High‐density lipoprotein in diabetes: Structural and functional relevance

Abstract

Low levels of high‐density lipoprotein‐cholesterol (HDL‐C) is considered a major cardiovascular risk factor. However, recent studies have suggested a more U‐shaped association between HDL‐C and cardiovascular disease. It has been shown that the cardioprotective effect of HDL is related to the functions of HDL particles rather than their cholesterol content. HDL particles are highly heterogeneous and have multiple functions relevant to cardiometabolic conditions including cholesterol efflux capacity, anti‐oxidative, anti‐inflammatory, and vasoactive properties. There are quantitative and qualitative changes in HDL as well as functional abnormalities in both type 1 and type 2 diabetes. Non‐enzymatic glycation, carbamylation, oxidative stress, and systemic inflammation can modify the HDL composition and therefore the functions, especially in situations of poor glycemic control. Studies of HDL proteomics and lipidomics have provided further insights into the structure–function relationship of HDL in diabetes. Interestingly, HDL also has a pleiotropic anti‐diabetic effect, improving glycemic control through improvement in insulin sensitivity and β‐cell function. Given the important role of HDL in cardiometabolic health, HDL‐based therapeutics are being developed to enhance HDL functions rather than to increase HDL‐C levels. Among these, recombinant HDL and small synthetic apolipoprotein A‐I mimetic peptides may hold promise for preventing and treating diabetes and cardiovascular disease.

HDL particles are highly heterogeneous, and there are quantitative and qualitative changes as well as functional abnormalities of HDL particles in both type 1 and type 2 diabetes. HDL dysfunction is common in diabetes and contributes to the increased cardiovascular risk in people with diabetes.

INTRODUCTION

Cardiovascular diseases are the leading cause of death in both type 1 and type 2 diabetes ¹ , and diabetic dyslipidemia is a major risk factor for atherosclerotic cardiovascular disease (ASCVD). Lowering low‐density lipoprotein cholesterol (LDL‐C) with statin therapy effectively reduces cardiovascular risks in people with and without diabetes ² but substantial residual risks still exist despite the documented benefits of statin ³ . In addition to elevated LDL‐C, a low level of high‐density lipoprotein cholesterol (HDL‐C) is another important risk factor for ASCVD. Earlier epidemiological studies have shown an inverse correlation between HDL‐C levels and the risk of coronary artery diseases ⁴ . Recent studies have further revealed a U‐shaped relationship between HDL‐C levels and cardiovascular events such that very high HDL‐C levels are also associated with increased cardiovascular risks ⁵ . In this review, we will discuss the epidemiology of HDL‐C and cardiovascular risks in diabetes, changes in HDL composition and function (Figure 1), and the therapeutic implications of modifying HDL in the context of diabetes.

Figure 1.

HDL structure and function in diabetes. The structure and composition of HDL particles, proteome, and lipidome of HDL are altered in both type 1 and type 2 diabetes, which in turn are associated with impaired functions of the HDL. Various factors including hyperglycemia, non‐enzymatic glycation, carbamylation, oxidation, inflammation, cardiometabolic risk factors, therapeutics for diabetes and dyslipidemia, and lifestyle factors such as smoking and alcohol play a part in such modifications.

HDL‐C LEVELS AND CARDIOVASCULAR RISKS

Epidemiological studies have unequivocally established low HDL‐C levels as a risk factor for ASCVD ⁶ , ⁷ . The concept of high HDL‐C being protective, the ‘HDL hypothesis’ ⁸ , is supported by preclinical studies showing that increasing HDL‐C levels in animal models reduced atherosclerosis ⁹ , ¹⁰ . As a result, there was intense interest in the development of HDL‐raising therapies. Unfortunately, the results were disappointing when randomized controlled trials with niacin ¹¹ and CETP inhibitors ¹² did not show a reduction in cardiovascular events, despite significantly increasing HDL‐C levels. Although the trial with anacetrapib did show a significant reduction in cardiovascular events, the benefit was mediated through the reduction in LDL‐C levels rather than the increase in HDL‐C levels ¹³ . Mendelian randomization studies also suggested that genetic variations that increased HDL‐C levels were not associated with lower cardiovascular risks ¹⁴ . Indeed, higher HDL‐C levels are not always protective ¹⁵ . Recent studies have reported the U‐shaped relationships between HDL‐C levels and cardiovascular risks in different populations. In the CANHEART study of people without pre‐existing cardiovascular diseases, high HDL‐C levels were associated with increased mortality ¹⁶ . Data from the UK Biobank and the Emory Cardiovascular Biobank revealed that among patients with confirmed coronary artery disease, very high HDL‐C levels were associated with increased risks of all‐cause and cardiovascular mortality ¹⁷ . This U‐shaped relationship may be amplified in people with diabetes ¹⁸ .

All the above evidence may seem to negate the ‘HDL hypothesis’. However, it is important to recognize that the ‘cardio‐protective’ effect of HDL is related to the functions of HDL particles, and HDL‐C is a poor surrogate marker of HDL functionality. HDL particles possess multiple biological activities which can influence glycemia, immunity and inflammation, atherosclerosis, and the function of vascular cells ¹⁹ . This review will focus predominantly on the vascular and metabolic effects of HDL. There is substantial evidence demonstrating the association between HDL dysfunction and the risk of ASCVD ²⁰ . Both type 1 and type 2 diabetes may induce changes in the HDL structure and composition leading to abnormalities in HDL functionality ²¹ . Understanding the changes in the HDL structure and function in type 1 and type 2 diabetes allows possible future development of therapeutic strategies based on increasing HDL subpopulations that are beneficial, or therapies that replace or enhance the beneficial roles or functions of particular HDL subspecies for the optimization of cardiometabolic risks ¹⁹ .

HDL COMPOSITION, SUBFRACTIONS, AND SUBSPECIES

HDL comprises a family of lipoprotein particles which differ in density, size, charge, protein, and lipid composition, and HDL particles are highly heterogeneous ²² . The precursors of spherical HDL are discoidal particles, which are synthesized in the liver and intestine or assembled in the plasma compartment. The apolipoprotein A‐I (apoA‐I), the main apolipoprotein constituent of HDL, acquires cholesterol and phospholipids through its interaction with ATP‐binding cassette transporter A1 (ABCA1), leading to the formation of pre‐β HDL particles ²³ . These particles gradually accumulate more cholesterol, and further maturation of HDL occurs through binding to ABCG1, generating the spherical HDL particles that predominate in normal human plasma ²² , ²⁴ . Following the esterification of cholesterol by the enzyme lecithin‐cholesterol acyltransferase (LCAT), cholesteryl ester is transferred to the core of the HDL particles, which may undergo clearance by the hepatic scavenger receptor ²³ , or be transferred to VLDL/LDL for catabolism via the cholesteryl ester transfer protein (CETP).

Spherical HDL particles consist of a central core of neutral lipids surrounded by a phospholipid monolayer in which apolipoproteins are embedded. HDL particles can act as a transport platform and it has been estimated that HDL transports at least 96 different proteins and over 200 different lipid species ²⁵ . The proteome and lipidome of HDL particles is highly complex and can influence the biological activity of a given HDL particle. For instance, other than apoA‐I and A‐II, enzymes (such as paraoxonase‐1 [PON1], platelet‐activating factor acetyl‐hydrolase [PAF‐AH], and LCAT), lipid transfer proteins (such as CETP), acute‐phase response proteins (such as serum amyloid protein [SAA]), complement components, and proteinase inhibitors (alpha‐1‐antitrypsin) ²² can be found in HDL particles. Incorporation of specific protein components into an HDL particle is subject to the structural confines posed by the size and surface constraints of the particle.

HDL particles can be categorized into subfractions by their physicochemical properties ¹⁹ . Analytical methods to subfractionate HDL particles include ultracentrifugation, differential ion mobility, two‐dimensional gel electrophoresis, and nuclear magnetic resonance (NMR) spectroscopy. Most HDL particles have a density between 1.063 and 1.21 g/mL ²⁶ , and can be subdivided into two main subfractions HDL₂ and HDL₃ ²⁷ , ²⁸ . HDL particles can be further subdivided by size into the smaller size particles HDL_3c, HDL_3b, HDL_3a, and the larger size HDL_2a and HDL_2b ²² , ²⁸ , or by charge into pre‐α, pre‐β, and α‐HDL ²⁸ . Based on chemical shifts, NMR spectroscopy separated HDL into large, medium, and small particles ²⁸ , ²⁹ . HDL subfractions isolated by different methods may not contain the same groups of HDL particles. Despite the use of more advanced high resolution separation methods, there is still substantial heterogeneity within HDL subfractions ³⁰ , which partly contributes to the inconsistencies in findings from studies investigating the association between HDL subfractions and ASCVD ¹⁹ .

HDL can also be classified into subspecies (or sub‐particles) based on a specific component of HDL ¹⁹ . Some of these HDL subspecies have been shown to execute specific biological functions. Pre‐β1 HDL subspecies consist of lipid‐poor HDL particles that contain predominantly apoA‐I¹⁹, and pre‐β1 HDL has been shown to stimulate ABCA1‐mediated cholesterol efflux ³¹ . Another example of HDL subspecies is a functional ternary complex formed by myeloperoxidase (MPO), PON1, and apoA‐I. The presence of MPO renders the HDL subspecies dysfunctional, as MPO is a source of reactive oxygen species and impairs PON1 and apoA‐I function ³² . HDL subspecies can also be identified based on their minor lipid component. HDL subspecies containing the bioactive sphingolipid sphinosine‐1‐phosphate (HDL‐S1P) can regulate vasodilation by activating the S1P‐signaling pathway which controls nitric oxide (NO) production and vasodilation ³³ . S1P is bound to apoM, a minor apolipoprotein in HDL.

HDL FUNCTIONALITY

HDL has multiple important functions that can affect the vasculature. Firstly, HDL and apoA‐I play an important role in cholesterol efflux. This is the first step in the reverse cholesterol transport (RCT) pathway. Excess cholesterol from macrophages in the arterial wall is acquired by HDL and apoA‐I and transported to the liver for excretion in bile ³⁴ . Larger, spherical HDL particles accept the cholesterol exported from the cells by the ATP‐binding cassette transporter, ABCG1, while the related transporter, ABCA1, exports cellular cholesterol to lipid‐free apoA‐I and small dense HDL particles. Many cohort studies have demonstrated an inverse association between cholesterol efflux capacity (CEC) of HDL and cardiovascular risks which is independent of HDL‐C levels ³⁵ .

Secondly, HDL can reduce inflammation in various cell types, including endothelial cells and macrophages. In endothelial cells, HDL inhibits inflammation by reducing activation of nuclear factor‐kappa B and 3β‐hydroxysteroid‐delta24‐reductase ³⁶ . Inflammasome activation is also reduced ³⁷ , whereas the cytoprotective enzyme heme oxygenase‐1 is activated ³⁶ . These effects are mediated partly by the interaction between S1P in HDL with S1P receptors ³⁸ . HDL also reduces inflammation in monocytes and attenuates the binding of monocytes to the adhesion molecules on the surface of activated endothelial cells ³⁹ . Epidemiological studies have shown a causal relationship between inflammation and incident cardiovascular diseases ¹⁵ , ⁴⁰ . The loss of the anti‐inflammatory activity of HDL may have detrimental effects.

Thirdly, the unregulated uptake of oxidized LDL by macrophages and the formation of foam cells lead to the development of atherosclerotic lesions. Lipids and apolipoprotein B in LDL are susceptible to modifications in the atherosclerotic lesions by oxidative agents, including reactive oxygen species, and MPO ⁴¹ . HDL particles exert their anti‐oxidative effect by reducing oxidative stress in LDL and other atherogenic lipoproteins by accepting lipid hydroperoxides and detoxifying them into lipid hydroxides that are cleared from the circulation by the liver. Small HDL particles inhibit oxidative stress more effectively than large HDL particles ⁴² . The anti‐oxidant enzyme PON1 ⁴³ , ⁴⁴ and PAF‐AH ⁴⁵ contribute to the anti‐oxidative capacity of HDL ⁴ .

Fourthly, HDL particles exert vasoactive functions. HDL particles can protect endothelial function and integrity by promoting junction closure ⁴⁶ , preventing the loss of endothelial glycocalyx ⁴⁷ and promoting the proliferation of vascular endothelial cells ⁴⁸ . They can induce endothelial NO synthase (eNOS) leading to vasorelaxation ⁴⁹ . They can also inhibit the expression of endothelial adhesion molecules such as vascular cell adhesion molecule 1 (VCAM‐1), preventing the adhesion of mononuclear leukocytes to the endothelium ⁵⁰ .

TYPE 2 DIABETES: CHANGES IN HDL‐C LEVELS AND HDL COMPOSITION

In type 2 diabetes, insulin resistance has a well‐known impact on lipid metabolism ⁵¹ . The lipid profile is often characterized by hypertriglyceridemia and low HDL‐C levels ⁵² (Table 1). Exposure to hyperglycemia and the abnormal milieu associated with type 2 diabetes modifies the kinetics, composition, and function of HDL particles ²⁸ . There is increased HDL turnover due to an increase in both the clearance and synthesis of apoA‐I. The triglyceride enrichment of HDL makes it susceptible to hepatic lipase action, causing HDL degradation and leaving apoA‐I free to be cleared by renal receptors. SAA deposition in HDL in type 2 diabetes also enhances the clearance of HDL particles ²⁸ .

Table 1.

Changes in the HDL‐C levels and HDL composition in type 1 and type 2 diabetes

Type 2 diabetes	Reference
Low HDL‐C levels	[52]
Increased HDL turnover	[28]
Triglyceride enrichment in HDL	[28]
Shift in favor of smaller cholesterol‐poor HDL subfractions: lower levels of medium and large HDL particles and increased level of small HDL particles	[53, 55, 56]
↑SAA, apoC, fibrinogen, MPO ↓ApoA, PON1, apoD, apoE, apoF, apoJ, apoM	[58, 59, 70, 71]
↑Phosphatidylethanolamine species ↓Ether‐phosphatidylcholines, lysophosphatidylcholines, phosphatidylinositols, sphingomyelins, S1P, and 18:2 species of cholesteryl ester	[59, 97]

Type 1 diabetes	Reference
Increased HDL‐C levels	[74]
Triglyceride enrichment in HDL	[74, 77]
Increased HDL particle size: more large HDL particles and fewer small HDL particles	[75, 76]
↓PON1, apoM ↑Complement factor H‐related protein 2, protease inhibitors	[70, 81]
↓S1P ↑Ratios of free cholesterol/phosphatidylcholine and sphingomyelins/phosphatidylcholine	[74, 77, 79]

Abnormalities in HDL subfractions include lower levels of cholesterol‐rich, larger HDL₂, and higher levels of cholesterol‐poor HDL₃ in people with type 2 diabetes ⁵³ . Two‐dimensional gel separation studies also showed a lower level of large α‐1, α‐2, and pre‐α1 particles, and higher levels of lipid‐poor α3 HDL particles ⁵⁴ . NMR studies showed that people with type 2 diabetes have reduced levels of medium and large HDL particles but increased levels of small HDL compared with those without diabetes ⁵⁵ , ⁵⁶ . Overall, there appears to be a shift in favor of smaller cholesterol‐poor HDL subfractions.

There are also changes in the HDL proteome and lipidome in type 2 diabetes. Phospholipids and sphingophospholipids play a major role in HDL functions, either by binding to a specific receptor such as S1P, or by modulating the physicochemical properties of HDL ⁵⁷ . Surface lipids (phosphatidylcholine, ether‐linked phosphatidylcholine, sphingomyelin, ceramides, and free cholesterol) are reduced in type 2 diabetes, leading to a change in the architecture and fluidity of HDL which can affect function ⁵⁸ . Changes in the protein contents of HDL in people with type 2 diabetes included a significant reduction in apoA‐I levels ⁵⁸ . A recent study of HDL proteome revealed an increase in 17 proteins and decrease in 44 proteins ⁵⁹ . Notable ones included increased SAA, fibrinogen, apoC‐II, and apoC‐III levels, and reduced apoA‐II, apoE, apoM, and PON‐1.

Specific HDL subspecies might have a relevant role in type 2 diabetes. The plasma pre‐β1 HDL level was significantly decreased in type 2 diabetes and was associated with reduced cholesterol efflux mediated by ABCA1 ⁶⁰ . ApoC‐III HDL was strongly associated with the risk of type 2 diabetes and lower insulin sensitivity ⁶¹ , partly related to the effect of apoC‐III on triglyceride metabolism. S1P‐HDL (or apoM‐HDL) levels inversely correlated with body mass index and insulin resistance in humans ⁶² , and may be due to apoM/S1P activating AKT and AMPK insulin signaling pathways through S1P receptors. In addition, apoM has been shown to improve mitochondrial function in liver and adipose cells.

TYPE 2 DIABETES: HDL DYSFUNCTION

Abnormalities in HDL structure and composition can lead to impairment in HDL functionality ⁶³ . CEC is the most extensively studied HDL function in type 2 diabetes because of its strong association with prevalent and incident atherosclerotic cardiovascular diseases ⁶⁴ , ⁶⁵ , ⁶⁶ . Our group ⁶⁷ and others ⁶⁸ have shown that the CEC of HDL particles is reduced in people with type 2 diabetes. He and colleagues further demonstrated that there was a selective attenuation of ABCA1‐mediated cholesterol efflux to small HDL on a per particle basis in people with type 2 diabetes ⁶⁹ . This was due to a reduced content of the serpin family A member 1 (SERPINA1), an anti‐protease and phospholipid‐binding protein, in the small HDL particles in type 2 diabetes ⁶⁹ . HDL dysfunction in type 2 diabetes also includes impaired anti‐oxidative activity contributed by the reduced activity of the HDL‐associated anti‐oxidant enzyme PON1 ⁷⁰ . Besides, anti‐inflammatory activity is also impaired, one of the contributing factors being the enrichment of HDL in SAA ⁷¹ . Further details are discussed in the section ‘Effects of diabetes on HDL function’.

Changes in the HDL composition significantly affect the anti‐atherogenic functions of HDL. A recent comprehensive study of structure–function relationships of HDL in type 2 diabetes clearly demonstrated that specific modifications of the proteomic and lipidomic components of HDL particles can trigger specific functional abnormalities in type 2 diabetes, and HDL dysfunction has diverse biological effects. Cardner and colleagues investigated the relationship between the proteome and lipidome of HDL and five distinct HDL functions in obese adults with type 2 diabetes who did not have coronary artery disease ⁵⁹ . The HDL functions being studied were CEC, inhibition of apoptosis of endothelial cells and β‐cells, rescue of mitochondrial membrane potential of myotubes, and mitochondrial respiration of adipocytes. Their study revealed both loss and gain of specific lipids and proteins in HDL particles, with depletion in apoA‐IV, PON1, PON3, apoD, apoE, apoF, apoJ, and apoM, and an increase in SAA1 and SAA2, apoC‐II, apo‐CIII, and fibrinogen. Hence, there is a loss of beneficial functional proteins and a gain of proteins with deleterious functions. For example, SAA‐enriched HDL particles have impaired anti‐inflammatory capacity and in fact become pro‐inflammatory, increasing TNF‐α secretion in peripheral blood mononuclear cells ⁷² . Loss of PON1 was associated with a reduced ability to protect LDL particles from oxidative damage ⁷³ . Changes in the lipidome included enrichment in phosphatidylethanolamine species and depletion in ether‐phosphatidylcholines, lysophosphatidylcholines, phosphatidylinositols, sphingomyelins, and the 18:2 species of cholesteryl ester. Although they did not observe any changes in CEC in their study, they reported that HDL was dysfunctional in all four other aspects that were studied ⁵⁹ . Their data showed that various cellular functions of HDL only weakly correlated with each other, and CEC could not be used as a proxy for other HDL functions. Different cellular functions of HDL were determined by different structural components ⁵⁹ . Non‐CEC functions were typically associated with minor HDL proteins.

TYPE 1 DIABETES: CHANGES IN HDL‐C LEVELS AND HDL COMPOSITION

On the whole, HDL metabolism and function is less extensively studied in type 1 diabetes compared with type 2 diabetes. The mechanism(s) related to insulin deficiency and dyslipidemia in type 1 diabetes are less well understood ⁵¹ . Data regarding the HDL‐C levels in type 1 diabetes are inconsistent, although most studies show increased plasma HDL‐C levels ⁷⁴ (Table 1). This might be explained by the elevated lipoprotein lipase/hepatic lipase ratio. The increased lipoprotein lipase activity could be related to peripheral hyperinsulinemia resulting from subcutaneous insulin administration. Regarding HDL subfractions, Colhoun and colleagues showed that people with type 1 diabetes had more large HDL particles and fewer small HDL particles with a resultant increase in the average HDL particle size ⁷⁵ . A more recent study also reported larger mean HDL particle size and lower HDL particle numbers in type 1 diabetes ⁷⁶ .

TYPE 1 DIABETES: HDL DYSFUNCTION

Previous studies of people with type 1 diabetes showed a significant reduction in the CEC of HDL independent of glycemic control ⁷⁷ . In contrast, Ahmed and colleagues reported that a larger mean HDL particle size was associated with an increased total HDL‐mediated CEC in recent‐onset type 1 diabetes. Even after accounting for the HDL particle size, the total HDL‐mediated CEC was increased, but not for ABCA1‐dependent CEC ⁷⁶ . The discrepancies in the findings between studies may be explained by the differences between the averaged activity of all particles in a total HDL fraction compared with the activity in certain subfractions or subspecies, the differences in the cell systems in the experiments, and the glycemic control and treatment regime of the patients ²¹ .

The anti‐oxidative capacity of the HDL was consistently seen to be impaired in type 1 diabetes. Furthermore, this impairment was independent of the degree of glycemic control ⁷⁷ . This effect may be related to the reduction in the HDL‐associated PON1 activity. Interestingly, a study of HDL function in people with longstanding type 1 diabetes showed that individuals without vascular complications had higher levels of medium‐sized HDL particles than those with vascular complications. They also had higher levels of HDL‐associated PON1 mass and activity which were in part transported by medium‐sized HDL particles ⁷⁸ . No other differences in the HDL functions were observed between the two groups ⁷⁸ . In addition, a significant reduction in S1P cargo was detected in the HDL₂ and HDL₃ subfractions in people with longstanding type 1 diabetes ⁷⁹ . As S1P plays a critical role in endothelial cell signaling and the stimulation of NO synthesis, this reduction may contribute to impaired endothelial vasorelaxation ⁷⁷ .

The anti‐inflammatory activity of HDL is also impaired in type 1 diabetes ⁷⁴ , partly related to the reduction in S1P/apoM ³⁸ . Chiesa and colleagues showed that adolescents with type 1 diabetes who had early signs of kidney involvement had increased inflammatory risk scores and HDL dysfunction ⁸⁰ . The capacity of HDL to inhibit in vitro superoxide anion production was impaired, and PON1 activity as well as HDL‐mediated NO production was reduced. Interestingly, systemic inflammation related to early kidney dysfunction rather than glycemic control was associated with a dysfunctional HDL phenotype in their study ⁸⁰ .

Further insights have been gained from studying alterations in the HDL proteome and lipidome in type 1 diabetes and the association with glycemic control ⁷⁷ , ⁸¹ . A small number of specific HDL proteins (such as complement factor H‐related protein 2) are altered in type 1 diabetes independent of glycemic control, while others (such as α‐1B glycoprotein) are partially or fully corrected with the optimization of glycemic control ⁸¹ . The changes in the proteome mainly involve the enrichment of protease inhibitors ⁸¹ , which may contribute to the increased cardiovascular risks in type 1 diabetes since protease regulator activity on HDL has been suggested to play an important role in ASCVD ²⁵ . Changes in the lipidome that have been reported include triglyceride enrichment and elevated ratios of free cholesterol/phosphatidylcholine and sphingomyelins/phosphatidylcholine ⁷⁴ , ⁷⁷ . As the sphingomyelin content is a critical determinant of the surface pressure in lipid membranes and lipoproteins, elevated sphingomyelins/phosphatidylcholine ratio indicates enhanced HDL surface rigidity which influences the activity of embedded proteins ⁸² . These changes may be associated with alteration in the HDL function, such as decreased LCAT activity ⁸³ , CEC ⁸⁴ , and capacity of the HDL to acquire LDL‐derived oxidized lipids to protect LDL from free radical induced‐oxidative damage ⁸⁵ . The elevated free cholesterol/phosphatidylcholine ratio may be the result of enhanced transfer of cholesterol‐rich triglyceride‐rich lipoprotein surface fragments during lipolysis, attenuated CETP‐mediated lipid transfer activity, diminished cholesterol esterification by LCAT, or a combination of these ⁷⁴ , ⁷⁷ .

EFFECTS OF DIABETES ON HDL FUNCTION

The diabetic milieu can alter the structure of HDL and induce HDL dysfunction. The protein and lipid components of the HDL particles are susceptible to modifications in diabetes by processes including non‐enzymatic glycation, carbamylation, oxidation, and inflammation ⁶³ , ⁸⁶ . These processes are common to both type 1 and type 2 diabetes, especially in individuals with poor glycemic control ⁷⁷ .

Hyperglycemia leads to an increase in the production of advanced glycation end‐products (AGEs) ⁸⁷ . Non‐enzymatic glycation of apoA‐I can occur due to the spontaneous interaction with reactive α‐oxoaldehydes, which impairs the cardioprotective functions of HDL including CEC ⁸⁸ , ⁸⁹ . Glycoaldehyde and glyoxal non‐enzymatically glycate apoA‐I and markedly impair the ability to promote cholesterol efflux from macrophages by destabilizing ABCA1 ⁹⁰ . Non‐enzymatic glycation of protein components of HDL is also associated with decreased anti‐oxidative capacities related to decreased PON1 activity ⁹¹ , ⁹² , ⁹³ , ⁹⁴ . HDL particles containing non‐enzymatically glycated apolipoproteins are less effective in inhibiting the oxidation of LDL particles ⁹¹ , ⁹² , ⁹⁵ and in counteracting the oxidative stress in diabetes ⁹⁶ . Furthermore, non‐enzymatically glycated HDL particles have reduced S1P levels. S1P, in association with apoM, protects endothelial cells from apoptosis, inflammation, and oxidative stress. Reduced S1P levels in people with diabetes may be associated with impaired endothelial function and contribute to the accelerated development of ASCVD ⁹⁷ , ⁹⁸ . Non‐enzymatic glycation of lysine residues of apoA‐I also leads to impairment in the anti‐inflammatory function of HDL in type 2 diabetes by reducing the ability to inhibit inflammation in THP‐1 macrophages. This may be due to the changes in conformation of the non‐enzymatically glycated apoA‐I, thereby diminishing its binding to the surface of the macrophages ⁹⁹ .

Carbamylation is another non‐enzymatic post‐translational modification process which is increased in diabetes ¹⁰⁰ . Carbamylation involves the non‐enzymatic binding of isocyanate to free amino groups of proteins. The carbamoyl moiety is added to the amino terminus residues of protein‐like lysine, forming ε‐carbamyllysine (homocitrulline). The source of isocyanate can be derived from urea dissociation or MPO‐mediated catabolism of thiocyanate ¹⁰¹ , ¹⁰² . In addition to the situation of kidney failure, carbamylation is also increased in people with type 2 diabetes and normal kidney function, mainly driven by increased MPO activity ¹⁰⁰ . Carbamylation renders HDL dysfunctional, and Holzer and colleagues have demonstrated that the formation of one carbamyllysine residue per HDL‐associated apoA‐I was sufficient to induce cholesterol accumulation and lipid‐droplet formation in macrophages by increasing cholesterol uptake via the HDL receptor scavenger receptor class B type I (SR‐BI). Hence, HDL carbamylation may contribute to foam cell formation in atherosclerotic lesions ¹⁰³ . We have shown that carbamylated HDL levels are increased in type 2 diabetes and can independently predict the progression of diabetic kidney disease ¹⁰⁴ , and all‐cause and cardiovascular mortalities in people with type 2 diabetes ⁸⁶ .

Oxidative stress is increased in diabetes ¹⁰⁵ and oxidative modifications of HDL also contribute to HDL dysfunction. ApoA‐I was a selective target for MPO‐catalyzed oxidative modification in human atheroma, and an increased level of HDL oxidation was found to be associated with diminished ABCA1‐dependent cholesterol efflux ¹⁰⁶ . Oxidation of apoA‐I also impairs LCAT activity by modifying two amino acids (Tyr‐166 and Met‐148) within the LCAT binding site and thus prevent LCAT binding and abolish its activity ¹⁰⁷ , ¹⁰⁸ . Furthermore, PON1, an HDL‐associated protein that has the ability to hydrolyze oxidized lipids, is inactivated following HDL oxidation ¹⁰⁹ .

The chronic inflammation associated with diabetes drives the changes in the HDL proteome, converting HDL from an anti‐atherogenic particle to a raft of immunological proteins ¹¹⁰ . Some changes in the HDL particles include a reduction in the apoA‐I content and replacement by the acute phase protein, SAA, during the acute phase response ¹¹¹ . We have shown that SAA is associated with impairment of SR‐BI‐mediated cholesterol efflux to serum in type 2 diabetes ¹¹² . In addition, the activities of the HDL anti‐oxidant enzymes, such as PON1, PAF‐AH, and LCAT, are diminished during the acute phase response ¹¹¹ .

Furthermore, there are many other potential factors that can affect and alter HDL structure and functionality in diabetes, such as smoking, alcohol, obesity, anti‐diabetic medications, and lipid‐lowering medications, which are not covered in detail in this review ²¹ , ⁵⁸ , ¹¹³ .

ANTI‐DIABETIC FUNCTION OF HDL

Although clinical trials with CETP inhibitors did not result in a reduction in cardiovascular events, glycemic control was improved in subjects with type 2 diabetes treated with torcetrapib in the ILLUMINATE trial ¹¹⁴ . Small randomized, double‐blind trials have revealed that raising plasma HDL‐C and apoA‐I levels either acutely by a single recombinant HDL infusion, or chronically by CETP inhibition, improves glycemic control by enhancing pancreatic β‐cell function and increasing insulin sensitivity ¹¹⁵ , ¹¹⁶ . HDL has several potential anti‐diabetic properties including increasing insulin secretion, promoting β‐cell protection, and reducing insulin resistance.

Drew and colleagues first reported the potential therapeutic values of HDL and apoA‐I in diabetes by infusing reconstituted HDL (rHDL), prepared with apoA‐I and soybean phosphatidylcholine, in 13 people with type 2 diabetes. rHDL reduced plasma glucose by increasing insulin secretion from pancreatic β‐cells and enhanced glucose uptake into skeletal muscle ¹¹⁵ . Animal studies showed that apoA‐I knockout mice had impaired glucose tolerance while mice overexpressing human apoA‐I demonstrated improved glucose tolerance ¹¹⁷ . In vitro studies of cultured skeletal muscle cells showed that incubation with lipid‐free apoA‐I could increase glucose uptake in an insulin‐dependent as well as insulin‐independent manner via increasing glycolysis and mitochondrial respiration in muscles ¹¹⁸ . Hence, these results suggest that HDL‐based therapies may improve glycemic control even in people with diabetes with a significant loss of β‐cell function ⁶³ .

Experimental evidence suggests that HDL may protect and preserve β‐cell function. In vitro studies have shown that apoA‐I and apoA‐II in lipid‐free and lipid‐associated forms increase insulin synthesis and glucose‐stimulated insulin secretion in the MIN6 and Ins‐1E pancreatic insulinoma β‐cell lines ¹¹⁹ , ¹²⁰ . Mechanistically, this involves the activation of the G‐protein–cAMP–PKA–FoxO1 pathway, internalization of lipid‐free apoA‐I into the pancreatic β‐cells, and increased expression of the β‐cell survival gene, pancreatic and duodenal homeobox 1 (Pdx1) ¹¹⁹ , ¹²¹ . As lipid‐free apoA‐I and apoA‐II can increase Pdx1 gene expression, these apolipoproteins may conserve β‐cell function and reduce the adverse effects of activated T cells in type 1 diabetes ¹²² , ¹²³ . HDL can protect β‐cells from apoptosis induced by high levels of glucose and free fatty acid levels by both endoplasmic reticulum stress‐dependent and independent mechanisms ¹²⁴ , ¹²⁵ . HDL and lipid‐free apoA‐I have also been shown to inhibit β‐cell apoptosis by reducing the expression of pro‐inflammatory cytokine interferon‐1β ¹²⁶ .

The potential translation of the anti‐diabetic functions of apoA‐I and apoA‐II has been demonstrated by in vivo studies using mice in which ABCA1 and ABCG1 were conditionally deleted in β‐cells ¹²⁷ . Loss of β‐cell ABCA1 and ABCG1 led to an increase in islet cholesterol levels and severely impaired β‐cell insulin secretory capacity. ApoA‐I treatment of these mice improved glucose‐stimulated insulin secretion. This was mediated by a cholesterol‐independent mechanism as islet cholesterol levels were not affected by apoA‐I treatment ¹²⁸ .

Furthermore, apoA‐I has been shown to reduce insulin resistance in mouse models of type 2 diabetes. For example, lipid‐free apoA‐I treatment improves insulin sensitivity by increasing glucose uptake by skeletal muscle in insulin‐resistant db/db mice ¹²⁹ , mice with diet‐induced obesity ¹³⁰ and rats with pregnancy‐induced insulin resistance ¹³¹ . Consistent with these animal studies, increased glucose uptake into skeletal muscle has been reported in people with type 2 diabetes whose circulating HDL levels were raised with a single infusion of rHDL ¹¹⁵ .

HDL‐BASED THERAPEUTICS

Given the important role of HDL in cardiovascular and metabolic health, there has been interest in developing HDL‐based therapeutics. Efforts have been directed to develop rHDL, which enhances HDL‐mediated RCT specifically rather than increasing HDL‐C levels, following the evidence from preclinical studies that showed intravenous infusions of HDL rapidly reversing atherosclerotic plaque ¹³² . Either purified or recombinant apoA‐I is combined with phospholipids in different rHDL preparations. For example, CSL112 is a formulation of rHDL that uses apoA‐I purified from plasma ¹³³ . CSL112 significantly increases CEC and is being evaluated in the ApoA‐I Event Reducing in Ischemic Syndromes II (AEGIS‐II) trial. This large phase 3 cardiovascular outcome trial will provide insight into whether enhancing CEC with CSL112 can reduce major adverse cardiovascular events in people with recent acute myocardial infarction ¹³⁴ . There are also small synthetic apoA‐I mimetic peptides which share many of the same biological functions of apoA‐I used in drug development, either alone or complexed with lipids in rHDL ¹³⁵ . These compounds are tested in the early‐stage clinical trials mainly for safety, with the potential future development depending on the results from CSL112 trials. Possible strategies to exploit the anti‐diabetic function of HDL particles and their apolipoproteins include the synthesis of peptides that mimic the anti‐diabetic functions of HDL‐associated apolipoproteins ¹³⁶ . It may be possible to repurpose the small apoA‐I mimetic peptides developed to mimic the cardioprotective functions of apoA‐I as a treatment option for diabetes, which has been demonstrated in animal studies ¹³⁷ . This approach may be promising, given recent advances in peptide design and their increased clinical utilization ¹³⁸ .

LCAT plays a key role in HDL metabolism and RCT. There is ongoing research studying the effect of increasing LCAT activity on the prevention of coronary artery diseases. A phase 2a study of people with stable coronary artery disease showed that recombinant human LCAT (MEDI6012) added to statin therapy increased HDL‐C and apoA‐I levels, enhanced non‐ABCA1‐mediated cholesterol efflux, and reduced apoB levels and small LDL particle numbers ¹³⁹ . A phase 2b study in people with acute ST‐segment‐elevation myocardial infarction did not show a significant reduction in infarct size or non‐calcified plaque volume at 12 weeks. MEDI6012 was well tolerated with no excess in serious adverse events ¹⁴⁰ . It remains to be elucidated whether MEDI6012 has any beneficial effect on the non‐calcified plaque volume over a longer period of exposure.

CONCLUSION

HDL is a nanoparticle with anti‐atherogenic, anti‐oxidative, and anti‐inflammatory properties, contributing to cardiovascular and metabolic health. The changes in HDL composition and structure in people with diabetes lead to alterations in HDL functions. Dysfunctional HDL is common in diabetes and contributes to the increased risk of cardiovascular disease. The anti‐atherogenic and anti‐diabetic properties of HDL suggest that enhancing HDL function may be a promising therapeutic strategy for preventing and treating diabetes and ASCVD. Further functional characterization of HDL subpopulations is necessary to facilitate targeted design of novel HDL‐based therapeutics.

DISCLOSURES

KCBT is an Editorial Board member of Journal of Diabetes Investigation and a co‐author of this article. To minimize bias, she was excluded from all editorial decision‐making related to the acceptance of this article for publication.

Approval of the research protocol: N/A.

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.