GPIHBP1: Two get tangled

The role of triglyceride-rich lipoproteins (chylomicrons and VLDL) in promoting cardiovascular disease (CVD) has been an area of debate over many years.¹ Epidemiological and genetic evidence supports the idea of raised triglycerides, or triglyceride-rich lipoproteins (TRLs) and their remnants, as an etiology of CVD. However, human intervention trials have failed to show conclusive effects, perhaps because other risk factors are often associated with elevated triglycerides, such as insulin resistance and diabetes, or because plasma triglycerides might be a marker of CVD rather than an independent risk factor. Both fasting and non-fasting plasma triglyceride levels vary widely, with concentrations of 2–10 mM (177–885 mg/dl) conferring an association with increased risk of CVD, and concentrations greater than 10 mM conferring increased risk of acute pancreatitis.² Large-scale clinical trials on well-defined subject populations are needed to shed light onto the role of triglycerides in CVD risk.

Triglyceride-rich chylomicrons are generated in the postprandial state by intestinal enterocytes, secreted into the intestinal lymph and then released into circulation through the thoracic duct.^3–4 After reaching the bloodstream, the triglycerides within chylomicrons and VLDL are bound and hydrolyzed by lipoprotein lipase (LpL) along the luminal surface of capillaries, mainly in heart, skeletal muscle, and adipose tissue. This process allows release of fatty acids that can be used as an energy source or stored for safekeeping to be used in times of energy deprivation. However, recent studies in mice have demonstrated that adipocyte-derived LpL plays a smaller role in adipose tissue triglyceride accumulation than previously thought.⁵ Two hypotheses, which are not mutually exclusive, have been put forward to explain how the smaller fraction of LpL bound to the luminal endothelium in larger arteries, as compared to the majority bound to the capillary endothelium, might promote atherosclerosis. The “remnant hypothesis” states that LpL hydrolyzes TRLs into smaller remnants, which are then able to traverse the endothelium and enter the artery wall, where they are engulfed by macrophages. The “lipolytic toxin hypothesis” states that LpL-mediated hydrolysis of TRLs release lipids, such as fatty acids and oxidized lipids, which in turn promote pro-atherosclerotic changes in endothelial cells.¹ The dogma in the field held that LpL was tethered to endothelial cells only through the interaction of its positively charged heparin-binding domains⁶ via negatively charged heparan sulfate glycosaminoglycans (HSPGs) on the surface of endothelial cells.^7–9 LpL, however, is not synthesized by endothelial cells, but rather by myocytes and adipocytes. How LpL transverses the capillary endothelium has only recently begun to be illuminated.

In 2007, the Young laboratory demonstrated that mice deficient in glycosylphosphatidylinositol-anchored high-density lipoprotein-binding protein 1 (GPIHBP1) exhibited a striking accumulation of chylomicrons in the plasma and had plasma triglyceride levels in excess of 1000 mg/dl (11 mM) without altered LpL expression in adipocytes and myocytes.¹⁰ Furthermore the vast majority of the plasma triglycerides were located in the large lipoprotein fraction suggesting a defect in lipolytic processing of chylomicrons. Further studies revealed that GPIHBP1 expression in capillary endothelial cells is critical for the transportation of LpL from the basolateral to the capillary apical surface,^11–12 as shown in Figure 1A. Moreover, through a series of elegant experiments, this group has demonstrated the HSPGs are neither required nor sufficient to mediate LpL hydrolysis of triglycerides on the apical surface,¹³ and recently that LpL-binding to GPIHBP1, rather than LpL-binding to HSPGs is the main determinant of TRL margination in heart capillaries.¹⁴ Interestingly, GPIHBP1 is expressed in capillary endothelium, but not in large vessel endothelium or brain capillaries.^10–11 Therefore, LpL tethering to the luminal surface of large arteries susceptible to atherosclerosis is presumably more dependent on HSPGs.

Figure 1. Many mutations in the Ly6 domain of GPIHBP1 associated with hypertriglyceridemia in humans result in dimerization or multimerization of GPIHBP1 and prevent its ability to transport LpL to the lumen of capillaries.

A. Normally, GPIHBP1 acts as a monomer that transports LpL from the interstitial space to the luminal capillary endothelial cell surface, where it is required for LpL-mediated hydrolysis of triglycerides (TGs) to fatty acids (FA) and glycerol in triglyceride-rich lipoproteins (TRLs), such as chylomicrons and VLDL. This process results in the formation of lipoprotein remnants (R). B. The study by Beigneuxet al.²⁴ reveals that many, but not all, mutations in GPIHBP1 resulting in hypertriglyceridemia cause dimerization or multimerization of GPIHBP1, thereby preventing LpLbinding. HSPG, heparan sulfate glycosaminoglycans

Mature human GPIHBP1 contains a signal peptide, an acidic/negatively charged amino-terminal domain, a short linker domain, a highly conserved lymphocyte antigen 6 (Ly-6) motif (residues 65–136), and a hydrophobic carboxyl-terminal that triggers the addition of a glycosylphosphatidylinositol (GPI) anchor.¹⁵ The aspartic and glutamic acid enriched acidic domain is involved in the binding of LpL.¹⁶ The acidic domain also mediates Apo-AV interaction with GPIHBP1. The Ly-6 protein domain is characterized by 10 conserved cysteine residues that have a defined disulfide-bonding pattern. These cysteine residues generate a three-fingered structural motif. The Ly-6 motif also contains a N-linked glycosylation site (Asn-78 in human, Asn-76 in mouse) that is critical for trafficking of GPIHBP1 to the cell surface and for binding LpL. GPIHBP1 is tethered to the surface of the plasma membrane by its GPI anchor, which can be cleaved by a phosphatidylinositol-specific phospholipase C.¹⁵

To date, more than 10 different rare mutations/deletion in the GPIHBP1 gene resulting in changes in the GPIHBP1 protein in patients with severe hypertriglyceridemia have been reported. One patient with total loss of GPIHBP1 had extremely high serum triglycerides (>280 mM; 25,000 mg/dl) at the age of 2 months.¹⁷ Another family had a deletion of exons 3 and 4 in four adult individuals with serum triglycerides ranging between <2000 and 9,000 mg/dl (23–102 mM).¹⁸ Whether there was some partial function of the remaining exons 1 and 2 was not determined, but deletion of exons 3 and 4 eliminates the GPI-membrane anchor as well as the Ly6 motif. The most common mutations in GPIHBP1 associated chylomicronemia are missense mutations in one of the conserved cysteines in the LpL-binding Ly6 motif,^19–22 such as the mutations C65S, C65Y, C68G, C68Y, C89F and Q115P. It was first hypothesized that mutations of the cysteine residues would result in misfolding and accumulation of GPIHBP1 in the ER, however, work by Beigneux and colleagues¹⁹ revealed that mutating cysteine residues in the Ly6 domain to alanines did not markedly affect the ability of GPIHBP1 to reach the cell surface. Instead, the cysteine-to-alanine substitution mutants were incapable of binding and transporting LpL.¹⁹ Recently, Plengpanich et al. identified a patient with severe hypertriglyceridemia who was homozygous for a GPIHBP1 point mutation that converted a serine in the Ly6 domain (S107) to a cysteine.²³ This point mutation resulted in an increase in multimerization of GPIHBP1, and follow-up studies revealed that GPIHBP1 needs to be in a monomeric form in order to bind LpL (Figure 1A).²³ However, whether the concept of monomeric GPIHBP1 requirement for LpL transport across the endothelium extended to other reported human point mutations associated with hypertriglyceridemia was unanswered.

In this present issue of Circulation Research, Beigneux and colleagues investigated the effects of several known GPIHBP1 point mutations on dimerization and multimerization of GPIHBP1 and its ability to bind LpL in CHO cells expressing wild type human GPIHBP1 or GPIHBP1 mutants, rat heart microvascular endothelial cells and Drosophila S2 cells, and also performed cell-free LpL–GPIHBP1 binding assays.²⁴ Many of the point mutations evaluated were discovered in chylomicronemic patients. Mutation of cysteine residues in the Ly6 domain resulted in increased multimerization and loss of LpL binding (Figure 1B). Several non-cysteine mutations that also increased GPIHBP1 multimerization were located in the critical second finger of the Ly6 domain. Thus, both cysteine and “non-cysteine” mutations reduce the formation of functionally active GPIHBP1 monomers. Furthermore, a mutation of W109 to one of several other amino acids resulted in a decreased rate of multimerization compared to wild type GPIHBP1, although was still defective in binding LpL, suggesting that this highly conserved residue may directly interact with LpL. Further studies with crystallography or nuclear magnetic resonance will be needed to determine the specific details of the interaction. Furthermore, because these studies are based on in vitro overexpression systems, analysis of GPIHBP1 dimerization and multimerization in animal models and humans would be a logical and interesting next step.

It is hard not to draw similarity between the current manuscript²⁴ and that of Plengpanich et al.²³ by the same group. In that paper, the serine to cysteine (S107C) mutation discovered in a Thai family was found to not affect trafficking of GPIHBP1 to the cell surface, but to result in GPIHBP1 multimerization. Furthermore, the authors demonstrated in that study that LpL greatly prefers to bind monomeric GPIHBP1 (wild type GPIHBP1 also forms multimers), moreover, they attributed the lack of LpL binding of two previously identified GPIHBP1 mutants (S107C and C68G²¹) to the reduction of monomers. In the current study by Beigneux et al.,²⁴ the authors studied more than 15 different mutations in the Ly6 domain with special emphasis on the cysteines. Together, both manuscripts^23–24 strongly suggest that the majority of GPIHBP1 mutations associated with chylomicronemia are correlated with multimerization of the protein and loss of LpL-binding capacity. The possibility that GPIHBP1 dimerization/multimerization or other mutations found in humans also interfere with binding to Apo-AV (a component of chylomicrons, VLDL and other lipoproteins), and the relative physiological relevance of such an effect visavi the LpL-binding effect need to be addressed in future studies.

Recently, a polymorphism (rs72691625) was identified in the GPIHBP1 gene promoter.²⁵ Carriers of this g.-469G>A polymorphism have a significantly higher risk of elevated triglycerides (2.0 mM; 177 mg/dl) than non-carriers. However, whether this polymorphism affects gene expression was not determined. Nonetheless, it suggests that modulation of GPIHBP1 expression or activity could offer a therapeutic treatment strategy for hypertriglyceridemia.

Finally, it is interesting to ponder the potential role of GPIHBP1 in CVD in humans. The Gpihbp1^−/− mouse, which develops severe chylomicronemia even on a low-fat diet,¹⁰ exhibits small macrophage-rich fatty streak lesions of atherosclerosis in the aortic sinus and coronary arteries at 11–12 months of age and larger aortic sinus lesions at 22 months of age.²⁶ These pro-atherosclerotic effects of GPIHBP1-deficiency are probably due to the markedly elevated levels of chylomicrons/VLDL, which might exacerbate atherosclerosis through increased remnant formation in large arteries and/or through increased generation of pro-atherogenic lipid products. It is clear, however, that the effect of GPIHBP1-deficiency on atherosclerosis is weak compared to that of LDL receptor-deficiency or ApoE-deficiency in mouse models, which develop large and complex advanced lesions when fed low-fat diets at the ages studied in the Gpihbp1^−/− mice.

The studies by Beigneux and colleagues²⁴ have undoubtedly advanced our understanding of how GPIHBP1 mediates LpL transport, and the effect of GPIHBP1 point mutations in humans with hypertriglyceridemia. Further understanding of the role of GPIHBP1 and TRLs in the development of CVD in humans is urgently needed.