Lipoprotein(a) and LPA Kringle IV-2 Copy-Number Variation

Lipoprotein(a) [Lp(a)] is a hepatocyte-secreted atherogenic, pro-inflammatory, and prothrombotic plasma lipoprotein particle discovered by the Norwegian geneticist Kåre Berg in 1963 that consists of a cholesterol-rich low-density lipoprotein (LDL)-like core particle (containing a single molecule of apolipoprotein B-100 [ApoB-100]) covalently tethered by a single disulfide bridge to a polymorphic hydrophilic glycoprotein, apolipoprotein(a) [apo(a)], encoded by the LPA gene on human chromosome 6q25.3–q26. Unlike standard LDL cholesterol—whose circulating concentration is strongly influenced by diet, physical activity, and hepatic LDLR recycling via PCSK9—plasma Lp(a) concentrations are >85% to 90% genetically determined (heritability h² > 0.90), reach mature adult levels by 1 to 2 years of age, remain stable across the lifespan, and vary more than 1,000-fold between individuals (from <0.1 mg/dL to >300 mg/dL, or <0.2 to >750 nmol/L).

Large-scale Mendelian randomization studies (UK Biobank, Copenhagen City Heart Study, INTERHEART) and genome-wide association studies prove that elevated plasma Lp(a) (>50 mg/dL or >125 nmol/L, present in ~20% of the global population, ~1.4 billion people) is an independent, causal, dose-dependent risk factor for atherosclerotic coronary heart disease (CHD), myocardial infarction, ischemic stroke, peripheral arterial disease, and calcific aortic valve stenosis (CAVS). This extreme inter-individual variance is driven primarily by a hypervariable 5.5-kb tandem copy-number variation (CNV) inside the LPA gene encoding the Kringle IV Type 2 (KIV-2) domain (ranging from 1 to >40 tandem copies per allele), where alleles with fewer KIV-2 repeats synthesize smaller apo(a) isoforms that fold rapidly in the hepatocyte endoplasmic reticulum (ER) and produce high plasma Lp(a) levels.

Primate Duplication of PLG and Loss of Protease Activity

Comparative mammalian phylogenetics demonstrates that the human LPA gene (Chr6q25.3–q26) originated ~35–40 million years ago during primate evolution via tandem duplication and domain remodeling of the adjacent plasminogen (PLG) gene (located only ~40 kb centromeric to LPA). Canonical plasminogen consists of an N-terminal tail domain, five triple-disulfide-looped Kringle domains (KI–KV, ~80–85 amino acids each), and a C-terminal trypsin-like serine protease domain that is cleaved by tissue plasminogen activator (tPA) at Arg561–Val562 to generate fibrinolytic plasmin.

During the evolutionary emergence of Old World monkeys, apes, and humans (with a striking convergent evolution of an independent KIII-based apolipoprotein(a)-like gene in European hedgehogs, Erinaceus europaeus), the duplicated LPA locus underwent four structural transformations:

1. Deletion of KI, KII, and KIII;
2. Diversification and Tandem Expansion of Kringle IV (KIV) into 10 distinct KIV subtypes (KIV-1 through KIV-10): KIV-1 and KIV-3 through KIV-10 are each present in a single invariant copy per allele, whereas KIV-2 (encoded by two 160-bp and 182-bp exons separated by a 4.0-kb intron within a 5.55-kb tandem repeat unit) underwent massive unequal sister-chromatid exchange to vary between 1 and >40 copies per allele (generating apo(a) polypeptide molecular masses ranging from ~250 kDa to >800 kDa);
3. Retention of a single copy of Kringle V (KV); and
4. Inactivation of the C-Terminal Protease Domain: An Arg→Ser substitution at the tPA activation cleavage site (homologous to PLG Arg561), combined with a 9-amino-acid deletion and mutations around the catalytic triad, rendered the apo(a) protease domain catalytically inactive (zymogen-like). Consequently, Lp(a) competes with plasminogen for lysine-binding sites on fibrin clots without lysing them, exerting an antifibrinolytic effect.

Two-Step Assembly of Lp(a) and Inverse ER Secretion Kinetics

Inside the hepatocyte ER and at the sinusoidal hepatocyte surface, mature Lp(a) assembles via a two-step structural mechanism:

• Step 1 (Non-Covalent Lysine-Binding Docking): Weak lysine-binding pockets (LBPs) within KIV-7 and KIV-8 of apo(a) bind non-covalently with nanomolar affinity to specific exposed lysine residues (Lys680, Lys690, Lys4372) on the surface of nascent ApoB-100.
• Step 2 (Covalent Disulfide Bond Formation): This non-covalent docking positions the single unpaired free cysteine residue in apo(a) KIV-9 (Cys4057, or Cys67 of KIV-9) directly adjacent to Cys4326 of ApoB-100, catalyzing the formation of the single inter-chain disulfide bond (catalyzed extracellularly by protein disulfide isomerase / oxidase).

Why do alleles with fewer KIV-2 repeats (e.g., 10–22 total KIV domains) cause 4- to 10-fold higher plasma Lp(a) concentrations than alleles with many KIV-2 repeats (>30 total KIV domains)? Every individual Kringle domain (342 bp / 114 amino acids) contains three intramolecular disulfide bridges (6 cysteines) and heavy N- and O-linked glycosylation. A small 15-kringle apo(a) isoform requires proper folding of 45 disulfide bonds inside the hepatocyte ER and exits into the Golgi rapidly (high secretion efficiency). Conversely, a large 40-kringle apo(a) isoform must fold 120 intramolecular disulfide bonds cleanly in series; it experiences prolonged ER retention time (calnexin/calreticulin/BiP chaperone cycling) and up to 70–85% pre-secretory degradation via ER-associated degradation (ERAD) before reaching the cell surface.

What makes Lp(a) roughly 6-fold more atherogenic on a per-particle (equimolar apoB) basis than LDL? As demonstrated by Sotirios Tsimikas, Joseph Witztum, and colleagues, Lp(a) is the preferential carrier of >85% of circulating pro-inflammatory oxidized phospholipids (OxPL) in human plasma. Specifically, a strong lysine-binding pocket inside Kringle IV Type 10 (KIV-10) binds and forms a covalent schiff-base/Michael adduct at His31 and Lys33 with oxidized sn-2 polyunsaturated fatty acid moieties on phosphatidylcholine (OxPL).

When Lp(a) infiltrates the arterial intima or the fibrosa layer of the aortic valve leaflets, it delivers this concentrated payload of OxPL alongside lipoprotein-associated phospholipase A₂ (Lp-PLA2) and autotaxin (ATX). Autotaxin hydrolyzes oxidized lysophosphatidylcholine (lysoPC) into lysophosphatidic acid (LysoPA), which engages LPAR1 and TLR2/TLR4/NF-κB receptors on valvular interstitial cells (VICs)—driving osteogenic transdifferentiation (BMP2, RUNX2, and alkaline phosphatase upregulation), hydroxyapatite calcium deposition, and progressive calcific aortic valve stenosis (CAVS).

Why should clinical laboratories measure Lp(a) in nanomoles per liter (nmol/L) rather than milligrams per deciliter (mg/dL)?

Because the number of KIV-2 repeats in apo(a) varies from 1 to >40 copies per allele, the molecular mass of the apo(a) protein varies from ~250 kDa to >800 kDa across individuals. Older mass-based immunoassays (mg/dL) using polyclonal antibodies directed against the repetitive KIV-2 domain systematically underestimate the true particle count of small-isoform Lp(a) (the highest-risk patients) and overestimate large-isoform Lp(a). Isoform-insensitive immunoassays calibrated in molar particle concentration (nmol/L, using monoclonal antibodies targeting the unique single-copy KIV-9 domain) provide an accurate 1:1 count of circulating Lp(a) particles (risk threshold >125 nmol/L, roughly ~50 mg/dL).

Why do statins fail to lower Lp(a)—and often increase it by 10% to 20%?

Despite containing an ApoB-100 core, Lp(a) is cleared poorly by the canonical hepatocyte LDL receptor (LDLR) because the bulky, hydrophilic apo(a) glycoprotein chain wrapped around ApoB-100 sterically hinders ApoB-100 binding to the LDLR ligand-binding repeats. Instead, plasma Lp(a) levels are governed primarily by its hepatic production rate rather than LDLR-mediated catabolism (statins upregulate LDLR but also mildly stimulate hepatocyte LPA transcription via HNF1α/sterol signaling, increasing plasma Lp(a) by 10–20%; whereas PCSK9 inhibitors lower Lp(a) modestly by 20–25% only at supra-physiological LDLR densities).

How does the oral drug muvalaplin inhibit Lp(a) assembly without lowering plasminogen?

Before apo(a) can form its permanent disulfide bond (KIV-9 Cys4057–ApoB Cys4326), its KIV-7 and KIV-8 domains must dock simultaneously onto three spatially separated lysine residues on ApoB-100. Muvalaplin is a symmetric, trivalent small molecule engineered with three lysine-mimetic warheads on a central core that binds simultaneously to consecutive KIV-7 and KIV-8 domains of nascent apo(a), sterically blocking ApoB-100 docking (IC50 < 1 nM) while sparing monomeric plasminogen.

1. McLean, J. W., Tomlinson, J. E., Kuang, W.-J., et al. (1987). "cDNA sequence of human apolipoprotein(a) is homologous to plasminogen." Nature, 330(6144), 132–137. https://doi.org/10.1038/330132a0
2. Clarke, R., Peden, J. F., Hopewell, J. C., et al. (2009). "Genetic variants associated with Lp(a) lipoprotein level and coronary disease." New England Journal of Medicine, 361(26), 2518–2528. https://doi.org/10.1056/NEJMoa0902604
3. Tsimikas, S. (2017). "A test in context: lipoprotein(a): diagnosis, prognosis, controversies, and emerging therapies." Journal of the American College of Cardiology, 69(6), 692–711. https://doi.org/10.1016/j.jacc.2016.11.042
4. Nicholls, S. J., Nissen, S. E., Fleming, C., et al. (2023). "Muvalaplin, an oral small molecule inhibitor of lipoprotein(a) formation: a randomized clinical trial." JAMA, 330(11), 1042–1053. https://doi.org/10.1001/jama.2023.16503