Ligand binding characteristics of the carboxyl-terminal domain of the cytokine receptor homologous region of the granulocyte colony-stimulating factor receptor.

The carboxyl-terminal domain (BC domain, roughly 100 amino acid residues) of the cytokine receptor homologous region in the receptor for murine granulocyte colony-stimulating factor was secreted as a maltose binding protein fusion into the Escherichia coli periplasm. The murine BC domain was prepared from the fusion protein by restriction protease factor Xa digestion and was purified to homogeneity. The purified BC domain specifically and stoichiometrically bound granulocyte colony-stimulating factor. This result indicates that the BC domain is also critical for ligand binding, as shown for the amino-terminal domain of the cytokine receptor homologous region (Hiraoka, O., Anaguchi, H., Yamasaki, K., Fukunaga, R., Nagata, S., and Ota, Y. (1994) J. Biol. Chem. 269, 22412-22419). The tertiary folding and the beta-sheet structure of the BC domain were confirmed by NMR spectroscopy. The disulfide bond pattern suggested from peptide mapping was Cys224-Cys271 and Cys242-Cys285. Disruption of the disulfide bonds suggested that both bonds are critical for maintaining the folding of the BC domain, although a BC domain lacking the second bond still retained ligand binding activity. Mutational analysis of the WSXWS sequence conserved in the cytokine receptor family suggested that this motif is critical for protein folding rather than for ligand binding.

Granulocyte colony-stimulating factor (G-CSF) 1 is a cytokine that plays an essential role in maintaining the number of neutrophilic granulocytes in peripheral blood and that is responsible for granulocytosis during inflammation (Nagata, 1990;Nicola, 1989;Demetri and Griffin, 1991). Since the administration of G-CSF causes granulopoiesis, it is used to treat patients suffering from granulopoenia (Lieschke andBurgess, 1992a, 1992b). Investigations of the ability of ligand binding to the G-CSF receptor to mediate a specific signal inside the cell may promote clinical applications of G-CSF. The cDNAs for the murine and human G-CSF receptors (ϳ800 amino acid resi-dues) have been isolated (Fukunaga et al., , 1990b. The extracellular region of the G-CSF receptor (ϳ600 amino acid residues) has a composite structure containing an immunoglobulin-like domain, a cytokine receptor homologous (CRH) region, and three fibronectin type III domains (Fukunaga et al., , 1990b. Among them, the CRH region (ϳ200 amino acid residues) was predicted to be the ligand binding region (Bazan, 1990). Thus, knowledge of the CRH region is crucial to understanding the binding mechanism of the G-CSF receptor.
The CRH region was predicted to be cysteine-rich ␤-structure, which was defined by its striking homology to receptors for various cytokines, such as interleukins 2-7, erythropoietin, growth hormone (GH), and prolactin (Bazan, 1990). The CRH region shares two distinctive features in the extracellular region: four conserved cysteine residues, known as the "cysteine motif," are located in the amino-terminal domain of the CRH region (BN domain), and its carboxyl-terminal domain (BC domain) contains a "WSXWS" motif (Bazan, 1990). It is very likely that the cysteine motif constitutes a domain for the interaction of the receptor with its cognate ligand (Ullrich and Schlessinger, 1990). Deletion analysis of the G-CSF receptor cDNA indicated that the deletion of the BN domain completely abolished ligand binding (Fukunaga et al., 1991). Recently, Hiraoka et al. (1994), using an Escherichia coli maltose binding protein (MBP, malE gene product) fusion system, showed that the BN domain of the G-CSF receptor was a discrete folding unit that could bind ligand. On the other hand, the G-CSF receptor with a deletion of the BC domain still retained the ligand binding activity, although the K d value was increased as much as 50-fold, as compared with the intact molecule (Fukunaga et al., 1991). It was reported that the WSXWS motifs of the interleukin 2 (Miyazaki et al., 1991) and erythropoietin (Yoshimura et al., 1992) receptors were essential for ligand binding and signal transduction. These data indicated that the function of the G-CSF receptor BC domain in ligand binding is still unclear and suggested that the expression and purification of the BC domain in quantities suitable for biochemical and structural analyses would be essential for understanding the ligand binding mechanism of this receptor.
In the present study, we expressed the BC domain of the G-CSF receptor, using the E. coli MBP fusion system, and purified the product. The purified BC domain specifically bound ligand. Furthermore, we characterized the functions of the WSXWS motif and the cysteine residues.

Construction of Expression Plasmids for the BC Domain and Its
Mutants-The murine BC (mBC) domain corresponds to exons 7 and 8 of the murine G-CSF (mG-CSF) receptor (Seto et al., 1992) and contains 4 cysteine residues . Since we had developed an MBP fusion protein secretion system for the expression of the murine BN (mBN) domain, which contains many cysteine residues (Hiraoka et al., 1994), we directed the mBC domain fusion protein into the E. coli * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed.  (Hiraoka et al., 1994). The DNA encoding the mBC domain residues was amplified by the polymerase chain reaction from plasmid pBLJ17 encoding the mG-CSF receptor , using the sequence 5Ј-GGGGATCCATC-GAGGGTAGGGGTTCTTCTTTGGAGCCTCCCATGCT-3Ј as the 5Јsense primer and the sequence 5Ј-GGGTCTAGATTACTTCATGGTAG-GCCTCAGCTGCAG-3Ј as the 3Ј-antisense primer. The amplified DNA was ligated within the BamHI/XbaI sites of pMAL TM -P (New England Biolabs) after digestion with the restriction endonucleases BamHI and XbaI. The resultant plasmid pMALp-mBC contains the DNA sequence for the inducible tac promoter followed by the coding sequences: the MBP with its signal, the IEGR of the restriction protease factor Xa recognition sequence (Maina et al., 1988), a GSS coding sequence as a spacer for factor Xa digestion, the mBC domain (Leu 203 -Lys 308 ), and the stop codon. Factor Xa cleaves at the carboxyl terminus of the Arg residue in its recognition sequence, and thus the cleaved product contains the extra residues GSS at the amino terminus of the mBC domain.
Expression and Purification of the mBC Domain-The plasmid pMALp-mBC was introduced into E. coli strain KS474 (Strauch et al., 1989). Transformed E. coli cells were grown in M9 medium (usually a 4-liter culture) containing 0.8% glucose at 23°C to an A 600 of 0.6, and the expression of the MBP-mBC fusion protein was induced by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM. After incubation for 18 h, the cells were pelleted, and the periplasmic fraction was prepared as described (Hiraoka et al., 1994). After the addition of ammonium sulfate with stirring to 20% saturation (11.4 g/100 ml), the solution was applied to a butyl-Toyopearl 650 M (2.5 cm, inner diameter, ϫ 6 cm; TOSOH) column equilibrated with 2 mM sodium phosphate, pH 6.0, containing ammonium sulfate at 20% saturation, and was eluted with a linear gradient of 20 -0% ammonium sulfate. The eluate containing the MBP-mBC fusion protein was dialyzed against 20 mM sodium citrate, pH 5.5, and was applied to an S-Sepharose (1.6 cm, inner diameter, ϫ 10 cm; Pharmacia Biotech Inc.) column equilibrated with 20 mM sodium citrate buffer, pH 5.5. The MBP-mBC fusion protein was eluted with a linear gradient of NaCl from 0 to 0.4 M and was pooled. After dialysis of the pooled fraction, the fraction was applied to a Q-Sepharose (1.6 cm, inner diameter, ϫ 5 cm; Pharmacia) column equilibrated with 20 mM sodium phosphate buffer, pH 6.0, and was eluted with a linear gradient of NaCl from 0 to 0.4 M. The fraction containing the fusion protein was concentrated, and the protein was cleaved by factor Xa protease (1/250 (w/w), New England Biolabs) at 16°C for 20 h in 0.1 M Tris-HCl buffer, pH 7.5. The reaction mixture was applied to a Q-Sepharose (1.6 cm, inner diameter, ϫ 5 cm) column equilibrated with 20 mM sodium phosphate buffer, pH 6.0, and was eluted with the same buffer. It was then applied to an S-Sepharose (1.6 cm, inner diameter, ϫ 2.5 cm) column and was eluted with a linear gradient of NaCl from 0 to 0.4 M. Fractions containing the mBC domain were pooled and concentrated.
Peptide and Disulfide Mapping Methods-A proteolytic digestion of the mBC domain with thermolysin (Sigma) was performed for 3 h at 40°C in 50 mM Pipes, pH 6.5, containing 10 mM CaCl 2 . Cyanogen bromide (1% (w/v)) treatment in 70% formic acid was performed for 24 h at room temperature. Thereafter, pepsin (Sigma) (1/100 (w/w)) digestion was performed for 8 h at 30°C in 5% formic acid. The peptides were separated by C18 reverse phase high performance liquid chromatography (HPLC) (TSKgel ODS-120T; 4.6 mm, inner diameter, ϫ 250 mm; TOSOH), equilibrated with 0.1% trifluoroacetic acid, and eluted with a linear gradient of acetonitrile from 0 to 40%. The fractions were collected manually and lyophilized. The amino acid sequence was determined by a protein sequencer (Applied Biosystems 477A sequencer equipped with a model 120A PTH analyzer). Disulfide bonds were identified by detecting diphenylthiohydantoin (PTH)-cystine in the corresponding cycle (Marti et al., 1987).

Initial Characterization of the Purified mBC Domain-The
MBP-mBC fusion protein (54 kDa) from the E. coli periplasmic fraction was digested with restriction protease factor Xa to separate the mBC domain from the MBP, and the mBC domain was purified to homogeneity as described under "Experimental Procedures" (Fig. 1). The molecular mass of the mBC domain was estimated to be 12.5 kDa by gel filtration HPLC and by 15% polyacrylamide gel electrophoresis (PAGE) in the presence of 0.1% SDS (Fig. 2). The purified mBC domain was Western blotted and detected with the anti-G-CSF receptor CRH antibody (anti-M1 serum) (Fukunaga et al., 1991). Stepwise Edman degradation of the purified mBC domain, using a gas-phase automated sequencer, indicated that the purified mBC domain has the expected amino-terminal sequence of GSSLEP. The NMR spectrum of the mBC domain is shown in Fig. 3. The high field resonance around 0 ppm shows the extremely high field shifted methyl proton resonances, which are due to the ring current effect (Wü thrich, 1986). Thus, the purified mBC domain has tertiary structure. The down field-shifted ␣-proton indicates that this protein has ␤-structure (Wü thrich, 1986), as predicted (Bazan 1990). The far UV CD spectrum of the mBC domain is shown in Fig. 6. It has positive ellipticity around 198 nm and negative ellipticity around 213 nm (Fig. 6, left panel The term Ig-like in the mG-CSF receptor indicates the immunoglobulin-like domain. Fn3 indicates the fibronectin type III region, and TM indicates the transmembrane region. The thin bars in the mBN domain represent conserved cysteine residues. The thick bar in the BC domain represents the WSXWS motif. Numbers correspond to the amino acid numbering of the mG-CSF receptor . 20°C), also suggesting the existence of ␤-structure (Chang et al., 1978). About 500 g of purified mBC domain were consistently obtained per liter of bacterial culture. The purified mBC domain retained its activity after storage at 4°C for over 2 months.
Competitive Ligand Binding Assay-To measure the ligand binding activity of the mBC domain, we performed a competitive ligand binding analysis using chemical cross-linking. As shown in Fig. 4A, when 125 I-G-CSF was incubated with the mBC domain and cross-linked, an extra band migrating at 30 kDa was revealed by SDS-PAGE. The size of the 30-kDa band corresponds to that of the receptor-ligand complex, which consists of one molecule of G-CSF (19 kDa) per molecule of mBC (12.5 kDa). As a control, a 125 I-labeled 30-kDa complex was also formed with the BN domain as described (Hiraoka et al., 1994). Both of the 125 I-labeled 30-kDa bands formed with either the BN or the BC domain were competed away by almost the same concentration of unlabeled G-CSF. These results suggest that the BC domain specifically and stoichiometrically bound to G-CSF and that its binding affinity is similar to that of the BN domain. The apparent K d , derived from several repeated competitive ligand binding analyses using chemical cross-linking and obtained with different preparations of the mBC domains, was 4 -10 ϫ 10 Ϫ8 M. The specificity of the ligand binding activity of the mBC domain was further studied. As shown in Fig.  4B, the presence of 1 M (about 1000-fold excess to the labeled G-CSF) of the human GH receptor BC domain, as well as either RNaseA or bovine serum albumin in the assay mixture as a competitor, did not inhibit the binding of 125 I-G-CSF to the purified mBC domain, whereas the same amount of unlabeled G-CSF competed with 125 I-G-CSF for binding to the purified mBC domain.
Limited Proteolysis and Identification of the Disulfides of the mBC Domain-The mBC domain encodes four cysteine residues (Cys 224 , Cys 242 , Cys 271 , and Cys 285 ). Titration of the mBC domain with 5,5Ј-dithiobis-(2-nitrobenzoic acid) in the presence of 6 M guanidine hydrochloride did not reveal any free thiols (data not shown), suggesting that all of the cysteine residues in the mBC domain formed disulfide bonds. To assign the individual disulfide bonds in the mBC domain, cystine-containing fragments of the mBC domain were isolated by HPLC after proteolytic digestion. Cystine residues were directly identified as di-PTH-cystine after release in the corresponding cycle during Edman degradation, as reported (Marti et al., 1987;Hiraoka et al., 1994). The mBC domain was digested with thermolysin in the absence of reductants, and a fragment (TL11) was separated by C18 reverse phase HPLC. Sequencing revealed that the TL11 fragment consisted of two major, equal quantities of regions, which were interpreted as being MEQEC 242 E (residues 238 -243) and LQMRC 285 (residues 281-285) (Fig. 5A), in which the C in the sequence denotes a cystine directly identified as a di-PTH-cystine in cycle 5. Digestion with pepsin was also performed after the treatment with cyanogen bromide, and two fragments (BP17 and BP22) The 1 H-NMR spectrum was measured on a 500-MHz spectrometer (Bruker AM-500) at 20°C. The purified mBC domain was dissolved in 20 mM sodium phosphate buffer, pH 5.5, at a concentration of 12.5 mg/ml (1 mM). The arrow a indicates the extremely high field-shifted methyl proton resonances. The arrow b indicates the down field-shifted ␣-proton resonances. The protein concentration of the purified mBC domain was calculated from the absorption at 280 nm (a value of 2.9). This value was calculated using 1,576 M Ϫ1 ⅐cm Ϫ1 for tyrosine and 5,225 M Ϫ1 ⅐cm Ϫ1 for tryptophan at 280 nm (Goodwin and Morton, 1946). were separated by C18 reverse phase HPLC. Sequencing of the BP17 fragment revealed two major, equal quantities of sequences, which were interpreted as being EQEC 242 ELRYQ-PQLKGAN (residues 239 -254) and RX 285 IRSSLPGF (residues 284 -293) (Fig. 5A). The X in the sequence denotes a Cys residue predicted from the known sequence but which was not evident during the Edman degradation. These data indicated that these two sets of sequences were connected by a disulfide bond between Cys 242 and Cys 285 and suggested that the Cys 224 and the Cys 271 residues should be connected by a disulfide bond (Fig. 5B). The fragment BP22 contains two equal quantities of sequences, interpreted as being LDIGPDVVSHQPG (residues 211-223) and ELX 271 GLHQAPVYT (residues 269 -280) (Fig.  5A). The di-PTH-cystine should be identified at the carboxyl terminus of the former fragment (residue 211-223) as residue 224 but was not analyzed because of the low purification yield of the fragment. However, these data are consistent with the disulfide pattern suggested above (Fig. 5B).
Cysteine Mutants-To determine the effect of each of the disulfide bonds of the mBC domain on the structure and the ligand binding activity of the receptor, we mutated each pair of cysteine residues to alanine (C224A/C271A and C242A/ C285A), thereby disrupting these disulfide bonds. Among them, only the second disulfide bond mutant (C242A/C285A) was purified to homogeneity. The far UV CD spectrum of the mutant was similar to that of the wild-type mBC domain (Fig.  6, left and middle panels, at 20°C). To compare the stabilities of the mutant and wild-type proteins, their apparent denaturations were measured by comparison of the CD spectra at various temperatures. The apparent thermal denaturation of C242A/C285A began around 35°C because the positive and negative peaks of the CD spectra started to decay around this temperature (Fig. 6, middle panel). In contrast, the apparent denaturation of the wild-type mBC domain began at 50°C (Fig.  6, left panel). This indicates that the second disulfide mutant (C242A/C285A) was more labile than the wild type. The second disulfide mutant (C242A/C285A) still retained the ligand binding activity. Its K d appears to increase more than 10-fold as compared to the wild-type mBC domain, but we could not determine it accurately because of the lability of the protein.
The 54-kDa band, which corresponds to the first disulfide bond mutant (C224A/C271A) fused to the MBP, was detected in whole cell extracts by SDS-PAGE and Western blotting. The intensities of the 54-kDa bands of the native and mutant proteins were similar, suggesting similar expression levels. However, a reduced level of the mutant 54-kDa protein was detected in the periplasm. The mutant 54-kDa protein eluted as a broad peak from the S-Sepharose column, whereas the wildtype mBC domain was obtained as a sharp peak. The mutant C224A/C271A mBC product was precipitated immediately after the restriction protease factor Xa digestion. It appears that both disulfide bonds are critical for the maintenance of the stably folded protein.
WSXWS Motif Mutants-To determine the function of WSXWS motif, we constructed four WSXWS motif mutants (W294A, S295A, W297A, S298A). The expression levels of the native and the four mutant proteins were similar. Among them, the W294A and the S298A mutants were purified to homogeneity. The mutant W294A retained ligand binding activity (Fig. 4A). It appeared that its apparent K d , obtained by several experiments, is somewhat higher than that of the wildtype mBC domain, but the difference is not remarkable. The far UV CD spectrum of the mutant W294A is similar to that of the wild type (Fig. 6). The beginning of the apparent thermal denaturation of the mutant W294A, determined by measuring the CD spectra at various temperatures, appears around 50°C (Fig. 6, right panel), indicating that the stability of the mutant W294A is similar to that of the wild-type mBC domain. The mutant S298A also retained ligand binding activity. However, we could not compare the ligand binding activity and the CD spectrum of the mutant S298A to those of the wild-type mBC domain because of its low purification yield and lability, such as precipitation during concentration. It appeared that the expression levels of the fused proteins (54 kDa) corresponding FIG. 4. Ligand binding activity of the mBC domain. Panel A shows an autoradiograph of the competitive ligand binding of 125 I-G-CSF and unlabeled G-CSF to the mBN and mBC domains using chemical cross-linking, as described (Hiraoka et al., 1994). Each reaction was performed with 125 I-G-CSF (100,000 cpm, Amersham International), unlabeled G-CSF (Kirin Brewery), and either the mBN or the mBC domain under the conditions indicated above. Complexes of the mBN or the BC with 125 I-G-CSF were treated with 1 mM dithiobis(succinimidyl propionate) (Pierce) for cross-linking and were resolved by 0.1% SDS, 15% PAGE. The gel was exposed to an imaging plate (Fuji imaging plate, Type BA 20X40; Fuji Photo Film). The autoradiograph (upper part) of the gel from the image analyzer data is shown (Bio-image analyzer BA100; Fuji Photo Film), and the intensities of the bands corresponding to the 125 I-G-CSF bound to either the mBN (closed circles) or the mBC (closed squares) domains were estimated (lower part) as described (Hiraoka et al., 1994).  (Asakura et al., 1994), RNaseA (Sigma), ovalbumin, or bovine serum albumin (indicated as BSA, Pharmacia). The mBC⅐ 125 I-G-CSF complex was treated with the chemical cross-linker and was resolved by 0.1% SDS, 15% PAGE. The retained radioactivity of the complex was counted as described (Hiraoka et al., 1994). to two other mutants (S295A, W297A) were similar to the wild type, but the amounts of these bands were reduced in the periplasmic fraction, as determined by SDS-PAGE and Western blotting. These two fused mutants were obtained as broad peaks by S-Sepharose chromatography, whereas the wild type was obtained as a sharp peak, and the final products were precipitated by the factor Xa digestion. Thus, it appears that the disruption of the WSXWS motif did not affect the ligand binding activity remarkably but instead affected the stable folding of the mBC domain, although the W294A substitution still allowed stable folding of the mBC domain.

DISCUSSION
The BC domain specifically and stoichiometrically bound to G-CSF. The tertiary folding of the BC domain was confirmed by NMR spectroscopy. These data indicated that this domain (approximately 100 residues) folds discretely. NMR and far UV CD spectra indicated that the BC domain has ␤-structure. Analysis of the disulfide bonds of the mBC domain of the G-CSF receptor FIG. 5. Disulfide mapping of the mBC domain. Panel A shows the amino acid sequence of the mBC domain from residues 203-308, in addition to the GSS residues at the amino terminus. The boxes indicate the peptides generated by thermolysin digestion (TL11) and pepsin digestion after treatment with cyanogen bromide (BP17, BP22). Panel B shows the identification of the disulfide bonds in the mBC domain. Sequences in capital letters indicate the results of the Edman degradation of the non-reduced peptide. C denotes the di-PTH of cystine. (C) denotes a Cys residue that should be identified as the di-PTH of cystine, which was predicted from the known sequence but not observed. X denotes a Cys residue predicted from the known sequence but not observed. Numbers indicate the amino acid number of the mG-CSF receptor .
FIG. 6. Far UV CD spectra of the wild-type and mutant BC domains. Far UV CD spectra of the wild-type, mutant C242A/C285A, and mutant W294A BC domains measured on a Jasco J-720 spectropolarimeter (Japan Spectroscopic) at various temperatures with solutions containing 0.12 mg/ml (10 M) of the protein in 10 mM sodium phosphate buffer, pH 6.0, are shown. The optical path length was 1 mm for the far ultraviolet CD spectrum (190 -250 nm). The concentration of the purified mBC domain was calculated as described in Fig. 3. showed that the 4 cysteine residues of this domain formed two disulfide bonds. The structure of the CRH region of the cytokine receptor was predicted to be ␤-strands held together by disulfide bonds (Bazan, 1990). These two disulfide bonds (Cys 224 -Cys 271 , Cys 242 -Cys 285 ) were mapped between the B and E ␤-strands and between the C and F ␤-strands, respectively, on the predicted topology (Fig. 7). These BC domain disulfide bonds are not conserved in the cytokine receptor family (Bazan, 1990). However, two disulfide bonds are essential for proper folding of the BC domain of the G-CSF receptor. Miyazaki et al. (1991) determined that the two Trp residues in the WSXWS motif are critical for proper ligand binding to the interleukin-2 receptor ␤-chain, with the use of a mutant cDNA transfection system in lymphoma cells. Rozakis-Adcock and Kelly (1992) reported that substitution of the WSXWS motif precluded high affinity ligand binding by the prolactin receptor. However, our data suggest that the WSXWS motif may play an important role in the proper folding of the BC domain rather than in the formation of the ligand binding site. It is likely that the disruption of the WSXWS motif leads to the formation of an abnormally folded receptor and to a reduction in the ligand binding activity. Recent x-ray crystallographic analyses of the GH and prolactin receptors showed that the WSXWS motif (which is not completely conserved in the GH receptor) is an irregular, extended chain immediately preceding the following short ␤-strand and is quite removed from the interfaces that contact ligand (de Vos et al., 1992;Somers et al., 1994). Interestingly, it was reported that the WSXWS box of the prolactin receptor is part of a much larger, highly organized pattern, and the side chains of the two Ser residues form hydrogen bonds to the main-chain atoms of a neighboring ␤-strand (Somers et al., 1994). In the present work, the removal of the side chains of the Ser residues in the two mutants S295A and S298A caused lability of the mBC domain. Thus, it can be speculated that these Ser side chains may form hydrogen bonds that anchor the strands preceding the ␤-structure and stabilize the BC domain. Yoshimura et al. (1992) suggested that the WSXWS motif of the erythropoietin receptor is critical not only for ligand binding but also for the ability of the receptor to exit from the endoplasmic reticulum. Abundant work on secretory proteins suggests that there is selective retention and/or degradation of unfolded, misfolded, or aggregated proteins in the endoplasmic reticulum (Pelham, 1989). These results are also consistent with our results. The reduction of the yield of our precipitable mutants (S295A and W297A) of the BC domain from the periplasmic fraction is compatible with these results.
The apparent K d of 4 -10 ϫ 10 Ϫ8 M indicates that the BC domain plays a critical role in the recognition and the binding of the ligand. This K d value is similar to that of 3-8 ϫ 10 Ϫ8 M exhibited by the purified mBN domain of the G-CSF receptor (Hiraoka et al., 1994). Both of the K d values of the BN and the BC domains are relatively high as compared to the K d of 3-4 ϫ 10 Ϫ9 M exhibited by the deletion mutant of the G-CSF receptor expressed on the surface of murine myeloid cells, which contains both the BN and the BC domains (Fukunaga et al., 1991). Our previous study indicated that the BN domain also folds discretely (Hiraoka et al., 1994). These results indicate that the CRH region of the G-CSF receptor is composed of two small BN and BC domains, as predicted (Bazan 1990), and both domains are required for high affinity ligand binding, such as ϳ10 Ϫ9 M. Probably, G-CSF recognition occurs via a generic binding trough within the CRH region formed by the BN and BC domains. Mutational and x-ray crystallographic analyses of the human GH receptor indicated that two amino acid residues (Trp 104 and Pro 106 ) in its BN domain are critical for human GH binding (Bass et al., 1991;de Vos et al., 1992). Such important amino acid residues for ligand binding were not detected in the BC domain of the human GH receptor, although a substantial contact surface with the ligand exists in the BC domain. However, most of the residues important for ligand binding in the GH receptor and the interleukin-6 receptor could be located in or near the hinge region of the two terminal domains (Bass et al., 1991;Yawata et al., 1993).
In earlier work, Fukunaga et al. (1991) reported the absence of detectable affinity of the deletion mutants (Y⌬(5-195)) of the mG-CSF receptor, which contains the BC domain but lacks the immunoglobulin-like and the BN domains, at the cell surface. Probably, the K d value of the mutant Y⌬(5-195) was below the level of detection among the series of mutants analyzed with their system. Recent analyses suggest that receptor oligomerization is the major consequence of ligand binding, and a 2:1 receptor-ligand complex was detected with the GH receptor (Cunningham et al., 1991;Fuh et al., 1992;de Vos et al., 1992). However, we did not detect a 2:1 mBC domain⅐G-CSF complex by SDS-PAGE using chemical cross-linking. This is reasonable, because our recent study indicated that both the BC and the immunoglobulin-like domains, in addition to the BN domain, are required for the oligomerization of the G-CSF receptor (Hiraoka et al., 1995). Structural analysis of the BC domain would aid further investigations of the mechanism and function of this receptor. The size of the purified BC domain (12.5 kDa) is sufficiently small for NMR analysis.