Berberine from Tibetan Medicine Dracocephalum tanguticum Maxim Suppresses Colorectal Tumor Growth and Inhibits CD8⁺ T Cells Ferroptosis via the NRF2–SLC7A11–GPX4 Axis

Abstract

Objective

This study aimed to identify berberine as the candidate active constituent of Dracocephalum tanguticum Maxim (D. tanguticum) and to investigate whether it suppresses colorectal tumor growth by inhibiting ferroptosis in cytotoxic T lymphocytes via the NRF2–SLC7A11–GPX4 axis.

Methods

The antitumor activity of D. tanguticum ethanol extract (DME) was evaluated in CT-26 murine colorectal cancer cells using CCK-8, wound healing, and colony formation assays. Flow cytometry was used to determine the proportion of CD8+ T cells in tumor-bearing mice and to assess exhaustion markers on tumor-infiltrating CD8+ T cells. Transcriptomic sequencing and molecular docking were performed to identify the interaction between berberine (BBR), tentatively identified as a candidate active constituent of DME, and GPX4, and the effect of BBR on erastin-induced ferroptosis in CD8⁺ T cells was subsequently validated.

Results

BBR, a putatively annotated candidate metabolite putatively annotated from DME, attenuated ferroptosis in tumor-infiltrating CD8⁺ T cells, as confirmed by rescue assays using ferroptosis inducers. Accordingly, BBR treatment increased the proportion of functional CD8⁺ T cells while reducing exhausted T-cell populations in both in vitro and in vivo models. Mechanistically, BBR activated the NRF2–SLC7A11–GPX4 axis, leading to GPX4 upregulation and ferroptosis suppression in CD8⁺ T cells.

Conclusion

Berberine, a candidate metabolite derived from D. tanguticum, enhances CD8+T cells function and suppresses ferroptosis via the NRF2–SLC7A11–GPX4 pathway, supporting its further development as a novel immunotherapeutic agent for colorectal cancer.

Introduction

Colorectal cancer (CRC) remains one of the most prevalent and lethal malignancies worldwide, ranking as the second most commonly diagnosed cancer and the fourth leading cause of cancer-related deaths in China.^1–3 CRC carcinogenesis primarily occurs via three distinct pathways: the adenoma-carcinoma sequence, the serrated pathway, and inflammation-associated tumorigenesis.¹ Although immunotherapy has transformed cancer treatment, its clinical efficacy in CRC remains limited by primary or acquired resistance, immune-related toxicity, and heterogeneous clinical responses.^4,5 A central mechanism underlying these limitations is the dysfunctional state of CD8⁺ T cells within the tumor microenvironment (TME). As the core effector cells responsible for direct tumor cell killing, CD8⁺ T cells frequently undergo functional exhaustion, characterized by sustained expression of inhibitory receptors like PD-1 and LAG-3, which impairs immune surveillance.^6,7 Furthermore, beyond exhaustion, CD8⁺ T cells in the TME are also vulnerable to ferroptosis, an iron-dependent form of regulated cell death driven by lipid peroxidation^8,9 demonstrated that inhibiting CD36-mediated ferroptosis in CD8⁺ T cells activates immunotherapeutic efficacy.¹⁰ Notably, CTLs represent the cytotoxic effector subset of CD8⁺ T cells, and the abundance and functional integrity of CTLs are closely associated with CD8⁺ T-cell-mediated antitumor immunity. Therefore, developing novel strategies to protect CD8⁺ T cells from exhaustion and ferroptosis represents a promising avenue to improve immunotherapy outcomes in CRC.

The Qinghai-Tibet Plateau, regarded as the cradle of Tibetan medicine, is an important source of traditional remedies for disease prevention and treatment. Dracocephalum tanguticum Maxim (D. tanguticum), known as “Zhiyangge” in Tibetan medicine, is traditionally used to treat gastrointestinal disorders, including abdominal pain and dyspepsia. This use is formally documented in modern authoritative pharmacopeias, such as the Tibetan Medicine Standard issued by the Chinese Ministry of Health and the Tibetan Volume of Zhonghua Bencao (State Administration of Traditional Chinese Medicine, 1999). Against this background, we investigated the antitumor mechanism of DME in colorectal cancer. According to the Encyclopedia of Traditional Chinese Medicine, the recommended clinical dose of D. tanguticum is 9 to 15 g per day. D. tanguticum has been reported to contain various bioactive constituents, including flavonoids and terpenoids, which may contribute to its antioxidant, anti-inflammatory, and immunomodulatory activities.^11–13 Its antioxidant activity has been attributed to the suppression of ROS-induced oxidative damage. Accumulating evidence indicates that flavonoids exert anticancer effects by inducing apoptosis, causing cell-cycle arrest, and inhibiting inflammatory signaling pathways, particularly NF-κB-related cascades. In addition, several bioactive compounds have been reported to modulate the tumor microenvironment and immune evasion, suggesting that their antitumor effects may also involve immunomodulatory mechanisms.^14–16 Based on the anti-inflammatory activity of D. tanguticum, the anticancer potential of its flavonoid constituents, and its antioxidant activity attributed to the suppression of ROS-induced oxidative damage, we hypothesized for the first time that D. tanguticum may suppress colorectal cancer progression by inhibiting ferroptosis and thereby preserving CD8⁺ T-cell viability and function. To test this hypothesis, we investigated whether DME directly inhibits CT-26 cell growth and whether it modulates CD8⁺ T-cell function in the tumor microenvironment. However, although several constituents of D. tanguticum have been reported, the chemical profile of the ethanol extract used in this study had not been fully characterized, and the bioactive constituent responsible for its anti-CRC activity remained unclear.

Therefore, this study aimed to screen the ethanol extract of D. tanguticum for antitumor activity, to characterize its chemical profile by LC-MS/MS for the tentative annotation of candidate constituents, and to investigate whether BBR, a putatively annotated constituent, could suppress colorectal cancer progression with concomitant regulation of CD8⁺T cells exhaustion and ferroptosis-related phenotypes through the NRF2–SLC7A11–GPX4 axis. Material and methods.

Preparation of D. tanguticum Extract

The botanical drug Dracocephalum tanguticum Maxim (Lamiaceae; whole plant) was purchased from a licensed commercial supplier (Qinghai Tibetan Medical Hospital, Xining City, Qinghai Province, China; Product Standard Number: GHT1091). The identity of the plant material was authenticated by [Hangjian Testing Technology Co., Ltd. (Beijing, China)] based on macroscopic and microscopic characteristics. A copy of the authentication report is provided in the Supplementary Material (Report No. BAT202512105376083756). The dried whole plant of Dracocephalum tanguticum Maxim (D. tanguticum) was mechanically ground using a grinder, passed through a 400-mesh sieve. The powder was combined with 70% ethanol at a solid-to-solvent ratio of 1:12 (w/v), followed by ultrasonication for 30 minutes and a 3-hour incubation. The mixture was then agitated overnight at 37 °C and 120 rpm using an orbital shaker (Shanghai Zhichu Instrument Co., Ltd., China). To improve extraction efficiency, the mixture was further processed by reflux boiling for 2 hours. After cooling, the extract was centrifuged at 10,000 × g for 10 minutes. The resulting supernatant was collected and concentrated using a rotary evaporator (Shanghai Ailang Instrument Co., Ltd., China) to yield the D. tanguticum ethanol extract (DME). The extraction yield was approximately 14% (w/w), corresponding to 14 g of dry DME per 100 g of dried plant material. For in vitro experiments, DME was initially dissolved in 50% DMSO as a stock solution and then diluted with cell culture medium to a final DMSO concentration of 0.5%, ensuring minimal solvent-related cytotoxicity.

Preparation of Blood Samples

Twelve male Wistar rats (body weight 180±20 g) were obtained from Xi’an Tianyuan Laboratory Animal Centre (China) and randomly divided into a control group (n=6) and a DME-treated group (n=6). Following a 7-day acclimatisation period (free access to food and water), animals received oral administration of DME (624mg/kg body weight) for 7 consecutive days. On day 7, blood samples were collected from the abdominal aorta 1 h after the final administration under anesthesia. Blood samples were subsequently collected from the abdominal aorta into sterile tubes and centrifuged at 6000×g for 15 minutes. Serum was immediately separated and stored at −80°C. Samples were thawed before analysis.

Cell Culture and Cell Coculture System

The CT-26 and CTL cell lines were obtained from Servicebio (Wuhan, China; catalog no. STCC20042P-1).CT-26 cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin purchased from Procell (Wuhan, China; catalog no. PM150110B).CTL cells were maintained in CTLL-2-specific culture medium. All cells were incubated at 37°C in a humidified atmosphere with 5% CO₂.

Cell Viability Assay

Cell viability was evaluated with the Cell Counting Kit-8 (CCK-8; MedChemExpress, USA). CT-26 cells were plated in 96-well plates at a density of 1 × 10⁴ cells per well and allowed to adhere overnight. The cells were then exposed to DME at final concentrations of 0, 50, 100, 150, 200, 250, 300, and 350 μg/mL. Following 24 hours of treatment, 100 μL of CCK-8 working solution was added to each well, and the plates were incubated at 37 °C for 1–4 hours. Absorbance was measured at 450 nm using a microplate reader (BMG Labtech, Germany).

The half-maximal inhibitory concentration (IC₅₀) was calculated using GraphPad Prism software (version 9.0). Cell viability data (expressed as percentage of control) were fitted by nonlinear regression to a log(inhibitor) versus response model with variable slope (four parameters). The IC₅₀ value and the corresponding 95% confidence interval were derived from the fitted curve.

The viability of CD8⁺ T cells was also evaluated using the cell viability assay. CD8⁺ T cells were seeded in 96-well plates at 2 × 10⁴ cells per well and stimulated with 20 µM Erastin (a ferroptosis inducer) for 24 hours. Cells were subsequently treated with different concentrations of either DME or BBR. After 24 hours of treatment, 100 μL of CCK-8 working solution was added to each well, followed by incubation at 37°C for 1–4 hours. Absorbance at 450 nm was measured using a microplate reader.

Colony Formation Assays

CT-26 cells were seeded as a single-cell suspension in 6-well plates at 184–500 cells per well and allowed to adhere overnight. The cells were then treated with DME (184 μg/mL) for 24 h. After treatment, the medium was replaced with fresh drug-free complete medium, and the cells were cultured for an additional 10–14 days, with medium replacement every two days. The assay was concluded when visible colonies had formed in the control wells. Cells were washed with phosphate-buffered saline (PBS), fixed with ethanol for 15–30 min, and stained with crystal violet for 20 min in the dark. After a final PBS wash, colonies were imaged and quantified under an optical microscope (Olympus, Japan).

Wound-Healing Migration Assay

Cell migration was assessed using a wound healing assay. CT-26 cells were seeded in 6-well plates at a density of 2.5×10⁵ cells per well and cultured until reaching approximately 80% confluency. A uniform scratch was introduced into the monolayer using a sterile 10 μL pipette tip. The medium was replaced with fresh medium containing DME (184 μg/mL), and cells were incubated for 0, 6, 12, 24, and 48 hours. At each time point, cells were washed, fixed, and imaged under an optical microscope (Olympus, Japan). Migration was quantified using ImageJ software.

Preparation of DME Samples

For component analysis, 100 mg of DME was dissolved in 1 mL of ultrapure water, mixed with 3 mL of ethanol, and sonicated for 10 minutes. After storage at 4°C for 12 hours, the mixture was centrifuged at 4,000 × g for 10 minutes at 4 °C. The supernatant was dried under nitrogen, reconstituted in 1 mL ultrapure water, and filtered through a 0.22 μm membrane. A 10 μL aliquot was analyzed by LC–MS/MS using a SHIMADZU-LC30 UHPLC system with an ACQUITY UPLC® HSS T3 column (2.1 × 100 mm, 1.8 μm; Waters, USA) maintained at 40 °C. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B), delivered at 0.3 mL/min with the following gradient: 0–1 min, 0% B; 1–2 min, 0–30% B; 2–12 min, 30–50% B; 12–21 min, 50–100% B; 21–26 min, 100% B; 26–26.1 min, 100–0% B; 26.1–30 min, 0% B. Mass spectrometry was performed on a TripleTOF 6600 system (AB Sciex) using electrospray ionization in both positive and negative modes.

Preparation of Serum Samples

Serum aliquots (100 μL) were mixed with 400 μL of ice-cold methanol (1:4, v/v). The mixture was vortexed for 1 min, followed by ultrasonication in an ice bath for 20 min. After incubation at −20°C for 1 h, the samples were centrifuged at 16,000 × g for 20 min at 4°C. The supernatant was collected and lyophilized. The dried residue was reconstituted in 100 μL of pre-cooled 50% methanol, followed by centrifugation at 20,000 × g for 15 min at 4°C. The resulting supernatant (50 μL) was stored at −80°C until analysis. For LC-MS analysis, 10 μL of each processed sample was injected.

LC-MS/MS Analysis

Chromatographic Conditions

All samples were held at 4°C in the autosampler before injection. Chromatographic separation was carried out using a SHIMADZU LC-30 UHPLC system equipped with an ACQUITY UPLC® HSS T3 column (2.1 × 100 mm, 1.8 μm; Waters, USA) maintained at 40°C. The mobile phase comprised solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile), delivered at a flow rate of 0.3 mL/min. T The gradient elution program was set as follows: 0–1 min, 0% B; 1–2 min, 0–30% B; 2–12 min, 30–50% B; 12–21 min, 50–100% B; 21–26 min, 100% B; 26–26.1 min, 100–0% B; and 26.1–30 min, 0% B for column re-equilibration. Detailed ion source parameters are provided in Supplementary Table 1.

Mass Spectrometry Conditions

Mass detection was performed using a TripleTOF 6600 system (AB Sciex) equipped with an electrospray ionization (ESI) source operating in both positive and negative ion modes. The ion source parameters were as follows: ion spray voltage ±5500 V (positive) and –4500 V (negative), temperature 500°C, ion source gases 1 and 2 at 60 psi, curtain gas at 45 psi, and declustering potential at 60 V. Full-scan TOF MS data were acquired across the *m/z* range of 90–1500 with a collision energy of 10 V. Information-dependent acquisition (IDA) was employed to trigger MS/MS scans for the 18 most abundant ions exceeding 100 cps. The MS/MS scans covered a product ion range of *m/z* 50–1500 with a collision energy of 40 V and a collision energy spread of 20 V. Dynamic background subtraction was applied throughout the 30-minute acquisition.

Data Processing and Statistical Analysis

Raw mass spectrometry data were processed with MS-DIAL software for peak alignment, retention time correction, and peak area extraction. This untargeted LC-MS/MS screening was specifically used to putatively annotate prototype constituents absorbed into serum after oral administration. Metabolite putative annotation was performed using the following criteria: precursor ion mass tolerance < 0.01 Da, MS/MS fragment mass tolerance < 0.02 Da, and a minimum MS/MS spectral matching score > 70%, putatively annotated metabolites were categorized by superclass and class. Pathway enrichment analysis was then conducted based on KEGG IDs with Fisher’s exact test (p < 0.05). All statistical analyses and graphical outputs, including pathway enrichment plots and metabolite classification charts, were generated using R (version x.y.z), primarily with the ggplot2 package and custom scripts. Representative base peak chromatograms of crude DME and serum samples from DME-treated rats acquired in positive and negative ion modes are shown in Supplementary Figure 1A–D. The proportional distributions of chemical classes in crude DME and serum samples, as well as representative MS/MS spectra used for metabolite annotation, are presented in Supplementary Figure 2A–C.

Animal Modeling and Drug Administration

Male BALB/c mice (6–8 weeks old) were obtained from the Experimental Animal Center of Xi’an, China. All animal procedures were performed in accordance with the ARRIVE 2.0 guidelines and the Guide for the Care and Use of Laboratory Animals, and were approved by the Institutional Animal Care and Use Committee of Qinghai University (approval No. PJ202502-03). A colorectal cancer (CRC) model was established by subcutaneous injection of 3 × 10⁶ CT-26 cells into the right flank of each mouse. Five days post-inoculation, tumor-bearing mice were randomly assigned to groups based on body weight (n = 5 mice per group). Treatment groups received oral administration of the D. tanguticum ethanol extract (DME) at low (156 mg/kg), medium (312 mg/kg), and high (624 mg/kg) doses. These values represent the dry extract weight per kilogram of body weight. Based on the measured extraction yield of 14% (w/w), the high dose of 624 mg DME/kg corresponds to approximately 4.46 g of crude dried plant material per kg body weight per day. In a parallel experiment designed to evaluate the contribution of the key candidate metabolite BBR, an additional group of tumor-bearing mice received oral administration of BBR (50 mg/kg).A positive control group received 5-fluorouracil (5-FU; 20 mg/kg, ig)., while the control group received saline. Although the equivalent crude drug dose exceeds the 1 g/kg/day benchmark recommended in the “Four Pillars of Best Practice” for well-characterized extracts, it was selected as an exploratory regimen during early-stage development based on preliminary efficacy screening and was well tolerated, with no acute toxicity observed (see Results 3.4). Tumor volume and body weight were recorded every three days, with tumor volume calculated as 0.5 × length × width². Mice were euthanized when tumors reached approximately 1 cm in diameter or earlier if signs of ulceration, impaired mobility, weight loss, or other distress were observed, and tumor tissues were collected for subsequent analysis.

Flow Cytometry for Immune Profiling

On day 15, mice were anesthetized with 2.5% isoflurane and euthanized. Tumors were excised, minced, and digested in RPMI-1640 containing collagenase IV (2 mg/mL; Biosharp) and DNase I (0.1 mg/mL; MedChemExpress). The resulting cell suspensions were filtered, red blood cells were lysed using lysis buffer (Biocloud Biotechnology), and lymphocytes were enriched by Percoll density gradient centrifugation (Solarbio). Cells were resuspended in RPMI-1640 medium with 5% FBS and stained with fluorochrome-conjugated antibodies against CD3e, CD4, CD8a, PD-1 (CD279), and LAG-3 (CD223; Thermo Fisher Scientific). Antibodies used for flow cytometry are listed in Supplementary Table 2. Flow cytometry was performed with a minimum of 20,000 events per sample, and data were analyzed using FlowJo software (v10.8; BD Biosciences). The sequential gating strategy was as follows: debris was excluded based on FSC/SSC characteristics, singlets were selected using FSC-A versus FSC-H, CD3⁺ T cells were gated, and CD8a⁺ T-cell and CD4⁺ T-cell subsets were subsequently identified. PD-1 and LAG-3 expression was then analyzed within the CD8a⁺ T-cell population. The gating strategy for tumor-infiltrating CD4⁺ T cells, CD8⁺ T cells, and exhausted PD-1⁺/LAG-3⁺ T-cell populations is shown in Supplementary Figure 3. All flow cytometry data are presented as percentages of the parent population rather than absolute cell numbers.

Transcriptomic Sequencing

Total RNA was extracted from DME-treated CTLL-2 using TRIzol reagent (Vazyme, China) according to the manufacturer’s instructions. Genomic DNA was removed during reverse transcription according to the manufacturer’s protocol. Briefly, cells were lysed in TRIzol and incubated at room temperature for 5 min to allow complete dissociation of nucleoprotein complexes. RNA purity and concentration were then determined using a NanoDrop spectrophotometer, and RNA integrity was assessed using an Agilent 2100 Bioanalyzer. Only RNA samples with an A260/A280 ratio of 1.8–2.1 and a RNA integrity number (RIN) ≥ 7.0 were used for subsequent library construction. RNA quality assessment, library preparation, and sequencing were performed by Bioprofile Technology Company (Shanghai, China). Differentially expressed genes were identified and subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis using the DAVID database (v6.8). The results were visualized as bar plots and bubble charts.

Western Blot

CTLL-2 were seeded in 6-well plates and treated with 30 μM Erastin for 24 h, followed by treatment with 184 μg/mL DME for an additional 24 h. Cells were then lysed on ice using RIPA buffer (Solarbio) supplemented with 1% PMSF (Proteintech). Protein concentrations were quantified using the Pierce BCA Protein Assay Kit (Servicebio). After denaturation at 100°C for 10 minutes, proteins were separated by 10% SDS-PAGE and transferred to a 0.45 μm PVDF membrane (Millipore). The membrane was blocked with 5% skim milk in TBST for 1 hour at room temperature, followed by incubation with primary antibodies overnight at 4°C. After washing, membranes were incubated with HRP-conjugated secondary antibodies, and protein bands were detected using an ECL kit (Affinity Biosciences).

In a parallel experiment, CTLL-2 were treated with 20 μM Erastin for 24 h in the presence or absence of DME (184 μg/mL) or BBR (6.25 μM). Subsequent protein extraction, Western blotting, and detection were performed as described above. Antibodies used here are listed in Supplementary Table 3.

Total Superoxide Dismutase and Ferrous Ion Assays

Superoxide dismutase (SOD) activity was measured using a commercial assay kit (WST-8 method; Beyotime Biotechnology), while ferrous iron (Fe²⁺) levels were determined with an Iron Assay Kit (Solarbio). All assays were performed according to the manufacturers’ protocols.

ROS Assays

CTLL-2 were plated in 6-well plates and treated with 30 μM Erastin for 24 hours, followed by 184 μg/mL DME for an additional 24 hours. Cells were then stained with a fluorescent ROS probe (Beyotime Biotechnology) according to the manufacturer’s protocol. Intracellular ROS levels were quantified by flow cytometry and analyzed using FlowJo software (BD Biosciences).

In a parallel experiment, CTLL-2 pretreated with 30 μM Erastin for 24 hours were co-treated with DME (184 μg/mL) or BBR (6.25 μM) for an additional 24 hours. Intracellular ROS levels were evaluated using the same staining protocol, flow cytometric detection, and analytical methods described previously. In a parallel experiment, CTLL-2 pretreated with 30 μM Erastin for 24 h were further treated for an additional 24 h with one of the following: DME (184 μg/mL), BBR (6.25 μM), or DME combined with ML385 (an NRF2 inhibitor,1.9 μM). Intracellular ROS levels were assessed using the same staining protocol, flow cytometric detection, and analytical methods described earlier.

Intracellular BODIPY 581/591 C11 Staining

Lipid peroxidation was evaluated using the BODIPY 581/591 C11 probe (Beyotime Biotechnology). Cells were seeded in 6-well plates at 5 × 10⁴ cells per well, treated with Erastin for 24 hours, and then exposed to DME (184 μg/mL) for an additional 24 hours. Following treatment, cells were incubated with 2 μM BODIPY 581/591 C11 in PBS at 37°C for 30 minutes in the dark. Fluorescence images were captured using a Nikon ECLIPSE 80i microscope, and the reduced/oxidized fluorescence intensity ratio (red/green) was quantified using ImageJ software.

In a parallel experiment, CTLL-2 were treated with Erastin for 24 hours, followed by exposure to either DME (184 μg/mL) or BBR (6.25 μM) for 24 hours. Lipid peroxidation was assessed using the same BODIPY 581/591 C11 staining, imaging, and analysis protocol described previously.

Molecular Docking

The three-dimensional structure of berberine (BBR; PubChem CID: 2353) was obtained from PubChem and prepared using Open Babel. The crystal structure of GPX4 (PDB ID: 5H5Q) was downloaded from the Protein Data Bank. Molecular docking was conducted with AutoDock Vina using a semi-flexible approach, with BBR as the flexible ligand and GPX4 as the rigid receptor. The resulting binding poses were visualized using PyMOL.

Transmission Electron Microscopy (TEM)

For TEM analysis, CTLL-2 were subjected to the indicated treatments. In brief, cells were treated with 30 μM Erastin for 24 h, followed by 184 μg/mL DME for an additional 24 h. In the parallel rescue assay, cells were treated with 20 μM Erastin for 24 h and then co-treated with DME (184 μg/mL) or BBR (6.25 μM). After treatment, cells were fixed in 2.5% glutaraldehyde and 1% osmium tetroxide, dehydrated in graded ethanol, embedded, sectioned, and stained according to standard protocols. Ultrastructural changes were examined using a JEM-1400 TEM (JEOL, Japan).

Isolation of CD8⁺ T Cells

CD8⁺ T cells were isolated from mouse spleens using a CD8⁺ T-cell isolation kit (Elabscience) according to the manufacturer’s instructions. Cell purity was assessed by flow cytometry after staining with an anti-mouse CD8α antibody and was consistently greater than 90%.

Quantitative Real-Time PCR

CTLL-2 were treated as indicated. Total RNA was extracted from 1 × 10⁶ CTLL-2 cells using RNA isolation reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. Genomic DNA was removed using DNase I (Thermo Fisher Scientific) during RNA extraction. RNA concentration and purity were assessed by spectrophotometry (A260/A280 ratio 1.8–2.1). Subsequently, cDNA was synthesized using the PrimeScript RT reagent kit (Takara Bio). No-RT controls were included to confirm the absence of genomic DNA contamination. The reaction mixture was incubated in a PCR instrument at 37°C for 10 min, followed by enzyme inactivation at 85°C for 5 s. The synthesized cDNA was stored at −80°C until further analysis.

qPCR was performed using TB Green Premix Ex Taq II Fast qPCR (2×; Takara Bio) in a 25 μL reaction system containing 12.5 μL of TB Green Premix Ex Taq II Fast qPCR (2×), 1 μL of forward primer (10 μM), 1 μL of reverse primer (10 μM), 1 μL of cDNA template, and nuclease-free water to a final volume of 25 μL. Amplification was carried out on an Archimed X6 Real-Time PCR System (RocGene, Beijing, China) under the following conditions: Primer amplification efficiency was determined using standard curves generated from serial dilutions of cDNA and was within the acceptable range (90–110%) with R² values greater than 0.99. Initial denaturation at 95°C for 30s, followed by 40 cycles of denaturation at 95°C for 5 s and annealing/extension at 60°C for 10s. No-template controls (NTC) were included in each qPCR run. After amplification, melting curve analysis was performed by heating at 95°C for 15s, cooling at 60°C for 30s, and reheating at 95°C for 15s. All amplicons showed a single melting peak, indicating specific amplification without detectable primer-dimer formation. Cq values were determined by the second derivative maximum method using Archimed Analyzer v2.0.2. Three independent biological replicates were performed. Relative gene expression was calculated using the 2^(-ΔΔCt) method, with each target gene normalized to the internal reference gene β-actin to obtain ΔCt values, and the control group used as the calibrator for ΔΔCt calculation. Primer sequences and amplification efficiency are provided in Table 1.

Table 1.

Primer Sequences and Amplification Efficiency Used for RT-qPCR Analysis

Gene	Forward Primer (5′–3′)	Reverse Primer (5′–3′)	Primer Length (bp)	Slope	Efficiency (%)	R2
NRF2	CTTTAGTCAGCGACAGAAGGAC	AGGCATCTTGTTTGGGAATGTG	140	3.32	100.1	0.998
SLC7A11	GGCACCGTCATCGGATCAG	CTCCACAGGCAGACCAGAAAA	160	3.28	101.8	0.995
GPX4	GTACTGCAACAGCTCCGAGT	ATGCACACGAAACCCCTGTA	140	3.41	96.4	0.997
β-actin	GTTACCAACTGGGACGA	CAGAGGCATACAGGGAC	210	3.36	98.3	0.999

Statistical AnalysisAll data are presented as mean ± SEM. For comparisons among multiple groups at each indicated time point, one-way ANOVA followed by Tukey’s post hoc test was used. Between two groups, Student’s t-test was used. A p-value < 0.05 was considered statistically significant. Here, n refers to the number of biologically independent mice per group.

Results

In vitro Anti‑tumor Activity of DME and Identification of Candidate Metabolites

The ethanol extract of Dracocephalum tanguticum Maxim (DME) was prepared through ultrasonic-assisted extraction followed by reflux with 70% ethanol. After centrifugation and concentration, DME was dissolved in 50% DMSO and diluted with culture medium to a final DMSO concentration of 0.5% for in vitro assays (Figure 1A).

Figure 1.

D. tanguticum extract (DME) inhibits colorectal cancer cell proliferation, migration, and colony formation. (A) Preparation scheme of DME. (B) cell viability assay showing the inhibitory effect of DME on CT-26 cell viability. (C) Colony-forming ability assay. (D) Wound-healing assay showing the effect of DME on CT-26 cell migratory ability. Red lines indicate the wound edges used for relative wound area quantification. (E–H) LC-MS/MS chromatograms and metabolite identification. Data are presented as mean ± SEM. ns, not significant; ** p < 0.01; *** p < 0.001.

To evaluate the antitumor activity of DME, CT-26 cells were treated with DME in vitro. The cell viability assay showed that 184 μg/mL DME significantly reduced CT-26 cell viability after 24 h (Figure 1B). Colony formation assays further demonstrated that DME markedly impaired the clonogenic capacity of CT-26 cells (Figure 1C). In addition, wound-healing assays showed that DME significantly suppressed CT-26 cell migratory ability (Figure 1D). Collectively, these findings indicate that DME exhibits antitumor activity by inhibiting CT-26 cell proliferation and migration. Having confirmed the antitumor activity of DME, we next analyzed its chemical composition to identify potential bioactive constituents. LC-MS/MS profiling putatively annotated 497 compounds in the DME extract, among which 29 prototype constituents were also detected in serum after oral administration. The DME used in this study was tentatively characterized by LC-MS/MS (Figure 1E–H). These findings provided the rationale for prioritizing BBR as the putatively annotated candidate metabolite responsible for the observed antitumor effects.

Network Analysis Reveals the Immunomodulatory Mechanisms of DME, A Potential Candidate Metabolite Against Colorectal Cancer

Using untargeted LC-MS/MS screening focused on serum-absorbed prototype constituents, we putatively annotated 29 blood-entering metabolites in DME (Table 2). Subsequently, we obtained the standard SMILES structures of these metabolites from PubChem and imported them into the SwissTarget Prediction platform for target analysis. After excluding targets with reliability scores of zero and removing duplicates, we identified 897 potential targets for DME’s candidate metabolites. Concurrently, from GeneCards and DrugBank databases, we collected 14,097 and 9,834 confirmed or potential CRC targets, respectively. All targets were screened and standardized via UniProt, followed by duplicate removal, ultimately yielding 13,029 targets. A total of 560 overlapping targets were identified between potential candidate metabolites and CRC (Figure 2A). Finally, KEGG and GO enrichment analyses showed significant enrichment of these overlapping targets in immune-related pathways and biological processes (Figure 2B and C), suggesting that DME exerts anti-CRC effects through immunomodulatory. The blood-absorbed constituents were putatively annotated and used for subsequent target prediction. DME Inhibits Colorectal Cancer Growth by Promoting CD8⁺ T Cell Infiltration and Reversing Exhaustion.

Table 2.

MS Data of Putatively Annotated Serum-Absorbed Prototype Constituents

Peak No.	Name	Molecular Form	m/z	tR/min	Ion Selected	Mean Peak Area
1	Acetylcholine	C7H16NO2+	146.11626	1.654	[M]+	3145.33333333333
2	Diosmetin 7-O-beta-D-glucuronopyranoside	C22H20O12	477.10281	5.8	[M+H]+	1018.16666666667
3	Citrusin C	C16H22O7	349.12711	4.519	[M+Na]+	56695.3333333333
4	N-Acetylneuraminic acid	C11H19NO9	308.09821	0.941	[M-H]−	43908.8333333333
5	Kaempferol 3-Glucuronide	C21H18O12	461.0677	4.21	[M-H]−	8863.5
6	Heneicosanoic acid	C21H42O2	325.30969	23.064	[M-H]−	310.5
7	9-Oxononanoic acid	C9H16O3	171.10316	5.773	[M-H]−	3825.16666666667
8	Abscisic acid	C15H20O4	263.13156	5.711	[M-H]−	3686.16666666667
9	9-hydroxy-7-(2-hydroxypropan-2-yl)-1,4a-dimethyl-2,3,4,9,10,10a- hexahydrophenanthrene-1-carboxylic acid	C20H28O4	331.19153	15.099	[M-H]−	3597.66666666667
10	Melilotic acid	C9H10O3	165.05745	4.605	[M-H]−	18243.3333333333
11	Kampferol-3,4’-dimethyl ether	C17H14O6	313.07205	11.181	[M-H]−	1747.66666666667
12	Manghaslin	C33H40O20	755.19867	0.927	[M-H]−	430.5
13	Stearidonic acid	C18H28O2	277.21524	16.69	[M+H]+	3152.66666666667
14	Soyasaponin I	C48H78O18	941.51361	10.133	[M-H]−	1680.16666666667
15	PC (18:0/20:4)	C46H84NO8P	810.60284	19.512	[M+H]+	34459573
16	PC (16:0/20:5)	C44H78NO8P	780.54504	18.455	[M+H]+	6411978.33333333
17	Myrcene	C10H16	159.11577	6.263	[M+Na]+	2222
18	Luteolin	C15H10O6	287.05466	5.751	[M+H]+	1814.83333333333
19	Germine	C27H43NO8	510.31497	11.263	[M+H]+	8116.66666666667
20	Ganoderic acid F	C32H42O9	593.27771	20.345	[M+Na]+	4191.5
21	Dihydroactinolide	C11H16O2	181.12296	8.589	[M+H]+	11230.6666666667
22	Chlorogenic acid	C16H18O9	355.10358	3.583	[M+H]+	15271.8333333333
23	Berberine	C20H18NO4+	336.11783	5.176	[M]+	1286.33333333333
24	Eupafolin	C16H12O7	317.06628	3.715	[M+H]+	8883.5
25	Nepitrin	C22H22O12	479.11844	3.75	[M+H]+	3501.66666666667
26	Phenylpyruvic acid	C9H8O3	163.04149	4.178	[M-H]−	72368.6666666667
27	MGMG 18:3	C27H46O9	559.30902	14.74	[M+FA-H]−	1187.66666666667
28	Corosolic acid	C30H48O4	471.34537	16.668	[M-H]−	2456.66666666667
29	4-Nonylphenol	C15H24O	219.17496	27.189	[M-H]−	21255.8333333333

Figure 2.

DME Network Analysis. (A) Venn diagram illustrating the overlap between blood-absorbed metabolites of DME and colorectal cancer-related targets. (B) GO pathway enrichment analysis of potential targets of DME. (C) KEGG pathway enrichment analysis of potential targets of DME.

A CT-26 tumor-bearing mouse model was established to evaluate the in vivo antitumor and immunomodulatory effects of DME on colorectal cancer. To this end, mice were randomly assigned to five groups (n = 5): model, low-dose DME (156 mg/kg), medium-dose DME (312 mg/kg), high-dose DME (624 mg/kg), and positive control (5-FU, 20 mg/kg). All treatments were administered orally once daily for 14 days, starting on day 0 (Figure 3A). As a result, DME treatment significantly suppressed tumor growth in a dose-dependent manner, with the high-dose group DME(H) showing the most pronounced effect. Notably, the antitumor efficacy of DME(H) was comparable to that of 5-FU (Figure 3B and C). Moreover, the DME(H) group showed minimal body weight loss, indicating favorable safety and reduced systemic toxicity (Figure 3D).

Figure 3.

DME suppresses tumor growth and modulates the tumor immune microenvironment in CT-26 tumor-bearing mice. (A) Schematic diagram of animal experimental design. CT-26 cells were inoculated into the right axillary region of mice, with treatment commencing when tumor volume reached approximately 100 mm³. (B) Diagram of tumor volume measurement (Volume = ½ × length × width²) and representative photographs of excised tumors from each group at the endpoint. (C) Tumor growth curves showing that DME treatment inhibits tumor progression in a dose-dependent manner. 5-FU was used as a positive control. (D) Body weight changes of mice during the treatment period. All groups maintained stable body weight, indicating no overt systemic toxicity. (E) Representative flow cytometry plots demonstrate the presence of CD8⁺T cells (CD3⁺, CD8⁺) following different treatments. (F) Representative flow cytometry plots showing the number of LAG-3⁺ exhausted T cells (CD3⁺, CD8⁺, LAG-3⁺) following different treatments. (G) Representative flow cytometry plots displaying the number of PD-1⁺ exhausted T cells (CD3⁺, CD8⁺, PD-1⁺) following different treatments. (H) The number of tumor-infiltrating CD8⁺ T cells were quantified by flow cytometry analysis (n = 5). (I) Flow cytometry analysis quantified the number of PD-1⁺ cells among CD3⁺CD8⁺ T cells (n = 5). (J) Flow cytometry analysis quantified the number of LAG-3⁺ cells among CD3⁺CD8⁺ T cells (n = 5). Data are expressed as the mean ± SEM (n = 5 mice per group). Statistical significances were calculated via one-way ANOVA. ns, not significant; * p < 0.05; ** p < 0.01; *** p < 0.001.

CD8⁺ T cells are pivotal effectors of antitumor immunity, directly responsible for eliminating malignant cells. Unfortunately, their cytotoxic function is often compromised within the immunosuppressive tumor microenvironment (TME). To determine whether DME can counteract this suppression, we analyzed immune cell infiltration by flow cytometry. We found that DME treatment significantly increased the number of tumor-infiltrating CD8⁺ T cells, suggesting a potent immunostimulatory role. (Figure 3E and H). Furthermore, DME markedly reduced the expression of exhaustion markers PD-1 and LAG-3 tumor-infiltrating CD8⁺ T cells (Figure 3F, I, G and J). Collectively, these findings demonstrate that DME not only promotes cytotoxic T lymphocyte (CTL) infiltration but also reverses their functional exhaustion, thereby restoring and enhancing antitumor immunity.

DME Attenuates Oxidative Stress and Inhibits Ferroptosis in CTLL-2 via Upregulation of GPX4

To investigate how D. tanguticum enhances cytotoxic T lymphocyte (CTL) activity, we performed RNA sequencing on purified CTLL-2. Subsequent KEGG pathway enrichment analysis showed that DME treatment significantly modulated the ferroptosis pathway (Figure 4A–C). GPX4 expression was significantly upregulated at the protein level (Figure 4D). These results suggest that DME inhibits CTLL-2 ferroptosis by enhancing GPX4 expression, thereby maintaining CTLL-2 abundance and function in the tumor microenvironment. In addition, DME did not significantly alter caspase-3 expression in CTLL-2, suggesting that its protective effect was not primarily attributable to apoptosis regulation (Supplementary Figure 4A). The corresponding original Western blot images for GPX4 and caspase-3 are provided in Supplementary Figure 5A and B. To directly assess ferroptosis, we measured its key biochemical hallmarks. As expected, the ferroptosis inducer Erastin significantly increased intracellular Fe²⁺ and reactive oxygen species (ROS), leading to the defining event of ferroptosis: elevated lipid peroxidation. Crucially, DME co-treatment effectively counteracted these effects, suppressing the Erastin-induced accumulation of Fe²⁺, ROS, and lipid peroxides (Figure 4E–G). We next probed the antioxidant defense system and found that DME significantly restored the activity of superoxide dismutase (SOD) (Figure 4H). This enhancement of a key antioxidant enzyme suggests that DME bolsters cellular defenses to resist ferroptosis.

Figure 4.

DME remodels the transcriptomic profile and inhibits ferroptosis in CTLs by upregulating GPX4, with BBR identified as a putatively annotated candidate metabolite. (A) Volcano plot of differentially expressed genes (DEGs) between DME-treated and control CTLs (filtered by |log2FC| > 1.0 and adjusted p-value < 0.05). Red and blue points represent up- and down-regulated genes, respectively. (B) Heatmap of DEGs from RNA-seq analysis showing expression patterns in control and DME-treated groups. (C) KEGG pathway enrichment analysis of DEGs. Significantly enriched pathways (p < 0.05) are displayed, with ferroptosis highlighted. (D) Western blot analysis showing GPX4 protein expression in CTLs after DME treatment. (E) Levels of reactive oxygen species (ROS) in CTLs treated with Erastin (20 μM) and DME (184 μg/mL). (F) Intracellular Fe2⁺ levels in CTLs. (G) Lipid peroxidation levels measured by C11-BODIPY 581/591 staining, presented as the ratio of oxidized to reduced fluorescence intensity. (H) Superoxide dismutase (SOD) activity in CTLs under the indicated treatments. (I) Chemical structure of berberine (BBR). (J) Molecular docking simulation showing the binding interaction between BBR and the GPX4 protein domain. Data are presented as mean ± SEM. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001.

Network analysis screening putatively suggested six potential candidate metabolites with high oral bioavailability (OB) and drug-likeness (DL) (Table 3). Molecular docking revealed that berberine had the strongest binding affinity for GPX4 (−7.82 kcal/mol), indicating its role as a putatively annotated candidate metabolite in DME for GPX4 regulation (Figure 4I and J). The docking scores represent predicted binding affinities based on structural complementarity, which do not equate to confirmed biological activity.

Table 3.

Predicted Binding Scores Between GPX4 and Candidate Metabolites from DME

Ranking	Candidate Metabolites	Binding Score (kcal/mol)
1	9-hydroxy-7-(2-hydroxypropan-2-yl)-1,4a-dimethyl-2,3,4,9,10,10a-hexahydrophenanthrene-1-carboxylic acid	−2.81
2	PC (18:0/20:4)	−5.6
3	PC (16:0/20:5)	−5.85
4	Luteolin	−4.7
5	Berberine	−7.82
6	4-Nonylphenol	−3.97

BBR Enhances CTLL-2 Survival by Suppressing Ferroptosis and Alleviating Oxidative Stress

We first determined the half-maximal inhibitory concentration (IC₅₀) of BBR, a putatively annotated candidate metabolite of DME, to be 30.66 μM using a cell viability assay (Figure 5A).Under Erastin-induced ferroptosis, 6.25 μM BBR optimally restored CTL viability and was therefore used in subsequent experiments (Figure 5B). Compared with the Erastin group, BBR significantly decreased intracellular Fe²⁺ and ROS levels and increased SOD activity (Figure 5C–E). Using the C11-BODIPY 581/591 probe, we observed that BBR markedly inhibited lipid peroxidation, as shown by reduced green fluorescence (Figure 5F). Transmission electron microscopy (TEM) analysis of CTLL-2 revealed that Erastin treatment induced the classic ultrastructural hallmarks of ferroptosis, including mitochondrial shrinkage, increased membrane density, and a notable reduction or even loss of mitochondrial cristae. In contrast, co-treatment with BBR substantially attenuated these morphological aberrations, preserving normal mitochondrial size, membrane integrity, and distinct cristae (Figure 5G). Together, these ultrastructural findings further support that BBR may protect CTLL-2 by inhibiting ferroptosis at the subcellular level. Furthermore, the NRF2 inhibitor ML385 partially reversed the BBR-mediated reduction in intracellular Fe²⁺ and ROS levels and the restoration of SOD activity, suggesting that NRF2 is involved in the protective effect of BBR against ferroptosis in CTLL-2 (Supplementary Figure 4B–D).

Figure 5.

DME remodels the transcriptomic profile and inhibits ferroptosis in CTLs by upregulating GPX4, with BBR identified as a putatively annotated candidate metabolite. (A) The effect of BBR on the proliferation of CTLs was determined using the cell viability assay. (B) Cell viability assays were conducted in CTLs following treatment with Erastin (20 µM) alone or in combination with BBR. Cell viability was calculated relative to untreated controls (set at 100%). Statistical graphs are based on data from three independent experiments.) (C) Detection of reactive oxygen species in CTLs following treatment with Erastin (20 µM) alone or in combination with BBR (6.25 µM) (D) Intracellular Fe2⁺ levels in CTLs. (E) Intracellular superoxide dismutase (SOD) levels in CTLs. (F) Detection of DME C11-BODIPY 581/591 fluorescence in CTLs via confocal microscopy. (G) Ultrastructure of CTLs. Data are presented as mean ± SEM. ns, not significant; **p < 0.01; ***p < 0.001.

BBR Enhances CD8⁺ T Cell Anti-Tumor Function by Activating the NRF2–SLC7A11–GPX4 Axis to Suppress Ferroptosis

To further assess whether BBR is a putatively annotated candidate metabolite in DME associated with the inhibition of CD8⁺ T cell ferroptosis and exhaustion, we first evaluated its in vivo antitumor efficacy. In CT-26 tumor-bearing mice, BBR treatment significantly suppressed tumor growth compared with the model control group (Figure 6A and B). Flow cytometry analysis further showed that BBR treatment increased the number of tumor-infiltrating CD8⁺ T cells and reduced the proportion of PD-1⁺LAG-3⁺ exhausted T cells in the tumor microenvironment (Figure 6C–E). The downregulation of these critical inhibitory receptors demonstrates that BBR promotes T cell infiltration and functional reinvigoration. These findings confirm that BBR is a putatively annotated candidate metabolite in DME that enhance anti-tumor immunity by preventing T cell exhaustion.

Figure 6.

BBR modulates the tumor immune microenvironment in vivo and upregulates the NRF2-SLC7A11-GPX4 axis in vitro. (A) Diagram of tumor volume measurement (Volume = ½ × length × width²) and representative photographs of excised tumors from each group at the endpoint. (B) The tumor growth curve indicates that BBR inhibits tumor progression (C). Representative flow cytometry plots showing tumor-infiltrating CD3⁺CD8⁺ T cells under different treatments (n = 5). (D) Representative flow cytometry plots showing the number of LAG-3⁺ cells among CD3⁺CD8⁺ T cells (n = 5). (E) Representative flow cytometry plots showing the number of PD-1⁺ cells among CD3⁺CD8⁺ T cells (n = 5). (F) Relative mRNA expression levels of NRF2, SLC7A11, and GPX4 in CTLs treated with Erastin in the presence or absence of BBR, as determined by qRT-PCR. (G) Western blot analysis of NRF2, SLC7A11, and GPX4 protein expression in CTLs treated with Erastin and BBR. β-Actin was used as a loading control. (H) Relative mRNA expression levels of NRF2, SLC7A11, and GPX4 in primary CD8⁺ T cells treated with Erastin in the presence or absence of BBR, as determined by qRT-PCR. (I) Western blot analysis of NRF2, SLC7A11, and GPX4 protein expression in primary CD8⁺ T cells treated with Erastin and BBR. β-Actin was used as a loading control. Data are presented as mean ± SEM (n = 5 mice per group). * p < 0.05; ** p < 0.01; *** p < 0.001.

To elucidate the mechanism by which BBR inhibits ferroptosis, we examined its effect on the core NRF2–SLC7A11–GPX4 defense pathway in both CTLL-2 and primary CD8⁺ T cells. In CTLL-2, Erastin treatment significantly downregulated the expression of nuclear factor erythroid 2–related factor 2 (NRF2), Solute carrier family 7 member 11 (SLC7A11), and Glutathione peroxidase 4 (GPX4) at both mRNA and protein levels. BBR co-treatment reversed this suppression, upregulating all three molecules at both transcriptional and translational levels (Figure 6F and G). To confirm the physiological relevance of this pathway, primary CD8⁺ T cells were isolated from mouse spleens. Consistent with the findings in CTLL-2, BBR similarly rescued the Erastin-induced downregulation of NRF2, SLC7A11, and GPX4 in primary cells at both mRNA and protein levels (Figure 6H and I). Quantitative analyses of NRF2, SLC7A11, and GPX4 protein expression in CTLL-2 and primary CD8⁺ T cells are shown in Supplementary Figure 6 A and B. The purity of isolated primary CD8⁺ T cells was verified by flow cytometry and consistently exceeded 90% (Supplementary Figure 6C). The original Western blot images for NRF2, SLC7A11, and GPX4 in CTLL-2 are provided in Supplementary Figure 7. The original Western blot images for NRF2, SLC7A11, and GPX4 in primary CD8⁺ T cells are provided in Supplementary Figure 8. Together, these coordinated results across different cellular models support the view that BBR may exert its anti-ferroptotic effect in part by modulating the NRF2–SLC7A11–GPX4 signaling axis.

Discussion

This study suggests that the ethanol extract of Dracocephalum tanguticum Maxim (DME) inhibits colorectal cancer progression by reprogramming the tumor immune microenvironment, primarily through protecting CD8⁺ T cells from ferroptosis. We identified BBR as a candidate constituent potentially associated with this effect and proposed its underlying mechanism: BBR upregulates the NRF2–SLC7A11–GPX4 axis, thereby alleviating oxidative stress and iron overload to contribute to ferroptosis resistance in CD8⁺ T cells.

The role of natural products in oncology is increasing. They are evolving from direct cytotoxic agents into immunomodulators that augment current therapies. For example, metabolites including baicalin, protopanaxadiol, and Ganoderma lucidum extracts can suppress proliferation and induce apoptosis in colorectal cancer.^17–20 Furthermore, some agents, such as Atractylenolide I, can synergize with anti-PD-1 therapy by directly modulating the tumor immune microenvironment.²¹ However, the therapeutic potential of such single metabolites is often limited by their single-target mechanisms, which may be inadequate for tackling the multifaceted nature of cancer. This limitation highlights the potential of multi-component botanical drug systems, which can engage multiple pathways simultaneously. Traditional Tibetan Medicine (TTM) is a prime example of this paradigm and a rich source of such multifunctional agents.²² Modern research has validated the therapeutic efficacy of D. tanguticum, which often acts through anti-inflammatory mechanisms and has a favorable safety profile.^23,24 Specific examples include Rhizoma Paridis, which induces cytoprotective autophagy in CRC cells,²⁵ and preparations of Magnolia officinalis, which alleviate 5-FU-induced gastrointestinal mucositis.²⁶ Despite the multi-target nature of TTM, its specific role in reprogramming the tumor immune microenvironment in CRC has received scant attention.

D. tanguticum, a medicinal plant endemic to the Qinghai-Tibet Plateau, is listed in the Tibetan Medicine Standard and clinically used for gastrointestinal disorders due to its anti-inflammatory, antioxidant, and mucosal protective properties.²⁷ Although anti-inflammatory metabolites have been identified in D. tanguticum, their role in colorectal cancer treatment remains unexplored. Our findings suggest DME as an immune modulator that targets CD8⁺ T cells ferroptosis and exhibits a multi-target profile that could be useful for reprogramming the tumor immune microenvironment. Thus, DME shows promise potential for combination immunotherapy, particularly against treatment-resistant “cold” tumors, and provides a preliminary basis for its future development.

LC-MS/MS analysis further characterized the chemical composition of DME and provided a basis for its biological effects. The chromatograms and metabolite identification data in Figure 1E–H allowed us to annotate the major constituents of the extract according to their retention times and fragmentation patterns. This phytochemical profiling is important because it connects the observed antitumor activity of DME with its multi-component nature and supports berberine as a key candidate responsible for the immunomodulatory effect. These findings suggest that DME may exerts its activity through a coordinated action of multiple metabolites rather than a single compound alone.

Immune checkpoint inhibitors targeting PD-1/PD-L1 are widely used to treat advanced CRC.⁴ As central effectors of antitumor immunity, CTLs directly kill target cells, regulate B cell function, and secrete cytokines.²⁸ Although CTLs are often present in the tumor microenvironment, their antitumor function is frequently impaired.⁷ In this context, CTLs represent the cytotoxic functional subset of CD8⁺ T cells, and changes in CD8⁺ T-cell infiltration are closely linked to the abundance of CTLs in the tumor microenvironment. Reduced phospholipid phosphatase 1 expression disrupts unsaturated phospholipid metabolism, leading to lipid peroxidation accumulation and ferroptosis in CD8⁺ T cells. Ferroptosis, which differs from apoptosis and necrosis morphologically and mechanistically, has been reported to impairs antitumor immunity and promotes immune escape.⁹ However, the biological consequences of ferroptosis inhibition are likely cell type-dependent. In CD8⁺ T cells, suppressing ferroptosis may preserve cytotoxic function and sustain antitumor immunity, whereas in tumor cells, ferroptosis inhibition could theoretically favor cell survival and weaken therapeutic efficacy.²⁹ Therefore, the net antitumor effect of DME/BBR may depend on whether these agents preferentially protect CTLs while leaving tumor cells susceptible to ferroptosis.

Iron is an important trace element abundant in the human body; iron overload is known to induce oxidative stress and promote cell death.³⁰ As a key regulator, GPX4 is a central inhibitor of ferroptosis. Under stress, NRF2 dissociates from Keap1, translocates to the nucleus, and transactivates antioxidant response element–driven genes such as SLC7A11 and GPX4.^31,32 Consistent with this mechanism, our transcriptomic sequencing showed that DME intervention was associated with upregulates GPX4 expression in CD8⁺ T cells. Further experimental validation revealed increased that DME intervention also markedly elevated the mRNA and protein levels of both NRF2 (the key transcriptional regulator of GPX4) and its downstream target, SLC7A11. Consistent with previous reports, our results suggest that DME may exert its anti-ferroptotic effect through the NRF2–SLC7A11–GPX4 axis. By promoting cysteine uptake, glutathione biosynthesis, and GPX4 activity, this pathway may help protect CD8⁺ T cells from ferroptosis.³³ Collectively, these data support the involvement of this signaling axis in DME-mediated ferroptosis resistance, although further studies are needed to fully validate this mechanism. GPX4 uses glutathione to counteract lipid peroxidation, reduce ROS and Fe²⁺ accumulation, enhance SOD activity, and restore mitochondrial integrity and cellular redox homeostasis. We found that both DME and its active metabolite, BBR, significantly attenuated these pro-ferroptotic changes in CD8⁺ T cells. Berberine, a well-studied isoquinoline alkaloid, is known for its broad pharmacological activities, including anti-inflammatory and antioxidant effects.^34–36 BBR exerts beneficial effects against colorectal cancer by inhibiting tumor growth and suppressing epithelial-mesenchymal transition (EMT).³⁷ In this study, BBR showed pronounced protective effects on CD8⁺ T cells by markedly reducing lipid peroxide and intracellular Fe²⁺ accumulation, restoring mitochondrial integrity, and enhancing SOD activity. These findings suggest that berberine is a candidate metabolite that may contribute to DME’s anti-ferroptotic activity. Collectively, our results suggest that D. tanguticum, with berberine as a key contributor, may suppresses ferroptosis in CD8⁺ T cells, thereby potentially preserving their anti-tumor capacity.

Notably, NRF2 and GPX4 serve as molecular nodes linking immunomodulation and metabolic reprogramming. Beyond regulating ferroptosis susceptibility, they may influence immune checkpoint expression, such as PD-L1, through cross-talk with signaling pathways. Network analysis of 29 blood-absorbed metabolites of DME also revealed enrichment in the PD-1/PD-L1 pathway, supporting the rationale for combining DME with immune checkpoint blockade. These findings were supported via transcriptomic sequencing, molecular docking, PCR, and Western blot, providing preliminary evidence for the role of D. tanguticum and BBR in modulating the tumor immune microenvironment.

A limitation of the present study is that the mechanistic conclusions were derived mainly from a single murine tumor model and preclinical network pharmacology analyses, and therefore require further validation in additional models. The absence of CD8⁺ T-cell depletion experiments also limit our ability to fully define the contribution of CTLs to the observed antitumor effects. In addition, the phytochemical complexity of D. tanguticum and the uncertain standardization of DME make definitive mechanistic interpretation challenging. Finally, further toxicity evaluation and pharmacokinetic characterization will be among our next research priorities to better define the translational potential of this approach.

Conclusion

This study suggests that BBR, a putatively annotated metabolite from Dracocephalum tanguticum Maxim (D. tanguticum), may inhibit CRC progression and preserve CD8⁺ T-cell function by modulating ferroptosis-related pathways. These findings indicate that BBR and D. tanguticum-based preparations are promising candidates for CRC immunotherapy, although CD8⁺ T dependence, target engagement, and safety remain to be confirmed.

Funding Statement

The authors declare that financial support was received for the research, authorship, and/or publication of this article. This work was supported by research funding from the National Natural Science Foundation of China (81960451 to YL), the Applied Basic Research Project Fund of the Qinghai Provincial Department of Science and Technology (2024-ZJ-710) and the Special Fund for Enhancing Research Capacity of Qinghai University, Research Team for Plateau Cardiovascular and Cerebrovascular Health and Wellness (2026KTST07)

Abbreviations

CTLs, cytotoxic T lymphocytes; CRC, colorectal cancer; D. tanguticum, Dracocephalum tanguticum Maxim; TTM, Traditional Tibetan Medicine; DME, Dracocephalum tanguticum Maxim ethanol extract; BBR, berberine; TME, tumor microenvironment; CCK-8, Cell Counting Kit-8; ESI, electrospray ionization; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; SOD, superoxide dismutase; Fe²⁺, ferrous iron; TEM, Transmission Electron Microscopy; ROS, reactive oxygen species; OB, oral bioavailability; DL, drug-likeness; IC₅₀, half-maximal inhibitory concentration; GPX4, Glutathione peroxidase 4; NRF2, Nuclear factor erythroid 2–related factor 2; SLC7A11, Solute carrier family 7 member 11.

Data Sharing Statement

The original contribution data presented in this study have been publicly released. The data can be accessed here: https://www.ncbi.nlm.nih.gov/, under accession numbers PRJNA1423454 of 16S rRNA sequencing and PRJNA1423454 of Transcriptome sequencing.

Disclosure

The authors report no conflicts of interest in this work.