Perverted responses of the human granulocyte-macrophage colony-stimulating factor receptor in mouse cell lines due to cross-species β-subunit association

Human granulocyte-macrophage colony-stimulating factor (hGM-CSF) is a pleiotropic cytokine that stimulates the proliferation, differentiation, and survival of myeloid precursors and induces the effector functions of mature myeloid cells.1,2 These multiple functions are mediated by binding to high-affinity receptors that comprise a GM-CSF–specific α chain (hGMR-α) and an affinity-converting, signal-transducing β subunit (hβ_c), which is shared with the interleukin-3 (IL-3) and IL-5 receptors.3,4 In the murine system there are 2 β subunits, mβ_c, which is analogous to hβ_c, transducing signals induced by mGM-CSF, mIL-3 or mIL-5, and mβ_IL-3, which is specific for mIL-3.5,6Mouse myeloid cell lines, for instance FDCP-1 and BaF-B03, express both mouse β subunits.

Given the importance of hβ_c in transducing signals that regulate immune responses, hematologic recovery, and, in some cases, leukemia, a significant amount of work has been devoted to structure-function analysis of these cytokine-receptor systems. Studies have led to the engineering of cytokine analogs with unique properties such as the hGM-CSF mutant E21R that behaves as a specific hGM-CSF antagonist.7 Characterization of the receptors has sought to identify functional regions involved in receptor activation and has identified regions in the cytoplasmic domain of hβ_c that couple to specific signaling molecules such as JAKs, STATs, and the ras/MAP kinase pathway.8-10 However results from these studies have in some cases been ambiguous or even conflicting.

Central to the analysis of cytokine and receptor mutants is the choice of experimental system. Predominantly, mouse myeloid cell lines, which are readily transfected with human receptor subunits, have been used. A major problem is the expression of endogenous receptors for these cytokines in these cell lines. For example, transfection of hGMR-α alone in murine FDCP-1 cells was initially reported to be sufficient to mediate a proliferative signal despite only displaying low-affinity hGM-CSF binding.11 However, it was later shown that functional reconstitution of hGMR required both hGMR-α and hβ_c subunits,12 and that the initial observation with hGMR-α alone was confounded by the recruitment of endogenous mβ_c.13 Likewise, interaction of an extracellular point mutant of hβ_c with an endogenous mGMR-α has been shown to lead to factor-independent proliferation14 and chimeras between a constitutively active erythropoietin receptor with the cytoplasmic domain of GMR-α promoted proliferation but only in presence of mβ_c.15 Here we show an abnormal response of the human GM-CSF antagonist E21R in transfected mouse cell lines and the molecular basis for human-mouse GM-CSF receptor cross-talk. These results emphasize the need for caution when interpreting data from experiments using transfectants and the desirability of using homologous systems.

The human erythroleukemic cell line TF1.8, the megakaryocytic leukemia human cell line, UT7, and the murine pro-B-cell line, BaF-B03 (transfected with human GMR-α and hβ_c), were grown as previously described.16,17 Following cytokine starvation overnight, the proliferation of cell lines in response to hGM-CSF or E21R was determined as previously described.18 E21R, a single point mutant of hGM-CSF (Gln21Arg) was donated by Bresagen (Adelaide, Australia).

Murine monoclonal antibodies 1C1 and 8E4, for detection of hβ_c, and 4H1 for immunoprecipitating hGMR-α were produced as previously described.16 1C1 was biotinylated using a cellular labeling and immunoprecipitation kit (Boehringer Mannheim, Rose Park, SA, Australia) and streptavidin–horseradish peroxidase (HRP) was purchased from Amersham Life Science (Little Chalfont, United Kingdom). Anti-mβ_c rabbit polyclonal K-17 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). HRP-conjugated antibodies specific for mouse or rabbit immunoglobulins were purchased from Pierce (Rockford, IL) and Dako (Botany, NSW, Australia), respectively. The HRP-conjugated antiphosphotyrosine monoclonal antibodies, PY20 and 4G10, were obtained from Transduction Laboratories (Lexington, KY) and Upstate Biotechnology (Lake Placid, NY), respectively.

Analysis of ligand-induced receptor complexes and their phosphotyrosine status was determined using starved cells that were stimulated with factors for 5 minutes at room temperature at indicated concentrations. Cells were lysed and immunoprecipitated proteins were run on reducing sodium dodecylsulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and subjected to immunoblotting as previously described.19 

Human GM-CSF supports the proliferation of human cells by activating endogenous hGMR.7 The antagonistic hGM-CSF mutant, E21R, however, cannot mediate a proliferative response in human cells and is unable to bind hGMR with high affinity.7 We observed that the biologic activity of E21R is strikingly different when analyzed in hGMR-transfected murine cell lines. First, the biologic activity of E21R was determined in comparison to wild-type hGM-CSF on the human cell line, TF1.8, and the murine cell line BaF-B03 transfected with both hGMR-α and hβ_c (hGMR BaF-B03). Human GM-CSF induces proliferation of TF1.8 cells with an effective dose (ED₅₀) of 0.03 ng/mL, whereas E21R is unable to elicit a proliferative response and antagonizes the activity of wild-type GM-CSF (Figure 1A).7Surprisingly, however, E21R induced a proliferative response in the hGMR BaF-B03 cell line at concentrations of E21R above 10 ng/mL (Figure 1B).

It has been previously described that hGMR-α can interact with either hβ_c or mβ_c to transmit a growth signal in response to hGM-CSF in BaF-B03 cells.13 However, the interaction of hGMR-α with mβ_c, unlike hβ_c, does not form a high-affinity receptor. The proliferation induced by E21R in hGMR BaF-B03 cells is also the result of a low-affinity interaction with a dissociation constant of approximately 4 nM (data not shown) as determined by Scatchard analysis of saturation binding,18 identical to the affinity of E21R measured on human neutrophils.7 

The abnormal proliferative response of hGMR BaF-B03 cells to E21R differs from the previously described cross-talk phenomenon between hGMR-α and mβ_c because it is dependent on the coexpression of both hGMR-α and hβ_c. This is highlighted by the inability of BaF-B03 cells expressing hGMR-α alone to respond to E21R, even at concentrations of 100 μg/mL (data not shown). Because the abnormal agonism was only observed in cell lines where mβ_c is endogenously expressed, it suggested that the unexpected activity demonstrated by E21R may be the result of a novel cross-species interaction, involving both hβ_c and mβ_c.

Together with the inability of E21R to mediate a proliferative response in human cells, it cannot stimulate tyrosine phosphorylation of hβ_c. The tyrosine phosphorylation status of hβ_c was investigated to determine if E21R can activate hβ_c in this murine system. Tyrosine phosphorylation of hβ_c was readily detected in the murine cell line hGMR BaF-B03 after stimulation with E21R (Figure2B), but not the human cell line expressing endogenous hβ_c, UT7 (Figure 2A), and TF1.8 (data not shown). In comparison, tyrosine phosphorylation of hβ_c was detected following stimulation with hGM-CSF in both human UT7 cells and murine hGMR BaF-B03 cells (Figure 2A,B). The dose response of E21R-induced tyrosine phosphorylation of hβ_c is consistent with the proliferation data where responsiveness occurs at a concentration of E21R above 10 ng/mL, with no response at 1 ng/mL (Figure 1B).

Because mβ_c functions when recruited to a hGM-CSF–hGMR-α complex, we investigated if this molecule plays a role in facilitating the tyrosine phosphorylation of hβ_cin response to hGM-CSF or E21R in hGMR BaF-B03 cells. Stimulation of hGMR BaF-B03 cells with either GM-CSF or E21R resulted in tyrosine phosphorylation of hβ_c (Figure 2C). Significantly, mβ_c was associated with hβ_c and this interaction appears to occur regardless of stimulation (Figure 2C). It is interesting to note that dimerization of a least 2 hβ_csubunits is required for hGMR activation.19,20 The interaction between mouse and human β_c may provide the molecular basis to support the agonistic activity of E21R and mediate this perverted receptor response observed in this murine cell line.

Phosphotyrosine immunoprecipitations were performed to address if preassociation between the human and murine β_c subunits allows mβ_c to be associated with an activated hGMR complex. Human β_c was immunoprecipitated by an antiphosphotyrosine antibody following stimulation of hGMR BaF-B03 cells with hGM-CSF but not when left unstimulated or with an isotype-matched control antibody (Figure 2D). Mouse β_cappears to be weakly phosphorylated in the absence of stimulation, but interestingly was strongly coimmunoprecipitated with antiphosphotyrosine antibody after stimulation with hGM-CSF (Figure2D), suggesting an increase in phosphorylation of mβ_c on stimulation with the human ligand.

The hGMR subunits GMR-α and hβ_c have previously been shown to exist as a preformed complex on human cells,20,21and we have now shown that mβ_c and hβ_c are also associated prior to ligand stimulation on hGMR BaF-B03 cells. To determine whether mβ_c is also a component of a preformed complex on hGMR BaF-B03 cells, hGMR-α was immunopreciptated from nonstimulated and GM-CSF–stimulated cells. Immunoblotting revealed that both hβ_c and mβ_c were associated with hGMR-α regardless of stimulation (data not shown) suggesting that mβ_c chain may be a component of a preformed hGMR receptor on hGMR BaF-B03 cells. Therefore activation of the receptor by GM-CSF and E21R may be mediated by a preformed hGMR-α–hβ_c–mβ_c complex in these cells.

The ability of human and mouse β_c subunits to interact and influence responses of a hGM-CSF variant introduces a new level of complexity in the analysis of ligand-receptor interaction and subsequent signaling capabilities. It also raises the question of the influence this interaction has had on previous hGMR studies that have been performed in murine cell lines. Clearly mβ_c can become associated within an active hGMR complex, and it is important to consider its contribution to the receptor's biologic response. This is of concern especially when investigating the activation of downstream signaling molecules.22 In light of the interaction between hβ_c and mβ_c it will be difficult to discriminate which signaling molecules have emanated from the hβ_c alone.23 In addition, the associated mβ_c subunit may facilitate signaling by an otherwise inactive hβ_c mutant. This may explain the surprising result seen where hβ_c deficient in intracellular tyrosines is still capable of responding to hGM-CSF.10 In addition, the observation that a cytoplasmically truncated hβ_c can still respond to hGM-CSF when expressed in BaF-B03 cells may be potentiated via signaling through mβ_c.24 The contribution this interspecies subunit interaction plays when distinguishing regions in hGMR responsible for differentiation or proliferation in murine myeloid cell lines remains unclear. It has been demonstrated previously that hβ_c can form homodimers that are activated in response to hGM-CSF.20,21 The association between mβ_cand hβ_c must result in a complex that has different properties to hβ_c homodimers because E21R is unable to activate the latter.

A number of receptors belonging to the cytokine receptor family show cross-species specificity between human and murine components. Because receptor dimerization is a common theme in receptor activation, it cannot be ruled out that interaction of interspecies signaling components may obscure results when studying a wide range of human receptors in murine cells and similar precautions may need to be taken. In addition to our findings that the use of transfected murine cell lines may not be the best approach when studying hGM-CSF ligand variants, other cytokines show similar discrepancies. An analogous situation has been reported for a human IL-4 mutant that can mediate either agonistic or antagonistic responses when studied in either a murine or human cell line, respectively.25 Similarly, mutational analysis of thrombopoietin shows conflicting data in the identification of residues functionally important in interaction with its receptor when screened on Mpl-transfected BaF-B03 cells compared to enzyme-linked immunosorbent assay or Biacore analysis.26,27 

The demonstration of an inherent association between mouse and human β_c shown here highlights the need for a careful selection of appropriate systems. In the case of the hGMR the availability of β_c knockout mice permits the use of cells from these animals for receptor reconstitution experiments as a better background to analyze structural and functional outcomes with the hGMR.