Butyrate rescued CPF-induced social deficit in zebrafish.

(A) Chlorpyrifos (CPF) induced social deficits through embryonic exposure in a dose-dependent manner. CPF’s effects were significant from 10-17.5 µM and peaked at 15 µM. (B) Selected gut microbial metabolites and liver metabolites were tested at 10 µM for their abilities to rescue social deficits. Colored dots represent the social scores of individual fish. Sodium butyrate effectively rescued social deficits induced by 15 µM CPF. UDCA: ursodeoxycholic acid; IPA: indole-3-propionic acid; CDCA: chenodeoxycholic acid. (C) Sodium butyrate (SB) rescued social deficits induced by 15 µM CPF in a dose-dependent manner. Significant rescue was observed at 10 µM. Significance was calculated by one-way ANOVA and Dunnett’s multiple comparison test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Class I HDAC inhibition phenocopied butyrate’s rescue effect.

(A) Valproic acid (VPA) (100 µM) but not trichostatin A (TSA) (200 nM) or nicotinamide (NAM) (100 µM) rescued social deficits induced by 15 µM CPF. VPA is an inhibitor of classes I and IIa HDACs, TSA is an inhibitor of classes I, II, and IV HDACs, and NAM is an inhibitor of class III (sirtuins) HDACs. Significance was calculated by one-way ANOVA and Dunnett’s multiple comparison test. (B) CRISPR-Cas9 induced F0 knockout (KO) of zebrafish class I HDACs (HDAC-I) rescued social deficits induced by 15 µM CPF. This experiment shares the same DMSO and CPF controls with the HDAC1 KO experiment shown in Figure 3A. Significance was calculated by two-way ANOVA with Fisher’s LSD test for post hoc analysis. (C) CRISPR-Cas9 induced F0 KO of zebrafish class IIa HDACs (HDAC-IIa) failed to rescue social deficits induced by 15 µM CPF. Significance was calculated by two-way ANOVA with Fisher’s LSD test for post hoc analysis. ns: not significant, *p<0.05, ****p<0.0001.

Selective inhibition of HDAC1 effectively rescued CPF-induced social deficit.

(A) CRISPR-Cas9 induced F0 KO of the zebrafish hdac1 gene robustly rescued social deficits induced by 15 µM CPF (CPF+hdac1 KO). Social behavior in DMSO control was also significantly boosted by hdac1 KO (DMSO+hdac1 KO). This experiment shares the same DMSO and CPF controls with the HDAC-I KO experiment shown in Figure 2B. Significance was calculated by two-way ANOVA with Fisher’s LSD test for post hoc analysis. (B) Knocking out the zebrafish hdac3 gene modestly rescued social deficits induced by 15 µM CPF (CPF+hdac3 KO). Significance was calculated by two-way ANOVA with Fisher’s LSD test for post hoc analysis. (C) Knocking out the zebrafish hdac8 gene rescued social deficits induced by 15 µM CPF (CPF+hdac8 KO). Significance was calculated by two-way ANOVA with Fisher’s LSD test for post hoc analysis. (D-F) Dose dependent rescue of CPF-induced (15 µM CPF) social deficits following overnight exposures of BRD-6929, a selective inhibitor of HDAC1 (HDAC1i) (D), RGFP966, a selective inhibitor of HDAC3 (HDAC3i) (E), and PCI-34051, a selective inhibitor of HDAC8 (HDAC8i) (F). Red dot marks the DMSO control of each experiment. The HDAC1 inhibitor BRD-6929 robustly rescued CPF-induced social deficit in a dose-responsive manner (D). The HDAC3 inhibitor RGFP966 modestly rescued social deficit in a dose-responsive manner (E). The HDAC8 inhibitor PCI-34051 failed to rescue CPF-induced social deficit. Significance was calculated by one-way ANOVA and Dunnett’s multiple comparison test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Butyrate partially rescued sustained downregulation of neuronal genes induced by early exposure to CPF.

(A) Gene Set Enrichment Analysis (GSEA) of RNA-seq data using Gene Ontology’s (GO) Biological Process (BP) terms identified 13 pathways (red arrows) related to neuronal projection, synaptogenesis, and learning among the top 20 downregulated pathways in CPF-treated samples as compared to the DMSO control samples. Significantly downregulated pathways (p < 0.05) are ranked by Normalized Enrichment Score (NES). (B) Hub genes were identified by network topology analysis of a protein-protein interaction network formed by protein products of significantly downregulated neuronal genes found in the CPF samples compared to the control samples. All genes belong to the 13 neuronal pathways shown in (A). The top 30 hub genes were ranked by node degree values. Colors represent hub gene rankings, with red denoting the top-ranking genes. The size of each colored circle represents the degree centrality (number of connections in the network). Blue underlines mark the SFARI ASD risk genes. (C) Gene expression heatmap shows butyrate (SB) partially reversing the downregulation of 30 neuronal hub genes induced by CPF-exposure (CPF+SB vs. CPF). (D) Comparing the average normalized expressions of 30 neuronal hub genes. Neuronal hub genes are significantly downregulated in CPF:No-SB samples compared to the DMSO:No-SB controls. Butyrate (SB) significantly upregulated the expression of neuronal hub genes in CPF-treated fish (CPF:SB), although not enough to match the level of expression in DMSO:No-SB control samples. Gene expression levels are not significantly different in CPF:SB samples as compared to DMSO:SB samples. Lines between treatment groups connect expression levels of the same gene. Significance was calculated by two-way ANOVA with Tukey’s multiple comparison test. ns: not significant, *p<0.05, **p<0.01,***p<0.001, ****p<0.0001.

Early CPF exposure induced sustained overexpression of key circadian genes.

(A) Gene Set Enrichment Analysis (GSEA) of RNA-seq data using Gene Ontology’s (GO) Biological Process (BP) terms. Showing the top 20 upregulated pathways in CPF-treated samples as compared to the DMSO control samples, among which 5 are related to circadian regulation (green arrows). Significantly upregulated pathways (p < 0.05) are ranked by Normalized Enrichment Score (NES). (B) Volcano plot showing the RNA-seq result comparing gene expression in CPF-treated samples and DMSO control samples. 11 of the top 21 significantly upregulated genes (ranked by adjusted p value) are circadian genes. (C) Quantitative PCR analyses conducted using samples from an independent experimental replicate validated RNA-seq results by confirming upregulation of key circadian genes following embryonic exposure to CPF. CPF-treated fish and DMSO control fish were dissected at the exact same time during the day by two experimenters working side-by-side. Significance was calculated by two-tailed Student’s t test. *p< 0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Early CPF exposure selectively impacted nitrogen metabolism-related pathways in the juvenile zebrafish gut.

(A-C) Enrichment analyses detected changes in gut metabolomic pathways in CPF-treated fish compared to DMSO control fish, based on untargeted metabolomics analysis of gastrointestinal tissues and gut contents. Surveying the RaMP-DB (A), SMPDB (B), and KEGG (C) databases consistently identified urea cycle, arginine metabolism, and aspartate metabolism among the top enriched pathways (red arrows). Both arginine and aspartate are intermediate metabolites of the urea cycle. Changes in nitric oxide (NO) metabolism and nitric oxide synthase (NOS) activity were detected by surveying RaMP-DB (A). Color coding represents levels of significance (p value). The size of each dot represents enrichment ratio. (D) Network view of the enrichment analysis result in (A). L-arginine is a precursor for the endogenous production of NO through NOS, thereby connecting arginine metabolism, aspartate metabolism, and urea cycle with the NO/NOS pathway. (E) Enrichment analysis comparing metabolomic changes between CPF+SB and CPF samples based on untargeted metabolomics analysis of gastrointestinal tissues and gut contents. Pathways related to the urea cycle, arginine metabolism, and aspartate metabolism (red arrows) were identified by surveying SMPDB.

Early CPF exposure led to enrichment of denitrifying bacteria in the juvenile zebrafish gut microbiome.

(A) Gut microbial taxonomic structure of the DMSO samples. Kraken generated read counts were classified at each level of the taxonomic tree and summed as we go up the taxonomy. (B) Gut microbial taxonomic structure of the CPF samples. Red arrow points to the genus Pseudomonas which is increased in abundance in the CPF samples as compared to the DMSO control samples. (C) Number of taxa that are differentially abundant at each taxonomic level. (D) Selected counts of differentially abundant species by genus. (E) β coefficient estimates and significance for the 10 species with the largest beta values (log fold change) and the 10 species with the smallest beta values. β coefficients are calculated for the test between CPF and DMSO samples. Error bars represent the confidence interval of each coefficient. Color denotes false discovery rate (FDR). There are a total of 139 significant species, only the 20 with the most extreme beta values are shown in this plot. (F) Nitrite concentration was significantly elevated in 6 dpf zebrafish larvae following embryonic (0-3 dpf) treatment of CPF. n=85 larvae per condition. Significance was calculated by two-tailed Student’s t test. **p<0.01.

A working hypothesis for CPF’s neurodevelopmental toxicity.

(A) A schematic demonstration of the NO-HDAC hypothesis. Chlorpyrifos (CPF) exposure increases the abundance of the denitrifying bacteria Pseudomonas in the gut microbiome, enhancing the production of nitric oxide (NO) in the gut through the denitrification pathway. CPF is also known to stimulate NO production by activating eNOS through acetylcholine (Ach) accumulation and promoting iNOS expression. Elevated NO levels result in the selective inhibition of HDAC8 through S-nitrosylation. HDAC1 and HDAC3 are resistant to S-nitrosylation and thus escape inhibition by NO. This selective inhibition skews the balance of class I HDAC activity in the developing zebrafish brain, altering histone acetylation patterns in a way that suppresses the expression of neuronal genes critical for social behavior, thereby leading to social behavioral deficits. Other biological effects of NO production include the downregulation of NO-sensitive bacteria of the genus Aeromonas and the upregulation of circadian genes, particularly the Per and Cry family genes. Butyrate inhibits all members of class I HDACs, including the zebrafish HDAC1, HDAC3, and HDAC8, thereby resetting the genome-wide balance of histone acetylation at gene loci targeted by class I HDACs. SNO: S-nitrosothiols. (B) Embryonic exposure (0-3 dpf) to the NO donor SNAP (10 µM) induced social deficits in Fishbook assay. Significance was calculated by two-tailed Student’s t test. *p< 0.05. (C) Embryonic exposure (0-3 dpf) to the non-selective PDE inhibitor IBMX (100 µM) did not induce measurable social deficits in Fishbook assay. Significance was calculated by two-tailed Student’s t test. ns: not significant.