With the cloning of the cDNA encoding murine GM-CSF in 1984, Dunn and colleagues (1) ushered the power of recombinant DNA technology into the exploration of the biology of this multifaceted cytokine. Prior to this seminal work, only vanishingly small amounts of purified protein, reflecting the output of a complex and lengthy isolation scheme, were available for study, severely limiting analysis of GM-CSF structure and function (2). With the production of large quantities of recombinant protein derived from the cloned cDNA, experiments were undertaken that definitively established the striking ability of GM-CSF to enhance the production of myeloid cells in vivo, a result replete with substantive clinical implications (3). Moreover, the capability to perform mutagenesis studies of the GM-CSF gene and, shortly thereafter, the gene encoding the cognate receptor led to the delineation of additional unexpected roles for the cytokine in tissue homeostasis and immunity (4–7), aspects that continue to intrigue investigators today.
Metcalf (8) of the Walter and Eliza Hall Institute in Melbourne discovered GM-CSF as part of a larger effort aimed at learning more about the mechanisms controlling hematopoiesis. Pioneering studies of Till and McCullough (9) and Becker et al. [(10), previously featured as a Pillar of Immunology], in which bone marrow cells were injected into lethally irradiated mice and the ensuing splenic nodules enumerated, had provided a quantitative system to investigate hematopoiesis in vivo. The regulation of blood cell formation was suspected to involve not only direct cell–cell interactions within the bone marrow microenvironment, but also paracrine and possibly endocrine pathways. To explore this issue, Bradley and Metcalf (11) and Pluznik and Sachs (12) (independently in Israel) developed elegant in vitro systems to propagate blood cells. The application of soft agar, rather than liquid media, in their approach ensured that the progeny of individual precursors would remain physically associated in a CFU, thereby allowing precise lineage relationships to be determined. Proliferation was not cell autonomous in this system, but depended on soluble mediators provided by feeder cells.
The search for colony stimulating factors involved testing many cell lines and tissue extracts for their activity as feeders, and this soon yielded the recognition that colonies elicited with different conditioned media were heterogeneous (13). Some colonies contained only neutrophils or monocytes/macrophages, whereas others consisted of both cell types, and yet others included additional elements such as eosinophils, mast cells, erythroid precursors, and megakaryocytes. Eventually, four distinct factors were linked with the different colonies: G-CSF, which yielded neutrophils only; M-CSF, which produced monocytes/macrophages only; GM-CSF, which promoted the appearance of both granulocytes and macrophages; and multi-CSF (better known as IL-3), which gave rise to the most diverse colonies.
Herculean biochemical efforts were undertaken to purify these factors. GM-CSF was found to be most abundant in conditioned media obtained from lungs of mice injected with LPS (2). The isolation scheme eventually devised in the laboratory of Burgess at the Walter and Eliza Hall Institute required almost 100,000 mice to yield a small amount of material that proved biologically active in vitro at picomolar levels (14, 15). Limited sequence data could be obtained for the N terminus of the protein, and these were used in the Dunn laboratory (1) to generate corresponding degenerate oligonucleotides for screening a cDNA library prepared from the lungs of endotoxin-treated mice. The isolated mRNAs were injected into Xenopus oocytes, and the corresponding translated protein displayed the appropriate colony-forming activity, confirming the successful cloning of the GM-CSF sequences. Southern analysis documented a single gene locus, which resolved the uncertainty regarding whether there was more than one GM-CSF protein, an inference that had been made based on the heterogeneity observed in purifying the factor from different sources (16). Sequence comparisons further showed that GM-CSF was distinct from other growth factors that had been characterized at the time, including IL-3, the multi-CSF that displayed some functional similarities.
The injection of rGM-CSF into mice led to marked increases in neutrophils, monocytes/macrophages, and eosinophils (3), demonstrating a key role for the cytokine in myelopoiesis and validating the overall scientific strategy that Metcalf had envisioned decades before. Clinical trials of recombinant human GM-CSF quickly followed in patients with impaired bone marrow function secondary to chemotherapy, immune deficiency, or intrinsic marrow disease (17, 18). Augmentation of myeloid cell production was consistently observed, with a concomitant reduction in the risk of serious infection, leading to regulatory approval of the factor for the acceleration of bone marrow recovery after cytotoxic therapy or bone marrow transplantation.
The cloning of the GM-CSF cDNA also afforded the opportunity to generate transgenic mice with constitutive production of the factor (19). These animals manifested significant tissue pathology due to the activities of robust myeloid cell infiltrates, findings that anticipated current efforts targeting GM-CSF for the treatment of inflammatory disease (20, 21). An intricate cross-talk between GM-CSF– secreting cells (such as lymphocytes and stromal elements) and myeloid cells appears to be critical for the induction of several autoimmune disorders, including rheumatoid arthritis, multiple sclerosis, thyroiditis, and cardiomyopathy, among others. Preliminary clinical trials of GM-CSF blockade (accomplished with mAbs targeting the receptor or the cytokine) show promise for attenuating disease activity (22).
Whereas these therapeutic studies established a role for GM-CSF in some forms of emergency hematopoiesis and tissue inflammation, Metcalf (8) sought to understand the regulation of steady-state blood cell formation as well. In this context, the generation and characterization of GM-CSF–deficient mice revealed that the cytokine has a rather limited function (at least as an individual knockout), with most indices of hematopoiesis remaining intact (4, 5). Nonetheless, detailed characterization of the mice uncovered a specific requirement for GM-CSF in the development of some nonlymphoid dendritic cell populations (23, 24) and the maturation of alveolar macrophages from fetal monocytes (and the maintenance of these cells in adulthood) (25, 26). This latter finding harkens back to the early observation that the lung was the best source for high levels of the cytokine (2). Indeed, the impaired generation of alveolar macrophages with GM-CSF deficiency compromises surfactant clearance and catabolism, triggering the human pathology pulmonary alveolar proteinosis, a disorder that may be ameliorated through exogenous GM-CSF administration (27).
Metcalf (8) also posited that the CSFs might contribute to leukemogenesis, and consistent with this idea, rare forms of pediatric chronic myeloid leukemia involve hypersensitivity to GM-CSF, reflecting alterations in downstream signaling pathways (28). GM-CSF–deficient mice additionally display susceptibility to a variety of hematologic and solid malignancies (in conjunction with IFN-γ loss), raising the possibility that the cytokine may have a broader role in tumor protection (29, 30). In accord with this notion, two GM-CSF–based cancer immunotherapies received regulatory approval for the treatment of advanced prostate carcinoma and melanoma (31, 32), and clinical trials of these in combination with checkpoint blockade are underway (33).
Notwithstanding the ability of GM-CSF to enhance tumor immunity and resistance to infection, the factor may also contribute to immune suppression through the induction of inhibitory myeloid cells and Foxp3+ regulatory T cells (34, 35). Correspondingly, GM-CSF deficiency appears to be involved in some autoimmune pathologies, including inflammatory bowel disease, where some patients harbor loss-of-function variants of the GM-CSF receptor or autoantibodies that block the activity of the cytokine (36, 37). This multiplicity of function is perhaps one of the most fascinating aspects of GM-CSF biology, an ongoing mystery that emerges from the rich heritage of curiosity-driven inquiry into the regulation of hematopoiesis.
G.D. is a current employee and stock holder in Novartis, a previous consultant to Novartis, and has received research support from Novartis and Bristol-Myers Squibb.