A Mutant Truncated Protein Disulfide Isomerase with No Chaperone Activity*

A mutant human protein disulfide isomerase with the COOH-terminal 51 amino acid residues deleted (abb′a′) has been expressed in Escherichia coli. Its secondary structures are very similar to those of the native bovine enzyme. The mutant enzyme shows neither peptide binding ability nor chaperone activity in assisting the refolding of denaturedd-glyceraldehyde-3-phosphate dehydrogenase but keeps most of the catalytic activities for reduction of insulin and isomerization of scrambled ribonuclease. It assists the reactivation of denatured and reduced proteins containing disulfide bonds, acid phospholipase A2, and lysozyme to different levels, which are significantly lower than those by the native bovine enzyme.

danese (20) as well as disulfide-containing proteins such as lysozyme (21) and acidic phospholipase A 2 (APLA 2 ) (22), although Lilie et al. (23) had reported that PDI showed no chaperone effect on the refolding of denatured immunoglobulin Fab with intact disulfides. A PDI mutant, inactive as an isomerase, has the same function as an essential subunit for the assembly of the fully active tetramer of prolyl-4-hydroxylase ␣ 2 ␤ 2 (24) and the dimer of microsomal triglyceride transfer protein complex (25), suggesting that the role of PDI in the above two proteins is independent of its isomerase activity but related to its chaperone-like peptide binding function. The deletion of the NH 2 -terminal 3 amino acid residues of the peptide binding region of human PDI indeed prevents prolyl-4-hydroxylase tetramer formation (25,26). It was shown that a mutant PDI with Ser substituted for Cys at the -CGHC-active sites and devoid of isomerase activity can increase the folding and secretion of lysozyme coexpressed in yeast (27). The yeast bearing a mutant PDI shortened from the COOH terminus and devoid of the putative peptide binding region in the middle part of the molecule can hardly survive (28). It has been shown in our previous work that the chaperone activity of PDI in assisting GAPDH refolding is suppressed by competitive peptide binding (29), which also inhibits its enzymatic activity (12,29).
In this paper a mutant human PDI, abbЈaЈ, with the deletion of the COOH-terminal 51 amino acid residues responsible for peptide binding, has been expressed in Escherichia coli. The mutant shows neither peptide binding ability nor chaperone activity in assisting the refolding of denatured GAPDH but displays most of the catalytic activities of PDI and assists the reactivation of denatured and reduced APLA 2 and lysozyme to an extent significantly lower than those by the native bovine enzyme.

EXPERIMENTAL PROCEDURES
Materials-The plasmid pBR322-PDI, containing the full-length of human PDI (hPDI) cDNA (11), is a generous gift from Prof. K. Kivirikko, University of Oulu, Finland. GAPDH was from rabbit muscle (30). APLA 2 was from Agkistrodon blomhoffii brevicaudus (Agkistrodon halys Pallas) (31). PDI was prepared from bovine liver (bPDI) essentially according to Lambert and Freedman (32) and showed one band on SDS-polyacrylamide gel electrophoresis with a specific activity of more than 800 units/g. S-Carboxymethylated A-chain of insulin was prepared according to Zheng et al. (33). Modified PDI (mPDI) carboxymethylated at thiols in the -CGHC-sequence of active sites, was prepared as described previously (29).
Restriction endonucleases, T4 DNA ligase, isopropyl 1-thio-␤-D-galactopyranoside, and dithiothreitol were from Promega. Vent R DNA polymerase and the large (Klenow) fragment of DNA polymerase I were from New England Biolabs, Inc. The prokaryotic gene fusion expression vector pGEX-4T-1, E. coli BL21 strain (F Ϫ , ompT, R B Ϫ , mB Ϫ ), the T7 sequencing kit, and glutathione-Sepharose 4B were from Pharmacia Biotech Inc. ␣-35 S-dATP was obtained from NEN Life Science Products. 5,5Ј-Dithiobis(2-nitrobenzoic acid) was from Fluka. Glutathione (GSH), glutathione disulfide (GSSG), NAD ϩ (98%), and NADPH (type III) were from Boehringer Mannheim. Thrombin, glutathione reductase (yeast, type III), glyceraldehyde 3-phosphate, 8-anilino-1-naphthalenesulfonic acid (ANS), bovine serum albumin (BSA, 98 -99% albumin, fraction V), * This work was supported by the Pandeng Project of the Chinese Commission of Science and Technology and Grant 39470163 from the China Natural Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Determinations-The concentrations of GdnHCl-denatured GAPDH, and denatured and reduced APLA 2 were determined by the method of Bradford (34) with BSA as a standard. The concentrations of other proteins were determined spectrophotometrically at 280 nm with the following absorption coefficients (A 1 cm 0.1% ): 0.94 for PDI, 0.66 for BSA, 0.98 for GAPDH, 1.03 for insulin, 1.3 for APLA 2 , 1.14 for S-carboxymethylated A-chain of insulin (35), 2.63 for native lysozyme, and 2.37 for denatured lysozyme (21). The A 1 cm 0.1% value of abbЈaЈ was determined to be 0.83. For the convenience of comparison, both GAPDH and PDI are considered as protomers in the calculation of molar ratios. Thiol groups of proteins were determined with 5,5Ј-dithiobis(2-nitrobenzoic acid) (36).
Construction of the Expression Plasmid for abbЈaЈ-The primer I (5Ј-CGGGATCCGACGCCCCCGAGGAG-3Ј) was designed to hybridize with the first 15 nucleotides at the 5Ј-terminus of the hPDI cDNA sequence and contains a BamHI site (underlined) just before the sequence. The reverse primer II (5Ј-GGGTTAGTAATCGATGAC-3Ј) with a stop codon (underlined) hybridizes with the sequence between 1309 and 1320 base pairs of the hPDI cDNA. The double-stranded 1.3kilobase DNA fragment coding for the sequence of the Asp 1 to Tyr 440 of PDI (abbЈaЈ) was generated by polymerase chain reaction using Vent R DNA polymerase with the above two primers and pBR322-PDI as a template. The polymerase chain reaction product was elongated to blunt ends in both termini with the large (Klenow) fragment of DNA polymerase I, digested with BamHI at the 5Ј-terminus, and ligated into pGEX-4T-1 digested with BamHI and SmaI in-frame with the glutathione S-transferase fusion codons to construct the expression plasmid pGEX-abbЈaЈ. The foreign DNA sequences cloned into the expression plasmid was verified by nucleotide sequencing.
Gene Expression and Purification of abbЈaЈ-Protein production was carried out in E. coli strain BL21 containing the pGEX-abbЈaЈ plasmid. Cells grown overnight at 37°C in LB medium were diluted 100-fold and grown at 28°C with vigorous shaking to an A 600 of 0.8, and then grown for another 4 h after adding 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside. Cell pellet was disrupted by sonication and then mixed gently with 1% Triton X-100 by stirring for 30 min at room temperature. The supernatant of cell lysate was loaded onto a 2-ml glutathione-Sepharose 4B RediPack column, and the fusion protein on the matrix was cleaved with 100 units of thrombin at 22°C for 16 h. The eluted abbЈaЈ fraction was loaded onto a Mono Q HR (5/5) column equilibrated with 20 mM phosphate buffer, pH 6.3, and eluted with a linear gradient from 0 to 0.3 M NaCl. The abbЈaЈ fraction was identified by 10% SDS-polyacrylamide gel electrophoresis analysis, dialyzed thoroughly, and lyophilized.
Activity Assay-The disulfide isomerase and thiol-protein oxidoreductase (TPOR) activities were assayed according to Lambert and Freedman (32). Lysozyme activity was determined at 30°C by following the absorbance decrease at 450 nm of a 0.25 mg/ml M. lysodeikticus suspension in 67 mM sodium phosphate buffer, pH 6.2, containing 100 mM NaCl (21,37).
Denaturation and Renaturation of Enzymes-Denaturation and assisted reactivation of GAPDH and APLA 2 by bPDI and/or abbЈaЈ were carried out according to Cai et al. (19) and Yao et al. (22), respectively. Lysozyme at 20 mg/ml was completely denatured and reduced in 0.1 M sodium phosphate, pH 8.0, containing 8 M GdnHCl and 150 mM dithiothreitol at room temperature for 4 h. The reaction mixture was brought to pH 2.0 with 6 N HCl, dialyzed first against 0.01 N HCl for 3 h and then against 0.1 N acetic acid at 4°C thoroughly. The denatured and reduced lysozyme was aliquoted and stored at Ϫ20°C. Oxidative refolding of reduced and denatured lysozyme was carried out by dilution into phosphate buffer containing 0.1 M sodium phosphate, 2 mM EDTA, 1 mM GSSG, and 2 mM GSH, pH 7.5, to a final concentration of 10 M. The activity recovery was completed and determined 2 h after dilution.
Circular Dichroism (CD) and ANS Fluorescence Spectra Analysis-CD spectrum determinations in the far ultraviolet region from 200 to 250 nm were carried out with a Jasco J720 spectropolarimeter at 25°C. ANS fluorescence spectra were measured at 25°C in a Hitachi F-4010 spectrofluorometer with an excitation wavelength of 365 nm.

RESULTS
Characterization of the Expressed abbЈaЈ-The inserted coding sequence for abbЈaЈ in the expression plasmid pGEX-abbЈaЈ was verified by DNA sequence analysis (data not shown). The constructed protein should contain residues 1-440 and an ad-ditional Gly-Ser at the NH 2 terminus of hPDI. Experimental conditions for sonication, growth temperature, and isopropyl 1-thio-␤-D-galactopyranoside concentration were optimized by 10% SDS-polyacrylamide gel electrophoresis analysis (data not shown). Fig. 1 shows the high yield of abbЈaЈ expression and the homogeneity of the purified product. The molecular weight of abbЈaЈ is about 48,000 as expected.
Secondary Structure of abbЈaЈ-As shown in Fig. 2, the CD spectrum of abbЈaЈ is almost the same as that of bPDI, indicating that the COOH-terminal shortened enzyme has secondary structures closely similar to that of bPDI. The same ANS fluorescence spectra of bPDI and abbЈaЈ indicated that the truncation of COOH-terminal 51 residues of PDI has little effect on the surface hydrophobicity of the intact molecule (data not shown).
Enzymatic Activities of abbЈaЈ-The mutant abbЈaЈ has about 69 Ϯ 3% of isomerase activity and 80 Ϯ 2% of TPOR activity compared with that of bPDI (Table I), suggesting that the truncation of the COOH-terminal 51 amino acid residues of PDI has only a minor effect on these enzymatic activities.
Effects of S-Carboxymethylated A-chain of Insulin on the TPOR Activity of abbЈaЈ-As shown in Fig. 3, the presence of S-carboxymethylated A-chain of insulin inhibits the TPOR activity of PDI; in sharp contrast, it has no effect on the TPOR activity of abbЈaЈ. The above result is highly suggestive that the S-carboxymethylated A-chain of insulin inhibits the TPOR activity of PDI by binding at its peptide binding site, which is lacking in the mutant abbЈaЈ.   Fig.  4 shows that the reactivation of GAPDH in the presence of bPDI increases from 4 to 19% with the increase of the molar ratio of PDI to GAPDH from 0 to 10. In contrast, with BSA used for comparison, abbЈaЈ at the same range of ratios shows no effect on the reactivation of denatured GAPDH. This result indicates the necessity of the peptide binding site of PDI for its chaperone activity in the reactivation of GAPDH.

Effects of abbЈaЈ on Reactivation of Denatured GAPDH-
Effects of abbЈaЈ on the Reactivation of GdnHCl-denatured and Reduced APLA 2 -It has been proposed that the foldase activity of PDI consists of both its isomerase and chaperone activities and the latter activity can be fully replaced by mPDI (22), which is devoid of isomerase activity but nearly as active as native PDI in its chaperone activity in assisting the reactivation of GAPDH (29). The reactivation of GdnHCl-denatured and reduced APLA 2 , containing seven disulfide bonds, upon dilution in the presence and absence of abbЈaЈ and/or mPDI was determined to examine the foldase activity of abbЈaЈ. Fig. 5A shows that the spontaneous reactivation of APLA 2 at 12 M is only about 1%, and the reactivation in the presence of abbЈaЈ takes 10 h to reach completion as in the presence of PDI. In the presence of abbЈaЈ the reactivation yield increases with increasing concentrations of abbЈaЈ to a maximal level of 15% when the molar ratio of abbЈaЈ to APLA 2 approaches 2 as shown in Fig.  5B. Higher ratios of abbЈaЈ has little further effect on the reactivation of APLA 2 . mPDI alone has no effect on the reactivation of APLA 2 ; however, the simultaneous presence of both mPDI and abbЈaЈ increases markedly the reactivation yield of APLA 2 compared with the same amount of abbЈaЈ alone in the refolding buffer. In presence of mPDI at 36 M together with abbЈaЈ at 120 M the reactivation yield of APLA 2 approaches 48%, about the same as the maximal level obtainable by native bPDI. Further increases in the concentrations of either mPDI or abbЈaЈ or both have no further effect on the reactivation of APLA 2 .
Effects of abbЈaЈ on the Refolding of Denatured and Reduced Lysozyme-As shown in Fig. 6 the very low spontaneous reactivation of denatured and reduced lysozyme in phosphate buffer increases with increasing concentrations of bPDI in the refolding buffer to a maximal level of around 71% at a stoichiometric amount of PDI. However, the maximal reactivation yield of lysozyme in presence of abbЈaЈ is reduced to 58% under the same condition. DISCUSSION From the nearly identical CD spectra and ANS binding, in addition to similar enzyme activities between bPDI and abbЈaЈ, it appears that the addition of 2 extra residues at the NH 2 terminus and the deletion of the COOH-terminal 51 amino acid residues show little effect on the secondary structures, the surface hydrophobicity, and the enzymatic activity of PDI. The NH 2 -terminal sequence of PDI seems to be of little importance as it has also been found that even an extension of 10 residues at the NH 2 terminus showed no deleterious effect on the properties of PDI (38). The mutant abbЈaЈ is structurally stable and retains most of the isomerase and oxidoreductase activities of PDI, suggesting that it is more or less independent from the c domain, but the involvement of the c domain in the above activities cannot be excluded, apart from the peptide binding site; it could also contribute to substrate binding. Alternatively, deletion of the c domain could alter somewhat the functional conformation of PDI, leading to decreased enzymatic activities. The peptide inhibition of the TPOR activity of PDI suggests its binding at the peptide binding site of PDI (12,29,33). In this respect, it is not surprising that the TPOR activity of abbЈaЈ is not inhibited by the presence of an excess of the peptide as abbЈaЈ no longer has the c domain, which is the major if not the only peptide binding site (11). The above results indicate that, contrary to the previous suggestion (29), the c domain does not appear to be essential for the enzymatic activities of PDI, and the peptide binding site and the substrate binding site are not identical. However, the above does not exclude the possibility  that occupation at the peptide binding site could interfere with substrate binding to the substrate binding site, and the c domain could partially contribute to the substrate binding site, thus the peptide and substrate binding sites could be close to and overlapping with each other. This appears to be different from the E. coli trigger factor, in which the active site of peptidyl-prolyl cis/trans-isomerase is separated widely enough from the peptide binding site so that the binding of an unfolded protein does not interfere with the catalysis of prolyl cis/trans isomerization in a small peptide (39).
As shown previously, the presence of a peptide in the refolding buffer suppresses GAPDH reactivation assisted by PDI, indicating that peptide binding in competition with the GAPDH folding intermediates prevents and suppresses the PDI-assisted refolding of GAPDH (29). In this respect, it is to be expected and is actually found that abbЈaЈ is unable to assist the reactivation of denatured GAPDH and thus provides a straightforward demonstration that the peptide binding site at the c domain of PDI is directly responsible for its chaperone activity.
It is widely accepted that PDI plays a critical role in nascent peptide folding by catalyzing the formation of native disulfide(s). However, PDI as a foldase not only catalyzes disulfide isomerization but is also intrinsically involved in peptide chain folding through its peptide binding site(s). Therefore PDI most likely binds with folding intermediates at different folding stages (see Scheme 1). The two processes of disulfide formation and peptide folding are intimately interdependent and work in cooperation in the generation of the native conformation of disulfide-containing proteins. The chemically modified PDI alkylated at active site cysteine residues is devoid of isomerase activity but retains its chaperone activity almost fully (29). In this report, the COOH-terminal truncated PDI shows most of the catalytic activity but is devoid of any chaperone activity. Comparisons of the properties and the possible roles of these derivatives in assisting protein folding are summarized in Table I and Scheme 1, respectively.
For spontaneous refolding (the central lines in a, b, and c) the denatured and reduced protein (U) would undergo a fast conformational change upon dilution to form intermediate (I 1 ), which could fold first to I 2 and finally to native molecule (N) through both folding steps and oxidative disulfide formation. Both intermediates I 1 and I 2 face two alternative folding pathways, correct folding to the native molecule (N) and misfolding leading to aggregation (A). The presence of PDI and/or its derivatives affect the relative proportion between the alternative pathways. As shown in Scheme 1a, PDI, being a chaperone, binds with I 1 at its peptide binding site of domain c and prevents the aggregation of I 1 . On the other hand, PDI also probably binds at the substrate binding site with I 2 , which is assumed to represents a folding intermediate better folded than I 1 with thiols properly paired to be oxidized to the native disulfide(s) and also decrease aggregation and increase reactivation. As shown in Scheme 1b, abbЈaЈ, devoid of a peptide binding site, no longer binds with I 1 but binds indeed with I 2 through its substrate binding sites and catalyzes disulfide formation, leading to the native molecule (N) with significantly lower efficiency than PDI does. This could also explain the fact that the reactivation of APLA 2 assisted by abbЈaЈ at all concentrations shows a lag phase for the first 30 min (Fig. 5A) in contrast to the apparent first order kinetics of PDI-assisted reactivation (22). Moreover, abbЈaЈ does not show anti-chaperone activity of PDI in assisting the reactivation of lysozyme in HEPES buffer. 2 As shown in Scheme 1c, mPDI, with full peptide binding activity of PDI, binds with both I 1 and I 2 , but without isomerase activity it is unable to catalyze the formation of disulfide and hence the native molecule from I 2 . After dissociated from the respective complexes with mPDI, the intermediate would aggregate irreversibly, and therefore mPDI does not increase the reactivation yield. However, with both abbЈaЈ (as enzyme only) and mPDI (as chaperone only) simultaneously present in the refolding buffer, the cooperative action of the two PDI derivatives could assist the reactivation of APLA 2 to the same maximal level as native PDI does, but the amount of abbЈaЈ required in this case for the maximal reactivation is much more than by native PDI alone. This could be accounted for by a higher efficiency of PDI with both functions in the same molecule compared with the combined action of two molecules in abbЈaЈ and mPDI. It is also significant that compared with thioredoxin, PDI acquired an additional -CXXCcontaining domain, domain aЈ, as well as the domain c during evolution so as to function as a foldase with both isomerase and chaperone activities, whereas thioredoxin shows neither peptide binding ability (3,33) nor chaperone activity (29) and has a much lower isomerase activity than PDI (2). Similarly, the trigger factor composed of peptidyl-prolyl cis/trans-isomerase and peptide binding sites has an efficiency for assisting protein folding 1,000-fold higher than that of peptidyl-prolyl cis/transisomerase (39).
As a foldase, PDI assists folding of different target proteins containing disulfide bond with efficiency different from that shown for APLA 2 (from 1 to 45%) and lysozyme (from 1 to 71%). This can be explained by the difference in the folding pathway and properties of folding intermediates of the substrates on the one hand and in the specificity of PDI itself as a chaperone or an isomerase on the other. For APLA 2 , the maximal reactivation level decreased from 45 to 15% when PDI was replaced by abbЈaЈ, whereas for lysozyme, the level decreased from 71 to 58%. It appears that for APLA 2 refolding the chaperone function of PDI is more important compared with the refolding of lysozyme.