Abstract Proprotein convertase subtilisin/kexin type 9 (PCSK9) regulates cholesterol metabolism by inducing the degradation of hepatic low-density lipoprotein receptor (LDLR). Plasma PCSK9 has two main molecular forms: a 62-kDa mature form (PCSK9_62) and a 55-kDa, furin-cleaved form (PCSK9_55). PCSK9_55 is considered less active than PCSK9_62 in degrading LDLR. We aimed to identify the site of PCSK9_55 formation (intra- vs. extracellular) and to further characterize the function of PCSK9_55 relative to PCSK9_62. Co-expressing PCSK9_62 with furin in cell culture induced formation of PCSK9_55, most of which was found extracellularly (97-times more). Under the same conditions, i) adding a cell-permeable furin inhibitor preferentially affected the formation of PCSK9_55 out of the cell, and ii) substituting PCSK9_62 for a mutant that cannot get secreted completely abolished furin action on PCSK9, even though the majority of furin locates intracellularly. Further, the expression of a recombinant PCSK9_55 in cells demonstrated that the intracellular pool of PCSK9_55 fails to be secreted and is retained intracellularly. Nevertheless, this intracellular pool of PCSK9_55 induced degradation of LDLR, though with 50% lower efficiency when compared with PCSK9_62. Collectively, our data show that PCSK9_55 is 1) generated in the extracellular space, and that the small intracellular pool of PCSK9_55 is 2) not secreted but is 3) capable of inducing LDLR degradation through an intracellular pathway.

Two biological replicates including the products of a negative control reaction (no furin; PCSK9 62-kDa) and the experimental reaction PCSK9_62 with furin (PCSK9 55-kDa) were evaluated separately in two independent LC-MS analyses to detect the expected furin cleavage product containing residues 153-218 (average mass = 7730.5). With no further digestion, aliquots of 2.2 µg of PCSK9 from a control sample (no furin) and a furin incubated sample were dried using a DNA120 SpeedVac (Thermo Scientific) and stored at -20 ̊C until analysis. Samples were solubilized in 22 µL of 8 M urea 24 h prior to analysis. A total of 1 µg of protein was analyzed using a LTQ Velos Pro linear ion trap (Thermo Scientific, San Jose, CA). Samples were injected onto a micro protein trap cartridge (Optimize Technologies, Oregon City, OR) at a flow rate of 20 µl/min in mobile phase containing 0.1% formic acid. After 5 min, the flow was diverted to a 1 x 250 mm C4 column (Vydac, SN 214MS51, Grace, Deerfield, IL). Protein was eluted by increasing the acetonitrile concentration from 2-7.5% over one minute, then 7.5-60% over 30 minutes and data collection on the mass spectrometer started 10 min into the separation. The instrument used a HESI-II probe fitted with a 34 gauge metal needle, 5.0 kV source voltage, 325 °C ion transfer tube temperature, sheath gas setting of 5, full MS scans in profile mode over a range of m/z = 400-2000, and averaging of 10 µscans. Spectra acquired during elution of the PCSK9 153-218 peaks was then averaged and deconvoluted using the Manual Respect module for isotopically unresolved data in Protein Deconvolution 4.0 software (Thermo Scientific). Single ion chromatographic traces were produced using Qual Browser software within the Xcalibur Suite (Thermo Scientific).

PCSK9 from different sources (cells overexpressing PCSK9 (3.4 µL of media or 3.4 µL of cell extract) or a purified recombinant form (1900 nM, Cat# 20631, Cayman Chemical) was set to react with furin (680 nM; Cat# 450-47, PeproTech) under optimized buffer, temperature and time conditions (4 mM CaCl2, 150 mM NaCl, 20 mM KCl and 50 mM Tris-HCl pH 7.4 in 10 µL of final volume at 37 ̊ C for 24 hour) to achieve the highest yield of PCSK9 proteolysis. Reactions were stopped by chelating calcium with 4 mM EDTA.