Contribution of Variants in the Small Heterodimer Partner Gene to Birthweight, Adiposity, and Insulin Levels
Mutational Analysis and Association Studies in Multiple Populations
Mutational Analysis and Association Studies in Multiple Populations
Loss of function mutations in the small heterodimer partner (SHP) gene have been reported to cause obesity and increased birth weight. We examined the relation between genetic variation in SHP and birth weight, adiposity, and insulin levels in three independent populations. The coding regions and 562 bases of the SHP promoter were screened for mutations in 329 subjects with severe early-onset obesity. Two novel missense mutations, R34G and R36C, were identified; these were not found in control subjects and did not cosegregate with obesity in family studies. Two common polymorphisms, G171A and −195CTGAdel, were found in 12 and 16% of subjects, respectively. Within the obese cohort, G171A and −195CTGAdel carriers had higher and lower birth weights, respectively, than wild-type subjects, the rare homozygotes for G171A being particularly large at birth. In a U.K. population-based cohort of 1,079 children, the 171A allele was associated with higher BMI (P < 0.05) and waist circumference (P = 0.001). Children carrying the G171A variant had higher 30-min insulin responses to a glucose load (P = 0.03). In conclusion, although mutations in SHP are not a common cause of severe human obesity, genetic variation in the SHP locus may influence birth weight and have effects on BMI, possibly through effects on insulin secretion.
Small heterodimer partner (SHP) is a 257-amino acid orphan nuclear receptor expressed in the liver, pancreas, spleen, small intestine, and adrenal gland in humans (1). SHP inhibits the transcriptional activity of hepatocyte nuclear factor-4α (HNF-4α). Nishigori et al. (2) recently screened a cohort of Japanese patients with maturity onset diabetes of the young (MODY) and found seven mutations that result in impaired inhibition of HNF-4α transcriptional activity in vitro. Genotyping of probands’ family members failed to show segregation of SHP mutations with MODY, but mutations appeared to segregate with obesity in these families. Probands with SHP mutations had birth weights at least 1 SD higher than the mean birth weight adjusted for gestational age in population-based control subjects and showed hyperinsulinemia and decreased insulin sensitivity, suggesting a possible physiological mechanism for the observed effects on birth weight and adiposity.
We have studied SHP in three independent populations: 1) a group of 329 unrelated subjects with severe early-onset obesity in which we have undertaken mutation detection and association studies (Genetics Of Obesity Study [GOOS]); 2) a population-based cohort of 1,079 U.K. children in whom detailed anthropometric and metabolic data from birth are available (Avon Longitudinal Study of Parents and Children [ALSPAC]) (3); and 3) a population-based cohort of ∼600 U.K. Caucasian adults from Ely, Cambridgeshire, U.K., in whom detailed metabolic phenotypic data are available (Ely Study). In the two population-based cohorts (ALSPAC and Ely), we have conducted association studies of two common polymorphisms that were detected in our mutational scanning studies in the obese cohort (GOOS).
The human SHP gene, located on chromosome 1p36.1 consists of two exons separated by an intron of 1.8 kb (1). The 5′ region including ∼600 nucleotides upstream from the transcription start site has been characterized to possess promoter activity. Promoter and coding regions including intron/exon boundaries were screened using denaturing high-performance liquid chromatography (DHPLC) and patterns characterized by direct nucleotide sequencing.
Two novel missense mutations were found in unrelated subjects with severe early-onset obesity. One subject was heterozygous for a missense amino acid substitution, R34G, which was not detected in 100 Caucasian control subjects. However, in extended family studies this variant did not cosegregate with obesity, as a sibling who also inherited the mutant allele was lean. An unrelated subject was found to be heterozygous for a missense mutation, R36C, which was not present in 100 control subjects; however, family members were not available for genotyping.
Three common polymorphisms were identified in the promoter and coding regions. A G171A variant was found in heterozygous form in 34 (10.3%) subjects and in homozygous form in 4 (1.2%) subjects. A 4-bp (CTGA) deletion 195 bp upstream of the transcription start site (−195CTGAdel) was found in 52 (15.8%) subjects in heterozygous and in 2 (0.6%) subjects in homozygous form. Additionally, a C-394T variant, 394 bp upstream of the transcription start site, was found in three (0.9%) obese subjects. The three heterozygous C-394T carriers were homozygous for G171A.
We tested the association between these variants and indexes of birth weight, adiposity, and insulin levels in the three populations. Neither polymorphism deviated from Hardy-Weinberg predictions, and no linkage disequilibrium between the two polymorphisms was observed.
In the GOOS cohort the G171A variant was associated with higher birth weight, and the −195CTGAdel allele with a lower birth weight, compared with subjects who were wild-type at both sites (Table 1). When the data were expressed as birth weight standard deviation scores (SDSs), to correct for differences in gestational age and sex, only the effect of the −195CTGAdel polymorphism remained significant. In the ALSPAC cohort, neither variant was associated with birth weight or birth weight SDS (Table 2). Birth weight was not available and thus not studied in the Ely study. The four 171A homozygotes in the GOOS study had a markedly elevated birth weight (mean birth weight SDS = 1.27 vs. 0.21 in wild-type subjects) and the three homozygotes in the ALSPAC cohort were also among the heaviest in that group (mean birth weight SDS = 0.34 vs. −0.01 in wild-type subjects). It was also noted that two homozygotes for the −195CTGAdel variant in the GOOS cohort had a lower birth weight (mean birth weight SDS = −1.22 vs. 0.21 in wild-type subjects), as did the four homozygotes from the ALSPAC study (mean birth weight SDS = −0.31 vs. −0.01 in wild-type subjects).
In the GOOS study there was no association between either polymorphism and BMI SDS. In the ALSPAC study, G171A carriers had a greater BMI and waist circumference at 7 years (Table 2). In the Ely study, however, the BMI of female −195CTGAdel carriers was increased (Table 3). Two homozygotes for G171A in ALSPAC had a higher mean BMI at 7 years compared with wild-type subjects (mean BMI = 19.4 vs. 16.1 kg/m2 in wild-type subjects), and the two homozygotes for the −195CTGAdel variant had a lower mean BMI (mean BMI = 15.4 vs. 16.1 kg/m2 in wild-type subjects). Similarly, in the Ely study the lone male G171A homozygote had a higher BMI (27.7 kg/m2) and the lone male 195CTGAdel homozygote had a lower BMI (20.9 kg/m2) compared with wild-type subjects (mean BMI = 26.68 kg/m2).
In the GOOS study, −195CTGAdel carriers had a lower fasting plasma insulin concentration (Table 1). In the ALSPAC cohort, G171A carriers had higher fasting insulin levels and 30-min insulin response (Table 2). In contrast, in the Ely population no association between either polymorphism and fasting or post-glucose load insulin was detected (Table 3). Although mean plasma insulin in G171A carriers did not differ from wild-type, the four G171A homozygotes in the GOOS study had higher fasting plasma insulin levels (mean = 168.86 vs. 96.88 pmol/l in wild-type subjects). The sole G171A homozygote in the ALSPAC study had a higher fasting insulin level (91.2 pmol/l) compared with wild-type subjects (mean = 29.4 pmol/l).
Thus, we have identified two novel missense mutations and three common polymorphisms in SHP. Family data from the proband with the R34G mutation would suggest that this variant is unlikely to cause obesity in a simple Mendelian manner. However, R34G and R36C are highly conserved in mammalian species, and these rare variants may contribute in part to the obesity in the two probands.
Our studies suggest that SHP variants may have age-dependent influences on birth weight, adiposity, and insulin levels that are subtle in heterozygotes but may be more marked in homozygotes. In the two childhood cohorts, carriers of the A allele showed a trend toward increased birth weight, increased adiposity, and higher plasma insulin levels, while carriers of the minor allele of the −195CTGAdel promoter polymorphism showed the opposite trend. It is unsurprising that no effects of the variants on BMI were seen in the GOOS cohort, as these subjects were highly selected for extreme obesity, and other major gene effects are likely to obscure any subtle effects of the SHP polymorphisms. Although the numbers of homozygotes for either polymorphism were small, it is notable that their phenotypes were consistent with an exaggerated effect on all parameters. This is consistent with recent observations on the association of homozygosity for the 171A variant with increased BMI and birth weight in other independent cohorts (A. Hattersley, personal communication). In contrast, we were unable to discern any effect of the G171A variant on BMI or insulin levels in an adult population. This discordance may reflect the fact that the genetic determinants of adiposity and insulin levels in childhood are not identical to those operating in adult life.
The site and nature of the two variants do make it plausible that they might have a direct effect on the function and/or expression of SHP. The Glycine 171 is located within a highly conserved region, which has been shown to determine repressor function in vitro (4). When compared with DAX-1, a closely related nuclear receptor, the Glycine at position 171 in SHP corresponds to the Glycine at position 383 in DAX-1, and mutations in this region have been reported to result in congenital adrenal hyperplasia (5). The other common polymorphism results in a four-nucleotide deletion in a 5′ region known to confer promoter activity and therefore may have an effect on gene expression (1).
Nishigori et al. (2) have suggested a model whereby SHP might influence apparently diverse phenotypes such as birth weight, adiposity, and insulin levels. Thus, SHP appears to negatively regulate a number of transcription factors, such as HNF-4α, which is involved in pancreatic β-cell differentiation and function. In utero, insulin levels are a major determinant of birth weight as exemplified by patients with glucokinase mutations (6). Thus, reduction in SHP expression or function might be expected to increase HNF-4α activity during development, resulting in fetal hyperinsulinemia with consequent increases in weight at birth and in later life. Continued hyperinsulinemia in postnatal life may also continue to drive adiposity and result in secondary insulin resistance. Data from SHP knockout mice will be very helpful in testing this hypothesis (7,8).
In summary, mutations in SHP are not a common cause of severe early-onset obesity in Caucasians, although a contributory role for rare mutations in occasional patients cannot be excluded. Common variants in SHP do show some evidence for association with indexes of adiposity, birth weight, and insulin levels, particularly in children; however, these effects are likely to be modest, and large-scale population genetic studies are needed to test these observations further.
RESEARCH DESIGN AND METHODS
Subjects with severe obesity of early onset (<10 years of age) have been recruited to GOOS. Ninety percent of the cohort is U.K. Caucasian. BMI (weight in kilograms/height in meters squared) SDS were calculated using 1990 growth reference data (9). The mean BMI SDS of probands is 4.2 ± 0.8.
The Ely Study.
The Ely Study is a prospective population-based cohort study of the etiology and pathogenesis of type 2 diabetes and related metabolic disorders in U.K. subjects aged 40–65 years (10,11). Glucose and insulin measurements fasting and post-oral glucose load (standard 75-g oral glucose tolerance test) were available.
ALSPAC is a prospective study of 14,541 pregnancies recruited from Bristol, U.K., between April 1991 and December 1992 (12). Altogether, 1,079 children were invited for anthropometry at age 7 years and for an oral glucose tolerance test with insulin measurement at age 8 years. The corrected insulin response was calculated from 30-min insulin and glucose levels after OGTT as a correlate of first-phase pancreatic insulin secretion (13).
PCR and sequencing.
Two primers, SHPFor (5′-AGG AAT CAG GCT GGG GAT AAG GA-3′) and SHPRev (5′-GCC AGG CTG AAT CAG CAC TGC CA-3′), were used in a PCR to amplify the promoter region and two exons from genomic DNA isolated from whole blood using a QIAamp blood kit (Qiagen, London). PCR was carried out under standard conditions, and products were sequenced using BigDye terminator chemistry (Perkin-Elmer, Foster City, CA) and analyzed on an ABI 377 automated DNA sequencer (Perkin-Elmer) (http://diabetes.diabetesjournals.org). DHPLC methods and genotyping using restriction fragment-length polymorphism are described in the online appendix.
GOOS, Ely, and ALSPAC data were compared using the unpaired Student’s t test. Significance level P < 0.05 was considered significant.
This work was supported by the Wellcome Trust (to I.S.F., N.J.W., and S.O.R.) and the Medical Research Council (MRC) (to S.O.R. and N.J.W.). The Ely Study was funded by Diabetes U.K. and the Anglia and Oxford Research Development Directorate, and ALSPAC is supported by the MRC, the Wellcome Trust, and the Department of Health.
We thank Jonathan Achermann for helpful discussions, and we are grateful to the subjects, referring physicians, and the staff of the ALSPAC and Ely study research clinics.
Address correspondence and reprint requests to Stephen O’Rahilly, University Department of Clinical Biochemistry, Addenbrooke’s Hospital, Cambridge, CB2 2QQ, U.K. E-mail:.
Received for publication 15 November 2002 and accepted in revised form 30 January 2003.
C.C.H. and I.S.F. contributed equally to this work.
Additional information can be found in an online appendix at http://diabetes.diabetesjournals.org.
ALSPAC, Avon Longitudinal Study of Parents and Children; DHPLC, denaturing high-performance liquid chromatography; GOOS, Genetics Of Obesity Study; HNF-4α, hepatocyte nuclear factor-4α; MODY, maturity onset diabetes of the young; SDS, standard deviation score; SHP, small heterodimer partner.