Isolation and characterization of a macrophage-derived heparin-binding growth factor

Human mononuclear cells were plated in culture, and the conditioned media of these cells were analyzed by heparin-Sepharose affinity chromatography. The fractions were tested for growth factor activity as measured by the stimulation of DNA synthesis in BALB/c 3T3 cells. After 2 d in culture, two peaks of heparin-binding growth factor (HBGF) activity were detected, one eluting with 0.5 M NaCI, which could be shown to be platelet-derived growth factor (PDGF)-like, and the other eluting with 1.0 M NaCI. After 7-11 d in culture, when monocytes had clearly differentiated into macrophages, >95% of the HBGF activity in conditioned medium consisted of the 1.0 M NaCI elution peak. This activity, which was designated macrophage-derived HBGF (MD-HBGF), was found to be a cationic heat-resistant polypeptide with a molecular weight in the range of 14-25 kDa. Analysis using Western blots and specific neutralizing antisera, as well as compara-tive heparin affinity analysis, indicated that MD-HBGF was not identical to other heparin-binding 3T3 cell growth factors known to be produced by macrophages, such as PDGF (AB, AA, and BB forms), acidic fibroblast growth factor, and basic fibroblast growth factor. In addition to stimulating mitogenesis in 3T3 cells, MD-HBGF also stimulated the proliferation of vascular smooth muscle cells, but did not stimulate the proliferation of vascular endothelial cells. units of growth factor activity present at that time in 500 ml of 1 x conditioned medium. The same results were repeated in a second experiment.


Introduction
Macrophages are responsible for many important functions including host defense, acute and chronic inflammation, and wound healing. Some of the biological activities of macrophages, such as the stimulation of connective tissue cell pro-* Corresponding author. liferation after injury, are thought to be mediated by macrophage-derived growth factors. Macrophages produce many growth factors, possibly more than any other cell type. Among the well-characterized macrophage-derived growth factors are platelet-derived growth factor (PDGF) (Shimokado et al., 1985;Mornex et al., 1986), basic fibroblast growth factor (bFGF) (Baird et al., 1985), transforming growth factoralpha (TGF-a) (Madtes et al., 1988), transforming growth factor-beta (TGF-fl) (Assoian et al., 1987), interleukin-1 (IL-1) (Hazuda et al., 1988;Schmidt et al., 1984), and tumor necrosis factoralpha (TN F-a) (Leibovich et al., 1988). Additional macrophage-derived growth factors have been reported but they have yet to be fully characterized and purified (Madtes et aL, 1988;Singh et al., 1988). With the exception of bFGF and IL-1, macrophage-derived growth factors are secreted and act ostensibly in a paracrine manner. bFGF is known to stimulate processes such as formation of granulation tissue and angiogenesis (Davidson et al., 1985;Shing et al., 1985;Folkman and Klagsbrun, 1987;Klagsbrun, 1989). However, as a cell-associated growth factor, the mechanism by which macrophagederived bFGF could stimulate these processes is unclear. Alternatively, these processes may be induced by other macrophage-derived growth factors, either characterized, e.g., TGF-/3, PDGF, or as yet uncharacterized.
Heparin affinity chromatography has been used widely as a method for purifying and characterizing growth factors, in particular, members of the FGF family which have a strong affinity for heparin (Shing et al., 1984;Klagsbrun and Shing, 1985;Folkman and Klagsbrun, 1987;Klagsbrun, 1989). Acidic FGF (aFGF) and bFGF bind to immobilized heparin columns and are eluted with 1-1.2 and 1.5-1.8 M NaCI, respectively. Several growth factors structurally homologous to aFGF and bFGF, including hst/Kfgf and KGF have an affinity for heparin as well (Delli-Bovi et al., 1988;Finch et al., 1989). PDGF also binds to immobilized heparin but with relatively low affinity and is eluted with 0.5 M NaCI, typical of cationic proteins in general (Shing et al., 1984). Given the recent success of heparin relatively little growth factor activity, .2% as 8 / \ much, was found in cell lysates. When mononuclear cell CM was harvested and analyzed at 48-h intervals, growth factor activity in CM was found to fluctuate with time in culture ( Figure  6 1). Growth factor activity was relatively high 2 d after plating, decreased at 4-6 d, increased at 8-12 d, and slowly decreased after that. The growth factor activity found a week after plating and beyond was strictly macrophage derived 4 since the nonadherent cells had been removed, and the remaining cells were 99.8% pure macrophages as assessed by staining for a-naphthyl acetate esterase (not shown). 2 CM collected after plating of mononuclear cells were analyzed by heparin affinity chromatography ( Figure 2). At 2 d, two peaks of heparin-binding growth factor (HBGF) activity were found in CM, one eluting at 0.5 M NaCI 0 4 8 12 16 20 and the other at 1 M NaCI ( Figure 2A). The relative amounts of these growth factors varied in TIME (days) the different preparations that were obtained from the blood of various hospital volunteers.
1. Time course of growth factor secretion by hu- The units of mitogenic activity in the peak eluting nononuclear cells. Human peripheral blood monoat 0.5 M NaCI were in the range of one to five kr t-, lia VVrI mG int nne nlne-G l in e all ti ltiira n.nwe FlUUMM:d LAPll5 Wt:lt: I:*UldLtUU dllU VIUta:U III UVII LUIUFtu:. %.u11ditioned medium (CM) was harvested every 2 d and assayed for stimulation of DNA synthesis in BALB/c 3T3 cells. Each point (in quadruplicate) represents the total number of units of growth factor activity present at that time in 500 ml of 1 x conditioned medium. The same results were repeated in a second experiment. affinity chromatography in isolating novel growth factors such as KGF (Finch et al., 1989) and vascular endothelial growth factor (VEGF) (Ferrara and Henzel, 1 989; Tischer et a/., 1 989), we analyzed monocyte/macrophage-conditioned medium for the presence of secreted heparin-binding growth factors. In this report, we demonstrate that short-and long-term cultures of adherent macrophages produce a heparin-binding growth factor mitogenic for 3T3 and smooth muscle cells but not for endothelial cells. This macrophage-derived heparin-binding growth factor (MD-HBGF) is neither PDGF nor FGF and may be novel.

Results
Secretion of heparin-binding growth factors by macrophages Conditioned media (CM) obtained from 1-2-dold cultures of human mononuclear cells contained high levels of growth factor activity (-2-10 U/ml), as measured by the stimulation of The positive control for BALB/c 3T3 is 20% serum (135 000 cpm), for smooth muscle cells it is 20% bovine serum (51 000 cpm), and for endothelial cells it is 2 ng/ml basic FGF (42 000 cpm).
times that of the peak eluting at 1 M NaCI. By 7-11 d after plating, the macrophages secreted the HBGF eluting at 1 M NaCI almost exclusively ( Figure 2B), and this pattern persisted thereafter. Cells grown in serum-free medium for 72 h also contained two HBGFs eluting from a heparin affinity column at 0.5 and 1 M NaCI showing that these HBGFs were macrophage-derived and not serum-derived. Growth factor activity that did not bind to heparin was also measured at intervals following plating. At 2 d, non-HBGF activity constituted -650/o of the total activity, but by day 12, non-HBGF was only -5%/o of the total (data not shown). Taken together, these results suggested that as the cultured monocytes differentiated into macrophages, most of their secreted growth factor activity, as measured by the stimulation of 3T3 DNA synthesis, was comprised of the HBGF eluting at 1 M NaCI. The fractions collected after heparin affinity chromatography of macrophage CM (7-11 d in culture) were also analyzed for their ability to stimulate smooth muscle cell and endothelial cell proliferation (Figure 3). The HBGF eluting at 1 M NaCI, previously shown to be mitogenic for 3T3 cells, was also mitogenic for smooth muscle cells but was not mitogenic for endothelial cells that were responsive to basic FGF. A dose-dependent stimulation of smooth mus-cle cell proliferation by the HBGF eluting at 1 M NaCI is shown in Figure 4. Analysis of macrophage CM for PDGF and FGF PDGF, aFGF, and bFGF have previously been shown to be HBGFs (Klagsbrun, 1989;Shing et aL, 1984;Klagsbrun and Shing, 1985). In addition, macrophages synthesize PDGF (Shimokado etaaL, 1985;Mornex etal., 1986) and bFGF (Baird et al., 1985). To characterize the macrophage CM-derived HBGFs in a preliminary way, their heparin column elution profiles were compared with those of PDGF, aFGF, bFGF, and other growth factors previously shown to be produced by macrophages ( Figure 5). TGF-a (5F), TGF-fl (5E), and IL-1 (5G) did not bind to heparin, TNF-a (5H) was eluted with 0.1-0.2 M NaCI, PDGF (5B) (AB, AA, or BB forms) was eluted with 0.5 M NaCI, aFGF (5C) with 1.5 M NaCI, and bFGF (5D) with 2 M NaCI. Of the two HBGFs secreted by macrophages (5A), the one eluting at 0.5 M NaCI coincided with PDGF (5B), whereas the HBGF eluting at 1 M NaCI did not coincide with any other of the growth factors. Antisera that inhibited the mitogenic activity of human PDGF (whether AB, AA, or BB dimeric forms) for BALB/c 3T3 cells inhibited the mitogenic activity of the HBGF eluting with 0.5 M NaCI but not of the HBGF eluting at 1 M NaCI (Table 1). These results suggested that the macrophage-derived HBGF eluting at 0.5 M NaCI was PDGF-like, whereas the HBGF eluting at 1 M NaCI was not.
bFGF and aFGF are cell-associated growth factors and are not secreted (Folkman and  . Nevertheless, the fractions collected after heparin affinity chromatography of macrophage CM were tested for the presence of aFGF and bFGF. The fractions were not mitogenic for endothelial cells (Figure 3), suggesting the absence of bFGF or aFGF, both potent endothelial cell mitogens. In a more direct analysis, it was found that neither the HBGF eluting at 0.5 nor 1 M NaCI was inhibited by anti-bFGF neutralizing antibodies (Table 1).
aFGF neutralizing antibodies were not available for such analysis. However, Western blot analysis of 100 U of HBGF eluting at 1 M NaCI using antisera that can easily detect 10 ng (20-50 U) amounts of acidic FGF and basic FGF, clearly indicated that it was neither aFGF nor bFGF ( Figure 6). Taken together, the lack of endothelial cell mitogenic activity and of interaction with anti-aFGF or anti-bFGF antibodies strongly suggested that macrophages did not secrete biologically active or immunoreactive FGF-like growth factors.

Biochemical characterization
The macrophage-derived HBGF eluting with 1.0 M NaCI was partially characterized. Incubation with 50 ,ug/ml trypsin for 30 min destroyed >960/o of the mitogenic activity of this growth factor. Both HBGF eluting at 0.5 M NaCI (PDGF) and HBGF eluting at 1 M NaCI were exposed to 100°C for 10 min and to 5 mM dithiothreitol (DTT). Both HBGF were resistant to the heat treatment. On the other hand, the HBGF eluting with 1.0 M NaCI was resistant to inactivation by DTT under conditions that inactivated PDGF (not shown). The macrophage-derived HBGF eluting with 1.0 M NaCI adhered to the cation exchange column Mono-S and was eluted with 0.7 M NaCI, suggesting it to be a highly cationic protein ( Figure 7A). This HBGF was shown to have a molecular mass of 14-25 kDa by gel filtration chromatography ( Figure 7B). A combination of heparin-Sepharose, Mono-S, and gel filtration chromatography (each of which has about a 20-300/o recovery) was insufficient to purify the higher affinity HBGF.

Discussion
Conditioned media of human peripheral blood monocytes cultured for 1-2 d contain two dis-CELL REGULATION F. Heparin-binding growth factors eluting at 0.5 and 1.0 M NaCI (HBGF-0.5 and HBGF-1.0, respectively) from a heparin affinity column (Figure 2A) were tested for their ability to stimulate DNA synthesis in BALB/c 3T3 cells in the presence (+) or absence (-) of blocking antisera to human PDGF (1: 1000 dilution) or human bFGF (1 :10 dilution). Control wells received the same concentration of antisera plus either 0.5 ng/ml human PDGF or 0.5 ng/ml bFGF. The data shows the mean ± SEM of triplicate determination of [3H]thymidine incorporation (in cpm), which in the absence of any addition of growth factor was 1000 ± 100 cpm. Nonimmune serum gave the same results as absence of neutralizing antisera. tinct growth factors that bind to heparin. One of these HBGFs elutes from immobilized heparin with 0.5 M NaCI and can be shown to be PDGFlike, confirming earlier reports of macrophagederived PDGF (Shimokado et al., 1985;Mornex et al., 1986). The evidence that the HBGF eluting at 0.5 M NaCI is PDGF-like includes coelution with human PDGF on immobilized heparin columns, neutralization of growth activity by antisera that neutralize PDGF, inactivation by disulfide reducing agents, and resistance to heating at 100°C. It is not known at present whether the macrophage-derived PDGF is the AB, AA, or BB heterodimeric form. The second HBGF has a stronger affinity for heparin and is eluted from immobilized heparin with 1 M NaCI. It is not neutralized by anti-PDGF antisera, suggesting that it is not a PDGF-like growth factor. At 1-2 d in culture, it is difficult to ascertain the cellular origin of these two HBGFs because at this point mononuclear cell preparations are contaminated by other cells such as platelets, which are known producers of growth factors such as PDGF (Shimokado et aL, 1985;Mornex et al., 1986) and platelet-derived endothelial cell growth factor (PD-ECGF) (Ishikawa et al., 1989). In general, there is about one to five times more PDGF compared with the higher affinity-binding HBGF in the conditioned medium of 1-2-d-old mononuclear cells. The variation might be due to nonmononuclear cell contamination. On the other hand, after longer term culture of 1 wk and beyond, when the cell population is virtually all adherent macrophages and potential contaminating cells are absent, almost of all the growth factor activity secreted into conditioned medium is due to the HBGF that has a higher affinity for heparin than PDGF. The decrease in PDGF activity with time in culture is of interest. PDGF found in early cultures may be produced by contaminating cells and removal of these cells would diminish PDGF levels. Alternatively, PDGF secretion may decrease as monocytes differentiate into macrophages in the absence of any inducer of PDGF synthesis. Although, it can not be definitively demonstrated that the HBGF eluting at 1 M NaCI is synthesized by monocytes, it certainly appears to be synthesized by adherent macrophages. The mechanism of activation of this HBGF is unclear at this point. Its synthesis is not stimulated any further by treatment of macrophages with PMA but other activating factors can not be ruled out.
anti-aFGF anti-bFGF silver 1 2 3 4 5 Figure 6. Western blot analysis with anti-FGF antibodies. Macrophage-derived HBGF eluting from heparin-Sepharose at 1 M NaCI as shown in Figure 2B was analyzed by SDS-PAGE. Western blot analysis was carried out using antiacidic FGF antisera (lanes 1 and 2) and anti-basic FGF antisera (lanes 3 and 4). Silver stain was used to confirm that the sample of macrophage-derived HBGF contained significant amounts of protein (lane 5). The following samples were applied: lane 1, 12 ng bovine brain acidic FGF; lanes 2, 4 and 5, 100 U of macrophage-derived HBGF eluting at . Biochemical characterization of macrophagederived heparin-binding growth factor. Macrophage-derived HBGF eluting from heparin affinity columns with 1 M NaCI ( Figure 2B) was (A) applied to an FPLC Mono-S column and eluted with a gradient of 0.02-1 M NaCI, or (B) applied to a Sephadex G75 column equilibrated in 0.6 M NaCI and 0.02 M Tris-HCI, pH 7.5. Fractions were collected (0.5 ml for Mono S and 1 ml for G-75) and tested (5 and 10 ,ul aliquots, respectively) for the ability to stimulate 3T3 cell DNA synthesis. The recovery from the Mono-S column was 506 units of growth factor activity out of the 2400 U applied (21%). The recovery from the G-75 column was 146 U of growth factor activity out of the 500 U applied (29%).
Because of its secretion by cultures of adherent macrophages, we have designated the HBGF eluting from heparin columns with 1 M NaCI as MD-HBGF, which is a cationic, 14-25 kDa polypeptide mitogenic for 3T3 cells and smooth muscle cells but not endothelial cells. It appears to be heat stable and is not inactivated by DTT. The affinity of MD-HBGF for heparin suggests that it may be related to aFGF and bFGF. However, this is highly unlikely since MD-HBGF is secreted, is not an endothelial cell mitogen, is not neutralized by anti-bFGF antisera, and is not immunoreactive with anti-aFGF or anti-bFGF antisera. It is also unlikely that MD-HBGF is another member of the FGF family, e.g., int-2, hstlKfgf, FGF 5, and KGF. int-2 has not been shown to be mitogenic for any cell type, hst/Kfgfand FGF 5 are mitogenic for endothelial cells, and KGF is specific for epithelial cells (Delli-Bovi et al., 1988;Finch et al., 1989;Zhan et al., 1988). Other known macrophage-derived growth factors differ from MD-HBGF in that either they do not bind to heparin, e.g., TGF-f, TGF-a, TNF-a ( Figure 5); they are endothelial cell mitogens, e.g., TGF-a (Schreiber et al., 1986); or they are endothelial cell inhibitors (TGF-3, TNF-a) (Baird and Durkin, 1986;Heimark et al., 1986;Frater-Schroder et aL., 1987;Muller et aL, 1987;Schweigerer et aL, 1987). The macrophage inflammatory proteins (MIP), although heparin-binding, are low molecular weight (8 kDa) anionic polypeptides that are pyrogenic and chemotactic for leukocytes but do not appear to be mitogenic (Wolpe et aL., 1988;Wolpe et aL., 1989). The monocyte-derived growth factor described by Singh and Bonin (1988) is an anionic 40-kDa polypeptide. Given all the evidence so far, direct and circumstantial, MD-HBGF does not appear to be one of the wellcharacterized macrophage-derived growth factors and may indeed be novel. However, in the absence of sequence data, we can not yet rule out that MD-HBGF may be a precursor or an altered form of one of the well-characterized macrophage-derived growth factors.
The physiological role of MD-HBGF remains to be determined but two possibilities are wound healing and smooth muscle cell proliferation in vivo. The ability of MD-HBGF to stimulate fibroblast growth suggests a possible role in macrophage-mediated formation of granulation tissue, whereas the ability to stimulate aortic smooth muscle cell growth suggests a possible role in blood vessel development. It has been suggested that the smooth muscle cell hyperplasia associated with atherosclerosis is induced by a smooth muscle mitogen, possibly of monocyte/macrophage origin (Ross, 1986;Schwartz and Reidy, 1987). Furthermore, it has been suggested that this mitogen is PDGF, which is produced by macrophages (Shimokado et al., 1985). However, we find that PDGF production diminishes relative to MD-HBGF as monocytes differentiate into macrophages. Thus, MD-HBGF might be an attractive candidate as a macrophage-derived stimulator of smooth muscle cell proliferation, both in normal physiology and in pathology. The affinity of MD-HBGF for heparin might be significant. Heparan sulfate proteoglycans (HSPG) are found in blood vessels and have been shown to be low affinity binding sites for bFGF (Vlodavsky et aL., 1987;Moscatelli, 1988;Bashkin et aL., 1989). We speculate that MD-HBGF could also bind to HSPG in blood vessels. Perhaps, the combination of heparin-binding growth factors such as bFGF or aFGF, which are endothelial cell mitogens, and MD-HBGF which is a smooth mus-cle cell mitogen, could act as stimuli for blood vessel development.

Materials and methods Materials
Bovine aFGF was a gift from Dr. Pat D'Amore, The Children's Hospital, Boston, MA. Recombinant human bFGF was supplied by the Takeda Chemical Company, Osaka, Japan. Human PDGF was obtained from PDGF Inc., Boston, MA. Recombinant homodimer and heterodimer forms of human PDGF (AA, AB, and BB), TGF-a and EGF were kindly supplied by Dr. Mark Charette, Creative Biomolecules, Hopkinton, MA. lodinated human granulocyte macrophage-colony stimulating factor (GM-CSF) was supplied by Dr. Colin Seiff, Dana-Farber Cancer Institute, Boston, MA. lodinated porcine TGF-# was obtained from R & D Systems, Minneapolis, MN. Human cell line-derived IL-1 and recombinant human TNF-a were obtained from Genzyme, Boston, MA. Neutralizing antibodies to bFGF (Kurokawa et al., 1989) were prepared in rabbits immunized with human recombinant bFGF. Neutralizing antibody to PDGF was obtained from Collaborative Research (Bedford, MA).

Preparation of macrophage-conditioned medium
Buffy coats prepared from whole blood donated by healthy human volunteers were obtained from The Children's Hospital, Boston, MA, layered onto Ficoll-Hypaque (Pharmacia, Piscataway, NJ), and centrifuged at 4 x g for 20 min at room temperature, using a slight modification of the technique of Boyum (1968). Mononuclear cells located at the Ficoll-Hypaque/PBS interface were harvested, washed twice in phosphate-buffered saline (PBS), and resuspended in RPMI 1640 medium (Gibco Inc., Grand Island, NY), containing 1 Yo GPS (glutamine, 29.2 mg/ml; penicillin-G, 10 000 U/ml; streptomycin sulfate, 10 000 ug/mI) (Irvine Scientific, Santa Anna, CA), to a final concentration of 1 x 107 cells/ml. Cells (50 ml, 5 x 108 cells/flask) were plated in T 150 culture flasks in RPMI 1640 containing 1/0% fetal calf serum (FCS) and 1% GPS and incubated at 370C. Conditioned medium was harvested every 2 d and replaced with fresh medium. Once macrophages were adherent to the tissue culture flasks (typically 5-7 d), nonadherent cells were washed away, yielding purified macrophage cultures. Cells were maintained in culture in this way for up to 4 wk, with continuous secretion of growth factors.

Cell proliferation
The proliferation of BALB/c 3T3 cells, bovine capillary endothelial cells, and bovine aortic smooth muscle cells was assayed by measuring the incorporation of [H]-thymidine into DNA. BALB/c 3T3 cells. Measurement of DNA synthesis in 3T3 cells was performed as previously described (Shing et al., 1984;Klagsbrun and Shing, 1985). Briefly, BALB/c 3T3 cells were plated in 96 well plates (20 000 cells/well, 6.4 mm in diameter per well, 200 Ml of medium). After 7 days, the cells formed a confluent quiescent monolayer. Samples (from 1-50 Al) were added along with 3[H]-thymidine (1 MCi/well,4 MCi/mI final concentration, 6.7 Ci/mmol, ICN). After 36-48 h with no change of media, cells were fixed with methanol, DNA was precipitated with 50/o trichloroacetic acid, and 0.3 M NaOH was added to solubilize the cells, which were then transferred to scintillation vials. One unit of 3T3 stimulation activity is defined as the amount of growth factor required to stimulate half-maximal DNA synthesis in BALB/c 3T3 cells. Capillary endothelial cells (CEC). CEC (Shing et al., 1984;Klagsbrun and Shing, 1985) were grown in DMEM, 100/o bovine calf serum (BCS), 1 % GPS, and plated into gelatinized plates (48 well, 1 1.3 mm diameter, 1 X 104 cells/well). After 24 h, the CEC were refed with DMEM, 20/o BCS, 0.5% bovine serum albumin (BSA), and 1 gcM cold thymidine and incubated for another 24 h. Samples containing growth factor diluted in 0.2%o BSA/PBS were added, and 18 h later the cells were pulsed with [HJ-thymidine (0.6 ACi/well) for 4-6 h. Tritiated thymidine incorporation into CEC DNA was measured in the same manner as 3T3 cells (Shing et al., 1984;Klagsbrun and Shing, 1985). Bovine aortic smooth muscle cells (ASMC). ASMC were obtained from Dr. H. Weich (Children's Hospital, Boston, MA). The ASMC were grown in Dulbecco's modified Eagle's medium (DMEM) containing 100/ BCS and 1Yo GPS and plated in 48-well plates at 1 x 104 cells/well. The cells were grown for 5 d (with refeeding on day 2), at which point they were confluent. On day 5 the medium was changed to DMEM containing 2%/ BCS and 1% GPS and cells were incubated for another 24 h. Growth factor samples were added and after 18 h of incubation, the ASMC were pulsed for 6 h with [3H]-thymidine (1 ,uCi/well) and incorporation of labeled thymidine was measured as for 3T3 cells and CEC. ASMC proliferation was also determined by counting cell number. In these experiments, ASMC were plated sparsely in 24-well plates in DMEM containing 10% BCS and 10/% GPS at 104 cells/well. Growth factor samples were added and after 3 d cells were removed by trypsinization and counted in a Coulter counter (Coulter, Hialeah, FL).

Chromatography
Heparin-Sepharose chromatography. CM was collected and centrifuged at 15 000 x g for 30 min at 40C. 500-1000 ml of 1 x CM was applied to a 2-ml column of heparin-Sepharose (H-S) (Pharmacia, Uppsala, Sweden), pre-equilibrated with 0.2 M NaCI, 0.01 M tris(hydroxymethyl)aminomethane (Tris)-HCI, pH 7.5. The column was washed with 10 column volumes of equilibration buffer, and then eluted with an 80 ml gradient of 0.2-2.0 M NaCI, 0.01 M Tris-HCI, pH 7.5 at 40C. (When indicated, gradients of 0.1-3.0 M NaCI were used.) Fractions (2.5 ml) were collected and assayed for stimulation of [3H]-thymidine incorporation in BALB/c 3T3 cells. TSK-heparin FPLC. Samples were applied to a TSK gel heparin-5PW column (7.5 cm x 8 mm) (TosoHaas Corp., Woburn, MA), using a Pharmacia FPLC system (Pharmacia LKB Biotechnology, Uppsala, Sweden). Elution was performed with a linear gradient of 0.2-2.0 M NaCI, and fractions (0.5 ml) were collected and assayed for biological activity. Gel filtration chromatography. Active fractions from the 1.0 M NaCI elution peak of H-S were pooled and concentrated using Centri 10 concentrators followed by Centricon 10 microconcentrators (Amicon Co., Danvers, MA) to a final volume of 200 ,l. The retenate was then applied to a Sephadex G-75 size exclusion column (Pharmacia), pre-equilibrated with 0.6 M NaCI, 0.02 M Tris-HCI, pH 7.5. Fractions (1 ml) were collected and assayed for 3T3 cell mitogenic activity. Cation exchange chromatography. Active fractions from the 1.0 M NaCI elution peak of H-S were pooled, diluted with 0.01 M Tris-HCI, pH 7.5 to reduce the NaCI concentration to 0.2 M NaCI, and applied to a Mono S cation exchange FPLC column (Pharmacia), previously equilibrated with 0.02 M 2(n-morpholino)ethanesulfonic acid (MES), pH 6. The column was washed with equilibration buffer and eluted with a linear gradient of 0.2-1.0 M NaCI in 0.02 M MES, pH 6. Fractions (0.5 ml) were collected and assayed for 3T3 cell mitogenic activity.

Biochemical characterization
Macrophage-derived HBGF partially purified by heparin-Sepharose chromatography was treated as follows: 1) incubation with 50 Ag/ml pancreatic trypsin (Sigma) in PBS for 30 min at 37°C; before bioassay on BALB/c 3T3 cells, trypsin was inactivated by addition of 140 Ag/ml soybean trypsin inhibitor; macrophage-derived HBGF in the absence of trypsin was not inactivated under these conditions; 2) heating to 100°C for 10 min; and 3) incubation with 5 mM DTT (Sigma).