Proteolytic activation of single-chain precursor macrophage-stimulating protein by nerve growth factor-gamma and epidermal growth factor-binding protein, members of the kallikrein family.

Promacrophage-stimulating protein (MSP) is an 80-kDa protein that acquires biological activity after cleavage at an Arg-Val bond to a disulfide-linked alpha beta heterodimer by serine proteases of the intrinsic coagulation cascade. These proteases, which include serum kallikrein, factor XIIa and factor XIa, are members of the trypsin family of serine proteases. We now report that two other members of the family, nerve growth factor-gamma (NGF-gamma) and epidermal growth factor-binding protein (EGF-BP), cleave and activate pro-MSP to the disulfide-linked alpha beta heterodimer. Cleavage of 1.5 nM pro-MSP by 1 nM NGF-gamma or EGF-BP at 37 degrees C was almost complete within 30 min. These concentrations of enzyme are about 2 orders of magnitude less than is required for cleavage by serum kallikrein or factor XIIa. Cleavage of pro-MSP to MSP was associated with a conformational change in the protein, because the cleaved product, but not pro-MSP, was detected by a sandwich enzyme-linked immunoassay. Cleavage caused the appearance of biological activity, as measured by chemotactic activity of MSP for resident peritoneal macrophages, by MSP-induced macrophage shape change, and by stimulation of macrophage ingestion of C3bi-coated erythrocytes. These findings suggest the possibility of cooperative interactions between NGF-gamma or EGF-BP and pro-MSP in inflammation and wound healing.

Macrophage-stimulating protein (MSP)' was first described as a serum protein that stimulated chemotactic responsiveness of murine resident peritoneal macrophages to endotoxin-activated mouse serum (1,2). Pure MSP is also a chemoattractant for resident peritoneal macrophages,' induces macrophage shape change, and causes ingestion of C3bi-coated erythrocytes (3). MSP was recently found to inhibit the induction by lipopo-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
A. Skeel and E. J. Leonard, unpublished observations. lysaccharide or inflammatory cytokines of nitric oxide synthase mRNA in murine resident macrophage^.^ We recently purified MSP from human plasma (3) and cloned from the Hep G2 liver tumor cell line a MSP cDNA that predicts a protein of 711 amino acids (4). The cloning of MSP was also reported by Shimamoto et al. (5). MSP is a glycoprotein belonging to the family of kringle proteins that includes hepatocyte growth factor/Scatter factor (6,7) and plasminogen (8). Proteins in this family are secreted as single-chain precursors, which acquire biological activity when they are cleaved, usually at a single arginyl bond, by specific serine proteases. Biologically active MSP comprises a disulfide-linked 53-kDa a-chain and a 25-kDa P-chain. By using MSP cDNA-transfected CHO cells, we obtained the single-chain form of pro-MSP and found that enzymes of the intrinsic coagulation cascade including serum kallikrein, factor XIa, and factor XIIa could cleave biologically inactive pro-MSP to yield a disulfidelinked +chain heterodimer, the biological activity of which was quantitatively comparable with that of natural MSP isolated from human plasma (9). These findings indicate that the biological effect of MSP on target cells is regulated by specific serine proteases responsible for conversion of pro-MSP into the heterodimeric form. Because MSP is produced by liver cells, circulates in blood, and acts on resident macrophages, we postulated that MSP might be involved in regulation of macrophage functions during inflammation (3). In view of the capacity of serum kallikrein to cleave and activate pro-MSP, we became interested in mouse NGF-y and EGF-BP, which are members of the glandular kallikrein subfamily of serine proteases (10, 11). Alignment of the amino acid sequence of NGF-7 and EGF-BP with the sequence of mouse serum kallikrein (12) shows sequence similarity to serum kallikrein of 56 and 51%, respectively; there are considerable stretches of sequence identity for the three proteins in the two regions that incorporate the His-Asp-Ser catalytic triad. Both NGF-y and EGF-BP are found in high concentrations in mouse submaxillary glands and saliva (10). NGF-y is part of a noncovalent complex with two other proteins, NGF-a and NGF-P, and is thought t o proteolytically activate pro-NGF. Likewise, EGF-BP activates pro-EGF (11).
In this paper, we show that nanomolar concentrations of NGF-y and EGF-BP each cleave pro-MSP to the biologically active disulfide-linked heterodimer.
The effective concentrations are about 2 orders of magnitude lower than that observed for proteases of the intrinsic coagulation system. This points to a possible role for NGF-y and EGF-BP as pro-MSP convertases in vivo.

EXPERIMENTAL PROCEDURES
Reagents-Human plasma MSP was purified as previously described (3). NGF-y and EGF-BP were prepared according to the method of Darling and Shooter (13). The concentration of NGF-y and EGF-BP were active-site-titrated with PNPGB by the method of Chase and Shaw (14). Cells-Peritoneal resident macrophages were obtained from C3W HeN mice by lavage of the peritoneal cavity with RPMI 1640 medium as previously described (3). CHO cells were obtained from American Type Culture Collection (Rockville, MD) and maintained in this laboratory. All cell cultures were maintained a t 37 "C in a humidified incubator containing 5% CO, in air.
Antibodies to Pro-MSP, MSE and MSP Synthetic Peptide-Rabbit antibodies against MSP were as described (15). Polyclonal rabbit IgG anti-MSP reacted with MSP or pro-MSP. Monoclonal antibody 2S, which recognizes MSP but not pro-MSP, was used in sandwich ELISA to study the conformational change in pro-MSP after cleavage.
Expression of Pro-MSP in CHO Cells-Ligation of a human MSP cDNA with pcDNA-3, transfection of CHO cells, and cloning of the MSP-producing cell CHO-MSP18 were as described (9). Pro-MSP was prepared from cultures of CHO-MSP18 cells in serum and cysteine-free Dulbecco's modified Eagle's medium with 100 pCi/ml of ["Slcysteine for 56 h at 37 "C. Culture fluids were harvested and used as a source of pro-MSP. The concentration of pro-MSP was measured with a sandwich ELISA after treatment with serum kallikrein (9).
Cleavage of Pro-MSP in Vitro-Culture fluid 50-1.11 aliquots, with a concentration of 150 ng/ml of [3sSlcysteine-labeled MSP, were incubated for 30 min at 37 "C in a water bath with NGF-y, EGF-BP, or serum kallikrein in 50 nM Tris buffer, pH 7.6, containing 0.1 M NaCl. To test the effect of C1-INH on cleavage of pro-MSP by serum kallikrein or NGF-y, C1-INH a t different concentrations was mixed with kallikrein or NGF-y a t 37 "C for 20 min, after which pro-MSP was added for another 30 min of incubation. After treatment, the radiolabeled products were immunoprecipitated, boiled in sample buffer containing 2-mercaptoethanol, run in SDS-polyacrylamide gel electrophoresis, and autoradiographed as previously described (9).
Assay for Macrophage Shape Change-The assay was performed as previously described (9). Murine resident macrophages in serum-free RPMI 1640 medium a t a concentration of 5 x 105/ml were added to wells of a 48-well tissue culture plate. Samples to be assayed were serialdiluted in RPMI 1640 medium and added to the wells. Purified human plasma MSP was the positive control. After incubation at 37 "C for different intervals, supernatants were carefully removed, and cells were fixed with 3% formaldehyde for 10 min at room temperature. The cells were stained with Diff-Quik and counted a t a magnification of 250x in two randomly selected areas. Macrophage shape was either round or extended; results are expressed as the percentage of macrophages with extended morphology.
Macrophage Chemotaxis Assay-The assay was conducted as previously described (9). Briefly, the bottom wells of a multi-well chemotaxis chamber were filled with purified MSP or samples to be tested and then covered with a 10-pm thick polycarbonate membrane with 5-pm holes. Upper wells were filled with 50 pl of macrophage suspensions. During the incubation period of 3 h a t 37 "C, macrophages migrated through the holes and remained attached to the attractant side of the membrane. After air drying and staining with Diff-Quik, migrated cells were counted with an image analyzer. MSP or samples were assayed in duplicate. The results were expressed as the percentage of input macrophages that migrated.
Assay for Macrophage Phagocytosis of EC3bi-The assay was performed as described previously (3) with slight modifications. Briefly, resident macrophages a t a concentration of 8 x 105/ml in Dulbecco's modified Eagle's medium were cultured in a 24-well culture plate for 45 min a t 37 "C. After the wells were washed twice to remove nonadherent containing 0.1 M NaCl were incubated with 10 nxl NGF-y, 10 nsl EGF-BP, 100 ny serum kallikrein, or without enzyme. After 30 min at 37 "C, pro-MSP and MSP heterodimer were precipitated with rabbit anti-MSP of cleavage by PPACK, NGF-y or EGF-BP was pretreated with 20 pxr IgG as described under"Experimenta1 Procedures." To study prevention PPACK for 10 min before pro-MSP was added. IgG for immunoprecipitation was rabbit anti-MSP, except for normal IgG in lune 1. Lanes 1 and 2, no enzyme; lanes 3 and 4, serum kallikrein and serum kallikrein plus PPACK; lanes 5 and 6 , EGF-BP and EGF-BP plus PPACK; lunes 7 and 8, NGF-y and NGF-y plus PPACK. cells, 0.5 ml of EC3bi were added to each well. The ratio of erythrocytes to macrophages was 50:l. After incubation for 30 min to allow erythrocyte binding to macrophages, MSP or tested samples were added to wells. After incubation for 15 min, nonadherent erythrocytes were washed away, and 0.5 ml of ammonium chloride lysis buffer was added. The buffer was removed after 2.5 min, and the monolayer was stained with Diff-Quik. The percentage of macrophages with ingested erythrocytes was determined by examination of the monolayer with a 100 x oil immersion objective.
Sandwich ELISA for MSP-The ELISA was previously described (15). It used monoclonal antibody anti-MSP 2s as capture antibody and polyclonal rabbit IgG anti-MSP as detection antibody. Although the polyclonal rabbit anti-MSP detects pro-MSP by immunoprecipitation, the sandwich ELISA does not detect pro-MSP, apparently because the epitope with which monoclonal antibody anti-MSP 2s reacts is not accessible.

RESULTS
Cleavage of Pro-MSP by NGFy and EGF-BP-We previously reported that pro-MSP could be cleaved by three enzymes that initiate the intrinsic coagulation cascade, serum kallikrein, factor XIIa, and factor XIa (9). Because NGF-y and EGF-BP belong to the family of glandular kallikreins (lo), we tested the capacity of these proteases to cleave pro-MSP. Fig. 1 shows that NGF-y and EGF-BP completely cleaved the single-chain form of pro-MSP into the a-and 0-chains that are characteristic of mature MSP. Cleavage by 100 nM serum kallikrein is shown for comparison. PPACK, an irreversible inhibitor of thrombin, blocked the effects of all three enzymes on cleavage of pro-MSP. Concentration-dependent cleavage of 1.5 nM pro-MSP by NGF-y and EGF-BP is shown in Fig. 2. Complete cleavage of pro-MSP within 30 min occurred with 1 nM NGF-y or EGF-BP. Comparison of results with enzyme concentrations of 0.1 nlr shows that NGF-y was slightly more efficient than EGF-BP. In contrast to these glandular kallikreins, we previously showed that 100 nM serum kallikrein was required for complete cleavage of pro-MSP (9).
Time Course of Conversion of Pro-MSP by NGF-?Pro-MSP a t a concentration of 1.5 nM was incubated with 2 nxf NGF-y at 37 "C for different time intervals. Fig. 3 shows about 12% cleavage of pro-MSP as early as 1 min, 60% a t 5 min, and 100% at 60 min. A trace of a 48-kDa digested band was seen at 60 min. Trace amounts of this band could also be seen after digestion with high concentrations of serum kallikrein or EGF-BP. The band apparently represents the a-chain after cleavage of a short peptide, possibly near the C terminus at residue 471.
Cleavage-dependent Pro-MSP Conformational Change-For quantifying serum MSP concentrations, we previously developed a sandwich ELISAthat used a mouse monoclonal anti-MSP capture antibody and a rabbit polyclonal anti-MSP detection antibody (4). As shown in Table I, pro-MSP was not detected by this ELISA, whereas the protein was detected after proteolytic conversion to the mature ap heterodimer by 50 nM serum kallikrein. Proteolytic conversion of the same concentration of pro-MSP by NGFy or EGF-BP resulted in ELISAvalues comparable with that of kallikrein-digested pro-MSP. Partial digestion of pro-MSP with 10 nar serum kallikrein (resulting in a mixture of pro-MSP and MSP as monitored by immunoprecipitation) was reflected in lower ELISA values. These results indicate that cleavage of pro-MSP by each of these three enzymes was associated with a conformational change that exposed an epitope recognized by the monoclonal capture antibody.

Effect of Cl-INH on NGF-y-mediated
Cleavage of Pro-MSP-It was reported that NGF can substitute for the activated first component of complement (Ci), in that it can proteolytically activate C4 and C2; like Ci, it is inhibited by C1 inhibitor, a protein that blocks enzymatic activity of C1 and serum kallikrein (16). We determined if pro-MSP cleavage activity of NGF-y could be regulated by C1-INH. Fig. 4 shows that C1-INH inhibited the capacity of NGF-y to cleave pro-MSP. However, 1 unit/ml C1-INH only partially counteracted 1 nu of NGF-y, and no inhibition was seen at C1-INH concentrations below 0.2 unitdm1 (lane 6). In contrast, 0.02 unitlml C1-INH totally blocked the enzymatic activity of 50 nM serum kallikrein (lane 4 ) .
Cleavage of Pro-MSP by NGF-y or EGF-BP Produces Biologically Active MSP-To test whether NGF-y and EGF-BP cleavage generates biologically active MSP, samples were tested for induction of macrophage shape change and phagocytosis of C3bi-coated erythrocytes. As shown in Table 11, pro-MSP from CHO-MSP18 cell cultures was inactive in both assays. After incubation of pro-MSP with NGF-y or EGF-BP, the product induced macrophage shape change and stimulated phagocyto- sis of EC3bi. Biological activity was comparable with that of plasma MSP and to pro-MSP cleaved by serum kallikrein. Fig.  5 shows that pro-MSP acquired chemotactic activity for macrophages after incubation with either NGF-y or EGF-BP. The activity was comparable with that of MSP purified from human plasma. DISCUSSION We previously reported that pro-MSP is cleaved by enzymes of the intrinsic coagulation cascade, including serum kal-

TARLE I1
Effects of pro-MSP on resident macrophage shape change and phagocytosis of EC3bi after treatment with NGF-y, EGF-BP, or serum kallikrein Pro-MSP (1 nsl) was incubated for 30 min at 37 "C with NGF-y (5 nM), EGF-BP (5 nk!), serum kallikrein (50 nsl), or no enzyme. After incubation, samples were tested for capacity to induce macrophage shape change or ingestion of ECSbi, as described under "Materials and Methods." Purified plasma MSP was the positive control. Enzymes at the concentration used for treatment of pro-MSP were also included in the assays. The results for the shape change assay are expressed as the percentage of macrophages with extended morphology. Phagocytosis assay results are expressed as the percentage of macrophages with a t least one ingested erythrocyte. likrein, factor XIa, and factor XIIa (9). The data in this paper show that pro-MSP can also be cleaved and activated by NGF-y and EGF-BP. These five enzymes are all members of the trypsin family of serine proteases, which have been grouped on the basis of highly conserved sequences in the region of the catalytic site histidine and serine residues (17). Although the configuration of these enzymes confers a high degree of substrate specificity, their specificity is not absolute, as illustrated by their capacity to cleave and activate pro-MSP. In Table 111, we have listed members of the trypsin serine protease family that we have tested for proteolysis of pro-MSP. The minimal concentrations of NGF-y or EGF-BP that cause complete conversion of 1.5 nM of pro-MSP are at least an order of magnitude less than that of the listed coagulation enzymes. Also of interest are the enzymes in this family that did not cleave and activate pro-MSP. Urokinase, which activates plasminogen, is also reported to activate prohepatocyte growth factor/Scatter factor (18) but fails to cleave pro-MSP, which has 4 5 8 sequence similarity to prohepatocyte growth factor/Scatter factor and comparable sequences around the scissile bond. Among the three listed enzymes that are inhibited by C1-INH (serum kallikrein, NGF-y, and Ci), C i did not cleave pro-MSP, possibly related to the lack of the trypsin family serine protease consensus pattern a t the active-site histidine. In the context of inflammatory reactions, additional enzymes in the family, such as the elastases and mast cell proteases, should be tested for their capacity to either act as pro-MSP convertases or to degrade and biologically inactivate pro-MSP or MSP.

Medium
Our data support the hypothesis that cleavage of pro-MSP by the serine proteases listed in Table I11 causes a conformational change in the protein, because cleavage is associated with exposure of an epitope recognized by a monoclonal anti-MSP antibody (Table I) a s well as the appearance of biological activity (Table 11, Fig. 5 ) . Although we have demonstrated binding of MSP to its target cells, we have not yet determined if binding is associated with the conformational change in pro-MSP. Despite the presence of other MSP Arg-Val bonds that could be potential sites of serine protease cleavage, with the exception of plasmin, the serum proteases tested did not cause degradation of MSP to biologically inactive fragments under the conditions described.
NGF-y and EGF-BP are both examples of proteolytic enzymes that occur in complexes with their respective proprotein substrates (10, 11). At high concentrations of the 116-kDa NGF complex, NGF-y lacks proteolytic activity. But on dilution to 0.5 p~ or less, NGF-y autocatalytically activates (19). Active NGF-y cleaves arginyl bonds at two sites of pro-NGF to yield NGF-P, which is the moiety that induces nerve growth in the in vitro assay by which NGF was first discovered (20). Although neither NGF-y nor EGF-BP can cleave their noncorresponding substrates (111, our work shows that they can both cleave and activate pro-MSP. In addition, NGF-y has been shown to proteolytically activate zymogens of two other protein systems, complement and plasminogen. NGF-y can mimic Ci, in that it can proteolytically activate C4 and C2; like Ci, it is inhibited by C1-INH, a protein that blocks enzymatic activity of C i and serum kallikrein (16). NGF-7 is also capable of cleaving precursor single-chain urokinase-type plasminogen activator to the proteolytically active two-chain u-PA, which mediates conversion of plasminogen to plasmin (21).
Activation by NGF-y of pro-MSP, a protein that stimulates macrophages, prompts consideration of possible roles of NGF in inflammation and wound healing. NGF augments responses of B-lymphocytes to mitogens (22). Increased concentrations of NGF-/3 have been reported in the serum of patients with systematic lupus erythematosus during exacerbations (23) and in the cerebrospinal fluid of multiple sclerosis patients during acute attacks (24). High concentrations of NGF and EGF are found in the salivary glands and saliva of mice (10). In 1979, it was reported that the rate of contraction and healing of a 1-cm2 wound cut into the back of a mouse was enhanced by cage mate communal licking. The rate was slowed if the mouse was housed alone or if cage mates were sialoadenectomized (25). I t was subsequently shown that wound contraction rates of sialoadenectomized mice housed together could be increased to normal by application of 116-kDa NGF at 12-h intervals (26). Enzymatic activity of NGF was required, because diisopropylfluorophosphate-treated NGF was ineffective. Do salivary enzymes and growth factors play a role in wound healing of higher animals? Although communal wound licking is not a characteristic behavior of contemporary human society, "licking their wounds" has persisted as a figure of speech in many languages. And NGF-y mRNA has been demonstrated in Northern blots of human salivary gland RNA, hybridized with mouse NGF-y cDNA (27). Transcripts of the whole NGF complex have not been detected, but this may be due to the lack of a species-specific probe. Thus, at least the NGF-y gene has persisted in evolution from mouse to man.
Based on these findings, we can enumerate possible roles of proteolytically active NGF-y in wound healing and tissue repair as follows. 1) In the promotion of wound contraction, the mechanism is unknown. 2) Activation of serum complement (16) occurs with the release of the C5a leukocyte chemoattractant. In contrast t o the classical complement pathway, which begins with antibody-induced activation of C1, there is no antibody requirement for initiation of the complement activation cascade by NGF-y. 3) There is induction of neutrophil migration to a site of injury. Because this occurs in C5-deficient mice, the NGF-y substrate is apparently not C5 (28). 4) Cleavage and activation of precursor single-chain u-PA can lead to the formation of plasmin and dissolution of fibrin clot (21). 5) Cleavage and activation of pro-MSP may cause stimulation of tissue macrophage motility, pinocytic, and phagocytic activity, including ingestion of C3bi-coated cells via the CR3 receptor (3). In.
vivo experiments in mouse models of tissue injury will be required to explore this possibility.