α2-Macroglobulin Functions as a Cytokine Carrier to Induce Nitric Oxide Synthesis and Cause Nitric Oxide-dependent Cytotoxicity in the RAW 264.7 Macrophage Cell Line

Nitric oxide (NO) is an important mediator of macrophage activities. We studied the regulation of macrophage NO synthesis by α2-macroglobulin (α2M), a proteinase inhibitor and carrier of certain growth factors, including transforming growth factor-β (TGF-β). Native α2M and the α2M receptor-recognized derivative, α2M-methylamine (α2M-MA), increased nitrite generation by the RAW 264.7 murine macrophage cell line. The level of nitrite accumulation, which is an index of NO synthesis, was comparable to that observed with interferon-γ. Native α2M and α2M-MA also increased inducible nitric oxide synthase (iNOS) mRNA levels and substantially reduced the number of viable cells, as determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium/succinyl dehydrogenase assay or trypan blue exclusion. At slightly higher α2M concentrations, [3H]thymidine incorporation was inhibited. All of these activities were counteracted nearly completely when the iNOS competitive inhibitor NG-monomethyl-L-arginine was included. By in situ nick translation, native α2M and α2M-MA increased the percentage of cells with detectable single strand chromatin nicks from 4 to 12 and 17%, respectively. This change suggested apoptosis; however, electron microscopy studies demonstrated variability in the morphology of injured cells. To determine the mechanism by which α2M increases macrophage NO synthesis, we studied proteolytic α2M derivatives that retain partial activity. A 600-kDa derivative that retains growth factor binding activity increased RAW 264.7 cell NO synthesis and iNOS mRNA levels comparable to native α2M and α2M-MA. The purified 18-kDa α2M receptor-binding fragment had no effect on NO synthesis or iNOS expression. Thus, the growth factor-carrier activity of α2M and not its receptor-binding activity is essential for NO synthesis regulation. A TGF-β-neutralizing antibody mimicked the activity of α2M, increasing RAW 264.7 cell NO synthesis and decreasing cellular viability. These studies demonstrate that α2M can regulate macrophage NO synthesis and profoundly affect cellular function without gaining entry into the cell and without binding specific plasma membrane receptors.

NO is a potent mediator of many macrophage activities including the cytolytic response to tumor cells and microorganisms (18,26,27). NO also regulates processes involved in atherogenesis (28 -32). When induced at high levels, iNOS can generate NO at concentrations that are cytotoxic to the cells that synthesize it and to neighboring cells (29,33). High levels of NO induce changes suggestive of apoptosis in mouse peritoneal macrophages and in the RAW 264.7 mouse macrophage cell line (33)(34)(35). Thus, regulation of NO synthesis may be important under normal physiologic conditions and in various pathophysiologic states.
Human ␣ 2 -macroglobulin (␣ 2 M) is a large (M r ϳ 718,000), homotetrameric glycoprotein found at high concentration in the plasma (2-5 M) and in extravascular spaces (36,37). Classically, ␣ 2 M has been described as a broad-spectrum proteinase inhibitor; however, more recent studies suggest an important role for ␣ 2 M in the regulation of cellular growth and physiology (37,39,40). ␣ 2 M exists in at least two well-described conformations. The native form expresses proteinase-inhibitory activity but is not recognized by cellular receptors (36). "Activated" ␣ 2 M is generated by reaction with proteinases or small primary amines that modify the ␣ 2 M thiol ester bonds. The structure of ␣ 2 M after reaction with proteinase or methylamine (␣ 2 M-MA) is equivalent, allowing the use of ␣ 2 M-MA as a model of the activated ␣ 2 M conformation (41,42). ␣ 2 M-MA and ␣ 2 M-proteinase complexes are recognized equivalently by ␣ 2 M-specific cellular receptors. One such receptor has been purified and characterized; this receptor is identical to low density lipoprotein receptor-related protein (LRP) (43)(44)(45)(46). ␣ 2 M may regulate cellular growth and physiology by at least two mechanisms, the first of which involves cytokine carrier activity. TGF-␤1, TGF-␤2, platelet-derived growth factor BB, nerve growth factor-␤, and interleukin-1␤ are among a growing list of cytokines reported to bind to ␣ 2 M (39,40,47). All of these cytokines, with the exception of TGF-␤2, demonstrate higher affinity for the activated ␣ 2 M conformation (47,48). Nevertheless, due to the large excess of native ␣ 2 M compared with activated ␣ 2 M in plasma and serum-supplemented cell culture medium, native ␣ 2 M is frequently responsible for the cytokine carrier activity observed in biological systems (39, 49 -51). Many cytokines, including IFN-␥, do not bind ␣ 2 M with significant affinity (47). Therefore, ␣ 2 M may alter the balance of cytokines within the microenvironment surrounding responsive cells. Binding of TGF-␤ isoforms to ␣ 2 M neutralizes the activity of the TGF-␤ toward various cells including endothelium (51), keratinocytes (52), and mink lung epithelium (49). The extent of TGF-␤ neutralization correlates with the affinity of the ␣ 2 M/growth factor interaction (39,47,51). The growth factor binding activities of tetrameric ␣-macroglobulins from different species, including human, mouse, and rat, are comparable (39). 2 The second mechanism by which ␣ 2 M may regulate cellular function is independent of cytokine-carrier activity. Misra et al. (53,54) proposed that mouse peritoneal macrophages express a second ␣ 2 M "signaling receptor." The receptor recognizes only activated ␣ 2 M but is distinct from LRP. In response to activated ␣ 2 M, the second ␣ 2 M receptor initiates signal transduction responses, including rapid phosphatidylinositol 4,5bisphosphate hydrolysis (53,54). Activated ␣ 2 M has been shown to affect superoxide anion production (55) and prostaglandin E 2 synthesis (56) in mouse peritoneal macrophages. The mechanism by which activated ␣ 2 M causes these changes has not been fully explored.
Macrophages and fibroblasts synthesize ␣ 2 M (36). Furthermore, with loss of vascular integrity, ␣ 2 M might be expected to reach high concentrations in the interstitial spaces (37). Therefore, we hypothesized that ␣ 2 M may be a significant regulator of macrophage activity at sites of infection or inflammation, in the intima of an injured artery, and in a vascularized tumor. The present study was undertaken to determine whether ␣ 2 M regulates macrophage NO production. Our results demonstrate that ␣ 2 M increases iNOS levels in RAW 264.7 cells, causing cellular death that is prevented by a specific competitive iNOS inhibitor. Regulation of iNOS was entirely due to the ability of ␣ 2 M to bind growth factors secreted by the macrophages. Furthermore, the activity of ␣ 2 M was mimicked by TGF-␤-specific neutralizing antibody. Based on these studies, we propose that ␣ 2 M may be a component of an important macrophage autocrine regulatory pathway involved in controlling cellular NO production.

MATERIALS AND METHODS
Reagents-Recombinant murine IFN-␥ was obtained from Genzyme Diagnostics (Cambridge, MA) or from Schering Co. (Kenilworth, NJ). TGF-␤2 was purchased from R&D Systems (Minneapolis, MN). Monoclonal antibody 1D11.16 (IgG 1 ), which neutralizes the activity of TGF-␤1 and TGF-␤2, was obtained from Genzyme Diagnostics. Monoclonal antibody AB 6.01, an IgG 1 that recognizes cytokeratins, was purified from ascites fluid generated with hybridoma UCD/AB 6.01 from the American Type Culture Collection (ATCC) (Rockville, MD). In this study, AB 6.01 was used as an IgG subclass-matched control for 1D11. 16. Porcine pancreatic papain was from Worthington. The iNOS inhibitor N G -monomethyl-L-arginine (NMMA) was purchased from Sigma.
Native ␣ 2 M was purified from human plasma according to the method of Imber and Pizzo (57). Final preparations were free of conformationally modified forms as determined by nondenaturing polyacrylamide gel electrophoresis and SDS-polyacrylamide gel electrophoresis. ␣ 2 M-MA was prepared by reacting native ␣ 2 M with 200 mM methylamine in 50 mM Tris-HCl, pH 8.2, for 6 h at 22°C. Unreacted methylamine was removed by dialysis against 20 mM sodium phosphate, 150 mM NaCl, pH 7.4 (PBS) at 4°C. Native ␣ 2 M and ␣ 2 M-MA preparations contained less than 0.1 ng/ml of endotoxin as determined by the limulus lysate assay (Associates of Cape Cod, Inc., Woods Hole, MA).
␣ 2 M-MA was digested with papain according to the method of Sottrup-Jensen et al. (58), as modified by Hussaini et al. (59). The resulting products, including the 18-kDa receptor-binding fragment (RBF) from the C terminus of each ␣ 2 M subunit and the truncated 600-kDa ␣ 2 M-MA derivative were purified by chromatography on Ultrogel AcA-22. Purified RBF competed with ␣ 2 M-MA for specific binding to RAW 264.7 cells, as expected since RBF retains the ␣ 2 M receptorrecognition site (58). Specific 125 I-␣ 2 M-MA binding was completely inhibited by 5 M RBF; the IC 50 was 250 nM. The 600-kDa derivative does not bind to ␣ 2 M receptors but retains growth factor binding activity (60). RBF completely lacks growth factor binding activity. 2 Cell Culture-RAW 264.7 cells were obtained from the ATCC and cultured in RPMI 1640 (Sigma) supplemented with 10% FBS (Sigma) and 1% penicillin-streptomycin (PS) (Sigma) at 37°C, 5% CO 2 , and 95% humidity. The concentration of L-arginine in RPMI 1640 is 1.2 mM. Fetal bovine heart endothelial (FBHE) cells were obtained from the ATCC and cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 20 ng/ml acidic fibroblast growth factor, and 80 ng/ml basic fibroblast growth factor.
Northern Blot Analysis of iNOS mRNA-The iNOS cDNA in pUC19 was generously provided by Drs. Qiao-wen Xie and Carl Nathan, Cornell University Medical College, New York, NY. A 645-nucleotide fragment of the cDNA was excised with HindII/BamHI and labeled with [␣-32 P]dCTP by random oligonucleotide-primed synthesis (Boehringer Mannheim, Random Primed DNA Labeling Kit) for use as a Northern blot probe. RAW 264.7 cells (plated at 2 ϫ 10 5 /well in 6-well plates) were treated with native ␣ 2 M (280 nM), ␣ 2 M-MA (280 nM), IFN-␥ (10 ng/ml), the 600-kDa derivative (200 nM), or RBF (200 nM) for 24 h. Total cellular RNA was isolated by Trizol extraction (Life Technologies, Inc.). Equal amounts of RNA (20 g) from each preparation were subjected to electrophoresis in 1.0% (w/v) agarose gels and electroblotted to Zeta probe nylon membranes (Bio-Rad). The labeled iNOS probe was incubated with the nylon membranes at 42°C in 5 ϫ SSPE, 5 ϫ Denhardt's solution, 50% formamide, 0.1% SDS, and 100 g/ml salmon sperm DNA for 24 h. The membranes were rinsed twice with 5 ϫ SSPE, 0.5% SDS at room temperature, followed by two washes with 0.1 ϫ SSPE, 1.0% SDS for 10 min each at 65°C. The membranes were then either exposed to Kodak X-Omat AR-5 film at Ϫ80°C or a PhosphorImager. As a control for RNA load, all blots were hybridized with a [␣-32 P]dCTPlabeled cDNA probe for phosphoglyceraldehyde dehydrogenase.
Cell Viability/Proliferation-Viable RAW 264.7 cells were quanti-2 D. J. Webb and S. L. Gonias, manuscript submitted for publication. tated using the Cell Proliferation Kit I marketed by Boehringer Mannhein. The assay detects succinyl dehydrogenase, a mitochondrial enzyme, only in living cells (62). The substrate for succinyl dehydrogenase is 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT). In our studies, 10 4 cells were plated in each well of a 96-well flat bottom tissue culture plate (Corning, NY). The culture medium was RPMI 1640, 10% FBS, 1% PS. After 24 h, the medium was replaced with SFM, and incubation was continued for an additional 24 h. Any of the following were then added: native ␣ 2 M (280 nM), ␣ 2 M-MA (0.14 -280 nM), RBF (200 nM), the 600-kDa derivative (200 nM), or antibody 1D11.16 (50 g/ml). After 24 h, viable cell number was determined by adding MTT directly into the cell culture wells, as described by the manufacturer. All incubation conditions were studied in quadruplicate.
In some experiments, ␣ 2 M and/or IFN-␥ were added to cultures together with NMMA (1-10 mM).
RAW 264.7 DNA synthesis was assessed by [ 3 H]thymidine incorporation. The cells were plated at 10 4 /well in 24-well plates and cultured first in serum-supplemented medium for 24 h and then in SFM for an additional 24 h. At the beginning of the third day, ␣ 2 M-MA was added and the cells were cultured for 24 h. [ 3 H]Thymidine (1 Ci/ml) was included during the last 6 h. Cells were then washed with Earle's balanced salt solution, 10 mM HEPES, pH 7.4, and fixed in 10% trichloroacetic acid. Cell-associated radioactivity was recovered by incubation in 1.0 M NaOH for 12 h. The pH was neutralized with 1 M HCl. Cell extracts were then combined with Ready-Safe scintillation fluid for counting in a Beckman scintillation counter.
RAW 264.7 cell viability was determined by trypan blue exclusion. Cells were plated at a density of 5 ϫ 10 4 /well in 6-well plates and cultured for 48 h. Native ␣ 2 M (280 nM), ␣ 2 M-MA (0.14 -280 nM), the 600-kDa derivative (200 nM), and RBF (200 nM) were then added. After an additional 24-h incubation, the cells were released, either with trypsin/EDTA (Life Technologies, Inc.) or with a cell scraper, incubated with trypan blue, and counted using a hemocytometer. Since the monolayers were not washed prior to trypsin-EDTA treatment or scraping, adherent cells, and cells that detached during incubation with ␣ 2 M were detected.
In Situ Nick Translation (ISNT) of RAW 264.7 Cell DNA-In ISNT, DNA polymerase I is used to label DNA that contains single strand nicks (63,64). The method detects programmed cell death but may also detect other forms of cell injury/death including what has been frequently referred to as necrotic death (65,66). In our experiments, RAW 264.7 cells (5 ϫ 10 4 ) were cultured on coverslips in 6-well plates for 24 h in SFM supplemented with native ␣ 2 M (280 nM) or ␣ 2 M-MA (280 nM). Cells were washed with Hanks' balanced salt solution, fixed with 3.7% paraformaldehyde in PBS, permeabilized with 0.5% Triton X-100, washed again, and air dried. Cells on coverslips were then incubated for 2 h at 37°C with 50 mM Tris-HCl, 5 mM MgCl 2 , 10 mM mercaptoethanol, 0.005% bovine serum albumin, 0.02 mM dATP, dCTP, dGTP, 13 M dTTP, 7 M digoxigenin-11-dUTP, and DNA polymerase-1 (Boehringer Mannheim). After 1 h, the coverslips were washed twice with 20 mM EDTA and once with 20 mM Tris-HCl, 150 mM NaCl, pH 7.4 (TBS). The cells were then incubated with TBS and 5% nonfat dried milk for 1 h followed by anti-digoxigenin Fab antibody conjugated to alkaline phosphatase (Boehringer Mannheim) for 12 h (1:200 dilution). The next day, cells were washed and stained with the chromogen, 5-bromo-4-chloro-3-indoyl phosphate p-toluidine/p-nitro blue tetrazolium chloride. Cellular labeling was viewed and photographed using Nomarski differential interference contrast optics in an Olympus BH2 Microscope. In control experiments, no labeling was observed when the digoxigenin-11-dUTP, the anti-digoxigenin Fab antibody, or the DNA polymerase were omitted from the various steps of the procedure.
Electron Microscopy-RAW 264.7 cells were examined by transmission electron microscopy after treatment with ␣ 2 M. Cells (5 ϫ 10 4 ) were cultured in 6-well plates for 24 h. The medium was SFM supplemented with native ␣ 2 M (280 nM) or ␣ 2 M-MA (280 nM). Osmium tetroxide (2% w/v) dissolved in 0.1 M sodium cacodylate was slowly added to the medium to a final concentration of 1% and allowed to fix the cells for 10 min. The medium and fixative were then aspirated and replaced with 2% (w/v) paraformaldehyde, 2% (v/v) glutaraldehyde in 0.1 M sodium cacodylate to fix overnight at 4°C. The cells were removed from the wells with the aid of a rubber policeman and pelleted in an Eppendorf microcentrifuge (16,000 ϫ g). The pellet was then processed by transfer sequentially into 70% acetone, anhydrous acetone, 1:1 acetone/epoxy resin, and pure resin. Finally, the pellet was transferred into a BEEM capsule filled with pure resin for polymerization at 60°C for 24 h. Thin sections were prepared, placed on 150 mesh nickel grids, stained with uranyl acetate/lead citrate, and analyzed using a ZEISS 902 electron microscope.

RESULTS
Effects of ␣ 2 M on NO Synthesis and iNOS mRNA Expression-RAW 264.7 cells that were exposed to native ␣ 2 M or ␣ 2 M-MA for 24 h synthesized significantly increased levels of NO, as determined by nitrite analysis (Fig. 1). The increase was comparable to that observed in cultures exposed to IFN-␥ (10 ng/ml), a known inducer of macrophage iNOS (18,67). The combination of IFN-␥ and native ␣ 2 M caused an additive NO synthesis response. IFN-␥ and ␣ 2 M-MA caused an NO synthesis response that was at least additive. The ability of the two ␣ 2 M conformations to induce RAW 264.7 cell NO production was not significantly different in the absence of IFN-␥. In the presence of IFN-␥, NO synthesis in response to ␣ 2 M-MA was significantly higher than that observed with native ␣ 2 M (p Ͻ 0.05).
Native ␣ 2 M and ␣ 2 M-MA increased iNOS mRNA levels in RAW 264.7 cells, as determined by Northern blot analysis (Fig.  2). The increase in iNOS mRNA caused by ␣ 2 M-MA was either equivalent to the increase caused by native ␣ 2 M or slightly greater in six separate experiments. In the particular study shown in Fig. 2, nitrite levels were measured in the cultures from which mRNA was harvested. The nitrite levels, like the iNOS mRNA levels, were increased similarly by native ␣ 2 M and ␣ 2 M-MA.
The increase in iNOS mRNA caused by native ␣ 2 M or ␣ 2 M-MA was significantly less than the increase observed with IFN-␥ (10 ng/ml) even though all three agents increased nitrite levels similarly (as shown in Figs. 1 and 2). Thus, while the increase in NO synthesis caused by ␣ 2 M is at least partially explained by the increase in iNOS mRNA, a second contributing mechanism may be involved. Feasible second mechanisms include an increased rate of iNOS mRNA translation and/or stabilization of the synthesized enzyme.
NO Synthesis and iNOS Expression in Cultures Treated with ␣ 2 M Derivatives-In the next series of experiments, we sought to determine the mechanism by which ␣ 2 M increases iNOS expression and NO synthesis in RAW 264.7 cells. Our experiments showing that native ␣ 2 M and ␣ 2 M-MA are comparably active in promoting NO synthesis provided some insight into the mechanism. For example, since native ␣ 2 M does not bind LRP (36) or the second ␣ 2 M signaling receptor (53,54), the increase in iNOS mRNA and NO synthesis could not be due to an ␣ 2 M receptor-mediated mechanism. If the activity of ␣ 2 M was due to its ability to neutralize an extracellular proteinase that regulates RAW 264.7 cell NO synthesis, ␣ 2 M-MA would be inactive since this form of ␣ 2 M retains no proteinase inhibitory activity (36). Thus, the most likely mechanism involved the cytokine binding activity of ␣ 2 M. We hypothesized that ␣ 2 M binds one or more cytokines secreted by the RAW 264.7 cells, thereby interrupting an important autocrine regulatory loop controlling NO synthesis.
To test this hypothesis, two ␣ 2 M derivatives with different activities were studied. The 600-kDa derivative retains growth factor binding activity but is not recognized by ␣ 2 M receptors (58,60). RBF does not bind growth factors (60) 2 but interacts with LRP and the second ␣ 2 M signaling receptor described by Misra et al. (53,54). The RBF/second ␣ 2 M signaling receptor interaction apparently causes the full spectrum of signal transduction responses observed with activated ␣ 2 M. Fig. 3 shows that the 600-kDa derivative increased NO synthesis in RAW 264.7 cells. The extent of the response, in the presence and absence of IFN-␥, was comparable to that observed with ␣ 2 M-MA. By contrast, RBF had no effect on NO synthesis. The 600-kDa derivative also increased the level of iNOS mRNA while RBF had little or no effect (Fig. 4). These results confirm that the ability of ␣ 2 M to increase NO synthesis in RAW 264.7 cells results from ␣ 2 M-cytokine interactions occurring in the medium and not from an ␣ 2 M receptor interaction.
RAW 264.7 Cell Proliferation and Viability-Incubation with ␣ 2 M-MA (280 nM) decreased the number of RAW 264.7 cells detected by MTT assay and this effect was time-dependent. At 2 and 6 h, cell number was decreased by less than 10% compared with SFM-treated control cultures; however, at 24 and 48 h, cell number in ␣ 2 M-MA-treated cultures was decreased by 72 and 84%, respectively (results not shown). To assess the role of NO synthesis in causing RAW 264.7 cell death, cultures were treated with various forms of ␣ 2 M for 24 h. All of the derivatives that increased NO synthesis, including native ␣ 2 M, ␣ 2 M-MA and the 600-kDa derivative, decreased the number of viable cells detected by MTT assay (Fig. 5). RBF did not reduce the number of viable cells. Therefore, cytokine carrier activity and/or NO synthesis was apparently responsible for the de-crease in cell number caused by ␣ 2 M.
To determine if the observed decrease in RAW 264.7 cell number was due to a change in the rate of proliferation or altered cellular viability, [ 3 H]thymidine incorporation and trypan blue exclusion experiments were performed. Increasing concentrations of ␣ 2 M-MA were added to the cultures. By MTT assay, an ␣ 2 M-MA concentration-dependent decrease in the number of viable cells was apparent (Fig. 6). This decrease was accompanied by a significant increase in the fraction of cells that did not exclude trypan blue in cultures treated with 10 -300 nM ␣ 2 M-MA. ing through S-phase of the cell cycle.
Cultures were treated with native ␣ 2 M, RBF, or the 600-kDa derivative for 24 h and studied by trypan blue exclusion. The ␣ 2 M derivatives that decreased cell number by MTT assay (as shown in Fig. 5) also decreased the fraction of cells that excluded trypan blue (Table I). By contrast, RBF did not decrease trypan blue exclusion. Equivalent results were obtained whether the cells were released from the wells by scraping or with trypsin/EDTA.
NO as the Mediator of ␣ 2 M-induced Cytotoxicity-Experiments were performed with the competitive iNOS active site inhibitor NMMA to determine whether NO synthesis was responsible for the observed loss of viability in ␣ 2 M-treated RAW 264.7 cells. Fig. 7 shows that NMMA blocked NO synthesis and the resulting nitrite accumulation in cultures treated with ISNT Experiments-RAW 264.7 cells, cultured on glass coverslips, were treated with native ␣ 2 M (280 nM) or ␣ 2 M-MA (280 nM) for 24 h. The cultures were then processed for ISNT and photomicrographed using Nomarski differential interference contrast optics. Fig. 8 shows representative fields for each treatment protocol. The number of cells remaining on the coverslip was decreased after treatment with native ␣ 2 M or ␣ 2 M-MA. Among the remaining cells, considerable heterogeneity in size was apparent; at least a few cells in each field were substantially enlarged compared with the control cultures. Occasional multinucleated cells were observed.
DNA fragmentation was detected in some cells of the control preparation by ISNT. Cells in the ␣ 2 M and ␣ 2 M-MA-treated cultures were labeled as well. The percentage of adherent cells that stained positively was increased after ␣ 2 M or ␣ 2 M-MA treatment (Table II), even though many of the cells that were injured or dead had already detached from the coverslips. DNA   was isolated from RAW 264.7 cells treated with native ␣ 2 M or ␣ 2 M-MA for 24 h and analyzed by agarose gel electrophoresis. Internucleosomal DNA fragmentation was not detected (data not shown). Detection of DNA ladders provides strong evidence for apoptosis; however, the absence of DNA ladders does not preclude apoptosis in part or all of a cell population (68,69).
Electron Microscopy Experiments-RAW 264.7 cells that were treated with native ␣ 2 M or ␣ 2 M-MA demonstrated considerable morphologic heterogeneity by electron microscopy (Fig. 9). Different patterns of cellular injury were observed. The central cell, labeled with an arrow, showed signs suggestive of apoptosis. The nuclear chromatin was condensed into a number of electron-dense bodies, and the cytoplasm had an amorphous appearance. The adjacent cell, labeled with an arrowhead, showed changes limited primarily to the cytoplasm. Organelles were severely dilated and the plasma membrane was discontinuous. Other cells showed additional morphologic changes, such as irregular nuclear contours and increased electron density of the heterochromatin; these changes may or may not have been associated with cellular injury. Occasional enlarged cells with irregularly shaped nuclei were observed (asterisk), consistent with the changes seen by light microscopy. The diverse morphologic appearance of cells treated with ␣ 2 M-MA suggests that the cells did not undergo a uniform mode of cell death. Equivalent results were obtained with native ␣ 2 M.
The TGF-␤-Carrier Activity of ␣ 2 M and iNOS Production-Of the 13 growth factors studied by our laboratory to date, TGF-␤2 binds to ␣ 2 M with the highest affinity (47,51). Furthermore, TGF-␤2 is the only growth factor that binds to native ␣ 2 M and ␣ 2 M-MA with equivalent affinity. Given the role of TGF-␤ isoforms as down-modulators of macrophage NO synthesis (21) and previous studies demonstrating TGF-␤ expression by macrophages (70 -72), we decided to determine whether the observed increase in iNOS expression and NO synthesis could be explained by the TGF-␤-carrier activity of ␣ 2 M. RAW 264.7 cells were treated for 24 h with antibody 1D11.16 (50 g/ml), which neutralizes TGF-␤1 and TGF-␤2. As shown in Fig. 10, the antibody caused a significant increase in nitrite accumulation, similar to ␣ 2 M-MA. Furthermore, the antibody significantly decreased the number of viable cells detected by MTT assay. The IgG subclass-matched control antibody AB 6.01 (50 g/ml) did not significantly affect nitrite levels relative to the control (results not shown). These studies demonstrate that the activities of ␣ 2 M described here can be fully mimicked by antibody that neutralizes TGF-␤ released by macrophages in culture.
Conditioned medium from RAW 264.7 cells was analyzed for TGF-␤ activity using the FBHE proliferation assay. The conditioned medium inhibited [ 3 H]thymidine incorporation by the FBHE cells, as expected for a TGF-␤-containing vehicle. Antibody 1D11.16 neutralized the activity in the conditioned medium, confirming that the FBHE growth inhibition was due to TGF-␤. The concentration of TGF-␤ in the RAW 264.7 cell conditioned medium, as determined by comparison to purified TGF-␤2, was 9 Ϯ 4 pM (n ϭ 4).

DISCUSSION
The growth factor-carrier activity of ␣ 2 M was originally identified in studies of whole plasma. O'Connor-McCourt and Wakefield (73) demonstrated that nearly all of the TGF-␤1 in plasma is associated with ␣ 2 M and Huang et al. (74) showed that TGF-␤1 is inactive while bound to ␣ 2 M. Based on these early studies, ␣ 2 M⅐TGF-␤ complex was initially referred to as latent TGF-␤, a term subsequently reserved for the precursor form of TGF-␤ secreted by cells in complex with a latent TGF-␤ binding protein (75). More recent studies have shown that TGF-␤1 (39, 50), platelet-derived growth factor-BB (39), and  nerve growth factor-␤ 3 bind to native ␣ 2 M when injected intravascularly in mice. After an initial rapid clearance phase, the complex with ␣ 2 M becomes the primary form of each growth factor present in the plasma. The ␣ 2 M⅐growth factor complex is relatively stable in the blood, presumably forming a pool of slowly releasable activity since complexes of growth factors with native ␣ 2 M remain primarily noncovalent and reversible.
These animal model and whole plasma experiments demonstrate that ␣ 2 M-growth factor interactions are physiologically significant in vivo.
In this study, we demonstrated that ␣ 2 M is an important regulator of RAW 264.7 cell NO synthesis. The underlying mechanism depends on the growth factor-carrier activity of ␣ 2 M. Since the experiments were performed in serum-free medium and an alternative source of cytokines was not provided, the critical interaction must have involved ␣ 2 M and one or more cytokines secreted by the RAW 264.7 cells. For a number of reasons, we examined the possible role of the TGF-␤ superfamily in the ␣ 2 M/NO-regulatory system. First, TGF-␤ isoforms inhibit macrophage NO synthesis (21). Furthermore, it is known that monocytes-macrophages synthesize TGF-␤ (70 -72), a result confirmed for RAW 264.7 cells in this study. Since TGF-␤ decreases iNOS mRNA translation and destabilizes the protein (22), a neutralizing interaction of ␣ 2 M with RAW 264.7 cell TGF-␤ is consistent with our observation that ␣ 2 M elevates NO synthesis disproportionately with iNOS mRNA. Finally, we have shown that the TGF-␤-neutralizing activity of ␣ 2 M in FBHE cultures closely correlates with the binding affinity (1/ K D ) of the ␣ 2 M/TGF-␤ interaction (51). We assume that the same principle holds for macrophage cultures. Since native ␣ 2 M was as active, or nearly as active, as ␣ 2 M-MA in the regulation of NO synthesis, the involvement of TGF-␤2 is suggested; TGF-␤2 is the only growth factor studied to date that binds native ␣ 2 M and ␣ 2 M-MA with equal affinity.
To test the hypothesis that TGF-␤-binding accounts for the NO synthesis induced by ␣ 2 M, we studied a TGF-␤-neutralizing antibody. The antibody completely mimicked the activity of ␣ 2 M, increasing cellular NO synthesis and decreasing cellular viability. Although the spectrum of cytokines that interact with ␣ 2 M in the macrophage culture medium may be complex, the antibody studies allow us to conclude that TGF-␤ binding is sufficient to account for the activities of ␣ 2 M observed here. Furthermore, these antibody studies identify an important autocrine regulatory loop for the RAW 264.7 macrophage cell line. Apparently, TGF-␤ synthesized by the cells themselves functions to suppress iNOS expression. By interrupting this autocrine pathway, extracellular mediators, such as ␣ 2 M or antibody, can profoundly affect cellular phenotype and function without gaining entry into the cell and without binding to plasma membrane receptors.
As a result of the shift in available cytokines in the culture medium, ␣ 2 M-treated RAW 264.7 cells underwent cell death. One possible explanation for this result is that ␣ 2 M withdraws a growth factor(s) that is otherwise available to the macrophages and necessary for continued growth. Alternatively, cell death might be mediated directly by NO, synthesized at cytotoxic levels in the ␣ 2 M-treated cultures. These two mechanisms are similar since both involve ␣ 2 M interacting with macrophage-secreted cytokines. Furthermore, the mechanisms are not mutually exclusive. To isolate the role of NO synthesis in cell death, among other changes in cellular function that may result from altered cytokine availability, we performed experiments with a specific, competitive iNOS inhibitor, NMMA. Inhibition of iNOS substantially reversed the loss of cellular viability in ␣ 2 M-treated cultures, demonstrating that NO is completely, or nearly completely, responsible for death of the macrophage cell line under these experimental conditions.
It has been reported that macrophages, which are induced to secrete high levels of NO by treatment with IFN-␥ and lipopolysaccharide, undergo apoptosis (33)(34)(35). Therefore, we performed ISNT and agarose gel electrophoresis experiments to examine the changes occurring in ␣ 2 M-treated RAW 264.7 cells. The ISNT experiments revealed evidence of singlestranded DNA fragmentation, at the single-cell level, in an increased percentage of the cells treated with native ␣ 2 M or ␣ 2 M-MA. The DNA-agarose gel electrophoresis experiments were negative. ISNT is more sensitive than agarose gel electrophoresis in detecting DNA fragmentation, especially when the change is limited to a subpopulation of the cells under examination (65). However, labeling of cells by ISNT may detect modes of cell death other than apoptosis. Therefore, we further explored the morphologic changes occurring in ␣ 2 Mtreated RAW 264.7 cells by electron microscopy. These studies showed a heterogeneous pattern of ultrastructural changes in each ␣ 2 M-treated preparation. While some cells showed morphologic changes suggesting apoptotic cell death, other injured and dying cells did not. One possible explanation for this result is that the cells are at various stages of dying; however, we believe that dissynchrony in our cultures does not entirely explain the difference in our results and the previously reported studies (33)(34)(35). A second explanation is that the various agents used to induce NO synthesis (␣ 2 M, IFN-␥, and lipopolysaccharide) have different effects on the manner in which the macrophage responds to high levels of NO.
Since NO has been implicated in many normal physiologic and pathophysiologic processes, studies on the regulation of NO synthesis have broad relevance. The experiments performed here identify ␣ 2 M as a regulator of macrophage NO synthesis for the first time. Additional work will be necessary to determine whether our model cell culture system is representative of monocytes and macrophages in vivo. Our results implicating ␣ 2 M in the regulation of NO synthesis are unique compared with many previous studies that focused on the biological activities of ␣ 2 M, since, in our system, native ␣ 2 M was active in addition to the activated form (␣ 2 M-MA). Due to the efficiency of the LRP clearance mechanism, concentrations of activated ␣ 2 M are very low in the blood (36,39,40). Higher levels of activated ␣ 2 M may accumulate in interstitial spaces and in non-vascular body fluids when elevated levels of active proteinase are available locally (37); however, LRP may limit concentrations of activated ␣ 2 M in these microenvironments as well. By contrast, the native form of ␣ 2 M is not subject to receptor-mediated clearance and therefore is stable in the presence of LRP-expressing cells, including macrophages, smooth muscle cells and fibroblasts. Thus, native ␣ 2 M, due to its growth factor binding activity and high concentration, may be the most significant form of the protein to consider as a potential regulator of NO synthesis in vivo.