COMMUNICATION:
Scanning and Escape during Protein-disulfide Isomerase-assisted Protein Folding*

(Received for publication, January 10, 1997)

Kenneth W. Walker and Hiram F. Gilbert Dagger

From the Verna and Marrs McLean Department of Biochemistry, Baylor College of Medicine, Houston, Texas 77030

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS AND DISCUSSION
FOOTNOTES
REFERENCES


ABSTRACT

During oxidative protein folding, efficient catalysis of disulfide rearrangements by protein-disulfide isomerase is found to involve an escape mechanism that prevents the enzyme from becoming trapped in covalent complexes with substrates that fail to rearrange in a timely fashion. Protein-disulfide isomerase mutants with only a single active-site cysteine catalyze slow disulfide rearrangements and become trapped in a covalent complex with substrate. Escape is mediated by the second, more carboxyl-terminal cysteine at the active site. A glutathione redox buffer increases the kcat for single-cysteine mutants by 20-40-fold, but the presence of the second cysteine at the active site in the wild-type enzyme increases the kcat by over 200-fold. A model is developed in which kinetic scanning for disulfides of increasing reactivity is timed against an intramolecular clock provided by the second cysteine at the active site. This provides an alternative, more efficient mechanism for rearrangement involving the reduction and reoxidation of substrate disulfides.


INTRODUCTION

With no known sequence cues to designate which cysteines should be cross-linked in a native protein, the question arises as to how correct pairs of cysteines are selected during folding and how this selection is made quickly enough to be useful in the cell. The uncatalyzed formation of native disulfides during in vitro protein folding is generally not fast enough to support the folding rate that is found in the cell (1). To overcome this, catalysis of disulfide formation and rearrangement in eukaryotes is provided by protein-disulfide isomerase (PDI),1 a 55-kDa protein of the endoplasmic reticulum (2).

As a catalyst, PDI must deal with different mechanisms for directing disulfide formation. With some proteins, such as bovine pancreatic trypsin inhibitor, the specification of cysteine connectivity occurs early in folding and is directed by the formation of native-like structures that are interconverted by intramolecular disulfide rearrangements (3). With other proteins, such as ribonuclease A (RNase), the identification of which cysteines to connect occurs late in folding resulting in a large collection of intermediates with random disulfides and substantial non-native structure that must be rearranged to give native connectivity (4). PDI is capable of accelerating folding that proceeds by either of these mechanisms (5, 9).

During the oxidative folding of reduced ribonuclease (RNase), PDI catalyzes the initial formation of substrate disulfides to yield a collection of inactive, RNase redox isomers that must be rearranged to the native disulfide pattern (6). PDI has two active sites that are housed in two, internally homologous thioredoxin domains, one near the amino terminus and the other near the carboxyl terminus (7). Both active sites contribute to PDI catalysis (8), but the two active sites are not equivalent (9). Each of the two active sites has two cysteine residues, found in the sequence WCGHCK (7). The two cysteines at each active site serve different functions. The cysteine nearer the amino terminus of the domain (WCGHCK) is essential for all catalytic activity (10-12), both oxidation and isomerization. Mutation of the cysteine closer to the carboxyl terminus (WCGHCK) selectively destroys the ability of PDI to catalyze disulfide formation. However, these mutants (NSO:COO and NOO:CSO)2 still retain a low but measurable amount of isomerase activity (10, 12), approximately 10-12% of the kcat of the wild-type enzyme (10).

Formally, the rearrangement of substrate disulfides does not require the net oxidation or reduction of PDI or the substrate (Scheme I). A disulfide rearrangement initiated by the attack of a single active-site cysteine of PDI could propagate through the substrate and terminate by expelling PDI (13). Consequently, we were surprised to find that PDI mutants with only one cysteine at an active site were considerably less effective than a wild-type active site in catalyzing rearrangements of RNase during oxidative folding of the reduced substrate (10).


Scheme I. Mechanism of an intramolecular, PDI-catalyzed rearrangement of substrate disulfides.
[View Larger Version of this Image (8K GIF file)]


The formation of "incorrect" disulfides that must be isomerized during folding creates logistical problems for a catalyst like PDI. Isomerization is initiated by the formation of a covalent complex between the active-site cysteine and the substrate protein. The incorrect structures must be searched and searched again, scanning, isomerizing, and breaking disulfides until the native structure is found. However, forming disulfides with a "random" collection of substrates could be dangerous, since the rate of intramolecular rearrangement would depend on the reactivity of the substrate cysteine with a substrate disulfide. If PDI were to find itself in a covalent complex with a substrate molecule that rearranged slowly, the enzyme could become trapped and unavailable to other substrate molecules. One solution to this problem could be provided by an active site with two cysteines, one to initiate substrate rearrangement and a second to enable an escape pathway. The experiments described below were performed to determine whether escape through the second active-site cysteine is a common event during PDI catalysis. The results suggest that escape is, in fact, the dominant mechanism for PDI-dependent isomerization of scrambled RNase.


EXPERIMENTAL PROCEDURES

Materials

Recombinant rat PDI and its mutants were constructed, expressed, and purified as described previously. Glutathione (GSH), glutathione disulfide (GSSG), bovine pancreatic ribonuclease A (RNase), urea, and dithiothreitol (DTT) were obtained from Sigma. Scrambled RNase was prepared by a minor modification of the method of Hillson et al. (14). RNase (10 mg/ml) was reduced at 37 °C for 2 h with 140 mM DTT in 6 M guanidine hydrochloride, 0.1 M Tris·HCl at pH 8.3. The DTT was removed by ultrafiltration, replacing the reducing buffer with buffered 6 M guanidinium hydrochloride. Oxygen was bubbled into the solution with stirring at room temperature for several days until the residual thiol content was <0.1 mol of SH/mol of RNase. After ultrafiltration to remove the residual guanidinium hydrochloride and to exchange the buffer for 0.1% acetic acid, the scrambled RNase was stored at -80 °C until use.

Activity Measurements

The activity of PDI and its mutants in catalyzing the rearrangement of sRNase was determined with a continuous spectrophotometric assay in which the concentration of active RNase was monitored by refolding in the presence of the substrate, cCMP at 25.0 °C in 0.1 M Tris·HCl, pH 8.0 (6). Assays were performed in duplicate or triplicate, including controls without PDI. Additional controls without cCMP showed that there was no significant absorbance contributed by RNase aggregation (9). After correction of the observed rate of active RNase formation for the uncatalyzed reaction, Km and kcat values were estimated by non-linear least squares fitting to the equation of a rectangular hyperbola. Errors reported are the standard deviations of the parameter estimates. The inhibition constant (Ki) for substrate inhibition was determined by fitting to the equation v = Vmax [S]/([S] + Km + [S]2/Ki) (15). In this case, fitting had to be constrained by fixing Vmax at that observed in the presence of redox buffer. End point assays of the overall yield of the refolding reaction were performed with a discontinuous assay using cCMP as substrate. Percent activity recovered was determined relative to a sample of the same concentration of wild-type RNase.

Other Methods

SDS-PAGE was performed by the method of Laemmli (16). Gels (12.5% polyacrylamide) were stained with Coomassie Brilliant Blue.


RESULTS AND DISCUSSION

When presented with a fully reduced RNase substrate and a glutathione redox buffer, mutants of PDI with a single, reactive cysteine (NSO:COO and NOO:CSO) do not catalyze significant RNase oxidation, but they still catalyze isomerization of the spontaneously formed disulfides at a low, but significant, rate (10-20% of the wild-type kcat) (10). To observe the isomerization reaction more directly without the complication of inefficient catalysis of substrate oxidation, the single-cysteine mutants, NSO:COO and NOO:CSO, were examined for catalysis of the isomerization of scrambled RNase where the substrate is completely oxidized (<0.2 mol of SH/mol of RNase). As with reduced RNase as a substrate, the single-cysteine mutants of PDI exhibit measurable, but low, activity at saturating RNase (Fig. 1). The kcat is only 10-12% that observed for the wild-type enzyme under the same conditions (Table I).


Fig. 1. Catalysis of the rearrangement of scrambled RNase by PDI mutants in the presence of a glutathione redox buffer. Reactions were performed at 25 °C, pH 8.0 (0.1 M Tris·HCl) in the presence of 1 mM GSH and 0.2 mM GSSG using 3.5 µM PDI. PDI mutants and scrambled RNase were prepared as described previously (10). The initial rate of active RNase formation was determined continuously by performing the refolding in the presence of cCMP (6). Individual points are the average of at least two independent determinations of activity. bullet , NOO:CSO; open circle , NSO:COO.
[View Larger Version of this Image (16K GIF file)]


Table I.

Kinetic constants for the rearrangement of scrambled RNase by wild-type PDI and PDI mutants

The velocity of native RNase formation was measured using a continuous spectrophotometric assay in which the formation of active RNase leads to an increase in the rate of cCMP hydrolysis (6). Refolding was performed at pH 8.0, 0.1 M Tris · HCl, 25°C. The PDI concentration was 3.5 µM.


Enzyme kcata Kma Kib Wild-type kcat

min-1 µM µM %
With redox buffer
  NSS:CSS 1.3  ± 0.1 65  ± 7 100
  NSO:COO 0.15  ± 0.03 83  ± 20 12
  NOO:CSO 0.13  ± 0.01 8  ± 5 10
  No PDI 0.0059  ± 0.0001
Without redox buffer
  NSS:CSS 1.3b 44  ± 4 4.2  ± 0.4 100
  NSO:COO 0.009  ± 0.001 32  ± 7 0.7
  NOO:CSO 0.003  ± 0.001 2.2  ± 1.2 0.3
  No PDI (9.2  ± 0.8) × 10-5 min-1

a Km and Vmax values were determined by unweighted, non-linear least squares fitting to the equation for rectangular hyperbola. Errors shown are standard deviations of the fitting parameters.
b In the absence of redox buffer, independent values for Vmax and Km could not be obtained when both parameters were allowed to float. However, at low concentrations of RNase, where substrate inhibition is not significant, the velocity (kcat/Km) is identical to that observed in a redox buffer. Consequently, the Vmax value was fixed at that value observed in redox buffer, and the Km and Ki values were used as fitting parameters.

The relatively low activity of a single-cysteine active site in catalyzing a simple rearrangement reaction is surprising. If the dominant isomerization mechanism involves an intramolecular rearrangement of the substrate induced by the attack of PDI the active-site cysteine of PDI (Scheme I), one active-site cysteine should be sufficient, and the presence or absence of a redox buffer would be expected to have little effect. However, when the glutathione redox buffer is omitted, the turnover numbers for the mutants, NSO:COO and NOO:CSO, are only 0.3-0.7% that of wild-type PDI (Fig. 2 and Table I). However, in the absence of a redox buffer, all (>90%) of the scrambled RNase is eventually converted to native enzyme suggesting that PDI inactivation by oxygen oxidation and the subsequent cessation of rearrangement is not responsible for the low activity. Furthermore, the decreased activity cannot be explained by an effect of mutation on the reactivity of the remaining cysteine, since Darby and Creighton (17) found that the reactivity of the more amino-terminal cysteine in the amino-terminal thioredoxin domain of PDI was decreased only 2-fold by mutation of the more carboxyl-terminal cysteine to alanine. Thus, the simple mechanism of disulfide rearrangement (Scheme I) does not appear to account for PDI catalysis of RNase rearrangements.


Fig. 2. Catalysis of the rearrangement of scrambled RNase by PDI mutants in the absence of a glutathione redox buffer. Reactions were performed at 25 °C, pH 8.0 (0.1 M Tris·HCl) using 3.5 µM PDI. Immediately before use, PDI was reduced with dithiothreitol and isolated by gel filtration (10). The initial rate of active RNase formation was determined continuously by performing the refolding in the presence of cCMP (6). Note the difference in scale compared with Fig. 1. Individual points are the average of at least two independent determinations. open circle , NSO:COO; bullet , NOO:CSO.
[View Larger Version of this Image (17K GIF file)]


If PDI were to indiscriminately scan folding intermediates for reactive disulfides, attack of the more amino-terminal, active-site cysteine on a random substrate disulfide might occasionally engage PDI in a slowly rearranging, covalent complex with substrate. In the wild-type active site, the second cysteine could prevent PDI from becoming trapped in these complexes by expelling the offending substrate, releasing PDI and the substrate for another round of catalysis. According to this model, the low activity of a single-cysteine active site would result from the inability of PDI to escape from such complexes. In such a system, GSH in the redox buffer would provide an intermolecular escape route and stimulate the activity of single-cysteine mutants. In support of this hypothesis, mutants with a single active-site cysteine (NSO:COO and NOO:COS) accumulate as covalent complexes between RNase and PDI, but the wild-type enzyme (NSS:CSS) does not (Fig. 3).


Fig. 3. Covalent complexes between scrambled RNase and PDI. After incubation of the various PDI species (4.0 µM) either alone or with 64 µM sRNase at pH 8.0, 25 °C for 30 min, N-ethylmaleimide (5 mM final concentration) was added to prevent subsequent thiol/disulfide exchange. The samples were displayed on non-reducing SDS-PAGE (12.5% acrylamide) and stained with Coomassie Brilliant Blue. Lane 1, wild-type PDI plus sRNase; lane 2, wild-type PDI; lane 3, NSO:COO plus sRNase; lane 4, NSO:COO; lane 5, NOO:CSO; plus sRNase; lane 6, NOO:CSO lane 7, sRNase; lane 8, low molecular weight standards (Bio-Rad). When 2-mercaptoethanol is included in the sample buffer, no covalent complex is observed, and both RNase and PDI run as single bands (not shown).
[View Larger Version of this Image (46K GIF file)]


Escape from the wild-type enzyme, mediated by the second active-site cysteine, would convert the active-site dithiol to a disulfide and release the RNase substrate with a reduced disulfide (Scheme II). In the absence of a redox buffer, where there is no alternative source of oxidizing or reducing equivalents, the oxidized PDI would have to find a reduced RNase substrate and reoxidize it to complete the reaction cycle. At higher concentrations of scrambled RNase this should become increasingly difficult, leading to substrate inhibition. As predicted, challenging reduced, wild-type PDI with higher and higher concentrations of sRNase, produces significant substrate inhibition if there is no redox buffer present (Fig. 4). However, if a redox buffer is available to recycle the oxidized PDI back to the reduced state and provide GSSG to oxidize the substrate, RNase rearrangements are not subject to inhibition by high scrambled RNase concentrations.


Scheme II. Reductive escape and reoxidation as an alternative mechanism for PDI-catalyzed rearrangement of substrate disulfides.
[View Larger Version of this Image (14K GIF file)]



Fig. 4. The effect of redox buffer on the catalysis of the rearrangement of scrambled RNase by wild-type PDI. Reactions were performed at 25 °C, pH 8.0 (0.1 M Tris·HCl) in the presence of 1 mM GSH and 0.2 mM GSSG using 3.5 µM PDI. The initial rate of active RNase formation was determined continuously by performing refolding in the presence of cCMP. Individual points are the average of at least two independent determinations of activity. bullet , with redox buffer; open circle , without redox buffer.
[View Larger Version of this Image (19K GIF file)]


During catalysis of substrate isomerization, reduced PDI scans the unfolded protein for a reactive disulfide. When the cysteine of one of the active sites of PDI attacks a substrate disulfide, the covalent intermediate that is formed can suffer three alternative fates (Scheme II). The substrate cysteine that has just been released from a disulfide bond scans the immediate environment for any disulfide that can be attacked. If it attacks the PDI-substrate disulfide, the disulfide that was originally present in the substrate simply forms again, and the substrate is released without rearrangement (Scheme II, path 3). Alternatively, rearrangement occurs when the substrate thiol attacks a sulfur that is different from the one to which it was originally paired (Scheme II, path 1). Scanning the substrate favors the attack of PDI on the most reactive disulfide, and subsequent scanning by the substrate thiol is biased toward replacing the more reactive disulfide with one that is kinetically less reactive (Scheme II, path 1). For reversible thiol/disulfide exchange, the kinetics of disulfide formation may parallel thermodynamic stability (18, 19); consequently, scanning by PDI would favor the more stable arrangement of disulfides.

The second (more carboxyl-terminal) cysteine at each of the active sites of PDI provides yet another choice to be made in the covalent enzyme-substrate complex. Superimposed on the bias toward substrate rearrangements that lead to increasing disulfide stability, the active-site cysteine provides a reductive escape mechanism to extricate the enzyme from slowly rearranging complexes. If scanning by the substrate thiol is too slow, the second active-site cysteine can discharge the substrate by an intramolecular reaction between the two active-site sulfurs (Scheme II, path 2).

Intramolecular disulfide formation at the active site provides a clock to ensure that the enzyme does not become trapped in slowly rearranging, covalent complexes. The disulfide between glutathione and PDI is displaced by an intramolecular reaction between the two active-site cysteines with a half-time estimated at 20-100 ms (20). Thus, if kinetic scanning by the substrate does not succeed rapidly, wild-type PDI can escape from the complex by discharging the substrate. This escape mechanism provides another advantage; the offending substrate disulfide is reduced, which may make additional rearrangement/reoxidation pathways more accessible. The very low activity that is observed for single-cysteine active sites when intramolecular and intermolecular escape pathways are not available suggests that the dominant pathway for PDI-catalyzed rearrangement of sRNase involves reductive escape and reoxidation of the substrate rather than direct isomerization. The decision among reversal, rearrangement, and escape must be made multiple times, each time a covalent PDI-substrate complex is formed. Multiple cycles of kinetic scanning and reductive escape will ultimately lead to a native disulfide arrangement that is either resistant to the attack of PDI or in which partitioning is always toward reforming the original disulfide (Scheme II, path 3).


FOOTNOTES

*   This work was supported by National Institutes of Health Grant GM-40379.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: Dept. of Biochemistry, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030. Tel.: 713-798-5880; Fax: 713-796-9438; E-mail:hgilbert{at}bcm.tmc.edu.
1   The abbreviations used are: PDI, recombinant, rat protein-disulfide isomerase; DTT, dithiothreitol; RNase, bovine pancreatic ribonuclease A; sRNase, scrambled RNase; PAGE, polyacrylamide gel electrophoresis.
2   The nomenclature for describing mutations of the four active-site cysteines of PDI uses S to represent a sulfur nucleophile (cysteine) at a given active-site position and O to represent an oxygen nucleophile (serine) at this position. The two thioredoxin domains of the enzyme are designated the amino-terminal domain (N) and the carboxyl-terminal domain (C). Thus, the wild-type enzyme would be designated NSS:CSS. A mutant PDI with a single cysteine at the more amino-terminal position in the more carboxyl-terminal thioredoxin domain would be designated NOO:CSO.

REFERENCES

  1. Goldberger, R. F., Epstein, C. J., and Anfinsen, C. B. (1963) J. Biol. Chem. 238, 628-635 [Free Full Text]
  2. Noiva, R. (1994) Protein Expression Purif. 5, 1-13 [CrossRef][Medline] [Order article via Infotrieve]
  3. Goldenberg, D. P. (1992) Trends Biochem. Sci. 17, 257-261 [CrossRef][Medline] [Order article via Infotrieve]
  4. Creighton, T. E. (1984) J. Mol. Biol. 129, 411-4315
  5. Weissman, J. S., and Kim, P. S. (1993) Nature 365, 185-188 [CrossRef][Medline] [Order article via Infotrieve]
  6. Lyles, M. M., and Gilbert, H. F. (1991) Biochemistry 30, 613-619 [Medline] [Order article via Infotrieve]
  7. Edman, J. C., Ellis, L., Blacher, R. W., Roth, R. A., and Rutter, W. J. (1985) Nature 317, 267-270 [Medline] [Order article via Infotrieve]
  8. Vuori, K., Myllyla, R., Pihlajaniemi, T., and Kivirikko, K. I. (1992) J. Biol. Chem. 267, 7211-7214 [Abstract/Free Full Text]
  9. Lyles, M. M., and Gilbert, H. F. (1994) J. Biol. Chem. 269, 30946-30952 [Abstract/Free Full Text]
  10. Walker, K. W., Lyles, M. M., and Gilbert, H. F. (1996) Biochemistry 35, 1972-1980 [CrossRef][Medline] [Order article via Infotrieve]
  11. LaMantia, M. L., and Lennarz, W. J. (1993) Cell 74, 899-908 [Medline] [Order article via Infotrieve]
  12. Laboissière, M. C. A., Sturley, S. L., and Raines, R. T (1995) J. Biol. Chem. 270, 28006-28009 [Abstract/Free Full Text]
  13. Saxena, V. P., and Wetlaufer, D. B. (1970) Biochemistry 9, 5015-5022 [Medline] [Order article via Infotrieve]
  14. Hillson, D. A., Lambert, N., and Freedman, R. B. (1984) Methods Enzymol. 107, 281-294 [Medline] [Order article via Infotrieve]
  15. Plowman, K. M. (1972) Enzyme Kinetics, p. 98, McGraw-Hill, New York
  16. Laemmli, U. K. (1970) Nature 227, 680-685 [Medline] [Order article via Infotrieve]
  17. Darby, N. J., and Creighton, T. E. (1995) Biochemistry 34, 16770-16780 [Medline] [Order article via Infotrieve]
  18. Gilbert, H. F. (1990) Adv. Enzymol. 63, 69-172 [Medline] [Order article via Infotrieve]
  19. Creighton, T. E. (1984) Methods Enzymol. 107, 305-329 [Medline] [Order article via Infotrieve]
  20. Darby, N. J., and Creighton, T. E. (1994) Biochemistry 34, 11725-11735

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