Genetic Association of Diagnostic Traits of Metabolic Syndrome with Lysosomal Pathways: Insights from Target Gene Enrichment Analysis

Abstract

Genome-wide association studies (GWAS) identified many association signals for metabolic syndrome (MetS). However, the understanding of its pathophysiology may be limited because of the complexity of the intertwined genetic factors that underlie diagnostic condition traits. We conducted an enrichment analysis of spatial expression genes (eGenes) associated with GWAS signals for MetS and its diagnostic condition traits. Consequently, eGenes associated with MetS were significantly enriched in 14 biological pathways (P_BH < 0.05, where P_BH is the p-value adjusted for Benjamini–Hochberg multiple testing). Moreover, 38 biological pathways were additionally identified in the enrichment analysis of the individual diagnostic traits (P_BH < 0.05). In particular, the lysosomal pathway was revealed for waist-to-hip ratio, glucose measurement, and high-density lipoprotein cholesterol (P_BH < 0.05), but not for MetS (P_BH > 0.05). It was inferred that lysosomal pathway-based control of cellular lipid metabolism and insulin secretion/resistance could result in eGene enrichment for these diagnostic traits. In conclusion, this target gene enrichment analysis of diagnostic traits of MetS uncovered a lysosomal pathway that may dilute its effects on the MetS. We propose that lysosomal dysfunction should be a priority for research on the underlying pathogenic mechanisms of MetS and its diagnostic traits. Experimental studies are needed to elucidate causal relationships of ribosomal pathways with metabolic syndrome and its diagnostic traits.

1. Introduction

Metabolic syndrome (MetS) is a cluster of at least three of five pathological conditions: abdominal obesity, hyperglycemia, hypertension, hypertriglyceridemia, and low high-density lipoprotein (HDL) levels, which increase the risk of cardiovascular diseases and chronic complications. The chronic nature of MetS raised great concerns about its risk factors, as it can reduce quality of life and impose a substantial economic burden. Risk factors include aging [1], diet [2], sleep disorder [3], low physical activity [4], sedentary behavior [5], excessive alcohol consumption [6], and psychotropic medication [7]. Early family and twin studies showed that susceptibility to MetS may be attributed to genetic factors [8,9].

The genetics of MetS and its complex nature were intensively examined through genome-wide association studies (GWAS) to identify candidate genetic loci [10,11,12,13,14,15,16,17,18,19]. Consequently, the genetic architecture of MetS is better understood. Nevertheless, there are apparent limitations in understanding biology or applying this knowledge to precision medicine. This is largely attributed to the complexity not only by the polygenic loci that have a small effect on MetS, but also by the intertwined genetic factors that underlie the individual diagnostic condition traits of MetS. Some genetic variants identified through GWAS were found to influence the specific diagnostic condition trait of MetS. For example, certain variants may be associated with insulin resistance (e.g., rs290487, an intronic variant of TCF7L2) [20], while others may be associated with dyslipidemia (e.g., rs6589566, an upstream variant of APOA1) [21]. Uncovering such causal genes and the related pathological mechanisms for each diagnostic condition of MetS, which is highly complex, will greatly help us understand its genetics. Conversely, we could not rule out the possibility of finding a GWAS signal for MetS that was not identified for any individual diagnostic condition trait because the effect was too small to be detected. In this study, we aimed to identify the expression genes (eGenes) regulated by GWAS signals for MetS and its diagnostic condition traits and perform enrichment analysis on these eGenes to reveal the pathological pathways underlying GWAS signals.

2. Materials and Methods

2.1. Colocalization Analysis

To identify target genes to be used for enrichment analysis, we colocalized GWAS signals for MetS and expression quantitative trait loci (eQTL). Single nucleotide polymorphisms (SNPs) were retrieved as GWAS signals from the GWAS Catalog of the National Human Genome Research Institute, European Bioinformatics Institute (https://www.ebi.ac.uk/gwas; accessed on 10 November 2021). All the SNPs analyzed in this study showed suggestive associations (p < 1 × 10⁻⁵ proposed by Hindorff et al. [22]) in the original GWAS. The experimental factor ontology (EFO) identifiers used in the current study were EFO_0000195, EFO_0004343, EFO_0004465, EFO_0004468, EFO_0006335, EFO_0006336, EFO_0004530, and EFO_0004612 for MetS and its diagnostic condition traits, namely waist–hip ratio (WHR), fasting blood glucose (FBG), glucose measurement (GM), systolic blood pressure (SBP), diastolic blood pressure (DBP), triglyceride (TG), and HDL cholesterol (HDLC), respectively. These diagnostic condition traits were selected as the most common diagnostic criteria for MetS as proposed by several clinical guidelines: ATP III [23], WHO [24], EGIR [25], and IDF [26]. Only independent GWAS signals were selected for each EFO. Colocalization of a GWAS signal was determined when a representative SNP within the signal matched or was strongly linked (r² > 0.95) with any eQTL mapped by the genotype-tissue expression (GTEx) consortium [27]. Files containing summary statistics of the associations between eQTLs and eGenes for all tissues were downloaded from the GTExPortal (v8; https://gtexportal.org; accessed on 17 November 2021). Summary statistics were estimated by linear regression to determine the best nominal association between genotypes and gene expression levels using FastQTL [28]. The GTEx consortium pre-filtered and normalized gene expression levels. Additional details can be found in the article by the GTEx consortium [26]. We used the false discovery rate (P_FDR) to determine the statistical significance of colocalization for each tissue, with a significance threshold of P_FDR = 0.05. The enrichment analysis excluded eGenes within the major histocompatibility complex (MHC) region because strong linkages in a broad region may lead to spurious enrichment [29].

2.2. Enrichment Analysis

We examined whether the eGenes resulting from the colocalization analysis for MetS and its diagnostic condition traits were enriched in biological pathway terms. This eGene enrichment analysis is a statistical method that discovers biological pathways and corresponding genes overrepresented in a pool of eGenes for each trait. Enrichment analysis with eGene groups for every trait was conducted for each of the Kyoto Encyclopedia of Genes and Genomes (KEGG) terms using Enrichr (https://maayanlab.cloud/Enrichr; accessed on 11 December 2021) [30]. KEGG biological pathways included genetic and environmental information processes, cellular processes, metabolisms, organismal systems, human diseases, and drug development. Enrichment analysis was examined by tissue. This tissue-specific enrichment analysis was to avoid spurious biological pathways generated from a combined analysis with mixed eGenes. Statistical significance for enrichment was determined by P_BH < 0.05, where P_BH is the p-value corrected for multiple testing using the Benjamini–Hochberg method.

This study used publicly accessible population-based secondary data, and the subjects were unidentifiable. Thus, the Institutional Review Board (IRB) confirmed that this study was eligible for exemption from IRB review.

3. Results

3.1. Colocalization Analysis

In total, 252 genetic associations of SNPs with MetS susceptibility were observed in the GWAS Catalog (p < 10⁻⁵;

Table 1. Number of eQTL and eGenes resulted from colocalization analysis.

	MetS	WHR	FBG	GM	SBP	DBP	TG	HDL
GWAS signal a
No. of raw signals	252	9919	538	2199	3674	2417	8414	10,774
No. of unique signals	225	4735	293	1499	2603	1626	3413	3718
Colocalized eQTL-eGene b
No. of eQTLs	92	3220	55	433	929	650	1197	1519
No. of eGenes+MHC	72	566	40	229	470	381	454	522
No. of eGenes	67	545	40	224	464	374	439	507

^a Number of unique signals is the count after excluding replicates. ^b Number of eGenes^+MHC is the count before excluding eGenes within the MHC region that were not included in the enrichment analysis. Abbreviations: eQTL, expression quantitative trait loci; eGene, expression gene; MetS, metabolic syndrome; WHR, waist–hip ratio; FBG, fasting blood glucose; SBP, systolic blood pressure; DBP, diastolic blood pressure; TG, triglyceride; and HDLC, high-density lipoprotein cholesterol.

). After excluding replicates, 225 unique MetS GWAS signals remained. All available GWAS signals and eQTL pairs were examined to determine whether they colocalized. Consequently, eGenes were found by tissue type (P_FDR < 0.05), and after excluding eGenes within MHC, the average number of eGenes was 67 (Table 1). The number of eGenes by tissue type is presented in Supplementary Table S1.

Similarly, 4735, 293, 1499, 2603, 1626, 3413, and 3718 unique GWAS signals were obtained for the individual diagnostic condition traits of MetS, namely WHR, FBG, GM, SBP, DBP, TG, and HDLC, respectively. Using these signals, colocalization analysis revealed an average of 545, 40, 224, 464, 374, 439, and 507 eGenes per tissue for WHR, FBG, GM, SBP, DBP, TG, and HDLC, respectively (P_FDR < 0.05; Table 1). The number of eGenes per tissue for individual diagnostic condition traits is presented in Supplementary Tables S2–S8.

3.2. Enrichment Analysis

Enrichment analysis revealed that the eGenes identified for MetS, WHR, FBG, GM, TG, and HDLC were significantly enriched in 46, 41, 59, 22, 26, and 51 biological pathways, respectively (P_BH < 0.05;

Table 2. Biological pathways identified from eGene enrichment analysis for metabolic syndrome (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini-Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Cholesterol metabolism	Heart left ventricle	1.46 × 10−6	1.14 × 10−4
Cholesterol metabolism	Skin—not sun-exposed suprapubic	1.73 × 10−5	5.72 × 10−4
Cholesterol metabolism	Thyroid	6.15 × 10−6	9.16 × 10−4
Cholesterol metabolism	Adipose—subcutaneous	6.69 × 10−5	1.07 × 10−3
Cholesterol metabolism	Muscle—skeletal	2.57 × 10−5	1.14 × 10−3
Cholesterol metabolism	Heart atrial appendage	8.21 × 10−5	1.50 × 10−3
Vitamin digestion and absorption	Heart left ventricle	1.09 × 10−4	4.25 × 10−3
Cholesterol metabolism	Esophagus—mucosa	3.91 × 10−4	4.76 × 10−3
PPAR signaling pathway	Testis	7.41 × 10−4	6.07 × 10−3
Biosynthesis of unsaturated fatty acids	Stomach	2.77 × 10−3	8.03 × 10−3
Cholesterol metabolism	Artery—aorta	1.73 × 10−3	8.14 × 10−3
Hepatitis C	Brain—amygdala	8.63 × 10−3	1.51 × 10−2
Cholesterol metabolism	Testis	2.63 × 10−3	1.58 × 10−2
Vitamin digestion and absorption	Brain—substantia nigra	1.31 × 10−2	1.59 × 10−2
Glycosphingolipid biosynthesis	Pituitary	7.06 × 10−3	1.80 × 10−2
PPAR signaling pathway	Heart left ventricle	3.03 × 10−3	1.82 × 10−2
Cholesterol metabolism	Liver	4.95 × 10−3	1.94 × 10−2
Cholesterol metabolism	Small intestine—terminal ileum	3.27 × 10−3	1.95 × 10−2
Oocyte meiosis	Brain—frontal cortex BA9	2.18 × 10−3	2.24 × 10−2
African trypanosomiasis	Brain—substantia nigra	2.02 × 10−2	2.34 × 10−2

,

Table 3. Biological pathways identified from eGene enrichment analysis for waist–hip ratio (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini–Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Lysosome	Esophagus—mucosa	1.58 × 10−6	5.52 × 10−5
Steroid hormone biosynthesis	Brain—putamen basal ganglia	5.23 × 10−5	9.57 × 10−4
Lysosome	Whole blood	4.27 × 10−5	1.31 × 10−3
Glyoxylate and dicarboxylate metabolism	Brain—cortex	6.00 × 10−5	1.50 × 10−3
Steroid hormone biosynthesis	Brain—hippocampus	1.45 × 10−4	1.84 × 10−3
Steroid hormone biosynthesis	Brain—hypothalamus	1.48 × 10−4	2.29 × 10−3
Glyoxylate and dicarboxylate metabolism	Pituitary	9.52 × 10−5	2.38 × 10−3
Steroid hormone biosynthesis	Kidney—cortex	6.67 × 10−4	2.85 × 10−3
Glyoxylate and dicarboxylate metabolism	Brain—caudate basal ganglia	3.49 × 10−4	4.92 × 10−3
Steroid hormone biosynthesis	Brain—amygdala	1.35 × 10−3	6.67 × 10−3
Steroid hormone biosynthesis	Brain—cerebellar hemisphere	6.73 × 10−4	8.04 × 10−3
Glyoxylate and dicarboxylate metabolism	Stomach	6.58 × 10−4	9.47 × 10−3
Collecting duct acid secretion	Artery—coronary	8.54 × 10−4	1.00 × 10−2
Steroid hormone biosynthesis	Brain—cerebellum	4.80 × 10−4	1.04 × 10−2
Steroid hormone biosynthesis	Muscle—skeletal	5.87 × 10−4	1.18 × 10−2
Ovarian steroidogenesis	Adipose—subcutaneous	4.71 × 10−4	1.40 × 10−2
Steroid hormone biosynthesis	Brain—nucleus accumbens basal ganglia	1.65 × 10−3	1.71 × 10−2
Steroid hormone biosynthesis	Brain—caudate basal ganglia	1.63 × 10−3	1.78 × 10−2
Lysosome	Skin—sun-exposed lower leg	8.70 × 10−4	1.94 × 10−2
Lysosome	Heart atrial appendage	1.20 × 10−3	2.33 × 10−2

,

Table 4. Biological pathways identified from eGene enrichment analysis for fasting blood glucose (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini–Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Thyroid hormone synthesis	Brain—hippocampus	2.06 × 10−4	1.85 × 10−3
Fructose and mannose metabolism	Prostate	5.44 × 10−4	9.79 × 10−3
Biosynthesis of unsaturated fatty acids	Stomach	6.03 × 10−4	1.07 × 10−2
Galactose metabolism	Stomach	7.97 × 10−4	1.07 × 10−2
Starch and sucrose metabolism	Stomach	1.07 × 10−3	1.07 × 10−2
Biosynthesis of unsaturated fatty acids	Brain—hypothalamus	9.41 × 10−3	1.88 × 10−2
Arachidonic acid metabolism	Brain—caudate basal ganglia	1.21 × 10−3	1.93 × 10−2
PPAR signaling pathway	Small intestine—terminal ileum	1.57 × 10−3	2.35 × 10−2
alpha-Linolenic acid metabolism	Pituitary	8.70 × 10−4	2.44 × 10−2
Amoebiasis	Brain—putamen basal ganglia	1.37 × 10−3	2.57 × 10−2
Leukocyte transendothelial migration	Brain—putamen basal ganglia	1.71 × 10−3	2.57 × 10−2
alpha-Linolenic acid metabolism	Ovary	9.96 × 10−3	2.69 × 10−2
Biosynthesis of unsaturated fatty acids	Ovary	1.08 × 10−2	2.69 × 10−2
Biosynthesis of unsaturated fatty acids	Brain—spinal cord cervical c-1	5.39 × 10−3	2.69 × 10−2
Glycolysis/Gluconeogenesis	Stomach	3.68 × 10−3	2.76 × 10−2
Cortisol synthesis and secretion	Brain—spinal cord cervical c-1	1.29 × 10−2	3.01 × 10−2
Morphine addiction	Brain—spinal cord cervical c-1	1.81 × 10−2	3.01 × 10−2
Cushing syndrome	Brain—spinal cord cervical c-1	3.06 × 10−2	3.06 × 10−2
Purine metabolism	Brain—spinal cord cervical c-1	2.56 × 10−2	3.06 × 10−2
Arachido + B2:F21nic acid metabolism	Brain—hippocampus	1.82 × 10−2	3.27 × 10−2

,

Table 5. Biological pathways identified from eGene enrichment analysis for glucose measurement (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini–Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Fructose and mannose metabolism	Skin—sun-exposed lower leg	1.60 × 10−5	4.20 × 10−3
Fructose and mannose metabolism	Whole blood	4.96 × 10−5	1.24 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Skin—sun-exposed lower leg	1.58 × 10−4	1.38 × 10−2
Fructose and mannose metabolism	Heart atrial appendage	7.78 × 10−4	1.41 × 10−2
Fructose and mannose metabolism	Heart left ventricle	6.20 × 10−4	1.46 × 10−2
Sulfur metabolism	Brain—spinal cord cervical c-1	4.05 × 10−4	2.05 × 10−2
Sulfur relay system	Brain—spinal cord cervical c-1	2.53 × 10−4	2.05 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Heart atrial appendage	1.57 × 10−3	2.32 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Heart left ventricle	1.39 × 10−3	2.37 × 10−2
Biosynthesis of unsaturated fatty acids	Adipose—visceral omentum	8.44 × 10−4	2.50 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Breast—mammary tissue	1.58 × 10−3	2.59 × 10−2
Fructose and mannose metabolism	Nerve—tibial	2.18 × 10−4	2.69 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Nerve—tibial	1.85 × 10−4	2.69 × 10−2
Fructose and mannose metabolism	Adipose—visceral omentum	1.83 × 10−3	2.95 × 10−2
Fructose and mannose metabolism	Stomach	2.45 × 10−4	3.36 × 10−2
Galactose metabolism	Skin—sun-exposed lower leg	1.03 × 10−3	3.38 × 10−2
Amino sugar and nucleotide sugar metabolism	Skin—sun-exposed lower leg	1.31 × 10−3	3.44 × 10−2
Neomycin, kanamycin, and gentamicin biosynthesis	Adipose—visceral omentum	2.47 × 10−3	3.67 × 10−2
Fructose and mannose metabolism	Brain—hippocampus	3.69 × 10−4	3.76 × 10−2
Lysosome	Whole blood	1.31 × 10−3	4.10 × 10−2

,

Table 6. Biological pathways identified from eGene enrichment analysis for triglyceride (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini–Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Cholesterol metabolism	Esophagus—mucosa	2.78 × 10−8	5.39 × 10−6
Metabolism of xenobiotics by cytochrome P450	Liver	7.45 × 10−7	1.41 × 10−4
Cholesterol metabolism	Skin—sun-exposed lower leg	3.83 × 10−6	1.57 × 10−4
Cholesterol metabolism	Nerve—tibial	8.22 × 10−6	7.07 × 10−4
Cholesterol metabolism	Heart atrial appendage	3.85 × 10−5	1.12 × 10−3
Cholesterol metabolism	Liver	5.09 × 10−5	1.60 × 10−3
Drug metabolism	Liver	9.85 × 10−5	2.07 × 10−3
Cholesterol metabolism	Thyroid	6.13 × 10−5	2.20 × 10−3
Cholesterol metabolism	Artery—tibial	4.81 × 10−5	2.24 × 10−3
Cholesterol metabolism	Esophagus—muscularis	1.18 × 10−4	4.01 × 10−3
Cholesterol metabolism	Testis	1.26 × 10−4	5.76 × 10−3
Cholesterol metabolism	Skin—not sun-exposed suprapubic	1.96 × 10−4	7.89 × 10−3
Cholesterol metabolism	Pancreas	6.89 × 10−4	9.27 × 10−3
Cholesterol metabolism	Adipose—subcutaneous	4.12 × 10−4	1.40 × 10−2
Cholesterol metabolism	Cells—cultured fibroblasts	3.84 × 10−4	1.42 × 10−2
Cholesterol metabolism	Lung	3.68 × 10−4	1.45 × 10−2
Cholesterol metabolism	Muscle—skeletal	6.99 × 10−4	1.58 × 10−2
Glutathione metabolism	Liver	9.54 × 10−4	1.71 × 10−2
Cholesterol metabolism	Esophagus—gastroesophageal junction	1.17 × 10−3	2.60 × 10−2
Cholesterol metabolism	Colon—transverse	1.72 × 10−3	3.27 × 10−2

and

Table 7. Biological pathways identified from eGene enrichment analysis for high-density lipoprotein cholesterol (P_BH < 0.05, where P_BH is the p-value adjusted for the Benjamini–Hochberg multiple testing).

Biological Pathway	Tissue	p	PBH
Cholesterol metabolism	Heart atrial appendage	3.36 × 10−9	8.79 × 10−7
Cholesterol metabolism	Skin—sun-exposed lower leg	3.74 × 10−7	2.78 × 10−5
Cholesterol metabolism	Esophagus—mucosa	1.68 × 10−7	3.11 × 10−5
Cholesterol metabolism	Esophagus—gastroesophageal junction	9.60 × 10−7	4.10 × 10−5
Cholesterol metabolism	Muscle—skeletal	2.23 × 10−6	1.08 × 10−4
Cholesterol metabolism	Esophagus—muscularis	1.84 × 10−6	1.21 × 10−4
Cholesterol metabolism	Skin—not sun-exposed suprapubic	3.63 × 10−6	1.52 × 10−4
Cholesterol metabolism	Testis	9.81 × 10−7	2.64 × 10−4
Cholesterol metabolism	Heart left ventricle	7.61 × 10−6	2.95 × 10−4
Cholesterol metabolism	Adipose—subcutaneous	7.83 × 10−6	4.59 × 10−4
Cholesterol metabolism	Spleen	2.16 × 10−6	5.48 × 10−4
Metabolism of xenobiotics by cytochrome P450	Minor salivary gland	4.61 × 10−5	1.48 × 10−3
Cholesterol metabolism	Nerve—tibial	5.21 × 10−6	1.54 × 10−3
Cholesterol metabolism	Cells—cultured fibroblasts	6.01 × 10−6	1.77 × 10−3
Cholesterol metabolism	Lung	2.57 × 10−5	1.81 × 10−3
Cholesterol metabolism	Liver	9.53 × 10−5	2.12 × 10−3
Metabolism of xenobiotics by cytochrome P450	Liver	1.38 × 10−4	2.73 × 10−3
Glycine serine and threonine metabolism	Colon—transverse	1.38 × 10−4	4.31 × 10−3
Cholesterol metabolism	Thyroid	2.96 × 10−5	4.33 × 10−3
Lysosome	Esophagus—mucosa	1.61 × 10−4	5.18 × 10−3

). In contrast, no enrichment of the eGenes identified for SBP and DBP was found in any biological pathway (P_BH > 0.05). Regardless of the tissue type, we discovered 52 biological pathways; however, eGenes for MetS were enriched in only 14 biological pathways (Figure 1). The lysosomal pathway was identified for three diagnostic condition traits: WHR, GM, and HDLC (P_BH < 0.05; Table 3, Table 5 and Table 6), but not for MetS (P_BH > 0.05; Table 2). In particular, among all the enrichment results, the largest significance was shown in the lysosomal pathway for WHR (esophagus mucosa, P_BH = 5.52 × 10⁻⁵; Table 3) except for cholesterol metabolism for TG (esophagus mucosa, P_BH = 5.39 × 10⁻⁶; Table 6) and HDLC (heart atrial appendage, P_BH = 8.79 × 10⁻⁷; sun-exposed skin (lower leg), P_BH = 2.78 × 10⁻⁵; esophagus mucosa, P_BH = 3.11 × 10⁻⁵; gastroesophageal junction, and P_BH = 4.10 × 10⁻⁵; Table 7).

4. Discussion

4.1. Enrichment Analysis and Its Implications

Target eGene enrichment analysis revealed 14 biological pathways for MetS and 38 additional biological pathways for its diagnostic condition traits. The nature of mixed medical conditions for MetS might dilute the enrichment effect. In other words, the complex array of eGenes that underpins the individual diagnostic condition traits of MetS may limit our understanding of its pathophysiology. Additionally, the current enrichment results show a considerable heterogeneity across tissues as shown in Figure 1. This demonstrates that mixed target genes obtained regardless of tissue may produce inaccurate results. The lysosomal pathway discovered in the WHR, GM, and HDLC enrichment analysis was not detected in the analysis of MetS. The enrichment of eGenes for WHR in the lysosomal pathway showed the greatest statistical significance (P_BH = 5.52 × 10⁻⁵), except for the obvious enrichment of eGenes for TG and HDLC in cholesterol metabolism.

The enrichment analysis results have three important implications. First, the genetic etiology of individual diagnostic condition traits of MetS should be emphasized to understand the genetic architecture of MetS. Second, it is necessary to pay attention to genes related to the biological pathways identified in this study, including those not identified in the colocalization analysis. Multiple testing can result in numerous false negatives, especially in genome-wide studies [31]. Using biological pathways to trace back candidate genes for susceptibility to MetS could reduce the burden of finding such false negative genes [32]. Lastly, and most importantly, dysfunction or malfunction of the lysosomal pathway might produce three abnormal conditions that can cause MetS.

In general, individuals with accumulated abnormal conditions may be exposed to a risk of cardiovascular disease that far exceeds the sum of the risks posed by each abnormality [33]. This implies invisible relations among the individual diagnostic conditions that contribute to the complexity of MetS. A predefined accumulation pattern of abnormal conditions was reported with frequent occurrences of abdominal obesity, insulin resistance, and hypertension [34]. Therefore, a common pathological mechanism is suspected to underlie these abnormal conditions. In this respect, the lysosomal pathway discovered in this study is a novel and common pathological mechanism underlying the abnormal WHR, GM, and HDLC in MetS.

4.2. Lysosomal Pathway and Diseases

Lysosomes, which degrade cellular waste via autophagy and endocytosis, play pivotal roles in cellular and organismal homeostasis through numerous cellular processes. Therefore, the lysosomal pathway closely interacts with various signaling systems in the nucleus and cytoplasm to control cellular activity [35]. Lysosomal dysfunction caused human Mendelian diseases called lysosomal storage disorders by mutations in genes encoding proteins relevant to lysosomal pathways, including lysosomal hydrolases, lysosomal membrane proteins (e.g., NPC1 [36] and TRPML1 [37]), and proteins responsible for the transport and post-translational modification of lysosomal enzymes (e.g., SUMF1 [38]). Moreover, Mendelian forms of inheritance are observed in common neurodegenerative diseases. For example, mutant forms of PSEN1 were most frequently observed to cause familial Alzheimer’s disease with abnormalities in lysosomal acidification and Ca²⁺ homeostasis [39,40,41]. Heterozygous disruption of the GBA1 gene encoding the lysosomal hydrolase β-glucocerebrosidase can be a predisposing factor for Parkinson’s disease [42]. Additionally, susceptibility to various cancers is associated with lysosomal pathway-related genes. Monoallelic deletion of BECN1 is associated with risk and poor prognosis of sporadic breast and ovarian cancer [43,44]. Mutations in PARK2 are associated with brain, lung, and colon cancers [45].

The lysosomal pathway was also implicated in the pathogenesis of complex metabolic diseases, such as diabetes and obesity. Lysosomes are involved in insulin secretion and glucose homeostasis by maintaining the health of pancreatic β-cells [46]. This lysosomal signaling function in insulin secretion and insulin resistance was supported by a recent study in which spatial eGenes colocalized with type 2 diabetes GWAS signals were enriched in the lysosomal pathway [29]. Enrichment analysis for GM in the current study also suggests a potential role of lysosomes in insulin secretion/resistance. This potential role of lysosomes cannot be neglected by the lack of significant enrichment for FBG. Inference on the negative result from enrichment analyses should be cautious. Such a negative result is attributed to the insufficient number of genes belonging to a specific biological pathway, but the biological pathway may be influenced by one or a few genes in that pathway. Moreover, the current enrichment was conservatively determined by multiple testing methods.

Mice fed a high-fat diet show lysosomal permeabilization and cathepsin B (a lysosomal cysteine protease) activation in adipocytes, ultimately resulting in adipocyte death and macrophage recruitment to the adipose tissue [47]. Thus, lipids regulate the lysosomal pathway, which regulates cellular lipid metabolism associated with obesity [48]. The latter bidirectional relation may explain the results of the current study. This directly explains eGene enrichment for WHR, an index of obesity. The enrichment for HDLC might be explained by the remaining lipids resulting from lipid metabolism, which are controlled by the lysosomal pathway.

Lysosomal lipase hydrolyzes lipoprotein cholesteryl esters, and free cholesterol is released with the help of lysosomal transport proteins NPC1 and NPC2 [49]. Lack or mutation of NPC1 or NPC2 (e.g., a missense mutation Ile1061Thr in NPC1) can cause a lysosomal disorder called Niemann–Pick type C, which has a Mendelian form of inheritance [50]. This disease causes extensive brain damage by disrupting lipid metabolism and accumulating cholesterol in the lysosomes. Lysosomes are recognized as metabolic control centers with functions such as sorting lipids and transmitting lipid signals to a lipid sensor mTORC1 [51]. Abnormal cholesterol levels might be caused by dysfunction of lysosomes that degrade lipids and signal cellular lipid availability.

4.3. Future Directions

The genetics of MetS are more complex than expected; consequently, we failed to find the enrichment of eGenes for MetS in the lysosomal pathway identified for its diagnostic condition traits. In addition, no definitive clues regarding the relationship between MetS and the lysosomal pathway were reported; however, both are of great concern. In particular, the knowledge about lysosome-mediated cellular degradation mechanisms drastically increased [52,53]. However, genetic studies on the lysosomal pathway are relatively limited. Most of these studies focused on identifying therapeutic targets for cancer and neurodegenerative diseases [54]. In addition, a relation between MetS and the lysosomal pathway was suggested in this study. We propose that the lysosomal pathway is the most likely biological process underlying the genetic factors in MetS. This was discovered conservatively by multiple testing despite the disadvantage of generating many false negatives. In addition, because it was discovered three times in the enrichment analysis of the diagnostic condition traits, the possibility of false positives was considered negligible.

The lysosomal pathway is an attractive candidate for various metabolic diseases. Research efforts are needed to understand the mechanisms underlying the lysosomal pathophysiology of therapeutic targets. Genetic studies of relevant genes and their sequence variants are essential for precision medicine of metabolic traits. Tracking functional nucleotide sequence variants within each signal is crucial for understanding gene regulatory mechanisms and realizing precision medicine. Although functional variants are likely to be in the linkage disequilibrium block with each signal [55], the difficulty in this task may be due to interactions among functional variants [56]. Moreover, the effects of multiple functional variants within a single signal place a heavy burden on accurate estimation. For example, a single eQTL for MRPL43 included three variants in a strong linkage, showing independent functions of regulating transcription factor binding affinity, altering splice site, and strengthening microRNA binding affinity [57]. An antagonistic regulatory effect of the linked alleles between transcription and translation was estimated, leading to further complexity. To reduce the burden of such complexity, employing more rational methods would be helpful in accurately estimating eQTL effects; that is, mixed model-based methods that may avoid confounding polygenic effects [58,59]. Moreover, to explain heterogeneity by sex in precision medicine [60], the genetic covariance between males and females was estimated using a modified mixed model [61].

5. Conclusions

This study provides novel insights into the biological pathways for MetS and its diagnostic condition traits. This highlights the lysosomal pathway discovered for the three individual diagnostic condition traits: WHR, GM, and HDLC. This novel biological pathway for MetS was not observed for MetS itself in this study. It was inferred that the lysosomal pathway-based control of cellular lipid metabolism and insulin secretion could enrich eGenes for the obesity index (WHR), lipid (HDLC), and glucose (GM). Lysosomal dysfunction was suggested as a candidate pathogenic mechanism underlying MetS and its diagnostic traits. Experimental studies are needed to establish their causal relationship and to specify lysosome-related genetic factors and signaling pathways that may influence susceptibility to MetS.