A Two-Stage Mechanism for the Reductive Unfolding of Disulfide-containing Proteins*

(Received for publication, July 17, 1996, and in revised form, September 16, 1996)

Jui-Yoa Chang Dagger

From the Pharmaceuticals Research Laboratories, Ciba-Geigy Ltd., Basel CH-4002, Switzerland

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
Acknowledgments
REFERENCES


ABSTRACT

Reductive unfolding of disulfide-containing proteins can be experimentally dissected into two distinct stages. In the presence of denaturant and thiol catalyst, native proteins unfold by reshuffling their native disulfides and convert to a mixture of scrambled structures. Subsequent reduction of the disulfide bonds of scrambled proteins requires only mild concentration of reductant (0.2-0.5 mM reduced dithiothreitol) and undergoes intermediates that consist of highly heterogeneous disulfide isomers. These properties have been characterized with three cystine-containing proteins, namely hirudin, tick anticoagulant peptide (TAP), and bovine ribonuclease A. In the cases of hirudin and TAP, most intermediates observed during the oxidative folding were found to exist along the pathway of reductive unfolding as well.


INTRODUCTION

Elucidation of the pathway of protein folding and unfolding has remained one of the most demanding tasks in protein chemistry (1, 2, 3, 4, 5, 6, 7). The challenge stems primarily from the difficulty of analyzing the transient intermediates involved in these pathways. For proteins containing disulfide bonds, unfolding and refolding are generally followed by reduction and oxidation of the native disulfides (8, 9). Since breaking and formation of disulfide bonds can be chemically trapped and characterized (10), the disulfide unfolding and folding pathways can thus be constructed on the basis of the heterogeneity and structures of the trapped intermediates (11, 12, 13, 14, 15).

Unfolding of a disulfide-containing protein can be achieved conventionally by either reduction of disulfide bonds in the absence of denaturant (reductive unfolding) (11, 15, 16, 17) or by denaturation (e.g. GdmCl) in the absence of reductant (disulfide-intact unfolding) (18, 19). In the latter case, the unfolded protein retains intact native disulfides. So far, the pathway of reductive unfolding has been investigated with limited numbers of proteins (11, 15, 20, 21, 22, 23). BPTI and RNase A are two most notable examples. These data, however, were largely obtained from the reductive unfolding performed in the absence of denaturant (11, 15).

During our recent analysis of the properties of scrambled hirudins (24), we have observed that denaturant and reductant can actually be applied in a two-step manner to unfold native protein. In an alkaline solution including strong denaturant and thiol catalyst, the native protein unfolds to form a mixture of scrambled species that admit mostly non-native disulfides but still retain the intact number of disulfide bonds. This is followed by reduction of the disulfides of scrambled proteins. This approach is distinguished from the conventional techniques described above. In this study, this new method was applied to characterize the unfolding pathways of three disulfide-containing proteins, namely hirudin (3 disulfides), tick anticoagulant peptide (TAP)1 (3-disulfides), and ribonuclease A (RNase A) (4 disulfides). The disulfide folding pathways of hirudin and TAP have been analyzed in our laboratory (25, 26). Many well populated folding intermediates of hirudin and TAP were isolated and structurally characterized (26, 27). Demonstration of whether those folding intermediates exist as well along the unfolding pathway will be crucial to our understanding of the relative mechanism of their folding/unfolding. Most importantly, this two-stage mechanism offers a unique insight into the intertwining dependence of the disulfide bonds and non-covalent forces that stabilize the native structure.


EXPERIMENTAL PROCEDURES

Materials

Hirudin (65 amino acids, CGP-39393) and tick anticoagulant peptide (60 amino acids, TAP) were recombinant proteins produced by Ciba Pharmaceuticals, Inc. In the case of hirudin, only the core domain, residues 1-49, was used for the study of unfolding and folding. The core domain of hirudin was prepared by selective removal of its C-terminal tail by chymotrypsin digestion (28) Ribonuclease A (RNase A) was obtained from Sigma. GdmCl, reduced dithiothreitol, and beta -mercaptoethanol were products of Merck with purity greater than 99%.

Unfolding in the Presence of Denaturant and Thiol Catalyst (Stage I)

The native protein (1 mg/ml) was incubated in the Tris-HCl buffer (0.1 M, pH 8.4) containing 0.25 mM beta -mercaptoethanol and selected concentrations of GdmCl. The unfolding was carried out at 23 °C for 16 h. Samples were then trapped by mixing with double volumes of 4% trifluoroacetic acid or by reaction with iodoacetic acid (0.2 M) for 20 min. Unfolded samples were analyzed by HPLC. End products contain a mixture of the native and scrambled species that exist in a state of equilibrium. Calculation of the equilibrium constant is based on the relative recoveries of scrambled proteins (collective) and the native protein. For hirudin and TAP, this can be readily determined from HPLC peak integration since the native hirudin and TAP are well separated from their scrambled isomers (see Fig. 1). The identity of native hirudin and TAP was also confirmed by their inhibitory activity as well as theromolysin-digested peptide mapping. In the case of RNase A, direct quantification was not possible due to the overlapping of the native and scrambled RNase A. This was overcome by additional experiments described in the following: unfolded samples of RNase A were removed from denaturant and beta -mercaptoethanol by gel filtration (NAP-5 column, eluted with 0.1% trifluoroacetic acid). The samples were freeze-dried, reduced with 0.5 mM DTT in the Tris-HCl buffer (0.1 M, pH 8.4) for 20 min, and subsequently analyzed by HPLC. Under this mild reducing condition, the native RNase A remained intact, but scrambled RNase A were quantitatively converted to the fully reduced species (see Fig. 4). The relative concentrations of the native and fully reduced (equals scrambled) RNase A can, therefore, be accurately determined for each unfolded sample.


Fig. 1. Left, time-course unfolding of hirudin in the Tris-HCl buffer (0.1 M, pH 8.4) containing 8 M GdmCl and 0.25 mM of beta -mercaptoethanol. Open arrows indicate the elution position of the fully reduced hirudin. III underlines the scrambled species. c and b are two well populated scrambled species. Both acid and iodoacetate-trapped samples exhibit identical HPLC patterns. Samples were analyzed by HPLC using the following conditions. Solvent A was water containing 0.1% trifluoroacetic acid. Solvent B was acetonitrile/water (9:1, by volume) containing 0.1% trifluoroacetic acid. The gradient was 14-32% solvent B, linear in 50 min. The column used was a Vydac C-18 for peptides and proteins, 4.6 mm, 10 µm. Column temperature was 23 °C. Middle, unfolding of TAP in the presence of indicated concentrations of GdmCl. Samples were trapped by acid after 16 h of unfolding and analyzed by HPLC. Chromatographic conditions were as those described in the case of hirudin except for using different gradient systems. For TAP, it was 28-45% solvent B, linear in 40 min. Right, unfolding of RNase A at different concentrations of GdmCl. The HPLC gradient was 30-50% solvent B, linear in 30 min. N indicates the elution position of the native species in all three cases.
[View Larger Version of this Image (15K GIF file)]



Fig. 4. Reduction of scrambled RNase A in the Tris-HCl buffer (0.1 M, pH 8.4) containing 0.5 mM of DTT. The reaction was performed at 23 °C. HPLC conditions were as those described in Fig. 1.
[View Larger Version of this Image (7K GIF file)]


Reduction of Scrambled Proteins (Stage II)

Scrambled proteins, prepared by the method described above (stage I), were separated from the denaturant and beta -mercaptoethanol by passing the sample through an NAP-5 column. Scrambled hirudin and TAP were further purified by HPLC. Reduction of scrambled proteins was performed at 23 °C in the Tris-HCl buffer (0.1 M, pH 8.4) containing varying concentrations of DTT (0.2-4 mM). The protein concentration was 1 mg/ml. The reaction was carried out both in the absence and presence of 6 M GdmCl. Reduced samples were trapped in a time course manner by mixing aliquots of the sample with 2 volumes of 4% trifluoroacetic acid.

Reduction of the Native Protein in the Absence of Denaturant

The native protein (1 mg/ml) was treated with different concentrations of DTT (5-200 mM) in the Tris-HCl buffer (0.1 M, pH 8.4). Reduction was carried out at 23 °C. Unfolding intermediates were trapped by mixing aliquots of the reaction sample with 2 volumes of 4% trifluoroacetic acid, followed by HPLC analysis.

Protein Analytical Methods

The disulfide content of scrambled proteins was determined by the dabsyl chloride pre-column derivatization method (29), which permits direct quantification of the disulfide bonds of proteins. The MALDI mass spectrometer was a home-built time of flight (TOF) instrument with a nitrogen laser of 337 nm wavelength and 3 ns pulse width. The apparatus has been described in detail elsewhere (30). The calibration was performed either externally or internally, by using standard proteins (hypertensin, M.W. 1031.19; synacthen, 2934.50, and calcitonin, 3418.91). The unfolding/folding intermediates were analyzed by HPLC using the conditions described in the legend of Fig. 1.


RESULTS

Unfolding of the Native Protein to the State of Scrambled Structures (Stage I)

In the presence of denaturant and trace concentration of thiol reagent (0.25 mM beta -mercaptoethanol), unfolding of the native protein was accompanied by reshuffling of the native disulfides, which led to the formation of scrambled structures. Under these conditions, beta -mercaptoethanol functioned as a thiol catalyst instead of reducing agent, and the unfolded proteins retained negligible residues of free cysteine. An example of time course unfolding of hirudin in the alkaline solution containing 8 M GdmCl and beta -mercaptoethanol (0.25 mM) is shown at the left column of Fig. 1. The unfolding reaction reached a plateau within about 4-5 h, and in this case, 85% of the native hirudin was unfolded and denatured. Prolonged incubation of up to 24 h or further replenishment of thiol catalyst neither altered the composition nor improved the yield of scrambled hirudins, suggesting that the native and scrambled hirudins had reached a state of equilibrium. Using this method, hirudin, TAP, and RNase A were unfolded in the presence of varying concentrations of GdmCl. The results are presented at the middle and right columns of Fig. 1.

The extent of unfolding, and hence the equilibrium constant (Keq) between scrambled species and the native protein, was shown to be dependent upon the strength of denaturant (Table I). In addition, there exist narrow ranges of the concentration of denaturant in which native proteins precipitously unfold. For instance, in between 1 and 3 M GdmCl, the equilibrium constant of unfolding of RNase A increased from <0.05 to >100. The same phenomenon was observed with TAP at 3-5 M GdmCl. These critical concentrations of denaturant appear to resemble "melting point" at which forces of non-covalent interactions that stabilize the native proteins become abruptly nullified.

Table I.

Equilibrium constant of scrambled proteins and the native protein in the presence of various concentrations of GdmCl

Calculation of the equilibrium constant is based on the relative recoveries of scrambled proteins (collective) and the native protein. The data should be allowed a standard deviation of ±5%.
GdmCl
1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M

Hirudin <0.01 ND 0.11 0.30 0.67 1.43 3.0 5.25
TAPa ND <0.05 <0.10 0.90 4.75 16.0 28.0 ND
RNase A <0.01 1.0 >100 >100 >100 >100 ND ND

a  ND, not determined.

All scrambled proteins contain intact numbers of disulfide bonds. Free cysteines or mixed disulfides with beta -mercaptoethanol have not been detected. These were confirmed by amino acid composition analysis and MALDI mass spectroscopic analysis of unfolded species that were treated with 0.1 M iodoacetic acid. The composition of scrambled proteins are heterogeneous. For hirudin and TAP that contain 3 disulfides, there exists 14 possible scrambled isomers. Eleven species of scrambled hirudins have been isolated and structurally characterized (27). Similar numbers of scrambled TAP were identified as well (26). In the case of RNase A, there are 104 possible scrambled isomers. Complete separation of these isomers will be a daunting task. It is not possible to estimate how many isomers of scrambled RNase A do exist since they are poorly separated by the HPLC conditions employed here. However, if the cases of hirudin and TAP are any indication, scrambled RNase A could be very heterogeneous as well.

It is relevant to mention that the selection of 0.25 mM beta -mercaptoethanol has been concluded from a systematic study of unfolding performed at 6 M GdmCl and varied concentrations of beta -mercaptoethanol. The end products (after 16 h of unfolding) were carboxymethylated, followed by amino acid composition analysis and molecular mass analysis. The results reveal that the optimized concentration of beta -mercaptoethanol ranges from 0.2 to 0.3 mM. At higher concentration of beta -mercaptoethanol, the disulfide bonds of proteins may become partially reduced (data not shown).

Reduction of Scrambled Proteins (Stage II)

The disulfide bonds of scrambled proteins were reduced with varying concentrations of DTT (0.2-4 mM). The reactions were trapped in a time-course manner, and intermediates that appeared along the course of reduction were analyzed by HPLC. The results obtained from the reduction of scrambled hirudins, both in the absence and presence of denaturant, are shown in Fig. 2. Using 0.5 mM DTT, complete reduction of scrambled hirudins was achieved within 20 min. The reaction undergoes complex intermediates that consist of 2- and 1-disulfide species. Judging from the HPLC pattern, the composition of these unfolding intermediates are indistinguishable from those found along the pathway of oxidative folding of hirudin (25, 31). Furthermore, the rate of reductive unfolding and the composition of unfolding intermediates are largely unaffected by the presence of 6 M GdmCl (Fig. 2), suggesting that 2- and 1-disulfide intermediates adopt minimal structural elements of non-covalent interactions that are sensitive to the denaturant. These data are consistent as well with that observed during the process of oxidative refolding of fully reduced hirudin in which the compositions of 1- and 2-disulfide intermediates are not affected by the presence of denaturant.


Fig. 2. Reduction of scrambled hirudins with 0.5 mM of DTT in the Tris-HCl buffer (0.1 M, pH 8.4) in the absence (left) and presence (right) of 6 M GdmCl. The reaction was carried out at 23 °C. c denotes one of the two well populated fractions of scrambled species. R indicates the fully reduced species. HPLC conditions were as those described in Fig. 1.
[View Larger Version of this Image (16K GIF file)]


Reduction of scrambled TAP also proceeds via heterogeneous 2- and 1-disulfide isomers. However, the patterns of 2- and 1-disulfide unfolding intermediates are visibly influenced by denaturant (Fig. 3), an indication that some of the 2- and 1-disulfide species of TAP must be stabilized by non-covalent interactions. Those stabilizing factors, however marginal, are also reflected by a roughly 2-fold increase of the rate of reduction when the reaction was carried out in the presence of 6 M GdmCl. Again, the composition of unfolding intermediates of scrambled TAP can not be distinguished from that observed along the course of oxidative folding of fully reduced TAP, both with or without the participation of denaturant (26). Scrambled RNase A could be similarly reduced and the unfolding intermediates also comprise highly heterogeneous species (Fig. 4)


Fig. 3. Reduction of scrambled TAP with 0.5 mM of DTT in the Tris-HCl buffer (0.1 M, pH 8.4) in the absence (left) and presence (right) of 6 M GdmCl. The reaction was carried out at 23 °C. R indicates the fully reduced species. HPLC conditions were as those described in Fig. 1.
[View Larger Version of this Image (13K GIF file)]


The most striking outcome is that disulfide bonds of three different sets of scrambled proteins exhibit very similar stability against DTT reduction. All of them could be fully reduced by 0.5-1 mM DTT within 10-20 min at room temperature (23 °C). The stability of scrambled disulfides is also comparable with the inter-disulfide bond that cross-links two intact hirudin monomers (32).

Reduction of the Native Protein in the Absence of Denaturant

In the absence of denaturant, it needs high concentration of reducing agent to break the native disulfides. For example, complete reduction of the three native disulfides of hirudin requires incubation with 100 mM DTT for longer than 2 h. Thus, the native disulfides of hirudin are approximately 500-fold more stable against reduction than non-native disulfides of scrambled hirudins. However, unlike their scrambled counterparts, the stability of native disulfides varies substantially among these three proteins. Judging from the concentration of DTT required to achieve the same kinetics of full reduction, the native disulfides of RNase A were about 8-fold and 64-fold more stable than that of hirudin and TAP, respectively (Fig. 5). These differences apparently reflect the extent of contribution of non-covalent interactions of individual proteins. When the factors of non-covalent forces are largely abrogated, as in the state of scrambled structures, the stability of disulfide bonds (against reduction) become more or less indistinguishable.


Fig. 5. Reduction of the native proteins in the absence of denaturant. The native protein was treated with indicated concentrations of DTT. Time course intermediates were trapped by acid and analyzed by HPLC using the conditions described in the legend of Fig. 1. A negligible amount of intermediate was found with hirudin and TAP. In the case of RNase A, two minor fractions of intermediates (denoted x and y) were observed.
[View Larger Version of this Image (16K GIF file)]


It is important to point out that the order of stability of native proteins exhibited by their resistance against reduction (RNase A > hirudin > TAP) (Fig. 5) differs from that measured against denaturant (hirudin > TAP > RNase A) (Table I). For instance, the native disulfide bonds of RNase A are about 64-fold more stable than those of TAP. On the other hand, complete denaturation of TAP requires double concentration of GdmCl (5-6 M) as compared with that needed for RNase A (3 M). This discrepancy is probably related to the nature of non-covalent interactions (e.g. the number of hydrogen bonds, etc.). While non-covalent interactions can be added up collectively to enforce the stability of a disulfide bond against reduction, their own stability in the presence of denaturant may not behave in the same fashion.

Examination of time course-trapped samples shows that reduction undergoes an apparent all or none mechanism in which only trace amounts of partially reduced intermediates accumulate. This phenomenon was observed with all three proteins at DTT concentrations ranging from 10 to 200 mM. With hirudin and TAP, about 1-4% of 1- and 2-disulfide intermediates was indeed detectable (Fig. 5). In the case of RNase A, two fractions of intermediates comprising approximately 5-15% of the total protein were observed (Fig. 5). Molecular mass analysis of carboxymethylated derivative has revealed that these two minor fractions contain mostly 3-disulfide RNase A. These two intermediates may correspond to those observed by Scheraga and colleagues (15). It has been suggested that the concentration of these intermediates is related to the efficiency of the trapping agent (15). Notwithstanding, the level of their accumulation has not exceeded 15% of the total protein in repeated experiments despite the fact that acid trapping represents one of the most efficient methods in quenching the disulfide formation and rearrangement (16).

An "all or none" mechanism suggests that the reduction of the first disulfide bond exists as a rate-limiting step. Following breakdown of the first disulfide, the remaining cystines become precipitously reduced. This phenomenon thus implies that disulfides of a protein are stabilized in a cooperative and concerted manner by the non-covalent forces of the folded structure. However, it must be emphasized that this mechanism does not represent a general property for disulfide-containing proteins. For large proteins or those consisting of multiple domains, it is most likely that stability of disulfides will reflect their local structures that are strengthened by varied degrees of non-covalent interactions. Many disulfide-containing proteins, even small proteins, are known to be reduced in a sequential manner (21, 33, 34, 35). The reality is that the pathway of disulfide reduction for most proteins probably lies in between the strict mode of all or none and sequential. The reduction of human antithrombin III is a typical example (36). Even hirudin, demonstrated here, does not follow an absolute all or none mechanism because of the presence of approximately 2% of intermediates.

Furthermore, the mode of reduction, either proceeding via a sequential manner or undergoing an all or none mechanism, bears crucial implications in interpreting the structure and function relationship of disulfide bonds. In searching for the functional role of disulfide bonds, proteins are generally treated with reducing agent in the absence of denaturant. Partially reduced proteins are irreversibly trapped by alkylation (e.g. carboxymethylation). The remaining biological activity of the modified protein is then correlated to the residual number of disulfides, conventionally determined by amino acid composition analysis. If the surviving numbers of disulfides and biological activity display a close to linear relationship, interpretation of the data will be dependent upon the mode of disulfide reduction. Take TAP as an example. It has been shown that reduction of one-third of the disulfide bonds of TAP results in a loss of approximately one-third of its biological activity (37). In case that disulfide reduction proceeds sequentially, this data would suggest that the intactness of one specific disulfide bond accounts for one-third of the biological function of the protein. In case that reduction undergoes an all or none pathway, then it is obvious that the modified protein contains two-thirds of the intact species and one-third of the fully reduced species that lack inhibitory activity. The data obtained here together with those observed by Sardana et al. (37) clearly indicate that TAP belongs to the latter case.

Fully Reduced Hirudin Refolds via the Same Pathway Regardless of Whether Reduction Is Carried Out in the Presence or Absence of Denaturant

Since the native protein can be fully reduced both in the absence and presence of denaturant, one important question is whether these two types of fully reduced proteins adopt the same extent of unfolding. Here, we have attempted to partly answer this question by comparing their refolding behaviors. These experiments were designed to examine whether the sample reduced without denaturant retains residual native-like, non-covalent structures that are not abrogated simply by the breakdown of disulfide linkages. If this is the case, one may expect it to refold with increased efficiency and decreased heterogeneity of intermediates, as compared with that unfolded in the presence of denaturant.

Two types of fully reduced hirudins, that unfolded either by reduction alone (Rstar ) or by reduction in the presence of 6 M GdmCl (R), were allowed to refold in the same alkaline buffer both in the presence and absence of thiol catalyst. The mechanism of their folding was compared by characterizing acid-trapped intermediates. Analysis of the time course-quenched intermediates shows that folding of both reduced species (Rstar and R) undergoes an initial stage of nonspecific disulfide formation that leads to the formation of scrambled 3-disulfide species as essential folding intermediates (Figs. 6 and 7). When folding was carried out in the buffer alone (Fig. 6), about 70% of the sample became permanently trapped at scrambled structures and unable to convert to the native structure due to the lack of free thiol to catalyze the reshuffling of non-native disulfides. As folding was performed in the presence of thiol catalyst, the extent of accumulation of scrambled species significantly reduced along the pathway and the native hirudin was quantitatively recovered (Fig. 7). The data clearly demonstrate that both reduced hirudins refold with comparable efficiency and via the same complexity of intermediates. Indeed, the sample reduced in the absence of denaturant refolds at a slightly slower rate. This difference of kinetics was consistently observed with three repeated experiments. These results thus strongly indicate that the sample unfolded without denaturant adopts no unique element of structure that would facilitate its refolding. However, one can not rule out the possibility that some native-like interactions may be retained, but they need to be broken and remade during the process of refolding. This can well explain why it refolds slightly slower than that unfolded in the presence of denaturant.


Fig. 6. Refolding of fully reduced hirudins. Rstar was prepared by reduction of the native hirudin with 100 mM DTT at 23 °C for 90 min. R was prepared in the presence of 30 mM DTT and 6 M GdmCl at 23 °C for 90 min. Both Rstar and R were desalted by passing the sample through a PD-10 column eluted with Tris-HCl buffer (0.1 M, pH 8.4) and were allowed to refold in the same Tris-HCl buffer. The final protein concentration was 0.5 mg/ml. The folding intermediates were trapped by 4% trifluoroacetic acid. Note that the starting material Rstar contains 5% native species. The end products of folding consist of about 30% native hirudin and 70% scrambled species (III) (This ratio is dependent upon the concentration of the protein of the folding). It is relevant to mention that the composition of scrambled hirudins existing along the folding pathway is not identical to those obtained from the unfolding (see Fig. 1). About half of the scrambled hirudins are denaturant-sensitive, and their concentrations decrease significantly in the presence of GdmCl.
[View Larger Version of this Image (22K GIF file)]



Fig. 7. Quantitative analysis of various disulfide species along the folding pathways of fully reduced hirudins. a, refolding of Rstar in the Tris-HCl buffer alone (original chromatograms are shown at the left column of Fig. 6). b, refolding of R in the Tris-HCl buffer alone (original chromatograms are shown at the right column of Fig. 6). c, refolding of Rstar in the Tris-HCl buffer containing 0.25 mM beta -mercaptoethanol. d, refolding of R in the Tris-HCl buffer containing 0.25 mM beta -mercaptoethanol. black-square, N; square , III; triangle , I + II; open circle , R. The method for quantitative analysis of various species of hirudin has been described previously (31).
[View Larger Version of this Image (37K GIF file)]



DISCUSSION

Correlation of the Pathways of Unfolding and Folding

Oxidative folding of hirudin and TAP has been shown to undergo an initial stage of nonspecific2 disulfide pairing that leads to the formation of scrambled species as essential folding intermediates. This is then followed by disulfide reshuffling of scrambled species in the presence of thiol catalyst to attain the native structure (25, 26, 31). This two-stage mechanism of oxidative folding can be experimentally reversed during the process of reductive unfolding, as demonstrated here. In the presence of denaturant and thiol catalyst, the native protein unfolds back to the state of scrambled structures. Subsequent reduction of scrambled protein generates the fully reduced species. Thus, both unfolding and folding are accomplished by a sequential abrogation (unfolding) or formation (folding) of the two major structures that stabilize the native protein, namely the non-covalent specific interactions and the disulfide bonds. The flow chart of folding and unfolding pathways of hirudin and TAP, and conditions that influence kinetics of these pathways are outlined in Fig. 8.


Fig. 8. The two-stage mechanism of reversible unfolding/folding of hirudin and TAP. N, III, II, I, and R stand for the native, 3-disulfide scrambled, 2-disulfide, 1-disulfide, and fully reduced (0-disulfide) species, respectively. III, II, and I all consist of equilibrated isomers. Folding undergoes an initial stage of nonspecific packing that leads to the formation of scrambled species as folding intermediates (1. This process is uniquely accelerated by oxidized glutathione and cystine. In the case of hirudin, the presence of denaturant exerts no apparent influence on the compositions of I and II, but about half of species III are sensitive to the denaturant. In the case of TAP, the compositions of all three classes of intermediates are affected by the presence of denaturant. The final stage of folding is characterized as scrambled species reshuffle their non-native disulfides in the presence of thiol catalyst and consolidate to form the native structure (2. This process is driven by the non-covalent specific interactions and, therefore, is sensitive to denaturant in both cases of hirudin and TAP. The first stage of unfolding, performed in the presence of strong denaturant and thiol catalyst, leads to a reversible conversion of the native structure to the scrambled species again (3). Reduction of scrambled structures converts the protein back to the fully reduced species (4). This process undergoes the same 1- and 2-disulfide intermediates observed in the first stage of folding.
[View Larger Version of this Image (9K GIF file)]


Most intermediates observed during the oxidative folding exist along the pathway of reductive unfolding as well. Indeed, the composition of 1- and 2-disulfide species that appeared during the first stage of folding (1 of Fig. 8) is indistinguishable from those found at the second stage of unfolding (4 of Fig. 8). This holds for both hirudin and TAP. However, the composition of 3-disulfide scrambled intermediates is distinguishable along the two opposite pathways. These differences are shown in Fig. 9. Some scrambled species that are well-populated along the direction of folding become diminished along the route of unfolding. This is because unfolding that leads to the scrambled intermediates has to be performed in the presence of denaturant. Scrambled species that are denaturant-sensitive presumably adopt structural elements that are stabilized by non-covalent interactions. These structures still remain to be elucidated. Even if they are native-like, their existence appears to bear no direct link to the native disulfides since most denaturant-sensitive scrambled species do not contain any native disulfide (24, 26). Species f (Fig. 9) of scrambled TAP is a vivid example. The concentration of species f decreases from 30% of the total scrambled TAP in the absence of denaturant (folding) to less than 3% in the presence of 6 M GdmCl (unfolding). Yet, species f contains three non-native disulfides (26).


Fig. 9. Distinction of the compositions of scrambled intermediates observed along the reversible pathways of unfolding and folding of hirudin (left column) and TAP (right column). For the folding experiment, the native hirudin and TAP were first reduced and denatured for 90 min in the Tris-HCl buffer (0.5 M, pH 8.4) containing M GdmCl and 30 mM DTT. The unfolded samples were removed from the excess reductant and denaturant by gel filtration (NAP-5 column, Pharmacia) and subsequently allowed to refold in the Tris-HCl buffer (0.1 M, pH 8.4) alone for 16 h at 23 °C. The end products of folding contain, in both cases of hirudin and TAP, about 60% of the scrambled species and 40% of the native species. The scrambled species were unable to convert to the native structure due to the absence of thiol catalyst. For the unfolding experiment, the native hirudin and TAP were incubated in the Tris-HCl buffer (0.1 M, pH 8.4) containing 6 M GdmCl and 0.25 mM beta -mercaptoethanol. The experiments were performed at 23 °C for 16 h. The end products of unfolding also contain a mixture of scrambled species and native species. In the case of hirudin the ratio is 60:40 and for TAP it is 94:6. N indicates the native TAP. In the case of hirudin, only scrambled species are presented here. The native hirudin is eluted 14 min ahead of species h and is not shown here (see Fig. 6). Scrambled species are marked alphabetically (a-i) in both examples. The disulfide structures of all marked scrambled hirudins (27) and four well populated scrambled TAP (a, d, f, and g) (26) have been determined.
[View Larger Version of this Image (26K GIF file)]


The Intertwining Dependence of the Disulfide Bonds and Non-covalent Forces

Disulfides are known to enforce the stability of native proteins (16, 38, 39, 40). Most disulfide-containing proteins lose essentially all of their biological activity as well as their tertiary structures when disulfide bonds are ruptured by reduction (41). Even with exceptional cases (42) in which partial activity remains after complete disulfide reduction, the stability of their native conformation usually becomes drastically diminished (43). Less certain is how disulfide bonds interact with non-covalent structures to stabilize the folded native conformation (40, 44). The data of reductive unfolding, together with those obtained from the oxidative folding of hirudin and TAP (25, 26), may have provided an important insight into the intertwining dependence of the disulfide bonds and non-covalent forces. These crossing dependences, based on those observed in the cases of hirudin and TAP, are summarized in the following. During the process of folding, 1) the non-covalent interactions do not actively drive or direct the folding until non-native scrambled disulfides are formed; and 2) on the other hand, conversion of scrambled disulfides to the native disulfides is driven and guided by the non-covalent interactions. Under conditions in which non-covalent interactions are abrogated (e.g. in the presence of strong denaturant), scrambled structures accumulate as the end products because they are unable to convert to the native structure. While during the process of unfolding, 3) nullifying the non-covalent interactions by denaturant, the native disulfides will collapse into a mixture of scrambled disulfides; and 4) breaking the native disulfides directly by reducing agent, the non-covalent interactions will be incapable of maintaining their native structure and will eventually disintegrate.

These findings are fundamentally consistent with the hypothesis (40, 45, 46) that asserts that the primary role of disulfide bonds is to cross-link the protein at its unfolded state, to decrease and limit the entropy allowed to the unfolded structure, and thus help shift the unfolding/folding equilibrium to the favor of the native structure. For proteins like hirudin and TAP, the force of non-covalent interactions alone is apparently insufficient to drive the unfolded protein to cross the energy barrier and guide the way to attain the native conformation. Furthermore, once it assumes the native structure, it still needs the constant support of disulfide bonds.


FOOTNOTES

*   The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dagger    To whom correspondence should be addressed: K-121,104, Ciba-Geigy Ltd., Basel, CH-4002 Switzerland. Fax/Tel.: 41-61-6968313.
1    The abbreviations used are: TAP, tick anticoagulant peptide; RNase A, ribonuclease A; GdmCl, guanidine hydrochloride; DTT, dithiothreitol; HPLC, high performance liquid chromatography.
2    This conclusion is based on the heterogeneity of 1- and 2-disulfide intermediates observed along the folding pathway of hirudin and TAP.

Acknowledgments

The author thanks Drs. T. Hawthorne and W. Maerki for supplying the recombinant TAP and support of this study.


REFERENCES

  1. Kim, P. S., and Baldwin, R. L. (1990) Annu. Rev. Biochem. 59, 631-660 [CrossRef][Medline] [Order article via Infotrieve]
  2. Creighton, T. E. (1990) Biochem. J. 270, 1-16 [Medline] [Order article via Infotrieve]
  3. 3441Richards, F. M. (1991) Sci. Am. 34-41
  4. Dill, K. A., and Shortle, D. (1991) Annu. Rev. Biochem. 60, 795-825 [CrossRef][Medline] [Order article via Infotrieve]
  5. Matthews, C. R. (1993) Annu. Rev. Biochem. 62, 653-683 [CrossRef][Medline] [Order article via Infotrieve]
  6. Fersht, A. R. (1993) FEBS Lett. 325, 5-16 [CrossRef][Medline] [Order article via Infotrieve]
  7. Dill, K. A., Bromberg, S., Yue, K., Fiebig, K. M., Yee, D. P., Thomas, P. D., and Chan, H. S. (1995) Protein Sci. 4, 561-602 [Abstract/Free Full Text]
  8. Anfinsen, C. B. (1973) Science 181, 223-230 [Medline] [Order article via Infotrieve]
  9. Haber, E., and Anfinsen, C. B. (1962) J. Biol. Chem. 237, 1839-1844 [Free Full Text]
  10. Creighton, T. E. (1986) Methods Enzymol. 131, 83-106 [Medline] [Order article via Infotrieve]
  11. Creighton, T. E. (1979) J. Mol. Biol. 129, 411-431 [Medline] [Order article via Infotrieve]
  12. Scheraga, H. A., Konishi, Y., and Ooi, T. (1984) Adv. Biophys. 18, 21-41 [CrossRef][Medline] [Order article via Infotrieve]
  13. Creighton, T. E. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 5082-5086 [Abstract]
  14. Weissman, J. S., and Kim, P. S. (1991) Science 253, 1386-1393 [Medline] [Order article via Infotrieve]
  15. Li, Y.-J., Rothwart, D. M., and Scheraga, H. A. (1995) Nat. Struct. Biol. 2, 489-494 [CrossRef][Medline] [Order article via Infotrieve]
  16. Creighton, T. E. (1984) Methods Enzymol. 107, 305-329 [Medline] [Order article via Infotrieve]
  17. Jaenicke, R., and Rudolph, R. (1990) in Protein Structure (Creighton, T. E., ed), pp. 191-224
  18. Radford, S. E., Dobson, C. M., and Evans, P. A. (1992) Nature 358, 302-307 [CrossRef][Medline] [Order article via Infotrieve]
  19. Houry, W. A., Rothwarf, D. M., and Scheraga, H. A. (1995) Nat. Struct. Biol. 2, 495-503 [CrossRef][Medline] [Order article via Infotrieve]
  20. Creighton, T. E., and Goldengerg, D. P. (1984) J. Mol. Biol. 179, 497-526 [Medline] [Order article via Infotrieve]
  21. Kuwajima, K., Ikeguchi, M., Sugawara, T., Hiraoka, Y., and Sugai, S. (1990) Biochemistry 29, 8240-8249 [Medline] [Order article via Infotrieve]
  22. Ewbank, J. J., and Creighton, T. E. (1993) Biochemistry 32, 3677-3693 [Medline] [Order article via Infotrieve]
  23. Mendoza, J. A., Jarstfer, M. B., and Goldenberg, D. P. (1994) Biochemistry 33, 1143-1148 [Medline] [Order article via Infotrieve]
  24. Chang, J.-Y. (1995) J. Biol. Chem. 270, 25661-25666 [Abstract/Free Full Text]
  25. Chatrenet, B., and Chang, J.-Y. (1993) J. Biol. Chem. 268, 20988-20996 [Abstract/Free Full Text]
  26. Chang, J.-Y. (1996) Biochemistry 35, 11702-11709 [CrossRef][Medline] [Order article via Infotrieve]
  27. Chang, J.-Y., Schindler, P., and Chatrenet, B. (1995) J. Biol. Chem. 270, 11992-11997 [Abstract/Free Full Text]
  28. Chang, J.-Y. (1990) J. Biol. Chem. 265, 22159-22166 [Abstract/Free Full Text]
  29. Chang, J.-Y., and Knecht, R. (1991) Anal. Biochem. 197, 52-58 [Medline] [Order article via Infotrieve]
  30. Boernsen, K. O., Schaer, M., and Widmer, M. (1990) Chimia 44, 412-416
  31. Chang, J.-Y. (1994) Biochem. J. 300, 643-640 [Medline] [Order article via Infotrieve]
  32. Chang, J.-Y., Grossenbacher, H., Meyhack, B., and Märki, W. (1993) FEBS Lett. 336, 53-56 [CrossRef][Medline] [Order article via Infotrieve]
  33. Kress, L. F., and Laskowski, M., Sr. (1967) J. Biol. Chem. 242, 4925-4929 [Abstract/Free Full Text]
  34. Wan-kyng, L., and Meienhofer, J. (1968) Biochem. Biophys. Res. Commun. 31, 467-473 [Medline] [Order article via Infotrieve]
  35. Hollecker, M., and Creighton, T. E. (1983) J. Mol. Biol. 168, 409-437 [Medline] [Order article via Infotrieve]
  36. Sun, X.-J., and Chang, J.-Y. (1989) J. Biol. Chem. 264, 11288-11293 [Abstract/Free Full Text]
  37. Sardana, M., Sardana, V., Rodkey, J., Wood, T., Ng, A., Vlasuk, G. P., and Waxman, L. (1991) J. Biol. Chem. 266, 13560-13563 [Abstract/Free Full Text]
  38. Villafranca, J. E., Howell, E. E., Voet, D. H., Strobel, M. S., Ogden, R. C., Abelson, J. N., and Kraut, J. (1983) Science 222, 782-788 [Medline] [Order article via Infotrieve]
  39. Wells, J. A., and Powers, D. B. (1986) J. Biol. Chem. 261, 6564-6570 [Abstract/Free Full Text]
  40. Wetzel, R. (1987) Trends Biochem. Sci. 12, 478-482
  41. Thornton, J. M. (1981) J. Mol. Biol. 151, 261-287 [Medline] [Order article via Infotrieve]
  42. Lehle, K., Wrba, A., and Jaenicke, R. (1994) J. Mol. Biol. 239, 276-284 [CrossRef][Medline] [Order article via Infotrieve]
  43. Lehle, K., Kohnert, U., Stern, A., Popp, F., and Jaenicke, R. (1996) Nature Biotechnology 14, 476-480 [Medline] [Order article via Infotrieve]
  44. Pace, N. C. (1975) CRC Crit. Rev. Biochem. 3, 1-43 [Medline] [Order article via Infotrieve]
  45. Johnson, R. E., Adams, P., and Rupley, J. A. (1978) Biochemistry 17, 1479-1484 [Medline] [Order article via Infotrieve]
  46. Schulz, G. E., and Schrimer, R. H. (1979) Principles of Protein Structure, pp. 27-65, Springer-Verlag New York Inc., New York

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.