Characterization of the antibiotic bacitracin as an inhibitor of Lp(a) assembly

Abstract

Cardiovascular disease (CVD) is a leading cause of premature death in New Zealand and in the developed world. Numerous epidemiological studies have identified elevated plasma levels (>500 mg/l) of the cholesterol-rich lipoprotein(a) [Lp(a)] as an independent risk factor for multiple forms of cardiovascular disease (CVD). Lp(a) has both atherogenic and thrombogenic properties. It is comprised of a low density lipoprotein (LDL-) particle, which is covalently bound to the plasminogen homologue apolipoprotein(a) [apo(a)]. Attempts to reduce Lp(a) plasma levels with the use of medication and lifestyle changes have been largely unsuccessful and no therapy exists to date to safely and effectively lower Lp(a).

It is well established that Lp(a) is assembled in a two-step fashion. In the first step, the apolipoprotein B (apoB) lysine residues on low density lipoprotein (LDL) associate non-covalently with lysine binding sites in apolipoprotein(a) [apo(a)]. The second step is the subsequent formation of a disulfide bond between apoBCys4326 and apo(a)Cys4057. Although numerous investigations were aimed at the biogenesis and metabolism of Lp(a), controversial results made it difficult to determine where Lp(a) is assembled and how it is removed from the circulation. The majority of evidence suggests that covalent Lp(a) is formed in circulation from newly synthesized and secreted apolipoprotein(a) and circulating LDL. Since the rate of spontaneous covalent Lp(a) formation from purified apo(a) and LDL is slow, a protein disulfide isomerase (PDI)-like activity present in plasma has been proposed to be involved in this step. Lp(a) levels seem to be chiefly determined by the rates of hepatic apolipoprotein(a) synthesis rather than the breakdown of Lp(a). This has led to efforts to develop Lp(a)-lowering strategies, which focussed on synthesis, secretion and assembly of this unique lipoprotein.

In the present study, we investigate the potential of the old peptide antibiotic bacitracin as an inhibitor of Lp(a) assembly. Bacitracin has been used as an inhibitor of PDI for many years and since a PDI- like enzyme is thought to be involved in the covalent step of Lp(a) assembly, it was investigated in our lab for its ability to inhibit Lp(a) assembly. Commercial bacitracin is a mixture of at least 22 structurally related peptides. The inhibitory activity of individual bacitracin analogues on PDI is unknown. For the present study, we purified the major bacitracin analogues, A, B, H and F by reversed phase HPLC and tested their ability to inhibit the reductive activity of PDI using an insulin aggregation assay. The mechanism of PDI inhibition by bacitracin has previously been unknown. Here, we show by MALDI TOF/TOF MS a direct interaction of bacitracin with PDI, which involves disulfide bond formation between an open thiol form of the bacitracin thiazoline ring and cysteines in the substrate binding domain of PDI. While bacitracin has been commonly used as a specific PDI inhibitor, we further showed that bacitracin binding is non-specific to PDI and applies to other peptides and proteins with free cysteines.

Prior to this study commercial bacitracin has been shown to inhibit the in vitro Lp(a) formation with an IC50 of 1.7 mM in in vitro Lp(a) formation assays using plasma from human apo(a) and human apoB transgenic mice. In this study, we tested the ability of purified individual bacitracin analogues to inhibit Lp(a) assembly to identify the most potent analogue. We further aimed to elucidate the mechanism, by which bacitracin inhibits Lp(a) assembly and whether the inhibition was due to the inhibition of the PDI-like enzyme involved in the covalent step of Lp(a) assembly. Because we showed that bacitracin is able to bind free cysteines, we hypothesized that it was the binding of free cysteines on apo(a) and/or apoB, which caused its inhibitory action on Lp(a) assembly. Indeed, we found bacitracin to bind apoB via a disulfide and we therefore further aimed to identify which of the nine free cysteines on apoB was involved in bacitracin binding.

We also investigated the ability of the most active bacitracin analogue, bacitracin F, to reduce Lp(a) plasma levels in vivo in Lp(a) transgenic mice upon intravenous administration. The rationale behind this study was that bacitracin would bind circulating LDL and prevent secreted apo(a) from binding. Indeed, we found it to circulate in the bloodstream in association with LDL with a half-life of 2 h and Lp(a) levels were slightly reduced at 1 h post injection. Additionally, bacitracin was shown to bind albumin and other plasma proteins in vivo. Furthermore, we established that Bacitracin F was cleared by the kidney and did not exhibit a nephrotoxic effect at the administered dose.