| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Medical Genetics (A.R., J.K.) and Neurology (P.T.), Biomedicum Helsinki, Finnish Genome Center (A.R.), and Department of Public Health (J.G.E.), University of Helsinki, FIN-00014 Helsinki, Finland; Department of Epidemiology and Health Promotion (A.R., J.G.E., T.F., E.K.), National Public Health Institute, FIN-00300 Helsinki, Finland; Department of Biosciences and Nutrition and Clinical Research Center (J.K.), Karolinska Institutet, SE-14157 Huddinge, Sweden; Hospital for Children and Adolescents (S.A., H.S., E.K.), Helsinki University Central Hospital, FIN-00029 HUS Helsinki, Finland; and Medical Research Council Epidemiology Resource Centre (C.O.) and Developmental Origins of Health and Disease Centre (D.J.P.B., D.I.W.P.), University of Southampton, Southampton SO16 6YD, United Kingdom
Address all correspondence and requests for reprints to: Eero Kajantie, M.D., Ph.D., Department of Epidemiology and Health Promotion, National Public Health Institute, Mannerheimintie 166, FIN-00300 Helsinki, Finland. E-mail: eero.kajantie{at}helsinki.fi.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: Our objective was to assess whether haplotypes of the glucocorticoid receptor (GR) gene modify this link.
Design and Participants: We conducted a birth cohort study that included 437 men and women born in Helsinki, Finland, during 1924–1933, whose birth measurements were recorded.
Main Outcome Measures: We studied how the oral glucose tolerance test and fasting plasma total and free cortisol concentrations and, in a subset of 162 women, a more detailed HPAA evaluation, are predicted by body size at birth and haplotypes of the GR locus. We also measured the haplotype-specific relative mRNA expression level for the haplotype of interest.
Results: One of the haplotypes was associated with lower birth weight and length and higher fasting plasma and mean 24-h salivary cortisol. Moreover, this haplotype modified the association of length at birth with adult phenotypes; in carriers, short length at birth was associated with increased fasting plasma cortisol, cortisol/corticosteroid-binding globulin ratio, impaired glucose tolerance or diabetes [1 cm decrease corresponded to 1.36-fold odds ratio; 95% confidence interval (CI), 1.09–1.70; P = 0.007], and higher 120-min glucose (5.8%; 95% CI, 2.5–9.1%; P = 0.0007), but no association was seen in noncarriers (P for interaction was 0.06, 0.01, 0.02, and 0.01, respectively). The mRNA expression level of this haplotype was 93.7% (95% CI, 90.5–96.8%; P = 2.2 x 10–4) of the expression level of the other haplotypes.
Conclusions: A common GR haplotype may contribute to and modify the association of short length at birth with adult glucose tolerance and HPAA function by a mechanism that affects regulation of GR expression.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Glucocorticoids that are produced in response to signals from the HPAA have an important role in the regulation of glucose metabolism and fetal development. Glucocorticoids, including cortisol, mediate their cellular action by complexing with the cytoplasmic glucocorticoid receptor (GR), which then translocates to the nucleus and functions as a transcription factor. Therefore, GR is an attractive candidate gene to be involved in the link between HPAA function and early-life conditions resulting in later-life metabolic diseases.
Response to glucocorticoids varies greatly between individuals, which has in part been attributed to variation in the GR gene, although the exact causative polymorphisms have not been confirmed. Such polymorphisms could influence the amount of intrauterine glucocorticoid exposure on fetal growth and, perhaps, glucocorticoid metabolism later in life. The following three polymorphisms in the GR gene locus have been studied extensively: BclI restriction site and N363S polymorphisms have been associated with increased glucocorticoid sensitivity (16, 17) and consequently with unfavorable cardiovascular and metabolic profiles, whereas the ER22/23EK seems to show the opposite associations, including lower fasting insulin and serum total and low-density lipoprotein cholesterol concentrations in elderly subjects (18) and even lower all-cause mortality (19). The N363S variant has been associated with increased salivary cortisol responses to psychosocial stress (20) and in several studies with obesity (16, 21, 22), although these associations have been questioned by a number of negative reports (23, 24, 25). Similarly, the BclI polymorphism has been associated with abdominal obesity, higher systolic blood pressure, and higher cortisol levels (26), but also negative studies have been published (27, 28). BclI has also been associated with a lower cortisol response to psychosocial stress (20). Thus far, none of these polymorphisms has been associated with type 2 diabetes or impaired glucose tolerance (21, 22).
We are unaware of any earlier studies incorporating early-life factors in assessing the effects of genetic variations in the GR gene. We have examined within the Helsinki Birth Cohort, with detailed birth measurements available, whether the relationship of size at birth with adult HPAA function and glucose tolerance is dependent on different haplotypes of the GR gene. We focused on the haplotype structure, instead of individual rare polymorphisms, to consider major common variation within the GR gene locus.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Of these 437 subjects, 162 women had in addition taken part in a more detailed evaluation of HPAA function consisting of an overnight 0.25-mg dexamethasone test followed by a low-dose (1 µg) ACTH1–24 test (10) and a 24-h collection of salivary cortisol samples in their normal daily environment at awakening, 15 and 30 min thereafter, and at 1200, 1700, and 2200 h and the following morning (11). This relatively labor-intensive protocol allowed us to assess the association of the GR genotype with HPAA function in more detail, although the sample size was not sufficient for a meaningful analysis of interactive effects with size at birth. Compared with the rest of the women, those who took part in the more detailed HPAA examination had similar size and gestational age at birth (all P > 0.3) and similar height, body mass index (BMI), fasting cortisol, and CBG concentrations at examination, although they were slightly younger at the time of the baseline examination (69.3 vs. 70.0 yr, P = 0.05).
Genotyping
Fifteen single-nucleotide polymorphism (SNP) markers, located in the GR gene locus on chromosome 5q31, were selected from the public Ensembl human genome browser database (http://www.ensembl.org/Homosapiens/). The selected SNPs were evaluated by genotyping 91 individuals with a single-nucleotide primer extension (SNuPe)-based method. Uninformative markers (minor allele frequency < 0.05) were discarded and the rest of the samples were genotyped with six polymorphic SNPs. PCR assays were performed as described (32), followed by the allelic discrimination reactions by using the MegaBACE SNuPe genotyping kit (Amersham Biosciences, Piscataway, NJ) as described (33). All primers were designed by using the Primer3 program (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) and are available upon request. Genotypes followed Hardy-Weinberg equilibrium.
Haplotype construction
SNP rs6188 was in complete linkage disequilibrium with rs258813 and was therefore excluded. Haplotypes were inferred with SNPHAP software (http://www-gene.cimr.cam.ac.uk/clayton/software/snphap.txt/) by using the genotypes of the following five SNPs: rs6196, rs258813, rs4986593, rs852984, and rs6195 (Fig. 1
). Haplotypes with confidence values over 0.9 were accepted in the association analysis.
|
Log transformation was used to symmetrize the distributions of glucose, insulin, and salivary cortisol concentrations. Where variables were adjusted, e.g. for age, sex, and BMI, this was done by adding the mean value of the unadjusted variable to the residuals from a multiple regression analysis with age, sex, and BMI as the predictor variables. Interactions between a binary haplotype and a continuous predictor were tested by including in a regression analysis both terms and their product. The product term measures directly the difference in slope of the outcome variable on the continuous predictor in the two subgroups defined by the genotype. All analyses were performed with SPSS version 12.0.1.
Expression analysis
Because haplotype 3, which was consistently associated with the phenotypic data, does not include any known amino acid changing SNPs, we measured the relative haplotype-3-specific mRNA expression level. DNA samples were screened to find three heterozygous individuals for a silent coding SNP that tags haplotype 3 (rs6196). RNA of these three heterozygous individuals was extracted from frozen human thymus tissue by TRIzol (Invitrogen, Carlsbad, CA). The extracted RNA was further purified using the RNAeasy total RNA isolation kit (QIAGEN, Chatsworth, CA), followed by DNase (QIAGEN) treatment to eliminate genomic DNA contamination. First-strand cDNA was synthesized from 0.5 µg total RNA in a 25-µl reaction volume by RT using random hexamers (Promega, Madison, WI) and M-MLV reverse transcriptase (Promega). A transcript containing rs6196 in exon 9
was PCR amplified by using 2 µl cDNA as a template and forward primer in exon 7 (TGAAAACCTTACTGCTTCTCTCTTC) and reverse primer in exon 9 (CGACTTTCTTTAAGGCAACCA). This prevents the amplification of remaining genomic DNA because the two introns are too long to be amplified by PCR. After the quality check by agarose gel electrophoresis, RT-PCR products were purified enzymatically with Exo-SAPit (Amersham Biosciences) by incubating at 37 C for 60 min and heat inactivated at 72 C for 15 min. Genomic DNA of the same individuals was PCR amplified by using the forward primer in intron 8 (TGAGATGTTCCCACTGACCA) and the reverse primer shown above. Purified products were then sequenced using the forward primers. RT was performed twice for each three individual RNA samples, each cDNA sample was PCR amplified twice, and each RT-PCR product was sequenced twice, producing finally eight measurements for each individual RNA sample. Each genomic DNA sample was PCR amplified four times, followed by sequencing in duplicate, giving eight independent measurements altogether. Allele peak height ratio of allele C (tags haplotype 3) and allele T was measured, and values of genomic DNA were used as a reference value to compensate for the unequal amplification of the alternative alleles. The C/T allele peak height ratios of genomic DNA were compared with cDNA measurements by using linear regression where dummy variables for individual samples were used as covariates (SPSS version 12.0.1).
The study protocol was approved by the ethics committees of Helsinki University Central Hospital and the National Public Health Institute, and informed written consent was obtained from each participant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Four common haplotypes covered 96.5% of the observed haplotypes. Table 1 shows the distribution of the haplotypes numbered from 1–6 by decreasing frequency. The whole GR gene is located within a single linkage disequilibrium block, as assessed from the HapMap data (HapMap data release no. 19) using the Haploview software (Fig. 1
). We were able to match our haplotypes to those shown by HapMap and by Stevens et al. (34).
|
Mean birth weight of men was 3454 g (range, 1980–4620 g) and length at birth 50.3 cm (44.0–54.5 cm), and those of women were 3289 g (1770–5070 g) and 49.7 cm (43.0–57.0 cm). Haplotype 3 of the GR locus was associated with size at birth; carriers of this haplotype had lower birth weight and length at birth compared with noncarriers (Table 2). GR haplotype was unrelated to ponderal index, which is a measure of newborn thinness, and gestational age at birth.
|
GR gene haplotype 3 was weakly associated with increased fasting serum total cortisol concentrations and free cortisol index (Table 2) but not with CBG concentration. More detailed assessment of HPAA function performed in 162 women showed that haplotype 3 was associated with increased morning salivary cortisol (awakening response and a single measurement next morning) and mean salivary cortisol concentration during a 24-h follow-up and, by marginal significance, with increased cortisol concentration after 1 µg ACTH1–24 stimulation (Table 2). No relationship was seen with overnight 0.25 mg dexamethasone suppression.
Interactive effects with size at birth on HPAA function
Table 3 shows that the relationship between GR haplotype and cortisol concentrations was dependent on length at birth. Because of the small number of haplotype 3 homozygotes (22), we present combined data for all haplotype 3 carriers. Haplotype 3 was associated with high fasting total and free cortisol only in subjects who were short at birth. Conversely, short length at birth predicted high cortisol only in carriers of haplotype 3. In these subjects, a 1-cm decrease in length at birth corresponded to 15.2 [95% confidence interval (CI), 0.3–30.2] nmol/liter increase in total cortisol, whereas in noncarriers, no relationship was seen (1 cm corresponding to –1.5; 95% CI, –8.5 to 11.5 nmol/liter change). The P values for interaction between the effects of length at birth and haplotype 3 were 0.06 for total cortisol and 0.01 for free cortisol index.
|
Within the total study group, GR haplotype was unrelated to the prevalence of diabetes, impaired glucose tolerance, or glucose concentrations during an oral glucose tolerance test. Because we have previously shown in this cohort that small body size at birth predicts impaired/reduced glucose tolerance and type 2 diabetes (4), we assessed whether this relationship is different in subjects with different GR haplotypes (Table 4). In subjects with GR haplotype 3, 1 cm of decrease in length at birth was associated with a 1.36-fold odds ratio (95% CI, 1.09–1.70; P = 0.007) for developing impaired glucose tolerance or type 2 diabetes and 5.8% (95% CI, 2.5–9.1%; P = 0.0007) higher 120-min glucose at an oral glucose tolerance test, whereas in noncarriers of this haplotype, there was no association, corresponding odds ratio being 0.99 (95% CI, 0.85–1.16; P = 0.9) and change in 120-min glucose –0.2% (–3.2 to 2.8%; P = 0.9). Interaction between the effects of length at birth and haplotype 3 were significant for impaired glucose tolerance or type 2 diabetes (P = 0.02) and 120-min glucose (P = 0.01). Also, higher fasting glucose concentrations were associated with short length at birth in haplotype 3 carriers (P = 0.03) but not in noncarriers (P = 0.8) (P for interaction = 0.07).
|
As calculated from sequencing electropherogram (Fig 2), the average C/T allele peak height ratios were 0.831 (SD, 0.043) for genomic DNA and 0.778 (SD, 0.047) for cDNA samples. Therefore, the expression level of allele C, which tags haplotype 3, is 93.7% (95% CI, 90.5–96.8%; P = 2.2 x 10–4) of the expression level of allele T, which refers to the other haplotypes.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The search for genes underlying multifactorial diseases has been unrewarding. One of the major critiques concerning the large-scale genetic association studies has been their inability to incorporate environmental factors. The developmental origins of health and disease hypothesis proposes that environmental factors during early life cause lifelong alterations in the functions of different organs that have large impact on disease risk, i.e. susceptibility to adult diseases is programmed during early life (1, 35). Attempts have also been made to explain the association between small size at birth and adult disease by the genetic hypothesis suggesting that these phenotypes are merely two independent outcomes of the same genotype. Originally, this hypothesis was presented as the fetal insulin hypothesis, which proposes that genetic factors determine insulin resistance, which results in low birth weight (36). In the current study, we propose a synthesis of these hypotheses suggesting that an individual’s susceptibility to specific mechanism of programming is affected by his/her genetic make-up.
Our present findings add to our previous observations in the Helsinki Birth Cohort Study that have shown interactions between the effects of body size at birth and certain well-established genetic variants on late-life glucose tolerance and related phenotypes. These variants include Pro12Ala polymorphism of the peroxisome proliferator-activated receptor 
(PPAR-2) (4), K121Q polymorphism in plasma cell glycoprotein 1 (PC-1) (37), and the insertion/deletion polymorphism of the angiotensin-converting enzyme (ACE) (32) genes. These findings can be interpreted as examples of interactions between genotype and intrauterine environment with resulting changes in gene expression.
Previous studies of GR gene have focused primarily on the following three polymorphisms: N363S (rs6195), ER22/23EK (two SNPs in complete linkage disequilibrium, rs6189 and rs6190), and intronic BclI restriction site polymorphism formerly known as 2.3- and 4.5-kb alleles, but recently identified to constitute a C to G substitution in intron 2 (17). Locations for these polymorphisms are shown in Fig. 1A. Instead of these individual polymorphisms, we chose to study the haplotype structure of the GR gene to encompass a major proportion of common variation in the gene. Furthermore, we excluded very rare polymorphisms because they have potential to explain only a very small fraction of variation in response to glucocorticoids seen between individuals.
Rs6195 (N363S) resides in the haplotype (haplotype 5) that differs from the most common haplotype only by this one SNP. In our material, rs6195 did not associate with any measured phenotype (data not shown), although the power of the present study was not sufficient to assess the correlates of this relatively rare polymorphism. SNPs rs6189 and rs6190 (ER22/23EK) have been reported to be relatively rare (1.6%) (34). In our pilot genotyping of 91 individuals, they were monomorphic and were therefore discarded from the current study. According to a recent comprehensive analysis of haplotype structure of the GR locus by Stevens et al. (34), the haplotype including the ER22/23EK polymorphism differs from our haplotype 4 only by these two SNPs. Although these researchers used a largely different set of SNPs in their haplotype analysis, we were able to match our haplotypes to theirs (shown in Fig. 1B in addition to haplotypes according to HapMap). Our observed haplotype frequencies correspond well to those shown by Stevens et al. (34) and HapMap data (Fig. 1B). We did not genotype the BclI polymorphism, but all carriers of haplotype 2 and haplotype 3 have been reported to carry also the BclI 4.5-kb allele (34) (see Fig. 1B). We are unaware of any studies assessing the effect of the noncoding BclI polymorphism on gene expression. It is possible that this polymorphism and haplotype 3 in the present study may reflect the same causative variant.
The rs6196 (N766N) polymorphism, which tags our haplotype 3, has not been studied as extensively as the above mentioned three polymorphisms. It has, however, been associated with systemic lupus erythematosus in the Japanese population (38). In the study by Stevens et al. (34), the haplotype that corresponds to our haplotype 3 was associated with low post-dexamethasone cortisol. Although we did not observe this association, our results are difficult to compare because no other assessment of HPAA function was performed in their study (34). As Fig. 1 shows, haplotype 3 is the most diverged one from the most common haplotype 1, which makes it even more interesting as a risk haplotype.
Because haplotype 3 does not include any known coding nonsynonymous SNPs, we assumed that its phenotypic effects are most probably based on the altered allele-specific expression level of GR. Indeed, we found that the mRNA expression level of haplotype 3 in human thymus was consistently lower than that of the other haplotypes. The expression level of GR is critical for cell function; a 30–50% reduction of GR expression level has been shown to result in severe neuroendocrine, metabolic, and immunological abnormalities in transgenic mice (39). Our findings suggest that even a lesser reduction in the expression level could play a role in explaining normal phenotypic variation between individuals. That the association of body size at birth with adult glucose tolerance and HPAA function was confined to carriers of haplotype 3 implies an interaction between genotype and intrauterine environment. Whether this is explained by a more pronounced haplotype-specific decrease of GR expression level in individuals who were born small remains to be confirmed. The haplotype 3 tagging SNP (rs6196) resides in exon 9; therefore, our expression studies exclude the GR-ß splice variant, which is transcriptionally inactive but serves as a dominant-negative inhibitor of the GR- isoform (40).
Because the negative feedback mechanism of the HPAA is dependent on GRs, decreased GR expression could lead to higher cortisol levels as a result of impaired feedback control. Superficially, this may seem to be in contrast with our finding of no difference in dexamethasone-suppressed cortisol and a previous finding of augmented dexamethasone suppression (34) associated with haplotype 3. However, the dexamethasone suppression test is mostly an indicator of suppression at the pituitary level because dexamethasone does not pass freely through the blood-brain barrier (41). Reduced GR expression at the higher levels of the axis such as hypothalamus or hippocampus could as well lead to increased HPAA activity, although this remains speculative because specific assessment of negative feedback at these levels is difficult by any functional HPAA tests. Similarly, for obvious reasons, we were unable to measure the haplotype-3-specific GR expression in these sites. Although we cannot exclude tissue-specific differences, thymus, which we used, is expected to be as valid as any other tissue, because GR is ubiquitously expressed. Moreover, we found that other measures of HPAA function, such as cortisol after ACTH stimulation and 24-h salivary cortisol, were higher in haplotype 3 carriers than in noncarriers, indicating hyperactivity of the HPAA. Although the reduced GR expression in peripheral tissues could be expected to attenuate the effects of such hyperactivity, glucocorticoids are well known to induce insulin resistance in humans (42, 43), and impaired glucose tolerance and type 2 diabetes are key features of frank glucocorticoid excess caused by Cushing’s disease (44) or pharmacological treatment (43). Recent findings have suggested that normal individual variations in HPAA function are also associated with differences in glucose tolerance (9, 26, 45, 46), making variations in HPAA function, whether genetically or environmentally determined, plausible mechanisms to link early-life events with impaired glucose tolerance and type 2 diabetes in adult life.
We conclude that a common GR haplotype, which affects gene expression, modifies the association between short length at birth and HPAA function and glucose tolerance in adult life. The success of most large-scale genetic studies on type 2 diabetes and related traits has been limited. Our results illustrate the importance of incorporating both the genetic factors and determinants of early life in the studies aiming at understanding the etiology of glucose tolerance and related traits.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure statement: The authors have nothing to disclose.
First Published Online August 8, 2006
Abbreviations: BMI, Body mass index; CBG, corticosteroid-binding globulin; CI, confidence interval; GR, glucocorticoid receptor; HPAA, hypothalamic-pituitary-adrenal axis; SNP, single-nucleotide polymorphism.
Received May 17, 2006.
Accepted August 1, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. B. Piccoli, A. Conijn, V. Consiglio, E. Vasario, R. Attini, M. C. Deagostini, S. Bontempo, and T. Todros Pregnancy in Dialysis Patients: Is the Evidence Strong Enough to Lead Us to Change Our Counseling Policy? Clin. J. Am. Soc. Nephrol., January 1, 2010; 5(1): 62 - 71. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Schreiner, M. Tozakidou, R. Maslak, U. Holtkamp, M. Peter, B. Gohlke, and J. Woelfle Functional glucocorticoid receptor gene variants do not underlie the high variability of 17-hydroxyprogesterone screening values in healthy newborns Eur. J. Endocrinol., April 1, 2009; 160(4): 667 - 673. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Kajantie, K. Feldt, K. Raikkonen, D. I. W. Phillips, C. Osmond, K. Heinonen, A.-K. Pesonen, S. Andersson, D. J. P. Barker, and J. G. Eriksson Body Size at Birth Predicts Hypothalamic-Pituitary-Adrenal Axis Response to Psychosocial Stress at Age 60 to 70 Years J. Clin. Endocrinol. Metab., November 1, 2007; 92(11): 4094 - 4100. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
