Identification and functional characterization of Pseudomonas fluorescens as a novel intratumoral bacterium in colorectal cancer

Abstract

Colorectal cancer (CRC) remains a major global health burden as one of the leading causes of cancer-related mortality. Recent research has highlighted the crucial role of gut microbiota in CRC development. Through high-throughput full-length 16 S rDNA sequencing of tumor and adjacent non-tumor tissues from 14 CRC patients, significant microbial differences were identified. At the phylum level, Firmicutes (52.59%), Bacteroidetes (18.51%), and Proteobacteria (14.89%) dominated both tissue types, while at the genus level, Bacteroides (8.02%) and Escherichia (4.50%) showed the highest abundance. Notably, 17 bacterial species exhibited differential abundance between tumor and normal tissues, with Anaerotignum faecicola and Pseudomonas fluorescens being significantly enriched in tumor tissues. Functional prediction analysis revealed the microbiota’s predominant involvement in carbohydrate metabolism, amino acid metabolism, and energy metabolism pathways. Subsequent validation in 20 additional patient samples confirmed P. fluorescens enrichment in tumor tissues, and in vitro experiments demonstrated its ability to promote CRC cell viability and proliferation. These findings provide valuable insights into CRC-associated microbial signatures and suggest P. fluorescens as a potential contributor to tumor progression, offering new directions for developing diagnostic markers and therapeutic interventions in CRC management.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-026-04827-w.

Introduction

Colorectal cancer (CRC) represents a significant and evolving global health challenge. A concerning trend is its rising prominence in younger populations: CRC has become the leading cause of cancer death in men under 50 and the second-leading cause in women under 50 (after breast cancer), marking a substantial shift from its fourth-place ranking in 1998 [1]. While overall CRC mortality rates declined by 57% between 1970 and 2020 (from 29.2 to 12.6 per 100,000), the changing demographic profile, characterized by increasing incidence in younger adults, coupled with rapid societal aging and economic development, portends potential future challenges in sustaining this positive trajectory. This underscores the critical and ongoing need for enhanced prevention, early detection, and therapeutic strategies to mitigate the growing burden of CRC.

The human microbiota, densely populated in the gut and inhabiting various body sites (e.g., skin, respiratory tract, reproductive system), plays a pivotal role in maintaining host homeostasis, contributing crucially to metabolism, immune function, and nutrient synthesis. However, this symbiotic relationship is complex, and a burgeoning body of evidence implicates gut microbiota dysbiosis in carcinogenesis, tumor progression, and modulation of therapeutic responses across multiple cancer types, including CRC [2, 3]. Recent technological advancements, particularly in high-throughput genome sequencing, have further revolutionized our understanding by revealing the presence of microbial communities within solid tumors themselves - termed the intra-tumor microbiome [4–6]. This discovery signifies a paradigm shift, suggesting that the tumor microenvironment (TME) harbors its own distinct microbial ecosystem [7, 8].

Accumulating evidence demonstrates that intra-tumoral microorganisms (ITM) exert profound influences on tumor development and metastasis through diverse mechanisms [9–12]. ITM have been mechanistically linked to tumor initiation and development by promoting genome instability, increasing mutation rates, modulating oncogenic signaling pathways, and inducing chronic inflammation [13–17]. Nejman D’s team observed variations in the abundance of specific tumor-associated bacteria (Gardnerella vaginalis) and found that the distribution patterns of these microorganisms within tumors corresponded to the gut microbiome characteristics of melanoma patients responsive to immune checkpoint inhibitors [18]. This suggests a potential association between tumor-associated bacterial communities and the efficacy of immunotherapy.

P. fluorescens is a ubiquitous, Gram-negative bacterium which is commonly found in diverse environments such as soil, water, and plant surfaces. And it is often associated with food spoilage [19, 20]. Although primarily considered an environmental microbe and an opportunistic pathogen in immunocompromised hosts, P. fluorescens has been isolated from various human clinical specimens, including blood (the most frequent site of reported infection), sputum, and feces [21–24]. The foodborne strain P. fluorescens ITEM 17,298 exhibited adaptive traits at 15 °C—including enhanced biofilm formation, extracellular polymeric substance (EPS) production, and pigment synthesis—which may facilitate its persistence under stressful environmental conditions [25]. Intriguingly, beyond overt infection, hypotheses suggest potential roles for P. fluorescens in the pathogenesis of chronic respiratory diseases and inflammatory bowel disease (IBD), implying a capacity to interact with host tissues in ways that may promote inflammation or dysbiosis.

Despite its documented presence in humans and potential links to inflammation, the relationship between P. fluorescens and human solid tumors, particularly its role as an intra-tumoral bacterium in CRC pathogenesis and progression, remains poorly understood. No specific mechanistic pathways linking P. fluorescens to colorectal carcinogenesis have been elucidated to date. This represents a significant knowledge gap. Therefore, to identify novel microbial signatures associated with CRC and potential diagnostic/therapeutic targets, we performed 16 S rDNA sequencing on matched CRC tissues and adjacent normal tissues. Through comprehensive microbial differential analysis and validation of key candidates, we aimed to characterize the intratumoral microbial landscape, with specific focus on identifying and validating the presence and potential significance of P. fluorescens and other differential taxa in CRC.

Materials and methods

Sample collection and patients’ cohort

A total of 28 tissue samples were collected from 14 colorectal cancer patients (both tumor tissue and tumor-adjacent tissue for each one) in Zhongda Hospital, Southeast University. All volunteers have signed an informed consent form before sampling. Subjects were excluded if they had other malignant tumors, autoimmune disorders, infectious diseases, renal dysfunction, a history of gastrointestinal surgery, chemotherapy, radiotherapy or antibiotic treatment in the previous year or were administered antibiotics for more than 3 days in the previous 3 months. Surgically resected tumor tissues and tumor-adjacent tissues (5 cm away from the tumor edge) from CRC patients were collected under sterile conditions, immediately frozen in liquid nitrogen, and properly stored at -80℃ for future use. This research proposal was reviewed and approved by the ethics committee of Zhongda Hospital, Southeast University (2023ZDSYLL100-P01).

DNA Extraction, Amplification, and sequencing of the Microbiome

Genomic DNA was extracted from tumor and tumor-adjacent tissues using the E.Z.N.A.^® Tissue DNA Kit (Omega Bio-tek, Norcross, GA, USA) following the manufacturer’s standardized protocol. The V1-V9 hypervariable regions of the bacterial 16 S ribosomal RNA gene were amplified using polymerase chain reaction (PCR) with universal primers 27F (5’-AGRGTTYGATYMTGGCTCAG-3’) and 1492R (5’-RGYTACCTTGTTACGACTT-3’), incorporating unique 8-base barcodes for sample identification (Pacific Biosciences, Menlo Park, CA, USA; PN: 102-135-500). The thermal cycling protocol consisted of an initial denaturation at 95 °C for 2 min, followed by 27 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 60 s, with a final extension at 72 °C for 5 min.

PCR amplification was performed in triplicate 20 µL reactions containing: 4 µL of 5× FastPfu Buffer, 2 µL of 2.5 mM dNTPs, 0.8 µL of each primer (5 µM), 0.4 µL of FastPfu DNA Polymerase (AP221-01, TransGen Biotech), and 10 ng of template DNA. Resulting amplicons were electrophoresed on 2% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA) according to the manufacturer’s specifications. Genomic DNA was amplified and converted into SMRTbell libraries via blunt-end ligation following the SMRTbell Prep Kit 3.0 protocol (Pacific Biosciences, Cat. 102-182-700). Purified SMRTbell libraries from the pooled and barcoded samples were sequenced on a single PacBio Sequel IIe cell. All amplicon sequencing was performed by Shanghai Biozeron Biotechnology Co. Ltd (Shanghai, China). Raw subreads were processed using SMRT Link v11.0 to generate demultiplexed CCS reads (minimum passes = 3, accuracy ≥ 0.99). Sequences were filtered by length (1,000–1,800 bp) and quality via SMRT Portal, followed by barcode/primer removal using lima pipeline (Pacific Biosciences).

Bioinformatics analysis

The qualified libraries were then loaded onto the PacBio Sequel sequencing platform and sequenced with the Sequel Sequencing Kit 2.1 in CCS (Circular Consensus Sequencing) mode, with a read length set to [1,000–1,800 bp] and a minimum of 10,000 CCS reads per sample to ensure the reliability of microbial community analysis. After raw sequencing data generation, quality control and filtering were conducted using SMRT Link v [11.0] software to remove sequences with lengths < 1200 bp or > 1800 bp, an average quality value (QV) < 30, ambiguous bases (N), or incompletely trimmed primer sequences, yielding high-quality Clean Reads. Subsequently, Clean Reads were clustered into Operational Taxonomic Units (OTUs) at a 97% sequence similarity threshold using Uparse v [7.0.1001] software; low-abundance OTUs with a relative abundance < 0.001% and those exclusively present in blank control groups were excluded to eliminate experimental contamination. Finally, representative OTU sequences were aligned against the Silva database (SSU rRNA database) (v[138.2]), and taxonomic annotation at the kingdom, phylum, class, order, family, genus, and species levels was performed using the RDP Classifier Bayesian algorithm, followed by statistical analysis of the relative abundance of species at each taxonomic level.

Alpha diversity metrics (Chao1, ACE, Shannon, Simpson) were calculated using QIIME2. Beta diversity analysis, based on Bray-Curtis dissimilarity, was performed and visualized via principal coordinate analysis (PCoA) using the R package ‘vegan’. All custom analysis code is available in the accompanying R package.

Bacterial strains and treatment

The P. fluorescens strain (BMZ339565) was acquired from Mingzhoubio company in China. It was cultured in beef extract peptone medium (Beef paste: 3.0 g; Peptone: 10.0 g; NaCl: 5.0 g; Water: 1000 ml; pH: 7.4–7.6) at 37 °C. Planktonic growth of P. fluorescens was tracked by assessing the optical density (OD) at 600 nm.

Cell culture and treatment

The human colon cancer HCT-116 cells were obtained from the American Type Culture Collection (ATCC, USA), and grown in DMEM medium (Gibco, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin (Gibco, USA) at 37 °C under a humidified 5% CO2 atmosphere of the incubator. Cells were gathered and used for the subsequent assays once they reached 80% confluency. To create a bacteria-cell co-culture, HCT-116 cells (> 85% confluency) were exposed to P. fluorescens at multiplicity of infection (MOI) of 1:100 for 12-24 h, while the control group was set with DMEM medium with no P. fluorescens.

The human colon cancer DLD-1 cells were obtained from the American Type Culture Collection (ATCC, USA), and grown in RMPI-1640 medium (KeyGEN BioTECH, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin (Gibco, USA). To create a bacteria-cell co-culture, DLD-1 cells (> 85% confluency) were exposed to P. fluorescens at MOI of 1:100 for 12-24 h, while the control group was set with RMPI-1640 medium with no P. fluorescens.

Cell viability detection

Cell viability was determined using the Cell Counting Kit-8 (CCK-8; Beyotime Biotechnology, C0038, Shanghai, China) according to the manufacturer’s protocol. Briefly, HCT-116 or DLD-1 cells were seeded in 96-well plates and treated according to experimental groups. After treatment, cells were washed twice with phosphate-buffered saline (PBS; pre-warmed to 37 °C) and incubated with 100 µL of CCK-8 solution (1:10 dilution in complete medium) at 37 °C in a 5% CO2 humidified atmosphere for 1 h. Absorbance was measured at 450 nm using a Synergy H1 multi-mode microplate reader (BioTek Instruments, Winooski, VT, USA). Appropriate controls were included: positive control wells containing CCK-8 solution without cells, and negative control wells containing untreated cells. All experiments were performed in biological triplicates with three technical replicates for each condition to ensure reproducibility.

Cell proliferation detection

The 5-Ethynyl-2′-deoxyuridine (EdU) assay (BeyoClick™ EdU-594, Beyotime, China) was used to detect the cell proliferation of HCT-116 and DLD-1 cells. Cells were seeded in 12-wells plate and treated with/without P. fluorescens for 24 h and incubated with the EdU reagent for another 120 min. Afterwards, cells were fixed with 4% paraformaldehyde for 30 min, followed by permeabilization in 0.3% Triton X-100 at room temperature for 15 min. Azide 488 was added and stained for 30 min. Then, nucleus was stained with Hoechst 33342. Eventually, images were acquired on an inverted fluorescent microscope (SteREO Discovery, 20x, DAPI+FITC, Zeiss, German).

DNA/RNA isolation and quantitative real-time PCR (RT-qPCR)

Genomic DNAs were extracted from tumor tissue and tumor-adjacent tissue samples using E.Z.N.A.^® Tissue DNA Kit (Omega Bio-tek, Norcross, GA, USA). Total RNA from tumor tissue and tumor-adjacent tissue samples were extracted using RNAiso-Plus (Takara, China). The RNA was then reverse transcribed into cDNA via PrimeScript™ RT Master Mix (Takara, China) following the manufacturer’s protocol and stored at -20 °C. Subsequently, the RT-qPCR was performed using ChamQ SYBR qPCR Master Mix (Vazyme, China) in a StepOne Plus System (Applied Biosystems, USA) following standard procedures. PCR amplification was performed under the following cycling conditions: initial denaturation at 95 °C for 30 s; followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. The fluorescence signal was collected at the end of each cycle. The primer concentration used is 10 µM. GAPDH was used as an internal reference, and the 2^−ΔΔT method was employed to calculate the difference in the abundance of P. fluorescens per cell between the tumor group and tumor-adjacent group.

Statistical analysis

All statistical analyses were performed using R (v[4.0.2]; R Foundation for Statistical Computing) and SPSS 25.0 (IBM Corp., USA). Continuous variables with non-normal distributions (e.g., α-diversity indices) were expressed as median (interquartile range, IQR), while categorical data (e.g., patient demographics) were presented as frequencies and percentages (n (%)). Microbial relative abundances were reported as percentages at phylum, genus, and species levels.

Group comparisons employed non-parametric tests due to skewed data distributions: α-diversity indices (Ace, Chao1, Shannon, Simpson) between tumor (T) and tumor-adjacent (TA) tissues were compared using the Wilcoxon signed-rank test;β-diversity was assessed via principal coordinate analysis (PCoA) based on Bray-Curtis distance matrices; Microbial composition differences (phylum to species level) were analyzed using the Wilcoxon signed-rank test, with significance defined as P < 0.05 and median relative abundance > 0.01%.

Biomarker identification integrated three approaches: Random Forest analysis identified potential biomarkers at phylum/genus levels using Variable Importance (Varimp) thresholds (Varimp ≥ 1.0 for phylum, ≥ 0.10 for genus), and the R package was randomForest v[4.6–14]; LEfSe (Linear Discriminant Analysis Effect Size) identified differentially abundant taxa (Linear Discriminant Analysis (LDA) score > 2) [26]; Diagnostic efficacy of biomarkers was evaluated by receiver operating characteristic (ROC) curves, calculating area under the curve (AUC), specificity, and sensitivity.

We analyzed data from experimental validation (qPCR, cell assays) using the non-parametric Wilcoxon signed-rank test. All statistical results and profiles were visualized using GraphPad Prism software (v[8.0]) and the ggplot2 package within the R programming environment.

Results

Descriptive data of volunteers in this study

This study enrolled 14 colorectal cancer (CRC) patients (10 males, 4 females) aged 41–77 years (Table 1). Colon cancer accounted for 71.4% of cases, with the remainder being rectal cancer. Most tumors infiltrated to the serosa and subserosal layers, and 42.9% exhibited lymph node metastasis.

Table 1.

Descriptive data of included volunteers in the study

Characteristics	Median (range)/number (%)
Gender
Male	10(71.4%)
Female	4(28.6%)
Age(y)	66(41–77)
<60	4(28.6%)
≥60	10(71.4%)
Tumor size(cm)
<4.5	8(57.1%)
≥ 4.5	6(42.9%)
Tumor site
Colon	10(71.4%)
Rectum	4(28.6%)
TNM
Ⅰ	1(7.1%)
Ⅱ	5(35.7%)
Ⅲ	7(50.0%)
Ⅳ	1(7.1%)
Lymph node metastasis
Yes	6(42.9%)
No	8(57.1%)
Tumor tissue type
Adenocarcinoma	12(85.7%)
Mucinous adenocarcinoma	2(14.3%)
Depth of tumor infiltration
Serosa layer	2(14.3%)
Subserous layer	10(71.4%)
Muscular layer	1(7.1%)
Muscularis propria	1(7.1%)
AFP (ng/mL)	3.26(1.47–9.52)
CEA (ng/mL)	3.08(0.71–50.1)
CA199(U/mL)	19.0(1-44.9)
CA724(U/mL)	2.15(1-6.78)

Analysis of α diversity and β diversity of gut microbiota in CRC patients

In the α-Diversity analysis (Fig. 1), no significant differences were observed between tumor tissues (T group) and tumor-adjacent tissues (TA group) in terms of species richness indices (Ace, Chao1) (P = 0.7). However, the larger variance in Ace and Chao1 values within the T group suggested higher heterogeneity in species abundance among tumor samples. Similarly, for species diversity indices (Shannon, Simpson), no statistical differences were detected between the two groups (P = 0.8), although greater variance in the TA group indicated higher heterogeneity in microbial distribution within adjacent tissues. In the β-Diversity analysis, PCoA revealed no significant separation between the T and TA groups (Supplementary material.1), which was consistent with the spatial proximity of the sampling sites and indicated that intra-individual microbial similarity was higher than inter-individual differences.

Fig. 1.

The α diversity of microbial community in the samples. The Ace, Chao1, Shannon and Simpson indexes between tumor (T) and tumor-adjacent (TA) group had no difference statistically (A-D). However, the variance values of Ace and Chao1 had a tendency to be larger in T group than in TA group (A-B), while the Shannon and Simpson index had the opposite tendency (C-D)

Characteristic bacterial structure of tissue samples from CRC patients

In phylum level, the result (Fig. 2) found that Firmicutes (52.59%) dominated both T and TA groups, followed by Bacteroidetes (18.51%), Proteobacteria (14.89%), Fusobacteria (2.26%), and Actinobacteria (0.56%). As for genus level, high-abundance genera included Bacteroides (8.02%), Escherichia (4.50%), Fusobacterium (2.22%), and Parvimonas (1.37%). Significant inter-individual variations were noted at both phylum and genus levels, whereas intra-individual differences between T and TA tissues were minimal.

Fig. 2.

Different gut microbiota in CRC patients on phylum, genus and species level. The Wilcoxon test was used to analyze the composition of microbiota in CRC patients. P<0.05 marked as *, P<0.01 marked as **, P<0.001 marked as ***. It showed that Actinobacteria and Fusobacteria had statistical difference on phylum level between T and TA group (A). On genus level, there were 12 different genus of bacteria which was statistically different between T and TA group (B). 17 different species of bacteria was observed differently between T and TA group, including Bifidobacterium longum, Pseudomonas fluorescens, etc.(C)

Differences of microbiota between tumor tissues and tumor-adjacent tissues

At the phylum level (Fig. 2A), Actinobacteria and Fusobacteria exhibited trending differences (though statistical significance at P < 0.05 was not reached). Genus-level analysis (Fig. 2B) identified 12 genera with significantly differential abundance (P < 0.05), including Halomonas, Muricomes, and Burkholderia. Species-level profiling revealed significant enrichment of A.faecicola and P. fluorescens in tumor tissues (T group), with 15 species showing decreased abundance (including Bifidobacteriom longum, 4 Burkholderia species and so on). Subsequent experimental investigations focused on elucidating the potentially pro-tumorigenic role of P. fluorescens. Among the taxa exhibiting differential abundance, we focused subsequent analysis on P.fluorescens as it demonstrated not only statistical significance (P < 0.05) but also the better magnitude of increase in tumor tissues compared to adjacent non-cancerous controls.

Random forest and Lefse analysis for the gut microbiota

The Random Forest model (Fig. 3A-B) identified 7 phylum-level biomarkers (e.g., Actinobacteria, Proteobacteria) with Varimp > 1.0 and 15 genus-level biomarkers with Varimp > 0.10 (including Halomonas and Afipia). Complementary LEfSe analysis (LDA score > 2, Fig. 3B) demonstrated significant enrichment of 7 taxa in tumor-adjacent tissue (TA group; e.g., Bradyrhizobiaceae, Weeksellaceae), while tumor tissue (T group) exhibited specific enrichment of 2 genera (Fenollaria, Alloprevotella) within the Bacteroidota phylum. These findings demonstrate distinct microbial community structures between groups, with Random Forest and LEfSe analyses providing mutually reinforcing evidence for potential microbial biomarkers.

Fig. 3.

Random Forest and Lefse analysis for the gut microbiota in CRC patients. A and B represented the results of random forest for potential biomarker at phylum and genus level. C represents the results of Lefse analysis, Species with LDA Score greater than the set value, i.e., statistically different biomarker between groups, were shown in the LDA value distribution histogram

Classification ability of potential biomarkers

Analysis at the phylum level (Fig. 4) indicated moderate diagnostic potential, with biomarkers achieving AUC values of approximately 0.59 (sensitivity range: 0.545–0.571). The performance improved at the genus level, where Afipia and Halomonas were correlated with AUC values exceeding 0.7. Notably, species-level biomarkers exhibited the highest median AUC values among the taxonomic levels analyzed. Specifically, Burkholderia cepacia showed an AUC of 0.893, and P. fluorescens demonstrated an AUC of 0.684. These findings suggest a taxonomic-level-dependent gradient in discriminatory power, although their clinical diagnostic utility requires validation in independent cohorts.

Fig. 4.

ROC analysis for potential biomarkers. A and B represent the results of ROC analysis for phylum/genus level and mainly potential bacteria at species level, respectively

P. fluorescens is enriched in CRC tissues

qPCR validation in 20 paired samples from CRC patients was shown in Fig. 5, the sequences of primers were listed in Table 2: 70% of tumor tissues (14/20) showed significantly higher P. fluorescens abundance than adjacent tissues (P < 0.05), consistent with sequencing data.

Fig. 5.

The RT-qPCR results of CRC tissues. A showed the melt curve of samples tested. B showed the different relative expression of P. fluorescens in 20 CRC tissues. The paired t-test was used to analyze the difference between tumor and tumor-adjacent tissues. P<0.05 marked as *, P<0.01 marked as **, P<0.001 marked as ***

Table 2.

Sequence of primers for RT-qPCR of P. fluorescens and internal control gene

Primer	Forward sequence	Reverse sequence
P. fluorescens	CATCCGTTGACGTACACCTTGTC	TCAGGATCGGGTACAGCAGCAC
GAPDH	GGAGCGAGATCCCTCCAAAT	GGCTGTTGTCATACTTCTCATGG

P. fluorescens has significant promotion in viability and proliferation of CRC cell

The CCK-8 assay (Fig. 6A-B) revealed that co-culture with P. fluorescens at an MOI of 100 significantly enhanced HCT-116 and DLD-1 cell viability than MOI of 0 (P < 0.001), while the EdU assay (Fig. 6C-F) demonstrated concurrent increases in total HCT-116 cell and DLD-1 count and proliferating cell proportion compared to the group without P. fluorescens (P < 0.001, P < 0.01), indicating robust proliferative stimulation in-vitro conditions.

Fig. 6.

A-B showed CCK-8 results of HCT-116 and DLD-1 at different MOI of P. fluorescens. C and E showed the EdU, DAPI and merged picture of HCT-116 and DLD-1 cell with or without P. fluorescens during EdU test. D and F showed the statistical results about the EdU test of HCT-116 and DLD-1 cell line. P<0.01 was marked as **, P<0.001 was marked as ***

Discussion

Our findings provide a nuanced characterization of the microbiota landscape within the tumor microenvironment of CRC and its adjacent mucosa. The absence of significant differences in α-diversity indices (Ace, Chao1, Shannon, Simpson) between tumor (T) and tumor-adjacent (TA) tissues underscores a fundamental similarity in overall microbial richness and evenness at these spatially proximate sites [27, 28]. This observation aligns with the PCoA results demonstrating a lack of distinct clustering between T and TA groups, reinforcing the concept that intra-individual microbial community variation is significantly less pronounced than inter-individual differences [29]. These results support the growing consensus that local niche factors and host-specific variations exert a dominant influence over broad microbial community structure within the colonic mucosa [30–32]. Structural parallels in CRC microbiota may mask functional divergence, as metabolic pathways often exhibit tumor-specific alterations despite conserved community profiles [33]. This paradox suggests strain-level heterogeneity or microenvironmental gene regulation. The spatial continuity of dysbiotic networks supports an “ecological field effect” extending beyond tumor margins [18, 34], analogous to urban ecosystem resilience. Standardized sampling and longitudinal designs are critical to distinguish host-specific effects from true tumor-microbiome associations [35]. Future studies should combine metatranscriptomics and spatial transcriptomics to resolve functional niches.

Despite this overall similarity, our analysis revealed crucial taxonomic stratification. While Firmicutes and Bacteroidetes remained the dominant phyla in both tissues, our data identified specific microorganisms exhibiting differential abundance. The trending differences in Actinobacteria and Fusobacteria at the phylum level, coupled with the significant enrichment of A. faecicola and most notably P. fluorescens in tumor tissues, highlight a selective restructuring within the tumor microenvironment [36, 37]. Critically, the identification of 12 differentially abundant genera between T and TA tissues (e.g., Halomonas, Muricomes, Burkholderia) further substantiates the existence of a distinct microbial signature associated with the tumor niche. The complementary application of Random Forest (identifying biomarkers like Halomonas, Afipia) and LEfSe analyses (revealing taxa enriched in TA, e.g., Bradyrhizobiaceae, and T tissue, e.g., Fenollaria, Alloprevotella) provided convergent, mutually reinforcing evidence for distinct microbial community structures associated with the different tissue types. It is important to note that our discussion centers on P. fluorescens due to the strength of its association in our dataset. This does not preclude the potential relevance of other modestly enriched or depleted taxa (such as A. faecicola), which may be equally important and should be investigated in larger, functionally oriented studies.

The potential diagnostic performance of specific biomarkers, especially at the species level (B. cepacia, AUC = 0.893; P. fluorescens, AUC = 0.684), highlights their potential translational value as non-invasive diagnostic or prognostic tools for CRC. The robust validation of P. fluorescens enrichment in T tissues via qPCR in independent samples (> 70% showing significant increase) solidifies this observation. More importantly, our functional investigations revealed that P. fluorescens significantly enhanced the proliferative capacity of HCT-116 and DLD-1 cells in vitro, as demonstrated by CCK-8 and EdU assays. This observed effect is confined to specific in vitro co-culture conditions and necessitates further in vivo validation. The underlying mechanism for this proliferation enhancement may involve direct stimulation of epithelial cell growth or modulation of cellular pathways, which could, upon future confirmation in more complex models, contribute to our understanding of its potential role in CRC pathogenesis. However, studies indicate that P. fluorescens exhibits inhibitory effects in pancreatic and gastric cancers, yet it is also regarded as a pathogen in inflammatory bowel disease (IBD). Collectively, Pseudomonas species possess unique virulence factors, biofilm formation mechanisms, and drug resistance [38–40], the role of P. fluorescens in cancer pathogenesis requires further investigation [41–43].

This study is significant for identifying P. fluorescens as a novel bacterium enriched in CRC tissues and demonstrating its functional impact on CRC cell proliferation. While historically recognized in environmental settings and occasionally associated with opportunistic infections [39], its specific enrichment in CRC tissue and direct stimulation of cancer cell proliferation represent novel and pathophysiologically relevant findings, suggesting a potential role beyond mere colonization. A meta-analysis identified 29 core microbial species significantly associated with colorectal cancer (CRC) (FDR<1E-5), all enriched in the feces of CRC patients [44], P. fluorescens was not among them, possibly due to factors such as geography, dietary habits, population, patient heterogeneity and sampling methods. Studies have also identified changes in the abundance of Pseudomonas species in cervical and pancreatic cancers, suggesting their potential research value.

Our observation of P. fluorescens enrichment within the CRC tumor microenvironment aligns with the emerging paradigm of tumors as specific ecological niches for select microorganisms [47]. From a symbiotic perspective, this colonization may offer dual advantages. For P. fluorescens, the nutrient-rich, immunosuppressive tumor milieu likely provides a competitive survival advantage [47, 48]. For the tumor, the persistent presence of bacteria could sustain a procarcinogenic, chronic inflammatory state and potentially interfere with antitumor immunity [48]. While our study identifies this association, future functional work is needed to delineate whether the relationship is commensal, parasitic, or even mutualistic in specific contexts, and to investigate if P. fluorescens produces metabolites that directly influence host cell physiology or genomic stability, akin to mechanisms described for other gut bacteria [49]. The enrichment of P. fluorescens within the tumor tissue prompts speculation regarding its potential functional role. Future investigations should explore whether this bacterium influences local inflammatory signaling, contributes to immune evasion, or engages in metabolic competition within the tumor microenvironment, thereby linking its presence to cancer biology [50].

It should be noted that the functional assays in this study were conducted using a commercial Pseudomonas strain (BMZ339565). While this standardized strain allowed for controlled experimentation, future studies employing strains isolated directly from patient tumor samples would be essential to strengthen the clinical relevance and confirm the biological significance of our findings. The sample size included in the overall study is relatively small, potentially introducing sampling bias. The inconsistency between the LEfSe and Random Forest results highlights the inherent challenge of biomarker identification in high-dimensional microbiome data, where methodological differences and potential overfitting may contribute to non-convergent feature selection.

Further expansion of the study sample is necessary to provide additional support for the conclusions. Future research must elucidate the precise molecular mechanisms by which P. fluorescens exerts its pro-tumorigenic effects (e.g., specific virulence factors, metabolites, immune modulation) and validate its abundance and role in larger, geographically diverse cohorts, including correlation with clinical stages and outcomes. Fusobacterium nucleatum modulates disease progression through synergistic and antagonistic interactions with other microbes (e.g., Porphyromonas gingivalis, Clostridioides difficile), influencing inflammatory diseases and malignancies such as colorectal cancer, periodontitis, and atherosclerosis [51, 52]. Furthermore, investigating potential interactions between P. fluorescens and other CRC-associated pathogens (e.g., F. nucleatum) will be crucial to understand the complex microbial dynamics within the tumor niche. Novel research methodologies, such as integrated omics approaches [53], will also represent new directions for future research. Future studies focusing on the detailed genomic and proteomic profiling of these patient-derived P. fluorescens isolates are warranted. Such characterization will be essential to identify specific bacterial virulence factors or metabolic pathways that could serve as novel targets for therapeutic intervention. Research on P. fluorescens may offer potential for non-invasive detection of CRC or targeted interventions on the microbiome.

Conclusions

This study analyzed the differences in gut microbiota between CRC tissues and tumor-adjacent tissues. The results found that the composition of bacteria in the intestine of patients with CRC, the correlation between bacteria, and the function have undergone major changes. The changes in the intestinal microecology may be an important reason for the occurrence of CRC. Specifically, P. fluorescens was identified as a differentially abundant ITM candidate in CRC and was preliminarily validated at the cellular level. However, the generalizability of these findings is constrained by the limited sample size. Future work will expand on these observations by investigating specific intratumoral microorganisms throughout CRC progression, with a focused exploration of the mechanistic role of P. fluorescens in tumor development.

Authors’ contributions

Yongqi Zhang and Weitao Shen: Formal analysis, Investigation, Methodology, Writing–original draft. Qiliu Qian and Nan Li: Methodology, Writing – review & editing. Shiya Zheng, Wenhao Li, Guangxuan Yin and Ruihua Shi: Conceptualization, Methodology. Suna Cha, Mingjun Sun, and Ping Ye: Methodology, Supervision, Writing – review & editing. Qiliu Qian and Mingyue Hu: Conceptualization.

Funding

This work was supported by the National Natural Science Foundation of China (82303959), National Natural Science Foundation of China Youth Foundation (81302162), Open Project of Zhenjiang Traditional Chinese Medicine Spleen and Stomach Disease Clinical Medicine Research Center (No. SSPW2022-KF06), and Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Construction Funds (GSP-ZXY20).

Data availability

The data were deposited into the NCBI database under BioProject number PRJNA1247740 and are available at the following URL: https://www.ncbi.nlm.nih.gov/sra/PRJNA1247740.

Declarations

Ethics approval and consent to participate

The studies involving humans were approved by the Ethics Committee of Zhongda Hospital affiliated to Southeast University (2023ZDSYLL100-P01). The research was carried out in compliance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.