Galanin Preproprotein Is Associated With Elevated Plasma Triglycerides
Objective— There is increasing physiological evidence in rodents connecting the neuropeptide galanin to triglyceride (TG) levels. We hypothesized that variation in the galanin preproprotein (GAL) gene may contribute to hypertriglyceridemia (HTG) in humans.
Methods and Results— We investigated GAL as a TG candidate gene by genotyping 4 tagSNPs in Dutch, Finnish, and Mexican familial combined hyperlipidemia (FCHL) families as well as in white combined hyperlipidemia cases/controls (n=2471). The common allele of rs2187331, residing in the promoter region of GAL, was significantly associated with HTG (probability value=0.00038). In an unascertained population sample of 4463 Finnish males, the rare allele of rs2187331 was associated with higher TGs (probability value=0.0028 to 0.00016). We also observed an allele specific difference with rs2187331 in reporter gene expression and nuclear factor binding in vitro. Furthermore, we detected differential expression of many key lipid genes in adipose tissue based on rs2187331 genotypes.
Conclusions— The SNP rs2187331 is associated with HTG in FCHL and white combined hyperlipidemia cases/controls and influences TG levels in the population. Further studies are warranted to elucidate the allelic difference observed between FCHL and the general population. Functional evidence shows that rs2187331 has an allele specific cis-regulatory function and influences the expression of lipid related genes in adipose.
Elevated plasma lipid levels are key risk factors for cardiovascular disease. Recent genome-wide association studies (GWAS) have been able to account for 5% to 8% of the variation in the levels of low density lipoprotein (LDL), high density lipoprotein (HDL), and triglycerides (TGs).1 This suggests that there are other genetic variants affecting plasma lipid levels still to be discovered.
Familial combined hyperlipidemia (FCHL; MIM:144250) is the most common dyslipidemia observed in patients with coronary artery disease (CAD) and their relatives. FCHL is characterized by familial segregation of elevated plasma triglycerides (TGs), total cholesterol (TC), or both.2 Another common characteristic of FCHL is postprandial lipemia, the accumulation of triglyceride-rich lipoprotein particles (chylomicron, very low-density lipoprotein [VLDL] and their remnants) in plasma approximately 6 to 10 hours after a fatty meal.
A recent study demonstrated a correlation between plasma TGs and transcript levels of the neuropeptide galanin preproprotein (GAL) in the paraventricular nucleus (PVN) of the hypothalamus in rats fed a high-fat diet.3 Furthermore, injection of GAL peptide directly onto the PVN of rats caused a significant upregulation of lipoprotein lipase (LPL) expression in adipose tissue and a significant decrease in LPL expression of skeletal muscle.3,4 As many as one third of FCHL patients are estimated to exhibit decreased LPL activity,5 suggesting that perturbations in LPL may contribute to FCHL.6
We hypothesized that GAL is a compelling functional candidate gene for predisposition to elevated TGs in individuals with FCHL. Coincidentally, GAL is located near the peak marker linked to TGs and FCHL in Dutch and British FCHL families on chromosome 11p.7,8 We selected tagSNPs capturing the common genetic variation of GAL and tested these SNPs for association with hypertriglyceridemia (HTG) in study samples ascertained for hyperlipidemia and with TG levels at the population level. We also tested for allelic effects on reporter gene expression, nuclear factor binding, and downstream differential expression of genes including LPL. Our data suggest that variation in the GAL promoter region influences TG levels in humans and that the associated variant is likely to be functional.
Materials and Methods
Each subject involved in this study provided a written informed consent. The study design was approved by the ethics committees of the participating institutions.
Dutch, Finnish, and Mexican FCHL Families
A total of 565 individuals from 33 Dutch FCHL families were collected at the Utrecht Academic University Hospital.7 A total of 724 individuals from 60 Finnish families were collected in the Helsinki and Turku University Hospitals.9 A total of 806 individuals from 55 mexican FCHL families were collected at Instituto National de Ciencias Médicas Y Nutritión, Salvador Zubirán. HTG status was determined using the 90th age-sex specific population percentiles for TGs. Detailed information about the family collection and determination of the affection status can be found in the supplemental Materials and Methods (available online at http://atvb. ahajournals.org).
White Combined Hyperlipidemia Case/Controls
The white combined hyperlipidemia cases and controls were collected both at the Academic Hospital Maastricht and at the University of California San Francisco, as described previously.10,11 Combined hyperlipidemic status was specified as both TC and TG levels >90th age-sex specific population percentiles. The controls had TC and TG levels <75th age-sex specific population percentiles. Detailed information about the case/control definition can be found in the supplemental Materials and Methods.
The METSIM study is an ongoing population sample of Finnish males. A total of 4463 males were included in this study (supplemental Table I). Further information about sample collection and quantitative trait measurements can be found in the supplemental Materials and Methods.
SNP Selection, Genotyping, and Sequencing
SNPs were selected to capture the common genetic variation in the GAL region (GAL coding region and 10Kbp up and downstream of the gene) using the CEPH HapMap Phase I genotype data and the haplotype tagging software LDSelect12 with the binning threshold set at r2≥0.85 and the minor allele frequency (MAF) filter set to ≥0.1. We identified 3 tagSNPs in the region (rs3136541, rs3181041, and rs6591350), and one singleton SNP (rs2187331) that was not in significant linkage disequilibrium (LD) (r2<0.85) with any other SNPs in the region. Genotyping and sequencing procedures are described in detail in the supplemental Materials and Methods.
We used the Family Based Association Testing (FBAT) software package to test for association in families with HTG.13 Logistic regression was used to test for association in the cases/controls using an additive model of inheritance in the statistical software package R. We performed a combined analysis of the family-based and case-control studies (n=2471) using the Z-method to combine the statistics for the 4 SNPs. Regression based analysis of the METSIM study utilized ANCOVA in R adjusting for age, BMI and smoking status (smoking or not). Power was calculated using QUANTO.14 Additional details of the statistical analyses can be found in the supplemental Materials and Methods.
Transient Transfections and Luciferase Assay
A region of 564 bp containing the SNP rs2187331 was amplified and cloned from genomic DNA template into the pGL3-Promoter vector (Promega). All experiments were carried out in HEK293T and HepG2 cells. The dual-luciferase reporter assay (Promega) was used according to manufacturer’s protocol. Fourty-eight hours after transfection, cells in each well were lysed using 50 μL passive lysis buffer and 15 μL of the lysate was used for the assay. Detailed description of construct, transfection, and experimental design can be found in the supplemental Materials and Methods.
Electrophoretic Mobility Shift Assay
The nuclear proteins were extracted from the HepG2 and PANC-1 cell lines by NE-PER nuclear extraction reagents (Pierce) and protein concentrations were quantified by the Bradford assay. Complementary 5′ biotinylated and unlabeled oligonucleotide probes representing both alleles of rs2187331 were obtained commercially (Operon Biotechnologies). Specifics of EMSA can be found in the supplemental Materials and Methods.
Microarrays and qRT-PCR
We collected 49 fat biopsies from umbilical subcutaneous adipose tissue under local anesthesia from unrelated Mexican FCHL cases/controls. Clinical characteristics are given in supplemental Table II. Microarray hybridization and qRT-PCR experiments are described in the supplemental Material and Methods. The CEL files were imported into R 2.6.0 using the justRMA function of the Affymetrix library in Bioconductor, which does background subtraction and quantile normalization. The quality of each probe for every individual was assessed using the MAS 5.0 software from Affymetrix. The normalized relative expression of each gene was used as the dependent variable in linear regression with rs2187331 genotypes (TT=0, TC=1, CC=2) and the FCHL status as the independent variables using the anova and lm functions from the stats package in R. Pathway enrichment was conducted using the DAVID software package.15 Microarray data can be accessed in MIAME compliant format from GEO (GSE13506).
Association of GAL SNPs With Elevated TGs
We genotyped 4 tagSNPs (rs2187331, rs3136541, rs3181041, rs6591350) covering GAL and tested them for association with HTG in both the FCHL families and combined hyperlipidemia cases/controls (n=2471; Table 1). We observed an association with HTG in the combined hyperlipidemia cases/controls for the common allele T of rs2187331 (probability value=0.0015). The FCHL families showed a trend in the same direction (probability value=0.063). Combining the association results from the combined hyperlipidemia cases/controls and the FCHL families increased the significance (probability value=0.00038, n=2471). The 12 tests conducted would require a Bonferroni corrected probability value of ≤0.0042 to maintain an experiment-wide type I error rate of 0.05. The association detected in the combined analysis of rs2187331 (probability value=0.00038) is thus considered experiment-wide significant. We did not observe a sex-specific effect for rs2187331 in the combined analysis when analyzing males and females separately (male probability value=0.011, female probability value=0.014), and the same allele was associated in both sexes (male Z=−2.53, female Z=−2.45). We did not observe any association with TC in the families (probability value >0.1).
Functional Evidence for rs2187331
The SNP rs2187331 is located 5.6K bp upstream of the GAL transcriptional start and is not in significant LD (r2>0.85) with any other SNPs in the region. Therefore, we hypothesized that rs2187331 may be the causal SNP and likely functions in a regulatory role. To test this hypothesis, we cloned a 564-bp region surrounding rs2187331 into a luciferase reporter gene vector and assayed for allele specific expression in the human HEK293T (embryonic kidney) cell line. This cell line was selected because GAL expression was detected in both kidney tissue and the HEK293T cell line (supplemental Figure I), and also because it is easy to transfect. We found that the region containing the SNP rs2187331 functions as a regulatory element, because the T allele had higher expression than the C allele (probability value=0.017, T:C fold-change=1.53; Figure 1). This effect was also observed when the insert was cloned in the opposite orientation (data not shown). We also observed the same effect in the hepatic HepG2 cell line (data not shown). Next we assayed for allele specific nuclear-factor binding with 31-mer double-stranded oligos, representing the 2 rs2187331 alleles using EMSA with nuclear extract from PANC-1 (pancreatic) and HepG2 (hepatic) cell lines (Figure 2). GAL was found to be expressed in both pancreas and the PANC-1 cell line (supplemental Figure I). We observed a similar allele specific binding difference in both cell lines (Figure 2). These data demonstrate that the alleles of rs2187331 alter the rate of transcription and affect differential binding of nuclear-factors in vitro.
Genes Differentially Expressed Based on rs2187331 Genotypes
The expression of GAL was too low to be accurately quantified by qRT-PCR (CT values >30) in the Mexican FCHL fat biopsies. However, it is known that LPL transcript levels in adipose tissue are sensitive to GAL peptide levels in the paraventricular nucleus (PVN) of the hypothalamus in rats.3,4 We hypothesized that LPL and other downstream genes may be differentially expressed based on rs2187331 genotypes. We analyzed 49 Mexican FCHL fat biopsies for differential expression by regressing rs2187331 genotypes on the normalized gene expression and adjusting for case/control status. We identified 1884 differentially expressed probe sets with a probability value <0.05 (corresponding to 1667 genes). Importantly, we found LPL to be differentially expressed with 2 independent probes sets (probability values=0.0052 to 0.0063). Using qRT-PCR we validated the differential expression of LPL and hormone sensitive lipase (LIPE; supplemental Figure II). Interestingly, higher expression in the TT genotype carriers was observed for 77% of the 1884 probes identified as differentially expressed based on rs2187331 genotypes (probability value <0.05; supplemental Table VI), suggesting that the direction of differential expression is important. The set of differentially expressed genes was significantly enriched for GO terms related to lipid metabolism (Benjamini corrected probability value <0.05; Table 2). A subset of interesting lipid genes that were differentially expressed are shown in supplemental Table III (probability value <0.05). Importantly, the same pathways that were identified in the DAVID analyses were also identified when the analysis included FCHL status, sex, age, and BMI as covariates (data not shown). These data suggest that rs2187331 genotypes influence expression of multiple genes of lipid metabolism in adipose, a tissue known to be responsive to GAL levels in the hypothalamus.
Resequencing of GAL Coding Regions
We resequenced the coding regions of GAL in 13 unrelated Dutch FCHL family members who had elevated plasma TGs (>90th age-sex-percentiles), and who were homozygous for the T allele of rs2187331. In addition we resequenced ≈250 bp upstream and downstream of the rs2187331. We did not detect any novel sequence variants in significant LD, defined as r2>0.85 that could explain the association with TG in the coding region of GAL (NM_015973.3). The variants identified through resequencing are presented in supplemental Table IV, and the primers used for resequencing are listed in supplemental Table V.
Investigation of GAL SNPs at the Population Level
We tested for association between quantitative TGs and the 2 SNPs with a probability value <0.10 in the study samples ascertained for hyperlipidemia (rs2187331 and rs3181041) in an unascertained population-based sample of 4463 males from Kuopio, Finland (Table 3). We observed an association between the rare C allele of rs2187331 and high levels of plasma TGs in the total study sample (probability value=0.0028 to 0.00016; Table 3). We did not observe an association with LDL-C or TC in the METSIM (probability values >0.1). In the nonstatin users (n=3963), the probability value was borderline significant (P=0.026 to 0.019), and no association was observed in T2DM subjects (n=658) (Table 3). These findings are in line with the recent meta-analysis of quantitative lipid phenotypes of type 2 diabetes (T2D) genome-wide association studies (GWAS), which did not identify rs2187331 as significantly associated with TGs (probability value=0.057).1 Nor were there any genome-wide significant signals within 500Kbp of GAL (minimum probability value=0.0095 for TGs). This meta-analysis included diabetic cases and excluded individuals on lipid-lowering therapy, primarily statins.
For the analysis of diabetics, we calculated that we have 89% power to detect an association while maintaining a type I error rate of 0.05 with the given sample size (n=658) and an additive model. In this power calculation, we used the same effect size as observed in the total METSIM study for TGs. It is worth noting that there are only 16 subjects with combined hyperlipidemia when using the TC and TG 90th age-sex specific Finnish population percentiles in the nonstatin user group, preventing an actual replication analysis with rs2187331. To conclude, we observed an effect of rs2187331 on TGs in an unascertained study sample of Finnish males. This effect is confounded by lipid-lowering therapy and diabetes.
We observed that the T allele of rs2187331 residing in a putative regulatory region upstream of GAL is experiment-wide significantly associated with HTG in the combined analysis of FCHL families and combined hyperlipidemia cases/controls. In a nonascertained population sample, the C allele of rs2187331 was associated with higher levels of TGs. We also provide evidence for the functional relevance of rs2187331 by demonstrating allelic differences in reporter gene expression, nuclear factor binding, and expression of lipid metabolism related genes, including LPL, a known downstream target of GAL. These data demonstrate an association to a clinically relevant trait, a plausible mechanism by which this SNP may functionally alter GAL expression, and allelic differential expression of genes that are likely to play a role in the increased susceptibility for HTG.
We observed that LPL and many other lipid metabolism related genes are differentially expressed based on rs2187331 genotypes. Furthermore, 77% of these genes show higher expression with the TT genotype. This is consistent with data in rats where GAL was found to regulate LPL expression both in adipose and muscle.3,4 Interestingly, not only genes from the anabolic fat storage pathway have higher expression with the TT genotype (CD36, FATP1, LPL, VLDLR), but also genes from the catabolic pathway (ATGL, LIPE, MGLL). The molecular mechanism(s) underlying the coregulation of these genes will need to be investigated in future studies.
We found the C allele of rs2187331 associated with higher levels of TGs in the population-based METSIM study, whereas in samples ascertained for hyperlipidemia the T allele was associated with HTG. Approximately 10% of the METSIM individuals were taking statins and only 16 of the nonstatin users had combined hyperlipidemia. Also 15% of the METSIM study are diabetic. The sample sizes of the HTG (n=2471) and population (n=4463) studies make it less likely that either finding is by chance alone. Furthermore, the functional evidence supports a regulative allele-specific role of rs2187331. Moreover, ultimately an individual’s risk to develop a complex cardiovascular trait is a combination of susceptibility variants, environmental factors, behavior, and chance. The combination of variants and environmental factors are likely to differ between subjects with HTG and subjects representing the unascertained general population, thus contributing to these types of allelic differences. The difference in direction of the association between the METSIM and hyperlipidemic study samples could be attributable to these confounding factors (ie, statins and diabetes status), and will have to be further investigated in independent study samples with refined quantitative phenotypes and highly detailed lifestyle variables. Extensive resequencing of the entire genomic region of GAL in different study populations is also warranted to further explore the possibility that differences in LD structure or several functional variants, and their interactions with rs2187331 may contribute to the disparity in the association signal.
Originally it was observed that injection of nanomolar levels of GAL peptide onto the paraventricular nucleus (PVN) of the hypothalamus stimulates feeding.16 Subsequent genetic evidence from GAL knockout mice suggests that GAL plays a role in fat intake and preference.17–19 Interestingly, a positive correlation was also observed between plasma glucose and circulating levels of galanin peptide in healthy and type 2 diabetic individuals.20,21 However, the inhibition of insulin secretion identified in other mammals was not observed in humans given a bolus of either porcine or human galanin.22,23 Variation near the GAL gene has previously been associated with alcoholism and anxiety,24 whereas obesity studies have not yielded significant findings for GAL.25,26
In normolipidemic individuals plasma TG levels are increased after the consumption of a high-fat meal. In response to the elevated plasma TG levels, GAL expression is upregulated in the PVN,3 which in turn signals the upregulation of genes involved in the lipolysis of TGs from triglyceride rich lipoproteins (TRLs) and import of free fatty acids for storage in adipose tissue. A secondary effect of lipolysis of TGs from TRLs is a reduction in the level of plasma TGs attributable to a relocation of TGs from plasma into another tissue. We hypothesize that in response to postprandial elevated plasma TGs individuals with HTG carrying the C allele of rs2187331 will have decreased expression or reduced upregulation of GAL in the PVN, which modifies the upregulation of lipolysis and storage related genes in adipose tissue.
In conclusion, our data suggest that rs2187331 residing in a putative regulatory region upstream of GAL contributes to hypertriglyceridemia in FCHL, TG levels in the general population, and has an allele-specific cis-regulatory function.
We thank the individuals who participated in this study. We thank E. Nikkola and M. Lupsakko for laboratory technical assistance.
Sources of Funding
This study was funded by National Institutes of Health (HL-28481, HL082762); American Heart Association (072523Y to A.H.-V., 0465005Y to B.E.A.); National Human Genome Research Institute (T32 HG02536 to C.L.P., and D.W.-V.); Academy of Finland (to M.K.); Clinical Research Institute HUCH Ltd (to M.-R.T.); and Finnish Heart Foundation (to M.-R.T.).
T.W.A.d.B. has been employed by GlaxoSmithKline. This work was not supported in full or part by GlaxoSmithKline.
Received July 29, 2008; revision accepted October 23, 2008.
Willer CJ, Sanna S, Jackson AU, Scuteri A, Bonnycastle LL, Clarke R, Heath SC, Timpson NJ, Najjar SS, Stringham HM, Strait J, Duren WL, Maschio A, Busonero F, Mulas A, Albai G, Swift AJ, Morken MA, Narisu N, Bennett D, Parish S, Shen H, Galan P, Meneton P, Hercberg S, Zelenika D, Chen WM, Li Y, Scott LJ, Scheet PA, Sundvall J, Watanabe RM, Nagaraja R, Ebrahim S, Lawlor DA, Ben-Shlomo Y, Davey-Smith G, Shuldiner AR, Collins R, Bergman RN, Uda M, Tuomilehto J, Cao A, Collins FS, Lakatta E, Lathrop GM, Boehnke M, Schlessinger D, Mohlke KL, Abecasis GR. Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet. 2008; 40: 161–169.
Babirak SP, Brown BG, Brunzell JD. Familial combined hyperlipidemia and abnormal lipoprotein lipase. Arterioscler Thromb. 1992; 12: 1176–1183.
Hoffer MJ, Bredie SJ, Boomsma DI, Reymer PW, Kastelein JJ, de Knijff P, Demacker PN, Stalenhoef AF, Havekes LM, Frants RR. The lipoprotein lipase (Asn291–>Ser) mutation is associated with elevated lipid levels in families with familial combined hyperlipidaemia. Atherosclerosis. 1996; 119: 159–167.
Naoumova RP, Bonney SA, Eichenbaum-Voline S, Patel HN, Jones B, Jones EL, Amey J, Colilla S, Neuwirth CK, Allotey R, Seed M, Betteridge DJ, Galton DJ, Cox NJ, Bell GI, Scott J, Shoulders CC. Confirmed locus on chromosome 11p and candidate loci on 6q and 8p for the triglyceride and cholesterol traits of combined hyperlipidemia. Arterioscler Thromb Vasc Biol. 2003; 23: 2070–2077.
Pajukanta P, Porkka KV, Antikainen M, Taskinen MR, Perola M, Murtomaki-Repo S, Ehnholm S, Nuotio I, Suurinkeroinen L, Lahdenkari AT, Syvanen AC, Viikari JS, Ehnholm C, Peltonen L. No evidence of linkage between familial combined hyperlipidemia and genes encoding lipolytic enzymes in Finnish families. Arterioscler Thromb Vasc Biol. 1997; 17: 841–850.
Mar R, Pajukanta P, Allayee H, Groenendijk M, Dallinga-Thie G, Krauss RM, Sinsheimer JS, Cantor RM, de Bruin TW, Lusis AJ. Association of the APOLIPOPROTEIN A1/C3/A4/A5 gene cluster with triglyceride levels and LDL particle size in familial combined hyperlipidemia. Circ Res. 2004; 94: 993–999.
Gauderman WJ. Sample size requirements for association studies of gene-gene interaction. Am J Epidemiol. 2002; 155: 478–484.
Sherman BT, Huang da W, Tan Q, Guo Y, Bour S, Liu D, Stephens R, Baseler MW, Lane HC, Lempicki RA. DAVID Knowledgebase: a gene-centered database integrating heterogeneous gene annotation resources to facilitate high-throughput gene functional analysis. BMC Bioinformatics. 2007; 8: 426.
Legakis I, Mantzouridis T, Mountokalakis T. Positive correlation of galanin with glucose in type 2 diabetes. Diabetes Care. 2005; 28: 759–760.
Gilbey SG, Stephenson J, O'Halloran DJ, Burrin JM, Bloom SR. High-dose porcine galanin infusion and effect on intravenous glucose tolerance in humans. Diabetes. 1989; 38: 1114–1116.
Schauble N, Reichwald K, Grassl W, Bechstein H, Muller HC, Scherag A, Geller F, Utting M, Siegfried W, Goldschmidt H, Blundell J, Lawton C, Alam R, Whybrow S, Stubbs J, Platzer M, Hebebrand J, Hinney A. Human galanin (GAL) and galanin 1 receptor (GALR1) variations are not involved in fat intake and early onset obesity. J Nutr. 2005; 135: 1387–1392.