Participant characteristics
A total of 114 subjects were recruited in our study, including 35 patients with IS-T2D, 50 patients with IS, and 29 healthy individuals. As shown by the data in Table 1, there were no significant differences in age and sex among the three groups. Risk factors such as hypertension, dyslipidemia, and CHD were not statistically significantly different between the IS-T2D and IS groups. The severity of infarction (admission NIHSS score) and short-term prognosis (90-Day mRS score) of patients with IS-T2D were worse than those with IS (P <0.05). IS-T2D and IS groups had significantly higher levels of leukocyte (LEU), NE, blood urea nitrogen (BUN), and serum creatinine (Scr) when compared with the HC group, and the IS-T2D group had the highest levels among the three groups (P <0.05). Moreover, the glucose and HbA1c of the IS-T2D group were significantly higher than those of the other two groups (P <0.001).
PAGln concentration was an independent risk factor of IS-T2D
We measured the plasma PAGln concentration of the subjects by the targeted liquid chromatography-mass spectrometry. The plasma PAGln levels of the patients with IS-T2D were significantly higher than IS patients and HC group (median, 2.77 vs 1.77μmol/L, P=0.001; median, 2.77 vs 1.04μmol/L, P<0.001) (Figure 1A). Furthermore, the PAGln levels of the IS group were also higher than the HC group (median, 1.77 vs 1.04μmol/L, P=0.027) (Figure 1A).
Next, we sought to explore the associations among IS, T2D, and the PAGln levels. The univariate logistic regression analysis showed that PAGln levels (OR 1.68, 95%CI 1.04 to 2.72, P=0.04) were a risk factor for patients with IS (Figure 1B). This significance still existed after adjusting gender, age, hypertension, CHD, smoking, LEU, NE, BUN, Scr, TG, TC, Hcy (OR 2.34, 95%CI 1.08 to 5.09, P=0.03) (Figure 1B). That could translate to an increase of 2.34-times in risk of IS for every 1 μmol/L increase in plasma PAGln levels. More importantly, elevated PAGln levels increased the risk of T2D for stroke patients by both univariate (OR 1.56, 95%CI 1.13 to 2.17, P=0.01) and multivariate logistic regression analyses (adjustment for gender, age, hypertension, CHD, smoking, LEU, NE, BUN, Scr, TG, TC, Hcy) (OR 1.53, 95% CI 1.01 to 2.30, P=0.044) (Figure 1C). This also means an increase of 1.53-times in risk of T2D in stroke patients for every 1 μmol/L increase in plasma PAGln levels. Therefore, increased plasma PAGln levels were an independent risk factor for IS patients and stroke patients with T2D.
The gut microbiota was disturbed in stroke patients withT2D
Previous studies have demonstrated that PAGln was a metabolite of gut microbiota. Moreover, PAGln was elevated in the plasma of patients with T2D [34, 35]. Therefore, to investigate whether PAGln-producing bacteria were enriched in the gut microbiota of stroke patients with T2D, we conducted the 16S rRNA high throughput sequencing.
The species accumulation boxplot showed that our fecal samples were sufficient and the species were abundant (Figure 2A). The Shannon index and Simpson index were calculated to evaluate the richness and evenness of intestinal microorganisms of each group. However, we found that there were no significant differences in α diversity among the three groups (Figure 2B - 2C ). Next, the β diversity of the three groups was analyzed by principal coordinate analysis (PCoA) based on the unweighted UniFrac distance. The results showed that the microbiome structures of both IS-T2D and IS patients were clearly separated from the HC group, but β diversity was similar between IS and IS-T2D groups (IS-T2D vs HC, R2= 0.050, P=0.001; IS vs HC, R2= 0.034, P=0.002; IS-T2D vs IS, R2= 0.017, P=0.059, Adonis test) (Figure 2D).
A total of 5360 OTUs were obtained from 114 fecal samples by 16S rRNA sequencing and classified into 63 bacterial phyla, 455 bacterial families, and 778 bacterial genera. The gut microbiota differed among the three groups at the phylum, family, and genus levels. At the phylum level, the gut microbiota was mainly composed of Firmicutes (51.0%), Proteobacteria (18.7%), Bacteroidota (21.3%), Actinobacteriota (4.2%), Fusobacteriota (1.7%), and Verrucomicrobiota (3.1%) (Figure 2E). The relative abundance of Firmicutes in patients with IS-T2D was significantly lower than that in the IS and HC groups (43% vs 52%, P=0.027; 43% vs 55%, P=0.005) (Figure 2F). The relative abundance of Proteobacteria and Verrucomicrobiota in the IS-T2D group was higher than that in the IS group (23% vs 17%, P=0.024; 4.3% vs 1.4%, P=0.05) (Figure 2F). At the family level, the harmful bacteria Enterobacteriaceae were enriched in IS-T2D patients with significant differences compared with IS and HC groups (21.6% vs 16.5%, P=0.027; 21.6% vs 13.0%, P=0.019) (Figure 2G). The relative abundance of Akkermansiaceae in the IS-T2D group was the highest among the three groups, but there was a significant difference only between the IS and HC groups (1.4% vs 3.3%, P=0.046) (Figure 2G). At the genus level, the beneficial bacteria Faecalibacterium, Dialister, and Roseburia were all significantly reduced in IS-T2D and IS groups when compared with the HC group (Faecalibacterium, 7.4% vs 15.4%, P<0.001; 9.6% vs 15.4%, P=0.003; Dialister, 3.0% vs 3.7%, P<0.001; 2.0% vs 3.7%, P=0.005; Roseburia, 0.9% vs 3.4%, P<0.001, 1.4% vs 3.4%, P=0.001) (Figure 2H). And the relative abundance of harmful bacteria Klebsiella significantly increased in the IS-T2D group compared with the HC and IS groups (4.5 vs 1.8, P=0.002; 4.5 vs 2.5, P=0.01) (Figure 2H). In summary, the diversity of gut microbiota in patients with IS-T2D decreased, with decreased beneficial bacteria and increased harmful bacteria.
In addition, to better identify the microbial markers among the three groups, we used the LEfSe tool. The results showed that there were a total of 20 gut microbiota taxa from phylum to species, which were differentially abundant bacterial taxa (LDA score >4) in three groups (Figure 2I). Specifically, there were 6 increased abundant taxa in the IS-T2D group, including o_Enterobacterales, f_Enterobacteriaceae, p_Verrucomicrobiota, c_Verrucomicrobiota, s_Klebsiella_pneumoniae, and g_Klebsiella.
Next, we performed PICRUSt analysis to predict potential functional pathways of the microbiome communities, and compared the differences between the IS-T2D and IS groups (Figure 2J). In total, there were 300 differential KEGG pathways (level 3), of which 49 were significantly different with the average relative abundance of one of the two groups no less than 0.1%. There were 29 pathways enriched but 20 pathways poor in the IS-T2D group. Comparing with the IS group, we found that amino acids metabolism (such as glycine, serine, and threonine metabolism, P=0.033; tyrosine metabolism, P=0.014; beta-alanine metabolism, P=0.020; tryptophan metabolism, P=0.002) was significantly more abundant in the IS-T2D group (Figure 2J).
PAGln levels were associated with characteristic microbiota related to IS-T2D, poor prognosis, and NETs-related inflammation
To investigate the association between PAGln levels and IS-T2D, we performed Spearman correlation analysis. As shown in Table 2, the PAGln levels were positively correlated with age, BUN, Scr, Glu, HbA1c, and Hcy in all subjects, and the correlations between age and PAGln were especially pronounced in stroke patients with T2D. Furthermore, it is worth noting that the PAGln levels were positively correlated with mRS score of 90 days after onset in all stroke patients. In addition, we found that the PAGln levels were positively correlated with NE in stroke patients with T2D, suggesting an association between inflammation and PAGln levels in stroke patients with T2D.
NET is a bactericidal substance, released extracellularly after NE activation, and its dysregulation can lead to inflammation [36]. Therefore, we measured the plasma NET concentration of all subjects. We evaluated the concentration of CitH3 in the plasma to represent the level of NETs [37]. Our results showed that the plasma CitH3 levels in IS-T2D patients were significantly higher than those in IS and HC groups (6.34 vs 5.57ng/ml, P<0.001; 6.34 vs 4.73ng/ml, P<0.001). The plasma CitH3 levels in the IS group were also significantly higher than those in the HC group (5.57 vs 4.73ng/ml, P<0.001) (Figure 3A). Moreover, Spearman correlation analysis revealed a significant correlation between PAGln and CitH3 levels (r=0.41, P<0.001) (Figure 3B). We further analyzed the distribution of CitH3 levels based on PAGln levels as assessed by quartiles. The concentration of plasma CitH3 showed a dose-dependent increase according to PAGln levels with the highest levels being observed in subjects with the highest PAGln concentrations (Q4: 6.15 vs Q1: 5.27ng/ml, P<0.01) (Figure 3C).
Next, we explored the relationships between PAGln, clinical indicators, and gut microbiota. We performed Spearman correlation analysis on the data of all stroke patients. Spearman correlation analysis showed that PAGln level was significantly positively correlated with 6 microbial markers (o_Enterobacterales, f_Enterobacteriaceae, p_Verrucomicrobiota, c_Verrucomicrobiota, s_Klebsiella_pneumoniae, and g_Klebsiella) related to IS-T2D. Interestingly, o_Enterobacterales, f_Enterobacteriaceae, p_Verrucomicrobiota, and c_Verrucomicrobiota were also positively correlated with NE. Meanwhile, o_Enterobacterales and f_Enterobacteriaceae were positively correlated with Glu and Hcy; s_Klebsiella_pneumoniae and g_Klebsiella were positively correlated with HbA1c (Figure 3D).
PAGln levels, NETs levels, and gut microbiota were diagnostic indexes for IS-T2D.
We next explored whether the plasma PAGln, gut microbiota, and NETs could be used as biomarkers of stroke with T2D. Based on the plasma PAGln levels, differential microbiota (defined as the relative abundance of bacteria in gut microbiota with LDA score >4), and NETs levels, we conducted ROC analysis. As shown in Figure 4, plasma PAGln levels (AUC: 0.7160, 95%CI: 0.6074-0.8246; P=0.0007), different gut microbiota (AUC: 0.8757, 95%CI: 0.8016-0.9462; P<0.0001) and NETs levels (AUC: 0.8229, 95%CI: 0.7353-0.9105; P<0.0001) could well distinguish patients with IS-T2D from stroke patients (Figure 4). Notably, when plasma PAGln levels and the differential microbiota were incorporated into the model construction, the AUC increased to 88.7 ± 3.54% (P < 0.0001). And when the PAGln and NETs levels were incorporated into the model construction, the AUC increased to 83.4 ± 4.38% (P < 0.0001). Finally, we included PAGln levels, NETs levels, and differential microbiota into the model construction, the area under the ROC curve was up to 94.7 ± 2.26% (P < 0.0001). In summary, the intestinal metabolite PAGln, differential microbiota, and inflammatory indicator NET could all be used as diagnostic indicators for stroke with T2D. Moreover, the combination of these biomarkers might improve diagnostic efficiency.
Elevated PAGln levels in IS-T2D patients could be transmitted through the gut microbiota
To investigate whether the gut microbiota of stroke patients with T2D contributed to the elevation of plasma PAGln levels, we treated antibiotic-treated rats with a fecal transplant from IS-T2D and IS patients (Figure 5A). Results demonstrated that compared with the rats receiving fecal microbes from patients with IS (FMT-IS group), the PAGln concentration of the rats receiving fecal microbes from patients with IS-T2D (FMT-IS-T2D group) was significantly increased and nearly doubled (FMT-IS-T2D vs FMT-IS, P=0.02) (Figure 5D). MCAO model was established in rats after FMT. 24 hours after stroke, the rats receiving fecal microbes from IS-T2D patients had more severe stroke than the rats receiving fecal bacteria from IS patients, with decreased neurological function score (FMT-IS-T2D vs FMT-IS, P=0.042) (Figure 5B) and increased infarct volume (FMT-IS-T2D vs FMT-IS, P=0.042) (Figure 5C). Meanwhile, after stroke, PAGln levels in rats receiving fecal microbes from patients with IS-T2D were also nearly twice as high as those receiving fecal microbes from patients with IS (FMT-IS-T2D vs FMT-IS, P=0.03) (Figure 5D). Notably, PAGln levels in rats receiving fecal microbes from patients with IS-T2D were further elevated after stroke (FMT-IS-T2D vs FMT-IS-T2D-M, P=0.021), while the PAGln levels in rats receiving fecal microbes from IS patients unchanged (FMT-IS vs FMT-IS-M, P=0.31).