Calcium-Independent Inhibition of PCSK9 by Affinity-Improved Variants of the LDL Receptor EGF(A) Domain

Abstract

LDL (low-density lipoprotein) receptor (LDLR) binds to its negative regulator proprotein convertase subtilisin/kexin type 9 (PCSK9) through the first EGF (epidermal growth factor-like) domain [EGF(A)]. The isolated EGF(A) domain is a poor antagonist due to its low affinity for PCSK9. To improve binding affinity, we used a phage display approach by randomizing seven PCSK9 contact residues of EGF(A), including the Ca(2+)-coordinating Asp310. The library was panned in Ca(2+)-free solution, and 26 unique clones that bind to PCSK9 were identified. Four selected variants demonstrated improved inhibitory activities in a PCSK9-LDLR competition binding ELISA. The Fc fusion protein of variant EGF66 bound to PCSK9 with a K(d) value of 71 nM versus 935 nM of wild type [EGF(A)-Fc] and showed significantly improved potency in inhibiting LDLR degradation in vitro and in vivo. The five mutations in EGF66 could be modeled in the EGF(A) structure without perturbation of the EGF domain fold, and their contribution to affinity improvement could be rationalized. The most intriguing change was the substitution of the Ca(2+)-coordinating Asp310 by a Lys residue, whose side-chain amine may have functionally replaced Ca(2+). EGF66-Fc and other EGF variants having the Asp310Lys change bound to PCSK9 in a Ca(2+)-independent fashion. The findings indicate that randomization of an important Ca(2+)-chelating residue in conjunction with "selection pressure" applied by Ca(2+)-free phage selection conditions can yield variants with an alternatively stabilized Ca(2+) loop and with increased binding affinities. This approach may provide a new paradigm for the use of diversity libraries to improve affinities of members of the Ca(2+)-binding EGF domain subfamily.