Processing of wild-type and mutant proinsulin-like growth factor-IA by subtilisin-related proprotein convertases.

Insulin-like growth factor I (IGF-I) is required for normal embryonic development and postnatal growth. Like most hormones and growth factors, IGF-I is synthesized as a proprotein that is converted to the mature form by endoproteolysis. Processing of pro-IGF-I to mature IGF-I occurs by cleavage within the unique pentabasic processing motif Lys-X-X-Lys-X-X-Arg71-X-X-Arg-X-X-Arg77. We have previously shown that human embryonic kidney 293 cells process pro-IGF-IA at Arg71 to generate IGF-I-(1-70) and at Arg77 to produce IGF-I-(1-76). Cleavage at each of these sites requires upstream basic residues, indicating that subtilisin-related proprotein convertases (SPCs) may be involved. In order to investigate the identity of the endogenous enzymes involved in maturation of pro-IGF-IA, we have expressed wild-type and mutant pro-IGF-IA in 293 cells and in the furin-deficient Chinese hamster ovary cell line, RPE.40. We have also co-expressed these constructs with SPCs that are thought to play a role in processing precursor proteins in the constitutive pathway: furin, PACE4, PC6A, PC6B, and LPC. The results show that furin is most active at cleaving wild-type and mutant pro-IGF-IA and can cleave these precursors at multiple sites within the pentabasic motif. PC6A and LPC are less active than furin but cleave only at Arg71. PACE4 and PC6B have very little activity on pro-IGF-IA precursors. Wild-type pro-IGF-IA was correctly processed to mature IGF-I in 10 of 10 cell lines that were tested. Since furin, PC6A, and LPC are known to have a broad pattern of tissue distribution and we have demonstrated expression of LPC in RPE.40 cells, our results suggest that these SPCs may be responsible for the endogenous pro-IGF-IA processing activity observed in a wide variety of cell lines.

Insulin-like growth factor I (IGF-I) 1 is expressed ubiquitously throughout embryonic development and postnatal growth of mammalian organisms (1). IGF-I is an important regulator of growth; mice that are homozygous for a disrupted IGF-I gene display a number of developmental abnormalities and most die shortly after birth (2,3). Overproduction of IGF-I has been associated with many cancers, in which IGF-I may act as an autocrine or paracrine growth factor (4) or an inhibitor of apoptosis (5). Regulation of the biological activity of IGF-I is therefore an important physiological process.
Most hormones and growth factors are initially synthesized as inactive propeptide precursors that are post-translationally processed to the biologically active mature peptide. The subtilisin-like proprotein convertases (SPCs) are a family of enzymes that have been shown to play an important role in the maturation of prohormones and growth factors as well as a number of other protein precursors. PC2 and PC3 (also known as PC1) are restricted in expression to endocrine and neuroendocrine cells and process substrates such as proinsulin and proopiomelanocortin. PC4 is expressed in the testis, and substrates for this processing enzyme have not yet been identified. In contrast, furin, PACE4, PC6A, PC6B, and LPC (also known as PC7, SPC7, and PC8) are expressed in many tissues and may have an important function in proteolytic processing of precursors in the constitutive secretory pathway (6 -12).
The mature, 70-amino acid IGF-I molecule can be generated from two different prohormone precursors, pro-IGF-IA or pro-IGF-IB. Pro-IGF-IA and pro-IGF-IB are derived from a single gene by alternative splicing of 3Ј-exons. The prohormone sequences are identical through the B, C, A, and D domains of mature IGF-I as well as the first 16 amino acids of the E domain or pro-region. Pro-IGF-IA and pro-IGF-IB, therefore, contain the same prohormone processing site at the junction of the D and E domains but diverge in sequence at the carboxyl terminus of the pro-region (1).
In our previous study on pro-IGF-I processing, we found that expression of pro-IGF-IA in human embryonic kidney 293 cells resulted in secretion of N-glycosylated pro-IGF-IA, pro-IGF-IA, IGF-I- , and IGF-I-(1-70). Using site-directed mutagenesis, we demonstrated that IGF-I-(1-76) is produced by cleavage at Arg 77 and that IGF-I-(1-70) is generated by cleavage at Arg 71 . Cleavage at both of these positions required a P4 basic residue (see Fig. 1). Furthermore, cleavage at Arg 71 occurred in the furin-deficient LoVo and RPE.40 cell lines, indicating that another processing enzyme must be involved in maturation of pro-IGF-IA (13). In an attempt to identify the enzyme(s) involved in maturation of pro-IGF-IA, we have expressed wildtype and mutant pro-IGF-IA in 293 and RPE.40 cells. We have also co-expressed these constructs with candidate pro-IGF-Iconverting enzymes, furin, PACE4, PC6A, PC6B, and LPC.

EXPERIMENTAL PROCEDURES
Expression Constructs-The expression vector pCMVigf1-FLAG has been described previously (13). The human furin cDNA was excised from pBluescript with EcoRI and SalI and cloned into EcoRI/SalIdigested pCMV6c to generate pCMV-furin. The human PACE4 cDNA was removed from pBluescript by digestion with XhoI and SalI. The purified PACE4 fragment was end-filled and blunt-end ligated to pCMV6c that had been digested with SmaI to generate pCMV-PACE4. Human pBSK-LPC was digested with SacI and 3Ј-overhangs were removed with T4 DNA polymerase under conditions favoring exonuclease activity. The purified fragment was blunt-end ligated to pCMV6c to create pCMV-LPC. pCMV-furin, pCMV-PACE4 and pCMV-LPC were analyzed by restriction site digestion and DNA sequencing. cDNAs coding for mouse PC6A (14) and PC6B (15) were subcloned into pRc-CMV (Invitrogen).
Cell Culture and Transfections-Human embryonic kidney 293 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc) with 10% fetal calf serum and 100 units/ml penicillin and 100 g/ml streptomycin. 293 cells were transfected by the calcium phosphate method (17). RPE.40 cells were grown in Ham's F12 medium with 5% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. RPE.40 cells were transfected using LipofectAMINE following the manufacturer's protocol (Life Technologies, Inc). For co-transfection experiments, 5 g of prohormone cDNA was co-precipitated with 5 g of SPC cDNA.
Immunoprecipitates were solubilized in SDS-sample buffer containing 2-mercaptoethanol and denatured by heating at 95°C for 5 min. Samples were electrophoresed on 16% Tricine-buffered polyacrylamide gels with a 4% stack at 4°C (18). Gels were fixed in 10% acetic acid, 25% isopropyl alcohol for 30 min and then soaked in a fluorographic solution (Amplify, Amersham) for 15 min. Dried gels were exposed to x-ray film (Kodak) with an intensifying screen for 4 h to 2 days unless stated otherwise. Quantitation of individual bands was done using a Phosphor-Imager (Molecular Dynamics).
Cloning and Expression of Hamster LPC-Total RNA was prepared from RPE.40 cells and transcribed to cDNA with Superscript (Life Technologies, Inc.). Hamster LPC was amplified using degenerate PCR primers LPC-3 and LPC-4 that were designed to anneal to unique sequences coding for the P domain. The LPC-3 primer (5Ј-TT(T/C)TG(T/ C)CCNAG(T/C)GGNATGATG) corresponds to the amino acid sequence 547 FCPSGMM 553 , and LPC-4 (5Ј-AT(G/C)GT(A/G)TANCC(A/G)TC(T/ C)TC(T/C)TC) encodes 654 EEDGYTI 660 . The 342-bp PCR product was end-filled, blunt end-ligated to pBluescript (Stratagene), and sequenced.
For Northern blot analysis, 4 g of poly(A) ϩ RNA from RPE.40 cells was electrophoresed through 1% agarose, blotted to a HyBond membrane (Amersham), and probed with a 32 P-labeled 342-base pair PCR product encoding the P domain of hamster LPC. Standard molecular biology protocols were followed (17).

RESULTS
Processing of Wild-type and Mutant Pro-IGF-IA by SPCs in 293 Cells-Our previous study on pro-IGF-IA processing uti-lized the pCMVigf1-FLAG expression vector. This vector directs expression of human prepro-IGF-IA containing the eightamino acid FLAG epitope between Ala Ϫ1 of the signal peptide and Gly ϩ1 of the B domain. We have shown that the FLAG peptide does not affect co-translational or post-translational processing of prepro-IGF-IA. Human embryonic kidney 293 cells transfected with pCMVigf1-FLAG secrete N-glycosylated pro-IGF-IA, pro-IGF-IA, IGF-I-(1-76), and IGF-I-(1-70) (13). In the present work, we sought to determine 1) if any of the known SPCs could process pro-IGF-IA and 2) if Arg for Lys substitutions at the P4 and/or the P7 positions would affect processing by endogenous or exogenous enzymes. The primary sequences of the cleavage sites of wild-type and mutant pro-IGF-IA-FLAG constructs used in this study are indicated in Fig. 1. The results of expression of these constructs in 293 cells and their processing by endogenous enzymes or SPCs are shown in Fig. 2 and Table I.
When wild-type pro-IGF-IA and pCMV6c (as a negative control) are co-transfected in 293 cells, two major bands corresponding to IGF-I-(1-70) and IGF-I-(1-76) are detected in conditioned medium ( Fig. 2A). The diffuse band that migrates at approximately 20 kDa corresponds to N-glycosylated pro-IGF-IA (see marker on Fig. 2E). Cells are continuously labeled for approximately 24 h after transfection, and under these conditions approximately 45% of the IGF recovered from conditioned medium is IGF-I-(1-70) and 45% is IGF-I-(1-76). When wild-type pro-IGF-IA is co-transfected with pCMV-furin, 89% of the precursor is processed to IGF-I-(1-70). An additional band is visible that migrates slightly faster than IGF-I-(1-70) and has a molecular mass of approximately 7 kDa. The 7-kDa band is not present in the absence of pCMV-furin. Furin has been reported to cleave at lysine residues (19), and this band may result from processing at Lys 65 -X-X-Lys 68 to generate IGF-I-(1-67). The pattern observed for processing of the K65R mutant by enzymes endogenous to 293 cells is qualitatively and quantitatively identical to processing of wild-type pro-IGF-IA. In the presence of furin the main processing product is again IGF-I-(1-70). A 7-kDa band is also present and is more intense than the the minor cleavage product produced by co-expression of wild-type pro-IGF-IA with furin. This band may result from cleavage at Arg 65 or Lys 68 . The K68R mutant shifts the Lys 68 -X-X-Arg 71 processing motif of pro-IGF-IA upstream by three amino acids to Lys 65 -X-X-Arg 68 and creates a new minimal furin site at Arg 68 -X-X-Arg 71 (see Fig. 1). When the K68R mutant is co-transfected with pCMV6c in 293 cells, IGF-I-(1-70) is the major product, and IGF-I-(1-76) is not detected ( Fig.  2A). A faint band is visible below IGF-I-(1-70) in Fig. 2A. However, this product is not produced consistently from K68R by the endogenous processing enzymes (see Fig. 2, B, C, D, and FIG. 1. Wild-type and mutant IGF-I prohormone-processing sites. The amino acid sequence of the wild-type pro-IGF-I-processing site is shown in single-letter code. Basic residues are numbered, and substitutions generated by site-directed mutagenesis are indicated by boldface type. X, stop codon. The Lys 68 -X-X-Arg 71 and Arg 74 -X-X-Arg 77 processing motifs utilized in wild-type pro-IGF-I are in brackets, and the putative scissile bonds are indicated by vertical arrows. The P1, P4, and P7 positions relative to the Lys 68 -X-X-Arg 71 processing motif are indicated at the top. E). In the presence of furin, 54% of the K68R precursor is processed to IGF-I-(1-70), and 34% is processed to a 7-kDa product that likely represents cleavage at the Lys 65 -X-X-Arg 68 site. The K65R/K68R mutant contains arginine residues at positions 65, 68, 71, 74, and 77. Processing enzymes endogenous to 293 cells generate two products from this precursor ( Fig. 2A). The upper band represents 65% of the precursor and corresponds to IGF-I-(1-70). The 7-kDa band represents 26% of the precursor and may result from cleavage at Arg 65 or Arg 68 . Furin processes 89% of the K65R/K68R precursor to the 7-kDa form, and IGF-I-(1-70) is not detected.
When PACE4 is co-expressed with wild-type pro-IGF-IA or the K65R mutant, IGF-I-(1-70) and IGF-I-(1-76) are the only cleavage products, lower molecular weight forms are not produced (Fig. 2B). PhosphorImager analysis indicates a slight increase in the production of IGF-I-(1-70) from the wild type and the K65R precursor in the presence of pCMV-PACE4. The K68R mutant is processed only to IGF-I-(1-70), and the amount of product produced is not affected by PACE4. As seen previously, the K65R/K68R mutant is processed primarily to IGF-I-(1-70) by endogenous enzymes, with a smaller peptide composing 27% of the final product. Processing of this mutant is unaffected by PACE4.
Co-expression of PC6A with either the wild-type pro-IGF-IA or the K65R mutant enhances production of IGF-I-(1-70), and lower molecular weight forms are not generated (Fig. 2C). PC6A processes 59% of the wild-type precursor and 71% of the K65R mutant to IGF-I-(1-70). PC6A does not qualitatively or quantitatively alter the processing of the K68R or K65R/K68R mutants.  Fifty-four percent of the wild-type precursor and 59% of the K65R mutant are processed to IGF-I-  in the presence of PC6B, compared with 45 and 48% IGF-I-(1-70) product by endogenous enzymes (Fig. 2D). PC6B does not qualitatively or quantitatively alter the processing of the K68R or K65R/K68R mutants.
In the presence of LPC, 73% of wild-type pro-IGF-IA was converted to IGF-I-(1-70), as compared with 45% conversion by endogenous enzymes (Fig. 2E). LPC was less active on the K65R, K68R, and K65R/K68R mutants. LPC did not generate peptides smaller than IGF-I-(1-70) from any of the precursors.
Processing of Wild-type and Mutant Pro-IGF-IA by SPCs in RPE.40 Cells-The co-expression experiments in 293 cells indicated that furin will efficiently process wild-type and mutant pro-IGF-IA to IGF-I-(1-70) and will also cleave the mutant precursors at other sites. Processing enzymes endogenous to 293 cells process all precursors to IGF-I-(1-70) and cleave only the K65R/K68R mutant at an alternate site. By co-expressing pro-IGF-IA constructs with SPCs in the furin-deficient RPE.40 cell line (20), we sought to determine 1) if furin was responsible for alternative cleavage of K65R/K68R and 2) the activity of PACE4, PC6A, PC6B and LPC for pro-IGF-IA in the absence of furin. The results of these experiments are shown in Fig. 3 and Table II. As we have shown previously (13), when pro-IGF-IA is expressed in RPE.40 cells, IGF-I-(1-70), but not IGF-I- (1-76), is generated. In RPE.40 cells, approximately 80% of the precursor is processed to IGF-I-(1-70), while 20% is secreted as N-glycosylated pro-IGF-IA (Fig. 3A). When co-expressed with furin, nearly 100% of the pro-IGF-IA is converted to IGF-I- (1-70). Approximately 80% of the K65R and K68R precursors ap-

Processing of Pro-IGF-IA by SPCs
peared as IGF-I-(1-70). In the presence of furin IGF-I-(1-70) is still the major product, but a smaller peptide is also generated from both mutants. The processing enzymes endogenous to RPE.40 cells generate only IGF-I-(1-70) from the K65R/K68R mutant. However, when co-expressed with furin, nearly 100% of the precursor is processed at an alternative site to generate the smaller 7-kDa form. When wild-type or mutant pro-IGF-IA is co-expressed with PACE4 (Fig. 3B), PC6A (Fig. 3C), or PC6B (Fig. 3D) in RPE.40 cells, IGF-I-(1-70) is the only cleavage product. Approximately 80% of each precursor is converted to mature IGF-I, and this figure is not affected by the presence of exogenous enzymes.
In the background of RPE.40 cells, LPC is nearly as efficient as furin at processing pro-IGF-IA. Eighty-five to 93% of each precursor is processed to IGF-I(1-70) when co-expressed with LPC, and smaller peptides are not produced (Fig. 3E).
Cleavage of Pro-IGF-IA at Arg 71 or Arg 77 by SPCs-In order to determine if SPCs discriminate between the Lys 68 -X-X-Arg 71 and Arg 74 -X-X-Arg 77 sites of pro-IGF-IA we co-expressed these enzymes with the R74A/R77A and R71A mutants. Each of these mutants contains only one of the two potential processing sites (Fig. 1). When the R74A/R77A mutant is expressed in 293 cells, 48% of the precursor is processed at Arg 71 (Fig. 4A). In the presence of furin, nearly 100% of the precursor is cleaved at Arg 71 . PACE4, PC6A, PC6B, and LPC cleave at Arg 71 to generate 59, 66, 60, and 58% IGF-I-(1-70), respectively. Seventyseven percent of the R71A precursor is cleaved by endogenous enzymes in 293 cells (Fig. 4B). Processing at Arg 77 increased to 96, 81, 91, 88, and 70% of the precursor in the presence of furin, PACE4, PC6A, PC6B, and LPC, respectively.
When expressed in RPE.40 cells, 47% of the R74A/R77A precursor is processed at Arg 71 . Furin processed this precursor very efficiently, but PACE4, PC6A, and PC6B had no apparent affect. In contrast, LPC had a moderate affect, processing 57% of the precursor (Fig. 4C). Sixty-eight percent of the R71A mutant is cleaved at Arg 77 in RPE.40 cells. In the presence of furin, cleavage at this site increases to 93% of precursor. PACE4, PC6A, PC6B, and LPC do not affect processing of this mutant in the RPE.40 cells (Fig. 4D). We often observe some additional processing or degradation of the R71A precursor in RPE.40 cells, resulting in faint bands below the IGF-I-(1-76) product.
Processing of the K68R/A70K Mutant by SPCs in RPE.40 Cells-In the previous experiments we have not observed processing of IGF-IA precursors by PACE4, PC6A, or PC6B in RPE.40 cells. In order to determine if these enzymes are active in RPE.40 cells, we co-expressed them with the K68R/A70K mutant. This mutant contains an Arg-X-Lys-Arg 71 cleavage site (Fig. 1), which is also found in other precursors that are processed by PACE4, PC6A, and PC6B (21)(22)(23)(24). The K68R/ A70K mutant was processed very efficiently by the endogenous RPE.40 processing enzymes, and it was therefore necessary to overexpose the gel to detect pro-IGF-IA. Overexposure of the gel revealed that processing is enhanced by furin, PACE4, PC6A, and PC6B (Fig. 6). This effect was observed in three independent experiments.
Cloning and Expression of Hamster LPC-The data on coexpression of pro-IGF-IA with SPCs in RPE.40 cells indicates that furin and LPC are most active in this cell line. The fact that both alleles of furin are known to be inactive in RPE.40 cells due to mutation (20) suggests that LPC may be responsible for the observed pro-IGF-IA processing. In order to determine if RPE.40 cells express LPC, we obtained a partial clone by PCR of RPE.40 cDNA. The 342-base pair cDNA encodes a 113-amino acid segment of the P domain that shares 92% sequence identity to human LPC and 95% sequence identity to rat PC7 (data not shown). When this cDNA was used to probe a Northern blot of RPE.40 poly(A) ϩ RNA, a single transcript of 4.4 kilobase pairs was detected (Fig. 7), indicating that LPC is expressed in RPE.40 cells.
Processing of Human Pro-IGF-IA in Various Cell Lines-Since IGF-I is synthesized by all tissues of the body, we were interested in determining if pro-IGF-IA-processing enzymes are also ubiquitous. We expressed pro-IGF-IA-FLAG in cell lines derived from several mammalian sources as well as one invertebrate cell line. Secreted IGF-I peptides were immunoprecipitated from conditioned medium and analyzed by gel electrophoresis, as described under "Experimental Procedures." The results have been summarized in Table III. All cell lines examined processed pro-IGF-IA to IGF-I-(1-70). IGF-I-(1-76) was also generated by all cells except the furin-deficient LoVo and RPE.40 cells and the Drosophila cell line. Pro-IGF-IA was also processed very efficiently. Endogenous enzymes were able to convert at least 50% of overexpressed precursor to final product in all cell lines (data not shown). DISCUSSION We have expressed wild-type and mutant pro-IGF-IA with furin, PACE4, PC6A, PC6B, and LPC in order to determine which of these enzymes might play a role in pro-IGF processing. In 293 cells, endogenous enzymes process wild-type pro-IGF-IA and the K65R mutant at Lys 68 -X-X-Arg 71 to yield IGF-I-(1-70) and at Arg 74 -X-X-Arg 77 to produce IGF-I-(1-76) (Fig.  2). The K68R mutant is processed at Arg 68 -X-X-Arg 71 to generate IGF-I-(1-70), and IGF-I-(1-76) is apparently not produced from this precursor by 293 enzymes. The lack of IGF-I-(1-76) could be due to a preference for Arg 68 -X-X-Arg 71 over Arg 74 -X-X-Arg 77 . It is also possible that pro-IGF-IA is processed sequentially from IGF-I-(1-76) to IGF-I-(1-70) and that the Arg for Lys 68 substitution creates a more efficient cleavage site or a site favorable to additional processing enzymes, resulting in complete processing to the final product. Pulse-chase analysis will be necessary to determine if the K68R precursor is sequentially cleaved. Endogenous processing enzymes in 293 cells produce two products from the K65R/K68R precursor (Fig.  2). The main product corresponds to IGF-I-(1-70) generated by cleavage at Arg 68 -X-X-Arg 71 . The smaller peptide likely corresponds to IGF-I-(1-67) resulting from cleavage at the newly created furin site, Arg 65 -X-X-Arg 68 . This suggestion is supported by the fact that the smaller peptide is not generated from K65R/K68R in the furin-deficient RPE.40 cell line (Fig. 3).

FIG. 7. Expression of LPC in RPE.40 cells.
Four micrograms of poly(A) ϩ RNA from RPE.40 cells was fractionated on a 1% agarose gel and blotted to a nylon membrane. The blot was hybridized with a 32 P-labeled nick-translated probe to hamster LPC and exposed to film for 3 days with an intensifying screen. Molecular size markers are indicated on the right. Kb, kilobase pairs.

TABLE III
Processing of human pro-IGF-IA in various cell lines Cells were transfected with pCMVigf1-FLAG, and the secreted IGF-I peptides were analyzed by gel electrophoresis, as described "Experimental Procedures." IGF-I-(1-70) was produced by all cell lines examined, as indicated by the plus sign. The presence (ϩ) or absence (Ϫ) of IGF-I-(1-76) is indicated in the column at the far right. A faint band migrating slightly faster than IGF-I-(1-70) is also produced and could represent cleavage at an upstream lysine. The fact that furin generates IGF-I-(1-70), but not IGF-I-(1-76), from pro-IGF-IA could be due to a preference for Lys 68 -X-X-Arg 71 over Arg 74 -X-X-Arg 77 or the result of sequential processing, with IGF-I-(1-70) being the major final product. Since each site is cleaved with high efficiency when the R74A/R77A and R71A mutants are co-expressed with furin in 293 or RPE.40 cells (Fig. 4, A and B), we suggest that the latter hypothesis is more likely. Furin processes the K65R and K68R mutants to IGF-I-(1-70) and a smaller band that may result from cleavage at one of the two upstream basic residues ( Fig.  2A). Only one product, which migrates slightly ahead of IGF-I-(1-70), is produced from the K65R/K68R mutant. This 7-kDa band likely represents IGF-I-(1-67) produced by cleavage at the Arg 65 -X-X-Arg 68 site. Again, cleavage may be sequential or reflect processing site specificity.
In contrast to furin, co-expression of PACE4, PC6A, PC6B, or LPC with pro-IGF precursors in 293 cells does not qualitatively change the pattern of processing products generated by the endogenous enzymes (Fig. 2, B, C, D, and E). However, some quantitative changes are apparent. PC6A and LPC are most efficient at generating IGF-I-(1-70) from pro-IGF-IA, followed by PC6B and then PACE4 (Table I). These enzymes appear to have a more strict substrate specificity than furin, since they produce only IGF-I-(1-70) from pro-IGF precursors. However, PACE4, PC6A, and PC6B are capable of cleaving at Lys 68 -X-X-Arg 71 or Arg 74 -X-X-Arg 77 (Fig. 4, A and B). Other studies comparing specificity and activity of SPCs have also found that furin has the widest substrate specificity and is more active than PACE4, PC6A, and PC6B. Furin will efficiently cleave both Arg-X-(Lys/Arg)-Arg and Arg-X-X-Arg motifs (25,26), while PACE4 and PC6A prefer substrates with a basic residue in the P2 position (14,19,27,28). Neurotropin precursors and human immunodeficiency virus gp160 contain Arg-X-(Lys/ Arg)-Arg motifs that are cleaved more efficiently by furin than PACE4 or PC6B, and not at all by PC6A (21,22,24). However, PC6A is more active than PACE4 at cleaving the Arg-X-(Lys/ Arg)-Arg site of receptor protein-tyrosine phosphatase (23), indicating that structural features are also important for processing activity. Although we have not extensively examined the substrate specificity of LPC, our data suggest that this convertase may prefer the Lys-X-X-Arg 71 site to the Arg-X-X-Arg 77 site (Fig. 4). Processing of pro-IGF by endogenous and transfected enzymes in 293 cells is summarized in Fig. 8A.
From our experiments in 293 cells it is clear that the endogenous processing enzymes cleave pro-IGF-IA very efficiently and that furin is more active on pro-IGF precursors than PACE4, PC6A, PC6B, or LPC. We therefore thought it would be informative to examine processing of pro-IGF-IA in a furindeficient cell line. We have used RPE.40 cells for this purpose. These cells were derived from CHO-K1 cells by exposure to the mutagen ethyl methane sulfonate and selection for resistance to Pseudomonas exotoxin A (29). RPE.40 cells fail to process several proprotein precursors due to mutations in both furin alleles (20). When expressed in RPE.40 cells, wild-type pro-IGF-IA is processed exclusively to IGF-I-(1-70) by endogenous enzymes (Fig. 3). Similar results were obtained previously in LoVo cells (13). It therefore appears that furin is required to generate IGF-I-(1-76), although co-expression of wild-type pro-IGF-IA and furin in 293 or RPE.40 cells results in processing to IGF-I-(1-70), not IGF-I-(1-76) ( Figs. 2A and 3A). This may be due to sequential processing of pro-IGF-IA by furin, as discussed above.
An unexpected result from the experiments with RPE.40 cells was that PACE4, PC6A, and PC6B do not process pro-IGF-IA precursors in this cell line ( Fig. 3 and Table II). This could be due to a requirement for furin to activate PACE4, PC6A, and PC6B. However, others have shown that these enzymes cleave prosomatostatin and neurotropin precursors in LoVo cells (21,22,30). Furthermore, we have found that PACE4, PC6A, and PC6B will cleave the K68R/A70K mutant in RPE.40 cells. This mutant contains a basic residue in the P2 position and may therefore be a better substrate for these enzymes (Fig. 6). It is also possible that PACE4, PC6A, and PC6B act primarily on IGF-I-(1-76) that would be generated by furin but is not produced in RPE.40 cells. This possibility was eliminated by showing that PACE4, PC6A, and PC6B process IGF-I-(1-76) to IGF-I-(1-70) in 293 cells but not in RPE.40 cells (Fig. 5). A third possibility is that expression levels of exogenous enzymes may vary in 293 and RPE.40 cells. Using ribonuclease protection assays, we determined that mRNA levels for PACE4, PC6A, and PC6B were 1.5-2.3-fold higher in 293 cells than RPE.40 cells (data not shown), suggesting that lower expression may account for their lack of activity in RPE.40 cells. Expression levels did not strictly correlate with processing activity, however. For example, PACE4 had the highest mRNA levels but the lowest activity in both cell lines. Processing of pro-IGF by endogenous and transfected enzymes in RPE.40 cells is summarized in Fig. 8B.
Our previous work clearly identified the processing site for final maturation of pro-IGF-IA as Lys 68 -X-X-Arg 71 (13). This dibasic motif has been shown to be cleaved by enzymes of the SPC family (8). The data presented here indicate that furin, PC6A, and LPC are capable of efficiently processing pro-IGF-IA to mature IGF-I. We have also found that processing activity can be influenced by enzyme expression levels and that all cells examined contain endogenous enzymes that efficiently and correctly process pro-IGF-IA. Furin, PC6A, and LPC are expressed in many tissues (6 -12), suggesting that these and perhaps other members of the SPC family that have not yet been described may account for our observation that pro-IGF-I processing is a function common to all cells.