Unfolding of human proinsulin Intermediates and possible role of its C-peptide in folding/unfolding Cheng-Yin Min, Zhi-Song Qiao and You-Min Feng Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China We have investigated the in vitro refolding process of human proinsulin (HPI) and an artificial mini-C derivative of HPI (porcine insulin precursor, PIP), and found that they have significantly different disulfide-formation pathways. HPI and PIP differ in their amino acid sequences due to the presence of the C-peptide linker found in HPI, therefore suggesting that the C-peptide linker may be responsible for the observed difference in folding behaviour. However, the manner in which the C-peptide contributes to this difference is still unknown. We have used both the disulfide scrambling method and a redox-equilibrium assay to assess the stability of the disulfide bridges. The results show that disulfide reshuffling is easier to induce in HPI than in PIP by the addition of thiol reagent. Thus, the C-peptide may affect the unique folding pathway of HPI by allowing the disulfide bonds of HPI to be easily accessible. The detailed processes of HPI unfolding by reduction of its disulfide bonds and by disulfide scrambling methods were also investigated. In the reductive unfolding process no accumulation of intermedi- ates was detected. In the process of unfolding by disulfide scrambling, HPI gradually rearranged its disulfide bonds to form three major isomers G1, G2 and G3. The most abun- dant isomer, G1, contains the B7-B19 disulfide bridge. Based on far-UV CD spectra, native gel analysis and cleavage by endoproteinase V8, the G1 isomer has been shown to resemble the intermediate P4 found in the refolding process of HPI. Finally, the major isomer G1 is allowed to refold to native protein HPI by disulfide rearrangement, which indi- cates that a similar molecular mechanism may exist for the unfolding and refolding process of HPI. Keywords: C-peptide; disulfide scrambling; disulfide stabil- ity; human proinsulin; unfolding. The protein folding process can be simply considered as a process in which a biologically inactive amino acid sequence becomes a uniquely structured molecule possessing a specifically biological activity. Conversely the unfolding of a protein can be considered as the other half of the protein folding process which causes a protein to lose its biological activity and become an ensemble of structurally denatured states [1–3]. The characterization of the protein folding and unfolding processes has become of great interest. It has been recognized that protein unfolding is a crucial step in protein degradation and protein translocation in vivo [4]. In addition, it has been observed that some unfolded proteins are capable of retaining some structural elements that may reflect folding initiation sites or inferred intermediates in the folding pathway [5]. Disulfide bond-containing proteins provide an advantage to the study of both protein folding and unfolding process due to the ability to capture intermediates contains partial disulfide bonds. Two conventional methods are often used in the investigation of disulfide bond-containing protein unfolding. One such method is denaturation, which requires the use of denaturants to unfold a protein in the absence of reducing reagent [6,7]. The other method is reductive unfolding, in which protein is unfolded by the additional reducing reagents (such as dithiothreitol) in the absence of denaturants [8–10]. The unfolding pathways of most proteins have been studied using denaturation, in which the disulfide bonds remain intact. Due to the cooperative and interdependent role of the disulfide bonds in maintaining the native conformation of the majority of proteins, the reductive unfolding pathway always results in an Ôall- or-noneÕ mechanism. Therefore, it is very difficult to capture the disulfide intermediates and to complete additional investigation of the molecular mechanism of the unfolding pathways. Currently, the unfolding pathway of a limited number proteins, such as bovine pancreatic trypsin inhi- bitor, RNaseA and a-lactoalbumin, have been well charac- terized by using the reductive unfolding method [11–14]. The recently established disulfide scrambling method of Chang et al. may make it possible to dissect experimentally the reductive unfolding of a disulfide-containing protein into two distinct stages [15]. During the first stage, in the presence of denaturant and trace thiol catalyst, native Correspondence to Y M. Feng, Shanghai Institute of Biochemistry, Chinese Academy of Sciences, 320 Yue-Yang Road, Shanghai 200031, China. Fax: + 86 021 54921011, Tel.: + 86 021 54921133, E-mail: fengym@sunm.shcnc.ac.cn Abbreviations: HPI, human proinsulin; PIP, porcine insulin precursor; IGF-I, insulin-like growth factor-I; TAP, tick anticoagulant peptide; PCI, potato carboxypeptidase inhibitor; LCI, leech carboxypeptidase inhibitor; GdnHCl, guanidine hydrochloride; IAA, sodium salt of idoacetic acid; GSH, reduced glutathione; GSSG, oxidized gluta- thione; frdHPI, fully reduced/dentured HPI; frHPI, fully reduced HPI; ESI-MS, electrospray ionization-mass spectrometry. Note: C Y. Min and Z S. Qiao contributed equally to this work. (Received 12 November 2003, revised 17 February 2004, accepted 9 March 2004) Eur. J. Biochem. 271, 1737–1747 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04079.x proteins are unfolded by reshuffling their native disulfide bonds and are thus converted into a mixture of disulfide isomers. In the subsequent stage, the disulfide bonds of the scrambled isomers could be readily reduced by a low concentration of reductive reagents, and intermediates with heterogeneous disulfide bonds could then be observed during this process. The unfolding pathway of many proteins, among them hirudin, tick anticoagulant peptide (TAP), RNase A, cardiotoxin III, potato carboxypeptidase inhibitor (PCI) and leech carboxypeptidase inhibitor (LCI) [15–18], have been studied by this method. Insulin is a two-chain protein hormone, designated A and B chain, respectively, containing three disulfide bonds. Two interchain disulfide bonds are A7Cys–B7Cys, A20Cys– B19Cys and one intrachain disulfide bond is A6Cys– A11Cys [19]. The disulfide linkages of insulin have been shown to be important in maintaining its native conforma- tion and biological activity [20–25]. The double-chain insulin is synthesized in vivo as a single-chain precursor (preproinsulin) and folded as proinsulin, in which a connecting peptide of 35 residues links the C terminus of the B chain and N terminus of the A chain. After digestion by a specific set of protein enzymes in the B-cell granule, proinsulin is converted into insulin and C-peptide of 31 amino acids [26]. Previous studies completed on the unfolding process of insulin or proinsulin were often carried out with disulfide bonds intact [27,28]. By using near- and far-UV CD, Brems et al. have investigated the guanidine hydrochloride- induced equilibrium denaturation of insulin and proinsulin [29,30]. The results of previous work on insulin are consistent with a two-state denaturation process that lack any appreciable equilibrium intermediates. The character- ization of the unfolding of insulin and proinsulin using the reductive unfolding method has not been thoroughly investigated. We have characterized the unfolding process of an artificial porcine insulin precursor (PIP), in which a dipeptide, AK, links the B and A chain, as shown in Fig. 1A, in denaturants containing a thiol catalyst. We observed that PIP reshuffled its native disulfide bonds to form disulfide isomers, with one major disulfide isomer present. The disulfide isomers of PIP could spontaneously refold to native PIP in the presence of a thiol reagent, clearly demonstrating that PIP has only one thermodynamically stable form [31]. Recently, the in vitro refolding process of human proinsulin (HPI) has been investigated in our laboratory. Four scrambled disulfide isomers with three intact disulfide bonds have been captured as intermediates [32]. To compare the disulfide isomers that appeared during the refolding and unfolding of HPI, we have investigated the process of unfolding by using disulfide scrambling method as well as the denaturation method. These results show a striking correlation between the oxidized refolding and unfolding of HPI by the disulfide scrambling method. HPIisthenativein vivo precursor of human insulin in which the B-chain and A-chain are connected by a flexible 31 residues connecting peptide (C-peptide), as shown in Fig. 1A. PIP is an artificial mini-proinsulin in which two amino acids, Ala, Lys, have been substituted for the C-peptide found in HPI. Thus, the only amino acid sequences difference between HPI and PIP are within the connecting peptide region. As the previous studies showed that the insulin A and B chains contain sufficient folding information for correct disulfide pairing [33,34], one may reasonably assume there should not be an obvious Fig. 1. Amino acid sequences and in vitro refolding pathway of PIP and HPI. (A) Amino acid sequences of PIP and HPI. Amino acids are shown in the one-letter code. The numbering of the residues in HPI and PIP are based on each chain separately. For examples, B19 denotes the nineteenth residue of B-chain and A1 denotes the first residue in A-chain. Disulfide bonds in the native HPI and PIP are indicated by dashed lines. For HPI, the corresponding insulin B- and A-chain are linked by the 31-residue C-peptide and two dibasic resi- dues, which are shown as dark circles. For PIP, the linker (KAA) between B29-K and A1-G is indicated by an asterisk. Please note that the B30 residue in HPI is Thr, while that in PIP is Ala. (B) Putative disulfide formation pathway of PIP in vitro. Intermediates are named using the disulfide bonds they contain. Arrows with dashed lines indicate the folding pathway for the first formation of the intra-A disulfide bonds. Another major folding pathway is indicated by the solid arrows, it begins with the A20–B19 disulfide bond formation and then involves the disulfide rearrangement [35]. (C) Schematic repre- sentation of the putative disulfide folding pathway of HPI in vitro. I–III represent the intermediates mixtures with one, two and three disulfide bonds, respectively. P1–P4 are the HPI disulfide isomers captured during the oxidized refolding process of HPI in vitro [32]. 1738 C Y. Min et al. (Eur. J. Biochem. 271) Ó FEBS 2004 difference between PIP and HPI in the refolding pathway. However, our studies of the oxidized refolding process of PIP and HPI in vitro [32,35] have found that these two proteins adopt two significantly different disulfide forming pathways as shown in Fig. 1B (PIP) and 1C (HPI). As a result, we can conclude that the connecting peptide in HPI partially controls its unique folding behaviour. However, the manner in which flexible C-peptide contributes to this folding process is still unknown. Compared with the step- by-step formation of the disulfide bonds in PIP (Fig. 1B), disulfide bond formation in HPI occurs by random formataion of intramolecular disulfide bonds at the begin- ning of oxidized refolding, and then rearrangement from non-native to native disulfide bonds. This different folding behaviour indicates that the energy state of the disulfide bonds in HPI and PIP may not be similar. During the HPI unfolding studies here, the disulfide scrambling method and redox equilibrium assays were used to test this hypothesis. The results confirm that the disulfide bond stability of HPI is lower than that of PIP, which indicates that the C-peptide may control the folding behaviour of HPI by making the disulfide bonds more accessible. Experimental procedures Materials Recombinant HPI and PIP were of > 98% purity as confirmed by RP-HPLC on a C8 column. Endoproteinase Lys-C and V8 were of sequencing grade (Sigma). The sodium salt of iodoacetic acid (IAA), reduced glutathione (GSH) and oxidized glutathione (GSSG) were ultra pure (Amersham Biosciences, Piscataway, NJ). Ultra pure dithiothreitol was from Sigma. Ultra pure urea and guanidine-HCl were from Promega. Acetonitrile and tri- fluoroacetic acid were of HPLC grade. All other reagents used in the experiment were of analytical grade. Reductive unfolding of the native protein in the absence of denaturant Native HPI was dissolved in buffer containing 100 m M Tris pH 8.7, 1 m M EDTA and various concentrations of dithiothreitol (ranging between 0.5 and 100 m M )atafinal protein concentration of 0.5 mgÆmL )1 . The reduction experiments were carried out at 25 °C for 16 h. To trap the unfolding intermediates, reduction was carried out at 25 °C in the presence of 1 m M dithiothreitol. At different time points during the reaction, 20 lL of the reaction sample was taken out and mixed with 80 lL0.3% trifluoroacetic acid to stop the reaction, followed by RP- HPLC on a C4 column. Fully reduced HPI (frHPI) was obtained by reducing the native HPI with 100 m M dithio- threitol in the above buffer for 16 h at 25 °C. To confirm the identity of the reduced protein, frHPI was modified by IAA and then separated by native PAGE. The native PAGE showed that there was only one single band, suggesting that disulfide bonds in HPI were fully reduced. The fully reduced/denatured HPI (frdHPI) was obtained by reducing the native HPI with dithiothreitol in the presence of 6.0 M guanidine hydrochloride (GdnHCl), as described in our previous work [32]. Unfolding of HPI in the presence of denaturant and thiol catalyst The native HPI was dissolved in buffer containing 100 m M Tris pH 8.7, 1 m M EDTA, 0.2 m M 2-mercaptoethanol and different concentrations of GdnHCl at a final protein concentration of 0.25 mgÆmL )1 . The unfolding reaction was carried out at 25 °C for 16 h. For the HPLC analysis, the reaction was terminated by adding trifluoroacetic acid and analysed by RP-HPLC on a C4 column. To observe the time-dependent distribution of the unfolding intermediates during this process, native HPI was dissolved in the unfolding buffer (100 m M Tris pH 8.7, 1 m M EDTA, 0.2 m M 2-mercaptoethanol, 6.0 M GdnHCl) at a final concentration of 0.25 mgÆmL )1 and the reaction was quenched by adjusting the pH to 1.0 with trifluoroacetic acid at different unfolding time point, followed by analysis on HPLC. Disulfide stability of the HPI and PIP in redox buffer HPI or PIP was dissolved in Tris buffer (0.1 M Tris, 1 m M EDTA pH 8.7) containing different redox potentials at the final concentration of 0.2 mgÆmL )1 . In the redox buffer, the ratio (m M /m M ) of GSH to GSSG was 1 : 10, 5 : 5, 10 : 1, 20 : 1, 30 : 1 and 50 : 1, respectively. Simul- taneously, a sample dissolved in the Tris buffer lacking both GSH and GSSG was used as a negative control. The reaction was carried out at 4 °C overnight. After incuba- tion, one-fifth of the volume of freshly prepared 0.5 M sodium iodoacetate solution was added to carboxymethy- late the free thiol groups of proteins. The carboxymehy- lation reaction was carried out at room temperature for 5 min. The modified mixture was then analysed by native PAGE. Isolation and purification of the scrambled disulfide isomers of HPI In the presence of denaturant and thiol catalyst as indicated above, HPI was converted into the mixture of native and scrambled disulfide isomers, which existed in a state of equilibrium. The mixture was adjusted to pH 1.0 with trifluoroacetic acid and separated using RP-HPLC on a C4 column (Sephasil peptide, ST 4.6/250 mm, Pharmacia). Unless otherwise indicated, the solvent A was 0.15% trifluoroacetic acid in water and solvent B was 60% acetonitrile containing 0.125% trifluoroacetic acid. The linear elution gradient was 50% B to 80% B in 30 min with a flow rate of 0.5 mLÆmin )1 . The detection wavelength was 280 nm. The partially isolated disulfide isomers of HPI were further purified by HPLC on a C8 column (Sephasil peptide, ST 4.6/250 mm, Pharmacia). The corresponding fraction was collected and lyophilized. Disulfide linkage analysis of the intermediates by enzyme digestion The endoproteinase V8 that cleaves at the C terminus of Glu residues was used to digest the disulfide isomers of HPI in order to elucidate their disulfide linkage patterns. Ten Ó FEBS 2004 Possible role of C-peptide in the folding/unfolding of HPI (Eur. J. Biochem. 271) 1739 micrograms of the isomer was dissolved in 10 lL 100 m M NH 4 HCO 3 (pH 8.0) and 0.5 lg endoproteinase V8 was added. HPI was used as a positive control in each enzyme digestion. The reaction was carried out at 25 °Cfor16h and quenched by addition of 90 lL of 0.3% trifluoroacetic acid. The digestion mixture was then immediately analysed by RP-HPLC on a C8 column (ZORBAX SB-C8, 5l, 4.6/150 mm; DuPont, San Diego, CA). The elution gradient was 25% B to 65% B linear in 35 min. The flow rate was 0.5 mLÆmin )1 and the detection wavelength was 210 nm. The characteristic peaks on HPLC were manually collected, lyophilized and their molecular masses measured by ESI MS. Refolding of scrambled disulfide isomer G1 To initiate the refolding process, the HPLC purified isomer G1 of HPI was dissolved in buffer containing 100 m M Tris pH 8.7, 1 m M EDTA and 0.2 m M 2-mercaptoethanol at final concentration of 0.1 mgÆmL )1 . The refolding reaction was carried out at 4 °C. Aliquots of the folding solution were removed at time intervals and mixed with an equal volume of 2% trifluoroacetic acid to stop the folding process. The mixture was then analysed by RP-HPLC on a C4 column (Sephasil peptide, ST 4.6/250 mm, Pharmacia) with a linear gradient of 50% B to 80% B in 30 min. The flow rate was 0.5 mLÆmin )1 and the detector wavelength was 230 nm. Protein analysis The protein concentration of HPI and the disulfide isomers were calculated by UV spectroscope using an absorption constant A 276 (1 cm, 1.0 mg mL )1 ) ¼ 0.65 according to the reference [36]. The molecular mass of the disulfide isomers of HPI and the enzyme-digested fragments were measured by ESI MS. The molecular mass of the mixture of scrambled isomers was measured by MALDI-TOF MS. CD studies CD measurements were performed on a Jasco-700 CD spectropolarimeter at 25 °C. The protein samples of disul- fide isomers and HPI were dissolved in 5 m M HCl at a concentration of 0.25 mgÆmL )1 . Samples were scanned from 190 nm to 250 nm and accumulated twice at the resolution of 1.0 nm with the scanning speed of 50 nmÆ min )1 . The cell length was 0.1 cm and the stepwise increment was 0.1 nm. Results Reductive unfolding of the native HPI in the absence of denaturant showed no obviously accumulated intermediates Native HPI was reduced by varies the concentration of dithiothreitol (0.1, 0.5, 1, 5, 10, 50 m M )at25°Cfor16h and then the reaction was stopped by addition of trifluoro- acetic acid. RP-HPLC was used to measure the amount of HPI that had been reduced. We found that the lowest concentration of dithiothreitol capable of completely redu- cing HPI is 1.0 m M . At concentrations less than 1.0 m M dithiothreitol, most of the HPI accumulated as disulfide- linked aggregates with only a small portion being reduced. The reduction of HPI by 1 m M dithiothreitol is shown in Fig. 2. The native disulfide bonds of HPI were rapidly reduced in a collective manner. After 20 min,  95% of the native HPI had been converted into frHPI. At early points (2 or 5 min) only a small fraction of intermediates existed between HPI and frHPI as measured by HPLC. At 10 s or 20 min during the reaction, we could not detect any visible unfolding intermediates by HPLC. Since there are no obviously accumulated intermediates during this reducing process, it is very difficult to study the reductive unfolding pathway by analysis of intermediates. Fig. 2. Reductive unfolding of HPI by the addition of 1 m M dithio- threitol in alkaline buffer. The reducing reaction was quenched at dif- ferent time points, as indicated at the right side of each HPLC chromatograph. The corresponding peaks of native HPI and reduced (frHPI) are indicted at the top of the peaks. HPLC conditions are described in Experimental procedures. 1740 C Y. Min et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Native HPI unfolds more readily than PIP using the disulfide scrambling method Because it is difficult to study the reductive unfolding of HPI in the absence of denaturant, we used the disulfide scrambling method to monitor the HPI unfolding process by RP-HPLC. In the presence of denaturant and a low concentration of thiol reagent (0.2 m M 2-mercaptoethanol), HPI will unfold by reshuffling the native disulfide bonds, which leads to the formation of scrambled disulfide isomers. The scrambled isomers each contain three disulfide bonds, of which at least two are non-native. The denatured states of HPI under varying concentrations of urea and GdnHCl are shown in Fig. 3. With increasing concentration of denatu- rant, an increasing amount of native HPI becomes conver- ted into disulfide isomers that accumulate as three major peaks designated G1, G2 and G3, respectively, on HPLC. The G1, G2 and G3 were collected, lyophilized and modified by iodoacetic acid, followed by molecular mass measurement by MALDI-TOF. The results show that the isomers all have a molecular weight of 9388, identical with native HPI. This indicates that there are no free cysteines in G1, G2 or G3. The scrambled isomers of HPI always equilibrated with native HPI after 1–2 h of unfolding. The denaturation curves, calculated from the fraction of native HPI retained during the unfolding process, are shown in Fig. 4. As a control, PIP was also unfolded using the same disulfide scrambling method with that of HPI and the denaturation curves are also shown in Fig. 4. Comparison of the denaturation curves of PIP with those of HPI show that PIP is significantly more stable than HPI, regardless of whether urea or GdnHCl is used as the denaturant. The native HPI fraction decreased rapidly even at the lowest concentration of denaturant, such that  30% of the native HPI fraction was retained when the concentration of urea or GdnHCl was 1.0 M . In contrast, almost 4.0 M GdnHCl or 8.0 M urea was needed to reduce the native PIP fraction to 30%, indicating that PIP is much more able to maintain its native structure and disulfide bonds than HPI. Moreover, PIP showed a cooperative unfolding process with increasing denaturant conditions, while HPI rapidly lost its native structure even at the lowest concentration of denaturant. Fig. 3. Unfolding of HPI in the presence of denaturant and thiol cata- lyst. Controls for the disulfide scrambling method include the lack of thiol catalyst and lack of denaturant in the buffer. As both controls give the same results, only one is shown. Native HPI exists stably in both control experiments after incubation at 16 °C for 16 h. The three major peaks containing scrambled disulfide isomers of HPI are desi- gnated G1, G2 and G3 separately, based on their elution sequence on HPLC. Fig. 4. Denaturation curve of HPI and PIP by disulfide scrambling method. The native fraction retained is the percentage of native HPI that is not converted into the scrambled isomers. Denaturation was carried out at 16 °C for 16 h in denaturing buffer containing 0.2 m M 2-mercaptoethanol and the indicated concentration of denaturant. Ó FEBS 2004 Possible role of C-peptide in the folding/unfolding of HPI (Eur. J. Biochem. 271) 1741 Redox-equilibrium assay shows that HPI has a lower disulfide stability than PIP in redox buffer To further address differences in the disulfide stability between HPI and PIP, we used a redox-equilibrium assay, which has been routinely used to compare the disulfide stability of different disulfide-containing proteins [20,37–39]. The redox-equilibrium assay involves dissolving the protein in a redox buffer that contains different ratios of GSH/ GSSG. The disulfide bonds of the proteins remain stable when the ratio of GSH/GSSG is lower than a fixed redox potential point, whereas if the GSH/GSSG ratio is above the redox potential point, an increasing amount of native disulfide bonds will be reshuffled or reduced with increases in GSH relative to GSSG (or with an increase in the reductive potency), until the disulfide bonds reasch an equilibrium. Thus, proteins with different disulfide stability will have a different redox potential point. The redox equilibrium assay results of PIP and HPI are shown in Fig. 5. For PIP, part of the protein began to form high molecular mass aggregates when the ratio of GSH/GSSG was 20 : 1, this indicates that the disulfide bonds are disrupted by the redox potency used. Whereas for HPI, the disulfide bonds begin to be disrupted when the ratio of GSH/GSSG was only 5 : 5. These results show that disulfide bonds in HPI are more sensitive to changes in the reduction potential of the redox buffer compared with that of PIP, hence the disulfide stability of HPI is lower than that of PIP. Physical and chemical properties of the HPI disulfide isomers during the disulfide scrambling process HPI was unfolded by disulfide scrambling in the presence of 6.0 M GdnHCl and 0.2 m M 2-mercaptoethanol. The reac- tion was quenched in a time-course dependent manner by removing aliquots of the reaction mixture and adjusting the pH to 1.0 with trifluoroacetic acid, the samples were then analysed immediately by HPLC (Fig. 6). There were three main intermediates during the unfolding process of HPI, designated G1, G2 and G3. There were no other significant intermediates observable that resembled the partially struc- tured isomers, such as P1, P2 or P3, found during the refolding study of HPI [32].The HPLC peaks corresponding to G1, G2 and G3 were collected, partially purified and analysed by native PAGE (Fig. 7A). The native gel shows that the proteins corresponding to peak G1 are much more homogeneous than those in peaks corresponding to G2 or G3. To compare the intermediates found here with the intermediates found during the refolding studies of HPI, a mixture of intermediates P3 and P4 captured during the refolding process were used as a marker. Intermediate P3 is a scrambled disulfide isomer that contains a disulfide bond B7-A20 and retains a few secondary structure elements, while intermediate P4 is an unstructured isomer with a disulfide bond B7-B19 [32]. G1 is similar to P4 in the mobility on native PAGE. The G2 and G3 isomers contain mainly the protein fraction similar to G1 plus some additional proteins similar to intermediates P3. As none of the intermediates and isomers contain additional charges relative to native HPI, their mobility on native PAGE may indirectly reflect their conformation, such that the more flexible conformation will result in a slower mobility. The similar mobility of G1 and P4 indicates that they both possess a more flexible conformation. Some of the G2 and G3 fractions migrate slower than G1, suggesting that these fractions have a more flexible conformation than G1. The far-UV CD spectra of G1 and G3 are shown in Fig. 7B, G2 has been omitted due to the high degree of similarity with G3. Compared with the native HPI and frdHPI, both G1 and G3 retained little secondary structure. At the helix-sensitive wavelength of 222 nm, the molar ellipticity value of G3 is not as negative as that of frdHPI, indicating a lower helix content of G3 than frdHPI. The predominant G1 unfolding intermediates were collected and purified by HPLC. V8 proteinase digestion was used to characterize the disulfide-linkage pattern of G1 as described previously [32]. Briefly, there are in total seven Glu residues (V8 cleavage site) in the sequence of HPI, hence eight fragments, designated F1–F8 from N to C terminus. Due to the presence of disulfide bonds, the V8- cleaved native HPI will generate fragments F1 and F7 linked by A7-B7 as well as F2 and F8 linked by A20-B19. In the disulfide isomers, peptide fragments are linked by different disulfide bridges, therefore peptide mapping by V8 digestion and HPLC may also be different from that of native HPI. The peptide mixture of V8-digested G1 and native HPI were separated by HPLC as shown in Fig. 7C. Fig. 5. Disulfide stability of PIP (A) and HPI (B) in redox buffer. Lane 1 is the native protein marker. Lanes 2–8 represent that the ratio of GSHtoGSSG(m M /m M ). When the disulfide reshuffling reached equilibrium, the reaction was terminated by addition of IAA to car- boxymethylate the free thiols. The samples were analysed on 15% native PAGE and the gel was stained by Coomassie brilliant blue R250. 1742 C Y. Min et al. (Eur. J. Biochem. 271) Ó FEBS 2004 Fig. 7. Physiochemical properties and disulfide-linkage patterns of the intermediates. (A)NativePAGE(15%acrylamide)ofthedisulfide isomers G1, G2 and G3. P34 represents the mixture of intermediates P3 (upper) and P4 (lower) captured during the oxidized refolding process of HPI. The frdHPI is the IAA modified reduced/denatured HPI. (B) Far-UV CD spectra of the disulfide isomers of HPI. G2 is not shown due to the high degree of similarity with G3. The protein concentration used was 0.25 mgÆmL )1 for all the samples. (C) Peptide mapping of HPI and G1 after digestion by endoproteinase V8. HPI and isomer G1 were digested with the endoproteinase V8, and the mixtures were analysed by HPLC on a reversed phase C8 column. The peptide in peak f5 of G1 has a molecular mass of 2347, which shows that G1 contains an intra-B chain disulfide bond. Fig. 6. Time-course unfolding of HPI in the denaturing buffer containing 6.0 M GdnHCl and 0.2 m M 2-mercaptoethanol. The unfolding reaction was quenched at different time points, as indicated to the right of each HPLC chromatograph, by adjusting the pH to 1.0 with trifluoroacetic acid; samples were then analysed by HPLC using the conditions des- cribed in Experimental procedures. The disulfide scrambling process of HPI under these conditions always reaches equilibrium after a reaction time of 2 h. Ó FEBS 2004 Possible role of C-peptide in the folding/unfolding of HPI (Eur. J. Biochem. 271) 1743 The fractions from each peak were collected and the peptides were identified by ESI-MS. Compared with native HPI, a remarkable peak designated f5 could be observed in the digestion mixture of G1. The molecular weight of the fragment in peak f5 was 2347.0, suggesting that the peptide in f5 corresponded to the fragments F1 and F2 linked by disulfide B7-B19. The profile of the enzyme digestion pattern of the G1 intermediates is almost identical with the intermediate P4 captured during the HPI refolding process [32], which indicates that G1 may have the same disulfide linkage as the P4 intermediate. Reverse refolding of disulfide isomer G1 to native HPI Given that the G1 isomer in the unfolding process may be the same intermediate as the P4 refolding intermediate, we questioned if the reverse refolding of the G1 isomer occurs by the same process as that of P4. To initiate the refolding of the scrambled G1 disulfide isomer, a low concentration of 2-mercaptoethanol was used as the thiol catalyst. In the presence of 0.25 m M 2-mercaptoethanol in alkaline buffer, G1 was able to spontaneously reshuffle its non-native disulfide bonds until native configuration of HPI was adopted, as shown in Fig. 8. During the refolding process of G1, only a few accumulated intermediates were observed by HPLC, among which one major peak corresponded, in elution time, to the refolding intermediates P2 of HPI. This indicates that possession of the A20-B19 disulfide bond in P2 is an important intermediate step during the disulfide reshuffling of G1 into native HPI. Taken together, the refolding process of the G1 isomer in Fig. 8 is very similar to that of P4 as reported previously, further indicating that G1 and P4 are the same intermediates during the unfolding and refolding process of HPI, respectively. The same inter- mediate captured in the unfolding and refolding process suggests that a correlation exists in the molecular mechan- ism of the unfolding or refolding of HPI. Discussion Although the insulin A- and B-chains contain sufficient structural information for the correct pairing of the disulfide bonds [33,34], our refolding studies of HPI have shown that both the A and B chain as well as the C-peptide contain the information necessary for proper protein folding. Com- pared with the cooperative, step-by-step formation of disulfide bonds and native conformation in the folding pathway of a mini-proinsulin (PIP) which lacks the C-peptide [35], HPI rapidly adopts a random formation of all the intramolecular disulfide bonds during an early stage of the oxidized folding process. As a result of the observed differences in the molecular folding process of PIP and HPI, as shown in Fig. 1B and C, we can conclude that the main function of the C-peptide in the folding process is to provide the necessary flexibility for the formation of intramolecular disulfide bonds. However, the manner in which the C-peptide is able to provide this flexibility is unknown. Although the three-dimensional structure of HPI and PIP has not been completed, many physicochemical data support that their core structure is similar to that of insulin [40–42], therefore the presence of the C-peptide should not influence dramatically the conformation of the A- and Fig. 8. Refolding of scrambled disulfide isomer G1 to native HPI. The isomer G1 was reconstituted in alkaline buffer containing trace amounts of 2-mercaptoethanol to initiate refolding. The refolding reaction was quenched at different time points and analysed on HPLC by using the conditions described in the Experimental procedures. The intermediates formed during the refolding reaction were identified based on similar elution time with the purified refolding intermediates such as P1, P2, P3 and P4. 1744 C Y. Min et al. (Eur. J. Biochem. 271) Ó FEBS 2004 B-chains. During this unfolding study, we compared the disulfide stability of HPI with PIP using disulfide scrambling methods and the redox-equilibrium assay. Our results show that the disulfide bonds in HPI are more easily disrupted by the addition of thiol reagents than those of PIP, indicating that the C-peptide of HPI reduces the stability of the disulfide bonds more than that of PIP. The reduced disulfide bond stability of HPI may explain why HPI can randomly form all of the intramolecular disulfide bonds at the beginning of the refolding process. We can therefore deduce that the C-peptide affects the HPI refolding process by influencing the stability of its disulfide bonds. Due to the absence of structural information for the C-peptide, we are not able to determine how the C-peptide interacts with the insulin A- or B-chains to make the disulfide bonds more accessible than that PIP. It’s possible that the longer linker between the B- and A-chains may make the C terminus of the B-chain more flexible. There are examples that conformational stability of a protein can be modulated by changing the lengths of loop or linker segments. For example, a four-helix bundle protein Rop has been shown to have inverse correlation between loop length and stability [43]. The effects of the linkers have generally been attributed to the increased entropic penalty associated with fixing the end positions of longer linkers. Considering the passive role of the linker in proteins like Rop, we may question whether the role of C-peptide in HPI refolding is also passive and simply a flexible longer linker. However, there are at least three examples that have shown that the 31-amino acid C-peptide does not act as a simple linker. First, replacing the native C-peptide of HPI with different short linkers always resulted in lower expression level and higher disulfide isomers formation in the mam- malian cells [44], thus the native C-peptide of HPI is important for its refolding in vivo. Secondly, either alanine scanning mutagenesis or deletion of three highly conserved acidic residues (EAED) at the N terminus of the C-peptide resulted in severe HPI aggregation during refolding [45]. This suggests that the amino acid composition of the C-peptide is also an important factor for its function. Finally, the in vitro refolding yield of HPI could easily be optimized, whereas it is difficult to efficiently refold PIP under the same conditions [32]. In summary, we may deduce that the C-peptide of HPI contains important folding information necessary for the correct pairing of disulfide bonds. Our work shows that HPI can be denatured and reshuffle its disulfide bonds to form a series of disulfide isomers in the presence of denaturant and a trace thiol catalyst, with isomer G1 being the most abundant isomers identified. The CD spectrum and native PAGE of G1 showed that it retained little secondary structure and adopts a flexible conformation, as observed for frdHPI. Together with the result that more than 95% of native HPI can be converted into the isomer G1 in the presence of strong denaturant (6.0 M GdnHCl) and thiol reagent, we may suggest that the G1 is the predominant fraction of unfolded HPI with three disulfide bonds. The disulfide linkage analysis of G1 shows that it contains the intra-B chain disulfide bond, B7-B19, and two intra-A chain disulfide bonds. Due to the absence of peptide sequencing analysis, we were not able to determine the disulfide linkage pattern in the intra-A chain. Insulin-like growth factor-I (IGF-I), which is homologous to HPI, has been investigated by using disulfide scrambling with the similar condition used for disulfide scrambling. Three major disulfide isomers of IGF-I, namely IGF-a, IGF-b1 and GF-b2, respectively, were identified and their disulfide linkage patterns were analysed by Edman sequen- cing and peptide mapping [46]. Comparison of the results of IGF-I with that of HPI, considering the high primary sequence homology and 3D structure of the insulin superfamily, we can deduce that the disulfide linkage pattern of G1 corresponds to the predominant disulfide isomer IGF-b1 or IGF-b2, which should be [B7-B19, A6-A11, A7-A20] or [B7-B19, A6-A20, A7-A11]. During the disulfide scrambling process of proteins, such as PCI [16], TAP [47], and hirudin [48], the predominant isomer always contains the disulfide linkage pattern in which the nearest cysteines in primary sequence pair and form the beads-form disulfide bonds. Although the isomer IGF-a, with a Cys47-Cys48 disulfide bond, is adopted to the pattern of consecutive disulfide linkage, it may be absent from the folding pathway of fully reduced IGF due to its poor solubility [49]. There may not be IGFa-like isomers during the unfolding of HPI because all the isomers are highly soluble. Maybe this is one of the reasons why IGF-I has a swap form while HPI/insulin has not. We have studied the oxidized refolding pathway of HPI and captured four disulfide isomers as intermediates. P4 was identified as the most unstructured intermediate and contains the disulfide bond B7-B19 [32]. In this study, we found that HPI was converted mainly into a disulfide isomer G1 during its unfolding in the presence of denaturant and a trace thiol reagent. The native electrophoresis, CD spectrum, disulfide linkage analysis and the HPLC beha- viour of G1 strongly suggest that it is identical to P4. The identical intermediate captured during the oxidized refold- ing and disulfide-scrambling unfolding suggests that the pathway of unfolding and refolding of HPI might be similar but in the reverse direction, which is consistent with the underlying mechanism of protein folding proposed by Chang [46]. This also indicates that disulfide-scrambling unfolding may be used as a reversible step to investigate the folding pathway of proteins. A key question in protein folding is how the folding initiates from a random-coiled peptide chain [50]. In order to solve this question, it is necessary to determine the 3D structure of the unfolded peptide. However, since the reduced/denatured protein adopts numerous conforma- tional isomers, it is very difficult to complete a 3D structure analysis. The isomer G1 obtained here may provide a proper reduced/denatured HPI state for 3D structure analysis by NMR. The reasons are as follows: (a) G1 is a major scrambled isomer with three stable intradisulfide bonds; (b) CD spectra show that G1 retains little secondary structure; (c) G1 can reshuffle its disulfide bonds until adopting the native conformation, which indicates that G1 may be function as the early intermediate in the oxidized refolding of HPI; (d) the 3D structure of insulin and several of its analogues have been well studied, which will provide a comparison for the G1 3D structure. The elucidation of G1 structure will help us to partially understand which amino acids in the random-coiled peptide have the potential to participate in the formation of folding initiation sites in Ó FEBS 2004 Possible role of C-peptide in the folding/unfolding of HPI (Eur. J. Biochem. 271) 1745 HPI, and to further learn the molecular mechanism of the initiation of HPI folding. Acknowledgements We thank Profs M. A. Weiss and Q X. Hua for providing human proinsulin and helpful discussion. We are grateful to K. Brazine at the Dana-Farber Cancer Institute for critical reading of this manuscript. This work was supported by the grants from the National Foundation of Natural Science (No.39670179) and Chinese Academy of Sciences (KJ951-B1-606). References 1. Shortle, D. (1996) The denatured state (the other half of the fold- ing equation) and its role in protein stability. FASEB J. 10, 27–34. 2. Rader, A.J., Hespenheide, B.M., Kuhn, L.A. & Thorpe, M.F. (2002) Protein unfolding: rigidity lost. Proc. Natl Acad. Sci. USA 99, 3540–3545. 3. Hespenheide, B.M., Rader, A.J., Thorpe, M.F. & Kuhn, L.A. (2002) Identifying protein folding cores from the evolution of flexible regions during unfolding. J. Mol. Graph. Model. 21, 195–207. 4. Matouschek, A. (2003) Protein unfolding – an important process in vivo? Curr. Opin. Struct. Biol. 13, 98–109. 5. Li, R., Battiste, J.L. & Woodward, C. (2002) Native-like inter- actions favored in the unfolded bovine pancreatic trypsin inhibitor have different roles in folding. Biochemistry 41, 2246–2253. 6. Radford, S.E., Dobson, C.M. & Evans, P.A. (1992) The folding of hen lysozyme involves partially structured intermediates and multiple pathways. Nature 358, 302–307. 7. Houry, W.A., Rothwarf, D.M. & Scheraga, H.A. (1995) The nature of the initial step in the conformational folding of disulphide-intact ribonuclease A. Nat. Struct. Biol. 2, 495–503. 8. Creighton, T.E. (1984) Disulfide bond formation in proteins. Methods Enzymol. 107, 305–329. 9. Creighton, T.E. (1979) Intermediates in the refolding of reduced ribonuclease A. J.Mol.Biol.129, 411–431. 10. Li, Y.J., Rothwarf, D.M. & Scheraga, H.A. (1995) Mechanism of reductive protein unfolding. Nat. Struct. Biol. 2, 489–494. 11. Mendoza, J.A., Jarstfer, M.B. & Goldenberg, D.P. (1994) Effects of amino acid replacements on the reductive unfolding kinetics of pancreatic trypsin inhibitor. Biochemistry 33, 1143–1148. 12. Ewbank, J.J. & Creighton, T.E. (1993) Pathway of disulfide- coupled unfolding and refolding of bovine alpha-lactalbumin. Biochemistry 32, 3677–3693. 13. Creighton, T.E. & Goldenberg, D.P. (1984) Kinetic role of a meta- stable native-like two-disulphide species in the folding transition of bovine pancreatic trypsin inhibitor. J. Mol. Biol. 179, 497–526. 14. Kuwajima, K., Ikeguchi, M., Sugawara, T., Hiraoka, Y. & Sugai, S. (1990) Kinetics of disulfide bond reduction in alpha-lactalbumin by dithiothreitol and molecular basis of superreactivity of the Cys6–Cys120 disulfide bond. Biochemistry 29, 8240–8249. 15. Chang, J.Y. (1997) A two-stage mechanism for the reductive unfolding of disulfide-containing proteins. J. Biol. Chem. 272, 69–75. 16. Chang, J.Y., Li, L., Canals, F. & Aviles, F.X. (2000) The unfolding pathway and conformational stability of potato car- boxypeptidase inhibitor. J.Biol.Chem.275, 14205–14211. 17. Salamanca, S., Villegas, V., Vendrell, J., Li, L., Aviles, F.X. & Chang, J.Y. (2002) The unfolding pathway of leech carboxy- peptidase inhibitor. J. Biol. Chem. 277, 17538–17543. 18. Chang, J.Y. & Kumar, T.K.C. (1998) Unfolding and refolding of cardiotoxin III elucidated by reversible conversion of the native and scrambled species. Biochemistry 37, 6745–6751. 19. Steiner, D.F. & Chan, S.J. (1988) An overview of insulin evolu- tion. Horm. Metab. Res. 20, 443–444. 20. Yan, H., Guo, Z.Y., Gong, X.W., Xi, D. & Feng, Y.M. (2003) A peptide model of insulin folding intermediate with one disulfide. Protein Sci. 12, 768–775. 21.Weiss,M.A.,Hua,Q.X.,Jia,W.,Chu,Y.C.,Wang,R.Y.& Katsoyannis, P.G. (2000) Hierarchical protein Ôun-designÕ:insu- lin’s intrachain disulfide bridge tethers a recognition alpha-helix. Biochemistry 39, 15429–15440. 22. Guo, Z.Y. & Feng, Y.M. (2001) Effects of cysteine to serine substitutions in the two inter-chain disulfide bonds of insulin. Biol. Chem. 382, 443–448. 23. Hua,Q.X.,Chu,Y.C.,Jia,W.,Phillips,N.F.,Wang,R.Y.,Kat- soyannis, P.G. & Weiss, M.A. (2002) Mechanism of insulin chain combination. Asymmetric roles of A-chain alpha-helices in disul- fide pairing. J. Biol. Chem. 277, 43443–43453. 24.Hua,Q.X.,Nakagawa,S.H.,Jia,W.,Hu,S.Q.,Chu,Y.C., Katsoyannis, P.G. & Weiss, M.A. (2001) Hierarchical protein folding: asymmetric unfolding of an insulin analogue lacking the A7–B7 interchain disulfide bridge. Biochemistry 40, 12299–12311. 25. Hua, Q.X., Gozani, S.N., Chance, R.E., Hoffmann, J.A., Frank, B.H. & Weiss, M.A. (1995) Structure of a protein in a kinetic trap. Nat. Struct. Biol. 2, 129–138. 26. Steiner, D.F. (1978) On the role of the proinsulin C-peptide. Diabetes 27 (Suppl. 1), 145–148. 27. Millican, R.L. & Brems, D.N. (1994) Equilibrium intermediates in the denaturation of human insulin and two monomeric insulin analogs. Biochemistry 33, 1116–1124. 28. Bryant,C.,Strohl,M.,Green,L.K.,Long,H.B.,Alter,L.A., Pekar, A.H., Chance, R.E. & Brems, D.N. (1992) Detection of an equilibrium intermediate in the folding of a monomeric insulin analog. Biochemistry 31, 5692–5698. 29. Brems, D.N., Brown, P.L., Heckenlaible, L.A. & Frank, B.H. (1990) Equilibrium denaturation of insulin and proinsulin. Biochemistry 29, 9289–9293. 30. Markussen, J. (1971) Structural changes involved in the folding of proinsulin. Int. J. Protein Res. 3, 201–207. 31. Guo, Z.Y., Qiao, Z.S. & Feng, Y.M. (2001) Unfolding of recombinant single-chain insulin in denaturants containing thiol reagents. ShengWuHuaXue.YuShengWuWuLiXue.Bao. (Shanghai) 33, 431–436. 32. Qiao,Z.S.,Min,C.Y.,Hua,Q.X.,Weiss,M.A.&Feng,Y.M. (2003) In vitro refolding of human proinsulin. Kinetic inter- mediates, putative disulfide-forming pathway folding initiation site, and potential role of C-peptide in folding process. J. Biol. Chem. 278, 17800–17809. 33. Tang, J.G. & Tsou, C.L. (1990) The insulin A and B chains con- tain structural information for the formation of the native mole- cule. Studies with protein disulphide-isomerase. Biochem. J. 268, 429–435. 34. Wang, C.C. & Tsou, C.L. (1991) The insulin A and B chains contain sufficient structural information to form the native molecule. Trends Biochem. Sci. 16, 279–281. 35. Qiao, Z.S., Guo, Z.Y. & Feng, Y.M. (2001) Putative disulfide- forming pathway of porcine insulin precursor during its refolding in vitro. Biochemistry 40, 2662–2668. 36. Frank, B.H., Veros, A.J. & Pekar, A.H. (1972) Physical studies on proinsulin. A comparison of the titration behavior of the tyrosine residues in insulin and proinsulin. Biochemistry 11, 4926–4931. 37. Guo, Z.Y., Shen, L. & Feng, Y.M. (2002) The different folding behavior of insulin and insulin-like growth factor 1 is mainly controlled by their B-chain/domain. Biochemistry 41, 1556–1567. 1746 C Y. Min et al. (Eur. J. Biochem. 271) Ó FEBS 2004 [...]...Ó FEBS 2004 Possible role of C-peptide in the folding /unfolding of HPI (Eur J Biochem 271) 1747 38 Guo, Z.Y., Shen, L & Feng, Y.M (2002) The different energetic state of the intra A-chain/domain disulfide of insulin and insulinlike growth factor 1 is mainly controlled by their B-chain/domain Biochemistry 41, 10585–10592 39 Guo, Z.Y., Tang, Y.H., Wang, S & Feng, Y.M (2003) Contribution of the absolutely... foldability of insulin Biol Chem 384, 805–809 40 Steiner, D.F (1973) Cocrystallization of proinsulin and insulin Nature 243, 528–530 41 Snell, C.R & Smyth, D.G (1975) Proinsulin: a proposed threedimensional structure J Biol Chem 250, 6291–6295 42 Weiss, M.A., Frank, B.H., Khait, I., Pekar, A., Heiney, R., Shoelson, S.E & Neuringer, L.J (1990) NMR and photo-CIDNP studies of human proinsulin and prohormone... Acidic residues on the N-terminus of proinsulin C-Peptide are important for the folding of insulin precursor J Biochem (Tokyo) 131, 855–859 46 Chang, J.Y., Li, L & Bulychev, A (2000) The underlying mechanism for the diversity of disulfide folding pathways J Biol Chem 275, 8287–8289 47 Chang, J.Y (1999) Denatured states of tick anticoagulant peptide Compositional analysis of unfolded scrambled isomers... of human proinsulin and prohormone processing intermediates with application to endopeptidase recognition Biochemistry 29, 8389–8401 43 Nagi, A.D & Regan, L (1997) An inverse correlation between loop length and stability in a four-helix-bundle protein Fold Des 2, 67–75 44 Liu, M., Ramos-Castaneda, J & Arvan, P (2003) Role of the connecting peptide in insulin biosynthesis J Biol Chem 278, 14798–14805... unfolded scrambled isomers J Biol Chem 274, 123–128 48 Chang, J.Y (1995) The properties of scrambled hirudins J Biol Chem 270, 25661–25666 49 Chang, J.Y., Marki, W & Lai, P.H (1999) Analysis of the extent of unfolding of denatured insulin-like growth factor Protein Sci 8, 1463–1468 50 Baldwin, R.L (1989) How does protein folding get started? Trends Biochem Sci 14, 291–294 . Unfolding of human proinsulin Intermediates and possible role of its C-peptide in folding /unfolding Cheng-Yin Min, Zhi-Song Qiao and You-Min Feng Institute. synthesized in vivo as a single-chain precursor (preproinsulin) and folded as proinsulin, in which a connecting peptide of 35 residues links the C terminus of the