The chaperonin GroEL binds nonnative substrate protein in the hydrophobic central cavity of an open ring. ATP and GroES binding to the same ring converts this cavity into an encapsulated, hydrophilic chamber that mediates productive folding. A 'rack' mechanism of initial protein unfolding proposes that, upon GroES and ATP binding, the polypeptide is stretched between the binding sites on the twisting apical domains of GroEL before complete release into the chamber. Here, the structure of malate dehydrogenase (MDH) subunit during folding is monitored by deuterium exchange, peptic fragment production and mass spectrometry. When bound to GroEL, MDH exhibits a core of partially protected secondary structure that is only modestly deprotected upon ATP and GroES binding. Moreover, deprotection is broadly distributed throughout MDH, suggesting that it results from breaking hydrogen bonds between MDH and the cavity wall or global destabilization, as opposed to forced mechanical unfolding.
Structural and mechanistic studies of the past few years have provided a general understanding of how the GroEL−GroES chaperonin system cycles in binding nonnative polypeptides and assisting folding1. However, much more is known about the machinery itself and its conformational states than about the substrate polypeptides and how their conformations are affected during the reaction cycle. For example, nonnative polypeptides are bound in the open ring of an asymmetric GroEL−GroES−ADP complex2,
3via hydrophobic contacts between substrate protein and multiple apical domains facing the central cavity1,
4, but the exact consequences of such binding on polypeptide conformation remain unknown. A partial unfolding action has been suggested, which could remove a polypeptide from a kinetically trapped state. This unfolding could operate either through catalyzed unfolding, observed in one case for a small protein incubated with GroEL5, or through a thermodynamic partitioning mechanism in which binding of less folded states is favored, shifting an equilibrium ensemble by mass action toward less folded forms6,
7. This latter partitioning action has been observed for several polypeptides7,
8,
9, but has not yet been demonstrated for a stringent — that is, GroEL−GroES−ATP-dependent — substrate protein.
Upon ATP/GroES binding to a polypeptide-bound ring, the substrate protein is ejected within a second into the GroES-encapsulated, now hydrophilic cis chamber where folding commences10,
11. A recent study has suggested that during the phase of ATP/GroES binding prior to complete ejection into the chamber, the substrate polypeptide is stretched on the elevating and twisting apical domains in a so-called 'rack' mechanism12. Thus, the energy of ATP/GroES binding would be directly used to support a forced mechanical unfolding of a kinetically trapped state.
The subsequent events of folding inside cis complexes are also not well understood. Why do only a few percent of such stringent substrate molecules, such as ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco) or mitochondrial malate dehydrogenase (MDH), reach native form in any given round in a cis chamber? What are the differences in conformation between those molecules that proceed productively and the majority that do not? How structured are the nonnative states that are released at the end of each cycle into the bulk solution? To begin to address some of these questions, we have carried out a pulsed deuterium exchange study of a chaperonin reaction involving the stringent substrate MDH13,
14,
15. At various times during folding of MDH inside the GroEL−GroES cavity, analysis of the accessibility of the backbone amide protons of MDH to hydrogen-deuterium exchange allowed determination of the presence of secondary structural elements. Analysis of both intact, exchanged MDH and peptides derived from it allowed identification of specific structured regions of the protein in binary complexes with chaperonin and during the course of folding inside ternary MDH−GroEL−GroES complexes. These data allowed conclusions to be drawn concerning the conformational states of MDH during a chaperonin-mediated folding cycle.
Results The single ring version of GroEL, SR1, provided efficient formation of folding-active cis ternary complexes, because GroES added to MDH−SR1 binary complexes obligately and quantitatively binds to the polypeptide-containing ring11 (Fig. 1). This contrasts with wild type, double ring GroEL, where GroES associates randomly with the rings of a polypeptide−GroEL complex to produce an 50:50 mixture of active cis and nonproductive trans complexes16. Moreover, the folding-active MDH−SR1−GroES ternary complexes formed upon addition of ATP/GroES are stable, because complex dissociation is prevented by the absence of the trans ring, which normally provides an allosteric signal for GroES release triggered by ATP binding to it10,
11,
16,
17. The substrate polypeptide, ejected within a second of ATP/GroES binding into the cis folding chamber where it commences folding, remains encapsulated inside SR1−GroES complexes, as compared with repeatedly departing and rebinding wild type GroEL−GroES complexes10,
11,
16. The kinetics of substrate refolding inside SR1−GroES, including that of the MDH subunit, have been shown to closely parallel those of the cycling wild type reaction11. Thus, SR1 behaves identically to wild type GroEL in forming stable binary complexes with nonnative substrate, binding GroES and ATP to the substrate-bound ring to form a cis complex and mediating cis folding for at least a time corresponding to one round of folding with wild type GroEL (t1/29 s). At later times, the SR1−GroES folding reaction proceeds inside an SR1−GroES complex that has undergone a single round of ATP hydrolysis to produce a stable cis SR1−GroES−ADP complex10. This complex likely still resembles a wild type complex, given its identical productivity, but any conformational changes associated with substrate release and rebinding to wild type GroEL do not occur.
Figure 1. Experimental scheme for examining secondary structure of MDH subunit during its chaperonin-mediated folding.
Binary complexes of MDH with the single ring version of GroEL (SR1) (left) and folding active ternary complexes of MDH−SR1−GroES (right) were examined by continuous deuterium exchange (binary complexes) or by pulsed deuterium exchange (both binary and ternary complexes). At the various time points of the chaperonin reaction, pulsed exchange with D2O was carried out, and either intact MDH or peptic peptides were examined by HPLC MS after quenching at pH 2.5 and 4 °C.
Deuterium exchange of MDH in a binary complex The experimental protocol for observing pulsed deuterium exchange of MDH in a binary complex with SR1 and during a folding reaction is diagrammed in Fig. 1. After each sample was exposed to 93% D2O for 1 s at pD 8, exchange was quenched by addition of acid. MDH was rapidly separated from SR1 on a C4 column (in protic buffer) and examined by mass spectrometry (MS). As a fully unfolded reference sample, MDH was unfolded in urea and equilibrated in 93% D2O; its mass spectrum showed that 243 deuterons were present. The difference between this value and the theoretical one of 292 (314 residues, 21 prolines) is due to labeling in 93% D2O and exchange during HPLC. The number of deuterons observed by MS in the various complexes was adjusted by the ratio of these values (292/243) before reporting them in the figures and tables. The mass spectrum of intact MDH exposed to a 1 s pulse of D2O while bound to SR1 is shown in Fig. 2 (top trace, 'Binary complex'); 247 deuterons (adjusted value) were incoporated into MDH, indicating protection of 45 amide protons (Table 1, line 2).
Figure 2. Mass spectra showing extent of deuterium incorporation into MDH after various times of SR1-assisted folding, measured after pulse-labeling in D2O for 1 s.
Deuterium incorporation is reflected by the increase of molecular mass. Times of 1 s−100 min refer to the time after addition of ATP/GroES to the MDH−SR1 binary complex when the pulse-labeling was carried out. Vertical dashed lines indicate the centers of reference peaks corresponding to unlabeled MDH (left), labeled native MDH (middle) and equilibrium labeled, fully unfolded reference MDH (right). Note that the same extent of labeling was obtained for the fully unfolded reference with pulse labeling.
Four most protected peptides are in a Rossmann fold To identify regions of MDH that were protected in the initial binary complex, a continuous-exchange experiment was performed. The MDH−SR1 binary complex was exposed to 93% D2O for varying times (from 1 to 2,100 s), followed by acid quenching, separation of MDH, preparation of peptic peptides from the MDH (at pH 2.5) and HPLC MS. Analysis of these fragments at 1 s of exchange showed that the MDH peptides from 50% of the backbone had deuterium levels <90%, indicating significant protection from exchange (note that at pH 8.0, fully unfolded MDH would be 99% exchanged in 0.5 s; ref. 18). The four most protected peptides exhibited levels of 60−80% deuteration at 1 s and 80−90% at 10 s (Fig. 3a, circles), and map to three -helices and two adjacent -strands that are part of the second Rossmann fold in the N-terminal NAD-binding domain of native MDH (Fig. 3b). The protected regions may comprise a native-like 'core' structure significantly populated in GroEL-bound MDH molecules.
Figure 3. Four regions of MDH in the MDH−SR1 binary complex that exhibit the greatest degree of protection from deuterium exchange.
a, Binary complexes of MDH−SR1 were exposed to D2O for various labeling times, the mixtures were acid quenched, MDH was recovered and peptic peptides were prepared and analyzed by HPLC MS. The level of deuterium incorporation in the four most protected derived peptides was expressed as a percentage of that exhibited by the same peptides derived from urea-unfolded, equilibrium-labeled MDH (circles). The behavior of these same regions during exchange of native MDH is shown for comparison (triangles). The dotted line is at the 90% level; at values below this, protection from exchange is considered to be significant (see text). b, Position of the four most protected regions (indicated in blue) in a ribbon model of the MDH subunit (PDB 1MLD)32 localized within the N-terminal NAD binding domain. The domain is composed of a six-stranded -sheet (A−F), flanked on both aspects by -helices. Protection was observed in D, the -helix following it, E (two middle residues of which are not represented in peptic peptides and are therefore left uncolored) and in the -helices preceding and following F. Model illustrated using Molscript33.
Modest deprotection upon ATP/GroES binding Within 1 s after addition of ATP/GroES to SR1−substrate binary complexes, the substrate is ejected into the central cavity and commences folding. (For example, the anisotropy of intrinsic tryptophans of Rubisco begins to rise by 1 s, after an initial drop11.) Here, 1 s pulsed-exchange experiments (Fig. 1) were carried out both on the binary complex and on the complex at various times after ATP/GroES addition (Fig. 2; Table 1). As indicated above, the mass spectrum of intact MDH in the binary complex after a pulse of D2O exhibited a net protection of 45 protons (247 deuterons incorporated). At 1 s after ATP/GroES addition, 32 protons were protected (260 deuterons incorporated). Thus, there was a net deprotection of 13 protons (4) upon the addition of ATP/GroES. To examine the nature of this deprotection, peptic fragments were prepared from MDH, both from pulse-exchanged MDH−SR1 binary complex and from pulse-exchanged ternary complex 1 s after ATP/GroES addition. In particular, the four peptides that exhibited the greatest protection in the binary complex were examined (Fig. 4). Peptide 71−89 exhibited essentially no change in its mass/charge. Peptides 101−112 and 149−156 exhibited some increase in the intensity of their higher mass/charge (right-hand) shoulders, reflecting greater exchange after ATP/GroES addition. However, neither peptide exhibited a mass/charge ratio equivalent to that of the fully unfolded reference, indicating that there had been only a partial deprotection. The fourth peptide, 115−129, showed a shift of its entire peak partway toward the position of the unfolded reference (Fig. 4), also indicating partial deprotection after ATP/GroES addition. In terms of deuterium content (Table 2), there was essentially no change in peptide 71−89, with five nonexchanged amide protons out of 15 in both the binary complex and 1 s after ATP/GroES addition. In peptides 101−112 and 149−156, approximately one additional deuteron was incorporated, with five and one of the total remaining nonexchanged, respectively, after ATP/GroES addition (Table 2). Finally, in peptide 115−129, there was also the exchange of one additional deuteron, leaving one proton nonexchanged. Thus, whereas peptide 71−89 remained protected to the same extent after ATP/GroES addition as in the binary complex, the other three peptides appeared to have been partly deprotected upon ATP/GroES binding.
Figure 4. Retention of exchange protection in the four regions of MDH exhibiting greatest protection in the MDH−SR1 binary complex upon addition of ATP/GroES.
Pulse-labeling was performed on MDH−SR1 binary complexes and ternary complexes 1 s after addition of ATP/GroES. The reactions were acid-quenched, MDH was recovered and peptic peptides were prepared and analyzed by HPLC MS. Mass/charge profiles of the peptides are displayed, along with those of the same peptides prepared from urea-unfolded equilibrium-labeled MDH ('unfolded reference') and near-native MDH monomer prepared by 1 s pulse-labeling 100 min after ATP/GroES addition. Note that peptide 101−112 appears to show a bimodal distribution between two populations, one unprotected and one partially protected, in both the 'Binary complex' and '1 sec' traces. The unprotected population represents about one-third of the total in the binary complex, increasing to about one-half at 1 s after ATP/GroES addition. However, the degree of protection of the partially protected population is largely unchanged at 1 s.
Deprotection is broadly distributed The sum of three additional exchanged amide protons observed in the four peptides described above 1 s after ATP/GroES addition (Table 2) does not account for the total of 13 additional exchanged amide protons observed in intact MDH at this time (Fig. 2; Table 1). Indeed, the additional deprotection of MDH appears to be broadly distributed, as observed from comparison of the deuterium content of all of the individual peptides from MDH at 1 s after ATP/GroES addition to the binary complex (Table 3, third column). Individual peptides spread throughout the primary sequence exhibited small net deprotection. In particular, five peptides were subject to deprotection amounting to between 0.4 and 0.9 deuterons, and six peptides exhibited a change of 1.0−2.0 deuterons. Thus, the extent of deprotection observed in three of the four most protected regions of the binary complex (Table 2) was similar to the degree of deprotection observed in other portions of the protein. Upon ATP/GroES addition there is either no deprotection (peptide 71−89) or only partial deprotection (peptides 101−112, 115−129 and 149−156) of what had been the most protected regions of MDH in the binary complex. In addition, there is a modest degree of deprotection spread broadly throughout the rest of the protein.
Table 3. Changes in deuterium levels of MDH peptic fragments
Structure of nonnative species at later times MDH−SR1−GroES ternary complexes were also examined by pulse labeling (1 s) at longer times of ATP/GroES-driven folding. Surprisingly, when exchange of MDH−SR1−GroES ternary complexes was examined at 3 and 9 s, we observed the same level of protection in MDH as that at 1 s. GroES and polypeptide are ejected from wild type GroEL after 9 s, corresponding to the half life of a wild type cis complex10,
11,
19. Because only a few percent of MDH molecules (below the level of detection here) reach native form in any given cycle15,
19 (Fig. 2; Table 1), the vast majority of molecules remain nonnative. The protection data indicate that these species do not have additional stable secondary structure. To date, it has been unclear how much structure nonnative forms acquire during a single folding/release cycle. Given their lack of additional amide proton protection in this experiment, most have apparently not formed any new stable secondary structure. Why these discharged MDH nonnative forms are so aggregation-prone upon leaving the cavity is clear: their unstable structures likely allow exposure of hydrophobic surfaces that dispose to multimeric association.
A near-native species lacks protection only at the dimer interface By 81 s (corresponding to 5−10 cycles of folding with wild type GroEL−GroES), analysis of intact MDH retained inside SR1−GroES revealed both broadening of the major peak observed at 9 s, reflecting production of one or more somewhat more protected intermediate species (Fig. 2), and the presence of a new peak with a molecular mass near to that of the native MDH subunit (Fig. 2, Table 1). At subsequent times, this latter, highly protected species became progressively more populated at the expense of the less protected ones (Fig. 2). Its rate of appearance corresponds well with the production of catalytically active MDH upon forced release of GroES from SR1 by brief exposure to 4 °C (Table 1, right column). Comparison of the peptic peptides from MDH at 100 min after addition of ATP/GroES (Fig. 2) with those from a peptic digest of similarly exchanged and processed native MDH dimer shows that this species with near-native protection indeed corresponds to a near-native form of MDH. There are two regions of the 100 min species that exhibited significantly less than native protection (Table 3, right column), amino acids 42−70 and 213−227, both of which map to the dimer interface (Fig. 5). Although the rest of the MDH monomer folding inside SR1−GroES reaches a stable native-like secondary structure, this portion of the dimer interface, composed of -helical structure, may not achieve stable structure until interaction occurs with the partner subunit. Alternatively, this region may fold to the same extent in the SR1−GroES−sequestered monomer as in the native dimer, but the lower deuterium level in the native dimer could result from shielding of amide groups in the subunit interface.
Figure 5. MDH refolded inside SR1−GroES lacks native protection from exchange only at the dimer interface.
A model of native MDH dimer (1MLD)32 shows the two regions lacking native protection in the left subunit (light blue) (see Table 2). The right subunit is colored yellow. The structural elements in the left subunit forming the contact sites are labeled. A corresponding set is present in the right subunit, related by two-fold symmetry. See ref. 32 for nomenclature and discussion of the interface.
Regarding production of the near-native species inside SR1−GroES, later events of folding inside the SR1−GroES complex appear likely to involve a cooperative conversion of the early, weakly protected species and later-appearing putative intermediate one(s) (observed at 81 s as a lower mass shoulder to the high mass peak) to the near-native species. However, the nature, population and productivity of the putative intermediates is difficult to assess. Whether any of these species are obligate in formation of the native state and whether these species are produced during a wild type GroEL reaction is unknown. For example, the shoulder species observed by 81 s are not appreciably populated at 9 s in the SR1−GroES reaction, corresponding to the half-time for release from a wild type GroEL−GroES reaction.
Discussion Observations here of the 'stringent' substrate protein, MDH, whose productive folding is absolutely dependent on GroEL, GroES and ATP, show that it exhibits regional, although unstable, secondary structure while bound to GroEL. This structure is partially but not completely deprotected upon binding ATP/GroES. These weakly structured conformations are the dominant, nonnative species present in folding-active cis complexes, giving rise in each round of folding to 2−3% of molecules that reach near-native conformation and lack native amide proton protection only at their dimer interfaces. The normal dissociation of folding-active cis complexes releases both the poorly structured species, which are prone to multimolecular interactions but are also likely to efficiently rebound by GroEL, and the near-native forms, which are no longer recognizable by GroEL — that is, are 'committed' — and can efficiently dimerize to form the native active enzyme.
Putative 'core' structure in MDH bound to GroEL The unstable secondary structure of MDH in binary complexes with GroEL agrees well with previous exchange studies of binary complexes with such proteins as cyclophilin20, -lactalbumin21, human DHFR22,
23 and chemically denatured -lactamase24 where only low levels of protection were also observed. However, the sensitivity of the present experiments and their ability to resolve at the level of derived peptides have allowed detection of specific, partially protected regions of MDH in the binary complex that correspond in the native state to a Rossmann fold in the NAD-binding domain. Although this region is unstable in the binary complex, with protection factors estimated at 100, such a structure could comprise a 'core' already present in the bound polypeptide. Presumably such a core would be present in the nonnative MDH intermediate before it becomes bound, although we have no evidence at present on this point. The low levels of protection in this putative core resemble those observed in a DHFR−GroEL binary complex exchange study, which mapped protection in DHFR to a region corresponding in the native state to the central parallel -sheet23.
Forced mechanical unfolding versus breaking bonds with the cavity wall The time of 1 s after ATP/GroES addition corresponds to the time by which the polypeptide is ejected off the apical domains into the encapsulated cis space underneath bound GroES10,
11. Here we observed that upon ATP/GroES addition, intact MDH became deprotected relative to its level of protection in the binary complex. This deprotection was partial, amounting to 13 more deuterons that were exchanged of the 45 protected in the binary complex. Mapping with peptic peptides indicated that the deprotection was distributed broadly across the protein, with the most protected, putative core region also exhibiting a mild degree of deprotection that amounted to three additional deuterons out of 10 available sites. Thus, the putative core structure may not be substantially perturbed through the action of ATP/GroES binding. Instead, there was a broad distribution of a small amount of deprotection. This seems explained best as the result of simple release of MDH from the GroEL cavity wall, which could involve direct breaking of hydrogen bonds formed between bound MDH and multiple GroEL apical domains in the binary complex. Although hydrophobic contacts between side chains of bound polypeptide and the apical binding surface are likely to be predominant contributors to polypeptide binding, hydrogen bonds between main chain amide protons of bound peptide and the GroEL apical surface have been observed in two different crystal structures in which short peptide segments are bound in extended conformation in the groove between apical -helices H and I. In the study of a seven-residue 'tag' peptide bound in an extended state in this groove in a structure of the apical domain alone, three such hydrogen bonds were observed25. In a similar study of a dodecameric peptide added exogenously both to isolated apical domains and to a GroEL tetradecamer, in which six residues of a -hairpin bound in this groove, one such hydrogen bond was observed26. It thus seems possible that the deprotection here of at least a portion, if not all, of the 13 deuterons upon ATP/GroES binding is a function of breaking such contacts. Alternatively, ejection of polypeptide off the cavity wall could indirectly produce deprotection by the exposure of amide protons that had been shielded from exchange by virtue of association of local structure with the cavity wall, or by a general destabilization of the polypeptide by its increased freedom of motion once released from the apical binding sites.
Although the present data do not exclude a rack mechanism as an explanation for deprotection of MDH upon ATP/GroES binding, it would have to be a selective action not affecting certain regions with demonstrable protected structure. For example, the region of MDH between residues 71 and 89 was not at all deprotected (Fig. 4; Table 2). Thus, the occurrence of broadly but unevenly distributed deprotection seems as consistent with breakage of hydrogen bonds with the cavity wall as it does with forced mechanical breakage of internal hydrogen bonds.
Although a rack mechanism of forced unfolding seems uncertain for MDH, this does not exclude that such behavior could be exerted on another stringent substrate, Rubisco, whose study evoked the model originally12. In that study examining tritium exchange of Rubisco, 12 tritiums were highly protected, both in a metastable intermediate form of Rubisco in solution and in the binary complex with GroEL, with such protection surviving times of manipulation in protic solvent of at least 30 min. Thus, this set of tritiums is two or more orders of magnitude more protected than the core protons of MDH studied here. Upon addition of ATP/GroES, deprotection of 10 of the 12 tritiums was observed at the earliest time tested (5 s). Yet the nature of this deprotection seems somewhat unclear, because the GroEL double ring was used instead of SR1. Instead of obligate and quantitative cis complex formation, only 50% of Rubisco molecules would have become encapsulated in cis complexes in the time of one turnover, because GroES would bind approximately randomly to the polypeptide-bound versus opposite GroEL ring. Why >50% of the tritiums became deprotected thus seems puzzling; reexamining Rubisco in the setting of SR1 using exchange methods similar to those applied both in that study and here may be of value.
How is MDH removed from kinetic traps? If MDH is not subject to a forced mechanical unfolding upon ATP/GroES binding and is to be rescued from kinetic traps, there must be another point in the chaperonin reaction cycle where such action is exerted. By elimination, this action must occur at the initial binding of nonnative MDH to an open GroEL ring. It could occur through a thermodynamic partitioning mechanism, in which GroEL binds less folded states with greater affinity and shifts an ensemble of nonnative states in equilibrium with each other toward less folded conformations6,
7,
8,
9. Although this type of behavior has been reported for a number of proteins, it has not been directly demonstrated for a stringent substrate. Nevertheless, in the case of MDH, an earlier kinetic study suggested that binding to GroEL shifts an ensemble of low order reversible aggregates toward the state of GroEL-bound monomer and away from conversion to a higher order, irreversibly aggregated state15. An alternative mechanism for removal of nonnative MDH from kinetically trapped states involves catalyzed unfolding5 — for example, exerted by association with multiple GroEL apical domains of an open ring4. To date, such behavior has only been reported for the small protein, barnase, which became globally exchanged during incubation with GroEL5. The observation here of at least partial protection of MDH in a binary complex argues against such a mechanism acting globally, but a regional action cannot be excluded.
Acquisition of structure Pulsed exchange of MDH−SR1−GroES ternary complexes at later times allowed two significant conclusions. First, the failure to observe any change in the extent of intact MDH protection at 3 and 9 s versus that at 1 s indicates that MDH intermediates failing to reach the native state inside cis complexes remain only weakly protected. After 9 s, corresponding to the approximate half life of a wild type GroEL−GroES complex, the polypeptide present in a wild type cis complex is ejected into solution. Because only a few percent of MDH molecules reach native form in any given cycle of the GroEL−GroES reaction, the majority of molecules inside SR1−GroES must be nonnative at this time point, and their degree of protection should report the likely state of nonnative MDH molecules after 9 s of attempted folding inside a wild type cis cavity. As judged from the lack of acquisition of any additional protection as compared with 1 s (Fig. 2; Table 1), it appears that there is no detectable additional secondary structure acquired by these molecules during this time period. With wild type GroEL, then, the molecules would be ejected into the bulk solution in this state. This unstable secondary structure and exposure of hydrophobic surfaces would account not only for the susceptibility of such species to aggregation in the bulk solution, but also for their ability to be rebound by chaperonin.
The second conclusion, from pulsed exchange of MDH−SR1−GroES ternary complexes at even later times up to an hour, corresponding to multiple turnovers of wild type GroEL, was that MDH reached a near-native state inside these complexes, lacking native protection only at the dimer interface. From this observation, we can extrapolate that this species corresponds to the form of MDH achieving the native state during any given cycle in cis complexes of wild type GroEL−GroES (or a form committed to the native state that will not rebind to GroEL). As mentioned, only 2−3% of a population of ternary complexes would produce such a native state in any given round of the wild type cycle. At present, however, we lack methods for structural examination of such a small fraction of molecules within a single turnover of wild type GroEL. It appears conveivable that less structured states in cis ternary complexes could already be committed to reaching the native state upon release, but the present study correlated production of native MDH only with appearance of the near-native species (Fig. 2; Table 1).
Methods Proteins. Pig heart mitochondrial MDH was obtained from Boehringer Mannheim and further purified by ultrafiltration on a Centricon 30 (Amicon). SR1 and GroES were expressed and purified as described16.
Formation of MDH−SR1 binary complex. MDH (6.25 nmol) was unfolded in 10 l of 6.6 M urea and 10 mM tris(carboxyethyl)phosphine (TCEP) at 20 °C. After 1 h, the mixture (1.25 mM in MDH subunit) was diluted rapidly into 1 ml containing 12.5 M SR1 in 20 mM Tris, pH 7.6, and 5 mM TCEP. After 15 min at 20 °C, binary complex was separated from unbound MDH by gel filtration on a Superose 6HR column (Pharmacia) equilibrated in 20 mM Tris, pH 7.6, and 40 mM KCl. The complexes were concentrated to 150 M by ultrafiltration (Centricon 30).
Continuous labeling of MDH−SR1 binary complexes. Labeling of MDH−SR1 binary complexes was initiated by 13.5-fold dilution of 7.5 nmol MDH−SR1 complex (starting in 50 l; 150 M) into D2O containing 20 mM Tris, pD 8.0, 40 mM KCl and 10 mM MgCl2 at 20 °C. After various labeling times, isotope exchange was quenched by decreasing the pH to 2.5 with 600 mM KH2PO4 in D2O and chilling to 0 °C (refs 27, 28). For short labeling times (1, 2, 4, 10, 27 and 81 s), labeling and quenching were performed with a continuous-flow mixing system consisting of two mixing tees (Upchurch) separated by delay lines of variable length and diameter. Solution flow rates were controlled by two syringe pumps (Harvard Apparatus), and the protein sample was injected into the system through a Rheodyne HPLC injector. For longer labeling times (240, 720 and 2,160 s), manual mixing was employed.
Pulse labeling of MDH−SR1 binary and MDH−SR1−GroES ternary complexes. Pulse labeling of MDH was carried out with a continuous flow mixing system consisting of three mixing tees joined by two delay lines. For ternary complex studies, 55 l of MDH−SR1 complex (150 M in 20 mM Tris, pH 7.6, and 40 mM KCl) was mixed with 27 l of GroES (375 M) plus 15 mM ATP and 30 mM MgCl2 in 20 mM Tris, pH 7.6, and 40 mM KCl. Under these conditions, over 99% of the MDH−SR1 complexes were bound to GroES. The flow rates, as well as the volumes of the delay line separating the first and second mixing tees, were selected to give refolding times of 1, 9, 27, 81 and 106 s. Isotope labeling was initiated by 13.5-fold dilution of the folding solution containing the MDH−SR1−GroES complexes (100 M) into D2O (20 mM Tris, pD 8.0, 40 mM KCl and 10 mM MgCl2) in the second mixing tee. Following labeling for 1 s, isotope exchange was quenched by mixing with chilled 600 mM phosphate in D2O in the third mixing tee to give a final pH 2.5 and temperature of 4 °C. For longer folding times (240, 480, 720 and 6,000 s), manual mixing was used to form MDH−SR1−GroES complexes, with pulse labeling and quenching as described above.
Isolation and analysis of labeled MDH. In all cases, isotopically labeled and acid-quenched MDH was immediately separated from SR1 or SR1 and GroES by reversed phase chromatography on a Vydac C4 column (50 4.6 mm) using a 35−47% (v/v) acetonitrile gradient, a flow rate of 3 ml min-1 and a mobile phase containing 0.05% (v/v) TFA. MDH eluted at 3 min and was collected into a tube containing 5 ml of ether. The aqueous phase was immediately separated from the organic phase and placed under a stream of nitrogen for 1 min to remove residual ether. This extraction procedure removed most of the acetonitrile, necessary for subsequent HPLC or digestion steps. Following extraction, the solution was buffered at pH 2.5 (200 mM phosphate) and stored at -70 °C until analysis by HPLC MS. The entire isolation and extraction procedure required 5 min. Because the reversed phase chromatography and subsequent steps were performed with protic solvents, rapidly exchanging deuterium in MDH was replaced with protium, leaving deuterium only at the peptide amide linkages27,
28,
29. Samples were analyzed within 3 d of preparation to avoid significant back exchange.
Intact MDH was analyzed by HPLC MS to determine deuterium levels and molecular distributions. MDH was first desalted by reversed phase chromatography on a Vydac C4, 50 1 mm column at 0 °C, directly connected to a Finnigan LCQ mass spectrometer and using a 35−65% (v/v) acetonitrile gradient and 0.05% (v/v) TFA in the mobile phase. Deuterium levels in MDH were determined from the average molecular mass indicated by mass spectra reconstructed from multiple charge states of MDH. Results obtained for undeuterated and totally exchanged control samples analyzed under the same conditions were used to adjust for deuterium back exchange during analysis28. Because rapidly exchanging deuterium was replaced with protium during HPLC, the deuterium level calculated from the average molecular mass was a direct measure of deuteration at amide linkages. Results for the totally exchanged sample of MDH indicated that 89% of the deuterium at peptide amide linkages was retained following isolation by the procedures described above.
MDH digestion and fragment analysis by HPLC MS. Deuterium levels along the MDH backbone were determined from peptic proteolytic fragments of labeled MDH analyzed by HPLC MS28. Isotopically labeled MDH was digested with pepsin at a 1:1 mass ratio for 1.5 min at 0 °C. Duplicate peptic digests were fractionated by reversed phase HPLC (Vydac C4, 50 1 mm, 0 °C) using two different ion pairing agents that changed the elution order to ensure that peptides representing nearly all of the MDH backbone were detected. One elution procedure used 0.05% (v/v) TFA in the mobile phase (0−37% (v/v) acetonitrile gradient over 9 min), whereas the other used 0.05% hexafluorobutyric acid (2−40% (v/v) acetonitrile gradient in 9 min). The HPLC column was attached directly to a Micromass AutoSpec magnetic sector mass spectrometer equipped with an electrospray source and a focal plane detector. Mass spectra of MDH peptic fragments were used to determine the deuterium levels and molecular distributions of deuterium. Results obtained for undeuterated and totally exchanged control samples analyzed in the same way were used to adjust for deuterium back exchange during analysis. Results for the totally exchanged sample of MDH indicated that 70% of the deuterium at peptide amide linkages was retained following isolation and digestion by the procedures described above. All fragments were identified by collision-induced-dissociation tandem mass spectrometry using a Finnigan LCQ mass spectrometer.
Sigler, P.B. et al. Structure and function in GroEL-mediated protein folding. Annu. Rev. Biochem.67, 581−608 (1998). | Article | ISI |
Sparrer, H. & Buchner, J. How GroES regulates binding of nonnative protein to GroEL. J. Biol. Chem.272, 14080−14086 (1997). | Article | ISI |
Rye, H.S. et al. GroEL−GroES cycling: ATP and nonnative polypeptide direct alternation of folding-active rings. Cell97, 325−328 (1999). | Article | ISI |
Farr, G.W. et al. Multivalent binding of nonnative substrate proteins by the chaperonin GroEL. Cell100, 561−573 (2000). | Article | ISI |
Zahn, R., Perrett, S., Stenberg, G. & Fersht, A.R. Catalysis of amide proton exchange by the molecular chaperones GroEL and SecB. Science271, 642−645 (1996). | ISI |
Zahn, R. & Plückthun, A. Thermodynamic partitioning model for hydrophobic binding of polypeptides by GroEL. II. GroEL recognizes thermally unfolded mature -lactamase. J. Mol. Biol.242, 165−174 (1994). | Article | ISI |
Walter, S., Lorimer, G.H. & Schmid, F.X. A thermodynamic coupling mechanism for GroEL-mediated unfolding. Proc. Natl. Acad. Sci. USA93, 9425−9430 (1996). | Article |
Clark, A.C. & Frieden, C. The chaperonin GroEL binds to late-folding non-native conformations present in native Escherichia coli and murine dihyrofolate reductases. J. Mol. Biol.285, 1777−1788 (1999). | Article | ISI |
Bhutani, N. & Udgaonkar, J.B. A thermodynamic coupling mechanism can explain the GroEL-mediated acceleration of the folding of barstar. J. Mol. Biol.297, 1037−1044 (2000). | Article | ISI | ChemPort |
Weissman, J.S., Rye, H.S., Fenton, W.A., Beechem, J.M. & Horwich, A.L. Characterization of the active intermediate of a GroEL−GroES-mediated protein folding reaction. Cell84, 481−490 (1996). | Article | ISI |
Rye, H.S. et al. Distinct actions of cis and trans ATP within the double ring of the chaperonin GroEL. Nature388, 792−798 (1997). | Article | ISI | ChemPort |
Miller, A.D., Maghlaoui K., Albanese, G., Kleinjan, D.A. & Smith, C. Escherichia coli chaperonins Cpn60 (groEL) and Cpn10 (groES) do not catalyse the refolding of mitochondrial malate dehydrogenase. Biochem. J.291, 139−144 (1993). | ISI |
Peralta, D., Hartman, D.J., Hoogenraad, N.J. & Hoj, P.B. Generation of a stable folding intermediate which can be rescued by the chaperonins GroEL and GroES. FEBS Lett.339, 45−49 (1994). | Article | ISI |
Ranson, N.A., Dunster, N.J., Burston, S.G. & Clarke, A.R. Chaperonins can catalyze the reversal of early aggregation steps when a protein misfolds. J. Mol. Biol.250, 581−586 (1995). | Article | ISI |
Weissman, J.S. et al. Mechanism of GroEL action: productive release of polypeptide from a sequestered position under GroES. Cell83, 577−588 (1995). | Article | ISI |
Todd, M.J., Viitanen, P.V. & Lorimer, G.H. Dynamics of the chaperonin ATPase cycle: implications for facilitated protein folding. Science265, 659−666 (1994). | PubMed | ISI | ChemPort |
Bai, Y., Milne, J.S., Mayne, L. & Englander, S.W. Primary structure effects on peptide group hydrogen exchange. Proteins Struct. Funct. Genet.17, 75−86 (1993). | ISI |
Ranson, N.A., Burston, S.G. & Clarke, A.R. Binding, encapsulation and ejection: substrate dynamics during a chaperonin-assisted folding reaction. J. Mol. Biol.266, 656−664. | Article |
Zahn, R., Spitzfaden, C., Ottiger, M., Wüthrich, K. & Plückthun, A. Destabilization of the complete protein secondary structure on binding to the chaperone GroEL. Nature368, 261−265 (1994). | Article | ISI |
Robinson, C.V. et al. Conformation of GroEL-bound -lactalbumin probed by mass spectrometry. Nature372, 646−651 (1994). | Article | ISI |
Gro, M., Robinson, C.V., Mayhew, M., Hartl, F.U. & Radford, S.E. Significant hydrogen exchange protection in GroEL-bound DHFR is maintained during iterative rounds of substrate cycling. Protein Sci.5, 2506−2513.
Goldberg, M.S. et al. Native-like structure of a protein-folding intermediate bound to the chaperonin GroEL. Proc. Natl. Acad. Sci. USA94, 1080−1085 (1997). | Article |
Gervasoni, P., Gehrig, P. & Pluckthun. Two conformational states of -lactamase bound to GroEL: A biophysical characterization. J. Mol. Biol.275, 663−675 (1998). | Article | ISI |
Buckle, A.M., Zahn, R. & Fersht, A.R. A structural model for GroEL-polypeptide recognition. Proc. Natl. Acad. Sci.USA94, 3571−3575 (1997). | Article |
Chen, L. & Sigler, P.B. The crystal structure of a GroEL/peptide complex: plasticity as a basis for substrate diversity. Cell99, 757−768 (1999). | Article | ISI |
Englander, J.J., Rogero, J.R. & Englander, S.W. Protein hydrogen exchange studied by the fragment separation method. Anal. Biochem.147, 234−244 (1985). | ISI |
Zhang, Z. & Smith, D.L. Determination of amide hydrogen exchange by mass spectrometry: a new tool for protein structure elucidation. Protein Sci.2, 522−531 (1993). | ISI |
Smith, D.L., Deng, Y. & Zhang, Z. Probing the noncovalent structure of proteins by amide hydrogen exchange and mass spectrometry. J. Mass. Spectrom.32, 135−146 (1997). | Article | ISI |
Miranker, A., Robinson, C.V., Radford, S.E., Aplin, R.T. & Dobson, C.M. Detection of transient protein folding populations by mass spectrometry. Science262, 896−900 (1993). | ISI |
Zhang, Z., Post, C.B. & Smith, D.L. Amide hydrogen exchange determined by mass spectrometry: application to rabbit muscle aldolase. Biochemistry35, 779−791 (1996). | Article | ISI |
Gleason, W.B., Fu, Z., Birktoft, J. & Banaszak, L. Refined crystal structure of mitochondrial malate dehydrogenase from porcine heart and the consensus structure for dicarboxylic acid oxidoreductases. Biochemistry33, 2078−2088 (1994). | Article | ISI |
Kraulis, P.J. Molscript: A program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr.24, 946−950 (1991). | Article | ISI |
Acknowledgments This work was supported by grants from the NIH, the Nebraska Center for Mass Spectrometry and the Howard Hughes Medical Institute. We thank members of the Horwich lab for helpful discussion and W. Fenton for critical reading of the manuscript.
50:50 mixture of active cis and nonproductive trans complexes
-helices and two adjacent 
-strands that are part of the second Rossmann fold in the N-terminal NAD-binding domain of native MDH (
4) upon the addition of ATP/GroES. To examine the nature of this deprotection, peptic fragments were prepared from MDH, both from pulse-exchanged MDH−SR1 binary complex and from pulse-exchanged ternary complex 1 s after ATP/GroES addition. In particular, the four peptides that exhibited the greatest protection in the binary complex were examined (
l of 6.6 M urea and 10 mM tris(carboxyethyl)phosphine (TCEP) at 20 °C. After 1 h, the mixture (1.25 mM in MDH subunit) was diluted rapidly into 1 ml containing 12.5 
4.6 mm) using a 35−47% (v/v) acetonitrile gradient, a flow rate of 3 ml min-1 and a mobile phase containing 0.05% (v/v) TFA. MDH eluted at 3 min and was collected into a tube containing 5 ml of ether. The aqueous phase was immediately separated from the organic phase and placed under a stream of nitrogen for 1 min to remove residual ether. This extraction procedure removed most of the acetonitrile, necessary for subsequent HPLC or digestion steps. Following extraction, the solution was buffered at pH 2.5 (200 mM phosphate) and stored at -70 °C until analysis by HPLC MS. The entire isolation and extraction procedure required 
, M., Robinson, C.V., Mayhew, M., Hartl, F.U. & Radford, S.E. Significant hydrogen exchange protection in GroEL-bound DHFR is maintained during iterative rounds of substrate cycling. Protein Sci. 5, 2506−2513.
