gate the biosynthesis and regulation of PN-I in human fibroblasts. Unlabeled PN-I could compete for the bind- ing of metabolically labeled PN-I to anti-PN-I, as.
THEJOURNALOF BIOLOGICAL CHEMISTRY 01986 by The American Society of Biological Chemists, Inc.
Vol. 261, No. 30, Issue of October 25, pp. 14184-14190,1986 Printed in U.S.A.
Biosynthesis of Protease Nexin-I* (Received for publication, May 5, 1986)
Eric W. Howard and Daniel J. KnauerS From the $Department of Developmental and Cell Biology, School of Biological Sciences, University of California, Irvine, California 9271 7
Protease nexin-I (PN-I) is representative of a newly described class of serine protease inhibitors secreted by human fibroblasts, the protease nexins. Protease nexins form covalent complexes with their targetproteases, subsequently binding tocells via specific receptors. PN-I preferentially binds thrombin, urokinase, trypsin, and plasmin, and its binding to thrombin is accelerated by heparin. We have previously described the production of a polyclonal antibody against PN-I which is able to block the binding of PN-1-proteinase complexes to cells and will immunoprecipitate metabolically labeled PN-I. Anti-PN-I was used to investigate thebiosynthesis and regulation of PN-I in human fibroblasts. Unlabeled PN-I could compete for thebinding of metabolically labeled PN-I toanti-PN-I, as shown by the elimination of the 43-kDa band representing PN-I on sodium dodecyl sulfate-polyacrylamide gel electrophoresis autoradiographs. Excision of this 43-kDa band from gels, followed by amino-terminal sequencing, showed a homogeneous protein that is homologous with that described by Scott et al. (Scott, R. W., Bergman, B. L., Bajpai, A., Hersh, R. T., Rodriguez, H., Jones, B. N., Barreda, C., Watts, s., and Baker, J. B. (1985)J. Biol. Chern. 260, 7029-7034). An analysis of the biosynthesis of the PN-I revealed that a lower M, precursor exists intracellularly. This apparent rough endoplasmic reticulum form appears as a doublet on sodium dodecyl sulfate gels, as does mature PN-I. The PN-I precursor was also sensitive to contains Nendoglycosidase H, suggesting thatit linked carbohydrates of the high mannose form. Mature PN-I is not sensitive to endoglycosidase H, but does contain 3 kDa of N-linked carbohydrate. PN-I appears to be constitutively secreted by fibroblasts. PN-I levels in conditioned media reach a steady state within 48 h, althoughPN-I synthesis maintains aconstant rate. This steady state is due to the continuous uptake of PN-I from medium, presumably through a specific receptor.
The recent interest in plasminogen activators, in large part due to their presumed roles in clot dissolution, cell migration, and tumor metastasis (2), has also stimulated the search for the physiological inhibitors of these proteases. A variety of tissues produce physiological inhibitors of plasminogen activators. Endothelial cells derived from aorta (3),vena cava (4), umbilical vein ( 5 ) , and cornea (6), as well as macrophages, ( 7 ) , glial cells (8), and fibroblasts (9), all produce 43-kDa
* This work was supported by Research Grant CA 22400 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
inhibitors of plasminogen activators. PN-I,’ a 43-kDa inhibitor secreted by fibroblasts and several other cell types, will inhibitthrombin, plasmin, and plasminogen activators at physiological concentrations by forming covalent complexes (1). Once bound to a protease, PN-I binds toa specific, heparin-competible receptor on fibroblasts, and the PN-I. protease complex is degraded in lysosomes (10). Antibodies developed against purified PN-I will immunoprecipitate complexes of [‘251]-thrombinwith PN-I from fibroblasts and cells derived from kidney tubule (CV-1) but not with the inhibitor found in endothelial cell cultures.* Anti-PN-I will also recognize a 43-kDa factor from rat brain glioma cells.3 The relationship between PN-I and other similar protease inhibitors is not yet known, but itis likely that their physiological roles are similar. They may represent a system that limits proteolysis at a local wound site, where damage to thevasculature causes the activation of several plasma proteases. Some cells which produce PN-I-like inhibitorsalso produce plasminogen activators, suggesting that much of the wound response is controlled locally (11). PN-I represents roughly 5% of the secreted proteins in human fibroblasts and can be highly purified using heparinSepharose chromatography (12). Two peaks representing PNI activity eluted off of heparin-Sepharose, at 400 and 600 mM NaC1. Upon gel electrophoresis, PN-I appears as a closely migrating doublet. Amino-terminal sequence analysis on electroeluted PN-I from SDS gels reveals that thepurified protein is homogeneous and is identical to the PN-I described elsewhere (1). We have recently described the properties of a polyclonal antibody against humanPN-I (13).Anti-PN-I is able to block the formation of complexes between PN-I and proteases, and can block the binding of these complexes to cells. This antibody will also immunoprecipitate metabolically labeled PN-I from conditioned media, as well as PN-1.protease complexes. In the present study, we have made use of these properties of anti-PN-I to investigate the biosynthesis and secretion of PN-I. We show that PN-I is synthesized from a lower M , precursor that runs as adoublet on SDS gels, suggesting that the doublet is not anartifact of PN-I purification. In addition, the regulation of PN-I levels in cultures has been investigated. The synthesis of PN-I appears to be constitutive. However, the level of PN-I in fibroblast-conditioned medium reaches a steady state within 48 h, while total protein secretion continues at a constantrate. This steady state is due to theconstant ‘The abbreviations used are: PN-I, protease nexin-I; endo H, endoglycosidase H endo F, endoglycosidase F HF, human foreskin fibroblasts; PBS, phosphate-buffered saline; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; RER, rough endoplasmic reticulum; BSA, bovine serum albumin; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonicacid; Hepes, 4(2-hydroxyethyl)-l-piperazineethanesulfonic acid. * E. W. Howard and D. J. Knauer, unpublished observations. E. W. Howard and D. J. Knauer, manuscript in preparation.
14184
PN-I Biosynthesis uptake of PN-I from the medium, either by a single receptor that recognizes both PN-I and PN-1.protease complexes, or by a separate mechanism. EXPERIMENTALPROCEDURES
Materials-HF cells were cultured as described previously (13). PN-I was purified from the concentrated conditioned medium of HF cells grown in microcarrier spinner flasks. Active PN-I was bound to a heparin-Sepharose column (Affi-Gel Heparin, Bio-Rad) by overnight circulation, and theneluted with a sodium chloride gradient, as described (1).The PN-I content of each fraction was assayed by its ability to inhibit thrombin's interactionwith a chromogenic substrate (14). Briefly, fractions from heparin-Sepharose chromatography were incubated with 1.6 pg/ml thrombin in PBS and 1 mg/ml ovalbumin (50-pl total volume). Samples were then cooled on ice, and 150 pl of buffer were added (10 mM Tris-HC1, pH 8.1, 150 mM NaCl, 1 mM EDTA, 100 pg/ml ovalbumin, 10 mg/ml polyethylene glycol, and 1.75 USP/mlheparin), followedby 100 p1 of1.9 mM chromozym-TH (Boehringer Mannheim). The reaction was stopped by the addition of 3.5 M acetic acid (100 pl), and theabsorbance was measured at 405 nm. Sequence Analysis-Purified PN-I was electroeluted from the 43kDa region of SDS gels, as described elsewhere (15). The eluted sample was resuspended in 1 mlof 0.1% (v/v) trifluoroacetic acid (buffer A), and then applied to a Vydac C, reversed-phase column (4.6 X 250 mm) equilibrated in 95% buffer A and 5%acetonitrile plus trifluoroacetic acid to balance the Az15. The column was developed with a linear gradient of 5-80% acetonitrile at 1 ml/min and at 1%/ min increase of acetonitrile. PN-I was eluted at 60% acetonitrile. This was then subjected to automated Edman degradation using an Applied Biosystems Model 470A protein sequenator. Phenylthiohydantoins were identified by high performance liquid chromatography using a Zorbax ODS (DuPont) silica column (4.1 X 250 mm), essentially as described (16). Metabolic Z~beling-[~H]Mannoselabeling was performed by incubatingconfluent 35-mmplateswith 200 pCi of[3H]mannose (Amersham Corp., trk. 675) in 1 ml oflow glucose medium (50 mg/liter), serum-free, for 16h. Pulse-chase analysis of PN-I processing was performed with confluent 35-mm plates that were washed with Dulbecco's PBS, incubated with serum-free, methionine-free medium for 45 min, and pulsed with 100 pCi of [Y3]rnethionine (Amersham, sj. 204) for 10 min in 1 ml of medium/well. Plates were chased by removing medium and replacing it with normal methionine medium (30 mg/liter) plus 1 mg/ml BSA. At specific time points, the media were isolated, along with cell lysates, and PN-I was immunoprecipitated. Continuous labeling of cells was performed in the presence of serum-free medium with 30 mg/liter unlabeled methionine and 1mg/ ml BSA. Cultures were incubated with 50 pCi of [35S]methioninein 500 p1 of medium in 24-well plates. The biosynthetic rate of PN-I was analyzed by labeling cells for 30 min with 100 pCi/ml [35S] methionine in serum-free medium plus 1 mg/ml ovalbumin and 300 pg/liter unlabeled methionine. Cells were washed with cold PBS, lysed in lysis buffer (10 mM CHAPS in 20 mM Hepes, 150 mM NaC1, 1 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, pH 7.5), and immunoprecipitated. Immunoprecipitation-Medium was isolated following labeling and treated with 2 mM phenylmethylsulfonyl fluoride and anti-PN-IIgG. Cells were washed with cold PBS, lysed with lysis buffer, and precipitated with anti-PN-I IgG. No difference was noted when the inhibitors chymostatin, pepstatin, antipain, leupeptin, and aprotinin were included. Following an overnight incubation with antibody at 4 "C, samples were treated with protein A-Sepharose at room temperature for 60 min, washed twice with lysis buffer and once with 20 mM Hepes, and solubilized with SDS sample buffer. Endoglycosidase Treatment-Samples to be treated with endo H were immunoprecipitated and washed, as described above, and then boiled in 25 p1 of 150 mM Tris, pH 7.8, plus 1%SDS and 1%2mercaptoethanol. The Microfuge tubes were centrifuged to remove protein A-Sepharose beads. Supernatants were added to 225 p1 of 150 mM citrate, pH 5.5, plus 75 ng of endo H (New England Nuclear) and incubated at 37 'C for 24 h (17). Samples were treated with 10 p1of 2% deoxycholate for 15 min and thenprecipitated with 100 pl of 50% trichloroacetic acid (18). Pellets were washed twice with ice-cold acetone and resuspended in reduced sample buffer. Endo F treatment was performed by resuspended immunoprecipitated PN-I in 0.1 M phosphate, pH 6.1, plus 50 mM EDTA, 0.5% Nonidet P-40, and 0.1% SDS. Samples were incubated for 24 h at 37 "C with 25 units/ml of
14185
endo F (New England Nuclear) and then acid-precipitated, as above (19). Electrophoresis, Fluorography, and Quantitation-Electrophoresis was performed using 10%polyacrylamide gels in the presence of SDS, as described by Laemmli (20). Gels were stained in Coomassie Blue, destained, impregnated with Fluorohance (Research Products), and exposed to Kodak XAR-5 film at -70 "C. Fluorographs were analyzed by densitometer scanning of underexposed film. These numbers were compared with counts of excised PN-I from the dried gel, which was obtained by solubilizing cut gel sections with TS-1 tissue solubilizer (Research Products) and water (l:l, v/v), heated to 50 "C overnight, added to scintillation fluid (3a70, Research Products), and counted. RESULTS
Analysis of PN-I Purity and Anti-PN-I Specificity-PN-I has been purified from the conditioned mediumof fibroblasts (1, 12). A grown onmicrocarrierbeadsinspinnerflasks polyclonal antibody againstthis material has been developed, and its propertieshave been described(13).As we have shown, anti-PN-I IgG will immunoprecipitate metabolically labeled PN-I fromconditionedmedium and fromcelllysates. Its utility in probing the biosynthesisof PN-I would depend on its cross-reactivity with other proteins, especially in the region where PN-I runs on SDS-PAGE. Therefore, to analyze the specificity of anti-PN-I, conditioned medium from cells labeled with [35S]methionine was immunoprecipitated with 30 pg of IgG in the presenceof purified, unlabeled PN-I. Densitometerscanning of theresultingautoradiograph(Fig. 1) shows that unlabeled PN-I successfully competed for antibody binding in the region of t h e gel associated with PN-I, which runs near the 43-kDa marker. This verifies the utility of antiPN-I antibody in the investigation of PN-I biosynthesis and regulation: anti-PN-I is the critical tool that makes these studies practical. Purified PN-I, as prepared by Scott et al. (l),has been sequenced at the amino terminus and is pure by this criterion. In order to verifythat the material purified in our laboratory was indeed the same as that above, we electroeluted PN-I fromgelslices of purified PN-I run on SDS-PAGE. This material was then sequenced to 21 amino acids (Table I)and was foundto be identical to the sequence previously published elsewhere. This analysis served two purposes. First, it verified that the material identified as PN-I was the same as published by Scott et al.(1). Second, it showed that the protein was homogeneous and not a group of co-migrating proteins. Since this material completely competed for the binding of metabolically labeled PN-I to anti-PN-I, it is apparent that the antibody is highly specific for PN-I and is a valid probe of PN-I biosynthesis. Biosynthesis of PN-I-Anti-PN-I was next used to probe the biosynthesisof PN-I i n HF cell cultures.It has been noted that the concentration of PN-I rapidly increases in HF cultures whose conditioned medium has recently been replaced with fresh medium.' One possibility explaining this phenomenon would be the release of intracellular pools of PN-I. T o test this possibility, pulse-chase labeling studies using [35S] methionine were performed. Cells were pulsed with 100 pCi/ ml [35S]methionine for 10 min, and then chased with fresh mediumwithnormalmethionineplus 1 mg/mlBSA.At specifiedtimepoints, PN-I wasimmunoprecipitatedfrom both the medium and the cell lysates. Analysis by SDS-PAGE reveals several cross-reactive bands in nonreduced gels (Fig. 2). Reduction with 2-mercaptoethanol eliminates most material at the top of the running gel but causes the doublet to migrate very closely, creating problems in photographic reproduction. The contaminants found at 68 and 30 kDa are not known and do not appear in conditioned medium immunoprecipitated after longer periods of labeling. The analysis
PN-I Biosynthesis
14186 " " "
.oo 1
B
.o 1
.1
1
10
PN-I (rg/ml)
FIG.1. Competition of metabolically labeled PN-I binding to anti-PN-I by unlabeled PN-I. Conditioned medium from H F cells incubated for 24 h in the presenceof 50 pCi/ml ["S]methionine was incubated with 60 pg/ml anti-PN-I IgG plus a given concentration of purified, unlabeled PN-I. Immunoprecipitated protein was electrophoresedinnon-reduced SDS gels, andautoradiographs were scanned by densitometer, followed by integration of peaks.
TABLE I Sequence of amino terminus X-His-Phe-Asn-Pro-Leu-Ser-Leu-Glu-Glu10
5
Leu-Gly-Ser-Asn-Thr-Gly-Ile-Gln-Val-Phe-Asn 15 20
Cells Mrx10-3
"
68-
"
-
-
0
. .
-
- .
43-
30-
Medium . .
"
-
-
10 20 30 40 60
0
10 20 30 40
60
Time of Chase ( m i d FIG.2. Biosynthesis of PN-I. Immunoprecipitatedproteins from pulse-labeled cell lysatesand media were subjected to nonreduced SDS-PAGE, followed by fluorography. Length of chase period is indicated below the lunes. Cells were pulsed for 10 min with 100 pCi/ml ['"S]methionine, then chasedwith fresh media containing normal methionine and 1 mg/ml BSA.
of intracellular PN-I reveals that a lower M,precursor exists. After 45 min, the precursor matures into a secreted form which migrates roughly 2 kDa larger, as seen in the panel designated "cells" in Fig. 2. This mature form,which appears as a doublet, is then completely secreted into themedium 10 min later. As is typical for constitutively secreted proteins, no intracellular pool of PN-I was detected after 60 min of chase. Therefore, it is probable that the rapid increase of PNI levels in cultures is due tonewly synthesized protein. The biosynthetic steps thatoccur during PN-I maturation probably follow the pathway of processing of other secretory andmembraneproteins. Typically, theseproteinsare cotranslationallyinsertedinto rough endoplasmic reticulum, where the initial glycosylations are also added to selected asparagine moieties (21). The 45-min period that PN-I precursor can be observed is comparable to other proteins that are known to transit the RER. The transition to mature PNI would be expected tooccur inthe Golgi, where modifications to carbohydrate groups lead to altered mobility on SDS gels compared to the RER form. The relatively short period of time that PN-I remains in this Golgi form is also typical of secreted proteins that are constitutively produced by eucaryotic cells. Also notable is the fact that intracellular PN-I appears asa doublet on SDSgels. This suggests that PN-I in conditioned medium exists as multiple isoforms that are a result of differential processing or different mRNA species and nota consequence of purification artifacts. Post-translational Modifications of PN-I-Most post-translational modification of secreted proteins involve carbohydratestructures,primarilythose linked toasparagine. As proteins move from the RER to the Golgi, they normally lose the high mannose glycosylations added via dolichol intermediates in the RER and acquire complex structures (22). PNI, as found in conditioned medium, consists of 6% carbohydrate. Of this, roughly half is sialic acid, which is added to proteins as they transit the Golgi (23). A reliable probe of the
PN-I Biosynthesis
14187
nature of the N-linked glycosylations on a protein is endo H, mediumwas immunoprecipitated with anti-PN-I. PN-I lawhich is able to cleave only high mannose, N-linked carbobeled with['H]mannose appearedto be identicaltothat hydrate structures but not complex-type structures. Immulabeled with [:"S]methionine, as seen inFig. 4. This provided noprecipitated PN-I fromcell lysates pulsed with [""Slmethi- independent proof that PN-I is a glycoprotein and that its onine were treated with endo H, as described above. Acid- appearance on polyacrylamide gels is not dependent on the precipitated protein was then analyzed by SDS-PAGE and labeling procedure. fluorography, as seen in Fig. 3. The precursor form of PN-I PN-I Secretion by Cultured Fibroblasts-PN-I represents was sensitive to endo H as expected of proteins associated roughly 5% of the total cell-secreted protein in H F cultures. with RER. The intracellular form thought to be associated Indirect observations have indicated that the concentration with the Golgi, as seen in the 30-min time point(Fig. 3), was of PN-I in conditioned medium reaches a plateau within 2 not sensitive to endo H. In addition, endo H was unable to days. When [''"I]thrombin was incubated for 1 h with the alter the migration of secreted PN-I. Mature PN-I, as seen medium from plates conditionedfor progressively longer time after 30 min of chase and in secreted form, contains complex periods, it wasfound that the amount of PN-Iathrombin carbohydrate structures resistant to endo H.However, it was complexformedreachedamaximum level after 48 h of sensitive to digestion by endo F, whichcleaves both high conditioning.It seemedpossible that anegative feedback mannoseand complexglycosylations. Whencultures were inhibition could be operating and that the concentration of incubated with tunicamycin, an inhibitor of dolichol-based PN-I in the medium directly affected PN-I synthesis. This glycosylationsin RER (24), a 40% decrease in PN-I levels form of regulation would require several levels of control, was noted (data not shown). This, along with theendoglyco- including secondary messengers and translational or transidase data, suggests that the primary modifications made toscriptional inhibitors. It was also possible that PN-I was not PN-I during its synthesis involve asparagine-linked carbobeing regulated a t all but that cell-secreted proteases were hydrate structures. accumulating in the conditioned media and competing with While the metaboliclabeling of cell-secreted proteins with ['251]thrombin for PN-I binding. H F cells produce a plasmin["'S]methionine has proven to be quite reliable and accurate in many systems, it was thought that labeling with another M, X I O - 3 radioactive tracer would confirm the above results obtained P , . " with [:"S]methionine. Therefore, H F cellswere incubated overnight with 100 /zCi/ml ["Hlmannose, and the conditioned Endo H
Endo F ,
.
. ..
94-
68-
4 34341-
30-
20-
- +
-
+ - +
- +
15 30 60 60 FIG. 3. Endoglycosidase digestion of PN-I. PN-I immunopreA B cipitated from cell lysates and media of [:'sS]methionine-labeled H F cells wastreated with endo H andendo F, as described under FIG. 4. [3H]Mannose-labeled PN-I. HF cells were incubated "Experimental Procedures." Cells were pulsed for15 min, then chased overnight with 100 pCi/ml ["Hlmannose, as described under "Experfor the indicated time periods. Intracellular PN-I (15- and 30-min imental Procedures." PN-I was immunoprecipitated from the conditime points) and secreted PN-I (60-min time point) are shown. tioned medium with anti-PN-I and nonspecific IgG.
PN-I Biosynthesis
14188
ogen activator that is secreted as a zymogen which, upon activation, binds PN-I (24). A simpler means of regulating the concentration of PN-I in the cellular surroundings would involve constant secretion accompanied by receptor-mediated uptake, which would ultimately result in a steady state level of PN-I. Such a mechanism would be dependent on receptor numbers, which may or may not be regulated, as well as the volume of medium surrounding the cells. To determine which, if any, form of regulation exists in PN-I biosynthesis, we assayed the levels of metabolically labeled PN-I in the media of cells over time. Cultures of HF cells were incubated in serum-free medium containing BSA and normal amounts of unlabeled methionine, plus 50 pCi/ ml [35S]methionine, for up to 72h. Isolated media were immunoprecipitated twice with anti-PN-I, and the proteins were subjected to gel electrophoresis. To quantitate PN-I levels, autoradiographs of the gels werescanned by densitometer, immunoprecipitates were counted, and gel slices of the PN-I region were digested and counted. All methods yielded similar results. Fig. 5 shows the amount of PN-I in conditioned medium reaching a constant level by 48 h, while total protein secretion remains linear. Since a large percentage of the secreted PN-I was probably unlabeled due to the conditions of the experiment, the media were immunoprecipitated twice to insure that competition of the labeled PN-I by unlabeled protein did not lead to misleading results. The second precipitates mirrored the first. It was next of interest to determine therate of PN-I secretion under different levels of conditioning. If some form of inhibition of PN-I synthesis were operating, it would be expected that a change in the biosynthetic rate occurs over time. Such a change could be measured by determining the amount of PN-I synthesized over a short timeperiod without changing the conditions that may cause the inhibition, namely, the constituents of the medium. The limitations of such a study include the competition by unlabeled PN-I of antibody binding to labeled PN-I. This can be eliminated by using a labeling period of 30 min and immunoprecipitating the PN-I from cell lysates. No labeled PN-I is secreted into the medium, since roughly 60 min is required for secretion. Since PN-I does not accumulate intracellularly, the immunoprecipitation conditions at different times would besimilar. Therefore, cultures were incubated with serum-free medium for 0-72 h and were pulsed with 100 pCi/ml [35S]methionine
V
for 30 min. Cells were washed with cold PBS and then lysed with buffer containing CHAPS detergent. The PN-I excised from SDS gels was counted and compared to the total acidprecipitable counts found in the lysates. No significant differences in synthetic rates were found between fresh plates and those conditioned for 72 h (Fig. 6). To attain a steady state level of PN-I in themedium, the synthetic ratewould have to equal the clearance rate. Without a rapid uptake of PN-I, small changes in the synthetic rate would not yield the steady state levels of PN-I observed but would result in a gradual accumulation of the protein. PN-I Uptake by Cells-The above data suggest that, in order to reach a steady state level of PN-I in conditioned medium, cells must remove PN-I in a rapid fashion. Previous models of PN-I interactions with the cell surface involved the formation of PN-1.protease complexes as a prerequisite for receptor binding. It was thought that free PN-I could not bind the receptor. If this were true and if the steady state level of PN-I were due to its uptake, then some protease must be secreted into the mediumwhich bindsPN-I, allowing its removal by its receptor. The removal of metabolically labeled PN-I from conditioned medium was studied in a number of ways. In one such experiment, cells were incubated for 24 h in serum-free medium plus [35S]methionineand then given a bolus of unlabeled methionine as a chase. The PN-I in the media was immunoprecipitated twice with anti-PN-I, and the percentage of labeled PN-I in the media was analyzed over time. Since some [35S]methionineremains in the medium, labeled proteins are continually made, including PN-I. Regardless of this complication, labeled PN-I disappears from the media of cultures rapidly, as seen in Fig. 7. The rate of this removal gradually slows as unlabeled PN-I is synthesized by cells. This newly made PN-I competes for the binding of labeled PN-I to thecells. Conditioned medium incubated for 24 h in theabsence of cells resulted in no loss of PN-I andno formation of higher M,complexes. In addition, PN-I does not accumulate at cell surfaces, as judged by lysing cells after the removal of medium, followed by immunoprecipitation with anti-PN-I. This suggests that PN-I is endocytosed by fibroblastsandthen degraded. Earlier studies involving PN-1. protease complexes revealed that, upon internalization by cells, these complexes were degraded by lysosomes (10). Such is probably the case with the metabolically labeled PN-I in this study. This removal does not appear to involve complexes with proteases, since no evidence of complex formation is
I!
1
!
48
24
72
Time (hrs)
FIG. 5. PN-I levels in fibroblast cultures. Fibroblasts were labeled with [3SS]methioninefor 8-72 h in serum-free medium, as described under “Experimental Procedures.” PN-I was immunoprecipitated from the conditioned media, excised from the resulting SDS gels, and scintillation counted. The inset graph indicates the total acid-precipitable protein in the medium over time.
0
24
1h
48
72
Tim (trs) FIG. 6. PN-I secretion rate over time. Cells were incubated with serum-free medium for given time periods, then labeled for 30 min with 100 pCi/ml [36S]methionine.Cell lysates were immunoprecipitated with anti-PN-I, and subjected to SDS-PAGE. PN-I was excised from gels, counted, and compared to totallabeled protein.
PN-I Biosynthesis
14189
FIG.7. Uptake of PN-I by fibroblasts. HF cultures were incubated for 24 h in the presence of serum-free medium containing 100 pCi/ml ["S]methionine and 1 mg/ml ovalbumin. At t = 0, unlabeled methionine was added to the cells, andthe mediawere isolated a t given times. PN-I was excised from SDS gels of immunoprecipitated media, and the percentage of PN-I ucrsus total labeled proteinwasplotted. The arrow indicates the amounts of PN-I found in medium incubated for 24 h in the absence of cells. Immunoprecipitated PNI a t 0, 8, and 24 h is represented in the autoradiograph.
A*
8 T i m e 24
0
B
present. Electrophoretic analysis of the PN-I in themedium reveals none of the high M , bands expected when PN-I is complexed with proteases (Fig. 7). It ispossible that free PNI is removed from media by a separate mechanism than that which removes PN-1.protease complexes. It was noted that the uptake of PN-I by cells was not inhibited by 200 pg/ml heparin. Such a concentration of heparin is known to completely eliminate the bindingof complexes to cells (13).However, the uptakewas completely inhibited by anti-PN-I, suggesting that the event is receptor-mediated. possible It is that a separatereceptorispresentonfibroblastsurfacesthat recognizes free PN-I, or one receptor that binds both free and complexed PN-I.
8
16
24
Time (hrs)
migrate identically on SDS gels to PN-I. thrombin, yetwill not interact with anti-PN-I. Sequence analysis will ultimately determine the relationshipbetween the various inhibitors but may not suggest a physiological purpose. Insight into the roles of these inhibitors in uiuo will be found in their mechanism of regulation. By studying the rate of PN-I secretion, its method of removal, andits response toenvironmental changes, one may learn the capabilities of the PN-I-based system of protease inhibition. We have investigated thissystem by using ahighly specific antibody developed against humanfibroblastPN-I in conjunctionwith metabolically labeled PN-I. PN-I is secreted in large quantities by H F cells; roughly 5% of total cell-secreted protein can be PN-I, depending upon DISCUSSION culture conditions. The intracellular events leading to PN-I The physiological roles of the various nonvascular protease secretion appearto be typical of other secreted and membrane inhibitors, such as the protease nexins, are not completely proteins (26, 27). Initially, a lower M , precursor, probably understood. Part of the answer lies in the specificites of the associated with RER, is seen in pulse-chase analysis. This different inhibitors toward target proteases. PN-I has been form of PN-I containshigh mannose glycosylations, as judged shown to inhibit many serine proteases but a high has binding by endo H sensitivity. Maturation occurs in the Golgi, where affinity towards thrombin and urokinase, although this affinthe carbohydrates assume acomplex configuration that is ity is modulated by heparin (1).However, PN-I has an even insensitive to endo H(22). Subsequently, PN-I is secreted higher affinity for trypsin, which is unlikely to be encountered into the medium as a 43-kDa glycoprotein; no pulse-labeled physiologically. The resolution of the preferred target proPN-I remains in intracellular pools. During a wound response, teases of many inhibitors is far from complete, even in the whereactivatedprotease from plasma become exposed to case of the plasma protease inhibitors. Thecells that secrete nonvasculartissue,onlythePN-Ialreadypresent in the PN-I-like proteins are abundant. Several of these inhibitors extracellular matrix would be available immediately. Addiare recognized by anti-PN-I, such as those from kidney and tional PN-I would bederivedfrom PN-I biosynthesis, not glioma cells. Others share common properties, yet will not released from stored protein. T o insure an adequate concencross-react with the antibody. Inhibitors from several endo- tration exists, cells appear to secrete large quantities of PNthelial cell lines will form complexes withthrombinthat I.
PN-I Biosynthesis
14190
The regulation of PN-I levels in HF cultures appears to involve few elements. Secretion of PN-I over time is constant, regardless of the amountof PN-I already in the medium. The removal of PN-I also appears to occur continuously. When the level of PN-I in the medium reaches a concentration at which both secretion and uptake rates are equal, a steady state occurs. The removal of PN-I by cells is probably receptor-mediated, since it is rapid, and is inhibited by anti-PN-I IgG. However, the receptor involved in the removal of PN-I may not be the same as that which binds PN-I-protease complexes. Heparin will compete for the cellular binding of complexes, possibly through its interaction with the heparinbinding domain of PN-I (13). Heparin does not, however, inhibit the removal of free PN-I from medium over time. We have recently labeled PN-I withiodinated Bolton-Hunter reagent; this 1251-labeledPN-I retains itsability to bind proteases. Preliminaryexperimentsindicate that PN-I binds specifically to the cell surface in a noncomplexed form.3 It is possible that two receptors exist on fibroblast surfaces, one that binds free PN-Iand one that binds complexes. An alternative would be a single receptor system, able to bind both free and complexed PN-I. Recent data indicate that this alternative model is probably accurate; free PN-I competes for the binding of complexed PN-I to cell surfaces. Because the PN-I level i n vivo would affect the activity of several regulatory serine proteases, its regulation is probably important. The high level of PN-I in conditioned media suggest that fibroblasts, and probably other cell types that secrete similar inhibitors, place a premium on limiting proteolysis. By limiting the activity of thrombin, PN-I could serve to limit the extentof the initialclot formation. Inhibition of plasminogen activators and of plasmin by PN-I would limit the extent of tissue damage by plasmin, as well as the rate of clot dissolution. PN-I appears as a doublet upon SDS-gel electrophoresis. The nature of this phenomenon is not understood, nor is the physiological role of multiple PN-I species clear. It has been reported that two isoforms of ATIII exist and that carbohydrate side chain differences result in the altered mobility of one form.Evidence that thetwo forms of ATIII bind heparin with different affinitieshas been presented (28). Such may be the case with PN-I. Elution of PN-1 from heparin-Sepharose with a salt gradient yields two peaks of activity (1). It is possible that the two forms of PN-I bind proteases, heparin, or receptors with different affinities. Acknowledgments-Wewish
tothank
Wilson Burgess, Revlon
Biotechnology Group, for the valuable sequence data obtained in minimum time, and Jeffrey Smith for his valued suggestions. REFERENCES 1. Scott, R. W., Bergman, B. L., Bajpai, A., Hersh, R. T., Rodriguez, H., Jones, B. N., Barreda, C., Watts, S., and Baker, J. B. (1985) J. Biol. Chem. 260,7029-7034 2. Christman, J. K., Silverstein, S. C., and Acs, G. (1977) in Proteinases in Mammalian Cells and Tissues (Barrett, A. J., ed) pp. 91-149, North-Holland Publishing Co., Amsterdam 3. Levin, E. G., and Loskutoff, D. L. (1982) J. Cell Bwl. 94, 631636 4. Loskutoff, D. L., and Edgington, T. S. (1981) J.Bid. Chem. 2 5 6 , 4142-4145 5. Astedt. B.. Lecander. I.. Brodin. T.. Lundblad., A,., and Low. K. (1985) Thromb. Huemostasis 5 3 , 122-125 6. Savion, N., Isaacs, J. D., Gospodarowicz, D., and Shuman, M. A. (1981) J. Biol. Chem. 2 5 6 , 4514-4519 7. Vassalli, J., Dayer, J., Wohlwend, A., and Belin, D. (1984) J. Exp. Med. 159,1653-1668 8. Monard, D. (1985) Cell Biol. Int. Rep. 9, 297-304 9. Baker, J. B., Low, D. A,, Simmer, R. L., and Cunningham, D. D. (1980) Cell 2 1 , 37-45 10. Knauer, D. J., Thompson, J. A., and Cunningham, D. D. (1983) J. Cell Physwl. 1 1 7 , 385-396 11. Eaton, D. L., Scott, R. W., and Baker, J. B. (1984) J. Bid. Chem. 259,6241-6247 12. Scott, R. W., and Baker, J. B. (1983) J. Biol. Chem. 258,1043910444 13. Howard, E. W., and Knauer, D. J. (1986) J . Biol. Chem. 2 6 1 , 684-689 14. Bartl, K., and Lill, H. (1980) Thromb. Res. 18,267-272 15. Hunkapillar, M. W., Lujon, E., Ostrander, F., and Hood, L. E. (1983) Methods Enzymol. 9 1 , 233-242 16. Luhas, T. J., Iversen, D., Schleicher, M., and Watterson, D. M. (1984) Plant Physiol. 7 5 , 788-795 17. Zilberstein, A., Snider, M. D., Porter, M., and Lodish, H.F. (1980) Cell 21,417-427 18. Bensadoun, A., and Weinstein, D. (1976) Anal. Biochem. 70, 241-250 19. Elder, J. H., and Alexander, S. (1982) Proc. Natl. Acad. Sci. U. S. A. 79,4540-4544 20. Laemmli, U. K. (1970) Nature 227,680-685 21. Lennarz, W. (1983) Methods Enzymol. 9 8 , 91-97 22. Kornfeld, R., and Kornfeld, S. (1985) Annu. Reu. Bwchem. 5 4 , 631-664 23. Sabatini, D. D., Kreibach, G., Morimoto, T., and Adesnik, M. (1982) J. Cell Biol. 92, 1-22 24. Schwartz, R. T., and Datema, R. (1982) Methods Enzymol. 83, 432-443 25. Deleted in proof 26. Schwartz, A. L., and Rup, D. (1983) J. Biol. Chem. 2 5 8 , 1124911255 27. Omary, M. B., and Trowbridge, I. S. (1981) J. Biol. Chem. 2 5 6 , 12888-12892 28. Peterson, C. B., and Blackburn, M. N. (1985) J. Biol. Chem. 2 6 0 , 610-615 ~
I~
