Variability of the Insulin Receptor Substrate-1, Hepatocyte Nuclear Factor-1α (HNF-1α), HNF-4α, and HNF-6 Genes and Size at Birth in a Population-Based Sample of Young Danish Subjects¹

REDUCED SIZE at birth has been proposed to be a risk factor for insulin resistance and type 2 diabetes (1–4). Initially, this association was suggested to be due to unfavorable intrauterine environment, making low birth weight only a marker of an yet unidentified intrauterine factor associated with increased risk of type 2 diabetes later in life. This hypothesis challenged the current idea of the existence of type 2 diabetes susceptibility genotypes for the development of the disease. However, an alternative explanation might be a coincidence of specific susceptibility genotypes for both reduced fetal growth and type 2 diabetes.

Recently, it was observed that mutations in the glucokinase gene associated with the type 2 form of maturity-onset diabetes of the young (MODY) result in a reduced birth weight, most likely caused by changes in fetal insulin secretion (5). Additionally, polymorphisms of the variable number of tandem repeats locus that influence insulin gene transcription have been shown to be related to altered birth weight (6), suggesting that genetic factors might be involved in intrauterine growth retardation. However, the molecular mechanism behind abnormal fetal growth is unknown, but might occur as a result of alterations in the insulin secretion or insulin signaling pathways. Such alterations could be a result of variability of genes known to be involved in the pathogenesis of type 2 diabetes, insulin resistance, impaired insulin secretion, or growth retardation. In this study we focused on previously identified variants in the insulin receptor substrate 1 (IRS-1), hepatocyte nuclear factor-1α (HNF-1α), HNF-4α, and HNF-6 genes in relation to birth weight and length and ponderal index.

Disruption of the IRS-1 gene causes a 50% reduction in intrauterine growth in mice (7, 8). Furthermore, we have recently demonstrated that a common Gly/Arg⁹⁷² polymorphism of the IRS-1 gene interacts with obesity to decrease insulin sensitivity in humans (9). A crucial role for the HNF-1α in controlling embryonic development was suggested by analysis of HNF-1α null mice, which were smaller at birth and were born at less than half the expected frequency compared to their wild-type littermates (10). Mutations in HNF-1α are responsible for the type 3 form of MODY, and three common amino acid polymorphisms were identified in this gene, Ile/Leu²⁷, Ala/Val⁹⁸, and Ser/Asn⁴⁸⁷ (11), of which the Ala/Val⁹⁸ variant was associated with insulin secretory dysfunction, as estimated during an oral glucose tolerance test (OGTT) (12, 13). Also, the MODY1 gene, HNF-4α, in which a Thr/Ile¹³⁰ was identified (14), plays an important part in early development, as mice lacking HNF-4α die early in embryogenesis (15). HNF-6 is a related transcription factor, which stimulates transcription of the HNF-4α gene (16). Therefore, this gene might also be of importance during fetal life, although we do not yet have indications from studies of knockout mice. A common Pro/Ala⁷⁵ variant was recently described in this gene (17).

The purpose of the present study was to analyze whether these amino acid polymorphisms are associated with decreased birth weight and length or ponderal index in a population-based sample of young Danish Caucasian subjects.

Subjects and Methods

Subjects

The study population comprised 380 subjects randomly recruited from a population of young Caucasians, aged 18–32 yr (18). The original birth length and birth weight data of 331 of 380 subjects were obtained from the midwife records stored in the Danish Provincial Archives for Zealand, Lolland/Falster, and Bornholm Islands. The measurements of birth length and weight were taken to nearest 1 cm and 50 g, respectively. Ponderal index was calculated as (birth weight)/(birth length)³. Physiological and anthropometric characteristics of this population sample have been presented previously (18).

Before participation, informed consent was obtained from all subjects. The study was approved by the ethical committee of Copenhagen and was performed in accordance with the principles of the Declaration of Helsinki II.

Screening for polymorphisms in the IRS-1, HNF-1α, HNF-4α, and HNF-6 genes

The assays for detection of the genetic variants have been described in detail previously: Gly/Arg⁹⁷² of IRS-1 (9); Ile/Leu²⁷, Ala/Val⁹⁸, and Ser/Asn⁴⁸⁷ of HNF-1α (11); Thr/Ile¹³⁰ of HNF-4α (14), and Pro/Ala⁷⁵ of HNF-6 (17).

Statistics

Differences in continuous variables between groups of subjects were tested with a generalized model, including genotype and gender as fixed variables. Subgroup analyses of the Ser/Asn⁴⁸⁷ variant were conducted using the generalized model, in which the genotype was included as a dichotomous variable (homozygous carriers vs. heterozygous/wild-type carriers). Data represent means (sd). The Statistical Package of Social Science (SPSS, Inc., Chicago, IL) for Windows (version 9.0) was used for statistical analysis.

Results and Discussion

Birth weight and length and ponderal index data of the present study population stratified in accordance with the genotype of Gly/Arg⁹⁷² of IRS-1, Thr/Ile¹³⁰ of HNF-4α, Pro/Ala⁷⁵ of HNF-6, and the Ile/Leu²⁷, Ala/Val⁹⁸, and Ser/Asn⁴⁸⁷ of the HNF-1α, respectively, are given in Table 1. Using a generalized linear model in which adjustments for gender were made, no differences in birth length, birth weight, or the ponderal index between groups of genotypes were obtained with respect to Gly/Arg⁹⁷² of IRS-1, Thr/Ile¹³⁰ of HNF-4α, Pro/Ala⁷⁵ of HNF-6, and Ile/Leu²⁷ or Ala/Val⁹⁸ of HNF-1α (Table 1). There was a significant difference (P = 0.04) in ponderal index across the three genotype groups of the Ser/Asn⁴⁸⁷ variant of HNF-1α. No differences in birth weight or length between these groups were observed. In this respect it should be noted that the ponderal index might be a more specific indicator of fetal malnutrition than low birth weight (19). Subgroup analysis revealed that Asn/Asn⁴⁸⁷ HNF-1α carriers had a 5% reduction in ponderal index (P = 0.02) compared to heterozygous and wild-type carriers. The power of the present study to detect this difference is actually more than 80%. However, as repeated analyses of three continuous variables on seven gene variants have been performed, we applied Bonferroni correction for multiple testing, which turned down the significant difference.

Previous studies have shown associations between low birth weight and impaired β-cell function (20). However, recent analyses in the present study population could not confirm this association (21). Nevertheless, the present study sample is young of age (mean age, 25 yr), and a potential influence of fetal growth retardation on insulin secretion might be undetectable until later in life. Recently, we observed in two independent study populations that carriers of the Ala/Val⁹⁸ variant of the HNF-1α had a 20% impairment in β-cell function as estimated during an oral glucose tolerance test (12, 13). However, the present study failed to show that this association is mediated through estimates of fetal growth retardation. A tentative difference in birth weight or ponderal index of 20% between carriers of the variant and wild-type carriers would have been detected with a power of more than 99%, thereby excluding the risk of false negative results.

Previous studies have demonstrated a direct relationship between impaired fetal growth and insulin resistance in adult life (3, 22). Interestingly, recent analyses in the present study population showed a positive association between low birth weight, but not low ponderal index, and low values of the insulin sensitivity index (21). Previously, we also demonstrated in this study group that obese carriers of the Gly/Arg⁹⁷² variant of IRS-1 gene had a 50% reduction in insulin sensitivity index compared to obese wild-type carriers. Furthermore, this variant was in 32D(IR) cell lines observed to cause a 32% decrease in thymidine incorporation into DNA, suggesting a role of the Gly/Arg⁹⁷² variant on mitogenesis (23). However, the negative results of the present study suggest that these effects of the Gly/Arg⁹⁷² variant are explained by postnatal events.

The fact that the present study failed to demonstrate any relationship between variability in four candidate genes for type 2 diabetes and low birth weight might have several explanations. Although there is accumulating evidence that low birth weight predisposes to the development of type 2 diabetes in adult life, it is unclear whether this association is causal, whether low birth weight per se is only a marker of an intrauterine or maternal factor associated with an increased risk of type 2 diabetes, or whether the association is explained by gene variants predisposing to both low birth weight and type 2 diabetes. There is no evidence of a major influence of any of the examined variants on type 2 diabetes susceptibility. Thus, only minor effects of these variants on size at birth would be expected, and despite the fact that this study is favored by very exact determinations of birth size, these effects might be undetectable. Furthermore, it is important to note that although the growth of a fetus obviously is influenced by its genes, the relative contributions of genes to the variation in size at birth are unknown, and several nongenetic maternal and fetal factors also contribute to the determination of size at birth (24–27).

From the present study, which elucidates the effect of amino acid polymorphisms of the IRS-1, HNF-1α, HNF-4α, and HNF-6 on human fetal growth, we conclude that common variability in these genes can be excluded as a major factor influencing size at birth among Danish Caucasian subjects.

We thank Annemette Forman, Lene Aabo, Dorte GøthJohansen, Bente Mottlau, Susanne Kjellberg, Lis Ølholm, and Maja Lis Halkjær for dedicated and careful technical assistance and Grete Lademann for secretarial support.