Vancomycin (Batch no. S17059), Metronidazole (S17079), Colistin (S17057), Gentamicin (V30125), and Clindamycin (B61748) were purchased from Shanghai Yuanye Bio-Technology Co., Ltd. The RNAprep Pure Tissue Total RNA Isolation Kit (DP431) was obtained from TIANGEN. The Fecal DNA Extraction Kit (D4015010000D01V051) was supplied by OMEGA. Premix ExTaq™ (1927827100), probes, and primers (AK81837A) were acquired from Sangon Biotech (Shanghai) Co., Ltd. First-Strand Synthesis Master Mix and cDNA reverse transcription kits (0202122021) were from Lamborlyde Biotechnology. qPCR SYBR Green Master Mix (Low Rox Plus) (H8216980) was purchased from YESEN. Pentobarbital sodium (P3761) was obtained from Sigma-Aldrich Co., Ltd. The Clostridium difficile 027 (BI/NAP1/027) strain originates from the Department of Clinical Laboratory, Beijing Friendship Hospital, Capital Medical University.

Zingiberis Rizoma Recens (Zingiber officinale Rosc., batch number: 20110404) was sourced from Beijing Renwei Herbal Pieces Factory, China. Zingiberis Rhizoma (Zingiber officinale Rosc., batch number: 20093003) and Jujubae Fructus (Ziziphus jujuba Mill., batch number: 201105005) were from Beijing Qiancao Chinese Herbal Pieces Co., Ltd. Ginseng Radix et Rhizoma (Panax ginseng C. A. Mey., batch number: 1906005) was obtained from Beijing Tongchuntang Hospital of Traditional Chinese Medicine. Glycyrrhizae Radix et Rhizoma Praeparata Cum Melle (Glycyrrhiza uralensis Fisch., batch number: 200516) was provided by Beijing Yanbei Herbal Pieces Factory. Coptidis Rhizoma (Coptis chinensis Franch., batch number: 1912205) and Scutellariae Radix (Scutellaria baicalensis Georgi, batch number: 1912204) were supplied by Beijing Shuangqiao Yanjing Chinese Herbal Pieces Factory. Pinelliae Rhizoma Praeparatum (Pinellia ternate (Thunb.) Breit., batch number: 1231144) was from Beijing Tongrentang Co., Ltd., China.

SXD was prepared according to established methods (Cui et al., 2023). The formulation contains eight herbal components: Zingiberis Rhizoma Recens 12 g, Glycyrrhizae Radix et Rhizoma Praeparata cum Melle 9 g, Ginseng Radix et Rhizoma 9 g, Zingiberis Rhizoma 3 g, Scutellariae Radix 9 g, Pinelliae Rhizoma 9 g, Coptidis Rhizoma 3 g, and Jujubae Fructus 9 g. The herbs are first immersed in pure water at a solid-to-liquid ratio of 1:8 (w/v) for 30 minutes, followed by decoction for 30 minutes. The resulting filtrate is collected, and the residue is boiled again with water at a ratio of 1:6 (w/v) for an additional 30 minutes. The two filtrates are then combined and concentrated to a specified concentration. The prepared solution was stored at 4°C until use.

Eighty-four male C57BL/6 mice aged 5 weeks and weighing 20 ± 2 g were obtained from Beijing SPF Biotechnology Co., Ltd. (Experimental Animal Production License No.: SCXK (Jing) 2019-0010), an accredited animal supplier. The experiment procedure was reviewed and approved by the Institution of Animal Care and Use Committee (IACUC); in addition, written consent for experimental use had been provided by supplier for animal. The animals were maintained under controlled conditions (20–22°C, 12:12 h light: dark) in their home cage. Food and water were available ad libitum in the form of regular rodent chow and normal tap water. All animal procedures were in accordance with the standards of the Animal Ethics Committee of Beijing Friendship Hospital, Capital Medical University. The C57BL/6 mice were randomly assigned to seven groups, with 12 animals per group: Control (Con) group, Model (Mod) group, Vancomycin (Van) group, Shengjiang Xiexin Decoction (SXD) group, Combined Drug Regimen 1 (CDR1) group, Combined Drug Regimen 2 (CDR2) group, and Combined Drug Regimen 3 (CDR3) group.

The entire experiment spanned 34 days, with the experimental procedure and dosing regimens detailed in Table 1 and Figure 1. The Control and Model groups received an equivalent volume of physiological saline. The experimental animal treatment process was as follows: Except for the Con group, which maintained a normal diet and water, all other groups of mice were initially given a mixed antibiotic solution ad libitum (Vancomycin: 0.0225 mg/mL; Metronidazole: 0.1075 mg/mL; Colistin: 425 U/mL; Gentamicin: 0.0175 mg/mL) for 3 consecutive days. This was followed by a 3-day period of normal drinking water. On the second day of normal water consumption, all groups except the Con group received a single intraperitoneal (i.p.) injection of 10 mg/kg clindamycin for pretreatment. After the modeling period, an infection model was established in all groups (except Con) by oral gavage with 1×10⁶ CFU/mL Clostridioides difficile spore suspension. A 10-day interval of drug treatment followed. After drug administration, an observation period of 10 days was determined. After sample collection at the recovery stage, 4 mice in each group (except Con) were directed to a 7-day re-challenge with 1 ×10⁵ CFU/mL C. difficile spore suspension to mimic clinically relapse. During the course of the experiment, mice were observed daily for mental state, activity and fecal texture, whereas body weight was monitored at regular intervals. Sample collections. Specific time points were organized: During the treatment period (days 2, 5, 8 and 10) and recovery phase (days 11, 14, 17 and 20), fecal samples from mice were collected. At the end of the treatment period, four mice were euthanized for tissue collection. Most remaining mice were euthanized at the end of the recovery period, with four mice per group reserved for the re-challenge observation phase. During the re-challenge period, fecal samples continued to be collected (days 1, 3, 5, and 7). At the experiment’s conclusion, all remaining mice were euthanized, and final samples were collected. For euthanasia and tissue collection, mice were anesthetized via i.p. injection of pentobarbital sodium (50 mg/kg). Experimental procedures, including the collection of colon tissue and cecal contents for subsequent analysis, commenced after the loss of the righting reflex. Euthanasia was confirmed by an additional i.p. injection of pentobarbital sodium (200 mg/kg) after the procedures (Sugita et al., 2012).

During the whole period of experimentation, mice were closely monitored at 9:00 AM every day for general health conditions by measuring their mental state, activity and feces. Clinical symptoms of CDAD were assessed based on the criteria outlined in Table 2. Two factors, the state of mice and consistency in feces, were consistently observed and then scored according to the guidelines (Fachi et al., 2020). More specifically, fecal consistency was evaluated and categorized using a three-point visual grading scale (as depicted in Figure 2E): a score of 1 was assigned to formed, brown feces that retained their shape; semi-formed or soft, yellow feces that did not flow easily received a score of 2; and liquid, yellow feces that flowed easily were assigned a score of 3.

Total RNA was extracted from intestinal tissues using the RNAprep Pure Tissue Total RNA Isolation Kit (TIANGEN, DP431). Subsequently, RNA was reverse-transcribed into cDNA using a reverse transcription kit (Beijing LABLEAD Biotechnology Co., Ltd., 0202122021). The reverse transcription protocol involved an initial incubation at 37°C for 2 minutes, followed by 55°C for 15 minutes, 85°C for 5 minutes, and then storage at 4°C. Quantitative PCR reactions were performed in a 20 μL system containing 2 μL of cDNA template, 10 μL of SYBR Green Master Mix, 1 μL of forward primer, 1 μL of reverse primer, and 6 μL of Nuclease-Free Water. The amplification program was carried out in two steps: initial denaturation at 95°C for 5 min, then the cycles went as follows: 40 cycles of denaturation at 95°C for 10 seconds and annealing/extension at 60°C for 30 seconds. The expression of IL-1β, CXCL-1, Occludin, Claudin-1, Muc-2, TNF-α and ZO-1 in colon tissues was measured by RT-qPCR; the internal reference gene was GAPDH. All primer and probe sequences utilized are detailed in Table 3.

Mice were individually moved to clean cages and fresh fecal pellets collected with sterile forceps at the appropriate time points, directly into 2 mL cryovials (approximately 200 mg or 6–8 pellets). These samples were immediately frozen in a bath of liquid nitrogen. Total DNA from mouse fecal samples was subsequently extracted strictly according to the manufacturer’s instructions for the Stool DNA Kit. A standard curve was generated through serial dilutions, and Clostridioides difficile load was quantified by relating Ct values to Log10 values of bacterial load. Real-time quantitative PCR (qPCR) was performed using specific primers and a TaqMan probe. The reaction conditions involved an initial denaturation at 95°C for 30 seconds, followed by 40 cycles of denaturation at 95°C for 3 seconds and annealing/extension at 50°C for 30 seconds. Absolute quantification was employed for the RT-qPCR analysis of mouse fecal samples.

Fresh distal colon tissues were meticulously prepared using the “Swiss roll” technique and then fixed in 4% paraformaldehyde in a fume hood before paraffin embedding. After sectioning, the tissues were stained with Hematoxylin and Eosin (H&E). Pathological changes in the colon were observed under a microscope, and representative images were captured for subsequent analysis. Imaging assistance was provided by Wuhan Seville Biotechnology Co., Ltd.

Full-length 16S rRNA gene sequencing was performed to obtain comprehensive information on microbial taxonomy and diversity by targeting the V1-V9 regions through multiplex PCR amplification. Cecal content samples were collected, and DNA was extracted using a fecal DNA extraction kit. Library construction involved PCR amplification of the V3-V4 hypervariable regions. Following library preparation, single-molecule sequencing of the full-length 16S amplicons was conducted on the PacBio sequencing platform. After detection, multiplex PCR amplification and purification were performed, and samples were loaded for sequencing. Post-sequencing, raw paired-end data were assembled using overlap reads and subjected to quality control to yield high-quality Clean Data. To minimize noise, raw sequencing data underwent rigorous filtering, and the DADA2 (Divisive Amplicon Denoising Algorithm) was utilized for further denoising to extract more accurate sequence information and improve data precision, resulting in Effective Tags. Based on these effective tags, OTU (Operational Taxonomic Unit) clustering/denoising and species classification analyses were performed to construct species abundance profiles and to analyze OTU abundance, diversity indices, and community structure. Furthermore, clustering analysis, statistical comparisons, and correlation analysis of OTU and species composition were conducted to identify inter-sample species differences. Advanced bioinformatics analyses, including alpha (α) diversity analysis, beta (β) diversity analysis, species taxonomic identification, and functional prediction, were also performed on the generated sequence data.

All quantitative data are presented as the mean ± standard deviation (SD). Statistical analyses were performed using SPSS 26.0 software, and graphs were generated with GraphPad Prism 9.0. For comparisons between two normally distributed samples, Student’s t-test was used. One-way ANOVA was employed to determine statistical significance for more than two independent samples following a normal distribution. When data did not conform to a normal distribution or exhibited unequal variances, the non-parametric Kruskal-Wallis rank-sum test was applied. A P-value of less than 0.05 (P < 0.05) was considered statistically significant.

For evaluating the efficiency of treatment, we analyzed firstly clinical symptoms and bacterial burden. All of the combination treatments (CDR1, CDR2, and CDR3) were more effective in mitigating CDI than were those with vancomycin alone (Van monotherapy), quickly improving clinical scores and fecal consistency to a comparable level with Van treatment (Figures 3B, C). Conversely, both treatment groups experienced durable and prompt reductions in fecal C. difficile load, consistent with the killing kinetics of the standard Van regimen (Figure 3A). By comparison, monotherapy with SXD performed relatively weak on clinical outcomes and had markedly less bacterial eradication than Van and the combination groups. These results establish that the combination therapies are as effective as vancomycin in controlling the acute infection.

The primary advantage of the combination therapies was their profound effect on restoring gut health. The three regimens all remarkably increased the mRNA levels of tight junction proteins (ZO-1, Claudin-1, Occludin) and attenuated critical inflammatory cytokines (TNF-α, IL-1β) (Figures 3D–I). Our molecular results were supported by our histopathological analysis, which demonstrated that combination-treated groups displayed more profound attenuation of inflammation and obvious epithelial repairing (Figure 3J). It is interesting to note that extracts from the CDR1 were shown to be significantly effective and also showed significant recovery of Muc-2 protein. In contrast, monotherapies failed to induce tissue repair. The bactericidal effect against the Van group could not avoid a remarkable inflammatory infiltration and epithelial injury. The SXD group only generated a certain degree of protection of the mucosa, but to overcome the etiological results.

In summary, the combination regimens, particularly CDR1, deliver a superior therapeutic outcome by addressing both the pathogen and the host pathology. They uniquely pair the potent bactericidal activity of vancomycin with the robust gut barrier restoration and anti-inflammatory effects of SXD. This dual action makes them more effective than either monotherapy alone. Furthermore, the other two combination regimens (CDR2 and CDR3) demonstrated a certain level of therapeutic effect during the acute treatment phase. Although their pathogen clearance efficacy was slightly inferior to that of CDR1, achieving this outcome with a halved dose of vancomycin highlights their potential clinical value. It is important to note that a comprehensive evaluation of these different treatment regimens requires an integrated analysis incorporating data from subsequent phases and the dynamic changes in the gut microbiota. While the CDR1 regimen exhibited the most comprehensive restorative effects on the gut, the robust performance of CDR2 in the acute phase is equally noteworthy. Specifically, its ability to maintain considerable therapeutic efficacy with a shorter course of vancomycin defines its advantage as a “potential alternative for acute-phase treatment in selected patient populations”—rather than a universally “safer option” for all patients—offering a targeted clinical choice for specific scenarios.

To investigate therapeutic mechanisms, we performed 16S rRNA sequencing. CDI induction significantly reduced alpha diversity (Shannon, Simpson, Chao1), which all combination therapies partially restored. Notably, among all treatment groups, CDR1 demonstrated the most substantial restoration, with its Chao1, Shannon, and Simpson indices showing significantly greater improvement than CDR2, CDR3, and monotherapy groups (Figures 4A–C). Beta diversity analysis (PCoA, PCA) confirmed that the microbial structure of the CDR1 group shifted most closely to that of the Control (Con) group, significantly outperforming other regimens in restoring the overall gut community structure (Figures 4D–F).

At the taxonomic level, Lactobacillus was the dominant genus in control and most treated groups. Systematic comparison revealed that CDR1 achieved the most effective restoration of key genera, including Blautia and Lactobacillus, with relative abundances most closely resembling the Con group. At the species level, CDR1 exhibited comprehensive advantages in restoring beneficial species: it not only recovered Akkermansia muciniphila and Bacteroides stercorirosoris to near-normal levels, but also showed significantly superior restoration of Bacteroides thetaiotaomicron—a species crucial for intestinal barrier repair—compared to all other treatment groups (Figures 5A–F).

Analysis of the restorative effects during the treatment phase demonstrated that CDR1 most effectively shifted the microbial structure and key species—notably Bacteroides thetaiotaomicron (a species known for its role in intestinal barrier repair)—toward normal (Con) levels (Supplementary Figures S1, 2) (Wang et al., 2023). This superior microbial restoration by CDR1 was further corroborated by significantly enhanced colonization resistance against recurrent challenge, outperforming other treatment regimens.

In summary, the conventional combined regimen (CDR1) demonstrated comprehensive advantages over other therapeutic approaches in ameliorating gut dysbiosis, most effectively restoring microbial diversity, re-establishing key beneficial taxa, enhancing protective functional pathways, and ultimately providing superior colonization resistance.

In the recovery period, we considered treatment results on a long-term basis. The vancomycin (Van) monotherapy group had a severe bacterial rebound, and day 17 was higher than that of the model group. In contrast, among the combination regimens, all three (CDR1, CDR2 and CDR3) suppressed this rebound so that significantly lower levels of bacteria could be retained at a more constant level, indicating less pronounced recurrence (Figure 6A). CDR1 The CDR1 group was the best in all comparisons, maintaining the lowest C. difficile burden after treatment. These microbiologic results were in agreement with clinical observations. Mice in the combination regimens had both more pronounced and sustained improvement of their clinical scores and stool consistency compared to the Mod group (P<0.05 - P<0.001), with CDR1 and particularly CDR3 most effectively restoring normal stool form (Figures 6B, C).

We then evaluated the long-term effects on gut health. Histopathological analysis showed ongoing healing with better-organized intestinal cells and an intact barrier in the CDR1 group compared to all other groups when evaluating the percentage of shallow lesions (Figure 6H). By contrast, the CDR2 and CDR3 groups, as well as both Van monotherapy and SXD monotherapy groups, still displayed apparent inflammatory infiltration and epithelial damage.

Molecular analysis shed more light on these findings. Both the CDR1 and CDR2 regimens sustained low post-treatment levels of the inflammatory cytokines TNF-α and IL-1β (P<0.001 vs. Mod group). While the CDR2 group showed a strong recovery in the expression of tight junction proteins, the superior histological outcome of the CDR1 group suggests it conferred the most comprehensive and durable protection (Figures 6D–G). The monotherapies showed split effects: the Van group was more effective at reducing inflammatory markers, whereas the SXD group was superior in maintaining intestinal barrier components.

In summary, during the critical recovery period, the combination therapies—and particularly CDR1—provided a more holistic and effective outcome than monotherapies by simultaneously controlling bacterial recurrence, suppressing inflammation, and promoting durable structural healing of the intestinal mucosa.

The diversity of mouse gut microbiota in the recovery period was assessed. To evaluate alpha diversity, Important indexes were investigated, contained Simpson, Shannon and Pielou. Compared with the Con group, the Simpson and Shannon indices results after recovery reflected that there was a remarkable difference between the Mod and Con groups (P < 0.05, P < 0.01, or P < 0.001), as well as the evaluation of the Pielou index after restoration. This indicates that the gut microbiome in untreated CDI mice did not recover spontaneously. All combination groups, however, tended towards normalization with indices close to those of the Con group. Notably, among all treatment regimens, CDR1 demonstrated the most significant restoration in alpha diversity indices, outperforming both CDR2 and CDR3 as well as monotherapy groups (Figures 7A–C).

To assess differences in microbial community composition, beta diversity analysis was performed. Principal coordinate analysis (PCoA) and principal component analysis (PCA) revealed significant differences between the Con and Mod groups, indicating substantial alterations in the gut microbiota following CDI (Figures 7D, E). Boxplots based on Bray-Curtis dissimilarity further illustrated the community structure differences between the treatment and control groups. After treatment and recovery, while both CDR1 and CDR3 groups showed proximity to the Con group, CDR1 exhibited the closest alignment with the normal microbial structure, demonstrating superior efficacy over all other treatment groups (Figure 7F).

The composition of gut microbiota at the genus and species levels was also analyzed. At the genus level, Lactobacillus was the dominant genus in the Con, CDR1, CDR2, and CDR3 groups (Figure 8A). Systematic comparison revealed that CDR1 achieved the most effective restoration of key genera, including Helicobacter and Lachnospiraceae, with their relative abundances most closely matching those in the Con group (Figures 8C, D). At the species level, Lactobacillus murinus was the dominant species in all groups (Figure 8B). Comparative analysis demonstrated that CDR1 showed superior recovery of beneficial species, including Lactobacillus reuteri and L. murinus, with abundances significantly closer to normal levels compared to other treatment groups (Figures 8E, F). These results indicate that the CDR1 and CDR3 could effectively ameliorate gut microbiota dysbiosis after recovery.

Phenomenally, the restoration of host-microbiota homeostasis upon CDR1 treatment was evidenced by prompt microbial reconstitution after therapy, including probiotic and pathogenic normalization, beneficial consortia stabilization (e.g., Bacteroides thetaiotaomicron and L. murinus), together with a convergent tendency of 15 differential species revealed in LEfSe analysis toward the Con group (Supplementary Figure S4). Functional investigation also revealed that increased metabolic activity and DNA repair explained this stability of the microbiota (Supplementary Figure S3).

Taken together, these results demonstrate that the CDR1 treatment provides comprehensive advantages over other therapeutic approaches, not only most effectively improving gut microbiota dysbiosis during active treatment but also maintaining ecological balance after treatment cessation. This provides substantial evidence for the long-term modulatory effects of CDR1 on the gut microbiome.

Because the restoration of colonization resistance is critical for preventing CDI recurrence (Lawley and Walker, 2013; Perez-Cobas et al., 2015), we next performed a C. difficile spore re-challenge to assess the long-term protective efficacy of each treatment. Upon re-challenge, the CDR1-treated mice demonstrated the most robust colonization resistance. Fecal C. difficile load progressively decreased in this cohort and was accompanied by a lack of further superinfection (Figures 2A–C). The other combination regimens, CDR2 and CDR3-treated mice, also did not develop severe relapse but showed weaker protection with transient or delayed control of bacterial loads. Conversely, the two monotherapies each yielded a recurrence. The Van group experienced a transient clinical deterioration and bacterial resurgence. The SXD group also experienced a bacterial rebound over time, with the final load exceeding its initial level (Figure 2A).

When re-challenged, an analysis of the colonic tissue further showed chronic effects on mucosal integrity associated with each regimen. The suppression in mucosal destruction paralleled the degree of this process that was present in Van monotherapy-induced disease, which exhibited extreme mucosal disruption with extensive shedding of epithelial cells. Groups CDR2 and CDR3 showed partial inflammatory infiltration or loss of epithelial cells. Notably, the CDR1 and SXD groups displayed the most effective mucosal protection, with the least tissue damage and inflammation observed among all groups (Figure 2D).

Together, these data show that the CDR1 regimen is the most effective way to prevent CDI recurrence long term by completely restoring colonization resistance and maintaining intact intestinal mucosal functions.

Assessment and comparison of microbiota composition after treatment indicated parallel but distinct outcomes between three regimens (CDR1, CDR2, CDR3). LEfSe analysis indicated that the low-dose schedule (CDR3) significantly enriched Parabacteroides goldsteinii as compared with CDR1, whereas a greater abundance of Firmicutes and Lactobacillales was present in CDR1 than in CDR2(Supplementary Figures S5A, B, S6A, B). The two regimens also differed in their predicted metabolic landscape, with CDR1 being more enriched for pathways associated with Firmicutes colonization, and CDR2 being more abundant in oxidative phosphorylation (Supplementary Figure S5C). CDR3 also had a higher abundance of the flagellar assembly protein, FlgJ, compared with CDR1, which may indicate an increase in colonization-associated taxa (Supplementary Figure S6C).

And, even in the recovery phase, the composition of microbiota was significantly different between regimens. CDR1 was characterized by higher Bacteroides thetaiotaomicron relative abundance compared to CDR2 and CDR3. CDR2 (short-course vancomycin) was enriched in Lactobacillus johnsonii and Lactobacillus reuteri compared with CDR1, while CDR3 (low-dose vancomycin) showed marked expansion of potentially pathogenic taxa such as Escherichia coli and Escherichia–Shigella (Supplementary Figures S7A, B, S8A, B). LEfSe analysis indicated that CDR2 and CDR1 were more similar in microbial structure, with fewer discriminatory taxa between them, whereas CDR3 differed substantially from both. The predicted metabolic profiles during the recovery phase also differed between the two treatment regimens. PICRUSt2-derived KEGG pathway prediction combined with STAMP analysis further demonstrated that the lipid metabolism pathways were at higher enrichment in CDR2 and CDR3 groups, while secondary metabolite biosynthesis was more abundant for predicted function comparisons between CDR1 and the other two samples. These findings suggest that different medication strategies may influence therapeutic outcomes by modulating nutrient metabolism patterns (Supplementary Figure S7C, S8C).

Overall, these findings indicate that the antibiotic dosing strategy significantly influences gut microbiota remodeling. CDR1 most completely returned to a balanced community, while CDR2 retained major probiotic groups but with partial similarity, and CDR3, although beneficial in some aspects, was also enriched with potential pathogens, indicating the need for additional improvement.