Catalytic Activity Is Not Required for Secreted PCSK9 to Reduce Low Density Lipoprotein Receptors in HepG2 Cells^🅂

EXPERIMENTAL PROCEDURES

RESULTS

Among the PC family of mammalian serine proteases, auto-catalytic cleavage of the prodomain is required for entry into the secretory pathway. Accordingly, mutating the conserved serine (Ser³⁸⁶) of the catalytic triad in PCSK9 prevents autocatalytic cleavage resulting in retention of the protein within the ER (8, 17). To determine whether catalytic activity is required for exogenous PCSK9 to destroy LDLRs, an alternative strategy was employed to obtain secreted catalytically inactive PCSK9. An expression plasmid was constructed that contained the signal peptide and prodomain of PCSK9 followed by a V5 epitope tag. A second expression plasmid was constructed that contained the signal sequence, catalytic, and C-terminal portions of PCSK9 followed by a FLAG tag. The two plasmids were co-transfected into HEK 293 cells to determine whether the PCSK9 fragments expressed in trans could associate and be secreted from the cell. As shown in Fig. 1A, the prodomain and catalytic fragments were similarly expressed in the transfected cells (lanes 3–5); however, secretion of either peptide only occurred when both fragments were co-expressed (lane 10). The ability to purify PCSK9 by co-expressing the prodomain and catalytic fragment of PCSK9 now provided a means to mutate key amino acid residues of the catalytic triad to determine whether catalytic activity was required for PCSK9 activity.

An HEK 293S cell line was subsequently established that stably expressed the “trans-PCSK9” or a trans-PCSK9 protein containing an alanine substituted for the serine at position 386 (S386A) of the catalytic triad. This amino acid change abolishes the autocatalytic cleavage of PCSK9 (3, 17). From these cell lines, large amounts of the recombinant purified protein could be obtained to characterize their binding properties and test for activity. Secreted trans-PCSK9 proteins were purified using the FLAG epitope as previously described (13). Fig. 1B shows that the catalytic fragment and prodomain co-purify with both proteins expressed in trans, indicating the two fragments interact and are secreted together in a manner similar to that of the wild-type PCSK9 purified from HEK 293S cells stably transfected with a single plasmid encoding the full-length PCSK9 protein. The alanine to serine change at position 386 of the purified protein was confirmed by sequencing using mass spectrometry (data not shown). Further evidence that the S386A mutation produces a catalytically inactive protein was gained from the immunoblot data using an antibody that recognizes the V5 tag present at the C terminus of the prodomain. Immunoblot of purified PCSK9 or trans-PCSK9 using an anti-V5 or anti-PCSK9 prodomain antibody revealed that only the trans- PCSK9(S386A) retained the V5 tag (Fig. 1C). This indicates that the wild-type trans-PCSK9 could cleave the V5 tag from the prodomain and the trans-PCSK9(S386A) could not. The sequence of the cleavage site in trans-PCSK9 (VFAQ^]152↓GKP) was verified by mass spectrometry and was found to occur at the same residue at which the prodomain is normally cleaved from full-length PCSK9 (VFAQ¹⁵²↓SIP).

We first determined whether the secreted recombinant trans-PCSK9 proteins could bind to the LDLR by ligand blotting using purified LDLR protein. Fig. 1D shows that the purified trans-XSXSPCSK9 and trans-PCSK9(S386A) proteins bound the LDLR with affinities similar to that measured with purified PCSK9 protein derived from a single plasmid encoding the full-length protein. We next tested whether catalytic activity was required for PCSK9 to degrade LDLRs when added to the medium of HepG2 cells cultured in sterol-depleted medium to induce LDLR expression (13). After incubation with purified PCSK9, the surface proteins of the cells were covalently modified with a cell-impermeable biotinylation reagent and then isolated using streptavidin beads. Total cellular LDLRs in whole-cell extracts and cell surface LDLRs were measured by SDS-PAGE and immunoblotting (13). The amounts of whole-cell and cell surface PCSK9 were also measured by SDS-PAGE and immunoblotting using an anti-FLAG antibody that detected only exogenously added PCSK9.

Fig. 2 shows that PCSK9 purified from the full-length plasmid (PCSK9), trans-PCSK9, and trans-PCSK9(S386A) had nearly identical potencies in reducing the number of LDLRs on the cell surface at all concentrations tested without affecting another cell surface receptor, the transferrin receptor (compare lanes 6–10 in Fig. 2, A–C). FLAG-tagged PCSK9 was detected in whole-cell extracts in a concentration-dependent manner (Fig. 2, A–C, lanes 2–5) but was not detected among the biotin-labeled cell surface proteins (lanes 7–10), suggesting that most of the cell-associated PCSK9 had been internalized. Of note, catalytically inactive PCSK9 functioned to degrade LDLRs at 0.5 μg/ml, a concentration similar to that found in human plasma (Fig. 2C, lane 7) (13).

These results indicate that catalytic activity of PCSK9 is not required to degrade LDLRs when PCSK9 is added exogenously to HepG2 cells. Previously, point mutations in human PCSK9 were identified that resulted in a gain-of-function and hypercholesterolemia (15). One gain-of-function PCSK9 mutation, PCSK9(D374Y), binds to the LDLR with ~25-fold greater affinity and is ~10-fold more active in reducing LDLRs than the wild-type protein (13, 14). If the hypothesis that binding of PCSK9 to LDLR, but not catalytic activity, is required for PCSK9 to degrade the LDLR is correct, then the introduction of the D374Y mutation into the catalytically inactive PCSK9(S386A) should still confer increased binding and enhanced degradation capacity. To test this possibility, an HEK 293S cell line was established that stably expressed PCSK9 in trans and that contained both the D374Y and S386A mutations in the catalytic fragment (trans-PCSK9(D374Y-S386A)).

The binding of the trans-PCSK9(D374Y-S386A) was first evaluated by ligand blotting (Fig. 3A). The affinity of trans-PCSK9(D374Y-S386A) for the LDLR was similar to that of the D374Y protein, both of which had an apparent higher affinity for LDLR than the wild-type PCSK9 protein. When added to the medium of HepG2 cells at one-tenth the concentration of wild-type PCSK9, the trans-PCSK9(D374Y-S386A) and the PCSK9(D374Y) proteins were able to degrade the LDLRs on the cell surface (Fig. 3, C and D, lanes 7–10). These data further support the conclusion that the binding of PCSK9 to the LDLR facilitates the degradation of the LDLR through a mechanism that does not require proteolytic activity of PCSK9.

DISCUSSION

In this report, we demonstrate that introducing a mutation in PCSK9 that abolishes catalytic activity but does not interfere with the secretion of the protein failed to alter the mutant protein’s ability to bind to the LDLR or to mediate the destruction of LDLRs when added to the medium of cultured HepG2 cells. These results support a model in which exogenous PCSK9 binds to the LDLR, which then either targets the LDLR to the lysosome for degradation or prevents the recycling of the receptor in a manner that is independent of inherent catalytic activity of the protein. These data indicate that unlike other PCs, PCSK9 is unique as a subtilisin-like serine protease in that the protein carries out a biological function that is independent of its proteolytic activity.

Studies have demonstrated that PCSK9 and the LDLR interact directly and that the association of PCSK9 with the cell surface and its subsequent internalization is dependent upon the presence of LDLRs (13, 22). In addition, both proteins co-localize to a late endocytic/lysosomal compartment, and internalization is required for PCSK9 to reduce LDLR protein levels because this activity is blocked in the absence of autosomal recessive hypercholesterolemia (ARH), an endocytic adaptor protein required in hepatocytes for LDLR internalization in clathrin-coated pits (13, 22). Inasmuch as PCSK9 is a member of the proteinase K subfamily of subtilisin-related serine endoproteases, it seemed likely that PCSK9 cleaved the LDLR, which then facilitated degradation. A second possibility was that PCSK9 cleaved another as yet unidentified protein that ultimately mediated the destruction of LDLRs. The data of Fig. 2 indicate that catalytic activity is not required for PCSK9 to mediate the destruction of LDLRs when added exogenously to HepG2 cells. The data of Fig. 3 further support the conclusion that binding of PCSK9 to the LDLR is sufficient to target LDLRs for degradation, because inactivating the catalytic site in the hyperactive protein PCSK9(D374Y) did not change the ~10-fold higher specific activity of the protein (13). These data also support the conclusion that the gain-of-function of PCSK9(D374Y) is a result of its increased affinity for the LDLR.

Previous studies have suggested that cleavage of PCSK9 is required for the protein to be secreted from cells (9). We have found that if catalytically inactive mutants are highly overexpressed, a significant amount of unprocessed PCSK9 is secreted into the medium. The secreted full-length but catalytically inactive protein was purified and tested in vitro as described in Figs. 1 and 2; however, the uncleaved PCSK9 protein did not bind to the LDLR nor did it reduce LDLRs in HepG2 cells (data not shown). Thus, cleavage of the prodomain may be required for PCSK9 to adopt a conformation that mediates LDLR binding.

The data of Fig. 1 demonstrates that the PCSK9 prodomain is capable of performing a chaperone function in trans, as evidenced by the ER exit and secretion of trans-PCSK9 and the uncompromised ability of the protein to bind to the LDLR. Previously, it has been shown that mature, active furin can be formed in monkey kidney BSC-40 cells by the co-expression of the furin prodomain and a furin fragment lacking the prodomain on separate plasmids (11). In the case of PCSK9, the ability of the prodomain to function in trans allowed for studies of the effect of catalytic function on LDLR degradation independent of effects on PCSK9 trafficking in the secretory pathway. The crystal structure of PCSK9 was recently solved at 2.0-Å resolution (14). The prodomain failed to contain an apparent target loop that is found in the prodomains of other proprotein convertases and typically is the site of a second cleavage event that results in exposure and activation of the protease. The absence of a target loop in the prodomain of PCSK9 has been interpreted as evidence that catalytic activity may not be involved in the LDLR-lowering function of PCSK9 (14). The current studies provide direct experimental evidence that exogenously added PCSK9 reduces LDLRs in a manner that requires LDLR binding but not catalytic activity.

Studies by Maxwell et al. (25) showed that overexpression of PCSK9 in HepG2 cells induces degradation of LDLRs intracellularly in a post-ER compartment, and Nassoury et al. (26) have suggested that the two proteins interact in the ER and co-localize in the Golgi apparatus in hepatocytes. The results of our studies do not preclude the possibility that PCSK9 can function intracellularly in a manner that requires catalytic activity.

Previously, Cohen et al. (19) reported that one of every 50 African-Americans inherits a nonsense mutation in PCSK9 that lowers LDL cholesterol levels by ~40%. Ina 15-year prospective study, non-sense mutations in PCSK9 reduced LDL cholesterol levels by 28%, which was associated with a reduction in the frequency of coronary heart disease of 88% in African-Americans (27). In addition, individuals that carry these cholesterol-lowering loss-of-function mutations in PCSK9 appear to have a normal life span (27). Thus, genetic data from humans with loss-of-function mutations in PCSK9 validate PCSK9 as a potential target for the treatment of hypercholesterolemia and suggest that inhibitors of the protease would be of therapeutic benefit. The lack of a requirement for PCSK9 catalytic activity in reducing LDLRs has important implications for the development of PCSK9 inhibitors. It has been shown previously that LDLR levels do not change when a catalytically dead enzyme that is not secreted is expressed in liver (17) or in cultured liver cells (28). Thus, the current data now suggest that inhibitors of PCSK9 catalytic activity will have to function intracellularly in the ER to block PCSK9 secretion and, hence, its ability to reduce LDLR protein levels.

Supplementary Material