Expression and purification of Shiga-like toxin II B subunits

INFECTION AND IMMUNITY, Jan. 1995, p. 301–308 0019-9567/95/$04.0010 Copyright q 1995, American Society for Microbiology Vol. 63, No. 1 Expression and Purification of Shiga-Like Toxin II B Subunits DAVID W. K. ACHESON,1* SABINE A. DE BREUCKER,1 MARY JACEWICZ,1 LISA L. LINCICOME,2 ARTHUR DONOHUE-ROLFE,1 ANNE V. KANE,2 AND GERALD T. KEUSCH1 Division of Geographic Medicine and Infectious Diseases1 and Center for Gastroenterology Research in Absorptive and Secretory Processes,2 New England Medical Center, Boston, Massachusetts 02111 Received 1 August 1994/Returned for modification 9 September 1994/Accepted 12 October 1994 We are interested in purifying large amounts of SLT B subunits both for further characterization and to investigate their use as a vaccine. Recombinant SLT-I B subunit has been purified (1, 21) and shown to be expressed in a multimeric form. We have cloned the gene for the SLT-I B subunit (sltI B) into a variety of hyperexpression vectors under the control of various promoters (1). We obtained high levels of SLT-I B subunit expression in E. coli by using a trc promoter and found that the levels of expression from this construct could be increased approximately 10-fold by transferring the plasmid to Vibrio cholerae O395-N1 (1). We further demonstrated that the recombinant SLT-I B subunit retained the properties of the authentic B subunit from holotoxin and was capable of stimulating in rabbits neutralizing antibodies which appear to be protective following oral challenge with SLT-I-producing RDEC strains (16, 17, 25). At present there is no evidence that antibodies to SLTs are protective in humans. However, in the long term it may be possible to incorporate appropriate portions of both SLT-I and SLT-II in a multicomponent vaccine designed to prevent infection with toxin-producing organisms and/or to prevent the local and systemic consequences of the toxins. Despite the observation that all members of the SLT family have been associated with diseases, there is evidence from both North America (19) and England (24) that SLT-II may be responsible for more clinical disease than SLT-I. We investigated a variety of systems with different vectors and host strains containing the whole SLT-II B subunit gene, containing portions thereof, and containing the gene as maltose-binding protein (MBP) fusion in order to hyperexpress the SLT-II B subunit. It is clear from our results that development of SLT-II B subunit expression vectors which are capable of producing high levels of readily purifiable SLT-II B is not as straightforward as development of such SLT-I B subunit expression vectors. The Shiga family of toxins is a group of closely related cytotoxins. The prototype toxin, originally described 90 years ago (6), is produced by Shigella dysenteriae type 1 strains and is termed Shiga toxin. The Shiga-like toxins (SLTs) (also known as verotoxins) are closely related to Shiga toxin and are produced by enterohemorrhagic Escherichia coli (2, 13). Enterohemorrhagic E. coli and SLTs have been epidemiologically linked with a wide variety of gastrointestinal (bloody and nonbloody diarrhea) and systemic complications, the most severe of which is hemolytic uremic syndrome (13). However, the precise way in which these toxins cause disease is unknown. SLTs can be divided into two groups: SLT-I and SLT-II. SLT-I is different from Shiga toxin by three nucleotides but only one amino acid and is neutralized by antiserum raised to Shiga toxin. In contrast, comparison of the nucleotide sequences of the A and B subunits of SLT-I and -II showed 57 and 60% homology, respectively, with 55 and 57% amino acid homology, respectively (12). Despite this degree of homology, SLT-I and -II are reported to be immunologically distinct when reacted with polyclonal antiserum. However, at least one monoclonal antibody has been isolated that reacts with the B subunits of both SLT-I and -II and is capable of neutralizing both toxins (7). All SLTs share important characteristics: they are composed of a single enzymatically active A subunit and multiple B subunits; the A subunit acts as an N-glycosidase on a specific adenine residue in the 28S rRNA of the 60S ribosomal subunit (9), with the result that protein synthesis is inhibited; and the B subunits form pentamers (26), in association with a single A subunit, and are responsible for the binding of toxin to a neutral glycolipid receptor (11, 23, 27). * Corresponding author. Phone: (617) 636-7001. Fax: (617) 6365292. 301 Downloaded from http://iai.asm.org/ on July 24, 2020 by guest Shiga-like toxins (SLTs), which are produced by certain strains of Escherichia coli, are composed of enzymatically active A and B subunit multimers responsible for the toxin’s binding. We have previously purified large amounts of the SLT-I B subunit by using a hyperexpression vector in Vibrio cholerae under the control of the trc promoter. In this study we examined various expression vectors to maximize yields of the SLT-II B subunit. The SLT-II B subunit has been expressed by using both the T7 promoter and the tac promoter in E. coli. When expressed from a plasmid containing the structural gene for SLT-II B deleted of the leader sequence, SLT-II B was able to form multimers when cross-linked, although SLT-II B production from this plasmid was unreproducible. SLT-II B expressed in all three systems appeared to form unstable multimers, which did not readily bind to a monoclonal antibody which preferentially recognizes B subunit multimers. SLT-II B expression was not increased by moving any of the plasmids into V. cholerae. Polyclonal antibodies raised to SLT-II B in rabbits recognized B subunit in SLT-II holotoxin yet were poorly neutralizing. SLT-II B was also expressed as a fusion protein with maltose-binding protein and could be cleaved from maltose-binding protein with factor Xa. Although the expression vectors were able to make large amounts of SLT-II B, as determined by Western blotting (immunoblotting), the levels of purified SLT-II B subunit were low compared with those obtained previously for SLT-I B subunit, probably because of instability of the multimeric SLT-II B subunit. 302 ACHESON ET AL. INFECT. IMMUN. TABLE 1. Bacterial strains and plasmids used in this studya Bacterial strain or plasmid Relevant characteristic(s) Bacterial strains E. coli JM105 E. coli JM107 pcnB E. coli BL21 (DE3) V. cholerae CVD103HgR E. coli C600 (933W) E. coli SR2 pET9 pCG807f pLL14 pDA63 pDA76 pDA78 a Ampr Ampr Ampr; contains a tac promoter Kanr; contains a T7 promoter Ampr; malE gene under lac promoter Kanr; sltII B under T7 promoter Ampr; genetic fusion between malE and sltII B, lac promoter Ampr; sltII B without leader sequence under tac promoter Ampr; whole sltII B under tac promoter Pharmacia LKB 15 Novagen J. Kaper 18 Laboratory stock Stratagene Laboratory stock Pharmacia-LKB Novagen New England BioLabs This study This study This study This study See Fig. 1 for a description of pLL10 to pLL13. MATERIALS AND METHODS Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are shown in Table 1. V. cholerae CVD103HgR was provided by James Kaper, University of Maryland, Baltimore. Bacteria were grown in Luria-Bertani medium containing appropriate antibiotics (ampicillin at 100 mg/ml or kanamycin at 25 mg/ml). Plasmid construction. Standard techniques were used for construction of plasmids (22). Restriction enzymes were obtained from New England BioLabs (Beverly, Mass.). Transformations into E. coli were done with standard CaCl2competent cells. Transformations into V. cholerae CVD103HgR were performed by electroporation. V. cholerae CVD103HgR cells were made electrocompetent as follows. V. cholerae CVD103HgR (1-liter culture) was grown to an optical FIG. 1. Construction of plasmids pLL10 to pLL14. The heavy line represents DNA encoding the SLT-II B subunit gene. The vector used was either pBluescript KS2 (KS2), pUC19, or pET9. Only pLL14 was constructed for expression with the T7 promoter. See the text for further details. S, SacI; B, BamHI; P, PstI; K, KpnI; E, EcoRI. The figure is not drawn to scale. Downloaded from http://iai.asm.org/ on July 24, 2020 by guest Plasmids pBluescript KS pUC19 pKK223-3 Reduces plasmid copy number T7 RNA polymerase under lacUV5 promoter ctxA deletion vaccine strain SLT-II containing bacteriophage lysogen Used for MBP–SLT-II B expression Source or reference density at 600 nm (OD600) of 0.7, and the cells were then pelleted by centrifugation. The cells were washed several times with diminishing volumes of 2 mM CaCl2 in 10% glycerol and finally suspended in 40 ml of 2 mM CaCl2 in 10% glycerol. Aliquots of the competent cells were stored at 2708C until required. Electroporation was performed with a Bio-Rad electroporator (2.5 kV, 25 A, 200V). The gene encoding the SLT-II B subunit (sltII B) was cloned from bacteriophage 933W isolated from E. coli C600(933W) (18) by standard techniques (22). Bacteriophage DNA was restriction digested with PstI, and the fragment containing sltII B (approximately 2.5 kb) was determined by Southern blotting with a [32P]-ATP-end-labeled oligonucleotide probe (GCTAAAGGTAAAATTG AG) directed toward sltII B. The PstI fragment was cloned into the PstI site of pBluescript KS to create pLL10 (Fig. 1). pLL10, however, turned out to contain two different 2.5-kb inserts. Therefore, pLL14 was constructed via three other intermediate plasmids (Fig. 1). First, plasmid pLL11 was created by deleting a 2-kb KpnI fragment from the insert in pLL10. Second, the unwanted 2.5-kb PstI fragment was deleted to create pLL12, which contained a 500-bp PstI-KpnI fragment in the pBluescript KS vector encoding the 15 carboxy-terminal amino acids of the SLT-II A subunit, the whole B subunit, and some downstream DNA. Third, a SacI-KpnI digest was made of pLL12, and the 500-bp fragment was inserted into the SacI and KpnI sites of pUC19 to provide BamHI sites on either side of the gene for SLT-II B (pLL 13). Finally, to construct pLL14, a BamHI fragment (approximately 500 bp) was removed from pLL13 and inserted into the BamHI site of pET9 (Novagen), which placed sltII B under the control of a T7 promoter. pLL14 was transformed into E. coli BL21, which contains an IPTG (isopropyl-b-D-thiogalactopyranoside)-inducible T7 RNA polymerase gene under the control of a lacUV5 promoter. pDA78 was constructed by removing an EcoRI-HindIII fragment of approximately 500 bp (encoding the 15 carboxy-terminal amino acids of the SLT-II A subunit, the whole B subunit, and some downstream DNA) from pLL13 (Fig. 1) and ligating it to the EcoRI and HindIII sites of pKK223-3. This placed sltII B under the control of a tac promoter. We were not able to make transformants of JM107 with pDA78 and therefore made transformants of JM107pcnB (a mutant of JM107 which results in a lower plasmid copy number [15]) with pDA78. pDA76 was constructed by PCR. An upstream primer (P1) (59 GCGAAT TCATGGCGGATTGTGCTAAAGG 39) was designed to anneal to the first base pair of the structural gene of the SLT-II B subunit with an ATG just 59 to the annealing site and an upstream EcoRI site. A downstream primer (P2) (59 GCGCGCAAGCTTGGGGGATTCACCATGCG 39) annealed to the DNA downstream of the sltII B stop codon and contained a HindIII site. PCR with P1 and P2 resulted in a 331-bp fragment as expected. The fragment was purified by low-melting-point agarose gel electrophoresis and after restriction with EcoRI and HindIII was ligated into the EcoRI and HindIII sites of pKK223-3, which placed the structural SLT-II B subunit gene under the control of a tac promoter. pDA76 was transferred into E. coli BL21. pDA63, containing a genetic fusion between the gene for maltose-binding protein (malE) and the structural gene for the SLT-II B subunit, was constructed by PCR. sltII B was amplified by using an upstream primer (P3) (59 GCGCG GTCTCGGCGGATTGTGCTAAAGG 39) which annealed to the DNA encoding the N terminus of the structural SLT-II B subunit gene and contained an upstream BsaI restriction site. The downstream primer used was primer P2 (see above). PCR with P2 and P3 resulted in a 333-bp fragment, as expected, which was purified with a low-melting-point agarose gel. The PCR fragment was then restricted with BsaI, and the 4-bp overhanging portion was filled in by using T4 DNA polymerase. The fragment was then cut with HindIII. The filled BsaI site and HindIII site were then ligated to the StuI and HindIII sites of pCG807fx, respectively. This resulted in an in-frame gene fusion between the 39 end of the malE gene and the 59 end of sltII B. VOL. 63, 1995 SLT-II B SUBUNIT EXPRESSION SYSTEMS 303 TABLE 2. SLT-II B subunit expression in small-scale culturesa SLT-II B (ng/ml) in: OD600 Time (h) after IPTG addition 0 1 2 4 20 Culture supernatant Polymyxin B extract With IPTG Without IPTG With IPTG Without IPTG With IPTG Without IPTG 0.56 0.61 0.48 0.60 5.33 0.58 1.35 1.48 2.51 3.78 3.2 11.5 22.5 29.8 ,0.3 3.3 3.7 4.6 19.5 2,300 0.4 6.8 2.7 ,0.3 ,0.3 0.4 1.2 8.9 81.0 295.0 a Cultures of E. coli BL21(pLL14) were grown to an OD600 of approximately 0.6, and IPTG (1 mM final concentration) was added to half of the cultures. The SLT-II B subunit in both the culture supernatant and polymyxin B cell extract was then quantitated by ELISA at various time points. Each value represents the mean from two experiments. the lanes were scanned with an LKB Ultroscan XL Enhanced Laser Densitometer. The SLT-II B subunit peaks were integrated by using Beckman system Gold software (Beckman Instruments, Fullerton, Calif.). Purification of MBP–SLT-II B fusion protein. MBP fusion proteins were purified from 500-ml cultures of E. coli SR2(pDA63). A 1% inoculum from a stationary-phase culture of E. coli SR2(pDA63) was made in 500 ml of LuriaBertani medium containing 100 mg of ampicillin per ml. The culture was induced with IPTG (0.1 mM final concentration) when the OD600 was 0.600 and was then grown for a further 3 h at 378C. After centrifugation, the bacterial pellet was redissolved in 20 ml of PBS (pH 7.4) and sonicated to approximately 80% clarity. The sonicate was centrifuged, and the supernatant was stored at 2208C until affinity chromatography was performed. The MBP–SLT-II B fusion protein was purified by amylose column affinity chromatography as described in the New England BioLabs protocol. Fractions containing MBP–SLT-II B fusion protein were determined by immunoblotting in the following way. Five microliters of each fraction was spotted onto a nitrocellulose filter, and MBP–SLT-II B fusion protein was detected by using a polyclonal antibody directed toward the SLT-II B subunit. The positive fractions were pooled. The MBP–SLT-II B fusion protein was cleaved by overnight digestion with factor Xa at 378C according to the New England BioLabs protocol. The MBP–SLT-II B fusion protein was characterized in terms of its ability to inhibit the cytotoxicity of SLT-II and its recognition by a number of antibodies (monoclonal and polyclonal) to both the SLT-II B subunit and SLT-II in Western blots. Interaction of SLT-II B and MBP–SLT-II B fusion proteins with HeLa cells. The binding of SLT-II B was tested in a number of ways. Initially both SLT-II and the SLT-II B subunit were labeled with iodine-125 by using a modification of the chloramine T procedure as previously described (8). The specific activities were 23,094 and 14,338 cpm/ng of protein for SLT-II and the SLT-II B subunit, respectively. The binding of the two labeled proteins to HeLa 229 cells was then compared as previously described (5). Competitive inhibition of 125I-SLT-II B with noniodinated SLT-II B was examined by adding increasing concentrations of noniodinated SLT-II B to a fixed amount (approximately 90,000 cpm) of 125 I-SLT-II B and determining the binding of radioactivity to HeLa cells. The ability of the SLT-II B subunit to inhibit SLT-II cytotoxicity in HeLa cells was measured by incubating HeLa cells with various concentrations of SLT-II in the presence and absence of the SLT-II B subunit (10 mg/ml). HeLa cell cytotoxicity was measured by inhibition of [3H]leucine incorporation (7). Purified MBP–SLT-II B fusion protein was tested for its ability to inhibit SLT-II cytotoxicity by preincubating HeLa cells with MBP–SLT-II B protein at 378C for 1 h and then adding SLT-II. After overnight incubation, the level of cytotoxicity was determined by measuring [3H]leucine incorporation (7). Raising immune serum to SLT-II B in rabbits. SLT-II B subunit purified from E. coli BL21(pLL14) was used to immunize a New Zealand White rabbit. The RIBI Adjuvant System (RIBI Immuno Chem Research, Inc., Hamilton, Mont.) was used according to the manufacturer’s protocol. One hundred micrograms of the SLT-II B subunit was used for the first and second immunizations, and the rabbit was bled 10 days later. The serum was tested in Western blots for reactivity with SLT-II and the SLT-II B subunit and in a cytotoxicity assay for its ability to neutralize SLT-II. The cytotoxicity assay was performed by incubating three dilutions of serum (1:100, 1:500, and 1:1,000) with various concentrations of SLT-II for 1 h at room temperature. HeLa cells were then exposed to the toxin-serum mixture overnight, after which [3H]leucine incorporation was determined. The immunized rabbit was then given a final injection of approximately 100 mg of cross-linked SLT-II B subunit prepared by using dimethyl suberimidate and was rebled 10 days later. The final serum was tested by Western blotting and neutralization assays as described above. Downloaded from http://iai.asm.org/ on July 24, 2020 by guest Production and purification of SLT-II B subunit. SLT-II B subunit expression from E. coli BL21(pLL14) was first examined in small-scale cultures. A 1% inoculum was made from an overnight culture into Luria-Bertani medium containing kanamycin (25 mg/ml). When the OD600 was 0.6, the cultures were divided and IPTG was added to one half of the cultures. Both cultures were then grown, and samples were taken at 1, 2, and 4 h and at 20 h (overnight) for determination of SLT-II B expression in both the culture supernatant and a polymyxin B extract (1) of the cell pellet. E. coli(pDA76) and E. coli(pDA78) were also grown in small-scale cultures, and the level of SLT-II B expression was determined at various time points. Following small-scale cultures, E. coli BL21(pLL14) and E. coli BL21(pDA76) were grown in a 9.9-liter fermentor culture (fermentor model MF128S; New Brunswick Scientific Co., Inc., Edison, N.J.). The SLT-II B subunit was harvested from the culture supernatant of an overnight culture of E. coli BL21(pLL14) and from the cell pellet of a late-log-phase culture of E. coli BL21(pDA76). Following the culture, bacteria were concentrated by Pellicon tangential-flow membrane filtration (0.45-mm-pore-size Durapore filter; Millipore Corp., Bedford Mass.). The supernatant from E. coli BL21(pLL14) was concentrated by passage over a 10-kDa-cut-off ultrafiltration membrane in a second Pellicon apparatus. The concentrate was washed with 1.5 liters of 10 mM Tris HCl (pH 7.5) and then treated with ammonium sulfate to 70% saturation. After overnight incubation at 48C, the precipitate was collected by centrifugation at 13,200 3 g for 20 min, resuspended in 10 ml of 10 mM phosphate-buffered saline (PBS) (pH 7.5), and dialyzed with three changes of 4 liters of the same buffer over 24 h. The pellet from the overnight cultures of E. coli BL21(pDA76) was lysed with a French press, and a supernatant was prepared by centrifugation at 39,000 3 g for 20 min. The resulting supernatant was then treated with ammonium sulfate to 70% saturation and processed as outlined above. SLT-II B from both E. coli BL21(pLL14) and E. coli BL21(pDA76) was purified by using a hydatid cyst material-Sepharose 4B (HCM-S4B) column as previously described (1, 7). The dialyzed, lyophilized eluate from the HCM-S4B column from E. coli BL21(pLL14) contained a contaminating protein and was further purified by anion-exchange chromatography with a Pharmacia/LKB fast protein liquid chromatography (FPLC) system. An aliquot of lyophilized HCMS4B eluate was reconstituted in 20 mM piperazine HCl (pH 5.5) and loaded onto a Mono-Q column equilibrated with the same buffer. After a wash with 52.5 mM NaCl in piperazine buffer to remove contaminating proteins, the B subunit was eluted as a single peak by increasing the NaCl concentration from 52.5 to 157.5 mM in a linear gradient. Biological activity of SLT-II B subunit. The purified SLT-II B subunit was characterized with the following assays: (i) cytotoxicity on tissue culture cells (14); (ii) binding of iodinated SLT-II B subunit; (iii) inhibition of cytotoxicity of native SLT-II B holotoxin to tissue culture cells; and (iv) covalent cross-linking using the homobifunctional cross-linking agent dimethyl suberimidate (Pierce Co.) (1). Detection of SLT-II B subunit. Recombinant SLT-II B was detected by enzyme-linked immunosorbent assay (ELISA) and by Western blotting (immunoblotting). The ELISA used was as previously described (3). Briefly, a monoclonal antibody (4D1) directed toward the B subunit of SLT-II (7) was used as the capture molecule. Bound SLT-II B was detected with a rabbit polyclonal antibody raised against the native SLT-II holotoxin, followed by an antirabbit alkaline phosphatase-conjugated antibody and substrate. SLT-II holotoxin was used as a standard in the ELISAs. Western blots were done with sodium dodecyl sulfate–15% polyacrylamide gel electrophoresis (SDS–15% PAGE) slab gels. Following transfers, filters were blocked with 5% nonfat milk and then incubated with polyclonal antibodies raised in rabbits to SLT-II (or SLT-II B), followed by an appropriate anti-rabbit alkaline phosphatase-conjugated antibody and then substrate. The SLT-II B subunit was quantitated on Western blots by comparing fermentor samples from cultures of E. coli BL21(pLL14) and E. coli BL21(pDA76) with known amounts of SLT-II B or SLT-II (holotoxin), the latter being approximately 50% B subunits by weight. Following development of the Western blots, 304 ACHESON ET AL. FIG. 2. SDS–15% PAGE gel stained with Coomassie blue. The gel shows SLT-II B purified from E. coli BL21(pLL14) and SLT-II holotoxin (containing both A and B subunits). Recombinant SLT-I B subunit (1) and SLT-I holotoxin are shown for comparison. MW, molecular weight (in thousands). INFECT. IMMUN. RESULTS Expression of SLT-II B subunit. Initially SLT-II B subunit expression from E. coli BL21(pLL14) was determined in smallscale cultures (Table 2). The results from this experiment demonstrated that although IPTG had an early positive effect on SLT-II B subunit production, for maximal yields it was better to grow the cultures overnight without the addition of IPTG and to harvest the culture supernatant (Table 2). The high levels obtained in culture supernatants at 20 h are probably due to bacterial cell lysis during stationary-phase growth. For the large-scale cultures, a 9.9-liter fermentor culture of E. coli BL21(pLL14) was made, the cell pellet was harvested, and the SLT-II B subunit was purified as described in Materials and Methods. After the initial purification with the HCMS4B affinity column, there was a minor contaminant (,5%) seen on Coomassie blue-stained gels of the affinity column eluate. Therefore, a second purification step, using FPLC as described in Materials and Methods, was added. We obtained approximately 1 mg of pure SLT-II B subunit from the 9.9-liter fermentor culture of E. coli BL21(pLL14) (Fig. 2). We were disappointed with the levels of SLT-II B production from pLL14 and believed that it should be possible to purify SLT-II B from 9.9-liter fermentor cultures in amounts similar to those we had previously obtained for SLT-I B. We therefore investigated the utility of the tac promoter to produce SLT-II B, simply because it was a different system and might behave differently. Given the difficulty in constructing a plasmid which expressed the SLT-II B subunit under the control of the tac promoter, we speculated that the expressed protein was toxic. We therefore adapted two strategies to obtain an alternative expression system. The first was to use a host strain which carried a reduced plasmid copy number (E. coli JM107pcnB) (15). The second was to construct a new plasmid (pDA76) in which the SLT-II B subunit was expressed without its leader sequence. We were able to adequately express SLT-II B, as determined by Western blotting and ELISA, by reducing the copy number of the plasmid pDA78 (which contained the whole of sltII B under control of a tac promoter). Maximal expression was seen in overnight cultures. The amount of SLT-II B detected in the ELISA was approximately 2.5 to 10 ng/ml of culture in both the polymyxin B extract and the sonicated cell suspension. By using the SLT-II B subunit purified from E. coli BL21(pLL14) as a standard in the Western blot, we estimated that polymyxin B extracts from overnight cultures of E. coli JM107pcnB(pDA78) were producing 25 to 50 mg of SLT-II B subunit per ml of culture, which was approximately 1,000-fold more than that measured by ELISA. Maximal expression was seen in over- night cultures. The same polyclonal antibody directed toward SLT-II B was used in both the ELISA and the Western blot. E. coli BL21(pDA76) produced SLT-II B readily detectable by both ELISA and Western blotting in the cellular sonicates. In contrast to B subunit expression from pLL14 and pDA78, maximal expression was seen in early stationary phase and the levels of the B subunit did not increase in overnight culture. As with pDA78, there appeared to be significantly more B subunit detected by Western blotting than by ELISA. In 9.9-liter cultures of E. coli BL21(pDA76), the SLT-II B subunit was readily detectable in the culture lysate by Western blotting. However, when the lysate was passed over an HCM column and pure SLT-II B was eluted from the affinity column, the yields of SLT-II B were highly variable. The maximum amount of SLT-II B obtained from any one 9.9-liter fermentor batch was approximately 1 mg. The SLT-II B subunits obtained from pDA76 and pLL14 both migrated on SDS–15% PAGE gels in the same way as the B subunit from SLT-II holotoxin. When purified SLT-II B from pDA76 was cross-linked by using dimethyl suberimidate, multimers of at least two to five B subunits were present, as determined by Western blotting with polyclonal antibodies raised to SLT-II B (data not shown). Transformation of pLL14, pDA76, and pDA78 into V. cholerae CVD103HgR was undertaken by electroporation, and the presence of the gene for SLT-II B was confirmed by PCR. We failed to obtain levels of SLT-II B expression measurable by ELISA in any of the V. cholerae transformants. However, V. cholerae(pDA76) did produce enough SLT-II B to be seen by Western blotting in small-scale cultures. SLT-II B was not FIG. 4. Purified MBP–SLT-II B fusion protein run on an SDS–15% PAGE gel and stained with Coomassie blue. MBP–SLT-II B is shown with (lane A) and without (lane B) overnight incubation (378C) with factor Xa. An MBP band can be seen in lane A just below the MBP–SLT-II B fusion protein band. Pure SLT-II B is shown in lane C for comparison. The lower band in all lanes is due to staining of the dye front. Downloaded from http://iai.asm.org/ on July 24, 2020 by guest FIG. 3. Overnight cultures of E. coli(pDA63) grown with (lane B) and without (lane C) IPTG induction. Approximately equal amounts of the sonicated bacterial cells were run on an SDS–15% PAGE gel and examined by Western blotting with a rabbit polyclonal antibody to SLT-II. Lane A, SLT-II holotoxin. Numbers are molecular weights in thousands. VOL. 63, 1995 SLT-II B SUBUNIT EXPRESSION SYSTEMS 305 detected in Western blots with either the pLL14 or pDA78 plasmid in small-scale cultures. In an attempt to quantitate the SLT-II B subunit detected by 4D1 ELISA compared with that detected by Western blots more accurately, the materials loaded onto the HCM column from 9.9-liter fermentor cultures of both E. coli BL21(pDA76) and E. coli BL21(pLL14) were compared with known standards. Western blots were made of dilutions (1/50, 1/100, and 1/200) of E. coli BL21(pLL14) culture supernatant preparation and of the same dilutions of E. coli BL21(pDA76) cell lysate from 9.9-liter fermentor cultures prior to loading onto the HCM affinity column. SLT-II (2.5 mg) was loaded as a standard. The areas under the B subunit curves following integration were 0.273, 0.187, and 0.085 for the three pLL14 dilutions and 0.268, 0.150, and 0.053 for the three pDA76 dilutions, respectively. A value of 0.446 was obtained for the SLT-II B standard. By 4D1 ELISA we obtained 44.3 mg/ml from the concentrate of the E. coli BL21(pLL14) culture. We calculated that approximately 4,600 mg of SLT-II B per ml of concentrate was detected by Western blotting with the pLL14 preparation. We were unable to detect any SLT-II B subunit in the 4D1 ELISA with the E. coli BL21(pDA76) lysate; however, according to the Western blot, there was 3,640 mg/ml of concentrate. Expression of MBP–SLT-II B fusion protein. Following the construction of pDA63, the plasmid was inserted into E. coli SR2 and the expression of MBP–SLT-II B was investigated. Cultures were grown to an OD600 of 0.5 and then divided. IPTG (to a final concentration of 0.1 mM) was added to one half, and aliquots were removed at 1, 3, and 16 h after addition of IPTG. The cell pellets from the aliquots were concentrated 10-fold, sonicated, and examined by Western blotting with a polyclonal antibody directed toward SLT-II holotoxin (Fig. 3). A 48-kDa reactive band was present, corresponding to the expected size of the MBP–SLT II B fusion protein. There was also a clear inducing effect of the IPTG (Fig. 3). Following purification over amylose, we found a predominant band corresponding to the MBP–SLT-II B fusion protein (Fig. 4, lane B), and 6.5 mg of MBP–SLT-II B fusion protein was purified from the 1-liter culture. When this material was digested with factor Xa (overnight at 378C), the SLT-II B subunit was cleaved from the fusion protein and the amount of MBP seen in the gel increased (Fig. 4, lane A). Both the MBP–SLT-II B fusion protein and the cleaved SLT-II B subunit reacted with polyclonal antibodies directed toward the B subunit in Western blots. MBP–SLT-II B fusion protein was also recognized by polyclonal antibodies raised to SLT-II (holotoxin) and the monoclonal antibody 4D1 (data not shown). Interaction of SLT-II B and MBP–SLT-II B fusion proteins with HeLa cells. Iodinated SLT-II B failed to bind significantly to HeLa cells compared with SLT-II (Fig. 5). Despite this lack of binding of iodinated SLT-II to HeLa cells, it was still able to inhibit the cytotoxicity of holotoxin on HeLa cells when the SLT-II B subunit was added in excess (approximately 106-fold) over SLT-II (Fig. 6). In studies of competitive inhibition of binding of iodinated SLT-II B to HeLa cells by noniodinated SLT-II B, low doses (25:1) of cold SLT-II B actually increased the binding of the iodinated protein, while higher doses inhibited 125I-SLT-II B binding (Fig. 7). Despite the increased binding, the total number of counts per minute bound was still low (approximately 0.5% of the input counts per minute). MBP–SLT-II B was not cytotoxic to HeLa cells and was unable to block the cytotoxity of SLT-II to HeLa cells even when the HeLa cells were preincubated with high concentrations (100 mg/ml) of MBP–SLT-II B for 1 h at 378C prior to the addition of SLT-II. Characterization of immune serum raised to SLT-II B sub- Downloaded from http://iai.asm.org/ on July 24, 2020 by guest FIG. 5. Binding of iodinated SLT-II (open circles) and SLT-II B subunit (closed circles) to HeLa cells determined by adding various amounts of iodinated protein to the cells and measuring the counts per minute which bound following multiple washes. FIG. 6. Different dilutions of either SLT-II alone (open circles) or SLT-II mixed with SLT-II B (10 mg/ml) (closed circles) were added to HeLa cells in order to determine whether SLT-II B was able to inhibit SLT-II cytotoxicity. Inhibition of protein synthesis in HeLa cells was then determined by measuring incorporation of tritiated leucine. The results are expressed as percent leucine incorporation compared with that in control wells not exposed to SLT-II or SLT-II B. Error bars indicate standard deviations from triplicate experiments. 306 ACHESON ET AL. INFECT. IMMUN. TABLE 3. Neutralization of SLT-II cytotoxicity by rabbit seraa Rabbit serum unit. Following the second immunization with SLT-II B, the rabbit immune serum clearly recognized the B subunit from both SLT-II holotoxin and recombinant SLT-II B by Western blotting (Fig. 8). Both SLT-II B antiserum and SLT-II antiserum recognized SLT-II B (Fig. 8). Despite the strength of reaction seen in the Western blot, the anti-SLT-II B immune serum was not able to significantly neutralize the cytotoxicity of SLT-II, even at a dilution of 1:100 (Table 3). Following the third immunization with cross-linked SLT-II B subunit, the immune serum was still unable to neutralize SLT-II cytotoxicity (anti-SLT-II B [second bleed] in Table 3). In contrast, the immune serum raised against SLT-II holotoxin was clearly capable of neutralizing SLT-II at a dilution of 1:1,000. In subsequent experiments with very low concentrations of SLT-II (1 to 10 pg/ml), we were able to show that anti-SLT-II B antiserum was able to neutralize SLT-II. However, its neutralizing capacity was much less than that of SLT-II antiserum, FIG. 8. Purified SLT-II (lane A) and SLT-II B (lane B) examined by Western blotting with rabbit polyclonal antiserum raised to SLT-II B (1:1,000 dilution). For comparison SLT-II (lane C) and SLT-II B (lane D) were blotted with polyclonal antiserum raised to SLT-II (1:2,000 dilution). 10 1 0.1 Normal 1:100 1:500 1:1,000 1.3 1.2 0.8 2.0 1.4 1.8 3.2 3.7 4.5 Anti-SLT-II B (first bleed) 1:100 1:500 1:1,000 1.4 0.8 0.7 1.7 1.4 1.3 3.9 3.3 3.3 Anti-SLT-II B (second bleed) 1:100 1:500 1:1,000 1.3 0.8 0.6 2.6 1.7 1.5 7.0 3.4 3.6 Anti-SLT-II 1:100 1:500 1:1,000 63.2 52.8 37.9 99.0 89.1 89.0 86.9 84.7 88.5 a SLT-II at three concentrations was preincubated with three dilutions of rabbit sera from normal animals, animals immunized with the SLT-II B subunit, or animals immunized with SLT-II toxoid. The toxin-serum mixture was incubated with HeLa cells overnight, following which [3H]leucine incorporation was determined. b Compared with control (non-toxin-treated) wells. High values represent low levels of cytotoxin. despite the strong reaction to the SLT-II B subunit in Western blots. DISCUSSION During the course of this study, we have examined a variety of systems for the expression and purification of SLT-II B subunits. We have successfully purified SLT-II B subunits both with and without a leader sequence as well as a fusion protein comprising MBP and the structural protein of the SLT-II B subunit. We have previously expressed and purified to homogeneity large amounts of the SLT-I B subunit by using similar vectors and host strains (1). However, while these systems express moderately large amounts of the SLT-II B subunit, we have not been able to purify the SLT-II B subunit as readily as we purified the SLT-I B subunit. We noted differences in the pattern of expression of SLT-II B in the different vectors. The amounts of SLT-II B produced by pLL14 and pDA78 were maximal in overnight cultures. In contrast, the SLT-II B subunit produced by pDA76 was detected maximally in late-logphase cultures and did not increase in overnight cultures. The SLT-II B subunit expressed from pDA76 was made without a leader sequence and was therefore detectable only in bacterial cell lysates. We do not know if the difference in SLT-II B expression is related to the lack of a leader sequence. Despite the absence of a leader sequence, the SLT-II B subunit was assembled into multimers as is seen with holotoxin. However, the variation in our success in purifying the SLT-II B subunit from pDA76 suggests that the multimeric B subunit made from constructs lacking the leader sequence (pDA76) may be less stable than those made from constructs with the leader sequence (pLL14), at least when produced in 9.9-liter fermentor cultures. In retrospect, we were not able to improve upon the amounts of purified SLT-II B subunit beyond the levels seen in our first expression system (pLL14) with the T7 promoter. SLTs normally move into the periplasm and are assumed to assemble into the final multimeric structure in that site. SLT-II B contains two cysteine residues, and it is likely that disulfide Downloaded from http://iai.asm.org/ on July 24, 2020 by guest FIG. 7. Competitive binding to HeLa cells of iodinated SLT-II B and unlabeled SLT-II B determined by incubating HeLa cells with solutions containing the same amount of iodinated SLT-II B (approximately 90,000 cpm) and increasing concentrations of unlabeled SLT-II B. Error bars indicate standard deviations from triplicate experiments. Dilution % Leucine incorporationb at the following SLT-II concn (ng/ ml): VOL. 63, 1995 307 the semiquantative Western blots) expressed by the vectors described in this study was in the same range as we had seen previously with the SLT-I B subunit expression system in V. cholerae (1). When we examined SLT-II B expression from the plasmids described in this study in V. cholerae CVD103HgR, however, the levels were barely detectable by Western blotting. Our previous study of SLT-I B expression used V. cholerae O395-N1 (1); however, we have expressed similarly high levels of the SLT-I B subunit from V. cholerae CVD103HgR (4). Therefore, the lack of SLT-II B expression from V. cholerae CVD103HgR is not thought to be due to differences in the V. cholerae strain used in the present study, and the explanation remains elusive. We were able to obtain large amounts of pure MBP–SLT-II B subunit fusion protein and to cleave SLT-II B from the fusion protein by using factor Xa. However, the yields of the SLT-II B subunit with this method were no greater than those with the other expression systems in which the SLT-II B subunit was expressed alone. We are aware of only one other study which examined the expression of SLT-II B as a fusion protein, in that instance with glutathione S-transferase (10). Those investigators did not purify the SLT-II B subunit alone and obtained only small amounts of the B subunit when their fusion protein was cleaved with thrombin. We have previously shown that it is possible to raise neutralizing antibodies to SLT-I in rabbits by using either holotoxin (SLT-I) or recombinant SLT-I B subunit (16, 17). In contrast, antiserum raised to SLT-II B in the present study reacted strongly in Western blots with the SLT-II B band present in the holotoxin but was only poorly neutralizing. Similar to the observation with SLT-I, the antiserum raised to intact SLT-II readily neutralized the holotoxin. The reasons for this difference are not clear. Initially, we speculated that because we were immunizing animals with a preparation of SLT-II B which was predominantly B subunit monomers, the immunodominant epitopes present on monomers were not recognized on holotoxin. This speculation is supported by the observation that the monoclonal antibody 4D1 (raised to holotoxin and capable of neutralizing SLT-II [7]) does not recognize monomers in Western blots. However, a booster dose of cross-linked SLT-II B did not enhance the titer of neutralizing antibody. Thus, it is possible that the predominant epitopes exposed in the cross-linked protein were still not the neutralizing epitopes. While the antiserum to holotoxin contains antibodies to the A subunit, antibodies to the A subunit are not required in order to neutralize SLT-I (16, 17), and we have previously shown that monoclonal antibodies to SLT-II B are capable of neutralizing SLT-II (7). We have already shown that antibodies to the SLT-I B subunit are protective in an in vivo rabbit model of enterohemorrhagic E. coli infection with strain RDEC-H19A (16, 17, 25). The data from the present study suggest that a recombinant SLT-II B subunit produced as described in this study will not have the same capability to serve as a vaccine antigen to protect against the effects of SLT-II. We are currently evaluating ways to stabilize the SLT-II B multimer, on the assumption that preservation of the multimeric SLT-II B subunit structure is important in maintaining neutralizing epitopes. ACKNOWLEDGMENTS This work was supported by grants A-16242 and AI-20325 from the National Institute of Allergy and Infectious Diseases and grant P 30 DK-34928 for the Center for Gastroenterology Research on Absorptive and Secretory Processes, New England Medical Center, from the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Md., and by a Health Sciences Downloaded from http://iai.asm.org/ on July 24, 2020 by guest bond formation is important in the correct folding of the B subunit pentamer. Although the reducing environment in the cytoplasm is likely to prevent disulfide bond formation, it is possible that this occurs in a slow and spontaneous fashion once the SLT-II B subunit is in a nonreducing environment (20). This non-protein isomerase-dependent disulfide bond formation is slow and prone to errors (20), which may be part of the explanation for the variation in our ability to purify SLT-II B from pDA76. One of the striking features seen in the various expression systems studied was the difference between the amount of SLT-II B detected by Western blotting and that detected by ELISA. There are at least two possible explanations for this. First, our data suggest an apparent instability of SLT-II B subunit multimers when synthesized without the A subunit. The ELISA described in this study uses the monoclonal antibody 4D1 as the capture molecule. 4D1 does not recognize B subunit monomers in a Western blot but does recognize crosslinked B subunit multimers (4). Therefore, any B subunit monomers in the bacterial preparations would likely not be detected in the ELISA but would be seen on a Western blot with polyclonal antiserum directed toward the B subunit, which, in contrast to the monoclonal antibody 4D1, does recognize SLT-II B subunit monomers. An alternative possibility is that the majority of the SLT-II B subunit produced remains as monomers and never assembles into multimers. Even if this is the case, we still believe that the purified SLT-II B subunit multimers are significantly less stable than SLT-I multimers. Our purification scheme uses a 10-kDa-cutoff filtration step, which would not retard B subunit monomers. Despite this, the filtered material does not appear to retain a multimeric nature. The small-scale experiments using pLL14 demonstrate an increase in the detected periplasmic SLT-II B levels which is disproportionate from the increase in OD (Table 2). Since the ELISA used in this assay detects multimeric SLT-II B subunit, it is possible that this disproportionate increase in SLT-II B is actually a reflection of increased multimerization of the SLT-II B subunit in the periplasm. The competitive inhibition experiment using iodinated and noniodinated SLT-II B supports this concept of SLT-II B subunit multimer instability. This experiment demonstrates that as more noniodinated SLT-II B subunit is added, the binding of iodinated B subunit increases before it then decreases at higher concentrations of unlabeled protein. We speculate that the initial increase in binding is due to an increase in the number of multimeric SLT-II B subunits which contain at least one iodinated B subunit and are therefore detected in the assay as an increase in the amount of radioactivity bound to HeLa cells. As the amount of nonradioactive B subunit added increases, the number of totally nonradioactive multimers increases, and these competitively inhibit the binding of the limited number of multimers containing an iodinated SLT-II B subunit. We also examined the binding of cell lysates from the SLT-II B expression systems to HCM used as the capture system in an ELISA format (3) and found that amounts similar to those seen with the 4D1 ELISA bound (data not shown), suggesting that binding to HCM occurs with multimeric B subunit. Since the purification scheme used in this study was based on binding to HCM coupled to Sepharose 4B, it is not surprising that we obtained only small amounts of purified SLT-II B subunit by using an HCM-based purification method. We are currently investigating methods to purify the SLT-II B subunit monomer, which may form multimers when purified and concentrated. The total amount of the SLT-II B subunit (as determined by SLT-II B SUBUNIT EXPRESSION SYSTEMS 308 ACHESON ET AL. for the Tropics Partnership in Research and Training grant from the Rockefeller Foundation, New York, N.Y. We thank J. Kaper (Center for Vaccine Development, University of Maryland, Baltimore) for providing V. cholerae CVD103HgR. REFERENCES 13. Karmali, M. A. 1989. Infection by verocytotoxin-producing Escherichia coli. Clin. Microbiol. Rev. 2:15–38. 14. Keusch, G. T., A. Donohue-Rolfe, M. Jacewicz, and A. V. Kane. 1988. Shiga toxin: production and purification. Methods Enzymol. 165:152–162. 15. Lopilato, J., S. Bortner, and J. Beckwith. 1986. Mutations in a new chromosomal gene of Escherichia coli K-12, pcnB, reduce plasmid copy number of pBR322 and its derivatives. Mol. Gen. Genet. 205:285–290. 16. Mullett, C., D. W. K. Acheson, L. Y. Tseng, and E. Boedeker. 1994. High titer serum antibody against the B subunit of Shiga-like toxin I protects in a model of enterohemorrhagic Escherichia coli (EHEC) colitis. Abstr. Gastroenterol. 106:A740. Abstracts of the Annual Meeting of the American Society of Gastroenterologists. 17. Mullett, C., C. Holls, D. W. K. Acheson, L. Y. Tseng, and E. Boedeker. 1993. Serum and mucosal immune response to Shiga-like toxin I (SLT-I) B subunit during infection of adult rabbits with Escherichia coli strain RDEC-H19A: a rabbit model for enterohemorrhagic E. coli colitis. Abstr. Gastroenterol. 104: A750. 18. O’Brien, A. D., J. W. Newland, S. F. Miller, R. K. Holmes, H. William-Smith, and S. B. Formal. 1984. Shiga-like toxin-converting phages from Escherichia coli strains that cause hemorrhagic colitis or infantile diarrhea. Science 226: 694–696. 19. Ostraff, S. M., P. I. Tarr, M. A. Neill, J. H. Lewis, N. Hargrett-Bean, and J. M. Kobayashi. 1989. Toxin genotypes and plasmid profiles as determinants of systemic sequelae in Escherichia coli O157:H7 infections. J. Infect. Dis. 160:994–998. 20. Pugsley, A. P. 1993. The complete general secretory pathway in gram-negative bacteria. Microbiol. Rev. 57:20–108. 21. Ramotar, K., B. Boyd, G. Tyrrell, J. Gariepy, C. Lingwood, and J. Brunton. 1990. Characterization of Shiga-like toxin I B subunit purified from overproducing clones of the SLT-I B cistron. Biochemistry 272:805–811. 22. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 23. Samuel, J. E., C. P. Perera, S. Ward, A. D. O’Brien, V. Ginsburg, and H. C. Krivan. 1990. Comparison of the glycolipid receptor specificities of Shigalike toxin type II and Shiga-like type toxin II variants. Infect. Immun. 58: 611–618. 24. Scotland, S. M., G. A. Willshaw, H. R. Smith, and B. Rowe. 1987. Properties of strains of Escherichia coli belonging to serogroup O157 with special reference to production of Vero cytotoxins VT1 and VT2. Epidemiol. Infect. 99:613–624. 25. Sjogren, R., R. Neill, D. Rachmilewitz, D. Fritz, J. Newland, D. Sharpnack, C. Colleton, J. Fondacaro, P. Gemski, and E. Boedeker. 1994. Role of Shiga-like toxin I in bacterial enteritis: comparison between isogenic Escherichia coli strains induced in rabbits. Gastroenterology 106:306–317. 26. Stein, P. E., A. Boodhoo, G. J. Tyrrell, J. L. Brunton, and R. J. Read. 1992. Crystal structure of the cell binding B oligomer of Verotoxin-1 from Escherichia coli. Nature (London) 355:748–750. 27. Waddell, T., S. Head, M. Petric, A. Cohen, and C. Lingwood. 1988. Globotriaosyl ceramide is specifically recognized by the Escherichia coli verotoxin. Biochem. Biophys. Res. Commun. 152:674–679. Downloaded from http://iai.asm.org/ on July 24, 2020 by guest 1. Acheson, D. W. K., S. B. Calderwood, S. A. Boyko, L. L. Lincicome, A. V. Kane, A. Donohue-Rolfe, and G. T. Keusch. 1993. Comparison of Shiga-like toxin I B-subunit expression and localization in Escherichia coli and Vibrio cholerae by using trc or iron-regulated promoter systems. Infect. Immun. 61: 1098–1104. 2. Acheson, D. W. K., A. Donohue-Rolfe, and G. T. Keusch. 1991. The family of Shiga and Shiga-like toxins, p. 415–433. In J. E. Alouf and J. H. Freer (ed.), Sourcebook of bacterial protein toxins. Academic Press Ltd., London. 3. Acheson, D. W. K., M. Jacewicz, A. V. Kane, A. Donohue-Rolfe, and G. T. Keusch. 1993. One step high yield purification of Shiga-like toxin II variants and quantitation using enzyme linked immunosorbent assays. Microb. Pathog. 14:57–66. 4. Acheson, D. W. K., A. V. Kane, A. Donohue-Rolfe, and G. T. Keusch. Unpublished data. 5. Calderwood, S. B., D. W. K. Acheson, M. B. Goldberg, S. A. Boyko, and A. Donohue-Rolfe. 1990. A system for production and rapid purification of large amounts of the Shiga toxin/Shiga-like toxin I B subunit. Infect. Immun. 58: 2977–2982. 6. Conradi, H. 1903. Uber losliche, durch aseptische autolyse erhalatene giftstoffe von Ruhr- und Typhus bazillen. Dtsch. Med. Wochenschr. 20:26– 28. 7. Donohue-Rolfe, A, D. W. K. Acheson, A. V. Kane, and G. T. Kuesch. 1989. Purification of Shiga toxin and Shiga-like toxins I and II by receptor analog affinity chromatography with immobilized P1 glycoprotein and production of cross-reactive monoclonal antibodies. Infect. Immun. 57:3888–3893. 8. Donohue-Rolfe, A., G. T. Keusch, C. Edson, D. Thorley-Lawson, and M. Jacewicz. 1984. Pathogenesis of Shigella diarrhea. IX. Simplified high yield purification of Shigella toxin and characterization of subunit composition and function by the use of subunit-specific monoclonal and polyclonal antibodies. J. Exp. Med. 160:1767–1781. 9. Endo, Y., K. Tsurugi, T. Yutsudo, Y. Takeda, T. Ogasawara, and K. Igarashi. 1988. Site of action of a Vero toxin (VT2) from Escherichia coli O157:H7 and of Shiga toxin on eucaryotic ribosomes. Eur. J. Biochem. 171:45–50. 10. Gunzer, F., and H. Karch. 1993. Expression of A and B subunits of Shiga-like toxin II as fusions with glutathione S-transferase and their potential for use in seroepidemiology. J. Clin. Microbiol. 31:2604–2610. 11. Jacewicz, M., H. Clausen, E. Nudelman, A. Donohue-Rolfe, and G. T. Keusch. 1986. Pathogenesis of Shigella diarrhea. XI. Isolation of a Shigella toxin-binding glycolipid from rabbit jejunum and HeLa cells and its identification as globotriaosyl ceramide. J. Exp. Med. 163:1391–1404. 12. Jackson, M. P., R. J. Neill, A. D. O’Brien, R. K. Holmes, and J. W. Newland. 1987. Nucleotide sequence analysis and comparison of the structural genes for Shiga-like toxin I and Shiga-like toxin II encoded by bacteriophages from Escherichia coli 933. FEMS Microbiol. Lett. 44:109–114. INFECT. IMMUN.