Factors regulating macrophage production and growth: identity of colony-stimulating factor and macrophage growth factor.

The activities of a colony-stimulating factor (CSF), which stimulates granulocyte-macrophage colony formation by mouse hemopoietic cells, and macrophage growth factor (MGF), which stimulates proliferation of activated peritoneal macrophages, have been demonstrated by various criteria to reside in the same molecular species. These criteria include occurrence in various sources and copurification of the activities in mouse L-cell-conditioned medium as well as the biological, physicochemical, and antigenic properties of the activities of L-cell-conditioned medium. CSF and MGF activities of L-cell-conditioned medium are ascribable to a glycoprotein of mol wt approximately 60,000 which migrates electrophoretically with alpha-globulin. Human urinary CSF, which also possesses MGF activity, has similar properties and can be neutralized by antiserum to highly purified L-cell medium CSF. A procedure is described for the partial purification of material from L-cell medium that has activity at 1 ng/ml in both MGF and CSF assays.


FACTORS REGULATING MACROPHAGE PRODUCTION AND GROWTH
MGF in L-cell-conditioned medium is resistant to inactivation by a variety of enzymes including nucleases and proteases and to heat treatment with periodate (3). CSF has been found in similar sources and also in the serum and urine of mouse and man. CSFs in the medium conditioned by L cells (11), mouse embryo fibroblasts (12), or human urine (8) share the properties mentioned above for MGF. In addition, it has been shown that CSFs from the latter sources are sialic acid-containing glycoproteins of mol wt 45,000-70,000 (8) with ion-exchange and gel-filtration characteristics similar to MGF (4,8,11,12).
In view of these physical similarities and because of the early appearance of macrophages in mouse colonies stimulated by CSF from these sources (13), it was of interest to determine whether the MGF and CSF assay systems were detecting the same molecule(s). Part of this interest stems from the observations that CSF and MGF are produced in many immunological reactions in which macrophages and possibly macrophage proliferation are important (14)(15)(16). This paper demonstrates an identity of sources, biological properties, physicochemical behaviour, and reactivity to antibody for both MGF and CSF. Methods for the purification of active material from serum-free L-cell-conditioned medium and the preparation and some properties of antibody to this material are also described.

Materials and Methods
MGF Assay. 6-12-wk-old male C57BL/6J mice (The Jackson Laboratory, Bar Harbor, Maine) were injected intraperitoneally with 1.5 ml of a 2% soluble starch suspension (Mann Research Labs, Inc., Orangeburg, N. J.) in phosphate-buffered saline (PBS). After 4 days, the mice were killed with ether, and their peritoneal cavities were washed with two 4-ml vol of minimal essential medium (MEM, Auto-Pow medium, Flow Laboratories, Inc., Rockville, Md.) containing 1% heparin. The cells were centrifuged (800g, 10 min), washed twice in 1 vol of PBS, resuspended in MEM containing 10% FCS, counted in leukocyte counting fluid, and seeded in Lab Tek 8 chamber slides (Miles Laboratories, Inc. Elkhart, Ind.) at 5 x 104 nucleated cells/cm 2. 30 min later the medium was suctioned off to reduce the numbers of lymphocytes and flbroblasts in the culture. The attached cells were washed twice with PBS and refed. 1-4 days after seeding, culture medium in wells was replaced with 0.5 ml of medium containing samples and [3H]thymidine (sp act 52.2 Ci/ mmol) at 0.2/~Ci/ml. Medium used for preparation and incubation of cultures was TES-HEPESbuffered Eagles medium pH 7.6 (17) containing 10% "pretested" (5) fetal calf serum (FCS, Flow Laboratories Inc.), penicillin (100 U/ml), and streptomycin (0.05 ~tg/ml). Duplicate culture wells were set up for each sample tested. After 5 days incubation at 37°C in 10% CO2 in air in a humidified incubator, cells were washed with PBS, fixed in Carnoy's solution for 20 min and washed with ethanol. Dry slides were dipped in Kodak NTB2 nuclear track emulsion (Eastman Kodak Co., Rochester, N. Y.), dried at room temperature, and left 2 days in sealed boxes containing a dessicant at 4°C for exposure before development with Kodak D-19 developer (Eastman Kodak). The developed slides were fixed in Amflx (May & Baker LTD., Montreal, Quebec) and stained with Giemsa. The percentage of labeled cells was determined microscopically under 400 magnification by counting at least 300 cells in each culture well.
CSF Assay. The techniques and media used in the agar culture of mouse bone marrow cells and the use of such cultures to assay CSF have been reported in full elsewhere (18). Sterile samples (<0.2 ml) were each mixed in duplicate 35-mm Petri dishes (Falcon Plastics, Oxnard, Calif.) with 1 ml of agar medium containing 75,000 nucleated bone marrow cells from adult LAF~ mice (The Jackson Laboratory). The medium used in these experiments was alpha-MEM (19) including 5% (vol/vol) horse serum, 10% (vol/vol) fetal calf serum (Flow Laboratories), 10% (vol/vol) trypticase soy broth (Baltimore Biological Laboratories, Cockeysville, Md.), 75 ~g/ml DEAE dextran (Pharmacia, Uppsala, Sweden), 5 x 10 -~ M mercaptoethanol, 0.4% Bacto-Agar (Difco Laboratories, Detroit, Mich.) and additional asparagine (20 ~tg added/ml of medium), and glutamine (0.29 mg added/ml of medium). The dishes were incubated at 37°C in a fully humidified atmosphere of 10% CO2 in air (pH 7.2). After 7 days of incubation colony counts were performed at 25 magnification using dark field on an Olympus dissecting microscope. Colony size ranged from 50-2,000 cells.
Conditioned Media and Sources of Activity. Serum containing media conditioned by the following cell lines were used: Human WI38, normal embryonic fibroblasts (20); NS20, a cholinergic clone of mouse neuroblastoma cells (21); IC21, an SV4o-transformed line of peritoneal macrophages derived from C57BL/6J mice (22); L-5178Y, a mouse lymphoma induced by methylcholanthrene (23); and embryo fibroblasts from BALB/c mice (24) and L 60 T cells, a clonal isolate of mouse L cells (25) (for standard L-cell-conditioned medium). Ehrlich ascites fluid was prepared by intraperitoneal injection of 5 × 10~-107 Ehrlich ascites cells (26) into 10-16-wk-old C57BL/6J male mice. The fluid was collected after 13 days, left at room temperature for 6 h, then centrifuged at 1,800 g, the supernate collected and filtered through a 0.22 ~m filter (Millipore Corp., Bedford, Mass.). Freund ascites fluid was prepared by intraperitoneal injection of 1.5 ml of complete Freund's adjuvant (CFA) (Difco Laboratories) into 10-16-wk-old C57BL/6J male mice. This ascites fluid was harvested in the same way as the Ehrlich ascites fluid.
The L-cell trypsin digest was prepared by treating confluent cultures in 160 cm 2 glass bottles with 0.1% trypsin (Flow Laboratories) 24 h after plating at 37°C. After 5 min, trypsinization of the cells was inhibited by addition of FCS; the medium was harvested and filtered through a 0.22 ~m filter (Millipore Corporation). Endotoxin mouse serum and normal mouse serum were separated from blood obtained from LAF~ mice (Jackson Laboratory). For endotoxin mouse serum the mice were given an intraperitoneal injection of 5 p~g Escherichia coli lipopolysaccharide (Difco Laboratories) and bled 4 h later (27). Stage VI human urinary CSF was purified from human urine as described elsewhere (8) and possessed an activity of 3 × 107 colonies per mg protein. Extract of pregnant mouse uterus was prepared from pregnant BALB/c mice as described earlier (28).
Purification of CSF from L-Cell-Conditioned Medium. L cells were passaged in spinner cultures of alpha medium containing 10% FCS (Flow Laboratories). For production of serum-freeconditioned medium for purification of CSF, cells were collected from the spinner flask, washed twice in serum-free alpha medium, and seeded at 2 x 105 cells/ml in 75 cm 2 plastic bottles (Falcon Plastics) each containing 40 ml of serum-free alpha medium. After 7 days incubation at 37°C in 10% CO2 in air in a humidified incubator, the medium was collected by decantation, concentrated approximately 10-fold by vacuum rotary evaporation (29), dialyzed against distilled water (3 × 20 volume changes, 4°C), centrifuged (1,500 g, 10 min), and the precipitate discarded. The supernatant fluid was further concentrated to a small volume by vacuum rotary evaporation. 5-10 liters of conditioned medium could be made and processed in this way to yield a final volume of approximately 10 ml. This material, although substantially purified from the unconcentrated conditioned medium, was designated as starting material for the purification. The starting material was dialyzed against 0.1 M Tris-HC1 pH 7.4 and chromatographed on a DEAE cellulose (Eastman Kodak) column (dimensions 107 × 2.2 cm) using linear salt gradient (0-0.8 M NaC1 in 0.1 M Tris-HC1 pH 7.4) elution at 4°C (Fig. 3). Active fractions were pooled (Stage I material), concentrated, dialyzed against 0.03 M Tris-HCl, pH 7,4, and subjected to Sephadex G-200 (Pharmacia Fine Chemicals) gel-filtration (column dimensions 130 x 2.6 cm) in that buffer at 4°C. At this stage the activity was usually found in two regions (elution volume to void volume ratio, V~/V,, = 1.3-1.5 and V,/V,, = 1.7-2.2) and fractions from each of these regions separately pooled (stages IIi and IIii, respectively), concentrated and dialyzed against Concanavilin A (Con A) buffer (0.1 M acetate pH 6.0 containing 1 M NaC1, 10 -:~ M MgC12, 10 :~ M MnCI2, 10 -'~ M CaCI2, and 0.001% polyethylene glycol (PEG, 6,000). Both stage IIi and IIii material were chromatographed on Con A-sepharose (Pharmacia Fine Chemicals) (column dimensions 36 × 1.0 cm) in Con A buffer at 4°C. After collection of the unadsorbed material, elution of material bound to the Con A-sepharose was accomplished using 0.1 M a-methyl-n-glucoside in Con A buffer (30). Activity was found in both the unadsorbed (IIIi) and bound (IIIii) fractions for pool IIi and similarly for pool IIii (IIIiii, unadsorbed; IIIiv, bound) (Fig. 3). All 4 stage III pools were concentrated to approximately 2.0 ml, dialyzed against distilled water, and stored at -20°C. Protein estimations (31) were carried out on all stages of each purification.
Each of the stage III fractions was subjected to gradient gel electrophoresis in a gradipore system (32) using preformed gradient gels (Ortec Inc., Oak Ridge, Tenn.). After electrophoresis of duplicate samples for 8 h at 75 V constant voltage, duplicate gel strips were cut out, one stained for protein with Coomassie Brilliant Blue (33), and the other fractionated using a razor blade device.

Kinetics of Disappearance of MGF and CSF from Cultures used in their Assay. Replicate MGF
cultures were set up with 20% L-cell-conditioned medium in the absence of :~HTdr, and the medium was removed for both MGF and CSF assays at various times afterwards. In addition, replicate 1ml CSF cultures were overlaid with 1 ml of liquid medium and contained 20% L-cell-conditioned medium throughout. Periodically, liquid overlayers were removed from cultures and assayed for MGF and CSF.
Treatments. The high molecular weight, Con A-Sepharose adherent fraction (stage IIIii L-cell CSF, 3.1 x 107 colonies/rag protein) was used for all experiments. Periodate treatment was carried out as described earlier (29). Incubation mixtures (0.9 ml) containing stage IIIii CSF (2.4 ~g) and K10~ (0.01 M) in 0.1 M acetate-HC1, pH 5.0, were incubated for various times at 4°C in the dark. Reaction was stopped by addition of 0.1 ml of 50% sucrose. For heat inactivation, incubation tubes contained 0.8/zg of stage IIIii CSF in 250 /~l of 0.03 M Tris-HCl, pH 7.3. Heating was at 90°C for various times. Treatment with a-chymotrypsin (Worthington Biochemical Corp., Freehold, N. J. ) was carried out as described earlier (35). Incubation tubes (28 ~l) contained 0.8 p,g of stage IIIii CSF and 16/zg a-chymotrypsin in 0.05 M CaC12 0.05 M Tris-HC1, pH 7.3. Incubation was at 37°C for various times, and the reaction was stepped by addition of 2.0 ml of 2% bovine serum albumin (BSA) and cooling to 4°C. . Fractions were counted for ~2'I before assay to establish that the trypsin was active under the conditions employed. Polyacrylamide disk electrophoresis of 5.0/zg stage IIIii CSF was performed in 7.8% polyacrylamide at a running pH of approximately 9.4 (36). Gradient gel electrophoresis of 12/zg (20 ~g for staining) of stage IIIii CSF, of 200 p.g of stage IIIi CSF, and the fractionation and assay of gels from the disk electrophoresis were carried out as described for stage III fractions from the purification.
Preparation and Testing of Rabbit Sera. An outbred rabbit was bled twice by heart puncture to provide 50 ml of prebleed serum. It was then injected at multiple intradermal sites with stage IIIii L-cell-conditioned medium CSF (0.14 mg) emulsified in an equal volume of CFA (Difco Laboratories) (total injected volume 0.4 ml). 2 wk later an identical booster injection and thereafter further boosters of 0.14 mg of stage IIIii material were given in various ways at 1-2-wk intervals. At the same time, small blood samples were collected via the ear veins for determination of the serum neutralization titre. Testing of the antiserum was carried out by mixing preparations of MGF or CSF with the antiserum or the PBS-diluted antiserum and incubating at room temperature for 30 min before assaying samples in both assay systems.

Results
Dose Response Curves. Typical dose response curves for the standard L-cellconditioned medium in both MGF and CSF assay systems are presented in Fig.  1. The CSF assay system at certain concentrations was approximately twice as sensitive as the MGF assay system. A sample causing almost maximal response in the CSF assay could show no or low response in the MGF assay because of the sigmoidal nature of the dose-response curve in the latter assay. Inhibition at high doses of conditioned medium could be seen in the MGF assay dose response (4). These differences between the two assay systems are important for interpretation of the results of some of the following experiments.

Sources of Activity. A number of sources for both factors have been reported.
Some sources such as ascites fluid and endotoxin mouse serum had been tested in only one system. Others such as normal mouse serum were only active in the CSF assay. To clarify the questions raised by these data, various sources that had been shown to be active or inactive in one of the two systems were tested at the same time for both CSF and MGF activity at different concentrations. The results are summarized in Table I. There was a complete correlation between assays for the presence or absence of activity in all of the various source materials assayed except for normal mouse serum where activity was low in the CSF assay and undetectable in the MGF assay.

Kinetics of Depletion of MGF and CSF Activities in both Culture
Systems. One biological test for identity of MGF and CSF is to compare the kinetics of disappearance of both activities in the different culture systems. Activities of MGF and CSF in their respective assay systems have been shown to decrease with time of culture (5,18). Transfer of medium from a test culture to fresh cells results in decreased activity in the recipient culture. Conditioned medium was incubated with either activated macrophages or mouse bone marrow cells, and the rate of disappearance of both CSF and MGF activities was measured. The results are summarized in Fig. 2. There was no significant difference between the kinetics of loss of CSF and the kinetics of loss of MGF during incubation with either cell population. Previous studies in both systems (5,18) have shown that under the same incubation conditions in the absence of cells, both activities remain at initial levels.  Copurification of CSF and MGF from L-Cell-Conditioned Media. Fractionation of L-cell-conditioned medium pools was carried out as summarized in Fig. 3. Activity of fractions was followed using the CSF assay. In two of three purifications, two peaks of activity were found at stage II, one of higher Fro. 3. Purification of CSF from serum-free L-cell-conditioned medium and analysis of the four active products by gradient gel electrophoresis. The banding of colony-stimulating activity for each of the stage III fractions is represented by the filled bands adjacent to the photographs of the stained sections of the gel slab. The migration of marker proteins included in the run is also indicated (Ova, ovalbumin; BSA, bovine serum albumin; (BSA)2, BSA dimer). apparent molecular weight (i) containing most of the activity and the other of lower apparent molecular weight (ii). Passage of stage II pools through Con Asepharose columns resulted in further fractionation. In all cases most of the activity was adsorbed to Con A-sepharose. Material that was unadsorbed (stages IIIi and IIIiii) failed to bind when passed through newly prepared Con Asepharose columns. Differences in the size characteristics of the activity and stainable protein in the four active fractions resulting from the fractionation can be seen in the gradient gel electrophoretograms presented in Fig. 3. Separation under the conditions used was almost entirely due to apparent size differences (32). The banding of activity in each electrophoretogram was broad and polydisperse compared with the banding of marker proteins. Results of a typical purification are summarized in Table II. It can be seen that most of the recovered activity was found in fraction IIIii (high molecular weight, Con A adherent), that it represented a substantial proportion of the starting activity, and was the most purified fraction with respect to protein. This fraction was used in the inactivation experiments, in the fractionation experiments, and to raise antibody for the neutralization experiments. Results of simultaneous assays for MGF and CSF on fractions from purifications from L-cell-conditioned medium and a tryptic digest of L cells are presented in Table III. It can be seen that, considering the inherent variability of the assays, there was a close correlation between results from both systems. For these experiments fractions were diluted for assay with a-medium which was shown to result in loss of both CSF and MGF activity. This loss of activity could be prevented by dilution in 0.01% PEG (8), e.g., stage IIIii in a-medium, 460 colonies/ml, and in PEG, 23,200 colonies/ml.

Kinetics of Inactivation of Purified L-Cell Material.
Previous studies on MGF (3) and CSF (12) from mouse sources have shown that they are inactivated by heat or periodate treatment and relatively resistant to treatment with achymotrypsin. There was no significant difference in the kinetics of inactivation between the MGF and CSF activities of stage IIIii material during incubation with a-chymotrypsin or periodate (Fig. 4) and during heating at 90°C (not shown). In contrast to results obtained with human urinary CSF (35), neither activity was lost on incubation with the high concentrations of a-chymotrypsin used. Similar kinetics to those for the inactivation by periodate were observed for the inactivation by heating at 90°C, 50% of both activities being lost after 45 min.
Fractionation of the CSF and MGF Activities of Purified L-CeU Material. The most purified fraction (stage IIIii) was used for these experiments. On Sephadex G-200 gel-filtration (Fig. 5 a) the band of MGF lay under the band of CSF. Austin et al. (11) reported an increase in elution volume on gel filtration of L-cell-conditioned medium CSF previously treated with trypsin. This finding was confirmed in experiments in which stage IIIii material was treated with trypsin before gel-filtration. Furthermore, trypsin treatment increased coordinately the elution volume of CSF and MGF activities (Fig. 5 b). Higher resolution analyses of the two activities in stage IIIii material were performed, separating according to charge by polyacrylamide disk gel electrophoresis and according to size by gradient gel electrophoresis. Cochromatography of MGF and CSF was observed on 6% polyacrylamide gel electrophoresis (Fig. 5 c) and on  gradient gel electrophoresis (Fig. 5 d) where the polydisperse activity bands coelectrophoresed with a polydisperse protein stained band. In addition, the MGF and CSF activities of stage IIIi material (high molecular weight, Con A nonadherent) were also shown to coelectrophorese on both disk electrophoresis and gradient gel electrophoresis (results not shown).
Kinetics of Antibody Production. Repeated injection of 0.14 mg of stage IIIii material with adjuvant into a rabbit produced an antiserum with neutralizing activity for L-cell-conditioned medium CSF and MGF and for human urinary CSF. The kinetics of production of the neutralizing activities of the antiserum are shown in Fig. 6. Neutralizing activity for CSF and MGF of L-cell-conditioned medium developed simultaneously. In contrast, the neutralizing activity for human urinary CSF took a longer time to develop and attained a lower but quite significant titre. There was no inhibitory effect of the prebleed rabbit serum on the CSF and MGF assays at the highest concentration of antiserum used. In addition, the untreated antiserum at the highest concentration used had no direct effect on colony formation if preincubated with the target cells for 1 h at 0°C in the presence or absence of guinea pig serum (as a source of complement), indicating that the neutralizing activity produced was directed against the factors and not the cells.
Neutralization by Antiserum. The results of a neutralization titration of high titre antiserum against L-cell-conditioned medium CSF and MGF and human urinary CSF are shown in Fig. 7. Titration curves for L-cell medium CSF and MGF were coincidental and distinct from the curve for human urinary CSF. Action of Antiserum on Murine Sources of CSF. The neutralizing activity of the antiserum for a number of sources of murine CSF was tested. With the exception of stage IIIiii (low molecular weight, Con A nonadherent), all the stage III purification fractions (including trypsin-treated stage IIIii) were neutralized at equivalent concentrations of antiserum (0.0001% antiserum (vol/vol) for 50% reduction of colony numbers). There was only partial neutralization (40%) of stage IIIiii material by the antiserum even at concentrations as high as 0.05%. The stimulating activities of mouse embryo-conditioned medium (24), extract of pregnant uterus (28), serum from mice injected with endotoxin (13), the Ehrlich ascites fluid, and the partially purified trypsin digest of L cells were completely neutralized by the antiserum. They exhibited identical neutralization curves with 50% reduction end points at 0.0001% antiserum.
Colony Formation by Peritoneal Cells. Lin and Stewart (37,38) reported that thioglycollate-induced peritoneal cells form macrophage colonies in agar in the presence of L-cell-conditioned medium. These colonies were distinct from the CSF-dependent colony forming units (CFU-C) seen in bone marrow cultures, being smaller and appearing later by approximately 3 wk of culture. To test whether the factor stimulating colony formation by peritoneal cells was related to CSF, starch-induced and thioglycollate-stimulated (38) peritoneal cells were  plated with the most purified CSF fraction, stage IIIii. The results (Table IV) indicate that purified CSF stimulates colony formation by both sources of peritoneal cells, and that there was no colony formation of the CFU-C type observable at 1 wk of culture. In addition, quantitatively similar results to those shown in Table IV were obtained using an unpurified L-cell-conditioned medium.

Discussion
Very close similarity of CSF and MGF has been demonstrated by a number of biological, physicochemical, and immunological criteria. In no instance could any significant difference between the two activities be demonstrated.
They were coordinately present or absent in the same sources. There was coordinate depletion or stability of activity of conditioned medium during incubation with peritoneal macrophages or bone marrow and after inactivation with heat, periodate, and a-chymotrypsin. Pooled, purified fractions from CSF and MGF purifications were active in both systems. Profiles of MGF and CSF activities of purified CSF from polyacrylamide gel electrophoresis and gradient gel electrophoresis were coordinate, as were elution profiles from G-200 chromatography before and after trypsin treatment. Antibody made in rabbits against the most purified material available appeared against CSF and MGF at the same time. In addition, the neutralization curves of CSF and MGF were identical using this antiserum. While absolute proof of identity must await chemical analysis of purified material, the data indicate that the MGF and CSF assays are probably detecting the same molecule. The chance that any two different glycoproteins would share the same source, heterogeneity on fractionation, physicochemical, biological, and antigenic properties, as do the molecules detected by the MGF and CSF assays, is quite remote. However, it is apparent that the CSF assay is more sensitive than the MGF assay, and this probably explains earlier findings (4) that indicated that certain preparations expected to contain CSF were inactive in the MGF assay.
The heterogeneity of the colony-stimulating activity of L-cell-conditioned medium may be related to its in vitro production, since other sources of CSF produced in vitro possess greater heterogeneity than CSF's from body fluids (12). L-cell-conditioned medium could be fractionated into four quite different molecular species described by their size and ability to bind Con A. While the lower molecular weight species (stages IIIiii and IIIiv) represented less than 5% of the total recovered activity its presence in L-cell -conditioned medium has not been previously reported. The relation of this fraction to the higher molecular weight fractions (stages IIIi and IIIii) is not understood. Its gel-filtration behaviour is similar to the low molecular weight CSF produced in vitro by mouse lungs in the presence of endotoxin (39)(40)(41). The specific activity of the most purified fraction, stage IIIii, (approximately 90% of the total activity recovered at stage III) was significantly higher than that reported for other mouse sources of CSF and MGF (5,11,12), except for the low molecular weight material in the medium conditioned by mouse lung in the presence of endotoxin (40). The gradient gel electrophoresis experiment indicates that stage IIIii material is not completely purified, although a protein band coelectrophoresed with the peak of activity. This band possessed an apparent mol wt of 70,000-120,000 daltons, whereas the high molecular weight, Con A-adherent fraction from another purification had an apparent mol wt of approximately 60,000 daltons. Both fractions have MGF activity and the 70,000-120,000 dalton band is the result of an association of the 60,000 dalton CSF with protein(s) (E. R. Stanley and P. M. Heard, manuscript in preparation). Similar behaviour has been observed for human urinary CSF (42).
The activity of highly purified human urinary CSF in the MGF assay is not surprising when one considers its physicochemical and immunological similarity to L-cell-conditioned medium CSF. These findings appear to enlarge the reported species specificity of MGF (4).
Our studies have been concerned with a particular type of CSF and should not be taken to indicate that all CSF's possess MGF activity. In human urinary CSF and L-cell-conditioned medium CSF-stimulated murine marrow cultures there is an early appearance of macrophages in colonies (13). Furthermore, L-cellconditioned medium CSF and human urinary CSF in their most purified states are glycoproteins of mol wt 45,000-70,000 which migrate electrophoretically with the a-globulins and albumin. They are physicochemically distinct from the CSF stimulating eosinophilic colony formation (43) which migrates electrophoretically with the ~/-globulins, a low molecular weight CSF (44,45), and a high molecular weight CSF (46) (Stanley, unpublished observations) stimulating human granulocytic colony formation. Inasmuch as stage IIIii material stimulates the formation of small macrophage colonies (37,38) from thioglycollate-or starch-induced peritoneal cells, it is probable that the factor in L-cell-conditioned medium responsible for stimulation in these systems is CSF (MGF). The relation of CSF (or MGF) to the factors in L-cell-conditioned medium responsible for stimulating the growth of alveolar macrophages in liquid cultures (47) has not been investigated, but it is possible that the one factor is responsible for growth of macrophages in all culture systems.
The results presented here indicate that macrophages, from an undifferentiated precursor cell to the mature differentiated macrophage, are under regulation by the same molecular species. If the molecular species responsible for CSF and MGF activity are the same, the difference then appears to reside in the test cells. Activated macrophages have a longer doubling time than CFU-C in the presence of the factor, and they respond maximally at twice the factor concentration required for maximal response of CFU-C. If receptor molecules are involved in the stimulation by CSF or MGF, an attractive hypothesis would be that their distribution or density change as the target cells progress to the final macrophage state. The differences in response would then not be due to different substances, but to cells of the same lineage, which, at different stages of their maturation, are programmed to respond by proliferating according to different concentrations of the same group of substances.
Positive regulation of macrophage production ought to be homeostatically linked to the requirements of the organism. In this context, it is of interest to consider the release of CSF and MGF. Peritoneal macrophages from unstimulated mice are a good source from which the release of CSF is largely antigen dependent (48). Maximal release of CSF from adherent macrophages occurs within 3 h of exposure to endotoxin, and the degree of release is related to endotoxin concentration (48). The release is only slightly inhibited by puromycin (49) compatible with a cell surface localization of CSF, as reported for MGF on L cells (10). It is attractive to consider the regulation of macrophage differentiation in terms of positive feedback via the antigen concentration-dependent release of a regulator from mature macrophages.
The CSF-MGF system offers a potentially valuable means of analyzing the precise operation of a regulator molecule on eukaryote cell differentiation and proliferation. In addition, since CSF levels are elevated quickly and dramatically on antigen stimulation, it is pertinent to the study of immunological systems and disease states.

Summary
The activities of a colony-stimulating factor (CSF), which stimulates granulocyte-macrophage colony formation by mouse hemopoietic cells, and macrophage growth factor (MGF), which stimulates proliferation of activated peritoneal macrophages, have been demonstrated by various criteria to reside in the same molecular species. These criteria include occurrence in various sources and copurification of the activities in mouse L-cell-conditioned medium as well as the biological, physicochemical, and antigenic properties of the activities of Lcell-conditioned medium. CSF and MGF activities of L-cell-conditioned medium are ascribable to a glycoprotein of mol wt approximately 60,000 which migrates electrophoretically with a-globulin. Human urinary CSF, which also possesses MGF activity, has similar properties and can be neutralized by antiserum to highly purified L-cell medium CSF. A procedure is described for the partial purification of material from L-cell medium that has activity at 1 ng/ml in both MGF and CSF assays.
Note Added in Proof. Further purification of L-cell-conditioned medium CSF has resulted in a preparation that is pure as assessed by isoelectric focusing, SDS polyacrylamide gel electrophoresis, and immunological criteria. The purified CSF possesses MGF activity and stimulates colony formation by thioglycolate-induced peritoneal cells (E. R. Stanley and P. M. Heard, manuscript in preparation).
We wish to thank Dr. E. Bikoff for her helpful comments in the preparation of this manuscript.
Received for publication 12 November 1975.