A multi-omic analysis delineates a causal protective role for Bifidobacteriaceae and implicates key host genes in inflammatory bowel disease

Abstract

Background

While gut microbiota dysbiosis is a hallmark of inflammatory bowel disease (IBD), the causal microbial drivers and their host-mediated mechanisms remain elusive. This study leverages an integrated multi-omics approach, combining Mendelian randomization (MR) and transcriptome analysis, to bridge the gap from microbial causality to host molecular pathways.

Methods

We performed a two-sample MR analysis using large-scale genome-wide association study (GWAS) data to identify specific gut microbiota taxa with a causal effect on IBD risk. Subsequently, we conducted a multi-level bioinformatic analysis of IBD patient transcriptomes to elucidate the downstream host genes, regulatory networks, and immune cell interactions modulated by these causal microbes.

Results

Our MR analysis established a robust causal protective effect of the family Bifidobacteriaceae against IBD. Integrating this finding with transcriptomic data, we identified three key host genes as potential mediators acting through distinct mechanisms: LCT, whose regulation may foster a protective prebiotic niche; MCM6, which appears to function as a hub driving the proliferation of pathogenic immune infiltrates; and UBXN4, a critical regulator of cellular proteostasis, the failure of which can precipitate inflammatory stress.

Conclusions

This study moves beyond association to delineate a causal protective role for Bifidobacteriaceae in IBD and pinpoints specific host genes (LCT, MCM6, UBXN4) through which this effect is likely orchestrated. These findings provide a novel mechanistic framework for host-microbiota interactions and highlight new pathways for therapeutic intervention in IBD.

Introduction

Inflammatory bowel disease (IBD), encompassing Crohn’s disease (CD) and ulcerative colitis (UC), represents a growing global health challenge characterized by chronic, relapsing intestinal inflammation (Caron, Honap & Peyrin-Biroulet, 2024). Patients often endure debilitating symptoms such as abdominal pain, persistent diarrhea, and intestinal bleeding, alongside systemic complications like malnutrition and an increased risk of colorectal cancer, all of which severely impair their quality of life (Bruner, White & Proksell, 2023; Diez-Martin et al., 2024). Despite significant advances, the etiology of IBD remains incompletely understood, believed to stem from a complex interplay between host genetic susceptibility, environmental triggers, an aberrant immune response, and dysbiosis of the gut microbiome (Lu et al., 2022; Piovani et al., 2019; Little et al., 2024). This multifactorial nature complicates treatment and underscores the urgent need for a deeper mechanistic understanding of its pathogenesis.

The human gastrointestinal tract harbors a vast microbial ecosystem critical for nutrient metabolism, immune system regulation, and maintaining intestinal homeostasis (Lynch & Pedersen, 2016). A hallmark of IBD is a profound imbalance in this community, a state known as dysbiosis, which is strongly associated with disease activity and progression (Glassner, Abraham & Quigley, 2020; Larabi, Barnich & Nguyen, 2020). Observational studies (Liu et al., 2021) have consistently linked reduced microbial diversity and the depletion of beneficial taxa, such as Faecalibacterium prausnitzii (Sokol et al., 2008), to IBD. However, these associations do not establish causality. It remains unclear whether dysbiosis is a primary driver of inflammation or merely a consequence of the established disease environment, a critical distinction for developing targeted therapeutic strategies (Graham & Xavier, 2020; Lloyd-Price et al., 2019).

The advent of large-scale genome-wide association studies (GWAS) of the gut microbiome, spearheaded by consortia such as MiBioGen (Kurilshikov et al., 2021), has paved the way for more rigorous causal inference. Consequently, several Mendelian randomization (MR) studies have recently been conducted to dissect these relationships (Liu et al., 2022; He et al., 2025). These investigations have started to identify putative causal links, implicating taxa from families such as Lachnospiraceae and genera like Ruminococcus in IBD risk, though findings can be inconsistent across IBD subtypes (Liu et al., 2022; Luo et al., 2025). Despite this progress, a critical knowledge gap persists: these studies establish potential causality but do not elucidate the underlying biological mechanisms. The specific host genes and molecular pathways through which these causal microbes exert their influence remain largely unknown.

To bridge this gap from causality to mechanism, we employed MR, a powerful genetic epidemiological method (Sekula et al., 2016). MR leverages randomly allocated genetic variants, typically single nucleotide polymorphisms (SNPs), as instrumental variables to infer the causal effect of an exposure (e.g., gut microbiota abundance) on an outcome (e.g., IBD risk) (Smith & Shah, 2003). By utilizing the random assortment of genes from parents to offspring, MR can mitigate confounding and reverse causation biases that plague traditional observational studies (Burgess, Butterworth & Thompson, 2013). In this study, we implemented a two-pronged approach: first, we performed a comprehensive MR analysis to identify specific gut microbial taxa that have a causal impact on IBD risk (Wang, Li & Ji, 2024). Second, we integrated these findings with transcriptome profiling of patients with IBD to uncover the potential host genes and molecular pathways through which these causal microbes might exert their influence (Zhu et al., 2016).

This integrated multi-omics strategy allows us not only to corroborate causal links between the gut microbiota and IBD but also to propose a mechanistic framework connecting specific microbial signatures to host gene regulation. By bridging the gap between microbial causality and host response, our findings provide novel insights into IBD pathogenesis and may unveil new therapeutic targets aimed at modulating the intricate host-microbiota dialogue.

Methods

An illustration of the analytical methods is presented in Fig. 1.

Figure 1. Schematic summary of the study.

Data acquisition

Gut microbiota GWAS data (exposure)

Summary-level data from a GWAS of the gut microbiome were obtained from the MiBioGen consortium (available at: https://mibiogen.gcc.rug.nl/) (Kurilshikov et al., 2021). As the largest multi-ethnic meta-analysis of the 16S rRNA gene to date, this resource includes data from 18,340 individuals across 24 cohorts. The analysis focused on microbial composition data derived from the V4, V3–V4, and V1–V2 hypervariable regions of the 16S rRNA gene. For our study, we included all available gut microbial taxa that were classified down to the phylum, class, order, family, and genus levels.

IBD GWAS data (outcome)

GWAS summary statistics for IBD were sourced from the IEU OpenGWAS database (GWAS ID: ieu-a-31) (De Lange et al., 2017). These data originate from a large-scale meta-analysis of individuals of European ancestry, comprising 12,882 IBD cases and 21,770 controls. This study provided robust genetic associations for use as the outcome in our MR analysis.

Gene expression datasets for mechanistic analysis

To investigate the downstream effects of identified genetic variants on host gene expression, two datasets were downloaded from the NCBI Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).

For bulk transcriptome analysis, we obtained the normalized expression data (Series Matrix File) for dataset GSE36807, which was generated using the Affymetrix Human Genome U133 Plus 2.0 Array (Platform ID: GPL570). This dataset contains expression profiles from colonic biopsies of 28 patients with IBD and seven healthy controls.

For single-cell resolution analysis, we acquired the raw count data for dataset GSE214695. This dataset contains single-cell RNA-sequencing (scRNA-seq) profiles from intestinal tissue biopsies of 18 patients with IBD, allowing for cell-type-specific expression analysis.

MR analysis

Instrumental variable selection and quality control

To investigate the causal effects of gut microbiota on IBD, a two-sample MR analysis was conducted. We implemented rigorous quality control criteria to select valid instrumental variables (IVs). First, independent SNPs associated with each microbial trait were identified at a locus-wide significance threshold of P < 1 × 10⁻⁵. Second, to ensure the independence of the selected IVs, we performed linkage disequilibrium (LD) clumping using a strict threshold of r² < 0.001 within a physical window of 10,000 kb, retaining only the SNP with the lowest P-value in each locus.

Third, to mitigate potential weak instrument bias, the strength of each candidate SNP was assessed using the F-statistic. As the effect size (β) and standard error (SE) were available from the exposure GWAS summary statistics, the F-statistic was calculated directly using the formula F=β²/SE². In accordance with established MR guidelines, SNPs with an F-statistic <10 were considered weak instruments and were excluded from the analysis. Finally, we performed a harmonization process to align the effect alleles and their effect sizes between the exposure and outcome datasets. During this step, palindromic SNPs with ambiguous allele frequencies were excluded to prevent strand alignment errors.

Causal effect estimation and sensitivity analyses

The primary causal estimate was derived using the inverse-variance weighted (IVW) method (Burgess, Butterworth & Thompson, 2013). To assess the robustness of this estimate and detect potential pleiotropy, we performed several sensitivity analyses, including: MR-Egger regression, to detect and adjust for directional pleiotropy (Bowden, Smith & Burgess, 2015); the weighted median method, which provides a valid estimate even when up to 50% of IVs are invalid (Bowden et al., 2016); and the weighted mode method, for robustness against outlier IVs (Hartwig, Davey Smith & Bowden, 2017). The validity of the MR results was further confirmed using Cochran’s Q test for heterogeneity and a leave-one-out analysis to identify influential SNPs. All analyses were performed using the TwoSampleMR package(version 0.6.14) in R (Hemani et al., 2018). The statistical significance of the causal estimates was determined after correcting for multiple comparisons, as detailed in the ‘Statistical analysis’ section.

Identification of potential causal genes

Following the MR analysis, we sought to identify potential host genes that might mediate the observed causal effects. We focused on the microbial taxon with the strongest evidence of a causal association, Bifidobacteriaceae. The lead instrumental SNPs for Bifidobacteriaceae were mapped to their corresponding genes using the get_variants function from the gwasrapidd R package (version 0.99.17). This mapping process identified LCT, MCM6, and UBXN4 as candidate causal genes for subsequent mechanistic investigation.

Bioinformatic analysis of transcriptome data

Gene Set Variation Analysis (GSVA)

To assess pathway-level alterations on a per-sample basis, we performed Gene Set Variation Analysis (GSVA) using the GSVA R package (Hänzelmann, Castelo & Guinney, 2013). This analysis transformed the normalized gene expression matrix into a gene set enrichment matrix, allowing for the quantification of pathway activity for each sample. The gene sets were obtained from the Molecular Signatures Database (MSigDB v7.0), specifically the Hallmark (H), Gene Ontology (C5: GO:BP, GO:CC, GO:MF), KEGG (C2:CP:KEGG), and Immunologic Signatures (C7) gene set collections. This comprehensive selection enabled a multi-faceted exploration of biological processes, canonical pathways, and immune-related functions across the samples.

Gene Set Enrichment Analysis

To identify biological pathways associated with the causal genes identified from our MR analysis, patients were stratified into high- and low-expression groups based on the median expression of each gene (LCT, MCM6, and UBXN4). GSEA was then performed separately for each stratification to identify enriched pathways between the respective high- and low-expression groups (Subramanian et al., 2005). We used the GO and KEGG gene sets from MSigDB as the reference. Pathways with a false discovery rate (FDR) q-value <0.05 were considered significantly enriched.

Construction of miRNA-target network

To explore potential post-transcriptional regulation of key genes identified in our analysis, we predicted targeting microRNAs (miRNAs) using the TargetScanHuman (v8.0) database. A network illustrating the regulatory relationships between these key genes and their predicted miRNAs was constructed and visualized using Cytoscape (v3.7.2).

Estimation of immune cell infiltration

To quantify the relative proportions of 22 types of tumor-infiltrating immune cells from the bulk RNA-seq data, we employed the CIBERSORT algorithm with the LM22 signature matrix (Newman et al., 2015). The analysis was run with 1,000 permutations. Subsequently, Pearson correlation analysis was used to evaluate the association between the expression of key genes and the abundance of infiltrating immune cells.

Single-cell RNA-sequencing analysis

The single-cell RNA-sequencing (scRNA-seq) dataset (GSE214695) was analyzed using the Seurat R package (v4.4.0) (Hao et al., 2021). Following initial quality control (nFeature_RNA > 50, percent.mt < 5), the data underwent standard preprocessing steps including normalization (NormalizeData), scaling (ScaleData), and principal component analysis (RunPCA). The top 11 principal components were selected for downstream analysis. Cell clusters were identified and visualized using t-distributed stochastic neighbor embedding (t-SNE). This allowed us to identify the primary cell types expressing the causal genes identified in the MR analysis.

Patient cohort and immunohistochemistry

Patient samples

A total of 10 IBD tissue samples and 3 normal intestinal tissue samples were collected from the Department of Pathology at Jiangyin Hospital, Affiliated to Nantong University. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Jiangyin Hospital, Affiliated to Nantong University (NO. 2024007). Written informed consent was obtained from all participants prior to sample collection.

IHC staining

Formalin-fixed, paraffin-embedded (FFPE) tissue sections (4 µm) were used for IHC analysis. After deparaffinization and rehydration, antigen retrieval was performed by boiling the sections in citrate buffer (10 mM, pH 6.0) for 15 min. The sections were then incubated with 3% hydrogen peroxide to block endogenous peroxidase activity, followed by blocking with 5% bovine serum albumin (BSA). Subsequently, the sections were incubated overnight at 4 °C with a primary antibody against MCM6 (rabbit anti-human, 1:2000 dilution; Abcam, ab201683). The following day, sections were incubated with a horseradish peroxidase (HRP)-conjugated secondary antibody. The signal was visualized using a diaminobenzidine (DAB) substrate kit, and the sections were counterstained with hematoxylin.

IHC quantification and scoring

The expression of MCM6 was evaluated based on the percentage of positively stained nuclei. For each slide, three representative high-power fields (HPFs, 400× magnification) were selected by two independent pathologists blinded to the clinical data. In each HPF, at least 100 intestinal epithelial cells were counted, and the percentage of MCM6-positive cells was calculated. The final positivity score for each sample was recorded as the average percentage from the three HPFs.

Statistical analysis

All statistical analyses were conducted using R software (version 4.2.2). Specific analyses utilized dedicated packages as detailed in their respective methods sections.

For the two-sample MR analysis, we investigated the causal effects of multiple bacterial taxa on IBD risk. To control for multiple comparisons, we applied the Benjamini–Hochberg procedure to calculate the FDR for the P-values obtained from the primary IVW analysis (Benjamini & Hochberg, 1995). A causal association was considered statistically significant if the FDR-adjusted P-value (q-value) was less than 0.05.

For other analyses, a two-sided P-value <0.05 was considered statistically significant unless otherwise specified (e.g., for GSEA where an FDR q-value <0.05 was used, or for MR IV selection where P < 1 × 10⁻⁵ was used). Differences in MCM6 protein expression (IHC scores) between IBD and normal intestinal tissues were assessed using the Mann–Whitney U test. The association between gene expression levels and immune cell abundance was evaluated using Pearson correlation analysis.

Results

Causal associations between gut microbiota and IBD

We performed a two-sample MR analysis to investigate the causal effects of gut microbiota on IBD. The analysis utilized summary statistics from a large-scale IBD GWAS (ieu-a-31), encompassing 12,882 cases and 21,770 controls (Figs. 2A–2B).

Figure 2. Overview of the Genome-Wide Association Study (GWAS) for IBD (ieu-a-31).

(A) Manhattan plot of the GWAS results. The x-axis represents the genomic position organized by chromosome, and the y-axis shows the -log10 transformed P-values for the association of each SNP with IBD. Alternating colors are used to distinguish adjacent chromosomes for visual clarity. Labels indicate the lead SNPs in loci reaching genome-wide significance. (B) Quantile–quantile (Q–Q) plot of the observed versus expected P-values from the GWAS.

After correcting for multiple comparisons using the Benjamini–Hochberg procedure, our primary analysis with the IVW method identified five gut microbial taxa with a statistically significant causal association with IBD risk (FDR q < 0.05, Table 1). Specifically, a genetically predicted higher abundance of the Lachnospiraceae ND3007 group was associated with an increased risk of IBD (OR = 1.706, 95% CI [1.099–2.648], q = 0.022). Conversely, four taxa demonstrated a protective effect. The strongest protective signal came from the family Bifidobacteriaceae (OR = 0.811, 95% CI [0.686–0.960], q = 0.022) and its corresponding order Bifidobacteriales, which shared the same instrumental variables and effect size. Additionally, the phylum Lentisphaerae (OR = 0.866, 95% CI [0.770–0.973], q = 0.022) and its class Lentisphaeria (OR = 0.871, 95% CI [0.770–0.986], q = 0.028) were also found to be causally protective against IBD. Given that Bifidobacteriaceae showed a robust protective effect and its members are well-documented probiotics with known immunomodulatory functions, we prioritized this family for in-depth downstream mechanistic analysis. The scatter plot illustrating the causal effect of Bifidobacteriaceae on IBD is presented in Fig. 3A.

Table 1. Causal associations of gut microbiota with IBD identified by two-sample MR.

Gut microbiota	Nsnp	Method	abs(B)	OR (odds ratio)	95%CI (confidence interval)	P value	q value
Lachnospiraceae ND3007 group	3	Inverse variance weighted	0.5341	1.706	1.099–2.648	0.017	0.022
Bifidobacteriaceae	12	Inverse variance weighted	0.2090	0.811	0.686–0.960	0.015	0.022
Bifidobacteriales	12	Inverse variance weighted	0.2090	0.811	0.686–0.960	0.015	0.022
Lentisphaerae	9	Inverse variance weighted	0.1442	0.866	0.770–0.973	0.016	0.022
Lentisphaeria	8	Inverse variance weighted	0.1383	0.871	0.770–0.986	0.028	0.028

Figure 3. MR and sensitivity analyses for the causal effect of bifidobacteriaceae on IBD.

(A) Scatter plot showing the causal effect of Bifidobacteriaceae on IBD. Each point represents a single instrumental SNP, and the slope of each fitted line corresponds to the causal estimate from a different MR method. (B) Funnel plot used to assess directional pleiotropy. The symmetrical distribution of SNPs around the summary estimate suggests an absence of significant pleiotropic bias. (C) Forest plot displaying the causal estimates from individual SNPs. Each point represents the causal effect estimated from a single SNP, with the overall IVW estimate shown at the bottom. (D) Leave-one-out sensitivity analysis. Each point represents the IVW causal estimate after excluding the indicated SNP, demonstrating that no single SNP disproportionately influences the overall result.

The robustness of these findings was confirmed through a series of sensitivity analyses focused on the lead taxon, Bifidobacteriaceae. First, the MR-Egger regression intercept did not significantly deviate from zero (intercept = −0.011, P = 0.582), indicating no evidence of directional pleiotropy. The symmetrical distribution of SNPs in the funnel plot further supported this conclusion (Fig. 3B). Second, Cochran’s Q test revealed no significant heterogeneity among the instrumental variables for the IVW analysis (Q = 8.793, P = 0.641), suggesting that the causal estimates were consistent across all SNPs (Fig. 3C). Finally, the leave-one-out analysis demonstrated that the overall causal estimate was not disproportionately driven by any single SNP, confirming the stability and reliability of our results (Fig. 3D). Collectively, these analyses strongly support a robust causal protective effect of Bifidobacteriaceae on IBD risk.

Transcriptome analysis

Identification and expression analysis of potential causal genes in IBD tissues

Following the identification of a robust protective causal effect of Bifidobacteriaceae on IBD, we sought to pinpoint the host genes that might mediate this interaction. As described in the Methods, the instrumental SNPs for Bifidobacteriaceae were mapped to their corresponding protein-coding genes, which identified three potential causal genes: LCT, MCM6, and UBXN4. To investigate their functional relevance in IBD, we subsequently analyzed their expression and co-expression patterns within the GSE36807 transcriptomic dataset.

First, we cross-referenced these three genes with a comprehensive list of known IBD-associated genes from the GeneCards database (Fig. 4A). We then performed a correlation analysis between the expression of LCT, MCM6, and UBXN4 and that of the top 20 IBD-associated genes (ranked by Relevance Score).

Figure 4. Transcriptomic and co-expression analysis of IBD-associated genes in the GSE36807 dataset.

(A) Expression heatmap of the top 20 IBD-associated genes, ranked by GeneCards Relevance Score, displaying Z-score normalized expression in IBD and normal control samples. (B) Multi-component co-expression analysis of the three MR-identified causal genes (LCT, MCM6, UBXN4) and the top 20 IBD-associated genes. The central panel is a correlation dot plot where color indicates the Pearson correlation coefficient (red: positive; purple: negative) and dot size is inversely proportional to the p-value (larger dots for P < 0.05). Flanking the central plot are detailed analyses of MCM6’s strongest correlations. The scatter plot on the right shows a strong positive correlation between MCM6 and IL1B (r = 0.594, P = 0.00017), while the scatter plot on the left shows a strong negative correlation with APP (r =  − 0.545, P = 0.00071). Adjacent box plots confirm the significant differential expression of all three genes in disease versus control samples (MCM6, P = 0.044; IL1B, P = 0.023; APP, P = 0.00041; by ANOVA).

This analysis revealed significant co-expression relationships, with MCM6 showing the strongest correlations (Fig. 4B). Notably, MCM6 expression was significantly positively correlated with the pro-inflammatory cytokine IL1B (r = 0.594, P < 0.05) and negatively correlated with Amyloid Precursor Protein (APP) (r =  − 0.545, P < 0.05). These findings suggest a potential regulatory link between MCM6 and key genes involved in IBD pathogenesis.

Functional pathways associated with causal gene expression

To elucidate the biological functions associated with the MR-identified genes, we performed GSVA on the GSE36807 dataset. Specifically, high expression of LCT was significantly associated with increased activity in key inflammatory and metabolic pathways, such as the IL-6/JAK/STAT3 signaling pathway (HALLMARK_IL6_JAK_STAT3_SIGNALING) and the xenobiotic metabolism pathway (HALLMARK_XENOBIOTIC_METABOLISM) (Fig. 5A). High expression of MCM6 was linked to pathways crucial for cell growth and immune response, including the mTORC1 signaling pathway (HALLMARK_MTORC1_SIGNALING) and the IL-2/STAT5 signaling pathway (HALLMARK_IL2_STAT5_SIGNALING) (Fig. 5B). Interestingly, the IL-2/STAT5 signaling pathway (HALLMARK_IL2_STAT5_SIGNALING) was also positively associated with the expression of UBXN4, which was additionally linked to the p53 pathway (HALLMARK_P53_PATHWAY) (Fig. 5C).

Figure 5. GSVA of MR-identified causal genes.

The analysis identifies Hallmark pathways whose activity scores are significantly correlated with the expression of (A) LCT, (B) MCM6, and (C) UBXN4 in the GSE36807 dataset. The bar plots display the top pathways that are positively (blue) and negatively (green) correlated with the expression of each respective gene.

Biological processes and pathways modulated by causal genes

To further define the biological functions of the causal genes, we performed GSEA based on their expression in the GSE36807 dataset.

The KEGG pathway analysis revealed that each gene is associated with distinct signaling networks. High expression of LCT was enriched in metabolic pathways, including the PPAR signaling pathway and retinol metabolism, as well as the pro-inflammatory IL-17 signaling pathway (Fig. 6A). In contrast, MCM6 was strongly linked to key inflammatory and cell death cascades, such as Necroptosis, the NF-kappa B signaling pathway, and the TNF signaling pathway (Fig. 6B). UBXN4 expression was associated with major intracellular signaling axes, including the JAK-STAT and PI3K-Akt signaling pathways (Fig. 6C).

Figure 6. GSEA enrichment plots for pathways and processes associated with LCT, MCM6, and UBXN4.

The top enriched (A–C) KEGG pathways and (D–F) GO Biological Process terms are shown for gene signatures associated with high expression of each respective gene. The plots illustrate the running enrichment score and the distribution of leading-edge genes.

Complementing these findings, Gene Ontology (GO) analysis highlighted specific biological processes. LCT was primarily involved in metabolic functions, such as ‘urea metabolic process’ and ‘lipid digestion’ (Fig. 6D). MCM6 was implicated in adaptive immune responses, including ‘phagocytosis’ and both ‘B cell receptor’ and ‘T cell receptor’ signaling (Fig. 6E). Finally, UBXN4 was associated with innate immunity and programmed cell death, showing enrichment in the ‘type I interferon signaling pathway’ and ‘pyroptosis’ (Fig. 6F).

Predicted miRNA regulatory network for causal genes

To explore potential post-transcriptional regulation of the MR-identified causal genes, we predicted their targeting microRNAs (miRNAs) using the TargetScanHuman database. This analysis identified 947 unique miRNAs that putatively regulate LCT, MCM6, and UBXN4, forming a complex network of 1,148 potential mRNA-miRNA interactions, which is visualized in Fig. S1. The extensive number of predicted interactions suggests that all three causal genes are subject to a complex layer of post-transcriptional control that may influence their function in IBD.

Association of causal genes with the immune microenvironment

To investigate how the causal genes might influence IBD pathology, we analyzed the correlation between their expression and the immune cell infiltration landscape in the GSE36807 dataset. Our analysis revealed significant and distinct associations for each gene.

Specifically, LCT expression was positively correlated with plasma cells but negatively correlated with several T cell subsets, including naive CD4+ T cells and follicular helper T cells (Fig. 7A). In contrast, MCM6 expression was strongly associated with pro-inflammatory and antigen-presenting cells, showing positive correlations with follicular helper T cells and M0 macrophages, and negative correlations with resting dendritic cells and resting memory CD4+ T cells (Fig. 7B). Notably, UBXN4 expression was negatively correlated with key effector and regulatory populations, including activated NK cells and regulatory T cells (Tregs) (Fig. 7C). The most significant of these correlations are visualized in scatter plots (Figs. 7D–7F).

Figure 7. Correlation of causal gene expression with immune cell infiltration in IBD.

(A–C) Lollipop plots showing the Pearson correlation coefficients between the expression of (A) LCT, (B) MCM6, and (C) UBXN4 and the abundance of 22 types of infiltrating immune cells. (D–F) Scatter plots visualizing the most significant correlation for (D) LCT, (E) MCM6, and (F) UBXN4, respectively. The line represents the linear regression fit, and the shaded area indicates the 95% confidence interval.

To further explore their roles, we assessed the relationship between these genes and a panel of immunomodulators using the TISIDB database. This revealed extensive correlations with key chemokines, receptors, and immune checkpoint molecules (Fig. 8). Taken together, these findings strongly suggest that the MR-identified causal genes are actively involved in shaping the IBD immune microenvironment, likely by influencing immune cell recruitment, activation, and function.

Figure 8. Correlation profiles of causal genes with key immunomodulatory molecules.

Bubble charts show the correlation between causal gene expression and (A) Chemokines, (B) Immunoinhibitors, (C) Immunostimulators, (D) MHC molecules, and (E) Receptors (data from TISIDB). The color of the bubble indicates the direction and magnitude of the correlation coefficient (r; red for positive, blue for negative). The size of the bubble represents the statistical significance and is inversely proportional to the p-value (larger bubbles for smaller p-values). The specific significance thresholds are denoted by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001).

Cellular localization of causal genes in the IBD microenvironment

To identify the specific cellular sources of the causal genes within diseased tissue, we analyzed a single-cell RNA-sequencing dataset (GSE214695) from patients with IBD. Unbiased clustering of the 18 samples revealed 18 distinct cell clusters (Fig. 9A), which were subsequently annotated into eight major cell populations: B cells, epithelial cells, neutrophils, T cells, monocytes, smooth muscle cells, endothelial cells, and tissue stem cells (Fig. 9B).

Figure 9. Single-cell analysis reveals the cell-type-specific expression of causal genes in IBD tissue.

(A) UMAP projection of all cells from the GSE214695 dataset, color-coded by unsupervised cluster. (B) UMAP projection with cells colored by their annotated cell type identity. (C) Dot plot showing the average expression (color scale) and the percentage of cells expressing the gene (dot size) for LCT, MCM6, and UBXN4 across the major annotated cell types. (D) Feature plots showing the expression level and distribution of each causal gene overlaid on the UMAP.

Mapping the expression of the causal genes onto this cellular landscape revealed highly specific localization patterns. LCT expression was almost exclusively confined to epithelial cells, consistent with its primary digestive function. In contrast, MCM6 was predominantly expressed in proliferative populations, including T cells and tissue stem cells. UBXN4 expression was most prominent within immune lineages, particularly in T cells and monocytes (Figs. 9C–9D). This single-cell resolution provides direct evidence linking the MR-identified genes to distinct cell types that are known to be pivotal in the pathogenesis of IBD.

Upregulation of MCM6 protein in tissues from patients with IBD

To validate our transcriptomic findings at the protein level, we performed IHC to assess MCM6 expression in intestinal tissues from patients with IBD and normal controls. The IHC analysis demonstrated that MCM6 protein expression was significantly higher in IBD tissues compared to normal tissues (Fig. 10A, p <  0.05). Furthermore, MCM6 staining was predominantly localized to the cell nucleus and was particularly strong in infiltrating inflammatory cells within the lamina propria, with weaker expression in intestinal epithelial cells (Fig. 10B). These results confirm that MCM6 is overexpressed in the IBD microenvironment, consistent with our bioinformatics predictions.

Figure 10. Immunohistochemical validation of MCM6 protein expression in IBD.

(A) Box plot quantifying MCM6 protein levels (IHC score) in intestinal tissues from normal controls (n = 3) and IBD patients (n = 10). The difference is statistically significant (*P < 0.05, Student’s t-test). (B) Representative IHC images of MCM6 staining. In normal tissue (left), MCM6 expression is low. In IBD tissue (right), strong nuclear staining (brown) is observed, particularly in infiltrating immune cells, with weaker staining in epithelial cells. Scale bars, 500 µm.

Discussion

While previous MR studies have established causal links between specific gut microbial taxa and IBD (Liu et al., 2022; Xu et al., 2023), the underlying host-mediated mechanisms remain largely undefined. Our study represents a significant advancement by integrating MR with multi-level transcriptomic analysis. This multi-omics approach allows us to move beyond simply identifying a causal microbe and begin dissecting how it may influence IBD pathogenesis through the regulation of specific host genes, thereby providing a novel framework for understanding microbe-host interactions in this disease.

Our primary MR analysis identified a robust, protective causal relationship between the Bifidobacteriaceae family and IBD. Within this family, the genus Bifidobacterium is the most abundant and well-characterized member, making it the most plausible driver of the observed protective effect. This interpretation is consistent with a substantial body of evidence highlighting its probiotic potential through multifactorial mechanisms, including immune modulation, restoration of microbial balance, and enhancement of intestinal barrier integrity (O’Callaghan & van Sinderen, 2016; Martin et al., 2023). For instance, specific Bifidobacterium species are known to suppress pro-inflammatory taxa (Martin et al., 2023; Yao et al., 2021) and bolster the intestinal barrier (Cui et al., 2023).

However, it is crucial to interpret this family-level finding with caution, as the Bifidobacteriaceae family also includes genera with markedly different or even detrimental clinical profiles. For example, Gardnerella is a key genus associated with bacterial vaginosis (Rosca et al., 2020), and Scardovia is linked to severe early childhood caries (Kameda et al., 2020). Therefore, the strong protective signal we observed for the Bifidobacteriaceae family is likely driven by the dominant beneficial effects of the highly abundant Bifidobacterium genus, which may statistically overwhelm or mask the neutral or potentially negative contributions of other less abundant members.

This distinction reinforces the therapeutic rationale for using precisely targeted, genus-specific probiotics. Our findings lend strong support to the continued clinical use and development of formulations based specifically on the Bifidobacterium genus, which are already applied in managing IBD, particularly Ulcerative Colitis (Nie et al., 2025; Matsuoka et al., 2018; Hu et al., 2022). Nevertheless, significant knowledge gaps persist. Disentangling the specific contributions of each genus and species within the family, and elucidating the role of their metabolites, such as SCFAs (Parada Venegas et al., 2019), in mediating these effects remain key areas for future research. Our subsequent analyses connecting these microbial signals to host gene expression begin to address this critical gap.

In addition to Bifidobacteriaceae, our MR analysis identified a novel protective causal role for the phylum Lentisphaerae against IBD. Although this phylum is less characterized in the context of IBD, our finding is strongly supported by evidence from other chronic inflammatory conditions. For instance, a recent MR study on childhood asthma revealed a causal relationship where the disease leads to a reduction in Lentisphaerae abundance, suggesting its protective role in airway inflammation (Li et al., 2025). Furthermore, an interventional study demonstrated that a nutraceutical blend promoting weight loss and reducing systemic inflammation significantly increased the abundance of the Lentisphaerae phylum (Santamarina et al., 2025). While the evidence for Bifidobacteriaceae is currently more substantial and mechanistically tractable, leading us to focus our downstream analysis on this family, the robust causal link we have established with Lentisphaerae represents a highly promising and novel avenue for future investigations into protective microbes in IBD.

Furthermore, the interplay between gut microbiota, host immunity, and vitamin D signaling represents another critical axis in IBD pathogenesis that complements our findings (Fakhoury et al., 2020; Aggeletopoulou et al., 2023). Vitamin D deficiency is highly prevalent in patients with IBD and is associated with intestinal dysbiosis and increased disease activity (Dell’Anna et al., 2025). The vitamin D receptor (VDR) is widely expressed in intestinal epithelial and immune cells, where it plays a pivotal role in maintaining barrier integrity, regulating antimicrobial peptide secretion, and modulating immune responses (Vernia et al., 2022). Intriguingly, this relationship is bidirectional. While vitamin D/VDR signaling can shape the gut microbiome by promoting beneficial species and suppressing pathogens, the microbiota can, in turn, influence vitamin D metabolism and function (Battistini et al., 2020; Murdaca et al., 2024). Specifically, beneficial bacteria, including members of the Bifidobacteriaceae family, are major producers of short-chain fatty acids (SCFAs) like butyrate. Butyrate has been shown to upregulate VDR expression in colonic cells, creating a potential positive feedback loop (Vernia et al., 2022). Based on this model, the protective Bifidobacteriaceae we identified may not only exert direct effects but also enhance host responsiveness to vitamin D, thereby amplifying anti-inflammatory pathways and strengthening the gut barrier. This hypothesis warrants further investigation, suggesting that the therapeutic effects of Bifidobacterium-based probiotics may be related to the host’s vitamin D levels.

Our transcriptome-wide MR analysis provides a compelling mechanistic link between the protective effects of Bifidobacteriaceae and the regulation of the host gene LCT. The LCT gene encodes the lactase enzyme, and its genetic variants are the primary determinant of adult-type lactose intolerance. The relationship between IBD and lactose intolerance has often been considered unidirectional (Mishkin, 1997), where intestinal inflammation damages the mucosa, leading to secondary lactase deficiency.

However, our findings, supported by recent large-scale population studies, suggest a more intricate and potentially beneficial interplay (Parizadeh & Arrieta, 2023; Brandao Gois et al., 2022). These studies have demonstrated a strong positive correlation between the LCT genotype associated with lactose intolerance and the abundance of Bifidobacterium. This points to a plausible prebiotic mechanism: in individuals with low lactase expression, undigested lactose travels to the colon where it can selectively fuel the growth of saccharolytic bacteria like Bifidobacteriaceae.

Therefore, our results reframe the role of LCT in IBD. Instead of simply being a marker of intestinal damage, genetically determined low LCT expression might create a favorable niche for protective microbes. This microbial advantage, fueled by dietary lactose, could contribute to the protective effect against IBD that we identified in our primary MR analysis. This novel hypothesis warrants direct investigation. Future studies should aim to confirm whether Bifidobacterium or its metabolites can directly modulate LCT expression and further clarify how lactose fermentation products impact intestinal barrier function and immune homeostasis in the IBD context.

Our study identifies MCM6 as a pivotal, yet previously underappreciated, player in IBD immunopathology. While its canonical function is in DNA replication (Wei & Zhao, 2016) and its genetic linkage to LCT is well-known (Lukito et al., 2015), our findings point towards an independent immunological function that has been largely overlooked.

This immunological role is first suggested by our immunoinfiltration analysis, which revealed a striking positive correlation between MCM6 expression and both T follicular helper (TFH) cells and pro-inflammatory M1 macrophages, alongside a negative correlation with anti-inflammatory M2 macrophages. This positions MCM6 within a pro-inflammatory axis that is central to IBD pathogenesis. The link to TFH cells is particularly compelling, as they are key drivers of Th17-mediated inflammation and are functionally inhibited by established IBD therapies like Ustekinumab (Globig et al., 2021), suggesting that MCM6 operates within a clinically relevant pathway.

Crucially, our IHC analysis provides direct experimental validation for this hypothesis at the protein level. The IHC results not only confirmed a significant upregulation of MCM6 protein in IBD tissues but, more importantly, revealed its precise cellular localization. The staining was not diffusely increased; instead, it was predominantly and intensely localized within the nuclei of infiltrating immune cells in the lamina propria. This specific localization allows us to distinguish its role from a mere marker of epithelial regeneration, which would also be proliferating. The intense signal within immune cells strongly suggests that MCM6 is actively involved in driving the proliferation of the inflammatory infiltrate itself.

Synthesizing these findings, we propose a refined model where MCM6 acts as a functional hub, sustaining chronic inflammation by fueling the clonal expansion of pathogenic lymphocytes and the proliferation of M1 macrophages. This dual role in orchestrating both adaptive (TFH) and innate (M1) immune responses makes MCM6 an attractive therapeutic target. Future investigations should focus on validating this mechanism and exploring whether MCM6 expression levels in immune infiltrates correlate with disease severity or predict therapeutic response, potentially paving the way for novel diagnostics and targeted immunotherapies for IBD.

Our analysis implicates UBXN4 as a potentially crucial regulator of cellular stress responses in IBD, a role that, while consistent with its known molecular functions, has not been previously explored in the context of intestinal inflammation. UBXN4 is a key modulator of the p97/VCP complex, a master regulator of cellular protein homeostasis (proteostasis) (Lim et al., 2009). This unique position places UBXN4 at the crossroads of two cellular pathways fundamentally implicated in IBD pathogenesis: endoplasmic reticulum-associated degradation (ERAD) (Tan et al., 2023) andautophagy (Dowdell & Colgan, 2021; Larabi, Barnich & Nguyen, 2020).The dysregulation of UBXN4 could fundamentally impair the cell’s ability to manage stress and maintain homeostasis.

We therefore hypothesize that UBXN4 acts as a central quality control node in intestinal cells. Its failure would precipitate a dual crisis, crippling both the clearance of misfolded proteins via ERAD and the removal of damaged organelles via autophagy. This combined failure provides a compelling mechanism for the sustained inflammation and barrier dysfunction seen in IBD. Consequently, validating UBXN4’s role as a master regulator of proteostasis could unveil novel therapeutic strategies aimed at restoring cellular resilience in the gut.

Conclusions

In conclusion, by integrating MR with multi-level transcriptomic analysis, this study provides robust causal evidence for a protective role of the Bifidobacteriaceae family against IBD. More importantly, we bridge the gap from microbial causality to host response by identifying three key mediating genes and proposing their distinct mechanistic roles: LCT, whose genetically determined low expression may foster a protective prebiotic niche; MCM6, which acts as a functional hub driving the proliferation of pathogenic immune infiltrates; and UBXN4, a central regulator of cellular proteostasis whose failure can precipitate inflammatory stress. These findings offer a novel framework for understanding the molecular dialogue between the gut microbiota and the host in IBD.

We acknowledge the limitations of our study, primarily that protein-level validation was confined to MCM6 due to antibody availability. The mechanistic links proposed herein, however, establish a clear roadmap for future research. Functional validation using in vitro co-culture systems and in vivo models with cell-type-specific gene perturbations is now required to confirm these pathways. Ultimately, dissecting these intricate host-microbiota interactions at the molecular level holds significant promise for developing novel biomarkers for risk stratification and precision therapies aimed at restoring intestinal homeostasis in IBD.

Supplemental Information

Supplemental Information 1. Predicted miRNA-mRNA regulatory network for IBD causal genes.

The network visualizes the predicted interactions between the three causal genes (LCT, MCM6, and UBXN4) and the miRNAs predicted to target them, based on data from the TargetScanHuman database. Nodes represent mRNAs (pink) and their targeting miRNAs (orange), while edges represent the predicted regulatory interactions.

Funding Statement

This study was funded by the Science and Technology Development Fund Project of the Affiliated Hospital of Xuzhou Medical University (Grant No. XYFC202402), the Major Project of the Wuxi Municipal Health Commission (Grant No. Z202410), and the Wuxi “Taihu talent plan” for excellent medical expert team (Grant No. 2021-9). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.