Evidence for Natural Selection and Ethnic Diversity at the Adenylate cyclase 3 Gene Associated with Obesity and Type 2 Diabetes

Background: Diabetes and obesity cause serious complications worldwide, including stroke and cardiovascular disease, and are a global health burden. Diabetes is strongly related with obesity and both are signicantly heritable. The prevalence of diabetes and obesity are higher in African populations than in European and Asian populations. In human evolution, natural selection is a key process of genetic survival over generations. Thus, the selection for diabetes- and obesity-related genes is a key mechanism for survival during times of feast and famine. Loss-of-function variations in the adenylate cyclase 3 ( ADCY3 ) gene are associated with obesity and diabetes, while mutations in ADCY3 are also associated with childhood obesity. ADCY3 -decient mice showed severe obesity, impaired insulin sensitivity, and reduced physical activity. Here, we researched evidence for natural selection at ADCY3 . Methods: We used a three-step genetic method to identify natural selection at ADCY3 using data on four populations from the 1000 Genomes Project and HapMap: Utah residents with Northern and Western European ancestry (CEU), the Yoruba in Ibadan, Nigeria (YRI), Han Chinese in Beijing (CHB) and Japanese in Tokyo (JPT). First we used Wright’s F-statistics (Fst) as a measure of population differentiation to nd ethnic diversity at ADCY3 . We then used a long-range haplotype (LRH) test to nd signicant long haplotypes, and then the integrated haplotype score (iHS) to nd natural selection at ADCY3 . Results: We observed high Fst values and signicant ethnic diversity at four ADCY3 body mass index (BMI)-associated variations (rs7586879, rs6545814, rs11676272 and rs10182181) between the non-African and African populations. Both LRH and iHS also provided evidence for natural selection at ADCY3 . Conclusions: These observations show evidence for natural selection and ethnic diversity at ADCY3 . Further exploration into the evolution of obesity- and Type 2 diabetes-associated genes is needed. evolution of diabetes- and obesity-related genes.


Background
Diabetes and obesity cause metabolic syndrome and coronary artery disease in particular. Approximately 30% of US adults have metabolic syndrome (1). The increased prevalence of this syndrome worldwide has become a global health burden.
Natural selection, as de ned by Charles Darwin (2), is the key process of genetic survival over generations. Thus, selection for diabetes-and obesity-related genes is a key mechanism for survival during times of feast and famine (3). In adapting to historical environmental changes in food availability and lifestyle, there was evidence for natural selection of genes related to glucose (4,5), obesity and energy (6).
The prevalence of diabetes and obesity are higher in African populations than in European and Asian populations. In the 2013−2016 US National Health and Nutrition Examination Survey (NHANES) (https://www.cdc.gov/nchs/data/databriefs/db319.pdf), the prevalence of total diabetes among US adults was signi cantly higher in non-Hispanic black (17.9%) and Hispanic (19.8%) adults than in non-Hispanic white adults (12.4%). The prevalence of total diabetes among non-Hispanic Asians was 15.3%. In the 2015−2016 US NHANES (https://www.cdc.gov/nchs/data/databriefs/db288.pdf ), the prevalence of total obesity was signi cantly higher among non-Hispanic black (46.8%) and Hispanic (47.0%) adults than among non-Hispanic white (37.9%) and non-Hispanic Asian (12.7%) adults. Signi cantly different incident rates of diabetes and obesity can be observed among ethnic groups worldwide.
Melanocortin-3 receptor (MC3R) gene is also one of the obesity genes (6,10). Two MC3R childhood obesity-associated SNPs (rs3746619 and rs3827103) (6,10) also showed the highest risk frequencies (0.51 and 0.48) in the African population. This evidence may explain the differences in obesity prevalent among ethnicities.
ADCY3-de cient mice showed severe obesity, impaired insulin sensitivity and reduced physical activity (18).
A loss-of-function variant of ADCY3, c.2433-1G > A, showed an increased risk of obesity and Type 2 diabetes in a Greenlandic population (17). This variant was a splice site variant and disrupted a splice-acceptor in exon 14 (17). There were two novel isoforms of ADCY3 in these variant carriers. One isoform showed exon 14 skipped transcript and the other isoform retained the intron between exon 13 and exon 14 (17). Variant carriers showed decreased ADCY3 RNA expression and the expression data of ADCY3 showed severely decreased AA carriers, while the GA carriers showed an intermediate expression levels (17). The AA carriers had higher body fat, a larger waist circumference and had higher levels of glucose after an oral glucose tolerance test (17).
The common loss-of-function missense variant (Ser107Pro) of rs11676272 in ADCY3 showed an association with increased body mass index (BMI) in populations with European ancestry (8), East Asian populations (15) and in children from British and Dutch populations (16). In a functional study, the rs11676272 risk G allele was associated with reduced expression of ADCY3 (16). This missense variant G (Ser107Pro) of rs11676272 was reported to disrupt the interaction between the two-helix bundle of ADCY3 and inhibit adenylcyclase activity (16). The missense variant of rs11676272 was also reported to be associated with fat mass in childhood (16).
The rs7586879 SNP in ADCY3 was also signi cantly associated with BMI in populations with African ancestry (19).
Meta analysis also identi ed the rs6545814 SNP in ADCY3 associated with BMI in East Asian populations (15).
Meta analysis identi ed the rs10182181 SNP in ADCY3 associated with overweight and BMI (BMI ≥ 30 kg/m 2 ) in populations with European ancestry (20).
Loss-of-function mutations in ADCY3 are associated with childhood obesity in Pakistani and European populations (21).
In this study, we examined evidence for natural selection at the ADCY3 by a three-step genetic method, including Wright's F-statistics (Fst) (22), the long-range haplotype (LRH) test (23) and the integrated haplotype score (iHS) (24,25), using data on four population from the 1000 Genomes Project and HapMap (26,27): Utah residents with Northern and Western European ancestry (CEU), the Yoruba in Ibadan, Nigeria (YRI), Han Chinese in Beijing (CHB) and Japanese in Tokyo (JPT).
In the LRH (23), we used the complete data set of all chromosome 2 from HapMap release 24 (27) with sixty CEU, sixty YRI, and ninety East Asians from CHB and JPT.

Statistical methods
We carried out a 3-step genetic method with the Fst and LRH tests, and the iHS to identify natural selection at the ADCY3 in population data from the 1000 Genomes Project and HapMap (26,27).

Fst test
We examined four BMI-associated SNPs (rs7586879, rs6545814, rs11676272 and rs10182181) of ADCY3 (  (Table 1). We found signi cant ethnic diversity between non-African (European and/or Asian populations) and African populations at these four ADCY3 BMI-associated SNPs. The risk alleles for these four ADCY3 BMI-associated SNPs are all ancestral alleles (Supplementary Table 1).

LRH test
We examined the ADCY3 region and found six signi cant core haplotypes (REHH percentile, 99.6-99.9) in the YRI population ( Table 2). There were 74 haplotypes of this ADCY3 locus in the CEU population, 109 haplotypes of this locus in the YRI population and 68 haplotypes of this locus in the East Asian population. Interestingly, we observed that one core major haplotype with two BMI risk alleles (rs11676272G and rs10182181G) and the other core major haplotype with one BMI risk allele (rs7586879T) in the YRI population showed evidence for selection (REHH percentile, 99.9). We also found that the core major haplotype with one BMI-related SNP (rs753529G) in the YRI population showed evidence for selection (REHH percentile, 99.9). This rs753529 intron SNP was reported to be associated with BMI in Chinese obesity (32).  (Table 3). Fourteen SNPs had a positive score ([iHS] score, 2.51-3.62) in the YRI population (Table 3). These 14 SNPs include 11 intron SNPs, one silent mutation, and two 3′ UTR SNPs. These SNPs may be within a gene regulatory region in ADCY3. Interestingly, these SNPs include BMIassociated rs753529 SNP (32) with a high score of 2.86.

Discussion
The prevalence of diabetes and obesity are increasing worldwide. Signi cantly different incident rates of diabetes and obesity can be observed among ethnic groups worldwide. The prevalence of diabetes and obesity are higher in African populations than in European and Asian populations. Interestingly, frequencies of four body mass index-associated SNPs (rs7586879, rs6545814, rs11676272 and rs10182181) at ADCY3 showed the highest risk frequencies (0.90, 0.89, 0.94 and 0.94, respectively) in the African population. This evidence may explain the differences in obesity prevalent among ethnicities.
As de ned by Charles Darwin in On the Origin of Species (2), natural selection is a key process to ensure genetic survival over generations. In human history, signi cantly important genes may have been the targets for selection for surviving different food supplies and geographies over more than 10,000 years. These selected genes could present signi cant historical footprints in the current human genome.
One genetic selection analysis of 65 loci associated with type 2 diabetes showed that positive selection did not have a powerful in uence on the prevalence of type 2 diabetes risk alleles (33). One more genetic selection analysis of genes associated with type 2 diabetes and obesity presented only some evidence for selection at speci c loci (30). The other genetic selection studies showed natural selection at the genes associated with type 2 diabetes (25,34,35).
Using the genotypes of selected diabetes and obesity genes could be helpful for decisions about clinical and hospital medication for health care of patients with diabetes and obesity.
The ADCY3 gene maps to chromosome 2p23.3 and is expressed in broad tissues including adipocyte, hypothalamus, skeletal muscle, and pancreatic islet (16). Prevention of diet-induced obesity was observed in ADCY3-gain-of-function mice. The common loss-of-function variant (Ser107Pro) of rs11676272 (A to G) showed an association with increased BMI (9,15,16) and the rs11676272 risk G allele was associated with reduced expression of ADCY3 (16). The missense variant of rs11676272 was also reported to be associated with fat mass in childhood (16).
The rs7586879 SNP in ADCY3 was also signi cantly associated with BMI in populations with African ancestry (19). Meta analysis also identi ed the rs6545814 SNP in ADCY3 associated with BMI in East Asian populations (15). Meta analysis identi ed the rs10182181 SNP in ADCY3 associated with overweight and BMI (BMI ≥ 30 kg/m 2 ) in populations with European ancestry (20).
In this study, we detected high Fst CEU−YRI values for rs7586879:0.58, rs6545814:0.49, rs11676272:0.50 and rs10182181:0.49, and identi ed high Fst JPT+CHB−YRI values for rs11676272:0.49 and rs10182181:0.48 (Table 1). These four ADCY3 BMI-associated SNPs showed signi cant ethnic diversity between non-African (European and/or Asian populations) and African populations. Both LRH and iHS analyses showed natural selection at ADCY3 (Tables 2, 3). The observed one core major haplotype with two BMI risk alleles (rs11676272G and rs10182181G) and the other core major haplotype with one BMI risk allele (rs7586879 T) provided evidence for selection (both REHH percentile, 99.9, and p-value, 0.0003 and 0.003, respectively) in the African population ( Table 2). The prevalence of diabetes and obesity are higher in African populations than in European and Asian populations. The differences in risk allele frequencies caused signi cant differences in the genetic risk and prevalence of diabetes and obesity, which were higher in African populations than in European and Asian populations.
While complex, the ndings on important genes that underwent natural selection, genetic drift, or migration may be signi cant for clinical and hospital medications.

Conclusions
We detected natural selection and ethnic diversity at ADCY3 and these results are signi cant for clinical prescriptions of medicine for patients with obesity and diabetes. Future studies should further examine the evolution of diabetes-and obesity-related genes.

Declarations
Con ict of interest: None Competing interests The authors declare that they have no competing interests.

Funding
Not applicable.