(Received for publication, October 7, 1996, and in revised form, December 23, 1996)
From the Department of Pharmacology, New York University Medical Center, New York, New York 10016
Most growth factors and cytokines activate their receptors by inducing dimerization upon binding. We have studied binding of the dimeric cytokine stem cell factor (SCF) to the extracellular domain of its receptor Kit, which is a receptor tyrosine kinase similar to the receptors for platelet-derived growth factor and colony-stimulating factor-1. Calorimetric studies show that one SCF dimer binds simultaneously to two molecules of the Kit extracellular domain. Gel filtration and other methods show that this results in Kit dimerization. It has been proposed that SCF-induced Kit dimerization proceeds via a conformational change that exposes a key receptor dimerization site in the fourth of the five immunoglobulin (Ig)-like domains in Kit. We show that a form of Kit containing just the first three Ig domains (Kit-123) binds to SCF with precisely the same thermodynamic parameters as does Kit-12345. Analytical ultracentrifugation, light scattering, and gel filtration show that Kit-123 dimerizes upon SCF binding in a manner indistinguishable from that seen with Kit-12345. These data argue that the fourth Ig-like domain of Kit is not required for SCF-induced receptor dimerization and provide additional support for a model in which bivalent binding of the SCF dimer provides the driving force for Kit dimerization.
Ligand-induced receptor dimerization provides the mechanism for
transmembrane signaling by many receptors with a single transmembrane domain (1-3). Crystallographic and other biophysical studies have
shown that this results from bivalent binding of the cognate ligand to
the receptors for human growth hormone
(hGH)1 (4, 5) and interferon- (IFN-
)
(6). hGH is a monomeric cytokine that has two receptor binding sites.
Each site binds to a separate receptor molecule, forming a 1:2
(ligand:receptor) complex (4), with a 1:1 monomeric receptor complex
occurring as an intermediate (4, 7). An x-ray crystal structure of the
hGH·hGH-R complex shows that, in addition to ligand-mediated interactions, receptor-receptor interactions also contribute to stabilization of the receptor dimer (5). IFN-
is a dimeric cytokine,
and one IFN-
dimer binds simultaneously to two molecules of the
-chain of its receptor. An x-ray crystal structure of the complex
between IFN-
and the extracellular domain of the IFN-
receptor
-chain (IFN-
R
) shows that the dimer is stabilized solely
through ligand-mediated interactions (6). In another case of
ligand-induced receptor oligomerization for which high resolution
structural information is available, that of the tumor necrosis factor
receptor, the ligand (tumor necrosis factor-
) is a trimer that binds
simultaneously to three receptor molecules. Each receptor extracellular
domain contacts two protomers of the ligand trimer, and no direct
receptor-receptor contact is seen in the trimeric complex (8).
Although no detailed structural information exists for activated receptor tyrosine kinases in complex with their ligands, many reports indicate that bivalent binding of growth factors is key for inducing receptor dimerization (reviewed in Refs. 2 and 3). For example, platelet-derived growth factor (PDGF) is a disulfide-linked dimer that binds to two PDGF-receptor molecules (9, 10). The ligands for the Trk receptors also form dimers to which two receptor molecules can bind, resulting in receptor dimerization (11). Finally, acidic fibroblast growth factor (aFGF), although itself monomeric and incapable of inducing dimerization of its receptor, forms a multivalent complex with heparan sulfate proteoglycans that can in turn bind to two or more receptors and thus stabilize active FGF receptor dimers (12).
In the process of our investigations of the mechanisms of growth factor-induced dimerization of receptor tyrosine kinases, we have studied the binding of stem cell factor (SCF) to the extracellular domain of its receptor, Kit, a class III receptor tyrosine kinase with similarity to the receptors for colony stimulating factor-1 (CSF-1) and PDGF (1). Kit contains a ligand-binding extracellular domain of 520 amino acids, a single transmembrane domain, and an intracellular tyrosine kinase domain. Like CSF-1 and PDGF, SCF is dimeric in solution (13), and has been shown to induce the dimerization of both the full-length Kit receptor (with resulting activation) and the isolated Kit extracellular domain (14, 15). As with the PDGF and CSF-1 receptors, the Kit extracellular domain consists of five immunoglobulin (Ig)-like domains. The first three of these domains are primarily responsible for SCF binding to Kit, as shown in interspecies domain swapping experiments (16) and studies of monoclonal antibodies that inhibit SCF binding (17). The remaining two Ig-like domains have been proposed to play a role in stabilizing the SCF-induced Kit dimer (18). A mechanism for Kit activation has been proposed in which monovalent binding of SCF to a single molecule of Kit induces a conformational change in the receptor that in turn exposes a receptor dimerization site (15). It was further reported (18) that the fourth Ig-like domain of Kit represents the key ligand-induced receptor dimerization site. Monoclonal antibodies against this domain inhibited SCF-induced Kit activation; deletion of the fourth domain abolished Kit function (and reduced SCF binding); and a form of the Kit extracellular domain containing only the first three Ig-like domains did not form cross-linked dimers upon SCF treatment.
In most studies, growth factor receptor dimerization has been analyzed in intact cells or in detergent-solubilized preparations, using approaches such as covalent cross-linking and antibody inhibition that do not permit a quantitative analysis. We have generated large quantities of the extracellular domains of several growth factor receptors for structural studies, but have so far been unable to obtain sufficiently good crystals for x-ray crystallographic analysis. In the absence of a detailed structural view of growth factor-induced receptor dimerization, we have used several biophysical methods to explore this process. In this report, we describe studies of SCF binding to the isolated Kit extracellular domain, using titration calorimetry, analytical ultracentrifugation, and size-exclusion chromatography. We show that one bivalent SCF dimer binds simultaneously to two molecules of Kit, and thus induces its dimerization. By contrast with the findings of Blechman et al. (18), we find that a version of Kit containing just the first three Ig-like domains (Kit-123) binds to SCF and dimerizes in an manner identical to that seen with the complete extracellular domain (Kit-12345). Our data provide further evidence for the importance of SCF bivalence in Kit dimerization, through an explicit measurement of binding stoichiometry, and refute the proposal that a significant ligand-induced Kit dimerization site is present in the fourth Ig-like domain of Kit.
cDNA encoding amino acids 1-141 of human SCF was subcloned into pET11a (Novagen) and expressed in Escherichia coli MGT7. After cell lysis, SCF in the insoluble pellet fraction was solubilized with 8 M urea, from which it was refolded by dilution in the presence of 1 mM reduced glutathione essentially as described by Langley et al. (19). The refolded material was purified by anion exchange chromatography in 10 mM Tris-HCl, pH 8.0, using a FastFlow Q column (Pharmacia Biotech Inc.) followed by gel filtration on a Superdex 75 fast protein liquid chromatography (Pharmacia) column in 50 mM Tris-HCl, pH 8.0, 250 mM NaCl. The resulting material was better than 95% pure as assessed by Coomassie staining of an overloaded SDS-polyacrylamide gel.
For expression of Kit-12345 (residues 1-520) and Kit-123 (residues
1-319), stop codons were introduced by site-directed mutagenesis at
the desired positions in the cDNA encoding human c-Kit. Mutated Kit
cDNA fragments were then subcloned between the BamHI and
HindIII sites of the BlueBac III baculovirus transfer vector
(Invitrogen), and plasmid was cotransfected with BaculoGold DNA
(Pharmingen) into Sf9 insect cells. Recombinant virus stocks were
amplified using HighFive cells (Invitrogen) to yield a high titer virus stock. Protein was secreted into medium by Sf9 cells infected with this
virus stock for 48-72 h. Kit extracellular domains were then purified
from conditioned medium using immunoaffinity chromatography (for
Kit-12345) or a column containing SCF on CnBr-activated beads (for
Kit-123). Kit protein was eluted using ActiSep, dialyzed exhaustively
against 20 mM HEPES, pH 7.5, and further purified by
anion-exchange chromatography on a Mono-Q column followed by gel
filtration using a Superose-12 fast protein liquid chromatography column (Pharmacia). The final protein product was seen to be greater than 95% pure in each case by Coomassie staining of overloaded SDS-polyacrylamide gels. Protein concentrations were measured spectrophotometrically, using calculated molar extinction coefficients (9970 M1 cm
1 for SCF, 36800 M
1 cm
1 for Kit-123 and 68250 M
1 cm
1 for Kit-12345).
Quantitative amino acid analysis of several samples with known
absorbance showed these values to be correct to within 10%.
The OMEGA instrument from MicroCal (20) was used, to which access was kindly provided by Professor Julian Sturtevant (Department of Chemistry, Yale University). Titrations were performed at 25 °C in 50 mM HEPES, 150 mM NaCl, pH 7.5, against which both SCF and Kit protein had been exhaustively dialyzed prior to titration. Controls for heats of dilution were performed as described previously (21) and found to remain constant throughout each titration. Values for the heat of dilution were subtracted from the heats measured in each titration prior to fitting the data. The value for [sites]/KD (c value) in each experiment was between 20 and 60. Experimental titration curves were fit using ORIGIN software (MicroCal) as described elsewhere (20), assuming a single class of binding site as indicated by the data.
Dynamic Light Scattering, Size-exclusion Chromatography, and Analytical UltracentrifugationAll experiments were performed at 25 °C in 50 mM HEPES, 150 mM NaCl, pH 7.5. Dynamic light scattering experiments employed the Protein Solutions instrument, to which access was kindly provided by Prof. Rashmi Hegde (Skirball Institute).
Size-exclusion chromatography was performed using a Bio-Sil SEC 250 column (300 × 7.8 mm) with a Bio-Sil guard column (80 × 7.8 mm) (Bio-Rad) attached to an LKB high performance liquid chromatography system, running at 0.5 ml/min.
Sedimentation equilibrium experiments employed the XL-A analytical ultracentrifuge (Beckman) in the laboratory of Professor Steven K. Burley (Rockefeller University), kindly aided by Elaine Halay. Concentration distributions were monitored at 230, 260, or 280 nm. Experiments were performed at 25 °C using six-channel cells at three different speeds (8500, 10200, and 10500 rpm) with identical results. Kit-123 and SCF were analyzed separately at the concentrations shown in Table II. For analysis of the SCF·Kit-123 complex, an equimolar mixture of the two components was run at the Kit-123 concentrations noted. Samples were equilibrated at each speed for greater than 18 h. Protein partial specific volumes were estimated from their amino acid compositions (0.74 ml/g for SCF, and 0.715 ml/g for Kit-123, assuming approximately 15% (w/w) oligosaccharide), and solvent density was taken as 1.003 g/ml. Data were fit using the Optima XL-A data analysis software (Beckman/MicroCal). In each case, a fit to a single ideal species gave the most randomly scattered residuals.
|
Using a baculovirus expression system we have generated and purified milligram quantities of the complete extracellular domain of human Kit (Kit-12345), as well as a form from which Ig-like domains 4 and 5 have been removed (Kit-123) (see "Experimental Procedures"). Human SCF (1-141) was expressed in E. coli, using a T7 expression system, and was refolded and purified using a procedure described by Langley et al. (19). Binding of SCF to the two different forms of the Kit extracellular domain was studied using isothermal titration calorimetry (ITC). SCF-induced dimerization of Kit was also monitored, using size-exclusion chromatography (SEC), analytical ultracentrifugation, chemical cross-linking and dynamic light scattering. We find that both Kit-12345 and Kit-123 behave identically with respect to both SCF binding and their resulting dimerization. A single bivalent SCF dimer binds simultaneously to two molecules of Kit-12345 or Kit-123, resulting in dimerization of the receptor extracellular domain.
Oligomeric State of SCF and Kit Extracellular DomainBefore
proceeding with an analysis of ligand-induced receptor dimerization, it
is important to determine the oligomeric state of the ligand and the
receptor extracellular domains under the experimental conditions
employed. Using SEC and analytical ultracentrifugation, we found that
both unliganded Kit-12345 and unliganded Kit-123 behave as monomers of
their expected molecular weight (plus mass of carbohydrate), and that
free SCF (1-141) behaves as a dimer of 28 kDa, as expected (see Table
II and Fig. 2).
Binding of SCF to Kit-12345
ITC was used to monitor SCF
binding to Kit-12345. The data obtained upon titration of SCF into a
solution of Kit-12345 (present in the calorimeter cell) demonstrate the
existence of a single class of thermodynamically equivalent binding
sites (Fig. 1, upper panel). Since heat is
released throughout the titration, the binding reaction is exothermic
(H =
8.7 kcal·mol
1) and
predominantly enthalpy-driven at 25 °C. Fitting of several titrations such as that in Fig. 1 gave an apparent binding constant (KB) for SCF binding to Kit-12345 of 1.8 × 107 M
1
(KD(app) = 55 nM) (Table
I), and a stoichiometry of 1:1 (SCF:Kit-12345). Since
SCF is a dimer under the conditions of this experiment, this measured
stoichiometry requires that a single dimer of SCF binds simultaneously
to two molecules of Kit-12345, thus inducing Kit-12345 dimerization.
Identical titrations were obtained when the experiment was reversed,
with Kit-12345 being titrated into a solution of SCF (not shown). The
apparent KD reported here (55 nM) is
higher (by around 20-fold) than values reported for SCF binding to Kit
on the surface of cells (KD
2 nM)
(15, 16) or to the immobilized Kit extracellular domain (14). Similar
differences occur with other receptors, for example those for aFGF
(12), EGF (22), and hGH (7) when ligand binding to the extracellular
domain (free in solution) is compared with binding to the intact
receptor. There are several likely explanations for this difference.
Studies of SCF binding to intact Kit have required the interpretation of nonlinear Scatchard plots (15-18), which is plagued with
quantitative difficulties (23, 24). The fact that the ligand in this
case (SCF) is also bivalent creates further difficulties in
interpretation (25) that have not been adequately addressed. In
addition to these considerations, the apparent affinity will be
elevated by an avidity effect that arises from the high local
concentration of receptor when restricted to the cell membrane or
immobilized artificially. Furthermore, each of the binding events is
coupled to receptor dimerization, and it can readily be shown that
dimerization of membrane-localized receptors is greatly favored over
their association when tumbling freely in solution (26). Therefore, it
is expected that the apparent KD for ligand
interacting with the free receptor extracellular domain will be higher
than that for binding to the membrane-localized receptor. In studies similar to those reported here, Philo et al. (27) determined a KD for SCF/Kit-12345 binding, when both components were free in solution, that agrees closely with our value.
|
Identical ITC studies were repeated
for Kit-123. Titrations of SCF into Kit-123, and of Kit-123 into a
solution of SCF, gave binding isotherms that were indistinguishable
from those obtained with Kit-12345 (Fig. 1, lower panel).
KD(app) for this interaction was 49.2 nM, H was
10.2 kcal·mol
1,
and the stoichiometry was close to 1:1 (SCF:Kit-123). These thermodynamic parameters are indistinguishable from those obtained with
Kit-12345 (Table I). Therefore, removal of Ig-like domains 4 and 5 from
the Kit extracellular domain has no influence upon the thermodynamics
of SCF binding. This result is in agreement with the reports that
Ig-like domains 1-3 are of primary importance in ligand binding to Kit
(16, 17) and shows that domains 4 and 5 do not contribute at all.
However, since SCF binding is coupled to the dimerization of both
Kit-12345 and -123 (see below), our results are inconsistent with the
hypothesis (18) that Ig-like domains 4 and/or 5 contribute in a
significant way to SCF-induced Kit dimerization.
To further test
our interpretation of the ITC results for SCF binding to Kit-12345, we
used chemical cross-linking and SEC to monitor dimerization of the
receptor extracellular domain upon ligand binding. In agreement with
previous reports (14, 18), we could detect significant
SCF-dependent formation of covalently linked Kit-12345
dimers in chemical cross-linking experiments (data not shown).
SCF-induced dimerization was also apparent in SEC studies (Fig.
2). Kit-12345 loaded alone eluted at a position corresponding well to its expected monomeric molecular weight (80 kDa).
Addition of SCF at the molar ratios noted in Fig. 2 resulted in the
appearance of a new peak, eluting at the position expected for dimeric
Kit-12345 with a single bound SCF dimer (190 kDa). Kit-12345 eluted
exclusively at this dimeric position upon addition of one molar
equivalent of SCF, confirming the 1:1 (one SCF dimer per two molecules
of Kit-12345) stoichiometry determined by ITC. Ligand added in excess
of this ratio elutes as free SCF, with no further alteration of either
the position or amplitude of the
(SCF)2:(Kit-12345)2 peak. To determine whether
a complex containing a single Kit-12345 molecule bound to an SCF dimer
could be detected, the SCF:Kit-12345 ratio in the loaded samples was further increased, up to 100:1. Even at these ligand:receptor ratios
the position and amplitude of the
(SCF)2:(Kit-12345)2 peak was unaffected (data
not shown). This result suggests that the stability of the dimeric
(SCF)2·(Kit-12345)2 complex is greater than
that of a monomeric (SCF)2·Kit-12345 complex under the
experimental conditions explored here.
Identical experiments
employing SEC were also performed with the truncated form of Kit
containing only the first three immunoglobulin domains (Kit-123). The
results obtained using Kit-123 were identical to those obtained with
Kit-12345 (Fig. 3), showing that one dimer of SCF binds
to Kit-123 and induces its dimerization. As with the complete
extracellular domain, no evidence for the formation of a complex
containing a single Kit-123 molecule bound to an SCF dimer was
obtained. These results were corroborated in studies employing dynamic
light scattering, which gave an approximate molecular mass of 62 kDa
for Kit-123 alone, and 114 kDa for the SCF·Kit-123 complex; again
suggesting SCF-induced Kit-123 dimerization. However, in agreement with
the finding of Blechman et al. (18), we could not detect
SCF-induced Kit-123 dimerization by chemical cross-linking analysis. We
therefore sought additional evidence for SCF-induced Kit-123
dimerization, performing sedimentation equilibrium experiments with an
analytical ultracentrifuge. As reported in Table II, SCF
(1-141) itself sedimented in the ultracentrifuge as a single species
that represents a dimer of 28.2 ± 1.2 kDa, while unliganded
Kit-123 sedimented as a monomer of 29.2 ± 1.5 kDa (slightly below
the expected molecular mass). When a series of concentrations of a 1:1
mixture of SCF and Kit-123 was analyzed, the residuals for the fits to
the data were most clearly random, for each of the three rotor speeds
employed, when a single ideal species was assumed (Fig.
4). This single species sedimented with a molecular mass
of 79.3 ± 1.5 kDa, which can be accommodated (given the known
stoichiometry) only by the formation of a complex containing one SCF
dimer and two molecules of Kit-123 (Table II). Thus, with the exception
of chemical cross-linking studies, each of the physical techniques
employed here shows that Kit-123, like Kit-12345, dimerizes
quantitatively upon binding of one SCF dimer to two molecules of
Kit-123.
The Fourth Immunoglobulin-like Domain of Kit Is Not Required for Dimerization
Since dimerization of the Kit extracellular domain is coupled to SCF binding (Figs. 2 and 3), and the thermodynamics of SCF binding are unaffected by the removal of Ig domains 4 and 5 (Table I), the fourth Ig-like domain of Kit cannot represent a thermodynamically significant receptor dimerization site. The model of Blechman et al. (18) proposes that one SCF dimer binds to one Kit molecule, and induces a conformational change that exposes a key receptor dimerization site in the fourth Ig-like domain. Kit dimerization is suggested not to occur when this site is blocked or removed. According to this model, Kit-123 should not dimerize upon SCF binding. Furthermore, its SCF binding should be weaker than that of Kit-12345 since bivalent ligand binding to the latter is coupled to receptor dimerization. Any loss of receptor-receptor interactions involving the putative dimerization site in domain 4 should therefore be clearly reflected in a comparison of the thermodynamics of SCF binding to Kit-12345 and Kit-123. We could detect no difference in SCF binding to the two forms of Kit, and our data show that Kit-123 dimerizes upon SCF binding in a manner indistinguishable from that seen for Kit-12345. The fact that we were unable to detect the binding of one SCF dimer to a single Kit-123 or Kit-12345 monomer does argue that receptor-receptor interactions may participate in Kit dimerization to some extent. However, these interactions are not significantly impaired upon removal of both Ig-like domains 4 and 5.
The role of the fourth Ig-like domain in Kit dimerization was primarily
suggested by the ability of a monoclonal antibody raised against this
domain to inhibit SCF-induced Kit activation and to prevent
cross-linking of SCF to full-length Kit dimers (18). Furthermore, it
was found that full-length Kit from which the fourth Ig-like domain had
been deleted (Kit-4) could not be activated and did not form
SCF-induced dimers in cross-linking experiments. Finally, Kit-123 could
not be shown to dimerize upon SCF binding in chemical cross-linking
experiments, while Kit-12345 could (18). The biophysical analyses
presented here, using four different methods, show clearly that Kit-123
dimerizes upon SCF binding, with parameters that are indistinguishable
from those seen with Kit-12345. We therefore suggest the possibility
that the monoclonal antibody against the fourth Ig-like domain employed by Blechman et al. (18) could exert its influence through
steric effects upon receptor dimerization. The inability to cross-link Kit-123 dimers in the presence of SCF, which we also experienced, could
simply reflect a difference in the availability of reactive groups in
the two forms of the Kit extracellular domain. Clearly, from our SEC,
dynamic light scattering, and centrifugation studies, SCF does induce
Kit-123 dimerization. The results reported with the Kit-
4 mutant
that would not dimerize or signal (18) are more difficult to explain,
but could reflect deleterious conformational alterations. Indeed, it
seems surprising that, while the extracellular forms Kit-12345,
Kit-1234, and Kit-123 are reported to bind SCF with similar affinities
(18), the full-length Kit-
4 mutant bound SCF 10-fold more weakly
than did intact wild type Kit.
The results
presented here demonstrate that a single SCF dimer binds to two
molecules of the Kit extracellular domain (either Kit-12345 or
Kit-123), thus inducing receptor dimerization. The fact that no other
binding events were detectable in ITC or SEC studies argues strongly
that bivalence of the SCF dimer is important in its induction of Kit
dimerization. If bivalent binding of SCF were the only driving force
for receptor dimerization, it should be possible to disrupt the
(SCF)2:(Kit)2 dimer into monomeric (SCF)2·Kit complexes by addition of excess SCF. Such
dimer disruption has been demonstrated for bivalent binding of hGH to
its receptor (4). We were unable to detect formation of the monomeric
(SCF)2·Kit complexes upon addition of excess SCF in our
SEC studies, which suggests that receptor-receptor interactions may
cooperate with bivalent SCF binding to stabilize the Kit extracellular
domain dimer. A schematic model for SCF-induced Kit dimerization is
presented in Fig. 5, where it is compared with models
for ligand-induced dimerization of several other receptors. Bivalent
SCF binding to two molecules of the Kit extracellular domain cooperates
with weak receptor-receptor interactions (that are not abolished upon removal of the fourth and fifth Ig-like domains) to drive dimer formation. The model is very similar to those for IFN- binding to
the
-chain of its receptor (6), and binding of an aFGF-heparin complex to the FGF receptor (12, 28). In each of these cases, a
multimeric ligand (HSPG-induced in the case of aFGF) stabilizes receptor oligomers by virtue of its multivalent receptor binding (12,
30). It is also very similar to the model for hGH-induced dimerization
of hGH-R, except that in this case the bivalent ligand is a monomer.
While receptor-receptor contacts are clearly observed in the crystal
structure of the hGH-induced hGH-R dimer (5, 29), no receptor-receptor
contacts are evident in the crystal structure of IFN-
bound to two
extracellular domains of IFN-
R
(6). We have also not observed
thermodynamically significant interactions between the two protomers of
the aFGF-induced dimer of the FGF receptor extracellular domain
(12).
Studies of SCF binding to CHO-cell derived Kit-12345, employing analytical ultracentrifugation, SEC, and ITC were recently reported by Philo et al. (27), with results very similar to those presented here, showing that one SCF dimer binds to two molecules of Kit-12345. The thermodynamic parameters described for CHO-cell derived binding to E. coli-derived SCF (1-165) are in accord with our measurements for E. coli-derived SCF (1-141) binding to baculovirus-generated Kit-12345. In agreement with our finding, Philo et al. (27) did not observe disruption by excess SCF of the (SCF)2·(Kit-12345)2 dimeric complex in SEC studies. However, detailed model-fitting of analytical ultracentrifugation data suggested that the (SCF)2·Kit-12345 monomeric complex does occur to a significant extent. It was thus concluded that there is little cooperativity in Kit-12345 binding to the SCF dimer, indicating that inter-receptor interactions are likely to be weak.
According to the conclusions of Blechman et al. (18), if the fourth Ig-like domain represents a key ligand-induced Kit dimerization site, significant cooperativity in Kit-12345 binding to dimeric SCF would be observed, and Kit-123 would not dimerize. Our results and those of Philo et al. (27) make a strong case against this assertion. Possible origins for this discrepancy are outlined above. Lev et al. (15) and Blechman et al. (18) have also argued against the importance of SCF bivalency in inducing Kit dimerization, largely on the basis of the absence of a bell-shaped dose-response curve for SCF (although this is seen in some experiments). Our inability to detect monomeric SCF·Kit complexes represents an analogous observation. However, as pointed out by Wells (30), a bell-shaped dose-response curve (interpreted as an indication of dimer disruption) may be difficult to detect in cases where bivalent ligand binding drives dimerization. With the cooperation of bivalent ligand/receptor interactions and receptor/receptor interactions, extremely high ligand concentrations may be required to disrupt the dimer. For example, the concentration of hGH required to disrupt an hGH-induced dimer of membrane-bound hGHR dimer is some 10,000-fold higher than the hGH concentration required for a maximal cellular response. Such high excesses have not been studied for SCF. As with hGH/hGH-R, therefore, we argue that the subunits of an activated Kit dimer are cooperatively linked both by a bivalent cross-linking ligand and receptor-receptor interactions.
We thank Julian Sturtevant (Yale University) for kindly allowing access to the MicroCal OMEGA titration calorimeter (supported by National Institutes of Health Grant GM04725); Stephen Burley and Elaine Halay (Rockefeller University) for access to and assistance with the Beckman XL-A analytical ultracentrifuge; and Rashmi Hegde for access to the dynamic light scattering instrument. We are also grateful to Sima Lev and James Rice (Sugen, Inc.) for their help with SCF production, and Kathryn Ferguson, James A. Wells, and Donald Engelman for their critical reading of the manuscript.