Portal de Periódicos da CAPES

Probing the Folding Pathways of Long R3Insulin-like Growth Factor-I (LR3IGF-I) and IGF-I via Capture and Identification of Disulfide Intermediates by Cyanylation Methodology and Mass Spectrometry

1999; Elsevier BV; Volume: 274; Issue: 53 Linguagem: Inglês

10.1074/jbc.274.53.37598

1083-351X

Ying Yang, Jiang Wu, J. Throck Watson,

Cancer, Hypoxia, and Metabolism

This report describes an integrated investigation of the refolding and reductive unfolding of insulin-like growth factor (IGF-I) and its variant, long R3 IGF-I (LR3IGF-I), which has a Glu3 to Arg3 substitution and a hydrophobic 13-amino acid N-terminal extension. The refolding performed in glutathione redox buffer was quenched at different time points by adjusting the pH to 2.0–3.0 with a 1 n HCl solution of 1-cyano-4-dimethylaminopyridinium tetrafluoroborate, which trapped intermediates via cyanylation of free sulfhydryl groups. The disulfide structure of the intermediates was determined by chemical cleavage followed by mass mapping with mass spectrometry. Six refolding intermediates of IGF-I and three refolding intermediates of LR3IGF-I were isolated and characterized. Folding pathways of IGF-I and LR3IGF-I are proposed based on the time-dependent distribution and disulfide structure of the corresponding trapped intermediates. Similarities and differences in the refolding behavior of IGF-I and LR3IGF-I are discussed. This report describes an integrated investigation of the refolding and reductive unfolding of insulin-like growth factor (IGF-I) and its variant, long R3 IGF-I (LR3IGF-I), which has a Glu3 to Arg3 substitution and a hydrophobic 13-amino acid N-terminal extension. The refolding performed in glutathione redox buffer was quenched at different time points by adjusting the pH to 2.0–3.0 with a 1 n HCl solution of 1-cyano-4-dimethylaminopyridinium tetrafluoroborate, which trapped intermediates via cyanylation of free sulfhydryl groups. The disulfide structure of the intermediates was determined by chemical cleavage followed by mass mapping with mass spectrometry. Six refolding intermediates of IGF-I and three refolding intermediates of LR3IGF-I were isolated and characterized. Folding pathways of IGF-I and LR3IGF-I are proposed based on the time-dependent distribution and disulfide structure of the corresponding trapped intermediates. Similarities and differences in the refolding behavior of IGF-I and LR3IGF-I are discussed. 1-cyano-4-dimethylaminopyridinium insulin-like growth factor long R3 IGF-I high pressure liquid chromatography matrix-assisted laser desorption-ionization Considerable insight into the folding and unfolding pathways of a protein can be obtained from trapping and characterizing intermediates involved in the dynamic process (1Kim P.S. Baldwin R.L. Annu. Rev. Biochem. 1990; 59: 631-660Crossref PubMed Scopus (1203) Google Scholar, 2Matthews C.R. Annu. Rev. Biochem. 1993; 62: 653-683Crossref PubMed Scopus (455) Google Scholar, 3Ptitsyn O.B. Curr. Opin. Struct. Biol. 1995; 5: 74-78Crossref PubMed Scopus (225) Google Scholar, 4Li Y.J. Rothwarf D.M. Scheraga H.A. Nat. Struct. Biol. 1995; 2: 489-494Crossref PubMed Scopus (94) Google Scholar). Disulfide-containing proteins provide an opportunity to capture thiol-containing intermediates by chemical reaction during the time course of folding or unfolding (5Creighton T.E. J. Mol. Biol. 1974; 87: 563-577Crossref PubMed Scopus (89) Google Scholar, 6Creighton T.E. Goldenberg D.P. J. Mol. Biol. 1984; 179: 497-526Crossref PubMed Scopus (262) Google Scholar, 7Scheraga H.A. Konishi Y. Ooi T. Adv. Biophys. 1984; 18: 21-41Crossref PubMed Scopus (55) Google Scholar, 8Weissman J.S. Kim P.S. Science. 1991; 253: 1386-1393Crossref PubMed Scopus (492) Google Scholar). The detailed folding pathways of several proteins (9Creighton T.E. Ewbank J.J. Biochemistry. 1994; 33: 1534-1538Crossref PubMed Scopus (53) Google Scholar, 10Chang J.Y. Biochemistry. 1996; 35: 11702-11709Crossref PubMed Scopus (63) Google Scholar, 11Glocker M.O. Arbogast B. Milley R. Cowgill C. Deinzer M.L. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 5868-5872Crossref PubMed Scopus (25) Google Scholar) have been studied in this way, including bovine pancreatic trypsin inhibitor (6Creighton T.E. Goldenberg D.P. J. Mol. Biol. 1984; 179: 497-526Crossref PubMed Scopus (262) Google Scholar, 8Weissman J.S. Kim P.S. Science. 1991; 253: 1386-1393Crossref PubMed Scopus (492) Google Scholar, 12Darby N.J. Norin P.E. Talbo G. Creighton T.E. J. Mol. Biol. 1995; 249: 463-477Crossref PubMed Scopus (74) Google Scholar, 13Weissman J.S. Kim P.S. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 9900-9904Crossref PubMed Scopus (97) Google Scholar) and ribonuclease A (14Xu X. Rothwarf D.M. Scheraga H.A. Biochemistry. 1996; 35: 6406-6417Crossref PubMed Scopus (83) Google Scholar, 15Rothwarf D.M. Scheraga H.A. Biochemistry. 1993; 32: 2671-2679Crossref PubMed Scopus (140) Google Scholar, 16Rothwarf D.M. Scheraga H.A. Biochemistry. 1993; 32: 2680-2689Crossref PubMed Scopus (65) Google Scholar, 17Rothwarf D.M. Scheraga H.A. Biochemistry. 1993; 32: 2690-2697Crossref PubMed Scopus (57) Google Scholar, 18Rothwarf D.M. Scheraga H.A. Biochemistry. 1993; 32: 2698-2703Crossref PubMed Scopus (50) Google Scholar). To isolate and characterize intermediates that are involved in folding or unfolding of proteins containing disulfide bonds, it is necessary to stop thiol/disulfide exchange reactions that convert one intermediate to another. One traditional approach involves trapping of thiol groups irreversibly by alkylation with iodoacetate, iodoacetamide, or vinylpyridine under alkaline conditions. However, rearrangement of intermediates during the trapping procedure with iodoacetate has been observed for both bovine pancreatic trypsin inhibitor (8Weissman J.S. Kim P.S. Science. 1991; 253: 1386-1393Crossref PubMed Scopus (492) Google Scholar) and ribonuclease A (19Rothwarf D.M. Scheraga H.A. J. Am. Chem. Soc. 1991; 113: 6293-6294Crossref Scopus (47) Google Scholar, 20Creighton T.E. BioEssays. 1992; 14: 195Crossref PubMed Scopus (53) Google Scholar). Another traditional method quenches thiol/disulfide exchange by lowering the pH to ≤2 by acidification. An advantage of acid quenching is its reversibility; intermediates trapped at low pH can be isolated and allowed to undergo further rearrangement or refolding after readjusting the pH in experiments designed to more completely characterize particular pathways (8Weissman J.S. Kim P.S. Science. 1991; 253: 1386-1393Crossref PubMed Scopus (492) Google Scholar, 10Chang J.Y. Biochemistry. 1996; 35: 11702-11709Crossref PubMed Scopus (63) Google Scholar). While quenching by acidification occurs at a diffusion-controlled rate, it does not completely stop thiol/disulfide exchange (21Pain R.H. Mechanisms of Protein Folding. Oxford University Press, New York1994: 112Google Scholar); further chemical modification is also necessary for the structural characterization of an acid-trapped intermediate. We recently developed a methodology to trap folding intermediates based on the cyanylation of thiol groups by 1-cyano-4-dimethylaminopyridinium (CDAP)1 tetrafluoroborate under acidic conditions (22Wu J. Yang Y. Watson J.T. Protein Sci. 1998; 7: 1017-1028Crossref PubMed Scopus (55) Google Scholar). This approach has several unique advantages. First, cyanylation of thiol groups in acidic solution quenches the refolding process and minimizes thiol/disulfide exchange. Second, cyanylation of thiol is already part of our procedure for structural elucidation of the intermediates, which involves partial reduction, cyanylation, chemical cleavage, and mass mapping (22Wu J. Yang Y. Watson J.T. Protein Sci. 1998; 7: 1017-1028Crossref PubMed Scopus (55) Google Scholar, 23Wu J. Watson J.T. Protein Sci. 1997; 6: 391-398Crossref PubMed Scopus (160) Google Scholar). Third, this methodology is fast, simple, and even applicable to disulfide structural analysis of proteins containing adjacent cysteines (24Yang Y. Wu J. Watson J.T. J. Am. Chem. Soc. 1998; 123: 5834-5835Crossref Scopus (24) Google Scholar). Insulin-like growth factor-I (IGF-I) (25Rinderknecht E. Humbel R.E. J. Biol. Chem. 1978; 253: 2769-2776Abstract Full Text PDF PubMed Google Scholar) is a single-chain polypeptide of 70 residues containing three intramolecular disulfide bonds, two of which involve adjacent cysteines (Fig. 1). IGF-I is postulated to be the mediator of growth hormone action on skeletal tissue as well as of mitogenic activity on several cell types (26Froesch E.R. Burgi H. Ramseier E.B. Bally P. Labhart A. J. Clin. Invest. 1963; 42: 1816-1834Crossref PubMed Scopus (204) Google Scholar, 27Rinderknecht E. Humbel R.E. Proc. Natl. Acad. Sci. U. S. A. 1976; 73: 2365-2369Crossref PubMed Scopus (249) Google Scholar). The refolding of IGF-I has been studied by several groups (28Hober S. Forsberg G. Palm G. Hartmanis M. Nilsson B. Biochemistry. 1992; 31: 1749-1756Crossref PubMed Scopus (104) Google Scholar, 29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar, 31Rosenfeld R.D. Miller J.A. Narhi L.O. Hawkins N. Katta V. Lauren S. Weiss M.A. Arakawa T. Arch. Biochem. Biophys. 1997; 342: 298-305Crossref PubMed Scopus (33) Google Scholar, 32Hober S. Uhlen M. Nilsson B. Biochemistry. 1997; 36: 4616-4622Crossref PubMed Scopus (31) Google Scholar). Hober et al. (28Hober S. Forsberg G. Palm G. Hartmanis M. Nilsson B. Biochemistry. 1992; 31: 1749-1756Crossref PubMed Scopus (104) Google Scholar) proposed a folding pathway for IGF-I, based on the structural analysis of refolding intermediates trapped by pyridylethylation at pH 8.7. Four intermediates were identified, including a native-like one-disulfide intermediate, two two-disulfide intermediates, and a mismatched three-disulfide intermediate. A different refolding pattern was obtained by acidic quenching (29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar); in addition to the four intermediates detected by pyridylethylation, a nonnative two-disulfide species and a mixed two-disulfide intermediate with GSH were also detected. Furthermore, a different equilibrium distribution of intermediates was obtained by each of the two trapping procedures. It has been documented that the oxidative refolding of IGF-I follows a pathway governed by thermodynamic rather than kinetic principles, resulting in two folding isomers of similar thermodynamic stability but different disulfide conformations (native and mismatched isomers) (28Hober S. Forsberg G. Palm G. Hartmanis M. Nilsson B. Biochemistry. 1992; 31: 1749-1756Crossref PubMed Scopus (104) Google Scholar, 29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar, 31Rosenfeld R.D. Miller J.A. Narhi L.O. Hawkins N. Katta V. Lauren S. Weiss M.A. Arakawa T. Arch. Biochem. Biophys. 1997; 342: 298-305Crossref PubMed Scopus (33) Google Scholar, 32Hober S. Uhlen M. Nilsson B. Biochemistry. 1997; 36: 4616-4622Crossref PubMed Scopus (31) Google Scholar). Milner et al. (33Milner S.J. Francis G.L. Wallace J.C. Magee B.A. Ballard F.J. Biochem. J. 1995; 308: 865-871Crossref PubMed Scopus (30) Google Scholar) proposed that a salt bridge between Glu3 and Arg56 in IGF-I might stabilize the mismatched isomer, accounting for the observation of more than one folding outcome of IGF-I. Recombinant human long R3 insulin-like growth factor-I (LR3IGF-I) is a variant of human insulin-like growth factor-I (IGF-I) in which glutamate 3 is replaced by arginine, and a 13-residue extension appears at the N terminus (Fig. 1). LR3IGF-I is substantially more potent than IGF-I in affecting carbohydrate metabolism and in stimulating the growth of fetal tissue in animals (34Tomas F.M. Knowles S.E. Chandler C.S. Francis G.L. Owens P.C. Ballard F.J. J. Endocrinol. 1995; 137: 413-421Crossref Scopus (65) Google Scholar). The refolding behavior of LR3IGF-I has been described as being significantly different from that of IGF-I (33Milner S.J. Francis G.L. Wallace J.C. Magee B.A. Ballard F.J. Biochem. J. 1995; 308: 865-871Crossref PubMed Scopus (30) Google Scholar); however, in that preliminary report, neither the disulfide structure of the intermediates nor the folding pathway of LR3IGF-I was determined. We report here on important aspects of the folding pathway of both LR3IGF-I and IGF-I by structural elucidation of disulfide intermediates involved in the folding and unfolding processes. Our results were obtained using cyanylation chemistry for trapping intermediates and specific cleavage/mass mapping for subsequent structural elucidation of the intermediates (22Wu J. Yang Y. Watson J.T. Protein Sci. 1998; 7: 1017-1028Crossref PubMed Scopus (55) Google Scholar, 23Wu J. Watson J.T. Protein Sci. 1997; 6: 391-398Crossref PubMed Scopus (160) Google Scholar, 24Yang Y. Wu J. Watson J.T. J. Am. Chem. Soc. 1998; 123: 5834-5835Crossref Scopus (24) Google Scholar). Similarities and differences in the folding patterns of LR3IGF-I and IGF-I are highlighted and rationalized on the basis of their structures. Recombinant human long R3 insulin-like growth factor-I (LR3IGF-I) was purchased from Sigma. IGF-I was obtained from Austral Biologicals (San Ramon, CA). The proteins were purified prior to use by reversed phase HPLC as described below. Tris(2-carboxyethyl)phosphine hydrochloride was purchased from Pierce. Guanidine hydrochloride was obtained from Roche Molecular Biochemicals. GSSG, GSH, citric acid, sodium citrate, hydrochloric acid, and CDAP were purchased from Sigma and used without further purification. Acetonitrile and trifluoroacetic acid were of HPLC grade. Tris(2-carboxyethyl)phosphine solution in 0.1 m citrate buffer at pH 3.0 was prepared as 0.10 m stock solution and stored under N2 at -20 °C for weeks with little deterioration. The 0.10 m CDAP solution in 0.1m citrate buffer at pH 3.0 was prepared prior to use. An overall scheme representing the refolding (and reductive unfolding) protocol is shown in Fig. 2. LR3IGF-I or IGF-I (0.1 mg) was dissolved in 0.5 ml of citrate buffer, pH 3.0, containing 6 m guanidine-HCl and 0.1 mTris(2-carboxyethyl)phosphine reducing agent. Reduction of the proteins was carried out at 37 °C for 2 h; the reduced/denatured LR3IGF-I or IGF-I was purified by HPLC, dried under reduced pressure, and stored at −70 °C. The refolding of reduced and denatured protein was initiated by diluting the reduced/unfolded protein sample with 0.10 mTris-HCl buffer (pH 8.7), containing 1 mm GSSG, 10 mm GSH, 0.2 m KCl, and 1 mm EDTA, to a final protein concentration of 0.1 mg/ml. The refolding intermediates were trapped at different times by the method described below. Native LR3IGF-I or IGF-I (0.1 mg/ml) was reduced/unfolded in 0.10m Tris-HCl buffer, pH 8.7, containing 0.25 mmcysteine, 0.20 m KCl, and 1.0 mm EDTA. After incubation for the indicated periods, an aliquot of reductive/unfolding solution (0.1 ml) was removed, and the intermediates were trapped, separated, and analyzed by the method described below. After the reductive unfolding reached the equilibrium, cysteine was added to a final concentration of 100 mm to drive the reductive unfolding reaction to completion. Refolding of the reduced, unfolded LR3IGF-I was quenched at 30 min by adding 1.0 mHCl to pH ∼2. After HPLC separation, the fractions containing a two-disulfide intermediate and a mismatched (three-disulfide) protein isomer were dried; reconstituted in 0.10 m Tris-HCl buffer (pH 8.7) containing 1 mm GSSG, 10 mm GSH, 0.2m KCl, and 1.0 mm EDTA to a final concentration of 0.1 mg/ml; and incubated for up to 30 min at room temperature. At designated time points, aliquots were removed, and refolding intermediates were trapped by the method described below. Refolding or reductive unfolding intermediates were trapped in a time course manner by removing aliquots (0.1 ml) of protein solution and mixing with 1.0 m HCl containing freshly prepared 0.2 m CDAP to give a solution of pH 2–3. Cyanylation of free thiol groups by the CDAP proceeded at room temperature for 10 min. The trapped intermediates were immediately separated by HPLC under the conditions described below. The HPLC fractions were collected manually and analyzed by MALDI-time-of-flight mass spectrometry. Those with 0-, 52-, 104-, or 156-Da increases over the mass of the intact protein correspond to three-disulfide (native or nonnative), two-disulfide, one-disulfide, and the completely reduced species, respectively. A mass shift of +306 Da is characteristic of the formation of a mixed disulfide bond between a thiol group and glutathione; a mass shift of +612 Da indicates the presence of two mixed disulfide bonds with two glutathiones. The disulfide structure of purified refolding or reductive unfolding intermediates was determined by the partial reduction/cyanylation/chemical cleavage/mass mapping approach, as described previously (22Wu J. Yang Y. Watson J.T. Protein Sci. 1998; 7: 1017-1028Crossref PubMed Scopus (55) Google Scholar, 23Wu J. Watson J.T. Protein Sci. 1997; 6: 391-398Crossref PubMed Scopus (160) Google Scholar). The separation of folding/unfolding intermediates was carried out by reversed phase HPLC with a linear gradient elution using Waters model 6000 pumps controlled by a PC. The UV detection was at 215 nm. The column was a Vydac C18 (catalog no. 218TP54; 10-μm particle size, 300-Å pore, 4.6 × 250 mm). Solvent A was 0.1% aqueous trifluoroacetic acid. Solvent B was acetonitrile/water (9:1, v/v) containing 0.1% trifluoroacetic acid. The linear gradient was 30–50% solvent B in 45 min at a flow rate of 1 ml/min. The HPLC fractions were collected manually, and the contents were then dried under reduced pressure for further use. MALDI mass spectra were obtained on a Voyager Elite time-of-flight mass spectrometer (PerSeptive Biosystems Inc., Framingham, MA) equipped with delayed extraction and a model VSL-337ND nitrogen laser (Laser Science, Newton, MA). The accelerating voltage in the ion source was set to 20 kV. Grid and guide wire voltages were 93.6 and 0.2% of the accelerating voltage, respectively. Data were acquired in the positive linear DE mode of operation. Time-to-mass conversion was achieved by external and/or internal calibration using standards of bradykinin (m/z1061.2), bovine pancreatic insulin (m/z 5734.5), and horse skeletal myoglobin (m/z 16952) obtained from Sigma. All experiments were performed using α-cyano-4-hydroxycinnamic acid (Aldrich) as the matrix. Saturated matrix solutions were prepared in a 50% (v/v) solution of acetonitrile/aqueous 1% trifluoroacetic acid, mixed in equal volumes with peptide or protein samples, and applied to a stainless steel sample plate. The mixture was allowed to air-dry before being introduced into the mass spectrometer. The distribution of intermediates during the refolding process is represented by HPLC chromatograms of cyanylated species trapped at designated times. Fig. 3, a and b, shows an array of chromatograms of LR3IGF-I and IGF-I intermediates trapped by reaction with CDAP under acidic conditions at various times after initiating refolding in the GSSG/GSH buffer. In addition to native LR3IGF-I (N) and completely reduced LR3IGF-I (R), three well populated species were observed at different time points during the refolding of LR3IGF-I (Fig. 3 a). The oxidative state of the trapped intermediates was determined by MALDI mass spectrometry based on the mass shift between intermediate and the intact protein. Thus, peaks I′ and II′ in Fig. 3 a correspond to intermediates containing one and two disulfide bonds, respectively, as evidenced by 104- and 52-Da shifts from the mass of the intact protein. Peak III′ represents a mixture of a mismatched three-disulfide intermediate (the same mass as intact protein) and a mixed two-disulfide intermediate in which two SH groups each formed a mixed disulfide bond with a glutathione (as indicated by a shift of +612 Da from the mass of the intact protein). Reduced/unfolded LR3IGF-I was first converted to a one-disulfide intermediate (peak I′) and then a two-disulfide intermediate (peak II′) within 1 min in the presence of 10 mm GSH, 1 mm GSSG. The three-disulfide intermediate and mixed two-disulfide intermediate coeluted as peak III′ by 2 min. Considerable native protein (N) was formed by 30 min, a time by which an equilibrium had been reached, and the ratio of the mismatched and native protein remained constant thereafter. Fig. 3 b shows the HPLC trace of refolding intermediates of IGF-I trapped at 10 s, 1 min, 2.5 min, and 30 min, respectively. During the course of refolding, a total of six well populated intermediates were trapped and identified. Mass analysis by MALDI showed that peak III represents a mismatched protein isomer containing three disulfide bonds; peaks IIA, IIB, and IIC are two-disulfide intermediates; peak IImixrepresents a mixed two-disulfide species with two glutathiones; and peak I is a one-disulfide species. The intermediate distribution at equilibrium (see bottom panel) is very similar to that previously reported from a study using acidic quenching (29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar,30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar). In order to prevent the formation of mixed disulfide intermediates as would occur with GSH/GSSG in the refolding process, the reductive unfolding of both proteins was performed in cysteine solution. The time-dependent distribution of the CDAP-trapped intermediates during reductive unfolding of LR3IGF-I is shown in Fig. 4 a. The pattern of intermediates was the same (but inverted) as that observed during the refolding of LR3IGF-I. The mismatched three-disulfide isomer was observed as the earliest intermediate, accounting for ∼10% of the total protein at 3 h. Prolonged incubation of up to 71 h did not alter the ratio of the native and mismatched protein isoforms, suggesting that the native and mismatched species had reached a state of equilibrium. However, the two-disulfide intermediate II′ did accumulate during the incubation. After 71 h of incubation, thermodynamic equilibrium among the three species had been reached. When the concentration of cysteine was increased to 100 mmand the mixture was incubated for another 30 min after reaching the 71-h equilibrium point under the initial conditions, the completely reduced/unfolded protein (R) became the predominant species, accompanied by only minor amounts of two- and one-disulfide intermediates. The reductive unfolding of IGF-I under the same conditions (Fig. 4 b) showed a similar pattern to that of LR3IGF-I, but the ratio of the mismatched isomer at equilibrium was significantly higher (35%) than that (10%) observed during the unfolding of LR3IGF-I. Unlike the refolding of IGF-I in which many more intermediates were observed at thermodynamic equilibrium, during reductive unfolding of IGF-I the mismatched three-disulfide species appeared as the only early intermediate to reach an equilibrium with the native protein within 2 h. Further incubation up to 20 h did not significantly change the pattern of unfolding intermediates. Apparently, the trace of cysteine (0.25 mm) in the solution was insufficient to effect further reduction of the disulfide bonds. Increasing the cysteine concentration to 100 mm and incubating for another 1 min shifted the equilibrium distribution of intermediates of IGF-I to the pattern shown in the bottom panel of Fig. 4 b; both the one-disulfide intermediate (I) and the two-disulfide intermediate (IIA) (but not IIB and IIC) were formed within 1 min. After 30 min, IIA and I were almost completely converted to reduced/unfolded IGF-I (R) (data not shown). The refolding intermediates of LR3IGF-I (corresponding to peaks II′ and III′ in Fig. 3 a, respectively) were trapped during a separate experiment by acid quenching and purified by HPLC. Secondary refolding experiments were initiated with each of these purified acid-trapped intermediates by reconstituting them in 10 mm GSH, 1 mm GSSG buffer, pH 8.7, for up to 30 min at room temperature. Fig. 5, a and b, shows HPLC chromatograms showing the distribution of intermediates after initiating refolding experiments with II′ and III′, respectively. Comparison of Figs. 5 a, 5 b, and 3 areveals a surprising similarity in the course of refolding whether it was started with the two-disulfide intermediate (II′), the mismatched isomer (III′), or the reduced/unfolded protein. In each case, the same equilibrium distribution of II′, III′, and native protein was obtained. The one-disulfide intermediate containing the stable Cys18–Cys61 structure (I′) was not observed in the refolding experiments starting with purified II′ or III′. The refolding experiments initiated from the native two-disulfide intermediate (II′) did not increase the yield of native LR3IGF-I. Analysis by HPLC (Fig. 5 a) showed formation of approximately 90% native LR3IGF-I at 30 min. On the other hand, refolding from the two-disulfide intermediate (IIA) of IGF-I yielded all of the intermediates (I, IIA, IIB, IIC, III, and N) (31Rosenfeld R.D. Miller J.A. Narhi L.O. Hawkins N. Katta V. Lauren S. Weiss M.A. Arakawa T. Arch. Biochem. Biophys. 1997; 342: 298-305Crossref PubMed Scopus (33) Google Scholar). Furthermore, the HPLC trace of the refolding intermediates from III′ (Fig. 5 b) showed a partially resolved peak from III′. Analysis of the corresponding material indicated a molecular weight increase of ∼612 Da, which identified it as a glutathione adduct containing two glutathiones per LR3IGF-I. The methodology developed in our laboratory (22Wu J. Yang Y. Watson J.T. Protein Sci. 1998; 7: 1017-1028Crossref PubMed Scopus (55) Google Scholar, 23Wu J. Watson J.T. Protein Sci. 1997; 6: 391-398Crossref PubMed Scopus (160) Google Scholar, 24Yang Y. Wu J. Watson J.T. J. Am. Chem. Soc. 1998; 123: 5834-5835Crossref Scopus (24) Google Scholar), based upon chemical cleavage at cyanylated cysteine residues and subsequent mass mapping of the fragments, was employed to determine the disulfide structure of the trapped intermediates. The disulfide structures of the respective folding intermediates are summarized in Table I.Table IDisulfide linkage of intermediates trapped during the refolding and reductive unfolding of IGF-I and LR 3 IGF-IProteinHPLC peakDisulfide structureIGF-II18–61IIA18–61, 6–48IIB18–61,6–47IIC18–61, 6–52IImixMixed two disulfidesIII18–61,6–47, 48–52N18–61, 6–48, 47–52LR3IGF-II′18–61II′18–61, 6–48III′18–61, 6–47,48–52N18–61, 6–48, 47–52 Open table in a new tab Proper trapping of folding intermediates and subsequent structural determination are critical steps in the elucidation of a folding pathway and in studies of the associated kinetics. The trapping method should stop thiol/disulfide exchange, which may occur on the microsecond time scale in refolding buffer. The cyanylation of thiol groups by CDAP in acidic solution effectively quenches the refolding process by blocking reactive thiol groups. The folding intermediates of IGF-I trapped by CDAP and their distribution pattern are similar to those obtained by acid trapping under the same refolding conditions (29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar). By mass mapping the cleavage products resulting from cyanylated intermediates, the disulfide structures of six well populated intermediates were identified, including a native one-disulfide intermediate (I, Cys18–Cys61), a native two-disulfide intermediate (IIA, Cys18–Cys61/Cys6–Cys48); two nonnative two-disulfide intermediates (IIB, Cys18–Cys61/Cys6–Cys47; IIC, Cys18–Cys61/Cys6–Cys52); a mismatched intact protein (III, Cys18–Cys61/Cys6–Cys47/Cys48–Cys52); and a mixed two-disulfide intermediate with glutathione (IImix) (nonnative disulfide residues underlined). The intermediates IIB and IImix were not observed by vinylpyridine trapping (28Hober S. Forsberg G. Palm G. Hartmanis M. Nilsson B. Biochemistry. 1992; 31: 1749-1756Crossref PubMed Scopus (104) Google Scholar). Furthermore, intermediate I was the major form captured by pyridylethylation (28Hober S. Forsberg G. Palm G. Hartmanis M. Nilsson B. Biochemistry. 1992; 31: 1749-1756Crossref PubMed Scopus (104) Google Scholar), while it was the minor form in the distribution of the folding intermediates trapped by the CDAP approach described herein and also by acid trapping as reported elsewhere (29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 30Miller J.A. Narhi L.O. Hua Q.X Rosenfeld R. Arakawa T. Rohde S.P. Lauren S. Stoney K.S. Tsai L. Weiss M.A. Biochemistry. 1993; 32: 5203-5213Crossref PubMed Scopus (105) Google Scholar). In order to compare the flow of intermediates in the folding pathway of LR3IGF-I and IGF-I, the refolding of both LR3IGF-I and IGF-I was performed under identical conditions in a GSSG/GSH buffer. Comparison of the HPLC pattern of folding intermediates for LR3IGF-I and IGF-I exhibited some similarities and discrepancies. All the trapped intermediates contain a native Cys18–Cys61 structure, which is the first disulfide bond formed in the refolding process (Fig. 3,a and b) and the last disulfide bond reduced in the unfolding process (Fig. 4, a and b). The most abundant two-disulfide intermediates (IIA in IGF-I and II′ in LR3IGF-I) and mismatched three-disulfide intermediates (III in IGF-I and III′ in LR3IGF-I) also have a homologous disulfide structure (Table I). Refolding kinetics indicate that formation of the native Cys18–Cys61 disulfide bond is very fast in both proteins, as is the subsequent formation of the native Cys18–Cys61/Cys6–Cys48intermediate. However, the formation of the last disulfide bond, Cys47–Cys52, is slower, supporting the conclusion that its formation is energetically unfavorable (29Hua Q.X. Narhi L. Jia W. Arakawa T. Rosenfeld R. Hawkins N Miller J.A. Weiss M. J. Mol. Biol. 1996; 259: 297-313Crossref PubMed Scopus (66) Google Scholar, 31Rosenfeld R.D. Miller J.A. Narhi L.O. Hawkins N. Katta V. Lauren S. Weiss M.A. Arakawa T. Arch. Biochem. Biophys. 1997; 342: 298-305Crossref PubMed Scopus (33) Google Scholar,32Hober S. Uhlen M. Nilsson B. Biochemistry. 1997; 36: 4616-4622Crossref PubMed Scopus (31) Google Scholar). The folding result of LR3IGF-I is different from that for IGF-I in that the native three-disulfide isomer (N) is the predominant species at equilibrium, and much less mismatched disulfide bond formation is observed. Oxidation of the native intermediate, Cys18–Cys61, results only in the formation of the native Cys18–Cys61/Cys6–Cys48intermediate in the refolding of LR3IGF-I, while as many as four two-disulfide intermediates are observed in the refolding of IGF-I, each containing a Cys18–Cys61 bond (Table I). Furthermore, the formation of early intermediates in IGF-I is much faster; at 2.5 min, ∼90% of the reduced IGF-I is oxidized, whereas only ∼50% of the reduced LR3IGF-I is oxidized in the same time period. Nevertheless, an equilibrium distribution of intermediates can be reached for both proteins within 30 min. Rosenfeld et al. (31Rosenfeld R.D. Miller J.A. Narhi L.O. Hawkins N. Katta V. Lauren S. Weiss M.A. Arakawa T. Arch. Biochem. Biophys. 1997; 342: 298-305Crossref PubMed Scopus (33) Google Scholar) observed that the oxidative refolding from the native two-disulfide intermediate of IGF-I (IIA in Fig. 3 b, with Cys18–Cys61/Cys6–Cys48linkage) formed an equilibrium mixture of all of the disulfide intermediates observed from the refolding of reduced/unfolded IGF-I. The refolding experiment starting with the homologous intermediate of LR3IGF-I (II′ in Fig. 3 a) resulted in an equilibrium mixture of starting intermediate, native protein, and mismatched protein isomer (coeluting with the glutathione adduct). The one-disulfide intermediate, I′, was not observed. The absence of I′ in the refolding from II′ could be attributable to the greater stability of the Cys6–Cys48 pair in LR3IGF-I, as compared with the homologous pair in IGF-I. Nevertheless, comparison of a and b in Fig. 5reveals that the refolding courses from the isolated two-disulfide intermediate and the mismatched isomer, respectively, resulted in an identical distribution of intermediates, suggesting that the refolding of LR3IGF-I is also thermodynamically controlled and all of the trapped intermediates are thus interconvertible. Three distinct features are recognized after comparing the unfolding processes of LR3IGF-I and IGF-I. First, the conversion of native IGF-I to the three-disulfide intermediate III (Fig. 4 b) is faster than the conversion of native LR3IGF-I to the three-disulfide intermediate III′ (Fig. 4 a). Second, no appreciable two-disulfide intermediate (neither IIA, IIB, nor IIC) was detected during the unfolding process of IGF-I in the presence of 0.25 mm cysteine (Fig. 4 b), while the two-disulfide intermediate II′ of LR3IGF-I was observed after incubation for up to 24 h under identical conditions (Fig. 4 a). This observation may indicate that the disulfide bond Cys47–Cys52 in IGF-I is more stable than its counterpart in LR3IGF-I. Third, the conversion of IIA → I during the “forced” unfolding experiment of IGF-I (bottom chromatogram of Fig. 4 b) was faster than the conversion of II′ → I′ in LR3IGF-I (bottom chromatogram of Fig. 4 a), implying that the disulfide bond Cys6–Cys48 in IGF-I is less stable after reduction of the disulfide bond Cys47–Cys52. The observed differences in the folding of IGF-I and LR3IGF-I can be rationalized to the variation in the thermodynamic stability of the folding intermediates. The LR3IGF-I variant contains two unrelated structural changes: a substitution of Glu by Arg at position 3 and an N-terminal extension of 13 hydrophobic amino acids. Milner et al. (33Milner S.J. Francis G.L. Wallace J.C. Magee B.A. Ballard F.J. Biochem. J. 1995; 308: 865-871Crossref PubMed Scopus (30) Google Scholar) proposed that a salt bridge between Glu3 and Arg56 in IGF-I accounted for the high proportion of mismatched protein isomer during the folding of IGF-I. After the charge substitution by Arg at position 3 in LR3IGF-I, the salt bridge between residues 3 and 56 was replaced by charge repulsion. Thus, the mismatched folding intermediates may be destabilized, leading to predominant formation of the native protein isomer. The N-terminal extension apparently causes steric hindrance, decreases the flexibility of the molecule, and thus limits the number of conformations leading to disulfide bond formation. As a result, the refolding kinetics is much slower for LR3IGF-I, and native folding isomers with minimum conformational energies are preferentially formed. Both structural changes may account for the difference in the folding pathway, and it is not possible to assign which change is dominant in determining the folding process. Based on the disulfide structure and kinetic analysis of folding intermediates, proposed folding pathways for both IGF-I and LR3IGF-I are shown in Fig. 6. Overall, the folding of the two proteins is a thermodynamically controlled process. The two proteins share a similar mechanism, leading to the formation of native and mismatched protein isomers; however, the most productive routes may be slightly different. According to our results, the folding of IGF-I is more flexible in the formation of nonnative two-disulfide intermediates and the mismatched protein isomer via rearrangement of disulfide bonds or thiol/disulfide exchange. The nonnative two-disulfide intermediates (IIB and IIC) can be formed by either direct oxidation of the native one-disulfide intermediate (I) or disulfide rearrangement of IIA. While IIB can lead directly to the mismatched protein isomer (III), both IIB and IIC can revert to productive intermediates (I or IIA). Although only native one- and two-disulfide intermediates were trapped for LR3IGF-I, the presence of the mismatched isomer at equilibrium implicates the involvement of nonnative disulfide intermediates, because neither II′ nor native protein can spontaneously convert to the mismatched isomer without breaking/reforming disulfide bonds. Although these intermediates were so short lived as not to be detected, they represent productive intermediates for the formation of mismatched species (35Chang J.Y. Schindler P. Ramseier U. Lai P.H. J. Biol. Chem. 1995; 270: 9207-9216Crossref PubMed Scopus (60) Google Scholar). We speculate that these intermediates may have a more native-like structure but are thermodynamically less stable (than the native two-disulfide intermediate II′). Therefore, they can rapidly convert to the mismatched isomer. The glutathione intermediates detected during the refolding experiment from II′ strongly support the participation of glutathione in the disulfide rearrangement, largely accounting for the formation of Cys18–Cys61/Cys6–Cys47and/or Cys18–Cys61/Cys48–Cys52. The involvement of glutathione in the rearrangement of the two-disulfide species was further supported by the energetics of late folding intermediates. The refolding of an IGF-I mutant protein showed that the formation of Cys47–Cys52 is energetically unfavorable (32Hober S. Uhlen M. Nilsson B. Biochemistry. 1997; 36: 4616-4622Crossref PubMed Scopus (31) Google Scholar). Therefore, the conversion of the native two-disulfide intermediate (II′ in LR3IGF-I and IIA in IGF-I) to other intermediates, such as glutathione adducts, was as favorable as the conversion to native protein. Another route leading to the formation of precursor intermediates of the mismatched isomer is the direct oxidation of the Cys18–Cys61 intermediate, which is obviously a route of choice for the refolding of IGF-I, as observed in Fig. 3 b. However, the absence of I′ during the refolding from II′ and III′ at least suggests that the direct oxidation of I′ is not essential for the formation of the nonnative two-disulfide intermediates, although we cannot exclude the possibility. It seems that a disulfide rearrangement between the two-disulfide intermediates is a more favorable route for the formation of the precursor intermediates of the mismatched three-disulfide species. The availability of chemical and mass spectrometric evidence for the disulfide structure of trapped intermediates provides a great advantage in studies of the folding process. Furthermore, investigations of both the folding and unfolding processes by the same methodology for a given protein and its mutant provide the basis for integrating time-dependent “snapshot” observations for insight into the dynamic phenomenon. Despite the great experimental challenges associated with providing concrete solutions to problems in protein folding, the data presented here should help provide an integrated view of the folding patterns for the family of insulin-like proteins.