Metabolome-microbiome convergence under circadian disruption accelerates mammary tumorigenesis: multi-omics integration insights

Abstract

Circadian rhythm disruption (CRD), common in shift work and jet lag, promotes mammary tumorigenesis through coordinated reprogramming of tumor metabolism and the local microbiome. CRD increased immunosuppressive metabolites - kynurenic acid, spermidine, and argininosuccinic acid - that suppressed effector T-cell activity and drove macrophage polarization toward anti-inflammatory phenotypes. 16S rRNA sequencing revealed enrichment of immune-modulatory Firmicutes and Bacilli within CRD tumors. Experimental and multi-omics integrative analyses strongly support metabolome-microbiome crosstalk driving immune suppression under CRD. Inhibition of arginase-1 (ARG1) with nor-NOHA restored cytotoxic T-cell responses and reduced lung metastases. These findings establish metabolome-microbiome interactions as a central mechanism of CRD-induced immunosuppression and metastasis, revealing potential therapeutic targets for circadian disruption-associated breast cancer.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12885-026-16546-6.

Introduction

Circadian rhythms regulate fundamental biological processes, including metabolism, immunity, and microbial composition [1]. Although much research has focused on the gut microbiota and its interaction with host circadian rhythms, the microbiome within tumor microenvironments (TME) remains an unexplored but crucial factor in cancer initiation and progression [2, 3]. Accumulating evidence indicates that the intratumoral microbiome exerts metabolic and immunomodulatory effects that can promote tumorigenesis and cancer progression, primarily through modulation of oncogenic signaling pathways, inflammatory processes, and interactions with the tumor microenvironment [4, 5].

CRD, which is caused by factors such as shift work, jet lag, or experimental disruption of the light-dark (LD) cycle, has been linked to increased cancer risk and tumor progression [6]. Beyond these systemic effects, CRD may also disrupt gut microbiome rhythms, leading to dysbiosis with altered microbial composition and metabolite availability [7]. Such disturbances can perturb immune-metabolic balance, and accumulating evidence shows that immunometabolic dysfunction and oxidative stress promote tumor development [8]. Tumors may exploit CRD-induced rewiring of polyamine biosynthesis as an adaptive metabolic strategy [9].

In this connection, epidemiological [10, 11] and experimental [12, 13] studies have established a strong link between circadian rhythm disruption (CRD) and elevated breast cancer risk, with long-term night shift work significantly increasing incidence rates [14]. Animal models corroborate these findings, showing that chronic CRD accelerates mammary tumor development [14]. Altogether, previous reports state that the circadian clock tightly coordinates metabolic pathways at the transcriptomic and metabolomic levels and that disruption of circadian genes leads to profound metabolic disturbances [15].

Despite this clear association, the specific metabolic and microbial pathways through which CRD promotes tumorigenesis remain largely unknown. Building on this framework, to bridge the knowledge gap, we aim to investigate the crosstalk between metabolome and microbiome in tumors subjected to CRD or a regular LD cycle (12:12). The results suggest that CRD promotes an immunosuppressive mammary TME. Building on our previous finding [13] of arginase-1 (ARG1) upregulation alongside immune inhibitor receptor LILRB4a under CRD conditions, in this study, we used a mouse model of breast cancer to explore the role of ARG1 in sustaining immune suppression within the TME. Targeted modulation of ARG1 alleviated the immunosuppressive state, restoring aspects of anti-tumor immunity. These findings emphasize the critical interaction between microbial metabolites and host immune regulation in CRD-driven tumor progression. By focusing on the microbiome of the TME rather than the gut microbiota, we aim to provide novel insights into the relationship among CRD, tumor metabolism, and local microbial influences. Our findings highlight new avenues for therapeutic intervention in CRD-driven breast malignancies.

Materials and methods

Animal husbandry

Nulliparous BALB/cJ female mice (aged 5 weeks) were purchased from Jackson Laboratory. The mice were housed under 12 h light and 12 h dark (LD 12:12) at room temperature (25 °C), and food and water were provided ad libitum. BALB/cJ mice (Jax stock #000651) were used to create a syngeneic mouse model by injecting 4T1 cells into these mice. All animal care and treatments were per the Institutional Animal Care and Use Committee of Texas A&M University under protocol # 2024-0095.

CRD conditions

Mice (9–10 for each group) were randomly assigned to either standard light-dark (LD) conditions (12:12 LD cycle) or circadian rhythm disruption (CRD) or jet lag (JL) conditions, which involved an 8-hour advance in the light phase every two days, as previously described [13]. All animals had ad libitum access to standard chow and water. Zeitgeber time (ZT) 0 was defined as light onset, while ZT12 corresponded to dark onset. For activity monitoring, both LD and CRD groups were provided with running wheels (Columbus Instruments) for voluntary exercise, with wheel rotation continuously recorded over several weeks. ActogramJ (ImageJ) was used to analyze voluntary running activity patterns [13].

4T1 cell mouse TNBC model

The mouse TNBC model was prepared with minor modifications to the standard protocol [16]. 4T1 cell lines were obtained from the American Type Culture Collection (ATCC, CRL-2539™) and cultured and maintained in RPMI-1640 (ATCC, 30-2001) with high glucose, L-Glutamine, and HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) buffer system, under 5% CO2 atmosphere at 37 °C. The cell culture medium was supplemented with 10% fetal bovine serum (FBS) (Alkali Scientific Inc, FB72) and 1% penicillin and streptomycin (ATCC, 30-2300). In each mouse, 10,000 cells suspended in sterile PBS were injected subcutaneously into the 4th mammary gland fat pad. After the development of palpable tumors, these mice were housed in LD or CRD conditions for 3–4 weeks before they were sacrificed to harvest the tumors.

Tumor burden

The tumors and the mammary glands were harvested using aseptic techniques. Tumor volumes were measured for each breast tumor using a caliper after every 3 days, throughout the experiment. Tumor volume =(length ×width^2 )/2 was used to measure the volume of the tumors, whereas the equation Tumor burden (%) =(Tumor weight /mice body weight)×100% was considered to determine the tumor burden [13].

Image-based metabolomics

Frozen tissue samples were carefully sectioned at a thickness of 12 μm using a Thermo CryoStar NX50 cryostat and placed onto indium tin oxide (ITO)-coated conductive microscope slides. To support both histological evaluation and molecular imaging, serial sections were collected, one designated for mass spectrometry imaging (MSI) and the other for hematoxylin and eosin (H&E) staining. Before the MSI acquisition, fiducial markers were added to the corners of each slide to assist with accurate alignment between optical and MSI data. High-resolution optical images were then captured at 4800 dpi using an Epson Perfection V600 flatbed scanner. For matrix application, the tissue sections were coated with a solution of 1,5-diaminonaphthalene (DAN) at 10 mg/mL in 50% acetonitrile, using an HTX M5 robotic sprayer. The matrix was applied over four uniform passes using the following parameters: a flow rate of 0.10 mL/min, track speed of 1200 mm/min, track spacing of 3 mm, nozzle height of 40 mm, nozzle temperature of 60 °C, nitrogen pressure of 10 psi, and a crisscross spray pattern. These settings were selected to ensure fine, consistent coating with minimal analyte delocalization, optimized specifically for metabolite imaging in negative ion mode. MSI data were acquired using a Bruker timsTOF fleX QTOF mass spectrometer in negative ion mode at a spatial resolution of 40 μm, with 300 laser shots accumulated per pixel. Acquisition parameters included an m/z range of 50-1000, Funnel 1 RF of 75.0 Vpp, Funnel 2 RF of 100.0 Vpp, Multipole RF of 150.0 Vpp, Collision Energy of 10.0 eV, Collision RF of 500.0 Vpp, Transfer Time of 35.0 µs, and Pre-Pulse Storage of 2.0 µs. The acquired ion images were visualized and analyzed using SCiLS Lab Pro 2024b software (Bruker). All data were root mean square (RMS) normalized, after which peaks of interest were manually selected from the averaged spectra. Putative metabolite identifications were generated using the MetaboScape 2024b (Bruker) plugin integrated with SCiLS, based on high mass accuracy within a tolerance of ± 10 ppm.

Metabolomics

For analysis of the metabolomics, equal fractions of breast tumor samples from mice under 12:12 light-dark (LD) or circadian rhythm disruption (CRD) conditions were collected and flash-frozen in liquid nitrogen. Tissues were homogenized in 80% methanol with internal standards and sonicated at 4 °C. After centrifugation, the supernatant was collected for derivatization. Sample analysis was carried out by MS-Omics as follows. The analysis was carried out using a Thermo Scientific Vanquish LC coupled to a Q Exactive™ HF Hybrid Quadrupole-Orbitrap, Thermo Fisher Scientific. An electrospray ionization interface was used as an ionization source. Analysis was performed in positive and negative ionization modes. The UPLC was performed using a slightly modified version of the protocol described by Doneanu et al. (UPLC/MS Monitoring of Water-Soluble Vitamin Bs in Cell Culture Media in Minutes, Water Application note 2011, 720004042en). Peak areas were extracted using Compound Discoverer 3.3 (Thermo Scientific). Identification of compounds was performed at four levels; Level 1: identification by retention times (compared against in-house authentic standards), accurate mass (with an accepted deviation of 3ppm), and MS/MS spectra, Level 2a: identification by retention times (compared against in-house authentic standards), accurate mass (with an accepted deviation of 3ppm). Level 2b: identification by accurate mass (with an accepted deviation of 3ppm), and MS/MS spectra, Level 3: identification by accurate mass alone (with an accepted deviation of 3ppm).

High-throughput 16 S rRNA gene amplicon sequencing and data analysis

Tumor tissues were collected and snap-frozen until microbiome analysis. Microbial DNA was extracted using the ZymoBIOMICS™ DNA Miniprep Kit (Zymo Research, USA), and quality was assessed via Qubit™ fluorometer (Thermo Fisher, USA). The full-length 16 S rRNA gene (27 F/1492R primers) was amplified using Kapa HiFi HotStart DNA polymerase (Roche, Switzerland), purified with Select-a-Size DNA Clean & Concentrator™ (Zymo Research, USA), and sequenced on the PacBio Sequel IIe System with HiFi reads in Circular Consensus Sequencing mode. Raw reads were processed with DADA2 [17] for error correction and ASV inference, with taxonomic classification assigned using Uclust (QIIME v1.9.1) and the Zymo Research curated 16 S database. Microbial composition was analyzed using QIIME v1.9.1, including α-diversity metrics (Shannon, InvSimpson, Ace, Chao1) and β-diversity via Principal Coordinates Analysis (PCoA), with ANOSIM used for statistical assessment of group differences [18]. Significant taxa were identified using LEfSe where applicable [19]. Quality controls included the ZymoBIOMICS^® Microbial Community Standard for DNA extraction, the ZymoBIOMICS^® Microbial Community DNA Standard for library preparation, and negative controls to assess background contamination.

Bacteria abundance and survival analysis in human breast tumors

The Bacteria in Cancer (BIC) database developed by Chen et al. (2022) [20] is used to evaluate the expression of intratumoral bacteria in human breast cancer. The relative abundance values of Firmicutes and Bacilli, as provided by the database, were extracted and compared across samples. The clinical relevance of these bacterial species was used to evaluate survivability in patients using the database’s predefined cutoffs and statistical pipelines. The results were downloaded from the software for visualization and interpretation. Finally, the bacteria-associated biological function was identified by fast gene set enrichment analysis (FGSEA) using the association between mRNA expression and bacterial abundance as gene rank.

Human breast tissue microbiome validation

To evaluate whether microbial alterations observed in the murine model were also detectable in human breast tissues, a publicly available breast tissue microbiome dataset deposited in Dryad was reanalyzed. Samples were categorized as healthy (H), premalignant (PD), adjacent normal (AN), or tumor (T) according to the accompanying metadata [21].

Taxonomic abundance tables were imported into R and aggregated at the phylum, class, and genus levels. Differences in Firmicutes and Bacilli abundance across disease groups were assessed using the Kruskal-Wallis test. For genus-level analyses, Bacilli-associated genera were converted to presence/absence data and prevalence was compared between healthy and tumor samples using Fisher’s exact test. Odds ratios and 95% confidence intervals were calculated for each genus. Multiple-testing correction was performed using the Benjamini-Hochberg false discovery rate (FDR) method. Statistical significance was defined as FDR < 0.05.

Cytokine array

This was performed using the Proteome Profiler Mouse XL Cytokine Array kit (R&D Systems, ARY028). The blood samples were kept undisturbed for 30 min at room temperature before being spun at 2000 x g for 10 min at 4 °C. The serum was separated for the cytokine array profiling according to the manufacturer’s protocol and was scanned using ChemiDoc Imaging System (BioRad, 12003153).

Flow cytometry

Dissociated tumor cells were counted and resuspended in 1 mL of cell staining buffer (CSB) (BioLegend, 420201). Zombie NIR™ Fixable Viability Kit (BioLegend, 423105) was applied on ice for 20 min to assess cell viability. After centrifugation at 400 × g for 4 min, cells were incubated with fluorochrome-conjugated antibodies, including Alexa Fluor^® 700 anti-mouse CD4, Pacific Blue™ anti-mouse CD8a, Alexa Fluor^® 488 anti-mouse CD86, PE anti-mouse CD163, Alexa Fluor^® 594 anti-mouse CD45, PE anti-mouse/human CD11b, and FITC anti-mouse Ly-6 C/Ly-6G. Cells were then fixed with 2% paraformaldehyde and permeabilized with 0.1% Triton X-100 in PBS, followed by intracellular staining with Alexa Fluor^® 488 anti-mouse/rat/human FOXP3 for 20 min at 4 °C in the dark. After staining, cells were rinsed with FACS buffer (1% FBS in PBS) and stored in FACS buffer until analysis on a Cytek Aurora spectral flow cytometer.

MIMOSA 2 analysis

MIMOSA2 [22] is a metabolic model-based framework for relating variation in microbiome composition to paired metabolite measurements [https://doi.org/10.1093/bioinformatics/btac003]. Briefly, MIMOSA2 uses the KEGG database [https://doi.org/10.1093/nar/27.1.29] to predict the metabolic potential of a community of microbes, linking genomes to genes to metabolic reactions. It then compares the community-wide metabolic potential to actual metabolite levels observed, and infers per metabolite whether it’s variation can be well-explained by changes in bacterial composition. Here, metabolite and microbial profiles of tumor samples (LD and CRD-induced) were inputted to the MIMOSA2 web app (http://elbo-spice.cs.tau.ac.il/shiny/MIMOSA2shiny/), with settings: 16 S rRNA sequence variants (ASVs) are mapped to KEGG using the pre-computed genome inferences from PICRUSt model. Metabolite names were mapped to their respective KEGG compound ID’s using MetaboAnalyst [https://doi.org/10.1093/nar/gkae253]. However, since MIMOSA2 relies on inferred metabolic functions and statistical modeling, the resulting associations should be interpreted as model-supported relationships rather than direct evidence of microbial metabolic activity or causal microbial regulation of metabolite production.

Statistical analysis

All statistical analyses were performed using GraphPad Prism (v9.0) and R (v4.2.3). Data normality was assessed using the Shapiro-Wilk test, and variance homogeneity was tested with Levene’s test. For comparisons between two groups, unpaired two-tailed Student’s t-tests were used for normally distributed data, while the Mann-Whitney U test was applied for non-normally distributed datasets. For multiple-group comparisons, one-way or two-way ANOVA was performed with Tukey’s post hoc test for pairwise comparisons. The Kruskal-Wallis test and, when necessary, Dunn’s multiple-comparison test was used to evaluate differences in microbiome abundance analyses involving more than two groups. Fisher’s exact test was used to compare detection frequencies between groups for prevalence-based analyses, and the Benjamini-Hochberg false discovery rate (FDR) correction was employed to adjust p-values for multiple testing.

Results

CRD accelerates aggressive tumorigenesis via metabolic rewiring

To investigate the impact of CRD on the TME, we used a murine model maintained under either standard light-dark (LD) or CRD conditions (Fig. 1A). Consistent with our recently published study [13], CRD profoundly altered tumor behavior, leading to increased tumor burden and enhanced metastasis (Fig. 1B-D). Untargeted metabolomic profiling revealed broad reprogramming of tumor metabolism under CRD. Notably, kynurenic acid (KA) was markedly upregulated in CRD tumors compared with LD controls (Fig. 1E). Lactate levels were also elevated under CRD (Fig. 1E). CRD tumors exhibited reduced D-galactose and glucosamine (Fig. 1E).

Fig. 1.

CRD reprograms tumor metabolism. A Schematic of the experimental design: BALB/cJ mice were injected with 4T1 cells and housed under either a standard LD 12:12 cycle or circadian rhythm disruption (CRD). B Tumor burden (%) was significantly higher in CRD mice compared to LD controls (n = 9–10). C Quantification of metastatic foci per lung revealed a markedly increased number of metastases in CRD mice (n = 5). D Representative histology of lung sections showing metastatic nodules. Scale bar = 350 μM. E Elevated KA, adenosine, XA, lactate, and reduced sugars. Scale bar = 3 mm. Data presented as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 represents the significance level from an unpaired t-test

To assess global metabolic remodeling, we employed pathway enrichment and correlation-based network analyses via MetaboAnalyst and KEGG. Several pathways were significantly enriched under CRD, including histidine metabolism, arginine biosynthesis, and nicotinate/nicotinamide metabolism (Fig. 2A, B). Principal component analysis confirmed distinct separation between LD and CRD tumors (Fig. 2C), and heatmap clustering of metabolites further demonstrated discrete profiles (Fig. 2D). These analyses support that CRD drives systemic metabolic rewiring that promotes both tumor growth and immune escape. In addition to these immunoregulatory metabolites, spermidine, N8-acetyl-spermidine, and argininosuccinic acid were also significantly upregulated in CRD tumors (Fig. 2E-G).

Fig. 2.

CRD drives polyamine metabolism. A Network diagram of enrichment analysis demonstrates significant perturbations in metabolic pathways. B Dot plot summarizing pathway enrichment, highlighting histidine metabolism, β-alanine metabolism, arginine biosynthesis, and nicotinate/nicotinamide metabolism. C Principal component analysis (PCA) showing separation between LD and CRD metabolomes (n = 6). D Heatmaps showing upregulation of spermidine and argininosuccinic acid. E–G Box plots of selected metabolites, showing increased spermidine, argininosuccinic acid, and N8-acetylspermidine in CRD tumors compared with LD controls, implicating dysregulated polyamine and arginine metabolism in shaping an immunosuppressive tumor microenvironment (n = 6), * P < 0.05, ** P < 0.01

Collectively, these findings indicate that CRD reshapes tumor metabolism to create an immunosuppressive landscape. Elevated metabolites such as KA, XA, adenosine, lactate, spermidine, and argininosuccinic acid converge through distinct but complementary pathways to inhibit anti-tumor immunity (Supplementary Table 1).

Spermidine orchestrates CRD-induced immunosuppressive tumor microenvironment

Spermidine, a polyamine critical for cell growth and autophagy, exerts potent immunomodulatory effects by dampening T-cell responses and inhibiting pro-inflammatory pathways (Fig. 3A) [23]. Our earlier study [13] and current study demonstrated that CRD enhances LILRB4a signaling, which in turn promotes ARG1 expression (Fig. 3B, C). This convergence of polyamine accumulation and ARG1 upregulation underscores how CRD-induced metabolic alterations promote immune suppression.

Fig. 3.

Effect of arginase-1 inhibition by nor-NOHA in CRD-induced tumors. A Diagrammatic representation of the molecular pathway of polyamine synthesis driven by arginase-1. Spermidine synthesis is influenced by the urea cycle, where Argininosuccinate is broken down to Arginine, which is hydrolyzed to Ornithine via arginase-1. Ornithine is diverted to the polyamine pathway by Ornithine Decarboxylase 1 (ODC1) to Putrescine, which is converted to spermidine. The link between the urea cycle and spermidine synthesis is implicated in tumor immunosuppression by activating Treg cells and MDSCs and by promoting M1-to-M2 macrophage polarization. B The transcript level of Lilrb4 in LD and CRD tumors was analyzed using real-time PCR. (n = 4). C The transcript level of Arg1 in LD and CRD-induced mammary tumors was analyzed using real-time PCR. (n = 3). There is a reduction in the number of metastasis foci in the lungs of CRD mice treated with nor-NOHA compared to CRD (control) (n = 5) (D). Nor-NOHA-treated CRD mice resulted in upregulation of CD8 positive cells (E), and reduction in regulatory T cells (F) and MDSCs (G). The pro-inflammatory macrophage (M1) is increased in nor-NOHA-treated CRD tumors, while the anti-inflammatory macrophage (M2) is decreased compared to control CRD tumors (H). M1/M2 ratio shown in (I) is high in CRD tumors treated with nor-NOHA, which explains a shift in the TME to an anti-tumor state (n = 4, E-I). Arg1 transcript level showed a significant reduction in CRD tumors following nor-NOHA treatment (n = 3) (J). Data presented as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 represents the significance level from an unpaired t-test

Given the central role of ARG1 in depleting arginine and maintaining immune suppression, we next tested whether ARG1 inhibition could restore anti-tumor immunity. Tumor-bearing mice were treated with nor-NOHA, a specific ARG1 inhibitor [24], under both LD and CRD conditions. Nor-NOHA treatment significantly reduced metastatic foci in the lungs of CRD mice (Fig. 3D, Supp Fig. 1A) and partially restored immune balance. Flow cytometry (gating strategy, Supp Fig. 2) revealed increased infiltration of cytotoxic CD8⁺ T cells and decreased frequencies of regulatory T cells, myeloid-derived suppressor cells, and M2 macrophages (Fig. 3E-I, Supp Fig. 1B-F). Interestingly, nor-NOHA inhibited the expression of Arg1 in CRD-induced tumor but not LD (Fig. 3J, Supp Fig. 1G) but did not significantly alter primary tumor volume (Supp Fig. 2H). This suggests that ARG1 inhibition primarily modulates the metastatic niche and immune function.

CRD alters tumor microbiome composition

We next investigated whether CRD-associated immunometabolic alterations were linked to intratumoral microbiome changes. Using 16 S rRNA sequencing, we compared tumor microbiomes from LD and CRD mice. Distinct differences were observed at the phylum level (Fig. 4A), with CRD tumors showing a higher relative abundance of Firmicutes and Bacteroidetes compared with LD controls. Firmicutes’ abundance was significantly elevated (Fig. 4B). At the class level, CRD tumors exhibited enrichment of Bacteroidia, Sphingobacteria, and Bacilli (Fig. 4C). Among these, Bacilli were significantly increased under CRD conditions (Fig. 4D).

Fig. 4.

CRD reshapes tumor microbiome. A Phylum-level microbiome profiles of tumor tissues from LD 12:12 and CRD mice, illustrating overall compositional differences across Bacteroidetes, Firmicutes, and Proteobacteria. B Firmicutes were significantly more abundant in CRD tumors compared with LD 12:12 tumors, indicating a phylum-level microbial shift associated with circadian disruption (n = 5). C Within the Firmicutes phylum, the class Bacilli was markedly enriched in CRD tumors relative to LD counterparts, highlighting a CRD-associated shift toward taxa with known immunomodulatory properties. D Quantitative comparison confirmed that Bacilli abundance was significantly elevated under CRD conditions (n = 5). Box-and-whisker plots show the abundance of Firmicutes (E) and Bacilli (F) in healthy breast tissue (H, n = 49), premalignant disease (PD, n = 15), adjacent normal tissue from breast cancer patients (AN, n = 39), and tumor tissue (T, n = 32). Differences were assessed using the Kruskal–Wallis test (P = 0.4666 for Firmicutes and P = 0.9726 for Bacilli). G Prevalence of the four significantly enriched Bacilli-associated genera in healthy breast tissue (H, n = 49) and tumor tissue (T, n = 32). Prevalence was calculated as the percentage of samples in which a genus was detected. Tumor tissues exhibited markedly higher prevalence of Facklamia (31.3% vs. 2.0%), Enterococcus (28.1% vs. 2.0%), Aerosphaera (43.8% vs. 8.2%), and Atopostipes (37.5% vs. 8.2%) compared with healthy controls. H Log2-transformed odds ratios derived from Fisher’s exact test demonstrate a significantly increased likelihood of detecting these genera in tumor tissues. Facklamia (OR = 21.03, P = 0.00027), Enterococcus (OR = 18.15, P = 0.00077), Aerosphaera (OR = 8.49, P = 0.00027), and Atopostipes (OR = 6.58, P = 0.00173) remained significant following Benjamini–Hochberg false-discovery-rate correction. I Serum IL-1β levels were significantly elevated in CRD mice compared with LD controls, consistent with Firmicutes- and Bacilli-driven induction of pro-inflammatory cytokines that contribute to immunosuppression within the tumor microenvironment (n = 2 independent preparations for each condition). Data presented as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 represents the significance level from an unpaired t-test

To explore whether microbial taxa identified in the murine CRD model may also be detectable in human breast cancer datasets, we first interrogated the Bacteria in Cancer (BIC) database [20]. Firmicutes were significantly more abundant in tumor tissues than in adjacent normal tissues (Supp Fig. 3A), and high Firmicutes abundance was associated with reduced overall survival (Supp Fig. 3C). These trends hold for Bacilli as well (Supp Fig. 3B and D). Given the observational nature of the dataset and the potential influence of cohort-specific and technical confounding factors, the BIC analysis was considered exploratory and was subsequently complemented by independent validation using a publicly available Dryad breast tissue microbiome dataset. Firmicutes and Bacilli abundance in this independent cohort did not change substantially between disease groups (Kruskal-Wallis test, P = 0.467 and P = 0.973, respectively) (Fig. 4E, F). However, genus-level prevalence analysis showed four Bacilli-associated genera (Aerosphaera, Facklamia, Enterococcus, and Atopostipes) that remained significantly enriched in tumor tissues after Benjamini-Hochberg correction (FDR < 0.05). Among these taxa, Enterococcus had a significant connection with tumor tissues (28.1% prevalence in tumors versus 2.0% in healthy tissues; Fisher’s exact test, P = 0.00077; odds ratio = 18.15) (Fig. 4G, H, Supp. Figure 3E, Supp. Table 2). Pathway analysis using KEGG, GO, and Reactome demonstrated that Firmicutes and Bacilli were present in pathways related to cell cycling, DNA repair, and DNA replication (Supp Fig. 4). Firmicutes exert immunomodulatory effects through small glycoconjugates that can enhance cytokine production, including IL-34 and IL-1β. Consistent with this, CRD mice displayed elevated serum IL-1β (Fig. 4I).

Crosstalk between the microbiome and metabolites in CRD-induced tumors

Finally, we applied MIMOSA2 [22], a metabolic model-based framework (Fig. 5A), to evaluate how changes in microbial communities relate to metabolic alterations within CRD-induced mammary tumors. This approach allowed us to integrate microbial abundance with estimated metabolic potential and compare the results to experimentally measured metabolite levels. MIMOSA2 analysis revealed significant associations between CRD-associated microbial community composition and several metabolites involved in arginine and polyamine metabolism. Notably, MIMOSA2 identified citrulline (Fig. 5B, C) and kynurenine (Fig. 5B) as metabolites whose variation was mostly explained by microbial composition variations. These findings indicate a potential link between microbiome remodeling and metabolic alterations within the tumor microenvironment, although causal relationships cannot be established from the present analysis. However, the MIMOSA2 findings support a functional microbiome-metabolite axis in CRD tumors, where bacterial taxa may contribute to the generation or regulation of immunomodulatory metabolites that alter the tumor microenvironment.

Fig. 5.

MIMOSA demonstrates a crosstalk between microbiome and metabolome profile. A Summary of the MIMOSA2 analysis pipeline. B Summary of MIMOSA2 results. For each metabolite (columns), upper bars show total variance explained by the model, and the heatmap below shows the fraction of variation explained by each specific taxon. C L-Citrulline is the metabolite best explained by microbial composition (p = 0.006), as identified by MIMOSA2. Scatter plots show the relationship between the microbial community-level metabolic potential scores (computed by MIMOSA2, x-axis) and actual metabolite measurements (y-axis). The model fit is drawn as a dashed line.”

Discussion

CRD has far-reaching consequences for metabolism, immunology, and microbial balance, all of which can have a significant impact on cancer progression. Our study reveals that CRD enhances mammary tumor growth by altering both the metabolic landscape of tumors and the microorganisms that inhabit them, resulting in a tumor microenvironment that inhibits immune protection. Our combined findings demonstrated that CRD promotes the accumulation of various immunosuppressive metabolites, including KA, spermidine, adenosine, and argininosuccinic acid, while also enriching bacterial groups such as Firmicutes and Bacilli, which are known to influence host immunity. We also identified arginase-1 (ARG1) as a critical metabolic circuit that links these alterations to decreased cytotoxic T-cell function. Importantly, inhibiting ARG1 partially restored anti-tumor immune responses and decreased metastatic dissemination. These findings show how CRD alters the metabolic-microbial balance within tumors and suggest new treatment approaches for minimizing the negative effects of circadian disturbance in breast cancer.

The increase in KA is consistent with other studies that report how KA stimulates immune evasion by reducing T-cell responses, hence fostering tumor development [25]. These findings highlight KA as a key immunosuppressive metabolite in the CRD-altered mammary TME, underscoring its potential as a biomarker in CRD-driven mammary tumorigenesis. Other immunoregulatory metabolites were also elevated in CRD tumors. Adenosine, acting through A2A and A2B receptors, was significantly increased, consistent with its established role in dampening cytotoxic T-cell activity, promoting regulatory T cells, and driving M2 macrophage polarization [26]. Consistent with this immunosuppressive profile, xanthurenic acid (XA) acts via aryl hydrocarbon receptor (AHR) to suppress dendritic cell maturation and T-cell priming [25]. Lactate levels were also found to be elevated in CRD tumors, and they are known to promote immune suppression by skewing macrophages toward an M2 phenotype [27]. CRD tumors exhibited reduced D-galactose and glucosamine, which are essential for glycosylation, antigen presentation, and T-cell receptor stability. These findings suggest that CRD shifts the balance toward immunosuppressive metabolites while depleting those critical for effective immune surveillance [28].

Spermidine, which has been reported to promote M2 macrophage polarization through mitochondrial ROS-mediated activation of AMPK and HIF-1α signaling [29], was elevated in CRD tumors. Subsequently, elevated argininosuccinic acid (L-arginine precursor) suggests the dysregulation of arginine metabolism, a key axis in immune suppression. L-arginine is an essential amino acid for T cell proliferation and function; in the TME, L-arginine is targeted for breakdown by ARG1 (Fig. 3A) [23].

Our study (Fig. 4) demonstrates that Bacteroidetes, Firmicutes, and Bacilli levels shift considerably between LD and CRD-induced tumor tissues. Previous studies have shown that Bacteroidetes can synthesize spermidine through alternative pathways, while many Firmicutes, although unable to produce polyamines, maintain active polyamine transport systems (e.g., potABCD) that allow them to scavenge spermidine and putrescine [30]. Within Bacilli, Bacillus subtilis encodes a speD-mediated S-adenosylmethionine decarboxylase, enabling conversion of putrescine to spermidine. These findings suggest that CRD-induced microbial shifts favor taxa capable of producing or accumulating polyamines, directly linking microbiome alterations to elevated spermidine levels in CRD tumors [30].

The BIC and DRYAD datasets were generated independently of the present study, and differ in patient cohorts, tissue collection procedures, sequencing methodologies, DNA extraction protocols, and bioinformatic analysis pipelines which could impact in cumulative way on the detection and estimation of abundance of microbes, across the studies. In addition, the used datasets are both observational and cross-sectional in nature, which does not allow a longitudinal study to establish any causality between specific microbial taxa/genera and breast cancer. Therefore, the human microbiome analyses presented here are exploratory and hypothesis-generating evidence rather than definitive proof of microbial involvement in CRD-induced breast cancer progression. Importantly, the primary objective of the study was to investigate how CRD-induced change in the TME is linked with the abundance of specific microbiome. The human dataset studies were carried out to assess if the microbial fingerprints identified in the mouse model could be recognized in independent clinical cohorts. In spite of the limitations of the re-analysis of the Dryad dataset, the observed enrichment of specific Bacilli-associated genera lends translational support to our experimental findings. Hence, the BIC analysis was considered exploratory and was subsequently complemented by independent validation using a publicly available Dryad breast tissue microbiome dataset.

These results highlight how circadian rhythm disruption drives tumor progression through coordinated metabolic and microbial changes. Elevated immunosuppressive metabolites converge with microbial enrichment to establish an immunosuppressive TME. This synergy promotes immune evasion, metastasis, and poor prognosis. Importantly, ARG1 inhibition partially reversed these effects, demonstrating the therapeutic potential of targeting metabolic–microbial crosstalk in CRD-associated breast cancer. Future work should dissect the causal pathways linking microbial metabolites to host immune suppression, paving the way for novel interventions in circadian disruption-driven malignancies (Fig. 6).

Fig. 6.

CRD promotes crosstalk between the microbiome and the metabolome, driving tumor progression and immune suppression. Schematic representation of how a disrupted circadian clock enhances tumor initiation and growth through microbiome-metabolome interactions. Circadian disruption induces microbial dysbiosis characterized by high abundance of Firmicutes and Bacilli, and alterations in the host metabolome. Bidirectional crosstalk between the dysregulated microbiota and metabolites results in elevated levels of kynurenic acid, xanthurenic acid, adenosine, lactic acid, and spermidine. These changes perturb key metabolic pathways, including tryptophan, purine, glucose, polyamine, and urea metabolism. Collectively, these metabolic alterations reshape the tumor immune microenvironment, promoting a shift from an immune-inflamed “hot” tumor (enriched in M1 macrophages, dendritic cells, and CD8⁺ T cells) toward an immune-suppressed “cold” tumor phenotype dominated by M2 macrophages and FOXP3⁺ regulatory T cells, thereby facilitating tumor progression

Conclusion

This study demonstrates that CRD reshapes the mammary tumor microenvironment through coordinated metabolic and microbial reprogramming that favors immune suppression and metastatic progression. CRD significantly elevated immunosuppressive metabolites, including kynurenic acid, adenosine, xanthurenic acid, lactate, spermidine, and argininosuccinic acid, each acting through distinct pathways to dampen anti-tumor immunity. These converging pathways establish a robust immunosuppressive metabolic axis under CRD. CRD also altered the intratumoral microbiome, enriching Firmicutes, Bacteroidetes, and Bacilli, taxa capable of producing or accumulating polyamines such as spermidine. Human tumor analyses validated their association with poor survival and immune-modulatory pathways. The elevation of IL-1β in CRD mice provides a mechanistic link, as Firmicutes-derived factors can stimulate systemic inflammatory signaling and enhance MDSC recruitment. Importantly, pharmacologic inhibition of ARG1 with nor-NOHA restored CD8⁺ T-cell infiltration and reduced lung metastases, highlighting ARG1 as a key mediator of CRD-driven immune escape. In conclusion, CRD promotes tumor aggressiveness through integrated metabolic-microbial dysregulation and targeting this immunometabolic network represents a promising therapeutic strategy for circadian disruption-associated breast cancer.

Biographies

Mrinmoy Sarkar

Mrinmoy Sarkar is a cancer researcher whose work focuses on elucidating the molecular and immunological mechanisms underlying tumor progression through integrated experimental and systems biology approaches. His research encompasses cancer metabolism, tumor microenvironment, microbiome-host interactions, multi-omics integration, and biomarker discovery to advance the understanding of cancer biology. He worked as Postdoctoral Research Associate in the Roy Sarkar Lab.

Olajumoke Ogunlusi

Olajumoke Ogunlusi is a Ph.D. candidate in the Department of Biology at Texas A&M University, working under the supervision of Prof. Tapasree Roy Sarkar. Her research focuses on cancer biology, circadian rhythm disruption, tumor microenvironment, and molecular mechanisms driving breast cancer progression, with an emphasis on translational oncology and preclinical cancer research.

Trenton Stewart

Trenton Stewart worked as a Graduate Student in the Roy Sarkar Lab at Texas A&M University. He assisted with microbiome data analysis and interpretation, and contributed to manuscript preparation and scientific writing for collaborative translational cancer research projects.

Efrat Muller

Efrat Muller is a researcher in the Department of Veterinary Medicine, University of Cambridge, UK, where her work focuses on microbiome science, metagenomics, and translational microbiome research. Her research integrates large-scale microbial genomic analyses with host-microbiome interactions to advance the understanding of human health and disease.

Erin H. Seeley

Dr. Erin H. Seeley is a researcher specializing in translational biomedical research and advanced analytical technologies. She is the Director of the Mass Spectrometry Imaging Core Facility at UT MD Anderson Cancer Center, Houston, Texas, USA. Her work integrates molecular imaging with multidisciplinary approaches to investigate disease biology and support collaborative studies aimed at advancing precision medicine.

Tapasree Roy Sarkar

Dr. Tapasree Roy Sarkar is an Assistant Professor in the Department of Biology at Texas A&M University, where she leads a multidisciplinary research program in translational oncology. Her laboratory combines molecular biology, cancer metabolomics, bioinformatics, mechanobiology, and tumor microenvironment research to investigate the mechanisms of cancer progression and metastasis and to develop innovative therapeutic strategies for precision medicine.

Authors’ contributions

M.S. and O.O. performed experiments, analyzed the data, and wrote the original draft. T.S. performed experiments and analyzed the data. E.M did data curation and formal analysis, E.S. did data curation and Formal analysis, TRS conceptualized the project, supervised, wrote the original draft, and secured funding.

Data availability

All the data generated or analyzed during this study are included in the manuscript and supporting files.

Declarations

Ethics approval and consent to participate

Animal experiments were performed according to the Use of Laboratory Animals approved by the Animal Care and Use Committee of Texas A&M University under protocol #2024-0095. All methods were performed in accordance with relevant guidelines and regulations.

Competing interests

The authors declare no competing interests.