Evaluation of the Crosstalk Between the Host Mycobiome and Bacteriome in Patients with Chronic Pancreatitis

Abstract

The human microbiome is a diverse consortium of microbial kingdoms that play pivotal roles in host health and diseases. We previously reported a dysbiotic bacteriome in chronic pancreatitis patients with diabetes (CPD) compared with patients with it's nondiabetic (CPND) phenotype. In this study, we extended our exploration to elucidate the intricate interactions between the mycobiome, bacteriome, and hosts' plasma metabolome with the disease phenotypes. A total of 25 participants (CPD, n = 7; CPND, n = 10; healthy control, n = 8) were recruited for the study. We observed elevated species richness in both the bacterial and fungal profiles within the CP diabetic cohort compared to the nondiabetic CP phenotype and healthy control cohorts. Notably, the CP group displayed heterogeneous fungal diversity, with only 40% of the CP nondiabetic patients and 20% of the CP diabetic patients exhibiting common core gut fungal profiles. Specific microbial taxa alterations were identified, including a reduction in Bifidobacterium adolescentis and an increase in the prevalence of Aspergillus penicilloides and Klebsiella sp. in the disease groups. In silico analysis revealed the enrichment of pathways related to lipopolysaccharide (LPS), apoptosis, and peptidase, as well as reduced counts of the genes responsible for carbohydrate metabolism in the CP groups. Additionally, distinct plasma metabolome signatures were observed, with CPD group exhibiting higher concentrations of sugars and glycerolipids, while the CPND cohort displayed elevated levels of amino acids in their blood. The fatty acid-binding protein (FABP) concentration was notably greater in the CPD group than in the HC group (4.220 vs. 1.10 ng/ml, p = 0.04). Furthermore, compared with healthy controls, disease groups exhibited fewer correlations between key fungal taxa (Aspergillus sp., Candida sp.) and bacterial taxa (Prevotella copri, Bifidobacteria sp., Rumminococcaceae). Our study unveils, for the first time, a dysbiotic mycobiome and emphasizes unique host bacterial-mycobial interactions in CP patient with diabetes, potentially influencing disease severity. These findings provide crucial insights for future mechanistic studies aiming to unravel the determinants of disease severity in this complex clinical context.

Supplementary information: The online version contains supplementary material available at 10.1007/s12088-024-01207-8.

Keywords: Chronic pancreatitis; Fungal dysbiosis; Gut microbiome; Metabolites; Microbial interactions.

Fig. 1

Flowchart representing the key features of the study

Fig. 2

Mycobial profiles of the studied participants. Violin plots illustrating the alpha diversity indices a Chao1 and b Shannon diversity indices in the fungal microbiome of CP and HC samples. c PCoA plot based on the Weighted Unifrac distance matrix depicting the beta/β-diversity of study participants. d Parallel plot representing the axis that captured most of the sample distribution in the PCoA plot. {Blue: CPND/CP-nondiabetic group, red: CPD/CP diabetic group, and green: matched Healthy participants} (PERMANOVA, pairwise: HC Vs CPND, p = 0.34; HC Vs CPD, p = 0.03 and CPND Vs CPD, p = 0.08)

Fig. 3

Bacterial profiles of study participants. Violin plots represent the alpha diversity indices a Chao 1 and b Shannon indices, c Principal component analysis (PCA) plot using Euclidean distance matrix represents the beta diversity of the participants. d Heatmap representing the intergroup and intragroup beta diversity, based on Euclidean distance matrix, among the study groups. Violin plots in panels a-d represent the top most abundant bacterial phyla viz. e Firmicutes, f Bacteroidetes, g Proteobacteria, h Actinobacteria. The p-values were derived using the Kruskal–Wallis test and are tabulated in Supplementary Table S2

Fig. 4

Plasma metabolomes of the studied participants. a Partial Least Squares-Discriminant Analysis (PLS-DA) plot based on the plasma metabolome of the participants (Accuracy: 0.72, R2: 0.830, Q2: 0.50485). b Heatmap constructed based on the PLS-DA matrix observed in the analysis. {The scale represents the raw z-scores}. c 2d plot based on the PLS-DA scores observed in the analysis. Box and whisker plots in the panel d–i represent classes of the metabolites detected in the studied participants. j A circular heatmap was constructed based on the median values of metabolites detected in the participants. {The scale represents the concentration values detected (µmol/L for amino acids, biogenic amines, and µM/L for lipids, hexoses, and others)}. All abbreviations have been spelled out in Supplementary Table S5

Fig. 5

Crosstalk between the host microbiome (bacteriome-mycobiome) and the metabolites in the studied samples. Corrplots and network plots depicting the interactions between the host bacteriome and mycobiome and host microbiome vs. plasma metabolome in a, d CPD, b, e CPND and c, f Healthy/HC groups. We used Spearman correlation to generate both Corrplots and Network plots. {Green—positive and pink—negative interactions}