Unveiling the interplay between microbiota and PD1/PD-L1 axis in tumor immunity and immunotherapy

Microbiota, through both direct and indirect interactions, plays an essential role in modulating the host’s immune system. The immune system’s development and function are closely influenced by microbial colonization, especially early in life (32–34). The gut microbiota is particularly important in shaping the immune system’s balance, including the distribution and function of immune cells such as regulatory T cells (Tregs), dendritic cells, and macrophages (35–40). These immune cells are key players in maintaining immune homeostasis, tolerance, and inflammation control (41, 42).

Microbiota influences immune responses by producing metabolites that directly impact immune cells (43, 44). Short-chain fatty acids (SCFAs), such as butyrate, propionate, and acetate, are metabolites produced by the fermentation of dietary fibers by gut bacteria (45–48). SCFAs have been shown to have significant anti-inflammatory effects and can promote the differentiation of Tregs, which help maintain immune tolerance and prevent excessive inflammation (49–53). These metabolites also influence the activity of dendritic cells and T cells, promoting immune responses against pathogens and tumors (54, 55).

In the context of cancer, the gut microbiota regulates the immune composition of the TME (56, 57). Microbial-induced changes in immune cell populations can influence the efficacy of the immune system in recognizing and eliminating tumor cells (58, 59). For instance, the balance between effector T cells and Tregs, as well as the infiltration of antigen-presenting cells, plays a crucial role in tumor immunity. Dysbiosis, an imbalance in microbial communities, can impair immune responses, contributing to immune escape and tumor progression (60–63). As a bridge connecting microbial signals to adaptive immunity, the function of innate lymphoid cells (ILCs) exhibits remarkable plasticity, serving as a classic example of context-dependency. For instance, while ILC3s typically maintain intestinal barrier integrity by producing IL-22, under conditions of chronic inflammation or specific microbial stimulation (such as from certain pathogenic bacteria), they may convert to a pro-inflammatory phenotype, producing abundant IL-17 and thereby promoting the development and progression of colorectal cancer (64). Conversely, in non-small cell lung cancer, the presence of ILC3s is associated with the formation of tertiary lymphoid structures and a favorable prognosis (64). This indicates that ILC function is not fixed but is “reprogrammed” by the local tissue microenvironment and specific microbial signals. Consequently, the effect of microbes on the PD-1/PD-L1 axis via ILCs is highly dependent on the functional state of the ILCs within the specific tumor context.

Tumors have evolved various mechanisms to escape detection and destruction by the host immune system, allowing for continuous growth and metastasis (65, 66). One of the most important mechanisms of immune evasion is the upregulation of immune checkpoint molecules, including PD1 and its ligand PD-L1 (67, 68). PD1/PD-L1 signaling inhibits T cell activation, thereby suppressing immune responses and promoting immune tolerance within the TME (69–71). This immune checkpoint pathway is exploited by tumors to evade immune surveillance (72, 73).

PD1/PD-L1 is an important immune checkpoint pathway. PD1 is mainly expressed on the surface of T cells and other immune cells, while its ligand PD-L1 is expressed on the surface of tumor cells and immune cells (74–76). By binding PD1 to PD-L1, T cell activation is inhibited, leading to immune tolerance and promoting tumor cells to escape immune surveillance (77, 78). This pathway plays a key role in tumor immune escape, inhibiting T cell function and weakening anti-tumor immune response. ICIs, such as anti-PD1 and anti-PD-L1 antibodies, have achieved significant efficacy in a variety of cancer treatments, but still face limitations such as mixed efficacy and immune-related side effects, which has driven deeper research to optimize their application and overcome resistance (79–81).

In addition to immune checkpoints, tumors can alter the TME by influencing immune cell recruitment, cytokine production, and immune cell dysfunction (82–87). The TME is characterized by various immune cells, including macrophages, myeloid-derived suppressor cells (MDSCs), and Tregs, which support tumor growth and immune suppression (66, 88). These immune cells can further enhance the tumor’s ability to escape immune surveillance and resist immune therapies.

Recent research has revealed that microbiota plays a significant role in tumor immune evasion by modulating immune checkpoint pathways, particularly the PD1/PD-L1 axis. The composition of the microbiota, especially gut microbes, can directly or indirectly affect immune responses, influencing how tumors evade immune surveillance. Specific microbial species can alter the expression of PD-L1 in tumor cells, immune cells, or both. For example, certain beneficial microbes such as Bifidobacterium have been found to modulate the immune system, leading to reduced tumor growth and enhanced immune responses (89–91).

In addition to modulating immune cells, microbiota-derived metabolites such as SCFAs can influence immune checkpoint regulation. Butyrate, for example, has been shown to enhance T cell function and promote the expression of PD-L1 in certain immune cell populations, thus playing a role in immune evasion within the TME (92–94). Furthermore, microbial metabolites can influence the production of pro-inflammatory cytokines and other signaling molecules that enhance or suppress immune responses, impacting the overall immune landscape within tumors (95, 96).

Interestingly, the interaction between microbiota and immune checkpoint signaling is not limited to the gut. Microbial species from other body sites, such as the skin, lungs, and even the tumor itself, have been shown to impact tumor immunity (97–101). For instance, the lung microbiome has been associated with the regulation of PD-L1 expression in lung cancer, while skin microbiota has been linked to immune responses in melanoma (102–104). This suggests that microbiota from various niches can collectively contribute to the modulation of immune checkpoints and tumor immune evasion.

Emerging evidence highlights the pivotal role of gut microbiota in modulating immune cell activity to influence PD1/PD-L1 axis dynamics. For instance, specific gut microbial communities enhance the efficacy of anti-PD-1 therapy by downregulating the PD-L2-RGMb pathway, a newly identified immunosuppressive axis. Blocking PD-L2-RGMb interactions or deleting RGMb in T cells significantly restored anti-PD-1 responsiveness in previously resistant models, including germ-free mice and fecal microbiota-transplanted mice, indicating microbiota-driven remodeling of the tumor immune microenvironment (TIM) (105). Similarly, the ketogenic diet (KD) and its metabolite 3-hydroxybutyrate (3HB) suppress myeloid cell PD-L1 upregulation induced by immunotherapy while expanding CXCR3+ T cell populations, thereby enhancing anti-PD-1/CTLA-4 efficacy (106). Additionally, interventions like MS-20 (Symbiota^®), a postbiotic, amplify CD8+ T cell infiltration into tumors and reduce PD1 expression, correlating with increased abundance of Ruminococcus bromii (107). Estrogen pretreatment in MC38 colon cancer models also shifts gut microbiota composition (e.g., reducing Enterobacteriaceae and enriching Lactobacillus), which may synergize with anti-PD-L1 therapy by altering immune cell dynamics (108). These findings collectively underscore microbiota’s capacity to orchestrate immune cell functions—spanning T cells, myeloid cells, and macrophages—to regulate PD1/PD-L1 signaling.

Gut microbiota-derived metabolites serve as critical mediators linking microbial activity to PD1/PD-L1 axis regulation. Butyrate, a SCFA, exemplifies this mechanism by enhancing histone H3K27 acetylation at the PDCD1 and CD28 promoters in CD8+ T cells, thereby elevating PD-1 expression and amplifying anti-PD-1 therapeutic responses in non-small cell lung cancer (NSCLC) (109). Similarly, 3HB from KD inhibits immunotherapy-induced PD-L1 upregulation in myeloid cells while promoting cytotoxic CD8+ T cell (CTL) expansion (106). Plant-derived nanoparticles (PNPs), such as ginger exosome-like nanoparticles (GELNs), enrich docosahexaenoic acid (DHA) via gut bacterial modulation (e.g., Lachnospiraceae and Lactobacillaceae). DHA binds to the PD-L1 promoter, suppressing its transcription by antagonizing c-Myc, thereby sensitizing tumors to PD-L1 inhibitors (110). Fecal microbiota transplantation (FMT) combined with salinomycin in colorectal cancer (CRC) models elevates propionate and butyrate levels, fostering CD8+ T cell infiltration and reducing PD-L1 expression (111). These metabolites act as molecular bridges, directly or indirectly fine-tuning immune checkpoint dynamics.

Nevertheless, interpreting the functions of microbial metabolites requires caution, as growing evidence indicates their effects are highly context-dependent and can even exhibit seemingly paradoxical dual roles. Taking the short-chain fatty acid butyrate as an example, this review mentions in several sections that it enhances CD8+ T cell function through epigenetic mechanisms, thereby improving responses to immunotherapy. Conversely, however, multiple studies have also shown that butyrate can induce an immunosuppressive state by promoting the differentiation and function of regulatory T cells (Tregs) or by directly upregulating PD-L1 expression on tumor cells and myeloid cells (51, 112, 113). This seemingly contradictory dual role is likely determined by the type of tumor and the local cytokine environment. In a microenvironment dominated by Th1-type cytokines (such as IFN-γ, IL-12), butyrate may be more inclined to enhance the function of effector T cells; while in a microenvironment enriched with TGF-β and other signals, it may drive immunosuppression. Similarly, bile acids can exert anti-inflammatory effects locally in the intestinal tract through the Farnesoid X receptor, but their elevated systemic levels are associated with immunosuppression in hepatocellular carcinoma (114). Such discrepancies may stem from various factors, including the local concentration of the metabolite, duration of exposure, the specific tumor microenvironment (e.g., cytokine milieu and immune cell composition), and the tumor type itself. Similarly, the role of specific bacterial species is not fixed. For instance, microbial communities that are beneficial in immunocompetent individuals may act as opportunistic pathogens in immunocompromised hosts, and their interactions with different anticancer regimens (such as chemotherapy and radiotherapy) may yield complex outcomes that cannot be predicted from immune checkpoint inhibitor models alone. These paradoxes highlight the complexity of the field and caution against simplistically categorizing microbes or their metabolites as purely “beneficial” or “detrimental.” Instead, there is a need to elucidate the precise conditions under which they exert specific functions.

Microbiota and their components engage specific signaling cascades to regulate PD1/PD-L1 expression. The PD-L2-RGMb axis, suppressed by commensal bacteria, represents a novel pathway through which microbiota enhance anti-PD-1 efficacy by alleviating T cell exhaustion (105). Bifidobacterium-derived extracellular vesicles (Bif.bev) exploit the TLR4-NF-κB pathway to upregulate PD-L1 in NSCLC cells while paradoxically enhancing CD8+ T cell infiltration, suggesting context-dependent modulation of immune checkpoints (115). GELN-derived aly-miR159a-3p inhibits bacterial phospholipase C, leading to DHA accumulation and subsequent PD-L1 transcriptional repression via c-Myc blockade (110). Furthermore, butyrate activates TCR signaling in CD8+ T cells, augmenting cytotoxic cytokine production and PD-1/PD-L1 checkpoint engagement (109). These mechanisms highlight the intricate interplay between microbial signals, host epigenetic regulation, and immune checkpoint pathways, offering actionable targets for overcoming immunotherapy resistance. However, the immunomodulatory effects mediated by Bif. bev exhibit significant context-dependency. The paradoxical effect of “upregulating PD-L1 while simultaneously enhancing anti-tumor immunity” observed in NSCLC models may stem from the unique composition of the tumor immune microenvironment. For instance, in an immunosuppressive microenvironment rich in myeloid-derived suppressor cells or regulatory T cells, the upregulation of PD-L1 might dominate and impair the anti-tumor response. Conversely, in an inflammatory microenvironment with pre-existing, pre-activated T cells, Bif. bev may primarily function as a “brake,” preventing excessive immune activation, thereby synergizing with and enhancing the efficacy of checkpoint inhibitors. Furthermore, differences such as bacterial subtypes, the dosage of vesicles, and their routes of administration are also key factors contributing to the discrepant observations across different studies (116). Therefore, categorizing Bifidobacteriumor its products simply as “beneficial” or “detrimental” is an oversimplification; their immunological and microbiological context must be taken into account.

Clinical and preclinical studies have established a strong correlation between specific gut microbial taxa and enhanced responses to anti-PD1/PD-L1 therapy. For instance, responders to PD-1 blockade exhibit higher abundances of Parabacteroides distasonis and Bacteroides vulgatus, which are associated with elevated levels of the microbial metabolite valerate. Valerate reduces immunosuppressive metabolites like L-kynurenine and the kynurenine/tryptophan (Kyn/Trp) ratio, thereby suppressing Tregs and activating effector T cells (117). Similarly, FMT from responders to refractory melanoma patients reshapes gut microbial communities, enriching immunostimulatory taxa (e.g., Clostridiales and Ruminococcaceae) while reducing IL-8-expressing MDSCs. This remodeling enhances CD8+ T cell activation and TME immunogenicity, achieving clinical benefit in 40% of previously resistant patients (56). Akkermansia muciniphila and its outer membrane protein Amuc_1100 further exemplify beneficial microbiota, promoting CTL expansion in colitis-associated CRC by downregulating PD-1 and upregulating TNF-α in mesenteric lymph nodes (118). Conversely, Fusobacterium nucleatum and its metabolite succinate drive resistance in CRC by inhibiting the cGAS–IFN-β pathway, impairing CD8+ T cell recruitment. Depleting F. nucleatum via metronidazole restores PD-1 inhibitor sensitivity, highlighting its role as a predictive biomarker for resistance (119).

Resistance to PD1/PD-L1 therapy is closely linked to dysbiotic gut microbiota and their metabolites. In CRC, Fusobacterium nucleatum-derived succinate suppresses CD8+ T cell infiltration by inhibiting cGAS–STING signaling, establishing an immunosuppressive TME (119). Similarly, gut microbiota alterations in non-responders to PD-1/PD-L1 inhibitors are characterized by reduced Faecalibacterium prausnitzii abundance. Low F. prausnitzii levels exacerbate ICIs-induced colitis and weaken anti-tumor immunity, whereas its supplementation alleviates intestinal inflammation while enhancing CTL activity and tumor control (120). In biliary tract cancer, poor clinical outcomes correlate with Bacilli, Lactobacillales, and Pyrrolidine, which are enriched in non-durable clinical benefit patients. These taxa are associated with suppressed anti-tumor immunity and shorter survival, contrasting with the survival-promoting effects of Alistipes in responders (121). Moreover, dysregulated microbial metabolism contributes to resistance: low butyrate-producing microbiota in non-responders reduce histone H3K27 acetylation in CD8+ T cells, dampening PD-1 expression and T cell receptor (TCR) signaling (122).

Targeted modulation of gut microbiota emerges as a promising strategy to improve PD1/PD-L1 immunotherapy outcomes. Probiotics and postbiotics: Lactobacillus gallinarum produces indole-3-carboxylic acid (ICA), which inhibits IDO1-mediated kynurenine production and blocks aryl hydrocarbon receptor (AHR)-dependent Treg differentiation, thereby restoring CD8+ T cell cytotoxicity and sensitizing CRC to PD-1 inhibitors (123). Dietary interventions: Inulin gel enriches SCFA-producing commensals (e.g., Bifidobacterium), fostering stem-like TCF1+PD-1+CD8+ T cells in the TME and enhancing systemic IFN-γ+CD8+ T cell responses (124). Similarly, KD or supplementation with 3HB suppress PD-L1 upregulation in myeloid cells while expanding CXCR3+ T cells, synergizing with anti-PD-1 therapy (125). Pharmacological and microbial engineering: CBM588, a microbial modulator, increases Ruminococcaceae abundance and IL-10 production, improving gut immune homeostasis and overcoming PD-1 blockade resistance in murine models (126). FMT: Transferring responder microbiota to refractory patients or germ-free mice restores anti-PD-1 responsiveness by elevating beneficial taxa (e.g., Akkermansia) and SCFAs like propionate and butyrate, which promote CD8+ T cell infiltration and suppress PD-L1 (56, 127).

While the gut microbiota has been extensively studied, emerging evidence underscores the critical role of non-gut microbiota—including intratumoral, lung, skin, and oral microbes—in shaping antitumor immunity and modulating the PD-1/PD-L1 axis. These microbial communities reside directly within tumor tissues or at barrier sites, allowing them to interact with local immune cells and influence checkpoint pathways through context-specific mechanisms. For instance, in lung cancer, intratumoral fungi such as Aspergillus sydowii recruit myeloid-derived suppressor cells (MDSCs) and induce an immunosuppressive microenvironment, thereby upregulating PD-L1 expression on tumor cells and promoting resistance to anti-PD-1 therapy (128). Similarly, intratumoral bacteria in lung adenocarcinoma—including butyrate-producing species like Roseburia—can enhance metastasis by stimulating PD-L1 via metabolite-driven pathways (e.g., butyrate-induced H19/MMP15 signaling) and polarizing macrophages toward an M2 phenotype (129). Beyond the lungs, intratumoral microbes in gastric cancer, such as Fusobacterium nucleatum, recruit tumor-associated neutrophils (TANs) and activate IL-17/NF-κB pathways to foster PD-L1-dependent immune evasion (130). Even in melanoma, skin microbiota and intratumoral fungi have been linked to differential T-cell infiltration and PD-1 expression, though their mechanisms are less defined (131, 132). Importantly, these non-gut microbes often operate independently of gut-centric effects, as demonstrated by single-cell analyses showing direct microbial-host interactions in tumor tissues that alter gene expression in T cells and macrophages. The therapeutic implications are significant: for example, targeting intratumoral fungi with antifungals or engineering microbes to deliver PD-1/PD-L1 modulators could overcome local resistance. However, challenges remain, including the low biomass of non-gut microbiota and the need for spatially resolved techniques to dissect their niche-specific functions. Future studies should prioritize mapping microbial influences across all body sites to develop holistic interventions that integrate gut and non-gut microbiota for precision immunotherapy (132).

Microbiota-derived metabolites serve as both biomarkers and mediators of immunotherapy efficacy. High serum butyrate levels correlate with improved PD-1 inhibitor responses in NSCLC by enhancing PDCD1 and CD28 promoter acetylation in CD8+ T cells (122). Conversely, succinate accumulation from F. nucleatum predicts resistance in CRC, while methylglyoxal (MG), a radiosensitizing metabolite, enhances PD-1 inhibitor efficacy by inducing immunogenic cell death and activating STING-dependent antitumor immunity (119, 125). Additionally, plant-derived nanoparticles (e.g., ginger exosome-like nanoparticles) increase DHA via Lachnospiraceae modulation, which binds the PD-L1 promoter to suppress its transcription and overcome resistance (125).

The gut microbiota profoundly influences PD1/PD-L1 immunotherapy outcomes through taxon-specific interactions, metabolite signaling, and immune pathway modulation. Enriching beneficial taxa (e.g., Akkermansia, Lactobacillus gallinarum) and depleting resistance-associated species (e.g., Fusobacterium nucleatum) via dietary, pharmacological, or microbial interventions holds transformative potential for personalized cancer immunotherapy (Figure 3). We summarize more relevant studies in Table 1.

Despite a wealth of convincing preclinical studies elucidating the mechanisms by which the microbiota regulates the PD-1/PD-L1 axis, its translation into clinical application remains challenging, often yielding inconsistent results (160, 161). This “translational chasm” stems from multiple factors. First, there are fundamental differences in research models. Preclinical studies predominantly use inbred mice reared under germ-free or standardized conditions, which have low-diversity and well-defined microbiomes (160). In contrast, the human microbiome is profoundly shaped by genetics, diet, antibiotic history, and environment, resulting in exceptionally high inter-individual heterogeneity (162, 163). Furthermore, mouse tumor models are often homogeneous and progress rapidly, whereas human tumors undergo long-term evolution, exhibiting high heterogeneity and a complex immune-edited microenvironment (160, 161). Consequently, single bacterial strains or simple consortia that show significant efficacy in mouse models often struggle to colonize or exert their intended effects within the complex ecosystem of humans (163). Second, the vast inter-individual heterogeneity poses a major obstacle to clinical translation. Factors such as the patient’s baseline microbiome composition, immune status, genetic background, and concomitant medications collectively lead to highly variable responses to microbial interventions (162). Research indicates that the colonization capacity of even the same probiotic strain depends on an individual’s inherent gut microbiota structure and physiological conditions (e.g., intestinal transit time) (163). This distinction between “permissive” and “non-permissive” hosts explains why clinical trial results are often inconsistent and underscores the necessity for personalized, precision interventions in the future. Finally, limitations in study design and methodology cannot be overlooked. Many preclinical studies have small sample sizes, failing to mimic the scale and population diversity of clinical trials. In microbiome analysis, the lack of standardized protocols for sample processing, sequencing, and data analysis makes direct comparison of results across studies difficult. Current 16S rRNA sequencing based on relative abundance cannot distinguish bacterial viability or absolute quantity, potentially failing to accurately reflect changes in the functional microbial community (163). To overcome these challenges, future research should focus on: 1) Developing animal models (e.g., humanized models) that more closely recapitulate human disease states; 2) Conducting large-scale, multi-center prospective clinical studies integrated with multi-omics technologies to deeply characterize causal functional bacterial strains and their metabolites; 3) Establishing standardized protocols for microbiome analysis and management; 4) Advancing precision stratification of patients based on their baseline microbial features to enable personalized interventions. Only through in-depth interdisciplinary collaboration can we successfully bridge this translational chasm and realize genuine microbiome-based precision immunotherapy.

Breakthrough discoveries from preclinical research are being rapidly translated into real-world therapeutic strategies through a series of clinical trials, aiming to validate the feasibility of enhancing the efficacy of PD-1/PD-L1 inhibitors by modulating the microbiota. Among these strategies, fecal microbiota transplantation (FMT) is the most extensively studied intervention. In patients with advanced melanoma, several early-phase clinical trials have demonstrated its potential. For PD-1 inhibitor-refractory patients who did not respond initially, infusion of fecal microbiota from responders (e.g., in trial NCT03341143↗) could reshape the gut microbiota and reverse the immunosuppressive tumor microenvironment in a subset of patients (approximately 40%), thereby overcoming drug resistance (56). More prospective explorations have combined FMT from healthy donors with PD-1 inhibitors as first-line therapy (NCT03772899↗), showing that this regimen has a favorable safety profile and achieved an objective response rate as high as 65%, suggesting FMT’s potential to prevent or delay the onset of resistance (164). Furthermore, small-scale studies in patients with anti-PD-1 refractory metastatic melanoma also confirmed the safety of FMT combination therapy and observed favorable changes in immune cell infiltration within the tumor microenvironment (165).

Beyond directly transplanting the entire microbial community, more precise microbial modulation strategies are under development. For instance, a multicenter, randomized, placebo-controlled Phase I trial (NCT03817125↗) evaluated the combination of SER-401, a spore-based formulation rich in Firmicutes, with nivolumab. Although this study was underpowered due to enrollment challenges during the pandemic, its translational analysis provided key insights: an improperly designed antibiotic pretreatment regimen intended to empty ecological niches might disrupt the microbiota and impair immunity, consequently weakening the subsequent efficacy of immunotherapy (166). Addressing the clinical challenge of antibiotic-induced dysbiosis, an innovative study tested the colon-targeted adsorbent DAV132 (NCT03678493↗). This device effectively reduced fecal antibiotic levels without affecting plasma antibiotic concentrations, thereby protecting gut microbiota diversity in healthy volunteers. Preclinical models further confirmed that fecal transplants from DAV132-protected volunteers could prevent antibiotic-associated resistance to anti-PD-1 therapy (167).

Notably, the influence of the microbiota may extend beyond traditionally “hot” tumors. In the ALICE trial for triple-negative breast cancer, higher baseline gut microbiota alpha diversity (Faith’s PD) was significantly associated with improved efficacy of atezolizumab (anti-PD-L1) combined with chemotherapy, suggesting that microbiome features could serve as predictive biomarkers, even in PD-L1-negative patients (146).

However, translating these preliminary successes into universal clinical protocols faces significant challenges. Existing trials are mostly early-phase, single-arm studies with small sample sizes and heterogeneous outcomes. Donor screening for FMT, standardization of preparations, long-term safety, and the substantial inter-individual heterogeneity of microbiota are pressing issues that need resolution (166, 168–171). Future directions require larger-scale randomized controlled trials, biomarker-based precise patient stratification, and the development of standardized synthetic microbial consortia composed of well-defined functional strains, ultimately enabling the precise application of microbiota modulation in cancer immunotherapy.