Unfolding/Refolding Studies of Smooth Muscle Tropomyosin EVIDENCE FOR A CHAIN EXCHANGE MECHANISM IN THE PREFERENTIAL ASSEMBLY OF THE NATIVE HETERODIMER*

The thermal and the urea-induced unfolding profiles of the coiled-coil a-helix of native and refolded

The thermal and the urea-induced unfolding profiles of the coiled-coil a-helix of native and refolded tropomyosin from chicken gizzard were studied by circular dichroism.
Refolding of tropomyosin at low temperature from a + @ subunits, dissociated by guanidinium chloride, urea, or high temperature, predominantly produced aa + j3@ homodimers in agreement with earlier studies of refolding from guanidinium chloride (Graceffa, P. (1989) Biochemistry 28, 1282-1287). both of which are relatively stable against chain exchange below -25 "C. Equilibrating the homodimer mixture at 37-40 "C for long times, however, produced the native a@ molecule via chain exchange. The equilibria involved indicate that the free energy of formation from subunits of a@ is much less than that of (aa + ,@)/2. In vivo folding of a;B from the two separate (Y and j3 gene products is, therefore, thermodynamically favored over the formation of homodimers and biological factors need not be considered to explain the native preferred a/3 composition. Smooth muscle tropomyosin, purified from chicken gizzard (GTm),' is composed of two subunits, (Y and p, which differ slightly in amino acid sequence (1,2) and are present in about equal amounts (3). Because tropomyosin (Tm) is a twosubunit molecule, assembly in vivo after chain biosynthesis or in vitro after chain dissociation, for a/@ = 1 can, in principle, either produce a 1:l mixture of homodimers ((Y(u$@), * This work was supported by grants from the National Institutes of Health and National Science Foundation. A preliminary report was presented (1989 J. Cell. Biol. 107, 200a all heterodimer (a/3), or a mixture of the three species. Evidence has been presented that GTm isolated under native conditions is a heterodimer (4) but that after refolding by dialysis at low temperature from GdmCl-dissociated chains, a 1:l mixture of homodimers is produced (5). This apparent discrepancy between the native and in vitro refolded composition raises the possibility that the in vivo assembly of the molecule may require biological factors rather than being determined by thermodynamics alone.
Tm is a coiled-coil a-helix that dissociates into subunits in parallel with a major cooperative helix-coil transition (6, 7), whose transition midpoints are characteristic of the species and muscle type as well as the molecular composition (8). Thus, information about Tm composition and subunit dissociation/association may be obtained by monitoring the characteristic unfolding profiles of the homodimers and heterodimer using circular dichroism (CD) techniques. Previous CD studies of the unfolding of GTm showed a single transition for the native molecule (4,8,9).
In this work we confirm that assembly of GTm at low temperature from unfolded (Y and p subunits in denaturants produces (~a! + p/3 (5) and show that rapid refolding at low temperature from the separated chains at high temperature also predominantly produces homodimers. In contrast, refolding by incubating at temperatures close to physiological produces the native (~/3 heterodimer. Thus, a/3 is the preferred species in vitro, and biological factors are not required to explain its preference in vivo. CD studies also provided information about the mechanism of assembly by showing: (i) that homodimers are initially formed from dissociated subunits in a rapid process; (ii) that (~/3 is produced from the mixture of homodimers between 37-40 "C via a chain exchange mechanism, during times that allow equilibrium to be reached.

RESULTS
In these studies, three samples of GTm were used: GTm ((YP),~ GTm refolded from 6 M GdmCl (CY(Y + BP), and an enriched CYLY sample (80% o(o(, 20% BP). SDS-polyacrylamide gel electrophoresis of the samples before and after crosslinking with aromatic disulfides at low concentration (to optimize cross-linking over the competitive blocking reaction) (12) verified their composition ( Fig. 1). It is seen that native GTm reacts with the disulfide reagents but is not cross-linked. This is because the a-chain contains a Cys at position 190 and the P-chain contains a Cys at position 36 (1, 2) and in the c@ molecule where the chains are parallel and in register, the 2 Cys are too far apart to be disulfide cross-linked. In contrast,

Refolded
Tm Native the refolded cup GTm is appreciably cross-linked and is composed of an equal mixture of ~yol and pp (5). The CD spectrum of native gizzard Tm is typical of an o(helix with negative ellipticity peaks at 222, 208, and a positive peak at 193 nm. The value of [O]222 nm -3.6 X lo* deg cm21 dmol at 15 "C for the mean residue ellipticity indicates >95% a-helix. In low salt solutions, the urea-induced unfolding of native GTm took place in a single cooperative transition with a midpoint at 0.85 M urea; refolded GTm unfolded in two approximately equal transitions at 0.4 and 0.85 M urea ( Fig.  2A). The first transition of refolded GTm is identified as being due to the unfolding of (YO(, since the major transition of enriched (Y(Y occurred at ~0.4 M urea. The second transition of refolded GTm must therefore be the unfolding of pp. Native GTm (a@) unfolds in a single transition with a midpoint similar to that of the second transition of refolded GTm (due to @) indicating that & has a stability similar to that of /3/3 in agreement with thermal unfolding studies at high salt (see below and Ref. 4). These CD data, which show that LYLY can be distinguished from cup and p/3 by its unfolding profiles in low salt solutions, verifies that refolding from denaturants at low temperature predominantly produces homodimers. Similar CD results were obtained on these samples in studies of thermal unfolding in low salt solutions (Fig. 2B). By comparison with the unfolding of native rabbit skeletal Tm which is known to be a mixture of (Y(Y and o/3 (10,12), these data show that the stability of the various species increases in the order: (YOI gizzard Tm < @3 skeletal Tm = cup gizzard Tm, @@ gizzard Tm < CY(Y skeletal Tm. From these data alone it is difficult to determine the order of stability of o/3 GTm and p/3 GTm. This order of stability at low salt is in approximate agreement with the number of optimum ionic interactions of charged residues across the chains (2). In contrast to refolding at low temperature where homodimers are preferentially produced, when native GTm was refolded at temperatures in the physiological region, c@ was preferentially produced. Thus, the thermal unfolding profile in low salt solutions of (Y + @ that had been refolded by dialysis at 35-38 "C showed only one transition similar to native GTm (Fig. 3, R 35-38 "C). In a parallel experiment, refolding by dialysis at 5 "C produced an appreciable amount of homodimers as evidenced by two transitions. Rapidly cooling (~2 min) of separated (Y + /3 chains from 50 to 20 "C in low salt solutions, also resulted in considerable homodimer formation (Fig. 3, R' 20 "C). Similar low salt unfolding studies performed on samples refolded by dilution from urea showed preferential formation of homodimers at 0 "C and heterodimer at 37 "C. These data indicate that refolding at physiological temperature results in the formation of o$l in contrast to refolding at lower temperatures which favors homodimers.
Further insight into the factors that determine the composition of the dimer was obtained by thermal unfolding/refolding studies of solutions containing salt concentrations close to physiological. The thermal unfolding profile of refolded GTm ((wcu + &3) between ionic strengths of 0.05 and 1.0 showed two transitions as observed at low salt concentrations (Fig.  2), but the first transition was less prominent and appeared truncated (Fig. 4A, curue 1) as compared with the corresponding low salt transition, in agreement with the data of Graceffa (5). The first part of the first transition coincided with the unfolding of enriched (YCY by heating up to about 39 "C, but after about 15-20% helix loss, the rest of the curve was very similar to the profile of native a@ GTm (Fig. 4, curue 2). In view of the above observation that a/3 is preferentially formed  during equilibration in the 35-40 "C temperature range in physiological buffer, it appears that the truncated first transition is due to two processes, CYLY = 2~y and 2cu + /3p = 2cu& i.e. chain exchange which converts homodimers into heterodimers, Q(Y + &3 = 2@. This can explain the smaller decrease in helix content observed in the first transition (the truncation), since partially unfolded (TLY is converted to fully folded CYP in this temperature range. The second transition must then be due to I$ = CY + p. The shape of the first transition would then depend on the relative rates of chain exchange uersuS heating, e.g. a faster heating rate would favor more ~yty unfolding before chain exchange. To obtain information on the rates involved in refolding and chain exchange, the temperature was quickly dropped to 37 "C from a higher temperature where the chains were separated, and the temperature and ellipticity were monitored with time. It was found that the ellipticity changed in two steps; an initial rapid decrease in parallel with the temperature drop and a very slow exponential decrease (increase in helix content) with a tl,$ of about 500 s after the temperature reached 37 "C (in about 60 s) (data not shown). After incubating at 37 "C for about 1200 S, the sample was cooled to 20 "C, to quench the exchange process. Its unfolding profile indicated that it was converted to (Y@ (Fig. 4A, curue 2). More definitive evidence for refolding to a@ in two steps was obtained by a repeat of the kinetic refolding studies at 40 "C, where the difference between the unfolding curves of homodimers and heterodimer is greatest (in Fig. 4A, compare curves 1 and 2). It is clearly seen that the negative ellipticity increased in two processes; a fast process which follows the temperature drop and a slow process with t Ih -500 s (Fig. 4B). It appears that the fast process is the production of homodimers in view of the smaller value of negative ellipticity initially obtained since DALY is ~50% unfolded at 40 "C. The slow process is chain exchange resulting in the production of a& Evidence for the fast process producing homodimers was also obtained by monitoring the unfolding profile of a sample which was cooled from 55 to 20 "C rapidly (0.3 "C/s). Although there was incomplete refolding during this fast temperature drop, it can be seen that an appreciable amount of homodimer was formed as evidenced by the presence of the truncated first transition (Fig. 4A, curve 3).

DISCUSSION
These data indicate that assembly of LY + /3 subunits by rapid cooling forms homodimers. This is most simply explained by formation of &3 first then CY(Y since /3/3 appears to be most stable. In the physiological temperature region, heterodimers are subsequently formed by slower chain exchange. If the temperature is too low for chain exchange to readily take place (below ~25-30 "C), homodimers remain kinetically trapped. The preference for homodimer formation by refolding by dialysis or dilution at low temperature is similarly explained. Even with slow dialysis where equilibrium would appear to be attained, homodimers will preferentially be produced due to the very slow rate of chain exchange below room temperature. Cooling a mixture of separated chains from high temperatures slowly enough, however, would be expected to produce &?. This explains the observation that the CD unfolding/refolding curves for (Y@ were reversible because the heating/cooling rate of 0.4 "C/min was slow enough to allow chain exchange to take place before the temperature was lowered to values where the exchange rate was too slow. In contrast to physiological salt solutions, in low salt solutions the CD unfolding profiles of the mixture of homodimers appeared normal, i.e. each homodimer unfolded independently without chain exchange. This may be expected in view of the unshielded charge repulsion of the negatively charged molecules in the neutral pH region resulting in a slower rate of chain exchange. The observation that heterodimers are obtained in vitro from a mixture of homodimers by equilibrating at physiological temperature indicates that thermodynamics can explain the native a/3 composition of GTm. The tendency for heterodimer preference over homodimers can be determined from a thermodynamic analysis of the equilibrium, 2a@ = (Y(Y + fib. The equilibrium constant, K,, for this exchange reaction is given by K, = (K&2/(K&&, where rC,,, K+ and I& are individual dissociation constants. At a given temperature, ap will be preferred over a mixture of LY(Y and /3/3 if Kaasp >> (Km&" (for a random mixture, K,,Kpu = (2K,,#).
The equilibrium dissociation constants are determined by the change in standard free energies upon formation of dimers from separated chains, AG. An alternate formulation is, therefore, that heterodimers will be preferentially formed if AG@ << I/Z (AGcrcr + AG&3) + RT In 2. Since a/3 and /!?fi have about the same stability, it appears that (~8 preferentially forms in order to minimize the total free energy at temperatures where (YLY is relatively unstable and thus would be partially dissociated. The heterodimer of the leucine zipper coiled-coil from FOS and Jun is also favored thermodynamically in an analogous manner (13).
In a recent study with frog skeletal Tm, chain exchange was also shown to be important in determining the native afl composition (14). In the case of rabbit skeletal Tm, refolding from a 50~50 mixture of (Y and p chains by cooling from high temperature gave a random composition (15) with a tendency toward heterodimer formation when refolded by dilution at physiological temperature (16). In view of the findings presented above it is possible that skeletal Tm heterodimers may be produced if incubation times at physiological temperature are long enough to allow chain exchange to reach equilibrium.
Several studies suggest functional differences of Tm which depend on the a/p ratio and therefore on the dimer composition for a particular muscle. The a//3 ratio of Tm from skeletal muscle has been shown to change during development (17,18) and in differentiating cells in uitro (19). There is an assembly preference of the Tm heterodimer in several rabbit skeletal muscles where a/p = 1 (20) as well as for gizzard smooth muscle (4). In many other smooth muscles the Tm cr/fl ratio was found to be ~1 (21) further suggesting the possibility of heterodimer preference. The only reported property that appears to differ from the homodimers compared with the heterodimer in addition to the different stabilities observed in this work, is the low salt polymerizibility, reflecting greater end-to-end interactions for the heterodimer (5).
In our early conformational studies with GTm (9), several observations were made which can now be understood in light of this work and the studies of Sanders et al. (4). In contrast to reactions of skeletal Tm with Nbsn where disulfide crosslinks were efficiently produced (12), we found that the native GTm molecule reacted with Nbsp at 25 "C without the production of disulfide cross-links but that Cu'+-catalyzed air oxidation at 35 "C during long incubations produced disulfidecross-linked homodimers (9). This latter observation was in agreement with earlier studies of Strasburg and Greaser (22). The lack of ability to disulfide-cross-link the native c@ with Nbs, is understandable, because the Cys are located at Cys-36 in the p-chain and at Cys-190 in the a-chain, i.e. too far apart in the molecule to be cross-linked and only Nbsn-blocked SH-groups were produced. At 35 "C, however, sufficient chain exchange may produce a small equilibrium concentration of homodimers. With air oxidation, there is no competing reaction with SH-groups as there is for oxidation by aromatic disulfides, and over long times, homodimers which have Cys in proximity could be cross-linked, removing them from the chain-exchange equilibrium and allowing further cross-linking. A similar explanation was proposed earlier (4).