Variation in the Gene for Muscle-Specific AMP Deaminase Is Associated With Insulin Clearance, a Highly Heritable Trait

  1. Mark O. Goodarzi12,
  2. Kent D. Taylor1,
  3. Xiuqing Guo1,
  4. Manuel J. Quiñones3,
  5. Jinrui Cui1,
  6. Xiaohui Li1,
  7. Tieu Hang1,
  8. Huiying Yang1,
  9. Edward Holmes4,
  10. Willa A. Hsueh3,
  11. Jerrold Olefsky5 and
  12. Jerome I. Rotter1
  1. 1Medical Genetics Institute, Departments of Pediatrics and Medicine, Steven Spielberg Pediatric Research Center, Cedars-Sinai Medical Center, Los Angeles, California
  2. 2Division of Endocrinology, Diabetes, and Metabolism, Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, California
  3. 3Division of Endocrinology, Diabetes, and Hypertension, Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles, California
  4. 4Department of Medicine, University of San Diego School of Medicine, San Diego, California
  5. 5Division of Endocrinology, Diabetes, and Hypertension, Department of Medicine, University of San Diego School of Medicine, San Diego, California
  1. Address correspondence and reprint requests to Mark O. Goodarzi, MD, PhD, Cedars-Sinai Medical Center, Division of Endocrinology, Diabetes and Metabolism, 8700 Beverly Blvd., Becker B-128, Los Angeles, CA 90048. E-mail: mark.goodarzi{at}cshs.org

Abstract

The rising prevalence of the insulin resistance syndrome in our society necessitates a better understanding of the genetic determinants of all aspects of insulin action and metabolism. We evaluated the heritability of insulin sensitivity and the metabolic clearance rate of insulin (MCRI) as quantified by the euglycemic-hyperinsulinemic clamp in 403 Mexican Americans. We tested the candidate gene AMP deaminase 1 (AMPD1) for association with insulin-related traits because it codes for an enzyme that has the potential to influence multiple aspects of insulin pharmacodynamics. By converting AMP to inosine monophosphate, AMPD1 plays a major role in regulating cellular AMP levels; AMP activates AMP kinase, an enzyme that modulates cellular energy and insulin action. We determined that nine AMPD1 single nucleotide polymorphisms (SNPs) defined two haplotype blocks. Insulin clearance was found to have a higher heritability (h2 = 0.58) than fasting insulin (h2 = 0.38) or insulin sensitivity (h2 = 0.44). The MCRI was associated with AMPD1 SNPs and haplotypes. Insulin clearance is a highly heritable trait, and specific haplotypes within the AMPD1 gene, which encodes a skeletal muscle−specific protein, are associated with variation in insulin clearance. We postulated that the processes of insulin action and insulin clearance in skeletal muscle are highly regulated and that AMPD1 function may play an important role in these phenomena.

The insulin resistance syndrome (also called the metabolic syndrome) is a clustering of factors associated with an increased risk of coronary artery disease (1). In the U.S., >20% of adults are affected by it (2). Mexican Americans have a high prevalence of hyperinsulinemia and insulin resistance as well as the highest age-specific prevalence of the insulin resistance syndrome (24). Thus, by studying a large family-based sample of Mexican-American subjects, we sought to elucidate the genetic determinants of insulin metabolism and action.

In pursuit of this goal, we evaluated insulin phenotypes using the euglycemic-hyperinsulinemic clamp study. Assessment of the glucose infusion rate (M) during the euglycemic-hyperinsulinemic clamp is regarded as the most direct physiological measurement of insulin sensitivity (5,6). The clamp also allows calculation of the metabolic clearance rate of insulin (MCRI). In this family-based study, we assessed the variation of these traits within families and observed for the first time a high heritability for MCRI. To begin to identify specific genes that mediate heritability of insulin-related phenotypes, we selected a candidate gene, AMP deaminase 1 (AMPD1), which has been shown to influence skeletal muscle adenine nucleotide levels (7) and which in turn could have effects on other enzymes, such as AMP-activated protein kinase (AMPK), that are known to influence insulin action (8). We show herein that polymorphisms in AMPD1 are associated with variation in the MCRI.

RESEARCH DESIGN AND METHODS

The University of California at Los Angeles/Cedars-Sinai Mexican-American Coronary Artery Disease Project enrolls families ascertained through a proband with coronary artery disease, as determined by evidence of myocardial infarction on an electrocardiogram or in a hospital record, evidence of atherosclerosis on a coronary angiography, or a history of coronary artery bypass graft or angioplasty (9,10). DNA is obtained from all available family members, and the adult offspring (age 18 years or older) of the proband and the spouses of those offspring are also asked to undergo a series of tests to characterize their metabolic and cardiovascular phenotype. In the present study, 832 subjects from 164 families were genotyped; of these, 403 adult offspring and offspring spouses from 99 families underwent the euglycemic- hyperinsulinemic clamp.

All studies were approved by the Human Subjects Protection Institutional Review Boards at the University of California at Los Angeles and Cedars-Sinai Medical Center. All subjects gave informed consent before participating.

Genotyping.

We genotyped 12 single nucleotide polymorphisms (SNPs) across the AMPD1 gene (Fig. 1). We selected six variants (rs926938, rs2010899, rs2268698, rs2269697, rs743041, and rs761755) based on the finding that they were commonly shared across different population groups (11). We also genotyped missense variants known to be associated with altered AMPD1 function (Q12X, P48L, and Q156H) (12,13). The remaining variants were selected from the National Center for Biotechnology Information SNP database (www.ncbi.nlm.nih.gov/SNP/). Q12X was incompatible with our genotyping assay, and two variants (Q156H and S269S) were not polymorphic in our Mexican-American population and therefore were not considered further. The nine remaining SNPs were successfully genotyped in 832 subjects from 164 families. For each SNP, 1 represents the major allele and 2 represents the minor allele. This large-scale genotyping was performed using the 5′-exonuclease (Taqman MGB) assay, as previously described (9,14). PCR primer and oligonucleotide probe sequences are listed in Table 1.

Phenotyping.

In all, 403 genotyped adult offspring and their spouses underwent a 3-day phenotyping protocol that included indexes of insulin resistance and clearance determined by a euglycemic clamp study on the 3rd day.

During the euglycemic-hyperinsulinemic clamp (5), a priming dose of human insulin (Novolin; Novo Nordisk, Clayton, NC) was given, followed by infusion for 120 min at a constant rate (60 mU · m−2 · min−1), with the goal of achieving a plasma insulin concentration of 100 μIU/ml or greater. Blood was sampled every 5 min, and the rate of 20% dextrose coinfused was adjusted to maintain plasma glucose concentrations at 95–100 mg/dl. The glucose infusion rate (M; given in milligrams per kilogram per minute) over the last 30 min of steady-state insulin and glucose concentrations reflects glucose uptake by all body tissues (primarily insulin-mediated glucose uptake in muscle) and is therefore a direct physiological measurement of tissue insulin sensitivity (5). Often, an insulin sensitivity index (Si) is calculated as M/I, where I is the steady-state insulin level. In this study, to clearly distinguish between insulin sensitivity and insulin clearance, we relied on M as the insulin sensitivity measure because the calculations of Si and insulin clearance both use steady-state insulin in the denominator.

The plasma insulin levels during the steady state of the clamp study are a direct reflection of the MCRI. The MCRI (milliliters per meter squared per minute) was calculated as the insulin infusion rate divided by the final steady-state plasma insulin level of the euglycemic clamp. This formula was chosen because the hyperinsulinemic infusion is known to suppress endogenous insulin secretion; furthermore, in vivo tissue clearance mechanisms do not distinguish between endogenously secreted insulin molecules and infused insulin. Because this measurement of MCRI assesses both, it provides the most accurately available measure of overall insulin clearance in this data set.

Data analysis.

Log-transformed trait values (BMI, fasting insulin, and MCRI) or square root−transformed values (M) were used to reduce skewness for all statistical analyses. Unpaired, two-sided t tests were used to compare trait values between men and women.

The pairwise relation of age, BMI, fasting insulin, M, and MCRI were individually assessed using simple regression. P values were derived using generalized estimating equations to account for familial relationships (15). Generalized estimating equations were used to assess the joint effects of age, BMI, sex, M, and MCRI on fasting insulin, adjusting for familial relationships.

Heritability estimates were obtained using the SOLAR (Sequential Oligogenic Linkage Analysis Routines) program (16) to implement a variance components approach. The total phenotypic variance in a trait (σ2P) was partitioned into the variance due to the additive effects of genes (σ2G) and environmental effects (σ2E). The genetic effect was assumed to be independent and normally distributed with zero mean and variance of σ2G. Heritability (h2) of a trait was calculated by the ratio of the genetic variance (σ2G) divided by the total phenotypic variation.

The program Haploview was used to determine haplotypes as well as to delineate haplotype blocks (17). Haploview constructs haplotypes by using an accelerated expectation maximization algorithm, similar to the partition/ligation method (18), which creates highly accurate population frequency estimates of the phased haplotypes based on the maximum likelihood derived from the unphased input genotypes. Haploview was used to calculate linkage disequilibrium (LD; the D′ statistic) between each pairwise combination of all nine SNPs used in the haplotype block determination. To determine haplotype blocks, Haploview searches for regions of strong LD (D′ > 0.8) running from one marker to another, wherein the first and last markers in a block are in strong LD with all intermediate markers.

Association was evaluated by quantitative transmission disequilibrium testing for individual polymorphisms and haplotypes using the QTDT program (19). The transmission disequilibrium test (TDT) was first developed for dichotomous traits in which alleles transmitted and not transmitted from parents to affected offspring are compared to determine whether one allele is associated with the disease in question (20). The TDT was later extended to quantitative traits (21). Abecasis et al. (19) developed a general approach for scoring allelic transmission that accommodates families of any size and uses all available genotypic information. Family data allow the construction of an expected genotype for every nonfounder, and orthogonal deviates from this expectation are a measure of allelic transmission. The QTDT program implements this general TDT using the orthogonal model of Abecasis et al. (22). In our study, age, sex, and BMI were specified as covariates. Environmental variance, polygenic variance, and additive major locus were specified in the variance model. The within-family component of association was evaluated to eliminate any effects of population stratification.

RESULTS

The clinical characteristics of the 403 subjects (168 men, 235 women) who underwent clamp assessment of insulin resistance are shown in Table 2. There were no significant differences between the men and women in anthropometric or insulin-related traits.

Both M and MCRI were negatively correlated with the fasting insulin concentration (P < 0.0001 for both comparisons) (Table 3). There was a weak correlation between MCRI and M (r = 0.085, P = 0.032). BMI was highly correlated with M (r = −0.58, P < 0.0001) but only weakly correlated with MCRI (r = −0.11, P = 0.016).

Age, sex, BMI, M, and MCRI were analyzed jointly to determine which were independent predictors of the fasting insulin level. Age, BMI, M, and MCRI were all highly significant (P = 0.0016, P < 0.0001, P < 0.0001, and P = 0.0006, respectively) predictors of fasting insulin in this joint analysis.

Fasting insulin and insulin resistance are known to be heritable traits (23). However, to our knowledge, the genetic contribution to MCRI has not been previously investigated. We used a variance component method to estimate the heritability of the clamp-derived indexes of insulin sensitivity and clearance (Table 4). In our population, the covariate-adjusted heritability of fasting insulin was 0.38 (P = 0.0011) and that of M was 0.44 (P < 0.0001). The heritability of Si was 0.40. In comparison, the heritability of MCRI was substantially higher at 0.58 (P < 0.0001).

The frequencies of the nine polymorphic AMPD1 SNPs are shown in Table 5. The genotype frequencies for all nine markers were in Hardy-Weinberg equilibrium. LD among the four markers (D′) ranged from 0.11 to 1.0 (average pairwise D′ of 0.86. Two haplotype blocks were identified, one major block spanning the 5′ end of the gene to intron 5 and a smaller haplotype block comprising the two SNPs in intron 6 (Fig. 2). The average pairwise LD within the 5′ haplotype block was 0.94.

The association of AMPD1 SNPs with insulin-related traits was evaluated using QTDT. No SNP showed a significant association with fasting insulin or M. SNP3, SNP6, and SNP7 were associated with MCRI (P = 0.037, 0.041, and 0.0091, respectively). We also evaluated the association of haplotypes from the large haplotype block that extends for 14 kb from upstream of the AMPD1 gene to intron 5. AMPD1 haplotypes were not associated with fasting insulin or M. However, the most common haplotype, haplotype 1, and the second most common haplotype, haplotype 2, were both significantly associated with MCRI (P = 0.017 and 0.015, respectively). Of note, the minor alleles of the associated SNPs lie on these haplotypes, with those of SNP3 and SNP6 lying on haplotype 1 and that of SNP7 lying on haplotype 2 (Fig. 2).

Figure 3 shows the mean MCRI levels according to haplotype carrier status and haplogenotype among 320 individuals of the offspring generation who were haplotyped and phenotyped. Haplotype 1 was associated with increased MCRI and haplotype 2 was associated with decreased MCRI. A dosage-response relation was observed whereby the number of chromosomes bearing haplotype 1 corresponded with increasing MCRI, and the number of chromosomes bearing haplotype 2 corresponded with decreasing MCRI.

The heritability of MCRI was recalculated with the AMPD1 haplogenotype as a covariate (Table 4) to assess the impact of the AMPD1 genotype on the heritability of MCRI. In this model, the heritability of MCRI was 0.49 (P < 0.0001), indicating that the AMPD1 genotype accounts for ∼15% of the heritability of MCRI and that other as-yet unidentified genes must also contribute to the heritability of MCRI.

DISCUSSION

In this study, we examined the genetic nature of various insulin-related phenotypes. We found that MCRI is a highly heritable trait and that specific haplotypes in the AMPD1 gene are closely linked to quantitative differences in the overall MCRI in our study population.

MCRI and M are independent predictors of fasting insulin concentration. Insulin-resistant nondiabetic subjects maintain normoglycemia by a compensatory increase in insulin secretion, which explains the negative correlation between insulin sensitivity and fasting insulin. In a similar vein, the negative correlation between MCRI and fasting insulin is consistent with the concept that once insulin clearance declines, insulin concentrations rise.

MCRI becomes of great interest because of the evidence presented herein that it is a highly heritable trait. In fact, it was more heritable in our study than M. To our knowledge, this is the first report assessing the heritability of MCRI. Insulin sensitivity/resistance is known to be heritable, as is evidenced by the observation of reduced insulin sensitivity in nondiabetic relatives of type 2 diabetic subjects (24,25). The heritability of insulin sensitivity is 0.28–0.44 when quantified by the frequently sampled intravenous glucose tolerance test (26,27) and is 0.37 when assessed by the euglycemic clamp (28). The heritability of M observed in our study is consistent with these reports.

As a major determinant of circulating insulin levels, MCRI is potentially of great importance in that insulin levels may play a role in modulating processes that influence the development of atherosclerosis. Proatherogenic effects of insulin include stimulation of proliferation of vascular smooth muscle cells, as well as production of plasminogen activator inhibitor 1 by these cells (29,30). In contrast, insulin may protect against atherosclerosis by antagonizing inflammatory transcription factors, inhibiting adhesion molecule expression, and promoting nitric oxide production in endothelial cells (3133). The elucidation of genetic determinants of MCRI will provide insight into not only how insulin is cleared but also the mechanisms of insulin action.

The AMPD1 gene (chromosome 1p13) codes for the muscle-specific form of the AMP deaminase enzyme (myoadenylate deaminase), which catalyzes the deamination of AMP to inosine monophosphate in skeletal muscle. Mutations in AMPD1, which are found in ∼20% of the Caucasian population, are frequently found in patients with exercise-induced myopathy. Inherited defects in AMPD1 that lead to decreased activity of this enzyme result in AMP accumulation in skeletal myocytes (7); the resultant alteration in adenylate energy charge has the potential to influence the activity of numerous enzymes. For example, a reduction in AMPD1 expression or function would lead to increased AMP levels, which activate AMPK.

AMPK serves as a cellular energy sensor, acting to maintain cellular ATP levels by phosphorylating metabolic enzymes and regulating gene expression (34). For example, AMPK phosphorylates and inactivates enzymes in the gluconeogenic pathway and inhibits gene expression of these enzymes (e.g., PEPCK, G6Pase) (35). AMPK stimulates muscle glucose uptake by increasing expression and translocation of GLUT4, stimulates fatty acid oxidation in muscle and liver, inhibits hepatic glucose production, and inhibits lipolysis and lipid synthesis (8,34,36). AMPK has emerged as a possible mediator of the effects of insulin-sensitizing medications (37,38). Of interest, biguanide and thiazolidinedione insulin sensitizers that activate AMPK also alter insulin clearance (39).

Alterations in AMPD1 activity, with its resultant effects on AMPK or other metabolic pathways, may not be uniformly manifest throughout the cell. AMPD1 binds reversibly to intracellular organelles, such as the myofibril (7); consequently, changes in the activity of this enzyme may alter metabolism differentially in localized regions of the myocyte. If the insulin receptor endocytic pathway, which is an energy-requiring, critical step in insulin clearance, was modified by local changes in the adenylate energy charge or if a protein component of this pathway were a substrate responsive to local changes in AMPK activity, then this might provide a mechanism for the observed alterations of MCRI in patients with the various AMPD1 genotypes.

AMPD1 is a tightly regulated, allosteric enzyme that contains unique regulatory domains in the nonconserved NH2-terminal region that interact with the catalytic and nucleotide regulatory sites located in the conserved COOH-terminal region (7,40). Figures 1 and 2 indicate that the boundary between the two haplotype blocks observed maps closely to the boundary between the nonconserved, isoform-specific 5′ region of the AMPD1 gene and the highly conserved 3′ region of this gene, which is shared with all members of this multigene family (7). Of note, the boundary between conserved and nonconserved region extends all the way to the yeast enzyme (41). The fact that we observed phenotype association of haplotypes only in the block that maps to the isoform-specific region of AMPD1 further supports the conclusion that variation in this gene influences MCRI.

These results may find practical application in the pharmacokinetics of insulin treatment. When given as a drug, the clearance of administered insulin is an important determinant of the amount of insulin required to attain an appropriate plasma concentration. Thus, the AMPD1 genotype may be one determinant of the insulin dosage required to achieve adequate glucose control in diabetic subjects.

It is important to note that, overall, in vivo measurements of insulin clearance primarily reflect the ability of the liver to extract and metabolize insulin. Renal excretion of insulin is also significant, and it has been estimated that together, hepatic and renal mechanisms account for up to 80% of total insulin clearance. Thus, skeletal muscle contributes a relatively small component of total insulin clearance, perhaps up to 20%, and changes in muscle insulin clearance will have only modest effects on total body insulin clearance. Because AMPD1 is a skeletal muscle−specific enzyme, any effect that a variation in AMPD1 expression or function has on insulin clearance must be exerted within skeletal muscle itself. With this line of reasoning, because the various measures of insulin clearance differ by 8–15% in patients with and without the AMPD1 haplotype 1, it is possible to infer that this haplotype may lead to a 30–50% variation in skeletal muscle insulin clearance in affected individuals.

In summary, we have examined the genetic regulation of various insulin-related phenotypes. We found that the MCRI is a highly heritable trait and that specific haplotypes in the AMPD1 gene are closely associated with quantitative differences observed in overall MCRI in our study group. AMPD1 is well located from a metabolic perspective to modulate other enzymes (e.g., AMPK) that are known to influence insulin action. The association we have described between AMPD1 and insulin clearance provides insight into new biochemical pathways that can modulate insulin action and clearance in skeletal muscle, a critical target organ. Interventions that alter adenine nucleotide levels and adenylate energy charge may represent new therapeutic targets for modifying insulin action in syndromes of insulin resistance.

FIG. 1.

AMPD1 gene organization and location of SNPs. The AMPD1 gene (chromosome 1p13) spans ∼22 kb in genomic length. The nine genotyped SNPs that were polymorphic in our population are listed on the top, and the three SNPs not genotyped are listed on the bottom. The width of the exons is slightly exaggerated for clarity. □, exons specific to AMPD1; ▪, conserved exons shared with AMPD2 and AMPD3.

FIG. 2.

LD plot showing haplotype blocks in the AMPD1 gene. D′ values (percent) are indicated in the LD plot (left panel). The solid blocks indicate D′ = 1 (100%) for the corresponding pair of variants. Right panel: Haplotype listing (in rows) with corresponding population frequencies. Each row represents a haplotype, and each column represents one SNP (1 = major allele; 2 = minor allele). The program Haploview was used to generate this figure.

FIG. 3.

Mean levels of insulin clearance by the AMPD1 haplogenotype. The three bars represent the mean MCRI for all subjects, haplotype 1 carriers, and haplotype 2 carriers, respectively. The line plots represent the mean MCRI for the haplogenotypes indicated. Vertical lines represent the SE.

TABLE 1

Primers and probe sequences used in the 5′-exonuclease assay

TABLE 2

Clinical characteristics of subjects

TABLE 3

Correlations among variables of insulin metabolism and action

TABLE 4

Heritability of indexes of insulin metabolism and action

TABLE 5

AMPD1 SNP frequencies

Acknowledgments

The Mexican-American Coronary Artery Disease project is supported in part by National Institutes of Health Program Project Grant HL-60030. Further support came from the Cedars-Sinai Board of Governors’ Chair in Medical Genetics (to J.I.R.), the Cedars-Sinai General Clinical Research Center Grant RR000425, and the Diabetes Endocrinology Research Center Grant DK63491.

We thank all the study participants and referring physicians.

Footnotes

    • Accepted January 10, 2005.
    • Received November 19, 2004.

REFERENCES

| Table of Contents