The complement component C1s is the protease that accounts for cleavage of insulin-like growth factor-binding protein-5 in fibroblast medium.

Cultured fibroblasts secrete an 88-kDa serine protease that cleaves insulin-like growth factor binding protein-5 (IGFBP-5). Because IGFBP-5 has been shown to regulate IGF-I actions, understanding the chemical identity and regulation of this protease is important for understanding how IGF-I stimulates anabolic functions. The protease was purified from human fibroblast-conditioned medium by hydrophobic interaction, lectin affinity, and heparin Sepharose affinity chromatography followed by SDS-polyacrylamide gel electrophoresis. An 88-kDa band was excised and digested with lysyl-endopeptidase. Sequencing of the high pressure liquid chromatography-purified peptides yielded the complement components C1r and C1s. To confirm that C1r/C1s accounted for the proteolytic activity in the medium, immunoaffinity chromatography was performed. Most of the protease activity adhered to the column, and the eluant was fully active in cleaving IGFBP-5. SDS-polyacrylamide gel electrophoresis with silver staining showed two bands, and IGFBP-5 zymography showed a single 88-kDa band. Amino acid sequencing confirmed that the 88-kDa band contained only C1r and C1s. C1r in the fibroblast medium underwent autoactivation, and the activated form cleaved C1s. C1s purified from the conditioned medium cleaved C(4), a naturally occurring substrate. The purified protease cleaved IGFBP-5 but had no activity against IGFBP-1 through -4. C1 inhibitor, a protein known to inhibit activated C1s, was shown to inhibit the cleavage of IGFBP-5 by the protease in the conditioned medium. In summary, human fibroblasts secrete C1r and C1s that actively cleave IGFBP-5. The findings define a mechanism for cleaving IGFBP-5 in the culture medium, thus allowing release of IGF-I to cell surface receptors.

Insulin-like growth factor-I (IGF-I) 1 is a potent trophic factor for multiple cell types (1). The mitogenic potential of IGF-I is controlled by a family of high affinity IGF binding proteins (IGFBPs) that are ubiquitously present in interstitial fluids (2). The concentrations of IGFBPs and their affinity constants are such that at equilibrium most of the IGF-I is bound (3). Factors that disrupt this binding, such as proteolysis, result in the release of IGF-I to receptors (4,5).
IGFBP-5 has been shown to be an important regulator of IGF-I actions in mesenchymal cell types (6). Both cultured human fibroblasts and smooth muscle cells secrete an IGFBP-5 protease that is specific for IGFBP-5 and cleaves it into a 22-kDa fragment, which has more than 1000-fold reduction in its affinity for IGF-I (7,8). Studies utilizing a protease-resistant IGFBP-5 mutant have shown that high concentrations of the mutant IGFBP-5 (at least a 5-fold molar excess over IGF-I) completely inhibited IGF-I-mediated receptor activation (9). In contrast, if a 1:1 molar ratio of native IGFBP-5 to IGF-I is added to extracellular matrix, IGFBP-5 can act to potentiate the mitogenic effect of IGF-I (10). Therefore, the factors that control proteolytic cleavage of IGFBP-5 represent an important mechanism for controlling the amount of IGF-I that is available to interact with receptors. Although some broad spectrum proteases, such as plasmin, thrombin, or matrix metalloproteases-2 and -9, have been shown to cleave several forms of IGFBPs, the identity of IGFBP protease activities that are specific for a single form of IGFBP has been difficult to determine (9,(11)(12)(13)(14). Recently, a partially purified fraction of fibroblast-conditioned medium containing the pregnancy-associated plasma protein-A protease was shown to specifically cleave IGFBP-4 (15). However a homogenous preparation of a protease that specifically cleaves IGFBP-5 has not been reported. For these reasons, we purified and characterized the IGFBP-5-specific protease that is present in fibroblast-conditioned medium.

EXPERIMENTAL PROCEDURES
Cell Culture-Human dermal fibroblasts (GM 10) were purchased from Coriell Institute (Camden, NJ). The cells were grown to confluency in 175-cm 2 tissue culture flasks (Falcon Labware, Fairfield, NJ) in minimum essential media (Life Technologies, Rockville, MD) supplemented with 10% bovine serum (Colorado Serum Co., Denver, CO). To collect conditioned medium, the monolayers were washed three times with PBS, and 150 ml of serum-free minimum essential medium was added per flask. The medium was collected after 48 h, centrifuged at 16,000 ϫ g for 20 min to remove cellular debris, and then stored at Ϫ70°C.
Protein Purification-Ammonium sulfate (Mallinckrodt, Baker, Paris, KY) was added to 12 liters of conditioned medium to 85% satu-ration, and the mixture was stored for 14 h at 4°C, then centrifuged at 23,500 ϫ g for 1 h. The pellet was extracted with 600 ml of 20 mM Tris, 4 mM CaCl 2 , 10 mM NaCl, pH 7.4. The extract was adjusted to 1.0 M ammonium sulfate, stirred for 2 h at 4°C, and centrifuged at 23,500 ϫ g for 45 min. The supernatant was applied to a 4.4 ϫ 3.5 cm butyl-Sepharose column-4 (Fast Flow, Amersham Pharmacia Biotech). The flow-through material was discarded, and the column was eluted with 20 mM Tris-HCl, pH 7.4, then tested for proteolytic activity. The fractions that were eluted were tested for activity by incubating between 1 and 10 l with 100 ng of IGFBP-5 for 14 h at 37°C in 50 l of 0.05 M Tris, 4 mM CaCl 2 , pH 7.2 (7). The products of the reaction were analyzed by SDS-PAGE with immunoblotting using a specific anti-IGFBP-5 antiserum, as described previously (16). The eluant, approximately 350 ml, was applied directly to a wheat germ-agarose column (1.6 ϫ 5.0 cm; Sigma Chemical Co., St. Louis, MO) that had been equilibrated with 20 mM Tris, 4 mM CaCl 2 , 0.4 M NaCl, pH 7.2. The fractions were eluted with the same buffer containing 0.5 M N-acetyl-D-glucosamine. The fractions containing proteolytic activity were determined as described above, and then the fractions with the greatest amount of activity were pooled (approximately 100 ml of eluant) and diluted 1:1 with 20 mM Tris, 4 mM CaCl 2 , 2 mM MnCl 2 , 0.4 M NaCl, pH 7.2. This solution was applied to a concanavalin A-agarose column (Sigma; 1 ϫ 9 cm) that had been equilibrated with the same buffer. The column was eluted with the same buffer containing 0.5 M ␣-methylmannopyranoside (Sigma), and the fractions were tested for activity as described previously. The pool of active fractions was diluted to 30 mM NaCl, 42 mM ␣-D-mannopyranoside, and applied to a 2.5 ϫ 6 cm heparin-Sepharose column (Amersham Pharmacia Biotech) that had been equilibrated with 20 mM Tris, 4 mM CaCl 2 , 30 mM NaCl, pH 7.2. The column was eluted with the same buffer containing either 0.5 or 1.0 M NaCl, and the fractions were tested for protease activity.
Immunoaffinity Chromatography-Antiserum to C1r and C1s was prepared using synthetic peptides that contained the following sequences: C1r (CLYPKEHEAQSNASLDVFLGHTNVEE) and C1s (CVEGNREPTMYVGSTSVQTSRLAKS). The peptides (6.1 mg of C1r and 8.8 mg of C1s) were each conjugated to 4 mg of maleimide-activated KLH (Pierce, Rockford, IL). Each peptide was linked with KLH in 1.4 ml of 0.83 M Na 2 H 2 PO 4 , pH 7.2, containing 0.9 M NaCl, 0.35 M EDTA. Following dialysis in the same buffer without EDTA, the mixture was lyophilized. New Zealand White rabbits were immunized with 1 mg of linked KLH peptide in complete Freund's adjuvant. The rabbits were injected with 0.5 mg of KLH peptide in incomplete Freund's adjuvant at monthly intervals. To prepare an affinity column, 6 ml of C1r and C1s antisera were each diluted to 12 ml with 20 mM NaH 2 PO 4 , pH 7.0, and each solution was applied to a protein G-Sepharose-4 (Fast-flow, 1.0 ϫ 7.0 cm; Amersham Pharmacia Biotech) affinity column. The IgG was eluted with 0.1 M glycine HCl, pH 2.5, and immediately neutralized with 1.0 M Tris to pH 7.2. The purity of the IgG was 95-99% based on SDS-PAGE with silver staining. 15.4 mg of the anti-C1r and 15.4 mg of anti-C1s IgG were mixed and dialyzed against 0.02 M sodium acetate, pH 5.0, containing 0.15 M NaCl, then oxidized for 1 h with 0.01 M NaIO 4 . Oxidation was terminated with glycerol, and the mixture was then dialyzed against 0.1 M sodium acetate, pH 4.5, containing 0.10 M NaCl. The solution was mixed with Affi-Prep Hz support (Bio-Rad, Hercules, CA) for 24 h at 4°C, then washed. Approximately 10 ml of the pool of active fractions that had been eluted from heparin-Sepharose was diluted to 20 mM Tris, 50 mM NaCl, 4 mM CaCl 2 , pH 7.2 and applied to the anti-C1r/C1s affinity column (1.5 ϫ 5.5 cm) at 3 ml/h. The active fractions were eluted in that buffer containing 1.0 M NaCl.
Amino Acid Sequence Analysis-The partially purified material (4 -5 ml) was concentrated ϳ50-fold on an Ultrafree 0.5-ml centrifugal filter (Biomax-10K NMWL, Millipore, Bedford, MA) prior to SDS-PAGE. Both material that had been purified through the heparin-Sepharose step and material that had been immunoaffinity-purified were sequenced. The proteins were separated by SDS-PAGE, 9% gel. The gel was stained with Coomassie Blue, R-250 (Sigma) then destained for 2 h, and bands corresponding to molecular mass estimates of 88, 180, 280, and Ͼ 300 kDa were excised. These gel slices were washed three times in 450 l of 50% acetonitrile/0.2 M Tris, pH 9.0. The gel pieces were incubated in 200 l of 0.1 M Tris, pH 9.2, containing 0.4 g of lysylendopeptidase (Wako Bioproducts, Richmond, VA) for 14 h at 37°C, and then they were sonicated for 30 min at room temperature. An additional 0.2 g of lysyl-endopeptidase was added, and the incubation was continued for 5 h at 37°C. The peptides were extracted by adding 100 l of 0.1 M Tris, pH 9.0, followed by 30 min of sonication. This was repeated using 400 l of buffer then followed by two additional extractions with 400 l of 60% acetonitrile, 0.2% trifluoroacetic acid. All four extracts were pooled and diluted to 6% acetonitrile with 0.04% trifluoroacetic acid. The pH was adjusted to 3.0, and the solution was applied to a reverse-phase HPLC C18 (2 mm ϫ 15 cm) column (VYDAC, Hesperia, CA). The column was eluted with 0.04% trifluoroacetic acid and a linear acetonitrile gradient (10 -60%) over 75 min. Selected peptides were repurified using the same solvents except that the gradient was extended to 125 min. Multiple internal peptides that were obtained from each band were sequenced. Automated Edman degradation chemistry was used to determine the NH 2 -terminal sequence for the unknown peptides (17). The following systems were employed for these analyses: I. Applied Biosystems model 494 protein sequencer, model 140C microgradient system, and model 785A absorbance detector. The PTH-derivatives were identified by reverse-phase HPLC analysis in an on-line fashion and utilized a PerkinElmer Life Sciences/Brownlee 2.1-mm inner diameter PTH-C18 column. II. PerkinElmer Life Sciences/Applied Biosystems model 492 cLC protein sequencer model 140D microgradient system and model 785A absorbance detector. The online reverse-phase HPLC analysis of the PTH-derivatives was performed on a PerkinElmer Life Sciences/Brownlee 0.8-mm inner diameter PTH-C18 column.
Gel Electrophoresis and Immunoblotting-Purified protein fractions or concentrated conditioned media samples were loaded onto 9% SDSpolyacrylamide gels and electrophoresed as described previously (18). Appropriate protein markers were run in a parallel lane to determine molecular weight. For some analyses, the gel was fixed with 5% acetic acid, 10% ethanol, and analyzed by silver staining. For other analyses, the proteins were transferred to polyvinylidene difluoride membranes, as described previously (19). The membranes were incubated with a 1:500 dilution of C1r or C1s antiserum in 2.0 ml of PBS containing 1% BSA, pH 7.0, for 14 h. After extensive washing in PBS, they were incubated for 3 h, diluted 1:1500 with goat anti-rabbit IgG conjugated to alkaline-phosphatase in TBS with 0.1% BSA, and then the immune complexes were detected as described previously (16). Pure C1r (400 ng) (Calbiochem, La Jolla, CA) or C1s (400 ng) (Enzyme Research, South Bend, IN) were run in parallel lanes to confirm that the antibodies recognized proteins of the correct molecular weight. For immunoblotting of C 4 , a 7.5% gel was used. The anti-C 4 antiserum (Calbiochem) was used at a 1:500 dilution. For immunoblotting for IGFBP-1, -2, -3, and -4, the specific antisera utilized and the conditions used have been described previously (7).
IGFBP-5 Zymography-Seven micrograms of IGFBP-5 was mixed with 4.0 ml of acrylamide gel solution (10% gel), and the gel was polymerized (5). The IGFBP-5 protease-containing fractions were concentrated 5-20ϫ by centrifugation using Ultrafree 0.5-ml centrifugal filters. The filter was exposed to 10 l of 3ϫ Laemmli sample buffer, and this was pooled with the concentrate (20 l) then electrophoresed at room temperature. The gel was washed in 2.5% Triton X-100 at 4°C for 1 h, followed by extensive washing in distilled water. The gel was incubated in 0.05 M Tris, 4 mM CaCl 2 , pH 7.4, overnight at 37°C to allow for proteolysis and for capillary transfer to a polyvinylidene difluoride membrane (9). The membrane was immunoblotted using a 1:1000 dilution of IGFBP-5 antiserum as described above. The electrophoretic mobility of the bands that were detected was compared with prestained molecular weight standards (Life Technologies).
Analysis of Protease Activation-To determine if the purified C1r could undergo autoactivation, 500 ng of material was incubated in 0.02 ml of 0.025 M MES, 125 mM NaCl, 2 mM EGTA, pH 7.2, for 15-90 min. The products of the reaction were then analyzed by SDS-PAGE (9% gel) using reducing conditions, 0.1 M dithiothreitol, followed by immunoblotting for C1r and C1s. To determine the ability of the purified C1s to be cleaved by C1r, 400 ng of the highly purified protease was exposed to 800 ng of C1r in 30 l of the buffer listed previously, incubated at 37°C for 2 h, and then analyzed by SDS-PAGE (9% gel) under reducing conditions, followed by immunoblotting for C1s. To determine if the purified IGFBP-5 protease had activity in cleaving C 4 , 180 or 540 ng of the highly purified protease was incubated with 600 ng of C 4 (Calbiochem) in 25 l of 25 mM MES, 125 mM NaCl, 2 mM EGTA, pH 7.2, for 18 h and the products analyzed SDS-PAGE (7% gel) with immunoblotting for C 4 . Duplicate tubes containing 200 or 500 ng of pure C1s (Enzyme Research) were also analyzed. To determine the ability of the purified IGFBP-5 protease to cleave IGFBP-5, between 3 and 50 ng of protein was incubated with 100 ng of IGFBP-5 in 60 l of 0.05 M Tris, pH 7.2, with 4 mM CaCl 2 for variable time periods up to 3.5 h at 37°C. To determine the effects of protease inhibitors on the ability of IGFBP-5 protease in conditioned media to cleave IGFBP-5, 25 l of fibroblast conditioned media was incubated with 100 ng of IGFBP-5 for 16 h at 37°C in the buffer listed previously. The following inhibitors were analyzed, PB-145 (10 Ϫ7 M), a synthetic peptide with sequence similarity to anti-thrombin III (20), heparin co-factor II (10 Ϫ7 M), antithrombin III (10 Ϫ7 M). In additional experiments, the ability of the purified protein to cleave other forms of IGFBPs was determined by incubating 50 ng of purified protease with 150 ng of pure IGFBP-1, -2, -3, and -4 as stated for IGFBP-5. The products of the reactions were then analyzed by SDS-PAGE (12.5% gel) followed by immunoblotting for each specific IGFBP.

RESULTS
Silver stain analysis of the material purified through the heparin-Sepharose step showed a prominent 88-kDa band, a high molecular weight band that did not enter the gel, and two other bands with molecular mass estimates of 180 and 280 kDa (Fig. 1A). Minor bands were detected at 92 and 240 kDa. Amino acid sequence analysis of the 88-kDa band showed that it contained the C1r and C1s sequences. Eleven peptides that were sequenced had sequences corresponding to human C1r, and eight peptides had sequences corresponding to human C1s (Fig. 2). Each of these peptides was distinct and was not a mixture of the two sequences. Likewise, single HPLC peaks yielded pure peptides that were either pure C1r or pure C1s, but, within a single digested band, peaks that contained either C1r or C1s were detected. When the faint band that was detected at 92 kDa was carefully excised from the gel and separated from the lower 88-kDa band, sequences corresponding to C1r only were obtained. The 180-kDa band gave the sequences of the ␣1 and ␣2 chains of collagen type VI. The 280-kDa band contained two peptides with sequences corresponding to tenascin-C, and the band that did not enter the gel contained multiple peptides with sequences encoding collagen type VI or thrombospondin-1. Both tenascin-C and thrombospondin-1 have been shown to bind to IGFBP-5 (21,22). Immunoblotting of the identical material with anti-C1r antiserum showed a prominent 88-kDa band that co-migrated with the intact C1r standard and a second band that had a molecular mass estimate of Ͼ200 kDa (Fig. 1B). When analyzed following reduction, the major C1r band was detected that had a molecular mass estimate of 94 kDa, and less abundant bands were detected at 62 and 38 kDa (Fig. 1B). These corresponded to size estimates of pure activated C1r. Immunoblotting for C1s showed prominent bands with molecular mass estimates of 88 and Ͼ200 kDa (non-reduced) and 86, 68, and 28 kDa (following reduction). IGFBP-5 zymography of this same material showed a prominent band of activity with an 88-kDa molecular mass estimate, and a less intense band was detected that had an estimated molecular mass of 190 kDa (Fig. 2C). Four independently run gels, followed by excision of the 88-kDa band and sequence analyses, showed only C1r and C1s, and no other proteins were present in this band. When IGFBP-5 protease activity was analyzed, 25 ng of this material (400 ng/ml) cleaved 100 ng of IGFBP-5 in 1.5 h (Fig. 2D). There was a concentration-dependent increase in activity, and cleavage was detected using protease concentrations as low as 3.5 ng (60 ng/ml).
Because the heparin-Sepharose-purified material had several bands that were detected by silver staining, the material was further purified by immunoaffinity chromatography. Silver stain analysis of the material that was eluted showed three detectable bands (Fig. 3A). Sequence analysis of the 88-kDa band confirmed the presence of C1r and C1s, and no other peptide sequences were detected. Sequence analysis of the band that did not enter the gel revealed thrombospondin-1. The 160-kDa band yielded rabbit IgG, suggesting that this band The arrows denote the positions of the four bands that were detected. For sequencing, 250 l of the active fraction was run in two lanes, that were stained with Coomassie Brilliant Blue R250. The same four bands that had been detected by silver stain analysis were excised, digested, and sequenced. B, immunoblotting for C1r and C1s. The fraction that was analyzed by silver staining was also analyzed by immunoblotting for C1r (lanes 2 and 6) and C1s (lanes 4 and 8) under non-reducing (lanes 1-4) and reducing (lanes 5-8) conditions. Pure C1r standard (lanes 1 and 5) and pure C1s standard (lanes 3 and 7) were also analyzed. The single arrow (lanes 1-4)  had been cleaved and eluted during immunoaffinity chromatography. Immunoblotting for C1s showed an intense 88-kDa band and faint bands of Ͼ200 kDa and 68 kDa. Following reduction, bands with estimated molecular masses of 92 and 70 and 30 kDa were detected (Fig. 3B). Immunoblotting for C1r showed two bands at 88 and 92 kDa and a band with an estimated molecular mass of 180 kDa. Analysis of IGFBP-5 protease activity showed that the material was fully active and yielded the expected 22-kDa band. Approximately 2 ng of the material degraded 100 ng of IGFBP-5 (2 g/ml) in 3.5 h (Fig.  3C). This result establishes that the material purified by this procedure that cleaved IGFBP-5 into a 22-kDa fragment contained only C1r and C1s, and no other proteases could be identified in the 88-kDa band or any of the protein contaminant bands.
C1r has been shown to cleave C1s, which is secreted as a zymogen, to generate its active form (23,24). To determine if pure C1r could activate our purified C1s, activated C1r was incubated with our purified material, and then it was analyzed by immunoblotting after reduction for evidence of activation. Immunoblotting showed full activation of our purified C1s that was comparable to activation of a purified C1s standard (Fig.  4a). To exclude the possibility that all of the IGFBP-5 protease activity was due to C1r, activated C1r was incubated with IGFBP-5. A high concentration (e.g. 2-5 g of C1r) was required to detect cleavage (Fig. 4b). Furthermore, cleavage proceeded slowly, and the fragments that were obtained had estimated molecular masses of 24 and 26 kDa. In contrast, a much lower concentration of activated C1r (e.g. 400 -800 ng) cleaved C1s into the 67-and 30-kDa forms (Fig. 4b). To determine if C1r in the purified samples could further autoactivate, and if its autoactivation was associated with C1s activation, we incubated the heparin-Sepharose-purified material for 0.5, 1, and 2 h at 37°C, then determined the degree of C1r and C1s activation by immunoblotting. As shown in Fig. 5, a time-dependent increase in activated C1r and C1s was detected. These findings suggest that C1r is functioning to activate C1s, which is cleaving IGFBP-5. To confirm that the purified protease could cleave C 4 , the known physiologic substrate for C1s, its capacity to cleave C 4 was compared with pure C1s. The purified protease had greater activity in cleaving C 4 than did pure C1s (Fig. 6). To determine the specificity of the protease for IG-FBP-5, 50 ng of material purified through the immunoaffinity chromatography step was incubated with pure IGFBP-1, -2, -3, -4, or -5. As shown in Fig. 7, only IGFBP-5 was degraded by the purified protease. Because the IGFBP-5 proteolytic activity in crude fibroblast-conditioned medium is specific for IGFBP-5 (7), this finding further supports the conclusion that we have purified the protease that accounts for the IGFBP-5 proteolytic activity in fibroblast-conditioned medium.
To further confirm that C1r and C1s accounted for IGFBP-5 proteases in fibroblast-conditioned medium, several serine protease inhibitors that had been shown to inhibit IGFBP-5 proteolysis to some extent were incubated with crude material that had only been purified through the first step, and their ability to inhibit IGFBP-5 cleavage was determined. As shown in Fig. 8, several of these inhibitors at least partially inhibited IGFBP-5 proteolysis. C1 inhibitor, the physiologic inhibitor of C1s activation (25), was the most potent inhibitor of the IG-FBP-5 protease activity contained in the fibroblast-conditioned medium. The other inhibitors that had less of an effect on C1s activation (data not shown) also had less of an effect on IG-FBP-5 proteolysis. DISCUSSION We have reported previously that human fibroblasts secrete an IGFBP-5 protease that cleaves IGFBP-5 into predominantly a 22-kDa fragment, and this fragment has a low affinity for IGF-I (8). This proteolytic activity is a serine protease and is specific for IGFBP-5 (7). Because IGFBP-5 is an important modulator of IGF-I bioactivity, the factors that regulate the activity of this protease have the potential to regulate the ability of IGFBP-5 to modulate IGF-I actions. For that reason, we were interested in determining the molecular identity of IGFBP-5 protease activity in fibroblast-conditioned medium.
To that end, we purified a large quantity of IGFBP-5 protease from human fibroblast-conditioned medium and subjected it to amino acid sequence analysis. Sequence analysis yielded several peptides that encoded C1r and C1s. Because these proteins could not be separated, it is not possible to definitively determine which protease is actually cleaving IGFBP-5; however, C1s is known to have a much broader range of substrates (25). Furthermore, a very low concentration (60 ng/ml) of the C1r/ C1s mixture cleaved IGFBP-5 rapidly. In contrast, IGFBP-5 FIG. 4. Activity of C1s. a, 400 ng of the heparin-Sepharose-purified material was incubated with 800 ng of C1r as described under "Experimental Procedures," and the products were analyzed by SDS-PAGE with dithiothreitol following immunoblotting for C1s (lanes 4 -6). The same amount of purified material was allowed to autoactivate for identical times (lanes 1-3). The incubations were 0 min (lanes 1 and 4), 30 min (lanes 2 and 5), and 60 min (lanes 3 and 6). The arrows denote the positions of intact C1s and its two major fragments. The addition of extra C1r resulted in more rapid cleavage of C1s at 30 and 60 min after analysis using reducing conditions. b, activity of C1r. To determine if C1r cleavage of IGFBP-5 was accounting for most of the activity of the purified protease, pure C1r (2 g) was incubated with IGFBP-5 for 18 h at 37°C, and then the fragments were analyzed by immunoblotting. As shown in lane 1, there was detectable IGFBP-5 cleavage, although the proteolytic fragments had estimated molecular masses of 26 and 24 kDa. To determine whether this C1r was fully active, 40 ng was incubated with C1s (200 ng), and the fragments were analyzed as in panel a. Lane 2, no incubation; lane 3, 1-h incubation. The material that had been immunoaffinity purified (50 ng) was tested for its capacity to cleave other forms of IGFBPs, as described under "Experimental Procedures." Each form of IGFBP (150 ng) was incubated with the protease for 16 h at 37°C, and then each was analyzed by immunoblotting using specific antisera. The preparation and specificity of the antisera that were used to detect IGFBP-1 through -4 have been described previously (7). No fragments were detected except when IGFBP-5 was used as a substrate. cleavage by C1r required a much higher concentration, proceeded slowly, and yielded fragments that had size estimates that were different from those that were detected after cleavage by the purified IGFBP-5 protease. These results do not exclude the possibility that C1r may be cleaving IGFBP-5 to some extent in the medium, but they suggest that C1s accounts for the major portion of the IGFBP-5 protease activity. Because C1s is secreted at least in part as an inactive zymogen, the primary role of C1r may be to cleave and activate C1s, which subsequently cleaves IGFBP-5. This conclusion is also supported by the known specificity of both enzymes for specific recognition sequences (25).
That these enzymes were the predominant protease activity for IGFBP-5 in fibroblast-conditioned medium is proven by several points. First, during purification, other proteases that have been shown to cleave IGFBP-5, such as ASPO-5 and MMP-2 and -9 were removed by various chromatographic steps; however, most of the proteolytic activity was retained. Furthermore, when pure MMP-2, -9, or ASPO-5 are incubated with IGFBP-5, the rate of proteolysis is slow and the fragment sizes that are generated are distinct from those generated by the IGFBP-5 protease activity (7). Finally, MMP-2 and -9 are not specific for IGFBP-5 and degrade other forms of IGF binding proteins. In contrast, the proteolytic activity in fibroblast media and the immunoaffinity-purified material were specific for IGFBP-5 and did not cleave other forms of IGF binding proteins. Second, immunoaffinity chromatography of partially purified conditioned medium showed that most of the IGFBP-5 protease activity in the medium could be accounted for by C1r and C1s that adhered to the antibody affinity column. Third, the specific C1s protease inhibitor, C1 inhibitor, inhibits IG-FBP-5 cleavage by crude conditioned medium, and its ability to inhibit C1r and C1s activation correlates with its ability to inhibit IGFBP-5 proteolysis. Furthermore, extensive sequencing of all of the protein bands that were detected in the most highly purified material did not yield sequences corresponding to any other protease. Following immunoaffinity chromatography of fibroblast medium, approximately 84% of the IGFBP-5 proteolytic activity adhered to the column (data not shown). Taken together, these data strongly suggest that C1s is the predominant protease cleaving IGFBP-5 in fibroblast medium and C1r is responsible for its activation.
Recently, it was shown that a part of the IGFBP-4 protease in fibroblast medium could be ascribed to PAPP-A, which is a metalloprotease (15). However, in that study PAPP-A was not purified to homogeneity, and therefore the results did not exclude the possibility that other proteases were present in fibroblast medium that could potentially degrade IGFBP-4. Furthermore, the extent to which any of these proteases is active is determined not just by the concentration of protease, but whether it is secreted as the zymogen, the percentage activation and whether protease inhibitors are present. When the secretion of C1r and C1s was analyzed using several test conditions, partial activation of these proteases was noted. Therefore, partially activated forms exist in fibroblast medium, making it more likely that they account for IGFBP-5 protease activity.
The physiologic significance of activated C1r/C1s in fibroblast medium and their roles in modulating IGF-I actions remains to be determined. However, we have shown that a protease-resistant form of IGFBP-5 inhibits IGF-I actions (11). Therefore, the fact that these proteases are present in the medium in partially activated forms that can cleave IGFBP-5 into fragments with very low affinity suggests that they are capable of modulating IGF-I bioactivity. Furthermore, inhibition of their activity with a specific inhibitor (e.g. C1 inhibitor) results in inhibition of IGFBP-5 cleavage, suggesting that activation can be modulated in physiologic fluids that contain this inhibitor. In additional studies, we have detected C1 inhibitor in fibroblast-conditioned medium by immunoblotting (data not shown). Therefore, the variables that regulate C1r and C1s synthesis and activation, as well as the secretion of C1 inhibitor, have the potential to alter IGFBP-5 cleavage and thereby modulate IGF-I actions.
Other investigators have reported that activated C1r and C1s occur at sites outside the liver (25,26) and that they have proteolytic functions other than complement activation (27)(28)(29)(30). Specifically, pleural fluid and joint fluid contain activated C1r and C1s, as do cell culture supernatants from glial cells (25,31). Several central nervous system cell types have been shown to contain C1s peptide and messenger RNA. Spleen, liver, brain, and kidney have been shown to contain C1r and C1s mRNA (25). Several studies have also shown that C1r and C1s activation may lead to cleavage of peptides other than the complement components (27)(28)(29)(30), and several of the proteins that have been shown to be substrates for C1s are not traditional components of the complement pathway. Taken together, these findings suggest that C1r and C1s may have roles other than complement activation.
For several years, investigators have hypothesized a linkage between inflammation (i.e. acute complement activation during injury) and subsequent cellular repair processes. Because IGF-I and IGFBP-5 are secreted by several cells, such as macrophages or fibroblasts, which are involved in repair that occurs in response to injury, and both peptides are present in pericellular fluids (32), on cell surfaces, and in extracellular matrix of several mesenchymal cell types (33), it is possible that there is a coordinated linkage between activation of C1r and C1s during acute injury and the subsequent release of IGF-I to receptors that are present on the surface of cell types that are involved in tissue repair. IGFBP-5, which is also secreted by the connective tissue cell types that are involved in tissue repair (34,35), could be an important component for controlling the amount of IGF-I that can be released to receptors after C1r and C1s activation (35). Whether such a linkage exists deserves further exploration. In summary, we have determined that the predominant protease component of fibroblast-conditioned medium is ascribable to the complement subcomponents C1r and C1s. C1s is the major protease that cleaves IGFBP-5, but its full activation requires the presence of activated C1r. Further studies are necessary to determine the physiologic role of these proteases in activating a cascade of events leading to IGF-I receptor stimulation.