Role of Intrachain Disulfides in the Activities of the CdtA and CdtC Subunits of the Cytolethal Distending Toxin of Actinobacillus actinomycetemcomitans

Abstract

The cytolethal distending toxin (Cdt) of Actinobacillus actinomycetemcomitans is an atypical A-B-type toxin consisting of a heterotrimer composed of the cdtA, cdtB, and cdtC gene products. The CdtA and CdtC subunits form two heterogeneous ricin-like lectin domains which bind the holotoxin to the target cell. Point mutations were used to study CdtC structure and function. One (mutC216^F97C) of eight single-amino-acid replacement mutants identified yielded a gene product that failed to form biologically active holotoxin. Based on the possibility that the F97C mutation destabilized a predicted disulfide, targeted mutagenesis was used to examine the contribution of each of four cysteine residues, in two predicted disulfides (C96/C107 and C135/C149), to CdtC activities. Cysteine replacement mutations in two predicted disulfides (C136/C149 and C178/C197) in CdtA were also characterized. Flow cytometry and CHO cell proliferation assays showed that changing either C96 or C149 in CdtC to alanine abolished the biological activity of holotoxin complexes. However, replacing C107 or C135 in CdtC and any of the four cysteines in CdtA with alanine or serine resulted in only partial or no loss of holotoxin activity. Changes in the biological activities of the mutant holotoxins correlated with altered subunit binding. In contrast to elimination of the B chain of ricin, the elimination of intrachain disulfides in CdtC and CdtA by genetic replacement of cysteines destabilizes these subunit proteins but not to the extent that cytotoxicity is lost. Reduction of the wild-type holotoxin did not affect cytotoxicity, and the reduced form of wild-type CdtA exhibited a statistically significant increase in binding to ligand. A diminished role for intrachain disulfides in stabilizing CdtA and CdtC may have clinical relevance for the A. actinomycetemcomitans Cdt. The cdt gene products secreted by this pathogen assemble and bind to target cells in periodontally involved sites, which are decidedly reduced environments in the human oral cavity.

Members of several genera of pathogenic gram-negative facultative bacteria, including Escherichia coli (21, 23), Campylobacter species (19, 24, 37), Shigella dysenteriae (22, 42), Haemophilus ducreyi (7), Helicobacter species (6, 27, 52), and Actinobacillus actinomycetemcomitans (34, 48, 50), produce a cytolethal distending toxin (Cdt). The primary mechanism of action of the Cdt is the induction of cell cycle arrest at the G₀/G₁ or G₂/M transition in various types of eukaryotic cells and cell lines (reviewed in references 9, 12, 13, 18, 29, 41, 43, 45, and 52). We have been interested in the Cdt of the periodontal pathogen A. actinomycetemcomitans as a potential virulence factor.

The Cdt of A. actinomycetemcomitans, like all members of this cytotoxin family, is an atypical A-B-type toxin consisting of a heterotrimer composed of 18- to 25-kDa (CdtA), 31-kDa (CdtB), and 21-kDa (CdtC) gene products (34, 48, 50). The deduced amino acid sequences of the A. actinomycetemcomitans CdtA, CdtB, and CdtC protein subunits are greater than 90% identical to those of the H. ducreyi Cdt (7, 34, 50). The CdtB subunit is most closely related to neutral nucleases of the type I DNase family (3, 10, 14, 16, 28, 29). In the crystal structures of the H. ducreyi and A. actinomycetemcomitans holotoxins, the CdtA and CdtC subunits form two heterogeneous ricin-like lectin domains that comprise a putative receptor-anchoring groove (20, 38, 39, 51). Purified recombinant CdtA and CdtC bind to cells in culture (2, 25, 30, 33, 35) and in an enzyme-linked immunosorbent assay of cells (CELISA) (5, 30). The cell surface receptor for the Cdt has not yet been identified. However, there is evidence that both CdtA and CdtC may be carbohydrate-binding proteins that recognize N-linked fucose moieties on the surfaces of HeLa cells (35). Other studies implicate gangliosides, such as GM1 and GM3, in toxin-cell recognition (36). In addition to being presumed to have a cell-binding function, CdtC may facilitate the transport of CdtB into the cell by an endosome-mediated process (1, 8, 17).

Examination of the crystal structure of the Cdt indicates that each of the CdtA and CdtC subunits contains two predicted intrachain disulfides (38, 51). The cysteines that comprise the disulfides are highly conserved in the CdtA and CdtC protein families (20). Intrachain disulfides are important for the proper folding and, in some instances, secretion of proteins. Ricin contains an interchain disulfide between the A and B chains that is essential for activity (4). However, the B chain also contains predicted intrachain disulfides suggested to be important for the protein to fold properly to maintain galactose (receptor)-binding activity (32).

Our recent studies have been focused on using mutagenic approaches to obtain more-detailed information about CdtA and CdtC structures and functions. In an earlier study, we constructed a library of proteins with randomly generated point mutations in cdtA (5). It was found that single conserved amino acid substitutions residing outside of predicted binding domains (aromatic-patch region) can significantly reduce the binding and biological activity of the holotoxin without necessarily affecting holotoxin assembly. In the present study, we used both random and targeted mutagenesis strategies to further dissect the molecular interactions of CdtC and CdtA. In the targeted strategy, we mutated each of the cysteine residues in both of these subunit proteins to evaluate the contributions of structurally predicted intrachain disulfides to subunit binding, Cdt assembly, and cytotoxicity.

MATERIALS AND METHODS

Random mutagenesis of cdtC.

The oligonucleotide primer pair cdtC-F/cdtC-R (Table 1) and pJDC2 plasmid DNA (5) were used with the GeneMorph random-mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer's instructions, to generate a set of mutations, at low frequency, in the cdtC gene. Escherichia coli BL21(DE3)(pJDC2) expresses the wild-type CdtC gene, and the gene product contains a His₆ tag at the carboxy-terminal end (5). Attempts were made to introduce mutations at a very low frequency to favor the recovery of mutants having a single nucleotide change (point mutation). Mutated cdtC DNA fragments were then cloned into the NcoI and BamHI sites of pET15b and transformed into Library Efficiency E. coli DH5α [supE44 ΔlacU169 (φ80lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1] (Invitrogen, Carlsbad, CA). Potential mutants were screened, identified, and characterized as described previously (5). Briefly, the transformants were pooled, and the plasmid DNA was extracted and used to transform E. coli BL21(DE3) [F⁻ ompT hsdS_B (r_B⁻ m_B⁻) gal dcm (DE3)] competent cells (Novagen, Madison, WI). Individual transformants were grown in LB broth medium containing 75 μg/ml of ampicillin. Isopropyl-β-d-thiogalactopyranoside (IPTG) (Sigma Chemicals, St. Louis, MO) was added to a final concentration of 1 mM when the cultures reached late logarithmic phase (optical density at 600 nm of 0.8 to 1.0) to induce gene expression. Bacteria were collected from the induced cultures, washed, and lysed at 4°C by sonication (three bursts of 1-min duration each at the low power setting; Braun-Sonic 2000, B. Braun Biotech, Inc., Allentown, PA). Inclusion bodies were collected by centrifugation and were dissolved and stored in 200 μl of a 6 M urea-phosphate-buffered saline buffer. These samples were used only for the initial screening of the transformants in the CELISA.

TABLE 1.

Oligonucleotide primers used for random and site-directed mutagenesis

Primer	Sequencea
CdtC-F	5′-TCCCGCCGGTTTTGTCCATGGTCGCTAAGGAGAATAC-3′
CdtC-R	5′-AATGGATCCGTGGTGGTGGTGGTGGTGGCTACCCTGATT-3′
CdtC-C96A-F	5′-GTCGGTGCAGCAAAAGCTTTTGCTTTCCCAGGCGAA-3′
CdtC-C96A-R	5′-TTCGCCTGGGAAAGCAAAAGCTTTTGCTGCACCGAC-3′
CdtC-C107A-F	5′-GAAGGCACAACTGATGCGAAAGATATTGATCATACTG-3′
CdtC-C107A-R	5′-CAGTATGATCAATATCTTTCGCATCAGTTGTGCCTTC-3′
CdtC-C135A-F	5′-GATGCCCTATTAGGATTTGCGATGACAAGCCATG-3′
CdtC-C135A-R	5′-CATGGCTTGTCATCGCAAATCCTAATAGGGCATC-3′
CdtC-C149A-F	5′-TTGAGGCTAGAACCTGCGGGAATTTCAGTGAGTGG-3′
CdtC-C149A-R	5′-CCACTCACTGAAATTCCCGCAGGTTCTAGCCTCAA-3′
CdtA-C136A-F	5′-AATCTTTAGGTACAGCGATTGAAGCTTACGGTAATGG-3′
CdtA-C136A-R	5′-CCATTACCGTAAGCTTCAATCGCTGTACCTAAAGATT-3′
CdtA-C149A-F	5′-AATTCATGATACTGCGAGTCTGGACAAATTAGCAC-3′
CdtA-C149A-R	5′-GTGCTAATTTGTCCAGACTCGCAGTATCATGAATT-3′
CdtA-C178A-F	5′-AAAAGTGTGTCACAAGGACGTGCAGTCACTTATAATCCTGTAAGTCC-3′
CdtA-C178A-R	5′-GGACTTACAGGATTATAAGTGACTGCACGTCCTTGTGACACACTTTT-3′

^aUnderlined bases mark the alanine codon.

Identification and expression of mutated CdtC-His₆ proteins were carried out as described previously (5). First, binding activity in the crude, solubilized inclusion body fractions was screened in a CELISA modified from the assay described by Lee et al. (30). Expression of the cdtC gene was assessed by examining these fractions on Western blots. Prestained molecular weight standards were obtained from Bio-Rad (Hercules, CA). Sonicated lysates from E. coli BL21(DE3)(pJDC2) and E. coli BL21(DE3)(pET15b) were used as positive and negative controls, respectively. Plasmid DNA was obtained from the mutant library cdtC clones using a QIAprep miniprep kit (QIAGEN, Valencia, CA). The plasmid DNA from each clone was digested with restriction endonucleases NcoI and BamHI according to the manufacturer's instructions. The digested samples were examined with 8% agarose gels to confirm the presence and size of the plasmid insert DNA. Automated cycle sequencing reactions were conducted by the Genetics Core Facility at the University of Pennsylvania using an Applied Biosystems 377 sequencer with dye primer chemistry. The expressed gene product was isolated from those transformants that had mutations confirmed by DNA sequencing and was used in the subsequent binding and activity assays. CdtC-His₆ expressed from the wild-type recombinant cdtC gene from E. coli BL21(DE3)(pJDC2) (5) served as a positive control.

Site-directed mutagenesis of cdtC and cdtA.

The cysteine residues C96, C107, C135, and C149 in CdtC and C136, C149, and C178 in CdtA were replaced using site-directed mutagenesis. Synthetic oligonucleotide primer pairs (Table 1) were used to change each cysteine residue to alanine. Mutant DNA strands were made using PfuUltra DNA polymerase in PCR (Stratagene, La Jolla, CA). Plasmid DNA preparations from pJDC2 and pJDA9 (5) were used as PCR templates for CdtC and CdtA cysteine mutagenesis, respectively. Methylated parental DNA strands were digested with DpnI (New England Biolabs, Beverly, MA) and transformed into E. coli TOP10 [F⁻ mcrAΔ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15ΔlacX74 recA1 araD139Δ(ara-leu)7697 galU galK rpsL(Str^r) endA1 nupG] chemically competent cells (Invitrogen). The mutations were confirmed by sequencing of the plasmid insert DNA. Plasmid DNA having the confirmed sequence was isolated and transformed into E. coli BL21(DE3) to express the mutant gene and to isolate the gene product.

Isolation of recombinant wild-type and mutant subunit proteins and reconstitution of holotoxin.

Recombinant clones E. coli BL21(DE3)(pJDA9), E. coli BL21(DE3)(pJDB7), and E. coli BL21(DE3)(pJDC2) were used to prepare the three wild-type CdtA-His₆, CdtB-His₆, and CdtC-His₆ proteins, respectively, by affinity chromatography as described previously (5). All three proteins have His₆ tags at their carboxy-terminal ends. The same method was used to obtain the mutated CdtC-His₆ and CdtA-His₆ gene products from the cdtC random-mutagenesis library and cysteine mutants, respectively. Yields were approximately 5 mg of protein/100 ml of culture. The final protein preparations were dialyzed to remove urea, passed through 45-μm filters, and quantified with a Micro BCA protein assay kit (Pierce, Rockville, IL) as described previously (5). Purity was assessed by analysis using 10 to 20% polyacrylamide gels. Aliquots of the quantified protein samples were stored at −70°C in a buffer containing 10 mM Tris-HCl (pH 7), 100 mM NaCl, 5 mM MgCl₂, and 5 mM imidazole for use in the binding and activity assays.

Wild-type holotoxin and holotoxin containing mutant CdtC-His₆ or CdtA-His₆ proteins were reconstituted as described previously (5, 33). The affinity-purified recombinant His₆-tagged proteins were mixed in equivalent proportions, by mass, and incubated for 1 h at 4°C in a reconstitution buffer containing 10 mM Tris-HCl (pH 7), 100 mM NaCl, and 5 mM MgCl₂. In some experiments, 10 mM dithiothreitol (DTT; Sigma Chemicals) was added to the reconstitution buffer to reduce the holotoxin.

Cytotoxicity assays.

Cell cycle arrest was determined by flow cytometry as described previously (25). Chinese hamster ovary K1 (CHO-K1) cells were grown in Ham's F-12 medium containing 5% fetal calf serum overnight at 37°C (34). Cultures were then treated with 10 μg (total protein) of reconstituted holotoxin/ml of culture medium. Reconstituted holotoxins contained either a CdtC-His₆ or a CdtA-His₆ mutant protein and the remaining two wild-type subunits. Cells were exposed to the holotoxin preparations in culture for 36 h. Propidium iodide-stained nuclei were prepared from 1 × 10⁶ cells and were analyzed on a FACSCalibur flow cytometer at the University of Pennsylvania Cancer Center Flow Cytometry and Cell Sorting Shared Resource facility. The data from 30,000 events were analyzed with ModFit 3.0 (Verity Software House, NH). Fluorescence-activated cell sorter analyses were repeated three times using three independent holotoxin-treated cultures.

To quantify the effects of the reconstituted mutant holotoxins on cell proliferation, CHO cells were grown and exposed to toxin as described previously (5, 33). Mutant holotoxin preparations were reconstituted without urea and typically contained 5 μg of total reconstituted protein/ml of culture medium. After 6 days of growth, colonies were fixed, stained, and counted as described previously (5, 33). The data were expressed as numbers of CFU. In those experiments that examined the effect of the reducing agent on Cdt activity, 10 mM DTT was added to both the refolding buffer and the cell culture medium. The DTT did not affect the growth of CHO cells. Proliferation assays were run a minimum of three times using independently grown cultures and various batches of reconstituted holotoxin.

Pulsed-field gel electrophoresis (PFGE) was performed as described previously (25). HeLa cell cultures (5 × 10⁶ cells) were exposed to 10 μg/ml of reconstituted holotoxin prepared with wild-type CdtA-His₆ and CdtB-His₆ and either a wild-type or a mutant CdtC-His₆ protein for 48 h. The treated cells were lysed, in situ, in agarose plugs. Electrophoresis was performed for 40 h at 4°C. The ratio of DNA in the well to that in the gel was estimated from digitized images of ethidium bromide-stained gels using the software program ImageJ version 1.34 (http://rsbweb.nih.gov/nih-image/index.html). The PFGE experiments were repeated twice.

Holotoxin complex formation in solution was determined by differential dialysis (5). Briefly, a dialysis membrane with a molecular mass exclusion limit of 100 kDa (Spectrum Laboratories Inc., Rancho Dominguez, CA) was soaked and rinsed in distilled water. Combinations of purified wild-type and mutant subunit proteins (100 μg of each protein or mutant protein) were suspended in a total volume of 1 ml of refolding buffer containing 0.3 M urea. Each sample was dialyzed against 800 ml of the same buffer for 48 h at 4°C with two buffer changes. An aliquot (35 μl) of each reconstituted protein sample was then examined before and/or after dialysis on Western blots as described previously (5). The amount of protein in each immunopositive band was quantified by analyzing digitized images with ImageJ. In those experiments that examined the effect of the reducing agent on complex formation, 10 mM DTT was added to the refolding and dialysis buffers. These experiments were performed two to three times on separate batches of purified recombinant proteins.

Binding assays.

A CELISA (5) was used to initially measure the binding activity in the crude inclusion body fractions from the CdtC-His₆ random-mutagenesis transformants. The CELISA was repeated with affinity-purified gene products expressed from the small number of CdtC-His₆ transformants that had confirmed point mutations. Purified wild-type CdtC-His₆ and mutant CdtC-His₆ mutant proteins were added to triplicate wells in 96-well plates (10 μg/well) containing attached CHO cells (1.5 × 10⁴ cells/well). Bound protein was detected with anti-His · Tag monoclonal antibody (1:3,000 dilution; Novagen) and antimouse immunoglobulin G horseradish peroxidase conjugate (1:3,000 dilution; Amersham Biosciences, Piscataway, NJ) as described previously (5). These experiments were run a minimum of three times.

To assess ligand and subunit binding activities of the CdtA-His₆ and CdtC-His₆ mutant proteins, a thyroglobulin ELISA (5) based on the immobilized glycoprotein binding experiments of McSweeney and Dreyfus (35) was used. In all experiments, microtiter plates (96 wells) were coated with 75 μg/well of thyroglobulin (Sigma Chemicals) overnight. To determine the subunit binding activity of CdtC-His₆ wild-type or mutant proteins, the thyroglobulin plates were first treated with wild-type CdtA-His₆ (4 μg/well). Saturation kinetics were determined for the binding of wild-type CdtC-His₆ to wild-type CdtA-His₆ by adding 0 to 16 μg/well of wild-type CdtC-His₆ to triplicate wells. Plates were washed, treated with anti-His · Tag monoclonal antibody and horseradish peroxidase conjugate, and processed as described above for the CELISA. In other experiments, each random CdtC-His₆ amino acid substitution or cysteine replacement mutant (3.5 μg/well) was added to triplicate wells, and the plates were processed as described above. In all CdtC-His₆ subunit binding experiments, an absorbance ratio was calculated by dividing the average absorbance value (of triplicate wells) of the CdtC-His₆ wild-type- or mutant-protein-containing well by the average absorbance value of wells containing only bound wild-type CdtA-His₆. To assess binding of the CdtA-His₆ cysteine replacement mutants to ligand, 10 μg of each mutant protein/well was added to the thyroglobulin-coated wells. The plates were incubated and processed as described above.

The thyroglobulin ELISA was also used to determine the effect of the reducing agent on wild-type CdtA-His₆ binding to ligand. The thyroglobulin and/or the wild-type CdtA-His₆ subunit protein was treated with a reducing buffer containing 10 mM DTT, 4 M guanidine-HCl, and 10 mM Tris-HCl (pH 8.5) for 10 min at room temperature (15). Separate ELISA plates were first coated with reduced and unreduced thyroglobulin. Reduced and unreduced wild-type CdtA-His₆ (0 to 16 μg/well) was then added to triplicate wells in both the reduced and unreduced thyroglobulin-coated plates. Reduced wild-type CdtA-His₆ was added in the presence of binding buffer containing 10 mM DTT. Plates were developed as described above. All experiments were repeated three times with different batches of purified recombinant proteins.

Computer analysis.

The European Molecular Biology Open Software Suite (EMBOSS, release 3.0; http://emboss.sourceforge.net) (46) was used to obtain amino acid sequence alignments and to calculate isoelectric points (pI). Deduced amino acid sequences of wild-type and mutant A. actinomycetemcomitans recombinant CdtA-His₆ and CdtC-His₆ were from nucleic acid sequences obtained during this study. The H. ducreyi CdtA and CdtC sequences were from GenBank (accession number U53215). The crystal structure of the A. actinomycetemcomitans Cdt (51) was modeled with the UCSF Chimera 1.2197 software program (http://www.cgl.ucsf.edu/chimera/) (44). Coordinates were obtained from the Protein Data Bank (accession number 2F2F). Bond distances in angstroms and surface-exposed residues were also determined with this program. Disulfide bonds were also predicted using the genomic disulfide analysis program (GDAP) (http://www.doe-mbi.ucla.edu/∼boconnor/GDAP/) (40).

RESULTS

Characterization of point mutations in cdtC.

A mutagenesis strategy similar to that used for the creation of a library of point mutations in cdtA (5) was employed in the present study. Of 304 transformants selected at random following mutagenesis, 134 exhibited binding activity in the CELISA lower than that of the wild-type CdtC-His₆-containing extract. Seventy of these 134 transformants exhibited a full-length gene product on Western blots. The plasmid DNA insert fragments from these 70 transformants were sequenced. Reliable DNA sequence data could not be retrieved from either of two sequencing runs for each of nine transformants. Forty-six transformants contained wild-type cdtC sequences. One transformant had three amino acid substitutions, and another transformant had two amino acid substitutions. Five transformants had single nucleotide changes but no corresponding amino acid substitutions. These transformants apparently gave false-positive results in the CELISA run with crude bacterial extracts. Eight transformants each had one or two nucleotide changes and a single amino acid substitution. These mutants all had altered CHO cell binding phenotypes and are described in Table 2. The amino acid substitutions in three of the mutants (mutC191, mutC278, and mutC378) resulted in a small increase (0.2 units) in the calculated pI of the proteins.

TABLE 2.

Summary of cdtC and cdtA mutations

Plasmid	Mutant designationa	Nucleotide change(s)	Amino acid change	Calculated pIb	Position in H. ducreyi CdtC/CdtAc
pMUT191cdtC	mutC191	A242G	D81G	7.3743	D81
pMUT216cdtC	mutC216	T253C/T290G	None/F97C	7.1310	F97
pMUT222cdtC	mutC222	T211C	Y71H	7.1769	Y71
pMUT268cdtC	mutC268	A43G	T15A	7.1396	A15
pMUT278cdtC	mutC278	A156G/T180A	N60K/none	7.3742	N60
pMUT300cdtC	mutC300	A231T	E77D	7.1395	E77
pMUT368cdtC	mutC368	T431C	L144S	7.1396	L144
pMUT378cdtC	mutC378	C50G/A501G	T17R/none	7.3743	A17
pMUTc96cdtC	mutCc96	T286G, G287C	C96A	7.1485	C96
pMUTc107cdtC	mutCc107	T319G, G320C, T321G	C107A	7.1485	C107
pMUTc135cdtC	mutCc135	T403G, G404C, T405G	C135A	7.1485	C135
pMUTc149cdtC	mutCc149	T445G, G446C, T447G	C149A	7.1485	C149
pMUTc136cdtA	mutAc136	T406G, G407C, T408G	C136A	8.3192	C137
pMUTc149cdtA	mutAc149	T445G, G446C, T447G	C149A	8.3192	C150
pMUTc178cdtA	mutAc178	T532G, G533C, T534A	C178A	8.3192	C179
pMUT65cdtAd	mutA65	T598A	C197S	8.3192	C198

^amutC and mutA represent mutations in cdtC and cdtA, respectively. A lowercase c indicates a cysteine.

^bWild-type CdtC-His₆ pI, 7.1396; wild-type CdtA-His₆ pI, 8.2164.

^cData are from Cope et al. (7). GenBank accession number U53215.

^dData are from Cao et al. (5).

CdtC-His₆ was isolated by affinity chromatography from each of the eight mutants (Table 2) that had confirmed, unique, single amino acid substitutions. Reconstituted holotoxins were prepared using the mutant CdtC-His₆ protein and wild-type CdtA-His₆ and CdtB-His₆ subunit proteins. Cultures of CHO cells were treated with the various mutant-CdtC-containing holotoxins, and cell survival was compared to that of cultures treated with holotoxin reconstituted with the wild-type CdtC-His₆ (Fig. 1A). Only holotoxin made with the mutant protein from mutC216 failed to cause a reduction in the number of CFU. In addition, no DNA damage was observed when HeLa cells were treated in culture with holotoxin made with CdtC-His₆ from this mutant (Fig. 1B). Holotoxins made with the other mutant CdtC-His₆ proteins induced DNA damage in HeLa cells (Fig. 1B; only mutC268 is shown). We recently showed that differential dialysis can be used to determine the ability of purified Cdt subunit proteins to form a heterotrimer complex (5). Using this method, the mutC216 protein showed much less complementation of heterotrimer formation in solution with wild-type CdtA-His₆ and CdtB-His₆ subunits than wild-type CdtC-His₆ (Fig. 1C). Three of the other mutant CdtC-His₆ proteins are shown for comparison.

FIG. 1.

Effect of point mutations in CdtC on the cytotoxic activities of reconstituted heterotoxin. (A) Effects on CHO cell proliferation. Affinity-purified recombinant wild-type CdtA-His₆, CdtB-His₆, and CdtC-His₆ or mutant CdtC-His₆ proteins were preincubated in folding buffer, and the mixture (5 μg/ml of culture medium) was added to CHO cell cultures. Cell colonies were stained and counted after 6 days of culturing, and the data are expressed as numbers of CFU. Amino acid substitutions are shown for each mutant. CdtC designates holotoxin reconstituted with wild-type CdtC-His₆. All samples were run in triplicate. Statistically significant differences between the numbers of CFU of untreated CHO cell cultures and of those treated with wild-type or mutant heterotoxins are marked with an asterisk (P < 0.000005). (B) Assessment of DNA damage by PFGE. Reconstituted wild-type or mutant heterotrimers (10 μg of total protein/ml) were added to HeLa cells, and the cultures were incubated for 36 h. The cells were then prepared for PFGE as described in Materials and Methods. The numbers represent the relative percentages of DNA retained in the well versus that in the gel. (C) Assessment of heterotoxin assembly by differential dialysis. Heterotrimers composed of preincubated mixtures of either wild-type or mutant CdtC-His₆ proteins were dialyzed for 48 h. The protein composition of the material remaining after dialysis was examined by Western blotting as described in Materials and Methods. CdtA′ represents the truncated form (17 to 18 kDa) of CdtA-His₆. Results of all experiments were typical of a minimum of three trials.

Thyroglobulin is a fucose-containing glycoprotein that mimics a putative Cdt cell surface receptor (35). An ELISA that takes advantage of this property of thyroglobulin (5) was used to examine the in vitro binding kinetics of CdtC-His₆ from mutC216. Bound protein was detected with an anti-His · Tag monoclonal antibody. We previously observed that, although wild-type CdtC-His₆ bound poorly to thyroglobulin, CdtC-His₆ bound in a stoichiometric ratio (1:1) to CdtA-His₆ (5). The same results were obtained if the thyroglobulin plates were precoated with CdtA-His₆ or if the two subunit proteins were added together. Wild-type CdtC-His₆ binding to wild-type CdtA-His₆ exhibited saturation kinetics (Fig. 2A) _. However, only CdtC-His₆ from mutC216 exhibited a statistically significant decrease in binding relative to that of the wild-type protein (Fig. 2B) (P < 0.005). The wild-type and other mutant CdtC-His₆ proteins bound to CdtA-His₆ in an almost 1:1 ratio (there was an increase in the absorbance ratio from 1.0 to 1.7 to 1.0 to 1.9 in this experiment).

FIG. 2.

Effect of point mutations on the binding activities of CdtC. (A) Saturation curve of the binding of wild-type recombinant CdtC-His₆ to CdtA-His₆ on thyroglobulin-coated wells from an ELISA. Bound Cdt protein was detected with anti-His · Tag monoclonal antibody (1:3,000 dilution) and antimouse immunoglobulin G horseradish peroxidase conjugate (1:3,000 dilution) as described in Materials and Methods. The dashed line marks the absorbance value of bound CdtA-His₆ at an input concentration of 4 μg/well. All CdtC-His₆ concentrations were run in triplicate. (B) Affinity-purified recombinant CdtC-His₆ and the mutant proteins (10 μg/well) were added to ELISA plate wells coated with thyroglobulin (white bars) or thyroglobulin and 4 μg/well CdtA-His₆ (black bars). mutC162 has a single nucleotide change (T46C) but no corresponding amino acid change. Bound protein was detected as in the experiment whose results are shown in panel A, and absorbance values were compared to those of wells containing only 4 μg/well CdtA-His₆ (middle dotted line). All samples were run in triplicate. Statistically significant differences between the absorbance values for the mutant and wild-type CdtC-His₆ proteins bound to wild-type CdtA-His₆-coated thyroglobulin are marked with an asterisk (P < 0.005). Results of all experiments were typical of three trials.

The results of the mutant holotoxin activity assays and the CdtA-His₆ binding kinetics were consistent in establishing that the amino acid substitution in mutC216^F97C was the only mutation identified in the random-mutagenesis library that had a cytotoxin-negative phenotype. The loss of biological activity is most likely due to the inability of this protein to form a heterotrimer complex with wild-type CdtA and CdtB. While mutC216^F97C is the most interesting mutant because of its holotoxin assembly-negative phenotype, it is important to note that altering the primary sequence of CdtC at a number of other sites (Fig. 3A and 4A) did not affect the biological activities of the protein.

FIG. 3.

Positions of the amino acid substitutions in CdtC and CdtA. (A) Alignment of the deduced amino acid sequences of A. actinomycetemcomitans recombinant CdtC-His₆ (5) and H. ducreyi CdtC (GenBank accession number U53215). Unique amino acid substitutions identified in the random-mutagenesis library and from the targeted cysteine mutagenesis are shown. Nonidentical residues are marked by shaded boxes. (B) Alignment of the deduced amino acid sequences of A. actinomycetemcomitans recombinant CdtA-His₆ (5) and H. ducreyi CdtA (GenBank accession number U53215). Amino acid substitutions for the four cysteine residues are shown. Nonidentical residues are marked by shaded boxes. The substitution C197S is in mutA65 from Cao et al. (5). (C) Alignment of the deduced amino acid sequences of A. actinomycetemcomitans recombinant CdtA-His₆ and CdtC-His₆, comparing the relative locations of the substituted cysteine residues. Identical amino acids are marked by shaded boxes. Cysteine residues in CdtA-His₆ and CdtC-His₆ not predicted to form disulfides are marked with an asterisk.

FIG. 4.

Locations of the amino acid substitutions in CdtC-His₆ and CdtA-His₆ in the crystal structure of the A. actinomycetemcomitans Cdt. The coordinates (Protein Data Bank accession number 2F2F) for the crystal structure were computer modeled. (A) The eight unique amino acid substitutions in the A. actinomycetemcomitans recombinant CdtC-His₆ mutants having a binding-deficient phenotype are shown at their corresponding positions in the A. actinomycetemcomitans crystal structure. Positions of the mutated residues within the structure are in orange. The position of the one substitution (F97C) that resulted in a noncytotoxic phenotype is underlined. The amino [N(E21)]- and carboxy [C(S186)]-terminal residues in CdtC are labeled. Only the backbone structure is shown. (B) Locations of predicted disulfide-forming cysteine residues in CdtC and CdtA in the A. actinomycetemcomitans crystal structure. There is one amino acid difference between the CdtA-His₆ shown in Fig. 3 (5) and that used to form the Cdt crystal structure (Protein Data Bank accession number 2F2F). Calculated distances between paired cysteines predicted to form disulfide bonds are shown in angstroms. Only the backbone structure is shown. Note that the structure in panel B is rotated for clarity relative to that in panel A. (C) Crystal structure of the A. actinomycetemcomitans Cdt showing the surface exposed residues in CdtC (blue) and CdtA (green). Exposed cysteines are yellow and are labeled. (D) Same structural model as that shown in panel C but rotated to show the exposed cysteines in CdtA.

Cysteine mutants of CdtC and CdtA.

Four cysteines are present in the A. actinomycetemcomitans CdtC (Fig. 3A and 4B). These cysteines are conserved in all members of the CdtC family (20). The GDAP (40) predicts two intrachain disulfides in the A. actinomycetemcomitans CdtC protein corresponding to cysteine pairs C96 and C107 (distance, 5.0 Å) and C135 and C149 (distance, 5.913 Å). These predicted distances are in good agreement with distance measurements taken from the A. actinomycetemcomitans Cdt structure in which cysteine pairs at positions 96 and 107 and at positions 135 and 149 have calculated distances of 5.23 Å and 5.30 Å, respectively (Fig. 4B). Disulfides are potentially important for the proper folding of some proteins. It is interesting that the replacement cysteine residue in the biologically defective mutant mutC216^F97C is adjacent to C96 in CdtC-His₆. Since a disulfide is predicted between C96 and C107, the presence of a third cysteine in this immediate location may destabilize the folding of the protein, thus affecting the activity of CdtC. We therefore replaced each of the four cysteines in CdtC-His₆ (positions 96, 107, 135, and 149) with alanine. A fifth cysteine residue exists at the end of the deduced protein sequence (position 196) as a consequence of the addition of the His₆ tag (Fig. 3A). A mutant with a substitution for this cysteine was not made since it is not predicted to be involved in disulfide bond formation.

The A. actinomycetemcomitans CdtA also has four cysteine residues that are predicted, using the GDAP algorithm, to form two intrachain disulfides corresponding to residues C137/C150 (distance, 5.068 Å) and C179/C198 (distance, 5.07 Å). These predicted distances are in reasonable agreement with distance measurements taken from the A. actinomycetemcomitans Cdt structure (Fig. 4B). Therefore, we also replaced three of the four cysteines (C136, C149, and C178) in CdtA-His₆ with alanine. The fourth cysteine (C197) had been changed to serine in a mutant (mutA65) recovered from the CdtA-His₆ random-mutagenesis library published earlier (5). A fifth cysteine, at position 16 in the CdtA-His₆-deduced amino acid sequence (Fig. 3B), was not replaced because it would be eliminated when the protein was posttranslationally modified by the bacterium to remove the signal sequence. The mutants containing cysteine substitutions in CdtC-His₆ and CdtA-His₆ are described in Table 2.

Holotoxin reconstituted with two of the CdtC-His₆ cysteine mutants (mutCc96 and mutCc149) lost the ability to inhibit the proliferation of CHO cells (Fig. 5A). The other two mutant proteins from mutCc107 and mutCc135 formed active holotoxins that exhibited a partial reduction, 28 and 23%, respectively, in the number of CFU relative to that of the untreated culture (P < 0.005 and P < 0.05, respectively). However, it was observed that the CHO cell colonies that formed following exposure to the mutant holotoxins made with protein from mutCc107 and mutCc135 were small in size relative to colonies in untreated cultures (Fig. 5A, inset). The noticeably smaller colony size was undoubtedly a consequence of an altered effect of the mutant holotoxin on cell growth, which led to a slower but inevitable progression to cell death. Therefore, replacement of either C107 or C135 with alanine still resulted in a CdtC-His₆ protein that formed a biologically active, but altered, holotoxin.

FIG. 5.

Effect of cysteine mutant holotoxins on the proliferation of CHO cells. (A) Holotoxin was reconstituted with either wild-type CdtC-His₆ or CdtC-His₆ cysteine mutant proteins and added to CHO cell cultures as described in the legend to Fig. 1. Cell colonies were stained and counted after 6 days of growth, and the data were expressed as numbers of CFU. All samples were run in triplicate. Statistically significant differences between the numbers of CFU in untreated CHO cell cultures and in those treated with wild-type or mutant heterotoxin are marked with asterisks (*, P < 0.005; **, P < 0.05). Images of stained colonies from the untreated and mutCc107 holotoxin-treated cultures are shown in the inset. nc, normal colonies; sc, small colonies. (B) Same experiment as that shown in panel A except that holotoxins were reconstituted with the wild-type and cysteine mutant CdtA-His₆ proteins.

CdtA-His₆ cysteine mutant proteins from mutAc136, mutAc149, and mutAc178 formed holotoxins that caused 100, 21, and 54% reductions, respectively, in the number of CFU relative to that of the untreated culture (P < 0.005) (Fig. 5B). As in the case of the CdtC-His₆ mutant holotoxins, the colonies were relatively small in size (data not shown). We previously found that replacing the fourth cysteine (mutA65^C197S) in CdtA-His₆ resulted in the formation of a mutant holotoxin that exhibited a 100% reduction in CHO cell proliferation under the same experimental conditions (see Fig. 3B in Cao et al. [5]).

Therefore, two of the CdtC-His₆ cysteine mutants and all four of the CdtA-His₆ cysteine mutant proteins formed holotoxins that exhibited partial to full cytotoxicity in the cell proliferation inhibition assay. The same mutant phenotypes were observed by flow cytometry (Table 3). An average of 96% of the CHO cells in cultures treated with wild-type recombinant toxin had a significant 4N DNA content after 36 h of exposure. The high percentage of cells having a 4N DNA content was indicative of cell cycle arrest at the G₂/M transition (5). Only 9 and 7% of the CHO cells in cultures exposed to holotoxin reconstituted with CdtC-His₆ from mutCc96 and mutCc149, respectively, had a 4N DNA content increase. In contrast, 81 and 38% of cells incubated with holotoxin reconstituted with CdtC-His₆ from mutCc107 and mutCc135, respectively, had a 4N DNA content. Holotoxin made with each of the CdtA-His₆ cysteine mutants arrested cells at the G₂/M transition (Table 3 in the present study and mutA65 in Table 4 of Cao et al. [5]).

TABLE 3.

Cell cycle analysis of CHO cells treated with reconstituted mutant holotoxin

Mutant protein(s) or mutant	% of CHO cells that were diploid in:			Coefficient of variance (%)
	G1 (2N DNA content)	G2 (4N DNA content)	S phase
None	73.69	7.06	19.25	4.80
CdtA-H + CdtB-H + CdtC-Ha	3.69	96.31	0	7.21
mutCc96	67.55	9.19	23.26	4.29
mutCc107	9.28	80.61	10.11	6.36
mutCc135	39.65	38.34	22.01	5.91
mutCc149	65.45	7.44	27.10	6.37
mutAc136	12.44	87.56	0	6.34
mutAc149	8.67	79.96	11.37	6.60
mutAc178	1.87	89.82	8.31	7.22

^aWild-type recombinant proteins from E. coli BL21(DE3)(pJDA9), E. coli BL21(DE3)(pJDB7), and E. coli BL21(DE3)(pJDC2), respectively.

The effect of the cysteine substitutions on the binding activity of the CdtC-His₆ subunit was also determined. The ability of the CdtC-His₆ cysteine mutant proteins to bind to CdtA was examined by immobilizing CdtA-His₆ on thyroglobulin and by looking for a stoichiometric increase (doubling) in the absorbance after addition of the mutant proteins. Each of the four CdtC-His₆ cysteine mutant proteins exhibited a statistically significant but a <100% reduction in binding to CdtA-His₆ (P < 0.005) (Fig. 6, gray bars). The absorbance ratios ranged from 1.2 to 1.4. These data showed that replacing the individual cysteines in CdtC-His₆ reduced but did not abolish binding to CdtA.

FIG. 6.

Effect of cysteine replacement on the ligand and subunit binding activities of CdtC-His₆ and CdtA-His₆ proteins. ELISA plates were coated with thyroglobulin. Wild-type CdtA-His₆ (4 μg/well) and wild-type CdtC-His₆ (3.5 μg/well) were added individually to the thyroglobulin-coated wells (white bars). Other wells received 4 μg/well of wild-type CdtA-His₆ followed by 3.5 μg/well of the individual wild-type and cysteine mutant CdtC-His₆ proteins (gray bars). Other thyroglobulin-coated wells received 10 μg/well of wild-type or cysteine mutant CdtA-His₆ proteins (black bars). The plates were developed, and the absorbance was determined as described in the legend to Fig. 2. All samples were run in triplicate. Statistically significant changes in the binding of the CdtC-His₆ and CdtA-His₆ cysteine mutant proteins from that of the corresponding wild-type proteins are marked by asterisks (*, P < 0.0001; **, P < 0.005). Absorbance ratios based on the binding of wild-type CdtA-His₆ to thyroglobulin were calculated as described in Materials and Methods. Dotted lines mark the absorbance values of wild-type CdtA-His₆ (4 and 10 μg/well) plus CdtC-His₆ (3.5 μg/well) bound to thyroglobulin. Results are representative of three experiments.

The effect of the cysteine substitutions in CdtA-His₆ on binding to thyroglobulin was also determined. Three of four CdtA-His₆ cysteine mutants (mutAc149, mutAc178, and mutA65) showed a statistically significant reduction in binding to thyroglobulin (P < 0.0001 or P < 0.005) (Fig. 5B, black bars). Therefore, replacing the individual cysteines in CdtA-His₆ had various effects on the binding of CdtA to thyroglobulin. In fact, replacing one of the cysteines (C136) in the C136/C149 potential disulfide-forming pair had no effect on binding. If this putative disulfide was essential for keeping the protein in a conformation required for binding, then replacement of C136 should have significantly reduced or abolished binding activity as in the case of C149.

All four of the CdtC-His₆ and CdtA-His₆ cysteine mutants formed nondialyzable heterotrimer complexes with the corresponding wild-type subunit proteins in the differential dialysis assay (data not shown). The amount of each subunit protein remaining in each of the mutant-containing samples was quantitatively the same, after dialysis, as that in the sample containing all three wild-type proteins. These results clearly indicated that, although binding of individual subunits may be reduced by the replacement of individual cysteines in both CdtA-His₆ and CdtC-His₆, the effects are not great enough to destabilize the subunit proteins to the extent that heterotrimer complex formation is prevented. The fact that heterotrimers are formed supports the data showing that replacing at least one cysteine in each disulfide-forming pair in both CdtA and CdtC results in a partially to fully active holotoxin.

Effect of reducing agent on holotoxin and subunit binding.

To further examine the functional significance of the predicted disulfides in CdtA and CdtC, we examined the ability of the wild-type subunit proteins to form a biologically active holotoxin complex under reducing and nonreducing conditions. CHO cells were grown under continuously reduced conditions by supplementing the culture medium with DTT. The reducing agent had no effect on the growth of the CHO cells (Fig. 7A). Holotoxin composed of wild-type subunits was reconstituted in refolding buffers with and without the reducing agent and was then added to the corresponding cell cultures growing under reducing and nonreducing conditions. There was a minimal, statistically significant difference (P = 0.02) between the number of CFU in CHO cell cultures treated with reduced reconstituted holotoxin and that of unreduced reconstituted holotoxin (Fig. 7A). Formation of a heterotrimer complex made with wild-type subunit proteins was also not affected by the reducing agent, when examined by a differential dialysis assay (data not shown).

FIG. 7.

Effects of a reducing agent on wild-type holotoxin and CdtA-His₆ activities. (A) Purified recombinant wild-type CdtA-His₆, CdtB-His₆, and CdtC-His₆ were reconstituted in refolding buffer under reducing and nonreducing conditions as described in Materials and Methods. These preparations were then added to CHO cell cultures growing in medium with and without 10 mM DTT. Colonies were stained and counted after 6 days of growth and were expressed as numbers of CFU. Black bars, unreduced culture and unreduced heterotoxin; white bars, reduced culture and reduced heterotoxin. Statistical differences between the effects of the unreduced and reduced samples are shown as P values. red, reduced. (B) ELISA plates were coated with either reduced or nonreduced thyroglobulin (thyroglobulin). Reduced or unreduced wild-type CdtA-His₆ was then added to the wells in increasing concentrations. Bound CdtA-His₆ was detected as described in the legend to Fig. 2. Unfilled squares, thyroglobulin and reduced CdtA-His₆; filled squares, reduced thyroglobulin and reduced CdtA-His₆; open circles, thyroglobulin and CdtA-His₆; solid circles, reduced thyroglobulin and CdtA-His₆. All samples were run in triplicate a minimum of three times.

To determine the effects of the reducing agent on the subunit binding activity of wild-type CdtA-His₆, saturation kinetics were determined for all four combinations of reduced and unreduced subunit proteins and thyroglobulin (Fig. 7B). Equivalent amounts of unreduced CdtA bound to both reduced and unreduced thyroglobulin. Surprisingly, there was a statistically significant increase in binding to both reduced and unreduced thyroglobulin when CdtA-His₆ was in the reduced state. A P value of <0.00002 was obtained for the binding of reduced CdtA-His₆ relative to that of the unreduced control at a saturating concentration of 8 μg of CdtA-His₆/well.

DISCUSSION

The CdtC protein subunit is an essential component of Cdt. It has been shown by insertional inactivation, using a chloramphenicol resistance cassette, that expression of CdtC is required to produce an active toxin in bacterial culture supernatants (31, 49). The crystal structures of recombinant H. ducreyi (38) and A. actinomycetemcomitans (51) Cdt indicate that the CdtC subunit interacts with CdtA to form lectin-like binding domains similar to that of the B chain in ricin. To date, few molecular genetic studies have been conducted to identify amino acid residues in CdtC that have key roles in maintaining the stability and binding activities of the protein. Lee et al. (30) found that an in-frame deletion of a region conserved in CdtC and CdtA (amino acids 115 to 136 in the C. jejuni CdtC sequence) did not alter CdtC binding. Nesić et al. (38) showed that a large deletion of the amino-terminal region of CdtC (residues 21 to 39) destabilized the holotoxin complex and reduced but did not abolish cytotoxicity. Similar effects were obtained with a relatively large deletion (residues 179 to 186) at the carboxy-terminal end of the protein. A larger deletion (residues 169 to 178) apparently had a much greater destabilizing effect on holotoxin formation. That same group also tested the activity of a reconstituted holotoxin containing targeted mutations in CdtC (R34K and Q49A) and CdtA (P103A and Y106A). These mutated residues reside along the putative binding groove formed between the CdtA and CdtC subunits. The mutations did not appear to significantly affect holotoxin assembly but did drastically reduce cytotoxicity.

Using a random-mutagenesis approach, we obtained eight CdtC mutants with unique point mutations that produced subunit proteins which exhibited a statistically significant reduction in binding to CHO cells. Two of these mutations resided in the amino-terminal region of the protein, while the other six mutations were distributed over the middle portion of the sequence. However, consistent with the findings of Lee et al. (30), no mutations leading to a binding-deficient phenotype were obtained within the 21-amino-acid conserved region in CdtA/CdtC. In addition, we did not find amino acid substitutions in the carboxy-terminal end of the protein that exhibited a loss or reduction in binding. In an earlier study, we observed that the insertion of a kanamycin resistance cassette in the 3′ end of the cdtC gene did not alter the cytotoxic activity of holotoxin made with the truncated gene product (34). Only one of eight cell binding-deficient mutants (mutC216^F97C) lost the ability to bind to CdtA in the thyroglobulin ELISA. In contrast to McSweeney and Dreyfus (35), we found that CdtC-His₆ by itself binds very poorly to fucose-containing glycoproteins (Fig. 1C) (5). However, a CdtC-His₆/CdtA-His₆ complex binds specifically and stoichiometrically to thyroglobulin in an in vitro ELISA assay (Fig. 1B) (5). These data are similar to those of Deng and Hansen (11), who found that detectable levels of CdtC attached to HeLa cells in culture only when added as a complex with CdtA. Holotoxin made with CdtC-His₆ from mutC216 also failed to inhibit the proliferation of CHO cells and induce DNA damage in HeLa cells. The loss of these typical Cdt activities appeared to be due to the inability of this mutant protein to form a holotoxin complex with wild-type CdtA and CdtB.

Computer models of the crystal structure of Cdt (38, 51) and the application of the GDAP algorithm (this study) predict the presence of two intrachain disulfides in CdtC. These disulfides reside between the cysteine pairs C96/C107 and C135/C149 in the deduced amino acid sequence of A. actinomycetemcomitans CdtC. The location of the predicted disulfides led to the hypothesis that the replacement of phenylalanine with cysteine at position 97 in mutC216 may interfere with the formation of the disulfide that may be formed by residues C96 and C107. To assess the potential contribution of the predicted disulfides to CdtC activity, we replaced each of the four cysteines with alanine. All four cysteines are highly conserved in CdtC (20), and it is generally accepted that intrachain disulfides are important stabilizing features of other carbohydrate-binding lectins such as the B chain of ricin (32). Therefore, we expected that independently replacing either cysteine in a potential disulfide-forming pair would extensively destabilize the CdtC protein, thus significantly altering binding properties and the formation of a biologically active heterotrimer complex. We employed primarily a genetic approach to examine the importance of disulfides for CdtC activities because of the inherent difficulties in biochemically establishing the presence of intrachain disulfides. There are no empirical data to date to establish that CdtC intrachain disulfides are formed in the heterotoxin. Yamada et al. (51) erroneously reported that Saiki et al. (47) used nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis to show that CdtC and CdtA contain intrachain disulfides. Only interchain disulfides can be detected by the mobility shift of a protein in reducing and nonreducing gels. Our data clearly showed that if C96 or C149 was replaced, then the mutated cdtC gene product formed biologically inactive toxin. However, replacing either C107 or C135 resulted in proteins that formed active holotoxins. Even though there was no question that the predicted disulfides were eliminated due to genetic replacement of a half cysteine, the mutant holotoxins were still cytotoxic as shown by flow cytometry and inhibition of cell proliferation. The fact that the specific activities of the mutant holotoxins made with CdtC-His₆ from mutCc107 and mutCc135 were lower than those of the wild-type heterotrimer indicates that the stability of the protein is affected to some degree by either the loss of the particular cysteine or the loss of the disulfide. The presence of relatively small, slow-growing colonies in cell cultures treated with heterotoxins containing protein from mutCc107 and mutCc135 supports the conclusion that these mutated proteins are only partially destabilized. The finding that the cysteine replacement mutations had a range of effects on subunit binding and formation of active holotoxin complexes is not surprising since similar results were obtained with CdtC deletion mutants in other studies (39).

The lectin or receptor-binding component of the B chain of ricin is composed of homogeneous subunits due to dimerization of the heterodimer (32). In contrast, the cell-binding component of the Cdt is composed of heterogeneous subunits (CdtA and CdtC). It is interesting that one of the similarities between CdtA and CdtC is the structural arrangement of the cysteine pairs and, as a consequence, putative disulfides. Like CdtC, there are four highly conserved cysteines in CdtA (20) which are predicted to form two intrachain disulfides (Fig. 3C and 4B). Because of these structural similarities, we also mutated three (C136, C149, and C178) of the four potential disulfide-forming cysteines in CdtA. A CdtA mutant in which the fourth cysteine (C197) had been replaced with serine was obtained in an earlier study (mutA65 in Cao et al. [5]). Characterization of the CdtA cysteine replacements revealed mutant phenotypes very similar to those observed with the CdtC cysteine replacements.

Thus, our genetic data indicated that putative intrachain disulfides may help stabilize both the CdtC and CdtA subunits but are not essential for the formation of an active heterotoxin, albeit with reduced specific activity. Specific individual cysteine residues may be required for proper folding of the CdtC and CdtA proteins in lieu of the formation of disulfide bonds. This could explain our finding that the activities of the proteins were not totally destroyed when both of the half cysteines in a disulfide-forming pair were separately eliminated. At least in the case of ricin, one of the most well-studied A-B toxins, disulfides appear to have a more active role in maintaining a cytotoxic complex. The strongest evidence is for an interchain disulfide that links the A and B chains (32). It has been suggested that four intrachain disulfides are required for the galactose-binding activity of the B chain. However, the experimental data to support this conclusion do not appear to be as strong as those for the interchain disulfide. Since there are no cysteines in the A. actinomycetemcomitans CdtB, there are no interchain disulfides between CdtB and either of the other two subunit proteins. In addition, the cysteines in CdtC are not close enough to those in CdtA to form interchain disulfides between these two subunit proteins (Fig. 4B). We also examined the effects of reducing agents on the ability of the wild-type subunit proteins to form a biologically active heterotrimer and on the ability of wild-type CdtA to bind to thyroglobulin (experimental fucose-containing glycoprotein-like receptor) in an attempt to support the genetic data. We expected that the disulfides would be disrupted by maintaining the heterotrimer or specific subunit under reducing conditions, since at least one of each cysteine in both disulfide pairs in both CdtC and CdtA are exposed on the toxin surface (Fig. 4C and D). In support of the genetic data, wild-type Cdt subunits formed a biologically active heterotrimer complex under both reducing and nonreducing conditions. Curiously, there was a statistically significant increase in the binding kinetics of wild-type CdtA-His₆ to thyroglobulin when the subunit protein was in the reduced state. These results are opposite to those that would be expected if disulfides were required to stabilize the protein for holotoxin assembly and binding to receptor. The most likely explanation for these results is that the formation of disulfides makes only a moderate contribution to the stability of CdtA and CdtC. This may make sense at least for the A. actinomycetemcomitans Cdt. This periodontal pathogen colonizes and presumably secretes the cdt gene products in periodontal pockets. Gingival sulci in healthy subjects and healthy sites in diseased subjects have reported mean oxidation-reduction (E_h) potentials of +74 ± 7 and +73 ± 12 mV, respectively (26). The mean reported E_h in periodontal pockets is −48 ± 16 mV (P < 0.001). Assembly and target cell binding of the A. actinomycetemcomitans Cdt would have to take place in this reducing environment.