Gene-gene interaction detection with deep learning

The extent to which genetic interactions affect observed phenotypes is generally unknown because current interaction detection approaches only consider simple interactions between top SNPs of genes. We introduce an open-source framework for increasing the power of interaction detection by considering all SNPs within a selected set of genes and complex interactions between them, beyond only the currently considered multiplicative relationships. In brief, the relation between SNPs and a phenotype is captured by a neural network, and the interactions are quantified by Shapley scores between hidden nodes, which are gene representations that optimally combine information from the corresponding SNPs. Additionally, we design a permutation procedure tailored for neural networks to assess the significance of interactions, which outperformed existing alternatives on simulated datasets with complex interactions, and in a cholesterol study on the UK Biobank it detected nine interactions which replicated on an independent FINRISK dataset.

South Asians: why are they at a higher risk for cardiovascular disease?

PURPOSE OF REVIEW

We comment on the high prevalence of cardiovascular disease (CVD) in South Asians (SA). The effect of various risk factors, for example biochemical, genetic, lifestyle, socioeconomic factors and psychosocial stress on CVD risk is discussed.

RECENT FINDINGS

'Prediabetes' is common in SA, but its relationship with coronary artery disease (CAD) is not significant unlike for the white population. At the same time, 'prediabetes' in SA is associated with an increased risk for cerebrovascular disease (CeVD). The differentiating factor could be the high lipids in Europeans and their relationship to CAD. Likewise, higher diastolic blood pressure in SA may explain the risk of CeVD. Small, dense, low-density lipoprotein (LDL), low high-density lipoprotein-cholesterol (HDL-C) concentration and high triglycerides may contribute to atherosclerosis. Thrombotic factors such as increased levels of plasminogen activator inhibitor, fibrinogen, lipoprotein (a) and homocysteine have been shown to be associated with increased CVD. Impaired cerebrovascular autoregulation and sympathovagal activity, increased arterial stiffness and endothelial dysfunction may increase CVD risk further. In addition, environmental and dietary factors may exaggerate the unfavourable cardiovascular profile through genetic factors.

SUMMARY

The implications of the findings suggest comprehensive screening of SA for CVD. Cultural differences should be considered while designing prevention strategies specifically targeting barriers for uptake of preventive service.

The rs2070895 (-250G/A) Single Nucleotide Polymorphism in Hepatic Lipase (HL) Gene and the Risk of Coronary Artery Disease in North Indian Population: A Case-Control Study

INTRODUCTION

Several Single Nucleotide Polymorphisms (SNPs) in lipid transport genes have been shown to be associated with Coronary Artery Disease (CAD). The Hepatic Lipase (HL)glycoprotein is a key component that catalyzes the hydrolysis of triglycerides and phospholipids in all major classes of lipoproteins.

AIM

We studied whether the HL gene-250G/A polymorphism affect blood lipid level and the CAD in a North Indian population.

MATERIALS AND METHODS

A total number of 477 subjects were enrolled in the study after approval of the Institutional Ethics Committee. Out of 477 subjects, 233 were with coronary artery disease as study group and 244 subjects without coronary artery disease as control group. All subjects recruited with matched ethnicity in age group of 40-70 years. Blood samples were collected in EDTA vials and genomic DNA was extracted from blood using the phenol-chloroform method. Lipid profile was estimated by using a commercially available kit. Polymorphisms in the HL (-250 G/A) gene were analysed by using restriction fragment length polymorphism-polymerase chain reaction (PCR-RFLP) method. The effect of this polymorphism on plasma lipids, lipoproteins and coronary artery disease was determined.

RESULTS

In Human Hepatic Lipase (LIPC)-250G/A genotype, the frequencies of GG, GA and AA genotype in CAD group was 80.69%, 15.45% and 3.86%, respectively; in the control group, the corresponding frequencies were 90.16%, 9.02% and 0.82%, respectively. A significant difference was found in the genotype (LIPC-250G/A) distribution between the two groups. Further logistic regression analysis indicated that the GA and AA genotypes in SNP-250G/A were significantly associated with CAD in all genetic models (In codominant model- GA vs. GG, OR=1.91, 95% CI=1. 09-3.37, p=0. 03 and AA vs. GG, OR= 5.26, 95% CI= 1.10-24.60, p=0.04; in dominant model- GA+AA vs. GG, OR=2.19, p=0.004 and in recessive model- AA vs. GG+GA, OR=5.26, p=0.04 whereas, A allele at nucleotide -250G/A in the LIPC gene had an association with increased risk of CAD (OR=2.33, p=<0.008).

CONCLUSION

Our findings indicated that the higher frequency of a dominant model (GA+AA) as well as mutant allele A of LIPC-250 G/A polymorphism is significantly associated with risk of CAD and the lipid profile can be used as a predictor of CAD.

Association between two common polymorphisms (single nucleotide polymorphism -250G/A and -514C/T) of the hepatic lipase gene and coronary artery disease in type 2 diabetic patients

BACKGROUND

Variations in the hepatic lipase (HL) gene are the potential candidate for coronary artery disease (CAD) especially in type 2 diabetes mellitus (T2DM) in diverse populations. We assessed the association of -514C/T and -250G/A polymorphisms in HL (LIPC) gene with CAD risk in Iranian population with type 2 diabetes.

MATERIALS AND METHODS

We evaluated 322 type 2 diabetic patients, 166 patients with normal angiograms as controls and 156 patients those identified with CAD undergoing their first coronary angiography as CAD cases. Genotyping of -514C/T and -250G/A polymorphisms in the promoter of the LIPC gene were studied by polymerase chain reaction (PCR)-restriction fragment length polymorphism technique.

RESULTS

Genotype distributions in CAD cases (73.7%, 20.5%, and 5.8% for -250G/A) and (62.2%, 32.7%, and 5.1% for -514C/T) were significantly different from those in controls (60.8%, 37.4%, and 1.8% for -250G/A) and (51.2%, 48.2%, and 0.6% for -514C/T). CAD cases had lower A-allele frequency than controls (0.131 vs. 0.196, P = 0.028). The odds ratio for the presence of -250 (GG + GA) genotype and A allele in CAD cases were 2.206 (95% confidence interval [CI] =1.33-3.65, P = 0.002) and 1.609 (95% CI = 1.051 -2.463, P = 0.029) respectively. Haplotype analysis demonstrated a significant association between especially LIPC double mutant (-250 A/-514 T) haplotype and presence of CAD.

CONCLUSION

Our findings indicated that -250 G/A polymorphism rather than -514 C/T polymorphism of LIPC gene is more associated with the increased risk of CAD particularly in women with T2DM.

Association of hepatic lipase gene polymorphisms with hypertriglyceridemia and low high-density lipoprotein-cholesterol levels among South Indian subjects without diabetes

AIM

The aim of this study was to investigate the association of four variants of the hepatic lipase (HL [or LIPC]) gene with various lipid parameters among South Indian subjects with normal glucose tolerance (NGT).

SUBJECTS AND METHODS

In total, 747 NGT subjects were randomly selected from the Chennai Urban Rural Epidemiological Study (CURES). Serum triglycerides, serum cholesterol, and high-density lipoprotein cholesterol (HDL-C) were measured using a Hitachi-912 autoanalyzer (Roche Diagnostics GmbH, Mannheim, Germany). Genotyping of HL gene variants was done by the polymerase chain reaction-restriction fragment length polymorphism method, and 20% of samples were sequenced to validate the genotypes obtained. Haplotype analysis was also carried out.

RESULTS

The TT genotype of the rs1800588 C/T (C-480T) polymorphism was significantly associated with hypertriglyceridemia, with an adjusted odds ratio of 2.58 (95% confidence interval 1.38-4.85, P=0.003), whereas those with the CC genotype of the rs6074 A/C (Thr479Thr) had significantly lower HDL-C levels (41.3±9.8 mg/dL) compared with the AA genotype (43.6±10.2 mg/dL, P=0.02). Haplotype analysis showed the TGC haplotype was significantly associated with low HDL-C levels.

CONCLUSIONS

Among South Indian subjects without diabetes, the rs1800588 C/T (C-480T) and rs6074 C/A (Thr479Thr) variants of the HL gene are associated with hypertriglyceridemia and low HDL-C, respectively. The TGC haplotype was significantly associated with low HDL-C.

Association studies of several cholesterol-related genes (ABCA1, CETP and LIPC) with serum lipids and risk of Alzheimer's disease

Objectives

Accumulating evidence suggested that dysregulation of cholesterol homeostasis might be a major etiologic factor in initiating and promoting neurodegeneration in Alzheimer’s disease (AD). ATP-binding cassette transporter A1 (ABCA1), hepatic lipase (HL, coding genes named LIPC) and cholesteryl ester transfer protein (CETP) are important components of high-density lipoprotein (HDL) metabolism and reverse cholesterol transport (RCT) implicated in atherosclerosis and neurodegenerative diseases. In the present study, we will investigate the possible association of several common polymorphisms (ABCA1R219K, CETPTaqIB and LIPC-250 G/A) with susceptibility to AD and plasma lipid levels.

Methods

Case–control study of 208 Han Chinese (104 AD patients and 104 non-demented controls) from Changsha area in Hunan Province was performed using the PCR-RFLP analysis. Cognitive decline was assessed using Mini Mental State Examination (MMSE) as a standardized method. Additionally, fasting lipid profile and the cognitive testing scores including Wechsler Memory Scale (WMS) and Wisconsin Card Sorting Test (WCST) were recorded.

Results and conclusions

We found significant differences among the genotype distributions of these three genes in AD patients when compared with controls. But after adjusting other factors, multivariate logistic regression analysis showed only ABCA1R219K (B = −0.903, P = 0.005, OR = 0.405, 95%CI:0.217-0.758) and LIPC-250 G/A variants(B = −0.905, P = 0.018, OR = 0.405, 95%CI:0.191-0.858) were associated with decreased AD risk. There were significantly higher levels of high-density lipoprotein cholesterol (HDL-C) and apolipoproteinA-I in the carriers of KK genotype and K allele (P < 0.05), and B2B2 genotype of CETP Taq1B showed significant association with higher HDL-C levels than other genotypes (F = 5.598, P = 0.004), while -250 G/A polymorphisms had no significant effect on HDL-C. In total population, subjects carrying ABCA1219K allele or LIPC-250A allele obtained higher MMSE or WMS scores than non-carriers, however, no significant association was observed in AD group or controls. Therefore, this preliminary study showed that the gene variants of ABCA1R219K and LIPC-250 G/A might influence AD susceptibility in South Chinese Han population, but the polymorphism of CETPTaq1B didn't show any association in despite of being a significant determinant of HDL-C.

Differential levels of apolipoprotein E and butyrylcholinesterase show strong association with pathological signs of Alzheimer's disease in the brain in vivo

Recently, we reported that 3 of the known risk factors of Alzheimer's disease (AD), i.e., advanced age, apolipoprotein E (ApoE) ε4, and female gender, are associated with differential levels of ApoE proteins and butyrylcholinesterase (BuChE) in the cerebrospinal fluid (CSF) of AD patients. The ApoE ε4 allele and certain BuChE polymorphisms synergistically affect the conversion rate of mild cognitive impairment (MCI) to AD. Here, we investigated interrelationships between ApoE and BuChE levels, and pathological markers of AD in vivo. CSF from patients with probable AD, assessed for cerebral glucose metabolism (CMRglc; n = 50) and Pittsburgh compound B (PIB) retention (β-amyloid [Aβ] load, n = 29) by positron emission tomography (PET), was used for measurement of BuChE, ApoE, Aβ, tau, phosphorylated tau (P-tau) and interleukin-1β (IL-1β) levels. Levels of ApoE and BuChE strongly correlated with CMRglc (fluorodeoxyglucose [FDG]-PET, r = 0.54, p < 0.0001, n = 50), cerebral Aβ load (PIB retention, r = 0.73, p < 0.0001, n = 29), and CSF P-tau (r = 0.73, p < 0.0001, n = 33). High ApoE protein was tied to low CMRglc and high PIB retention and P-tau. BuChE levels had opposite relationships. Other CSF covariates were levels of interleukin-1β and Aβ(42) peptide. The pattern of the patients' cognitive Z-scores strongly supported these observations. High ApoE protein was also linked to changes in 3 of the biodynamic properties of BuChE. In vitro analysis indicated that high ApoE protein levels were related to an increased pool of dormant BuChE molecules with an abnormally high intrinsic catalytic rate in CSF, which was "turned on" by excess Aβ peptides. The findings suggest that abnormally high levels of ApoE may play a causative role in the pathological events of AD, particularly those involving the early cholinergic deficit in the AD brain, through modulation of cholinesterases activities, hence disturbing the acetylcholine-dependent activity of neurons and nonexcitable cells such as glial cells.

The effect of whole grain wheat sourdough bread consumption on serum lipids in healthy normoglycemic/normoinsulinemic and hyperglycemic/hyperinsulinemic adults depends on presence of the APOE E3/E3 genotype: a randomized controlled trial

Background

Epidemiological studies associate consumption of whole grain foods, including breads, with reduced cardiovascular disease (CVD) risk; however, few studies have compared wheat whole grains with wheat refined grains.

Methods

This study investigated effects of 6-week consumption of whole grain wheat sourdough bread in comparison to white bread on fasting serum lipids in normoglycemic/normoinsulinemic (NGI; n = 14) and hyperglycemic/hyperinsulinemic (HGI; n = 14) adults. The influence of single-nucleotide polymorphisms, 3 within the APOE gene (E2, E3, E4) and 2 within the hepatic lipase gene promoter (LIPC -514C>T, LIPC -250G>A) were considered.

Results

At baseline, HGI participants had significantly higher body weight, waist circumference, body fat, and fasted glucose, insulin, homeostasis model assessment of insulin resistance (HOMA-IR), glucagon, triacylglycerols (TAG) and TAG:HDL-cholesterol, compared to NGI participants; however, none of these in addition to none of the other serum lipids, differed between bread treatments, within either participant group. For participants with the APOE E3/E3 genotype, LDL-cholesterol (P = 0.02) increased in the NGI group (n = 7), and TAG (P = 0.03) and TAG:HDL-cholesterol (P = 0.04) increased in the HGI group (n = 10), following consumption of whole grain wheat sourdough compared to white bread.

Conclusions

In summary, 6-week consumption of whole grain wheat sourdough bread did not significantly modulate serum lipids in NGI or HGI adults; however, it significantly increased LDL-cholesterol, TAG and TAG:HDL-cholesterol in participants with the APOE E3/E3 genotype. These data add to limited literature comparing wheat whole grains to wheat refined grains on CVD risk and highlight the need to consider genetic variation in relation to lipoprotein lipid content and CVD risk.

Association of LIPC -250G>A polymorphism and several environmental factors with serum lipid levels in the Guangxi Bai Ku Yao and Han populations

Background

The association between -250G>A polymorphism in the promoter region of the hepatic lipase gene (LIPC) and plasma high-density lipoprotein cholesterol (HDL-C) concentration is contradictory in diverse ethnics. Bai Ku Yao is an isolated subgroup of the Yao minority in China. This study was designed to detect the association of LIPC -250G>A (rs2070895) polymorphism and several environmental factors with serum lipid levels in the Guangxi Bai Ku Yao and Han populations.

Methods

A total of 778 subjects of Bai Ku Yao and 648 participants of Han Chinese aged 15-80 were randomly selected from our previous stratified randomized cluster samples. Genotyping of the LIPC -250G>A was performed by polymerse chain reaction and restriction fragment length polymorphism combined with gel electrophoresis, and then confirmed by direct sequencing.

Results

The levels of serum total cholesterol (TC), HDL-C, low-density lipoprotein cholesterol (LDL-C) and apolipoprotein (Apo) AI were lower in Bai Ku Yao than in Han (P < 0.01 for all). The frequencies of GG, GA and AA genotypes were 50.0%, 43.3% and 6.7% in Bai Ku Yao, and 35.7%, 50.6% and 13.7% in Han (P < 0.01); respectively. The frequencies of G and A alleles were 71.7% and 28.3% in Bai Ku Yao, and 61.0% and 39.0% in Han (P < 0.01). The levels of HDL-C and the ratio of ApoAI to ApoB in Bai Ku Yao were lower in GG genotype than in GA or AA genotype (P < 0.05-0.01). The levels of TC, HDL-C, LDL-C and ApoB in Han were lower in GG genotype than in GA or AA genotype (P < 0.05-0.01). The levels of HDL-C and the ratio of ApoAI to ApoB in Bai Ku Yao, and the levels of HDL-C, LDL-C and ApoB in Han were correlated with genotype and/or allele (P < 0.05 for all). Serum lipid parameters were also correlated with age, sex, alcohol consumption, cigarette smoking, blood pressure, body weight, and body mass index in both ethnic groups.

Conclusions

The differences in the serum lipid profiles between the two ethnic groups might partly result from different genotypic frequency of LIPC -250G>A or different LIPC-enviromental interactions.

The apolipoprotein E ε4 allele plays pathological roles in AD through high protein expression and interaction with butyrylcholinesterase

The apolipoprotein E (ApoE) ε4 allele has consistently been established as an Alzheimer's disease (AD) risk factor, but its pathological contribution to AD is obscure. Certain butyrylcholinesterase (BuChE) polymorphisms together with the ApoE ε4 allele synergistically increase the risk of AD. In addition, AD risk factors, i.e. advanced age, female gender and ApoE ε4 are associated with different levels of CSF BuChE in AD patients, and BuChE protein attenuates Aβ fibrillization in vitro. Here we investigated the roles of ApoE and BuChE gene products as modulators of pathological features of AD in vivo. We found that AD risk factors were associated with different levels of ApoE protein in the CSF of AD patients (n=115). Women and ApoE ε4 carriers had the highest levels of ApoE protein (up by 50-120%, p<0.01-0.0001), which were increased with age (r=0.30, p<0.0006). The CSF surrogate markers of pathological features of AD, i.e. high tau and P-tau, low Aβ(42) and high tau/Aβ(42) ratio, were associated with high levels of ApoE protein. Intriguingly, high ApoE protein levels were not only associated with low amounts of BuChE, but they also altered the aging and activity of this enzyme in concentration- and isoform-dependent manners, particularly in the presence of Aβ peptides. Both ApoE and BuChE levels were also differentially related to levels of the proinflammatory cytokine IL-1β. In conclusion, ApoE ε4 might impart its pathological role through high protein expression and interaction with BuChE, which in turn might modulate central cholinergic activity and Aβ load in the brain.