Long-Term Intestinal Epithelial Remodeling Induced by Acute Protein-Energy Malnutrition

Abstract

Protein-energy malnutrition (PEM) is a global health burden with lasting effects that extend well beyond the initial nutrient deficiency. To systematically investigate the long-term effects of a single episode of PEM on the structure and function of the intestinal epithelium and its associated microbiota, we employed a comprehensive multi-omics approach, including (spatial) transcriptomics, DNA methylation analysis, fecal metagenomics, and metabolomics. Our findings show that PEM persistently alters the intestinal epithelium by depleting Paneth cells and suppressing antimicrobial gene expression - changes linked to DNA methylation that persist despite dietary recovery. In germ-free mice, the sustained epithelial phenotype after was absent. We identified the microbial lipid metabolite 9-HODE and epigenetically deregulated PPAR-driven GDF15 expression as key molecular drivers of the persistent PEM-induced Paneth cell dysfunction. Targeting microbial lipid production and its link to the host GDF15 pathway could offer novel therapeutic strategies for long-term consequences of malnutrition and other Paneth cell-associated diseases.

Keywords: GDF15; Paneth cell; Protein energy malnutrition; intestinal epithelium; intestinal inflammation; microbiome.

Figure 1:. PEM induces long-lasting changes in intestinal physiology independent of caloric intake.

A) Scheme of the PEM regimen in mice (n=10 mice per group). 12-week-old C57BL6 mice were given either isocaloric control or protein-reduced food for 28 days to induce an acute PEM, and the PEM-mice were then switched to control diet as well for an additional 42 days to monitor recovery from PEM. B) Body weight curve shows drastic weight loss during PEM and incomplete recovery after switch to control diet. *p < 0.05, ***p < 0.001 and ****p < 0.0001 using two-way ANOVA. C) Food intake of the experimental groups. ns = not significant using Mann-Whitney U test. D) Lengths of the small and large intestine during PEM and recovery. PEM shortened both the small and large intestine and the small intestinal length could not be restored during recovery. **p < 0.01, ***p < 0.001 and ns = not significant using Mann-Whitney U test. E-F) Fewer proliferating cells per crypt (E) and apoptotic cells per villus (F) in PEM mice as determined by Ki-67 immunostaining and TUNEL assay including representative images. The scale bar represents 50 μm. *p < 0.05, **p < 0.01 and ns = not significant using Mann-Whitney U test.

Figure 2:. Transcriptome analysis of purified intestinal epithelial cells reveals lasting antimicrobial and metabolic alterations.

A, C. Gene ontology (GO) terms enriched in upregulated (A) and downregulated (C) genes only at 4 weeks (PEM only), 10 weeks (Recovery only) and at both 4 and 10 weeks (Shared). Dot size is proportional to the gene ratio and color corresponds to the p-value of enrichment. Top selected terms are visualized. B, D. Heatmaps showing the expression of DEGs corresponding to selected GO terms enriched in upregulated (B) and downregulated (D) shared genes. Scaled normalized gene expression counts across all samples are plotted. Note that due to the GO setup, genes can belong to multiple GO categories.

Figure 3:. PEM impairs Paneth cell differentiation and leads to spatially confined gene expression changes along the crypt-villus axis.

A) Log2-fold change (y axis) of differentially expressed genes (DEGs) between mice under Ctrl and PEM diet at 4 weeks and 10 weeks (x axis) with n=10 per group. Color discriminates genes that are upregulated or downregulated only at 4 weeks (PEM only), 10 weeks (Recovery only) and at both 4 and 10 weeks (Shared), and point size represents statistical significance (adjusted p value). Paneth cell related genes are highlighted in blue. B) Boxplots showing average number of paneth cells (Lyz+) per crypt in duodenum and ileo-jejunal regions of the small intestine of mice under Ctrl and PEM diets along with representative histological images. ***: p < 0.001 Ctrl_4w versus PEM_4w and Ctrl_10w versus Rec_10w using Mann-Whitney U test. C) Classification of spots into multiple regions: crypts represented in green, intermediate villi in purple, and villi tips in orange. Identified spots were used for DEGs analysis across PEM vs Ctrl conditions. D-E) Top enriched gene ontology terms of significant upregulated (D) and downregulated (E) DEGs discovered in crypts are presented. DEGs were sorted into “up” (log2foldchange > 0) and “down” (log2foldchange < 0) categories with an adjusted p-value (padj) < 0.05. Similar steps of gene ontology analysis were then applied to the Villi-Intermediate and Villi-Tip regions.

Figure 4:. PEM-dependent persistent DNA methylation changes and ex vivo stability of gene expression changes.

A-B) Comparison of transcriptional PEM effects in intestinal epithelial cells (IEC, blue) and organoids (yellow) isolated of mice from two separate dietary PEM-recovery intervention experiments during (A) acute PEM and (B) after recovery. The dot plot depicts the Log₂[fold-changes PEM_4w/10w vs Ctrl_4w/10w] whereas the Venn diagram depicts the overlap in up- and downregulated DEGs. C) Number of differentially (hypo- and hyper-) methylated positions (DMPs) detected by BeadCHiP array in small intestinal epithelial cells from mice during acute PEM (4 weeks) and after recovery (10 weeks) and respective controls. D) Shared DMPs between acute PEM (4 weeks) and recovery (10 weeks). E) Predicted transcription factor binding sites enriched in observed DMPs. Dot size is proportional to the odds ratio and color corresponds to the p-value of enrichment. Top selected transcription factors are visualized. F) Schematic depiction of gene expression - DNA methylation integration analysis. Spearman’s rank correlation coefficient between the methylation intensity and gene expression values was calculated between the DMPs located within a range of 5kb before or after the transcription start site of the respective DEG. G) Venn Diagram depicting the overlaps between shared DEGs between PEM_4w and Rec_10w and DEGs correlated with their nearby DMPs, identifying 31 candidate DEGs linked with DMPs including GDF15 that has been linked to body mass regulation.

Figure 5:. Acute PEM drastically alters microbiome composition and function.

A) Stool sample collection scheme for microbiome analyses. Stool samples were collected weekly during the 10 weeks of diet experiment (normal or reduced protein diet) and used for 16S amplicon sequencing analysis and taxonomy characterization. Shot-gun metagenomic sequencing and functional potential analysis was performed only on stool samples from weeks 4 and 10 (Ctrl_4w, PEM_4w, Ctrl_10w and Rec_10w, each n=10 per group). B) Longitudinal alpha diversity of stool samples from mice under normal and reduced protein chow. Shannon index was computed at ASV level. Longitudinal cross-sectional comparisons were performed using Wilcoxon signed rank test. ns: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001. C) Principal Coordinate Analysis based on Bray-Curtis’s distances of stool fecal microbiomes of mice during PEM and recovery to explore further differences in microbiome composition. Differences in the centroids of groups were tested with PERMANOVA with 10000 permutations. D) Linear discriminant analysis (LDA) effect size (Lefse) of the fecal microbiomes at peak acute PEM (day 28, PEM_4w versus Ctrl_4w). The cladogram depicts the phylogenetic distribution of differential taxa and the bar plot depicts differential taxa ranked by LDA. Differences are shown by the color of the most abundant class, with red indicating PEM_4w and blue indicating Ctrl_4w. The diameter of each circle corresponds to the abundance of the taxon. This analysis uses the SILVA v138 taxonomy. The bar plot represents the top 30 taxa based on LDA scores. The letters in brackets denote the abbreviation of the taxon shown in the cladogram. E) Differential amino acid and fatty acid metabolic microbiome pathways (PEM_4w vs Ctrl_4w and Rec_10w vs Ctrl_10w). The triangles show the hues and direction of the effect size (b coefficient). Color intensity and size represent the magnitude of the effect size and FDR significance of the specific pathway, respectively.

Figure 6:. Absence of microbial signals attenuates the Paneth-cell phenotype and fecal metabolic alterations upon PEM.

A) Body weight curves of GF mice without a microbiome under 4-week PEM regimen followed by 6-week recovery with regular diet compared to CONV-R mice with a regular microbiome. Note that GF mice displayed milder disease symptoms with significantly less body weight loss during PEM and a full recovery (n=10–11 mice per group). B) Paneth (LYZ1-positive) cell count in small intestinal sections of GF mice. C) Top 20 most significant lipid metabolites identified in feces of GF and CONV-R mice fed a CTRL or PEM for 28 days. D) Levels of 9-HODE and 12,13diHOME in feces of GF and CONV-R mice during the PEM & recovery experiment. E) Bacterial lipoxygenase and epoxide hydrolase (EH) genes identified in stool metagenomics of GF and CONV-R mice during PEM & recovery. F) Lipoxygenase and EH genes in feces collected on day 0, 90 and 120 of malnourished children undergoing a nutritional intervention with either a Microbiota Directed Complementary Food (MDCF-2) or standard Ready-to-Use Supplementary Food (RUSF) using published metagenomics data. Data in (A) were compared by two-way ANOVA and in (B) and (D-F) using Wilcoxon signed rank test. ns: not significant, *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001.

Figure 7:. Microbiome and PEM dependent factors block Paneth cell differentiation.

A) Expression of Gdf15, Pparg and Pparbd were assessed in small intestinal tissue of both CONVR and GF mice subjected to PEM and after recovery. *p < 0.05, ***p < 0.001 and ****p < 0.0001, ns = non-significant. Significance testing was performed using Wilcoxon-Mann-Whitney-Test. B) Schematic drawing of organoid intervention experiment. Murine intestinal organoids were first cultured in ENR-CV stem cell organoid medium and then Paneth cell differentiation was induced using ENR-CD or PEM-CD medium. Stimulants (recombinant GDF15 [1 μg/ml], pan-PPAR-agonist Lanifibranor [20 μM] or the lipid 9-HODE [1 μM] were added during initial ENR-CD and PEM-CD culture and kept throughout the experimental duration. All experiments were performed independently at least twice in triplicates. C-E) Relative Lyz1, Defa5 and Gdf15 mRNA expression in organoids during Paneth cell differentiation and stimulated with (C) recombinant GDF15, (D) the Pan-PPAR-agonist Lanifibranor or the lipid (E) 9-HODE. Note that PPAR-activation boosts GDF15 levels thereby blocking Paneth cell differentiation and the bacterial PEM metabolite 9-HODE also suppresses Paneth cell differentiation. *: p < 0.05, **: p < 0.01, ***: p < 0.001, ****: p < 0.0001 and ns = not significant using Mann-Whitney U test.