High-density lipoprotein (HDL) and infections: a versatile culprit

This editorial refers to ‘U-shaped relationship of HDL and risk of infectious disease: two prospective population-based cohort studies’†, by C.M. Madsen et al., on page 1181. For several years, HDL has been referred to as ‘good cholesterol’, exerting a plethora of beneficial effects mainly on cells of the vascular tree. HDL prevents endothelial dysfunction, reduces pro-inflammatory cell activation, and promotes reverse cholesterol transport.1 However, many studies questioned this principle of HDL’s protective functions and could intriguingly document that a variety of diseases render HDL dysfunctional.1 In patients with coronary artery disease, diabetes, or autoimmunological diseases, HDL loses its vasoprotective properties and, in chronic kidney disease, even turns into noxious lipoprotein inducing endothelial dysfunction and increasing arterial blood pressure.2 , 3 Therefore, it is not surprising that neither increased HDL-cholesterol (HDL-C) levels nor single nucleotide polymorphisms (SNPs) associated with elevated HDL-C levels were linked to improved outcomes in patients with prevalent cardiovascular diseases, which is in marked contrast to studies of the general population.4 , 5 In line with these epidemiological findings, studies of HDL-raising therapies by inhibition of the cholesterol ester transfer protein (CETP) yielded neutral (for torcetrapib, dalcetrapib, and evacetrapib) or weakly positive (for anacetrapib) effects.6 Based on the presence of dysfunctional HDL, particularly in patients with coronary artery disease or myocardial infarction, it makes sense that increasing such ‘dysfunctional HDL’ might not be related to improved outcomes in those patients. Therefore, HDL remains an enigmatic lipoprotein, which cannot be adequately therapeutically targeted. Besides increasing HDL-C, preventing changes of the composition of HDL (such as changes of the proteome or post-translational protein modifications), which are directly linked to the functionality of HDL, has at least the potential to put HDL back in the focus as a therapeutically targetable lipoprotein. However, these drawbacks and recent advances in LDL-lowering therapies have pushed HDL somewhat out of the spotlight of cardiovascular medicine. In the current issue of the journal, Madsen and colleagues have opened up a new field of interest in HDL.7 In two well-characterized and large-scale cohorts of the general population, the authors assessed the association between HDL-C levels and infection-related hospitalization. According to experimental studies demonstrating anti-inflammatory effects of HDL, low HDL-C was associated with increased infection risk. Surprisingly, very high HDL-C levels (>93 mg/dL) were also related to an increased risk for infectious disease events, resulting in a U-shaped association between HDL-C and (bacterial) infections. To strengthen their findings, the authors were able to demonstrate that SNPs in genes encoding hepatic lipoprotein lipase, LIPC, and cholesterol ester transfer protein, CETP, both HDL-C-increasing alleles, were consistently associated with a reduced risk for infections. Because of the low number of viral infection events, the study did primarily identify an association between HDL-C and bacterial infections. It is not unreasonable to explain the association between low HDL-C and increased number of infectious events. HDL may directly reduce the pathogenicity of Gram-positive and Gram-negative bacteria.8 For 20 years, it has been known that HDL serves as a ‘sponge’, binding bacterial lipopolysaccharide (LPS) from Gram-negative and lipoteichoic acid (LTA) from Gram-positive bacteria. Thereby, HDL has the potential to promote their clearance from the circulation and to prevent an excess of endotoxin. In addition, HDL may also exert an immunomodulatory function by limiting pathogen-mediated activation of the innate immune system. HDL increases the expression of the transcription factor ATF3 in macrophages, which attenuates the production of pro-inflammatory cytokines after challenge of Toll-like receptors (TLRs) with their ligands such as LPS.9 Moreover, in endothelial cells, HDL stimulates endothelial nitric oxide (NO) production, which prevents the expression of cell adhesion molecules and, thereby, the interaction between monocytes and endothelial cells, a crucial step in the pathogenesis of an inflammatory response.8 Furthermore, sphingosine-1 phosphate (S1P), a component of HDL, modulates adaptive immunity by, for instance, increasing the population of regulatory T lymphocytes.8 Distinct anti-inflammatory and pro-inflammatory effects of HDL modulating risk of infectious disease. CAMs, cell adhesion molecules. NO, nitric oxide; eNOS, endothelial NO synthase; ApoC3, apolipoprotein C-III; SAA, serum amyloid A; Casp11, caspase-11; ATF3, cAMP-dependent transcription factor; MAPK, MAP kinases; NFκB, nuclear factor kappa-light-chain-enhancer of activated B cells; AP-1, activator protein 1. In contrast, it is certainly ambitious to understand the association between high HDL-C and increased infectious risk. As discussed by the authors, high HDL-C levels may be caused by SNPs in HDL-increasing alleles. Recently, it has been documented that the rare variant P376L (minor allele frequency ∼0.0003) in the SCARB1 locus is associated with substantially increased HDL-C levels.10 Notably, compared with non-carriers of this variant, the composition and functionality of those HDL particles were dramatically altered. Pro-inflammatory apolipoprotein C-III accumulated in those HDLs, and scavenger receptor (SR)-BI-mediated cholesterol uptake was profoundly reduced. This indicates that despite high HDL-C levels, the functionality of such HDLs may be impaired. However, this might only apply to a small proportion of participants in the Copenhagen General Population Study. Approximately 8% of the participants had HDL-C levels ≥93 mg/dL, of which 82% were women. Therefore, it is unlikely that a higher prevalence of SNPs in loci encoding HDL-increasing genes can completely explain the association between HDL-C and infectious events. HDL binds large amounts of endotoxin.8 Therefore, it is plausible that HDL serves as a cellular shuttle for such bacterial components. Recently, it has been shown that intracellular LPS activates caspase-11 independently of its cell surface receptor TLR4, leading to non-canonical inflammasome activation and subsequent production of the pro-inflammatory cytokine IL-1β.11 Thus, receptor-mediated uptake of endotoxin-loaded HDL might elicit a pro-inflammatory programme in immune cells. Additionally, it has been reported that inflammation leads to structural alterations of HDL particles. In particular, inflammation causes accumulation of the acute phase protein serum amyloid A within the protein moiety of HDL.12 As a consequence, the functionality of serum amyloid A (SAA)-loaded HDL is substantially impaired. SAA-HDL thereby loses its vasoprotective properties, e.g. it inhibits endothelial NO production, increases endothelial expression of cell adhesion molecules, and promotes monocyte–endothelial adhesion. Moreover, SAA-enriched HDL loses its anti-inflammatory properties and even promotes pro-inflammatory activation of macrophages and translocation of TLR4 into lipid rafts.13 Importantly, it has been shown that serum SAA concentrations serve as a perfect surrogate to predict the fraction of ‘biologically active’ HDL.12 Besides alterations of the proteome, post-translational protein modifications of HDL-associated proteins might occur in an inflammatory microenvironment. In particular, leucocyte-derived myeloperoxidase has been shown to induce carbamylation of lysine residues of HDL- as well as LDL-associated proteins, which directly abolishes their vascular functions.14 , 15 These alterations of the functionality of HDL are particularly present in patients with impaired kidney function, in whom HDL not only loses its vasoprotective properties but even turns into a noxious lipoprotein.3 , 16 Importantly, estimated glomerular filtration rate (eGFR) was significantly lower in participants of the Copenhagen General Population Study with HDL levels exceeding 93 mg/dL, which indicates the presence of a substantial proportion of subjects with reduced kidney function within these strata of HDL. This may at least partially explain the association between high HDL-C and infection risk. Although these considerations are merely speculative and not definitely proven by direct experimental evidence, it is reasonable that higher HDL-C represents more HDL, which can be the subject of modifications leading to impaired or inversed functionality of this lipoprotein. Overall, the present study provides novel evidence on the link between HDL and infection-related outcomes. Although the authors cannot definitely prove a causal relationship between low and high HDL-C and infections, it is highly likely that HDL may indeed modulate the outcome of and the susceptibility for infectious diseases. Nevertheless, despite the large study size and the sensitivity analyses provided, the presence of additional confounders or reverse causation cannot be fully excluded. The association between HDL-C and increased infectious risk was only seen in the Copenhagen General Population Study, but not in the replication cohort, the Copenhagen City Heart Study. Moreover, the analyses of SNPs in HDL-increasing alleles also did not confirm the association between high HDL-C and higher infection risk. Therefore, further studies are necessary to prove this association definitely. While these findings are of great interest from a scientific and clinical point of view, the implications of this study are limited since HDL cannot be adequately therapeutically targeted. The most interesting and unexpected finding is the association between high HDL-C levels and increased infection risk. This highlights persons with very high HDL-C as an important population (in this study ∼10%), which requires further experimental and epidemiological studies. In particular, a broader understanding of the phenotype of HDL in those subjects is needed. Madsen et al. answered many important questions on HDL and in parallel they gave rise to many new directions in the field of lipoprotein biology with HDL as a versatile culprit. Conflict of interest: none declared.