Role of Kinetic Intermediates in the Folding of Leech Carboxypeptidase Inhibitor*

The oxidative folding and reductive unfolding pathways of leech carboxypeptidase inhibitor (LCI; four disulfides) have been characterized in this work by structural and kinetic analysis of the acid-trapped folding intermediates. The oxidative folding of reduced and denatured LCI proceeds rapidly through a sequential flow of 1-, 2-, 3-, and 4-disulfide (scrambled) species to reach the native form. Folding intermediates of LCI comprise two predominant 3-disulfide species (designated as III-A and III-B) and a heterogeneous population of scrambled isomers that consecutively accumulate along the folding reaction. Our study reveals that forms III-A and III-B exclusively contain native disulfide bonds and correspond to stable and partially structured species that interconvert, reaching an equilibrium prior to the formation of the scrambled isomers. Given that these intermediates act as kinetic traps during the oxidative folding, their accumulation is prevented when they are destabilized, thus leading to a significant acceleration of the folding kinetics. III-A and III-B forms appear to have both native disulfides bonds and free thiols similarly protected from the solvent; major structural rearrangements through the formation of scrambled isomers are required to render native LCI. The reductive unfolding pathway of LCI undergoes an apparent all-or-none mechanism, although low amounts of intermediates III-A and III-B can be detected, suggesting differences in protection against reduction among the disulfide bonds. The characterization of III-A and III-B forms shows that the former intermediate structurally and functionally resembles native LCI, whereas the III-B form bears more resemblance to scrambled isomers.

The new view of protein folding, which has emerged in the recent years from a combination of experimental work and theoretical approximations, postulates the folding process as a parallel flow of molecules that follow multiple folding routes to reach the native state (1,2). As folding proceeds, some semi-stable conformations corresponding to local free energy minima (intermediates) may be transiently accumulated, acting as kinetic traps (3). Thus, understanding protein folding requires identification of the intermediate(s) that form(s) along the preferential pathways leading from the unfolded state to the native form (4). Unfortunately, characterizing kinetic folding intermediates is usually a difficult issue because of their short half-life. An important part of our knowledge about the role and nature of intermediates along the folding process comes from studies of disulfide-rich proteins in which transient folding forms have been trapped and characterized (5).
Oxidative folding is one of the well-established methods used to analyze the folding of disulfide-containing proteins (6 -14). For these proteins, the folding pathway is characterized and defined by the heterogeneity and structures of the disulfide isomers that accumulate along the folding process. Folding intermediates can be trapped by acidification of the protein solution and separated by reversed-phase high performance liquid chromatography (RP-HPLC), 1 which allows their further structural characterization. Application of the oxidative folding and acid-trapping method has allowed the elucidation of folding pathways of several 3-disulfide proteins such as hirudin (15,16), potato carboxypeptidase inhibitor (PCI) (17,18), tick anticoagulant peptide (TAP) (19,20), epidermal growth factor (21,22), insulin-like growth factor (IGF-1) (23,24), and the extensively investigated model of bovine pancreatic trypsin inhibitor (BPTI) (6, 7, 9 -11). However, few models aside from ribonuclease A (RNase A) and ␣-lactalbumin (␣LA) have been studied in detail among 4-disulfide proteins (25)(26)(27)(28)(29)(30)(31)(32). In these cases, analysis of the folding pathway represents another level of technical challenge because of the increase in the number of possible disulfide intermediates.
The above-mentioned studies have not indicated any predominant folding scenario, and even among small 3-disulfide proteins the folding mechanism varies substantially. For proteins as BPTI, intermediates with native disulfide bonds and native-like structures prevail along the folding pathway (10,11). The non-covalent interactions that stabilize the native BPTI play a crucial role in guiding the early folding events and hence dictate the formation of a limited number of intermediates that admit the prevalence of native disulfides. In the case of hirudin and PCI, two other 3-disulfide proteins, folding proceeds through an initial nonspecific disulfide pairing (packing) that leads to the formation of a heterogeneous population of 3-disulfide scrambled isomers; this is followed by disulfide reshuffling (consolidation) of these intermediates to finally acquire the native form (15,17). For the latter proteins, noncovalent interactions do not seem to participate significantly in guiding protein folding during the early phase of nonspecific packing. Within this context, folding studies of novel protein models are required to better understand the underlying causes of such a diversity of disulfide folding pathways.
Leech carboxypeptidase inhibitor (LCI) is a 66-residue cysteine-rich protein that folds in a compact domain consisting of a five-stranded antiparallel ␤-sheet and a short ␣-helix, as reported by our group (Fig. 1) (33). The molecule is stabilized by four disulfide bridges, which are all located within secondary structure elements (Fig. 1). LCI is a potent metallocarboxypeptidase inhibitor that binds tightly to pancreatic carboxypeptidases A1, A2, B (CPA1, CPA2, CPB) and to plasma CPB, also called thrombin-activable fibrinolysis inhibitor (TAFI) (34). Assuming that leeches secrete LCI during feeding, LCI may participate in the elimination of blood clots by inhibiting TAFI, an enzyme shown to retard fibrinolysis (35,36). LCI could help to maintain the liquid state of the blood during feeding and possibly block the host defense mechanisms involving mast cell proteases (33). The profibrinolytic effect of LCI has been demonstrated in vitro, suggesting a potential pharmacological application in thrombotic diseases. 2 We have recently described both, the unfolding pathway and thermodynamic stability (37), and the oxidative folding process of this protein (38), showing that 3-and 4-disulfide intermediates act as kinetic traps along its folding pathway. In the present work, we study in depth the kinetic, thermodynamic, conformational, and functional properties of several disulfide intermediates along the pathways of oxidative folding and reductive unfolding of LCI.

EXPERIMENTAL PROCEDURES
Materials-Recombinant LCI was obtained by heterologous expression in Escherichia coli with an added glycine at the N terminus. The protein was purified by ion-exchange chromatography on a TSK-DEAE column (Tosohaas), followed by RP-HPLC (34). The recombinant protein was more than 99% pure, as judged by HPLC analysis. The chromogenic substrates N-(4-methoxyphenylazoformyl)-Phe-OH and N-(4methoxyphenylazoformyl)-Arg-OH were obtained from Bachem. Bovine CPA was purchased from Sigma. Human CPA1, CPA2, and CPB were prepared following described procedures (39). Dithiothreitol (DTT), guanidine hydrochloride (GdnHCl), thermolysin (P-1512), and 2-mercaptoethanol were purchased from Sigma with purities greater than 99%.
Oxidative Folding of Fully Reduced LCI-Native LCI (1 mg) was reduced and denatured in Tris-HCl buffer (0.1 M, pH 8.4) containing 8 M GdnHCl and 50 mM DTT, at 22°C for 2 h. To initiate folding, the sample was passed through a PD-10 column (Sephadex-25, Amersham Biosciences), previously equilibrated with Tris-HCl buffer (0.1 M, pH 8.4). Reduced and denatured LCI was recovered in 1.2 ml and immediately diluted to a final protein concentration of 0.5 mg/ml in the same Tris-HCl buffer, both in the absence (control, Ϫ) and presence (control, ϩ) of 0.25 mM 2-mercaptoethanol. Folding intermediates of LCI were trapped in a time course manner at selected times by mixing aliquots of the sample with 2% trifluoroacetic acid. Trapped folding intermediates were analyzed by RP-HPLC.
Analysis of the Folding Intermediates of LCI by RP-HPLC-Analysis and isolation of folding intermediates of LCI were achieved by RP-HPLC using the following conditions. Solvent A was 0.1% trifluoroacetic acid and solvent B acetonitrile containing 0.1% trifluoroacetic acid. The column used was a 4.6 mm Protein C4 (Vydac). A linear 20 -40% gradient of solvent B was applied over 50 min, with a flow rate of 0.75 ml/min.
Stop/Go Folding-Acid-trapped intermediates were isolated by RP-HPLC, freeze-dried, and allowed to carry on the folding by dissolving the sample (0.5 mg/ml) in Tris-HCl buffer (0.1 M, pH 8.4), both in the absence and presence of 0.25 mM 2-mercaptoethanol. Folding intermediates were trapped with 2% trifluoroacetic acid and analyzed by RP-HPLC. Scrambled isomers of LCI were separated from 3-disulfide intermediates by treatment with vinylpyridine and further isolation by RP-HPLC.
Oxidative Folding of LCI in the Presence of Denaturants-The procedures of unfolding and refolding were as described in the oxidative folding experiments. Immediately after the desalting of unfolded LCI through a PD-10 column, selected concentrations of denaturants (0.5-5 M GdnHCl, 1-8 M urea) were added. Folding intermediates were similarly trapped by acidification and analyzed by RP-HPLC.
Reductive Unfolding-Native LCI and the 3-disulfide intermediates (0.5 mg) were dissolved in 1 ml of Tris-HCl buffer (0.1 M, pH 8.4) with different concentrations of DTT (0.1-100 mM). Reduction was carried out at 22°C. To monitor the kinetics of unfolding, time course aliquots of the samples were trapped with 2% trifluoroacetic acid, and analyzed by RP-HPLC. In addition, native LCI was dissolved in the above- with 10 mM tributylphosphin and analyzed by MS to identify their peptidic composition. The N-terminal sequence of selected peptides was also analyzed by automated Edman degradation.
Deuterium to Proton Exchange Followed by MS-Acid-trapped intermediates were isolated by RP-HPLC and freeze-dried. The samples (50 g) were resuspended in deuterated glycine buffer (20 mM, pD 2.5), incubated at 90°C for 2.5 h to exchange all labile protons and then maintained at room temperature for 1 h to promote protein refolding. The deuterated proteins were diluted 1:4 with ammonium citrate (50 mM, pH 4.0) to start the hydrogen exchange. Aliquots were taken at different time points and analyzed by matrix-assisted laser desorption/ ionization time-of-flight/mass spectrometry (MALDI-TOF/MS) until an exchange plateau was reached. Samples were prepared by mixing equal volumes of the protein solution and matrix solution (sinapic acid in 30% acetonitrile with 0.1% trifluoroacetic acid). At each exchange time, six samples were analyzed by duplicate. The average of the mass values, corresponding to the centroid of the peaks, was calculated for each exchange time and compared with an external unlabeled control, whose mass was determined by duplicate measurements.
Mass Spectrometry and Amino Acid Sequencing-The molecular masses of disulfide-containing peptides were determined by MALDI-TOF/MS on a Bruker Ultraflex spectrometer. Samples for the deuterium to proton (D/H) exchange experiments were analyzed by the same spectrometer. The amino acid sequences of selected thermolytic peptides were analyzed by automatic Edman degradation using a Beckman LF3000 Protein Sequencer.
Circular Dichroism and NMR Spectroscopy-Samples for circular dichroism (CD) spectroscopy were prepared by dissolving the protein to a final concentration of 0.2 mg/ml in 0.1% trifluoroacetic acid (pH 2.0). CD analyses were carried out in a Jasco J-715 spectrometer at 25°C using a cell of 2-mm path length. Protein samples for 1 H NMR experiments were prepared by dissolving the protein in H 2 O/D 2 O (9:1 ratio, v/v) with a concentration of 1 mg/ml at pH 2.0. NMR spectra were acquired on a Bruker AMX 500-MHz spectrometer at 25°C.
CP Inhibitory Activity-The inhibitory activity of selected LCI folding intermediates was assayed by measuring the inhibition of the hydrolysis of the chromogenic substrate N-(4-methoxyphenylazoformyl)-Phe-OH by CPAs and N-(4-methoxyphenylazoformyl)-Arg-OH by CPB. The assay was performed in Tris-HCl buffer (50 mM, pH 7.5) containing 100 mM NaCl, with a substrate concentration of 100 M. The inhibition constants (K i ) for the complexes of LCI intermediates with different carboxypeptidases were determined at the presteady-state as described for tightly binding inhibitors (40). The protein concentration of the LCI intermediates was determined from the A 280 of the solution (LCI extinction coefficient: E 0.1% ϭ 2.12).

Accumulation of 3-Disulfide Intermediates and Scrambled
Isomers Along the Oxidative Folding Pathway of LCI-Oxidative folding of fully reduced LCI was carried out in the Tris-HCl buffer in the absence and presence of 2-mercaptoethanol as thiol catalyst. The RP-HPLC profiles of acid-trapped folding intermediates at selected time points are shown in Fig. 2. A high degree of heterogeneity of intermediates is observed at the beginning of the folding reaction, with identical RP-HPLC profiles in both refolding conditions (control Ϫ and control ϩ). This initial stage is followed by the accumulation of two fractions (III-A and III-B) of major intermediates that act as kinetic traps. At this point (at 8 h), the RP-HPLC patterns are similar regardless of the presence of a reducing agent. The last stage of the folding process is characterized by an accumulation of a heterogeneous population of intermediates, which is more pronounced when the refolding is performed in the absence of a thiol catalyst (control Ϫ).
Purified intermediates from the RP-HPLC analyses were derivatized with vinylpyridine, and analyzed by MALDI-TOF/MS to evaluate the disulfide bond content of the folding intermediates. Folding of LCI was shown to undergo a sequential conversion through 1-, 2-, 3-, and 4-disulfide intermediates to reach the native structure (data not shown). Both 3-disulfide (III-A and III-B) and a mixture of non-native 4-disulfide (scrambled) isomers co-exist as folding intermediates and major kinetic traps of LCI folding. The folding of LCI cannot reach completion in the absence of a thiol catalyst, indicated by the fact that only ϳ30% of the protein was recovered in the native form after 48 h of refolding (Fig. 2). In the presence of 2-mercaptoethanol, the recovery of native LCI was more than 90%, confirming the role of this redox agent in promoting the disulfide reshuffling and the conversion of scrambled forms to their native conformation.
Evolution of the 3-Disulfide Intermediates and Scrambled Isomers along the Oxidative Folding Pathway of LCI-Our previous study on the oxidative folding of LCI revealed the presence of at least two 3-disulfide intermediates (III-A1 and III-A2) in fraction III-A and one 3-disulfide intermediate in fraction III-B after 8 h of refolding (38). Assignment of their disulfide pairings showed that isomers III-A2 and III-B contain three native disulfide bonds: Cys 11 -Cys 34 , Cys 18 -Cys 62 and either Cys 19 -Cys 43 or Cys 22 -Cys 58 , respectively, while isomer III-A1 contains one native and two non-native disulfide bonds: Cys 11 -Cys 34 , Cys 19 -Cys 62 , and Cys 18 -Cys 43 .
In the present work, purified fractions III-A and III-B from different refolding time points were derivatized with vinylpyridine (at pH 8.4), and analyzed by RP-HPLC in order to know their composition in disulfide isomers along the folding process. The analysis showed that fraction III-B only contains one predominant 3-disulfide intermediate along the reaction, with the three native disulfide bonds previously described (data not shown). In contrast, within fraction III-A other 3-disulfide bonded forms were detected apart from the two previously characterized species (III-A1 and III-A2). This heterogeneity was not observed when the derivatization with vinylpyridine was performed at pH 6.4 with a lower protein concentration (data not shown). This fact suggests that such heterogeneity could be an artifact caused by the working pH (8.4), the high protein concentration and the conformation of the intermediates, since all these factors might affect the disulfide exchange rate. Structural analysis of the only species observed at pH 6.4 shows that it corresponds to the folding intermediate III-A2. Therefore, from now on, this species will be designated in the text as III-A. Thus, both 3-disulfide kinetic traps that populate LCI folding (III-A and III-B) correspond to species containing three native disulfide bridges.
To further assess the kinetic role of the 3-disulfide intermediates and scrambled 4-disulfide isomers, we performed stop/go experiments on these species. Acid-trapped intermediates III-A and III-B were isolated and allowed to resume the folding in the absence and presence of 2-mercaptoethanol. The data presented in Fig. 3 clearly show how such intermediates interconvert along the folding reaction reaching an equilibrium that is slightly biased toward the III-A intermediate and finally form the 4-disulfide scrambled isomers. At this initial stage the RP-HPLC profiles are indistinguishable regardless of the presence of the thiol catalyst, suggesting that scrambled forms are not yet formed. The equilibrium, which is reached faster beginning from the III-B form than from the III-A form, would represent a rate-limiting step for the folding of LCI. Acidtrapped 4-disulfide (scrambled) isomers were also isolated, separated from 3-disulfide intermediates by treatment with vinylpyridine (which only modifies the latter) followed by RP-HPLC, and allowed to resume the folding in the absence and presence of 2-mercaptoethanol. The reshuffling of non-native 4-disulfide isomers into the native disulfide-bonding pattern takes place directly and supposes yet another rate-limiting step for the folding of LCI (Fig. 3). As expected, in the stop/go experiments of scrambled forms the presence of 2-mercaptoethanol strongly promotes rearrangements, allowing conversion of more than 90% of the scrambled forms into native LCI, while in the absence of the redox agent only ϳ7% of the protein is recovered as native form at the end of the process.
Oxidative Folding of LCI in the Presence of Denaturants-Oxidative folding of LCI was performed in the presence of increasing concentrations of GdnHCl or urea in order to evaluate the influence of denaturant on the prevalence of the 3-disulfide intermediates formed during the LCI folding process (Fig. 4). The comparison of these results with the control folding experiments in Fig. 2 shows that the accumulation of 3-disulfide intermediates decreases in the presence of higher denaturant concentrations. Intermediates III-A and III-B still accumulate under low denaturing conditions (up to 2 M urea concentration), being an indication of the high stability of these species (Fig. 4A). We can also observe a higher prevalence of intermediate III-A under these folding conditions. Interestingly, the higher recovery of native LCI in the presence of 0.5-1 M GdnHCl correlates with a lower accumulation of 3-disulfide intermediates at these conditions (Fig. 4B). In contrast, the refolding process performed in the presence of 1-2 M urea does not alter native LCI recovery.
The folding pathway of LCI drastically changed when the refolding experiments were carried out at high denaturing conditions (more than 2 M GdnHCl or 4 M urea). Examination of time course trapped intermediates revealed that 3-disulfide species no longer accumulate in these conditions and that the reshuffling of the accumulated heterogeneous 4-disulfide scrambled isomers becomes the rate-limiting step of the folding reaction (Fig. 5). Important differences in native LCI recovery are observed in the absence and presence of 2-mercaptoethanol. For instance, when fully reduced LCI is allowed to refold in the presence of 4 M GdnHCl and 2-mercaptoethanol, ϳ35% of the protein attains the native structure after 24 h of refolding, while less than ϳ2% is obtained in absence of the thiol catalyst (Fig. 5).
Reductive Unfolding of Native LCI and 3-Disulfide Intermediates-Reductive unfolding of native LCI was performed at pH 8.4 using various concentrations of DTT as reducing agent. Reduction undergoes an apparent all-or-none mechanism in which only low amounts of partially reduced intermediates accumulate (Fig. 6). The unfolding intermediates were trapped in a time course manner by acidification and were analyzed by RP-HPLC. Two different fractions of 3-disulfide intermediates (assessed by treatment with vinylpyridine and molecular mass analysis) were detected. These species subsequently convert to the fully reduced LCI (R) without a significant buildup of 1-or 2-disulfide intermediates along the pathway. The same behavior was observed when the analysis was performed in the presence of different concentrations of DTT ranging from 2 to 100 mM. The two 3-disulfide intermediates fractions have a RP-HPLC elution equivalent to that of the intermediates III-A and III-B observed along the pathway of oxidative folding (Fig. 2).
The unfolding intermediates were isolated to carry out structural analyses. They were treated with vinylpyridine, further purified by RP-HPLC, and digested with thermolysin. Thermo- lytic peptides were isolated by RP-HPLC and analyzed by MALDI-TOF/MS and Edman sequencing to identify the structures of the disulfide-containing peptides. The results confirm that these unfolding intermediates are indeed identical to the predominant 3-disulfide oxidative folding intermediates of LCI (data not shown). These intermediates accumulate little along the reductive unfolding pathway. At early stages of the process, intermediate III-A comprises only ϳ1% of the total protein, while species III-B represents about 3-4%. When the experiment was performed in the presence of a high concentration of denaturant (4 M GdnHCl), these intermediates did not accumulate at all.
Reductive unfolding of purified intermediates III-A and III-B was also performed at pH 8.4 using various concentrations of DTT. In all conditions, the reduction of the three native disulfides takes place in a cooperative and concerted manner, and both forms unfold to the fully reduced LCI without further accumulation of 1-or 2-disulfide intermediate species (Fig. 6). However, the interconversion between both intermediates can also be observed along the reduction process. Since the interconversion process is slower than the reduction reaction and the intermediate III- liar, with a well-defined ellipticity minimum at 210 nm and a maximum at 228 nm (Fig. 7). The former may be related to the presence of a high percentage of residues in ␤-structure, and the latter to both ␤-structures and loops or to an asymmetric environment of Tyr 64 (34,41). Previous CD spectroscopy measurements showed that the degree of LCI denaturation correlates with the decrease in ellipticity at 228-nm, and complete disappearance of this signature is observed when the protein is completely unfolded (34). The shape of CD spectrum of the III-A species is similar to that of the native protein. However, the 228-nm maximum is only about 30% of that of the native LCI and the minimum in ellipticity shifts to 205 nm (Fig. 7). The CD spectra of the III-B species and scrambled isomers exhibit clear differences from that of the native form. At 228 nm both species show negative values, and the minimum is located at about 200 nm (Fig. 7), indicating that they are less structured intermediates than III-A.
The conformational stability of the above-mentioned LCI folding intermediates was also investigated by D/H exchange experiments followed by MALDI-TOF/MS (42). The extent of hydrogen exchange was quite different for the native form and its folding intermediates. Native LCI retains 27 deuterons at the end of the reaction, whereas the intermediates III-A and III-B, and the scrambled isomers retain 16 deuterons (S.D. Ϯ5%). This ϳ40% of decrease in protected deuterons probably reflects the low level of conformational packing in the interme-diates. However, hydrogen exchange is far from being free and a slow exchange core exists in all these kinetic traps, as expected if they are, at least, partially structured.
Another adequate approach to assess protein conformation is NMR. The 1 H NMR spectra of native LCI and intermediate III-A display very similar signal dispersion, peak sharpness, and upfield/downfield shifted resonances, a clear indication that this intermediate corresponds to a properly folded species (Fig. 8). In contrast, intermediate III-B and scrambled isomers spectra exhibit a clear band broadening and peak collapse (Fig.  8). However, chemical shift dispersion is appreciably greater than expected for a random coil conformation, an additional indication that these intermediates correspond to partially folded species.
CP Inhibitory Activity of the 3-Disulfide Intermediates and Scrambled Forms of LCI-LCI is a tightly binding, competitive inhibitor of carboxypeptidases A and B (34). Equilibrium dissociation constants for the complexes of the 3-disulfide intermediates and scrambled forms with different CPs were determined (Table I). Surprisingly, the inhibitory activities of native LCI and intermediate III-A are practically identical, both in the nanomolar range. In contrast, the inhibitory capabilities of intermediate III-B and scrambled forms are, respectively, one and two orders of magnitude lower than that of the native form. These results correlate well with the conformational properties of the LCI folding intermediates described above and with the deduced degree of "nativeness" of them.

Folding Pathways Among 4-Disulfide Proteins-Oxidative
folding is the process by which a reduced and unfolded disulfide-containing protein gains both its native disulfide bonds and its native structure (8). The disulfide folding pathways of several model proteins have been characterized using the oxidative folding approach, exhibiting an unexpected diversity (32,38). A protein disulfide folding pathway can be characterized by the level of heterogeneity of its folding intermediates, the occurrence of predominant intermediates, and the accumulation of fully oxidized scrambled isomers.
In general, 4-disulfide proteins display a higher extent of secondary structure than 3-disulfide proteins. This fact affects the accessibility, proximity and reactivity of the thiols of the former proteins, drastically interfering in the rates of disulfide bond rearrangement and thus, complicating the folding landscape (43). Therefore, only few 4-disulfide models have been studied in detail. In the case of RNase A, its oxidative folding is characterized by an initial stage of sequential oxidation of the disulfide intermediates, leading to the formation of 1-, 2-, 3-, and 4-disulfide ensembles without any prevalent accumulation (27,31,44). Its rate-limiting step is the formation of two nativelike species containing three native disulfide bonds, which are located in the protein core, protected from reduction and reshuffling. However, their thiols remain accessible to solvent and are subsequently oxidized to form the native protein (Fig. 9).
Another 4-disulfide protein that has been extensively characterized is ␣-lactalbumin, with a folding pathway dependent on the presence of calcium (29,30,32,(45)(46)(47). In its absence, oxidative folding proceeds through heterogeneous 1-to 4-disulfide intermediates, with a final conversion of 4-disulfide scrambled species to the native structure, which represents the major rate-determining step. No native-like conformations are predominant along the folding pathway (Fig. 9). Binding of calcium favors the formation of the ␤-sheet domain of ␣-LA, and then only two major disulfide intermediates with two and three native bonds accumulate along the folding. The formation of the fourth bond accounts for the rate-limiting step of folding in these conditions (Fig. 9).
Oxidative Folding Pathway of LCI-In this context, LCI represents a new 4-disulfide model, which could give us further insight into the oxidative folding pathways. Previously, we have elucidated that denatured and reduced LCI folds through a heterogeneous mixture of 1-and 2-disulfide intermediates, leading to the formation of two populations of intermediates, 3-disulfide species and 4-disulfide (scrambled) isomers, which apparently act as kinetic traps (38). In the present work, we have stated that, as it happens for RNase A and ␣-LA, both predominant 3-disulfide intermediates (III-A and III-B) posses native disulfides, which are directly formed by oxidation from the 2-disulfide ensemble without any detectable accumulation of other 3-disulfide species (Fig. 9). This would suggest that as it happens for RNase A (48), the 2-disulfide ensemble of LCI may be enthalpy-biased toward native disulfide bonds relative to the populations predicted taking into account entropic factors, allowing a faster and preferential formation of the third native disulfide bond.
The use of stop/go experiments clearly demonstrates that the rate of interconversion between the two 3-disulfide intermediates of LCI is much faster than their rate of conversion into scrambled forms (Fig. 9). It also shows that the rate of interconversion is faster from III-B to III-A intermediate, which is found at a slightly higher concentration at equilibrium, probably because of its higher thermodynamic stability and nativeness. III-A and III-B are probably metastable forms equivalent to what Scheraga and co-workers (43) have defined as disulfide-insecure intermediates. Inside such kind of intermediates, thiol groups are as well protected as their disulfide bonds; therefore such thiols cannot be simply exposed and oxidized by a local unfolding process. Structural fluctuations that expose the thiol groups are also likely to expose the disulfide bonds and promote their reshuffling instead of oxidation of the free thiols to the native pairing. In the two LCI intermediates both, disulfide bonds and free thiols, are similarly protected from the solvent. The presence of the external thiol reagent does not affect the first stages of the stop/go experiments of these intermediates, showing that the protein free thiols are not solventaccessible and thus cannot interact with the external reagent.
In RNase A or ␣-LA intermediates, local fluctuations may occur around the thiol groups of the fourth native disulfide bond allowing its oxidation without affecting the overall protein conformation. But LCI has a lower secondary structure content than RNase A or ␣-LA, and the unfolding events in LCI 3-disulfide intermediates are likely to affect the whole core of the molecule leading to an overall rearrangement that also exposes the disulfide bonds.
LCI 3-disulfide intermediates probably differ from the pre-viously described disulfide-insecure species in some aspects. First, they are able to interconvert in a fast way, and so minor local fluctuations may possibly allow solvent-independent disulfide interchange. This disulfide interchange is an internal process in which all the reacting groups are protected from the exterior, since neither the rate of intermediates interconversion nor the concentration of the species at the equilibrium are affected by the presence of external thiols. Secondly, whereas 3-disulfide-insecure species described to date preferentially reshuffle to an unstructured 3-disulfide ensemble forming metastable dead-end pathways, LCI 3-disulfide intermediates III-A and III-B simultaneously oxidize, reshuffle and convert into a heterogeneous population of 4-disulfide scrambled isomers.
Reshuffling of non-native 4-disulfide isomers into the native state is the last stage of the LCI oxidative folding and can be considered as the strongest rate-determining step (Fig. 9). Unlike that of the 3-disulfide intermediates, the disulfide bonds of unstructured scrambled forms are solvent-accessible and the addition of an external thiol group strongly accelerates the kinetics of native-disulfide formation from the scrambled population.
Effect of 3-Disulfide Intermediates Stability on the Folding Pathway of LCI-Despite the absence of a disulfide bond that might stabilize key secondary structural elements, both 3-disulfide intermediates display a striking stability in denaturing environment; so they are located in strong thermodynamic local minima that slow down LCI folding pathway. When LCI folding reaction is performed under mild denaturing conditions promoting partial unfolding of the intermediates, the rate and efficiency of LCI folding pathway increase. Under these conditions, the intermediates are understabilized, being less effective kinetic traps and accumulating to a lesser extent (49,50); native disulfides and free cysteines are probably more solventaccessible, and can easily convert into the scrambled forms by local unfolding events.
By adding enough denaturant to strongly destabilize III-A and III-B species, LCI oxidative folding pathway changes completely. The 3-disulfide intermediates no longer accumulate and LCI folding proceeds through a sequential oxidation of 1-, 2-, 3-, and 4-disulfides forms, which accumulate as scrambled isomers. The disulfide reshuffling of the scrambled intermediates to finally attain the native form becomes the only ratelimiting step of the reaction. Because of the high denaturant concentrations, the relative abundance among the scrambled isomers differs from that observed in the absence of denaturant. Probably, the scrambled isomers displaying more open and relaxed conformations, for instance the beads-form, show a higher prevalence, as observed for other proteins such as PCI, TAP, and IGF-1 (51-53). These scrambled isomers display a higher difficulty to attain the native bond pairing, hence this last stage of LCI folding becomes extremely slow under these conditions. The "simplified" LCI folding pathway observed in the presence of high concentrations of denaturants shows much resemblance to those exhibited by less structured 3-disulfide proteins (i.e. PCI, hirudin) (15,17), suggesting that the differences among their folding processes are caused by the higher extent of regular secondary structure displayed by LCI and not to the different number of disulfide bonds.
Reductive Unfolding Pathway of LCI-Proteins with their native disulfide bonds reduced collectively in an all-or-none mechanism, without detectable partially reduced species, display both a high degree of heterogeneity of folding intermediates and the accumulation of scrambled isomers, as observed for hirudin or PCI (54,55). On the other hand, a sequential reduction of the native disulfide bonds is generally associated with the presence of predominant folding intermediates, as in the case of BPTI or RNase A (55, 56). Reinforcing the above-mentioned theory, transient accumulation of two intermediate species during the reductive unfolding of LCI was detected, which corresponds to the 3-disulfide intermediates that act as kinetic traps in the oxidative folding pathway: the III-A and III-B forms. In LCI, these intermediates accumulate at a lesser extent than in the case of RNase A or BPTI, in agreement with the different characteristics of the 3-disulfide intermediates and folding pathways. In the oxidative folding reaction of BPTI or RNase A, one may expect a preferential protection toward reduction of those native bonds hidden in the protein core. Thus, the less stable and more solvent-accessible disulfide bonds can be preferentially reduced by local unfolding events, with the accumulation of the correspondent intermediate. Global unfolding only occurs after reduction of the covalent bonds hidden in the protein core. In the case of LCI, the disulfide bonds between Cys 18 -Cys 62 and Cys 11 -Cys 34 appear to be slightly more stable and protected than Cys 22 -Cys 58 and Cys 19 -Cys 43 . This allows the detection of intermediates in which the former bonds are still formed, and one of the two other disulfides is also present. It explains why the two 3-disulfide intermediates can still interconvert prior to their complete reduction in the presence of moderate concentrations of reducing agent. However, in LCI, the differences in protection against reduction between disulfides bonds are too small to allow "locking in" intermediate forms before the total reduction of the polypeptide and, on the overall, they are reduced almost in a concerted manner following an all-or-none mechanism.
Conformation and Functionality of LCI Folding Intermediates-LCI three-dimensional structure consists of a fivestranded anti-parallel ␤-sheet and a short ␣-helix ( Fig. 1) (33). The protein is stabilized by four disulfides, all of them located within secondary structure elements ( Fig. 1) (33). The III-A intermediate has two free cysteines, Cys 22 and Cys 58 , which in the native form connect the C-terminal end of ␤2 and the N-terminal end of ␤5. The III-B species lacks the disulfide bridge formed between Cys 19 and Cys 43 , which links the ␤2and the ␣-helix.
The III-B species and the scrambled population are marginally structured forms, while maintaining yet some conformational order and activity. In contrast, the III-A intermediate corresponds to a structured and properly folded species, as assessed by NMR and CD spectroscopy. In addition, it has an RP-HPLC elution time very similar to that of native LCI, indicative of similar hydrophobicity. Besides its inhibitory capability is indistinguishable from that of the native LCI for all tested carboxypeptidases. One question is why has LCI evolved to be a 4-disulfide protein instead of a protein with 3-disulfide bonds, with the same inhibitory efficiency and a less complicated and faster folding pathway? Although proteins perform in a very efficient way their role in vivo, it is now clear that they are not fully optimized. They only fulfill the minimum requirements in terms of stability and folding efficiency that allow them to operate properly in the cell (57)(58)(59). Thus, in the case of LCI, one may assume that a 3-disulfide bonded variant would not be stable enough to perform efficiently its functions in vivo. This assumption makes sense if one takes into account that LCI is a protease inhibitor from leech saliva, evolved to act in blood, a fluid very rich in proteases. Despite its nativeness, the III-A intermediate displays higher fluctuation and lower conformational stability than native LCI, as shown by the lower protection to D/H exchange. By analogy, a native 3-disulfide bonded LCI would be probably more susceptible to proteolytic attacks.
Our results, and the comparison made with others, clearly indicate that the folding pathway of disulfide-containing pro-teins hinge critically on the presence of localized stable structures. The different structural content of the 3-disulfide intermediates characterized in the present work suggests that the accumulation of kinetic intermediates along the disulfide folding reaction relies mainly on their ability to protect their native disulfide bridges from rearrangement in the interior of totally or partially folded protein conformations.