Serum Resistin and Polymorphisms of Androgen Receptor CAGn and GGNn and Aromatase TTTAn
There is evidence that androgens are regulators of insulin resistance (IR), and may be involved in the regulation of resistin, a cytokine that has been related with IR. Earlier studies found that androgen receptor length polymorphisms CAGn and GGNn and the aromatase polymorphism TTTAn may influence receptor or enzyme activity and serum concentrations of androgens. This study was designed to determine whether polymorphism length was related to serum resistin concentration and to other variables related with IR. In 1,580 persons chosen randomly from the general population of the Canary Islands (Spain), we measured polymorphism length, waist circumference, waist/hip ratio, BMI, and serum glucose concentration. In smaller subgroups, we also measured C-peptide (n = 677), resistin (n = 583), and leptin concentration (n = 754) and estimated IR (homeostasis model assessment-IR (HOMA2-IR)). In men, polymorphism length correlated with resistin concentration (CAGn, r = 0.13, P = 0.031; TTTAn, r = 0.15, P = 0.005; GGNn, r = −0.15, P = 0.026), and the correlations were confirmed in multivariate regression models. The length of CAGn and TTTAn correlated inversely with C-peptide (r = −0.13, P = 0.016 and r = −0.21, P < 0.001, respectively) and with estimated IR (r = −0.12, P = 0.032 and r = −0.19, P = 0.001, respectively). In men, length of the CAGn, GGNn, and TTTAn was associated with serum resistin concentration. These results support the hypothesis that androgens may be involved in the regulation of resistin. Resistin may be a link between IR and androgens.
In addition to the important role of sex hormones in reproduction, there is evidence that they also influence glucose metabolism. Clinical and epidemiological findings have related androgen levels with insulin resistance (IR). Several studies have reported that serum testosterone concentration in men correlates inversely with insulin concentration (1,2). It has also been found that low testosterone concentrations increase the risk of type 2 diabetes (3,4,5), and that the administration of androgens in hypogonadal men appears to improve insulin sensitivity (6).
In addition to sex hormones, certain cytokines such as leptin, adiponectin, and resistin have been associated with the appearance of IR and the maintenance of metabolic homeostasis (7). Resistin, a cytokine belonging to the family of small cystine-rich proteins, is produced mainly by adipocytes in mice (8) and by macrophages and monocytes in humans (9). Initially, resistin was reported to antagonize the action of insulin in rodents, and resistin levels were found to be increased in diabetic and obese mice (8). These early findings suggested that resistin might be a link between obesity and IR. However, later studies in rats and humans have yielded contradictory findings (10,11,12,13,14,15,16), and the role of resistin in the appearance of IR remains controversial.
Although the factors involved in resistin regulation are poorly understood, some studies have indicated that sex hormones may be involved (17,18,19). Two studies in rodents suggested that androgens can increase resistin expression in adipose tissue (17,20). Two other studies reported a direct correlation between serum concentrations of testosterone and resistin in women with polycystic ovary syndrome (21,22).
Androgens exert their action mainly by binding to the androgen receptor (AR), which is encoded by a single gene located on the X chromosome. Exon 1 of the gene contains two trinucleotide repeat polymorphisms, CAGn and GGNn, which encode a stretch of glutamine and glycine residues, respectively. A number of studies have shown that the number of repeats in these polymorphisms, located in the transactivation domain, correlates inversely with activity or concentration of AR (23,24,25,26). Some studies have also noted that these polymorphisms may influence serum androgen concentrations (27,28,29).
A key enzyme in sex steroid hormone metabolism is aromatase, which catalyses the conversion of androgens (testosterone and androstenedione) into estrogens (estradiol and estrone) in several tissues and organs. Intron 4 of the aromatase gene (CYP19) contains a tetranucleotide repeat polymorphism TTTAn. A number of studies have related polymorphism length of CYP19 with concentrations of sex steroid hormones (30,31).
To date, only two studies have analyzed the possible influence of the CAGn polymorphism on glucose metabolism. The first study, in healthy men, found a direct correlation between the number of CAGn repeats and concentrations of leptin and insulin (32). The second study, in women with polycystic ovary syndrome, found that CAGn repeat length modified the impact of testosterone on IR (33). We are unaware of earlier studies about the possible influence of these polymorphisms on serum resistin concentration. Accordingly, we designed this study to investigate whether the number of repeats in the polymorphic sequence was related with resistin concentrations or with other biochemical and anthropometric variables related with IR, such as serum concentrations of glucose, C-peptide and leptin, homeostasis model assessment-IR (HOMA2-IR) index, waist circumference, BMI, and waist-to-hip ratio.
Methods and Procedures
Participants in this cross-sectional, descriptive study were the first 1,580 individuals enrolled in the “CDC de Canarias,” a cohort study that recruited >6,700 individuals from 2000 to 2005. Some of the findings for this cohort have been reported earlier (34,35).
Participants, between 18 and 75 years old, were recruited from the general population of the Canary Islands (Spain). All participants were chosen randomly from the census of health-care system affiliates, because public health care covers 99% of the population of the Canary Islands. All participants in the “CDC de Canarias” study were sent a letter informing them about the aims of the research and inviting them to take part, and the final participation rate was 70%. Consent was obtained from each participant after complete explanation of the purpose and nature of all procedures used. The research was approved by the ethics committee of La Candelaria Hospital.
Anthropometric data were recorded for different numbers of participants as detailed in Table 1. Hip circumference was measured by placing a flexible tape measure around the patient's hips at the level of the prominences over the greater trochanters of both femurs. Waist circumference was measured by placing a flexible tape measure around the patient's waist at the level of the navel and halfway between the lower rib margin and the iliac crests. BMI was calculated by dividing weight by height in meters squared, and the quotient was expressed as kg/m2. Waist-to-hip ratio was calculated as an indicator of abdominal obesity.
Table 1. General characteristics of population sample by gender
Serum hormones and glucose
After informed consent had been obtained, fasting samples of venous blood were drawn and serum aliquots were stored at −80 °C. Serum glucose concentrations were recorded for all participants within 24 h after the blood was obtained; measurements were made with a Hitachi 917 autoanalyzer and expressed as mmol/l. One aliquot of frozen serum was used to determine C-peptide, leptin, and resistin concentration. Because of limited financial resources for this research, the concentration of these hormones was measured in smaller subgroups rather than the entire sample of participants. All determinations were done consecutively beginning with the first of the 1,580 participants to obtain as many analyses as funds permitted. Enzyme-linked immunosorbent assays were used to determine C-peptide (Biosource, ng/ml, intra-assay coefficient of variation 6.3%, interassay coefficient of variation 4.7%), resistin (Bio-Vendor, ng/ml, intra-assay coefficient of variation 7.0%, interassay coefficient of variation 7.2%), and leptin concentrations (Biosource, ng/ml, intra-assay coefficient of variation 3.6%, interassay coefficient of variation 6.8%). IR was estimated from baseline serum glucose and C-peptide concentrations using a computer-based HOMA system (HOMA2-IR) provided by the Oxford Centre for Diabetes Endocrinology and Metabolism (http:www.dtu.ox.ac.ukhoma).
Analysis of CAG, GGN, and TTTA polymorphisms
DNA was extracted from blood samples (200 μl) using High Pure PCR Template Preparation Kit (Roche Applied Science). Fragments of exon 1 of the AR gene containing either the CAGn or GGNn repeats and intron 4 of the aromatase gene containing TTTAn were amplified with a pair of primers (one of which was labeled with a fluorescent dye) as previously described (36,37,38). PCR amplifications were performed in a 25-μl reaction volume containing ∼100 ng of genomic DNA, 200 μmol/l of each deoxyribonucleotide triphosphate, 1× Fast Start Taq DNA polymerase buffer (Roche Applied Science), 1× GC-rich buffer solution (GG-rich buffer was not used for TTTAn polymorphism amplification), 1 U Fast Start Taq DNA polymerase, 2.5 mmol/l MgCl2, and 1.2 μmol/l of each primer. The conditions for PCR amplifications were initial denaturation at 95 °C for 5 min followed by 30 cycles of 95 °C for 45 s, 56 °C for 30 s and 72 °C for 30 for CAGn and TTTAn, and 30 cycles of 95 °C for 1 min, 55 °C for 2 min and 72 °C for 2 min for GGNn. A final extension was performed at 72 °C for 5 min. The PCR product was diluted 1:100 in distilled water and 1 μl of the dilution was mixed with 10 μl formamide and 0.3 μl GeneScan 500 LIZ Size Standard (Applied Biosystems), denatured at 95 °C for 3 min and cooled on ice. Fragments were separated by automated capillary electrophoresis using an ABI Prism 3100 Genetic Analyzer (Applied Biosystems), and repeat length was determined with Gene Scan Analysis Software (version 3.7) (Applied Biosystems). Genotyping analyses for these polymorphisms were repeated with blind DNA samples and the results were 100% coincident. Repeats number was confirmed by sequencing several samples harboring alleles of different sizes for these polymorphisms with the Big Dye Terminator Sequencing Kit (Applied Biosystems).
The overall characteristics of the sample are reported as the mean ± s.d. Correlations were estimated as the nonparametric Spearman's correlation coefficient (ro). Between-group comparisons (all two tailed) were performed with student t-test. Bivariate associations were corroborated with multivariate linear regression models. All analyses were done with the Statistical Package for Social Sciences (version 13.0 for Windows).
The general characteristics of men and women who participated in the study are summarized in Table 1. All variables differed significantly between men and women except for serum resistin concentration. Women were older than men and had a higher mean BMI, serum C-peptide and leptin concentrations, and HOMA2-IR index. Men had a higher mean waist, waist-to-hip ratio, and serum glucose concentration.
For the AR polymorphism CAGn, the population studied here showed 23 different alleles ranging from 10 to 36 repeats. Distribution of the alleles approached normality, and the most frequent alleles were those with 21 (18%), 19 (13%), and 24 repeats (13%). For the AR polymorphism GGNn, we found 19 different alleles ranging from 12 to 30 repeats. The most frequent alleles were those displaying 23 (54%) and 24 (26%) repeats. For the aromatase receptor polymorphism TTTAn, we observed 8 alleles ranging from 5 to 13 repeats; the most frequent alleles were those with 7 (56%) and 11 (32%) repeats.
Table 2 shows the results of bivariate correlation analysis in men and women with polymorphism length and serum concentrations of glucose, C-peptide, leptin and resistin, HOMA2-IR index, and each of the three anthropometric indexes. In men, the number of CAGn repeats in the AR gene correlated directly with serum resistin, and inversely with serum C-peptide and HOMA2-IR index. The number of TTTAn repeats in the aromatase gene showed the same pattern, correlating directly with serum resistin and inversely with serum C-peptide and HOMA2-IR index. In contrast, the number of GGNn repeats in the AR correlated inversely with serum resistin.
Table 2. Spearman correlation coefficients for repeat-length polymorphisms and anthropometric and biochemical measures
In women, the number of GGNn repeats correlated inversely (although the correlation just failed to reach statistical significance) with serum resistin and the number of CAGn repeats showed a significant inverse correlation with serum glucose concentration.
Of the anthropometric indexes we studied, only waist-to-hip ratio in women correlated weakly with the number of TTTAn repeats; this correlation approached statistical significance in men. No correlations were found for any of the polymorphisms and any of the other parameters. Waist-to-hip ratio was the only index that correlated significantly with resistin in men and in women.
These associations between polymorphism length and resistin concentration found in the bivariate analysis were corroborated with multivariate linear regression analysis, with resistin as the dependent variable and age, waist-to-hip ratio, serum C-peptide concentration, and each polymorphism as explanatory variables (variables that showed an association with resistin concentration). In men, CAGn y GGNn repeats remained in the model (P = 0.030 and P = 0.032, respectively) (Table 3). In contrast, the correlations for TTTAn repeats were no longer statistically significant (P = 0.070). When IR (HOMA2-IR) was included in the multivariate models as an independent variable, instead of C-peptide, both AR and aromatase polymorphism length were associated with serum resistin (Table 3).
Table 3. Standardized β-coefficients of multivariate regression models in men
The rest of the simple correlations between polymorphism repeat length and other variables such as C-peptide and HOMA2-IR index in men or glucose and waist-to-hip ratio in women were not corroborated in the multivariate models.
We found an association between serum resistin concentration and length of the AR CAGn, GGNn, and aromatase TTTAn polymorphisms. On a broader scale, there appears to be a relationship between these polymorphisms and glucose metabolism. This relationship was evident in men, in whom we observed correlations between polymorphism length and variables such as serum C-peptide and resistin concentrations and HOMA2-IR index, but was less evident in women. Multivariate analyses corroborated the associations between polymorphism length and serum resistin in men, although the associations with serum C-peptide concentration and HOMA2-IR index could not be confirmed. Our findings support the hypothesis of Nogueiras et al. (39) that resistin may be a link between androgens and IR.
The lack of significant associations in women may be attributable to differences between the sexes in sex hormone concentrations, function and metabolism, particularly with regard to androgens. In addition, the location of the AR gene on an X chromosome may have biased the results toward the null in women: because the relative contribution of each allele is not known, we used the average number of repeats for both.
Two earlier studies in rodents support the theory that androgens are able to regulate the expression of resistin in adipose tissue (17,20). One study found an increase in resistin expression in mice with elevated androgen concentrations, whereas the other reported a decrease in resistin mRNA levels in orchidectomized mice. Two other studies, however, found that resistin increased androgen production by the human ovary and testosterone secretion by the rat testicle (22,39). A direct correlation has also been reported between testosterone and serum resistin concentrations in women with polycystic ovary syndrome (21,22). Our findings are consistent with earlier research that reported regulation of serum resistin levels by androgens, although some authors have suggested that the two may exert a reciprocal regulatory effect (39).
A recent study in baboons (40) and two other studies in humans found that genetic background had a strong influence on resistin levels (41,42). The contribution of hereditary factors to variations in resistin levels were estimated as 70% in the former species and 66% in humans. According to these findings, variation in the resistin gene and other genes unrelated with IR pathways might influence the concentrations of this cytokine. In view of the apparent relationship between androgens and resistin, it is plausible that variations such as those we documented in genes directly related with androgen activity and metabolism might influence resistin concentrations.
The action of androgens is mediated by AR, which directly regulates the transcription of several genes. A number of studies have shown that CAGn polymorphism length influences receptor activity (24,26), which decreases as the number of sequence repeats increases. Some studies have reported an inverse correlation between the number of repeats of the CAGn polymorphism and serum androgen concentration in premenopausal (27) and postmenopausal women (28). Other studies, in contrast, found no relationship between the number of repeats and the concentrations of androgens or testosterone (43,44).
The GGNn polymorphism has been studied to a less extent than the CAGn polymorphism. One of the earliest studies found an inverse correlation between the number of GGNn sequence repeats and concentrations of AR protein (25). A more recent study reported nonlinear variations in AR function depending on repeat length (45). A direct association has been also found between GGNn length and the risk of developing cryptorchidism and penile hypospadias, both conditions considered consequences of low androgenicity (46).
Some studies of the aromatase polymorphism TTTAn reported an association between repeat length and concentrations of steroid hormones, although there is as yet no consensus regarding how these results should be interpreted (31,47). Other studies, in contrast, failed to find any relationship between these variables (48,49). To our knowledge, the only functional study published to date found increased aromatase activity in fibroblasts from individuals who carried the high-repeat genotype alleles compared with low-repeat alleles (30).
The results of our study contrast with those of Zitzmann et al. (32), who found a direct correlation between CAGn repeat length and the concentrations of insulin and leptin in men. However, we found an inverse correlation between repeat length and serum C-peptide concentration and were unable to document any correlation between CAGn repeat length and leptin concentration. These discrepancies between the two studies may reflect differences in population characteristics, because our data were from >300 men who had been sampled from the general population, whereas Zitzmann et al. studied a sample of 110 men in which diabetes, hypertension, and atherosclerosis (among other diseases) were the exclusion criteria.
Our results suggest that men with higher repeat alleles for the AR CAGn and the aromatase TTTAn polymorphisms are more likely to have higher serum concentrations of resistin, lower concentrations of C-peptide, and IR. As noted above, functional studies indicate that higher repeat alleles for CAGn and TTTAn may be translated as reduced AR and aromatase activity. Does this mean that men with reduced AR or aromatase activity have higher serum levels of resistin and lower IR? This would seem to contradict findings that androgen concentrations correlate directly with resistin levels, and with reports that low androgen concentrations in men facilitate IR. However, reduced AR activity secondary to longer repeat lengths could coexist with increased serum concentrations of androgens. In this connection, a recent study in men found that CAGn repeat length is positively associated with serum total testosterone (29). Although our results are internally consistent, the associations we found are hard to interpret in the light of earlier research, given that to date the relationships between androgens, resistin, and IR have not been firmly established, and the relationships between serum androgen concentrations and repeat length of the AR and aromatase polymorphisms we studied also remain unclear.
In conclusion, the number of repeats in the AR CAGn and GGNn and aromatase TTTAn polymorphisms is related with serum resistin concentration. This relationship is consistent with the findings of other studies which reported a relationship between resistin and androgens. Our results are compatible with the hypothesis that androgens regulate resistin levels, and reinforce the suspicion that resistin may act as a link between androgens and IR.
This study was supported by the grant 66/04 from FUNCIS (Fundación Canaria de Investigación y Salud, Canary Islands Research and Health Foundation). We thank Marta Batista Medina for her help with the serum samples and K. Shashok for translating parts of the original manuscript into English.
The authors declared no conflict of interest.