A Case-Control Study of the Association of Leptin Gene Polymorphisms with Plasma Leptin Levels and Obesity in the Kerala Population

Background Over the last few years, the importance of leptin in energy metabolism has been extensively studied in both animal models and in humans. Very few results are available on the association between human leptin gene (LEP) variants and obesity traits in India. We designed this study to analyse the polymorphisms in human leptin gene and the association of sequence variants with obesity among the population in Kerala, South India. Methods In this case-control design of 148 study participants, data were collected on socioeconomic aspects and anthropometric measurements. Plasma glucose, insulin, leptin, and lipid profile were measured. Genotyping was done by automated DNA sequencing. Results The common Single Nucleotide Polymorphism (SNP) of 5′-UTR of LEP − 2548G/A was found to be present in the study population with “A” variant as dominant allele. A novel synonymous mutation Thr5Thr of exon 2 of LEP was identified in heterozygous form in one subject with morbid obesity with hyperleptinemia. A novel missense mutation Phe17Leu was observed in two subjects with obesity in heterozygous condition. A novel missense mutation Lys36Arg in exon 2 of LEP was observed in one subject with abdominal obesity and decreased serum leptin level. Conclusion LEP − 2548G/A at 5′-untranslated region was found to be common with the mutant “A” variant in the study population. SNPs of exons in LEP were found to be rare but associated with morbid obesity and altered levels of serum leptin in the study population in Kerala, India.


Introduction
Obesity, a systemic disease due to excessive and abnormal accumulation of body fat leading to adverse health efects, has become a major public health problem in India, especially in Kerala. Obesity is associated with higher rate Heritable factors account for approximately 70% of the cases with increased BMI in adult life [2,3].
Leptin is a major adipokine that acts as a hormone to signal the brain to regulate the nutritional status by infuencing feeding behaviour, metabolism, and energy balance. Leptin gene is located on chromosome 7q32.1, spanning approximately 20 kb with three exons separated by two introns [4]. Te frst intron is approximately 10.6 kb long and is located in the 5′-untranslated region, 29 bp upstream of the start codon. Te coding region of LEP is 501 nucleotides long and is located mainly in exons two and three, which are separated by an intron of approximately 2.3 kb in size. Te third exon consists of the downstream coding region and the 3′-untranslated region of the human LEP gene.
Over the last few years, the importance of leptin in energy metabolism has been extensively studied in both animal models as well as in humans. Several polymorphisms of LEP were studied in diferent populations for their potential association with obesity and metabolic complications [5,6]. Among these polymorphisms, LEP−2548G/A and LEP 19G/A have been studied in detail [7,8]. Tese fndings were replicated across diferent populations, revealing contradictory fndings [9,10]. A few reports are available regarding the association between human leptin gene variants and obesity traits in India [11][12][13]. Te state of Kerala stands next to Punjab in the second position in obesity prevalence in India. However, literature regarding the association between obesity and leptin gene polymorphisms among the predominantly Dravidian population of Kerala is very scarce. Tis study was designed to analyse the polymorphisms in human leptin gene and to study the association of variants with the incidence of obesity among the population in Kerala.

Methods
Te study was approved by Institutional Human Ethics Committees of Sree Gokulam Medical College and Research Foundation, Tiruvananthapuram, Govt. Medical College, Tiruvananthapuram and Rajiv Gandhi Centre for Biotechnology. A detailed written informed consent was obtained from each participant before they could enter the study and the data were kept confdential. Study subjects with a BMI ≥ 25, with no history of any prior medication were considered as cases (obesity group). Te subjects with BMI ≤ 24.9, which included normal and overweight individuals, were considered as controls. Te study subjects were recruited prospectively. All study subjects were Malayalam speaking Dravidian population group from the state of Kerala in India. Tey belonged to various religions and subcommunities. Subjects with a history of drug intake afecting the study variables (BMI, blood glucose, insulin, lipid profle, and so on) were excluded from the study. Te subjects under the age of 20 years, pregnant and postpartum women, those with polycystic ovarian diseases (PCODs) and thyroid diseases, and those taking oral contraceptive pills were excluded from the study. All the participants answered a life style questionnaire. Information about age, gender, race, socioeconomic factors, lifestyle habits, and family history of diseases was gathered using questionnaire.
We collected the blood samples from the study participants in the setting of the Obesity Clinic, Govt. Medical College, Tiruvananthapuram, Kerala, and Sree Gokulam Medical College and Research Foundation, Tiruvananthapuram, Kerala, for biochemical and gene sequencing analysis. ELISA experiments were done at Sree Gokulam Medical College and Research Foundation and genotyping by direct DNA sequencing at Rajiv Gandhi Centre for Biotechnology, Tiruvananthapuram. Even though 150 study subjects were selected for the study, sample size for the diferent SNPs that were examined difered and was fewer than that of the entire study due to variation in the availability of high quality DNA and incomplete successful genotyping.
10 ml of blood was collected by venepuncture into sterile vacutainer tubes with or without EDTA. Blood sugar and lipid profle were done on the same day of sample collection in a Siemens Dimension autoanalyzer using Flex reagent cartridges (Siemens Healthcare Diagnostics Ltd., Camberley, UK). Te serum and plasma were stored in aliquots of 1 ml in a deep freezer at −80°C for future analysis of leptin [14] and insulin [15] by the ELISA method.

Selection of Single Nucleotide Polymorphisms (SNPs) of
Leptin Gene. SNPs in leptin gene, which showed an association with obesity phenotypes, were identifed by literature survey 5 −13 . For selecting the SNPs for the present study, we used data from the International HapMap Project (https:// www.hapmap.org), National Centre for Biotechnology Information (https://www.ncbi.nlm.nih.gov), and Gene Cards (https://www.genecards.org). Eight SNPs were selected with LEP − 2548A/G as the most studied SNP located in the promoter region, 2 SNPs located at the 5′-untranslated region with a minimum Minor Allele Frequency (MAF) of 0.2, and 5 SNPs in the exons.  (Table 1). PCR primers were designed using Primer 3 software and Primer BLAST software based on the known reference sequence information from NCBI Gen-Bank. Peripheral blood mononuclear cells (PBMCs) were isolated from 2 ml of freshly collected blood collected in EDTA tubes by using Histopaque columns [16]. Genomic DNA was extracted from PBMC by the chemical extraction method [17,18] using extraction bufer (1% SDS in 0.5 M NaCl), isopropanol, 70% ethanol, and Tris-EDTA bufer. PBMC was suspended in extraction bufer and sonicated mid-range 3 cycles of 10 s each. After centrifuging for 4 min, the supernatant was transferred to new microfuge tube, an equal volume of isopropanol was added, and the contents were mixed gently by inversion. Te DNA pellet was washed with 70% ethanol. Te pellet was allowed to air dry, and the Journal of Obesity 3 DNA was dissolved in 50 μl double distilled water. Te concentration of DNA was measured spectrophotometrically at 260 nm using Nanodrop spectrophotometer (Termo Fisher Scientifc, Inc.), and the purity was assessed using the same by calculating the 260/280 nm ratio. PCR was carried out individually for the eight SNPs being analysed. Te primer pairs (Sigma Aldrich, Bangalore, India) used in this study are listed in Table 1. Each PCR reaction was composed of 1 μg of the genomic DNA template, 10X reaction bufer containing 1.5 mM MgCl 2 , 0.2 mM dNTPs (New England Biolabs, MA, USA), 0.5 units thermostable Taq polymerase (New England Biolabs, MA, USA), and 1 μM each forward and reverse primer in a 22 μl reaction mixture. Tis was followed by direct DNA sequencing on Applied Biosystem 3730 series capillary automated sequencer (AB Applied Biosystems, CA, USA) using the Big Dye Terminator v3.1 Cycle sequencing kit. Te sequencing PCR reaction was composed of 1 μl primer (Forward or Reverse primer), 4 μl 5X sequencing bufer, 1 μl Big Dye Mix(AB Applied Biosystems, CA, USA), and 3 μl template in a 20 μl reaction mixture. Te sequencing PCR products were precipitated using 2 ml sodium acetate bufer and washed with 70 μl 100% ethanol. Te tubes were kept in ice for 30 min and then centrifuged at 18000 × g for 20 min, and the pellet was washed with 70% ethanol. Te tubes were kept in an inverted position over night before subjecting to automated DNA sequencing. Te sequences were analysed by NCBI nucleotide-nucleotide BLAST (https://blast.ncbi.nlm. nih.gov/Blast.cgi). All sequence variants were determined and verifed from the derived DNA sequence as well as by the visual inspection of the electropherogram. Te results were analysed by NCBI BLAST sequence analysis initially and then by doing Multiple Sequence Alignment (MSA) of all samples.

Statistical Analysis.
Data were analysed using Statistical Package for Social Sciences (SPSS) for Windows version 17.0 (SPSS Inc, Chicago, IL, USA). Allele frequencies were estimated by the gene-counting method, and the distribution of genotypes was tested for Hardy-Weinberg equilibrium and compared using 2 × 2 contingency tables. Testing for Hardy-Weinberg equilibrium and the signifcance of allelic and genotype distributions between the groups were performed by the chi-square test. To test an association between each SNP variant and phenotypes of adiposity and associated risk factors, the overall genotype test of association and genetic models (dominant, recessive, additive, allele, and homozygote) were computed. Te association of phenotype traits of obesity with genotypes was analysed by Analysis of Variance (ANOVA). Te statistical signifcance was defned by P < 0.05.

Results
LEP − 2548A variant was found to be common in the study population in Kerala with a frequency of 66%. About 30% of control group and 33% of obesity group were homozygous with AA genotype ( Table 2). A representative electropherogram showing −2548G/A variants in the study subjects is shown in Figures 1(a)-1(f ). Te mean serum leptin level in the obesity group with dominant homozygous AA genotype (56.6 ng/ml) was found to be high when compared with other obesity groups with AG (42.8 ng/ml) and GG (49.3 ng/ml) genotypes (Table S1). No statistically signifcant diference in the anthropometry measures and serum leptin was observed among the obesity groups with diferent genotypes (Table S1). Separate analysis by gender showed that the frequency of "A" allele was same in both genders (67%) in the study population in Kerala (Table S2). Frequency of "A" allele in both obesity and control group of men was found to be 65%, while in women, obesity group had a frequency of 69% and control group had 66%. Te frequency of AA genotype in obesity group was 31% while that of control group 26%.
LEP rs28954080 (−633C/T) at 5′ UTR is one of the leptin gene SNPs that played signifcant role in obesity in diferent populations during literature review. However, this SNP was found to be monomorphic with C allele in our study population while it was reported to have a global minor allele frequency of T � 0.036. On gene sequencing analysis, a novel SNP rs570757178 at 5′UTR of LEP with C to G transition about 90 base pair upstream of the reported SNP rs28954080 was observed in our study. Two subjects with obesity (with identifcation number P291 and P288) and two overweight (C063 and C155) individuals in the study population were found to be heterozygous carriers with CG genotype (Figures 1(g)-1(l)). Low serum leptin level was observed in one obesity subject (P291) (BMI 26.3 kg/m 2 with serum leptin 7.41 ng/ml) and one overweight subject (C155) (BMI 23.4 kg/m 2 with serum leptin 4.36 ng/ml) ( Table 3). Te decrease in serum leptin levels in these two subjects was found to be statistically signifcant in comparison with the other heterozygotes. Te WHR of these two subjects was found to be high in these two subjects, which refects the association of this SNP with abdominal obesity. No family history of obesity was observed with these heterozygotes of LEP rs570757178C/G.
Another Novel SNP rs6976701 of LEP 5′-UTR was observed with its heterozygous genotype (AG) in fve subjects with obesity (P055, P203, P286, P288, and P289) and homozygous recessive genotype (AA) in one subject with obesity (P292) (Figures 2(a)-2(h)). All other study subjects were found to have dominant homozygous GG genotype. Te recessive allele "A" was found to be associated with obesity. Te homozygote with recessive genotype AA was found to have strong family history of obesity. Of the fve heterozygote (AG) subjects, two were observed with a strong family history of obesity (Table 3).
For LEP exon 1 SNP + 19G/A, no recessive homozygotes were found in the present study population (Table 2). Te frequencies of homozygous dominant and heterozygous genotypes were found to be similar in both control and obesity groups. Te minor allele frequency of "A" variant was found to be similar in both groups (33%). Te recessive "A" variant was found to be strongly associated with adiposity measures such as BMI, hip circumference, and body fat in both control and obesity groups (Table 4). In the control group, a signifcant increase in waist circumference, which is a measure of abdominal adiposity, was observed in individuals with +19A variant when compared with those of +19G allele (P � 0.002). Among the obesity group, the increase in waist to hip ratio (WHR) in individuals with mutant "A" allele is found to be statistically signifcant in comparison with those having "G" allele (P � 0.001). Te mean serum leptin level showed a strong association with LEP + 19 A variant when compared to dominant "G" allele in both control (P � 0.01) and obesity groups (P � 0.004).
No missense mutation at LEP exon 2 + 45A/G (rs145044661) was observed in the study population. Only  Journal of Obesity the ancestral variant "A" is present in the study population. Homozygous dominant genotype AA was observed in all case and control subjects. However, novel mutations were observed near this SNP in exon 2 viz, Tr5Tr, Phe17Leu, and Lys36Arg. A novel synonymous mutation Tr5Tr (rs142904532) in exon 2 of LEP was found to be associated with morbid obesity. In this study population, only one individual with obesity (ID number P184) had mutant allele "G" in heterozygous condition (Figures 2(i) and 2(j)). Tis study subject had a BMI of 44.8 kg/m 2 and the serum leptin level of 137 ng/µl (Table 3). Father and mother of this study subject were reported to have obesity.
A novel missense mutation Phe17Leu (LEP rs201067336) was found to be linked with obesity in the study population. In this study population, only two subjects' obesity (P169 and P202) had mutant "G" allele in heterozygous condition (Figures 2(k)-2(n)). Serum leptin levels of these two subjects were found to be abnormal (Table 3). However, no family history of obesity was reported with this subject.
Te novel missense mutation Lys36Arg (LEP rs111650508) was observed in one subject with obesity (P171) (Figures 2(o)-2(r)), and this subject was found to be associated with abdominal obesity, decreased serum leptin level, and family history of obesity (Table 3).
In this study population, no missense mutation +103C/A or +103C/G or Asn103Lys or synonymous mutation +103C/ T (Asn103Asn) was observed. However, new sequence variants were observed near this region in the exon 3 of LEP. Analysis of rs28954114 (+328G/A) in the 3′-UTR of LEP revealed that the mutant allele A was observed in one subject with obesity (P076) and two control subjects (C145 and C248) in heterozygous condition (Figures 3(a)-3(f )). All other subjects had the dominant allele in homozygous form.
Te subject with +328A variant in heterozygous condition is reported to have family history of obesity and high serum leptin level (Table 3). Te two control subjects with mutant A allele in heterozygous condition had normal serum leptin level.

Discussion
LEP − 2548G/A (rs7799039) is one of the most studied common SNPs of human leptin gene (LEP) in diferent populations [5,8,19]. LEP − 2548G/A is a G to A transition at nucleotide position 2548 bp upstream of the ATG start site in the 5′ promoter region of the leptin gene. In the present study, frequency of the "A" allele was found to be 66% in control group and 67% in obesity group among the population in Kerala. Li et al. reported the higher frequency of variant "A" allele in Asians at least among Malaysia/Peninsular Bumiputras, Chinese, and Indians [20]. Mammès et al. reported the strong association of G allele with overweight [21] while in our study population, the "A" allele was found to be in association with obesity. Te inconsistency in various reports from diferent populations maybe due to diferent ethnicity, characteristics of subjects, or diverse covariate factors. Te dominant AA genotype was found to be more common in this study population in Kerala with a frequency of 30% in control and 33% in obesity group ( Table 2). Te people with obesity having AA genotype showed an increase in the mean serum leptin level in comparison with AG and GG genotypes (Supplementary Table 1). Tis might explain the infuence of the −2548A variant on the rate of transcription of leptin gene resulting in an increased level of plasma leptin. Hofstedt et al. reported an increase in DNA binding afnity with the presence of allele A at −2548 position, which can infuence leptin expression at the transcriptional level [22]. Hyperleptinemia Study subjects identifcation number is given with "C" or "P" as prefx. "C" represents control and "P" case (people with obesity). 6 Journal of Obesity Journal of Obesity 7 [21,22] and hypoleptinemia [23] have been reported in individuals carrying homozygous AA genotype by diferent studies. In this study population, an increase in the mean serum leptin level was observed in obesity group with dominant AA genotype (LEP − 2548G/A) (Supplementary Table 1). Various studies in diferent populations reported that there was no association of −2548G/A variants with plasma leptin levels [24,25]. Tere were inconsistent reports on its efect on leptin levels with the AA genotype in Indians [6], while the G allele in Mexicans has been reported in association with hyperleptinemia [26]. Studies reported that a functional binding site for an important transcription factor involved in adipocyte differentiation named CCAAT/enhancer binding protein α (C/ EBP-α) can also positively regulate the leptin gene promoter region [27,28]. Gong et al. reported that −2548G/A polymorphism is close to a SP-1 transcription factor binding site as well as two repetitive sequences WER-11 and Alu that may regulate leptin gene transcription [29]. DNA binding afnity was found to be increased with the presence of nucleotide "A" at −2548 position, which can infuence leptin expression at the transcriptional level, and thus the serum leptin level might be increased [21].
Tere was no association of LEP − 2548A variant with adiposity measures such as BMI, waist circumference, WHR, and body fat percentage. In our study population, no association of LEP-2548 A variant was observed with adiposity measures such as BMI, waist circumference, WHR, and body fat percentage. Tis is in agreement with Romanian and Polish subjects [30,31]. No signifcant diference in serum lipoprotein levels was observed in this study population. A recent study in a Pakistani cohort reported the association of −2548A variant with HDL cholesterol level [32].
We analysed another SNP at 5′ UTR of LEP rs28954080 (−633C/T), which was reported to have a global minor allele frequency of T � 0.036. However, in our study population, these loci were found to be monomorphic with C allele. We observed a novel SNP rs570757178 at 5′UTR of LEP with C to G transition about 90 base pair upstream of the reported SNP rs28954080. Two individuals with obesity and two overweight individuals in the study population were heterozygous carriers with CG genotype. Low serum leptin level was observed in two individuals. Another new SNP rs6976701 was observed at 5′UTR of LEP with G to A transition about 85 base pair upstream of the reported SNP rs28954080. Five subjects with obesity were heterozygous carriers of GA genotype and one subject with obesity with recessive AA genotype. "A" variant was found to be associated with hyperleptinemia.
Te common SNP + 19G/A (rs2167270) in the untranslated exon 1 of LEP showed a similar allele frequency in both obesity and control groups. Te LEP A19G variant (rs2167270) is a single base transition from A to G at nucleotide position 19 in the 5′ untranslated region (UTR) of exon 1 of the LEP gene [33]. Tis study observed a strong association between the LEP 19A variant with obesity related variables such as BMI, waist circumference, WHR, and body fat percentage in both obesity and control groups. However, most previous studies in diferent populations did not fnd such an association [10,24,34,35]. In our study population, a strong association of LEP 19A variant with serum leptin levels was observed, which is not in agreement with previous studies [10,24].
A novel synonymous mutation rs142904532 (Tr5Tr) of exon 2 of LEP locates at chromosome 7 : 128252033 was observed in one subject with obesity. Tis mutation in heterozygous form was associated with morbid obesity with very high serum leptin level.
Te missense mutation Phe17Leu (rs201067336) located on exon 2 of LEP (at chromosome 7:128252067) with the ancestral allele T was observed in two obese subjects in heterozygous condition. Te allele T to G transition leads to missense mutation (Phe17Leu). Te codon TTT codes for Phenyl alanine and TTG codes for Leucine. Of these two heterozygous individuals, one had low leptin level while the other had very high leptin level.
Novel missense mutation Lys36Arg (rs111650508) in exon 2 of LEP is observed in one subject with obesity in the study population. Te single base substitution from A to G, resulted in change of the 36 th residue of leptin from Lysine to Arginine. Te triplet codon AAG codes for Lysine and AGG codes for Arginine. Te obese subject (BMI 26.9 kg/m 2 ) with missense mutation Lys36Arg was associated with abdominal obesity with waist circumference of 104 cm and waist to hip ratio of 1.0. Serum leptin level was 6.6 µg/ml. Previous studies reported that mutations in the leptin gene cause  Journal of Obesity severe obesity and may also contribute to the complications associated with obesity [36,37]. SNP rs28954114 (+328G/A) in the 3′-UTR of LEP was observed in one subject with obesity and two control subjects in heterozygous condition. Te subject with obesity having "A" variant in heterozygous condition had increased serum leptin level with family history of obesity while the control subjects with mutant "A" allele in heterozygous condition had normal serum leptin level.
Since the LEP − 2548G/A region falls within the functional transcription factor binding sites, this polymorphism might afect the transcript levels of LEP and could account the diference in energy regulation feedback systems in diferent individuals. In our study, rare SNPs of mutations Tr5Tr, Phe17Leu, and Lys36Arg were detected and were found to be associated with low leptin levels and morbid obesity. Mutations of the exon regions of leptin gene generally cause leptin defciency due to the formation of leptin protein with altered structure and thus the function due to defect in secretion, transport, and binding with specifc receptors in target tissues [38]. Te change in amino acid (missense mutation) during synthesis results in drastic reduction in the biological activity of the mutant leptin protein. Such mutations could play a signifcant role in the development of morbid obesity. Te detection of such mutations in those subjects with strong family history of obesity or those subjects who are more prone to be obese could be helpful to indicate the genetic predisposition to obesity so that the incidence of obesity can be prevented or delayed by proper lifestyle modifcations such as increasing the physical activity and diet control.
Increased prevalence of obesity seems to be a consequence of modern life style habits with access to excess calorie rich food plus limited physical activity. However, with the same environment, individuals respond diferently. Our study could account the mechanism of such diferent responses in gaining weight or being healthy in the same energy rich environment that this may be largely due to genetic variation between individuals. Te genetic variation and environmental factors interact to cause obesity. Te genetic variations could explain why are the intervention based on diet and exercise more efective for some people only. More than that detection of such SNPs in people who are more susceptible to obesity may be benefcial for delaying or preventing this condition by adopting dietary restriction and increased physical activity.

Conclusion
DNA sequencing of the leptin gene identifed 3 genetic variants in the 5′ region of LEP; two of them, viz., −2548G/A (rs7799039) and rs6976701 were found to be associated with hyperleptinemia indicating a key role in the regulation of rate of transcription. Te −2548 G/A is found to be a common polymorphism associated with obesity among the population in Kerala. Mutations in the exons of leptin gene were found to be rare but associated with morbid obesity and altered levels of serum leptin.

Data Availability
All datasets and materials used and/or analysed during this study can be obtained from Sudharmadevi K. Manju upon reasonable request by emailing at manjuromel@gmail.com.

Additional Points
We shall address the limitation such as small sample size utilizing only 8 SNPs.

Conflicts of Interest
Te authors declare that there are no conficts of interest.