Effect of Thai traditional formulary medicine on gut microbiota and gene expression profiles in breast cancer cell line-derived xenograft mouse models

Preparation of TTFM extraction

The dried natural products and plant materials detailed in Table 1 were bought from a Thai traditional medicine shop (Chao-Krom-Poe Dispensary Pharmacy, Bangkok, Thailand). These materials, along with sulfur, were boiled in the supernatant derived from boiling 2.75 L of lac resin at 80-100°C for 30 min. The resulting mixture was then filtered through sterile voile fabric and centrifuged at 1,610 x g at 25°C for 20 min. The pellet was discarded, while the supernatants were lyophilized and kept at -20°C until use.

High performance liquid chromatography (HPLC) fractionation

In accordance with previously described protocols, HPLC fractionation was employed prior to liquid chromatography with tandem mass spectrometry (LC-MS/MS) analysis. A total of 20 mg of dried TTFM extracts were incubated in 90% methanol at 25°C and shaken at 1,300 rpm for 20 min. Then, the mixture was centrifuged at 17,000 x g at 4°C for 10 min. Reversed-phase HPLC was performed using an Agilent 1200 HPLC device coupled with a 1260 Infinity II photodiode array detector (Agilent Technologies, CA, USA). The solvents for HPLC fractionation included ultrapure water (solvent A) and HPLC-grade acetonitrile (solvent B). The column used was graphitic porous carbon (Hypercarb; 100x2.1 mm; particle size, 3 μm; Thermo Fisher Scientific, MA, USA) maintained at 65±0.8°C. The mobile phase flow rate was 0.2 mL/min with specific gradient elution steps.²⁶ Collected fractions were then analyzed by LC-MS/MS.

Ultra HPLC-MS/MS characterization

A Dionex Ultimate 3000 HPLC coupled with an Orbitrap Q Exactive Focus mass spectrometer (Thermo Fisher Scientific, MA, USA) was employed for untargeted metabolomics studies.²⁶ The Heated Electrospray Ionization source settings included a sheath gas flow rate of 30 arbitrary units, auxiliary gas flow of 10 L/min, spray voltage of 3 kV, capillary temperature of 350°C, S-lens RF level of 60 and auxiliary gas heater temperature of 300°C. Polar metabolites were separated using an Acclaim™ Polar Advantage II column (250x3 mm; particle size, 3 μm; Thermo Fisher Scientific, MA, USA). The mobile phases were as previously described. The LC gradient included: 0-5 min, isocratic elution of 1% B; 5-40 min, gradient elution of 1.5% B/min; 40-47 min, isocratic elution of 100% B (column wash); 47-65 min, isocratic elution of 1% B (reconditioning of the HPLC column).

Differential peak identification was performed using MS-Dial software version 4.90,²⁷ with raw files converted to ABF format via an ABF file converter. In MS-Dial, converted files were processed with default parameters, using public MSP-formatted libraries for both positive and negative ionization modes. MS/MS analysis conditions included a minimum peak height of 1,000 amplitude, m/z search tolerance of 0.01 Da, data acquisition in centroid mode and peak alignment filtering before feature removal based on blank information.

Cell culture

The mouse 4T1 cell line (ATCC^®; cat. no. CRL-2539™), representing stage IV human BC, was cultured in RPMI-1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 4,500 mg/L glucose, 1,500 mg/L sodium bicarbonate, and 1% antibiotic-antimycotic, containing 10,000 units/mL of penicillin, 10,000 μg/mL of streptomycin, and 25 μg/mL of Gibco Amphotericin B (Gibco; Thermo Fisher Scientific, MA, USA). The MDA-MB-231 BC cell line (gift from Ketchart laboratory; Chulalongkorn University) was cultured in DMEM with 10% heat-inactivated FBS and 1% antibiotic-antimycotic (Gibco; Thermo Fisher Scientific, MA, USA). The normal breast cell line EpH4-Ev (ATCC^®; cat. no. CRL-30639™) was used as a control and cultured in DMEM with 10% Calf Bovine Serum (ATCC, VA, USA), 1.2 μg/mL puromycin dihydrochloride (Merck, Darmstadt, Germany), and 1% antibiotic-antimycotic (Gibco; Thermo Fisher Scientific, MA, USA). All cells were maintained at 37°C under 5% CO₂.

Cell viability assay

A cell viability assay using MTT was performed on 4T1, MDA-MB-231 and EpH4-Ev cells. Cells at a total density of 1x10⁴ were seeded in 96-well plates and incubated for 24 h. TTFM extracts were then applied at a concentration range of 0-400 μg/mL. After 5 days of incubation, cells were treated with MTT (Abcam, Cambridge, UK) for 3 h, and the resulting formazan crystals were dissolved in DMSO (Merck, Darmstadt, Germany). Absorbance was measured at 570 nm, and IC₅₀ values were calculated by nonlinear regression (curve fit) using GraphPad Prism (version 9.3.0; Dotmatics).

Apoptosis assay

A total of 1 × 10⁵ cells were seeded in 24 well-microplates and incubated for 24 h, followed by treatment with TTFM extracts for 72 h. The mock cells were treated with PBS, while the positive control cells were treated with the apoptosis inducer, 50 μg of dihydrochloride hydrate (Merck, Darmstadt, Germany). After incubation, cells were harvested, washed with cold PBS and stained with a FITC Annexin V Apoptosis detection kit with propidium iodide (BioLegend, London, UK) for 15 min in the dark. Fluorescent intensity was evaluated using a BD™ LSR II flow cytometer, and data were analyzed with BD FACSDiva™ (version 6.1.3; BD Biosciences, NJ, USA).

RNA preparation and RNA sequencing (RNA-Seq)

Total RNA was extracted from TTFM-treated and -untreated cells using the RNeasy mini kit (Qiagen, Hilden, Germany). The concentration and integrity of RNA were assessed with Qubit RNA assay kits (Invitrogen; Thermo Fisher Scientific, MA, USA), and quality was confirmed using a Bioanalyzer (Agilent Technologies, CA, USA). Library preparation and RNA-Seq were performed commercially by Vishuo Biomedical Pte., Ltd., adhering to standard procedures. A total of 1 gm of total RNA was used for poly(A) mRNA isolation with oligo (dT) beads, followed by mRNA fragmentation and first- and second-strand cDNA synthesis. Adaptors were ligated to the double-stranded cDNA using T-A ligation, followed by size selection. Libraries were validated and quantified before multiplexing and sequencing with a NovaSeq 6000 SP Reagent Kit (version 1.5; Illumina, CA, USA) for paired-end (2 × 150) sequencing.

RNA-Seq data analysis

The RNA-seq data were processed to analyze differentially expressed genes (DEGs).²⁸ Trimmed reads were aligned to the mouse reference genome using HISAT2 v.2.1.0,²⁹ and transcript prevalence was quantified with Cufflinks v.2.2.1.³⁰ Differentially expressed genes were examined using DESeq2 v.1.24.0³¹ with a significance cut-off of p < 0.05. Gene Ontology (GO) and biological pathways were identified using GOSeq v1.34.1³² and the Kyoto Encyclopedia of Genes and Genomes (KEGG).³³

Establishment of cancer cell line-derived xenograft (CLDX) mouse models

A total of 24 six-week-old female BALB/c mice (~18-20 g) purchased from Nomura Siam International, Bangkok, Thailand, were held humanely in pathogen-free conditions (temperature, 21 ± 1°C; humidity, 50 ± 20%; 12 h light/dark cycle) and acclimated for 1 week before the experiment. The study procedures were approved by the Institutional Animal Care and Use Committee of Chulalongkorn University Laboratory Animal Center (approval no. 2173024). All animal experiments were conducted in accordance with ARRIVE guidelines (https://arriveguidelines.org/arrive-guidelines ). The sample size was calculated using the n4Studies program.³⁴ For anesthesia, mice were anesthetized using a mixture of oxygen and isoflurane, at a flow rate of 0.5 L/min and 2-3% isoflurane vaporization. A total of 12 mice were injected with 3 × 10⁵ 4T1 cells into the left mammary fat pad via subcutaneous injection under aseptic techniques.

In vivo cytotoxicity analysis

After a total of 2 weeks post-transplantation, mice were randomized into four groups (n = 6 per group): i) Untreated control (NC); ii) untreated CLDX (MD); iii) non-tumor treated with TTFM; and iv) CLDX treated with TTFM. Mice in the NC and MD groups were treated with a 0.9% sodium chloride solution (normal saline, A.N.B. Laboratories, Bangkok, Thailand) administered via oral gavage three times a week. For TTFM treatment groups, the animal equivalent dose (AED) was determined based on body surface area, using a Km ratio of 12.3, resulting in a TTFM dosage of 205 mg/kg for mice. TTFM, typically 16.67 mg/kg (1,000 mg/60 kg/day) for human cancer treatment, was administered in non-pharmaceutical grade form, dissolved in normal saline, sterilized and given via oral gavage thrice weekly. Treatments followed the protocol outlined in Figure 4A. Fecal samples were collected at weeks 0 and 3, stored in NAPseq™ (Bioentist, Bangkok, Thailand) at -80°C. At the end of the experiment, mice were euthanized via CO₂ inhalation. Tumor tissues were collected, rinsed with PBS, stored in DNA/RNA Shield™ (Zymo Research, CA, USA), snap-frozen in liquid nitrogen and kept at -80°C until use.

Full-length 16S library preparation and sequencing

Total DNA was extracted from mice fecal samples using the ZymoBIOMICS DNA Kit (Zymo Research, CA, USA). The full-length 16S rDNA (V1-V9 regions) of bacteria was amplified using in-house methods for DNA library preparation.³⁵ PCR included 0.25 μM primer (each), 0.2 mM dNTPs, 0.2 U/μL Phusion™ Plus DNA Polymerase (Thermo Fisher Scientific, MA, USA) and 10 ng DNA template. The thermal amplification involved initial denaturation at 98°C for 30 sec, followed by 25 cycles (98°C for 10 sec, 60°C for 10 sec, 72°C for 45 sec) and a final extension at 72°C for 5 min. Subsequently, PCR was performed on each amplicon to attach multiplexing Nanopore barcodes using the PCR Barcoding Expansion 1-96 kit (cat. no. EXP-PBC096; Oxford Nanopore Technologies, Oxford, UK). The thermocyclers were set to: Initial denaturation at 98°C for 30 sec, five cycles of amplification (98°C for 10 sec, 60°C for 10 sec, 72°C for 60 sec), and a final extension at 72°C for 5 min. Following agarose gel electrophoresis and purification with the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany), DNA libraries were quantified using the Qubit™ DNA HS assay kit (Invitrogen; Thermo Fisher Scientific, MA, USA). Barcoded DNA was pooled, purified with 0.5x Agencourt AMPure XP beads (Beckman Coulter, CA, USA), and subjected to adaptor ligation before final library construction with the Sequencing Ligation Kit (cat. no. SQK-LSK112; Oxford Nanopore Technologies, Oxford, UK). The final adaptor-ligated DNA library was loaded onto flow cell R10.4 (cat. no. FLO-MIN112; Oxford Nanopore Technologies, Oxford, UK) and sequenced on MinION Mk1C (Oxford Nanopore Technologies, Oxford, UK).

Bioinformatic analysis of full-length 16S rDNA sequence data

All data were analyzed based on a previous study.³⁵ Briefly, the raw FAST5 data of full-length 16S sequences from nanopore sequencing were basecalled using the super-accuracy mode of guppy basecaller (version 6.1.2; Oxford Nanopore Technologies, Oxford, UK). Sequence quality was checked with MinIONQC. Demultiplexing and adaptor trimming were performed using Porechop (version 0.2.4; https://github.com/rrwick/Porechop). The demultiplexed filtered reads were subsequently processed by NanoCLUST³⁶ for clustering, polishing and taxonomic identification, using the Ribosomal Database Project. The data were analyzed with the QIIME2 pipeline (version 2021.2)³⁷ and visualized using MicrobiomeAnalyst (https://www.microbiomeanalyst.ca).

Gene expression analysis based on nanoString

Total RNA was isolated from tumor tissue using the Qiagen RNeasy Kit (Qiagen, Hilden, Germany). Gene expression profiling was performed with the nCounter^® Mouse PanCancer Pathways Panel (nanoString Technologies, WA, USA), assessing 750 cancer-related genes. RNAs (100 ng) were hybridized and processed per the manufacturer’s protocols. mRNA counts were analyzed using nSolver and normalized against housekeeping genes, with data visualized as a heatmap. Functional enrichment analysis was conducted with g:Profiler software³⁸ and volcano plots were generated using GraphPad Prism (version 9.3.0; Dotmatics).

Statistical analysis

Cell viability was assessed using an unpaired t-test, while the apoptosis assay was evaluated with one-way ANOVA followed by Tukey’s post hoc test. Statistical analysis was performed with GraphPad Prism. p < 0.05 was considered to indicate a statistically significant difference.

Metabolite profiles of TTFM extracts

Metabolite profiling of TTFM extracts was conducted using MS-Dial, integrating mass accuracy, isotope ratios and retention-time prediction. After HPLC purification, the LC-MS/MS analysis identified 302 compounds. Positive ion mode (ESI+) enabled ionization of a larger spectrum of chemicals, while negative ion mode (ESI−) minimized background noise, detecting 270 and 32 compounds, respectively (refer to extended data-Supplementary Table 1 and 2). Notably, several identified metabolites, such as 6-methoxydihydrosanguinarine, astaxanthin, costunolide, kaempferol, hinokitiol, berberine, 1-methylnicotinamide, benzyl isothiocyanate, dexamethasone, vincamine, spironolactone, (9Z)-9-octadecenoic acid, hexanedioic acid, N,N-dimethylglycine, d-limonene, pyridoxine, thymol, ginkgolide A and azelaic acid, are recognized for their anti-cancer properties across diverse cancer types ( Table 2).

Effects of TTFM extracts on BC cell viability

The impact of TTFM extracts on BC cell viability was assessed using an MTT assay. The BC cell lines 4T1 and MDA-MB-231, along with normal breast cells (EpH4-Ev), were treated with varying concentrations of TTFM extracts for 5 days. A dose-response relationship was observed in the 4T1 and MDA-MB-231 cells, with increased extract concentrations associating with decreased cell viability ( Figure 1B and C). By contrast, EpH4-Ev cells demonstrated a high tolerance to TTFM extracts ( Figure 1A). The IC₅₀ values for TTFM extracts against 4T1, MDA-MB-231 and EpH4-Ev cells were determined to be 14.13, 20.59 and 117.30 μg/mL, respectively ( Figure 1D).

Figure 1. Inhibitory effect of TTFM on the proliferation of 4T1, MDA-MB-231 and EpH4-Ev cells.

(A-C) Cell viability was assessed using the MTT assay and (D) the IC₅₀ values for proliferation after 5 days of TTFM treatment are shown. Data are presented as mean ± standard deviation (n=3). * p < 0.01, ** p < 0.0001 (Unpaired t-test). n represents the number of independent experiments performed. TTFM, Thai traditional formulary medicine.

Effects of TTFM extracts on apoptosis

Results showed that TTFM extracts induced apoptosis in BC cells. In the NC group, 84.4% of 4T1 cells were alive, with 7.3% in late apoptosis, 6.8% in early apoptosis and 1.5% dead ( Figure 2A and H). Addition of the apoptosis inducer significantly increased early and late apoptotic cells by 1.98- and 4.7-fold, respectively (p < 0.0001; Figure 2B and H). Low TTFM concentrations, 25 and 50 μg/mL, increased early apoptosis by 1.6- and 1.7-fold, respectively (p < 0.001; Figure 2C-D and H). Higher concentrations, 200 and 400 μg/mL, raised early apoptosis by 2.7- and 6.4-fold, respectively (p < 0.0001; Figure 2E and F). Late apoptosis increased by 1.5-fold at 400 μg/mL (p < 0.0001; Figure 2G and H). TTFM also induced apoptosis in MDA-MB-231 cells with increasing concentrations (Supplementary Figure 1).

Figure 2. Effect of TTFM on 4T1 cell apoptosis, assessed by flow cytometry.

Apoptosis was analyzed after 72 hours of TTFM treatment at the following concentrations: (A) 0 μg/mL, (C) 25 μg/mL, (D) 50 μg/mL, (E) 100 μg/mL, (F) 200 μg/mL and (G) 400 μg/mL. (B) Positive control used 50 μg dihydrochloride hydrate. (H) Quantitative comparison of apoptosis rates. * p < 0.05, ** p < 0.001, *** p < 0.0001 indicate statistical significance differences when compared with mock (one-way ANOVA followed by Tukey’s post hoc test). TTFM, Thai traditional formulary medicine.

Effects of TTFM extracts on BC cell pathways and entire mRNA transcripts

RNA-Seq analysis was conducted to elucidate the effects of TTFM on gene expression in BC cells. Cells were treated with TTFM extracts at a concentration of 25 μg/mL for 72 h, and total RNA was collected in triplicate. Sequencing identified 55,359 protein- and non-protein-coding genes in EpH4-Ev and 4T1 cells. TTFM treatment in EpH4-Ev cells resulted in the significant upregulation of 200 genes and downregulation of 66 genes, while in 4T1 cells, two genes were upregulated and eight genes were downregulated > 2-fold (|log₂fold-change| > 1; p < 0.05) compared with the untreated control ( Figure 3A and B).

Figure 3. Impact of TTFM on RNA expression in EpH4-Ev and 4T1 cells.

(A) Venn diagram of upregulated genes. (B) Venn diagram of downregulated genes. (C) The volcano plot illustrates the significantly upregulated and downregulated genes in 4T1 cells, generated using DESeq2 with a significance cut-off of p < 0.05. * p < 0.05, ** p < 0.01, *** p < 0.001, *** p < 0.0001 indicate statistical significance differences when compared with the 4T1-treated or untreated groups. TTFM, Thai traditional formulary medicine.

These findings indicate that TTFM regulates five known genes in 4T1 BC cells ( Figure 3C; refer to extended data - Supplementary Table 3). TTFM treatment increased the expression of Acyp2, while decreasing the expression of Klf12, Dtnbos, 4930579G18Rik and Txlnb. The significantly DEGs impacted by TTFM were categorized into four major biological processes based on KEGG pathway enrichment analysis, primarily metabolism. Notably, TTFM treatment significantly reduced the expression of Klf12 ( Figure 3C; refer to extended data Supplementary Table 3). On the other hand, the top 5 upregulated genes following TTFM treatment were Dclk1, Slc1a3, Cadm2, Anks1b, and Tmod2, while Slc15a3, 7SK, Mir221, Gbp10, and Slfn4 were downregulated in EpH4-EV cells (refer to extended data Supplementary Table 3).

Effects of TTFM on tumor growth in 4T1 xenografted mice

After 8 weeks of treatment, TTFM had no significant effect on the body weight of normal mice compared with the NC group ( Figure 4B). After 3 weeks of treatment, xenografted mice in the TTFM group had higher body weight than those in the MD group (p = 0.0051; Figure 4C; refer to extended data Supplementary Table 4). TTFM treatment appeared to inhibit tumor growth in 4T1 xenograft mice ( Figure 4D and G; refer to extended data Supplementary Table 4). The tumor weight and volume in the TTFM group were lower than those in the MD group, but the differences were not statistically significant. At the end of the experiment, the tumor volume in the TTFM group was 800.98±195.74 mm³, compared with 1086.77±199.03 mm³ in the MD group ( Figure 4D, E and G). The tumor weight in the TTFM group was 0.95±0.2 g, while in the MD group it was 1.32±0.17 g ( Figure 4F).

Figure 4. Effect of TTFM on antitumor properties in 4T1 xenografted mouse models.

(A) Experimental design. (B) Body weight of normal group. (C) Body weight of CLDX group. (D) Tumor volume. (E) Tumor size, (F) weight and (G) volume at week 3. An unpaired t-test was used for statistical analysis. NC, normal control; MD, model; TTFM, Thai traditional formulary medicine; CLDX, cell line-derived xenograft.

Characteristics of the gut microbiota composition and impact of TTFM

Fecal samples from experimental mice were subjected to full-length 16S rDNA sequencing to investigate the impact of TTFM on gut microbiota diversity and composition. Data quality was validated by rarefaction analysis, confirming adequate sequencing depth (Supplementary Figure 2). α-diversity metrics such as Chao1 and Shannon indices were used to assess community diversity. A total of 2 weeks post-cancer cell transplantation, the CLDX group exhibited significantly higher Chao1 and Shannon indices compared with the normal group ( Figure 5A). However, Bray-Curtis dissimilarity-based PCoA plot analysis showed no significant group-level differences ( Figure 5B).

Figure 5. Gut microbiome profiles.

Fecal samples of normal and CLDX mice collected at week 0: (A) α-diversity analysis (Chao1 and Shannon indexes); (B) β-diversity analysis (Permutational multivariate ANOVA, PERMANOVA); (C) F/B ratio (Mann-Whitney U test); (D) Average relative bacterial abundance at species level (Top 20). Effects of TTFM at week 3: (E) α-diversity analysis (Chao1 and Shannon indexes); (F) F/B ratio (Mann-Whitney U test); (G) Average relative abundance of F at phylum level. (H) Average relative abundance of B at phylum level. (I) Average relative bacterial abundance at species level (Top 20). TTFM, Thai traditional formulary medicine; CLDX, cell line-derived xenograft; F, Firmicutes; B, Bacteroidetes.

At the phylum level, Bacteroidetes (B) and Firmicutes (F) dominated the bacterial composition, with a notable increase in B and decrease in F abundance in the CLDX group compared with the normal group (p = 0.0472; Supplementary Figure 2). The F/B ratio was significantly lower in the CLDX group (p = 0.0351) than in the normal group ( Figure 5C). Species-level analysis ( Figure 5D) indicated that tumor development altered the gut microbiome, with increased abundance of species such as Bacteroides coprophilus (p = 0.0197), Barnesiella intestinihominis, Tannerella forsythia and others, while some species like Barnesiella viscericola, Clostridium xylanolyticum and others decreased, though not significantly between groups.

Following TTFM treatment, α-diversity indices (Chao1 and Shannon) did not significantly change across groups ( Figure 5E), but Bray-Curtis dissimilarity-based PCoA analysis revealed distinct clustering (Supplementary Figure 3). At the phylum level (Supplementary Figure 3), B and F remained dominant. The F/B ratio in TTFM-treated CLDX mice was 2.21-fold higher than in untreated CLDX mice ( Figure 5F), indicating increased F and decreased B prevalence ( Figure 5G-H). At the species level ( Figure 5I), notable changes included increased Butyrivibrio hungatei and decreased Clostridium saccharolyticum abundance in TTFM-treated CLDX mice.

Effects of TTFM on the tumor gene expression profiles

NanoString gene expression analysis was employed to assess the impact of TTFM on tumor-related gene expression in 4T1 xenografted mice. Following sacrifice, tumor tissues were isolated for total RNA extraction and subsequent analysis of 750 tumor-related genes. Expression of DEGs was visualized using volcano plots (log₂fold-change > 1; p < 0.05) compared with untreated controls ( Figure 6A; refer to extended data Supplementary Table 5 and 6). TTFM treatment notably upregulated Pias1 expression ( Figure 6A and B), while the top five downregulated genes included Ccno, IL-1r2, IL-1β, IL-2 and Nkd1 ( Figure 6A and B). Functional pathway analysis using gProfiler revealed enriched biological processes such as regulation of protein metabolic processes, regulation of cell death, IL-1 signaling pathway, positive regulation of protein catabolic processes and positive regulation of the p38MAPK cascade ( Figure 6C and refer to extended data Supplementary Table 7).

Figure 6. Tumor gene expression profiles were assessed using nSolver data analysis software.

(A) Volcano plot of DEGs following TTFM treatment (log₂fold-change > 1; p < 0.05). (B) Heatmap of significantly downregulated (green) and upregulated (red) genes. (C) Functional enrichment analysis was conducted with g:Profiler software. DEGs, differentially expressed genes; TTFM, Thai traditional formulary medicine.