Drosophila Histone Demethylase KDM5 Regulates Social Behavior through Immune Control and Gut Microbiota Maintenance

Abstract

Loss-of-function mutations in the histone demethylases KDM5A, KDM5B, or KDM5C are found in intellectual disability (ID) and autism spectrum disorders (ASD) patients. Here, we use the model organism Drosophila melanogaster to delineate how KDM5 contributes to ID and ASD. We show that reducing KDM5 causes intestinal barrier dysfunction and changes in social behavior that correlates with compositional changes in the gut microbiota. Therapeutic alteration of the dysbiotic microbiota through antibiotic administration or feeding with a probiotic Lactobacillus strain partially rescues the behavioral, lifespan, and cellular phenotypes observed in kdm5-deficient flies. Mechanistically, KDM5 was found to transcriptionally regulate component genes of the immune deficiency (IMD) signaling pathway and subsequent maintenance of host-commensal bacteria homeostasis in a demethylase-dependent manner. Together, our study uses a genetic approach to dissect the role of KDM5 in the gut-microbiome-brain axis and suggests that modifying the gut microbiome may provide therapeutic benefits for ID and ASD patients.

Keywords: H3K4me3; KDM5; demethylase; gut microbiome; social behavior.

Figure 1. Loss of KDM5 causes intestinal epithelial barrier disruption and impaired social behavior.

A. Levels of KDM5, H3K4me3 and Histone H3 in intestine tissues from wildtype (wt, w¹¹¹⁸) and kdm5^K6801/10424 flies using Western blot analysis. B. Transcription levels of kdm5 in intestine tissues from wt and kdm5^K6801/10424 flies using RNA-seq analysis. C. Representative images of the digestive tract from wt and kdm5^K6801/10424 flies (left panel, 4×). The red dotted box highlights a portion of the midgut showing intestinal defects with bubbles in kdm5^K6801/10424 flies compared to wt (right panel 20×). D. Percentage of flies with intestinal defects with bubbles in wt and kdm5^K6801/10424 flies at the age of 3–5 days. In total, 30–40 flies each group were examined. E. Transmission electron micrographs of the epithelial barrier in the anterior (top-panel) and posterior (bottom-panel) midgut from wt and kdm5^K6801/10424 flies (2.9K magnification). F. Intestinal permeability analysis for wt and kdm5^K6801/10424 flies was assessed by measuring the post-feeding distribution of a non-absorbable blue food dye (top-panel). Quantification of the percentage of body area with distribution of non-absorbable blue food dye relative to whole body (bottom panel). G. The digestive tract of kdm5^K6801/10424 flies are highly permeable. Conventional fluorescence microscopy revealed that FITC-labeled beads remained in the lumen of wild-type flies after feeding. In contrast, FITC signals were diffuse in the gut of kdm5^K6801/10424 flies. H. Protein Levels of KDM5, H3K4me3 and Histone H3 in heads and intestine tissues from Myo1A-Gal4^TS/+ and Myo1A-Gal4^TS/kdm5^RNAi flies using Western blot analysis (top panel). Quantification of intensity for western blot is shown in the bottom panel. I. Post-feeding distribution of a non-absorbable blue food dye in Myo1A-Gal4^TS/+ and Myo1A-Gal4^TS/kdm5^RNAi flies to measure intestinal permeability. Quantification of the percentage of body area with distribution of non-absorbable blue food dye relative to whole body. J. kdm5^K6801ñ0424 flies showed an increase in social space compared to wt and kdm5 rescued flies. K. Social avoidance analysis showed a decrease activity for kdm5^K6801/10424 flies compared to wt and kdm5 rescued flies. L. Distribution of social space in control Myo1A-Gal4^TS and Myo1A-Gal4^TS/kdm5^RNAi flies. Myo1A-Gal4^TS/kdm5^RNAi flies showed an increase in social space compared to control. M. Social avoidance analysis showed decreased activity in Myo1A-Gal4^TS/kdm5^RNAi flies compared to control. N. Direct social contacting time was reduced in female kdm5^K6801/10424 flies compared to wt. O. Direct social contacting time was reduced in male Myo1A-Gal4^TS/kdm5^RNAi compared to control. Error bars represent minimum and maximum (J and L). Other data are shown as mean ± SEM. n = 3 (J–M). n>8 (N and O). *p < 0.05, **p < 0.01, and ***p < 0.001. In total, 10–15 flies in each group were examined (F, G, and I). Data shown in (D) and (F)–(I) are representative of two independent experiments.

Figure 2. Loss of KDM5 protein alters gut microbiota composition.

A. Observed bacterial species richness in gut samples from wt and kdm5^K6801/10424 flies. P value<0.05 by Wilcoxon rank sum test. B. Overlap of observed bacterial OUTs in gut samples from wt and kdm5^K6801/10424 flies. C. Hierarchical clustering of wt and kdm5^K6801/10424 flies using Bray-Curtis dissimilarity indices at the phylum level by the unweighted unifrac distances. D. Heatmap of normalized relative abundance levels of OTUs significantly changed between wt and kdm5^K6801/10424 flies grouped into 17 orders and colored at the phylum level (Wilcoxon rank sum test, P<0.05). E. Lactobacillus spp. loading was determined by plating 3-day old wt, kdm5¹⁰⁴²⁴, kdm5^K06801, kdm5^K6801/10424, and kdm5^k6801;gkdm5-HA fly intestine samples on De Man, Rogosa, and Sharpe (MRS) agar at 37°C. F. Gram-negative bacteria loading was determined by plating 4-day old wt, kdm5¹⁰⁴²⁴, kdm5^K06801, kdm5^K6801/10424, and kdm5^k6801;gkdm5-HA fly intestine samples on selective Gram-negative bacteria agar at 25°C. G. The ratio of gram-negative to gram-positive bacteria in intestinal samples of kdm5^K6801/10424 and wt flies. H. Q-PCR analysis showing the abundance level of L. piantarum L168 and P. rettgeri P6 in intestinal samples of kdm5^K6801/10424 and wt flies. I. The number of PH3 positive staining stem cells in mid-gut samples from kdm5^K6801/10424 and wt flies. J. The number of PH3 positive staining stem cells in mid-gut samples from Myo1A-Gai4^TS/kdm5^RNAi and wt flies. Data are shown as mean ± SEM. n ≥ 3 (A-G). 10–20 flies were examined each group (I and J). Data shown in (H-J) are representative of two independent experiments.*P <0.05; **P < 0.01 and ***P < 0.001.

Figure 3.. Gut bacterial and behavior analysis of GF wt and GF kdm5^K680110424 flies reconstituted with microbiome isolated from wt flies

A. Gram negative bacteria abundance in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies at 25°0. B. The ratio of gram-negative to gram-positive cells in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. C. Lactobacillus spp. bacteria abundance in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. D. Real-time PCR analysis of abundance of L. piantarnm L168 and P. rettgeri P6 in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. E. Social space in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. F. Social avoidance (performance index) in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. G. Direct social contacting time in 1 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. H. Gram negative bacteria abundance in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. I. The ratio of gram-negative to gram-positive cells in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. J. Lactobacillus spp. bacteria abundance in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. K. Real-time PCR analysis of abundance of L. plantarum L168 and P. rettgeri P6 in 4 day old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. L. Social space in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. M. Social avoidance (performance index) in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. N. Direct social contacting time in 4 days old GF wt and kdm5^K6801/10424 flies reconstituted with microbes from wt flies. Error bars represent min and max (E and L). Other Data are shown as mean ± SEM. n ≥ 3 (A-C, E, F,H-J, L and M). n≥8 (G and N). n=2 (D,K). *P < 0.05, **P < 0.01, ***P <0.001 and ****p < 0.0001. ns, not significant.

Figure 4. KDM5 manipulates the microbial composition to affect social behavior.

A. Social space in GF Myo1A-Gal4^TS/kdm5^RNAi and GF control flies. B. Pathogenic-like bacteria isolated from kdm5^K6801/10424 flies fed with antibiotic fortified food or standard food cultured on nutrient agar, mannitol agar and gram-negative bacteria agar at 25°C. C. Intestinal defects with bubbles in kdm5^K6801/10424 flies fed with antibiotic fortified food or standard food. D. Social space in kdm5^K6801/10424 flies fed with antibiotic fortified food or standard food. E. Social avoidance (performance index) in kdm5^K6801/10424 flies fed with antibiotic fortified food or standard food. F. Abundance of Lactobacillus spp. isolated from gut samples of kdm5^K6801/10424 flies maintained on food supplemented with Lactobacillus spp. or standard food. G. Intestinal permeability assay in kdm5^K6801/10424 flies maintained on food supplemented with L. plantarum L168 or standard food. H. Intestinal defects with bubbles in kdm5^K6801/10424 flies maintained on food supplemented with L. plantarum L168 or standard food. I. Social space in kdm5^K6801/10424 flies maintained on food supplemented with L. plantarum L168 or standard food. J. Social avoidance (performance index) in kdm5^K6801/10424 flies maintained on food supplemented with L. plantarum L168 or standard food. K. Social space in Myo1A-Gal4^TS/kdm5^RNAi flies maintained on food supplemented with L. plantarum L168 or standard food. L. Social avoidance (performance index) in Myo1A-Gal4^TS/kdm5^RNAi flies maintained on food supplemented with L. plantarum L168 or standard food. M. Social space in Myo1A-Gal4^TS/kdm5^RNAi flies treated with or without antibiotics. N. Social avoidance (performance index) in Myo1A-Gal4^TS/kdm5^RNAi flies treated with or without antibiotics. O. Concentration of 5-HT in abdomens samples from wt and kdm5^K6801/10424 flies. P. Concentration of 5-HT in abdomens samples from kdm5^K6801/10424 flies treated with or without antibiotics. Q. Concentration of 5-HT in abdomens samples from kdm5^K6801/10424 flies treated with L. plantarum L168 or standard food. R. Concentration of 5-HT in heads of Myo1A-Gal4^TS/kdm5^RNAi and control flies. S. Concentration of 5-HT in heads of Myo1A-Gal4^TS/kdm5^RNAi flies treated with L. plantarum L168 or standard food. Error bars represent min and max (A, D, I, K, M). Other data are shown as mean ± SEM; n ≥ 3 (A, B, D-F and I-S). 30–40 flies each group were examined (C and H). Data shown in C, G and H are representative of two independent experiments. *P <0.05, **P < 0.01 and *** P<0.001.

Figure 5. KDM5 negatively regulates IMD/Rel signaling.

A. Hierarchical clustering analysis of differentially expressed genes comparing wt and kdm5^K6801/10424 fly intestinal tissues. Up-regulated genes are shown in red; down-regulated genes in blue (P<0.05; fold-change ≥1.5). B. Gene ontology biological process enrichment analysis of up- and down-regulated genes comparing wt and kdm5^K6801/10424 fly intestinal tissues. C. Protein interaction network analysis of differentially expressed genes comparing wt and kdm5^K6801/10424 fly intestinal tissues. Node colors increasing from yellow to red indicate expression level changes in kdm5^K6801/10424 flies compared to wt. D. Expression levels of the antimicrobial and innate immune response genes in the intestinal tissues of kdm5^K6801/10424 flies relative to wt flies were confirmed by real-time PCR. E. Relative expression levels of antimicrobial genes in the intestinal tissues of kdm5^K6801/10424 female flies with L. plantarum L168 feeding or antibiotic treatment compared to that of kdm5^K6801/10424 respectively. F. PGRP-LC gene structure and transcriptional profiling of intestinal samples from wt and kdm5^K6801/10424 flies. G. ChIP-qPCR analysis of KDM5 levels at the promoter regions of imd, PGRP-LC-RA and PGRP-LC-RE genes in wt and kdm5^K6801/10424 flies intestine tissue. H. ChIP-qPCR analysis of H3K4me3 levels at promoter regions of imd (left panel), PGRP-LC-RA and PGRP-LC-RE in wt and kdm5^K6801/10424 flie intestine tissue (right panel). I. RT-PCR analysis of full-length PGRP-LC-RA and PGRP-LC-RE in S2 cells and intestinal tissue of kdm5^K6801/10424 and Myo1A-Gal4^TS/kdm5^RNAi flies. J. Immunofluorescence staining of Relish in intestinal tissue of wt and kdm5^K6801/10424 flies. K. ChIP-qPCR analysis of Relish levels at promoter regions of Relish target genes in wt and kdm5^K6801/10424 flies. IgG is included as a negative control (*P < 0.05). Data are shown as mean ± SEM; n ≥ 2 (A, D-I and K). *P < 0.05, **P < 0.01 and *** P < 0.001.

Figure 6. KDM5 demethylase activity regulates IMD/Rel pathway activity and gut microbiome-brain function

A. Expression levels of the antimicrobial and innate immune response genes of intestinal tissues from 3–5 day old kdm5^k6801;gkdm5 JmjC*-HA (abbreviation: kdm5^Jmjc*) relative to kdm5^k6801;gkdm5-HA flies. B. Lactobacillus spp. bacteria abundance in 3–5 day old kdm5^k6801;gkdm5-HA and kdm5^JmJC* mutant flies. C. Real-time PCR analysis of abundance of L. plantarum L168 and P. rettgeri P6 in kdm5^JmjC*and kdm5^k6801;gkdm5-HA flies. D. Number of PH3 positive staining cells in mid-gut from kdm5^JmjC* and kdm5^k6801;gkdm5-HA flies. 10–20 flies each group were examined at the 3–5 days. E. Intestinal permeability analysis for 3–5 day old kdm5^k6801;gkdm5-HA and kdm5^JmJC* flies was assessed (top-panel). Quantification of the percentage of body area with distribution of non-absorbable blue food dye relative to whole body (bottom-panel). F. The digestive tract of kdm5^JmjC* flies are highly permeable. FITC-labeled beads remained in the lumen of kdm5^k6801;gkdm5-HA flies. In contrast, FITC signals were diffuse in the gut of kdm5^JmjC*flies. G. Social space in kdm5^JmjC* and kdm5^k6801;gkdm5-HA flies. H. Social avoidance activity (performance index) in kdm5^JmjC* and kdm5^k6801;gkdm5-HA flies. I. Intestinal permeability in kdm5^JmjC*flies treated with or without L. plantarum L168. J. Social space in kdm5^JmjC* flies treated with or without L. plantarum L168. K. Social avoidance (performance index) in kdm5^JmjC* flies treated with or without L. plantarum L168. L. Concentration of 5-HT in abdomens tissues from kdm5^JmjC* and kdm5^k6801;gkdm5-HA flies. M. Concentration of 5-HT in abdomens tissues from kdm5^JmjC* flies treated with or without L. plantarum L168. Error bars represent min and max (G and J). Data are shown as mean ± SEM. n≥3 (B and G-M). n=2 (A and C). Data shown in D-F are representative of two independent experiments. *P < 0.05; **P < 0.01 and ***P < 0.001.

Figure 7. Inhibition of IMD signaling rescues gut dysbiosis and social behavior induced by loss of KDM5

A. The ratio of Gram-negative to Gram-positive bacteria in the gut of four groups. B. Real-time PCR analysis of abundance of L. plantarum L8168 and P. rettgeri P6 in gut tissue from four groups. C. Intestinal permeability in four groups was assessed (left panel). Quantification of the percentage of body area with distribution of non-absorbable blue food dye relative to whole body (right panel). D. Social space analysis in four groups. E. Social avoidance (performance index) in four groups. F. Direct social contacting time in four groups flies. G. Concentration of 5-HT in body from four groups flies. The genotypes of the four groups represented in Figures A-G are as follows: 1. Myo1A-Gal4^TS/+. 2. Myo1A-Gal4^TS/+;kdm5^RNAi/+. 3. Myo1A-Gal4^TS/+; kdm5^RNAi/imcf^RNAi. 4. Myo1A-Gal4^TS+;imd^RNAi/+. Error bars represent min and max (D). Other data are shown as mean ± SEM; n ≥ 3 (A, D-G). n=2 (B). Data shown in C are representative of two independent experiments. *P < 0.05; **P < 0.01 and ***P < 0.001.