Brief Genetics Report Association Between a Novel Variant of the Human Type 2 Thr92Ala and Insulin Resistance Evidence of Interaction With the Trp64Arg Variant of the ␤-3-Adrenergic Receptor Daniela Mentuccia,1,2 Laura Proietti-Pannunzi,3 Keith Tanner,1 Vincenzo Bacci,4 Toni I. Pollin,1 Eric T. Poehlman,5 Alan R. Shuldiner,1,6 and Francesco S. Celi3

Thyroid hormone action is an important determinant of energy and glucose metabolism. T4 metabolism is regu- lated by the of which type 2 is expressed in he cluster of obesity, hypertension, insulin resis- humans in skeletal muscle and brown adipose tissue, tance, and glucose intolerance/diabetes (Syn- where its transcription is stimulated by the ␤-3 adren- drome X) is a polygenic multifactorial condition ergic pathway. We performed molecular scanning of the in which the phenotype is the net result of the human type 2 deiodinase (DIO2) gene and evaluated a T interaction between environmental factors and genetic novel variant for associations with obesity and insulin resistance, assessing both the main effect and interac- predisposition deriving from variations in encoding tion with the Trp64Arg ␤-3–adrenergic receptor (ADRB3) involved in metabolic pathways (1). In fact, variant. Molecular scanning of DIO2 in 50 obese Cauca- polymorphisms in several candidate genes, such as the ␤-3 sians demonstrated a Thr92Ala variant. Association adrenergic receptor (ADRB3) (2,3), insulin receptor sub- studies in 972 nondiabetic patients, 135 of whom under- strate-1 (4), peroxisome proliferator–activated receptor-␥ went euglycemic-hyperinsulinemic clamps, showed that (5), fatty acid binding -2 (6), and perhaps others, subjects with the Thr92Ala variant had lower glucose have shown association with various traits of Syndrome X. ⅐ ؊1 ⅐ ؊1 ؎ disposal rate (0.54 0.02 mg min kg fat-free mass hormones are important elements in the regu- Ala92 homozygotes vs. 0.44 ؎ 0.02 Ala92 heterozygotes Asso- lation of metabolic rate (7,8) and, although the molecular .(0.0088 ؍ vs. 0.42 ؎ 0.04 Thr92 homozygotes, P ciation analysis of the entire group showed significant and cell biology of this process is not yet fully understood, evidence for a synergistic effect between the Thr92Ala evidence suggests that the stimulate DIO2 and Trp64Arg ADRB3 variants on BMI (both vari- resting metabolic rate (RMR) by increasing ATP expendi- -ants 34.3 ؎ 0.9 kg/m2 vs. neither variant 33.1 ؎ 0.4 kg/m2, ture and through the regulation of expression of uncou .(for interaction). To our knowledge, Thr92Ala pling proteins in the mitochondria of fat and muscle (9 0.04 ؍ P is the first description of a missense mutation of DIO2. Thyroid hormone also modulates adrenergic receptor This variant strongly associates with insulin resistance number and thus responsiveness to catecholamines, and, in subjects with the Trp64Arg ADRB3 variant, an which are also regulators of metabolic rate (10). Thus, increased BMI, suggesting an interaction between these thyroid hormone action is an extremely important deter- two common gene variants. Diabetes 51:880–883, 2002 minant of the maintenance of the energy homeostasis. Furthermore, thyroid hormone influences carbohydrate From the 1Division of Endocrinology, Diabetes and Nutrition, Department of metabolism in skeletal muscle and adipose tissue via the Medicine, University of Maryland, Baltimore, Maryland; the 2Graduate Pro- positive transcriptional regulation of the muscle/fat spe- gram in Endocrinology, University of Rome “Tor Vergata,” Rome, Italy; the 3Department of Experimental Medicine and Pathology, University of Rome cific GLUT4 (11,12). “La Sapienza,” Rome, Italy; 4Division of Nutrition, University of Rome “La The deiodinases play a key role in the maintenance of Sapienza, Rome, Italy; the 5Department of Nutrition, University of Montreal, circulating and tissue levels of thyroid hormones. The Montreal, Canada; the 6Baltimore Veterans Administration Geriatric Research and Education Clinical Center, Baltimore, Maryland. pro-hormone T4 is converted in the periphery to its active Address correspondence and reprint requests to Francesco Saverio Celi, form, T3, or to its inactive metabolite, reverse T3, by the Division of Endocrinology, Diabetes and Nutrition, University of Maryland action of these enzymes (13). The deiodinases are seleno- School of Medicine, 660 West Redwood St., Room 484, Baltimore, MD 21201. E-mail: [email protected] or [email protected]. enzymes characterized by a selenocysteine in the catalytic Received for publication 1 October 2001 and accepted in revised form 5 domain of the enzyme encoded by a UGA codon in the pres- December 2001. Ј Additional information for this article can be accessed at http://diabetes. ence of a characteristic 3 untranslated region stem loop diabetesjournals.org. structure, the selenocysteine insertion sequence (SECIS) ADRB3, ␤-3–adrenergic receptor; DIO2, human type 2 deiodinase; RFLP, (14). Type 2 deiodinase appears to be a tissue-specific restriction fragment–length polymorphism; RMR, resting metabolic rate; SE- CIS, selenocysteine insertion sequence; SSCP, single strand conformation regulator of the intracellular T3 concentrations in brown polymorphism. fat, brain, and pituitary. In humans, type 2 deiodinase is

880 DIABETES, VOL. 51, MARCH 2002 D. MENTUCCIA AND ASSOCIATES

TABLE 1 euglycemic clamp studies were performed in a subgroup Study population characteristics of 135 nondiabetic women. In this subgroup, there were no Age Men/ significant differences between genotypes in age, weight, n (years) BMI (kg/m2) Women BMI, fasting glucose levels, or RMR (Table 2). However, we observed a marked decrease in the glucose disposal Gene scanning 50 38.5 Ϯ 11 46.0 Ϯ 5 11/39 rate in subjects with the Thr92Ala variant (Ala92 homozy- Ϯ Ϯ Association analysis 972 56.0 13 32.8 8 71/901 gotes 17.2 Ϯ 1.6 mmol/min vs. Ala92 heterozygotes 18.2 Ϯ (total) Ϯ Association analysis 135 42.3 Ϯ 16.7 27.1 Ϯ 7.8 0/135 0.8 mmol/min vs. Thr92 homozygotes 22.6 0.9 mmol/min, (hyperinsulinemic P ϭ 0.0006), consistent with a dominant model (Table 2). clamp) A trend toward an increase of fasting insulin level was observed in carriers of the Ala92 DIO2 allele (Ala92 homo- Ϯ Data are means SD and n. zygotes 105.0 Ϯ 18.6 pmol/l vs. heterozygotes 72.0 Ϯ 9.6 pmol/l vs. Thr92 homozygotes 57.0 Ϯ 10.2 pmol/l, P ϭ also expressed in skeletal muscle (15). The type 2 deiodi- 0.0814) consistent with greater insulin resistance. nase gene (DIO2) is localized on 14q24.3, and In the larger group (n ϭ 922), the Thr92Ala DIO2 the entire gene is encoded by three exons (16). Experi- polymorphism did not associate with body weight or BMI. mental evidence demonstrates that DIO2 activity is regu- lated by metabolic stressors, such as cold exposure and However, significantly higher weight and BMI were ob- adrenergic stimulation, via the generation of intracellular served in carriers of both the Thr92Ala DIO2 and Trp64Arg cAMP (17,18). Furthermore, the action of this enzyme ADRB3 variants compared with subjects with neither or provides the supply of T3 for local use by the tissue itself, either variant alone, suggesting an interaction between the thus creating a type of autoregulatory, autocrine feedback two (Table 3). loop (19). We thus decided to study DIO2 as a candidate To our knowledge, this is the first description of a gene for obesity and insulin resistance in a population of missense variant of the human DIO2 gene. Although the morbidly obese Caucasians with normal thyroid function crystal structure of the type 2 deiodinine is not yet known, and to perform association studies of the gene variants in it is worth noting that this nonconservative amino acid a well-characterized nondiabetic Caucasian population change (aliphatic for polar group), which is not located (Table 1). within the conserved deiodinase catalytic domain, could By performing gene scanning of the entire coding region potentially affect its activity (21). However, this region of and the SECIS element of DIO2 with PCR–single strand the enzyme is not phylogenetically conserved. The homol- conformation polymorphism (SSCP) analysis, followed by ogous amino acid is represented by a proline in rodents DNA sequence analysis, we identified a common noncon- and by a glycine in chick. By contrast, humans and am- servative variant, 274 A3G, that predicts a nonconserva- phibians share a threonine in this position. The functional tive Thr92Ala substitution (Fig. 1). No other variants were consequences of this common missense mutation are not observed. The allele frequency of the Thr92Ala DIO2 yet known. variant was 0.35 in the study population with a distribution Our results indicate that the DIO2 Thr92Ala variant meeting Hardy-Weinberg equilibrium. This variant was strongly associates with insulin resistance as measured by common in various ethnic groups, particularly in Pima the hyperinsulinemic-euglycemic clamp. An inactivating Indians and Mexican-Americans with allele frequencies of mutation in DIO2 could lead to decreased intracellular 0.75 and 0.42, respectively (see online appendix at http:// availability of active thyroid hormone. A reduction in T3 diabetes.diabetesjournals.org). would, in turn, decrease the transcription of GLUT4 in Association studies were performed in a large, well- insulin-sensitive tissues, such as skeletal muscle and adi- characterized nondiabetic Caucasian population (20) re- pose tissue, contributing to insulin resistance. On the cruited for energy balance studies. Hyperinsulinemic- other hand, although DIO2 is not known to be expressed in

TABLE 2 Association of Thr92Ala DIO2 and glucose disposal euglycemic hyperinsulinemic clamp (n ϭ 135 women) Thr/Thr Thr/Ala Ala/Ala P (ANOVA) Age (year) 41.9 Ϯ 2.3 (52) 43.9 Ϯ 2.1 (66) 37.2 Ϯ 4.0 (17) 0.33 Weight (kg) 72.3 Ϯ 2.3 (52) 73.3 Ϯ 2.1 (66) 74.3 Ϯ 4.1 (17) 0.90 BMI (kg/m2) 26.6 Ϯ 0.8 (52) 27.4 Ϯ 0.7 (66) 27.6 Ϯ 1.4 (17) 0.70 Fasting glucose 4.6 Ϯ 0.1 (51) 4.7 Ϯ 0.1 (64) 4.8 Ϯ 0.1 (17) 0.35 (mmol/l) Fasting insulin 57.0 Ϯ 10.2 (49) 72.0 Ϯ 9.6 (56) 105.0 Ϯ 18.6 (16) 0.081 (pmol/l) Glucose disposal 22.6 Ϯ 0.9 (52) 18.2 Ϯ 0.8 (66)* 17.2 Ϯ 1.6 (17)* 0.0006 (mmol/min) Glucose disposal 0.54 Ϯ 0.02 (52) 0.44 Ϯ 0.02 (66)† 0.42 Ϯ 0.04 (17)† 0.0088 per kilogram fat free mass (mg ⅐ minϪ1 ⅐ kgϪ1) *Data are means Ϯ SEM (n). All variables adjusted for age. *P ϭ 0.0001 for dominant model; †P ϭ 0.0002 for dominant model. Thr/Thr ϭ Thr92 homozygotes; Thr/Ala ϭ heterozygotes; Ala/Ala ϭ Ala92 homozygotes.

DIABETES, VOL. 51, MARCH 2002 881 Thr92Ala DIO2 AND INSULIN RESISTANCE

TABLE 3 Interaction between Thr92Ala DIO2 and Trp64Arg ADRB3 Genotype Neither variant Ala92 DIO2 only Arg64 ADRB3 only Both variants P (interaction) n 364 474 52 70 Age (years) 55.7 Ϯ 0.7 55.5 Ϯ 0.6 58.3 Ϯ 1.7 56.5 Ϯ 1.5 0.501 Weight (kg) 87.5 Ϯ 1.1 86.1 Ϯ 1.0 83.2 Ϯ 2.9 92.4 Ϯ 2.5 0.011 BMI (kg/m2) 33.1 Ϯ 0.4 32.4 Ϯ 0.4 31.8 Ϯ 1.1 34.3 Ϯ 0.9 0.041 *Data are means Ϯ SE. Adjusted for age and sex. Subjects homozygous and heterozygous for each variant were reported together. liver, we cannot rule out the possibility of a decreased rule out the possibility of stratification bias and cannot hepatic insulin action. comment on the generalizability of our findings to other Several studies have demonstrated that the Trp64Arg ethnic populations. Third, without functional studies, it is ADRB3 variant generates, upon stimulation, a reduced possible that the Thr92Ala variant is in linkage disequilib- amount of cAMP compared to the wild-type receptor (22). rium with another pathogenic variant in the DIO2 gene or Furthermore, we and others (23,24) have demonstrated a gene close by. Finally, our results could represent a type that a functional cyclic AMP-responsive element is present 1 error. However, the P values for clamp measurements in the promoter region of the DIO2 gene. The interaction remains significant at the 0.01 level, even if Bonferroni between the Thr92Ala DIO2 and the Trp64Arg ADRB3 corrected for all 29 variables in our database. This finding, variants and a higher BMI could thus be due to a reduction along with the scientific plausibility of an association, in the transcription of an already functionally defective provides evidence against type 1 error. enzyme, ultimately generating a reduction of adrenergic- In conclusion, we have identified a novel common driven thyroid hormone–mediated lipolysis in adipose missense mutation of the DIO2 gene, which strongly tissue. No significant difference was observed in thyroid associates with insulin resistance and, in the presence of function parameters between subjects with or without the Trp64Arg ADRB3, increased BMI. Further association and Thr92Ala DIO2 variant in the subgroup of Italian obese in vivo and in vitro studies are needed to evaluate the subjects (data not shown). functional and epidemiological importance of this variant This study has several strengths. First, the number of alone and combined with other candidate gene variants for subjects studied was quite large, enabling us to examine obesity and insulin resistance. gene/gene interactions. Furthermore, we had good power to detect relatively modest differences in phenotypes. RESEARCH DESIGN AND METHODS Second, all subjects were Caucasian, thus reducing the Study population. DIO2 molecular scanning was performed in 50 unrelated risk of false positive/negative associations due to stratifi- morbidly obese Caucasian subjects. Abnormal thyrotropin (Ͼ5 mUI/l) or cation bias. Third, a subset of individuals underwent secondary causes of obesity were considered as exclusion criteria. A well- euglycemic-hyperinsulinemic clamps, which enabled us to characterized (20) Caucasian population of 972 unrelated subjects (91 men and 901 women participating in energy balance and molecular genetics studies evaluate direct measures of insulin sensitivity rather than of obesity) were used for association studies. Inclusion criteria were as relying on indirect and less precise measures, as is the follows: nondiabetic, nonsmokers, not taking any medication, without cardio- case in other studies of this kind. Despite these strengths, vascular disease or hypertension. Glucose disposal rate, determined by Ϫ there were also limitations. First, the vast majority of the hyperinsulinemic-euglycemic clamp (25) (insulin infusion rate 40 mU ⅐ m 1 ⅐ minϪ1) on a subset of 135 women, was performed as previously described subjects in our study were women, and no men underwent (20). All study subjects provided informed consent. euglycemic-hyperinsulinemic clamp. Thus, we do not Mutation screening and genotype analysis. The entire coding region, the know if these findings apply to men. Second, we cannot intron-exon junctions, and the SECIS element of DIO2 were screened with

FIG. 1. Thr92Ala Dio2 polymorphism. A: PCR-SSCP. B: Dydeoxy DNA sequencing. C: PCR-RFLP with Bsg-I. 1, Thr92 homozygote; 2, heterozygote; 3, Ala92 homozygote.

882 DIABETES, VOL. 51, MARCH 2002 D. MENTUCCIA AND ASSOCIATES

SSCP analysis. The gels (MDE; FMC, Rockland, ME) were run at four different small dose changes in patients on chronic thyroid hormone replacement. conditions, i.e., 25°Cor4°C in the presence and absence of 10% glycerol. In J Endocrinol Metab 82:1118–1125, 1997 our hands, the sensitivity of this method approaches 100%. The aberrantly 9. Reitman ML, He Y, Gong DW: Thyroid hormone and other regulators of migrating bands were confirmed with a second experiment. Direct dideoxy uncoupling proteins. Int J Obes Relat Metab Disord 23 (Suppl. 6):S56–S59, DNA sequencing on both strands of aberrantly migrating PCR products was 1999 performed manually according to established methods. The polymorphism 10. Rubio A, Raasmaja A, Silva JE: Thyroid hormone and norepinephrine was confirmed by PCR restriction fragment–length polymorphism (RFLP) signaling in brown adipose tissue. II. Differential effects of thyroid hor- analysis by digesting the PCR product resulting from a 5ЈCTCAGGGCTG- mone on beta 3-adrenergic receptors in brown and white adipose tissue. GCAAAGTCAAG3Ј sense primer and a 5ЈCCACACTCTATTAGAGCCAATTG3Ј Endocrinology 136:3277–3284, 1995 antisense primer with BsgI. The same PCR-RFLP strategy was used for 11. Torrance CJ, Devente JE, Jones JO, Dohm GL: Effects of thyroid hormone population screening. PCR-RFLP of the Trp64Arg ADRB3 variant was per- on GLUT 4 glucose transporter gene expression and NIDDM in rats. formed as previously described (2). Endocrinology 138:1204–1214, 1997 Statistical analysis. Data were analyzed by age-adjusted and, where appro- 12. Richardson JM, Pessin JE: Identification of a skeletal muscle-specific priate, sex-adjusted general linear models using SAS version 7.0 (SAS Insti- regulatory domain in the rat GLUT4/muscle-fat gene. J Biol Chem 268: tute, Cary, NC). To examine the main effect of the DIO2 Thr92Ala variant, the 21021–21027, 1993 three genotypes (Thr/Thr, Thr/Ala, and Ala/Ala) were considered separately, 13. Kohrle J: Local activation and inactivation of thyroid hormones: the followed by pooling the Thr/Ala and Ala/Ala groups when the glucose disposal deiodinase family. Mol Cell Endocrinol 151:103–119, 1999 data provided evidence of a dominant effect. In the analysis of interaction with 14. Berry MJ, Banu L, Chen Y, Mandel SJ, Kieffer DJ, Harney JW, Larsen PR: Trp64Arg ADRB3, each of the two variants was modeled as dichotomous Recognition of UGA as a selenocysteine codon in type I deodinase requires variables representing the presence or absence of at least one copy of the sequences in the 3Ј untranslated region. Nature 353:273–276, 1991 respective variant. 15. St Germain DL, Galton VA: The deiodinase family of selenoproteins. Thyroid 7:655–658, 1997 16. Celi FS, Canettieri G, Mentuccia D, Proietti-Pannunzi L, Fumarola A, Sibilla ACKNOWLEDGMENTS R, Predazzi V, Ferraro M, Andreoli M, Centanni M: Structural organization The financial support of Telethon-Italy (Grant E.763) is and chromosomal localization of the human type II deiodinase gene. Eur J Endocrinol 143:267–271, 2000 gratefully acknowledged. F.S.C. is a recipient of the Amer- 17. Silva JE, Larsen PR: Adrenergic activation of production ican Heart Association Grant 0160358U. in brown adipose tissue. Nature 305:712–713, 1983 18. Safran M, Farewell AP, Leonard JL: Catalytic activity of type II iodothyro- nine 5Ј-deiodinase polipeptide is dependent upon cyclic AMP activation REFERENCES factor. J Biol Chem 271:16363–16368, 1996 1. Stern MP: Strategies and prospects for finding insulin resistance genes. 19. Bianco AC, Silva JE: Nuclear 3,5,3Ј-triiodothyronine (T3) in brown adipose J Clin Invest 106:323–327, 2000 tissue: receptor occupancy and sources of T3 as determined by in vivo 2. Walston J, Silver K, Bogardus C, Knowler WC, Celi FS, Austin S, Manning techniques. Endocrinology 120:55–62, 1987 B, Strosberg D, Stern MP, Raben N, Sorkin J, Roth J, Shuldiner AR: Time 20. Garcia-Rubi E, Starling RD, Tchernof A, Matthews DE, Walston JD, of onset of non-insulin-dependent diabetes mellitus and genetic variation Shuldiner AR, Silver K, Poehlman ET, Calles-Escandon J: Trp64Arg variant of the B3-adrenergic-receptor gene. N Engl J Med 333:343–347, 1995 of the beta3-adrenoceptor and insulin resistance in obese postmenopausal 3. Widen E, Lehto M, Kanninen T, Walston J, Shuldiner AR, Groop LC: women. J Clin Endocrinol Metab 83:4002–4005, 1998 Association of a polymorphism in the beta 3-adrenergic-receptor gene with 21. Buettner C, Harney JW, Larsen PR: The role of selenocysteine 133 in features of the insulin resistance syndrome in Finns. N Engl J Med catalysis by the human type 2 . Endocrinology 333:348–351, 1995 141:4606–4612, 2000 4. Almind K, Bjorbaek C, Vestergaard H, Hansen T, Echwald S, Pedersen O: 22. Kimura K, Sasaki N, Asano A, Mizukami J, Kayahashi S, Kawada T, Fushiki Aminoacid polymorphisms of insulin receptor substrate-1 in non-insulin- T, Morimatsu M, Yoshida T, Saito M: Mutated human beta3-adrenergic dependent diabetes mellitus. Lancet 342:828–832, 1993 receptor (Trp64Arg) lowers the response to beta3-adrenergic agonists in 5. Beamer BA, Yen CJ, Andersen RE, Muller D, Elahi D, Cheskin LJ, Andres transfected 3T3–L1 preadipocytes. Horm Metab Res 32:91–96, 2000 R, Roth J, Shuldiner AR: Association of the Pro12Ala variant in the 23. Canettieri G, Celi FS, Baccheschi G, Salvatori L, Andreoli M, Centanni M: peroxisome proliferator-activated receptor-␥2 gene with obesity in two Isolation of human type 2 deiodinase gene promoter and characterization Caucasian populations. Diabetes 47:1806–1808, 1998 of a functional cyclic adenosine monophosphate response element. Endo- 6. Baier LJ, Sacchettini JC, Knowler WC, Eads J, Paolisso G, Tataranni PA, crinology 141:1804–1813, 2000 Mochizuki H, Bennett PH, Bogardus C, Prochazka M: An amino acid 24. Bartha T, Kim SW, Salvatore D, Gereben B, Tu HM, Harney JW, Rudas P, substitution in the human intestinal fatty acid binding protein is associated Larsen PR: Characterization of the 5Ј-flanking and 5Ј-untranslated regions with increased fatty acid binding, increased fat oxidation, and insulin of the cyclic adenosine 3Ј,5Ј-monophosphate-responsive human type 2 resistance. J Clin Invest 95:1281–1287, 1995 iodothyronine deiodinase gene. Endocrinology 141:229–237, 2000 7. Silva JE: Thyroid hormone control of thermogenesis and energy balance. 25. DeFronzo RA, Tobin JD, Andres R: Glucose clamp technique: a method for Thyroid 5:481–492, 1995 quantifying insulin secretion and resistance. Am J Physiol 237:E214–E223, 8. al-Adsani H, Hoffer LJ, Silva JE: Resting energy expenditure is sensitive to 1979

DIABETES, VOL. 51, MARCH 2002 883