Metabolic effects of periodontal pathogens Fusobacterium nucleatum and Porphyromonas gingivalis in colorectal cancer cells

Abstract

Background

Periodontal pathogens are increasingly recognized as key regulators of cancer biology. However, their roles in modulating metabolism-related gene expression in colorectal cancer remain poorly understood.

Methods

Human colorectal cancer (HCT)116 cells were exposed to Fusobacterium nucleatum and Porphyromonas gingivalis for 6 h and 24 h. Expression levels of genes involved in lipid metabolism (ACAT1 and PLD2), amino acid metabolism (PAH), polyamine synthesis (SMS), kynurenine pathway (KYNU), and detoxification (ALDH2) were evaluated using quantitative real-time PCR (qPCR). The expression profiles of these genes were also compared between colorectal tumor and normal tissues using The Cancer Genome Atlas (TCGA) RNA-seq data.

Results

F. nucleatum and P. gingivalis induced distinct, time-dependent metabolic changes in HCT116 cells. At 6 h, F. nucleatum upregulated KYNU, PLD2, and ACAT1, whereas P. gingivalis broadly suppressed SMS, PAH, KYNU, ACAT1, and ALDH2. At 24 h, F. nucleatum and P. gingivalis maintained predominantly suppressive effects on KYNU, SMS, and PAH. mRNA expression analysis indicated the downregulation of ACAT1 and ALDH2, and upregulation of KYNU, PLD2, PAH, and SMS in colorectal cancer, and partial overlap with F. nucleatum-induced changes.

Conclusion

The distinct impacts of the two bacterial species demonstrated the complexity of colorectal cancer host–microbiome interactions. These findings suggest that microbial species may influence tumor biology through different metabolic processes, leading to diverse rather than uniform metabolic changes in colorectal cancer cells.

Introduction

Periodontitis is a chronic inflammatory disease caused by a dysbiotic polymicrobial biofilm, where key pathogenic species such as Fusobacterium nucleatum (F.nucleatum) and Porphyromonas gingivalis (P.gingivalis) play central roles. These organisms are considered integral parts of the periodontal “pathogenic consortium” because of their ability to weaken host immune responses and promote inflammatory tissue damage. F. nucleatum plays a unique ecological role among periodontal pathogens in facilitating biofilm formation by bridging early and late colonizing bacteria in dental plaque.^1,2 This bacterium possesses numerous virulence factors that enhance adherence to epithelial and endothelial cells, enable host tissue invasion, and modulate immune responses.¹ P. gingivalis is a Gram-negative anaerobic bacterium that produces many virulence factors including gingipains, lipopolysaccharides, and fimbrial proteins, which help it to evade the immune system and sustain inflammation.³

Recent advances in microbiome research have uncovered a crucial connection between the oral microbiome and gut health called the oral–gut axis.^4,5 Oral bacteria from periodontal infections may enter the gastrointestinal tract via saliva or blood and potentially colonize the intestinal lining. In particular, F. nucleatum is one of the most frequently detected bacteria in colorectal tumor tissues.⁶ Genomic and metagenomic studies have shown that colorectal cancer (CRC) tissues contain significantly higher levels of F. nucleatum than nearby non-tumorous tissue, indicating its tendency to accumulate in the tumor environment.^2,7 This selective colonization suggests that F. nucleatum may actively contribute to CRC development instead of simply coexisting as a bystander. F. nucleatum has been shown to promote carcinogenesis via molecular pathways that affect epithelial cell signaling and immune responses, especially through its adhesin FadA, which binds to E-cadherin on colon cells, activating β-catenin signaling and enhancing oncogenic gene transcription associated with proliferation, inflammation, and tumor growth.7, 8, 9 Other periodontal pathogens such as P. gingivalis are also associated with CRC, mainly by inducing chronic inflammation and impairing epithelial barriers. P. gingivalis triggers Toll-like receptor pathways and produces enzymes that damage tissue and modify immune responses, fostering the formation of a pro-tumor environment.^3,10,11 These microbial activities promote chronic inflammation, immune modulation, and alterations in epithelial communication, which all facilitate the development and progression of CRC.

Recent data from the Global Cancer Observatory and World Health Organization indicate that CRC is the third most common cancer worldwide and the second leading cause of cancer-related deaths.¹² Recent GLOBOCAN estimates approximately 1.9 million new cases of CRC and over 900,000 deaths worldwide in 2022.¹² The prevalence of CRC varies considerably among different regions, and it is closely linked to lifestyle changes and socioeconomic development. Known risk factors, including diets that are high in red and processed meats, insufficient fiber intake, obesity, sedentary lifestyles, tobacco use, and alcohol consumption contribute to the rising incidence of CRC in shifting populations.¹³ CRC is characterized by significant metabolic reprogramming, which is a hallmark of malignant transformation that enables cancer cells to sustain rapid proliferation and resist metabolic stress.14, 15, 16 The Warburg effect is a well-documented metabolic feature of CRC and it describes how tumor cells primarily utilize aerobic glycolysis rather than oxidative phosphorylation to produce energy.¹⁷ In CRC, bacteria such as F. nucleatum and P. gingivalis can influence key metabolic pathways that support tumor growth and survival.^11,18 These microbes can modulate host signaling networks through inflammatory mediators, bacterial metabolites, and direct host–microbe interactions. For instance, F. nucleatum has been shown to activate oncogenic pathways including β-catenin and NF-κB signaling, which then alter the expression of genes linked to glycolysis, glutamine metabolism, and lipid biosynthesis.^9,19 These shifts facilitate the modulation of metabolism-related gene expression in cancer cells, enhancing glucose uptake, lactate production, and biosynthetic activities. Importantly, recent data suggest that microbial colonization can directly regulate the transcription of metabolism-related genes in colorectal epithelial and cancer cells.²⁰ Tumor-associated bacteria have been linked to changes in the expression of genes involved in glycolysis, fatty acid synthesis, amino acid metabolism, and redox homeostasis, indicating that microbial signals may trigger metabolic adaptations favoring tumor progression and therapy resistance.^18,20 These findings highlight a critical link between the microbiome and cancer metabolism, where microbial components act as key regulators of host metabolic networks. Investigating the effects of periodontal infections such as F. nucleatum and P. gingivalis on the expression of metabolism-related genes in CRC cells could provide vital insights into how microbial dysbiosis promotes carcinogenesis. Understanding these interactions might reveal novel metabolic vulnerabilities and lead to potential therapeutic targets for microbiome-influenced CRC.

Materials and Methods

Bacterial culture

The bacterial strains P. gingivalis ATCC 53977 and F. nucleatum subsp. nucleatum ATCC 25586 were cultured anaerobically in brain heart infusion (BHI) broth supplemented with hemin (5 mg/mL) and menadione (5 mg/mL) (Himedia, India). Cultures were incubated at 37 °C for 48 h in a Whitley A20 anaerobic workstation under a gas mixture of 80% N₂, 20% CO₂, and 0.1% O₂. Freshly prepared bacterial suspensions were standardized to 0.5 McFarland turbidity units before infection, ensuring comparable initial bacterial loads of both F. nucleatum and P. gingivalis. Cells were exposed to bacteria under identical culture conditions at a multiplicity of infection (MOI) of 100 for 6 h or 24 h according to a previously described standard protocol.21, 22, 23

Cell infection assay

Human colorectal cancer 116 (HCT116) CRC cells were seeded in 12-well plates at a density of 5 × 10⁵ cells/mL and allowed to adhere overnight at 37 °C, reaching 70–80% confluence prior to infection. At 1 h before bacterial exposure, the medium was replaced with antibiotic-free Dulbecco's modified Eagle medium. Cells were then infected separately with P. gingivalis or F. nucleatum at an MOI of 100 according to established protocols.21, 22, 23 Wells treated with phosphate-buffered saline were used as negative controls. Post-infection, both supernatants and cell lysates were harvested at 6 h and 24 h for RNA isolation and subsequent downstream analyses.

Gene expression analysis by quantitative real-time PCR (qPCR)

Total RNA was extracted using an RNeasy Mini Kit (50) (Qiagen 74104) according to the manufacturer's protocols. The concentration and purity of RNA were quantified using a NanoDrop system. RNA was then converted into cDNA using a QuantiTect Rev. Transcription Kit (50) (Qiagen 205311) according to the manufacturer's protocols. qPCR was performed using HOT FIREPol EvaGreen qPCR Supermix to target the following metabolism-associated genes: SMS (spermine synthase), PAH (phenylalanine hydroxylase), KYNU (kynureninase), ACAT1 (acetyl-CoA acetyltransferase), ALDH2 (aldehyde dehydrogenase 2), and PLD2 (phospholipase D2). PCR was amplified in a total reaction mixture volume of 20 μL under the following cycling conditions: initial denaturation for 12 min at 95 °C, followed by 40 cycles of denaturation for 15 s at 95 °C and annealing for 30 s at 60 °C, and extension for 30 s at 72 °C. qRT-PCR was performed using a QuantStudio™ 5 Real-Time PCR system. GAPDH served as the housekeeping reference gene, and relative gene expression levels were quantified using the 2^−ΔΔCt method.

Statistical analysis

Comparisons were conducted between bacterial treatments at the same time points, where data were expressed as the mean ± standard deviation based on three independent biological replicate experiments (n = 3). For each biological replicate, qRT-PCR reactions were performed as technical replicates and averaged before statistical analysis. Comparisons were performed between two groups using the unpaired two-tailed Student's t-test, whereas comparisons were made among multiple groups using one-way analysis of variance. Relative gene expression levels were calculated using the 2^−ΔΔCt method, where ΔCt was determined by subtracting the Ct value for the housekeeping gene (GAPDH) from the Ct value for the target gene, and ΔΔCt was calculated relative to the untreated control group. Gene expression levels were expressed as fold changes (2^−ΔΔCt). Statistically significant differences were accepted at P < 0.05. In the figures, significant differences are indicated as: P < 0.05 (∗), P < 0.01 (∗∗), and P < 0.001 (∗∗∗).

RNA sequencing data acquisition and processing

RNA sequencing gene expression data and the corresponding sample annotations for CRC were obtained from the National Cancer Institute Genomic Data Commons, including The Cancer Genome Atlas (TCGA) Colon Adenocarcinoma (TCGA-COAD) and Rectal Adenocarcinoma (TCGA-READ) projects. Only TCGA samples were included in the present analysis, and samples from the Genotype-Tissue Expression project or other external transcriptomic data sets were not incorporated.

TCGA-COAD and TCGA-READ were jointly analyzed because the primary objective was to evaluate broad CRC-level tumor–normal expression patterns for six prespecified metabolism-related genes rather than anatomical site-specific molecular subtypes. Previous integrated TCGA analyses found highly similar genomic alteration and gene expression patterns between non-hypermutated colon and rectal adenocarcinomas, supporting their pooled assessment in CRC-level analyses. Pooling also increased the number of available adjacent normal samples and improved the statistical power of the targeted comparisons. However, the combined analysis was not intended to imply complete molecular equivalence between colon and rectal adenocarcinomas, and potential anatomical site- and subtype-related heterogeneity was considered when interpreting the findings.

In total, 698 RNA-seq data sets were initially retrieved. Only primary tumor and solid-tissue normal samples with available gene-level expression data and valid TCGA sample annotations were considered eligible. Repeated samples or multiple aliquots derived from the same patient were identified according to the corresponding TCGA case and sample identifiers, and curated to avoid duplicate observations. Samples were excluded that lacked the required RNA-seq expression data or valid sample-type information. After data curation, 616 primary tumor tissues and 51 adjacent normal tissues were retained for downstream analysis.

Gene-level fragments per kilobase of transcript per million mapped reads (FPKM) values generated through the harmonized Genomic Data Commons RNA-seq processing workflow were used for the analysis. FPKM normalization accounts for differences in sequencing depth and transcript length. Expression values were subsequently log₂(FPKM + 1) transformed to reduce right-skewness and allow the visualization of genes with different expression ranges.

The analysis was restricted a priori to six metabolism-related genes: ACAT1, ALDH2, KYNU, PAH, PLD2, and SMS. Therefore, global low-expression transcript-filtering threshold was not applied. Zero-expression values were retained as valid observations, whereas samples with missing expression values for an individual gene were excluded only from the corresponding gene-specific comparison. Consequently, the number of available samples could differ slightly among genes.

Differences in gene expression between primary tumor and adjacent normal tissues were evaluated using the two-sided Wilcoxon rank-sum test because the two groups were independent and the expression values were not assumed to follow a normal distribution. A P value < 0.05 was considered statistically significant. All included samples were obtained from TCGA and processed using the same harmonized Genomic Data Commons RNA-seq workflow. Moreover, GTEx or other external data sets were not integrated into the analysis. Therefore, additional cross-platform batch-effect correction was not applied.

Results

Lipid metabolism genes

ACAT1 was strongly induced by F. nucleatum at 6 h, where the expression levels were significantly higher than those in untreated control cells. By contrast, P. gingivalis exposure led to the notable suppression of ACAT1 at the same time point. By 24 h, ACAT1 expression approached baseline under both bacterial treatments, whereas PLD2, which is involved in producing phosphatidic acid and lipid signaling, was substantially upregulated by F. nucleatum at both 6 h and 24 h. Conversely, treatment with P. gingivalis led to the significant downregulation of PLD2 at both time points (Figure 1). These data suggest that F. nucleatum may promote lipid metabolic activation and signaling pathways, potentially supporting adaptive or survival responses, whereas P. gingivalis appears to suppress these processes, indicating divergent strategies in terms of modulating host lipid metabolism and associated signaling networks.

Figure 1.

Relative expression levels of SMS, PAH, KYNU, ACAT1, PLD2, and ALDH2 in HCT116 colorectal cancer cells following infection with Porphyromonas gingivalis (PG) or Fusobacterium nucleatum (FN) for 6 h and 24 h. Gene expression was quantified by qRT-PCR and expressed as the fold change (2^−ΔΔCt) relative to uninfected controls. PG predominantly suppressed metabolic gene expression, whereas FN induced differential responses, including increasing the expression of ACAT1, PLD2, and KYNU at 6 h. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001.

Amino acid and tryptophan metabolism genes

SMS expression was greatly reduced by P. gingivalis at both 6 h and 24 h, whereas F. nucleatum caused a relatively smaller but statistically significant decrease compared with the control cells. The pattern was similar for PAH. At both time points, P. gingivalis significantly lowered PAH expression, and F. nucleatum caused a smaller but still significant decrease. F. nucleatum treatment led to a significant increase in KYNU expression at 6 h, whereas P. gingivalis decreased KYNU expression. Both bacterial treatments resulted in lower KYNU levels compared with the controls after 24 h (Figure 1). These results suggest that P. gingivalis exerts a stronger and more sustained inhibitory effect on host metabolic pathways, while F. nucleatum induces transient, pathway-specific modulation, particularly for kynurenine metabolism, potentially reflecting different strategies for modulating host metabolism-related gene expression and immune regulation.

Detoxification and aldehyde metabolism genes

P. gingivalis markedly decreased ALDH2 expression at both time points, suggesting transcriptional modulation of a gene involved in aldehyde detoxification F. nucleatum treatment caused a moderate reduction in ALDH2 expression compared with the control cells (Figure 1).

TCGA CRC data analysis

Analysis of TCGA CRC data sets indicated the significant modulation of metabolism-related gene expression characterized by the downregulation of ACAT1 (p = 1.53 × 10⁻²⁷) and ALDH2 (p = 1.29 × 10⁻⁷), as well as the upregulation of KYNU (p = 6.84 × 10⁻³), PAH (p = 3.26 × 10⁻¹⁷), PLD2 (p = 1.85 × 10⁻²), and SMS (p = 1.72 × 10⁻¹⁹). These findings indicate the suppression of acetyl-CoA metabolism aldehyde detoxification genes, and the concurrent activation of kynurenine, amino acid, polyamine, and phospholipid metabolism in CRC (Figure 2).

Figure 2.

Differential expression of metabolism-related genes in colorectal cancer tissues and adjacent normal tissues. The mRNA expression levels of ACAT1, ALDH2, KYNU, PAH, PLD2, and SMS were compared between primary colorectal cancer tissues and adjacent normal tissues obtained from the TCGA-COAD and TCGA-READ projects. Gene expression values are presented as log₂(FPKM + 1). Each point represents an individual tissue sample, and the number of available samples is indicated below each group. Differences between the two independent groups were evaluated using the two-sided Wilcoxon rank-sum test. ACAT1 and ALDH2 were significantly downregulated in colorectal cancer tissues, whereas KYNU, PAH, PLD2, and SMS were significantly upregulated. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001.

Discussion

The results obtained in the present study demonstrated that oral pathogens differentially influenced metabolic gene expression in CRC cells over time, highlighting their plausible roles in modulating tumor-associated metabolic pathways. F. nucleatum primarily stimulated genes related to lipid metabolism whereas P. gingivalis had a predominantly suppressive effect on multiple metabolic regulators. These findings suggest that microbial species do not exert uniform effects on tumor metabolism but instead they may induce distinct metabolic phenotypes, potentially reflecting differences in their virulence strategies, metabolic outputs, and interactions with host signaling networks.

F. nucleatum induced the upregulation of PLD2 and ACAT1, particularly at early time points, indicating activation and phospholipid remodeling processes that affect membrane dynamics and signaling. The observed upregulation of PLD2 following exposure to F. nucleatum in the present study aligns with previous findings of PLD2 elevation in CRC.²⁴ Previous studies reported increases in PLD2 expression and activity in colorectal tumor tissues compared with normal mucosa, implicating this enzyme in tumor-associated signaling processes.²⁴ PLD2 is involved in phospholipid hydrolysis and the generation of bioactive lipid mediators, which contribute to membrane dynamics, cell proliferation, and migration. However, more recent evidence suggests that reduced PLD2 expression may be associated with enhancement of the epithelial–mesenchymal transition, increased invasiveness, and poorer clinical outcomes, indicating a potentially protective role in certain contexts.²⁵ Indeed, the F. nucleatum-induced upregulation of PLD2 observed in our model may reflect the modulation of lipid signaling pathways that influence tumor cell behavior, highlighting the complex and context-specific role of PLD2 in CRC biology. The temporary rise in ACAT1 after 6 h, followed by a decline at 24 h, further indicates a dynamic and complexly regulated host lipid metabolic response to bacterial stimulation. Conversely, P. gingivalis had a primarily suppressive impact on these two genes, as well as SMS, PAH, and ALDH2, suggesting more extensive inhibitory effects on cellular metabolic processes, which may differ fundamentally from those of F. nucleatum. This temporal pattern may reflect an early increase in acetyl-CoA related metabolic process followed by adaptive feedback regulation in response to sustained microbial exposure.

Cancer reprograms amino acid metabolism via polyamine metabolism and tryptophan catabolism.^26,27 Polyamines, especially spermine, are required for cell proliferation, and CRC has been linked to dysregulation of polyamine-producing enzymes.^26,27 In the present study, P. gingivalis exposure dramatically reduced SMS expression, but F. nucleatum produced a smaller effect, showing that microbial exposure may affect polyamine metabolism in different ways. Reduced SMS expression may indicate altered polyamine availability, potentially influencing DNA stability and cellular proliferation, although the functional consequences require further investigation. The kynurenine route of tryptophan metabolism produces compounds that activate immune regulatory mechanisms such as the aryl hydrocarbon receptor, contributing to immunosuppression and tumor growth in CRC.^28,29 The biphasic regulation of KYNU observed following F. nucleatum exposure in the present study, with early upregulation at 6 h followed by downregulation at 24 h, suggests a dynamic temporal response by the kynurenine pathway to microbial stimulation. The initial increase in KYNU expression may reflect the rapid activation of immune–metabolic signaling pathways, whereas the subsequent decline at 24 h may indicate adaptive cellular feedback, pathway exhaustion, or compensatory regulation after prolonged exposure. This temporal shift demonstrates that microbial effects on tryptophan metabolism are not static and may vary according to the duration of exposure. These findings support the hypothesis that F. nucleatum may modulate CRC biology through time-dependent effects on interconnected metabolic and immune regulatory pathways.

The gene expression changes induced by F. nucleatum partially overlapped with tumor-associated profiles, suggesting that microbial exposure may contribute to metabolism-related alterations in CRC. However, the lack of complete concordance, particularly for genes such as SMS and PAH, indicates that microbial effects are selective and context dependent, and may be influenced by the tumor microenvironment, host genetics, and interactions with other microbial species.³⁰ The concordant upregulation of genes such as KYNU and PLD2 supports their potential relevance to microbiome-associated metabolic responses in CRC, whereas discordant patterns may reflect temporal dynamics or differences between acute in vitro exposure and chronic in vivo conditions. These findings are consistent with evidence that F. nucleatum is enriched in CRC tissues, and associated with tumor progression and altered host gene expression. By contrast, the role of P. gingivalis in CRC remains less clearly defined and may involve distinct or context-dependent mechanisms. Direct evidence linking P. gingivalis to metabolic gene regulation in CRC remains limited, highlighting the exploratory nature of the present comparison. The different effects of the two bacterial species further demonstrate the complexity of host–microbiome interactions, and suggest that individual oral pathogens may elicit distinct transcriptional responses in CRC cells. In addition to signaling-mediated effects, microbial-derived metabolites may influence host metabolic pathways. Bacterial products such as short-chain fatty acids and one-carbon intermediates, including formate, can affect mitochondrial function, redox balance, and biosynthetic processes in cancer cells, potentially contributing to metabolic alterations.^31,32 The temporal changes observed in gene expression further highlight the dynamic nature of host responses to microbial exposure. The early responses at 6 h may reflect acute transcriptional activation, whereas the changes at 24 h could indicate adaptation, compensatory regulation, or sustained microbial effects.^33,34 The qPCR expression values were normalized with respect to a stable reference gene, minimizing the influence of differences in total RNA or cell number, but cell viability was not independently assessed. Therefore, a contribution of cellular stress or cytotoxicity to the broader transcriptional suppression observed following P. gingivalis exposure cannot be completely excluded. Overall, our findings provide insights into how oral pathogens may influence metabolism-related gene expression in CRC, and highlight microbial–metabolic interactions as potential targets for future investigation. Further mechanistic and translational studies are required to determine whether modulating these interactions could have therapeutic value in microbiome-associated CRC.³⁵

Our study had some limitations. One limitation is that all of the experiments were performed using a single CRC cell line (HCT116). HCT116 is a well-established model for investigating CRC biology and host–microbial interactions, but the responses to bacterial exposure may vary among CRC cell lines with different molecular characteristics. Furthermore, the analysis was restricted to mRNA expression, and the corresponding protein levels were not assessed. F. nucleatum and P. gingivalis were applied at an identical starting MOI, but the bacterial viability and relative viable burden were not quantified at the 24-h endpoint, and thus differences in bacterial persistence during prolonged co-culture cannot be excluded. The possibility of residual technical variation within the TCGA cohort could also not be completely excluded and was considered when interpreting the findings. The selected genes were predefined based on biological relevance and analyzed independently, but validation is required in additional experimental models to confirm the reproducibility of these gene-specific findings. Broader -omics approaches and investigations of multi-species microbial interactions will further clarify the roles of microbiome-driven metabolic alterations in CRC.

Conclusion

In conclusion, F. nucleatum and P. gingivalis induced distinct, time-dependent changes in the expression of selected metabolism-related genes in HCT116 CRC cells. Some of these transcriptional changes partially overlapped with tumor–normal expression differences observed in TCGA CRC data. However, this comparison does not imply causal relationships between bacterial exposure and gene expression in clinical tumors. Moreover, the investigated transcripts are not specific to microbial effects and may be regulated by tumor genetics, inflammation, cellular composition, and other microenvironmental factors. Therefore, the present findings should be regarded as preliminary evidence of species-specific modulation of metabolism-related gene expression rather than supporting their use as microbiome-associated biomarkers. Further validation is required in additional experimental models and clinical samples by integrating microbial, transcriptomic, proteomic, and metabolomic data.

Ethical approval

Not applicable.

Authors contributions

DG: Conceptualization, Data curation, Investigation, Formal analysis, Writing – original draft. BS: Investigation, Methodology, Writing – review and editing. PS: Data curation, Visualization. MM and ZL, Investigation, Visualization, Resources, Writing – review and editing. All authors have critically reviewed and approved the final draft and are responsible for the content and similarity index of the manuscript.

Source of funding

The study was funded by Ajman University under Research Grant 2023-IRG-DEN-11 received by the primary author Dr Divya Gopinath. A preliminary version of the findings of this study was presented at the 2026 CED/NOF-IADR Oral Health Research Congress, September 3–5, 2026. The presenting author (Divya Gopinath) was funded by an Ajman University conference grant.

Conflict of interest

The authors have no conflict of interest to declare.