Introduction of a Disulfide Bond into Ricin A Chain Decreases the Cytotoxicity of the Ricin Holotoxin*

Wild type ricin A chain (RTA) contains two cysteine residues (Cys17' and C Y S ~ ~ ~ ) . CysZ5' forms the interchain disulfide bond of ricin holotoxin with Cys4 of ricin B chain (RTB). We have used site-directed mutagenesis of RTA cDNA to convert Cys171 to Ser and to introduce a disulfide bond into RTA by converting SeP" and MetzBB to Cys residues. Mutant RTA was expressed in Escherichia coli and directed to the oxidizing environment of the periplasmic space where the Cys21s-Cys26s disulfide bond was formed. The disulfide-containing RTA mutant had an in vitro catalytic activity similar to that of an identical form of recombinant RTA that lacked the S215C and M255C mutations. In the presence of glutathione and protein disulfide isomerase, this RTAvariant reassociated with RTB to form ricin holotoxin. Incuba- tion of this holotoxin with increasing concentrations of dithiothreitol showed that the interchain disulfide bond joining RTA and RTB was more readily reduced than the intrachain disulfide bond in RTA. Ricin in which the RTAmoiety contained the disulfide bond was 15-18-fold less cytotoxic to HeLa or Vero


Introduction of a Disulfide Bond into Ricin
Wild type ricin A chain (RTA) contains two cysteine residues (Cys17' and C Y S~~~) .
CysZ5' forms the interchain disulfide bond of ricin holotoxin with Cys4 of ricin B chain (RTB). We have used site-directed mutagenesis of RTA cDNA to convert Cys171 to Ser and to introduce a disulfide bond into RTA by converting SeP" and MetzBB to Cys residues. Mutant RTA was expressed in Escherichia coli and directed to the oxidizing environment of the periplasmic space where the Cys21s-Cys26s disulfide bond was formed. The disulfide-containing RTA mutant had an in vitro catalytic activity similar to that of an identical form of recombinant RTA that lacked the S215C and M255C mutations. In the presence of glutathione and protein disulfide isomerase, this RTAvariant reassociated with RTB to form ricin holotoxin. Incubation of this holotoxin with increasing concentrations of dithiothreitol showed that the interchain disulfide bond joining RTA and RTB was more readily reduced than the intrachain disulfide bond in RTA. Ricin in which the RTAmoiety contained the disulfide bond was 15-18-fold less cytotoxic to HeLa or Vero cells than ricin in which the RTA did not contain the stabilizing disulfide crosslink. Since these ricin molecules had identical RTB cell binding and RTA catalytic activities, we suggest that the observed reduction in cytotoxicity caused by the introduced disulfide bond resulted from a constraint on the unfolding of RTA, indicating that such unfolding is necessary for the membrane translocation of RTA during its entry into the cytosol.
Ricin is a well characterized member of a group of plant and bacterial toxins capable of entering and killing mammalian cells (1). The toxic components of these proteins are enzymes that catalyze the irreversible inhibition of cellular protein synthesis ( 2 ) . The stepwise cytotoxic mechanism involves binding to the mammalian cell surface, entry by endocytosis (normally via coated pits), and inactivation of a cellular component essential for protein synthesis. Ricin and related toxins catalyze the specific depurination of 28 S ribosomal RNA (3,4), whereas bacterial toxins such as diphtheria toxin or Pseudomonas exotoxin A catalyze the ADP ribosylation of elongation factor-2 (5).
Since the target substrates are located in the cytosol and the toxin molecules are endocytosed, the catalytically active toxin polypeptide or a fragment derived from it must traverse the membrane of an intracellular compartment in order to act. *This work was supported by Grant GWG00877 from the United Kingdom Science and Engineering Research Council Biotechnology Directorate. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 To whom correspondence should be addressed.
In ricin, the toxic polypeptide (the A chain, or RTA)' is joined to a galactose-binding polypeptide (the B chain, or RTB) by a disulfide bond. RTB is responsible for binding ricin to the cell surface and is also believed to facilitate endocytic transport to the cellular compartment most conducive for the subsequent membrane translocation step (6). Since there is no evidence for the translocation of RTB along with RTA (indeed, RTB can be successfully replaced by other cell binding proteins such as antibodies ( 2 ) ) , it is generally assumed that ricin becomes reductively cleaved within the endomembrane system prior to membrane traversal. Neither the site nor the mechanism of membrane translocation have yet been identified. We assume, however, that the translocation step requires that RTA becomes at least partially unfolded. If this is the case then any constraint on unfolding might be expected to reduce the apparent cytotoxicity of ricin if the amount of RTA in the translocationally competent compartment is limiting. To test this we have engineered an intrachain disulfide bond into RTA. In the present paper we show that ricin containing mutant recombinant RTA with an introduced disulfide bond is less cytotoxic than ricin containing an identical recombinant RTA but lacking this stabilizing disulfide cross-link.

MATERIALS AND METHODS
Creation ofMutunts-The expression vector pKK223.3 was digested with EcoRI and HindIII, and the large (4555-base pair) fragment was isolated. An irrelevant DNA fragment was inserted to maintain the restriction sites, the plasmid was digested with BumHI and SaZI, and 5' overhangs were removed and religated to produce plasmid pKH50, which no longer possessed tetracycline resistance or BamHI and Sal1 sites. Two complementary synthetic oligonucleotides with 5' overhangs, encoding the Escherichia coZi ompF signal sequence and EcoRI, XhoI, BumHI, and HindIII restriction sites, were annealed, end-tilled, and cloned into p H 5 0 to form pKI-IlOO. An RTA cDNA clone, containing an XhoI site at the -15-position of the preproricin N-terminal extension and a stop codon directly after the RTA coding sequence (7) was cloned into pKHl00 via the XhoI and BamHI sites to create pFAC.
Mutagenesis of the RTA cDNA was performed by standard techniques using an oligonucleotide-directed in vitro mutagenesis kit (U. S. Biochemical Carp.) as instructed by the manufacturer. Mutations were verified by dideoxy sequencing. Mutagenic oligonucleotides were synthesized on an Applied Biosystems model 380B DNA synthesizer and were designed such that the mismatch(es) to be introduced was flanked on either side by at least seven bases complementary to the wild type sequence.
Expression and Purification of Recombinant Proteins-E. coli TG2 cells (8) transformed with the appropriate vector were inoculated in defined M9 medium to give an A,,, of 0.1. The culture was grown with shaking at 37 "C until the A,,, reached 0.7-0.8. Expression was induced by adding isopropyl-1-thio-P-D-galactopyanoside to a final concentration of 1 mM, and the culture was shaken for 3 h at 30 "C. After har-DTA and DTB, diphtheria toxin A and B fragment respectively; TBS,

Leu Glu Asp Asn Asn
The amino acids encoded by the spacer between the ompF signal and ricin A chain are shown.
vesting, cells were resuspended in 580 mM sucrose, 300 m~ Tris, pH 8.0, 1 mM EDTA, 0.5 mM MgCI,, allowed to stand on ice for 10 min, and then repelleted by centrifuging at 7500 rpm (6200 x g) and 4 "C for 5 min. Cells were resuspended in 1 m~ Tris, pH 7.5, and after a further 10 min on ice were centrifuged at 19,000 rpm (28,000 x g) and 4 "C for 15 min.
The supernatant, which contained the periplasmic fraction, was dialyzed three times against 5 m~ sodium phosphate, pH 6.5. Recombinant protein from a 1-liter culture was purified by ion-exchange chromatography on a 1.6 x 40-cm CM-Sepharose column in 5 mM sodium phosphate, pH 6.5, and bound proteins were eluted with a 0-300 mM NaCl gradient. Fractions containing recombinant product, determined by SDS-polyacrylamide gel electrophoresis followed by silver staining or Western blots probed with sheep anti-RTA antibodies, were pooled, and protein was precipitated by adding ammonium sulfate to 70% (w/v) saturation. After centrifugation the precipitated proteins were redissolved in 20 mM Tris, pH 6.8,60 mM NaCl, dialyzed, and stored a t -70 "C after adding glycerol to a final concentration of 15% (v/v).
Detection of Disulfide Bonds-Disulfide bonds were detected using a DIG protein detection kit (Boehringer Mannheim). Protein samples were subjected to electrophoresis on a nonreducing SDS-polyacrylamide gel and then transferred onto a Hybond-C nitrocellulose filter by Western blotting. Free sulfhydryl groups were blocked by adding 20 ml of 10 mM N-ethylmaleimide in 50 mM potassium phosphate buffer, pH 7.0, for 30 min. The filter was washed three times, each time for 10 min, with 50 ml of 50 mM potassium phosphate, pH 7.0, before disulfide bonds in the protein were reduced with 2% (v/v) 2-mercaptoethanol in 50 mM potassium phosphate, pH 7.0, for 30 min. After washing the filter as before, it was incubated for 60 min with 5 pl of digoxigenin-3-0-succinyl-[2-(N-maleimido)]ethylamide in 10 ml of 50 mM potassium phosphate, pH 7.0, containing 0.01% Nonidet P-40. The filter was washed twice, each time for 5 min, in Tris-buffered saline (TBS; 50 mM Tris, pH 7.5, 150 mM NaC1) containing 0.1% Tween 20 and once in TBS without the Tween 20. The filter was blocked for at least 30 min by incubation in protein-blocking solution, washed three times in TBS, and then incubated for 1 h with 10 pl of sheep anti-digoxigenin-alkaline phosphatase conjugate in 10 ml of TBS. After washing in TBS, the filter was developed using 37.5 pl of 5-bromo-4-chloro-3-indolyl phosphate and 50 pl of nitro blue tetrazolium in 10 ml of 100 mM Tris, pH 9.5, 100 mM NaCl, 50 m~ MgC1, to stain the antibody conjugates. I n Vitro Protein Synthesis Inhibition Assays-Nuclease-treated rabbit reticulocyte lysate (Promega), supplemented with 25 m~ dithiothreitol, was incubated with various toxin concentrations (in triplicate) for 15 min at 30 "C. Translation was initiated by adding Brome mosaic virus mRNA and [35S]methionine. Aliquots were removed after 5, 10, and 15 min, and 35S-labeled protein was determined by scintillation counting. 28 S Ribosomal RNA Depurination Assay-The activity of recombinant proteins was determined by assessing their ability to depurinate the 28 S rRNA of ribosomes purified from non-nuclease-treated rabbit reticulocyte lysates (Promega). Reaction mixtures contained 30 pg of ribosomes in 25 mM Tris, pH 7.6, 25 mM KC1, 5 m~ MgCl,. After equilibration a t 30 "C for 5 min, wild type or mutant RTA was added, and the depurination reactions were carried out at 30 "C for 5 min in a total volume of 30 pl. RNA was isolated from the ribosomes and the degree of depurination assessed using the aniline assay as described previously (9).
Reassociation of RTA and RTB-Known amounts of purified rRTA, FACO-RTA, or FAC4-RTA were incubated at room temperature with a 2-fold excess of purified RTB (Inland Laboratories, Austin, "X). Where appropriate, the reassociation mixtures were supplemented with 10 m~ glutathione (in a reduced-to-oxidized ratio of 2:1, mimicking that found in the endoplasmic reticulum lumen (lo)), 10 mM glutathione plus 1.4 p~ protein disulfide isomerase, or ?A volume of reticuloplasm. Reticuloplasm, the soluble components of the endoplasmic reticulum lumen, was purified from MOPC-315 cells as described previously (11). In order to separate holotoxin from free RTA or RTA dimerdaggregates, reassociation mixtures were passed down a small column containing 2 ml of SeLectin-2 beads (lactose immobilized on acrylamide) (Pierce). The column was washed with PBS before bound holotoxin was eluted with PBS containing 50 mM galactose. Purified holotoxin was quantified by densitometry of silver-stained SDS-polyacrylamide gels, on which its RTA component was compared to standards of known amounts of purified rRTA, using a Molecular Dynamics computing densitometer.
Cytotoxicity Assays-Vero cells or HeLa cells grown i n Dulbecco's modified Eagle's medium containing 5% fetal calf serum, 2 mM glutamine, 50 unitdm1 penicillin, and 50 pg/ml streptomycin were plated out i n 96-well microtiter plates a t a density of -1.5 x lo4 celldwell. After allowing time for cells to adhere (at least 4 h) and washing with PBS, various toxin concentrations in 50 pl of media were added to each well, and the cells were incubated at 37 "C for various times or overnight. The cells were then washed with PBS, and protein synthesis was measured after adding 1 pCi of [35Slmethionine in 50 1.11 of PBS to each well and incubating at 37 "C for 2 h. Labeled proteins were precipitated by three washes with 5% (w/v) trichloroacetic acid and after three further washes in PBS were released with 0.2 M NaOH, and the incorporated radioactivity was determined by scintillation counting.
Other Methods-DNA manipulations, SDS-polyacrylamide gel electrophoresis, and Western blotting were carried out using standard published procedures (12, 13).

RESULTS
The RTA construct used for mutagenesis and subsequent expression is shown in Fig. 1. The RTA coding sequence was preceded by the E. coli ompF signal sequence. In constructing the fusion (utilizing the XhoI site at -15-position in preproricin (7)), the RTA sequence codes for 5 additional amino acid residues at the N terminus of RTA that are derived from the preproricin leader sequence. A stop codon was introduced immediately after the RTA coding sequence. Expression was directed by the isopropyl-1-thio-/3-D-galactopyranoside-inducible tuc promoter. The plasmid was designated pFAC for ompE-RTAgsteine mutant.
Wild type RTA contains 2 cysteine residues at positions 171 and 259. To avoid possible thioVdisulfide interchange reactions with introduced disulfides, cysteine 171 was converted to serine by standard site-directed mutagenesis of the encoding DNA ((3171s mutation to produce the construct pFAC0). Cysteine 259 was not changed since this residue is required for formation of the interchain disulfide bond linking RTA and RTB in the ricin holotoxin. pFAC0 encodes a variant form of RTA that differs from native RTA in containing five extra N-terminal amino acids and the C171S substitution. RTA encoded by pFAC0 (FACO-RTA) therefore provided the control toxin for comparison with mutant derivatives carrying additional pairs of cysteine residues, since the pFAC0 construct was used as the template for the introduction of these additional cysteine codons. The pairs of FACO-RTA residues that were converted to cysteines are shown in Table I. Plasmid pFAC and the mutant plasmids listed in Table I were expressed in E. coli TG2 cells.
Recombinant products were directed to the periplasmic space by the E. coli ompF signal sequence to facilitate disulfide bond formation (14). RTA bands were visualized on Western blots of periplasmic preparations by adding sheep anti-RTA antibodies followed by a donkey anti-sheep IgG-alkaline phosphatase conjugate and by color development (data not shown). The Western blots showed that pFAC, pFAC0, and pFAC4 expressed well. Recombinant product was not seen for E. coli transformed with pFACl or pFAC3, whereas very small amounts of cytosolic pFAC2 product were occasionally observed. FACO-RTA and FACCRTA were purified to homogeneity from E. coli periplasmic extracts by CM-Sepharose column chromatography as described elsewhere (15). Fig. 2 shows a silver-stained SDS-polyacrylamide gel of the purified products. FACO-RTA had the same electrophoretic mobility under either reducing or nonreducing conditions (Fig. 2). Significantly FAC4-RTA had an increased electrophoretic mobility under nonreducing conditions (Fig. 21, which we assumed to be caused by the presence of the engineered disulfide bond. The presence of a disulfide bond in FAC4-RTA was confirmed by running the protein, together with appropriate controls, on a nonreducing SDS-polyacrylamide gel. After electrophoresis the proteins were transferred to nitrocellulose, free thiol groups were blocked with N-ethylmaleimide, and disulfide bonds were reduced with P-mercaptoethanol. Free thiols resulting from this reduction were detected by adding digoxigenin reagent followed by anti-digoxigenin antibody conjugated to alkaline phosphatase. As shown in Fig. 3, FAC4-RTA gave a positive signal (lane 6), as did the disulfide-containing control proteins fetuin, RTB, and a-sarcin (lanes 2-4, respectively). No obvious signals were observed for subtilisin or recombinant wild type RTA (lanes 1 and 51, which do not contain disulfide bonds. The sensitivity of FACCRTA when incubated for 30 min on ice with a range of trypsin concentrations (0-100 pg/ml) was compared with that of FACO-RTA and wild type RTA. No difference in proteinase sensitivity was observed (data not shown), indicating that the introduced disulfide bond did not significantly affect the folding and conformation of the protein.
Recombinant FAC-RTA, FACO-RTA, and FAC4-RTA were all catalytically active. RTA and related toxins act by depurinating 28 S rRNA at a specific site close to the 3' end of the molecule. Depurination renders isolated rRNA susceptible to amine-cata- iyzed hydrolysis of the phosphodiester bonds on either side of the modification site. This cleavage generates a small RNA fragment of -390 ribonucleotides from rabbit reticulocyte 28 S rRNA. Reticulocyte ribosomes incubated with crude periplasmic extracts containing FAC-RTA, FACO-RTA, and FACCRTA released the same characteristic RNA fragment upon aniline treatment of isolated rRNA (Fig. 4, lanes 2, 4, and 8, respectively) that was also released by wild type RTA (Fig. 4, lane IO).
The in vitro activity of the mutants was quantified by measuring the inhibitory effects of various toxin concentrations on protein synthesis by nuclease-treated rabbit reticulocyte lysates programmed with Brome mosaic virus mRNA. In an attempt to mimic the reducing environment of the cytosol where RTA acts in cytotoxicity assays, translation mixtures were supplemented by adding 25 mM dithiothreitol. This reducing environment did not affect the activity of wild type RTA in comparison to its activity in unsupplemented lysates (data not shown). FACO-RTA and FACCRTA gave similar inhibition curves (Fig. 51, indicating that they had similar catalytic activities. Clearly the introduction of the disulfide bond into FACCRTA did not affect catalytic activity in a reducing environment. Both mutant proteins were slightly less active than wild type rRTA. The apparent ID,, values (concentration causing 50% inhibition of protein synthesis) based on the data shown in Fig. 5 were 0.55 nM for FAC4-RTA, 0.39 n~ for FACO-RTA, and 0.22 nM for rRTA, indicating that the extra five Nterminal residues and/or the C171S mutation did affect catalytic activity to a small degree.
Reassociation experiments with purified FACCRTA and biochemically purified RTB produced very little holotoxin when the two proteins were mixed without further additions. This contrasts with a high level of reassociation under the same conditions when native RTA and RTB were mixed (data not shown). The RTA cysteine that forms the disulfide bond with RTB (Cys2,') is close to the C terminus of RTA and, in FAC4-RTA, is also close to the introduced C y~"~-C y s~'~ disulfide bond. It is possible that the introduced disulfide bond caused a structural constraint near the FACCRTA C terminus that precluded ready reassociation between FAC4-RTA and RTB. However, attempts to reassociate FACO-RTA with RTB were equally un- successful, indicating that the presence of the engineered disulfide bond was not itself limiting reassociation. The addition of 10 mM glutathione (in a reduced-to-oxidized ratio of 2:l (lo)), glutathione plus 1.4 J~M protein disulfide isomerase, or Y 4 volume of isolated reticuloplasm (the soluble components of the endoplasmic reticulum lumen from MOPC-315 cells (11)) increased the degree of reassociation significantly. Reassociation in the presence of both glutathione and protein disulfide isomerase produced the highest level of holotoxin (Fig. 6). Holotoxin containing FACCRTA was isolated by affinity chromatography on immobilized lactose and eluted with galactose (Fig. 6). This ensured that the holotoxin preparation did not contain FAC4-RTA dimers or aggregates. Likewise, holotoxins containing either wild type rRTA or FACO-RTA were obtained and affinity purified.
In holotoxin containing FAC4-RTA, the interchain disulfide bond was more easily reduced by dithiothreitol than the intrachain disulfide in FACCRTA. At low dithiothreitol concentrations, the small amount of FAC4-RTAreductively released from the holotoxin ran on the gel with a mobility characteristic of the disulfide cross-linked polypeptide (Fig. 7). As the dithiothreitol concentration was increased, more FACCRTA was released from the holotoxin, and an increasing proportion of the released These antibodies did not cross-react with RTB. Lane 7 shows molecular mass markers (106,80,49.5,32.5,27.5,and 18.5 kDa).
protein ran with the lower mobility characteristic of the fully reduced polypeptide (Fig. 7). In the presence of 25 mM dithiothreitol the holotoxin dissociated completely and all of the released FAC4-RTA was reduced (data not shown). When holotoxin containing FACO-RTA was likewise treated with a range of dithiothreitol concentrations, the released FACO-RTA ran as a single band whose mobility was identical to that of reduced FAC4-RTA (data not shown). The cytotoxicities of the holotoxins towards Vero cells were then compared (Fig. 8). Holotoxin containing wild type rRTA was more cytotoxic (with a n IC,, value of 0.12 ng/ml (Table 11)) than holotoxin containing FACO-RTA (ICs, 1.4 ng/ml). Significantly, holotoxin containing FACO-RTA was 18-fold more cytotoxic than holotoxin containing FAC4-RTA (ICs,, for FAC4-RTA holotoxin, 25 ng/ml). Similar results were obtained in cytotoxicity assays using HeLa cells (Table 11). For the cytotoxicity assay (Fig. 8), each point is the mean value from triplicate samples, and the cytotoxicity profiles obtained were observed in three separate experiments.

DISCUSSION
In the present paper we describe the replacement of pairs of RTA residues with cysteine residues in an attempt to engineer novel disulfide bonds into the protein. The criteria for choosing these cysteine pairs were based on the coordinates of cysteine residues in naturally occurring disulfide bonds from known protein x-ray structures. These criteria were as follows: (i) the chosen cysteine pairs were separated by more than 20 residues of the polypeptide chain; (ii) the Ca-Ca distances fell within the range for naturally occurring disulfide bonds (4.6-6.8 A) (16); (iii) the target Cps were reasonably juxtaposed so as to allow disulfide bond formation without serious disruption of the backbone; (iv) there was no obvious potential steric hindrance to disulfide bond formation; and (v) neither of the target residues selected for mutagenesis was strictly conserved in the sequence alignment of plant toxin A chains or single chain ribosome-inactivating proteins (17). Such conservation could imply an essential structural or functional role for the residue. Similarly potential disulfide bonds were rejected if one of the target residues was located near the RTA active site cleft.  .1 (lane 31, 0.5 (lane 4 ) , 1 (lane 5 ) and 2.5 (lane 6 ) mM dithiothreitol. 10 mM iodoacetamide was then added and, after a further 10 min a t 37 "C, the samples were run on a nonreducing SDS-polyacrylamide gel electrophoresis and Western blotted, and RTA was visualized using sheep anti-recombinant RTA antibodies. These antibodies did not cross-react with RTB.  A further consideration was based on the presumed mechanism of ricin entry into cells. Before translocation into the cytosol occurs, RTA may dissociate from RTB following reduction of the interchain disulfide bond. This implies that the solvent at the site of RTA translocation may carry a reducing potential. The residue pairs listed in Table I were chosen because they should generate disulfide bonds in the polypeptide that are inaccessible to solvent. In this way an engineered disulfide bond ought to represent a stable constraint on protein unfolding.
Of the four cysteine pairs that were introduced into RTA (Table I), only one pair (S215C,M255C) led to the production of significant amounts of soluble, stable, and biologically active mutant product (designated FACCRTA) in E. coli. Fig. 9 shows the RTA backbone and the position of the engineered disulfide bond of FAC4-RTA. Both of the introduced cysteines are in reasonably close positions to the naturally occurring CysZ5'. From the x-ray structure of RTA (18) we concluded that neither CYS"~ nor Cys255 were likely to form a disulfide bond with C~S~~' (191, leaving it free to interact with Cys4 of RTB during holotoxin formation. Both Cys215 and Cys255 are buried in the RTA molecule, and if a disulfide bond had formed between either of these residues and CysZ5', the remaining free cysteine residue would also be buried and would be unable to interact with Cys4 disulfide bond introduced into FAC4-RTA of RTB. On the other hand, FACCRTA requires the addition of protein disulfide isomerase and glutathione before it associates to a significant extent with RTB. One possible explanation for this is that the disulfide bond in FACCRTA forms between one of the introduced cysteine residues and CYS"~ and that this form cannot associate with RTB. Thiol exchange, catalyzed by protein disulfide isomerase, might rearrange the disulfide bond to the Cy~~'~-Cys~'' form, which is then capable of associating with RTB via Cys2". However, since FACO-RTA, which only contains a single cysteine residue (CysZ5'), also requires protein disulfide isomerase and glutathione to associate effectively with RTB, we feel it is unlikely that the disulfide bond in FAC4-RTA involves CysZ5'. We are confident, therefore, that the disulfide bond introduced into FAC6RTA joins CysZ1' and Cys'". The introduced disulfide bond should be relatively inaccessible until the protein becomes at least partially unfolded. The current model for ricin cytotoxicity assumes that the reductive release of RTA from the holotoxin precedes the RTA membrane translocation step; RTB is not thought to cross the vesicle membrane. If the internal disulfide bond in FAC4-RTA subsequently constrains its unfolding, then this disulfide bond should be less sensitive to reduction than the disulfide bond joining RTA and RTB. Consistent with this, we observed that the interchain disulfide bond in holotoxin containing FAC4-RTA was more readily reduced than the intrachain disulfide bond present in FACCRTA itself (Fig. 7). In keeping with the conclusions of earlier studies, we assume that the folded conformation of RTA would be stabilized by the introduction of the disulfide bond (20,21). This effect would presumably be largely the result of a decrease in the configurational chain entropy of the unfolded molecule (22,23).
The introduction of the disulfide bond did not appear to perturb the RTA active site in any way, since FAC6RTA had catalytic activity similar to that of FACO-RTA, which lacked the disulfide cross-link (Fig. 5). "he translational system used to measure in vitro protein synthesis inhibition contained dithiothreitol, and it is possible that this reducing agent could have broken the disulfide cross-link in FAC6RTA (Fig. 5). The presence of an intact disulfide bond may well have reduced the catalytic activity of FACCRTA in comparison with that of