Triphala extract reshapes microbial and host proteomes in an obese descending colon model

2.1 Study approval and participant recruitment

The study was approved by the Ethics Committee of Kasetsart University, Thailand (COA64/068) and registered with the Thai Clinical Trials Registry (TCTR20220204007). The patient, a 25-year-old Thai male (BMI 35.0 kg/m², waist >80 cm; Supplementary Table S1), provided written informed consent.

2.2 Herbal extract preparation

Equal parts of dried T. bellirica, T. chebula, and P. emblica fruits from Songkhla Province, Thailand, were milled, boiled in distilled water, filtered, concentrated, and dried to yield Triphala extract (TPE), following a published method (Kwandee et al., 2025). Extracts were stored at 28 °C in airtight containers. TPE composition analysis showed concordance with a previous report (Kwandee et al., 2025).

2.3 Fecal sample handling and inoculation

Fresh stool was collected in anaerobic jars, transported within 30 min, and blended with PBS to form a 4% (v/v) slurry for inoculation (Kwandee et al., 2025).

2.4 In vitro descending colon simulation

A Biostat® B-DCU bioreactor simulated small intestine (600 mL), ascending colon (500 mL), and descending colon (800 mL) compartments under anaerobic, stirred conditions (37°C). Daily feeding included nutrient medium (600 mL) plus pancreatic juice and ox bile (120 mL each) (Mok et al., 2025). After a 2-week stabilization, TPE (2 g/day) was added to the small intestine compartment for 4 weeks, followed by a 2-week washout (Supplementary Table S2). Samples collected weekly (W0–W6) were centrifuged, washed in PBS, and frozen at −80°C.

2.5 Protein extraction and quantification

Proteins were extracted using SDS-based lysis, deoxycholate/TCA precipitation, acetone washing, and re-solubilization in 0.5% SDS. Concentrations were determined by the Lowry assay. Samples were digested with trypsin (Kruaweangmol et al., 2025).

2.6 Mass spectrometric analysis

Peptides were analyzed in triplicate on an Ultimate3000 LC system coupled to a ZenoTOF 7600 mass spectrometer. Separation was achieved on C18 columns with a 30-min gradient at 0.3 µL/min. Data were acquired in DDA mode with a Top-60 method and Zeno trap enabled (Kruaweangmol et al., 2025).

2.7 Data processing and functional annotation

Raw files were processed in MaxQuant (v2.1.4.0) (Tyanova et al., 2016) against Homo sapiens and bacterial UniProt proteomes (1% FDR; ≥2 peptides). Non-gut species proteins were excluded (Kwandee et al., 2025). Data were log₂-transformed, zeros imputed, and analyzed in MetaboAnalyst 6.0 (Pang et al., 2024). PCA and volcano plots (|FC| ≥ 2, p ≤ 0.05) were generated. Variance-based top 20 proteins were standardized and visualized in Seaborn heatmaps. DEPs were defined as |log₂FC| ≥ 2 with p < 0.05 (Welch’s t-test). GO enrichment identified weekly “Top 5” terms across categories, displayed as stacked bar charts. For host–microbiome interactions, the top 15 DEPs per week were ranked by |log₂FC| × –log₁₀p. Spearman correlations (ρ ≥ 0.6, p < 0.05) between bacterial and host proteins were calculated, visualized as clustered heatmaps. Analyses used Python (pandas, NumPy, SciPy, seaborn, matplotlib).

3.1 Principal component analysis of treatment dynamics

PCA revealed distinct proteomic shifts during TPE supplementation (Supplementary Fig. S1). In the bacterial proteome, baseline (W0) separated from treatment weeks (W1-W4) along PC1 (20.9%), with W3-W4 furthest from baseline. Post-treatment, W5 partially returned toward baseline, while W6 diverged along PC2 (18.1%), suggesting a rebound response. In the human proteome, W0 was distinct from treatment and post-treatment samples. Treatment weeks clustered along PC1 (19.9%), with W3 showing the strongest displacement. Post-treatment responses diverged on PC2 (16.6%), as W5 partly reverted but W6 formed a unique cluster, indicating delayed host responses.

3.2 Volcano plot analysis of differentially expressed proteins

Volcano plots highlighted distinct proteomic changes in both bacterial and human compartments (Fig. 1; Supplementary Tables S3–S4). In the bacterial proteome (Figs. 1a-f), 376-633 proteins were differentially regulated across treatment and post-treatment phases. At W1 vs W0, 181 were downregulated and 222 upregulated; W2 showed 164 down and 224 up; W3, 210 down and 229 up; and W4, the highest upregulation (250 up vs. 187 down). Post-treatment, W5 had 243 down and 219 up, while W6 showed the strongest imbalance (281 down vs. 218 up), suggesting continued microbial shifts after withdrawal. The human proteome (Figs. 1g-l) exhibited broader responses, with 4,993-5,798 proteins detected per comparison. At W1, 1,759 were downregulated and 1,854 upregulated; W2 favored upregulation (2,028 up vs. 1,562 down); W3, more downregulation (2,030 down vs. 1,780 up); W4 was balanced (1,989 down vs. 1,913 up); W5 again showed predominance of downregulation (2,084 down vs 1,741 up); and W6 had the most pronounced downregulation (2,539 down vs. 1,593 up, with 1,666 insignificant). Together, these results indicate that TPE supplementation induced moderate but dynamic bacterial shifts with rebound post-treatment, while the host proteome showed large-scale and sustained remodeling, dominated by downregulation in later phases, especially W6.

Close

Fig. 1.

Volcano plots showing differentially abundant proteins compared with baseline (W0) in (a-f) bacterial and (g-l) host proteomes across treatment and post-treatment weeks. Proteins significantly increased (red) or decreased (blue) were identified using thresholds of FDR < 0.05 and |log₂FC| > 2.

Export to PPT

3.3 Temporal dynamics of the most variable proteins

The 20 most variable proteins (Supplementary Tables S5-S6) were visualized in a heatmap (Fig. 2). In the bacterial proteome (Fig. 2a), highly variable proteins included metabolic enzymes (phosphoglycerate kinase, tryptophan synthase subunit beta, methionyl-tRNA formyltransferase) and stress proteins (GroEL, GrpE), all upregulated during treatment (W1-W4). After withdrawal (W5-W6), proteins such as RecX and D-erythrose-4-phosphate dehydrogenase increased, indicating microbial adaptation and reprogramming.

Close

Fig. 2.

Heatmaps presenting the top 20 most variable proteins in (a) bacterial and (b) human proteomes. Week 0 (W0) indicates the pre-treatment control, W1-W4 represent the TPE treatment phase, and W5-W6 indicate the post-treatment period.

Export to PPT

In the human proteome (Fig. 2b), variable proteins comprised structural and regulatory proteins (prelamin-A/C, CCN1, exocyst component 6) and metabolic enzymes (hydroxyacylglutathione hydrolase, lipid-associated enzymes). Treatment generally upregulated cytoskeletal, transcriptional, and lipid regulators, while post-treatment showed rebound expression, notably RGS14 and F-box protein 46.Overall, Triphala supplementation induced distinct temporal remodeling, with microbial adaptation focused on amino acid/nucleotide metabolism and stress responses, and host modulation on lipid metabolism, structural organization, and signaling.

3.4 Functional classification analysis

Functional distributions of bacterial and human proteomes have been shown in Fig. 3. In the bacterial proteome (Fig. 3a), amino acid metabolism fluctuated, peaking at W1–W2 (∼9.5%) and W5 (∼7.9%), while carbohydrate metabolism (≈5-8%) reached its maximum post-treatment (W6: 8.4%). Translation contributed 7-10% (highest at W3–W4), and nucleotide metabolism varied between 4-6%. Cofactor/vitamin metabolism appeared only at W1 (∼2.0%). Cell motility & cytoskeleton proteins were seen early (W1-W3: 2-3%), then disappeared. Defense/stress response dropped during treatment but reemerged at W5 (∼4.6%), suggesting renewed microbial stress adaptation. In the human proteome (Fig. 3b), transcription dominated (≈14-16%) and remained stable throughout. Transport (≈8-9%) increased slightly at W1-W2 and W6. Folding/sorting/degradation, signaling, and nucleotide metabolism each contributed ≈6–7%, while motility/cytoskeleton and amino acid metabolism remained ≈5–6%. Stress response proteins appeared only at W2 (∼2.1%) and W4 (∼2.0%), reflecting transient activation. Overall, the human proteome was stable, dominated by transcription and broad metabolic functions, whereas the bacterial proteome showed marked temporal variability, particularly in amino acid and carbohydrate metabolism, with transient vitamin, motility, and stress-related functions.

Close

Fig. 3.

Functional categorization of identified proteins from (a) bacterial and (b) human proteomes across different study phases. Week 0 (W0) represents the pre-treatment control, W1-W4 correspond to the treatment phase, and W5-W6 indicate the post-treatment phase.

Export to PPT

3.5 Alterations in differentially expressed proteins across functional categories in bacterial and human proteomes

DEPs were identified by comparing each treatment and post-treatment week (W1-W6) to baseline (W0) (Fig. 4). In bacterial proteome (Figs. 4a-f), responses were gradual: W1-W2 were dominated by downregulation in transport and transcription, W3-W4 showed a balanced pattern, and recovery emerged post-treatment, with W5-W6 marked by upregulation, especially in transport (272 proteins) and amino acid metabolism (261 proteins). In the human proteome (Figs. 4g-l), early treatment (W1-W2) showed balanced or transient suppression, while W3-W6 displayed predominant upregulation, strongest at W6. Functional categorization revealed transcription (5,096 proteins), signaling (4,579), transport (3,389), and folding/degradation (1,804) as most responsive, shifting from early downregulation to sustained activation. Thus, the host proteome underwent robust, persistent upregulation, whereas the bacterial proteome exhibited a delayed but distinct rebound after withdrawal, reflecting compensatory remodeling.

Close

Fig. 4.

Distribution of DEPs by functional categories in bacterial and human proteomes. (a–f) show bacterial proteome comparisons (W1-W6 vs. Week 0), while (g–l) present human proteome comparisons (W1-W6 vs. Week 0).

Export to PPT

3.6 Top enriched Gene Ontology categories in bacterial and human proteomes

Gene ontology (GO) enrichment analysis revealed distinct bacterial and human responses (Supplementary Fig. S2). In the bacterial proteome (Supplementary Fig. S2A), enriched biological processes included proteolysis (110 proteins), cell wall organization (88), and antibiotic biosynthesis (55), with cellular components dominated by plasma membrane (507), cytoplasm (441), and cytosol (315). Molecular functions emphasized ATP binding (380), metal ion binding (292), and DNA binding (194), reflecting stress adaptation and energy metabolism. In the human proteome (Supplementary Fig. S2B), biological processes were dominated by transcriptional regulation and signaling, regulation of transcription by RNA polymerase II (1,964 proteins), positive regulation (1,373), and signal transduction (1,189). Cellular components were enriched in the nucleus (7,545), cytosol (7,353), and cytoplasm (6,797). Molecular functions were led by metal ion binding (3,573), ATP binding (2,566), and RNA binding (2,176), consistent with widespread transcriptional and metabolic reprogramming. Overall, human proteomes showed strong nuclear and signaling activation, while bacterial responses reflected localized adaptations in cell wall remodeling, proteolysis, and energy metabolism.

3.7 Host-microbiome proteomic correlations during TPE supplementation

Correlation analysis revealed significant cross-talk between bacterial and human proteins during TPE supplementation (Fig. 5; Supplementary Table S7). Positive associations linked bacterial metabolic enzymes (e.g., ribonuclease Z, acyl carrier protein phosphodiesterase, NAD-specific glutamate dehydrogenase) with host proteins involved in metabolism, signaling, and stress responses (adipolin, prostaglandin E synthase, annexin A3, activin receptor type-1) (r = 0.40–0.76, p < 0.05). In contrast, negative correlations connected bacterial nucleotide or stress-related enzymes (Poly(A) polymerase I, Cardiolipin synthase 2, cis-aconitate decarboxylase) with host structural and nuclear proteins (annexin A3, reticulon-4) (r = –0.31 to –0.52, p < 0.05). Overall, these patterns indicate that TPE supplementation not only altered microbial and host proteomes independently but also reshaped their interdependent interactions, with positive links aligning microbial metabolism to host transcriptional regulation and negative ones reflecting counterbalanced structural and nuclear functions.

Close

Fig. 5.

Correlation analysis depicting host–microbiome interactions based on proteomic profiles.

Export to PPT

4.1 PCA and volcano plots reveal distinct remodeling

PCA showed distinct temporal shifts in both microbial and host proteomes during TPE supplementation, with rebound responses after withdrawal. The strongest divergence appeared mid-supplementation, while withdrawal samples clustered separately, indicating sustained remodeling. Such rebound patterns reflect “ecological memory,” where prior exposures shape later responses, as seen in human and artificial-gut studies. Washout phases in colon simulators often show partial rather than full reversion, consistent with recent polyphenol-fiber intervention findings (Letourneau et al., 2022; Reider et al., 2022). Volcano plots showed moderate microbial DEPs with balanced regulation during treatment but strong imbalance after withdrawal, suggesting renewed stress. The host proteome, with thousands of DEPs, exhibited broader remodeling, marked by late downregulation of regulatory proteins. Together, these patterns highlight flexible microbial adaptation versus sustained host reprogramming.

4.2 Microbial functional responses

Microbial proteomes were enriched for amino acid and nucleotide metabolism, reflecting the nutrient landscape of the descending colon, where residual proteins and host-derived glycans dominate (Muegge et al., 2011). This reliance on nitrogen-rich substrates is associated with proteolytic fermentation products such as branched-chain fatty acids, phenols, and indoles, which may impose stress on both microbes and host cells. During TPE supplementation, however, microbial profiles shifted toward increased translation and carbohydrate metabolism, suggesting partial redirection away from proteolysis toward saccharolytic activity. Polyphenol-rich substrates in Triphala may thus provide alternative fermentative routes, alleviating the need for protein catabolism. This interpretation agrees with recent studies showing that polyphenol-containing botanicals reduce proteolytic metabolites while enhancing SCFA production, including butyrate and acetate, which support epithelial function (Rodríguez-Daza et al., 2021; Whitman et al., 2024; Kumar et al., 2025). The concurrent decline in stress-associated proteins during supplementation suggests reduced microbial stress burden, possibly linked to lower proteolytic byproducts or improved cross-feeding among saccharolytic taxa. Their reappearance during withdrawal underscores the resilience and adaptive flexibility of distal gut microbes to changing dietary conditions. Collectively, these results suggest that TPE acts as both a metabolic substrate and an ecological modulator, rebalancing fermentation toward more favorable functional outputs.

4.3 Host functional responses

Host proteomic changes were pronounced, with dominant enrichment in transcriptional regulation, signaling, and cytoskeletal remodeling. During supplementation, upregulation of mitochondrial and lipid metabolism proteins suggests enhanced epithelial energy utilization, which may reflect adaptation to a remodeled metabolite environment characterized by increased SCFAs, reduced proteolytic products, and altered bile acid pools. Polyphenol-derived metabolites can modulate mitochondrial activity and lipid handling, thereby improving epithelial energy efficiency and redox balance (Gurjar et al., 2025; Maksimović et al., 2025). In later stages (W5-W6), suppression of host proteins became predominant, indicating a compensatory dampening of proliferative and stress-associated pathways after supplementation ceased. Such rebound effects have been observed in in vitro gut models where phytochemical withdrawal leads to re-exposure to proteolytic metabolites, triggering epithelial readjustment (Whitman et al., 2024). These temporal dynamics are consistent with mounting evidence that microbial metabolites directly regulate host transcription and signaling. SCFAs such as butyrate and propionate activate G-protein coupled receptors (GPCRs; FFAR2/3) to influence barrier integrity and immune signaling (Parada Venegas et al., 2019; Lee et al., 2024). In parallel, butyrate acts as a histone deacetylase (HDAC) inhibitor, inducing epigenetic reprogramming of host epithelial and immune genes (Liu et al., 2023; Wang et al., 2024). Moreover, bile acids and indolic compounds derived from proteolysis engage nuclear receptors such as FXR and AhR, further shaping host transcriptional networks and inflammatory tone (Cheng et al., 2024; Ma et al., 2024). Taken together, these results highlight that Triphala supplementation not only modulates microbial outputs but also reprograms host epithelial functions through metabolite-mediated signaling and epigenetic pathways, with implications for barrier function and metabolic homeostasis in obesity.

4.4 Functional enrichment and compartmental priorities

Functional enrichment analyses revealed clear compartmental priorities. Microbial proteins were enriched in proteolysis, cell wall remodeling, and energy metabolism, consistent with reliance on residual proteins and host mucins as dominant substrates in the descending colon. The persistence of proteolytic functions during supplementation highlights the intrinsic nitrogen-rich environment, where saccharolysis can be stimulated but not fully replace proteolysis. Such activity generates branched-chain fatty acids, phenols, and indoles, compounds often linked to epithelial stress in obesity-associated gut ecology (Rowland et al., 2018; Mancabelli et al., 2024).

In contrast, host proteomes were dominated by nuclear and cytosolic reprogramming, reflecting broad transcriptional and signaling adjustments rather than direct metabolic shifts. This divergence mirrors findings from recent metaproteomic studies, where microbes exhibit substrate-adaptive functions while host cells primarily modulate barrier integrity and regulatory pathways (Sun et al., 2024; Valdés-Mas et al., 2025). Collectively, these results suggest that TPE supplementation accentuates distinct but interdependent microbial and host functions, rebalancing colonic homeostasis under obesity-linked conditions.

4.5 Host-microbiome interactions

Correlation network analyses highlighted that TPE modulates not only microbial and host proteomes individually, but also the functional dialogue between them. Both cooperative and antagonistic associations were detected, underscoring the complex interplay between microbial metabolism and host regulation. Positive correlations, particularly between bacterial enzymes involved in carbohydrate and amino acid metabolism and host signaling proteins, suggest that microbial metabolites such as SCFAs, vitamins, and indole derivatives can act as signaling molecules to stimulate host transcriptional and immune pathways. Such cooperative dynamics are consistent with studies showing that microbial fermentation products engage epithelial GPCRs and nuclear receptors to regulate barrier integrity, inflammatory tone, and metabolic homeostasis (Parada Venegas et al., 2019). Conversely, negative correlations involving microbial stress proteins and host cytoskeletal or nuclear regulators point to potential trade-offs, where microbial adaptation to oxidative or proteolytic stress may impose strain on host epithelial stability. These antagonistic associations may reflect situations in which microbial stress metabolites, including phenols or secondary bile acids, challenge host barrier function and trigger compensatory transcriptional reprogramming (Shi et al., 2023; Ma et al., 2024). Together, these findings show that TPE reshapes both microbial and host proteomes and recalibrates their interactions, supporting the view of host-microbiome communication as a dynamic, bidirectional system where dietary botanicals can enhance resilience through cooperative regulation (Ullah et al., 2024).

The obese donor background is relevant, as obesity involves heightened proteolysis, altered bile acid metabolism, and impaired barrier function (Turnbaugh et al., 2009; Portincasa et al., 2022). Proteomic data show that TPE partly countered these effects by enhancing carbohydrate metabolism and reducing microbial stress, though rebound after withdrawal suggests benefits may require sustained use. These results extend prior microbiome and metabolome studies by uncovering protein-level mechanisms in the distal colon (Gurjar et al., 2025).

4.6 Strengths, limitations, and implications

A major strength of this study is its focus on the descending colon, an underexplored site compared to proximal gut regions. Using time-resolved metaproteomics, we showed how TPE modulates microbial metabolism and host transcriptional responses in a proteolysis-dominated, obesity-associated environment. Integration of microbial and host proteomes revealed cooperative and antagonistic interactions not captured by microbiome or metabolome profiling alone. Limitations include reliance on a single obese donor and in vitro simulations that cannot fully model systemic responses. Future studies should validate these findings in multiple donors and integrate metaproteomics with metabolomics and metagenomics for a more complete view of distal colon ecology. Although metaproteomics indicated a proteolysis-dominant context (proteolysis/AA-metabolism enrichment), indole, branched-chain fatty acids, and sulfide were not directly quantified; targeted metabolomics of these markers will be incorporated in future work.

Overall, TPE exerted region-specific effects by shifting microbial activity toward saccharolysis, reducing stress proteins, and modulating host pathways linked to energy metabolism and transcription. These adaptations may help counteract the proteolytic burden of the distal colon, highlighting TPE’s potential as a targeted dietary strategy to enhance gut resilience under obesity-related conditions.