| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 34, 35458-35468, August 20, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the Department of Obstetrics and Gynecology, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson (Rutgers) Medical School, Piscataway, New Jersey 08854
Received for publication, March 18, 2004 , and in revised form, May 14, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-subunit that has been likened to a "seatbelt" because it surrounds 
-subunit loop 2 and its end is "latched" by an intrasubunit disulfide bond to the -subunit core. As shown here, assembly begins when parts of the NH2 terminus, cysteine knot, and loops 1 and 3 of the -subunit dock reversibly with parts of the NH2 terminus, cystine knot, and loop 2 of the hCG -subunit. Whereas the seatbelt can contribute to the stability of the docked subunit complex, it interferes with docking and/or destabilizes the docked complex when it is unlatched. This explains why most hCG is assembled by threading the glycosylated end of -subunit loop 2 beneath the latched seatbelt rather than by wrapping the unlatched seatbelt around this loop. hCG assembly appears to be limited by the need to disrupt the disulfide that stabilizes the small seatbelt loop prior to threading. We postulate that assembly depends on a "zipper-like" sequential formation of intersubunit and intrasubunit hydrogen bonds between backbone atoms of several residues in the -subunit cystine knot, -subunit loop 2, and the small seatbelt loop. The resulting intersubunit -sheet enhances the stability of the seatbelt loop disulfide, which shortens the seatbelt and secures the heterodimer. Formation of this disulfide also explains the ability of the seatbelt loop to facilitate latching during assembly by the wraparound pathway. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-subunits is surrounded by a strand of their -subunits like a "seatbelt" (1–3). With the exception of some teleost fish follitropins, the seatbelts of most vertebrate hormones is "latched" by an intrasubunit disulfide to a cysteine in the -subunit core. The unusual structures of these heterodimers and the fact that their assembly can be studied within cells makes them useful for identifying factors that affect protein folding in the ER1 (4, 17–19). Assembly of most hCG, hFSH, and hTSH in the ER occurs after the seatbelt is latched by a process in which the glycosylated end of loop 2 is "threaded" through a hole in the -subunit (17). This process is facilitated by disruption of a disulfide that we have termed the "tensor" because it stabilizes a small loop within the seatbelt that regulates its length (18). Disruption of the tensor disulfide prior to assembly enlarges the hole in the -subunit and facilitates threading; reformation of the tensor disulfide following threading tightens the seatbelt around loop 2, which stabilizes the heterodimer. Alternatively, the hCG heterodimer can be assembled by a process in which the seatbelt is wrapped around loop 2 before the seatbelt latch disulfide is formed; formation of the seatbelt latch disulfide completes assembly. The "wrapping" mechanism, which was first proposed on the basis of pulse-chase analysis (4), appears to be used infrequently relative to the threading pathway for hCG assembly. Wrapping is required for assembly of hCG analogs in which the seatbelt is latched to a cysteine in the NH2-terminal end of the -subunit, a site comparable with that of the FSH -subunit found in salmon and many other teleost fish (19). It is also required for the assembly of heterodimers in which the seatbelt is latched to the -subunit (5).
Studies described here were designed to learn why most hCG is assembled by a threading mechanism rather than a wraparound mechanism. Conceivably, the latched seatbelt is a component of the subunit docking site. As a result, formation of the seatbelt latch disulfide would increase the affinity of the -subunit for the -subunit, thereby facilitating assembly by a threading route. Alternatively, the unlatched hCG seatbelt might occupy positions near the subunit interface where it would be in a position to interfere with subunit docking or where it can destabilize the docked complex. This would reduce the amount of docked complex and interfere with assembly by a wraparound route. The finding that the end of the seatbelt scans the -subunit to find its latch site (18) supports the notion that the seatbelt could contact the subunit interface. Efforts to distinguish these possibilities led us to study how the hCG - and -subunits dock and to determine how the latching of the seatbelt affected this process.
Using intersubunit disulfide bonds to stabilize partially assembled intermediates, we found that the hCG -subunit docks with the -subunit in similar but not identical fashions when its seatbelt is latched or unlatched. Most docked complexes appear to dissociate before the heterodimer can be assembled by either pathway, a phenomenon that would favor threading by providing more time for the -subunit to latch its seatbelt. Furthermore, the unlatched seatbelt appears to hinder docking and/or promote dissociation of the docked complex, a phenomenon that would also favor threading. By comparing the apparent positions of the subunits in the docked complexes with the structures of the assembled heterodimers, we devised models of heterodimer assembly. These suggest that while both threading and wrapping depend on the formation of similar intrasubunit and intersubunit hydrogen bonds, these would appear to form more readily by threading when the seatbelt is latched than by wrapping when it is unlatched.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-R8C,C93A,C100A represents an analog of the hCG -subunit in which codons for Arg8, Cys93, and Cys100 were replaced with cysteine, alanine, and alanine, respectively. The relative locations of antibody binding sites used in the hormone sandwich immunoassays are illustrated by Xing et al. (17). Briefly, most heterodimers were captured to microtiter plates using an antibody (A113) to the -subunit and detected using a radioiodinated antibody (B110 or B111) to the -subunit. Molecular modeling and cartoon illustrations were prepared with the aid of the programs Sybyl (Tripos, St. Louis, MO) and Sculpt (MDL Information Systems, Inc., San Diego, CA).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-subunits that cannot latch their seatbelts (i.e. hCG-C26A,C110A) or form the tensor disulfide (i.e. hCG-C93A,C100A). While neither was capable of being assembled into stable heterodimers with the native -subunit (6), both were incorporated into heterodimers that are cross-linked by an intersubunit disulfide (Fig. 2). This property formed the basis of our strategy for identifying portions of the subunits likely to contact one another during assembly. We assumed that the ability of a disulfide to "rescue" complexes containing docked subunits that cannot otherwise be assembled into a heterodimer would be proportional to the time that its component cysteines are adjacent. Based on our expectation that contacts between the -subunit and these -subunit analogs during wrapping and threading would be at least somewhat similar to those in the heterodimer, we introduced cysteines into each subunit at sites that were capable of cross-linking the heterodimer. This approach may have caused us to overlook transient contacts that do not lead to assembly, but these were of lesser interest for these studies.
|
carbons and between the C carbons of every residue in the - and -subunits (not shown). We assumed that the most favorable disulfides would be at positions in which the maximum distances between these atoms would be 
6–6.5 and 3.8–4.2 Å, respectively. Based on these considerations, we substituted cysteines for residues in the hCG - and -subunits to produce heterodimers having the potential to form the following disulfides: Q5C-R8C, Q27C-V44C, V76C-V44C, C7S-Y37C, R35C-A35C, Y37C-I33C, and K51C-D99C (Fig. 1). Note that the conversion of Cys7 to serine in -C7S disrupted its ability to form an intra-subunit disulfide between -subunit residues 7 and 31. The resulting free -subunit cysteine (i.e. Cys31) can form a disulfide with the cysteine introduced into the -subunit in place of hCG-Tyr37 (8). As described next, each of these analogs formed cross-linked heterodimers that were stable at pH 2, 37 °C, a condition known to promote hCG dissociation (9). Formation of these disulfides in complexes lacking the abilities to latch their seatbelts or to stabilize the small seatbelt loop indicated that these portions of the subunits were also adjacent at some time during the period in which the subunits were docked to one another.
To learn how interactions between the NH2-terminal portions of the subunits might affect subunit docking, we studied the formation of analogs that were cross-linked by the 5-8 disulfide. Co-expression of -Q5C and hCG-R8C led to the formation of an acid-stable heterodimer, indicating that it contained an intersubunit disulfide. When -Q5C was co-expressed with hCG-R8C,C26A,C110A and hCG-R8C,C93A, C100A, -subunits that cannot latch their seatbelts or form their tensor disulfides, respectively, we observed that 125 and 39% as much heterodimer was formed as that containing hCG-R8C (Fig. 2, left and right). This showed that an intersubunit disulfide had formed between the NH2-terminal ends of the - and -subunits while the subunits were docked with one another, even though the seatbelt latch and tensor disulfides could not be formed. It also revealed that the NH2-terminal portions of both types of -subunits appear to contact the NH2-terminal portions of the -subunit during subunit docking. The relative differences in the amounts of cross-linked docked subunits observed are impossible to interpret, however. They might indicate that contacts between the NH2-terminal ends of the subunits are favored more when the seatbelt is unlatched than when the tensor disulfide is disrupted. Then again, they might also reflect the tendency of seatbelt residue Cys110 to form a disulfide with Cys8 in hCG-R8C,C93A,C100A (18), which would have rendered Cys8 incapable of forming an intersubunit disulfide with Cys5. Consequently, only that fraction of hCG-R8C,C93A,C100A in which the seatbelt is latched to Cys26 would be capable of being cross-linked by a disulfide.
Parts of loop 2 contact loops 1 and 3 in hCG (1, 2) and in hFSH (3). To learn if these portions of the hCG -subunit might participate in docking, we tested the abilities of cysteines that had been introduced in place of loop 1 residue Q27C and loop 3 residue V76C to form intersubunit disulfides with -subunit analogs that contain a cysteine in place of hCG loop 2 residue V44C. Co-expression of -Q27C or -V76C with hCG-V44C led to the formation of cross-linked heterodimers, all of which were acid stable. The 27-44 and 76-44 disulfides each stabilized 28% of the complexes containing the -subunit that cannot latch its seatbelt. They were not as effective as the disulfide in the subunit NH2 termini (Fig. 2, left). This suggested that when the hCG seatbelt is unlatched, the subunits dock in orientations that favor the formation of NH2-terminal contacts relative to those between loop 2 and loops 1 and 3.
The 27-44 disulfide also rescued heterodimers that are unable to form the tensor disulfide. As a result, co-expression of -Q27C and hCG-V44C,C93A,C100A led to 54% as much heterodimer that was formed when -Q27C was expressed with hCG-V44C (Fig. 2, right). This was as good or better than the 5-8 disulfide (i.e. 39%). Remarkably, the 76-44 disulfide between V76C in loop 3 and V44C in loop 2 did not rescue docked complexes containing hCG-V44C,C93A,C100A (Fig. 2, right), an observation that remains puzzling. We did not expect to find that the 27-44 disulfide would rescue docked complexes containing -subunits that cannot form their tensor disulfides better than those that cannot latch their seatbelts. This was because hCG-V44C,C93A,C100A has the potential to latch its seatbelt to Cys44 and, as a consequence, this cysteine would not be capable of being cross-linked to residue Cys27 in -Q27C. In contrast, it is not possible for hCG-C26A,V44C,C110A to latch its seatbelt to Cys44. Thus, the lower ability of the 27-44 disulfide to rescue docked complexes containing an unlatched seatbelt may indicate that before it is latched, the seatbelt may impede the formation of contacts between the loops 1 and 3 with loop 2.
We studied potential contacts between loop 2 and either the cystine knot or loop 1 using analogs that can form intersubunit disulfide bonds between residues 35-35 and 37-33. Both disulfides have been shown to cross-link the subunits in hCG (7). Each rescued docked complexes containing -subunits having latched seatbelts and disrupted tensor disulfides better than complexes containing -subunits with unlatched seatbelts and intact tensor disulfides. Thus, the 35-35 disulfide rescued 66% of the heterodimer containing -R35C and hCG-A35C,C93A,C100A (Fig. 2, right), but only 8.5% of the heterodimer containing -R35C and hCG-C26A,A35C,C110A (Fig. 2, left). The 37-33 disulfide rescued 25% of the heterodimer containing -Y37C and hCG-I33C,C93A,C100A (Fig. 2, right), but only 3.9% of the heterodimer containing -Y37C and hCG-C26A,I33C,C110A (Fig. 2, left). The observations that the 35-35 and 37-33 disulfides formed more readily in analogs of hCG-C93A,C100A, which have latched seatbelts and disrupted tensor disulfides (Fig. 2, right), than in analogs of hCG-C26A,C110A, which have unlatched seatbelts and intact tensor disulfides (Fig. 2, left), suggested that loop 2 residues Arg35 and Tyr37 are more highly constrained during threading than wrapping. This would lead to increased contacts between the subunits, which would be expected to increase the stability of the docked complex.
Efforts to detect other potential contacts that might stabilize the docked complex prior to assembly by the wraparound pathway led us to test the ability of the 31-37 disulfide to secure the heterodimer. This disulfide was found to stabilize an hCG analog formed by co-expressing hCG-Y37C with -C7A (8) and was observed to rescue 15.6% of this material when hCG-C26A,Y37C,C110A was co-expressed with -C7S (Fig. 2, right). We did not repeat these studies with analogs that are unable to form the tensor loop, because disulfides on either side of 31-37 (i.e. 27-44, 35-35, and 37-33) had already been found to rescue a much larger fraction of the heterodimer (Fig. 2, right). Considered together, these studies indicated that contacts near the cystine knots were likely to make a greater contribution to the threading pathway than to the wraparound pathway. The finding that these areas of the subunits are less likely to contact one another when the seatbelt is unlatched may reflect the ability of the unlatched seatbelt to disrupt contacts between the subunits, a topic to be considered later.
We anticipated that the seatbelt would make extensive contacts with residues in loop 2 during the threading pathway and that we would not be able to distinguish these from contacts made during docking. Indeed, we had already found that disulfide bonds appear to form between unpaired cysteines in loop 2 and the tensor disulfides during threading (18). To identify contacts between the seatbelt and the -subunit that might facilitate docking in the wraparound pathway, we took advantage of a disulfide that forms between the seatbelt and loop 2, i.e. 51-99 (7, 8). This disulfide rescued heterodimers containing -K51C with hCG-C26A,D99C,C110A to 25.8% of the level observed when -K51C was co-expressed with hCG-D99C (Fig. 2, left). This suggested that the tensor loop can participate in contacts with loop 2 while the seatbelt is unlatched. As noted later, we anticipate that hydrogen bonds between the backbone atoms of loop 2 residues Val53-Glu56 and seatbelt residues Thr98-Gly101, which include part of the tensor loop, are necessary for efficient completion of assembly by the wraparound pathway. These contacts do not appear to stabilize NH2-terminal portions of loop 2 that contact the -subunit cystine knot, however, because analogs that are unable to latch their seatbelts were not rescued effectively by the 31-37, 35-35, or 37-33 disulfides (Fig. 2, left).
The abilities of intersubunit disulfides to rescue docked complexes containing -subunits that cannot latch their seatbelts or form the tensor disulfide suggest that the manner in which subunits dock is similar but nonidentical during assembly by threading and wrapping mechanisms. The finding that the 5-8 disulfide rescued the formation of both types of heterodimers supported the notion that contacts in the NH2-terminal region are important for the assembly of hCG and other lutropins. This is consistent with the observation that deletion of residues in the NH2 terminus of the -subunit reduced secretion of hCG analogs in which the seatbelt is latched normally (10, 11).
The ability of the 5-8 disulfide to stabilize heterodimers lacking the abilities to latch their seatbelts suggested that contacts between the NH2-terminal portions of the subunits may have a dominant role during assembly by the wraparound pathway. To test this possibility, we monitored the abilities of analogs to form cross-linked heterodimers during assembly that can occur only by the wraparound pathway (Table I). Elimination of residues 1–7 or 2–8 reduced heterodimer secretion substantially (Table I). They also reduced assembly of heterodimers in which the seatbelt is latched to either -subunit residue 37 (Table I, study 1) or to -subunit residue 43 (Table I, study 2). This suggests that NH2-terminal contacts are likely to have a role in subunit docking even though they are not essential for docking.
|
-subunits have only one and two residues in their NH2-terminal ends, respectively. Therefore, the finding that hTSH and hFSH were unable to form heterodimers by a wraparound mechanism (17) may indicate that contacts between the NH2-terminal portions of both subunits are more important for assembly by a wrapping mechanism than for assembly by threading.
Docking Is Readily Reversible, a Phenomenon That May Delay Most hCG Assembly Until the Seatbelt Is Latched—Experiments described next were performed during efforts to identify rate-limiting steps in glycoprotein hormone assembly and thereby learn why most hCG is assembled by a threading mechanism. These studies depended on our abilities to identify the relative rates of subunit docking, threading, and wrapping in the ER. We monitored these processes using the 5-8 disulfide because of its ability to trap and rescue docked complexes in which the seatbelt is unlatched, a phenomenon that was required to detect early stages in the wraparound pathway. During these studies we compared the abilities of hCG to compete with hCG-R8C and hCG-R8C,C26A,C110A for -Q5C for heterodimer formation. We reasoned that if threading or wrapping were not rate-limiting and occurred immediately after the subunits dock, then hCG would compete efficiently with hCG-R8C or hCG-R8C,C26A,C110A for heterodimer formation. Consequently, a substantial fraction of the heterodimer formed in the presence of hCG would lack an intersubunit disulfide and dissociate at pH 2, 37 °C. In contrast, if the subunits docked and undocked faster than the heterodimer became stabilized by threading or wrapping mechanisms, then hCG-R8C and hCG-R8C,C26A,C110A would constitute most of the -subunit in the heterodimer. This is because formation of the 5-8 disulfide cross-link, a reaction unlikely to be reversed quickly, would permit heterodimers containing these -subunit analogs to accumulate at the expense of heterodimers lacking the ability to form this disulfide.
To determine the relative rates of docking, undocking, and assembly, we compared the abilities of hCG to inhibit the formation of acid-stable heterodimers containing -Q5C and hCG-R8C or hCG-R8C,C26A,C110A. Preliminary studies showed that heterodimers containing hCG and -Q5C are acid labile, making them readily distinguished from those that contained -Q5C and hCG-R8C or -Q5C and hCG-R8C,C26A,C110A. For example, 3 days following the co-transfection of COS-7 cells with hCG and -Q5C, we observed that only 1.9 ± 0.6% of the total heterodimer in the medium (i.e. 13.6 ± 1.15 ng/50 µl) survived treatment at pH 2 for 30 min at 37 °C. In contrast, all the heterodimer formed by co-transfecting -Q5C and hCG-R8C was stable after this treatment (Table II, data row 1), showing that it was cross-linked by an intersubunit disulfide, as expected. Most of the heterodimer formed following co-transfection of -Q5C with a mixture of hCG-R8C and hCG was acid stable, indicating that hCG-R8C out competed hCG for assembly of heterodimers with -Q5C (Table II, data row 2). We usually observed that 85% or more of the heterodimer was acid stable when equal amounts of hCG and hCG-R8C were used during transfection and, in some experiments, we did not detect any competition of hCG.
|
competed poorly with hCG-R8C,C26A,C110A, an analog that cannot latch its seatbelt (Table II, data rows 3 and 4). In most studies, heterodimers formed when -Q5C was co-transfected with mixtures of hCG and hCG-R8C,C26A,C110A were at least 90% as stable as those that had been formed when -Q5C was co-transfected only with hCG-R8C,C26A,C110A. This suggested that they contained only small amounts of hCG, a finding that was confirmed by their abilities to be recognized by antibody B111. B111 is an antibody that recognizes a site on the hCG -subunit that includes -subunit residues near Cys26 and Cys110. This antibody can recognize single chain hCG analogs in which both of these latched cysteines are replaced by alanine, albeit not as well as hCG (5). It does not recognize analogs in which the seatbelt is latched to the -subunit (5) or those in which residues near Cys110 are derived from hLH, hFSH, or hTSH. The ability of hCG-R8C,C26A,C110A to out compete hCG shows that docking occurs efficiently before the seatbelt is latched and that docked complexes containing unlatched or latched -subunits are likely to dissociate before being assembled into stable heterodimers by a wraparound mechanism.
The ability of hCG-R8C,C26A,C110A to out compete hCG indicated that formation of the 5-8 disulfide occurs rapidly when the seatbelt is unlatched. Thus, it is conceivable that the ability of hCG-R8C to out compete hCG might reflect its ability to dock with -Q5C before its seatbelt is latched. To test this possibility, we monitored the ability of hCG-R8C to compete with hCG-R8C,C26A for heterodimer formation with -Q5C. We used hCG-R8C,C26A, an analog that cannot latch its seatbelt, in place of hCG-R8C,C26A,C110A in these studies because it combined with -Q5C to form heterodimers that were not recognized as well by antibody B111 (Table II, data rows 3 and 5), a phenomenon that increased the sensitivity of this assay. Heterodimers formed when -Q5C was co-expressed with a mixture of hCG-R8C and hCG-R8C,C26A contained more hCG-R8C than hCG-R8C,C26A (Table II, data row 6). The dominance of hCG-R8C in this assay suggested that analogs in which the seatbelt is latched dock better than those in which the seatbelt is unlatched. Control studies showed that hCG-R8C,C26A competed effectively with hCG-R8C,C26A, C110A, a related analog that also lacks the ability to latch its seatbelt (Table II, data row 7). Considered together, these findings suggested that hCG-R8C and hCG are likely to compete for -Q5C after their seatbelts are latched. Thus, the ability of hCG-R8C to out compete hCG suggests that a significant fraction of hCG dissociates from the -subunit before it can be incorporated into heterodimers by a threading mechanism. The finding that docking is reversible whether the seatbelt is latched or not suggests that threading and wrapping are rate-limiting steps in assembly. Repeated dissociation of the docked complex would provide additional time for the seatbelt to become latched, a phenomenon that would cause most assembly to take place by a threading mechanism as was found (17).
Whereas the Seatbelt Is Not Essential for Docking, It Can Interfere with Docking When It Is Unlatched—To learn if the seatbelt is essential for docking, we tested the abilities of hCG-R8C,
92,C26A, hCGR8C,101, hCG-R8C,93:100D, and hCG-R8C,111 to form cross-linked heterodimers when they were co-expressed with -Q5C. The first of these -subunit analogs lacks the tensor loop, the seatbelt strap, and the COOH terminus. The second analog lacks the seatbelt strap and COOH terminus, and the third analog lacks the tensor loop. All three analogs form cross-linked heterodimers, albeit not as well as hCG-R8C,111, a -subunit analog that lacks the hCG -subunit COOH terminus (Table III). These data showed that the seatbelt is not essential for docking even though it may enhance docking, affect formation of the 5-8 disulfide, or promote the secretion of the cross-linked complex. They also showed that the carboxyl-terminal end of the seatbelt is not essential for docking, a phenomenon that is consistent with the finding that this portion of the -subunit is not required for efficient heterodimer secretion (12).
|
-subunit analogs to disrupt hCG assembly when their seatbelts were unlatched or latched to non-native sites. We anticipated that movements of the unlatched seatbelt might offset contributions made by interactions of the tensor loop with -subunit loop 2 that enabled formation of the 51-99 disulfide (Fig. 2, left). By altering the position of the seatbelt latch site, we anticipated that we would reduce the inhibitory influence of the seatbelt caused by its mobility. We assumed that if the ability of the unlatched seatbelt to inhibit docking was greater than its contribution to docking, these analogs would dock with the -subunit better than the parental analog in which the seatbelt is unlatched, i.e. hCG-C26A. Because these analogs cannot be incorporated into the heterodimer, we expected that their relative abilities to dock with the -subunit could be estimated by their abilities to inhibit hCG secretion. We observed that several -subunit analogs in which the seatbelt was latched to alternate sites inhibited hCG assembly by 40% or more (Table IV). In contrast, hCG-C26A, the -subunit analog lacking the ability to latch its seatbelt, did not inhibit assembly (Table IV, data rows 1 and 2). This supported the notion that analogs having an unlatched seatbelt had reduced abilities to dock with the -subunit, making them less able to inhibit the interactions between the native hCG - and -subunits. The finding that latching the seatbelt to several sites on the -subunit appears to facilitate subunit docking shows that the ability of the unlatched seatbelt to disrupt subunit docking outweighs its positive contributions to subunit interactions. Combined with the finding that docking is reversible, which would provide additional time for the seatbelt to become latched, the finding that the unlatched seatbelt can interfere with docking or destabilize the docked complex would readily account for the dominance of the threading mechanism relative to the wraparound mechanism during hCG assembly (17).
|
-Subunit Loop 2 Appear to Contribute to Latching during the Wraparound Pathway—We have observed that formation of the tensor disulfide appears to facilitate formation of the seatbelt latch disulfide during the folding of the free hCG -subunit (18). The finding that residues of the tensor loop became cross-linked to loop 2 when -K51C was expressed with hCG-C26A,D99C,C110A suggested that these portions of the - and -subunits interact. Analysis of hydrogen bonds in the hCG and hFSH heterodimers showed that several are formed between residues in loop 2 and the tensor loop (Fig. 3, rightmost panel). We anticipated that formation of these bonds and/or other contacts in this region may facilitate latching. To test this possibility, we compared the ability of B111 to recognize hCG analogs that lack the abilities to latch their seatbelts. As noted earlier, this antibody recognizes a conformation of the end of the seatbelt when it is latched to Cys26 but not to other cysteines in either the - or -subunits (5, 17, 18). B111 can also recognize hCG analogs in which the seatbelt has a conformation similar to that in the heterodimer even when the position of the end of the seatbelt is not stabilized by the 26-110 seatbelt latch disulfide. For example, we found that B111 can recognize single chain hCG analogs in which Cys26 and Cys110 are converted to alanine 70% as well as hCG (5). We have also found that B111 can recognize heterodimers that are stabilized by an NH2-terminal disulfide (i.e. between 5-8) such as those that contain -Q5C and hCG-R8C,C26A,C110A (Table V, data row 2). We anticipated that if the contacts between the tensor loop and -subunit loop 2 contributed to the stability of the seatbelt, elimination of the tensor disulfide would adversely affect the ability of B111 to recognize hCG analogs lacking the abilities to latch their seatbelts. Control studies showed that elimination of the tensor disulfide by itself did not affect B111 binding (Table V, data row 3). Removal of both the tensor disulfide and the seatbelt latch disulfide reduced the ability of B111 to recognize the heterodimer (Table II, data row 4). This can be seen by comparing the ability of heterodimers containing hCG-R8C,C26A,C110A and those containing hCG-R8C,C26A,C93A,C100A,C110A to be recognized by B111 (Table V, data rows 2 and 4). These studies suggest that interactions between the seatbelt and the -subunit that occur before the seatbelt is latched contribute to the ability of the seatbelt to find its latch site during assembly that occurs by a wraparound mechanism. This would also explain the reduction in heterodimer formation that occurs when hCG analogs lacking the ability to form the tensor disulfide are forced to latch their seatbelts to a cysteine in the -subunit (Table V, data rows 5–7).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-subunit analogs that cannot latch their seatbelts or that cannot form the tensor disulfide. These analogs should mimic the structures of the -subunit at the time assembly is occurring by wrapping and threading mechanisms, respectively.
We considered two technical difficulties before beginning these studies, the first of which involved the selection of disulfide cross-links. We used only those cross-links that can form in the native heterodimer. While this may have caused us to miss early transient subunit interactions, it should have enabled us to detect contacts that form between the time the subunits dock productively and the time that wrapping and threading are nearly complete. Second, we expected this approach to work best with -subunit analogs in which Cys110 was replaced with alanine. The mobility of the hCG seatbelt prior to docking enables Cys110 to become latched to cysteines that have been added to the -subunit rather than Cys26, its normal site (18). Based on our earlier findings (Ref. 18, Table I), we were particularly concerned that this might affect -subunit constructs containing I33C and A35C, residues needed to observe contacts with loop 2. Because Cys110 was present only in constructs that were used to monitor threading, this problem would have affected our ability to monitor the threading pathway, not the wraparound pathway. Fortunately, we were able to detect these contacts readily despite the potential difficulty of doing so (Fig. 2, right).
Some areas of the subunits appeared to contact one another well in both assembly mechanisms. These include interactions between residues in the - and -subunit NH2 termini and interactions between parts of loops 1/3 and parts of loop 2 (Fig. 2, left and right). Other contact regions, such as those involving loop 2 and residues near the cystine knots, differ significantly in complexes that are thought to be precursors of threading and wrapping. For example, a larger portion of loop 2 appears to contact the -subunit core during threading (i.e. when the seatbelt is latched and the tensor disulfide is disrupted) than during wrapping (i.e. when the tensor disulfide is formed and the seatbelt is unlatched). Consequently, intersubunit disulfides 35-35 and 37-33 rescued complexes containing -subunits that cannot form the tensor disulfide much better than those that cannot latch their seatbelts (Fig. 2). This phenomenon may contribute to the dominance of the threading pathway during assembly and is particularly remarkable because these contacts were more likely to be underestimated in complexes containing -subunits that cannot form the tensor disulfide.
Earlier reports that deletion of the NH2-terminal portion of the hCG -subunit disrupted heterodimer secretion (10, 11) indicated that intersubunit contacts involving this portion of the -subunit might participate in docking. This is supported by the finding that -subunits that cannot latch their seatbelts or form the tensor disulfide were stabilized efficiently by the 5-8 intersubunit disulfide (Fig. 2). Because the follitropin and thyrotropin -subunits lack an NH2-terminal extension, contacts in these regions appear to be limited to lutropins. The requirement for these NH2-terminal residues was lessened in analogs containing portions of the hFSH seatbelt (11). The inability of the hFSH and hTSH -subunits to form these contacts may be responsible for our failure to detect any assembly of hFSH and hTSH by a wraparound mechanism (17).
Intersubunit disulfides between loops 2 and 1 rescued heterodimer formation in both pathways, suggesting that contacts between these regions also contribute to subunit docking during assembly by either mechanism (Fig. 2, left and right). Formation of contacts that involve loop 2 might appear surprising because this is one of the least conserved portions of the -subunit and differs substantially in lutropins, follitropins, and thyrotropins (13). Contacts in these regions of the hCG heterodimer may be stabilized by hydrogen bonds between backbone atoms of residues Cys28-Thr42, Gly30-Thr40, and Cys32-Cys38 (1, 2). These correspond to parts of the -subunit cystine knot and the NH2-terminal end of loop 2. Other hydrogen bonds are likely to involve backbone atoms of Ser34-Gly36, residues in the NH2-terminal end of loop 2 and the -subunit cystine knot. Thus, both cystine knots appear to contribute to docking, particularly when the tensor disulfide is disrupted. In addition, interactions between the hydrophobic side chains of loop 2 residues hCG-Val44 and hFSH-Val38 with a hydrophobic patch on a concave surface of loops 1 and 3 that includes human -subunit residues Phe17, Phe18, Phe74, Val76, and possibly Ile25 and Val70 appear to contribute to the stability of the docked subunits when the tensor disulfide is disrupted or the seatbelt is unlatched (Table VI). Sequence alignments show that hydrophobic residues are found at these positions in most vertebrate -subunits and in most gonadotropin -subunits. We anticipate that contacts in this region of the thyrotropin -subunit involve hydrogen bonds between residues corresponding to hTSH-Asn37 and the -subunit or hydrophobic interactions between the -subunit and residues corresponding to hTSH-Leu40,Phe41,Leu42. Thus, contacts between loops 1/3 and loop 2 would be expected to contribute to the stability of the docked complex despite the differences in loop 2 found in lutropins, follitropins, and thyrotropins.
|
2 is disordered in the free -subunit (14). Following heterodimer assembly, the portion of this loop near the -subunit cystine knot forms two anti-parallel strands (1–3). These are stabilized by residues in the -subunit cystine knot and by parts of the tensor loop (Fig. 3). The abilities of disulfides to stabilize intermediates that serve as prototypes for the wraparound and threading pathways (Fig. 2, left and right) suggest how loop 2 acquires this position. The 35-35 and 37-33 disulfides rescued analogs thought to model threading (Fig. 2, right) better than those thought to model wrapping (Fig. 2, left). This implies that residues near the NH2-terminal end of loop 2 are nearer the -subunit cystine knot during threading than wrapping. During the wrapping pathway (Fig. 2, left), the 51-99 disulfide rescued heterodimer formation better than the 35-35 and 37-33 disulfides (Fig. 2, left). Thus, tensor loop residue Asp99 is nearer loop 2 residue Lys51 during wrapping than cystine knot residue Ala35 is to Arg35 and Ile33 is to Tyr37. These data suggest how assembly is driven in the threading and wraparound pathways. During threading, the -sheet begins forming near the cystine knot and terminates with the formation of hydrogen bonds that stabilize loop 2 near tensor loop residue Cys100 (Fig. 3, lower panels). This constrains Cys100 near Cys93 and enables the tensor disulfide to reform, which completes assembly. This sequence of events is reversed in the wraparound pathway (Fig. 3, upper panels), which begins while the intact tensor disulfide is near Lys51 and terminates with the formation of hydrogen bonds between the NH2-terminal end of loop 2 and the cystine knot. Formation of the latter may require the seatbelt to be latched, albeit not necessarily to Cys26, because the wraparound pathway can be used to assemble heterodimers in which the seatbelt is latched to cysteines in the -subunit (5).
Docking Appears to Be Reversible Before or After the Seatbelt Is Latched, a Phenomenon That Would Favor Assembly by Threading—Previously, we found that the seatbelt is latched in most hCG -subunit molecules before it is incorporated into the heterodimer by a threading mechanism (17). Efforts to determine how turnover of the docked complex affected assembly led us to monitor the abilities of -subunit analogs to compete for the formation of heterodimers that were stabilized by the 5-8 disulfide. These studies showed that the subunits dock readily before the hCG -subunit seatbelt is latched. The poor ability of hCG to compete with hCG-R8C,C26A,C110A showed that formation of the 5-8 disulfide occurred more rapidly than threading or wrapping, processes needed to stabilize heterodimers containing hCG and -Q5C. The inability of the wraparound pathway to compete with the threading pathway for hCG assembly (17) shows that docked complexes having unlatched seatbelts are more likely to d
