Hum Genet DOI 10.1007/s00439-015-1533-x
REVIEW PAPER
Obesity and genomics: role of technology in unraveling the complex genetic architecture of obesity Yamunah Devi Apalasamy · Zahurin Mohamed
Received: 9 October 2014 / Accepted: 2 February 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract Obesity is a complex and multifactorial disease that occurs as a result of the interaction between “obesogenic” environmental factors and genetic components. Although the genetic component of obesity is clear from the heritability studies, the genetic basis remains largely elusive. Successes have been achieved in identifying the causal genes for monogenic obesity using animal models and linkage studies, but these approaches are not fruitful for polygenic obesity. The developments of genome-wide association approach have brought breakthrough discovery of genetic variants for polygenic obesity where tens of new susceptibility loci were identified. However, the common SNPs only accounted for a proportion of heritability. The arrival of NGS technologies and completion of 1000 Genomes Project have brought other new methods to dissect the genetic architecture of obesity, for example, the use of exome genotyping arrays and deep sequencing of candidate loci identified from GWAS to study rare variants. In this review, we summarize and discuss the developments of these genetic approaches in human obesity.
Introduction Obesity has reached epidemic proportions globally, with an estimation of more than 2.16 billion overweight people and 1.12 billion obese people by 2030 (Kelly et al. 2008). The World Health Organization reported that 65 % of the world’s population resides in countries in which obesity
Y. D. Apalasamy (*) · Z. Mohamed Department of Pharmacology, Pharmacogenomics Laboratory, Faculty of Medicine, University of Malaya, 50603 Kuala Lumpur, Malaysia e-mail: [email protected]
and overweight have led to death. As obesity is a major health and economic burden, the American Medical Association has considered it a disease in 2013 (American Medical Association 2013). It has a significant health burden, as obesity is also associated with other comorbidities such as diabetes, cardiovascular disease, liver and gallbladder disease, cancer, gynecological problems, sleep apnea, and osteoarthritis (Pi-Sunyer 2009). Obesity is a chronic condition in which excessive fat accumulates as a result of imbalance between energy consumption and expenditure (Galgani and Ravussin 2008). Interactions among environmental, genetic, and behavioral factors lead to a complex pathogenesis (Chaput et al. 2014). The modern lifestyle has partially led to a rapid increase in consumption of energy-dense food and physical inactivity. But even in the presence of an obesogenic environment, genetic components play an important role in contributing to individual risk of obesity. Twin, family, and adoption studies have shown that heritability of obesity is high, ranging from 40 to 70 % (Allison et al. 1996; Maes et al. 1997; Stunkard et al. 1990). Various genetic approaches such as linkage analysis and candidate gene association studies have been performed to identify susceptibility genetic loci in an attempt to dissect the biological mechanisms underlying body-weight regulation; however, these attempts have met with limited success. Study designs to dissect the genetic architecture of obesity have progressively shifted from linkage and candidate gene association studies to genome-wide association studies (GWAS) after 2005. In the early studies of obesity, in vivo, in vitro, linkage, and candidate gene association studies were employed to study obesity-susceptibility genes. The Human Genome Project and the International HapMap Project are the major developments which have significantly enhanced our knowledge on genetic variations in
13
the human genome. The human genome sequencing was completed in 2003, and the subsequent developments such as characterization of the LD patterns among the SNPs through the HapMap Project endeavor have led to the breakthrough discovery of the genetic variants of human diseases through GWAS (Schmutz et al. 2004). Rapid developments in technologies at a relatively low cost, such as high-throughput genotyping arrays and next generation sequencing (NGS) technologies, together with advanced statistical methods, have enabled GWAS and copy number variation (CNV) studies to identify numerous genetic loci associated with obesity (Naidoo et al. 2011). To date, GWAS have identified at least 58 novel loci associated with obesity/body mass index (BMI) in European and non-European populations (Lu and Loos 2013). The availability of well-curated Human Gene Mutation Databases has also facilitated large-scale meta-analyses of diseaseassociated variants These databases include the Human Gene Mutation Database, A Catalog of Published GWAS, Online Mendelian Inheritance in Man, ClinVar, and many other locus-specific mutation databases (Amberger et al. 2009; Sherry et al. 2001; Stenson et al. 2013). Nevertheless, the SNPs that have been identified thus far collectively explain only a relatively small proportion of the heritability. Unraveling the complex genetic architecture of obesity remains a challenging task for researchers. In this review, we summarize recent advances in genetic studies of obesity. We also discuss various technological approaches used to dissect the genetics of obesity, especially those that leverage high-throughput genotyping and NGS technologies.
In vivo and in vitro models of obesity Molecular mechanisms of body-weight regulation and obesity development were first discovered through animal models (Zhang et al. 1994; Zucker 1975). Mouse models are critical in obesity research in pre- and post-GWAS era, as reviewed in recent articles (Cox and Church 2011; Speakman et al. 2008). The quantitative genetic approach with mice models offers an excellent tool to study the genetic architecture of obesity. Studies in rodent genetic models to understand the pathways of body-weight regulation and feeding behavior have been useful in demonstrating the monogenic type of obesity (Leibel et al. 1997). For example, the localization of mutant genes in animal model has led to the discovery of monogenic genes such as leptin, a mutation in the LEP gene causing obesity in the ob/ob mouse (Zhang et al. 1994). Apart from single-gene mutations, fine-mapping of quantitative trait loci in mouse models also unraveled more than 200 novel targets related to obesity for candidate gene studies (Snyder et al. 2004).
13
Hum Genet
Polygenic animal models have been crucial in various studies of environmental effects such as epigenetics, responses to high-fat or sugar, and low-calorie and other weightloss diets, as well as in the development of pharmaceutical agents such as leptin and cannabinoid receptor type 1 antagonist agents for treating obesity (Speakman et al. 2008). In addition, adipogenic cellular models have been useful in understanding adipogenesis, energy homeostasis, and insulin resistance in obesity. In vitro approaches such as primary cell cultures, coculture models, and three-dimensional cell cultures have also shed light on the important components of obesity such as growth factors, hormones, and potential pharmacological compounds (Armani et al. 2010; Green and Meuth 1974). For instance, in vitro studies have shown that PPARG is extensively regulated by CCAAT/ enhancer-binding proteins, which subsequently provide a path towards the development of PPARG in the treatment of obesity and type 2 diabetes (Mikkelsen et al. 2010). Notably, in vivo and in vitro models are important in the study of genetics for both monogenic and polygenic obesity (Nilsson et al. 2012). Monogenic obesity is caused by a single-gene mutation, whereas polygenic obesity is associated with multiple genetic factors that interact with an “at-risk” environment (Mutch and Clement 2006). Despite the success in terms of development of some therapeutic targets, there are limited successes in other instances in translating the molecular compounds discovered by in vivo and in vitro studies into usable drugs for humans. For example, treatments of rodents with leptin and MC4R (MTII) antagonists resulted in weight loss, but these compounds were shown to be not clinically useful because they were short acting and also the occurrence of adverse effects (Gibson et al. 2004; Hadley 2005). The major shortcoming of the currently available in vitro and in vivo models is that these models only resemble certain aspects of human obesity; therefore, it would be useful if animal- and cell-based models of therapeutic approaches could be improved in future to reflect the pathophysiology of obesity in humans.
Monogenic obesity Monogenic obesity refers to a rare form of severe obesity that resulted from mutations with a large effect size. These mutations are known to cause severe early onset obesity with hyperphagia as a key feature and multiple endocrine anomalies (Table 1) (Clement 2006). To date, eight candidate genes (LEP, LEPR, PSCK1, POMC, MC4R, SIM1, BDNF, and NTRK2) listed in Table 1, have been identified for monogenic obesity, which have shed light into the biological pathways of the disease. Considerable success in
Hum Genet Table 1 Genes in monogenic obesity Gene
Main clinical characteristics
Severe early onset of obesity; hypogonadism; very low circulating leptin levels LEPR Severe early onset of obesity; hypogonadism; very low circulating leptin levels POMC Severe obesity; hyperphagia; hypoadrenalism; hypopigmentation MC4R Severe early onset of hyperphagia; rapid rise in fat mass BDNF Severe obesity; hyperphagia; hyperactivity; impaired cognitive function; developmental delay NTRK2 Severe obesity; hyperphagia; hyperactivity; impaired cognitive function; developmental delay PSCK1 Hyperphagia; hypogonadism; hyperglycemia; elevation in proinsulin to insulin ratio
LEP
SIM1
Discovery method
References
Animal model
Montague et al. (1997), Strobel et al. (1998)
Animal model
Clement et al. (1998), Farooqi et al. (2007)
Animal model
Farooqi et al. (2006), Krude et al. (2003)
Animal model
Animal model
Dubern et al. (2007), Farooqi et al. (2003), Huszar et al. (1997), Vaisse et al. (2000) Gray et al. (2006), Han et al. (2008), Kernie et al. (2000), Rios et al. (2001) Yeo et al. (2004)
Animal model
Benzinou et al. (2008), Jackson et al. (1997)
Animal model
Severe early onset obesity; Prader–Willi-like (PWL) Cell-based in vitro study Holder et al. (2000), Hung et al. (2007) syndrome; developmental delay
understanding monogenic obesity is derived from animal models; the discovery of all candidate genes were made on rodent models, except for SIM1, of which the genetic aberrations were identified using in vitro cell lines (Holder et al. 2000). Contribution of genes involved in monogenic obesity to polygenic obesity in the general population is largely unknown. The arrival of NGS and the development of exome sequencing have offered a new approach to identify causal genes in monogenic diseases. This approach has been shown to be highly successful in identifying causal genes for monogenic diseases with hitherto unknown genetic etiologies (Ku et al. 2012). Furthermore, exome sequencing works well for small sample sizes without the need of a large pedigree. Thus, exome sequencing can be applied to individual cases from different families, and the variants identified are then filtered accordingly to identify the mutations causing monogenic diseases which are likely residing in protein. Various bioinformatic tools are also developed and used to predict the functional effect on the protein and evolutionary conservation. Although exome sequencing has yet to be applied in monogenic obesity, it is anticipated that it will achieve similar success as with other monogenic diseases. The presentation of clinical features such as early onset obesity, hyperphagia, hypoadrenalism, and low levels of leptin concentration despite severe obesity and hypothyroidism is useful to identify monogenic obesity cases for exome sequencing approach. The genes identified in monogenic obesity such as BDNF, MC4R, and PSCK1 have also been found to be strongly associated with obesity in GWAS, suggesting the importance of using exome sequencing to explore additional monogenic loci.
The focus on the single-gene defect in obesity as a molecular target for pharmacological interventions had shown to be effective. This is well exemplified by the development of leptin treatment. The LEP mutation is the first and the only genetic form of human obesity to be successfully treated pharmacologically (Hainerova and Lebl 2013). Leptin treatment results in weight loss that is mainly due to loss of fat mass in patients with leptin deficiency, but it has no therapeutic effect on those with LEPR deficiency (Farooqi et al. 2002). Eventually it became clear that most of these obese subjects are not leptin deficient; rather, they have high levels of leptin but are leptin resistant (Considine et al. 1996). Despite leptin is not an effective treatment for polygenic obesity, the reversal effect of leptin on hyperglycemia has provided a path for the development of other potential therapeutic entities for diabetes in the future (Perry et al. 2014).
Polygenic obesity Although studies on monogenic obesity have provided significant insights into the biology of obesity, particularly the severe phenotype, the genetic loci of polygenic obesity remain elusive until the GWAS era. Different genetic approaches such as candidate gene association studies, genome-wide linkage studies (GWLSs), and GWAS have been used to investigate the polygenic basis of obesity, of which the results are summarized and discussed as follows. Candidate gene studies Candidate gene association studies have been widely used to study obesity-susceptibility loci, but achieved limited
13
successes. This is a hypothesis-driven approach of which the loci/genes to be studied are selected based on a priori knowledge. Small sample sizes with limited statistical power have been used to identify associations with common variants with small effect sizes, which led to irreproducible results. To date, candidate gene studies have investigated various genes involved in energy balance regulation such as those encoding factors that regulate food intake and energy homeostasis. For example, LEP, LEPR, GHRL, CCK, NPY, CRH, POMC, AGRP, MC1R, MC3R, MC4R, MC5R, and CART have been commonly studied for their associations with obesity (Loktionov 2003). Genes involved in peripheral regulation of energy expenditure such as ADRB1, ADRB2, ADRB3, UCP2, and UCP3 have also been shown to be associated with obesity (Loktionov 2003). It is well known that the link between obesity and diabetes is based on insulin resistance. Despite this tight relationship, these conditions do not seem to share a common genetic background, as only a few common susceptibility loci have been discovered such as PPARG2, PGC-1, CEBPE, and GIPR (Grarup et al. 2014; Ling et al. 2004; Loktionov 2003; Oberkofler et al. 2004). These genes in the PPARs pathway have a regulatory role in lipid metabolism, glucose homeostasis, and adipogenesis (Barbier et al. 2003; Kim et al. 2011; Lehrke and Lazar 2005; Tontonoz and Spiegelman 2008). Genome-wide analysis has demonstrated that most of the genes induced in adipogenesis are bound by both PPARs and CEBPE (Lefterova et al. 2008). Although candidate gene studies have limited success in detecting risk variants, certain genes with known biological functions in obesity have provided a path to the development of therapeutic interventions in metabolic abnormalities. This was evidenced in the use of a PPAR ligand, thiazolidinedione (TZD), in type 2 diabetes to reduce insulin resistance (Gastaldelli et al. 2007). Currently, the use of TZD had been withdrawn due to its adverse effects; a safer new class of PPAR-targeted antidiabetic drugs is still being explored, however, this is the only antidiabetic agent available currently for increasing insulin sensitivity (Choi et al. 2011; Fuentes et al. 2013; Monsalve et al. 2013; Soccio et al. 2014). GWLS GWLS is a hypothesis-generating method, where the whole genome is interrogated in an agnostic approach (Altmuller et al. 2001). This requires a large pedigree to study the segregation patterns between the genotype and phenotype, and hence to identify the loci and mutations. However, this study design might not be applicable to individual cases without a multigenerational pedigree, and it is not robust to phenotypic and genetic heterogeneity.
13
Hum Genet
GWLSs have thus far identified many genetic loci associated with obesity but most were not reproducible. For example, this was shown in a meta-analysis of 37 genomewide linkage scans that reported no evidence of significant linkage to any of the chromosomal regions being studied (Saunders et al. 2007). PSCKI1 is the only region reported by linkage studies to be strongly linked to obesity to date (Benzinou et al. 2008). Attempts to identify novel loci using linkage mapping have been largely unsuccessful, although novel loci such as TBC1D22A and THUMPD2 have been recently reported, but further validation would be needed (Liu et al. 2014). GWAS In contrast to the candidate gene association study design, GWAS are a non-hypothesis-driven approach in which a large number of genetic markers (SNPs) spanning the entire genome are interrogated for their associations with the phenotypes of interest. This approach has been very successful in identifying common SNPs that contribute to a relatively low risk (as measured by odds ratio [OR]
REVIEW PAPER
Obesity and genomics: role of technology in unraveling the complex genetic architecture of obesity Yamunah Devi Apalasamy · Zahurin Mohamed
Received: 9 October 2014 / Accepted: 2 February 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract Obesity is a complex and multifactorial disease that occurs as a result of the interaction between “obesogenic” environmental factors and genetic components. Although the genetic component of obesity is clear from the heritability studies, the genetic basis remains largely elusive. Successes have been achieved in identifying the causal genes for monogenic obesity using animal models and linkage studies, but these approaches are not fruitful for polygenic obesity. The developments of genome-wide association approach have brought breakthrough discovery of genetic variants for polygenic obesity where tens of new susceptibility loci were identified. However, the common SNPs only accounted for a proportion of heritability. The arrival of NGS technologies and completion of 1000 Genomes Project have brought other new methods to dissect the genetic architecture of obesity, for example, the use of exome genotyping arrays and deep sequencing of candidate loci identified from GWAS to study rare variants. In this review, we summarize and discuss the developments of these genetic approaches in human obesity.
Introduction Obesity has reached epidemic proportions globally, with an estimation of more than 2.16 billion overweight people and 1.12 billion obese people by 2030 (Kelly et al. 2008). The World Health Organization reported that 65 % of the world’s population resides in countries in which obesity
Y. D. Apalasamy (*) · Z. Mohamed Department of Pharmacology, Pharmacogenomics Laboratory, Faculty of Medicine, University of Malaya, 50603 Kuala Lumpur, Malaysia e-mail: [email protected]
and overweight have led to death. As obesity is a major health and economic burden, the American Medical Association has considered it a disease in 2013 (American Medical Association 2013). It has a significant health burden, as obesity is also associated with other comorbidities such as diabetes, cardiovascular disease, liver and gallbladder disease, cancer, gynecological problems, sleep apnea, and osteoarthritis (Pi-Sunyer 2009). Obesity is a chronic condition in which excessive fat accumulates as a result of imbalance between energy consumption and expenditure (Galgani and Ravussin 2008). Interactions among environmental, genetic, and behavioral factors lead to a complex pathogenesis (Chaput et al. 2014). The modern lifestyle has partially led to a rapid increase in consumption of energy-dense food and physical inactivity. But even in the presence of an obesogenic environment, genetic components play an important role in contributing to individual risk of obesity. Twin, family, and adoption studies have shown that heritability of obesity is high, ranging from 40 to 70 % (Allison et al. 1996; Maes et al. 1997; Stunkard et al. 1990). Various genetic approaches such as linkage analysis and candidate gene association studies have been performed to identify susceptibility genetic loci in an attempt to dissect the biological mechanisms underlying body-weight regulation; however, these attempts have met with limited success. Study designs to dissect the genetic architecture of obesity have progressively shifted from linkage and candidate gene association studies to genome-wide association studies (GWAS) after 2005. In the early studies of obesity, in vivo, in vitro, linkage, and candidate gene association studies were employed to study obesity-susceptibility genes. The Human Genome Project and the International HapMap Project are the major developments which have significantly enhanced our knowledge on genetic variations in
13
the human genome. The human genome sequencing was completed in 2003, and the subsequent developments such as characterization of the LD patterns among the SNPs through the HapMap Project endeavor have led to the breakthrough discovery of the genetic variants of human diseases through GWAS (Schmutz et al. 2004). Rapid developments in technologies at a relatively low cost, such as high-throughput genotyping arrays and next generation sequencing (NGS) technologies, together with advanced statistical methods, have enabled GWAS and copy number variation (CNV) studies to identify numerous genetic loci associated with obesity (Naidoo et al. 2011). To date, GWAS have identified at least 58 novel loci associated with obesity/body mass index (BMI) in European and non-European populations (Lu and Loos 2013). The availability of well-curated Human Gene Mutation Databases has also facilitated large-scale meta-analyses of diseaseassociated variants These databases include the Human Gene Mutation Database, A Catalog of Published GWAS, Online Mendelian Inheritance in Man, ClinVar, and many other locus-specific mutation databases (Amberger et al. 2009; Sherry et al. 2001; Stenson et al. 2013). Nevertheless, the SNPs that have been identified thus far collectively explain only a relatively small proportion of the heritability. Unraveling the complex genetic architecture of obesity remains a challenging task for researchers. In this review, we summarize recent advances in genetic studies of obesity. We also discuss various technological approaches used to dissect the genetics of obesity, especially those that leverage high-throughput genotyping and NGS technologies.
In vivo and in vitro models of obesity Molecular mechanisms of body-weight regulation and obesity development were first discovered through animal models (Zhang et al. 1994; Zucker 1975). Mouse models are critical in obesity research in pre- and post-GWAS era, as reviewed in recent articles (Cox and Church 2011; Speakman et al. 2008). The quantitative genetic approach with mice models offers an excellent tool to study the genetic architecture of obesity. Studies in rodent genetic models to understand the pathways of body-weight regulation and feeding behavior have been useful in demonstrating the monogenic type of obesity (Leibel et al. 1997). For example, the localization of mutant genes in animal model has led to the discovery of monogenic genes such as leptin, a mutation in the LEP gene causing obesity in the ob/ob mouse (Zhang et al. 1994). Apart from single-gene mutations, fine-mapping of quantitative trait loci in mouse models also unraveled more than 200 novel targets related to obesity for candidate gene studies (Snyder et al. 2004).
13
Hum Genet
Polygenic animal models have been crucial in various studies of environmental effects such as epigenetics, responses to high-fat or sugar, and low-calorie and other weightloss diets, as well as in the development of pharmaceutical agents such as leptin and cannabinoid receptor type 1 antagonist agents for treating obesity (Speakman et al. 2008). In addition, adipogenic cellular models have been useful in understanding adipogenesis, energy homeostasis, and insulin resistance in obesity. In vitro approaches such as primary cell cultures, coculture models, and three-dimensional cell cultures have also shed light on the important components of obesity such as growth factors, hormones, and potential pharmacological compounds (Armani et al. 2010; Green and Meuth 1974). For instance, in vitro studies have shown that PPARG is extensively regulated by CCAAT/ enhancer-binding proteins, which subsequently provide a path towards the development of PPARG in the treatment of obesity and type 2 diabetes (Mikkelsen et al. 2010). Notably, in vivo and in vitro models are important in the study of genetics for both monogenic and polygenic obesity (Nilsson et al. 2012). Monogenic obesity is caused by a single-gene mutation, whereas polygenic obesity is associated with multiple genetic factors that interact with an “at-risk” environment (Mutch and Clement 2006). Despite the success in terms of development of some therapeutic targets, there are limited successes in other instances in translating the molecular compounds discovered by in vivo and in vitro studies into usable drugs for humans. For example, treatments of rodents with leptin and MC4R (MTII) antagonists resulted in weight loss, but these compounds were shown to be not clinically useful because they were short acting and also the occurrence of adverse effects (Gibson et al. 2004; Hadley 2005). The major shortcoming of the currently available in vitro and in vivo models is that these models only resemble certain aspects of human obesity; therefore, it would be useful if animal- and cell-based models of therapeutic approaches could be improved in future to reflect the pathophysiology of obesity in humans.
Monogenic obesity Monogenic obesity refers to a rare form of severe obesity that resulted from mutations with a large effect size. These mutations are known to cause severe early onset obesity with hyperphagia as a key feature and multiple endocrine anomalies (Table 1) (Clement 2006). To date, eight candidate genes (LEP, LEPR, PSCK1, POMC, MC4R, SIM1, BDNF, and NTRK2) listed in Table 1, have been identified for monogenic obesity, which have shed light into the biological pathways of the disease. Considerable success in
Hum Genet Table 1 Genes in monogenic obesity Gene
Main clinical characteristics
Severe early onset of obesity; hypogonadism; very low circulating leptin levels LEPR Severe early onset of obesity; hypogonadism; very low circulating leptin levels POMC Severe obesity; hyperphagia; hypoadrenalism; hypopigmentation MC4R Severe early onset of hyperphagia; rapid rise in fat mass BDNF Severe obesity; hyperphagia; hyperactivity; impaired cognitive function; developmental delay NTRK2 Severe obesity; hyperphagia; hyperactivity; impaired cognitive function; developmental delay PSCK1 Hyperphagia; hypogonadism; hyperglycemia; elevation in proinsulin to insulin ratio
LEP
SIM1
Discovery method
References
Animal model
Montague et al. (1997), Strobel et al. (1998)
Animal model
Clement et al. (1998), Farooqi et al. (2007)
Animal model
Farooqi et al. (2006), Krude et al. (2003)
Animal model
Animal model
Dubern et al. (2007), Farooqi et al. (2003), Huszar et al. (1997), Vaisse et al. (2000) Gray et al. (2006), Han et al. (2008), Kernie et al. (2000), Rios et al. (2001) Yeo et al. (2004)
Animal model
Benzinou et al. (2008), Jackson et al. (1997)
Animal model
Severe early onset obesity; Prader–Willi-like (PWL) Cell-based in vitro study Holder et al. (2000), Hung et al. (2007) syndrome; developmental delay
understanding monogenic obesity is derived from animal models; the discovery of all candidate genes were made on rodent models, except for SIM1, of which the genetic aberrations were identified using in vitro cell lines (Holder et al. 2000). Contribution of genes involved in monogenic obesity to polygenic obesity in the general population is largely unknown. The arrival of NGS and the development of exome sequencing have offered a new approach to identify causal genes in monogenic diseases. This approach has been shown to be highly successful in identifying causal genes for monogenic diseases with hitherto unknown genetic etiologies (Ku et al. 2012). Furthermore, exome sequencing works well for small sample sizes without the need of a large pedigree. Thus, exome sequencing can be applied to individual cases from different families, and the variants identified are then filtered accordingly to identify the mutations causing monogenic diseases which are likely residing in protein. Various bioinformatic tools are also developed and used to predict the functional effect on the protein and evolutionary conservation. Although exome sequencing has yet to be applied in monogenic obesity, it is anticipated that it will achieve similar success as with other monogenic diseases. The presentation of clinical features such as early onset obesity, hyperphagia, hypoadrenalism, and low levels of leptin concentration despite severe obesity and hypothyroidism is useful to identify monogenic obesity cases for exome sequencing approach. The genes identified in monogenic obesity such as BDNF, MC4R, and PSCK1 have also been found to be strongly associated with obesity in GWAS, suggesting the importance of using exome sequencing to explore additional monogenic loci.
The focus on the single-gene defect in obesity as a molecular target for pharmacological interventions had shown to be effective. This is well exemplified by the development of leptin treatment. The LEP mutation is the first and the only genetic form of human obesity to be successfully treated pharmacologically (Hainerova and Lebl 2013). Leptin treatment results in weight loss that is mainly due to loss of fat mass in patients with leptin deficiency, but it has no therapeutic effect on those with LEPR deficiency (Farooqi et al. 2002). Eventually it became clear that most of these obese subjects are not leptin deficient; rather, they have high levels of leptin but are leptin resistant (Considine et al. 1996). Despite leptin is not an effective treatment for polygenic obesity, the reversal effect of leptin on hyperglycemia has provided a path for the development of other potential therapeutic entities for diabetes in the future (Perry et al. 2014).
Polygenic obesity Although studies on monogenic obesity have provided significant insights into the biology of obesity, particularly the severe phenotype, the genetic loci of polygenic obesity remain elusive until the GWAS era. Different genetic approaches such as candidate gene association studies, genome-wide linkage studies (GWLSs), and GWAS have been used to investigate the polygenic basis of obesity, of which the results are summarized and discussed as follows. Candidate gene studies Candidate gene association studies have been widely used to study obesity-susceptibility loci, but achieved limited
13
successes. This is a hypothesis-driven approach of which the loci/genes to be studied are selected based on a priori knowledge. Small sample sizes with limited statistical power have been used to identify associations with common variants with small effect sizes, which led to irreproducible results. To date, candidate gene studies have investigated various genes involved in energy balance regulation such as those encoding factors that regulate food intake and energy homeostasis. For example, LEP, LEPR, GHRL, CCK, NPY, CRH, POMC, AGRP, MC1R, MC3R, MC4R, MC5R, and CART have been commonly studied for their associations with obesity (Loktionov 2003). Genes involved in peripheral regulation of energy expenditure such as ADRB1, ADRB2, ADRB3, UCP2, and UCP3 have also been shown to be associated with obesity (Loktionov 2003). It is well known that the link between obesity and diabetes is based on insulin resistance. Despite this tight relationship, these conditions do not seem to share a common genetic background, as only a few common susceptibility loci have been discovered such as PPARG2, PGC-1, CEBPE, and GIPR (Grarup et al. 2014; Ling et al. 2004; Loktionov 2003; Oberkofler et al. 2004). These genes in the PPARs pathway have a regulatory role in lipid metabolism, glucose homeostasis, and adipogenesis (Barbier et al. 2003; Kim et al. 2011; Lehrke and Lazar 2005; Tontonoz and Spiegelman 2008). Genome-wide analysis has demonstrated that most of the genes induced in adipogenesis are bound by both PPARs and CEBPE (Lefterova et al. 2008). Although candidate gene studies have limited success in detecting risk variants, certain genes with known biological functions in obesity have provided a path to the development of therapeutic interventions in metabolic abnormalities. This was evidenced in the use of a PPAR ligand, thiazolidinedione (TZD), in type 2 diabetes to reduce insulin resistance (Gastaldelli et al. 2007). Currently, the use of TZD had been withdrawn due to its adverse effects; a safer new class of PPAR-targeted antidiabetic drugs is still being explored, however, this is the only antidiabetic agent available currently for increasing insulin sensitivity (Choi et al. 2011; Fuentes et al. 2013; Monsalve et al. 2013; Soccio et al. 2014). GWLS GWLS is a hypothesis-generating method, where the whole genome is interrogated in an agnostic approach (Altmuller et al. 2001). This requires a large pedigree to study the segregation patterns between the genotype and phenotype, and hence to identify the loci and mutations. However, this study design might not be applicable to individual cases without a multigenerational pedigree, and it is not robust to phenotypic and genetic heterogeneity.
13
Hum Genet
GWLSs have thus far identified many genetic loci associated with obesity but most were not reproducible. For example, this was shown in a meta-analysis of 37 genomewide linkage scans that reported no evidence of significant linkage to any of the chromosomal regions being studied (Saunders et al. 2007). PSCKI1 is the only region reported by linkage studies to be strongly linked to obesity to date (Benzinou et al. 2008). Attempts to identify novel loci using linkage mapping have been largely unsuccessful, although novel loci such as TBC1D22A and THUMPD2 have been recently reported, but further validation would be needed (Liu et al. 2014). GWAS In contrast to the candidate gene association study design, GWAS are a non-hypothesis-driven approach in which a large number of genetic markers (SNPs) spanning the entire genome are interrogated for their associations with the phenotypes of interest. This approach has been very successful in identifying common SNPs that contribute to a relatively low risk (as measured by odds ratio [OR]
