Graphical abstract graphic file with name fx1.jpg [47]Open in a new tab Highlights * • Patients with T1D show gut microbiota and bile acid dysregulation compared to HCs * • Secondary bile acids are associated with islet function in patients with T1D * • Gut microbial markers and bile acid profiles demonstrate T1D diagnostic potential * • WMT improves the islet function in T1D with the elevated LCA and isoLCA levels __________________________________________________________________ Liu et al. identify distinct gut microbial signatures and dysregulated bile acid metabolism in T1D. The enhanced synthesis of secondary bile acids, notably LCA and isoLCA, is associated with improved islet function following WMT, underscoring the importance of bile acid metabolism in both the pathogenesis and therapeutic approaches for T1D. Introduction Type 1 diabetes (T1D) is a chronic autoimmune condition characterized by the destruction of insulin-producing pancreatic β cells, triggered by an immune-associated mechanism.[48]^1 The classic triad of symptoms associated with disease onset, including polydipsia, polyphagia, and polyuria, along with overt hyperglycemia, remain the diagnostic hallmarks of T1D. An immediate requirement for exogenous insulin replacement therapy is also a hallmark of the condition, necessitating lifelong treatment.[49]^2 Epidemiological data from the International Diabetes Federation have revealed a concerning global trend of increasing T1D incidence and prevalence in recent years.[50]^3 While insulin replacement therapy remains the mainstay of T1D management, it does not address the underlying autoimmune dysregulation that causes the disease. Hence, substantial research efforts have been devoted to developing novel anti-diabetic agents and emerging therapies, such as immunomodulatory approaches, cell-based interventions, and pancreatic β cell transplantation, with the aim of providing new avenues for treating T1D and potentially curing it.[51]^4 The rapid global rise in the incidence of T1D cannot be fully explained by genetic factors alone, strongly implicating the involvement of environmental influences.[52]^5 One environmental factor that has garnered significant research interest is dysbiosis of the gut microbiota. As the largest immune organ interfacing with the external environment, the gut microbiome is intricately regulated by various host factors, including hormones, inflammatory mediators, the nervous system, and dietary components.[53]^6 Disruptions to this delicate gut microbial balance, such as those induced by excessive antibiotic exposure or infections, may contribute to the development of autoimmune and inflammatory disorders.[54]^7^,[55]^8 Compared to healthy controls (HCs), individuals with T1D have consistently demonstrated reduced gut microbial diversity, decreased production of short-chain fatty acids (SCFAs), and increased intestinal permeability.[56]^5^,[57]^9 Notably, similar gut dysbiosis has also been observed in high-risk individuals who are positive for islet autoantibodies but have not yet progressed to overt clinical T1D,[58]^10 underscoring the potential role of the gut microbiome in T1D pathogenesis. Furthermore, animal studies have validated the critical importance of establishing a stable gut microbiome early in life for the proper development and education of the immune system.[59]^11 Metabolites produced by the gut microbiota, such as secondary bile acids, SCFAs, B group vitamins, and indoles, actively participate in regulating diverse physiological processes of the host.[60]^12^,[61]^13^,[62]^14 Disruption of the dynamic equilibrium between the gut microbiota and their metabolic activities may contribute to the development of metabolic, immune, and inflammatory disorders.[63]^15 Accumulating evidence has implicated the potential involvement of gut microbial metabolites, particularly secondary bile acids, in the pathogenesis and progression of T1D. Studies using animal models have revealed altered bile acid profiles in mice with early-stage of T1D, and certain secondary bile acids, such as ursodeoxycholic acid, have been shown to reduce the risk of developing T1D by improving β cell function and regulating glucose metabolism.[64]^16 Recent human studies have also reported dysregulated metabolism of secondary bile acids even before the onset of islet autoimmunity in high-risk individuals.[65]^17 Therefore, targeting the gut microbiome may represent a promising therapeutic approach for T1D, such as through dietary fiber supplementation,[66]^18 administration of probiotics,[67]^19 or fecal microbiota transplantation (FMT) to restore gut microecological homeostasis.[68]^20 As a direct method for gut microbiome restoration, FMT has demonstrated clinical efficacy in treating gastrointestinal, infectious, and metabolic disorders.[69]^21 A randomized controlled trial on patients with new-onset T1D found that autologous FMT could ameliorate autoimmunity.[70]^22 Furthermore, Zhang’s team has developed an automated washed microbiota transplantation (WMT) system that strictly quantifies and quality-controls the washing process. This advancement significantly enhances the safety profile of microbiota transplantation.[71]^23 Our previous study demonstrated that WMT can reduce the frequency of hypoglycemic episodes and glucose fluctuations in patients with unstable diabetes.[72]^24 Based on this background, the present study aims to first establish a well-characterized cohort of individuals with T1D to explore gut microbiome biomarkers associated with the disease. Subsequently, WMT treatment will be performed on eligible T1D participants to potentially improve their gut microecological environment. Factors that may influence the therapeutic efficacy will also be investigated, providing a foundation for further optimization of gut microbiome-targeted therapies and precision treatment approaches for T1D. Results Study participants Our study enrolled a total of 111 participants: 44 patients with T1D and 34 HCs in the discovery cohort and 33 participants (21 T1D and 12 HCs) in the internal validation cohort. In the discovery cohort, the T1D group had a mean disease duration of 5.5 ± 6.0 years, fasting plasma glucose (FPG) of 8.11 ± 4.47 mmol/L, and glycated hemoglobin (HbA1c) of 10.13% ± 2.59%. The internal validation T1D group had a mean disease duration of 7.0 ± 9.0 years, FPG of 8.55 ± 3.91 mmol/L, and HbA1c of 8.76% ± 3.14%. No significant differences were found between the T1D and HC groups in either cohort regarding gender, age, body mass index (BMI), or blood lipid profiles (p > 0.05). Baseline characteristics are summarized in [73]Table 1. Table 1. Baseline characteristics of participants in the internal cohort Characteristics Discovery set __________________________________________________________________ Validation set __________________________________________________________________ HC (n = 34) T1D (n = 44) p value HC (n = 12) T1D (n = 21) p value Age (at diagnosis, years) 31 ± 14 32 ± 15 0.81 29 ± 8 33 ± 11 0.19 Gender (male/female) 11/23 19/25 0.33 8/4 11/10 0.42 BMI (kg/m^2) 20.91 ± 2.76 20.75 ± 3.63 0.82 20.97 ± 2.03 20.38 ± 2.79 0.50 HbA1c (%) 5.27 ± 0.34 10.13 ± 2.59 <0.01 5.25 ± 0.36 8.76 ± 3.14 <0.01 FPG (mmol/L) 4.53 ± 0.39 8.11 ± 4.47 <0.01 4.65 ± 0.22 8.55 ± 3.91 <0.01 Fasting C-peptide (ng/mL) 1.51 ± 0.22 0.20 ± 0.35 <0.01 1.38 ± 0.19 0.31 ± 0.40 <0.01 C-peptide AUC (ng/mL · min) – 1.19 ± 2.05 – – 1.18 ± 1.18 – TC (mmol/L) 4.47 ± 0.77 4.61 ± 1.25 0.57 4.59 ± 0.97 4.68 ± 0.99 0.80 TG (mmol/L) 1.14 ± 0.48 1.12 ± 0.74 0.90 1.21 ± 0.45 1.05 ± 0.37 0.32 HDL-C (mmol/L) 1.33 ± 0.23 1.42 ± 0.42 0.28 1.40 ± 0.24 1.40 ± 0.42 0.98 LDL-C (mmol/L) 2.69 ± 0.67 2.80 ± 1.02 0.56 2.73 ± 084 2.93 ± 0.94 0.54 Duration of diabetes (years) – 5.5 ± 6.0 – – 7.0 ± 9.0 – [74]Open in a new tab Plus-minus values are means ± SD. Wilcoxon rank-sum test utilized for continuous variables; chi-squared test utilized for analyzing gender differences. Abbreviations: T1D, type 1 diabetes; HC, healthy control; BMI, body mass index; HbA1c, glycated hemoglobin; FPG, fasting plasma glucose; C-peptide AUC, area under the curve of C-peptide; TC, total cholesterol; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol. In the external validation cohort from Pingjiang, China, the T1D group had a mean disease duration of 2.5 (0.7–4.3) years, FPG of 8.07 ± 3.52 mmol/L, and HbA1c of 8.05% ± 1.98%. Consistent with the previous cohorts, no significant differences were observed between T1D and HC groups in terms of gender, age, BMI, or blood lipid profiles (p > 0.05) ([75]Table S1). Gut microbiome dysbiosis in patients with T1D Metagenomic sequencing of fecal samples from the 78 participants in the discovery cohort revealed lower bacterial alpha diversity in the T1D group compared to HCs. The Chao1 richness estimator and Shannon diversity index, representing community richness and evenness, respectively, were significantly decreased in T1D ([76]Figure 1A, p < 0.05). Furthermore, principal coordinate analysis (PCoA) based on Bray-Curtis distance showed distinct clustering of gut microbiome compositions between the two groups ([77]Figure 1B, p = 0.0006, permutational multivariate analysis of variance [PERMANOVA] test with 10,000 permutations). Figure 1. [78]Figure 1 [79]Open in a new tab Dysbiosis of the gut microbiome in patients with T1D (A) Alpha diversity analysis showing reduced richness (Chao1 index) and evenness (Shannon index) of gut microbiota in individuals with T1D compared to HCs. Data are presented as boxplots with overlaid individual points, showing the median with IQR. ∗p < 0.05, ∗∗p < 0.01 (t test). (B) PCoA based on Bray-Curtis distances, showing distinct clustering of gut microbiome compositions between T1D and HC groups. (C) Differential abundance analysis of microbial species was performed using the DESeq2 method. The volcano plot displays log[2] fold changes versus −log[10] adjusted p values (padj). Microbes with |log[2] fold change| > 1 and padj < 0.05 were considered significantly differentially abundant. (D) PCoA based on Bray-Curtis distances, showing distinct clustering of KEGG ortholog abundances between T1D and HC groups. (E) KEGG pathway enrichment analysis of gut microbial functions. Color indicates upregulation (red) or downregulation (blue) in T1D. (F) Heatmap showing Spearman’s correlations between T1D-associated bacterial species and clinical indicators. ∗p < 0.05, ∗∗p < 0.01. Abbreviations: T1D, type 1 diabetes; HC, healthy control; IQR, interquartile range; PCoA, principal coordinate analysis; HbA1c, glycated hemoglobin; FPG, fasting plasma glucose; AUC, area under the curve; TIR, time in range; TC, total cholesterol; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol. Differential abundance analysis identified several significantly altered bacterial taxa in T1D ([80]Figure 1C; [81]Table S2). Notably, Veillonella parvula (CAG0286) was highly enriched in the T1D group (log2FC = 3.84, padj < 1e−9). Other enriched bacteria included Escherichia coli (CAG0017, CAG0398, and CAG0410), Streptococcus salivarius (CAG0453 and CAG0294), Bacteroides salyersiae (CAG0125), Clostridium bolteae (CAG0043), and Klebsiella pneumoniae (CAG0034). In contrast, metagenomics species including Faecalibacterium prausnitzii (CAG0310 and CAG0093), Clostridium leptum (CAG0030), Clostridium scindens (CAG0098), Roseburia faecis (CAG0172), and Dorea longicatena (CAG0302) were significantly depleted. Functional analysis by annotating metagenomic sequences to KEGG Orthology (KO) revealed significant differences in functional profiles between the two groups ([82]Figure 1D, beta diversity). 2,106 KO entries were upregulated while 232 were downregulated in T1D ([83]Figure S1). Enriched pathways in T1D included bacterial secretion systems, biofilm formation, phenylalanine metabolism, antimicrobial peptide resistance, and flagellar assembly. In contrast, secondary bile acid biosynthesis pathways were decreased ([84]Figure 1E). Finally, correlation analysis revealed significant associations between identified microbial biomarkers and clinical parameters in T1D participants ([85]Figure 1F). Roseburia faecis (CAG0172) positively correlated with fasting C-peptide and C-peptide area under the curve (AUC), while Klebsiella pneumoniae (CAG0034) showed negative correlations. Firmicutes sp. (CAG0063), Oscillibacter sp. ER4 (CAG0105), and Dorea longicatena (CAG0302) positively correlated with C-peptide AUC. Moreover, Bacteroides salyersiae (CAG0125) and Firmicutes sp. (CAG0350) correlated positively with time in range (TIR), whereas Clostridium bolteae (CAG0043) and Faecalibacterium prausnitzii (CAG0093) correlated negatively. These findings suggest potential mechanistic links between dysregulated gut microbiomes and metabolic dysfunction in T1D. Dysregulation of bile acid metabolism is associated with impaired islet function in T1D Further analysis revealed significantly lower relative abundances of 5 bile-acid-induced (bai) operon genes (baiB, baiCD, baiE, baiF, and baiI) and 2 hydroxysteroid dehydrogenase (HSDH) genes (3α-HSDH and 3β-HSDH) involved in secondary bile acid synthesis in the T1D group ([86]Figure 2A). Figure 2. [87]Figure 2 [88]Open in a new tab Dysregulation of fecal bile acid metabolism associated with impaired β cell function in T1D (A) Boxplots showing the square root-transformed RPKM of bai and HSDH in fecal metagenomes of T1D and HC groups. Data are presented as median with IQR. Statistical comparisons were performed using the Wilcoxon rank-sum test. ∗p < 0.05, ∗∗p < 0.01. (B) Stacked bar plot displaying the average proportion of secondary to primary bile acids in each group. (C) Boxplots showing the relative abundances of fecal bile acids in T1D and HC groups. Data are presented as median with IQR. Wilcoxon rank-sum test was used for comparisons. ∗p < 0.05, ∗∗p < 0.01. (D) Heatmap showing Spearman’s correlation coefficients between fecal bile acids and clinical indicators. ∗p < 0.05, ∗∗p < 0.01. (E) Boxplots showing the relative abundances of selected fecal bile acids between T1D[POS] and T1D[NEG] subgroups. Data are represented as median with IQR. Statistical significance was determined by the Wilcoxon rank-sum test. ∗p < 0.05, ∗∗p < 0.01. Abbreviations: RPKM, reads per kilobase of transcript per million mapped reads; bai, bile-acid-induced operon gene; HSDH, hydroxysteroid dehydrogenases genes; IQR, interquartile range; CDCA, chenodeoxycholic acid; GCA, glycocholic acid; GCDCA, glycochenodeoxycholic acid; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; CA, cholic acid; 12-ketoLCA, 12-ketolithocholic acid; 3-DHCA, 3-oxocholic acid; 6-ketoLCA, 6-ketolithocholic acid; 7-DHCA, 7-dehydrocholic acid; 7-ketoLCA, 7-ketolithocholic acid; ACA, allocholic acid; bCA, 3β-cholic acid; bDCA, 3β-deoxycholic acid; bUDCA, isoursodeoxycholic acid; GDCA, glycodeoxycholic acid; GLCA, glycolithocholic acid; GLCA-3S, glycolithocholic acid 3-sulfate; GUDCA, glycoursodeoxycholic acid; HCA, hyocholic acid; HDCA, hyodeoxycholic acid; DCA, deoxycholic acid; LCA, lithocholic acid; isoLCA, isolithocholic acid; LCA-3S, lithocholic acid 3-sulfate; NorCA, norcholic acid; NorDCA, nordeoxycholic acid; TDCA, taurodeoxycholic acid; TLCA, taurolithocholic acid; UDCA, ursodeoxycholic acid. Targeted metabolomics on fecal samples confirmed reduced proportions of secondary bile acids in patients with T1D compared to HCs ([89]Figure 2B). Further analysis revealed significantly increased relative abundance of the primary bile acid cholic acid (CA) but decreased abundances of secondary bile acids like deoxycholic acid (DCA), lithocholic acid (LCA), and isolithocholic acid (isoLCA) in the T1D group ([90]Figure 2C). Consistent with these findings, absolute abundance analysis also showed elevated CA levels and reduced LCA and isoLCA levels in the T1D group ([91]Figure S2). Additionally, targeted metabolomic analysis of serum bile acids showed significantly lower absolute abundance of DCA and LCA in the T1D group ([92]Figure S3). Moreover, we observed strong positive correlations between the abundances of DCA, LCA, and isoLCA in serum and fecal samples ([93]Figure S4). We then conducted Spearman correlation analysis between fecal bile acid profiles and clinical parameters in the participants. In the T1D group, the decreased levels of the absolute abundances of certain secondary bile acids (e.g., glycolithocholic acid, 3β-deoxycholic acid, and nordeoxycholic acid) as well as DCA, LCA, and isoLCA showed positive correlations with fasting C-peptide levels and C-peptide AUC. Conversely, the absolute abundances of the primary bile acid chenodeoxycholic acid and some secondary bile acids (glycoursodeoxycholic acid, ursodeoxycholic acid, 7-dehydrocholic acid, and 3-dehydrocholic acid) were negatively correlated with C-peptide AUC ([94]Figure 2D). To further explore the relationship between secondary bile acid biosynthesis and residual pancreatic β cell function in T1D, we stratified the patients into two subgroups based on their residual islet function status. The T1D[POS] and T1D[NEG] groups were comparable in terms of age, gender, BMI, HbA1c levels, and disease duration, with no significant differences between the two groups ([95]Table S3). In comparison to the T1D[POS] group, the relative abundances of secondary bile acids, such as DCA, LCA, and isoLCA, were significantly decreased in the T1D[NEG] group ([96]Figure 2E). Additionally, the T1D[NEG] group exhibited alterations in the composition of gut microbial species relative to the T1D[POS] group ([97]Figure S5; [98]Table S4). These results suggest that dysregulation of bile acid metabolism, particularly reduced biosynthesis of secondary bile acids, is associated with impaired residual islet function in individuals with T1D. Gut microbial signatures and bile acid profiles exhibit promising diagnostic potential in T1D To explore the diagnostic potential of the identified gut microbiome and bile acid signatures, we constructed comprehensive predictive models that incorporated the key microbial taxa and metabolites that were associated with T1D status and correlated with β cell function and glycemic control. The microbial taxa incorporated into the diagnostic models were Faecalibacterium prausnitzii (CAG0093), Roseburia faecis (CAG0172), Clostridium leptum (CAG0030), Clostridium scindens (CAG0098), and Escherichia coli (CAG0410). As shown in [99]Figure 3A, these microbial signatures achieved impressive AUCs of 0.962 (95% confidence interval [CI]: 0.926–0.998), 0.889 (95% CI: 0.751–0.998), and 0.749 (95% CI: 0.613–0.884) in the training, internal validation, and external validation datasets, respectively. Notably, the diagnostic models built using the significantly decreased secondary bile acids in T1D, including DCA, LCA, and isoLCA, also demonstrated robust performance, with AUCs of 0.977 (95% CI: 0.951–0.998), 0.853 (95% CI: 0.723–0.983), and 0.752 (95% CI: 0.602–0.903) in the same datasets ([100]Figure 3B). Intriguingly, integrating the key microbial taxa and bile acid metabolites further enhanced the diagnostic performance, with the combined model achieving AUCs of 0.988 (95% CI: 0.968–0.999), 0.931 (95% CI: 0.841–0.999), and 0.814 (95% CI: 0.672–0.955) ([101]Figure 3C). This synergistic approach leveraging both gut microbiome and metabolomic signatures showcased the robust diagnostic potential of these multi-omics biomarker panels. Figure 3. [102]Figure 3 [103]Open in a new tab Gut microbial signatures and bile acid profiles exhibit promising diagnostic potential for T1D (A–C) ROC curves depicting the diagnostic accuracy of microbial taxa and bile acid biomarkers in discriminating between T1D and healthy control groups. The curves were generated using a random forest model, and the AUC values for models based on microbial taxa (A), fecal secondary bile acids (B), and the multi-omics model (C) were computed across training, internal validation, and external validation datasets. (D–F) ROC curves depicting the diagnostic accuracy of microbial taxa and bile acid biomarkers in discriminating between T1D[POS] and T1D[NEG]groups. The curves were generated using a random forest model, and the AUC values for models based on microbial taxa (D), fecal secondary bile acids (E), and the multi-omics model (F) were computed across training, internal validation, and external validation datasets. Abbreviations: ROC, receiver operating characteristic; T1D[POS] group, individuals with peak C-peptide levels greater than 0.01 ng/mL during the steamed bread meal test (SBMT); T1D[NEG] group, indicative of exhausted islet function, included participants with peak C-peptide levels less than 0.01 ng/mL. We then evaluated the ability of these biomarkers to discriminate T1D subgroups with varying degrees of residual islet function ([104]Figures 3D–3F). A panel comprising 5 microbial taxa, including Bacteroides fragilis (CAG0039), Clostridium innocuum (CAG0136), Alistipes finegoldii (CAG0217), Alistipes putredinis (CAG0189), and Bifidobacterium longum (CAG0493) exhibited AUCs of 0.967 (95% CI: 0.919–0.998), 0.827 (95% CI: 0.647–0.998), and 0.747 (95% CI: 0.547–0.947) across cohorts, in distinguishing the T1D[POS] and T1D[NEG]subgroups. The diagnostic model based solely on the secondary bile acids DCA, LCA, and isoLCA also demonstrated good performance, with AUCs of 0.920 (95% CI: 0.836–0.998), 0.833 (95% CI: 0.504–0.998), and 0.798 (95% CI: 0.576–0.998) in the respective datasets. Notably, combining the microbial and bile acid signatures further improved the AUCs to 0.966 (95% CI: 0.919–0.999), 0.908 (95% CI: 0.763–0.999), and 0.839 (95% CI: 0.653–0.999). These comprehensive findings highlight the remarkable diagnostic potential of gut microbial signatures and bile acid metabolic profiles, both individually and synergistically, for differentiating individuals with T1D from HCs, as well as for stratifying T1D disease subtypes based on their residual islet function status. Clinical efficacy of WMT in T1D To evaluate the safety profile of WMT as a therapeutic approach for T1D, we closely monitored all participants receiving the WMT intervention. Notably, no adverse events, such as gastrointestinal symptoms (e.g., abdominal bloating, diarrhea, and constipation), or systemic reactions (e.g., fever), were reported during the treatment period and 3-month follow-up. Next, we conducted statistical analyses on the clinical parameters of the participants before and after the WMT intervention to evaluate its therapeutic efficacy. As illustrated in [105]Figure 4A, while FPG did not exhibit statistically significant improvements (p > 0.05), TIR was significantly increased at 1 week (63% ± 17%), 1 month (65% ± 22%), and 3 months (60% ± 20%) post-WMT, compared to the pre-treatment TIR of 39% ± 27% (p = 0.0053, 0.019, and 0.028, respectively). The coefficient of variation for blood glucose also showed significant reductions at the 1-month and 3-month follow-ups (p = 0.0069 and 0.031, respectively). Notably, the daily insulin requirement was significantly decreased after the WMT intervention (p = 0.004, 0.007, and 0.026 at 1 week, 1 month, and 3 months, respectively). At the 3-month time point, we further assessed the participants’ HbA1c levels and conducted a steamed bread meal test to evaluate C-peptide responses. As depicted in [106]Figure 4B, HbA1c significantly decreased from 9.4% ± 2.31% to 8.04% ± 1.62% (p < 0.028) after 3 months of WMT intervention, while fasting C-peptide and C-peptide AUC did not show statistically significant changes (p > 0.05). Figure 4. [107]Figure 4 [108]Open in a new tab Clinical efficacy and microbiota-associated characteristics of WMT in patients with T1D (A) Longitudinal monitoring of FPG, SMBG_TIR, SMBG_CV, and insulin requirements at T0, T1W, T1M, and T3M. Paired t tests were used for comparisons. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ns, p > 0.05; ∗p < 0.05; ∗∗p < 0.01. (B) Longitudinal monitoring of HbA1c, fasting C-peptide, and C-peptide AUC at T0 and T3M. Paired t tests were used for comparisons. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ns, p > 0.05; ∗p < 0.05. (C) Alpha diversity analysis revealing decreased richness (Chao1 index) of gut microbiota in patients with T1D compared to donors and at T1W, T1M, and T3M compared to T0. Data are presented as boxplots with overlaid individual points, showing the median with IQR. ∗p < 0.05, ∗∗p < 0.01. (D) PCoA of microbiome compositions based on Bray-Curtis distances. Circles and error bars represent means and SEM, respectively. Adjusted p values for the comparison of gut microbiota compositions in donors versus T1D and T1W, T1M, and T3M versus T0 are 0.001, 0.0375, 0.0040, and 0.2286, respectively (PERMANOVA with 10,000 permutations). ∗p < 0.05, ∗∗p < 0.01. (E and F) Longitudinal monitoring of relative abundances of bacterial species before WMT and after WMT, evaluated using paired Wilcoxon rank-sum tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ∗p < 0.05, ∗∗p < 0.01. (G) Longitudinal monitoring of relative abundances of bai before WMT and after WMT, evaluated using paired Wilcoxon rank-sum tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ∗p < 0.05, ∗∗p < 0.01. Abbreviations: WMT, washed microbiota transplantation; FPG, fasting plasma glucose; SMBG, self-monitoring of blood glucose; TIR, time in range; CV, coefficient of variation; IQR, interquartile range; AUC, area under the curve; SEM, standard error of the mean; T0, baseline before WMT; T1W, 1 week after WMT; T1M, 1 month after WMT; T3M, 3 months after WMT. Collectively, these findings suggest potential insights into the metabolic effects of WMT, indicating preliminary evidence of impacts on glycemic control status in T1D. However, further longitudinal studies are required to validate these observations and establish definitive therapeutic potential. Gut microbiome restoration following WMT in T1D To investigate the impact of WMT on gut microbiome restoration in patients with T1D, we characterized the changes in gut microbial profiles of the T1D participants before and after the WMT intervention. Metagenomic analysis of fecal samples revealed that microbial diversity was significantly increased in T1D participants following WMT treatment ([109]Figure 4C). Furthermore, PCoA demonstrated substantial shifts in the gut microbial community structure, with participant samples moving closer to the donor microbiome profile post-WMT ([110]Figure 4D). We then employed paired Wilcoxon signed-rank tests to identify specific microbial taxa and functional genes that were significantly altered post-WMT. At the species level, the relative abundances of Escherichia coli (CAG0410), Streptococcus salivarius (CAG0294), and Clostridium bolteae (CAG0043) were significantly decreased after WMT treatment ([111]Figure 4E). In contrast, the abundances of Faecalibacterium sp. (CAG0134), Roseburia faecis (CAG0172), and Dorea formicigenerans (CAG0328) were significantly increased post-WMT ([112]Figure 4F). Functional analysis of the gut metagenome revealed upregulated relative abundances of bile acid biosynthesis genes, including baiE, baiF, and baiI, in post-WMT samples ([113]Figure 4G). These microbiome alterations, including changes in specific taxa and bile acid biosynthesis pathways, may underlie the observed therapeutic effects on glycemic control and immunomodulation. WMT modulates inflammatory responses and immune tolerance in T1D To investigate the potential immunomodulatory effects of WMT in T1D, we conducted a comprehensive analysis of inflammatory markers and immune cell populations. We first measured serum inflammatory cytokines before and 3 months after WMT intervention. As shown in [114]Figure 5A, the levels of pro-inflammatory cytokines interleukin (IL)-17A and IL-1β were significantly downregulated in participants after WMT treatment (IL-17A: p = 0.011; IL-1β: p = 0.013), both of which are critical in T1D pathogenesis. Notably, no significant changes were observed in the other ten cytokines examined ([115]Figure S6). Figure 5. [116]Figure 5 [117]Open in a new tab Immunomodulatory effects of WMT in patients with T1D (A) Longitudinal monitoring of serum cytokine levels at T0 and T3M, evaluated using paired t tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ∗p < 0.05. (B) Representative flow cytometry plots of Treg cells (CD4^+CD25^+CD127^low/-), Th1 cells (CD4^+IFN-γ^+), and Th17 cells (CD4^+IL-17A^+) in PBMCs. (C) The frequency of CD4^+ T cell subsets (Treg, Th1, and Th17) was quantified at T0 and T3M. Paired t tests were used to assess changes over time. ∗p < 0.05. Abbreviations: IQR, interquartile range; Treg cells, regulatory T cells; Th1, T helper 1 cells; Th17, T helper 17 cells; PBMCs, peripheral blood mononuclear cells. Flow cytometry analysis of peripheral blood mononuclear cells (PBMCs) further revealed the immunomodulatory potential of WMT in T1D patients. While there was a trend toward increased regulatory T cells (Tregs) post-treatment, this did not reach statistical significance. The most striking finding was a significant decrease in Th17 cell populations and their signature cytokine IL-17A following WMT (p < 0.01). In contrast, Th1 cell subsets remained relatively stable, with minimal alterations observed ([118]Figures 5B and 5C). These immunological changes provide insight into potential mechanisms by which WMT might modulate inflammatory responses in T1D. Gut microbiome and bile acid profiles associated with improved islet function following WMT in T1D Upon analyzing the follow-up outcomes, we observed two distinct subgroups. Patients with no residual pancreatic β cell function at baseline (T1D[NEG], n = 4) showed no recovery of β cell function after WMT treatment. However, among those with residual islet function pre-treatment (T1D[POS], n = 11), WMT elicited varying degrees of β cell functional changes, as assessed by C-peptide AUC responses ([119]Figure 6A). Within the T1D[POS] group, we further classified participants as responders (increased C-peptide AUC post-WMT) or non-responders. The responder and non-responder subgroups were well matched for baseline characteristics ([120]Table S5). Figure 6. [121]Figure 6 [122]Open in a new tab Gut microbiome restoration following WMT in patients with T1D (A) Boxplots depicting changes in C-peptide AUC before and after WMT in the negative, responders, and non-responders, analyzed using paired Wilcoxon rank-sum tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ns, p > 0.05; ∗p < 0.05. (B and C) Boxplots depicting changes in relative abundances of bacterial species before and after WMT in the responder and non-responder groups, analyzed using paired Wilcoxon rank-sum tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ns, p > 0.05; ∗p < 0.05. (D) Boxplots depicting changes in relative abundances of fecal bile acids before and after WMT in the responder and non-responder groups, analyzed using paired Wilcoxon rank-sum tests. Data are presented as boxplots, showing the median with IQR. Gray lines indicate paired measurements from the same individual. ns, p > 0.05; ∗p < 0.05. Abbreviations: responders, individuals showing an increase in C-peptide AUC at T3M after WMT compared to baseline levels; non-responders, individuals with no increase in C-peptide AUC at T3M after WMT compared to baseline levels; IQR, interquartile range. Comparative analysis of gut microbiome profiles revealed significant increases in the abundances of Firmicutes sp. (CAG0108), Roseburia faecis (CAG0172), and Clostridium leptum (CAG0030), along with a concomitant decrease in Escherichia coli (CAG0410) abundance, in responders after WMT intervention. In contrast, non-responders did not exhibit such microbial shifts ([123]Figures 6B and 6C). Examination of fecal bile acid profiles showed increased relative abundances of the secondary bile acids isoLCA and LCA post-WMT in responders, while both subgroups lacked significant changes in DCA levels ([124]Figure 6D). The favorable modulations in specific gut commensals, such as enrichment of Firmicutes and Roseburia species, coupled with selective restoration of isoLCA and LCA metabolism, may underlie the improved residual pancreatic β cell function observed in WMT responders. Discussion In our study, we identified and validated gut dysbiosis in patients with T1D across two cohorts spanning different time periods. Notably, through functional pathway analysis and targeted metabolomics, we revealed a decrease in secondary bile acid biosynthesis, such as DCA, LCA, and isoLCA, in T1D subjects. Furthermore, we established and validated the positive correlation between secondary bile acids DCA, LCA, and isoLCA and pancreatic β cell function through association analyses, subgroup analyses, and classifier models. To directly test the therapeutic implications of these findings, we implemented WMT as a targeted microbial intervention. Our results demonstrate that WMT-mediated gut microbiota reconstruction significantly attenuated glycemic variability in T1D patients. Additionally, the improvement in pancreatic β cell function appears to be closely associated with the extent of gut microbiota reconstruction, as well as the increase in secondary bile acid levels. These results highlight the therapeutic potential of microbiota-based interventions in T1D management. Our evidence for the association between the gut microbiome and T1D is consistent with earlier studies,[125]^25^,[126]^26^,[127]^27 where the abundances of various gut bacteria producing SCFAs, such as Faecalibacterium prausnitzii, Roseburia faecis, Roseburia intestinalis, Roseburia inulinivorans, and Dorea longicatena, were significantly decreased in the gut of patients with T1D. Studies have suggested that gut immune dysregulation in T1D is related to the reduced relative abundance of SCFA-producing bacteria.[128]^28 In addition to SCFA-producing bacteria, we also observed a decrease in the abundance of Clostridium leptum (CAG0030) and Clostridium scindens (CAG0098). These bacteria play a critical role in the biotransformation of primary bile acids into secondary bile acids, such as DCA and LCA.[129]^29^,[130]^30 Furthermore, we found an increased abundance of potential pathogens, such as Escherichia coli, Veillonella parvula, Streptococcus anginosus, and Streptococcus salivarius, in the gut microbiota of patients with T1D. Bile acids, traditionally known for their roles in lipid digestion and antimicrobial defense, are now recognized as key signaling molecules that regulate metabolism and inflammation.[131]^31 Primary bile acids synthesized and secreted by the liver are converted into secondary bile acids, primarily DCA and LCA, by gut bacteria and key enzymes. Kostic et al. followed up on infants susceptible to T1D and found that gut microbiome and metabolite dysregulation preceded disease onset.[132]^32 In a prospective cohort study spanning three years, Swedish researchers monitored 74 children at elevated risk for developing T1D. Notably, they observed a decline in the abundance of key bacterial enzymes involved in the secondary bile acid metabolic pathway prior to the emergence of islet autoantibodies, which are early markers of autoimmune activity against pancreatic β cells. Furthermore, as islet autoimmunity progressed toward clinical diagnosis, the levels of secondary bile acids, particularly those detected in fecal samples, exhibited a decreasing trend.[133]^17 Our study demonstrated that patients with T1D exhibit significant downregulation of metabolic pathways, bai operons, and HSDH genes associated with secondary bile acid biosynthesis compared to HCs. Further fecal bile acid testing confirmed that the production of secondary bile acids in the gut was reduced in patients with T1D, and the abundances of secondary bile acids DCA, LCA, and isoLCA were significantly lower than those in HCs (p < 0.05). Our study once again demonstrated the potential role of secondary bile acids, particularly DCA, LCA, and isoLCA, in T1D pathogenesis and progression. Correlation analyses revealed that fecal bile acids were significantly positively correlated with pancreatic β cell function. Further subgroup analysis showed that the T1D[NEG] group had significantly lower levels of the secondary bile acids compared to the T1D[POS] group. This supports our hypothesis that reduced secondary bile acids may be associated with β cell dysfunction and disease progression in T1D. We screened important microbial taxa and metabolites associated with the risk of T1D onset and pancreatic β cell function from the discovery cohort and constructed an integrated prediction model. This model included microbial markers, as well as metabolite markers including the secondary bile acids DCA, LCA, and isoLCA. Compared to single microbial or metabolite markers, the combined multi-omics model demonstrated superior diagnostic performance, achieving the highest predictive accuracy in the training, internal validation, and external validation cohorts, which is consistent with previous study.[134]^33 Additionally, the combined model also achieved the best AUC performance in distinguishing residual islet function status among patients with T1D. Our findings, supported by previous studies, underscore the translational potential of microbiome and metabolomic biomarkers in clinical applications.[135]^34^,[136]^35^,[137]^36 Interventions to modulate the gut microbiota include dietary modifications (such as increasing dietary fiber intake), oral probiotic and prebiotic supplementation, and FMT.[138]^37 FMT is considered one of the most direct and effective approaches, rapidly reshaping the recipient’s gut microbiota.[139]^21 Studies have reported the potential of FMT in diabetes. In a trial of FMT intervention for type 2 diabetes (T2D), researchers found that 17 T2D subjects had significantly improved metabolic indicators such as blood glucose, blood lipids, and uric acid 12 weeks after treatment, and their 2-h postprandial C-peptide increased (from 4.503 ± 0.600 to 5.471 ± 0.728 ng/mL, p < 0.01).[140]^38 In 2021, in a randomized controlled trial conducted by Professor Max Nieuwdorp’s team, 20 newly diagnosed patients with T1D were randomly assigned to receive autologous or allogenic FMT treatment. The autologous FMT treatment group had better islet function preservation at 12 months of follow-up, possibly due to immune adaptation effects.[141]^22 Professor Zhang Faming’s team upgraded the traditional manually prepared FMT to an automated WMT using an intelligent fecal bacteria isolation system, significantly improving the safety of microbiota transplantation.[142]^23 We believe it may have better effects on immune adaptation. Our research team previously conducted WMT intervention in patients with brittle diabetes, not only reducing insulin dose requirements but also significantly decreasing the frequency of hypoglycemic episodes.[143]^24 Based on this experience, we applied WMT to the clinical treatment of T1D. Building upon our initial findings, we conducted a comprehensive investigation into the potential therapeutic effects of WMT in the clinical management of T1D. Our clinical investigation demonstrates that WMT offers promising therapeutic potential for T1D, exhibiting both immunomodulatory specificity and metabolic benefits. WMT significantly improved glycemic control and reduced daily insulin requirements while selectively attenuating key inflammatory pathways central to T1D pathogenesis. Flow cytometry analysis revealed a marked reduction in Th17 cells and their signature cytokine IL-17A—critical drivers of β cell destruction—alongside decreased IL-1β levels. Although Treg expansion showed a non-significant upward trend, these findings suggest partial immune rebalancing. Notably, Th1 cells remained largely unaffected, highlighting WMT’s targeted immunomodulation. However, not all T1D subjects receiving WMT treatment achieved improvement in islet function. While some patients (responders) exhibited varying degrees of improvement, others (non-responders) did not. Both groups had similar baseline clinical indicators but diverged in their gut microbiomes and metabolite profiles post-WMT. Responders showed significant increases in butyrate-producing species (e.g., Firmicutes sp., Roseburia faecis, and Clostridium leptum) and reduced Escherichia coli levels. Fecal bile acids (LCA and isoLCA) also increased significantly in responders but not in non-responders. These results suggest that WMT may improve pancreatic β cell function in responders by reshaping specific gut microbiota and enhancing secondary bile acid metabolism. This finding reveals the synergistic involvement of the gut microbiome and metabolites in WMT’s effects on pancreatic β cell function in T1D and provides a biological basis for explaining the interindividual response variability. Our study lays the foundation for developing individualized treatment strategies based on precise modulation of microbiota and their metabolites. These findings align with emerging insights into the role of secondary bile acid metabolism in immune and metabolic regulation. Recent studies have identified LCA derivatives—3-oxoLCA and isoLCA—as pivotal regulators of Th17/Treg equilibrium. 3-oxoLCA potently inhibits Th17 differentiation (reducing IL-17A), while isoLCA enhances Treg development via Foxp3 upregulation, mediated by gut bacteria such as Eubacterium species.[144]^39^,[145]^40 Complementing these discoveries, groundbreaking work by Professor Lin Shengcai’s team identified LCA as a key metabolite that mimics the anti-aging effects of calorie restriction by activating the AMPK pathway.[146]^41 This dual capacity of bile acids to regulate both immune and metabolic pathways provides a mechanistic framework for WMT’s therapeutic effects, suggesting that its benefits may stem from enhanced secondary bile acid production. Based on our research findings, the future may allow for precise modulation of the body’s immune status and islet function by regulating specific bacterial strains and their secondary bile acid metabolites. This innovative approach holds significant promise for the development of novel interventions aimed at alleviating or treating autoimmune diseases, particularly T1D. Limitations of the study The current work has several limitations. Firstly, our predictive model, developed from a cross-sectional study of the Chinese population, may be constrained by its reliance on data from a specific ethnic group and geographic region, potentially limiting its applicability to other ethnicities. Moreover, the model may not be generalizable to all age groups, given that the population used for model development was predominantly composed of adults. Secondly, the follow-up endpoint in this study was 3 months after WMT, which necessitates longer-term follow-up to evaluate the durability of treatment effects. Thirdly, ethical considerations precluded the inclusion of a placebo control group in the WMT study, introducing certain design flaws. Additionally, this study has not yet fully elucidated the specific mechanisms by which secondary bile acids ameliorate islet function in T1D. To address this gap, our research team is currently performing foundational experiments to explore the underlying pathways and molecular interactions. We anticipate that future studies will address these limitations and provide more robust evidence regarding the role of secondary bile acids in the pathogenesis and treatment of T1D. Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Yu Liu (drliuyu@njmu.edu.cn). Materials availability This study did not generate new unique reagents. Data and code availability * • The accession number for the metagenomics data reported in this paper is BioProject: PRJNA1226940. Clinical metadata for matched sequencing files of this study are available for academic use under confirmation of the [147]lead contact. * • The source code used in this study is publicly accessible through the repositories listed in the [148]key resources table. A comprehensive description of the analytical procedures is provided in the [149]method details section. * • Any additional information required to reanalyze the data reported in this work paper is available from the [150]lead contact upon request. Acknowledgments