Purification of two forms of colony-stimulating factor from mouse L-cell-conditioned medium

A modified procedure for the purification of the colony-stimulating factors (CSFs) in mouse L-cell-conditioned medium is used to isolate two forms of CSF, which are separable by reversed-phase high performance liquid chromatography with 300-A pore size supports. The specific biological activity of these CSFs (2 X 10(9) colonies/mg) was considerably higher than has been achieved by other methods. Even at high concentration (200 pM) both molecules stimulated predominantly more macrophage than granulocyte colonies; however, the less hydrophobic form appeared to stimulate the formation of more pure granulocytic colonies. Almost twice as much of the less hydrophobic CSF was recovered from L-cell-conditioned medium. Analysis using sodium dodecyl sulfate-polyacrylamide gel electrophoresis indicated that both forms of L-cell CSF had apparent molecular masses of approximately 70,000 daltons. However, on reduction with 2-mercaptoethanol, while both forms generated a 39,000-dalton subunit, the less hydrophobic form also yielded a 32,000-dalton subunit. Storage of either form of L-cell CSF at pH 2.1, in the presence of acetonitrile or isopropanol, destroyed the biological activity. Electrophoretic analysis of the L-cell CSFs stored under these conditions indicated that this was associated with a spontaneous dissociation of the CSF dimer into the inactive subunits. There was some charge heterogeneity (pI 3.5-4.7) indicating different degrees of glycosylation. The unique N-terminal amino acid sequences of both forms of CSF were the same: (Lys-Glu-Val-Ser-Glu-His-X-Ser-His-Met-Ile-Gly-Asn). Thus, the polypeptide chains appear to be identical for the subunits of both forms of L-cell CSF.     

The Wellcome Foundation lecture, 1986. The molecular control of normal and leukaemic granulocytes and macrophages

The development of semisolid culture methods supporting the clonal proliferation and maturation of granulocytes and macrophages led to the discovery of a group of specific glycoproteins, the colony-stimulating factors (CSFs), whose function it is to control the proliferation and functional activity of granulocytes, macrophages and associated blood cells. The four known CSFs in the mouse and man have been purified and complementary DNAs (cDNAs) for each have been cloned. The injection of bacterially synthesized recombinant CSF into mice has demonstrated that these CSFs can function in vivo to regulate granulocyte and macrophage formation. A major physiological role played by these CSFs is to control resistance to invading microorganisms through mechanisms capable of extremely rapid activation. Because the CSFs are the only known proliferative factors for these cells, the CSFs are involved in the initiation and the emergence of myeloid leukaemia but, conversely, at least one of the CSFs, G-CSF, is able to suppress myeloid leukaemic populations because of the ability of the CSFs to initiate differentiation commitment in responding granulocytic and macrophage populations. The CSFs are promising agents for clinical use in the treatment of infections in patients with depressed granulocyte-macrophage formation and possibly in the management of some types of myeloid leukaemia.     

Glucocorticoids and prostaglandins inhibit the induction of macrophage DNA synthesis by macrophage growth factor and phorbol ester

The tumor-promoting phorbol ester, 12-0-tetradecanoyl-phorbol-13-acetate (TPA), stimulates starch-elicited mouse peritoneal macrophages to undergo DNA synthesis in vitro, apparently without the generation of an endogenous macrophage growth factor (MGF). No evidence was found for any synergistic interaction between TPA and exogenous colony stimulating factors (CSFs) for macrophage DNA synthesis. Low concentrations of glucocorticoids and also prostaglandins E1 and E2 suppress both the CSF-1-stimulated and the TPA-stimulated macrophage DNA synthesis; these same drugs inhibit the CSF-1-mediated and TPA-mediated enhancement of macrophage plasminogen activator (PA) activity. Thus glucocorticoids and prostaglandins E1 and E2 oppose the action of growth factors and the tumor promoter on macrophage and precursor cell function.     

Importance, localization and functional properties of the cell-associated form of plasminogen activator in mouse peritoneal macrophages

In inflammatory macrophages, plasminogen activator exists in two active forms, a soluble form released into the extracellular medium and a cell-associated form. This communication describes some properties of the cellular form of plasminogen activator, in intact macrophages and in cell lysates. Cellular plasminogen activator is a membrane protein, associated with the outer face of the plasma membrane; in intact macrophages, it participates in the activation of exogenous plasminogen and, thus, has to be considered as an ectoenzyme. A plasminogen activator activity can be detected in cell lysates (macrophage monolayers lysed in 0.1% Triton X-100) only when plasmin production is followed by the use of small synthetic substrates because a soluble inhibitor, released during extraction, blocks plasmin fibrinolytic activity. In these lysates, plasminogen activator molecules exist as high molecular weight unstable complexes exhibiting a high affinity for plasminogen.     

Colony-stimulating factors in inflammation and autoimmunity

Although they were originally defined as haematopoietic-cell growth factors, colony-stimulating factors (CSFs) have been shown to have additional functions by acting directly on mature myeloid cells. Recent data from animal models indicate that the depletion of CSFs has therapeutic benefit in many inflammatory and/or autoimmune conditions and as a result, early-phase clinical trials targeting granulocyte/macrophage colony-stimulating factor and macrophage colony-stimulating factor have now commenced. The distinct biological features of CSFs offer opportunities for specific targeting, but with some associated risks. Here, I describe these biological features, discuss the probable specific outcomes of targeting CSFs in vivo and highlight outstanding questions that need to be addressed.     

Degradation of the P0, P1, and Pr, Proteins in Peripheral Nervous System Myelin by Plasmin: Implications Regarding the Role of Macrophages in Demyelinating Diseases

Activated macrophages secrete a variety of neutral proteinases, including plasminogen activator. Since macrophages are implicated in primary demyelination in the peripheral nervous system (PNS) in Guillain-Barré syndrome and experimental allergic neuritis, we have investigated the ability of plasmin and of conditioned media from cultured macrophages, in the presence of plasminogen, to degrade the proteins in bovine and rat PNS myelin. The results indicate that (a) the major glycoprotein P0 and the basic P1 and Pr proteins in PNS myelin are extremely sensitive to plasmin, perhaps more so than is the basic protein in CNS myelin; (b) the initial product of degradation of P0 by plasmin has a molecular weight higher than that of the "X" protein; (c) large degradation products of P0 are relatively insensitive to further degradation; and (d) the neuritogenic P2 protein in PNS myelin is quite resistant to the action of plasmin. Results similar to those with plasmin were obtained with conditioned media from macrophages and macrophage-like cell lines together with plasminogen activator, and the degradation of the PNS myelin proteins, Po and P1, under these conditions was inhibited by p-nitrophenylguanidinobenzoate, an inhibitor of plasmin and plasminogen activator. The results suggest that the macrophage plasminogen activator could participate in inflammatory demyelination in the PNS.     

Pertussigen in vivo enhances antigen-specific production in vitro of lymphokine that stimulates macrophage procoagulant activity and plasminogen activator

Pertussigen is a protein toxin of Bordetella pertussis that acts as a powerful stimulator of the intensity and duration of delayed-type hypersensitivity (DTH) in mice. This study describes the potent in vivo effect of pertussigen on the levels of antigen-specific macrophage-activating lymphokine(s); lymphokine(s) was measured by the stimulation of macrophage procoagulant activity (mPCA), or plasminogen activator (PA) activity. Lymphoid cells were removed from immunized animals and cultured with specific antigen, keyhole limpet hemocyanin, ovalbumin, or human gamma-globulin. The culture supernatants were then incubated with the monocyte-like cell line WEHI-265 to measure mPCA or with WEHI-265 or resident peritoneal macrophages to measure PA activity. Mice were given pertussigen at the time of immunization, and the subsequent generation by lymphocyte supernatants of both of these macrophage activities proved to be greatly enhanced; the effect of pertussigen was antigen specific. Pertussigen thus induces an increase in lymphokine(s) production responsible for the in vitro increase in macrophage mPCA and PA activity and which may be responsible for some of the potent immune effects of this agent in vivo.     

Elevation of plasminogen activators in cerebrospinal fluid of mice with eosinophilic meningitis caused by Angiostrongylus cantonensis

A hallmark of parasitic meningitis is the infiltration of eosinophils into the subarachnoid space. Infection with Angiostrongylus cantonensis in mice induced proteinase activity in parallel with the pathological changes of eosinophilic meningitis. Zymogram analysis demonstrated that 70 and 55 kDa proteinases from cerebrospinal fluid (CSF) were active against the casein/plasminogen substrate. The proteinase activities were clearly inhibited by phenylmethanesulphonyl fluoride but not by ethylenediamine tetraacetic acid, 1,10-phenanthroline or leupeptin. Western blotting confirmed these enzymes to be tissue-type plasminogen activator and urokinase-type plasminogen activator, respectively. High activities of tissue-type plasminogen activator and urokinase-type plasminogen activator were detected in the CSF of mice with eosinophilic meningitis, and correlated positively with CSF eosinophil numbers and total protein, respectively. Immunohistochemistry demonstrated that tissue-type plasminogen activator and urokinase-type plasminogen activator localised in the endothelial cells of blood vessels, in blood clots and in infiltrated leukocytes. These results suggest that tissue-type plasminogen activator and urokinase-type plasminogen activator may be play a role in the pathogenesis of eosinophilic meningitis of angiostrongyliasis.     

Human macrophage colony-stimulating factor inhibits bone resorption by osteoclasts disaggregated from rat bone

Colony stimulating factors (CSFs) regulate the survival, proliferation and differentiation of haemopoietic progenitor cells, as well as the functional activity of mature cells. Because the osteoclast is derived from haemopoietic tissue, and because osteoblastic cells produce CSFs, we tested the effects of several CSFs on bone resorption by osteoclasts disaggregated from neonatal rat long bone. We found that recombinant macrophage (M)-CSF was a potent inhibitor of bone resorption, causing significant inhibition at concentrations similar to those required to support the growth of macrophage colonies in agar. Unlike other inhibitors of osteoclastic resorption, M-CSF did not alter cytoplasmic motility in time-lapse recordings, suggesting that M-CSF may inhibit osteoclasts through a different transduction mechanism. None of the remaining cytokines tested (granulocyte-macrophage CSF, interleukin 3, interleukin 6, or interferon γ) influenced bone resorption. M-CSF production may be a mechanism by which osteoblastic cells, which produce M-CSF, may regulate osteoclastic function. Alternatively, inhibition of osteoclastic resorption by a CSF that is responsible for amplification of the macrophage compartment may reflect a close lineage relationship between mononuclear phagocytes, in which M-CSF induces a diversion of lineage resources away from osteoclastic function.     

Plasmin-mediated macrophage reversal of low density lipoprotein aggregation

Evidence suggests that aggregated low density lipoprotein (AgLDL) accumulates in atherosclerotic lesions. Previously, we showed that AgLDL induces and enters surface-connected compartments (SCC) in human monocyte-derived macrophages by a process we have named patocytosis. Most AgLDL taken up by these macrophages in the absence of serum is stored in SCC and remains undegraded. We now show that macrophages released AgLDL (prepared by vortexing or treatment with phospholipase C or sphingomyelinase) from their SCC when exposed to 10% human lipoprotein-deficient serum (LPDS). Macrophages also took up AgLDL in the presence of LPDS, but subsequently released it. In both cases, the released AgLDL was disaggregated. Although the AgLDL that macrophages took up could not pass through a 0.45-micrometer filter, >60% of AgLDL could pass this filter after release from the macrophages. Disaggregation of AgLDL was verified by gel-filtration chromatography and electron microscopy that also showed particles larger than LDL, reflecting fusion of LDL that aggregates. The factor in serum that mediated AgLDL release and disaggregation was plasmin generated from plasminogen by macrophage urokinase plasminogen activator. AgLDL release was decreased >90% by inhibitors of plasmin (epsilon-amino caproic acid and anti-plasminogen mAb), and also by inhibitors of urokinase plasminogen activator (plasminogen activator inhibitor-1 and anti-urokinase plasminogen activator mAb). Moreover, plasminogen could substitute for LPDS and produce similar macrophage release and disaggregation of AgLDL. Because only plasmin bound to the macrophage surface is protected from serum plasmin inhibitors, interaction of AgLDL with macrophages was necessary for reversal of its aggregation by LPDS. The released disaggregated LDL particles were competent to stimulate LDL receptor-mediated endocytosis in cultured fibroblasts. Macrophage-mediated disaggregation of aggregated and fused LDL is a mechanism for transforming LDL into lipoprotein structures size-consistent with lipid particles found in atherosclerotic lesions.     

Defining GM-CSF- and macrophage-CSF-dependent macrophage responses by in vitro models

GM-CSF and M-CSF (CSF-1) induce different phenotypic changes in macrophage lineage populations. The nature, extent, and generality of these differences were assessed by comparing the responses to these CSFs, either alone or in combination, in various human and murine macrophage lineage populations. The differences between the respective global gene expression profiles of macrophages, derived from human monocytes by GM-CSF or M-CSF, were compared with the differences between the respective profiles for macrophages, derived from murine bone marrow cells by each CSF. Only 17% of genes regulated differently by these CSFs were common across the species. Whether a particular change in relative gene expression is by direct action of a CSF can be confounded by endogenous mediators, such as type I IFN, IL-10, and activin A. Time-dependent differences in cytokine gene expression were noted in human monocytes treated with the CSFs; in this system, GM-CSF induced a more dramatic expression of IFN-regulated factor 4 (IRF4) than of IRF5, whereas M-CSF induced IRF5 but not IRF4. In the presence of both CSFs, some evidence of "competition" at the level of gene expression was observed. Care needs to be exercised when drawing definitive conclusions from a particular in vitro system about the roles of GM-CSF and M-CSF in macrophage lineage biology.     

Exposure of cryptic domains in the alpha 1-chain of laminin-1 by elastase stimulates macrophages urokinase and matrix metalloproteinase-9 expression

Degradation of the extracellular matrix leads to the release of fragments, which elicit biological responses distinct from intact molecules. We have reported that alpha1:Ser(2091)-Arg(2108), a peptide derived from the alpha1-chain of laminin-1, triggers protein kinase C-dependent activation of MAPK(erk1/2), leading to the up-regulation of macrophage urokinase type plasminogen activator and matrix metalloproteinase (MMP)-9 expression. Since intact laminin-1 failed to trigger these events, we hypothesized that alpha1:Ser(2091)-Arg(2108) is cryptic or assumes a conformation not recognized by macrophages. Here we demonstrate that elastase cleavage of laminin-1 generates fragments, which stimulate proteinase expression by RAW264.7 macrophages and peritoneal macrophages. In contrast, fragments generated by MMP-2, MMP-7, or plasmin had no effect on macrophage proteinase expression. Elastase-generated laminin-1 fragments were fractionated by heparin-Sepharose chromatography. Heparin-binding fragments stimulated macrophages' proteinase expression severalfold greater than nonbinding fragments. The heparin binding fragments reacted with antibodies directed against regions of the alpha1-chain including alpha1:Ser(2091)-Arg(2108) and the globular domain. A peptide from the first loop of the globular domain (alpha1:Ser(2179)-Ser(2198)) triggered the phosphorylation of MAPK(erk1/2) and stimulated the expression of macrophage urokinase type plasminogen activator and MMP-9. Moreover, a heparin-binding fraction isolated from an aortic aneurysm contained fragments of alpha1-chain and stimulated macrophages' proteinase expression. Based on these data, we conclude that cryptic domains in the COOH-terminal portion of the alpha1-chain of laminin are exposed by proteolysis and stimulate macrophages' proteinase expression.     

Mechanisms of cardiac fibrosis induced by urokinase plasminogen activator

Human hearts with end-stage failure and fibrosis have macrophage accumulation and elevated plasminogen activator activity. However, the mechanisms that link macrophage accumulation and plasminogen activator activity with cardiac fibrosis are unclear. We previously reported that mice with macrophage-targeted overexpression of urokinase plasminogen activator (SR-uPA+/o mice) develop cardiac macrophage accumulation by 5 weeks of age and cardiac fibrosis by 15 weeks. We used SR-uPA+/o mice to investigate mechanisms through which macrophage-expressed uPA causes cardiac macrophage accumulation and fibrosis. We hypothesized that: 1) macrophage accumulation and cardiac fibrosis in SR-uPA+/o mice are dependent on localization of uPA by the uPA receptor (uPAR); 2) activation of plasminogen by uPA and subsequent activation of transforming growth factor-beta1 (TGF-beta1) and matrix metalloproteinase (MMP)-2 and -9 by plasmin are critical pathways through which uPA-expressing macrophages accumulate in the heart and cause fibrosis; and 3) uPA-induced cardiac fibrosis can be attenuated by treatment with verapamil. To test these hypotheses, we bred the SR-uPA+/o transgene into mice deficient in either uPAR or plasminogen and measured cardiac macrophage accumulation and fibrosis. We also measured cardiac TGF-beta1 protein (total and active), Smad2 phosphorylation, and MMP activity after the onset of macrophage accumulation but before the onset of cardiac fibrosis. Finally, we treated mice with verapamil. Our studies revealed that plasminogen is necessary for uPA-induced cardiac fibrosis and macrophage accumulation but uPAR is not. We did not detect plasmin-mediated activation of TGF-beta1, MMP-2, or MMP-9 in hearts of SR-uPA+/o mice. However, verapamil treatment significantly attenuated both cardiac fibrosis and macrophage accumulation.     

Phorbol ester and mitogens stimulate human fibroblast secretions of plasmin-activatable plasminogen activator and protease nexin, an antiactivator/antiplasmin

Tumor-promoting phorbol esters have been reported to greatly increase plasminogen activator (PA) activity produced in numerous cell types. Many of these studies have employed a widely used fibrinolysis assay for PA activity that involves large-scale dilution of cell lysates or conditioned medium (CM) into buffer containing plasminogen and the plasmin substrate 125I-fibrin. This assay indicates that phorbol ester and the mitogens epidermal growth factor (EGF) and thrombin all stimulate secretion of PA activity in our human foreskin fibroblast cultures. However, these effects are not observed in a modified fibrinolysis assay employing undiluted conditioned culture medium unless the medium is first treated at pH 3, which inactivates the secreted protease inhibitor, protease nexin (PN). Moreover, a direct assay for plasminogen activator activity based on cleavage of 125I- plasminogen indicates that conditioned culture medium contains little if any active plasminogen activator either before or after treatment of the cultures with phorbol ester or EGF. Phorbol ester and mitogens do stimulate secretion of (a) an inactive PA that can be activated by plasmin and (b) PN, which inhibits both the activated form of the PA and plasmin. Secretions of the inactive PA and PN are further correlated in that release of both is stimulated most by phorbol ester, somewhat less by EGF, and least by thrombin. Significantly, these effects are not accompanied by increases in total protein secretion. We propose that fibroblasts secrete PA in an inactive form in the presence of PN to confine PA activity to an as yet undefined location or event.     

Occurrence of a monocyte/macrophage colony-stimulating factor in the continuous ambulatory peritoneal dialysis fluids and its chromatographic behaviors and antigenicity compared with human urinary colony-stimulating factor

Continuous ambulatory peritoneal dialysis (CAPD) fluid from three patients with chronic renal failure exhibited the activity of colony-stimulating factor (CSF) in amounts varying from 5 to 40 units per ml. Like the CSF obtained from normal human urine, the peritoneal CSF predominantly produced monocyte/macrophage colonies in soft-agar culture of mouse bone marrow cells. Semipurified peritoneal CSF showed its isoelectric point at pH 3.6 and 4.9 before and after the treatment with neuraminidase. Under the same conditions, the urinary CSF was focused at pH 3.1 and 4.6. The position of elution of the peritoneal and urinary CSF in ordinary gel-filtration chromatography corresponded to a molecular weight of 62,000 and 117,000, whereas both CSFs exhibited a molecular weight of 28,000 upon gel-filtration in the presence of 6 M guanidine HCl. Furthermore, the two CSFs from the human sources were neutralized by antimouse L cell CSF serum in the same manner. We conclude that the peritoneal CSF is a sialoglycoprotein which is nearly identical with the urinary CSF despite processing of the latter through kidneys.     

Acetyl-LDL stimulates macrophage-dependent plasminogen activation and degradation of extracellular matrix

The ability of acetyl-LDL to stimulate macrophage-dependent plasminogen activation and degradation of extracellular matrix was examined. We have found that expression of plasminogen activator activity in response to the scavenger receptor ligand varied among cell populations. Exposure to acetyl-LDL stimulated plasminogen activator expression by cells which constitutively released low levels of activator. These include a virally transformed macrophage-like cell line (RAW246.7), concanavalin A and C. parvum-activated macrophages. The stimulation of plasminogen activator activity was independent of cellular lipid accumulation since nonlipoprotein inhibitors of acetyl-LDL binding to the scavenger receptor stimulated activator expression in great excess to that observed with acetyl-LDL. In contrast, acetyl-LDL was unable to induce soluble plasminogen activator activity in cells which normally do not express it. These include a macrophage-like cell line (J774A.1) and resident peritoneal macrophages. Furthermore, acetyl-LDL was unable to modulate the copious secretion of activator by inflammatory macrophages elicited with thioglycolate. When macrophages were tested for their ability to degrade smooth muscle cell derived matrix, solubilization by resident, elicited, and activated cells was variously increased in the presence of plasminogen. Furthermore, exposure to acetyl-LDL enhanced plasmin-dependent degradation by resident cells and activated cells, whereas matrix degradation by elicited cells was unaffected.     

Granulocyte-macrophage colony-stimulating factor (CSF) and macrophage CSF-dependent macrophage phenotypes display differences in cytokine profiles and transcription factor activities: implications for CSF blockade in inflammation

GM-CSF and M-CSF (CSF-1) can enhance macrophage lineage numbers as well as modulate their differentiation and function. Of recent potential significance for the therapy of inflammatory/autoimmune diseases, their blockade in relevant animal models leads to a reduction in disease activity. What the critical actions are of these CSFs on macrophages during inflammatory reactions are unknown. To address this issue, adherent macrophages (GM-BMM and BMM) were first derived from murine bone marrow precursors by GM-CSF and M-CSF, respectively, and stimulated in vitro with LPS to measure secreted cytokine production, as well as NF-kappaB and AP-1 activities. GM-BMM preferentially produced TNF-alpha, IL-6, IL-12p70, and IL-23 whereas, conversely, BMM generated more IL-10 and CCL2; strikingly the latter population could not produce detectable IL-12p70 and IL-23. Following LPS stimulation, GM-BMM displayed rapid IkappaBalpha degradation, RelA nuclear translocation, and NF-kappaB DNA binding relative to BMM, as well as a faster and enhanced AP-1 activation. Each macrophage population was also pretreated with the other CSF before LPS stimulation and found to adopt the phenotype of the other population to some extent as judged by cytokine production and NF-kappaB activity. Thus, GM-CSF and M-CSF demonstrate, at the level of macrophage cytokine production, different and even competing responses with implications for their respective roles in inflammation, including a possible dampening or suppressive role for M-CSF in certain circumstances.     

Properties of colony-stimulating factors produced by macrophage cell lines and hybrid cells

Colony-stimulating factors (CSFs) produced by two simian virus 40(SV40) transformed macrophage cell lines (BAM1 and BAM3), and three hybrids (HM3-11, HM3-12, and HM3-14) derived from fusion between BAM3 and a Chinese hamster cell line (hs222-16) were examined. HM3-11 and HM3-14 produce two molecular species of CSF, which are not found in the conditioned media from cultures of BAM1 and BAM3 or lipopolysaccharide (LPS), phorbolmyristate-acetate (PMA), and zymosan-stimulated BAM3. HM3-12, which is classified into another group in terms of CSF secretion, does not produce these two CSFs. On the basis of various criteria, one of these CSF species (peak 1-CSF) was characterized as a macrophage-colony-stimulating factor (M-CSF). The other CSF (peak 2-CSF) induced a group of bone marrow cells in granulocytes and macrophages as well as growth of a mast cell line, IC2. This CSF has an apparent molecular weight of 18,000, estimated by SDS-polyacrylamide gel electrophoresis. Unlike interleukin 3 (IL3) from WEHI-3 cells, the growth factor activity of peak 2-CSF binds to DEAE-Sephacel. Thus, peak 2-CSF is similar to a granulocyte-macrophage colony-stimulating factor (GM-CSF) rather than to IL3. The anti L cell CSF serum does not inhibit the CSF activity in Chinese hamster fibroblast conditioned medium, and the IC2 cells do not respond to Chinese hamster lung conditioned medium (CHLCM), suggesting that peak 1- and peak 2-CSF are of mouse origin.     

Growth factors that stimulate human granulocyte—macrophage colony formation produced by the cell line CM-S

CM-S is an autonomous cell line of human hemopoietic precursor cells inducible to monocyte-macrophage differentiation in response to appropriate inducing agents. CM-S cells produce factors that stimulate their own growth and proliferation, and are also capable of stimulating clonal proliferation of human, but not mouse, monocytic and granulocytic bone marrow progenitor cells in viscous medium. Preliminary purification steps have demonstrated at least two species, one of which (MW 30,000–50,000) retains both these activities, while the other (MW ≤ 10,000) apparently retains only the autostimulatory activity. CM-S cells could thus be a useful source for the purification of human colony stimulating factors (CSFs). CM-S cells also respond to factors present in human placenta conditioned medium, known to contain human CSF. This suggests that CM-S cells could provide a homogeneous target cell population for testing CSFs from other human sources.