Tumor-suppressing multi-enterobacteria enhance the anti-PD-1/PD-L1 efficacy in microsatellite stable colorectal cancer

Abstract

Gut microbiome plays a pivotal role in modulating immunotherapy responses in colorectal cancer (CRC) treatment. While individual enterobacteria have been identified as enhancers of anti-PD-1/anti-PD-L1 therapy, the synergistic effects of multiple probiotic strains remain insufficiently explored. In this study, we investigated the therapeutic potential of Tumor-Suppressing Multi-Enterobacteria (TSME), a consortium of nine beneficial intestinal probiotic strains, in enhancing anti-PD-1/anti-PD-L1 therapy for microsatellite stable (MSS) CRC. Using a tumor-bearing mouse and employing techniques including flow cytometry, immunohistochemistry, ELISA, and genomic sequencing, we found that TSME significantly improved the efficacy of immune checkpoint inhibitors (ICIs) by optimizing tumor immune and microbe microenvironment. Specifically, the addition of TSME increased CD8⁺ T cell infiltration and reshaped cytokine profiles, including reducing pro-inflammatory cytokines (IL-17, IL-1β, IL-6, and TNF-α) while elevating anti-inflammatory factors (IFN-γ). Moreover, TSME significantly up-regulated key immune pathways, including TNF signaling, cytokine-cytokine receptor interaction, and JAK-STAT signaling. In addition, TSME restructured the gut microbiome, increasing the abundance of beneficial bacteria such as Akkermansia and Alistipes. These findings highlight the synergistic effect of the multi-strain probiotics in enhancing ICI efficacy. Well-formulated probiotic consortia offer a promising strategy for enhancing immunotherapy outcomes in MSS CRC and advancing broader implementation of microbiome-assisted precision oncology.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44494-5.

Introduction

Colorectal cancer (CRC), ranking as the third most common cancer and the second leading cause of cancer death worldwide, represents a significant health challenge¹. Despite advancements in treatments such as surgery, systemic chemotherapy, local radiotherapy, and targeted therapy, the mortality rate of CRC remains high. A remarkable aspect of CRC development is that it can evade the surveillance and destruction of the immune system². The emergence of novel immunotherapy has offered new prospects for patients with metastatic CRC, notably immune checkpoint inhibitors (ICIs). In the past decades, ICIs have demonstrated efficacy in treating multiple cancers, such as malignant melanoma³ and non-small cell lung cancer⁴(NSCLC), and have become a focal point of research in CRC treatment⁵. Studies have shown that for patients with metastatic CRC who have deficient mismatch repair (dMMR) or microsatellite instability-high (MSI-H), anti-programmed cell death protein 1 antibody (αPD-1)/anti-programmed cell death-ligand 1 antibody (αPD-L1) alone or in combination with anti-cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) antibody can achieve durable efficacy^6–9. However, these benefits are limited to a small subset of CRC patients, as most cases are characterized by proficient mismatch repair (pMMR) or microsatellite stable (MSS). CRC patients with pMMR/MSS typically have poor responses to immune monotherapy, highlighting the need to improve the efficacy of ICIs in this larger patient group.

Recent sequencing studies revealed a prevalent imbalance in the gut microbiota of CRC patients, characterized by a decreased proportion of beneficial bacteria and an increased proportion of potentially harmful bacteria^10–12. These findings underscore the intricate connection between gut microbiota and the occurrence, development, and treatment of CRC^13,14. Addressing the imbalance of intestinal flora in CRC patients can bolster both systemic and local immunity, thus amplifying the anti-tumor immune response. Therefore, leveraging the symbiosis between intestinal flora and immunotherapy to enhance anti-tumor immune responses emerges as a promising strategy for cancer treatment¹⁵. Several lines of evidence from preclinical to clinical research have gradually established that the gut microbiota can modulate antitumor immunity and affect the efficacy of cancer immunotherapies, especially ICIs^16–19. Variations in gut microbiota can cause changes in the tumor microenvironment, subsequently affecting the efficacy of immunotherapy²⁰. The gut microbiome’s critical role in modulating tumor immune surveillance and enhancing the efficacy of ICIs positions it as a key target for innovative cancer therapeutic strategies²¹.

Recent advances have underscored the pivotal role of gut microbiomes in modulating responses to CRC treatment, particularly immunotherapies targeting the PD-1/PD-L1 axis²². While existing studies have unrivalled the potential of specific enterobacteria or probiotics in improving the anti-PD-1/PD-L1 efficacy for CRC treatment^23–25, the exploration of multi-enterobacteria synergies remains limited. Notably, Huang et al.²⁶ reported that fecal microbiota transplantation (FMT), characterized by enrichment of Bifidobacterium thetaiotaomicron, B. fragilis, and reduction of B. ovatus, synergistically enhanced efficacy of anti-PD-1 therapy. Montalban-Arques²⁷ emonstrated that supplementation with a mixture of four Clostridiales strains outperformed anti-PD-1 therapy in CRC mouse models. In this context, we investigated the collective impact of TSME—a consortium of nine strains of beneficial intestinal probiotics, representing a taxonomically diverse composition distinct from previously reported consortia, on the efficacy of αPD-1/αPD-L1 therapy in a microsatellite stable (MSS) CRC mouse model.

The interplay between gut microbiota and CRC is complex, influencing not only the balance and diversity of intestinal flora, but also playing an important role in colorectal carcinogenesis and anti-tumor immune response. In this context, our research focuses on elucidating the underlying mechanisms through which TSME synergistically enhances anti-tumor effects in combination with ICIs. Our findings aim to provide foundational insights into how TSME, by modulating tumor immune and microbe microenvironment, improves the efficacy of αPD-1/αPD-L1 in the treatment of MSS CRC.

Materials and methods

Tumor-suppressing multi-enterobacteria (TSME)

Following the extensive pre-experimental screening, we developed TSME formulation by selecting nine probiotic species at their optimal concentrations and dosages. Strain selection was guided by three key criteria: (1) Physiological resilience, with the ability to withstand acidic and cholate-rich environments, ensuring survival during gastrointestinal passage; (2) Commensal relevance, enrichment of beneficial strains are commonly observed in the gut microbiota of healthy individuals; (3) Functional diversity, exhibiting unique, complementary functional properties. The selection rationales were based on the reported functionality of each strain (e.g., tumorigenesis suppression and immunotherapy efficacy enhancement), which has been detailed in our earlier publication²⁸. The final TSME formulation comprised Bifidobacterium adolescentis, B. animalis, B. bifidum, B. longum, Lactobacillus acidophilus, L. casei, L. reuteri, L. rhamnosus, and Streptococcus thermophilus. These probiotics work synergistically to modulate the intestinal flora, strengthen the gut micro-ecology, restore digestive function, activate antitumor immune responses, and promote overall systemic health.

TSME is a food-grade probiotic complex certified under Food Production License No. SC10632117100037. Prior to its approval by the State Food and Drug Administration, it underwent a thorough toxicity assessment in line with the regulatory requirements, ensuring the product’s safety when consumed at recommended doses.

Mouse model and treatment regimens

A tumor-bearing mouse model of MSS CRC was established using the mouse-derived CT26 cell line, known for its high immunogenicity and aggressive growth in syngeneic BALB/c mice²⁹. Six-week-old female BALB/c mice, weighing 18–20 g, of specific pathogen-free (SPF) grade, were purchased from Nanjing Jicui Company. These mice were housed in the SPF Barrier Center of the Animal Center of the Scientific Research Building, Tongji Hospital, with unrestricted access to food and water. At the end of 1 week of quarantine, the mice were shaved, ear-tagged, and weighed. CT26 cells, cultured in RPMI1640 medium with 10% inactivated fetal bovine serum, were maintained in an incubator with constant temperature (37.0℃) and 5% carbon dioxide. For the experiment, mice were subcutaneously inoculated with a relatively high concentration of CT26 cells (1 × 10⁶) on the right back.

Tumor growth was closely monitored and tumor volume was measured with a vernier caliper. Upon reaching a tumor volume of 50–100 mm³, the mice were stratified into six groups, each consisting of 10 mice. Careful allocation was undertaken to ensure that every group contained a well-balanced distribution of various tumor sizes. The six treatment regimen groups comprised: control group, intraperitoneal injection of normal saline and oral gavage with normal saline (Group 1); TSME group, intraperitoneal injection of normal saline and gavage with TSME (Group 2); αPD-1 group, intraperitoneal injection of αPD-1 (Leinco Technologies, RRID: AB_2749826) and oral gavage with normal saline (Group 3); αPD-1 + TSME group, intraperitoneal injection of αPD-1 and oral gavage with TSME (Group 4); αPD-L1 group, intraperitoneal injection of αPD-L1 (Leinco Technologies, RRID: AB_273757) and oral gavage with normal saline (Group 5); αPD-L1 + TSME group, intraperitoneal injection of αPD-L1 and oral gavage with TSME (Group 6). Each mouse received αPD-1 or αPD-L1 three times per week at a dose of 5 mg/kg. In addition, either 200 µl of normal saline or 0.5 mg/ml TSME was administered daily by gavage for 14 days. Tumor volume and weight of mice were continuously monitored. When the tumor volume reached 2000 mm³, indicating an excessive tumor load, the mice were euthanized. At the end of the experiment, samples including serum, feces, spleen, and tumor tissue were collected from each group for analysis, using techniques such as flow cytometry, enzyme-linked immunosorbent assay (ELISA), immunohistochemistry (IHC), 16 S ribosomal DNA (16 S rDNA), and Transcriptome sequencing (RNA-seq). To be noted, sample size for each assay were determined by the number of mice in each group that developed tumors of adequate size and yielded sufficient tissue volume to meet the assay requirements.

Prior to sample collection, mice were anesthetized using isoflurane. Specifically, two units of isoflurane were drawn into an insulin syringe and injected into a piece of cotton to ensure even absorption. The cotton was then placed inside a containment box specially designed for small animal handling. Each mouse was carefully placed inside the box which was promptly sealed. The mice were closely monitored, and anesthesia was confirmed once the animals exhibited a complete lack of reflexive response. Following anesthesia, blood samples were collected by removing the eyeball via retro-orbital puncture. Immediately after blood collection, the mouse was euthanized using the cervical dislocation. This procedure was conducted swiftly and humanely to ensure minimal suffering. Once euthanasia was confirmed, tumor dissection and subsequent analyses were carried out. All animal experiments were conducted in strict adherence to the ARRIVE guidelines and the Guidelines for the Care and Use of Laboratory Animals established by Tongji Hospital, affiliated with Huazhong University of Science and Technology. The experimental protocol for this study was approved by the Animal Ethics Committee of Tongji Hospital, under the Ethics Approval Number TJH-202,208,018. We confirm that all methods were performed in strict accordance with relevant international and institutional guidelines and regulations, including the American Veterinary Medical Association (VMA) Guidelines for the euthanasia of animals.

Flow cytometry, immunohistochemistry, and ELISA

Freshly isolated tumor and spleen tissues from mouse models were finely chopped and homogenized. The homogenate was digested using a mixture of digestive enzymes, including type IV collagenase (1 mg/ml), hyaluronidase (0.01 mg/ml), and DNase1 (0.002 mg/ml), followed by incubation at 37℃ for 45 min, with intermittent shaking every 15 min. After enzymatic digestion, red blood cell lysate was added for lyse erythrocytes, and a single cell suspension was made after multiple centrifugations, which was then filtered through a 200-mesh strainer and washed using PBS to remove impurities. Live/dead antibody was utilized for flow cytometry analysis before gating CD45⁺ leukocytes. Fixable Viability Stain 510 was applied to differentiate live cells from dead ones. Subsequently, cells were stained with anti-mouse antibodies from Biolegend: CD45 antibody (APC-CY7; Clone 30-F11), CD3 antibody (APC; 145-2C11), CD4 antibody (PE; GK1.5), CD8a antibody (FITC; 53 − 6.7), IFN-γ antibody (BV421; XMG1.2), and IL-17 antibody (BV650; TC11-18H10). A stepwise gating strategy was employed to identify the target cell populations. First, live/dead staining was applied to exclude dead cells, and only live cells were gated for further analysis. Within the live cell population, CD45⁺ cells were identified and gated. From the CD45⁺ population, CD3⁺ cells were selected. Finally, CD4⁺ and CD8⁺ cells were gated within CD3⁺ population. The same gating strategy was rigorously applied to both tumor and spleen samples to ensure consistency. Staining samples were analyzed using a CytoFLEX-3 Laser 13-color flow cytometer (Beckman Coulter) and data analysis was performed using FlowJo software version 10^® following the established gating scheme (Figures S2–S3).

Subcutaneous xenograft tumors taken from mice were embedded in paraffin and sectioned into 4 μm slices. These sections were stained with hematoxylin-eosin (H&E) to confirm the presence of tumors. For IHC staining, paraffin-embedded slides were deparaffinized with xylene, rehydrated with alcohol, and immersed in 3% H₂O₂ for 15 min to eliminate endogenous peroxidase activity. The cells were then blocked with 5% goat serum at 37℃ for 20 min, followed by incubation with CD4 primary antibody (Abcam, Cat: ab9657, 1:2000 dilution) and CD8 primary antibody (Abcam, Cat: ab180780, 1:2525 dilution) at 4℃ overnight. The next day, sections were incubated with horseradish peroxidase-labeled secondary antibodies for 1 h at room temperature and then stained using a DAB kit. After dehydration and drying, the sections were fixed with neutral glue and observed under a microscope. After dehydration and drying, the dissections were fixed with neutral glue and observed under a microscope. A white light scanner (Hamamatsu Electronics; NanoZoomerS360) and NDP.View2 software were used to analyze the images.

Serum samples were collected from mice for ELISA experiments. ELISA kits were used to detect IL-1β (Neobioscience, Cat: EMC001b.96), IL-6 (Neobioscience, Cat: EMC004.96) and TNF-α (Neobioscience, Cat: EMC102a.96) in the mouse serum. To assess IL-17 and IFN-γ expression in immune cells, blood samples were collected using sodium heparin as an anticoagulant. After centrifugation, the cellular fraction was isolated, and the supernatant was discarded. The cell pellet was resuspended in fresh culture medium and stained with specific fluorophore-conjugated antibodies: IFN-γ (BV421; BioLegend, Clone XMG1.2) and IL-17 (BV650; BioLegend; TC11-18H10). Stained cells were subsequently analyzed using flow cytometry to quantify cytokine expression levels in the target immune cell populations.

16 S rDNA sequencing

Mouse fecal samples were flash-frozen in liquid nitrogen and stored in a −80℃ refrigerator until use. DNA was extracted from mouse feces using the OMEGA Soil DNA Kit (D5625-01) (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s instructions. The V3-V4 variable region of the microbial 16 S rDNA was amplified using specific primers: 338 F (F terminal: ACTCCTACGGGAGGCAGCA) and 806R (R terminal: GGACTACHVGGGTWTCTAAT). After the raw data were unloaded, the Cutadapt software was used to cut out the primer sequence. Subsequent data processing involved DADA2 for quality control analysis, including quality filtering, noise reduction, splicing and removal of chimeras of the qualified double-ended raw data. This process was in accordance with the default parameters of QIIME 2 to obtain representative sequences and ASV abundance tables. QIIME 2 software package was used to select the representative sequences of each ASV, and all the representative sequences were compared with the database. Silva (version138) database was used for comparison, and species alignment annotations were analyzed using classify-sklearn software.

RNA sequencing

For RNA sequencing, tumor tissue samples were processed to enrich mRNA using Oligo(dT) beads. The polyA-tailed mRNA was then fragmented using divalent cations in NEB Fragmentation Buffer. Library construction and quality inspection were performed according to the common method of NEB library construction. The library construction kit was NEBNext^®Ultra™RNA from Illumina^®. After library construction, initial quantification was performed using a Qubit 2.0 fluorometer. The library was then diluted to a concentration of 1.5ng/µl, and then the insert size of the library was checked using an Agilent 2100 bioanalyzer. The effective concentration of the library was accurately quantified by qRT-PCR (effective concentration of the library > 2nM) to ensure the quality of the library. After library inspection, different libraries were pooled and combined according to their effective concentrations and target data volume requirements, followed by Illumina sequencing. Sequencing was performed on an Illumina NovaSeq 6000 (Illumina, USA) platform. The resulting data were subjected to comprehensive analysis for insights into the gene expression patterns in the tumor tissues.

Statistical methods

Statistical analyses were performed using two approaches, depending on the number of groups compared. The Student’s t-test was employed for comparisons between two groups, while One-way ANOVA was applied for comparison among more than two groups. Data analysis and visualization were conducted using GraphPad Prism 8.0.2. Results were expressed as mean ± standard deviation. Statistical significance was denoted P-values with the following thresholds: not significant (ns) for P ≥ 0.05; a significance level marked by * for P < 0.05; ** for P < 0.01; *** for P < 0.001; **** for P < 0.0001.

All key resources, including reagents, proteins, assays, software and algorithms, are summarized in Table S1.

Results

TSME significantly enhanced the anti-tumor efficacy of αPD-1/αPD-L1 in MSS CRC tumor-bearing mice

Tumor inhibition experiments were conducted as outlined in Fig. 1A. Tumor volume comparisons revealed that the combination groups (αPD-1 + TSME group and αPD-L1 + TSME group) significantly outperformed the antibody-only groups in suppressing tumor growth (Fig. 1B-D, Figure S1 and Table S2). The synergistic effect of TSME was further supported by the tumor growth curve, which showed a marked reduction in tumor progression in the combination groups compared to monotherapy (Fig. 1E). In addition, mice in all treatment groups exhibited steady weight gain throughout the experiment (Fig. 1F), indicating that TSME was well-tolerated and safe. These findings indicate that TSME can effectively enhance the anti-tumor efficacy of αPD-1 and αPD-L1 therapies in MSS CRC tumor-bearing mice.

Fig. 1.

TSME significantly enhanced the anti-tumor efficacy of αPD-1/αPD-L1 in MSS CRC tumor-bearing mice. (A) Experimental design: Mice were subcutaneously implanted with 1.0 × 10⁶ CT26 cells and treated with αPD-1 or αPD-L1 5 mg/kg by intraperitoneal injection, administered three times per week from Day 0 for a total of six injections. TSME was administered via gavage for two weeks. (B) Tumor growth: Tumor volumes in each treatment group over time. (C) Representative tumor images: Tumors collected at the end of the experiment. (D) Tumor volume comparison: Final tumor volumes in each group at the end of the experiment. (E) Tumor growth trends: Growth curves subcutaneous tumors for each treatment group. (F) Body weight monitoring: Changes in body weight across all groups during the experiment period. Note: n = 9 mice per group for subfigures (B)–(F).

Combined αPD-1/αPD-L1 with TSME promoted CD8⁺ T cell infiltration and affected the expression of cytokines

Flow cytometry was used to detect CD4⁺ and CD8⁺ T cells in tumor tissues and spleen (Table S3). Results revealed that CD8⁺ T cell infiltration in tumor tissues of αPD-1 + TSME group and αPD-L1 + TSME group was significantly higher than that of αPD-1 and αPD-L1 monotherapy group (Fig. 2A, C–F). However, no significant differences in CD4⁺ and CD8⁺ T cell numbers were observed among the treatment groups in the spleen (Fig. 2B, G–J). Immunohistochemical analysis of CD8⁺ T cells in tumor tissues further supported these findings, showing significantly increased CD8⁺ T cell infiltration in the combination therapy groups (αPD-1 + TSME and αPD-L1 + TSME) compared with the respective monotherapy groups (Fig. 2K–M, Table S4). Cytokine analysis revealed distinct patterns of modulation. The levels of proinflammatory factor IL-17 were significantly reduced in the treatment groups, with the combination therapy groups showing lower levels than the monotherapy groups (Fig. 2N). Conversely, levels of the anti-inflammatory factor IFN-γ were significantly higher in the treatment groups than in the control group, with the combination groups exhibiting higher levels than the monotherapy groups (Fig. 2O). Similarly, the ELISA measurements of serum pro-inflammatory cytokines (IL-1β, IL-6, and TNF-α) showed that their concentrations were significantly reduced in the treatment groups than in the control group, and lower in the combination groups compared with the monotherapy groups (Fig. 2P–R, Tables S5–S7).

Fig. 2.

Combined αPD-1/αPD-L1 with TSME promoted CD8⁺ T cells infiltration and modulated cytokine expression. (A) Representative flow cytometry plots showing CD8⁺ and CD4⁺ T cells (gated on CD45⁺ CD3⁺ cells) in tumor tissues. (B) Representative flow cytometry plots showing CD8⁺ and CD4⁺ T cells (gated on CD45⁺ CD3⁺ cells) in spleen. (C–F) Quantitative analysis of the proportion of CD8⁺ and CD4⁺ T cells (gated on CD45⁺ CD3⁺ cells) in tumor tissues (n = 5). (G–J) Quantitative analysis of the proportion of CD8⁺ and CD4⁺ T cells (gated on CD45⁺ CD3⁺ cells) in spleen (n = 5). (K) Immunohistochemical (IHC) staining of CD8⁺ T cells in tumor tissues. Scale bar = 100 μm. (L–M) Quantitative analysis of CD8⁺ T cell infiltration in the tumor immune microenvironment across treatment groups (n = 4). (N–O) Levels of IL-17 and IFN-γ in tumor tissues, measured by flow cytometry (n = 3). (P–R) Serum levels of IL-1β, IL-6, and TNF-α, detected by ELISA (n = 9).

TSME regulated the composition of gut microbiota in CRC tumor-bearing mice

16 S rDNA sequencing analysis was performed on the fecal samples of CT26 tumor-bearing mice to understand the changes in gut microbiota following TSME and αPD-L1 treatment. Since αPD-1 and αPD-L1 had similar tumor inhibition effects, when used alone or in combination, with consistent results across flow cytometry, ELISA, and immunohistochemistry analysis, we focused on four groups for analysis: Control, TSME, αPD-L1, and αPD-L1 + TSME (Combination group) for 16 S rDNA sequencing analysis of mouse feces and RNA-seq analysis of tumor tissues. The intestinal microbial communities of the four groups were found to be relatively diverse in 16 S rDNA sequencing analysis. A total of 1809 Amplicon Sequence Variants (ASVs) were identified after noise reduction using DADA2, with unique and shared distributions: 348 ASVs in the control group, 276 in the TSME group, 310 in the αPD-L1 group, and 361 in the combination group, and a total of 233 ASVs were common to the four groups (Fig. 3A). When alpha diversity was analyzed in each group, abundance-based coverage estimator (ACE) indexes showed that there was no significant difference among the groups (Fig. 3B). Beta-diversity analysis using Bray-Curtis dissimilarity revealed distinct differences in gut microbiota community composition among the four groups (Fig. 3C). We also detected specific changes in the gut microbiota in groups at different taxonomic levels. At the phylum level, Bacteroidetes was the dominant phylum in the combination group, whereas Firmicutes predominated in the αPD-L1 group (Fig. 3D). At the genus level, the combination group demonstrated a more balanced distribution of gut microbiota (Fig. 3E). Heatmaps revealed increased clustering of key genera, including Akkermansia (AKK), Alistipes, and Turicibacter in the combination group (Fig. 3F). From STAMP differential analysis, significant differences were observed in Firmicutes, Verrucomicrobiota, Actinobacteriota, and Proteobacteria across groups at the phylum level (Fig. 3G). At the phylum level, Lactobacillus, AKK, Enterorhabdus, UCG-010 and Erysipelatoclostridi aceaedium were significantly altered among the groups (Fig. 3H). The KEGG functional prediction^30–32, visualized using a heatmap, highlighted distinct microbial functions enriched in each group (Fig. 3I).

Fig. 3.

TSME regulated gut microbiota composition in CRC tumor-bearing mice. (n = 4) (A) Venn diagram showing the total number of shared and unique ASVs across the four groups. (B) Alpha diversity analysis based on ACE indices, illustrating microbial richness across groups. (C) Bray-Curtis principal coordinates analysis (PCoA) beta diversity plot. (D–E) Plot of intestinal community structure composition at phylum level (D) and genus level (E) of samples from four groups, showing the relative abundance of species in each group. (F) community heatmap at the genus level of four groups. (G–H) STAMP differential analysis at the phylum level (G) and genus level (H). (I) KEGG function distribution heatmap^28–30 of four groups.

TSME combined with αPD-L1 further activated immune-related pathways

RNA-seq analysis was performed on CT26 tumor tissues to investigate the mechanism underlying the enhanced anti-tumor effect of TSME combined with αPD-L1 treatment. Compared with the control group, 478, 211, and 525 differentially expressed genes (DEGs) were identified in the TSME, αPD-L1, and combination (TSME + αPD-L1) group, respectively (Fig. 4A). A Venn diagram further visualized the overlap and uniqueness of DEGs across the groups compared with the control group (Fig. 4B). Notably, between the combination group and the αPD-L1 group, 228 up-regulated and 81 down-regulated DEGs were observed (Fig. 4C). Gene Ontology (GO) Analysis: Functional enrichment analysis of DEGs in the combination group versus the αPD-L1 group revealed significant enrichment in immune-related biological processes. The top 30 GO terms, visualized as bar graphs, included pathways involved in inflammatory response, immune system activation, and cytokine regulation (Fig. 4D). KEGG Pathway Analysis: Key pathways enriched in the combination group compared to the αPD-L1 group included: TNF signaling pathway, cytokine-cytokine receptor interaction, and JAK-STAT signaling pathway (Fig. 4E). Key Upregulated Genes: Genes associated with immune activation and inflammation, such as Vcam1, Fas, Bcl3, and Csf2, were significantly upregulated in the combination group (Fig. 4F).

Fig. 4.

TSME combined with αPD-L1 further activated immune-related pathways. (n = 3) (A) Heatmap of significantly differentially expressed genes (DEGs) across groups (p < 0.05). (B) Venn diagram showing the total number of significantly differentially expressed genes and their overlap among groups. (C) Volcano plot of DEGs comparing the combination group vs. αPD-L1 group, highlighting the upregulated and downregulated genes. (D) GO enrichment analysis bar plot displaying the top 30 significantly enriched GO terms for DEGs in the combination group vs. αPD-L1 group. (E) KEGG enrichment scatter plot, showing key immune-related pathways in the combination group vs. αPD-L1 group. (F) Expression levels of Vcam1, Fas, Bcl3 and 13 additional immue-gene expression levels across four groups.

Discussion

In this study, we investigated the efficacy of a well-formulated probiotic consortium—TSME—in augmenting the anti-tumor immune response of αPD-1/αPD-L1 treatment in MSS CRC. Our study revealed, for the first time, how the integration of multiple beneficial enterobacteria altered both the gut microbiota composition and tumor tissue transcriptome in CT26 tumor-bearing mice. We discovered that the addition of TSME enhanced the efficacy of ICIs by optimizing immune and microbe microenvironments. Specifically, the combinational therapy (TSME + αPD-1/αPD-L1) resulted in increased CD8⁺ T cells infiltration, modulation of immune-related cytokine expression (elevated anti-inflammatory factors and reduced pro-inflammatory cytokines), upregulation of key immune pathways, and rebalancing of the gut microbiota composition, thereby synergistically promoting anti-tumor activity.

Our results elucidated the transformative impact of gut microbiota on the tumor microenvironment, notably turning the “cold” (immune-desert) into a “hot” (immune-inflamed) tumor microenvironment, thereby enhancing the efficacy of checkpoint blockade in CRC. This aligns with existing literature that positions gut microbiota as a potent enhancer of human immune function and tumor response to ICIs^33–35. It has been found that the infiltration of CD8⁺ T cells in the tumor tissues of CRC tumor-bearing mice was significantly increased after the treatment of αPD-1 combined with gut microbiota modulator³⁶. Similar results were obtained in our study, as flow cytometry and immunohistochemistry results of tumor tissue in the CT-26 model showed that the number of CD8⁺ T cell infiltration in the tumor tissue of the combined treatment group of mice was significantly increased after αPD-1/αPD-L1 combined with TSME treatment.

The primary focus of this study is the activation of CD8⁺ T cells, which serve as the primary effector cells in tumor cell elimination. Interestingly, we observed a significant reduction in the CD4⁺ T cells in mice treated with TSME compared with the control group (Fig. 2D-F). CD4⁺ helper T cells are traditionally recognized for their critical role in promoting CD8⁺ T cell activation. However, CD8⁺ T cells can be activated through multiple pathways, as evidenced by studies highlighting the direct contributions of IL-8³⁷, IL-2³⁸, and other inflammatory mediators that enhance CD8⁺ T cell activity independently of CD4⁺ T cell interactions. The mechanism underlying the decreased CD4⁺ T cell population in TSME-treated mice remains to be fully elucidated. Further studies incorporating expanded flow cytometry panels, single-cell transcriptomics, and functional assays will be critical to delineating the specific subsets involved and their contributions to the overall antitumor immune response.

Our findings suggest that the expression levels of immune-related cytokines can affect the anti-tumor effect of αPD-1/αPD-L1. The observed reduction in these cytokines in the combined treatment group, as compared to the monotherapy group, suggests a direct correlation between TSME-mediated gut microbiota modulation and the expression of immune-related cytokines. The reduction of cytokines such as IL-1β and TNF-α in the TSME-treated mice is consistent with the known immunomodulatory effects of probiotics^39,40. Our findings echo existing studies, including that changes in gut microbiota can cause changes in IFN-γ expression, and the gut microbiota improvement could inhibit the expression of cytokines IL-1β, IL-6, and TNF-α to obtain better anti-tumor effect^41,42.

The results of 16 S rDNA sequencing analysis in this study unveiled a compelling shift in bacteria phyla prevalence between treatment modalities, highlighting Bacteroidetes as dominant in the combination group (TSME + αPD-L1), whereas Firmicutes prevailed in the αPD-L1 monotherapy group. At a granular genus level, the intestinal flora in the combination group exhibited a more balanced composition and distribution. AKK has been recognized for its important role in regulating metabolic disorders and improving immune function^43–48. Our research corroborates this, with the combination group demonstrating substantial tumor reduction and a marked increase in AKK abundance, thus underscoring the link between AKK enrichment and potentiated antitumor immune responses⁴⁹. Alistipes is another focal point of our study. Alistipes plays a beneficial role in cancer immunotherapy by regulating the tumor microenvironment. One of the immunotherapy approaches is to induce myeloid cells in the tumor immune microenvironment to produce TNF, which causes tumor tissue to undergo TNF-dependent hemorrhagic necrosis to prevent tumor growth and tumor regression⁵⁰. Iida et al. verified a direct correlation between Alistipes abundance and enhanced TNF-medicated anti-tumor effects⁵¹. This underscores the beneficial effects of Alistipes in cancer immunotherapy, further substantiated by studies linked to improved responses to immunotherapies⁵². Our findings resonate with these observations, with significant upticks in both AKK and Alistipes in the combination group, signaling a synergistically enhanced response to αPD-L1 immunotherapy and a superior anti-tumor effect.

RNA-seq results of CT-26 tumor tissue in this study showed significant up-regulation of multiple immune-related pathways, including TNF signaling pathway, cytokine-cytokine receptor interaction and JAK-STAT signaling pathway, in the combined treatment group. These pathways are intrinsically related to the state of the tumor immune microenvironment, suggesting a profound regulatory role of the intestinal flora. Historical and current research underscores the multifaceted anti-tumor mechanisms of gut microbiota, ranging from direct secretion of proteins to metabolite production, like butyrate, which can inhibit the proliferation of CRC cells by down-regulating the WNT signaling pathway^53–55. These discoveries illuminate the potential of gut microbiota as an effective adjuvant for CRC immunotherapy.

Our study demonstrates that TSME significantly enhances the therapeutic efficacy of αPD-1/αPD-L1 treatments by optimizing both immune and microbe environments— increased CD8⁺ T cell infiltration, alteration of cytokine expressions, modulation of the gut microbiota composition, and up-regulation of immune-related pathways. These findings highlight the critical role of multi-enterobacteria in cancer immunotherapy and offer a promising translational avenue for the development of innovative microbiome-driven treatment strategies for MSS CRC. In addition, interindividual variability in gut microbiomes may influence TSME efficacy. The nine-strain formulation was designed to accommodate the heterogeneity, and future clinical trials will integrate microbiome profiling to identify response determinants and enable personalized regimens tailored to distinct microbial community types when needed.

This study has some limitations. Firstly, while our study demonstrated that the consortium of nine enterobacterial strains significantly enhanced the efficacy of anti-PD-1/PD-L1 therapy in MSS CRC, the precise contribution of individual strains remains unclear. Future ablation studies, systematically evaluating single strains and smaller defined subsets, will be critical to delineate the fundamental mechanisms and interactions underlying the observed synergy. Secondly, although the subcutaneous implantation CRC model offers technical simplicity, it does not full recapitulate the native tumor microenvironment interactions within the colon. To address this limitation, we plan to explore orthotopic CRC models in our future efforts. Thirdly, although our study delves into the mechanism of TSME in modulating microbiota composition and cytokine expression in the animal model, its translation to clinical practice presents both opportunities and challenges due to the intricate interplay of individual patient characteristics. Lastly, while the findings of this study provide compelling preclinical evidence, it is crucial to conduct well-designed randomized controlled trials before integrating such therapeutic approaches into routine clinical practice.

Conclusion

In summary, our study underscores the therapeutic potential of TSME as a microbiome-based intervention to enhance immunotherapy outcomes in MSS CRC. By modulating gut microbiota composition, increasing beneficial bacteria such as Akkermansia and Alistipes, and reshaping immune-related cytokine profiles, TSME promotes a favorable tumor microenvironment. The consortium notably up-regulates key immune pathways while increasing the infiltration of CD8⁺ T cells into tumor tissues. These findings emphasize the synergistic effects of multi-strain probiotics in optimizing the efficacy of ICIs, offering a novel avenue for microbiome-driven precision oncology. Future research should focus on validating the clinical applicability of TSME and its potential to revolutionize personalized cancer immunotherapy strategies.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

16S rDNA:

16 S ribosomal DNA

AKK:

Akkermansia

ASVs:

Amplicon Sequence Variants

αPD-1:

Anti-programmed cell death protein 1 antibody

αPD-L1:

Anti-programmed cell death-ligand 1 antibody

CRC:

Colorectal cancer

CTLA-4:

Cytotoxic T-lymphocyte-associated antigen 4

DEG:

differentially expressed genes

dMMR:

deficient mismatch repair

ELISA:

Enzyme-linked immunosorbent assay

ICIs:

Immune checkpoint inhibitors

IHC:

Immunohistochemistry

MSI-H:

Microsatellite instability-high

MSS:

Microsatellite stable

NSCLC:

Non-small cell lung cancer

pMMR:

Proficient mismatch repair

RNA-seq:

RNA-sequencing

SPF:

Specific pathogen-free

TSME:

Tumor-Suppressing Multi-Enterobacteria

Author contributions

X.T.S. and J.J. performed the experiments, interpreted data, and drafted the manuscript. Y.H., Z.L., and C.X.C. performed the experiments. H.Y.H. and X.G.W. conducted the experimental data analysis. F.L. critically revised the manuscript. Z.Q.D. supervised the experiments. M.S.Z. designed and supervised the study and critically revised the manuscript. All authors reviewed and approved the final manuscript.

Data availability

The RNA-seq and DNA-seq data generated in this study are publicly available in the Gene Expression Omnibus (GEO) under accession number GSE271663. Additional information or materials required to reanalyze the data are available from the corresponding author Dr. Mingsheng Zhang ([zms75@163.com](mailto: zms75@163.com)) upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.