Differential Association of Plasma Angiopoietin-Like Proteins 3 and 4 With Lipid and Metabolic TraitsSignificance
Objective—Angiopoietin-like protein 3 (ANGPTL3) and 4 (ANGPTL4) are secreted proteins that inhibit lipoprotein lipase in vitro. Genetic variants at the ANGPTL3 and ANGPTL4 gene loci are significantly associated with plasma lipid traits. The aim of this study was to evaluate the association of plasma ANGPTL3 and ANGPTL4 concentrations with lipid and metabolic traits in a large community-based sample.
Approach and Results—Plasma ANGPTL3 and ANGPTL4 levels were measured in 1770 subjects using a validated ELISA assay. A Pearson unadjusted correlation analysis and a linear regression analysis adjusting for age, sex, and race were performed. ANGPTL3 levels were significantly positively associated with low-density lipoprotein cholesterol and high-density lipoprotein cholesterol levels (both P<2×10−5) but not triglycerides. In contrast, ANGPTL4 levels were significantly negatively associated with low-density lipoprotein cholesterol and high-density lipoprotein cholesterol (both P<2×10−5) and positively associated with triglycerides (P=0.003). In addition, ANGPTL4, but not ANGPTL3, levels were significantly positively associated with fasting blood glucose and metabolic syndrome.
Conclusions—Despite having similar biochemical effects in vitro, plasma ANGPTL3 and ANGPTL4 concentrations have nearly opposite relationships with plasma lipids. ANGPTL4 is strongly negatively associated with low-density lipoprotein cholesterol and high-density lipoprotein cholesterol and positively with multiple features of the metabolic syndrome including triglycerides, whereas ANGPTL3 is positively associated with low-density lipoprotein cholesterol and high-density lipoprotein cholesterol and not with metabolic syndrome traits including triglycerides. Although ANGPTL3 and ANGPTL4 both inhibit lipoprotein lipase in vitro and influence lipoprotein metabolism in vivo, the physiology of these related proteins and their effects on lipoproteins is clearly divergent and complex.
- cholesterol, HDL
- cholesterol, LDL
- diabetes mellitus
- lipid metabolism
The family of angiopoietin-like proteins (ANGPTLs) are secreted proteins characterized by key structural motifs including an N-terminal signal peptide directing secretion, a N-terminal coiled-coiled domain, a linker region, and a C-terminal fibrinogen-like domain.1 Two of the members that have been extensively studied are ANGPTL3 and ANGPTL4. Functional studies in vitro have shown that ANGPTL3 reversibly inhibits and ANGPTL4 irreversibly inhibits lipoprotein lipase (LPL), suggesting a role in lipoprotein metabolism.1 Transgenic mouse models have shown that ANGPTL3 and ANGPTL4 overexpression leads to increased triglyceride levels.1–4 ANGPTL3 knockout mice have at least a 50% reduction in both plasma triglyceride and high-density lipoprotein cholesterol (HDL-C).5 ANGPTL4 knockout mice have a 65% to 90% decrease in triglyceride levels, lower circulating very-low-density lipoprotein cholesterol levels, and increased LPL activity.6,7
Consistent with these data, genome-wide association studies identified that common variants at the ANGPTL3 locus are associated with triglycerides and low-density lipoprotein cholesterol (LDL-C) and common variants at the ANGPTL4 locus are associated with HDL-C levels.8,9 Population-based resequencing in the Dallas Heart Study found that nonsynonymous loss-of-function variants in both proteins were associated with lower triglyceride levels.10,11 Exome sequencing in a family with familial combined hypolipidemia identified loss-of-function mutations in ANGPTL3 that were causally associated with decreased triglycerides, LDL-C, and HDL-C.12
Despite the biological interest in these proteins, the fact that they are secreted proteins detectable in the plasma, and their potential for therapeutic targeting, there are limited data on plasma levels of ANGPTL3 and ANGPTL4 and their relationship with plasma lipids and metabolic parameters.3,5,13–18 We therefore undertook the first large-scale study to assay plasma ANGPTL3 and ANGPTL4 levels and address their association with lipids and metabolic traits.
Materials and Methods
Materials and Methods are available in the online-only Supplement.
Distribution and Demographic Associations With Plasma ANGPTL3 and ANGPTL4 Levels
The characteristics of the study population (n=1770) are described in Table 1. The sample was predominantly white, majority male, and baseline median age was 53 years. Approximately half of the sample had metabolic syndrome or type 2 diabetes mellitus.
Higher levels of ANGPTL3 and ANGPTL4 levels were observed with increasing age quartiles as shown in Figure 2 (r=0.17 and r=0.31, respectively; P<1×10−13 for both) based on Pearson correlation.
ANGPTL3 levels were significantly lower in men (mean, 227±82 ng/mL) than in women (mean, 253±94 ng/mL; P=2.38×10−9) in a logistic regression analysis. There was no significant difference in ANGPTL4 levels between men (mean, 160±108 ng/mL and median, 130 ng/mL) and women (mean, 153±96 ng/mL and median, 126 ng/mL). Although the black population was small (6%), there were significant racial differences in ANGPTL3 and ANGPTL4, with black subjects having lower plasma levels for both proteins (P<0.01 for both) in a logistic regression analysis.
Plasma levels of ANGPTL3 and ANGPTL4 were weakly but significantly correlated with each other. Figure 3 shows a scatter plot of the measured ANGPTL3 and ANGPTL4 concentrations. Based on Pearson correlation, ANGPTL3 and ANGPTL4 were correlated (r=0.18; P<0.0001). When adjusted for age, sex, and race, the same findings held true (P<0.0001).
ANGPTL3 and ANGPTL4 Are Associated With LDL-C and HDL-C in Opposing Directions
Plasma ANGPTL3 levels were significantly positively associated with LDL-C (P<0.002), HDL-C (P<10−8), and total cholesterol (P<10−7), but were not associated with triglycerides or very-low-density lipoprotein cholesterol (Table 2). The age-, sex-, and race-adjusted linear regression analysis supported these findings (Table I in the online-only Data Supplement). In a second model, a linear regression analysis adjusting for demographics and statin use showed no significant change in the associations observed (Table II in the online-only Data Supplement). Lastly, the same associations were observed between ANGPTL3 and lipid phenotypes among patients with and without diabetes mellitus. Apolipoprotein A-I was positively associated with plasma ANGPTL3 levels based on Pearson correlation and linear regression analysis. Apolipoproteins A-II, B, and C-III were not associated with ANGPTL3 levels in a demographic-adjusted linear regression model (Table I in the online-only Data Supplement).
In contrast, plasma ANGPTL4 levels were significantly negatively associated with LDL-C, HDL-C, and total cholesterol (all P<10−4) and positively associated with triglycerides and very-low-density lipoprotein cholesterol (P<0.001; Table 2). The age-, sex-, and race-adjusted linear regression analysis supported these findings (Table I in the online-only Data Supplement). A second model adjusting for age, sex, race, and statin use showed no significant change from the association observed for the demographic-adjusted models (Table II in the online-only Data Supplement).
An interaction analysis was performed between ANGPTL4 and diabetes mellitus status to determine whether diabetes mellitus was affecting the association of ANGPTL4 with lipid parameters. For HDL-C, there was a significant interaction with ANGPTL4 and diabetes mellitus status (P=0.02). Interaction analysis of ANGPTL4 and diabetes mellitus was not significant when performing analysis for triglycerides and LDL-C. Therefore, a stratified analysis was performed which showed that among those without diabetes mellitus, there was no significant association between ANGPTL4 and HDL-C, LDL-C, total cholesterol, and triglycerides. Among those with diabetes mellitus, there was a significant negative association between ANGPTL4 and HDL-C (P<10−3) and a positive association with triglycerides (P=0.02).
In addition, associations between ANGPTL4 and lipid parameters were analyzed among patients with and without metabolic syndrome. First, there was no association observed between lipid parameters and ANGPTL4 in those without diabetes mellitus with or without metabolic syndrome. All the patients with diabetes mellitus had metabolic syndrome; therefore, a substratified analysis could not be performed.
In support of the lipid findings, apolipoproteins A-I and B were also negatively associated with ANGPTL4 levels (P<0.01), whereas apolipoprotein C-III, a marker of triglyceride-rich lipoproteins, was positively associated with ANGPTL4 levels (P<0.02; Table 2). In a demographic-adjusted linear regression analysis, apolipoproteins A-I, B, and C-III showed similar findings as the Pearson correlation analysis (Table I in the online-only Data Supplement).
ANGPTL4 Has a Much Stronger Relationship With Metabolic Parameters Than ANGPTL3
In an unadjusted model, ANGPTL4 had a more significant positive association with body mass index (BMI) and waist circumference (P<2×10−16 for both) compared with ANGPTL3 (P<10−4; Table 3). In addition, ANGPTL4, but not ANGPTL3, was significantly and positively associated with fasting plasma glucose (P<2×10−16). ANGPTL4 levels in those with diabetes mellitus were 2 times higher than those without diabetes mellitus (181 [interquartile range, 136–237] versus 93 [interquartile range, 69–134] ng/mL; P<2.0×10−16). ANGPTL4 was also significantly and positively associated with insulin, free fatty acids, and leptin, whereas plasma adiponectin levels were significantly and negatively associated with ANGPTL4 levels. In an age-, sex-, and race-adjusted model, linear regression analysis supported the above findings (Table III in the online-only Data Supplement).
In contrast, although modest positive associations were observed with ANGPTL3 and waist circumference, BMI, hemoglobin A1C, insulin, free fatty acids, and leptin on unadjusted analyses, after adjustment for demographic parameters, ANGPTL3 was only modestly significantly associated with BMI, waist circumference, adiponectin, and leptin (Table III in the online-only Data Supplement).
Given the significant associations observed with all factors of metabolic syndrome—waist circumference, systolic blood pressure, blood glucose, triglycerides, and HDL-C, there was a clear relationship between plasma ANGPTL4 levels and the number of metabolic syndrome parameters, as observed in Figure 4B (P<2×10−16). In contrast, no association was observed between plasma ANGPTL3 levels and metabolic syndrome (Figure 4A).
In the largest study of plasma concentrations of ANGPTL3 and ANGPTL4 to date, we identified highly significant correlations of each protein with lipid and metabolic traits. Remarkably, despite their known similar biochemical properties (specifically inhibition of LPL), plasma concentrations of these 2 proteins have relatively weak associations with plasma triglyceride levels and opposite relationships with LDL-C and HDL-C levels. ANGPTL3 concentrations are significantly positively correlated with LDL-C and HDL-C levels; conversely, ANGPTL4 concentrations are significantly negatively correlated with LDL-C and HDL-C levels. Although ANGPTL4 has an expected (albeit modest) positive correlation with triglyceride levels, ANGPTL3 has no correlation with triglycerides. Strikingly, ANGPTL4 is strongly associated with multiple features of the metabolic syndrome, whereas ANGPTL3 has much less association. Previous studies had small sample sizes or less accurate assays, which resulted in inconsistent associations between these proteins and phenotypic parameters.3,5,13–18 Thus, these results provide new information about these 2 related proteins that are clearly important in lipoprotein biology in humans.
In humans, ANGPTL3 expression occurs predominantly in the liver. ANGPTL3 expression is markedly induced by activation of liver X receptor, which binds to its response element in the ANGPTL3 promoter. ANGPTL3 is proteolytically cleaved by proprotein convertases to yield the biologically active N-terminal fragment and an inactive C-terminal fragment. The N-terminal fragments form multisubunit complexes to protect against degradation. N-terminal ANGPTL3 modestly suppresses LPL catalytic activity in vitro.1 Studies of gain of function of ANGPTL3 in mouse models have shown elevation of fasting triglyceride levels because of suppression of very-low-density lipoprotein clearance via inhibition of LPL, which is more pronounced in the fed state.19 Loss of function in mice results in reduction of both triglyceride and HDL-C levels.5
Common variants at the ANGPTL3 gene locus are significantly associated with LDL-C and triglycerides. Sequencing studies of the coding region of ANGPTL3 revealed that the variant M259T among blacks (mean allele frequency, 5%) was significantly associated with lower triglyceride levels.10 The M259T variant was not associated with other metabolic parameters such as glucose and BMI. Most compellingly, exome sequencing of a large family with pan-hypolipidemia revealed 2 rare ANGPTL3 nonsense mutations that cosegregated with reduced plasma lipid levels.12 Although increased LPL activity attributable to lack of ANGPTL3 might explain the reduced triglyceride levels in this family, it does not explain the reduced LDL-C or HDL-C levels. ANGPTL3 also inhibits endothelial lipase, which hydrolyzes HDL-C phospholipids, and the effect on HDL-C levels could be related to ANGPTL3 inhibition of endothelial lipase activity.5 However, the effect on LDL-C levels remains unexplained.
Studies of plasma ANGPTL3 are sparse and provide conflicting results. One study with 250 Finnish subjects showed that ANGPTL3 was positively associated with HDL-C and negatively associated with triglycerides.14 In a second population of predominantly end-stage kidney disease and healthy subjects (n=394), ANGPTL3 was positively associated with HDL-C and LDL-C on univariate analysis.15 Our finding in a significantly larger community-based sample that plasma levels of ANGPTL3 are strongly and positively associated with LDL-C and HDL-C levels is novel and largely consistent with the human genetic data. However, we did not observe the expected positive association with triglycerides. This suggests that there are other factors involved in modulating the relationship between ANGPTL3 levels and triglyceride metabolism.
In humans, ANGPTL4 expression occurs in the liver, adipose tissue, small intestine, and heart. ANGPTL4 expression is stimulated by peroxisome proliferator-activated receptor α, which binds to response elements in the ANGPTL4 promoter. Oligomerization occurs with intermolecular disulfide bonds followed by secretion of the protein. ANGPTL4 is proteolytically cleaved via proprotein convertases into an active N-terminal fragment and inactive C-terminal fragment containing the fibrinogen-like domain. N-terminal ANGPTL4 interacts with the extracellular matrix through heparin sulfate proteoglycans and has been shown to irreversibly inhibit LPL.1 Studies of gain and loss of function of ANGPTL4 in mouse models have shown both elevation and reduction of triglycerides, respectively.20
Common variants at the ANGPTL4 gene locus are significantly associated with HDL-C. Population-based resequencing of the coding regions in 3551 subjects from the Dallas Heart Study found 1 nonsynonymous sequence variant, E40K, in European Americans (mean allele frequency, 1.3%) associated with lower plasma triglyceride and LDL-C levels and higher HDL-C levels (P=0.004).11 These findings were reproduced in 2 larger patient populations.11 This variant was not genotyped in our study population. ANGPTL4 also inhibits hepatic lipase, which is involved in HDL-C and LDL-C metabolism.21 Unlike ANGPTL3, no patients with genetic deficiency of ANGPTL4 have been reported.
Thus far, the largest scale study of plasma ANGPTL4 levels was in 666 healthy men and showed that plasma ANGPTL4 levels were inversely associated with HDL-C, but not associated with LDL-C or triglycerides.17 Smaller scale studies in a Finnish population showed no correlation of plasma ANGPTL4 levels with triglycerides, LDL-C, or HDL-C. Metabolic and inflammatory parameters such as BMI, free fatty acids, and C-reactive protein were positively associated with the protein.14,17 Our current study showed that plasma ANGPTL4 concentrations are significantly inversely associated with LDL-C and HDL-C and positively associated with triglycerides. Although the directional associations with triglycerides and HDL-C are consistent with predictions from the human genetics, the inverse association with LDL-C is not. This suggests that there are other factors involved in modulating the relationship between ANGPTL4 levels and LDL metabolism.
The biological explanation for the difference between the associations of ANGPTL3 and ANGPTL4 levels with lipid traits is not obvious. Their mechanisms of inhibition of LPL are different.1 ANGPTL3 inhibits endothelial lipase, and ANGPTL4 inhibits hepatic lipase, and although it is not clear that these are distinct qualitative differences between the 2 proteins, this could provide some explanation. Of note, there is remarkable variation in their amino acid sequence, and only 30% similarity between the 2 proteins exists.20 They differ in tissue expression, with ANGPTL4 being more broadly expressed including in intestine and adipose, whereas ANGPTL3 is more specific to liver. Finally, regulation of expression differs, with ANGPTL3 more responsive to liver X receptor and ANGPTL4 to peroxisome proliferator-activated receptor α.
Furthermore, we found that ANGPTL4 levels were highly significantly associated with all aspects of the metabolic syndrome (increased BMI and waist circumference, increased systolic blood pressure, elevated triglycerides, low HDL-C, elevated glucose, and hemoglobin A1C). There are limited data on the effects of ANGPTL4 and glucose metabolism. ANGPTL4 is directly stimulated by free fatty acids via activation of peroxisome proliferator-activated receptor α. ANGPTL4 inhibits extracellular LPL-mediated lipolysis to prevent excess free fatty acid uptake into the cell. ANGPTL4 knockout mice studies show that ANGPTL4 deficiency has a significant inhibitory effect on macrophage foam cell production.7 However, in transgenic mice, ANGPTL4 overexpression decreased macrophage content by 41% (P<0.05). Human recombinant ANGPTL4 significantly decreased macrophage uptake of oxidized LDL-C.22
By an unknown mechanism, ANGPTL4 stimulates intracellular lipolysis in adipocytes and myocytes, particularly in the fasting state.20 At hyperinsulinemic–euglycemic clamp in 24-hour fasting mice, transgenic ANGPTL4 mice showed reduced glucose utilization compared with controls (125% versus 59%; P<0.05). These data suggest that there is peripheral insulin resistance seen with ANGPTL4 overexpression.21 Further mechanistic studies are needed to better elucidate the metabolic functions of ANGPTL4.
One of the limitations of this study was the predominantly white population; therefore, the observed associations may not be generalizable across races. The study population was enriched for diabetes mellitus and metabolic syndrome. In addition, the assays that were used for measuring plasma ANGPTL3 and ANGPTL4 concentrations may have included truncated portions of the protein, which are not involved in lipid metabolism. Because this is as a cross-sectional analysis, causal relationships of ANGPTL3 or ANGPTL4 cannot be inferred with lipid and metabolic parameters; only associations can be drawn.
The goal of this study was to determine relationships between ANGPTL3 and ANGPTL4 with various lipid and metabolic parameters. These circulating proteins clearly influence human physiology and may be potential targets for lipid-lowering therapies.
Sources of Funding
This work was supported by grants R37 HL055323 and RC2 HL101834 from the National Heart, Lung, and Blood Institute.
This manuscript was sent to Qingbo Xu, Consulting Editor, for review by expert referees, editorial decision, and final disposition.
The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.113.302802/-/DC1.
- Nonstandard Abbreviations and Acronyms
- angiopoietin-like protein
- body mass index
- high-density lipoprotein cholesterol
- low-density lipoprotein cholesterol
- lipoprotein lipase
- Received November 1, 2013.
- Accepted February 24, 2014.
- © 2014 American Heart Association, Inc.
- Mandard S,
- Zandbergen F,
- van Straten E,
- Wahli W,
- Kuipers F,
- Müller M,
- Kersten S
- Köster A,
- Chao YB,
- Mosior M,
- Ford A,
- Gonzalez-DeWhitt PA,
- Hale JE,
- Li D,
- Qiu Y,
- Fraser CC,
- Yang DD,
- Heuer JG,
- Jaskunas SR,
- Eacho P
- Shimamura M,
- Matsuda M,
- Yasumo H,
- et al
- Yin W,
- Romeo S,
- Chang S,
- Grishin NV,
- Hobbs HH,
- Cohen JC
- Kersten S,
- Lichtenstein L,
- Steenbergen E,
- Mudde K,
- Hendriks HF,
- Hesselink MK,
- Schrauwen P,
- Müller M
- Robciuc MR,
- Tahvanainen E,
- Jauhiainen M,
- Ehnholm C
- Smart-Halajko MC,
- Robciuc MR,
- Cooper JA,
- et al
- Zhu P,
- Goh YY,
- Chin HF,
- Kersten S,
- Tan NS
- Lichtenstein L,
- Berbée JF,
- van Dijk SJ,
- van Dijk KW,
- Bensadoun A,
- Kema IP,
- Voshol PJ,
- Müller M,
- Rensen PC,
- Kersten S
- Georgiadi A,
- Wang Y,
- Stienstra R,
- Tjeerdema N,
- Janssen A,
- Stalenhoef A,
- van der Vliet JA,
- de Roos A,
- Tamsma JT,
- Smit JWA,
- Tan NS,
- Muller M,
- Rensen PCN,
- Kersten S
Angiopoietin-like protein 3 (ANGPTL3) and 4 (ANGPTL4) are secreted proteins that inhibit lipoprotein lipase in vitro. Genetic variants at the ANGPTL3 and ANGPTL4 gene loci are significantly associated with plasma lipid traits. However, few data exist regarding the relationship of plasma levels of ANGPTL3 and ANGPTL4 with lipid and metabolic traits. The current study is the largest scale study that highlights the associations between these 2 biomarkers and lipid and energy metabolism. Despite having similar biochemical effects in vitro, plasma ANGPTL3 and ANGPTL4 concentrations have nearly opposite relationships with plasma lipids. ANGPTL4, but not ANGPTL3, is strongly negatively associated with low-density lipoprotein cholesterol and positively with multiple features of the metabolic syndrome, whereas ANGPTL3 is positively associated with low-density lipoprotein cholesterol and high-density lipoprotein cholesterol. These novel biomarkers may be potential targets for lipid-lowering therapies.