Predicting novel candidate human obesity genes and their site of action by systematic functional screening in Drosophila

Abstract

The discovery of human obesity-associated genes can reveal new mechanisms to target for weight loss therapy. Genetic studies of obese individuals and the analysis of rare genetic variants can identify novel obesity-associated genes. However, establishing a functional relationship between these candidate genes and adiposity remains a significant challenge. We uncovered a large number of rare homozygous gene variants by exome sequencing of severely obese children, including those from consanguineous families. By assessing the function of these genes in vivo in Drosophila, we identified 4 genes, not previously linked to human obesity, that regulate adiposity (itpr, dachsous, calpA, and sdk). Dachsous is a transmembrane protein upstream of the Hippo signalling pathway. We found that 3 further members of the Hippo pathway, fat, four-jointed, and hippo, also regulate adiposity and that they act in neurons, rather than in adipose tissue (fat body). Screening Hippo pathway genes in larger human cohorts revealed rare variants in TAOK2 associated with human obesity. Knockdown of Drosophila tao increased adiposity in vivo demonstrating the strength of our approach in predicting novel human obesity genes and signalling pathways and their site of action.

Fig 1. Identification of genes harbouring homozygous rare variants in severely obese individuals.

(A) Schematic of the approach used to identify candidate human obesity genes and the prioritisation of genes containing rare exonic SNVs and indels for screening in Drosophila. (B) All probands had severe obesity and presented in early childhood, as exemplified in 3 family pedigrees shown. Males (squares) and females (circles) with obesity (filled symbols) are indicated. BMI in children under 18 years is adjusted for age and sex and shown as BMI SDS. Arrows indicate probands who were sequenced; slashed lines indicate deceased individuals; double lines indicate consanguineous union. Family numbers refer to S1 Table. (C) Cumulative medium to large ROHs in affected cases and prioritised homozygous exonic rare SNVs and indels. The autosomal karyotype heatmap illustrates the cumulative coverage of medium to large ROHs (>4 Mb) among the affected individuals, based on estimates of ROHs using WES data (displayed in bins of size 10⁵ bp). Markers show the location and consequence of homozygous rare exonic SNVs and indels within any ROHs (>100 kb). Stop codons and frameshifts are displayed, and missense variants where >1 affected person had a missense variant in the same gene. Variant markers are shown per gene (S2 Table). BMI, body mass index; indel, insertion/deletion; ROH, run of homozygosity; SDS, standard deviation score; SNV, single-nucleotide variant; WES, whole-exome sequencing.

Fig 2. Experimental design of the Drosophila screen to identify novel obesity genes.

(A) Schematic of the functional screen in Drosophila. (B, C, D) TAG levels normalised to level of protein upon ubiquitous adult specific knockdown of Drosophila orthologues of human genes from severely obese people. Each data point corresponds to an average of 3 technical replicates of TAG normalised to protein levels obtained from 10 male flies. Results obtained from multiple replicates are shown for knockdown of each gene. Error bars represent SEM; *p < 0.05, **p < 0.01 and ***p < 0.001 versus control by Mann–Whitney U Test. The underlying data for this figure can be found in S7 Table. Fig 2A was “created with BioRender.com.” TAG, triacylglyceride.

Fig 3. Dachsous, Fat, Four-jointed, and Hippo regulate Drosophila obesity.

(A) A schematic of Dachsous and Fat signalling pathways. (B) TAG levels normalised to level of protein upon ubiquitous knockdown of Drosophila dachsous, fat, or fj in adult flies. Each data point corresponds to an average of 3 technical replicates of TAG normalised to protein levels obtained from 10 male flies. Results obtained from multiple replicates are shown for knockdown of both genes. (C) Fly weight upon ubiquitous knockdown of Drosophila dachsous, fat, or fj. Each data point corresponds to the weight of 8 to 10 males normalised to number of flies. TAG levels normalised to level of protein in adult male flies upon adult specific (D) fat body and (E) pan-neuronal knockdown of dachsous, fat, and hpo. Adult-specific pan-neuronal knockdown of ds, fat, or hpo decreases food intake and excretion on fat and hpo knockdown (F, G), increases survival upon starvation after fat knockdown (H), and does not affect climbing ability (I). Results obtained from multiple replicates are shown for all genes. Error bars represent SEM; *p < 0.05, **p < 0.01 and ***p < 0.001 versus control by Mann–Whitney U Test. The underlying data for this figure can be found in S7 Table. Fig 3A was “created with BioRender.com.” TAG, triacylglyceride.

Fig 4. Rare variants in DCHS1, FAT4, and the Hippo pathway in severely obese people and healthy controls.

(A) Locations of very rare (MAF < 0.025%) nonsynonymous variants in DCHS1 or FAT4. Variants are shown in severely obese cases (n = 927 people; red, upper triangles) or INTERVAL controls (n = 4,057 people; blue, lower triangles). Filled triangles indicate VEP IMPACT = HIGH or SNV predicted damaging. A sliding window approach was used to perform burden tests (SKATBinary, method = “burden”) within localised regions (window size = 500 aa, shift = 200 aa); asterisk (*, black, R1-R2) indicates windows with burden test p-value < 0.05 after BH adjustment for multiple sliding windows; grey (R3-R4), nominal p-value < 0.05. (B) Cartoon illustrating DCHS1-FAT4 and the location of regions R1-R4, as (A). (C, D) TAOK2 sliding window analysis of very rare variants (MAF < 0.025%) (S4 Table). (E) TAG levels normalised to level of protein in adult male flies upon adult specific ubiquitous knockdown of Drosophila tao. Error bars represent SEM; ***p < 0.001 versus control by Mann–Whitney U Test. The underlying data for this figure can be found in S7 Table. aa, amino acid; BH, Benjamini–Hochberg; MAF, minor allele frequency; SNV, single-nucleotide variant; TAG, triacylglyceride.