The Binding between the Stem Regions of Human Growth Hormone (GH) Receptor Compensates for the Weaker Site 1 Binding of 20-kDa Human GH (hGH) than That of 22-kDa hGH*

Despite the lower site 1 affinity of the 20-kDa human growh hormone (20K-hGH) for the hGH receptor (hGHR), 20K-hGH has the same hGHR-mediated activity as 22-kDa human GH (22K-hGH) at low hGH concentration and even higher activity at high hGH concentration. This study was performed to elucidate the reason why 20K-hGH can activate hGHR to the same level as 22K-hGH. To answer the question, we hypothesized that the binding between the stem regions of hGHR could compensate for the weaker site 1 binding of 20K-hGH than that of 22K-hGH in the sequential binding with hGHR. To demonstrate it, we prepared 15 types of alanine-substituted hGHR gene at the stem region and stably transfected them into Ba/F3 cells. Using these cells, we measured and compared the cell proliferation activities between 20K- and 22K-hGH. As a result, the activity of 20K-hGH was markedly reduced than that of 22K-hGH in three types of mutant hGHR (T147A, H150A, and Y200A). Regarding these mutants, the dissociation constant of hGH at the first and second step (KD1 and KD2) in the sequential binding with two hGHRs was predicted based on the mathematical cell proliferation model and computational simulation. Consequently, it was revealed that the reduction of the activity in 20K-hGH was attributed to the change of not KD1 but KD2. In conclusion, these findings support our hypothesis, which can account for the same potencies for activating hGHR between 20K- and 22K-hGH, although the site 1 affinity of 20K-hGH is lower than that of 22K-hGH.

Previous analyses showed that 20K-hGH formed an active 1:2 (hGH:hGHR) complex in a sequential manner as well as 22K-hGH, that is, the first hGHR binds to site 1 on hGH (this is designated "STEP1") and next the second one binds to site 2 (STEP2) (9,(11)(12)(13)(14). An active 1:2 (hGH:hGHR) complex formation is followed by tyrosine phosphorylation of hGHR and activation of intracellular signal transducer molecules (15)(16)(17). Fig. 1 shows the schematic illustration of the mode of receptor dimerization of both hGH isoforms. Here we tentatively designate the stem region of hGHR, which is considered to be involved in receptor dimerization, "siteBP." Concerning the binding affinity between hGH and hGHR, we reported that KD1(22K) was smaller than KD1(20K) in the biosensor analysis (9) and that KDsite2(20K) was considered to be almost the same as KDsite2(22K), because the site 2 region of 20K-hGH was conformationally similar to 22K-hGH (11). On the other hand, the activity of 20K-hGH via hGHR is the same as that of 22K-hGH at low hGH concentration and even higher at high hGH concentration in cell proliferation assay (11). Therefore, we hypothesized that binding of 20K-hGH to the first hGHR could result in the transient 1:1 complex, which has a more favorable conformation for forming an active 1:2 complex especially at the siteBP region of hGHR.
In this study, to elucidate the difference of siteBP contribution to the 1:2 complex formation with hGHR between 20K-and 22K-hGH, we made an hGHR expression plasmid with conversion of several amino acids at the siteBP region to alanine. Using mouse pro B cell line (Ba/F3) stably expressing the mutant or wild type hGHR, cell proliferation assay of both hGH isoforms was performed. As a result, in three mutants (T147A, H150A, and Y200A), 20K-hGH had markedly reduced activity than 22K-hGH. In addition, we estimated KD1 and KD2 of both hGH isoforms by theoretically simulating the cell proliferation curve, and it was highly speculated that alanine substitution at siteBP could mainly affect the affinity of 20K-hGH at STEP2 (KD2(20K)). By lacking 15 amino acids, 20K-hGH partly lost the site 1 affinity but the binding between siteBPs possibly compensates for the loss.

EXPERIMENTAL PROCEDURES
Materials-Recombinant 20K-hGH with an authentic amino acid sequence was prepared as described previously (6). As for the 22K-hGH sample, a commercially supplied recombinant one with an authentic amino acid sequence (Genotropin, Amersham Pharmacia Biotech, and Upjohn AB, Sweden) was used. IL-3-dependent mouse pro B cell line (Ba/F3) was from the RIKEN Cell Bank (Ibaraki, Japan).
* 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.
Construction of Mutant hGHR Complementary DNA (cDNA) Expression Vector-We ligated the full-length wild type hGHR cDNA (10) to cloning vector pUC18 (pUC18-hGHR) and substituted alanine for each of 15 amino acids by polymerase chain reaction method using oligonucleotides encoding the desired mutation. After confirming that the only desired alanine mutation was incorporated into the pUC18-hGHR by dideoxy sequencing on a DNA sequencer (ABI 373, Perkin-Elmer), pUC18-hGHR with an alanine mutation was digested with restriction enzyme. Each full-length wild type and mutant hGHR cDNA was separated from the pUC18 on an agarose gel and blunt-ended with T4 DNA polymerase. Next, they were subcloned into expression plasmid pCXN2Zb at the EcoRV cloning site. Finally, a sequence covering both ends of hGHR cDNA ligated to the pCXN2Zb vector was confirmed by dideoxy sequencing.
Cell Proliferation Assay-Cells were cultured in the selection medium B to logarithmic phase (ϳ1 ϫ 10 6 cells/ml). Cells were incubated in the assay medium (RPMI 1640 supplemented with 5% fetal calf serum, 4 mM L-glutamine, 50 M 2-mercaptoethanol, and antibiotics) for 4 h and were suspended in the assay medium at densities of 8 ϫ 10 5 cells/ml. The hGH sample solution (50 l) and cell suspension (50 l) were mixed together into the well of a 96-well plate and incubated for 20 h. The measurement of cell proliferation was achieved using Tetra-Color ONE (Seikagaku Corporation, Tokyo, Japan) according to the manufacturer's protocol, and absorbance at 450 nm (reference wavelength: 595 nm) was measured.
Sequential Binding Model-Human GH sequentially binds to hGHR resulting in the formation of a dimer consisting of one molecule of hormone and two molecules of receptor (12)(13)(14). Uchida et al. (9) studied the sequential binding of hGH with hGHR on the cell surface of Ba/F3 expressing hGHR and calculated the 1:2 complex concentration ([hGH: (hGHBP)2]) as functions of STEP1 and STEP2 affinity (KD1 and KD2), and [hGH:(hGHBP)2] can be represented as follows.
Because we are basically using the same Ba/F3 cells, expression plasmid, and cell proliferation assay method as Uchida et al. (9), we adopted the same definition, assumptions, and prerequisites of the sequential binding model to estimate the hGH 1:2 complex concentration ([hGH:(hGHBP)2]) on the cell surface of Ba/F3 expressing hGHR.
Cell Proliferation Model-During the logarithmic phase of cell proliferation, cell population is represented as follows.
Here, t is the period of time for cell proliferation (h), N0 is the initial cell population size, N is the size of the cell population after a period of growth, and is the specific growth rate (h Ϫ1 ).
In the cell proliferation model, we made the assumptions as follows. 1) The strength of intracellular growth signal only depends on the cell surface 1:2 complex concentration ([hGH:(hGHBP)2]) and is in direct proportion to it. 2) The strength of intracellular growth signal derived from one 1:2 complex is identical among the wild type and each mutant hGHR.
Based on above assumptions, is represented as follows.
is a proportionality constant. Finally, the fold induction (N/N0) in the cell proliferation assay is represented as follows.
Here, ␣ and t are constants that are determined by the experimental condition.  (13). Based on the tertiary crystal structure of the 1:2 complex between 22K-hGH and hGHR-ECD registered in Protein Data Bank (PDB ID: 3HHR), in addition to above the eight amino acids we selected seven amino acids (Leu-142, Val-144, Ile-149, Gln-154, , which were presumed to be located at the contact surface area between siteBPs. Because alanine substitution is minimally perturbing for the secondary and tertiary structure, it is generally regarded as the best means of selective removal of interactive residues involved in salt bridges and hydrogen bonds (19,20). We constructed 15 hGHR expression plasmids with a single alanine substitution at each of the 15 amino acids.

Selection of Amino
Cell Proliferation Assay-As already reported (11), Ba/F3 cells expressing hGHR proliferate in response to hGH in a dose-dependent manner and are a useful and convenient tool for measuring hGHR-mediated activity. The expression plasmid containing the wild type or each alanine-substituted hGHR cDNA was stably transfected into the IL-3-dependent mouse pro B cell line (Ba/F3) by the electroporation method, respectively. The first selection was performed with antibiotics G418 and mouse IL-3, and cells that retain the neomycinresistant gene deriving from hGHR expression plasmid were selected. Next, the second selection was performed with antibiotics G418 and hGH instead of mouse IL-3, and cells dependent on hGH stimulation were selected. Cells expressing the mutant hGHR (D152A) proliferated in the first selection step but not in the second one. Each cell expressing the wild type or mutant hGHR except for mutation D152A proliferated in response to 20K-and 22K-hGH in a dose-dependent manner; however, their dose-response curves were classified into three patterns as shown in Table I. Fig. 2 shows the representative cell proliferation curve of each pattern.
In pattern I, 20K-and 22K-hGH had equal potency at lower hGH concentration (Ͻ10 -30 nM), and the difference between both hGH isoforms was detected only at high hGH concentration. The cell proliferation curve of the wild type hGHR is shown in Fig. 2 as an example of pattern I, into which mutations L142A, V144A, S145A, L146A, Q154A, I192A, V197A, P198A, and S201A are classified. In pattern II, to which only mutation N143A belonged, cells proliferated even without hGH stimulation, and hGH had relatively lower fold induction than the other mutations probably because of the higher basal proliferation activity at 0 nM hGH. In pattern III, 20K-hGH had diminished activity compared with 22K-hGH. The cell proliferation curve of mutant hGHR (Y200A) represents this pattern III to which mutation T147A, I149A, and H150A belonged. In the assay of mutant T147A, H150A, and Y200A, the maximum fold induction level of 20K-GH was decreased by 8.4, 22, and 33% relative to that of 22K-hGH, respectively, even at higher 20K-hGH concentration than 22K-hGH. Regarding the mutation I149A, the maximum fold induction level was the same between both hGHs, but EC 50 of 20K-hGH was about three times larger than that of 22K-hGH.
Computer-aided Simulation of Cell Proliferation Curve-Human GH sequentially binds to two hGHRs, resulting in the formation of a 1:2 (hGH:hGHR) complex (12)(13)(14). This sequential binding mechanism was studied by a mathematical model (21), and 1:2 complex concentration ([hGH:(hGHBP)2]) on the cell surface of Ba/F3 expressing hGHR was calculated as functions of KD1 and KD2 (9). In this report, we have developed a mathematical cell proliferation model via hGHR based on the sequential binding model of hGHR.
The binding affinity of 22K-hGH at STEP1 (KD1(22K)) for the wild type hGHR has been studied using the hGHR-ECD

FIG. 2. Representative cell proliferation curve induced by 20K-or 22K-hGH stimulation.
The fold induction was calculated as absorbance in the presence of hGH divided by absorbance in the absence of hGH. Pattern I, cells were GH-dependent, and the activity of 20K-hGH was the same as that of 22K-hGH at low hGH concentration and higher at high hGH concentration. Pattern II, cells proliferated even without hGH stimulation. Pattern III, cells were GH-dependent and proliferated by 20K-GH to the less extent than by 22K-hGH. Each data point represents the mean Ϯ S.D. from triplicate wells.  I Summary of proliferation patterns of Ba/F3 cells expressing the wild type or each mutant hGHR All the proliferation curves were classified into three patterns. Cells in pattern I were GH-dependent, and the activity of 20K-hGH was the same as that of 22K-hGH at low hGH concentration and higher at high hGH concentration. Cells expressing N143A proliferated without hGH and were classified in pattern II. Cells in pattern III were GH-dependent and proliferated by 20K-GH to a lesser extent than by 22K-hGH. Cells expressing D152A did not proliferate in the presence of hGH and were depicted as N.D.

Mutation
Cell proliferation pattern Wild type I  L142A  I  N143A  II  V144A  I  S145A  I  L146A  I  T147A  III  I149A  III  H150A  III  D152A N.D. Q154A  I  I192A  I  V197A  I  P198A  I  Y200A  III  S201A  I by some groups, and it is considered to be 1-3 nM (9,22). Moreover, Pearce et al. (23) mentioned in their discussion that the KD2(22K) value for the wild type hGHR was estimated to be ϳ0.5-5 nM. Therefore, we first adopted KD1(22K) ϭ 2 (nM) and KD2(22K) ϭ 2 (nM) as initial values to determine the initial value of ␣ in the Equation 4. The calculated fold induction (N/N0) was fitted to the actual experimental values of cell proliferation assay with 22K-hGH and the wild type hGHR when the period of time for cell proliferation (t) was 20 h. Consequently, the initial value of ␣ was calculated as 2.30 ϫ 10 5 h Ϫ1 ⅐M Ϫ1 . Secondly, KD1(22K), KD2(22K), and ␣ were optimized using a systematic search on a grid, by minimizing the root mean square deviations between observed data and model calculations. The grid size and range used for KD1(22K) and KD2(22K) were 0.1 and 0 -150 nM, respectively, and those for ␣ were 1.0 ϫ 10 3 h Ϫ1 ⅐M Ϫ1 and 2.20 -3.00 ϫ 10 5 h Ϫ1 ⅐M Ϫ1 , respectively. As a result, as shown in Fig. 3A, when 5.1 nM, 2.9 nM, and 2.41 ϫ 10 5 h Ϫ1 ⅐M Ϫ1 were adopted as KD1(22K), KD2(22K), and ␣, respectively, the simulated curve gave the best fit to the experimentally obtained cell proliferation curve with 22K-hGH and the wild type hGHR. Finally, KD1 and KD2 of 20K-hGH via the wild type hGHR and those of both hGH isoforms via mutant hGHR (T147A, I149A, H150A, and Y200A) in pattern III were also optimized using a systematic search on a grid. The same grid size and range were used for them. ␣ was kept constant at 2.41 ϫ 10 5 h Ϫ1 ⅐M Ϫ1 as in the case of 22K-hGH and the wild type hGHR during the systematic search, because the experimental condition for the cell proliferation assay was identical. The adopted KD1 and KD2 value of 20K-and 22K-hGH are summarized in Table II. Relative values (R1(20K), R2(20K), R1(22K), and R2(22K)) of four mutants to KD1(20K-wild type), KD2(20K-wild type), KD1(22K-wild type), and KD2(22K-wild type) were calculated. R2(20K) was 10, 122, and 68 in mutants T147A, H150A and Y200A, respectively, and was changed more significantly than R1(20K), R1(22K), and R2(22K) by introducing alanine mutation to the siteBP region.

DISCUSSION
In this study, we have revealed that alanine substitution at Thr-147, Ile-149, His-150, and Tyr-200 in the siteBP region of hGHR reduced the cell proliferation activity of 20K-hGH as compared with that of 22K-hGH. The involvement of the siteBP region in the complex formation of 22K-hGH and hGHR has been studied by several groups. DeVos et al. (13) first reported that there was a substantial contact surface between the C-terminal domains of hGHR-ECD, namely siteBP regions, when 22K-hGH formed a 1:2 complex with hGHR-ECD. Crystallographic study showed the eight residues (Asn-143, Ser-145, Leu-146, Thr-147, His-150, Asp-152, Tyr-200, and Ser-201) were involved in this domain. Furthermore, Clackson et al. (24) showed that hGHR dimerization stabilized loop structure from Val-144 to Gly-148 at siteBP region. Chen et al. (25) converted several amino acids at siteBP of rabbit GHR to alanine, aspartate, lysine, or cysteine and presented that residues Ser-145, His-150, Asp-152, Tyr-200, and Ser-201 were required for effective signal transduction through the dimerization domain. These previous studies are well consistent with our finding except for Ile-149, because the involvement of Ile-149 in siteBP interaction has been demonstrated for the first time in this study.
In patients with Laron syndrome (familial GH resistance characterized by severe dwarfism and metabolic dysfunction), a point mutation in the siteBP region was identified resulting in the substitution of a highly conserved aspartate residue by histidine at position 152 (D152H) of hGHR (26). Duquesnoy et al. (26) reported that the hGHR with mutation D152H displayed subnormal GH binding activity but hGHR-ECD with D152H substitution was unable to dimerize. In this report, not only 22K-hGH but also 20K-GH had no cell proliferation activity in mutation D152A nor D152H (data not shown). These results mean that Asp-152 in hGHR plays a quite important role in the binding between siteBP regions.
Especially, the mutation T147A, H150A, and Y200A of hGHR resulted in a drastic decrease of cell proliferation activity of 20K-hGH compared with that of 22K-hGH. Similar results were observed also in the rat serine protease inhibitor 2.1 gene promoter activation assay in CHO-K1 cells transiently expressing each three mutants (data not shown). These data indicate that Thr-147, His-150, and Tyr-200 considerably contribute to the same hGHR-mediated activity of 20K-hGH as 22K-hGH regardless of its reduced site 1 affinity. To clarify the mechanism of how Thr-147, His-150, and Tyr-200 enable 20K-hGH to form an active 1:2 complex to the same degree as 22K-hGH, we are now under investigation of the x-ray crystal structure of the complex of 20K-hGH and hGHR-ECD.
To elucidate the binding affinity at STEP1 and STEP2 resulting from alanine substitution, we estimated the dissociation constant KD1 and KD2 of 20K-and 22K-hGH by modeling the cell proliferation curve mathematically. By fitting the calculative proliferation curve to the experimental data of the wild type hGHR, KD1(20K) and KD1(22K) were predicted to be 13 and 5.1 nM, respectively, and the binding affinity of 20K-hGH at site 1 was weaker than that of 22K-hGH. These estimates are reasonable compared with the results experimentally obtained by biosensor analysis using hGHR-ECD, where KD1(20K) (16 nM) is lager than KD1(22K) (2 nM) (9). Concerning the mutation T147A, H150A, and Y200A, KD2(20K) increased 10-, 122-, and 68-fold, respectively, compared with KD2(22K) (1.8-, 17-, and 5.3-fold). Generally alanine substitution is minimally perturbing for the secondary and tertiary structure, and three alanine mutations are located only at the siteBP region of hGHR. Therefore, mutations T147A, H150A, and Y200A are considered to have no direct influence on the contact surface structure between the site 2 region on hGH and the second hGHR in sequential binding, that is, the binding affinity KDsite2(20K) in Fig. 1. This suggests that the drastic change of KD2(20K) results from the change of KDsiteBP(20K).
In our calculation, the change of hGH bioactivity was in good concordance with the change of the binding affinity (KD2(20K)). However, Rowlinson et al. (27) reported that GH with mutation at site 1 markedly lost its bioactivity without loss of site 1 binding affinity. Recently, Pearce et al. (23) reported that the EC 50 of cell proliferation via wild type hGHR was unaffected until site 1 affinity of 22K-hGH was reduced about 30-fold from that of wild type 22K-hGH. Under the condition used in our sequential binding and cell proliferation model, it is predicted that an ϳ30-fold reduction in site 1 affinity should yield about 30-fold higher EC 50 for cell proliferation, which means the experimental data of Pearce et al. (23) cannot be simulated by our model. Unfortunately, at present we cannot explain such disagreement, and some modifications might be necessary for more accurate simulation by adding some new parameters to the sequential binding model or the cell proliferation model or by altering some prerequisites.
In conclusion, we have shown that some alanine substitutions at the siteBP region of hGHR caused the markedly decreased activity of 20K-hGH compared with 22K-hGH. This means the siteBP region is involved in 1:2 complex formation in a different manner between 20K-and 22K-hGH, and the binding between the siteBP regions compensates for the weaker site 1 binding of 20K-hGH than that of 22K-hGH.