue. Briefly, intestinal inflammation was scored on a scale of 0-4 by the degree of lamina propria immune cell infiltration, goblet cell depletion, crypt hyperplasia, crypt abscess, and architecture distortion and were calculated as average of proximal, mid and distal colon regions as described previously.¹⁰

Members of the consortium were individually purchased from DSMZ (German Collection of Microorganisms and Cell Cultures GmbH; Braunschweig, Germany) representing four major bacterial phyla in the mouse gut. Bacteroides acidifaciens (B. acidifaciens) (DSM 100502), Bacteroides caecimuris (B. caecimuris) (DSM 26085), Blautia caecimuris (B. caecimuris) (DSM 29492), Thomasclavelia ramosa (T. ramosa) (DSM 29357), Parabacteroides goldsteinii (P. goldsteinii) (DSM 29187), Akkermansia muciniphila (A. muciniphila) (DSM 26127) and Paraclostridium bifermentans (P. bifermentans) (DSM 29423). The bacterial strains Parabacteroides distasonis (P. distasonis) and Bacteroides uniformis (B. uniformis) were isolated by the Jobin’s laboratory from the stools of healthy mice. All the bacterial strains described above were grown from a glycerol stock in MEGA media¹⁶ for 3 days under anaerobic conditions. Escherichia coli (E. coli) (DSM 28618) was grown aerobically in Luria Bertani broth. Staphylococcus xylosus (S. xylosus) (DSM 28566) was grown in tryptic soy broth aerobically. Enterococcus hirae (E. hirae) (DSM 28619) was grown in tryptic soy broth and Ligilactobacillus murinus (L. murinus) (DSM 28683) was grown in MRS medium microaerophically created by BD GasPak Gas generating systems. Campylobacter jejuni (C. jejuni) 81-176 was grown in Campylobacter Selective Medium (Fisher Cat #R01274) for 48 hours at 37 °C using the BD GasPak EZ Campy pouch generating systems (Fisher Scientific Cat #B260685). CFU estimation was done using nanodrop for culture OD. Each isolate was pooled using equal CFU amounts to reach a total 1 × 10⁶ CFU/mouse dosage in 200 µL.

GF mice were maintained under sterile conditions using the ISOcage P Bioexclusion system. UF ACS staff transferred GF mice in a transfer disk to the biosafety cabinet. The mice were then carefully transferred using sterilized forceps to the pre-sterilized cages filled with autoclaved water and food. Cages and other materials were sterilized with Exspor (chlorine-dioxide sterilant activator and base) as described previously.¹⁷ For the stool collection, beakers, pre-labeled tubes, and forceps were autoclaved one day before collection. Post autoclave, the bag was immediately sealed with clear tape. On the day of stool collection, the same procedure was followed for Exspor sterilization for the cages, biosafety cabinet, and the autoclaved materials for stool collection. Freshly passed stools were collected in the tubes and were flash-frozen in liquid nitrogen and stored at -80 °C until used for analysis.

DNA extraction was performed as described previously.¹⁸ One-half stool pellet per mouse was used for fecal DNA extraction, and samples were aliquoted in a 96-well PowerBeard Pro Plate (QIAGEN). Samples were lysed using Tissue Lyser II (QIAGEN) and then processed using QiAcube HT as described.¹⁹ Following total fecal DNA extraction, the 16S rRNA V1-V3 hypervariable region was amplified using barcoded primer pairs 27F (5’-AGAGTTTGATCCTGGCTCAG-3’) and 534 R (5’-ATTACCGCGGCTGCTGG-3’) with universal Illumina paired-end adapter sequences. PCR products were purified, quantified, and pooled as described previously and sequenced in a single run of Illumina MiSeq.²⁰ Demultiplexed reads were fed to the DADA2²¹ pipeline for primer sequence removal, quality filtering, correction of Illumina amplicon sequencing errors and dereplication followed by amplicon sequence variants (ASVs) generation and chimera removal. Taxonomic classification was then performed using DADA2 assignTaxonom and addSpecies functions utilizing an in-house database composed of SILVA reference dataset (v. 138.1), GTDB (v. bac120_arc122_ssu_r89) and RefSeq-RDP16S (v. 3). We then removed any ASV that was classified as non-bacterial and all singleton ASVs. This resulted in a total of 1,042,049 reads, with an average of 69,470 reads per sample (min = 58,880; max = 81,295).

Differential taxa abundance analysis was performed using MaAsLin2²² R package with group set as fix effect and cage set as a random effect. For barplot representation of the proportional relative abundance of taxa, we agglomerated the ASVs at the species level using phyloseq²³ package and then used R ggplot package for plotting.

Total RNA was isolated from frozen mouse stool day 6 samples as described.²⁴ Briefly, lysis and disruption steps were performed with 1:1 mix of 1mm acid-washed glass beads and 0.1 mm zircona beads using a bead beater for efficient lysis and disruption; Precellys24 (Bertin Instruments Cat #EQ03119-200-RD000.0). RNA purification was further performed according to the manufacturer’s protocol using mirVana miRNA isolation kit, with phenol (Fisher Scientific AM1560). RNA eluted was treated with Turbo-DNA-free kit to remove any DNA impurities (Fisher Cat #AM1907). Colonic tissue RNA was extracted using TRIzol and a bead beater as previously described.²⁵ RNA concentration was quantified using a NanoDrop 2000.

Quality control, rRNA depletion and cDNA library preparation were performed by the University of Florida’s Interdisciplinary Center for Biotechnology Research (ICBR) Gene Expression Core (https://biotech.ufl.edu/gene-expression-genotyping/ RRID:SCR_019145). Briefly, RNA was quantified using the QUBIT fluorescent method (Fisher Biotech), and sample quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). Ribosomal RNAs were removed from total RNA using Illumina Ribo-Zero Plus rRNA Depletion Kit following the manufacturer’s protocol and eluted into 5 µl EB buffer. Then, the RNA-seq library was processed using NEBNext® Ultra™ Directional RNA Library Prep Kit for Illumina (NEB, U.S.A) following the manufacturer’s recommendations. The libraries were pooled at equal molar concentrations and sequenced by NovaSeq X 10B flow cell with 2 × 150 cycles run. Sequencing was performed at the ICBR NextGen Sequencing (https://biotech.ufl.edu/next-gen-dna/, RRID:SCR_019152).

For RNA-seq analysis, reads were quality trimmed at Q20 (sequencing quality score of 20) and filtered to remove sequencing adapters using Trimmomatic²⁶ (v.0.39). Alignment and quantification were done by STAR²⁷ (v.2.7.11b) using mouse reference genome (GRCm39) with its corresponding annotation file obtained from the National Center for Biotechnology Information (NCBI). An average of 14,367,266 reads per sample (min = 645,786; max = 40,516,378) were uniquely mapped to GRCm39. Gene counts were then imported into edgeR²⁸ for normalization and differential expression analysis. We considered a gene differentially expressed if its edgeR P_adj -value is less than 0.05. Principal Component Analysis (PCA) was done using R prcomp function using the output of R DESeq2 rlog (regularized log transformation) function, which produces a matrix of log transformed and normalized counts, with the option blind set to true. We tested for differences in samples clustering using gls (Generalized Least Squares model) function in R package nlme.

The pathway analysis was generated using QIAGEN IPA (QIAGEN Inc., https://digitalinsights.qiagen.com/IPA)²⁹ software from UF-Health Cancer Center ICBR Bioinformatics core (RRID:SCR_019120).

For Bacterial RNA-seq analysis, trimmed and adapter free RNA-seq reads from above were fed into Kraken2³⁰ and classified using the nt database v.k2_nt_20231129 (a large collection of GenBank, RefSeq, TPA and PDB sequences) obtained from https://benlangmead.github.io/aws-indexes/k2. We then collected reads that were classified as bacteria (average of 28,230,937 bacterial reads per sample; min = 7,762,062, max = 37,199,160) and used them for subsequent bacterial RNA-seq analysis, which guarantee that reads from host and other organisms do not introduce any contamination into our analysis. Those reads were then aligned to a reference composed of the genomes of our bacterial consortium (obtained from ftp://ftp.ncbi.nlm.nih.gov/genomes) using bowtie2³¹ aligner (average of 91% of input reads successfully aligned; min = 81%; max = 96%). Gene expression was quantified using featureCounts³² from subread package. Counts were imported to edgeR and analyzed as described under Mouse RNA-seq analysis above.

KEGG Orthology (KOs) abundances were generated using fmh-funprofiler³³ from the same reads used for bowtie2 alignment above, then pathway analysis was conducted through GAGE³⁴ using KEGG pathways. We considered a pathway significant if its GAGE P_adj-value is less than 0.05.

Neutrophils in colon tissues were detected using immunohistochemistry for the marker Ly-6G/Ly-6C (Gr-1; Clone RB6-8C5) as described previously.³⁵ Colonic tissues were deparaffinized, blocked, and incubated overnight with Ly-6G/Ly-6C (1:100). The next day, tissues were incubated with anti-rabbit biotinylated (NBP1-75414 -1:500), ABC (Vectastain ABC Elite kit, Vector Laboratories), DAB (Dako), and hematoxylin. The images were captured using an Olympus VS200 slide scanner (Evident Scientific) at 40X magnification to assess brown staining. The signals were quantified on ImageJ, where five fields of vision were counted.

FISH assay was performed as previously described.¹¹ Briefly, tissue sections from all groups were deparaffinized, hybridized, washed, and mounted in DAPI medium and imaged using a Leica DM6000B upright microscope. Cy3- tagged 5'AGCTAACCACACCTTATACCG3' targetting C. jejuni 23S rRNA³⁶ was used to detect C. jejuni localization in the colons.

The RNAscope 2.5HD Duplex Detection kit (Cat no.322430), RNAscope probe for Enterobacteriaceae-16S mRNA (Cat no. 509831), and mouse tumor necrosis factor (Tnfa) mRNA (Cat no.311081-C2) were purchased from Advanced Cell Diagnostics (ACD-Biotechne). RNA scope was performed according to the manufacturer's instructions. The slides were captured using an Olympus VS200 slide scanner (Evident Scientific) at 40X magnification to assess positive cells. The signals were quantified on ImageJ, where five fields of vision were counted. Qualitatively, fewer positive signals were observed in low regions compared to high regions of the mid-colon, where a denser distribution was evident, providing the basis for their classification as ‘low’ and ‘high’ regions. We also utilized slide scanning service (Olympus VS200) operated by UF Molecular Pathology Core (RRID:SCR_016601).

Formalin-fixed, paraffin-embedded (FFPE) tissue blocks were used. Prior to sectioning, RNA quality was assessed by extracting RNA from test sections and calculating the DV200 value (>30% required for optimal spatial transcriptomics). FFPE blocks were faced at 10 μm to expose the tissue, then sectioned at 5 μm thickness using a microtome. Sections were dried at 42 °C for 3 hours and stored overnight in a desiccator at room temperature. Spatial transcriptomics was performed by UF ICBR (RRID:SCR_019145), according to the manufacturer's protocol. Spatially barcoded cDNA was extracted and used for library construction following the Visium for FFPE library prep protocol. Libraries were quantified using a Bioanalyzer (Agilent Technologies) and sequenced on an Illumina NovaSeq 6000 platform using 150 bp paired-end reads, targeting 50,000 read pairs per capture spot resulting in more than 48 million paired-end reads per sample.

FASTQ files were processed using Space Ranger (v3.1.1) using mouse reference genome (mm10) resulting in more than 99% of reads mapped to probe set, more than 97% of reads mapped confidently to probe set, more than 97% of reads mapped confidently to the filtered probe set, more than 92% of barcodes are valid, more than 99% of UMIs are valid, more than 96% of reads in squares under tissue and less than 1.2% of UMIs are estimated from Genomic DNA. The resulting Space Ranger outputs were imported to Seurat R package (v5.2.1) for analysis following Seurat spatial transcriptomics analysis workflow: https://satijalab.org/seurat/articles/spatial_vignette.html. Briefly, we included spots with a percentage of mitochondrial genes less than 20% and have counts greater than 100. Data were normalized and variance stabilized using Seurat’s SCTransform. For analysis involving different slices, Harmony (v.1.2.3)³⁷ was used for batch effect correction, followed by PrepSCTFindMarkers. For region of interest analysis, the selected areas contained roughly the same number of bins. We used single-cell RNA sequencing data from Mayassi et al.)³⁸ as a reference for cell-type deconvolution in our spatial transcriptomics data using spacexr v.2.2.1.³⁹

500 ng of RNA was reverse transcribed using the iScript cDNA Synthesis Kit (Bio-Rad 170-8891). qPCR was performed using a Bio-RAD CFX384 Real-Time PCR system using primers described in Table 1 (E. coli virulent transcripts). Amplification was performed using the following conditions: initial denaturation at 95 °C for 3 minutes, followed by 39 cycles of 95 °C for 10 seconds, 56 °C for 10 seconds, and 72 °C for 20 seconds. The melting curve was performed by increasing the temperature from 65 °C to 95 °C in 0.5 °C increments, with a 5-second hold at each step. The universal bacterial 16S ribosomal gene was used as a control. Fold change was calculated using the 2^–ΔΔCt method⁴⁰ using SYBR green dye (Applied Biosystems).

500 ng of RNA was reverse transcribed and qPCR was performed using a Bio-RAD CFX384 Real-Time PCR system using primers described in Table 2 (host gene transcripts). Amplification was performed using the following conditions: initial denaturation at 95 °C for 3 minutes, followed by 39 cycles of 95 °C for 10 seconds, 56 °C for 10 seconds, and 72 °C for 10 seconds. The melting curve was performed by increasing the temperature from 65 °C to 95 °C in 0.5 °C increments, with a 5-second hold at each step. Gapdh was used as a housekeeping control. Fold change was calculated using the 2^–ΔΔCt method (81) using SYBR green dye (Applied Biosystems).

We used R's two-sample Wilcoxon test (also known as Mann-Whitney test) and for tests involving more than two comparisons, we corrected the p-values for multiple testing using R’s p.adjust function with method set to BH (Benjamini and Hochberg).⁴¹

To establish a tractable system to study host response to C. jejuni infection, we assembled a consortium of 13 bacterial species representing 4 major phyla present in the intestine. Since susceptibility to C. jejuni infection is influenced by the microbiota,³ we first examined the impact of the interaction between C13 and C. jejuni on host response using GF Il10^–/– mice, a well-established model of campylobacteriosis.^12,42 GF Il10^–/– mice were divided into three groups: GF, C13 and C13 containing C. jejuni. Each group was colonized with their respective bacterial community or no gavage (single oral gavage, 10⁶ CFU/mouse) and housed in an ISOcage P Bioexclusion cage system. Six days post-oral gavage, stool, and distal colon samples were collected from all the mice (Figure 1A). To determine engraftment of the consortium and impact of C. jejuni on C13, we profiled bacterial composition using 16S rRNA gene sequencing. We observed effective colonization of each member of the consortium in both C13 and C13 + C. jejuni-colonized Il10^–/– germ-free mice (Figure 1B). We observed significant expansion of E. coli and P. bifermentans and contraction of P. distasonis relative abundance in the C13 + C. jejuni group compared to C13 alone. Consistent with our findings, previous 16S rRNA sequencing studies have reported an increased abundance of Enterobacteriaceae following C. jejuni infection of antibiotic-treated mice,⁴³ supporting the role of C. jejuni in driving microbial community shifts that may contribute to inflammation.

C. jejuni infection triggers rapid colitis in GF Il10^–/– mice but remains healthy when mice are colonized with microbiota.³ Therefore, we next compared the severity of inflammation between C13 and C13 + C. jejuni colonized mice. While Il10^–/– mice colonized with C13 displayed minimal inflammatory response, the addition of C. jejuni to the consortium provoked significantly increased inflammation as quantified by histological score with increased immune cell infiltration, loss of goblet cells and crypt abscesses (Figure 2A, B). Similarly, a significant upregulation of inflammatory genes such as Ifng, Il1b, Il6, Tnfa, and Il17a mRNA was observed in C13 + C. jejuni-infected mice compared to C13 mice (Figure 2C–G). Thus, a defined-bacterial consortium allows the study of campylobacteriosis in Il10^–/– mice.

To gain a better understanding of the changes in host physiology caused by C. jejuni infection, we performed bulk RNA-seq from stool specimens. This approach allowed us to study both gene expressions from exfoliated gut luminal epithelial cells as well as associated microbial gene transcripts. PCA showed that C. jejuni infection induced marked changes in the host transcriptional response (Figure 3A). We observed 4,085 significant differentially expressed host genes, with 2,221 upregulated and 1,864 downregulated in C13 + C. jejuni colonized mice (Figure 3B, Additional file 1: https://figshare.com/s/b84ac4a3c99882822195). We noted increased expression of Tgm3 in C13 colonized mice, a gene implicated in mucosal barrier function, preventing pathogenic infections,⁴⁴ and Gsdmc2 and Gsmdc3 genes which play a role in Lgr5⁺ stem cell maintenance and regeneration by regulating mitochondrial homeostasis.⁴⁵ In addition, we observed that the Slc7a11 gene, which plays a role in extracellular cysteine uptake and is essential for maintaining redox balance,⁴⁶ is upregulated in C13 + C. jejuni mice compared to C13 colonized mice. We also noted increased chemokine Cxcl5, a chemoattractant for neutrophil infiltration⁴⁷ and Duoxa2 expression, a gene necessary for activation of the dual oxidases DUOX2 implicated in anti-microbial host response.⁴⁸ Using Qiagen Ingenuity pathway analysis (IPA), we found that different immune response pathways, including neutrophil degranulation pathway, pathogen-induced cytokine storm, leukocyte extravasation, and NOD1/2 pathway were significantly upregulated upon C. jejuni infection (Figure 3C).

We next focused on the impact of C. jejuni infection on bacterial gene expression profiles. PCA showed distinct changes in the bacterial transcriptional profile following the addition of C. jejuni to the C13 (Figure 4A). Volcano plot showed a plethora of differentially expressed genes between C13 and C13 + C. jejuni groups (Figure 4B, Additional file 2: https://figshare.com/s/b84ac4a3c99882822195). As compared to the C13 group, we observed 8,754 downregulated and 7,327 upregulated bacterial genes following inclusion of C. jejuni. We observed a decrease in bacterial expression genes such as Blautia carbohydrate ABC transporter permease, Ligilactobacillus alpha glucoside-specific transporter, and Bacteroides RagB/SusD outer membrane protein that play a role in nutrient uptake and in carbohydrate metabolism that aids in bacterial growth and survival.^49,50 In contrast, C. jejuni colonized mice show expression of various bacterial genes, including C. jejuni PorA implicated in adaptation to change in the intestinal environment,⁵¹ C. jejuni and P. bifermentans peroxiredoxin (Prxs) and C. jejuni Dps family protein that confers protection from oxidative damage.^52,53 Other notable altered genes included aspartate-ammonia lyase, which promotes E. coli growth under anaerobic conditions in the gut.⁵⁴ These findings suggest that the presence of C. jejuni modified the bacterial transcriptomic profile in the gut to promote pathogen survival.

Using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, we observed that various bacterial metabolic pathways, such as the biosynthesis of amino acids, biosynthesis of co-factors, ABC transporters, purine, and pyruvate metabolism, were altered in C13 + C. jejuni mice compared to C13 (Figure 4C). We observed that among significantly altered bacterial genes, 74.51% belonged to metabolic pathways, 17.65% to virulence pathways, and 7.84% to cell cycle and repair pathways in the C13 + C. jejuni colonized mice.

To validate these findings, we identified genes corresponding to virulent pathways present in E. coli such as biofilm, stress response, defense, motility, and quorum sensing. We selected a candidate gene for each pathway and normalized it to the universal bacterial 16S ribosomal gene using RT-qPCR. We observed a significant increase in E. coli AraC (biofilm), FliA (motility), RpoE (stress response) and Qseb (quorum sensing) transcripts in C13 + C. jejuni as compared to C13 (Figure 4D−G). Although we observed an increase in the E. coli BasR gene (defense), it was not statistically significant (P_adj-value = 0.057, Figure 4H). These data suggest that C. jejuni infection alters host inflammatory pathways and upregulates bacterial virulence pathways across the consortium.

We previously observed significant expansion of E. coli in the stool samples of C13 + C. jejuni-infected mice (Figure 1B). Thus, we next investigated the C. jejuni and E. coli distribution in the colon tissue. Fluorescence-in-situ hybridization (FISH) assay revealed C. jejuni infiltration in the mid colons of C13 + C. jejuni-infected mice compared to C13 mice (Figure 5A). Interestingly, RNAscope analysis showed an increased abundance of Tnfα and Enterobacteriaceae mRNA transcripts in the C13 + C. jejuni group compared to C13 (Figure 5B). Infiltration of E. coli following C. jejuni infection was also localized in the mid colon while Tnfα was dispersed throughout the colon (Figure 5B). We previously demonstrated that neutrophils play a key role in mediating campylobacteriosis in Il10^–/– mice,^11,35 therefore we next sought to determine neutrophil infiltration in the colon of C13 + C. jejuni mice using immunohistochemistry.^11,35 We observed significant neutrophil infiltration (Ly-6G⁺ marker) in the colons of C13 + C. jejuni compared to C13 (Figure 5C). These findings demonstrate that the C13 is a tractable bacterial consortium enabling the study of host response to C. jejuni infection.

The region-specific tissue colonization of C. jejuni and Enterobacteriaceae in the C13 + C. jejuni-infected mice prompted us to characterize the transcriptional reprogramming of these tissues. Clustering of our spatial transcriptomics revealed 19 (Clusters 0−18) clusters (Figure 6A) corresponding to distinct transcriptional signatures in the colonic tissue. We observed a similar clustering pattern between GF and C13-colonized mice, whereas the presence of C. jejuni affected several clusters (Figure 6A–C, Additional file 3: https://figshare.com/s/b84ac4a3c99882822195). For example, in the C13 + C. jejuni group, Clusters 2, 3, and 17 exhibited distinct gene expression patterns and spatial distributions throughout the colon, as visualized by spatial mapping (Figure 6C, lower panel). Differential gene expression analysis identified cluster-specific marker genes for Clusters 2, 3, and 17, which were further characterized using IPA to assess enriched functional pathways.

Cluster 2 showed significant upregulation of inflammatory genes associated with the S100 family signaling pathway, Th2 pathway, neutrophil degranulation, Stat-3, and Interleukin signaling. Conversely, pathways involved in epithelial regulation and tumor suppression were downregulated, suggesting a localized pro-inflammatory environment (Supplemental Figure S1A, Additional file 4: https://figshare.com/s/b84ac4a3c99882822195). Cluster 3 shared substantial overlaps with Cluster 2; to focus on distinct features, we selected pathways that were uniquely enriched in Cluster 3. These included pathways like Th1, PI3K/AKT, and NOD1/2 signaling pathways, suggesting coordinated activation of innate and adaptive immune responses in this cluster. Additionally, downregulation of lipid metabolism, LXR/RXR activation, and amino acid transport pathways pointed toward a metabolic shift (Supplemental Figure S1B).

Cluster 17 demonstrated a distinct immunological and metabolic signature. Upregulated pathways included neutrophil extracellular trap formation, PTEN signaling, HMGB1, and glycolysis and gluconeogenesis, consistent with a stress-responsive environment. In contrast, downregulated pathways such as extracellular matrix organization, Rho family GTPase signaling, senescence, and phagosome formation implying reduced inflammation within this cluster (Supplemental Fig. S1C). Interestingly, Cluster 17 exhibited downregulation of pathways associated with pathogen-induced cytokine storm, contrasting with the upregulation observed in other clusters. Cluster 17 is predominantly localized to the distal colon, while C. jejuni abundance is higher in the mid colon, corresponding to clusters such as 2 and 3, which display increased inflammatory pathway activity. This spatial distribution highlights region-specific differences in gene expression during infection.

These results illustrate how host gene expressions are organized spatially in response to GF, C13, and C13 + C. jejuni infection. Overall, these data suggest that C. jejuni infection in the context of a defined consortia induces both widespread and localized transcriptional responses across the colonic tissue. Furthermore, transcriptomic analysis of the entire colon revealed additional insight into overall gene expression changes across the different experimental groups. A total of 2,773 genes were significantly differentially expressed, of which 1,788 were downregulated and 985 were upregulated in C13 compared to the GF group (Figure 6D). We compared our GF and C13 transcriptomic datasets to recently published spatial transcriptomic data from SPF mice.³⁸ Interestingly, our C13 transcriptomic data overlaps with the published SPF mice dataset of Mayassi et al.,³⁸ as evidenced by increased gene expression of microbe-driven host genes Reg3g, Ang4, Itln1, and Wfdc18 in our C13 dataset, which play a role in host defense and mucosal homeostasis, compared to GF mice. The presence of C. jejuni in C13 drastically modified the transcriptomic landscape with upregulation of genes involved in antimicrobial defense (Saa3, Reg3β, Reg3γ), immune cell recruitment (Ifnγ, S100a8, S100a9), inflammation and barrier maintenance (Prss22, Chil3), tissue repair (Il22, Mmp8), neutrophil function (Retnlg), and immune metabolism (Acod1) (Figure 6E-F). These overlapping responses suggest that C13 + C. jejuni maintains a distinct gene expression profile relative to both GF and C13 mice.

Our data showed tissue-specific enrichment of C. jejuni and Enterobacteriaceae in C13 + C. jejuni mice (Figure 5A-B). Therefore, we performed region-specific transcriptomic analysis to determine gene expression profiles for those specific regions. Our region of interest (ROI) was selected based on the presence (+) or absence (–) of C. jejuni (Figure 7A). We observed an increased expression of cytokines (Il17a, Il1a, Il22), antimicrobial protein in Neutrophils (Ngp), epithelial repair and antimicrobial defense (Saa3) and immune modulation (Marco) in C. jejuni (+) region, whereas genes related to barrier function and microbial regulation (Lypd8, S100a14, Muc13) were predominantly expressed in C. jejuni (–) regions. These spatial expression patterns indicate that the mid colon mounts distinct transcriptomic responses depending on local bacterial abundance, with barrier and defense genes enriched in C. jejuni (–) zones (Figure 7B, Additional file 5: https://figshare.com/s/b84ac4a3c99882822195). Strikingly, pathway enrichment analysis using Qiagen IPA revealed a significant upregulation of the atherosclerosis signaling pathway (Figure 7C). This pathway encompasses key genes such as Col3a1 (collagen type III alpha 1), Ifng (interferon gamma), Il1a and Il1b (interleukin 1 alpha and beta), Itgb2 (integrin beta 2), Lyz1/Lyz2 (lysozyme 1 and 2), Pla2g7 (phospholipase A2 group VII), and S100a8. These genes are associated with extracellular matrix remodeling, immune cell recruitment, cytokine-driven inflammation, and dysregulation of lipid metabolism.

RNAscope analysis of the Enterobacteriaceae revealed regional differences in signal intensity across the C13 + C. jejuni colon. Spatial analysis on Enterobacteriaceae high region exhibited elevated expression of genes (Ngp, Ila, Ilb, Ubd, S100a8, S100a9 and Nos2) involved in host defense, pathogen recognition, and cellular stress, compared to Enterobacteriaceae low regions (Figure 8A−C, Additional file 6: https://figshare.com/s/b84ac4a3c99882822195). Common inflammatory pathways between C. jejuni and Enterobacteriaceae ROI were significantly enriched, including the Il-17a and Il-17f-mediated differential regulation of cytokine production in both intestinal epithelial cells and macrophages, and the pathogen-induced cytokine storm signaling pathway, indicating epithelial and systemic immune activation. Furthermore, both the classical and alternative activation signaling pathways of macrophages were upregulated. The alternative activation pathway was characterized by increased expression of genes such as Cebpb, Chil3/Chil4, Dusp1, Il1a, Il1b, and Marco, while the classical activation pathway featured elevated levels of Ccl5, Cxcl9, Ifng, Il17a, Il1a, and Il1b. Unique pathways include activation of the cGAS–STING pathway and cytokine signaling by immune cells in C. jejuni high regions, reflecting intracellular pathogen sensing and immune modulation. In contrast, Enterobacteriaceae high areas exhibited upregulation of antimicrobial peptides and Toll-like receptor (TLR) signaling pathways, indicative of classical mucosal defense mechanisms against extracellular bacteria. These findings, together with the upregulation of signaling pathways, reflect a complex and diverse spectrum of macrophage activation states within the tissue (Figure 7C).

Using spatial transcriptomic data, we performed cell-type deconvolution to examine differences in cellular composition across GF, C13, and C13 + C. jejuni mice (Supplemental Fig. S2 A–C). In both GF and C13 groups, we observed broad expressions of genes associated with enterocytes and goblet cells, indicating preservation of the epithelial compartment. In contrast, a substantial loss of these epithelial and goblet cell signatures was observed throughout the colon of C13 + C. jejuni group. The most striking change was a notable increase in genes associated with myeloid cell populations, reflecting innate immune response as previously reported.³⁵ These findings suggest that C. jejuni colonization drives a major remodeling of the cellular landscape in the colon.

While this study provides valuable insights into host–microbe and pathogen interactions using spatial transcriptomics and gnotobiotic models, several limitations should be acknowledged. One of the limitations is the inability to fully distinguish whether the increased inflammatory transcripts detected in fecal samples result from transcriptional induction within individual cells are primarily due to the increased shedding of epithelial and immune cells during inflammation. Spatial transcriptomic analysis offers a static snapshot of gene expression, limiting our ability to capture dynamic changes in microbial localization and host responses over time. Additionally, the resolution of current spatial transcriptomics technology and the sequencing depth achieved may not fully capture cellular heterogeneity or detect low-abundance transcripts, potentially overlooking rare but important cell populations or subtle transcriptional changes. Although bacterial transcriptional programs were modulated during infection, the potential role of host-derived factors in driving these changes remains speculative and requires further investigation. Notably, whether C. jejuni itself can alter host-derived signals that in turn regulate microbial behavior is an exciting question that warrants further study.