Search

Follistatin (FST)-type proteins are important antagonists of some members of the large TGF-β family of cytokines. These include myostatin, an important negative regulator of muscle growth, and the closely related activin A, which is involved in many physiological functions, including maintenance of a normal reproductive axis. FST-type proteins, including FST and FST-like 3 (FSTL3), differentially inhibit various TGF-β family ligands by binding each ligand with two FST-type molecules. In this study, we sought to examine features that are important for ligand antagonism by FST-type proteins. Previous work has shown that a modified construct consisting of the FST N-terminal domain (ND) followed by two repeating follistatin domains (FSD), herein called FST ND-FSD1-FSD1, exhibits strong specificity for myostatin over activin A. Using cell-based assays, we show that FST ND-FSD1-FSD1 is unique in its specificity for myostatin as compared with similar constructs containing domains from FSTL3 and that the ND is critical to its activity. Furthermore, we demonstrate that FSD3 of FST provides affinity to ligand inhibition and confers resistance to perturbations in the ND and FSD2, likely through the interaction of FSD3 of one FST molecule with the ND of the other FST molecule. Additionally, our data suggest that this contact provides cooperativity to ligand antagonism. Cross-linking studies show that this interaction also potentiates formation of 1:2 ligand-FST complexes, whereas lack of FSD3 allows formation of 1:1 complexes. Altogether, these studies support that domain differences generate FST-type molecules that are each uniquely suited ligand antagonists.

Follistatin (FS) is an important regulator of pituitary FSH secretion through its potent ability to bind and bioneutralize activin. It also represents a prototype for binding proteins that control bioavailability of other TGFbeta-related growth factors such as the bone morphogenetic proteins. The 288-residue FS molecule has a distinctive structure comprised principally of three 10-cysteine FS domains. These are preceded by an N-terminal segment shown by us previously to contain hydrophobic residues essential for activin binding. To establish the contribution of the FS domains themselves to FS's bioactivity, we prepared mutants with deleted or exchanged domains and intradomain point mutations. Mutants were expressed from mammalian (Chinese hamster ovary) cells and evaluated for activin binding and for biological activity in assays measuring differing aspects of FS bioactivity: activin-mediated transcriptional activity and suppression of FSH secretion in primary pituitary cell cultures. The N-terminal domain (residues 1-63) alone could not bind activin or suppress activin-mediated transcription, either alone or combined in solution with the FS domain region (residues 64-288). Deletion of FS domains 1 or 2 abolished activin binding and biological activity in both assays, whereas deletion of domain 3 was tolerated. Bioactivity was also reduced or eliminated after exchange of domains (FS 2/1/3 and FS 3/1/2) or doubling of domain 1 (FS 1/1/3) or domain 2 (FS 2/2/3). Several hydrophobic residues clustered within the C-terminal region of FS domains 1 and 2 are highly conserved among all FS domains. Mutation of any of these to Asp or Ala either reduced or eliminated FS bioactivity and disrupted distant epitopes for heparin binding (FS domain 1) or antibody recognition (FS domain 2), suggesting their role in maintaining the conformational integrity of the domain and possibly the FS molecule as a whole. These results are consistent with the importance of domain conformation as well as the overall order of the domains in FS function. A continuous sequence comprising the N-terminal domain and followed by FS domains 1 and 2 fulfills the minimum structural requirement for activin binding and FS bioactivity.

A highly conserved ten amino acid proregion separates the peptidylglycine alpha-hydroxylating monooxygenase (PHM) domain of the bifunctional peptidylglycine alpha-amidating monooxygenase (PAM) protein from the NH2-terminal signal peptide; propeptides with amino acid sequences similar to the PAM proregion have been identified in other secreted proteins. In AtT-20 cells, but not in human embryonic kidney (hEK)-293 cells, an endogenous endoprotease acting at a site distal to the trans-Golgi network efficiently removes the propeptide from stably transfected monofunctional PHM (PHMs). We constructed a mutant PHM protein (delta ProPHMs) in which the proregion was deleted and the signal peptide joined directly to the monooxygenase domain. Newly synthesized, enzymatically active delta ProPHMs was secreted from both AtT-20 cells and hEK-293 cells more slowly than PHMs. In endocrine cells, the proregion was not required for storage in regulated secretory granules. We transferred the PAM proregion to prohormone convertase 2 (PC2), another soluble constituent of secretory granules, to determine whether the effect of the proregion were transferrable. In both AtT-20 cells and hEK-293 cells, the PAM/PC2 fusion molecule was able to exit the endoplasmic reticulum more rapidly than PC2.

The structure of the glycoprotein hormones (LH, CG, FSH, and TSH) and their mechanism of receptor recognition are problems of long-standing interest and speculation. Here we describe the two-dimensional [1H]nuclear magnetic resonance ([1H]NMR) analysis of a linear peptide model for the intercysteine sequence (38-57) from the beta-subunit of human (h) LH. This sequence contains functional determinants for receptor binding and postreceptor activation and is predicted by computer-based modeling to fold as a compact minidomain containing a central amphipathic helix. To test this prediction, an Arg-extended disulfide-free (38-57) analog of enhanced solubility was prepared for complementary circular-dichroic and two-dimensional NMR studies. The linear peptide retains ovarian membrane receptor-binding activity. Although the peptide is not highly structured in aqueous solution, circular-dichroic analysis shows partial alpha-helix formation in a lipophilic medium (50% trifluoroethanol). Complete sequential assignment is obtained in 50% trifluoroethanol based on homonuclear and [15N]edited heteronuclear NMR methods. alpha-Helix-related (i,i + 3) connectivities are observed by nuclear-Overhauser effect spectroscopy that define an amphipathic alpha-helical segment (residues 41-48). Additional long range nuclear-Overhauser effects are observed in the C-terminal region that are consistent with beta-turns involving one or more proline residues; these may serve to reverse the direction of the peptide chain. A nuclear-Overhauser effect contact is identified between residues 38 and 55 at opposite ends of the linear sequence, suggesting that a loop configuration is significantly populated in this solvent system. These results, taken together, characterize elements of ordered structure in the 38-57 peptide, which appear to be distinguishing features of hLH (and the homologous region of hCG). We propose that the structure of this peptide provides a model for the structure of the corresponding region of native hLH in the hormone-receptor complex.

Mouse pro-ACTH/endorphin (or POMC) contains in its sequence each of the four possible pairs of basic amino acids recognized as potential cleavage sites in the production of bioactive peptides from higher mol wt precursors: KR (lysine-arginine), RR, RK, and KK. To examine the structural requirements for processing and routing in one region of pro-ACTH/endorphin, a reporter mutation was introduced into the mouse pro-ACTH/endorphin cDNA; a methionine residue was mutated to an isoleucine residue to allow biosynthetic double labeling with [3H]Ile and [35S]Met. Analysis of stable cell lines expressing the reporter cDNA indicated that this mutation did not affect processing or secretion. Therefore, additional mutations were introduced on the reporter background to investigate important structural features of the precursor. First, the tripeptide signal for N-linked glycosylation in the N-terminal glycopeptide (Asn65,Ser66,Ser67) was disrupted by the conservative substitution of asparagine65 with a glutamine residue. Secondly, O-glycosylation was prevented by substitution of threonine45 with an alanine residue. Finally, lysine50 was mutated to an arginine residue, transforming the RK doublet preceding the gamma 3MSH sequence into an RR doublet. The results show that the enzymatic machinery of AtT-20 cells fails to cleave efficiently at the Arg-Lys (RK) site even after elimination of any possible structural hindrance by carbohydrate side-chains. Elimination of O-linked oligosaccharides to the N-terminal side of gamma 3MSH did not allow cleavage at the RK site, and elimination of N-linked oligosaccharides did not alter the processing and routing of pro-ACTH/endorphin in AtT-20 cells. However, mutation of the RK sequence to RR allowed extensive cleavage regardless of the occurrence of O- or N-glycosylation.

Several regions on both the alpha- and beta-subunits of human LH comprise the receptor-binding domain of the hormone. One of these, a disulfide loop peptide containing residues 38-57 on the beta-subunit, also stimulated steroidogenesis in rat Leydig cells. Circular dichroism analysis and a Schiffer-Edmundson helical wheel projection of beta-(38-57) revealed the possibility of an amphipathic alpha-helical structure through its N-terminal region. Secondary structure prediction algorithms do not predict alpha-helix in beta-(38-57), but, rather, suggest a sheet-coil-sheet topology. Homology searches between this peptide and proteins with known structure revealed that the two best matches are with prealbumin-(10-30) and melittin-(1-26). Based on hydrophobic moment calculations, we suggest that beta-(38-57) more closely resembles melittin, a known example of an amphipathic helix. Molecular models were constructed that included an alpha-helix between Pro-39 and Pro-50 producing a hydrophilic face involving Thr-40, Arg-43, and Gin-46. Loop closure was performed either visually or by an incremental minimization procedure, using distance constraints to patch in a disulfide bond. Molecular dynamics at 300, 360, and 1000 K were used to explore the local conformational space, and dynamic structures were minimized. The most reasonable structures were found with the 300 and 360 K simulations, with those at 360 K consistently producing structures with lower conformational energies. In each of these simulations, the N-terminus of the alpha-helix unraveled to form a reverse turn (predicted by the GOR algorithm) which include Cys-38, Pro-39, Thr-40, and Met-41. Simulations at 1000 K produced the most variation in structure, but these were deemed unreasonable. Although not all possible conformations were explored, several models were found that comply with the assumption of an amphipathic helix in the N-terminal half of the peptide.

Many bioactive peptides terminate with an amino acid alpha-amide at their COOH terminus. The enzyme responsible for this essential posttranslational modification is known as peptidyl-glycine alpha-amidating monooxygenase or PAM. We identified cDNAs encoding the enzyme by using antibodies to screen a bovine intermediate pituitary lambda gt11 expression library. Antibodies to a beta-galactosidase/PAM fusion protein removed PAM activity from bovine pituitary homogenates. The 108,207 dalton protein predicted by the complete cDNA is approximately twice the size of purified PAM. An NH2-terminal signal sequence and short propeptide precede the NH2 terminus of purified PAM. The sequences of several PAM cyanogen bromide peptides were localized in the NH2-terminal half of the predicted protein. The cDNA encodes an additional 430 amino acid intragranular domain followed by a putative membrane spanning domain and a hydrophilic cytoplasmic domain. The forms of PAM purified from bovine neurointermediate pituitary may be generated by endoproteolytic cleavage at a subset of the 10 pairs of basic amino acids in the precursor. High levels of PAM mRNA were found in bovine pituitary and cerebral cortex. In corticotropic tumor cells, levels of PAM mRNA and pro-ACTH/endorphin mRNA were regulated in parallel by glucocorticoids and CRF.

The intercysteine loop sequence (93-100) in the beta-subunit has been postulated to be important for receptor binding and specificity in the glycoprotein hormones, LH and human CG (hCG). To demonstrate this directly, and to characterize the structural features essential for activity, we prepared a series of synthetic peptides and analogs incorporating this determinant loop region. Peptides were assayed for inhibition of labeled hCG binding to ovarian membrane receptors and stimulation of testosterone production in Leydig cells. Peptides with the native (93-100) sequence from hCG and hLH inhibited hCG binding half maximally at 2.18 and 2.62 x 10(-4) M, respectively, while the sequence from FSH was inactive. Isosteric substitution of Ala for Cys resulted in an inactive peptide, indicating that the (93-100) disulfide bridge is essential for activity. Optimal binding activity requires at least one net positive charge among the side chains, as shown by loss of activity in hybrid analogs with neutral or negative charges conferred by progressive replacement of Arg by Asp at 94 and 95 or by introduction of Asp at 96 and 97. Despite binding to receptors, the native sequence did not promote testosterone production at doses up to 10(-2) M. This contrasts with a second receptor binding sequence, beta (38-57) that activates testosterone production. There are differences between the (93-100) and (38-57) loop sequences in their chemical and physical properties, biological activity and antigenicity. While the cumulative evidence suggests that they associate with counterpart sites in alpha-subunit to form a topographical binding domain in the whole hormone, our results suggest that each sequence may contribute in different ways to activation of postreceptor events.

The ability of purified bovine neurointermediate pituitary peptidyl glycine alpha-amidating monooxygenase to catalyze the conversion of peptide substrates (D-Tyr-X-Gly) into amidated product peptides (D-Tyr-X-NH2) was evaluated. The pH optimum of the reaction was pH 8.5 when X was Val, Trp, or Pro, but 5.5 to 6.0 when X was Glu. Similar maximum velocity (Vmax) values were obtained for the Val, Trp, and Pro substrates while the Glu substrate had a substantially higher Vmax. The Michaelis-Menten constant (Km) of the enzyme for the peptide substrate increased in the order Trp less than Val less than Pro much less than Glu. Increasing levels of ascorbate brought about parallel increases in Km and Vmax, suggesting the presence of an irreversible step separating the interaction of the enzyme with the two substrates. The effect of copper on enzyme activity was dependent on the peptide substrate and the reaction pH. With the Val substrate, exogenous copper was required for optimal activity; no other metal ion tested could substitute for copper. With the Glu substrate, exogenous copper was not required for optimal activity; however, diethyldithiocarbamate, a copper chelator, inhibited activity and only copper could reverse this inhibitory effect. The ability of various cofactors to stimulate alpha-amidating monooxygenase activity was also dependent on assay conditions. With the Val or Glu substrate in the presence of exogenous copper, a variety of cofactors in addition to ascorbate were capable of supporting activity. With the Glu substrate in the absence of exogenous copper, the requirement of the enzyme for ascorbate was more strict. In keeping with the proposed reaction mechanism, nearly 1 mol ascorbate was consumed for each mole of D-Tyr-Glu-NH2 produced.