Natural variation in protein kinase D modifies alcohol sensitivity in Caenorhabditis elegans

Abstract

Differences in naïve alcohol sensitivity between individuals are a strong predictor of later life alcohol use disorders (AUD). However, the genetic bases for alcohol sensitivity (beyond ethanol metabolism) and pharmacological approaches to modulate alcohol sensitivity remain poorly understood. We used a high-throughput behavioral screen to measure acute behavioral sensitivity to alcohol, a model of intoxication, in a genetically diverse set of over 150 wild strains of the nematode Caenorhabditis elegans. We performed a genome-wide association study to identify loci that underlie natural variation in alcohol sensitivity. We identified five quantitative trait loci (QTL) and further show that variants in the C. elegans ortholog of protein kinase D, dkf-2, likely underlie the chromosome V QTL. We found that resistance to intoxication was conferred by dkf-2 loss-of-function mutations as well as partly by a PKD inhibitor in a dkf-2-dependent manner. Protein kinase D might represent a conserved, druggable target to modify alcohol sensitivity with application towards AUD.

Keywords: GWAS; alcohol sensitivity; protein kinase D.

Figure 1:. GWAS of natural variation in alcohol sensitivity of 152 C. elegans wild strains.

(A) Overview of high-throughput behavioral screen for naïve alcohol sensitivity in 152 C. elegans wild isolates. In each assay, basal egg-laying rates were measured using ten clonal adults, and strain average phenotypes were calculated as the average phenotype score across ~10 replicates for each genotype. Distribution of alcohol intoxication sensitivity in the population is depicted in top right corner. (B) Manhattan plot of GWAS for alcohol sensitivity. Each panel represents a chromosome (I-V & X); each point represents an SNV marker. X-axis represents genomic position (Mb), and y-axis denotes significance (−log₁₀(p)). The black dotted line denotes genome-wide significance cutoff after correcting for the number of independent association tests. Highlighted with a vertical red box is the ChrV QTL that was pursued for fine mapping.

Figure 2:. Fine mapping chromosome V QTL associated with decreased sensitivity.

(A) Each point in boxplots represents the average alcohol sensitivity of a single strain. Y-axis denotes alcohol intoxication sensitivity, and genotype at SNV is shown along the X-axis (REF or ALT). Average phenotypes of all strains were segregated by their genotype at the denoted allele. Top graph shows strains segregated by genotype at the genome-wide significant SNV identified by GWAS. Middle and bottom graphs show phenotypes segregated by genotype at the two high impact mutation candidate variants identified by fine mapping. (B) Results from fine mapping. Each line represents an SNV in the significant interval (V:16,638,047–18,330,166) from GWAS. Y-axis denotes significance. Color fill of each SNV represents predicted variant effect, where red denotes high-impact variants, gray represents low-impact variants, and light gray represents intergenic/linker variants. (C) Left shows the genotype of six strains at the two high-impact variants predicted in each candidate gene nearest the GWAS peak marker, dkf-2, and immt-2. In a small subset of strains (left; pink and blue multicolored bars), candidate variants in dkf-2 were not in strong LD with candidate variants in immt-2. (Right) Average intoxication sensitivity of each genotype arranged from least to most sensitive (left) (mean & SEM; n=10–15). Color denotes genotype shown in left panel, where red denotes ALT-ALT, orange denotes ALT-REF, and pale pink denotes REF-REF N2 lab strain.

Figure 3:. Independent deletion alleles in the highly conserved C. elegans orthologue of PKD (dkf-2) predictably reduce alcohol sensitivity.

(A) Protein kinase D is functionally conserved from humans to C. elegans. The human gene products (PKD1, PKD2, PKD3) are compared to one isoform of the C. elegans ortholog (DKF-2-A). Orthologous functional domains are labeled. (B) Known interactions of PKD. PKDs are most often downstream effectors of DAG-PKC signaling cascades. Activated PKDs mediate a wide array of cellular processes e.g., chromatin dynamics, cell growth and proliferation, innate immunity and inflammatory response, general stress tolerance and Golgi vesiculation. Recent studies show that PKD can activate via PKC-independent autophosphorylation of the conserved ULD domain (C) Mean alcohol sensitivities (SEM) of three dkf-2 deletion alleles (vx40, ok1704, and tm4076) were significantly lower than intoxication sensitivity of dkf-2 lab wild-type (N2) animals (mean & SEM; n=15; p=0.0001; p=0.0104; p=0.0167). (D) Deletion mutants in a wild isolate (NIC1) harboring candidate hypomorph allele(s) in dkf-2 do not cause a reduction in alcohol sensitivity (mean & SEM; n=18; p=0.35; p=0.1573), whereas deletions in a strain without the candidate hypomorph alleles cause significant reductions in alcohol sensitivity (mean & SEM, n=18; p=0.0001; p=0.0052). (E) Treatment with a PKD inhibitor (CID 755673) trends towards lower alcohol sensitivity in the N2 lab strain (mean & SEM, n= 18–20, two-sided t-test p=0.07, Cohen’s d: 0.584 -medium effect) but shows no effect on alcohol sensitivity in a dkf-2 deletion mutant (RB1468(ok1704)). (F) Strains that are hemizygous for ALT genotype at dkf-2 are resistant to alcohol, while strains that are hemizygous for the REF genotype at dkf-2 are sensitive to alcohol (mean & SEM; n=6; p=0.01). NIC1 (ALT) and PS2025 (REF) were mated to a dkf-2 deletion mutant (tm4076). F₁ progeny from these crosses were assayed for their alcohol sensitivity. Alcohol sensitivity was higher in PS2025 (REF)xdkf-2(−/−) F₁’s relative to NIC1(ALT)xdkf-2(−/−) F₁’s.