| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
- and ßA-Subunit DimersDepartment of Neurobiology and Physiology (M.A., M.S., T.K.W.), Northwestern University, Evanston, Illinois 60208; Department of Molecular Biology and Pharmacology (I.B.), Washington University School of Medicine, St. Louis, Missouri 63110; Department of Medicine (T.K.W.), Northwestern University Medical School, Chicago, Illinois 60611; and Robert H. Lurie Comprehensive Cancer Center of Northwestern University (T.K.W.), Chicago, Illinois 60611
Address all correspondence and requests for reprints to: Teresa K. Woodruff, Ph.D., Department of Neurobiology and Physiology, Northwestern University, O. T. Hogan 4-150, 2205 Tech Drive, Evanston, Illinois 60208. E-mail: tkw{at}northwestern.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
/ß) and activin homodimer (ß/ß) assembly and secretion from ovarian granulosa cells is not fully understood. Here, we examined the role of inhibin subunit glycosylation in the assembly and secretion of mature inhibin A and activin A. Inhibition of subunit glycosylation by tunicamycin treatment of - and ßA-expressing CHO cell lines reduced inhibin but not activin secretion. Dimeric inhibin A is preferentially secreted from parental isogenic wild-type (wt) cell lines (wtßwt). Mutation of a single glycosylation site at asparagine 268 (
268ßwt) reduces inhibin secretion by 78% and permits ß/ß assembly and secretion. Conversely, gain of a glycosylation (GOG) site in the analogous region of the ßA-subunit (wtßGOG327) enhances inhibin A secretion. The present study demonstrates that N-linked glycan sites direct heterodimer vs. homodimer assembly, and prevention of glycosylation abrogates inhibin secretion. These data support a definitive role for site-specific N-glycosylation in governing inhibin/activin dimer assembly and secretion. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
- and ß-subunits. Like many endocrine and paracrine hormones, inhibin must be assembled from large multimeric precursor proteins, with the mature protein secreted into the extracellular environment in a regulated manner. The intracellular assembly of inhibin heterodimers must be tightly controlled, because ß-subunits also homodimerize to produce the inhibin antagonist activin. Although much of this process has been delineated in a broad way, the details of inhibin and activin hormone assembly and secretion are still not completely understood.
The inhibin - and ßA-subunits are synthesized as pre-prohormones. The human -subunit precursor is 366 amino acids, consisting of a 18-amino acid signal peptide sequence, 43-amino acid proregion, 171-amino acid N-terminus (N) region, and a 134-amino acid C-terminus (C) mature region (1) (Fig. 1A
). The -subunit has seven conserved cysteine residues in the mature domain forming a cysteine knot motif crucial to the formation of the three-dimensional monomer and dimer protein structure, characteristic of all TGFß superfamily members (for review, see Ref. 2). The human -subunit contains three N-linked glycosylation sites, one in the pro-N region (Asn146) and two in the mature C region of the protein (Asn268, Asn302) (1, 3). All other vertebrate species contain only two N-linked glycosylation sites in the -subunit and lack the third N-glycan consensus sequence (Asn302).
|
). The mature ßA-subunit contains nine cysteine residues necessary for both intermolecular and intramolecular disulfide bonds. The ßA-subunit has one glycosylation site in the precursor domain of the protein (Asn165).
The addition of N-linked glycans to the inhibin - and ßA-subunits starts in the lumen of the endoplasmic reticulum (ER), with the transfer of oligosaccharides to the asparagine residues in the sequence Asn-X-Ser/Thr (NXS/T) in the nascent polypeptide chain (4). Like other TGFß superfamily proteins, the inhibin subunits undergo disulfide-linked dimerization in the ER before transport to the Golgi. Dimeric inhibin precursors are then cleaved by furin-like proconvertases at an RXRR site that separates the propeptides from the mature domains of each subunit (Fig. 1, C and D). It has been suggested that the mature -subunit protein arises after cleavage of the precursor at amino acids 232–233; however, the role of protein cleavage in mature protein production is complex, because noncleavable high-molecular-weight inhibin A is biologically active, whereas noncleavable activin A is not (1). The dimeric, mature hormones are secreted from the cell and regulate FSH secretion from the pituitary in a classic endocrine negative feedback loop (5).
Although it is unknown whether glycosylation of inhibin affects assembly, secretion, or biological activity, mutagenesis of the multiple glycosylation sites on TGFß has been shown to result in nonsecreted and inactive protein products (6). Another superfamily member, bone morphogenetic protein-1 (BMP-1), contains six glycosylation sites, one in the prodomain and five in the mature active protein (7). Mutagenesis of N-linked glycosylation sites in specific domains of the mature BMP-1 peptide result in secretion and stability defects. Moreover, the gonadotropins, particularly human chorionic gonadotropin (hCG), depend on N-linked glycosylation for dimer stability (8). Based on these related proteins, we suggest that N-linked glycosylation of the inhibin subunits influences heterodimer (/ß) or homodimer (ß/ß) assembly, secretion, and stability of human inhibin A and activin A.
To directly examine this hypothesis, the N-linked glycosylation sites in both inhibin - and ßA-subunits were eliminated by site-directed mutagenesis to create a series of -subunit glycomutants. Additionally, gain of glycosylation (GOG) ßA-subunit mutants were generated. We predicted that additional carbohydrate moieties would increase the secretion and stability of inhibin dimers. The assembly, secretion, and bioactivity of wild-type (wt), glycomutant, and GOG mutants of inhibin A isogenic cell lines were investigated. The data prove that N-linked glycosylation in the inhibin A molecule is necessary for dimer assembly and secretion. The elimination of all glycosylation sites in the -subunit reduces the secretion of inhibin dimers. A single glycan site (Asn268) ensures the proper folding of these polypeptides and directs the overall production of heterodimers and homodimers. These studies provide new insights into inhibin/activin assembly and secretion as well as provide an important model for intracellular processing of other multisubunit secreted proteins.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
- and ßA-subunits (9). To examine the secreted proteins derived from these cells, conditioned medium was trichloroacetic acid (TCA) precipitated and subjected to SDS-PAGE under reducing conditions. Immunoblot analysis using an antibody specific to the C mature protein demonstrated two large C-containing precursor proteins, pro-N-C of 52 and 49 kDa (Fig. 2A), in which the difference in size was dependent on its triglycosylated or diglycosylated status as has been reported previously (1, 3). Two distinct mature C bands are detected at 24 and 21 kDa and represent diglycosylated and monoglycosylated forms of the mature C protein (1). The secreted products were also examined for the expression of the ßA-subunit using an anti-ßA-subunit polyclonal antibody (Fig. 2B). The cells secrete two distinct immunoreactive bands of this protein: a larger pro-ßA precursor protein (54 kDa) with one N-glycan site within the prodomain of the protein and a mature ßA-subunit (14 kDa).
|
). Medium was immunoprecipitated with an anti--subunit antibody (lanes 1 and 3) or an anti-ßA-subunit antibody (lanes 2 and 4) and subjected to SDS-PAGE under reducing conditions. The inhibin precursor (
49 kDa) and mature (21 and 24 kDa) -subunit proteins were detected in the medium after immunoprecipitation with either antibody (lanes 1 and 2). The anti-ßA-subunit antibody also immunoprecipitated both the inhibin - and ßA-subunits (14 kDa), suggesting that these subunits were immunoprecipitated as a dimer (lane 2).
With tunicamycin treatment, mature -subunit (21 and 24 kDa) accumulation in the medium was greatly diminished, and precursor forms of this subunit shifted from approximately 49 kDa to approximately 43 kDa because of the absence of N-linked oligosaccharides (lanes 3 and 4). These data suggest that the secretion of mature -subunit is inhibited in the absence of its sugar moieties. Immunoprecipitation with the anti-ßA-subunit antibody resulted in the detection of mature ßA-subunit protein (14 kDa) and the precursor -subunit (43 kDa) (lane 4). These observations suggest that nonglycosylated mature inhibin A is not secreted from CHO cells, whereas activin A secretion is not affected by glycosylation-deficient conditions. The functional role of each glycosylation site in the - and ßA-subunits with respect to their assembly and secretion was further investigated.
Generation and Analysis of Inhibin - and ßA-Subunit Glycomutant CHO Cells
Three independent isogenic cell lines for each of the wt and inhibin A subunit glycomutants shown in Table 1 were generated. All lines were verified for isogenic integration, propagated, and characterized. Studies of the -subunit glycomutants are described first, followed by a discussion of studies on the ßA-subunit glycomutant and GOG mutants.
|
-subunit glycomutants was performed using specific monoclonal antibodies or polyclonal antisera to the - and ßA-subunits, respectively (Fig. 3). All media were subjected to TCA precipitation for immunoblot analysis attributable to the overall low levels of secretion for all isogenic cell lines (for overall secretion levels, see Fig. 4). The wt inhibin A (wtßwt) protein profile by immunoblot analysis under reducing conditions is shown in Fig. 3A (lane 1). The precursor protein gave rise to predominately a diglycosylated precursor form that migrated at approximately 49 kDa, as well as smaller amounts of triglycosylated (52 kDa) and monoglycosylated (46 kDa) precursor forms. Mature C proteins appear as monoglycosylated (21 kDa) and diglycosylated (24 kDa), confirming data reported previously (1).
|
|
268ßwt, 146,268ßwt, and 146,268,302ßwt cell lines did not secrete mature C proteins (Fig. 3A, lanes 3, 5, 7). Interestingly, immunoreactive subunit protein bands in 146,268ßwt and 146,268,302ßwt migrated at a high molecular mass (60 and 100 kDa), suggesting that these may be inhibin -subunit precursor proteins (Fig. 3A, asterisks). Glycomutant 146ßwt has a similar precursor protein profile as wt but lacked the triglycosylated precursor protein attributable to the single deletion of an N-glycan site (lane 2). 302ßwt secreted both diglycosylated and monoglycosylated precursor forms (lane 4), whereas glycomutant 146,302ßwt secreted only a monoglycosylated precursor protein (lane 6). Both of these glycomutants secreted a monoglycosylated mature -subunit protein attributable to the removal of one glycosylation site in the mature domain. We were not able to generate any stable cell lines for 268,302ßwt after numerous attempts. Any clonal isolates of this glycomutant in CHO-Flp cells could not be sustained during the selection and propagation process for an undetermined reason (data not shown).
To monitor inhibin ßA-subunit protein secretion from each -subunit glycomutant cell line, the blots were stripped and reanalyzed with a polyclonal antibody to the ßA-subunit (Fig. 3B). The wt protein wtßwt was expressed as both a precursor (54 kDa) and a mature (14 kDa) protein (lane 8). Most glycomutants produced mature ßA-subunit, with the exception of mutants 146,268ßwt and 146,268,302ßwt (Fig. 3B, lanes 12 and 14).
Peptide N-Glycosidase F (PNGase F) Treatment of Inhibin -Subunit Glycomutants
To measure the amount of inhibin subunit protein retained in each cell line, wt and -subunit glycomutant cell lysates were analyzed (Fig. 3C). Mature -subunit proteins were rarely detected in cell lysates, suggesting that any mature proteins that were generated were secreted into the media. The lysates were also treated with PNGase F, which cleaves all high mannose and complex forms of N-linked oligosaccharides, to verify glycosylation had occurred. PNGase F digestion of wtßwt generated a faster migrating -subunit precursor protein (43 kDa) as a result of the removal of all sugar chains (Fig. 3C). The removal of a single -subunit N-glycan site (146ßwt, 268ßwt, and 302ßwt) resulted in proteins of approximately 46 kDa. A shift of 6 kDa (3 kDa/glycan site) was observed with the addition of PNGase F, suggesting that these mutant proteins were modified at the two remaining glycosylation sites. A smaller shift down in protein size was observed for 146,302ßwt (from 46 to 43 kDa), which contained only one glycan site in the protein. 146,268ßwt was resistant to PNGase F treatment (similar to that observed for 146,268,302ßwt), suggesting that the removal of these two glycan sites impedes the attachment of additional oligosaccharides at the Asn302 site of the -subunit protein (asterisk).
Each of the cell lysates were assessed for the expression of the inhibin ßA-subunit protein (Fig. 3D). We confirmed that the pro-ßA protein is glycosylated only at Asn165 of the precursor domain. In the presence of PNGase F, these proteins shifted from approximately 54 kDa to approximately 49 kDa in size. Interestingly, glycomutants 146,268ßwt and 146,268,302ßwt had a higher molecular mass and were insensitive to PNGase F treatment, suggesting that these proteins are not glycosylated (asterisk). It is possible that the removal of two N-glycan sites from the -subunit proteins alters the overall structure and folding of the -subunit protein, resulting in the higher molecular aggregates for both the - and ßA-subunits observed in 146,268ßwt and 146,268,302ßwt in all immunoblots shown in Fig. 3.
Overall, the most striking finding from the above immunoblot experiments were that 1) no mature -subunit protein accumulated in medium from mutant 268ßwt CHO cells, and 2) no mature forms of either of the inhibin subunits accumulated in the media of mutants 146,268ßwt and 146,268,302ßwt.
Secretion of Inhibin A from Stable Isogenic Glycomutant Cell Lines
To investigate whether mutations in glycosylation sites result in less efficient secretion of bioactive inhibin A, the secretion of inhibin A glycomutants was measured. Serum-free media and lysates were collected from cell lines after 48 h, and inhibin A levels were measured by ELISA. Media was normalized by counting the number of cells plated in each well and corrected for total protein content. The 48 h inhibin A secretion rate from wt cells was 5730 ± 530 pg/ml (Fig. 4). The secretion levels of all glycomutants were significantly different from wtßwt, with the exception of 146ßwt, the mutant that lacks the glycosylation site in the precursor protein. Mutation of a single glycosylation site at Asn268 reduced inhibin A secretion by 78% (analyzed further in Fig. 7; see below). The two other mutants lacking Asn268 (146,268ßwt and 146,268,302ßwt) did not have any measurable levels of inhibin A. These results correlate well with the immunoblot data that showed that these same glycomutants lacked any mature C protein bands in TCA-precipitated culture media (Fig. 3A). A single mutation in 302ßwt resulted in significantly higher levels of hormone secretion compared with wt inhibin A, whereas the double glycomutant 146,302ßwt produced significantly lower levels of secreted inhibin A when compared with wtßwt. As stated above, three independent clonal cell lines were generated for wt inhibin A and each glycomutant. Characterization of each cell line by immunoblot analyses and ELISA techniques demonstrated that each of these cell lines behaved identically. The relationship between the secretion and bioactivity data varied by approximately 18% among the additional clonal lines. These results show that the removal of N-glycan sites on the -subunit alters the overall secretion of dimeric inhibin A.
|
146,268ßwt and 146,268,302ß Are Retained Intracellularly-subunit glycomutants 146,268ßwt and 146,268,302ßwt had a secretion profile that was unique among all other glycomutants (Fig. 3), with proteins that migrated as a higher molecular mass (60 kDa) than seen in their wt counterpart. In addition, the ELISA data presented above (Fig. 4) suggested that these glycomutants are unable to secrete mature inhibin A.
To test whether dimeric or single-chain mutant inhibin proteins are retained intracellularly in these glycomutants, cell lysates were harvested at the same time that the conditioned medium was collected from six-well plates after a 48-h period and measured in an inhibin A ELISA. wtßwt cell lysates contained 976 pg/ml inhibin A dimers (Fig. 5A), which is approximately 5-fold less inhibin A protein than is secreted (Fig. 4). In contrast, the two glycomutants 146,268ßwt and 146,268,302ßwt had significantly greater amounts of intracellular dimeric inhibin A (4170 and 4020 pg/ml, respectively), with levels reaching approximately four times that of wt cells. These data suggest that the mutant - and ßA-subunits are capable of dimerization but are secretion deficient.
|
-subunit, and analyzed by SDS-PAGE and autoradiography. The pulse-chase data from media is graphically represented in Fig. 5B. By 4 h of chase, the mature -subunit was detected in media collected from wtßwt cells. However, no mature -subunit protein was detected in media collected from 146,268ßwt and 146,268,302ßwt cells. Over a 5-h time course, the wt (wtßwt) protein continued to accumulate, whereas the lack of secretion of the two glycomutants (146,268ßwt and 146,268,302ßwt) confirmed that dimeric inhibin A was retained intracellularly.
Subcellular Localization of Secretion-Deficient Glycomutants
The failure of proteins to attain their proper conformational state after biosynthesis results in the selective elimination of these misfolded proteins by ER quality control mechanisms (10). N-linked glycans play a critical role in the folding of proteins, and our data suggested that glycomutants 146,268ßwt and 146,268,302ßwt may be misfolded and thus retained in the cell. To investigate this possibility, indirect immunofluorescence experiments were conducted. wtßwt CHO cells had low immunofluorescent staining for the inhibin -subunit, confirming our previous data suggesting that these proteins are secreted from the cell (Fig. 5Ci). 146,268ßwt and 146,268,302ßwt had abundant -subunit staining in the cytoplasm of each cell (Fig. 5Cii and Ciii).
The subcellular localization of the glycomutant -subunit proteins was determined by dual-labeling immunofluorescence with antibodies to the -subunit (Fig. 5Di) and the ER-specific protein calnexin (Fig. 5Di). -Subunit immunoreactivity in 146,268,302ßwt overlapped extensively with the immunoreactivity of the calnexin protein (Fig. 5Diii). The observed immunofluorescence for the inhibin -subunit extended beyond the staining for the ER, suggesting that this subunit may also be in the Golgi complex. Thus, these data further demonstrate that these -subunit glycomutants accumulate intracellularly and are retained in the ER and possibly Golgi.
Mutagenesis of Asn268 Triggers the Preferential Assembly of Activin A at the Expense of Inhibin A
The three mutants 268ßwt, 146,268ßwt, and 146,268,302ßwt, which lack Asn268, secreted significantly lower levels of inhibin A compared with wt. Interestingly, Asn268 is six amino acids C-terminal to a conserved cysteine residue (Cys262) located in the predicted intramolecular cysteine knot motif of inhibin -subunit. Whereas inhibin A was not detected in media collected from the 146,268ßwt and 146,268,302ßwt glycomutants, lower levels of inhibin dimers were measured for the 268ßwt mutant (Fig. 4). The 268ßwt glycomutant, unlike 146,268ßwt and 146,268,302ßwt, was glycosylated at the two remaining glycosylation sites, as determined by the PNGase F experiments (see above). Although this glycomutant did not produce observable mature inhibin C protein, the mature ßA-subunit protein was easily detected. We examined whether these mutants were capable of making activin A. Media from all glycomutants were collected, and an activin A ELISA was performed. Of all cell lines tested, only 268ßwt produced significant amounts of activin A (Table 2). This mutation caused the preferential assembly of two ßA-subunits to give rise to bioactive activin A. The bioactivity of concentrated glycomutant 268ßwt medium was tested using an activin-responsive FSH ß-luciferase reporter gene (Fig. 6A). This glycomutant behaved very similarly to wt activin A by stimulating FSHß reporter expression. These data suggests that a disruption of Asn268 in the -subunit protein impedes assembly of - and ßA-subunit heterodimers and favors the production of ßA-subunit homodimers.
|
|
AAS), an additional mutant was generated. The Ile269 residue was converted to an alanine residue (NIS NAS), giving rise to an 269ßwt mutant. Analysis of the secretion profile of these mutant and wt proteins in CHO cells under nonreducing conditions is shown in Fig. 6B. The 269ßwt mutant secreted dimeric inhibin A similar to that observed for wtßwt, in contrast to 268ßwt, which did not secrete any dimeric forms of inhibin A. 269ßwt secreted inhibin precursor proteins and both diglycosylated and monoglycosylated mature C protein (Fig. 6C). This mutant also secreted pro-ßA and mature ßA proteins (Fig. 6D). This data suggests that the disruption of Asn268 and specifically attachment of asparagine-linked oligosaccharides inhibit dimerization and secretion of inhibin proteins.
Single or Double N-Linked Glycomutants Retain Biological Activity
The bioactivity of glycomutants with single or double N-glycan sites deleted, as in 146ßwt, 302ßwt, and 146,302ßwt, was assessed quantitatively by measuring their ability to antagonize the actions of activin A on the activin-responsive FSH ß-luciferase reporter gene. Removal of the N-glycan site in the precursor domain of the inhibin -subunit (146ßwt) and in conjunction with the Asn302 site (146,302ßwt) did not alter bioactivity, and these mutants antagonized activin signaling in a dose-dependent manner (Table 3).
|
-subunit protein sequences reveals that the Asn302 glycosylation consensus sequence is present only in the human inhibin -subunit. The removal of this glycosylation site does not impede dimerization, although secretion was enhanced (Fig. 4). This glycomutant was less efficient at antagonizing activin A activity than wt inhibin A (Table 3).
The combined data from the 146ßwt, 302ßwt, and 146,302ßwt glycomutant studies demonstrate that the determinants for the bioactivity are uncoupled from those involved in assembly.
Deletions or Additions of N-Linked Oligosaccharides in the Inhibin ßA-Subunit
The absence of a N-linked glycan site at Asn268 in the -subunit disrupted the ability of this subunit to heterodimerize with a ßA-subunit. We predicted that the addition of N-linked glycan chains on the analogous site of the inhibin ßA-subunit would also affect the ability of the ßA-subunit to heterodimerize or homodimerize. An alignment of the human - and ßA-subunit amino acid sequences identified two sites in the ßA-subunit mature domain, which, by a single amino acid transition, could yield consensus N-linked glycosylation sites (Fig. 7A). To determine whether these additional glycosylation sites affect the assembly and secretion of inhibin A, two GOG mutants were generated on the ßA-subunit: wtßGOG327 and wtßGOG365, which are homologous to Asn268 and Asn302 of the -subunit, respectively. We postulated that the addition of N-linked glycans to the ßA-subunit might favor the formation of inhibin A dimers over activin A dimers.
To test this hypothesis, isogenic cell lines for mutants of the inhibin ßA-subunit were generated (Table 1). An additional glycomutant was generated in which the N-linked glycosylation site in the pro-ßA domain was eliminated (wtß165). The expression of ßA-subunit precursor and mature proteins for each of these glycomutants was assessed by immunoblotting (Fig. 7B). The elimination of a glycan site in the pro-ßA domain (wtß165) caused a shift down in precursor protein size compared with wt (Fig. 7B, lane 2). As expected, the protein profile for the ßA-subunit of wtßGOG327 was predominately diglycosylated pro-ßA and monoglycosylated mature ßA-subunit (lane 3). wtßGOG365 resulted in diglycosylated pro-ßA and both nonglycosylated and monoglycosylated mature ßA-subunit forms (lane 4). The expression of the -subunit protein did not differ in any of these glycomutants (Fig. 7C).
The cell lysates for these glycomutants were then treated with PNGase F to ensure that the N-linked glycan sites were removed or incorporated into the ßA-subunit sequence. Immunoblot analysis of wt ßA-subunit protein revealed that the precursor protein was approximately 54 kDa, and the single N-linked glycan could be cleaved with PNGase F treatment to give rise to an approximately 49 kDa protein (Fig. 7D). The addition of a N-linked glycan site in the mature domain of the ßA-subunit caused the diglycosylated precursor protein to migrate at approximately 57 kDa, which was reduced to 49 kDa with PNGase F treatment. The -subunit expressed in these isogenic cell lines were all glycosylated normally (49 kDa) and were reduced to approximately 43 kDa in the presence of PNGase F (Fig. 7E). The secretion of the inhibin -subunit was unaffected by alterations to the ßA-subunit in these glycomutants.
Addition of an N-Linked Glycan at a Specific Site Affects Inhibin A Secretion and Stability
The accumulation of inhibin A dimers in the media of the inhibin ßA-subunit glycomutants was measured by ELISA. All mutant cell lines secreted inhibin A, and significant increases in the overall secretion of inhibin A were observed for wtß165 and wtßGOG327 but not for wtßGOG365 (Fig. 8A). Surprisingly, a GOG site at amino acid Phe327, which is homologous to Asn268 of the inhibin -subunit, became a "supersecretor" of inhibin A (21,700 pg/ml), secreting approximately four times more inhibin A than wt (5730 pg/ml). An activin A ELISA was also performed on these mutants. Any activin A produced by these glycomutants was below the detectable levels of the ELISA (data not shown). The addition of a glycan site at Ser365 in wtßGOG365, which would be analogous to Asn302 site of the -subunit, had no significant effect on the secretion of inhibin A. The bioactivity of these molecules was tested, and there was no significant decrease in the ability of any of these ligands to antagonize the action of activin A on the FSHß-luciferase reporter gene (Fig. 8B).
|
). The secretion rates of inhibin A from wtß165 or wtßGOG327 did not differ from wt. However, mature wtßGOG327 was secreted approximately 2.3 times faster than wt inhibin. wtßGOG327 was remarkably stable throughout the chase period, more so than wt (Fig. 8C). Together, these results demonstrate that the addition of an N-linked glycan favors the assembly of inhibin heterodimers and stabilizes the mutant protein. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Position of N-Glycan Sites Is Critical to Posttranslational Modifications that Instruct Homodimer or Heterodimer Assembly
A series of loss and gain of glycosylation inhibin - and ßA-subunit mutants were generated to explore the role of glycosylation on homodimer or heterodimer assembly and secretion. Loss of a single N-linked glycan site in the -subunit at position Asn268 resulted in greatly reduced heterodimer formation. In the absence of this glycan, the homodimer activin A was secreted and accumulated in the medium of isogenic 268ßwt cell lines expressing both - and ß-subunit mRNAs. The intramolecular cysteine knot structure of the -subunit has not been solved by x-ray crystallography; however, it is known from studies of activin and other members of the TGFß superfamily that the first cysteine residue of the -subunit, Cys262, is involved in a key intramolecular disulfide bond pair between Cys262 and Cys328 (1). Moreover, it is known that this intramolecular structure supports the four-finger fold of the monomeric protein. The position of the key N-glycan site in the -subunit, Asn268, that is necessary for heterodimer secretion is located six amino acids from the first cysteine residue of the mature protein. When a similarly positioned N-glycan site at Asn52 in the hCG -subunit, another cysteine knot protein, was deleted, a critical intramolecular disulfide bridge and the capacity to assemble heterodimers were lost (14). Likewise, the absence of the carbohydrate moiety at Asn268 in the inhibin -subunit leads to loss of heterodimer formation and the preferential assembly of activin homodimers. Thus, inhibin dimer assembly relies on glycosylation on Asn268 of the inhibin -subunit.
Presence or Addition of an N-Glycan Site Proximal to Cys262 Results in More Rapid Assembly of Heterodimers
The data support the notion that N-glycans are necessary for the assembly and secretion of the appropriate dimer when strategically located near a specific cysteine residue of the knot motif of the inhibin ßA-subunit. However, the addition of N-linked carbohydrate moieties did not interfere with the proper assembly and secretion of inhibin A. Results from bioassays demonstrated that wtßGOG327 was capable of antagonizing activin A similar to wt inhibin A. This is not surprising, because the addition of N-glycans to other hormones, such as FSH, does not alter their bioactivity (15). The additional glycosylation site in wtßGOG327 presumably stabilizes the intramolecular disulfide bond, as was suggested in the case of the -subunit mutant, and increases the efficiency of secreted heterodimer formation. The addition of this sugar moiety in wtßGOG327 caused a significant increase in the assembly of inhibin dimers compared with wt. This is also supported by studies of hCG, in which the addition of N-glycosylation in the C-terminal domain of the -subunit increased the efficiency of heterodimer formation (16).
We predict that this GOG site in the ßA-subunit would not interfere with activin A assembly. Activin A production in this isogenic cell line was below detectable levels. However, previous studies from our laboratory demonstrate that activin A GOG mutants are bioactive (17). Studies of an additional member of the TGFß superfamily, nodal, also support these findings, because insertion of an N-glycan site in the C-terminus increased the stability of nodal protein and its signaling range (18). The addition of an N-glycan near Cys321 residue of the mature protein appears to have significant effects on inhibin A assembly compared with the wtßGOG365 mutant. wtßGOG365 was very similar to wt in terms of bioactivity and rate of secretion, suggesting that the location of glycan addition is important for enhanced dimer assembly. Together, the data support a model in which the presence of N-glycans near a conserved specific cysteine allows for more efficient monomer folding and stability, followed by immediate dimer assembly.
Glycosylation-Deficient Mutants Are Not Secreted from the Cell
Treatment of inhibin A cell lines with tunicamycin or mutating the glycan sites in the -subunit results in secretion-deficient mutants. Misfolded proteins are typically sorted from normal proteins by ER quality control and targeted for degradation (for review, see Ref. 10). Interestingly, inhibin dimers from glycomutants 146,268ßwt and 146,268,302ßwt are retained intracellularly. Consistent with this, mature subunits were not detected in lysates or secreted media. The prodomain of proteins in the TGFß superfamily are proteolytically processed by subtilisin-like proprotein proconvertases. Proteolytic maturation of inhibin subunits is necessary for inhibin A secretion (Antenos, M., and T. K. Woodruff, unpublished data), and secretion may be tightly coupled to the addition of N-linked glycans to the inhibin -subunit.
Evolution of a TGFß Antagonist
Our studies open intriguing new questions about the generation of the only known TGFß superfamily antagonist inhibin. Other heterodimers that exist in the TGFß family are strong agonists of specified pathways (19). Additionally, few members of this superfamily, outside of inhibin and BMP, contain glycosylation sites in their mature domains. Inhibin is also the only known true endocrine hormone of the superfamily. It is synthesized by the ovary under the influence of FSH and travels through the peripheral circulatory system to the pituitary in which it acts to block activin, thereby inhibiting FSH synthesis and secretion (5). This is a classically defined negative feedback system that is engaged in early puberty and exists throughout the fertile lifespan of an individual. Inhibin and activin may act as functional antagonists because of their structural similarities. However, the mechanism by which cells assemble homodimers or heterodimers has remained somewhat elusive until now. The stoichiometry of the subunits was thought to underlie preferential hormone assembly. Certainly, this is still a factor in this process; however, our data extend this basic premise by implicating differential glycosylation in the selection of heterodimer or homodimer assembly. These findings could be important clinically for conditions such as polycystic ovarian syndrome for which a variety of changes at both the neuroendocrine level and the level of the ovary result in the loss of normal cyclicity. In patients with this syndrome, follicular arrest is associated with deficient inhibin A and B biosynthesis (20). An unexplored contribution of biosynthetic regulation of inhibin secretion may contribute to this disease and will be the focus of additional experiments.
The present study reveals the evolutionary significance of the precise positioning of the N-linked glycans on the -subunit as it relates to the biosynthesis of the appropriate homodimeric or heterodimeric hormones (Fig. 9). Glycosylation adjacent to a critical and conserved cysteine involved in the assembly of the intramolecular cysteine knot ensures the heterodimerization of subunits, because removal of this critical site on the -subunit resulted in the preferential assembly of activin. Thus, when FSH stimulates -subunit expression, the resulting gene product is glycosylated, ensuring that inhibin heterodimers are produced to maintain the necessary negative feedback to the pituitary gland.
|
-subunit, which in turn alter the ability of these glycoforms to bind to betaglycan, a coregulatory protein important for inhibin action (3). These findings point to a very refined level of control over the evolution of inhibin subunits, which involved new gene duplication event followed by cotranslational modifications that ensure cycle-dependent biosynthesis and secretion. |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
- and ßA-subunits was constructed using pIRES plasmid (Clontech, Mountain View, CA). The pIRES plasmid ensures enhanced transcription and translation of the inhibin -subunit cDNA cloned into the first multiple cloning site (MCS) over the production of the ßA-subunit in the second cloning site, allowing for the preferential production of inhibin A rather than activin A. The inhibin -subunit cDNA, amplified from InhA vector (Genentech, South San Francisco, CA) was inserted into the first MCS at the XhoI site of pIRES. The human cDNA for the full-length human ßA-subunit was amplified from pRSV-ßA (Genentech) and inserted into the XbaI site of the second MCS of pIRES. The inhibin -IRES-ßA insert was then digested with NheI and NotI restriction enzymes and ligated into the Invitrogen (Carlsbad, CA) pcDNA5-FRT cloning vector of the Flp-In System.
Mutagenesis and Incorporation of N-Linked Glycosylation Sites
Site-directed mutagenesis was performed by PCR as described previously (17). Primers used for mutagenesis are listed in Table 4. Mutants were confirmed by DNA sequence analysis at the Northwestern University Biotech Core Facility.
|
- and ßA-subunit proteins. Three independent clones for wt and each glycomutant were selected for propagation and additional characterization. To analyze the secretion of each inhibin subunit from isogenic cell lines by immunoblotting procedures, TCA protein precipitation was performed on serum-free medium collected after a 48-h period. Briefly, 500 µl serum-free medium were precipitated with 20% TCA for 30 min on ice. Samples were microfuged for 10 min at 4 C, and the pellets were then washed with cold acetone and centrifuged, for a total of two acetone washes. Pellets were further dried at 80 C and resuspended in 1x sample buffer (with or without ß-mercaptoethanol) and loaded onto polyacrylamide gels.
Protein Production
After assessment of all glycomutant secretion rates, all cell lines were grown in Nunclon surface triple flasks (VWR, West Chester, PA) to maximize the total amount of conditioned media to be collected for additional bioactivity tests. At confluence, cells were grown for 5 d in Ham’s F12 serum-free media before collection. Media was filtered and concentrated using Amicon Ultra Centrifugal Filter Units (Millipore, Billerica, MA). Concentrated protein was dialyzed with 50 mM Tris and 150 mM NaCl (pH 8). Purified human inhibin A from DSL Laboratories (Webster, TX) was used as a standard to quantify the amount of inhibin A secreted from each glycomutant. Human activin A ligand (purified in our laboratory) was used as a standard to quantify the amount of activin A secreted from glycomutant 268ßwt. Immunoblot analysis was performed using the mouse antihuman inhibin -subunit antibody that recognizes amino acids 1–32 of the inhibin -subunit (Serotec, Raleigh, NC) or a polyclonal antibody specific for the inhibin ßA-subunit (provided by Dr. W. Vale, The Salk Institute, La Jolla, CA), followed by the appropriate secondary antibody conjugated to horseradish peroxidase (GE Healthcare, Little Chalfont, UK), and detected using ECL plus (GE Healthcare). The concentration of wt and mutant inhibin A solutions was determined by measuring immunoreactivity, and the signal was quantified using the Kodak 4000MM Digital Imaging System and analyzed using Kodak Imaging Software, version 4.0.1 (Eastman Kodak, Rochester, NY).
N-Glycosylation Inhibitor
Inhibin A-secreting CHO cells (NU-INHA1-BR) were maintained as described previously (9). Cells were pretreated with 2.5 µg/ml tunicamycin (Sigma, St. Louis, MO) for 2 h and metabolically labeled for 18 h with [35S]methionine/cysteine (MP Biomedicals, Irvine, CA). Immunoprecipitation was performed as described previously (21) using the antihuman inhibin -subunit antibody.
Deglycosylation Experiments
Deglycosylation with PNGase F was performed according to the recommendations of the manufacturer (New England Biolabs, Beverly, MA).
Inhibin A and Activin A ELISAs
Levels of inhibin A or activin A in serum-free media and cell lysates collected from six-well plates (2.5 x 106 cells per well) were determined after a 48-h incubation period and assessed in triplicate in three independent experiments by ELISA (DSL Laboratories) as described previously (22). The limit of detection for the inhibin A assay is 5 pg/ml, with interassay and intraassay variations of 3.1 and 16.8%, respectively. The limit of detection for the activin A assay is 5 ng/ml, with interassay and intraassay variations of 1.2 and 7%, respectively.
Indirect Immunofluorescence
Experiments were performed as described previously (21). Anti-inhibin R1, anti-inhibin ßA, and anti-calnexin (Abcam, Cambridge, MA) antibodies were used and detected with species-specific secondary antibodies conjugated with fluorescein isothiocyanate (FITC) or cyanine 3 (Jackson ImmunoResearch, West Grove, PA). Coverslips were mounted with Vectashield (Vector Laboratories, Burlingame, CA) and viewed with a Nikon Eclipse E600 microscope (Diagnostic Instruments, Sterling Heights, MI). Omission of the primary antibodies was used as a control.
Metabolic in Vivo Labeling and Subunit Immunoprecipitations
[35S]Cysteine labeling of continuous isogenic cultures and pulse-chase experiments were performed as described previously (23). Inhibin -subunit antibody, inhibin ßA-subunit antiserum, or normal rabbit IgG was used for immunoprecipitation.
Luciferase Assay
The bioactivity of wt or glycomutant inhibin A concentrated media and activin A (purified in our laboratory) was determined using LßT2 gonadotrope cells stably transfected with the –338 region of the rat FSHß promoter conjugated to a luciferase reporter (24). Cells were lysed after treatment with varying concentrations of inhibin A-conditioned media and challenged with 10 ng/ml activin A for 6 h. Luciferase activity was measured for 30 sec using an AutoLumat Luminometer (Berthold Technologies, Oak Ridge, TN).
Statistics
Values are reported as means ± SEM and analyzed using Prism (version 4.0a; GraphPad Software, San Diego, CA). ANOVA, followed by the appropriate post hoc test (Tukey’s or Kruskal-Wallis), was used to evaluate differences between glycomutants. Statistical significance was reported if P < 0.05.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online April 24, 2007
Abbreviations: BMP, Bone morphogenetic protein; C, 134-amino acid C terminus; ER, endoplasmic reticulum; FITC, fluorescein isothiocyanate; GOG, gain of glycosylation; hCG, human chorionic gonadotropin; MCS, multiple cloning site; N, 171-amino acid N terminus; pIRES, plasmid containing internal ribosomal entry site; PNGase F, peptide N-glycosidase F; TCA, trichloroacetic acid; wt, wild type.
Received for publication January 25, 2007. Accepted for publication April 20, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
subunit. J Biol Chem 264:17113–17118-subunit. Endocr J 50:245–253[CrossRef][Medline]-subunit is critical for secretion and stability of the human thyrotropin ß-subunit. Mol Endocrinol 2:95–100This article has been cited by other articles:
 |
|
 |
|
  E. F. Ozdowski, Y. M. Mowery, and C. Cronmiller stall Encodes an ADAMTS Metalloprotease and Interacts Genetically With Delta in Drosophila Ovarian Follicle Formation Genetics, November 1, 2009; 183(3): 1027 - 1040. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Makanji, P. D. Temple-Smith, K. L. Walton, C. A. Harrison, and D. M. Robertson Inhibin B Is a More Potent Suppressor of Rat Follicle-Stimulating Hormone Release than Inhibin A in Vitro and in Vivo Endocrinology, October 1, 2009; 150(10): 4784 - 4793. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Xia and A. L Schneyer The biology of activin: recent advances in structure, regulation and function J. Endocrinol., July 1, 2009; 202(1): 1 - 12. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Antenos, J. Zhu, N. M. Jetly, and T. K. Woodruff An Activin/Furin Regulatory Loop Modulates the Processing and Secretion of Inhibin {alpha}- and {beta}B-Subunit Dimers in Pituitary Gonadotrope Cells J. Biol. Chem., November 28, 2008; 283(48): 33059 - 33068. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Makanji, K. L. Walton, M. C. Wilce, K. L. Chan, D. M. Robertson, and C. A. Harrison Suppression of Inhibin A Biological Activity by Alterations in the Binding Site for Betaglycan J. Biol. Chem., June 13, 2008; 283(24): 16743 - 16751. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
) and dimerization (
) are shown in the 
