The clinical laboratory parameters in T2DM and T2DM-DF

We recruited 33 individuals with T2DM (average age 46, 57.5% male), 32 individuals with T2DM-DF (average age 54, 75% male), and 32 non-obese subjects without any known disease as an HC (mean age 43 years, 27% male). All participants submitted blood and fecal samples. As expected, patients with T2DM had a significantly higher body mass index (BMI) than HC. Conversely, those with T2DM-DF did not show a clear distinction in BMI compared to HC but did have a significantly lower BMI than individuals with T2DM. It is well known that a higher BMI represents a higher obesity index. In contrast, patients with T2DM-DF have a lower BMI than T2DM, which may be due to disorders of glucose metabolism, insulin resistance, prolonged dietary control, or medication side effects resulting from a prolonged diabetes process.

Some clinical laboratory parameters were significantly altered in T2DM and T2DM-DF (Table 1). Among these, AST and BUN were only markedly higher in patients with T2DM-DF than in HC. Elevated AST indicates liver damage, while BUN indicates kidney function. A proportion of patients with T2DM-DF are associated with liver and kidney damage, which may be caused by diabetic neuropathy, diabetic vasculopathy, diabetic nephropathy, and co-infections. TBIL and TG were significantly lower in T2DM-DF than in T2DM. TBIL is also a marker to evaluate liver function. The decrease in TG is due to inadequate lipid intake in patients with T2DM-DF due to a more prolonged unbalanced diet or due to excessive use of hypoglycaemic drugs. And UBIL, TC, LDL-C, RBC, and Hb were noticeably expressed only in T2DM-DF (Table 1).

Table 1 Clinical characteristics of HC, T2DM, T2DM-DF.

This suggests that T2DM and T2DM-DF showed significant differences in relevant clinical laboratory parameters.

The effect of diversity and richness of gut fungal community between HC, T2DM and T2DM-DF

To determine whether the disease severity or other clinical metadata of T2DM and T2DM-DF are related to gut fungi, total DNA samples were extracted from fecal specimens obtained from all groups, followed by Illumina 18s rRNA amplicon sequencing and fungal taxonomic classification. All samples were correctly sequenced. After filtering the qualified reads, all samples’ quality-filtered reads were 7,026,239, and the average amplicon length was 241 bp. The raw sequencing data are available on request. The results showed that the sequencing reads were enough to represent most of the fungal community.

Next, differences in the richness and diversity of fungal communities among the three groups were assessed using α-diversity. The α-diversity was presented as four indexes: Chao index and ACE index estimate the community richness; Simpson index and Shannon index estimate the community diversity. Compared with the HC, T2DM and T2DM-DF showed a significant decrease in both the Chao index and ACE index, but the reduction in T2DM-DF was more prominent. Simpson index and Shannon index of T2DM were markedly changed compared with T2DM-DF, while there was an inapparent from HC. However, T2DM-DF was notably different from the HC (Fig. 1a).

Fig. 1: T2DM and T2DM_DF alter gut fungal richness and diversity.

The T2DM and T2DM_DF gut fungal changes on (a) α-diversity were determined using Chao index, Ace index, Simpson index and Shannon index; b Venn diagram illustrated overlap of OTUs in gut fungi among the samples; c, d β-diversity was determined using Euclidean distance based principal component analysis (PCA) and weighted UniFrac distance based principal coordinates analysis (PCoA). Data are expressed as means ± SEM. *p < 0.05; **p < 0.01 and ***p < 0.001.

A Venn diagram overlaps provided a better understanding of the shared richness among groups. Only 135 out of the total 1596 OTUs were found in all groups, with the remaining OTUs distributed among three groups, two groups, or within each group (Fig. 1b). Moreover, OTUs in T2DM and T2DM-DF were significantly reduced compared with HC (Fig. 1b).

The β-diversity reflects the difference in fungal community composition among the three groups. A principal component analysis (PCA) (Fig. 1c) plot and principal coordinates analysis (PCoA) (Fig. 1d) were conducted. The variance analysis showed a significant difference in the overall fungal community composition among various groups.

These data indicate that the T2DM and T2DM-DF remarkably reduced the richness and diversity of fungal communities, and it was also different in T2DM-DF compared to T2DM.

Compositional changes of the gut fungal community in patients with T2DM and T2DM-DF

The gut fungal profiling from HC, T2DM, and T2DM-DF revealed two significantly altered phyla, Ascomycota and Basidiomycota, of which two accounted for 99% of the fungal community (Fig. 2a). In the HC, Ascomycota was the dominant phylum, followed by Basidiomycota (Fig. 2a). The abundance of Ascomycota in T2DM-DF was significantly higher than that in HC and T2DM (Fig. 2b). In addition, the abundance of Basidiomycota in T2DM-DF was considerably lower than in HC and T2DM (Fig. 2c). Interestingly, the Ascomycota/Basidiomycota ratio was notably altered in both T2DM and T2DM-DF compared with HC, and a significant increase in the Ascomycota/Basidiomycota ratio occurred in T2DM-DF compared with T2DM (Fig. 2d). That is to say, the gut fungi in T2DM-DF were changed more drastically at the phylum level.

Fig. 2: The gut fungal composition in patients with T2DM and T2DM_DF.

a The relative abundance of the major fungal phyla; b Relative abundance of Ascomycota; c Relative abundance of Basidiomycota; d The Ascomycota to Basidiomycota ratio; e Relative abundance of the top 30 different families; f Relative abundance of the top 30 different genera; g Relative abundance of the top 15 different species. h Network of fungi. Analyzed by Spearman rank correlation analysis. The size of the circle represents the average abundance of the species; the line represents the correlation between the two species; the thickness of the line represents the strength of the correlation; and regarding the color of the line, red represents a positive correlation. Data are expressed as means ± SEM. *p < 0.05; **p < 0.01 and ***p < 0.001 means HC versus T2DM; &p < 0.05, &&p < 0.01 and &&&p < 0.001 means HC versus T2DM_DF; #p < 0.05; ##p < 0.01 and ###p < 0.001 means T2DM versus T2DM_DF; or *p < 0.05; **p < 0.01 and ***p < 0.001.

Among the top 30 families, compared with HC, 6 families were significantly changed in T2DM, and 17 families in T2DM-DF (Fig. 2e). Among them, it was observed that Aspergillaceae, Sporidiobolaceae, Filobasidiaceae and Cladosporiaceae were significantly decreased in both T2DM and T2DM-DF, whereas Trichosporonaceae and Dipodascaceae were only drastically increased in T2DM. Sporormiaceae, Debaryomycetaceae, unclassified_o__Saccharomycetales, Sclerotiniaceae, Auriculariaceae, unclassified_p__Ascomycota, unclassified_o__Hypocreales, Niessliaceae, unclassified_k__Fungi, Cordycipitaceae, Nectriaceae and Hypocreales_fam_Incertae_sedis were only notable reduced in T2DM-DF. Remarkably, Schizophyllaceae completely disappeared in the gut of T2DM-DF (Fig. 2e). It can be seen that gut fungi of T2DM-DF were more altered at the family level. To further determine the changes, we compared gut fungi of T2DM and T2DM-DF at the family level and found that 7 families had significant changes in T2DM-DF compared with T2DM. More interesting is that Saccharomycetales_fam_Incertae_sedis alone significantly increased, while Psathyrellaceae alone significantly decreased (Fig. 2e). Therefore, these two fungi are expected to be biomarkers for the diagnosis of T2DM-DF and also provide the possibility of targeting gut fungi for the treatment of T2DM-DF.

In the top 30 genera, compared with HC, a total of 5 genera were significantly changed in the T2DM, and 15 genera in T2DM-DF (Fig. 2f). Among them, it was observed that Xeromyces, Naganishia, and Rhodotorula were significantly decreased in both T2DM and T2DM-DF, whereas unclassified_f__Dipodascaceae only drastically increased in T2DM. It’s worth noting that Cutaneotrichosporon is absent in the gut of HC, but it is significantly increased in T2DM and T2DM-DF (Fig. 2f). It reminds us that this fungus is expected to be a biomarker for diagnosing T2DM and T2DM-DF. Unclassified_o__Saccharomycetales, unclassified_p__Ascomycota, unclassified_o__Hypocreales, Niesslia, unclassified_k__Fungi, Debaryomyces, Didymella, Cladosporium, Aspergillus were only notable reduced in T2DM-DF; and Penicillium was only notable increased in T2DM-DF. Remarkably, Schizophyllum completely disappeared in the gut of T2DM-DF (Fig. 2f). It can be seen that gut fungi of T2DM-DF were more altered at the genera level. To further determine the changes, we compared gut fungi of T2DM and T2DM-DF at the genera level and found that 8 genera had significant changes in T2DM-DF compared with T2DM. More interesting is that Candida significantly increased alone, while unclassified_f__Dipodascaceae significantly decreased alone (Fig. 2f). Therefore, these two fungi are expected to be biomarkers for diagnosing T2DM-DF and also provide the possibility of targeting gut fungi for treating T2DM-DF.

Following this, an examination was conducted on the variations in the gut fungal makeup of the highest-ranking 15 species. Significant differences in the relative abundance of all 15 species were discovered among HC, T2DM, and T2DM-DF. Compared with HC, T2DM had 3 species increased, which were Cutaneotrichosporon_guehoae, unclassified_g_Cutaneotrichosporon, unclassified_f_Dipodascaceae, and 12 species decreased, which are Rhodotorula_mucilaginosa, unclassified_g_ Penicillium, Naganishia_sp, Aspergillus_minisclerotigenes, Cladosporium_delicatulum, unclassified_g_Aspergillus, Aspergillus_penicillioides, Xeromyces_bisporus, Aspergillus_amstelodami, Candida_solani, Filobasidium_sp, Penicillium_brevicompactum (Fig. 2g). Whereas in T2DM-DF, there had 5 species increased, which were unclassified_g_Candida, unclassified_g_ Penicillium, Cutaneotrichosporon_guehoae, unclassified_g_Aspergillus, unclassified_g_Cutaneotrichosporon, and 10 species decreased, which were Rhodotorula_mucilaginosa, Naganishia_sp, Aspergillus_minisclerotigenes, Aspergillus_penicillioides, Aspergillus_amstelodami, Cladosporium_delicatulum, Xeromyces_bisporus, unclassified_g_Didymella, unclassified_g_Debaryomyces, Aspergillus_cibarius (Fig. 2g). We found that some fungal species showed different trends of increases and decreases in T2DM and T2DM-DF compared to HC. To further determine the differences, we analyzed their gut fungal species. Compared with T2DM, only one species increased in T2DM-DF, which was unclassified_g_Candida, and the remaining 14 species were all reduced (Fig. 2g). It’s worth noting that unclassified_g_Candida showed a most prominent increase in T2DM-DF, but it was not in the top 15 species of HC and T2DM by the analysis.

A fungal network created by Spearman was utilized to examine the connections among these fungal species. Subsequent network analysis indicated that Aspergillus_cibarius, Saccharomyces, unclassified_g_Aspergillus, and Aspergillus_minisclerotigenes may potentially be critical players in the interactions (Fig. 2h).

Furthermore, the composition of the fungal community, which displayed significant variations between the HC, T2DM, and T2DM-DF groups, was examined using the linear discriminant analysis (LDA) effect-size method (LEfSe) (Fig. 3a, b). Then, it was figured out that taxa in different levels had differential abundance in three groups. We observed that Sporidiobolales, Microbotryomycetes, Sporidiobolaceae, Rhodotorula_mucilaginosa, Rhodotorula, Eurotiomycetes, Eurotiales, Aspergillaceae, Filobasidiaceae, Filobasidiales, Aspergillus, Naganishia_sp, Naganishia, Dothideomycetes, Aspergillus_minisclerotigenes, Agaricomycetes, Debaryomycetaceae, Capnodiales, Cladosporiaceae, Cladosporium were identified as significant biomarkers based on our observations in HC group. However, Trichosporonales, Trichosporonaceae, Cutaneotrichosporon, Sordariomycetes, Dipodascaceae, unclassified_f_Dipodascaceae, Sordaria, Sordariaceae, unclassified_g_Sordaria, Pleosporales, unclassified_f_Dipodascaceae, Aspergillus_cibarius played a significant role and could sever as biomarkers in T2DM group. Furthermore, unclassified_g_Candida, Saccharomycetales, Saccharomycetes, Ascomycota, Penicillium, unclassified_g_Penicillium, unclassified_g_Cutaneotrichosporon were essential and could serve as indicators in T2DM-DF group.

Fig. 3: Fungal biomarker in patients with T2DM and T2DM_DF.

LDA effect size (LeFSe) analysis identifies discriminant taxa among the four groups. a Cladogram of the fungal community. Significant discriminant taxon nodes of the HC, T2DM, and T2DM_DF are represented by red, blue, and green, respectively. At the same time, nondiscriminant taxon nodes are represented by yellow. Branch areas are shaded according to the highest-ranked variety for that taxon. b The LDA score indicates the level of differentiation among the three groups. A threshold value of 4.0 was used as the cutoff level. Horizontal bar chart showing discriminant taxa. Significant discriminant taxa of the HC, T2DM, and T2DM_DF are represented by red, blue, and green, respectively. p < 0.05.

The gut fungal function of T2DM and T2DM-DF

Both the gut microbiota and gut fungi play crucial roles in the host’s physiological functions, as is widely recognized. Furthermore, this significant capacity impacts overall body metabolism and plays a vital role in developing endocrine disorders. Thus, PICRUSt was utilized to forecast the functional capabilities of fungi in individuals with T2DM and T2DM-DF, followed by additional examination within the Kyoto Encyclopedia of Genes and Genomes (KEGG) database framework. Subsequently, based on the results, the focus was on identifying the enzyme associated with KEGG functional categories. Compared with HC, only one enzyme had significant changes in T2DM, which was unspecific monooxygenase. This enzyme also decreased dramatically in T2DM-DF compared to HC and T2DM. Moreover, three enzymes were significantly altered in T2DM-DF alone: DNA-directed RNA polymerase and glucan 1, 4-alpha-glucosidase was notably decreased compared to HC, and histone acetyltransferase was remarkably reduced compared to HC or T2DM (Fig. 4). To sum up, the gut fungal functions had more changes in T2DM-DF than T2DM.

Fig. 4: The enzyme function of fungal community-based Kyoto Encyclopedia of Genes and Genomes (KEGG) database in patients with T2DM and T2DM_DF.

*p < 0.05; **p < 0.01 and ***p < 0.001 means HC versus T2DM; &p < 0.05, &&p < 0.01 and &&&p < 0.001 means HC versus T2DM_DF; #p < 0.05; ##p < 0.01 and ###p < 0.001 means T2DM versus T2DM_DF.

Signature fungal species in HC, T2DM, T2DM-DF biomarkers in three groups

Based on the compositional differences and functional analyses of gut fungal species in each group, and to more accurately identify the signature fungi, we used random forest to rank the top 15 importance of all fungi in each group. Results show that compared with HC, four of the top 15 most important fungal species in T2DM and T2DM-DF were the same; 11 fungal species could be used as biomarkers to distinguish T2DM or T2DM-DF from healthy individuals. In T2DM, that is Naganishia_sp, Candida_solani, Filobasidium_sp, unclassified_f_Dipodascaceae, unclassified_g_Cutaneotrichosporon, Wallemia_ichthyophaga, unclassified_g_Erythrobasidium, Trichosporon_coremiiforme, unclassified_g_Penicillium, Cystobasidium_laryngis, Mucor_racemosus (Fig. 5b). And in T2DM-DF, that is Candida_solani, unclassified_k_Fungi, Cladosporium_delicatulum,

Fig. 5: The fungal species importance ranking of signature species in patients with T2DM and T2DM_DF.

It is based on a random forest and reflects different species’ ability to affect a classification model’s accuracy. The different fungi between (a) HC versus T2DM versus T2DM_DF, (b) HC versus T2DM, (c) HC versus T2DM_DF, and (d) T2DM versus T2DM_DF. The y-axis is the importance of species, the X-axis is equal to the importance measurement value of species/standard deviation value, and The Y-axis corresponds to species names in order of importance. e–g ROC analysis of three groups: HC versus T2DM, HC versus T2DM_DF, and T2DM versus T2DM_DF. ROC analysis is often used to evaluate the performance of classification models in microbiome data, for example, to distinguish the state of samples before and after processing, determine differences between different biological samples, etc. The X-axis of Specificity is 1–0, The Y-axis is Sensitivity, and the coordinate axis is 0–1. The point marked on the curve is the optimal critical value (the values in parentheses are the corresponding Specificity and Sensitivity of the point). “bar” is the confidence interval of Specificity and Sensitivity corresponding to the point. The AUC marked in the figure is the area under the corresponding curve; The area value under the ROC curve is usually between 1.0 and 0.5. The closer the AUC is to 1, the better the diagnostic effect is. AUC has low accuracy at 0.5–0.7, sure accuracy at 0.7–0.9, and high accuracy above 0.9.

Candida_albicans, unclassified_f_Mycosphaerellaceae, Aspergillus_amstelodami, Niesslia_exilis, Penicillium_brevicompactum, unclassified_g_Aspergillus, unclassified_o_Hypocreales, unclassified_g_Debaryomyces (Fig. 5c). This result also corroborates part of the LEfSe analyses. Next, to better distinguish the differential fungal species between T2DM and T2DM-DF, we used the random forest to rank the signature distinct species between the two groups. Among the top 15 fungi, 2 were identical to T2DM, and 5 were identical to T2DM-DF when compared to HC. Hence, 8 fungi species could be used as biomarkers to distinguish DF in T2DM, they were unclassified_g_Candida, unclassified_g_Lophiostoma, Anteaglonium_sp, Candida_quercitrusa, unclassified_g_Sordaria, Rhodotorula_babjevae, Trichothecium_roseum, Exophiala_dermatitidis (Fig. 5d). Identifying these fungal species offers a foundation for forecasting the development of DF in individuals with T2DM through fungal detection techniques, potentially enabling earlier DF treatment. Finally, to further identify the signature important fungal species in the three groups, we conducted a combined analysis of three groups, which showed that 2 were not seen in the previous two-by-two comparison; there were unclassified_o_Saccharomycetales and Thermomyces_lanuginosus (Fig. 5a). The receiver operating characteristic curve (ROC) also showed a better accuracy of the predicted results (Fig. 5e–g).

Gut fungi-associated obesity index, liver function and blood parameters

The heat map correlation analysis revealed that 8 of the top 15 fungal species showed significant correlations with obesity index, liver function, and blood parameters (Fig. 6). There, 5 fungal species were significantly associated with one factor only. Aspergillus_cibarius was obviously and positively correlated with hemoglobin (Hb); Aspergillus_minisclerotigenes was obviously and positively correlated with lymphocyte (LYM); unclassified_g_Penicillium was obviously and positively correlated with blood urea nitrogen (BUN), whereas Dipodascaceae_sp was noticeably and negatively correlated with body mass index (BMI); unclassified_g_Cutaneotrichosporon was noticeably and negatively correlated with total bilirubin (TBIL) (Fig. 6). In addition, 3 fungal species were significantly associated with various factors that could potentially contribute to the onset of T2DM and T2DM-DF. Rhodotorula_mucilaginosa was dramatically and positively correlated with lymphocyte percentage (L%) but also obviously and negatively correlated with FBG and HbA1c. Naganishia_sp showed a clear and robust positive relationship with high-density lipoprotein cholesterol (HDL-C) but dramatically and negatively correlated with fasting blood glucose (FBG), glycated hemoglobin (HbA1c), and absolute neutrophil count (ANC) (Fig. 6). It means that Naganishia_sp may alleviate T2DM and T2DM-DF. unclassified_g_Candida was dramatically and positively correlated with direct bilirubin (DBIL), FBG, HbA1c, white blood cell (WBC), neutrophil granulocyte percentage (N%), and ANC, but dramatically and negatively correlated with HDL-C, L%, red blood cell (RBC) and Hb (Fig. 6). It means that unclassified_g_Candida may induce T2DM and T2DM-DF development. Overall, the results show that these 8 species of fungi are crucial in developing T2DM and T2DM-DF.

Fig. 6: The relationship between obesity index, liver function, blood parameters, and the 15 top fungal species is estimated by Spearman’s correlation analysis.

*p < 0.05; **p < 0.01; ***p < 0.001. BMI body mass index, ALT alanine aminotransferase, AST aspartate aminotransferase, TBIL total bilirubin, DBIL direct bilirubin, UBIL unconjugated bilirubin, TC total cholesterol, TG Triglyceride, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, FBG fasting blood glucose, HbA1c glycated hemoglobin, BUN blood urea nitrogen, WBC white blood cell, N% neutrophil granulocyte percentage, L% lymphocyte percentage, ANC absolute neutrophil count, LYM lymphocyte, RBC red blood cell, Hb hemoglobin, PLT blood platelet.