Murine alpha-macroglobulins demonstrate divergent activities as neutralizers of transforming growth factor-beta and as inducers of nitric oxide synthesis. A possible mechanism for the endotoxin insensitivity of the alpha2-macroglobulin gene knock-out mouse.

a2-Macroglobulin null mice demonstrate increased resistance to endotoxin challenge (Umans, L., Serneels, L., Overbergh, L., Van Leuven, F., and Van den Berghe, H. (1995) J. Biol. Chem. 270, 19778–19785). We hypothesized that this phenotype might reflect the function of murine a2M (ma2M) as a neutralizer of transforming growth factor-b (TGF-b) and inducer of nitric oxide synthesis in vivo. When incubated with wild-type mouse plasma, TGF-b1 and TGF-b2 bound only to ma2M. Alternative TGF-b-binding proteins were not detected in plasma from a2M(2/2) mice. Wild-type mouse plasma, but not plasma from a2M(2/2) mice, inhibited TGF-b1 binding to TGF-b receptors on fibroblasts. Purified ma2M bound TGF-b1 and TGF-b2 with similar affinity; the KD values were 28 6 4 and 33 6 4 nM, respectively. Murinoglobulin, the second murine a-macroglobulin, bound both TGF-b isoforms with 30-fold lower affinity. Ma2M counteracted the activities of TGF-b1 and TGF-b2 in an endothelial cell growth assay. Ma2M also induced NO synthesis when incubated with RAW 264.7 cells, an activity which probably results from the neutralization of autocrine TGF-b activity. Human a2M induced NO synthesis comparably to ma2M; however, MUG had no effect. These studies demonstrate that the ability to neutralize TGF-b is a property of ma2M, which is not redundant in the murine a-macroglobulin family or in murine plasma. Ma2M is the only murine a-macroglobulin that promotes NO synthesis. The absence of ma2M, in a2M(2/2) mice, may allow TGF-b to more efficiently suppress excessive iNOS expression following endotoxin challenge.

Bacterial endotoxin is a lipopolysaccharide complex derived from the outer membranes of Gram-negative bacteria and a major mediator of the pathophysiologic changes referred to as septic shock (1,2). Endotoxin triggers release of inflammatory cytokines, including tumor necrosis factor-␣ and interleukin-1␤, from a number of cell types, including macrophages and endothelial cells (1,3). These cytokines, in addition to endotoxin, induce increased cellular synthesis of nitric oxide (4 -7). NO 1 plays a significant role in mediating the profound vasodilation, hypotension, and hyporeactivity to pressor agents, which are characteristic of severe septic shock (2,(5)(6)(7)(8)(9).
NO is produced by three enzymes of the nitric oxide synthase (NOS) family (4,10). Two constitutive forms of NOS (NOS-I, NOS-III) produce low levels of NO in response to agonists which increase intracellular calcium. The resulting NO is important in the regulation of normal homeostatic processes, including blood pressure and neuronal transmission (10). The inducible form of NOS (iNOS or NOS-II) produces high levels of NO, independently of intracellular calcium. The NO produced by iNOS functions in the immune response to pathogens and may also provide cytotoxic activity toward cancer cells (10). Since iNOS activity is not calcium-dependent, regulation of iNOS expression is critical and many cytokines are involved in this process in various cell types (11).
Intravenous administration of endotoxin in experimental animals causes an initial, transient increase in NO that is probably due to stimulation of constitutive NOS followed by a sustained increase in NO produced by iNOS (8). Mice that are iNOS-deficient, due to gene disruption in embryonic stem cells, demonstrate resistance to doses of endotoxin that are lethal in wild-type animals (12,13). iNOS(Ϫ/Ϫ) mice are also less effective at eliminating infections caused by Listeria monocytogenes or Leishmania major. Macrophages that are isolated from iNOS(Ϫ/Ϫ) mice are less effective killers of cancer cells in vitro (12).
Human ␣ 2 -macroglobulin (h␣ 2 M) is a proteinase inhibitor and specific cytokine carrier that induces iNOS expression and NO synthesis in the murine macrophage-like cell line, RAW 264.7 (14). The mechanism of iNOS induction involves the neutralization of autocrine cytokine activity by h␣ 2 M. Since RAW 264.7 cells synthesize, secrete, and activate transforming growth factor-␤ (TGF-␤) and because TGF-␤-neutralizing antibody induces RAW 264.7 cell NO synthesis, similarly to h␣ 2 M, we proposed that neutralization of autocrine TGF-␤ is most likely responsible for the increase in iNOS expression in h␣ 2 M-treated RAW 264.7 cells (14). H␣ 2 M exists in at least two well defined conformations (15,16). The native form is active as a proteinase inhibitor, but not recognized by the ␣ 2 M receptor, low density lipoprotein receptor-related protein (LRP) (17). Thus, native ␣ 2 M is stable in the blood and interstitial spaces. The activated form of ␣ 2 M, generated by reaction with proteinase, is LRP-recognized. In the RAW 264.7 cell system, native ␣ 2 M and activated ␣ 2 M are comparably effective as inducers of NO synthesis (14).
The murine ␣-macroglobulin family includes at least five distinct genes, two of which encode for proteins that have been purified and characterized (18 -20). Murine ␣ 2 M (m␣ 2 M) is a tetrameric protein (M r ϳ 720,000) and a close homologue of human ␣ 2 M. H␣ 2 M and m␣ 2 M share comparably broad proteinase-inhibitory specificities and undergo equivalent proteinase-induced conformational changes. The second member of the murine ␣-macroglobulin family is murinoglobulin (MUG), a single chain (M r ϳ 180,000) proteinase inhibitor which, like m␣ 2 M, is found at high concentrations (1 mg/ml) in adult murine plasma (21,22).
Umans et al. (23) recently reported disruption of the m␣ 2 M gene in embryonic stem cells and the generation of homozygous ␣ 2 M-deficient mice. The ␣ 2 M(Ϫ/Ϫ) mice developed normally and were fertile; however, these mice responded differently from wild-type animals when challenged with certain exogenous agents. ␣ 2 M(Ϫ/Ϫ) mice showed increased resistance to the lethal effects of endotoxin. We hypothesized that the endotoxin resistance of the ␣ 2 M(Ϫ/Ϫ) mice might be related to the in vitro observations that h␣ 2 M neutralizes TGF-␤ and induces iNOS expression (14). TGF-␤ suppresses expression of a number of genes associated with inflammatory reactions, including iNOS (24,25). If neutralization of TGF-␤ at sites of inflammation is an important function of ␣ 2 M, then m␣ 2 M deficiency would allow endogenous TGF-␤ to function more effectively in protecting against excessive iNOS induction following endotoxin challenge.
Levels of MUG in the plasma of non-pregnant ␣ 2 M(Ϫ/Ϫ) mice are unchanged compared with wild-type animals (23), raising the question of whether MUG and m␣ 2 M are redundant as TGF-␤ neutralizers or NO inducers. The goals of this study were to: (i) compare the binding and neutralization of TGF-␤ by proteins from wild-type and ␣ 2 M(Ϫ/Ϫ) mouse plasma; (ii) characterize the binding of TGF-␤1 and TGF-␤2 to purified m␣ 2 M and MUG; and (iii) determine whether m␣ 2 M and/or MUG promote macrophage NO synthesis. Our results demonstrate that m␣ 2 M is unique among murine ␣-macroglobulins, and murine plasma proteins in general, since m␣ 2 M is the only major protein that binds TGF-␤ and inhibits TGF-␤-receptor interactions. Purified m␣ 2 M neutralized TGF-␤ in endothelial cell growth assays and promoted macrophage NO synthesis, while purified MUG was inactive in both cell culture systems. These studies identify TGF-␤ neutralization as a non-redundant activity of m␣ 2 M and suggest a mechanism for the endotoxin insensitive phenotype of ␣ 2 M(Ϫ/Ϫ) mice.

MATERIALS AND METHODS
Reagents and Proteins-TGF-␤1 was from R&D Systems (Minneapolis, MN). TGF-␤2 was from Genzyme (Cambridge, MA). Trypsin was purchased from Worthington and active-site titrated according to the method of Chase and Shaw (26). Chloramine T, fetal bovine serum (FBS), and bovine serum albumin were from Sigma. Bis(sulfosuccinimidyl)suberate (BS 3 ) and Iodo-Beads were from Pierce. Na 125 I was from Amersham. Trypsin-EDTA, Dulbecco's modified Eagle's medium, RPMI 1640, and Earle's balanced salts solution were from Life Technologies, Inc.
H␣ 2 M was purified by the method of Imber and Pizzo (27). M␣ 2 M was isolated by Ni 2ϩ -affinity chromatography, followed by gel filtration on AcA 22 (28). Analysis of purified m␣ 2 M by SDS-PAGE under reducing conditions revealed three bands, with apparent masses of 180, 165, and 35 kDa, as expected. For some experiments, m␣ 2 M was radioiodinated with Iodo-Beads. The specific radioactivity was approximately 1.0 Ci/g. MUG was recovered in the flow-through fractions of the Ni 2ϩ -affinity chromatography column and further purified by gel filtration on Ultrogel AcA 34 and ion exchange chromatography on Q-Sepharose FF (22). SDS-PAGE analysis of purified MUG revealed a single band with an apparent mass of 180 kDa. MUG and m␣ 2 M preparations contained less than 0.3 ng/ml endotoxin, as determined by the limulus amebocyte lysate assay (Associates of Cape Cod Inc., Woods Hole, MA), and were diluted at least 10-fold when introduced into cell culture systems. In control experiments, we determined that endotoxin (Sigma), at concentrations of 1.0 ng/ml or less, does not induce NO synthesis by RAW 264.7 cells.
Nondenaturing PAGE-TGF-␤1 and TGF-␤2 were radioiodinated according to the method of Ruff and Rizzino (29). Specific activities were 100 -200 Ci/g. M␣ 2 M (0.7 M) was incubated with 125 I-TGF-␤1 or 125 I-TGF-␤2 (each at 0.5 nM). The reaction products were subjected to nondenaturing PAGE on 5% slabs (30). The mobility of the m␣ 2 M, as determined by Coomassie staining, provides an index of whether the m␣ 2 M is in the native or activated conformation (31). 125 I-TGF-␤ binding to m␣ 2 M was detected by autoradiography.
Binding of 125 I-TGF-␤ to Murine Plasma Proteins-125 I-TGF-␤1 and 125 I-TGF-␤2 (0.5 nM) were incubated with pooled plasma isolated from ␣ 2 M(Ϫ/Ϫ) or wild-type C57B1 female mice. The plasma was diluted to 2.5 or 25% in 20 mM sodium phosphate, 150 mM NaCl, pH 7.4. Incubations were conducted for 30 min at 37°C. BS 3 (in H 2 O) was then added to a final concentration of 5 mM and incubated for 1 min. Cross-linking reactions were stopped by acidification with 15 mM HCl. Samples were denatured in 2.0% SDS for 30 min at 37°C. The BS 3 was further neutralized with 100 mM Tris-HCl, pH 7.4. Equal amounts of radioactivity from each sample were subjected to SDS-PAGE and autoradiography/PhosphorImager analysis.

Affinity Labeling of TGF-␤ Receptors in the Presence of Murine
Plasma-Cellular TGF-␤ receptors were affinity-labeled by the method of Massague (32), as modified by LaMarre et al. (33). Briefly, AKR-2B fibroblasts were plated at 2 ϫ 10 5 cells per well in 35-mm wells and cultured for 48 h in Dulbecco's modified Eagle's medium with 10% FBS. The cells were then washed with Earl's balanced salt solution supplemented with 25 mM Hepes, pH 7.4, and 2 mg/ml bovine serum albumin (EHB). 125 I-TGF-␤1 (0.1 nM) was preincubated in 10% plasma from ␣ 2 M(Ϫ/Ϫ) or wild-type mice or in EHB without plasma for 30 min. The samples were then transferred to the AKR-2B cell cultures. Incubations were allowed to proceed for 4 h at 4°C. The cultures were then washed and incubated with 0.25 mM BS 3 for 20 min at 4°C. After washing again, the cells were detached by scraping, pelleted by centrifugation at 2500 ϫ g for 10 min, and resuspended in 10 mM Tris-HCl, 125 mM NaCl, 1 mM EDTA, 1.0% (v/v) Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 50 g/ml aprotinin, 100 g/ml soybean trypsin inhibitor, 100 g/ml benzamidine HCl, pH 7.4. Insoluble cellular debris was removed by centrifugation at 12,000 ϫ g for 15 min. Cell extracts were subjected to SDS-PAGE, under reducing conditions (4 -12% polyacrylamide gradient slab), and then to autoradiography.
Determination of Apparent Equilibrium Dissociation Constants-K D values for the binding of 125 I-TGF-␤1 and 125 I-TGF-␤2 to m␣ 2 M and MUG were determined by the BS 3 rapid cross-linking method, as described previously (34 -36). The K D values are based on a model in which the interaction of ␣-macroglobulin with TGF-␤ is initially noncovalent and reversible, with slow, irreversible conversion of noncovalent complex into covalent complex: and AC* is covalent ␣ 2 M-TGF-␤ complex. For TGF-␤ isoforms and h␣ 2 M, reversible binding occurs rapidly and k 2 is sufficiently small so that it may be ignored in the determination of K D values (34,37).
In the experiments performed here, various concentrations of native m␣ 2 M or MUG were incubated with 125 I-TGF-␤1 or 125 I-TGF-␤2 (1.0 nM) in phosphate-buffered saline with 75 M bovine serum albumin, at 37°C for 1 h. BS 3 in H 2 O, or an equal volume of H 2 O, was added to each tube for 1 min. The final concentration of BS 3 was 5 mM. Cross-linking was terminated instantaneously by acidification (34). Samples were denatured in 2.0% SDS and subjected to SDS-PAGE and autoradiography. Amounts of free 125 I-TGF-␤ and covalent m␣ 2 M-or MUG-TGF-␤ complex were determined by slicing the gels and measuring the radio-activity in each section. AC e is the concentration of BS 3 -stabilized, noncovalent m␣ 2 M-or MUG-TGF-␤ complex detected by PAGE (corrected for the presence of AC* by analysis of non-cross-linked samples). C e is the concentration of free TGF-␤ detected by PAGE (including free TGF-␤ and AC which is not cross-linked by BS 3 ). These values were plotted according to the following equation (34), which yields the K D and the BS 3 cross-linking efficiency, z, which is a constant (0ϽzϽ1): Apparent K D values, determined by this analysis, assume one TGF-␤binding site per ␣-macroglobulin. The K D values are "whole molecule" constants, accounting for both affinity and multiplicity of binding sites within a single binding protein, which is considered optimal for comparing two binding proteins. If there are two or four equivalent TGF-␤-binding sites per ␣-macroglobulin, then the "isolated site" K D is 2-or 4-fold higher than the reported K D .
Endothelial Cell Growth Assays-Fetal bovine heart endothelial (FBHE) cells were maintained in complete medium, as described previously (36). For experiments, cells were harvested in trypsin-EDTA and plated at 2 ϫ 10 4 cells/well in 24-well plates. After incubation for 15 h in complete medium, the cultures were washed and incubated with 10 pM TGF-␤1 or TGF-␤2, in the presence or absence of native h␣ 2 M or native m␣ 2 M (120 nM), in Dulbecco's modified Eagle's medium with 0.2% (dilute) FBS. Incubations proceeded for 30 h and then for an additional 18 h in the presence of [ 3 H]thymidine (1 Ci/ml). [ 3 H]Thymidine incorporation was determined as described previously (36). TGF-␤1 and TGF-␤2 are nearly equipotent inhibitors of FBHE growth when added in the presence of dilute FBS and in the absence of supplementary ␣ 2 M (36,38).
Nitric Oxide Synthesis Assay-RAW 264.7 cells were plated at 10 4 cells/well in 96-well plates and cultured, first in RPMI 1640 with 10% FBS for 24 h, and then in RPMI 1640 without FBS (SFM) for 24 h. Native h␣ 2 M, native m␣ 2 M, and MUG (each at 280 nM) were added separately to the cultures in SFM. After 24 h, conditioned culture medium samples were recovered. The stable NO oxidation product, nitrite, was measured in the conditioned medium samples, as described previously (14,39). The concentration of the iNOS substrate, arginine, in RPMI 1640, is 1.2 mM. We have previously shown that the increase in cellular NO synthesis, induced by h␣ 2 M, is inhibited by the NOS inhibitor, N G -monomethyl-L-arginine (14).

TGF-␤-binding Proteins in Murine
Plasma-125 I-TGF-␤1 and 125 I-TGF-␤2 were incubated with plasma from wild-type or ␣ 2 M(Ϫ/Ϫ) mice. BS 3 was then added to covalently stabilize 125 I-TGF-␤-native ␣-macroglobulin complexes, which are known to be primarily noncovalent (34), and/or other TGF-␤plasma protein complexes (Fig. 1). As a control, both 125 I-TGF-␤ isoforms were incubated for 30 min at 37°C in the absence of plasma. Free 125 I-TGF-␤ was recovered in a distinct band with an apparent mass of 25 kDa. After BS 3 treatment, additional free 125 I-TGF-␤ was detected diffusely across intermediate mobility positions (primarily by PhosphorImager analysis), suggesting some cross-linking of free 125 I-TGF-␤ into multimers. The extent of BS 3 -induced TGF-␤ multimer formation was not altered by adding bovine serum albumin (results not shown) and the mobilities of the resulting products did not overlap with that of purified ␣ 2 M, as previously reported (34).
When incubated with wild-type murine plasma, 125 I-TGF-␤1 and 125 I-TGF-␤2 were primarily recovered in two bands corresponding exactly in mobility to purified 125 I-m␣ 2 M. Free 125 I-TGF-␤ was still observed by SDS-PAGE, as expected since the efficiency of BS 3 cross-linking was less than 100%; however, TGF-␤ that was incubated with wild-type plasma did not form intermediate mobility products after BS 3 treatment. This result suggests that significant levels of free 125 I-TGF-␤ were not present in these samples and explains why the 125 I-TGF-␤ recovered in the 25-kDa band was not diminished to a greater extent after incubation with plasma.
When incubated with plasma from ␣ 2 M(Ϫ/Ϫ) mice, 125 I-TGF-␤1 and 125 I-TGF-␤2 did not form detectable complexes with MUG or any other plasma protein, suggesting that m␣ 2 M is the only significant TGF-␤-binding protein in murine plasma. Another explanation for these data is that TGF-␤1 and TGF-␤2 bind to a plasma protein other than m␣ 2 M, forming a complex that is not stabilized by the cross-linking reagent.
Affinity Labeling of TGF-␤ Receptors-As a second method for comparing the TGF-␤ binding activities of proteins in normal and ␣ 2 M-deficient mouse plasma, AKR-2B cells were affinity-labeled with 125 I-TGF-␤1 (Fig. 2). In the absence of plasma, the three well characterized TGF-␤ receptors in fibroblasts were identified, including ␤-glycan, a proteoglycan which migrates in a diffuse low-mobility band, the type II receptor, with an apparent mass of 75 kDa, and the type I receptor, with an apparent mass of 53 kDa (corrected for bound TGF-␤). When 125 I-TGF-␤1 was added to cultures in the presence of plasma from wild-type mice, labeling of all three of the TGF-␤ receptors was decreased. By contrast, ␣ 2 M(Ϫ/Ϫ) mouse plasma was ineffective at inhibiting 125 I-TGF-␤1 binding to cellular receptors. These studies confirm the role of m␣ 2 M, in wild-type mouse plasma, as a major TGF-␤-neutralizing protein. Other plasma proteins do not substitute for m␣ 2 M in this capacity.
Binding of TGF-␤1 and TGF-␤2 to Purified m␣ 2 M-The purified m␣ 2 M preparations used in this study consisted entirely of protein in the native conformation as determined by nondenaturing PAGE (Fig. 3). The mobility of the m␣ 2 M was increased after reaction with trypsin, reflecting the transition to the activated conformation (31).
Purified native m␣ 2 M bound 125 I-TGF-␤1 or 125 I-TGF-␤2, as determined by nondenaturing PAGE and autoradiography. The autoradiography bands completely aligned with the Coomassie-stained native m␣ 2 M bands, indicating that trace contamination of the m␣ 2 M preparation with activated forms was not responsible for the observed TGF-␤-binding. No attempt was made to quantitate the amount of 125 I-TGF-␤ recovered in association with the m␣ 2 M since recovery is influenced by Determination of Apparent K D Values-125 I-TGF-␤1 and 125 I-TGF-␤2 bound to purified m␣ 2 M and MUG, as determined by the BS 3 -rapid cross-linking method. In preliminary time course experiments (not shown), binding of the TGF-␤ isoforms to both ␣-macroglobulins maximized in less than 1 h. Fig. 4 shows representative autoradiographs in which 125 I-TGF-␤2 was incubated with different concentrations of m␣ 2 M or MUG. Significant amounts of TGF-␤-␣-macroglobulin complex were detected only after adding BS 3 . This result indicates that formation of covalent complex by thiol-disulfide exchange (AC* in Equation 1) was negligible, as expected since thiol-disulfide exchange requires free Cys residues which are absent in the native forms of m␣ 2 M and MUG (18). Minimal AC* formation (k 2 ) justifies the rapid equilibrium assumption in determining The results of six separate TGF-␤ equilibrium binding experiments were averaged to generate the graphs shown in Fig.  5. Using similar plots, separate K D values were determined from each individual study and averaged to obtain the mean K D values presented in Table I. The most important finding was the substantial difference in binding affinity of m␣ 2 M and MUG for the TGF-␤ isoforms. The K D values for MUG binding to TGF-␤1 and TGF-␤2 were 30-fold higher than the corresponding constants for m␣ 2 M. Thus, MUG is not a significant TGF-␤-binding protein capable of substituting for m␣ 2 M in ␣ 2 M(Ϫ/Ϫ) mice, as was suggested by our experiments with ␣ 2 M(Ϫ/Ϫ) mouse plasma.
A second unexpected finding was the high affinity of native m␣ 2 M for TGF-␤1. Studies of bovine ␣ 2 M (41) and h␣ 2 M (34,41) have shown that these ␣-macroglobulins selectively bind TGF-␤2 compared with TGF-␤1. The K D for the binding of TGF-␤2 to native h␣ 2 M is 11 nM, compared with a K D of 330 nM for the binding of TGF-␤1 to native h␣ 2 M (34). The ability of native m␣ 2 M to function as an equally effective binding protein for TGF-␤1 and TGF-␤2 suggests a potentially enhanced role for m␣ 2 M in the regulation of TGF-␤1 in this species.
FBHE Proliferation Assays-TGF-␤1 and TGF-␤2 are nearly equipotent inhibitors of FBHE proliferation when the cells are cultured in dilute FBS (36,38). ␣-Macroglobulins counteract the activity of TGF-␤ and promote FBHE growth to an extent that depends on the affinity of the TGF-␤/␣-macroglobulin interaction (36). Due to the unprecedented low K D determined for the interaction of native m␣ 2 M with TGF-␤1, FBHE studies were performed to compare the TGF-␤ neutralizing activities of native m␣ 2 M and native h␣ 2 M in a biological system. As shown in Table II  in its capacity to neutralize TGF-␤1.
Regulation of Nitric Oxide Synthesis by M␣ 2 M and MUG-Induction of NO synthesis by h␣ 2 M, m␣ 2 M, and MUG was compared in RAW 264.7 cells (Fig. 6). Each ␣-macroglobulin was present at the same concentration (280 nM); however, only h␣ 2 M and m␣ 2 M significantly increased nitrite levels in the conditioned media. The responses elicited by m␣ 2 M and h␣ 2 M were equivalent. By contrast, MUG did not cause a detectable increase in NO synthesis. We have previously shown that h␣ 2 M, at 280 nM, induces a response comparable to that observed with 10 ng/ml interferon-␥ (14), a well known stimulator of macrophage iNOS (42,43). In further control experiments, the ␣-macroglobulin preparations were boiled for 20 min and then added to the RAW 264.7 cell cultures. The boiled preparations did not increase NO synthesis, eliminating the possibility that endotoxin contributed to the observed responses. These studies demonstrate that the induction of NO synthesis, like the ability to neutralize TGF-␤, is an activity of m␣ 2 M which is non-redundant among murine ␣-macroglobulins.

DISCUSSION
TGF-␤ functions within the context of autocrine pathways to regulate many of the properties of cells in culture, especially when the culture medium is not serum-supplemented. In colon carcinoma cells, autocrine TGF-␤ increases integrin ␣ 5 gene expression (44); in rat VSMCs, autocrine TGF-␤ suppresses platelet-derived growth factor ␣-receptor expression (45) and in RAW 264.7 cells, autocrine TGF-␤ suppresses iNOS expression (14). In each of these studies, the importance of autocrine TGF-␤ was demonstrated using a pan-isoform TGF-␤-neutralizing antibody (14,45) or a constitutively expressed TGF-␤ antisense cDNA construct (46). H␣ 2 M binds a number of cytokines (34); however, in the absence of exogenously added cytokines, many of the activities of h␣ 2 M in cell culture may be attributed to specific neutralization of TGF-␤ (14,45). Whether ␣-macroglobulins counteract TGF-␤ activity in vivo is undetermined.
Gene knock-out experiments in mice have demonstrated that TGF-␤1 functions in vivo to provide homeostatic suppression of immune/inflammatory responses (47). Endotoxin challenge causes a form of inflammatory response which can be lethal. iNOS(Ϫ/Ϫ) mice are insensitive to endotoxin challenge (12), reflecting the important role played by iNOS in mediating endotoxin-associated shock (5-9). Since ␣ 2 M(Ϫ/Ϫ) mice also demonstrate decreased sensitivity to endotoxin (23), we decided to examine the effects of the murine ␣-macroglobulins on TGF-␤ activity and iNOS activity. Although normal murine plasma has two major members of the ␣-macroglobulin family, only m␣ 2 M binds TGF-␤1 and TGF-␤2 with significant affinity. Murine ␣ 2 M, but not MUG, inhibits TGF-␤ binding to its cellular receptors and the biological response of FBHE cells to TGF-␤. Furthermore, among the murine ␣-macroglobulins, only m␣ 2 M induces macrophage NO synthesis. Thus, while the proteinase-inhibitory specificities of m␣ 2 M and MUG are at least partially redundant, the growth factor-carrier activities of these two proteins are not. In ␣ 2 M(Ϫ/Ϫ) mice, MUG is expressed at normal levels and, during pregnancy, at increased levels (23); however, as shown with in vitro experiments in Figs. 1 and 2, neither this protein nor any other protein in m␣ 2 M-deficient plasma substitutes for m␣ 2 M as a TGF-␤ carrier and neutralizer. The studies described here suggest that m␣ 2 M, in wild-type mice, may counteract TGF-␤ locally, at sites of inflammation, and thereby permit augmented iNOS expression in response to endotoxin. Further in vivo testing is warranted to test this hypothesis.
Our analysis of NO synthesis was performed using RAW 264.7 cells. Although this is a passaged cell line, RAW 264.7 cells demonstrate excellent conservation of differentiated macrophage properties and can be primed/activated for tumorcidal and bacteriocidal activity, similarly to primary cultures of macrophages (48,49). Human macrophages, like the RAW 264.7 cells, secrete TGF-␤ (50) which functions as a potent inhibitor of human macrophage activation (51,52). Thus, RAW 264.7 cells provide an accurate and reproducible model for studying the responses of normal macrophages. Danielpour and Sporn (41) performed experiments with serum and ␣ 2 M, purified from human and bovine plasma, to demonstrate selective binding of TGF-␤2, compared with TGF-␤1. The K D for TGF-␤1 binding to native h␣ 2 M is 30-fold higher than the K D for the TGF-␤2/native h␣ 2 M interaction (34). Thus, the ability of m␣ 2 M to bind TGF-␤1 and TGF-␤2 with equal affinity is novel, compared with other characterized ␣-macroglobulins. In addition to m␣ 2 M and h␣ 2 M, we have also studied the binding of TGF-␤1 to rat ␣ 1 M, the constitutively synthesized homologue of h␣ 2 M, and to rat ␣ 2 M, an acute-phase reactant (53). Based on these studies, we can rank order the ␣-macroglobulins according to their affinities for TGF-␤1 as follows: m␣ 2 M Ͼ rat ␣ 2 M Ͼ rat ␣ 1 M Ն h␣ 2 M Ͼ Ͼ MUG.
In our analysis of TGF-␤ interactions with m␣ 2 M and MUG, we focused on the native forms of each protein. Activated ␣-macroglobulins are rapidly taken up by cells that express LRP (17,31). As a result, levels of activated ␣-macroglobulins are negligible in the plasma under most circumstances. In the pericellular spaces, higher concentrations of activated ␣-macroglobulins may accumulate unless the tissue includes cells that express LRP. One concern regarding our NO synthesis experiments was whether the RAW 264.7 cells might secrete sufficient levels of proteinases to convert the ␣-macroglobulins into activated forms during the 24-h incubation. We previously demonstrated that this does not occur in cultures of vascular smooth muscle cells (45). To test for proteinase secretion, we incubated radiolabeled native h␣ 2 M with the RAW 264.7 cells for 24 h. The h␣ 2 M was then subjected to nondenaturing PAGE. We also measured cell-associated radioactivity and the amount of trichloroacetic acid-soluble radioactivity in the medium. None of our tests revealed detectable conversion of h␣ 2 M into the activated conformation (results not shown). Thus, the induction of NO synthesis by native m␣ 2 M and h␣ 2 M does not occur subsequent to proteinase modification of these proteins.
When challenged with bleomycin, the lungs of ␣ 2 M(Ϫ/Ϫ) mice were not substantially affected while the lungs of wildtype mice developed inflammatory infiltrates and early connective tissue deposition (23). TGF-␤, which is secreted by alveolar macrophages and activated locally, has been implicated in the regulation of this process (54). It is intriguing to speculate that the ␣ 2 M(Ϫ/Ϫ) mice are bleomycin tolerant due to enhanced function of TGF-␤ as a general anti-inflammatory; however, TGF-␤ may actually promote certain aspects of the bleomycin response, such as the deposition of extracellular matrix proteins (55). Understanding the relationship between TGF-␤, ␣ 2 M, the development of immune cell infiltrates, and fibrosis, will require further investigation.
In addition to its ability to neutralize TGF-␤ locally, ␣ 2 M serves as a major carrier of TGF-␤ in the plasma, delaying TGF-␤ clearance and providing a potential pool of reversibly bound active factor (56 -58). The unique and non-redundant activity of ␣ 2 M as a TGF-␤ carrier in murine plasma suggests that TGF-␤ plasma pharmacokinetics may be altered in ␣ 2 M(Ϫ/Ϫ) mice, in addition to TGF-␤ activity. In our cell culture systems, we have shown that m␣ 2 M induces iNOS activity in macrophages while MUG does not. These in vitro results provide a model for explaining some of the phenotypic properties of the ␣ 2 M(Ϫ/Ϫ) mouse.