Endothelial lipase mediates efficient lipolysis of triglyceride-rich lipoproteins

Abstract

Triglyceride-rich lipoproteins (TRLs) are circulating reservoirs of fatty acids used as vital energy sources for peripheral tissues. Lipoprotein lipase (LPL) is a predominant enzyme mediating triglyceride (TG) lipolysis and TRL clearance to provide fatty acids to tissues in animals. Physiological and human genetic evidence support a primary role for LPL in hydrolyzing TRL TGs. We hypothesized that endothelial lipase (EL), another extracellular lipase that primarily hydrolyzes lipoprotein phospholipids may also contribute to TRL metabolism. To explore this, we studied the impact of genetic EL loss-of-function on TRL metabolism in humans and mice. Humans carrying a loss-of-function missense variant in LIPG, p.Asn396Ser (rs77960347), demonstrated elevated plasma TGs and elevated phospholipids in TRLs, among other lipoprotein classes. Mice with germline EL deficiency challenged with excess dietary TG through refeeding or a high-fat diet exhibited elevated TGs, delayed dietary TRL clearance, and impaired TRL TG lipolysis in vivo that was rescued by EL reconstitution in the liver. Lipidomic analyses of postprandial plasma from high-fat fed Lipg-/- mice demonstrated accumulation of phospholipids and TGs harboring long-chain polyunsaturated fatty acids (PUFAs), known substrates for EL lipolysis. In vitro and in vivo, EL and LPL together promoted greater TG lipolysis than either extracellular lipase alone. Our data positions EL as a key collaborator of LPL to mediate efficient lipolysis of TRLs in humans and mice.

Fig 1. Association of the LIPG p.Asn396Ser variant with plasma lipids in humans.

A-D. Association of LIPG p.Asn396Ser with plasma total cholesterol (A), HDL-C (B), LDL-C (C), and triglycerides (D) in the UK Biobank (UKBB) whole exome sequencing subset, UKBB genome-wide genotyping subset, Global Lipids Genetics Consortium (GLGC), and Million Veteran Program (MVP) cohorts, and a fixed-effects meta-analysis of these cohorts. Effect estimates after log-transformation for each lipid trait in standard deviation (S.D.) units and 95% confidence intervals are plotted.

Fig 2. Association of LIPG p.Asn396Ser variant with serum lipoprotein and lipid metabolites by NMR.

A. Association of LIPG p.Asn396Ser variant with lipoprotein subclass concentrations measured through NMR spectroscopy from human serum from a European cohort comprising up to 24,925 individuals as described previously [31] (see Materials and Methods). B. association of LIPG p.Asn396Ser variant with phospholipid concentrations in serum and in lipoprotein subclasses from analysis in (A). Effect estimates for each log-transformed lipid trait in standard deviation (S.D.) units and 95% confidence intervals are plotted. Single red star (*) indicates measures with nominally significant associations between LIPG p.Asn396Ser carriers and noncarrier controls (P<0.05); double red star (**) indicates measures with experiment-wide significant differences between carriers and controls (FDR < 0.05).

Fig 3. Genetic loss-of-function of Lipg and plasma triglycerides in mice.

A. Plasma TC, HDL-C, nonHDL-C, TGs, phospholipids, and nonesterified fatty acids from WT vs. Lipg^-/- mice after a 4 hour fast and again after 4 hours of refeeding a chow diet. Lipids were measured by chemical autoanalyzer. B. Percent change in each plasma lipid measure from (A) in 4 hour refed vs fasting measures. C. TG from plasma of fasted mice (left) and four hour-refed mice (right) fractionated by fast-protein liquid chromatography (FPLC) from pools from WT or Lipg^-/- mice from groups in (A). D. Total cholesterol, HDL-C, nonHDL-C, TGs, phospholipids, and nonesterified fatty acids in WT and Lipg^-/- mice after initiation of feeding adult mice a diet composed of 45% kilocalories fat for the indicated timepoints. Plasma was collected 4 hours after fasting and lipids were measured by autoanalyzer. E. Fasting plasma lipids after 7 weeks of high fat feeding were normalized to the levels at zero weeks of feeding and expressed as a percentage change. F. Triglycerides from fasted (left) and refed (mice) from (D-E) after FPLC-fractionation of plasma. A, B, E: *P<0.05, **P<0.01, ***P<0.001, Student’s unpaired T-test. D: **P<0.01, ***P<0.001, repeated factor 2-way ANOVA compared to WT group. Data is expressed as mean ± S.E.M.

Fig 4. Postprandial chylomicron clearance in Lipg-deficient mice.

A. Plasma triglycerides (left) and area under the triglyceride clearance curves (right) in EL WT vs. Lipg^-/- mice fed a regular chow diet after overnight fasting and oral fat tolerance testing with olive oil gavage. B. Cholesterol and triglycerides from FPLC fractions of pooled plasma from the 7 hour timepoint from the olive oil gavage in A. C. Plasma triglycerides (left) and area under the triglyceride clearance curves (right) in EL WT vs. Lipg^-/- mice fed a high fat diet for 12 weeks after overnight fasting and oral fat tolerance testing with olive oil gavage. D. Cholesterol and triglycerides from FPLC fractions of pooled plasma from the 7-hour timepoint from the olive oil gavage in (C). E. Plasma TGs (left) and areas under the curve (right) after olive oil gavage in EL WT mice treated with Null adeno-associated virus vector (WT-AAV Null), Lipg^-/- mice treated with Null vector (Lipg^-/—AAV Null), or Lipg^-/- mice treated with AAV expressing murine Lipg directed to the liver (Lipg^-/—AAV WT). A left, C left, E left: P<0.05, repeated factor 2-way ANOVA, A right, C right, E right: *P<0.05, **P<0.01, student’s unpaired T-test. Data is expressed as mean ± S.E.M.

Fig 5. Lipg deficiency delays TRL lipolysis in mice.

A.³H oleate plasma clearance curves from EL WT vs. Lipg^-/- mice administered ³H-triolein labeled human TRLs. Mice were administered radiolabeled TRLs and plasma ³H levels were measured over 15 minutes and normalized to the 1 minute timepoint. B. Clearance of ¹²⁵I tyramine cellobiose labeled human TRLs in EL WT vs. Lipg^-/- mice fed a high fat diet for 12 weeks. Plasma clearance of ¹²⁵I-TRL apolipoprotein B (apoB) after intravenous TRL administration was measured over 15 minutes. C. ³H-triolein-labeled TRL clearance in WT mice treated with AAV Null vector, Lipg^-/- mice treated with AAV Null, or Lipg^-/- mice treated with AAV expressing murine Lipg and all fed a high-fat diet for 12 weeks and then treated with radiolabeled TRLs as described in (A). D. Uptake of ³H-oleate into the indicated tissues after 15 min of administration of ³H-triolein labeled TRLs in the mice from (A). ³H activity in a fixed amount of tissue was normalized to the 1 minute plasma ³H activity for each mouse. Normalized tissue ³H activity is expressed for each tissue relative to the mean of the WT group. E. Hepatic mRNA expression of Lpl and Lipc in mice from (A). Quantitative real-time PCR cycle number for each gene was normalized to that of β-actin. F. Immunoblots of LPL and HL from liver lysates of mice in (A). Immunoblots of the proteins from the lysates for β-actin are shown as a loading control. A, C: *P<0.05, ****P<0.0001, repeated factor 2-way ANOVA. D & E: *P<0.05, **P<0.01, ****P<0.0001, student’s unpaired T-test. Data is expressed as mean ± S.E.M.

Fig 6. EL and LPL cooperate to promote TRL lipolysis in vivo.

A. Volcano plot of 77 plasma triglyceride (TG) species from plasma of WT vs Lipg^-/- mice after overnight fasting (filled black circles) and 7 hours (white circles) after oral fat tolerance test. Lipids were measured by liquid-chromatography tandem mass spectrometry (LC-MS) as described in the Materials and Methods. Plot shows fold change in Lipg^-/- mice vs WT (x-axis; logarithmic base-2 scale) with the p-value (y-axis; logarithmic base-10 scale). Vertical dotted lines indicate two-fold change, and horizontal dotted line corresponds to P<0.05. B. (left and right) Plasma concentration of TG species in the red line from (A) as measured by LC-MS at the indicated timepoints. C. Volcano plot of 28 plasma phosphatidylcholine (PC) species from plasma of WT vs Lipg^-/- mice after overnight fasting (filled black circles) and 7 hours (white circles) after oral fat tolerance test. Lipids were measured by liquid-chromatography tandem mass spectrometry (LC-MS) as described in the Materials and Methods. D. (left and right) Plasma concentration of PC species in the red line from (C) as measured by LC-MS at the indicated timepoints. E. In vitro TG lipase activity of EL and LPL conditioned media on ³H-triolein labeled TRL-like emulsions. Emulsions were incubated with the indicated volumes of EL or LPL-containing conditioned medium from COS7 cells infected with EL- or LPL-expressing adenovirus. Release of ³H-oleate was used to measure TG lipase activity. F. In vitro phospholipase activity on ³H-triolein labeled TRL-like emulsions in reactions with LPL of the indicated volume and increasing volumes of EL WT or EL S169A conditioned media. G. In vitro phospholipase activity of EL and LPL conditioned media on ¹⁴C-DPPC-labeled TRL-like emulsions, as shown in (A-B). H. TG-lipase activity on ¹⁴C-DPPC labeled TRL-like emulsions in reactions with LPL of the indicated volume and increasing volumes of EL WT or EL S169A conditioned media. I. TG-lipase activity on ³H-triolein labeled human VLDLs incubated with recombinant EL WT, LPL or a combination of EL WT and LPL. Red dotted line indicates the sum of the mean activities for the EL WT (red bar) and LPL (blue bar) groups. B and C: *P<0.05, *student’s unpaired T-test comparing the indicated groups. E-H: ****P<0.0001, two-way ANOVA for interaction term of indicated groups and sample volume term (X-axis). I: ****P<0.0001, one-way ANOVA comparing EL WT + LPL group (white bar) with other groups. Data is expressed as mean ± S.E.M.

Fig 7. Impact of LPL inhibition with ANGPTL4 overexpression in Lipg^-/- mice.

A-D. Plasma HDL-C, nonHDL-C, phospholipids, and TGs from WT vs. Lipg^-/- mice treated with AAV Null vector (WT-AAV Null; Lipg^-/-—AAV Null) or AAV ANGPTL4 vector (WT-AAV ANGPTL4; Lipg^-/-—AAV ANGPTL4) after a 4 hour fast. Lipids were measured by chemical autoanalyzer. E-F. Plasma TGs 7 hours after oral fat tolerance testing in mice from (A). TGs in mg/dl are shown in (E). The relative difference in percent (%) increase above the WT–AAV Null group in plasma TGs from each group in (E) is shown in (F). In (F) the red dotted line shows the mean summed % increase in 7 hour plasma TGs for the Lipg^-/-—AAV Null and the WT-AAV ANGPTL4 group. For A-D, *P<0.05, **P<0.01, ***P<0.001, two-way ANOVA comparing the indicated groups to the WT AAV-Null group. For E and F, ****P<0.0001, two-way ANOVA across groups comparing interaction of genotype (WT vs Lipg^-/-) and AAV (Null vs ANGPTL4). All error bars indicate mean ± S.E.M.

Fig 8. Schematic of role of EL in TRL catabolism.

A-B. In the setting of EL expression (A), phospholipid lipolysis on TRL in postprandial or high fat-fed state allows for synergistic and efficient TG lipolysis by LPL, which promotes FA and TRL remnant uptake by the liver. Multiple FAs including long chain PUFAs are internalized by the liver. In EL-deficient mice (B), lipolysis of phospholipids on the outer shell of TRLs is impaired, resulting in less efficient access of LPL to TRL TGs and reduced TG lipolysis and remodeling of TRLs to smaller remnants, thus reducing TRL remnant uptake by the liver. In addition, the lack of ability of EL to catalyze lipolysis of phospholipids containing its preferred substrates of long chain PUFAs results in decreased PUFA uptake by the liver.