Cystine Knot of the Gonadotropin α Subunit Is Critical for Intracellular Behavior but Not for in Vitro Biological Activity
1997; Elsevier BV; Volume: 272; Issue: 29 Linguagem: Inglês
10.1074/jbc.272.29.18098
ISSN1083-351X
AutoresAsomi Sato, Emerald Perlas, David Ben-Menahem, Masataka Kudo, Mary R. Pixley, Madoka Furuhashi, Aaron J.W. Hsueh, Irving Boime,
Tópico(s)Reproductive Biology and Fertility
ResumoThe common α subunit of glycoprotein hormones contains five disulfide bonds. Based on the published crystal structure, the assignments are 7–31, 59–87, 10–60, 28–82, and 32–84; the last three comprise the cystine knot, a structure also seen in a variety of growth factors. Previously, we demonstrated that the efficiency of secretion and the ability to form heterodimers by α subunits bearing single cysteine residue mutants in the cystine knot were significantly reduced. These results suggested that the cystine knot is critical for the intracellular integrity of the subunit. To assess if the presence of the free thiol affected the secretion kinetics, we constructed paired cysteine mutants of each disulfide bond of the α subunit. The secretion rate for these monomers was comparable with wild type except for the α-10–60 mutant, which was 40% lower. The recovery of the α7–31 and α59–87 mutants was greater than 95%, whereas for the cystine knot mutants, it was 20–40%. Co-expression of the wild-type chorionic gonadotropin β subunit with double cysteine mutants did not enhance the recovery of α mutants in the media. Moreover, compared with wild-type, the efficiency of heterodimer formation of the α10–60 or α32–84 mutants was less than 5%. Because subunit assembly is required for biological activity, studies on the role of these disulfide bonds in signal transduction were not possible. To bypass the assembly step, we exploited the single chain model, where the α and β subunits are genetically fused. The recovery of secreted tethered gonadotropins bearing mutations in the cystine knot was increased significantly. Although dimer-specific monoclonal antibodies discriminated the conformation of single chain α10–60 and α32–84 mutants from the native heterodimer, these mutants were nevertheless biologically active. Thus, individual bonds of cystine knot are important for secretion and heterodimer formation but not for in vitro bioactivity. Moreover, the data suggest that the native heterodimer configuration is not a prerequisite for receptor binding or signal transduction. The common α subunit of glycoprotein hormones contains five disulfide bonds. Based on the published crystal structure, the assignments are 7–31, 59–87, 10–60, 28–82, and 32–84; the last three comprise the cystine knot, a structure also seen in a variety of growth factors. Previously, we demonstrated that the efficiency of secretion and the ability to form heterodimers by α subunits bearing single cysteine residue mutants in the cystine knot were significantly reduced. These results suggested that the cystine knot is critical for the intracellular integrity of the subunit. To assess if the presence of the free thiol affected the secretion kinetics, we constructed paired cysteine mutants of each disulfide bond of the α subunit. The secretion rate for these monomers was comparable with wild type except for the α-10–60 mutant, which was 40% lower. The recovery of the α7–31 and α59–87 mutants was greater than 95%, whereas for the cystine knot mutants, it was 20–40%. Co-expression of the wild-type chorionic gonadotropin β subunit with double cysteine mutants did not enhance the recovery of α mutants in the media. Moreover, compared with wild-type, the efficiency of heterodimer formation of the α10–60 or α32–84 mutants was less than 5%. Because subunit assembly is required for biological activity, studies on the role of these disulfide bonds in signal transduction were not possible. To bypass the assembly step, we exploited the single chain model, where the α and β subunits are genetically fused. The recovery of secreted tethered gonadotropins bearing mutations in the cystine knot was increased significantly. Although dimer-specific monoclonal antibodies discriminated the conformation of single chain α10–60 and α32–84 mutants from the native heterodimer, these mutants were nevertheless biologically active. Thus, individual bonds of cystine knot are important for secretion and heterodimer formation but not for in vitro bioactivity. Moreover, the data suggest that the native heterodimer configuration is not a prerequisite for receptor binding or signal transduction. Human chorionic gonadotropin (CG), 1The abbreviations used are: CG, chorionic gonadotropin; hCG, human CG; LH, lutropin; WT, wild type. lutropin (LH), follitropin, and thyrotropin are members of the glycoprotein hormone family that share a common α subunit but differ in their hormone-specific β subunits. The amino acid sequence of the α subunit is identical within a species (1Pierce J.G. Parsons T.F. Annu. Rev. Biochem. 1981; 50: 465-495Crossref PubMed Scopus (1940) Google Scholar), suggesting that the tertiary structure of the α subunit requires a degree of flexibility to adapt to a unique conformation relative to the β domain(s). The human α subunit has 10 highly conserved cysteine residues, which form five disulfide bonds. Based on the crystal structure of the α subunit in hCG, the proposed pairs are 10–60, 28–82, 32–84, 7–31, and 59–87 (Fig. 1) (2Lapthorn A.J. Harris D.C. Littlejohn A. Lustbader J.W. Canfield R.E. Machin K.J. Morgan F.J. Isaacs N.W. Nature. 1994; 369: 455-461Crossref PubMed Scopus (853) Google Scholar, 3Wu H. Lustbader J.W. Liu Y. Canfield R.E. Hendrickson W.A. Structure. 1994; 2: 545-558Abstract Full Text Full Text PDF PubMed Scopus (462) Google Scholar). Bonds 28–32 and 32–84 comprise a ring structure penetrated by a bond bridging cysteine residues 10 and 60 resulting in a core that forms three hairpin loops. This structure is a common feature to the large growth factor family including tumor growth factor-β, activins, nerve growth factor, and platelet-derived growth factor (4McDonald N.Q. Hendrickson W.A. Cell. 1993; 73: 421-424Abstract Full Text PDF PubMed Scopus (482) Google Scholar, 5Sun P.D. Davies D.R. Annu. Rev. Biophys. Biomol. Struct. 1995; 24: 269-291Crossref PubMed Google Scholar, 6Brunner A.M. Lioubin M.N. Marquerdt H. Malacko A. Wang W-C. Shapiro R. Neubauer M. Cook M. Madisen L. Purchio A.F. Mol. Endocrinol. 1992; 6: 1691-1700Crossref PubMed Scopus (50) Google Scholar, 7Mason A. Mol. Endocrinol. 1994; 8: 325-332PubMed Google Scholar, 8Kenney W.C. Haniu M. Herman A. Arakawa T. Costigan V. Lary J. Yphantis D. Thomason A.R. J. Biol. Chem. 1994; 269: 12351-12359Abstract Full Text PDF PubMed Google Scholar). Chemical reduction (1Pierce J.G. Parsons T.F. Annu. Rev. Biochem. 1981; 50: 465-495Crossref PubMed Scopus (1940) Google Scholar) or site-specific mutations of individual cysteine residues in either the common α or CGβ subunit, alters dimer assembly and secretion rate (9Suganuma N. Matzuk M. Boime I. J. Biol. Chem. 1989; 264: 19302-19307Abstract Full Text PDF PubMed Google Scholar, 10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). Mutants lacking either the 7–31 or 59–87 bond were secreted (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar), and heterodimers containing the α7–31 mutation bound to the LH/hCG receptor with an affinity comparable with the wild-type hCG, whereas the 59–87 mutant interacted with a lower binding affinity. However, mutants comprising the cystine knot (10–60, 28–82, and 32–84) were not secreted in sufficient quantities to allow examination of the biological activity. To address this issue, we used a single chain gonadotropin model where the CGβ subunit was genetically fused to the α subunit (11Sugahara T. Pixley M.R. Minami S. Perlas E. Ben-Menahem D. Hsueh A.J.W. Boime I. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 2041-2045Crossref PubMed Scopus (112) Google Scholar, 12Narayan P. Wu C. Puett D. Mol. Endocrinol. 1995; 9: 1720-1726Crossref PubMed Google Scholar) to promote more efficient secretion of proteins bearing the α and β subunit domains in the same complex. Using this approach, the rate-limiting step of subunit assembly could be circumvented, and mutations that otherwise block dimer formation could be evaluated. Here we show that single chains containing mutations in the cystine knot (α10–60, 28–82, and 32–84) are secreted and biologically active, implying that quaternary relationships between the subunits in the heterodimer primarily influence the intracellular behavior rather than receptor binding/signal transduction. Enzymes for preparing vectors and for polymerase chain reaction were purchased from Promega (Madison, WI), Boehringer Mannheim, or Stratagene (La Jolla, CA). Oligonucleotides were prepared by the Washington University Sequencing Facility. Cell culture media and reagents were prepared by the Washington University Center for Basic Cancer Research (St. Louis, MO). Fetal bovine serum, dialyzed fetal bovine serum, and the neomycin analogue, G418, were purchased from Life Technologies, Inc. For metabolic labeling, [35S]cysteine-methionine (Promix) (Amersham Corp.) or [35S]cysteine (ICN, Irvine, CA) were used. Monoclonal antibodies A407 and B109 were gifts from Dr. Robert Canfield and Dr. Steve Birken (Columbia University, New York). Previously, the intracellular behavior of single cysteine mutants was assessed (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar,13Bielinska M. Boime I. Mol. Endocrinol. 1992; 6: 261-271PubMed Google Scholar). Double mutants were only constructed that deleted the 7–31 and 59–87 disulfide bridges. Here, we designed all of the analogs to contain double substitutions for each disulfide bond that were based on the assignments from the recently published crystal structure (2Lapthorn A.J. Harris D.C. Littlejohn A. Lustbader J.W. Canfield R.E. Machin K.J. Morgan F.J. Isaacs N.W. Nature. 1994; 369: 455-461Crossref PubMed Scopus (853) Google Scholar, 3Wu H. Lustbader J.W. Liu Y. Canfield R.E. Hendrickson W.A. Structure. 1994; 2: 545-558Abstract Full Text Full Text PDF PubMed Scopus (462) Google Scholar). Thus, both monomeric α subunit and the single chain analogs containing such modifications were constructed. The α cysteine mutants previously constructed in vector pM2 were used (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). Exon III of the α subunit contains a unique XbaI restriction site within the sequence encoding amino acid residues 34 and 35 (Fig.2 A). Because the vector also contains a singleXbaI site, fragments bearing α10, α28, or α32 mutations were exchanged with XbaI-digested pM2containing α60, α82, or α84 substitutions. This resulted in double cysteine mutants α10–60, α28–82, and α32–84, respectively. Construction of the pM2 α7–31 and pM2 α59–87 were described previously (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). Single chain variants were constructed with the carboxyl end of the β subunit fused to the amino end of the α subunit using overlap polymerase chain reaction mutagenesis (11Sugahara T. Pixley M.R. Minami S. Perlas E. Ben-Menahem D. Hsueh A.J.W. Boime I. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 2041-2045Crossref PubMed Scopus (112) Google Scholar) (Fig. 2 B). All constructs generated by polymerase chain reaction were sequenced to ensure that the final products contained no misincorporated residues. Transfection, selection of stable CHO clones, metabolic labeling, and immunoprecipitation with subunit-specific antiserum were previously described (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar, 14Graham F.L. van der Eb A.J. Virology. 1973; 52: 456-467Crossref PubMed Scopus (7037) Google Scholar, 15Matzuk M.M. Krieger M. Corless C.L. Boime I. Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 6354-6358Crossref PubMed Scopus (84) Google Scholar). Analysis by Western blotting was performed using conditioned media that were concentrated with Centriprep-10 columns (Amicon, Beverly, MA). Samples were loaded on 12.5% SDS-polyacrylamide gels and electroblotted to nitrocellulose (Hybond ECL, Amersham International, UK), and the proteins were detected by the Western Light kit (Tropix, Bedford, MA). To determine the biological activity, the variants were quantitated either by an hCG RIA (Diagnostic Products Corporation, Los Angeles, CA) containing CGβ polyclonal antiserum or an hCG dimer-specific enzyme-linked immunosorbent assay kit (Organon, Oss, The Netherlands). Both assays gave comparable results. Conditioned media were incubated with human fetal kidney 293 cells stably transfected with human LH/CG receptor, and the binding affinities (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar) and cAMP accumulation were determined (16Davoren J.B. Hsueh A.J.W. Biol. Reprod. 1985; 33: 37-52Crossref PubMed Scopus (125) Google Scholar). Iodination of hCG (CR-127; 14,900 IU/mg) was performed as described (17Thorell J.I. Johansson B.G. Biochim. Biophys. Acta. 1971; 251: 363-369Crossref PubMed Scopus (1089) Google Scholar); the specific activity and maximum binding of125I-labeled hCG, as determined by radioligand receptor assay (18Rao M.C. Richards J.S. Midgley Jr., A.R. Reichert Jr., L.E. Endocrinology. 1977; 101: 512-523Crossref PubMed Scopus (87) Google Scholar), were 53,000 cpm/ng and 40%, respectively. Nonspecific binding was determined by adding a 1000-fold excess of unlabeled ligand (Pregnyl, Organon); specific binding routinely was 10–12% of total125I-labeled hCG added. The cAMP levels were also determined in 293 cells expressing human LH receptors by radioimmune assay (16Davoren J.B. Hsueh A.J.W. Biol. Reprod. 1985; 33: 37-52Crossref PubMed Scopus (125) Google Scholar). The proposed disulfide bonds in the α subunit are at positions 7–31, 10–60, 28–82, 32–84, and 59–87 (Fig. 1). Previously, we demonstrated that α subunits containing single mutations at Cys-10, -28, -60, -82, and -84 were not secreted and in most cases were degraded intracellularly (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). These mutants also failed to assemble with hCGβ subunit, whereas mutants with alterations at Cys-7, -31, -32, -59, or -87 were secreted and assembly-competent. Because a free thiol group could alter intracellular behavior (9Suganuma N. Matzuk M. Boime I. J. Biol. Chem. 1989; 264: 19302-19307Abstract Full Text PDF PubMed Google Scholar, 10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar, 19Alberini C.M. Bet P. Milstein C. Sitia R. Nature. 1990; 347: 485-487Crossref PubMed Scopus (107) Google Scholar, 20Isidoro C. Maggioni C. Demoz M. Pizzagalli A. Fra A.M. Sitia R. J. Biol. Chem. 1996; 271: 26138-26142Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar), double mutants in the α subunit were constructed using the assignments based on the crystallographic analysis (2Lapthorn A.J. Harris D.C. Littlejohn A. Lustbader J.W. Canfield R.E. Machin K.J. Morgan F.J. Isaacs N.W. Nature. 1994; 369: 455-461Crossref PubMed Scopus (853) Google Scholar, 3Wu H. Lustbader J.W. Liu Y. Canfield R.E. Hendrickson W.A. Structure. 1994; 2: 545-558Abstract Full Text Full Text PDF PubMed Scopus (462) Google Scholar). To assess secretion/stability of these mutants, transfected cells were labeled with [35S]cysteine and subjected to pulse-chase analysis. The intracellular (lysate) and extracellular (media) forms were immunoprecipitated and resolved on SDS gels (Fig. 3). Except for the 10–60 variant, the secretion rate of the mutants and wild-type subunit were comparable. As described previously, the αWT, 7–31, and 59–87 mutants were efficiently secreted because greater than 95% of the subunits were recovered from the media (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). In contrast, the recovery of the α10–60, 28–82, and α32–84 double mutants was 21, 41, and 20%, respectively (Fig. 3; TableI). The decreased recovery was not due to lower synthesis but rather to enhanced degradation, since at zero time of chase the intracellular accumulation of the mutants is comparable (Fig.3). These data suggest the cystine knot structure is critical for maximum α subunit secretion. Although the secreted non-combined αWT is heterogeneous, the media form of α10–60 is more homogeneous, suggesting that a post-translational modification is affected by this mutation.Table ISecretion of α subunit cysteine mutantsVariantt 1/21-aTime when half of the maximal signal is detected in the media with free α subunit only.RecoveryMonomer1-bThe amount of hormone retrieved from the media at steady state and expressed as a fraction of the total (lysate + medium).Dimer1-cClones co-expressing hCGβ WT and α subunits were pulse-labeled and chased for up to 24 h. The samples from each clone were divided into two aliquots and immunoprecipitated with α or hCGβ antisera. Recovery is determined by comparing the quantity of α subunit in the secreted dimer (detected with β antiserum) with the total α subunit pool (lysate + media) using polyclonal α subunit antiserum.min%αWT130 ± 1292 ± 483 ± 17–31135 ± 181-dData from previous study (10).>951-dData from previous study (10).77 ± 610–60195 ± 3421 ± 6<528–82123 ± 301-eMean ± range of two experiments.41 ± 61-eMean ± range of two experiments.49 ± 532–84114 ± 61-eMean ± range of two experiments.20 ± 41-eMean ± range of two experiments. 951-dData from previous study (10).90 ± 5Values represent the mean ± S.E. of at least two experiments.1-a Time when half of the maximal signal is detected in the media with free α subunit only.1-b The amount of hormone retrieved from the media at steady state and expressed as a fraction of the total (lysate + medium).1-c Clones co-expressing hCGβ WT and α subunits were pulse-labeled and chased for up to 24 h. The samples from each clone were divided into two aliquots and immunoprecipitated with α or hCGβ antisera. Recovery is determined by comparing the quantity of α subunit in the secreted dimer (detected with β antiserum) with the total α subunit pool (lysate + media) using polyclonal α subunit antiserum.1-d Data from previous study (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar).1-e Mean ± range of two experiments. Open table in a new tab Values represent the mean ± S.E. of at least two experiments. To evaluate the efficiency of dimer formation, the α constructs were co-transfected with the hCGβ gene (21Matzuk M.M. Boime I. J. Cell Biol. 1988; 106: 1049-1059Crossref PubMed Scopus (186) Google Scholar), and clones synthesizing excess β subunit were selected to ensure that it would not limit assembly. The cells were labeled, and equal aliquots of lysate/medium were precipitated with subunit-specific polyclonal antisera (Fig.4, Table I). Recovery of the 10–60 and 32–84 mutants in the secreted heterodimer was less than 5% although synthesis of all the mutants was at significant levels. In the case of the 28–82 mutant (50% recovery), the combination efficiency is relatively unaffected, since the amount of α subunit precipitated by either α or β subunit antiserum was comparable. Pulse-chase experiments demonstrated that the α7–31, 28–82, and 59–87 mutants combined efficiently with the β subunit, and their recovery paralleled the uncombined α subunit synthesized in the absence of co-transfected β subunit (TableI). The low amount of dimers containing the α10–60 or α32–84 mutant reflects a decrease in combination efficiency and/or an enhanced intracellular degradation of the mutated α subunit. The data show that the disulfide bonds 10–60 and 32–84 (and to a lesser extent 28–82) are critical for assembly. Due to decreased dimer accumulation in the media, studies on the biologic action of the heterodimers containing the 10–60 or 32–84 mutation are difficult. However, if the β subunit is covalently linked to the α subunit, the rate-limiting assembly step is bypassed, and sufficient material could be obtained to examine signal transduction. The disulfide bond mutants described above were linked to the wild-type β subunit to form single chain analogs (see “Materials and Methods”), and the intracellular behavior was examined by pulse-chase experiments. The secretion kinetics of the single chain variants (Fig. 5, Table II) paralleled the corresponding monomers. The variants including the nonmutated tether (CGβα) were secreted at a comparable t½ (CGβα t½ = 106 min), while the single chain α10–60 mutant is secreted much slower (t½ = 244 min; see Table II). Thus, even in the single chain construct, the 10–60 bond is critical for secretion. However, compared with the corresponding heterodimers (compare Tables Iand II), recovery of the mutants 10–60 and 32–84 dramatically increases when incorporated in a single chain.Table IISecretion of α mutants in CG tetherVariantt 1/2Recoverymin%CGβα106 ± 474 ± 57–31107 ± 581 ± 510–60244 ± 3636 ± 328–82102 ± 1555 ± 532–84138 ± 3245 ± 659–87102 ± 1387 ± 4See the legend of Table I for description of quantitation. CGβα corresponds to the nonmutated CG single chain. Open table in a new tab See the legend of Table I for description of quantitation. CGβα corresponds to the nonmutated CG single chain. Because little heterodimer is formed with α mutants 10–60 and 32–84, we suspected that the conformation of the single chain bearing these mutated α domains would be changed. To address this issue, conditioned media were examined by Western blot and probed by two dimer-specific monoclonal antibodies: A407 and B109 (22Krichevsky A. Birken S. O'Connor J.F. Bikel K. Schlatterer J.P. Canfield R.E. Endocrine. 1994; 2: 511-520Google Scholar, 23Moyle W.R. Campbell R.K. Rao S.N.V. Ayad N.G. Bernard M.P. Han Y. Wang Y. J. Biol. Chem. 1995; 270: 20020-20031Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar, 24Krichevsky A. Armstrong E.G. Schlatterer J. Birken S. O'Connor J. Bikel K. Silverberg S. Lustbader J.W. Canfield R.E. Endocrinology. 1988; 123: 584-593Crossref PubMed Scopus (42) Google Scholar, 25Moyle W.R. Matzuk M.M. Campbell R.K. Cogliani E. Dean-Emig D.M. Krichevsky A. Barnett R.W. Boime I. J. Biol. Chem. 1990; 265: 8511-8518Abstract Full Text PDF PubMed Google Scholar, 26Cosowsky L. Rao S.N.V. Macdonald G.J. Papkoff H. Campbell R.K. Moyle W.R. J. Biol. Chem. 1995; 270: 20011-20019Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar) (Fig.6). An A407 epitope resides in amino acid residues 5 and 6, 11–13, and 81 of the α subunit (23Moyle W.R. Campbell R.K. Rao S.N.V. Ayad N.G. Bernard M.P. Han Y. Wang Y. J. Biol. Chem. 1995; 270: 20020-20031Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar). Under nonreducing conditions, A407 recognizes 7–31, 28–82, 59–87, nonmutated CGβα single chain, and CG heterodimer (Fig. 6 A). However, 10–60 and 32–84 are barely detected (Fig. 6 A). These low signals are not due to a difference in the recovery of the blotted protein, since reprobing with polyclonal β antiserum shows comparable signals for all mutants, including 10–60 and 32–84 (Fig. 6 B). Moreover, it is apparent that the migration of the mutants under nonreduced conditions is altered by the mutations, implying that the conformation of the single chain mutants and CGβα are not the same. Similar data were obtained with B109, which is specific for the dimer form of the β subunit (24Krichevsky A. Armstrong E.G. Schlatterer J. Birken S. O'Connor J. Bikel K. Silverberg S. Lustbader J.W. Canfield R.E. Endocrinology. 1988; 123: 584-593Crossref PubMed Scopus (42) Google Scholar, 25Moyle W.R. Matzuk M.M. Campbell R.K. Cogliani E. Dean-Emig D.M. Krichevsky A. Barnett R.W. Boime I. J. Biol. Chem. 1990; 265: 8511-8518Abstract Full Text PDF PubMed Google Scholar, 26Cosowsky L. Rao S.N.V. Macdonald G.J. Papkoff H. Campbell R.K. Moyle W.R. J. Biol. Chem. 1995; 270: 20011-20019Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar) (Fig. 6 C). Tethers containing mutations 7–31, 28–82, and 59–87 and the nonmutated βα are recognized as well as the heterodimer, but single chains containing either the 10–60 or 32–84 mutation show very weak signal compared with the same blotted membrane using CGβ antiserum (Fig.6 D). Both A407 and B109 recognized higher molecular weight proteins in CGβα 28–82 and 59–87. They are likely noncovalently linked because when the samples are boiled under nonreducing conditions, the “aggregates” disappeared (data not shown). (It is not clear why the aggregates are detected by monoclonal antibodies but not by polyclonal antiserum; we are currently purifying these proteins for extensive characterization.) These results suggest that the conformation of single chain mutants 10–60 and 32–84 differ substantially from the nonmutated single chain and the heterodimeric forms. Because we could now obtain sufficient quantities of single chains bearing the cystine knot mutants, the influence of the disulfide bonds on receptor binding/signal transduction was examined using human kidney 293 cells expressing the LH/hCG receptor. For these bioassays, conditioned media were quantitated with a β-specific polyclonal based radioimmune assay (see “Materials and Methods”). It is apparent that except for the 59–87 mutant, the binding affinity of all of the mutants was comparable with the nonmutated tether (Fig.7 A, Table III). The binding affinity of single chain 59–87 was reduced 10-fold, similar to that for the heterodimer containing this mutation in the monomeric α subunit as previously reported (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar).Table IIIBiological activity of α cysteine mutants in CG single chainBinding IC503-aConcentrations required to displace 50% of the maximal 125I-hCG bound. Values are presented as mean ± S.E. of at least three independent experiments.cAMP ED503-bConcentration inducing half-maximal cAMP levels. Data are the mean ± S.E. of at least two independent experiments.Coupling factor (IC50/ED50)3-cThe ratio of IC50 versus the ED50. Numbers are shown as mean ± S.E. of at least two independent experiment sets using same samples.ng/mlng/mlCGβα3-dCGβα corresponds to the nonmutated CG single chain.23.3 ± 7.94.3 ± 0.83.4 ± 1.47–316.7 ± 1.77.4 ± 1.81.0 ± 0.310–6026.9 ± 8.411.9 ± 3.53-eMean ± range of two experiments.2.9 ± 1.23-eMean ± range of two experiments.28–826.5 ± 2.16.9 ± 0.51.0 ± 0.232–8414.1 ± 3.112.9 ± 3.61.6 ± 0.659–87254.3 ± 55.443.2 ± 15.58.1 ± 2.73-a Concentrations required to displace 50% of the maximal 125I-hCG bound. Values are presented as mean ± S.E. of at least three independent experiments.3-b Concentration inducing half-maximal cAMP levels. Data are the mean ± S.E. of at least two independent experiments.3-c The ratio of IC50 versus the ED50. Numbers are shown as mean ± S.E. of at least two independent experiment sets using same samples.3-d CGβα corresponds to the nonmutated CG single chain.3-e Mean ± range of two experiments. Open table in a new tab Adenylate cyclase activation by the single chain was also determined (Fig. 7 B). The data demonstrate that signal transduction parallels receptor binding affinity, suggesting that the determinant(s) necessary for bioactivity is conserved between CG heterodimer and single chain (Fig. 7 B, Table III). Thus, despite differences in the intracellular behavior and immunoreactivity to conformationally sensitive monoclonal antibodies, the mutants are nevertheless biologically active in vitro, with coupling of receptor binding to signal transduction unaffected by these structural modifications. Within a species, the amino acid sequence of the α subunit is identical (1Pierce J.G. Parsons T.F. Annu. Rev. Biochem. 1981; 50: 465-495Crossref PubMed Scopus (1940) Google Scholar), and it is the unique structure of the β subunit that ensures binding of each dimer to its cognate receptor. The α subunit, therefore, must be very flexible, since it combines with each of the four glycoprotein β subunits and β subunits from different species (27Chen F. Puett D. J. Biol. Chem. 1991; 266: 6904-6908Abstract Full Text PDF PubMed Google Scholar). Further evidence for this adaptability is the conformational changes occurring during dimer formation (for review, see Refs. 1Pierce J.G. Parsons T.F. Annu. Rev. Biochem. 1981; 50: 465-495Crossref PubMed Scopus (1940) Google Scholar and28Hojo H. Ryan R.J. Endocrinology. 1985; 117: 2428-2434Crossref PubMed Scopus (27) Google Scholar, 29Benkirane M.M. Bon D. Costagliola S. Paolucci F. Darbonnet B. Prince P. Carayon P. Endocrinology. 1987; 121: 1171-1177Crossref PubMed Scopus (34) Google Scholar, 30Ehrlich P.H. Moustafa Z.A. Krichevsky A. Birken S. Armstrong E.G. Canfield R.E. Am. J. Reprod. Immunol. Microbiol. 1985; 8: 48-54Crossref PubMed Scopus (43) Google Scholar, 31Garnier J. McKerns K.W. Structure and Function of the Gonadotropins. Plenum Publishing Co., New York1978: 381-414Crossref Google Scholar, 32Strickland T.W. Puett D. Endocrinology. 1982; 111: 95-100Crossref PubMed Scopus (31) Google Scholar, 33Corless C.L. Bielinska M. Ramabhadran T.V. Daniels-McQueen S. Otani T. Reitz B.A. Tiemeier D.C. Boime I. J. Biol. Chem. 1987; 262: 14197-14203Abstract Full Text PDF PubMed Google Scholar, 34Corless C.L. Matzuk M.M. Ramabhadran T.V. Krichevsky A. Boime I. J. Cell. Biol. 1987; 104: 1173-1181Crossref PubMed Scopus (92) Google Scholar). Presumably, the configuration of the disulfide bonds of the α subunit is critical for its interaction with the β subunit. Previous studies on these bonds in the monomeric α subunit demonstrated that single mutations in the cystine knot blocked secretion and/or inhibited heterodimer formation (10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar). Although the results implicated a critical role for these bonds in the intracellular behavior of the subunit, we could not exclude the possibility that the free thiol, rather than the disrupted bond per se, led to the observed changes (9Suganuma N. Matzuk M. Boime I. J. Biol. Chem. 1989; 264: 19302-19307Abstract Full Text PDF PubMed Google Scholar, 10Furuhashi M. Ando H. Bielinska M. Pixley M.R. Shikone T. Hsueh A.J.W. Boime I. J. Biol. Chem. 1994; 269: 25543-25548Abstract Full Text PDF PubMed Google Scholar, 19Alberini C.M. Bet P. Milstein C. Sitia R. Nature. 1990; 347: 485-487Crossref PubMed Scopus (107) Google Scholar, 20Isidoro C. Maggioni C. Demoz M. Pizzagalli A. Fra A.M. Sitia R. J. Biol. Chem. 1996; 271: 26138-26142Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar). To address this issue, double cysteine mutations of all the proposed pairs were constructed in the α subunit. It was apparent that mutations in the knot altered the recovery of the subunit from the media. Recovery of variants containing mutations outside the cystine knot, α7–31 and α59–87, was comparable with the wild type α subunit. These changes in the secretion patterns presumably reflect alterations in the folding of the subunit. Consistent with this hypothesis are the experiments showing that, compared with the wild-type α subunit, the cystine knot variants 10–60 and 32–84 combined much less efficiently to the β subunit, with less than 5% of these mutants recovered as dimers. Thus, assessing the role of the α10–60 and α32–84 bonds in the biological action of hCG was precluded. Because tethers containing either mutation at 10–60 or 32–84 were recovered in the media, we could determine their biologic activity. Unexpectedly, it was observed that all of the cystine knot and the α7–31 mutants exhibited high affinity binding for the receptor. When corrected for binding, signal transduction was unchanged regardless of the mutation. These results imply that disulfide bonds of the cystine knot are required for maximal secretion/assembly but that for receptor recognition, not all of the bonds in the core are needed. The single chain bearing the α59–87 mutant exhibited a 10-fold decrease in receptor binding. Several studies have demonstrated the importance of the C-terminal residues 88–92 in the α subunit for maximum binding affinity (1Pierce J.G. Parsons T.F. Annu. Rev. Biochem. 1981; 50: 465-495Crossref PubMed Scopus (1940) Google Scholar, 35Bielinska M. Pixley M.R. Boime I. J. Cell Biol. 1990; 111: 330aGoogle Scholar, 36Chen F. Wang Y. Puett D. Mol. Endocrinol. 1992; 6: 914-919PubMed Google Scholar, 37Zeng H. Ji I. Ji T.H. Endocrinology. 1995; 136: 2948-2953Crossref PubMed Google Scholar, 38Grossmann M. Szkudlinski M.W. Zeng H. Kraiem Z. Ji I. Tropea J. Ji T.H. Weintraub B.D. Mol. Endocrinol. 1995; 9: 948-958Crossref PubMed Google Scholar). It is likely that disrupting the 59–87 bond perturbed the configuration of the carboxyl-terminal region, leading to the observed decrease in receptor binding. That the altered intracellular behavior of these tethers is not accompanied by a significant change in biologic activity, especially the 10–60 and 32–84 mutants, suggests that the receptor could recognize different conformations of the hormone. While this conclusion is supported by the absence of recognition by the dimer-specific monoclonal antibodies, we cannot exclude a greater liability of the epitopes in the 10–60 and 32–84 mutants to the SDS-polyacrylamide gel electrophoresis conditions used. Given that perturbing the cystine knot affects the intracellular behavior of both the monomeric α subunit and the single chain molecule, the conformation of this region is likely to be relatively conserved regardless of the associated hormone-specific β subunit. This fixed core determinant may be necessary for the “escort” (42Willnow T. Armstrong S.A. Hammer R.E. Herz J. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 4537-4541Crossref PubMed Scopus (248) Google Scholar,43Kim, P., and Arvan, P. (1997) Endocr. Rev., in press.Google Scholar) function of the α subunit, which is critical for rescuing the pituitary β subunits from the endoplasmic reticulum because few, if any, uncombined LH, follitropin, and thyrotropin β subunits are secreted in the absence of co-expressed α subunit (34Corless C.L. Matzuk M.M. Ramabhadran T.V. Krichevsky A. Boime I. J. Cell. Biol. 1987; 104: 1173-1181Crossref PubMed Scopus (92) Google Scholar, 39Keene J.L. Matzuk M.M. Otani T. Fauser B.C.J.M. Galway A.B. Hsueh A.J.W. Boime I. J. Biol. Chem. 1989; 264: 4769-4775Abstract Full Text PDF PubMed Google Scholar, 40Matzuk M.M. Kornmeier C.M. Whitfield G.K. Kourides I.A. Boime I. Mol. Endocrinol. 1988; 2: 95-100Crossref PubMed Scopus (62) Google Scholar). Thus, we would propose that the major conformational changes of the α subunit resulting from assembly with the β subunit occurs in regions outside of the cystine knot, i.e. the three hairpin loops (2Lapthorn A.J. Harris D.C. Littlejohn A. Lustbader J.W. Canfield R.E. Machin K.J. Morgan F.J. Isaacs N.W. Nature. 1994; 369: 455-461Crossref PubMed Scopus (853) Google Scholar, 3Wu H. Lustbader J.W. Liu Y. Canfield R.E. Hendrickson W.A. Structure. 1994; 2: 545-558Abstract Full Text Full Text PDF PubMed Scopus (462) Google Scholar, 23Moyle W.R. Campbell R.K. Rao S.N.V. Ayad N.G. Bernard M.P. Han Y. Wang Y. J. Biol. Chem. 1995; 270: 20020-20031Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar, 41DeBeer T. VanZuylen C.W.E.M. Leeflang B.R. Hard K. Boelens R. Kaptein R. Kamerling J.P. Vliegenthart J.F.G. Eur. J. Biochem. 1996; 241: 229-242Crossref PubMed Scopus (77) Google Scholar). Consistent with this interpretation, the recovery of the variants with the 7–31 and 59–87 mutations, which lie outside the knot, was comparable with the wild type. That the intracellular fate of single chains with mutations of paired cysteines and corresponding monomers parallel each other suggests that a misfolded epitope is not recognized by an intracellular transport component and/or that the mutants are trapped by a chaperone, which leads to enhanced degradation. Although the β domain increases the recovery of the mutated single chains, e.g. in CGβα10–60 and CGβα32–84, they are still less than that of the wild type. Thus, even the presence of the β subunit domain is not sufficient to completely override the α Cys mutations. Other members of the cystine knot growth factor superfamily display similar characteristics (6Brunner A.M. Lioubin M.N. Marquerdt H. Malacko A. Wang W-C. Shapiro R. Neubauer M. Cook M. Madisen L. Purchio A.F. Mol. Endocrinol. 1992; 6: 1691-1700Crossref PubMed Scopus (50) Google Scholar, 7Mason A. Mol. Endocrinol. 1994; 8: 325-332PubMed Google Scholar, 8Kenney W.C. Haniu M. Herman A. Arakawa T. Costigan V. Lary J. Yphantis D. Thomason A.R. J. Biol. Chem. 1994; 269: 12351-12359Abstract Full Text PDF PubMed Google Scholar). Mutations of the paired cysteine residues within the knot of tumor growth factor-β1 or activin dramatically reduced secretion (6Brunner A.M. Lioubin M.N. Marquerdt H. Malacko A. Wang W-C. Shapiro R. Neubauer M. Cook M. Madisen L. Purchio A.F. Mol. Endocrinol. 1992; 6: 1691-1700Crossref PubMed Scopus (50) Google Scholar, 7Mason A. Mol. Endocrinol. 1994; 8: 325-332PubMed Google Scholar). Thus, together with our data presented here, the cystine knot is apparently the basic frame of the glycoprotein hormone subunits and several growth factors and represents a critical determinant for secretion and assembly of functional ligand. Receptor recognition is apparently less dependent on the configuration of this structure. The data also show the importance of the single chain approach to study structure-function of multisubunit proteins where dependence on the assembly step is essential for biologic action. We thank Drs. David Ornitz, Mesut Muyan, and Edward Grotjan for critical review of the manuscript and Dr. Steven Birken for helpful discussions. Also, we are grateful to Susan Carnes for invaluable assistance in preparing this manuscript.
Referência(s)