Thiol-disulfide exchange of ribonuclease inhibitor bound to ribonuclease A. Evidence of active inhibitor-bound ribonuclease.

Ribonuclease Inhibitor (RI) has been purified from pig testis. It contains 30 half-cystines whose oxidation affects its ability to bind and inhibit ribonuclease (RNase). By N-terminal sequence analyses testis RI showed to be identical to that from porcine liver, for which a characteristic all-or-none type of SH-oxidation by 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) has been reported (Fominaya, J. M., and Hofsteenge, J.(1992) J. Biol. Chem. 257, 24655-24660). Under comparable reaction conditions, testis RI bound to RNase A did not exhibit this particular type of oxidation; instead, bound RI got intermediate oxidation degrees (up to 14 thiols oxidized per RI moiety) without dissociating from RNase. Moreover, RNase bound to partially oxidized RI was able to express some (15%) of its potential activity (active complex). Only when DTNB treatments accounted for complex dissociation (>14 thiols oxidized per RI moiety) the released RI molecules exhibited the all-or-none oxidation behavior. By both kinetic and circular dichroism analyses, conformational changes have been evidenced for the transition from the inactive to the active form of RI-RNase complex. Relaxation of RI-RNase binding without major alterations in RI structure is proposed as responsible for complex activation. The results are discussed in terms of a model for the reversible regulation of RNase activity mediated by the redox status of RI.

Ribonuclease Inhibitor (RI) has been purified from pig testis. It contains 30 half-cystines whose oxidation affects its ability to bind and inhibit ribonuclease (RNase). By N-terminal sequence analyses testis RI showed to be identical to that from porcine liver, for which a characteristic all-or-none type of SH-oxidation by 5,5-dithiobis(2-nitrobenzoic acid) (DTNB) has been reported (Fominaya, J. M., and Hofsteenge, J. (1992) J. Biol. Chem. 257, 24655-24660). Under comparable reaction conditions, testis RI bound to RNase A did not exhibit this particular type of oxidation; instead, bound RI got intermediate oxidation degrees (up to 14 thiols oxidized per RI moiety) without dissociating from RNase. Moreover, RNase bound to partially oxidized RI was able to express some (15%) of its potential activity (active complex). Only when DTNB treatments accounted for complex dissociation (>14 thiols oxidized per RI moiety) the released RI molecules exhibited the all-or-none oxidation behavior. By both kinetic and circular dichroism analyses, conformational changes have been evidenced for the transition from the inactive to the active form of RI-RNase complex. Relaxation of RI-RNase binding without major alterations in RI structure is proposed as responsible for complex activation. The results are discussed in terms of a model for the reversible regulation of RNase activity mediated by the redox status of RI.
Ribonuclease Inhibitor (RI) 1 is an intriguing protein which is practically ubiquitous in mammalian tissues. Its three-dimensional structure, as well as that of its 1:1 complex with RNase A, have recently been reported (1,2). Although the first piece of evidence about the existence of this inhibitor was gained a long time ago (3), the biological functionality of RI remains to be clarified yet (for reviews on early work on RI see Refs. 4 and 5). Its name suggests that RI is involved in the control of cytoplasmic RNases, thus having a potential role in determining levels of gene expression (6). In fact clear correlations between cellular metabolic states and levels of RI have been reported (7)(8)(9)(10)(11)(12)(13). Nevertheless, objections have been raised about a cytoplasmic functionality for RI (14). The absence of experimental evidence about the in vivo existence of cytoplasmic RI-RNase complexes, together with the known high affinity of RI for enzymes of the noncytoplasmic RNase superfamily (14), leave open the question about the environment where RI plays its role, if intracellular or extracellular. Whichever the answer to this question it might be, or if eventually both were true, a second question arises. This is derived from the tight binding between RI and RNase A, which is the model ligand mostly used to study the interaction properties of RI (15)(16)(17). The binding between these two molecules stands out as one of the tightest reported for protein-protein interactions (dissociation constant in the femtomolar range) (18,19). This adds more interest to the RI molecule whose amino acid sequence is characterized by its leucine-rich repeats (20 -22), a structural motif found for other molecules involved in protein-protein interactions (23). The extremely high affinity between RI and certain RNases raises the question of which fate could have the respective RI-RNase complexes. Aside from taking them as dead complexes, any other functional implication should require to consider that binding affinity may be modulated (24). In this sense, the numerous half-cystines of the RI molecule (around 30 for the different RIs studied) (20 -22, 25, 26) may be involved in its regulation. In fact it is well established that these residues must be as free thiols for RI to exert its inhibitory activity; in addition, RNase dissociates from RI when complex preparations are treated with sulfhydryl reagents, p-hydroxymercuribenzoate being the most employed one for such a purpose (4,5,27). This close relationship between the status of the SH groups and RI activity allows to propose some kind of redox control over the RI-RNase complex (24). In this context, a reasonable working hypothesis is to consider the RI-RNase complex as a heterodimeric enzyme, one of its subunits would be catalytic, the RNase, and the other one would be regulatory, the RI.
Thiol-disulfide exchange, as a general mechanism of enzyme control (28), might be involved in the proposed regulation. Inactivation of porcine RI by exchanging of its thiol groups with the Ellman's disulfide (DTNB) has been reported (29). This investigation, performed on free RI, evidenced an interesting "all-or-none" mechanism of inactivation by which, in the presence of amounts of DTNB that did not account for a complete oxidation of the thiol groups present in a RI preparation, the resulting RI molecules did not show intermediate oxidation degrees. Instead, a fraction of RI molecules resulted with all their half-cystines oxidized, whereas the rest maintained all of them as free thiols.
According to our working hypothesis, we were interested in * This work was supported by Grant C232/91 from the Comunidad de Madrid. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

Materials
Ribonuclease Inhibitor was purified from pig testis as described below. Testes were from pigs intended for slaughter (6 months old or older) and were directly collected at the abattoir (GYPISA, Pozuelo de Alarcón, Madrid) immediately after the animal death. Only testes weighing more than 200 g were selected, immediately frozen in liquid nitrogen, and stored at Ϫ80°C until required. Under these storage conditions no decrease of the inhibitor activity was observed, at least after 6 months.
Bovine pancreatic RNase A (type XII-A), DTNB, and dithiothreitol were purchased from Sigma; cyanogen bromide and SDS were from Serva (Heidelberg, Germany); iodoacetic acid was obtained from Merck (Darmstadt, Germany); and formic acid was from Carlo Erba (Milano, Italy). All other chemicals were at least of reagent grade. The buffers were boiled and degassed by bubbling argon both while cooling down and before use. All of them were 5 mM in dithiothreitol unless otherwise stated.

RI Purification
The rapid procedure described by Blackburn (30) for the purification of RI from human placenta was essentially employed. Testes, once thawed and stripped of skin, were homogenized as described for placenta. After centrifugation of homogenates a high fat content remains in the supernatants, making it necessary to perform the subsequent salt-fractionation twice. For such a purpose the 35%-60% salt saturation precipitate is redissolved in 20 mM Tris-HCl, pH 7.5, containing 2 mM EDTA, and dialyzed against 20 volumes of the same buffer for 6 and 12 h. The resulting solution is salt-fractionated again under the same conditions. The final recovery of RI was 70% or higher, being its final specific activity of 100,000 units/mg, as found for placenta inhibitor (27).
RI eluted from the affinity column was pooled and immediately desalted by gel filtration through Sephadex G-25 M (PD-10 column; Pharmacia LKB, Uppsala, Sweden) equilibrated and eluted with 20 mM Tris-HCl, pH 8.0, containing 2 mM EDTA, 150 mM NaCl, and 15% (v/v) glycerol. The pools so obtained were stored at 4°C in vials hermetically stoppered by Mininert valves (Alltech; Deerfield, IL) provided with septa, through which access with a syringe was possible without exposing the content to the atmosphere. Before storage, a positive pressure of argon (0.5 Kg/cm 2 ) was bubbled into the vials through the septa. Periodically the inert atmosphere was renewed in order to minimize spontaneous oxidation during storage.

Primary Structure Studies
RI preparations intended for N-terminal sequence were desalted as above described, but replacing the equilibration buffer of PD-10 columns by 0.575 M Tris-HCl, pH 8.6, containing 5 mM EDTA, 8 M urea, and 0.2 M 2-mercaptoethanol. The fractions containing RI were incubated for reduction and carboxymethylation with iodoacetic acid as described (31). Excess reagent was removed by gel filtration on PD-10 columns now equilibrated with 50 mM (NH 4 )HCO 3 . The mixtures, once lyophilized, were dissolved in 70% (v/v) formic acid at around 1.5 mg/ml protein concentration. Then 0.7 M CNBr in 70% (v/v) formic acid was added up to a 3,000 molar excess of the reagent over RI. The mixture was briefly flushed with nitrogen and allowed to stand at room temperature for 48 h in the dark. The mixture was then 35-fold diluted with glass-distilled water and lyophilized. The digest was redissolved in electrophoresis buffer and subjected to SDS-PAGE (32) in a Mini-PROTEAN II system (Bio-Rad). The electrophoretic pattern was visualized by reverse staining with imidazole-Zn 2ϩ (33). The transparent bands of interest were excised and incubated in 2% (w/v) citric acid for 10 min in order to mobilize the peptides. Then the excised gel portions were incubated for 15 min in a solution containing 48 mM Tris, 39 mM glycine, 0.375% (w/v) SDS, and 20% (v/v) methanol. Afterwards they were electroeluted in a semidry transblot system (Novablot LKB 2117-250) following the instructions of the manufacturer (0.8 mA/cm 2 for 75 min). Once electroeluted onto polyvinylidene difluoride membranes (Immobilon-P; Millipore) the cyanogen bromide peptides were detected by staining with 0.1% (w/v) Coomassie Blue R-250 in 40% (v/v) methanol containing 1% (v/v) acetic acid for 1 min, followed of destaining with 50% (v/v) methanol. The peptides of interest so immobilized and stained were finally subjected to automatic Edman degradation by using an Applied Biosystems model 477A sequencer, equipped with an on-line phenylthiohydantoin 120-A analyzer. The standard program supplied by the manufacturer was employed (34).

RI-RNase A Complex Formation
When required for subsequent studies, the complex between RI and RNase was formed immediately after elution of RI from the affinity chromatography by putting it into contact with commercial RNase A. Thus, a molar excess of RNase A (1.2 mol of RNase A/mol of RI) was added to each fraction of the affinity chromatography eluate. The required volume of a 10 mg/ml solution of RNase A was calculated for each fraction by taking into account its RI content (A 280 measurements) and the 1:1 stoichiometry of the RI-RNase complex. The mixtures so resulting were subjected to gel filtration on PD-10 columns (Pharmacia) as described for free RI, and the complex was getting formed as the proteins become desalted. Identical conditions to those described for free RI were employed for the storage of the RI-RNase complex.

DTNB Reaction
RI-RNase complex fractions eluting from PD-10 columns equilibrated in the buffer described for free RI, but without dithiothreitol, were pooled in order to obtain complex concentrations as high as 10 -12 M. These pools were poured into vials which were capped with Mininert valves (Alltech; Deerfield, IL). The vials were placed inside a diafiltration chamber lacking a filtration membrane and kept at room temperature. The solvent outlet of the chamber was closed and through the solvent inlet a pressure of 10 kg of argon/cm 2 was supplied. Periodically the solvent outlet was opened in order to allow efficient air displacement. At fixed times the diafiltration chamber was opened, and 1 mM DTNB in 50 mM phosphate buffer, pH 7.0, was injected into the vials through the Mininert septa. In each vial, the injected volume accounted for a DTNB/complex molar ratio of 0.5. In order to monitor the thioldisulfide exchange, prior to DTNB addition aliquots of the vials were withdrawn and subjected to the same DTNB reaction than the rest remaining in the vials. The aliquots were continuously monitored at 412 nm in a Beckman DU-7 spectrophotometer until reaching constant absorbance values. Then, further additions of DTNB were performed to account for additional DTNB/complex ratios of 0.5 and so on. In this way, the accumulated DTNB/complex molar ratios of the reaction mixtures were increased 0.5 at a time. DTNB additions to the vials were repeated until all of the thiol groups in the mixtures were oxidized.

Quantification of Thiol Groups
The number of thiol groups oxidized after each DTNB addition step was evaluated by TNB quantification. For such a purpose an extinction coefficient of 13,600 M Ϫ1 ⅐cm Ϫ1 at 412 nm was used for TNB (35). Additionally, the number of thiol groups remaining after each reaction step was measured in aliquots of the reaction mixtures by the method of Ellman (36), as described by Glazer et al. (37). 1.5% (w/v) SDS was employed as denaturing agent.

Enzymatic Assay
Ribonucleolytic activity, both free and latent (inhibited by RI), was evaluated by using cyclic 2Ј,3Ј-CMP as substrate (30). In all cases the amount of active RNase A in the assay mixture was in the 0.2-0.8 g range.

Chromatographic Analyses
Gel filtration on a Superdex 75 HR 10/30 column (Pharmacia) was employed to evaluate the dissociation degree of complex preparations, whereas ion-exchange on a Spherisorb-TSK DEAE 5PW column (75 ϫ 7.5 mm) (Beckman) was useful when RI-RNase complex had to be separated from released RI. In both cases a high performance liquid chromatography equipment (System Gold from Beckman) was employed.
Gel Filtration-The column was equilibrated and eluted with 50 mM phosphate buffer, pH 7.0, containing 0.15 M NaCl and 10% (v/v) glycerol. Salt was included in order to avoid nonspecific interactions between the acidic RI and the alkaline RNase A, whereas the presence of glycerol was mandatory for quantitative recovery of RNase A. Samples of 100 l were applied onto the column. A flow rate of 0.5 ml/min was employed, and the absorbance of the eluate was continuously monitored at 280 nm. For each evaluation two samples were run. They were prepared by incubating for 10 min at room temperature 90 l of the complex preparation with either 10 l of water or 10 l of 10 mM pHMB in water. The respective elution profiles were integrated at the RNase position and the dissociation degree evaluated by referring the area of the RNase peak of the first sample (without 1 mM pHMB) to that of the second one (with 1 mM pHMB).
Ion-Exchange Chromatography-The employed buffers were: A: 20 mM Tris-HCl, pH 8.0, containing 2 mM EDTA, 50 mM NaCl, and 15% (v/v) glycerol and B: same as A but containing 300 mM NaCl instead of 50 mM. The column was equilibrated in buffer A. After sample loading (500 l) the column was washed for 5 min with buffer A at a flow rate of 1 ml/min. The absorbance was continuously monitored at 280 nm. Complex and free RI were successively eluted with a linear gradient of 0%-50% buffer B in 1 min, followed of isocratic elution with 50% buffer B over 1 min; and a second linear gradient of 50%-100% buffer B in 1 min also followed of isocratic elution with 100% buffer B over 1 min. Finally, initial conditions were restored by a linear gradient of 100 -0% buffer B in 1 min, followed of isocratic elution with 100% buffer A until constant base line. Under these conditions RI-RNase complex and free RI eluate as well resolved peaks (w1 ⁄2 Ϸ 0.35 min) 11.3 and 14.4 min after sample injection, respectively.

CD Measurements and Secondary Structure Analyses
Circular dichroism spectra were registered in the 240 -200-nm wavelength range by using a Jobin Yvon mark III dichrograph fitted with a 250-watt xenon lamp. The scanning speed was of 0.5 nm⅐s Ϫ1 and 0.05-cm optical path cells were employed. Ellipticity values were expressed in units of degree⅐cm 2 ⅐dmol Ϫ1 of residue, considering that the mean residue mass for RI-RNase complex is 107.5 Da, as calculated from its amino acid composition. Prior to the spectroscopic measurements, DTNB-treated samples were chromatographed on Sephadex G-25 M (PD-10 columns, Pharmacia) in order to remove the released TNB. Under the employed DTNB reaction conditions no conformational changes should be expected for RNase A. Therefore, its calculated contribution was subtracted from each CD spectrum in order to simplify the subsequent conformational analyses. Thus, a set of CD curves (one curve for each accumulated DTNB/complex ratio) was generated, mostly reflecting the CD behavior of the RI subunit as increasingly oxidized. This set was appended to a set of 240 -200 nm CD spectra from 18 reference proteins (Table VIII in Ref. 38) in order to deconvolute the whole data set according to the convex constraint analysis (CCA) method (39). For such a purpose three conformational components were allowed, in accordance with the known three-dimensional structure of RI (1, 2), which can be taken as a repetition of three structural motifs. The shapes of the three pure CD curves so resulting were in good agreement with the known CD spectra of ␣-helix, ␤-sheet, and unordered secondary structure (Fig. 1). Goodness of the deconvolution analysis was additionally checked with six proteins of known three-dimensional structure, whose CD spectra were among the reference data set: myoglobin, lactate dehydrogenase, lysozyme, ribonuclease A, ␣-chymotrypsin, and elastase. Correlations between their secondary structure percentages, as totalled from x-ray diffraction results (Table III in Ref. 40), and the respective estimations from the applied CCA method were quantified for each secondary structure through the Pearson product-moment correlation coefficient (41). Very good correlation was found for ␣-helix estimations (r ϭ 0.995); on the other hand, the correlation coefficients for ␤-sheet and unordered form, 0.802 and 0.587, respectively, though poorer, were still significant. For these calculations ␤-turn percentages were not taken into account. The CD contribution of this structural element should be obscured in the pure CD curves of the other conformations and so its conformational weights added in some extent to the CCA-estimated percentages. Most likely this contamination affects the estimated contributions of ␤-sheet and unordered form, so explaining their lower correlation coefficients.

Protein Determination
Concentrations of pure RI, RNase A, and RI-RNase A complex were calculated from absorbance measurements. An extinction coefficient (E 278 nm 0.1% ) of 0.71 was employed for RNase A (42). Through amino acid analysis, performed as described (25), we have evaluated extinction coefficients (E 280 nm 0.1% ) of 0.88 and 0.86 for free RI and RI-RNase A complex, respectively.

Kinetics of the Reactions in Scheme I
The differential equations that govern the production of both TNB molecules and the evolution of the intermediate mixed disulfide are as follows.

Whereas Equation 1 can easily be integrated, integration of Equation 2
requires to previously integrate Equation 3. In both cases, an integral similar to that numbered as 501 in the integral table of "CRC Handbook of Chemistry and Physics" (43) must be solved. The exact integrated equation contains a sum of squared and higher order terms that can be discarded because of their nonsignificant effect on the final results. Thus, the following integrated equations are obtained.
zero subindex refers to concentrations at zero time.
For each DTNB addition, the k 1 and k 2 values could be obtained by nonlinear regression fitting of the experimental data (A 412 versus time) to the following equation.

Statistical Models for DTNB Oxidation of RI-RNase Complex
As a first approach, the same reactivity can be assumed for all the SH groups of the RI-RNase complex. Actually this is not the case, as deduced from the fluctuations detected in the evaluated rate constants (Table I under "Results"); however, it can be demonstrated (data not shown) that, while dissociation is not significant, the slight fluctuations measured for the rate constants are compatible with such an assumption. Accordingly, the probability of a particular SH group to be modified by DTNB will depend on the DTNB/SH molar ratio. Thus, for each global DTNB/complex ratio, r, reached in the reaction mixture, the probability P(m) of finding a complex molecule with m of its 30 thiol groups modified, whichever they can be into the amino acid sequence, will follow a binomial distribution.
where g is the average oxidation degree of the reaction mixture. Obviously, g ϭ 2r/30, since each DTNB molecule reacts with two thiol groups Each curve has been assigned to a secondary conformation by comparison of its shape with the known CD spectra for the different secondary structures (38).
(Scheme I). Let v be the minimum number of thiol groups that must be oxidized in a complex molecule for dissociation to occur. For a complex preparation subjected to DTNB treatment, the percentage of dissociated complex molecules (the dissociation degree) will be given by the summation of the probabilities of finding molecules with a number of modified groups ranging from v to 30. Thus, where P(m) can be calculated from Equation 7 by using the particular DTNB/complex ratio (r) accumulated in the reaction mixture.
The activation percentage would also be given by Equation 8 if only dissociation accounted for the release of ribonucleolytic activity. But this is not the case (see "Results"); the formation of active complex species should additionally be considered. Therefore, in order to predict the activation percentage of a complex preparation subjected to DTNB treatment, Equation 8 should be modified by including the contribution of such active complex molecules to the ribonucleolytic activity. This is taken into account in the following equation, where u is the minimum number of thiol gropus that must be modified in the complex for it to become active, and f is the fractional activity of the RNase in the active complex relative to its activity as free enzyme.

RESULTS
RNase Inhibitor from Pig Testis-We have centered our studies on the RI purified from pig testis. As reported previously (25), testis is a very suitable source of RI, what has been explained by its particularly high biosynthetic activity (spermatogenesis). Hence, from 1 kg of pig testis we routinely obtain 9 -10 mg of RI, which appears as a single band in silver-stained SDS-PAGE ( Fig. 2A). The availability of this biological material, as well as the simplicity of the employed purification scheme, together with its high yields in pure RI (70%), led us to use this molecule from pig testis in order to study the regulatory properties of RI bound to RNase.
When these studies were under way, investigations on thioldisulfide exchange properties of free RI from pig liver were reported (29). Although a high similarity between both porcine RI could be expected, we decided to assess this issue by com-paring the N-terminal sequence of testis RI with that of liver RI. As already found for the latter (20), RI from pig testis was not susceptible to automatic Edman degradation, so suggesting that its N-terminal residue was blocked. The amino acid analysis of testis RI revealed the occurrence of two methionyl residues (data not shown). Therefore, we could use the same strategy employed for the determination of the amino acid sequence of liver RI (20). Thus, CNBr cleavage of testis RI yielded two major peptides of 36 and 14 kDa, which could be separated by SDS-PAGE (Fig. 2B). Two similar peptides were also seen after CNBr cleavage of liver RI (20). In addition, this pattern can be taken as an evidence of the occurrence of a blocked methionine as the N-terminal residue; upon CNBr hydrolysis this methionine would migrate, as blocked homoserine, with the front in SDS-PAGE. The two peptides, once electroeluted from the gel were subjected to automatic Edman degradation, rendering the following N-terminal sequences: CB1, Asn-Leu-Asp-Ile-Cys-Glu-Gln-Leu and CB2, Leu-Thr-Gln-Asn-Lys-His-Leu-Leu-Glu-Leu-Gln-Leu. These two sequences are identical to the respective N-terminal sequences of the fragments CB2 and CB3 of liver RI (20). These results constitute strong evidences about the identity of both porcine inhibitors, although they proceed from different organs. Consequently, the knowledge gained in this work about the redox properties of porcine RI bound to RNase A can be taken as an extension of the thiol-disulfide studies carried out with free porcine RI (29).

Thiol-Disulfide Interchange between RI-RNase A Complex and DTNB-RI-RNase
A complex preparations were subjected to successive steps of reaction with substoichiometric amounts of DTNB (0.5 mol of DTNB/mol of complex), as described under "Experimental Procedures." In this way a collection of "A 412 versus time" curves were obtained; as many as addition steps were required to get the complete modification of the 30 cysteinyl residues of bound RI (the eight cysteine residues of RNase A are forming four disulfide bridges; see Ref. 41). For each reaction step the absorbance at 412 nm reached a final value which accounted for the formation of 2 mol of TNB/mol of added DTNB. This behavior was also seen by Fominaya and Hofsteenge (29) for the reaction of free RI with DTNB. These authors already indicated that such a behavior can only be interpreted as the consequence of a two-step reaction between DTNB and the protein thiol groups (Scheme I). As a consequence each DTNB molecule transforms two protein thiol groups into one disulfide bridge. This point was confirmed by determination under denaturing conditions of the thiol groups remaining in the complex after completion of the successive reactions with substoichiometric amounts of DTNB. The results of these evaluations are plotted in Fig. 3. The number of unmodified thiol groups plus the number of the modified ones (as calculated from the increase in the absorbance at 412 nm of Kinetic Evaluation of SH Reactivity-A careful analysis of the A 412 versus time curves obtained for each DTNB modification step revealed subtle but significant differences among them. This fact suggested that the reactivity of the thiol groups of the complex varies as its oxidation degree increases. These variations can be evaluated by fitting the experimental curves to the integrated rate equation which account for the production of total TNB (Equation 6 under "Experimental Procedures"). The results of these measurements for both rate constants are shown in Table I. As can be seen these values fluctuate as the complex is increasingly modified by the successive DTNB additions. While [DTNB]/[complex] ratios were lower than 6.0, these fluctuations were small, 1.5 and 4.0 being the ratios for which both rate constants reached minimum values. At higher [DTNB]/[complex] ratios, major changes could be detected. Since such high ratios accounted for considerable complex dissociation, these major changes should be interpreted as due to the unfolding of released RI after massive oxidation (see next paragraphs).
Leaving aside the effects of complex dissociation, one would predict that at [DTNB]/[complex] ratios for which dissociation could be neglected, thiol groups of the complex should result sorted out in their reaction with DTNB; that is, the higher reactivity of a cysteinyl residue the quicker its modification and vice versa. Therefore, a continuous decrease of the fitted k 1 and k 2 values could be expected. Alternatively, if the SH reactivity was not affected by the complex conformation, nonsignificant variations in the rate constants should be expected all trough the modification. However, the observed behavior in which SH reactivity fluctuations are detected should be interpreted as a consequence of conformational changes of the complex induced by thiol-disulfide exchange.
Circular Dichroism Studies of RI-RNase Complex as Increasingly Oxidized by DTNB-In order to evaluate the conformational changes of the complex predicted by the observed transitions in the fitted rate constants, aliquots of the DTNB complex reaction mixtures were withdrawn after reaction completion for each step of DTNB addition. Once dialyzed, their CD spectra were registered in the 240 -200 nm range and analyzed as described under "Experimental Procedures." Thus, conformational weights of ␣-helix, ␤-sheet, and unordered form were obtained for each discrete step throughout the gradual DTNB modification. These results are plotted in Fig. 4. In order to facilitate their comparison with the evolution of the fitted rate constants, the values in Table I are also plotted in Fig. 4. The first transition observed for the rate constants (accumulated DTNB/complex ratio ϭ 1.5-2.5) seems to correlate with a conformational change in which the initial decrease in ␣-helix occurs at the expense of an increase in unordered structure, whereas the subsequent increase in ␣-helix is parallel to a decrease in both ␤-sheet and unordered structure. On the other hand, the second transition kinetically detected (accumulated DTNB/complex ϭ 4.0 -5.0) is correlated with a loss of ␣-helix at the same time as unordered structure increases.
The three-dimensional structure of RI molecule, both free and bound to RNase, is basically formed by the repetition of three structural elements (1, 2): 16 ␣-helices (12 residues long in average), 17 ␤-strands (3 residues long in average) forming a curved parallel ␤-sheet, and 32 loops (containing between 4 and 9 amino acids) connecting the individual ␣ and ␤ segments; in these loops several types of ␤-turns are present. From these figures it can be calculated that ␣-helices contain 42% of all RI residues, whereas only 11% are in the parallel ␤-sheet. Although the percentages of ␣-helix estimated from the CD measurements are in good agreement with the x-ray results, the same cannot be said for ␤-sheet, which is overestimated probably due to some contribution from bends (see "Experimental Procedures").
Effects of Oxidation with DTNB on the Ribonucleolytic Activity and Association Degree of RI-RNase Complex-The dissociation degree of RI-RNase complex as increasingly oxidized by  b Goodness of fits was evaluated by the root-mean-square (rms) deviations between the experimental "A 412 versus time" curves and the fitted ones, as can be seen these mean deviations were less than 0.003 absorbance units in all cases. DTNB was evaluated by gel filtration on a Superdex 75 HR column. The results of these evaluations together with the percentages of released RNase activity are shown in Fig. 5. As can be seen, both activation and dissociation percentages do not evolve in a parallel fashion. In other words, RNase dissociation does not give an entire account of RNase activation. The obtained results point to the occurrence of partially oxidized RI that remains bound to RNase without completely abolishing its activity. We have called this complex showing ribonucleolytic activity, active complex. The minimum number of thiol groups that have to be modified for the complex to become active should then be less than the minimum number required for it to dissociate. In order to investigate this, we have modeled the oxidation of RI bound to RNase.
Our first model assumed a random oxidation mechanism, in which all the thiol groups in the complex have the same reactivity. According to this model predicted activation and dissociation curves can be calculated by Equations 8 and 9. Dotted and dashed lines in Fig. 5 show, respectively, these predicted curves when f ϭ 0.15, u ϭ 5, and v ϭ 15. Although they fit well, the experimental data up to a global DTNB/complex ratio of 5.0, a clear discrepancy is observed beyond this value. Thus, the experimental data grow slower than the predicted ones as the global ratio increases. It should be noticed that such a discrepancy becomes patent when the dissociation percentage reaches a significant value (Ͼ5%). This fact suggests that random modification may not be a valid model for DTNB oxidation of released RI.
The all-or-none type of reaction reported for the DTNB oxidation of free porcine liver RI (29) suggests that in the reaction mixtures the first RI molecules partially oxidized become more susceptible to subsequent DTNB oxidation. Thus, partially oxidized RI acts as a DTNB monopolizer until all of its thiol groups result modified. This particular behavior of free RI may explain the discrepancies of our model. Free RI molecules with some partial oxidation degree will appear in our reaction mixtures as a result of accumulated DTNB oxidation. They would behave as DTNB monopolizers in a subsequent DTNB addition, so lowering the amount of DTNB available to increase the modification degree of the remaining complex molecules. As a consequence, activation and dissociation of complex preparations will be slowed down once they reach accumulated DTNB/ complex ratios, which account for a significant dissociation percentage (e.g. r ϭ 5.0).
It is feasible to calculate theoretical activation and dissociation curves in which this monopolizer effect of the released RI is considered. Thus, for each DTNB/complex ratio the fraction of molecules with v or more of their thiol groups modified (free RI fraction) is calculated by using Equation 8. To determine the activating and dissociating effect of a subsequent DTNB addition, top priority as DTNB consumer is given to this population. For such a purpose, the amount of DTNB that this population consumes to fully modify its remaining thiol groups is subtracted from the DTNB added, so yielding the DTNB which is really available to increase the oxidation degree of the complex molecules. The activation and dissociation percentages can finally be calculated through Equations 8 and 9 by using the "effective" DTNB so determined.
In Fig. 5 the predicted activation and dissociation curves so calculated are also plotted. The best results were obtained when f, u, and v took the previous values of 0.15, 5, and 15, respectively. The accuracy of the fit supports the proposed model of random DTNB oxidation for RI-RNase complex, in conjunction with the all-or-none type of reaction for released FIG. 4

. Gradual modification with DTNB: effects on secondary structure (A) and SH reactivity (B) of RI while bound to RNase.
The ordinate scales of A refer to conformational percentages for each of the three pure components (Fig. 1) to which the applied CCA method deconvoluted the set of CD spectra calculated for RI (see "Experimental Procedures"). Deconvolution analyses were restricted to reaction mixtures in which the accumulated [DTNB]/[complex] ratio did not account for significant dissociation degrees. Conformational analyses of mixtures with higher oxidation degrees were hindered by the important CD contributions of disulfide bridges from the released and therefore fully oxidized RI molecules. In B the results of the kinetic analyses summarized in Table I are plotted in this restricted range of oxidation degrees.

FIG. 5. Gradual modification with DTNB of RI-RNase A complex: effects on RNase inhibition (q) and RNase-binding (E).
Experimental data are, respectively, expressed as percentages of the ribonucleolytic activity or the RNase dissociation found for preparation aliquots subjected to the same DTNB modification and then treated with 1 mM pHMB; ----and ⅐ ⅐ ⅐ ⅐ represent theoretical activation and dissociation curves, respectively, as calculated by assuming a random mechanism of SH modification, with u ϭ 5, v ϭ 15, and f ϭ 0.15 (see text); OO and ----correspond to the respective theoretical curves with the same set of parameters but taking into account a random mechanism of SH modification only for the RI fraction that remains bound to RNase and assuming an all-or-none mechanism of oxidation for the RI fraction that results dissociated after each step of DTNB modification (see text).
RI. Additional support would require experimental evidence about the presence in the reaction mixtures of complex species showing variable oxidation degrees, whereas released RI, if present, should have all its thiol groups oxidized.
Chromatographic Analysis of Complex Partially Oxidized by DTNB-The RI-RNase complex can be resolved from released RI by anionic exchange. A complex preparation subjected to an accumulated DTNB/complex ratio of 7.5 (mean oxidation degree of 0.5) was chromatographed on a TSK-DEAE 5PW column. Fig. 6 shows the elution profile. For comparison purposes the elution profiles of native RI and RNase-RI complex are also shown. The oxidized complex elutes as an asymmetrical and broad peak. This behavior is consistent with a heterogeneous population of molecules in terms of their oxidation degrees. Fractionation of the eluate, and determination of both the thiol and the protein content of each fraction, allowed us to calculate the mean number of free thiols per molecule along the elution profile. The results of these determinations are overlaid in Fig.  6 as bar plots. RI bound to RNase exhibits different degrees of oxidation, with a maximum of 14 SH groups oxidized per molecule (thus leaving 16 free thiols). These facts constitute not only an experimental evidence of the proposed model of oxidation at random, but also corroborate that the maximum number of SH groups that can be oxidized in the complex without dissociation is 14. The oxidation degree along the RI elution peak is constant and, within margins of experimental error, compatible with a complete oxidation of released RI. DISCUSSION The role of RI in the regulation of intracellular RNases remains to be proved. In fact, it has been questioned whether RI-RNase complexes have any implication in the catabolism of RNA, or, on the contrary, if their in vitro detection is only an artifact due to organelle disruption during tissue homogeniza-tion (14). Further investigation of this will be required to accurately assess the cytoplasmic location of RI-RNase complexes. On the other hand, an extracellular role of RI in the regulation of noncytoplasmic RNases has been proposed (14), which raises questions about how RI is transported out of the cytoplasm.
Any consideration about a potential regulatory role of RI would benefit from knowing the mechanism, if any, through which the inhibitory activity of RI can be modified. As mentioned previously in the introductory statement, RI is a good candidate to be controlled in vivo by thiol-disulfide exchange reactions, in response to changes in cellular redox status. Whether or not such a mechanism of regulation can be viable depends on both the susceptibility of RI to sulfhydryl oxidation and the susceptibility of oxidized RI to reduction (45). Studies of the first issue, performed on free porcine RI (29), indicated an extremely high susceptibility of this molecule to become totally oxidized (15 disulfide bridges resulting from its 30 thiol groups) and inactivated through an all-or-none type of reaction. The resulting RI is irreversibly denatured. Hence, in our hands, no reduction treatment was able to restore any of the inhibitory activity lost by reaction of free RI with DTNB, although the employed treatments were effective in the reduction of the 15 disulfide bridges (data not shown). At a first glance, these results seem to be discouraging as to the consideration of some type of reversible in vivo control of RI mediated by the intra-or extracellular thiol-disulfide redox status. But if this control really occurred, one should take into account the effect of thiol-disulfide exchange, not on free RI, but on the RI-RNase complex, which supposedly would be the species candidate for the in vivo control of the ribonucleolytic activity.
The results presented herein demonstrate that RI bound to RNase is oxidized by DTNB in a different way from that found for free RI. Thus, partially oxidized RI can be obtained when preparations of the RI-RNase complex are treated with substoichiometric amounts of DTNB. This partially oxidized RI (with up to 14 thiol groups oxidized) remains bound to RNase, only inhibiting a fraction of its activity (85%). In this complex showing ribonucleolytic activity (active complex), the RI has an altered conformation, as deduced from circular dichroism measurements (Fig. 4). The most remarkable fact of the conformational change associated with the transition from inactive complex to active complex is an increase in the percentage of ␣-helix at the expense of a decrease in that of ␤-sheet and bends. These results are compatible with the recently reported data about the three-dimensional structure of free RI (1) and RI-RNase complex (2). These studies have shown that the binding of RNase A takes place in an extensive RI area, which is mainly formed by the parallel ␤-sheet and loops connecting ␤-strands and ␣-helices. Therefore, the loss of ␤-sheet/␤-turns detected at the activation of the complex could account for some relaxation of the binding between RI and RNase, which would explain how RNase can exert some of its potential ribonucleolytic activity without dissociating.
Our preliminary results on the reversibility of the effects of DTNB oxidation (data not shown) allow us to anticipate that some of the ribonucleolytic activity released from the complex, seemingly that from the active complex, can be reinhibited by reduction treatments. More detailed studies on this particular, currently under way, are required, but for the moment a suggestive hypothesis arises about the possibility of considering the active complex as one of the forms between which the RI-RNase complex can be reversibly switched by the redox status. Fig. 7 summarizes a model based on the results obtained that account for this hypothesis.
In the scheme shown in Fig. 7, the formation of disulfide bridges is considered as occurring between cysteinyl residues which are close to each other in the RI molecule. This consideration has been derived from the results of the kinetic analyses carried out for the reaction between RI-RNase complex and DTNB (Table I). Two values of rate constants were evaluated for each addition of substoichiometric amounts of DTNB: the second order constant, k 1 , for the formation of a mixed disulfide between RI and DTNB and the first order constant, k 2 , for the formation of an intramolecular disulfide. Both rate constants fluctuate only slightly, provided that the complex does not dissociate. Thus, the second order constant fluctuates around 40 M Ϫ1 ⅐min Ϫ1 and the first order constant around 0.4 min Ϫ1 . One would expect that DTNB oxidation would yield mixed disulfides only if their subsequent transformation to intramolecular disulfides was very slow. The intramolecular reaction will be the rate-determining step if k 2 Ͻ k 1 ⅐[DTNB]. Therefore, mixed disulfides would accumulate if [DTNB] Ͼ 10 mM (0.4 min Ϫ1 /40 M Ϫ1 ⅐min Ϫ1 ). This is not the case in the present study, since in order to assess substoichiometric levels of DTNB, the concentrations of this reagent were always in the micromolar range. In such conditions, the formation of intramolecular disulfides is around 3 orders of magnitude faster than that of mixed disulfides, so explaining that only intramolecular disulfides are measurably formed. In the study carried out on free RI (29), the employed DTNB concentrations were also in the micromolar range, which would explain the disulfide formation also found.
This formation of intramolecular disulfide bridges allows us to conclude that in the mixed disulfide intermediates the adjacent thiol groups behave as if their "effective concentrations" were higher than 10 mM. This should be the consequence of the high number of thiol groups in the RI molecule, which occurs mostly at constant positions in the internal repeats of RI (20 -22). The susceptibility of RI to form internal disulfide bridges could be considered as an interesting property for its regulation by the cellular redox status. As stated previously (46), under the reducing intracellular conditions, the oxidation of an intracellular protein should be faster than its reduction if the oxidized form plays some role in vivo. Thus, the formation of intramolecular disulfide bridges in RI can act as a driving mechanism that increases the rate of oxidation in comparison with that of reduction, so allowing both active and inactive forms of the RI-RNase complex to coexist at equilibrium.
The experimental evidence obtained about the existence of an "active RI-RNase complex" allows us to maintain that RI, binding RNases, can have a role in reversibly switching them between active and inactive forms. Certainly, the active form may be considered as a poor enzyme, since it only expresses 15% of its potential activity; but, does the cell need the high ribonucleolytic activity that its RNases are able to exert? On the other hand, if the RI-RNase complex really exists in vivo, would the cell ever reach the strong oxidizing conditions required for dissociation? These questions will probably be answered when redox conditions similar to those found in vivo are employed in studying thiol-disulfide exchange of the RI-RNase complex. With this in mind, we are currently investigating the oxidation of the RI-RNase complex by biological disulfides (e.g. GSSG), as well as their thiol counterparts (e.g. GSH) for the reverse reaction. FIG. 7. Thiol-disulfide exchange of RI-RNase complex. As a result of oxidizing treatments, pairs of thiol groups of RI bound to RNase are progressively transformed into intramolecular disulfide bridges. The RI subunit has 30 thiol groups that can eventually be oxidized into 15 disulfide bridges; for the sake of clarity, not all of them have been represented. While the number of formed disulfides bridges is kept below eight, RI can remain bound to the RNase subunit. If this number is in the range of three to seven, RI is still able to mostly inhibit RNase; however, a small fraction (15%) of the potential RNase activity is expressed. This fractional activity accounts for the proposed denomination of this partially oxidized complex (ACTIVE COMPLEX). The release of ribonucleolytic activity without complex dissociation points to some binding relaxation that can result from the conformational change detected by CD studies. Such a conformational change corresponds to a decrease in ␤-sheet/bend content of RI. Both secondary elements are involved in the interaction of RI with RNase A (2); consequently the transition from the inactive to the active complex is schematized by a diminution of the contact areas between both subunits. Binding to RNase preserves the structure of RI to such an extent that it does not follow the all-or-none mechanism of oxidation found for free RI. Even more, on the basis of some preliminary results, we postulate that this structure preservation could allow reversible transition between the active and the inactive forms of the complex, depending on the redox conditions. If oxidizing conditions are further increased, accounting for the formation of eight or more disulfide bridges, binding relaxation turns into complex dissociation. Then free RI will exhibit its characteristic mechanism of oxidation by which it becomes completely oxidized and irreversibly denatured.