Sunflower oil-loaded lycopene protects intestinal barrier integrity following traumatic brain injury

Abstract

Traumatic brain injury (TBI) is a major cause of morbidity, mortality, and long-term neurological impairment. Growing evidence suggests that the gut-brain axis plays a crucial role in TBI pathophysiology, with gut microbiota dysbiosis potentially inducing inflammation and aggravating brain injury. Lycopene (Lyc), a potent antioxidant derived from tomatoes with neuroprotective and anti-inflammatory properties, is considered a promising potential therapy for TBI. Sunflower oil (SO) can effectively enhance Lyc bioavailability. We established a TBI mouse model and treated it with SO-loaded Lyc (SO-Lyc). Serum and fecal samples were collected for metabolomic analysis, 16S rRNA sequencing for gut microbiota changes, ELISA for inflammatory cytokines and injury markers, H&E staining for gut/brain histopathology, and immunohistochemistry for intestinal barrier-related gene expression. Furthermore, we used 16S rRNA data from patient samples with TBI to perform association analyses. The results showed that SO-Lyc supplementation may mitigate damage to intestinal and cerebral tissues by restoring intestinal barrier function and suppressing systemic inflammation through modulation of key metabolic pathways, such as those involving linoleic acid, sphingolipids, and tryptophan. Its effects manifest as reduced serum inflammatory cytokines (TNF-α, IL-6, CRP, IL-1β), intestinal injury markers (DAO, D-LA, I-FABP, ET), and brain injury markers (NFL, GFAP, S100B, NSE, UCH-L1), while enhancing the expression of intestinal barrier genes (occludin and ZO-1). Furthermore, this supplement promotes the growth of gut microbiota with anti-inflammatory properties. Differences in inflammation and gut microbiota have been observed among TBI patients with varying degrees of severity, suggesting that Lyc containing SO may be an effective treatment for TBI patients.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s10735-026-10980-3.

Introduction

Traumatic brain injury (TBI) is a force-induced cerebral injury that can cause temporary or chronic impairments, including cognitive dysfunction, motor disorders, and emotional disturbances (Capizzi et al. 2020). The lasting damage from TBI is considered the consequence of a cascade of processes, such as inflammation and oxidative stress, followed by disruption of the blood-brain barrier and neuronal dysfunction (Killen et al. 2019; Yang et al. 2024). Although acute TBI is often properly managed, options to prevent or mitigate its long-term effects remain limited (Capizzi et al. 2020).

Recent research suggests that the gut-brain axis, a bidirectional communication pathway between the gut and brain, plays a crucial role in the pathophysiology of TBI (El Baassiri et al. 2024). TBI induces profound alterations in the gut, including disruption of the intestinal barrier, dysbiosis of the gut microbiota, and disturbances in the enteric nervous system and autonomic nervous system (Hanscom et al. 2021; El Baassiri et al. 2024). This triggers a runaway vicious cycle, exacerbating systemic inflammation via the gut-brain axis, thereby worsening brain injury and inducing neurodegeneration (Bansal et al. 2009; Hanscom et al. 2021). Therefore, restoring intestinal integrity and modulating the inflammatory response may offer novel therapeutic strategies for treating TBI.

Lycopene (Lyc), a carotenoid abundant in tomatoes, is known as a potent antioxidant with anti-inflammatory properties (Khan et al. 2021). Previous studies have shown that Lyc protects against neurodegeneration and ischemic injury by alleviating the inflammatory response and restoring gut microbiota balance (Kaur et al. 2011; Saini et al. 2020; Chen et al. 2025). However, as a potent natural fat-soluble antioxidant, the bioavailability of Lyc is limited during digestion in the gastrointestinal tract (Kulawik et al. 2023). Research has indicated that Lyc not only enhances the oxidative and thermal stability of vegetable oils, endowing them with a richer array of bioactive compounds, but also that vegetable oils may improve the bioavailability of Lyc (Liang et al. 2019; Flais et al. 2025; Lin et al. 2025). Sunflower oil (SO) exhibits the highest solubility for Lyc, and multiple studies have used SO as its carrier for oral therapeutic administration (Bandeira et al. 2017; Campos et al. 2019).

Sunflowers (Helianthus annuus) are among the most important oilseed crops worldwide. Their seeds are a vital source of edible oil and an important dietary component, rich in polyunsaturated fatty acids (linoleic acid), monounsaturated fatty acids, vitamin E, phenolic acids, flavonoids, and trace elements, all of which exhibit potent antioxidant properties (de Mello Silva Oliveira et al. 2016; Guo et al. 2017; Desai et al. 2024). Interestingly, several studies have shown that SO and its primary constituents exert protective effects against cerebral diseases, such as ischemia–reperfusion injury, Alzheimer’s disease, and ischemic stroke (La Russa et al. 2022; Lee et al. 2022; Desai et al. 2024). This protective effect is primarily attributed to the antioxidant and anti-inflammatory properties of SO and its key components. Furthermore, SO regulates gut microbiota. Aldamarany et al. discovered that perilla, sunflower, and tea seed oils, as potential dietary supplements, may alleviate inflammation, oxidative stress, and fatty liver degeneration by modulating the gut microbiota composition in mice fed a high-fat diet (Aldamarany et al. 2023).

However, it remains unclear whether Lyc-enriched SO has therapeutic effects on TBI. We hypothesized that it may exert therapeutic effects in TBI by regulating the gut microbiota to protect intestinal barrier integrity. Therefore, this study employed a multi-omics strategy combining TBI mouse models with clinical samples to systematically integrate metabolomic and microbiomic data. By assessing the intestinal barrier integrity and neuroinflammatory responses, we thoroughly investigated the effects of SO-loaded Lyc (SO-Lyc) on the gut-brain axis in TBI. This study aimed to elucidate the mechanisms of the gut-brain axis in TBI and propose innovative therapeutic strategies for TBI treatment.

Materials and methods

Clinical samples and ethical approval

Serum and fecal samples were collected from three healthy controls and nine patients with mild, moderate, or severe TBI (three patients in each group). All samples were obtained from Quanzhou First Hospital. This study was approved by the Ethics Committee of Quanzhou First Hospital (No. 2023K055), and all participants provided written informed consent. For clinical details, please refer to the Supplementary Materials.

Inclusion and exclusion criteria for clinical samples

Inclusion criteria for TBI patients: (1) age between 18 and 80 years, regardless of gender; (2) a clear history of head trauma meeting the diagnostic criteria for TBI; (3) admission within 12 h of injury; (4) first-time patients who have not received treatment; and (5) written informed consent from the patients or their family members.

Inclusion criteria for the healthy control group: (1) age between 18 and 80 years, regardless of gender; (2) no history of neurological, psychiatric, or gastrointestinal disorders; (3) no history of acute infection, fever, or diarrhea within one month prior to enrollment; and (4) signed informed consent.

Exclusion criteria: (1) concurrent neurological disorders; (2) severe injuries to other parts of the body; (3) severe gastrointestinal dysfunction or inability to tolerate oral or nasogastric feeding; (4) unstable vital signs; (5) pregnant or breastfeeding women; (6) recent (within 1 month) consumption of foods or medications containing probiotics; (7) use of antibiotics, immunosuppressants, or glucocorticoids within the past 3 months; (8) history of malignant tumors, severe cardiac, hepatic, or renal insufficiency, or uncontrolled endocrine or metabolic disorders.

Traumatic brain injury severity grading

The severity of TBI was graded comprehensively based on the Glasgow Coma Scale (GCS) score combined with the duration of loss of consciousness (LOC) and post-traumatic amnesia (PTA). Mild TBI was defined as a GCS score of 13–15, LOC lasting less than 30 min, and PTA lasting less than 24 h. Moderate TBI was characterized by a GCS score of 9–12, LOC ranging from 30 min to 24 h, and PTA lasting 24 h to 7 days. Severe TBI was indicated by a GCS score of 3–8, LOC exceeding 24 h, and PTA lasting longer than 7 days. For patients with unassessable verbal responses due to intubation or tracheotomy, the verbal response was noted as Vt, and total GCS scores ranged from 2 to 10 T, calculated from eye and motor responses, and marked with a T suffix (e.g., E3M5Vt was recorded as 8 T). TBI severity was assessed using the assessable E+M scores combined with LOC and PTA duration. All GCS evaluations were conducted by trained clinicians within 24 h post-injury, and confounding factors were recorded.

Animals

Male BALB/c mice, aged 8–10 weeks, were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China). The mice were kept under specific pathogen-free (SPF) conditions at a temperature of 20–24 °C, humidity of 40–60%, and a 12 h light-dark cycle. The animals were provided with an unlimited supply of drinking water and fed a standard laboratory diet. All animal care, use, and experimental protocols were approved by the Ethics Committee of Quanzhou Medical College (No. 2024060) and were performed according to the 3Rs principle to minimize the number of animals used and reduce suffering.

Traumatic brain injury model

Before model construction, mice were fasted for 12 h with free access to water. All animals were randomly assigned to different groups using Microsoft Excel: Sham (n = 6), TBI (n = 6), and TBI + SO-Lyc (n = 6). The controllable cortical impact (CCI) approach was used to induce TBI (Schwulst and Islam 2019; Cai et al. 2022). Mice were anesthetized (3% sodium pentobarbital, i.p. 50 mg/kg), followed by a 4-mm diameter craniotomy performed using a portable drill 1 cm to the right of the midline over the sensorimotor cortex and above the hippocampus. A controlled impact was delivered to the cortex using a pneumatic impactor (TBI-0310; Precision Systems and Instrumentation, Fairfax, VA, USA) with the following parameters: 1.5 m/s velocity, 100 ms dwell time, and 1.5 mm impact depth. The sham control group underwent a craniotomy without CCI. Subsequently, bone wax was used to seal the bone window, and the scalp layers were sutured sequentially. Body temperature (37.5 ± 0.5 °C) was maintained using an electric heating pad (69,001, RWD Life Science, China) throughout the procedure until the mouse regained consciousness and was able to walk.

Postoperatively, the mice were observed for locomotor activity and mental state. For the first three days, each mouse received 2 mg/kg of the analgesic meloxicam (1533, Qilu Animal Health Products Co., Ltd., China), along with a daily prophylactic dose of 100,000 U/kg penicillin (1292, HuaXu Animal Medicine Co., Ltd., China), to prevent infection. The sham and TBI groups did not receive any oral gavage intervention, whereas the TBI + SO-Lyc group received Lyc dissolved in SO (4 mg of Lyc dissolved in 1 mL of SO, yielding a final concentration of 4 mg/mL) via oral gavage at a dose of 10 mg/kg per day at a fixed time once a day for 14 consecutive days (Fu et al. 2020; Zhang et al. 2026). On the day following the end of the experiment, the mice were euthanized under anesthesia (2% sodium pentobarbital solution, i.p. 100 mg/kg). Blood, feces, ileal tissue (3 cm segment), and brain tissue were collected for subsequent analysis. All outcome assessments were conducted under blinded conditions; the researchers were unaware of the allocation of the experimental animals, and the results were analyzed by a dedicated statistician.

ELISA for inflammatory markers and gut mucosal integrity

Pro-inflammatory cytokines (tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), C-reactive protein (CRP), and interleukin-1 beta (IL-1β)), markers of intestinal mucosal integrity (diamine oxidase (DAO), D-lactic acid (D-LA), intestinal fatty acid-binding protein (I-FABP), and endotoxin (ET)), and markers of brain injury (neurofilament light chain (NFL), microtubule-associated protein tau (T-tau), glial fibrillary acidic protein (GFAP), S100 calcium binding protein B (S100B), neuron-specific enolase (NSE), and ubiquitin C-terminal hydrolase L1 (UCH-L1)) were detected using ELISA kits. The details of the kits used are provided in Table 1. The ELISA was conducted according to the manufacturer’s instructions. First, the test samples were diluted fivefold using a universal diluent. Subsequently, add 50 μL of the diluted sample and the corresponding standard solutions of varying concentrations to each well. Add 50 μL of the universal diluent to the blank well. Concurrently, add 100 μL of horseradish peroxidase-labeled detection antibody directly to each well. Cover the microplate with a sealing membrane and incubate at 37 °C for 60 min. Following washing, add 100 μL of HRP-conjugated working solution to each well. Cover the microplate with a sealing membrane and incubate at 37 °C for 30 min. Finally, the plate was washed, and the color was developed using the substrate solution. Absorbance was measured at 450 nm using a microplate reader. A linear fit was applied, using the formula y = ax + b (where x is the OD value). After obtaining the calibration curve from standard samples (diluted in a 2-fold gradient), the concentration of the sample was calculated (the sample was diluted 5-fold). Statistical analyses were performed using the mean and standard deviation, with two replicates per animal.

Table 1.

ELISA kit

Name of reagent	Manufacturer	Cat. No
Mouse TNF-α ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0132M1
Mouse IL-1β ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0040M1
Mouse IL-6 ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0163M1
Mouse CRP ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0074M1
Mouse I-FABP ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-44955M1
Mouse DAO ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0228M1
Mouse D-LA ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0885M1
Mouse endotoxin ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-44709M1
Mouse GFAP ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-0376M1
Mouse NFL ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-44894M1
Mouse NSE ELISA Kit	Jiangsu Meimian Industrial Co., Ltd	MM-54136M1
Mouse S100B ELISA Kit	Elabscience	E-EL-M1033
Mouse UCH-L1 ELISA Kit	Shanghai Hengyuan Biotechnology Co., Ltd	HB1895-Mu
Mouse T-TAU ELISA Kit	Elabscience	E-EL-M1289

qPCR

To detect mRNA expression, total RNA was extracted and purified using the RNA Isolate Total RNA Extraction Reagent (Vazyme, R401-01). The extracted RNA was quantified using a NanoDrop spectrophotometer (Invitrogen, USA). Subsequently, RNA was reverse-transcribed into cDNA using the HiScript II 1st Strand cDNA Synthesis Kit (Vazyme, R212-01). Next, qPCR was performed on an ABI 7300 Plus instrument (Applied Biosystems, USA) using the ChamQ SYBR Color qPCR Master Mix (Vazyme, Q431-02). The relative expression levels of the target genes were determined using the 2^{^(−ΔΔCt)} method, with 18S rRNA serving as the internal control. At least three biological replicates were used for each experiment. The primer sequences are listed in Table 2.

Table 2.

qPCR primer sequences

Gene name	Forward primer 5′-3′	Reverse primer 5′-3′
Fabp2	CACGTGGAAAGTAGACCGGA	GTCATGAGCTCCAAGCTTCCT
Aoc1	GCGCTATGTGGAAGGCTACTTC	CTCCAACTGCGTATTTCTGAGC
Gfap	CTAGCTACATCGAGAAGGTCCG	GAAGCTCCGCCTGGTAGACATCA
Nefl	CGCCGAAGAGTGGTTCAAGA	CTTTCCGACACCTCGTCCTT
Eno2	GATGTGGCTGCCTCTGAGTT	GTCGTCCTGGTCAAATGGGT
S100b	TTGCCCTCATTGATGTCTTCCAC	GGAAGTGAGAGAGCTCGTTGTTG
Uchl1	CGGCCCAGCATGAAAACTTC	TGTCTTGGTTGTTGGCCACT

H&E staining

Hematoxylin and eosin (H&E) staining was performed on paraffin-embedded tissue sections to evaluate tissue morphology. The tissue sections were first deparaffinized by incubating at 65 °C for 30 min, followed by sequential immersion in xylene (20 min each) and then in anhydrous ethanol (5 min each for two changes). After clearing with ethanol, the sections were stained with hematoxylin for 3–5 min, washed with 0.5% hydrochloric acid alcohol for 10 s, and rinsed with running water. The sections were then stained with eosin for 1 min, dehydrated using graded ethanol (75, 85, 95, and 100% for 5 min each), and cleared in xylene. Finally, the sections were mounted with neutral balsam and coverslips. The stained slides were examined using a Nikon E100 biological microscope. Statistical analysis of intestinal tissue damage was performed using Chiu’s grading score (Chiu et al. 1970; Heuer et al. 2025), whereas brain damage was assessed by counting the number of neurons.

Immunohistochemistry analysis

For immunohistochemistry (IHC), paraffin-embedded tissue sections were deparaffinized by immersion in xylene for 15 min (Xylene I) and another 15 min (Xylene II). Next, the slides were soaked in anhydrous ethanol for 5 min (Ethanol I), then in anhydrous ethanol for 5 min (Ethanol II), followed by 95% ethanol for 5 min, 85% ethanol for 5 min, and 75% ethanol for 5 min. Finally, the sections were washed three times with water and subjected to antigen retrieval for 30 min using 1× citrate buffer (pH 6.0) in a high-pressure cooker. The slides were then boiled at 90 °C, removed, and cooled to room temperature.

The sections were incubated with 3% hydrogen peroxide for 20 min at room temperature to block endogenous peroxidase activity. The slides were washed three times for 5 min each with phosphate-buffered saline (PBS), blocked with 15% bovine serum albumin (BSA) at room temperature for 30 min, and washed three times with PBS.

Primary anti-occludin antibody (Proteintech, 27,260–1-AP, RRID: AB_2880820) and zonula occludens-1 (ZO-1) antibody (Proteintech, 21,773–1-AP, RRID: AB_10733242) were diluted with PBS and applied to the sections at a final concentration of 10 µg/mL, followed by incubation overnight at 4 °C. The slides were then washed three times with PBS. HRP-conjugated secondary antibodies (HRP goat anti-rabbit, Pinuofei Biotechnology, PN0046, RRID: AB_3683113) were applied to the slides and incubated at 37 °C for 1 h. After three rounds of 5 min washes with PBS, the chromogenic reaction was performed by adding DAB chromogen solution. The reaction was stopped by rinsing the slides with water and washing for 5 min. Counterstaining was performed using hematoxylin for 3–5 min, followed by washing with water for 2 min. The slides were dehydrated in graded ethanol solutions (75%, 85%, 95%, and 100%) and cleared in xylene. Finally, the sections were mounted with neutral balsam and coverslips.

Images were captured using a Nikon ECLIPSE-Ci microscope. IHC results were obtained and analyzed using Image-Pro Plus 6.0 software. The integrated optical density (IOD) was calculated for each section, and the average optical density was used to measure occludin and ZO-1 expression levels.

16S rRNA sequencing for gut microbiota analysis

16S rRNA sequencing was performed on 12 clinical samples (three samples per group) and nine animal samples (three samples per group). Fecal samples were processed for genomic DNA extraction, and their quality was confirmed by 1% agarose gel electrophoresis. The V3–V4 hypervariable regions of the prokaryotic 16S rRNA gene were amplified using universal primers and TransStart FastPfu DNA Polymerase (TransGen, AP221-02). PCR products were checked by 1% agarose gel electrophoresis to ensure the size of the expected product and were purified using the Agencourt AMPure XP kit. The primer pair targeting the V3‑V4 region was 341F_806R: 341 Forward primer-5′- CCTAYGGGRBGCASCAG-3′, 806 Reverse primer-5′- GGACTACNNGGGTATCTAAT-3′, which target the V3–V4 region.

For library construction, the “Y”-shaped adapters were ligated to the PCR products, and adapter-dimer fragments were removed with magnetic beads. Library templates were enriched via PCR amplification. Sodium hydroxide was used during the denaturation step to produce single-stranded DNA fragments. The final libraries were subjected to the G99 sequencing platform using paired-end sequencing (PE300, 2 × 300 bp). The average sequencing depth was approximately 50,000 reads per sample.

Raw FASTQ sequencing files were processed using the QIIME2 pipeline, including quality filtering and denoising procedures. Representative sequences were clustered into Operational Taxonomic Units (OTUs), and taxonomic annotation was performed using the RDP Classifier algorithm against the SILVA database (Release 128, http://www.arb-silva.de), with a confidence threshold of 0.7. Alternative classification methods, such as BLAST (v2.6.0 +) with an E-value threshold of 1e−5 and the UCLUST consensus taxonomy assigner, were also applied where appropriate. Taxa enrichment analysis was performed using linear discriminant analysis effect size (LEfSe), and functional prediction was conducted using PICRUSt2.

Metabolomic profiling

Untargeted metabolomic profiling was performed on 12 clinical samples (three samples per group) and nine animal samples (three samples per group). The frozen fecal samples (50 mg each) were melted at 4 °C and homogenized using two steel beads in 400 µL of prechilled methanol–water (4:1, v/v). Samples were then homogenized at − 10 °C for 5 min, stored at − 20 °C for 30 min, and the supernatants were collected after centrifugation at 16,000 g for 20 min at 4 °C. Liquid chromatography–mass spectrometry (LC–MS/MS) was performed using a SHIMADZU-LC30 UHPLC system and an ACQUITY UPLC® HSS T3 column (2.1×100 mm, 1.8 µm). The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). Gradient elution was performed using mobile phases in the positive and negative ion modes on a QEPlus (Thermo Scientific) mass spectrometer coupled with electrospray ionization (ESI). The samples were analyzed in a full scan (75–1050 m/z), MS2 (normalized collision energies of 20, 30, and 40). Peaks were aligned, retention times were corrected, and peak areas were extracted using MSDIAL software (for peak alignment). Metabolites were identified through a combination of accurate mass matching and secondary spectral matching using the public databases HMDB, MassBank, and GNPS.

Metabolites were identified based on accurate mass (mass tolerance < 10 ppm) and MS/MS spectra (mass tolerance < 0.02 Da), and were matched against HMDB, MassBank, and GNPS. Features with > 50% nonzero values in at least one group were retained.

The data were mean-centered and Pareto scaled. PCA, PLS-DA, and OPLS-DA models were constructed, and model overfitting was assessed by 200-permutation tests. The model quality was evaluated using cumulative R²X, R²Y, and Q² values, with the permuted models showing lower R² and Q² values than those of the original models.

Differential metabolites were identified using VIP scores (> 1.0) from OPLS-DA and a two-tailed Student’s t-test (p < 0.05). The fold change was calculated as the logarithm of the mean area ratio between the groups. Differential metabolites were subjected to clustering and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses using Fisher’s exact test. p values were adjusted for multiple testing using the Benjamini–Hochberg procedure, and an adjusted p < 0.05 was considered significant.

Correlation analysis between metabolites and gut microbiota

To analyze the association between the key metabolites in the linoleic acid metabolism pathway (12,13-epoxy-9-hydroxy-10-octadecenoate and 12,13-diHOME) and microbial taxa, distance matrices were constructed using Euclidean distance for metabolites and Bray–Curtis distance for microbes. Correlations between the matrices were evaluated using the Mantel test with 1,000 permutations. p values were adjusted for multiple testing using the Benjamini–Hochberg procedure, and associations with an adjusted p < 0.05 were considered significant.

Statistical analysis

Statistical analysis was performed using GraphPad Prism (version 8.0) software, and comparisons between multiple groups were conducted using one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. All experiments were repeated three times, with each group comprising at least three biological replicates. Data were presented as mean ± standard deviation (SD), and where data did not meet the assumption of normal distribution, non-parametric tests (such as the Mann–Whitney U test) were employed. For the 16S rRNA sequencing data analysis, Kruskal–Wallis and Wilcoxon rank-sum tests were used for non-normally distributed data. For metabolomic data analysis, both univariate (Student’s t-test) and multivariate (OPLS-DA) approaches were applied. Significant metabolites were identified using the VIP scores from the OPLS-DA model. A p value less than 0.05 was considered statistically significant.

Results

Sunflower oil-loaded lycopene mitigates intestinal barrier damage following traumatic brain injury

To determine whether SO-Lyc alleviates intestinal barrier damage in TBI, we performed a histopathological analysis of the ileal tissue using H&E staining (Fig. 1A). The ileum in the sham group showed intact histology, and normal villi were present, whereas the TBI group exhibited extensive damage to the villi due to cellular degeneration and loss of villus integrity. Chiu’s grading scores further supported this finding. However, following SO-Lyc treatment, the TBI + SO-Lyc group showed partial repair of the villus structures, indicating protection of intestinal integrity after TBI. We then stained the proteins responsible for maintaining the integrity of the epithelial barrier (occludin and ZO-1) (Fig. 1B). In the TBI group, the expression levels of occludin and ZO-1 were significantly reduced. In contrast, the TBI + SO-Lyc group showed improved expression levels of both proteins, as evidenced by increased average optical density (AOD) values (p < 0.05). These results indicated that SO-Lyc enhanced the intestinal mucosal barrier function following TBI.

Fig. 1.

Lycopene treatment mitigates intestinal barrier damage in traumatic brain injury. A H&E staining of ileal tissue from the Sham (SD = 0.52), TBI (SD = 0.63, p = 0.0004, Cohen’s d = 6.35, vs. Sham), and TBI + SO-Lyc (SD = 0.84, p = 0.0465, Cohen’s d = 3.37, vs. TBI) groups. Magnification: 400× (scale bar = 50 μm, top), 200× (scale bar = 100 μm, bottom). B Immunohistochemical staining of occludin and ZO-1 in ileal tissue from the Sham (SD = 0.52 and 0.08), TBI (SD = 0.10 and 0.07, both p < 0.0001, Cohen’s d = 3.13 and 4.00, vs. Sham), and TBI + SO-Lyc groups (SD = 0.06 and 0.07, p = 0.0416 and 0.0258, Cohen’s d = 1.52 and 1.78, vs. TBI). Magnification: 200×. scale bar = 100 μm. C ELISA analysis of serum biomarkers of intestinal barrier dysfunction (DAO, D-LA, I-FABP, and ET). TBI versus Sham: all p-values were < 0.0001, Cohen’s d values were 4.39, 4.51, 8.15 and 8.12 respectively. TBI versus TBI + SO-Lyc: p = 0.0155, < 0.0001, < 0.0001 and < 0.0001, respectively; Cohen’s d values were 1.93, 2.91, 4.03 and 3.76 respectively. D qPCR analysis of mRNA expression of Aoc1 (p < 0.0001) and Fabp2 (p < 0.0001), markers of intestinal barrier dysfunction. n = 6; one section per animal; statistics based on three fields of view. p values were calculated using one-way ANOVA with post-hoc Tukey HSD tests. *p < 0.05, ***p < 0.001, ****p < 0.0001

The serum concentrations of intestinal barrier biomarkers (DAO, D-LA, I-FABP and ET) were measured by ELISA, further confirming the protective action of SO-Lyc (Fig. 1C). The TBI group exhibited significantly higher levels of DAO, D-LA, I-FABP, and ET, biomarkers of intestinal barrier dysfunction. In the TBI + SO-Lyc group, the levels of these biomarkers were significantly lower (p < 0.05). Similarly, qPCR revealed that Aoc1 and Fabp2 were significantly lower in the TBI + SO-Lyc group compared to the TBI group, reflecting the levels of DAO and I-FABP, respectively, in accordance with the ELISA results (Fig. 1D). These results indicated that SO-Lyc exerted a protective effect against TBI-induced intestinal barrier damage.

Sunflower oil-loaded lycopene reduces neuronal loss and serum levels of brain injury biomarkers

To clarify the influence of SO-Lyc on the gut-brain axis, we analyzed its effect on brain tissue damage. Histological staining of brain tissue revealed severe neuronal injury, neurodegeneration, and loss of integrity (Fig. 2A and B). Compared to the TBI group, the TBI + SO-Lyc group exhibited neuronal recovery (p < 0.05) with less severe neuronal damage and preservation of neural architecture, suggesting that SO-Lyc may exert a neuroprotective effect in the context of TBI.

Fig. 2.

Neuroprotective effects of lycopene in traumatic brain injury-induced brain injury. A H&E staining of brain tissue from the Sham, TBI, and TBI + SO-Lyc groups. B Statistical analysis of neurons in panel A; TBI (SD = 0.63) versus Sham (SD = 0.52): p = 0.0005, Cohen’s d = 6.35. TBI versus TBI + SO-Lyc (SD = 0.84): p = 0.0497, Cohen’s d = 3.37. Magnification 200×, scale bar = 100 μm. C Levels of inflammatory biomarkers reflecting brain injury (NFL, T-tau, GFAP, S100B, NSE, and UCH-L1). TBI versus Sham: all p-values were < 0.0001 except for T-tau (p = 0.0001); Cohen’s d values were 5.10, 3.65, 7.68, 6.16, 5.78 and 10.08 respectively. TBI versus TBI + SO-Lyc: p < 0.0001, = 0.2343, = 0.0003, < 0.0001, < 0.0001 and < 0.0001 respectively; Cohen’s d values were 3.96, 0.98, 3.30, 3.10, 5.70 and 7.54 respectively. D qPCR analysis of mRNA expression of the inflammatory biomarkers Nefl (NFL), Gfap, S100b, Eno2 (NSE) and Uchl1, associated with brain injury. All p-values were < 0.0001. E Levels of pro-inflammatory cytokines (TNF-α, IL-6, CRP, and IL-1β). TBI versus Sham: all p-values were < 0.0001; Cohen’s d values were 8.71, 6.02, 8.81 and 9.04 respectively. TBI versus TBI + SO-Lyc: all p-values were < 0.0001 except for IL-6 (p = 0.0002); Cohen’s d values were 3.99, 2.95, 4.14 and 5.04 respectively. n = 6; one section per animal; statistics based on three fields of view. p values were calculated using one-way ANOVA with post-hoc Tukey HSD test. *p < 0.05, ***p < 0.001, ****p < 0.0001

Additionally, compared to the sham group, serum levels of biomarkers reflecting brain injury (NFL, T-tau, GFAP, S100B, NSE, and UCH-L1) were elevated in the TBI group (Fig. 2C). Following SO-Lyc treatment in the TBI + SO-Lyc group, biomarker levels decreased significantly, with p values below 0.05 for NFL, GFAP, S100B, NSE, and UCH-L1. Furthermore, qPCR analysis revealed that the mRNA expression levels of Nefl, Gfap, S100b, Eno2, and Uchl1 were consistent with the ELISA results, indicating that the levels of NFL, GFAP, S100B, NSE, and UCH-L1 were significantly lower in the TBI + SO-Lyc group than in the TBI group (Fig. 2D). Meantime, levels of all pro-inflammatory cytokines (TNF-α, IL-6, CRP, and IL-1β) were elevated, suggesting a strong inflammatory response in the TBI group (Fig. 2E). All inflammatory marker concentrations were significantly reduced in the TBI + SO-Lyc group (p < 0.05), indicating that SO-Lyc successfully downregulated the inflammatory response after TBI. This further suggests that SO-Lyc may exert neuroprotective effects against TBI.

The association between inflammatory metabolites and gut microbiota in traumatic brain injury treated with sunflower oil-loaded lycopene

To investigate the mechanism of action of SO-Lyc in the treatment of TBI, we performed metabolomic analysis and 16S rRNA sequencing on serum and fecal samples from animal models treated with SO-Lyc.

Firstly, we conducted rigorous quality control and multivariate analysis on the metabolomics and 16S rRNA sequencing data. The metabolomics analysis revealed metabolites that were significantly altered between the different groups (Fig. 3). Meanwhile, the results of the 16S rRNA analysis indicated that we captured a large number of microbial communities, with distinct differences in microbial diversity and community structure observed between the different groups (Fig. 4). Subsequently, a combined analysis of metabolomic and microbiomic data was applied to identify significantly altered microorganisms and metabolites in TBI. KEGG pathway enrichment analysis based on microbiomics indicated that TBI mainly affected pathways related to amino acid metabolism, oxidative stress, and inflammation (Fig. 5A). These pathways included taurine and hypotaurine metabolism, alanine, aspartate and glutamate metabolism, and glutathione metabolism. Taurine and hypotaurine metabolism plays an important role in antioxidant defense and inflammatory regulation (Mizota et al. 2022). Taurine acts as an endogenous protective molecule by reducing oxidative stress and modulating immune cell activity (Marcinkiewicz and Kontny 2014). Therefore, altered taurine metabolism after TBI may indicate reduced antioxidant capacity and increased inflammatory responses. Reduction of this pathway disturbance after SO-Lyc treatment suggests that Lyc supplementation may help restore redox balance in TBI. Glutathione metabolism is another key antioxidant pathway involved in maintaining intracellular redox homeostasis (Hristov 2022), which was significantly disrupted in TBI. Since oxidative stress is a major mechanism underlying TBI progression, restoration of glutathione metabolism may contribute to the protective effects of SO-Lyc.

Fig. 3.

Multivariate analysis and QC of metabolomic profiles for mouse fecal samples. A Principal component analysis (PCA) of all samples, showing clustering patterns among groups. B T2 quality control plot of all samples. C–D Two-dimensional score plots of PLS-DA (C) and OPLS-DA (D) comparing TBI mouse group (F-T) versus control mouse group (F-S). E VIP-selected metabolites from F-T versus F-S, highlighting significantly altered features. F–G PLS-DA (F) and OPLS-DA (G) score plots comparing lycopene treated TBI mouse group (F-T-L) versus F-T group. H VIP-selected metabolites from F-T-L versus F-T

Fig. 4.

Quality control and diversity analysis of gut microbiota based on 16S rRNA sequencing for mouse fecal samples. A Distribution of clean tags across all samples after quality filtering, indicating sequencing depth. B Venn diagram showing shared and unique OTUs among different groups. C Rarefaction curves demonstrating sequencing depth sufficiency and species richness. D Species accumulation curve indicating that the sampling effort was sufficient to capture the majority of microbial diversity. E Rank abundance curves illustrating species richness and evenness across samples. F Alpha diversity indices (e.g., Shannon, Chao1) comparing within-sample microbial diversity among groups. G–I Beta diversity analyses based on PCA (G), PCoA (H), and NMDS (I), showing differences in microbial community structure among groups. J–K Phylogenetic distance matrices based on unweighted J and weighted K UniFrac metrics. L Relative abundance bar plot showing taxonomic composition of gut microbiota at the genus level. M Heatmap displaying variation in microbial composition across samples

Fig. 5.

Metabolomic and microbiota analysis in animal models. A Metabolic pathways associated with inflammation and immune regulation. B Correlation between metabolic alterations and gut microbiota composition in animal models

We next applied the Mantel test to investigate the correlations between changes in the microbial community and metabolic profiles. Taurine and L-glutamate were significantly correlated with changes in microbial populations (Fig. 5B), supporting that gut microbiota changes may participate in metabolic recovery identified in microbiomics after SO-Lyc supplementation. These findings indicate that the effects of SO-Lyc may involve regulation of both gut microbial composition and host metabolic pathways, rather than relying only on direct antioxidant effects.

An analysis of the association between metabolomics and microbiomics in patients with TBI

To further investigate whether inflammatory metabolites and the gut microbiota play a significant role in patients with TBI, we collected serum and fecal samples from patients with TBI of varying severity (controls (N-F), heavy (H-F), medium (M-F), and light (L-F)) (Fig. S1).

Metabolomics quality control and multivariate analysis revealed significant differences in metabolite abundance among patients with varying degrees of the condition (Fig. 6). The results of 16S rRNA analysis also indicated that microbial abundance varied among patients with TBI of varying severity (Fig. 7). Metabolic profiling of human TBI patients revealed systemic inflammatory metabolic changes, including alterations in lipid mediator linoleic acid metabolism, sphingolipid signaling, and tryptophan metabolism (Fig. 8A). Linoleic acid metabolism is involved in the production of lipid mediators and inflammatory responses (Hennig et al. 2006). Changes of this pathway may alter the levels of pro-inflammatory and inflammation-resolving lipid mediators after tissue injury. Sphingolipid signaling also plays a role in immune activation, apoptosis, and neuroinflammation (Nixon 2009). Changes in sphingolipid metabolism have been reported in neurological disorders and may promote inflammatory responses after brain injury (Mondal et al. 2024). Tryptophan metabolism was reported to be associated with immune responses through metabolites involved in immune cell activation and immune tolerance (Seo and Kwon 2023). Alterations in metabolism observed in TBI patients may reflect systemic immune dysregulation associated with traumatic injury.

Fig. 6.

Multivariate analysis and QC of metabolomic profiles for human fecal samples. A Principal component analysis (PCA) of all samples, showing clustering patterns among groups. B Sample distribution in T2 quality control plots. C–D Score plots of PLS-DA (C) and OPLS-DA (D) comparing the heavy TBI patient group (H-F) and the light TBI patient group (L-F). E–F Volcano plot (E) and VIP-selected metabolites (F) for H-F versus L-F, highlighting significantly altered features. G–H PLS-DA (G) and OPLS-DA (H) score plots for H-F versus healthy controls (N-F) comparison. I VIP-selected metabolites for H-F versus N-F. J–K PLS-DA (J) and OPLS-DA (K) score plots for L-F versus N-F. L VIP-selected metabolites for L-F versus N-F. M–N PLS-DA (M) and OPLS-DA (N) score plots for medium TBI patient group (M-F) versus N-F comparison. O Volcano plot showing differential metabolites for M-F versus N-F

Fig. 7.

Quality control and diversity analysis of gut microbiota based on 16S rRNA sequencing for human fecal samples. A Distribution of clean tags across all samples, showing sequencing depth after quality filtering. B OTU Venn diagram illustrating the number of shared and unique OTUs among different groups. C Rarefaction curves demonstrating sequencing depth sufficiency and species richness. D Species accumulation (specaccum) curve indicating that the sample size is sufficient to capture the majority of microbial diversity. E Rank abundance curves showing species richness and evenness across samples. F Alpha diversity indices (e.g., Shannon, Chao1) comparing within-sample microbial diversity among groups. G–I Beta diversity analyses based on PCA (G), PCoA (H), and NMDS (I), illustrating differences in microbial community structure among groups. J–K Distance matrices based on unweighted (J) and weighted (K) UniFrac metrics, reflecting phylogenetic differences among microbial communities. L Taxonomic composition of gut microbiota at the genus level shown as a relative abundance bar plot. M Heatmap of differential taxa illustrating variations in microbial composition across samples

Fig. 8.

Metabolomic and microbiota analysis in clinical TBI samples. A KEGG pathway enrichment analysis of differential metabolites between the TBI and normal control groups. B-C Correlation between 16S rRNA sequencing results of gut microbiota composition and metabolomic data

Meantime, we applied LEfSe analysis to identify taxa that were differentially enriched in the different groups. Two members of the Ruminococcaceae family, Ruminococcus gnavus and Ruminococcaceae UBA1819 were significantly enriched in the medium TBI patient group (M-F) (Fig. S2A). Ruminococcus gnavus has been reported to be associated with inflammatory bowel diseases (Crost et al. 2023), while Ruminococcaceae UBA1819 has been reported to be increased in patients with Parkinson’s disease (Proano et al. 2023). We also used PICRUSt2 to predict functional alterations in microbial metabolic and signaling pathways across groups. Notably, several inflammation-related pathways showed increased abundance in the medium (M-F) and heavy (H-F) TBI patient groups, including lipopolysaccharide biosynthesis, the NOD-like receptor signaling pathway, bacterial invasion of epithelial cells, and the bacterial secretion system (Fig. S2B and C). Although not all differences were statistically significant, a consistent increasing trend was observed in patients with TBI compared with healthy controls (Fig. S2D). These findings suggest that microbiota with pro-inflammatory properties, such as Ruminococcus gnavus, may play a significant role in the intermediate stages of disease progression, thereby contributing to the initiation or exacerbation of inflammation. In particular, linoleic acid metabolism was highly correlated with differential microbiota composition. The clinical cohort was not designed to assess the therapeutic effects of SO-Lyc. However, several metabolic pathways altered in human TBI overlapped with those regulated by SO-Lyc in mice. This overlap suggests that these metabolic pathways may represent potential targets for TBI treatment. We observed that the metabolites involved in linoleic acid metabolism, especially 12,13-DiHOME and related lipid metabolites, were highly correlated with microbial community differences (p < 0.05), particularly when comparing TBI and control groups (Fig. 8B and C). This suggests that gut microbiota composition may be involved in the metabolic alterations associated with SO-Lyc treatment in TBI.

The clinical cohort was not designed to evaluate the therapeutic efficacy of SO-Lyc. However, patient-derived samples allowed us to examine whether metabolic pathways associated with inflammatory response and immune modulation rescued by SO-Lyc in mice were also altered in human TBI. Both mouse and human datasets showed similar changes in inflammation-related metabolic pathways. In mice, TBI-associated microbial functional changes were mainly linked to antioxidant and amino acid metabolism pathways, including taurine, glutathione, and glutamate metabolism. In human TBI patients, metabolomic analysis revealed changes in lipid mediator-related pathways, including linoleic acid metabolism and sphingolipid signaling.

These findings indicate that TBI causes metabolic disturbances involving oxidative stress, immune responses, and microbiota-related metabolism.

Discussion

TBI is a leading cause of morbidity and mortality worldwide and typically induces severe neurological and functional deficits (Coronado et al. 2012; Roozenbeek et al. 2013). Previous studies have shown that the pathophysiology of TBI involves protracted processes related to inflammation, oxidative stress, and gut microbiota dysbiosis (Zhu et al. 2018). This study used animal models and clinical samples from patients with TBI to investigate the combined effects of SO loaded with the antioxidant Lyc on inflammatory responses, intestinal barrier function, and dysbiosis. These findings provide crucial evidence for elucidating how SO-Lyc mitigates inflammation and microbial disruption in the context of TBI.

Lyc is a carotenoid with a carbon skeleton of approximately 40 atoms, featuring a straight unsaturated hydrocarbon chain composed of 11 conjugated and two non-conjugated carbon–carbon (C=C) double bonds (Shi and Le Maguer 2000). It has been shown to possess potent antioxidant properties, anti-inflammatory effects, and the ability to regulate blood lipids (Yang et al. 2018; Ugbaja et al. 2021). Several studies have used SO as a solvent for Lyc, which can improve its stability and bioavailability (Flais et al. 2025; Lin et al. 2025). However, its combined effects in TBI remain unclear. Our findings indicate that SO-Lyc significantly mitigated the inflammatory response following TBI in mice, as evidenced by reduced levels of TNF-α, IL-6, CRP, and IL-1β. Multiple studies have shown that TNF-α, IL-6, CRP, and IL-1β are key pro-inflammatory mediators in the neuroinflammation triggered by brain injury (Xu et al. 2022; Yan et al. 2022). Importantly, alleviation of neuroinflammation remains a promising therapeutic approach for TBI. Therefore, our research further demonstrates that SO-Lyc may represent a potential therapeutic strategy for treating TBI. This agrees with the findings of Zhao et al. (Zhao et al. 2018; Chen et al. 2024; Zhang et al. 2026).

Meanwhile, multi-omics data analysis showed that SO-Lyc modulated metabolic pathways related to inflammation and oxidative stress. In the mouse model, TBI altered taurine, glutathione, amino acid, and lipid metabolism, indicating impaired antioxidant defense and immune regulation. Surprisingly, we found that administration of SO-Lyc to mice reversed these metabolic alterations, including those in linoleic acid metabolism, sphingolipid signalling and tryptophan metabolism—pathways closely associated with immune and inflammatory responses, oxidative stress, and free radical activity (Maceyka and Spiegel 2014; Viladomiu et al. 2016; Etienne-Mesmin et al. 2017). Specifically, linoleic acid can be further metabolized into arachidonic acid, leading to the synthesis of potent inflammatory mediators, such as thromboxane, prostaglandin, and leukotriene (Granström, 1984). These findings further suggest that SO-Lyc may exert a protective effect against TBI by modulating metabolic pathways in response to inflammation and oxidative stress.

In addition to its anti-inflammatory effects, alterations in the gut microbiota and improved intestinal barrier function in TBI mice may also constitute the primary effects of SO-Lyc. Chen et al. found that Lyc improves gut microbiota composition by promoting the growth of beneficial bacteria, such as the Akkermansia genus. It enhances intestinal barrier integrity by increasing the expression of tight junction proteins and reduces local inflammation by inhibiting the LPS-TLR4/MyD88 signaling pathway (Chen et al. 2025). Aldamarany et al. found that SO alleviates inflammation and oxidative stress by modulating gut microbiota composition in high-fat diet-fed mice (Aldamarany et al. 2023). In this study, we found that SO-Lyc effectively restored intestinal villus integrity by upregulating the expression of tight junction proteins, including occludin and ZO-1. Occludin and ZO-1 are crucial for maintaining epithelial barrier integrity (Anderson 2001). Concurrently, our 16S rRNA sequencing data revealed changes in gut microbiota composition following TBI. Interestingly, SO, rich in Lyc, restored gut microbiota to a more balanced state, reducing the abundance of bacteria with pro-inflammatory properties, while promoting the growth of beneficial microorganisms. This aligns with the findings of other researchers that gut microbiota influences brain function via the gut-brain axis (Agirman Hsiao 2021), further supporting the idea that restoring healthy gut microbiota may play a crucial role in TBI recovery.

The clinical significance of our results was further supported by the characterization of clinical samples that showed similar metabolic changes and microbial alterations in patients with TBI. Similar to animal models, patients with TBI exhibited significant alterations in high-level metabolic pathways associated with systemic inflammation. Moreover, compositional changes in the gut microbiota and inflammatory metabolites found in clinical samples were similar to those observed in animal models. This reveals that inflammatory metabolites and the gut microbiota play a significant role in patients with TBI, whilst SO-Lyc can influence inflammatory metabolites and the gut microbiota in mice with TBI; in view of this, we believe that SO-Lyc may be one of the potential approaches for treating TBI.

However, this research still has a multitude of limitations. Studies have shown that Lyc retains its neuroprotective effects even after prolonged administration. In a study by Wang et al., mice treated with Lyc for up to three months still showed observable effects on neuronal cells, revealing that Lyc could improve neuronal degeneration, mitochondrial dysfunction, and synaptic damage (Wang et al. 2024). However, the treatment in this study lasted only two weeks. Therefore, further studies to elucidate the long-term effects of SO-Lyc under TBI conditions and its potential neuroprotective responses are necessary. Moreover, although our animal model data indicate that SO-Lyc possesses beneficial effects, further research is needed to elucidate the specific molecular mechanisms by which SO-Lyc modulates inflammatory responses, metabolism, and gut microbiota. This includes determining whether it regulates inflammation- and oxidative stress-related pathways, such as NF-κB, Nrf2, or MAPK. In particular, some studies have shown that Lyc has neuroprotective effects, helping to alleviate inflammation and oxidative stress, and improve learning and memory (Fu et al. 2020; Ugbaja et al. 2021; Chen et al. 2025); therefore, this study should also further investigate the effects of SO-Lyc on behavioural and neurological outcomes (such as cognitive and motor function). Concurrently, employing more sophisticated microbiome analytical approaches, such as metagenomic sequencing and functional microbiome profiling, may elucidate the precise mechanisms through which SO-Lyc exerts these beneficial effects, specifically which microbial taxa interact with metabolites. Such insights could provide a foundation for the development of microbiota-targeted therapies for TBI. Another matter that should not be overlooked is that, as we have previously pointed out, several components in SO possess antioxidant properties; future research should investigate which active substance in SO-Lyc plays the principal role in TBI.

In summary, our work shows that SO-Lyc exerts protective and anti-inflammatory effects after TBI by regulating inflammatory pathways, the gut barrier, and the recovery of the gut microbiota. Metabolomic and microbiome analyses provide a comprehensive view of the mechanisms of action of SO-Lyc. The positive effects in both animal models suggest that SO-Lyc could represent an innovative, effective, and therapeutic intervention for the treatment of TBI and other neuroinflammatory diseases. Further research is required to validate these findings and to explore the specific roles of SO-Lyc in clinical applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

TBI

Traumatic brain injury

Lyc

Lycopene

SO

Sunflower oil

GCS

Glasgow coma scale

LOC

Loss of consciousness

PTA

Post-traumatic amnesia

SPF

Specific pathogen-free

CCI

Controllable cortical impact

TNF-α

Tumor necrosis factor-alpha

IL-6

Interleukin-6

CRP

C-reactive protein

IL-1β

Interleukin-1 beta

DAO

Diamine oxidase

D-LA

D-lactic acid

I-FABP

Intestinal fatty acid-binding protein

ET

Endotoxin

NFL

Neurofilament light chain

T-tau

Microtubule-associated protein tau

GFAP

Glial fibrillary acidic protein

S100B

S100 calcium binding protein B

NSE

Neuron-specific enolase

UCH-L1

Ubiquitin C-terminal hydrolase L1

H&E

Hematoxylin and eosin

IHC

Immunohistochemistry

PBS

Phosphate-buffered saline

BSA

Bovine serum albumin

ZO-1

Zonula occludens-1

IOD

Integrated optical density

OTUs

Operational taxonomic units

LEfSe

Linear discriminant analysis effect size

LC–MS/MS

Liquid chromatography–mass spectrometry

ESI

Electrospray ionization

KEGG

Kyoto encyclopedia of genes and genomes

ANOVA

Analysis of variance

SD

Standard deviation

AOD

Average optical density

PCA

Principal component analysis

Author contributions

Tianlai Lin:Conceptualization, Data Curation, Formal Analysis, Software, Writing—Original Draft, Funding acquisition. Qiang Liu and Mouying Liu: Formal analysis, Visualisation, Validation, Investigation, Resources. Xizhe Chen, Ling Huang and Sican Wei: Methodology, Investigation, Data Curation. Hongling Zhang: Conceptualization, Project Administration, Supervision, Writing—Review & Editing and Funding Acquisition.

Funding

This work was supported by the grant from Natural Science Foundation of Fujian Province (Grant number: 2023J011769, 2023J011779), Health and Healthcare Science and Technology Program of Fujian Province (Grant number: 2024GGB14), Science and Technology Planning Project of Quanzhou (Grant number: 2024NY018) and Fujian Normal University & Quanzhou First Hospital Medical-Education Integration Open Fund (Grant number: 2025YJRHQH09).

Data availability

The datasets generated and/or analyzed during the current study are available from the China National Bioinformatics Centre repository in raw FASTQ, “URL: https://ngdc.cncb.ac.cn/gsa/s/Otd6eX6C and https://ngdc.cncb.ac.cn/gsa/s/83aFKwr9”. Should you have any questions, please contact the corresponding author [zhanghongling@fjmu.edu.cn].

Declarations

Conflict of interest

There is no conflict of interest between all the authors.

Ethical approval

The treatment of animals in this experiment complied with the Guidelines for the Care and Use of Laboratory Animals. The animal experimentation protocol was reviewed and approved by the Quanzhou Medical College Laboratory Animal Ethics Committee (Approval No. 2024060). This study was reported in accordance with the ARRIVE guidelines. Ethical approval for the collection of clinical samples was obtained from the Ethics Committee of Quanzhou First Hospital (Approval No. 2023K055). All procedures were conducted following the acquisition of written informed consent from patients and adhered to the principles of the Declaration of Helsinki.

Consent for publication

All authors consent to publication. All images published have been released with the patients’ consent.