Acetylation of O-specific lipopolysaccharides from Shigella flexneri 3a and 2a occurs in Escherichia coli K-12 carrying cloned S. flexneri 3a and 2a rfb genes

Most of the Shigella flexneri O-specific serotypes result from O-acetyl and/or glucosyl groups added to a common O-repeating unit of the lipopolysaccharide (LPS) molecule. The genes involved in acetylation and/or glucosylation of S. flexneri LPS are physically located on lysogenic bacteriophages, whereas the rfb cluster contains the biosynthesis genes for the common O-repeating unit (D.A.R. Simmons and E. Romanowska, J. Med. Microbiol. 23:289-302, 1987). Using a cosmid cloning strategy, we have cloned the rfb regions from S. flexneri 3a and 2a. Escherichia coli K-12 containing plasmids pYS1-5 (derived from S. flexneri 3a) and pEY5 (derived from S. flexneri 2a) expressed O-specific LPS which reacted immunologically with S. flexneri polyvalent O antiserum. However, O-specific LPS expressed in E. coli K-12 also reacted with group 6 antiserum, indicating the presence of O-acetyl groups attached to one of the rhamnose components of the O-repeating unit. This was confirmed by measuring the amounts of acetate released from purified LPS samples and also by the chemical removal of O-acetyl groups, which abolished group 6 reactivity. The O-acetylation phenotype was absent in an E. coli strain with an sbcB-his-rfb chromosomal deletion and could be restored upon conjugation of F' 129, which carries sequences corresponding to a portion of the deleted region. Our data demonstrate that E. coli K-12 strains possess a novel locus which directs the O acetylation of LPS and is located in the sbcB-rfb region of the chromosomal map.

polyvalent 0 antiserum. However, 0-specific LPS expressed in E. coli K-12 also reacted with group 6 antiserum, indicating the presence of O-acetyl groups attached to one of the rhamnose components of the 0-repeating unit. This was confirmed by measuring the amounts of acetate released from purified LPS samples and also by the chemical removal of O-acetyl groups, which abolished group 6 reactivity. The 0-acetylation phenotype was absent in an E. coli strain with an sbcB-his-rJb chromosomal deletion and could be restored upon conjugation of F'129, which carries sequences corresponding to a portion of the deleted region. Our data demonstrate that E. coli K-12 strains possess a novel locus which directs the 0 acetylation of LPS and is located in the sbcB-rjl region of the chromosomal map. Shigella flexneri is a major cause of bacillary dysentery in developing countries (25). Like the lipopolysaccharides (LPS) of other gram-negative enteric bacteria, S. fle-xnen LPS is composed of three portions that are covalently bound: lipid A, core oligosaccharide, and 0-polysaccharide chain or 0-specific antigen (20). S. flexneri strains are currently subtyped into 13 serotypes on the basis of antigenic determinants on the 0-specific polysaccharide (see reference 38 for a review). This typing scheme arises from a variety of combinations of type-and group-specific antigens which have been identified chemically and immunologically (6,12,(21)(22)(23)38). In all S. flexneni serotypes except type 6, the 0-antigen repeating unit is composed of a common tetrasaccharide with the following structure: -3)-13-D-Glcp-NAc-(1-2)-Ct-L-Rhap["RhaI"]-(1-2)-a-L-Rhap["RhaII"]-(l-3)-ao-L-Rhap["RhaIII"]- (1-(23). This common structure has been associated with the group antigen 3,4 and is found in S. fle-xneri serotype Y (Fig. 1). The other S. flexneri serotypes, X, la, lb, 2a, 2b, 3a, 3b, 4a, 4b, 5a, and Sb, arise from the attachment of a-D-glucosyl and/or O-acetyl residues to different specific positions on the common repeating unit (38). Such substitutions appear to be the result of postpolymerization modifications of the 0-specific LPS determined by various lysogenic bacteriophages (38).
Genetic mapping studies have revealed that S. flexneri rfb (rjbSf), encoding the common tetrasaccharide structure (corresponding to serotype Y in Fig. 1), maps adjacent to the his locus (17), whereas the bacteriophages responsible for acetylations and glucosylations of the 0-polysaccharide chain are * Corresponding author. Electronic mail address: 37_1510@uwo vax.uwo.ca. integrated near the pro-lac region on the chromosome (36). Bacteriophage Sf6, a lysogenic phage identified in S. flexneri 3a, has been shown to encode a gene involved in the 0-acetylation of the 0-polysaccharide chain at the RhaIII (9,18,26,47), thus leading to the expression of the group 6 antigen (Fig. 1). Group 6 antigen also occurs in S. flexneri serotypes lb, 3b, and 4b but it is not found in the remaining serotypes (38).
Our laboratory is interested in the molecular study of genes involved in the biosynthesis of the 0 side chain of LPS in some enteropathogenic bacteria. In this article, we report the identification of a novel function encoded by the Escherichia coli K-12 chromosome involved in determining the acetylation of the S. flexneri 0-specific LPS resulting in group 6-specific reactivity.

MATERUILS AND METHODS
Bacterial strains, plasmids, chemicals, and antisera. The bacterial strains and plasmids used in this study are described in Table 1 ,ug/ml, respectively. Strain KL704 was grown on M9 minimal medium containing the required amino acid supplements except histidine to maintain selective pressure for the F'129 carrying the his genes and surrounding sequences. S. flexnei polyvalent, group 3,4-, group 6-, group 7,8-, and type IIspecific antisera were purchased from the Beijing Institute of Biological Products, People's Republic of China. Antisera were adsorbed extensively with E. coli K-12 strain HB101 cells as described by Edwards and Ewing (15). The presence of S. flexneri 0-specific LPS antigens was determiined by slide agglutination.
Phage methods. A cell lysate containing bacteriophage Sf6 was obtained from strain FH10(Sf6) by induction with 1 ,ug of mitomycin C per ml added to a culture with an optical density at 660 nm of 0.3, followed by incubation for 3 h with vigorous shaking. The sensitivity of bacterial strains to Sf6 was determined by spot tests as described previously (18). For the isolation of Sf6 DNA, phage particles were concentrated by precipitation with 10% (wt/vol) polyethylene glycol and extracted with phenol-chloroform and the DNA was precipitated with ethanol. Recombinant DNA methods. Chromosomal DNA for the construction of genomic libraries was prepared as described previously (29). A rapid miniscale isolation of whole genomic DNA (35) was used for hybridization experiments (see below). Plasmids were extracted and purified by using a commercial kit (Quiagen Inc., Chatsworth, Calif.) and also in some cases by the method of Birnboim and Doly (4) followed by ultracentrifugation in cesium chloride-ethidium bromide density gradients (43). Small-scale plasmid preparations were carried out as described by Xu et al. (49). S. fletxneri 3a and 2a genomic DNA libraries were constructed by using cosmid pHC79 linearized with BamHI, treated with alkaline phosphatase, and ligated with partially digested (Sau3A) chromosomal DNA fragments from S. flexneri 3a strain SF51575 and S. flexneri 2a strain SF51250. The ligated DNA was in vitro packaged into A phage particles by using a commercial kit (Packagene; Promega, Madison, Wis.) and transduced into E. coli K-12 HB101. Gene libraries were screened by colony immunoblots (29), using a 1:1,000 dilution of adsorbed S. fleneri polyvalent antiserum (primary antiserum) and goat anti-rabbit immunoglobulin G (secondary antiserum) coupled to horseradish peroxidase. Electrophoresis of plasmid DNA cleaved with restriction endonucleases was performed as previously described (28,43,46). Transformations were carried out by the calcium chloride method (10) and in some cases by electroporation with a Gene Pulser apparatus (Bio-Rad Laboratories Ltd., Mississauga, Ontario, Canada), using 0.1-cm cuvettes and conditions described elsewhere (13).
Hybridizations. Genomic and plasmid DNA fragments previously cleaved with restriction endonucleases were separated by electrophoresis and transferred to nitrocellulose filters as previously described (43,46). Southern blot hybridization experiments were carried out by using undigested bacteriophage Sf6 DNA as a probe. The probe was labeled with [32P]dATP (Amersham, Arlington Heights, Ill.) by oligonucleotide synthesis (16), and hybridizations were carried out at 37°C for 16 to 18 h under conditions described previously (44,45). LPS analysis. LPS was extracted by the hot phenol-water method of Westphal and Jann (48) and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (32). LPS was visualized by a silver-staining procedure (41) and also examined by immunoblotting under conditions described elsewhere (39). LPS immunodot blots were carried out by spotting 10 ,uI of LPS suspensions (containing 1 to 2 jig of LPS) on nitrocellulose paper. De-O-acetylated LPS for immunological analysis with group 6 antiserum was obtained by alkaline treatment as described by Carlin et al. (6). Acetate released from LPS samples was detected and quantitated by high-pressure liquid chromatography (HPLC) as described by Dupont and Clarke (14).
Conjugations. F' matings were done as described previously (33) with KL704(F'129) as the donor and the E. coli rfb-deleted strain CLM4 carrying the recombinant plasmids pYS1-5 or pEY5 as the recipient (Table 1). Exconjugants were selected on M9 agar medium containing ampicillin and the required supplements except histidine.
Immunoelectron microscopy. Protein A-colloidal gold particles (diameter, ca. 15 to 20 nm) were provided by C.-S. Guo (Department of Vibrio cholerae, Institute of Epidemiology and Microbiology, Chinese Academy of Preventive Medicine). Bacterial cells were fixed at room temperature for 5 min in 1% glutaraldehyde diluted in 0.1 M phosphatebuffered saline (PBS; pH 7.2) and centrifuged at 6,000 x g for 15 min. Washed bacterial pellets were resuspended in 2 ml of PBS. Bacterial suspension (50 ,ul) was mixed with 50 ,u1 of S. flexneri polyvalent antiserum (1:100 dilution in PBS), and the mixture was incubated at 37°C for 60 min and washed once with PBS. Cells were resuspended in 100 ptl of PBS, mixed with 50 pl of protein A-colloidal gold (1:10 dilution in PBS), incubated at 37°C for 30 min, and given two washes with PBS. Finally, cells were resuspended in 50 pl of PBS. Copper grids were floated onto a drop of the bacterial suspension for 1 min and then onto a drop of 1% (wt/vol) uranyl acetate (pH 4.0) for 1 min and were then air dried. Grids were examined in a Philips 300 electron microscope at an operating voltage of 130 kV.

RESULTS AND DISCUSSION
Cloning of the S. flexneri 3a and 2a rjb regions. Chromosomal DNA fragments from S. flexneri 3a strain SF51575 and S. flexneri 2a strain SF51250 were cloned in the cosmid vector pHC79. Cosmids pY1214 and pEY5 were found to VOL. 174, 1992 b In the nomenclature of the S. flexneri serotypes, the type antigen is indicated in roman numerals. The other numbers indicate the group antigens. carry rfb sequences of S. flexnei 3a and 2a, respectively, as revealed by colony immunoblotting, slide agglutination, and immunoelectron microscopy with an S. flexneri polyvalent antiserum ( Fig. 2; data not shown). pY1214 and pEY5 were transformed into E. coli HB101 and the E. coli rfl-deleted strain CLM4. Transformants in both strains gave a positive slide agglutination with the S. flexneri polyvalent antiserum. These results indicated not only that the antigens recognized by the antiserum were mediated by these cosmids but also that the rfb region from E. coli K-12 did not contribute to the expression of S. flexneri 0 antigen. LPS expression was examined by SDS-PAGE followed by silver staining and immunoblotting with the S. flexneri polyvalent antiserum (Fig. 3). LPS extracted from HB101 (pY1214) and HB101(pEY5) revealed a typical ladder-like banding pattern with a bimodal distribution of 0-specific polysaccharide chains similar to that seen in the parent S. flexneri SF51575 (Fig. 3, lanes A, B, D, and E), whereas no 0-specific polysaccharide chains were detected in the host strain, HB101 (data not shown). 0-specific LPS was also detected in CLM4(pY1214) and CLM4(pEY5); however, the polysaccharide chains showed a unimodal distribution (Fig.  3, lanes C and F). These experiments demonstrated that pY1214 (carrying the rfba region) and pEY5 (carrying the rflsf2. region) directed the biosynthesis of 0-specific LPS in E. coli K-12 strains, which normally do not express any 0-specific antigen. 0-specific LPS determined by pY1214 and pEY5 was also detected in E. coli CLM4, thus confirming that the entire rjb regions of S. flexneri 3a and 2a were present in these two plasmids.
Previous studies have shown the existence of a gene determinant involved in 0-specific polysaccharide length distribution which maps on a region adjacent to the rjb genes in E. coli 075, E. coli 0111, E. coli 07, Salmonella enterica group C2, and Yersinia pseudotuberculosis (1-3, 5, 24, 42). This gene, designated rol by one group (3), has recently been subcloned and sequenced and shown to map immediately upstream from the last gene of the histidine operon in an E. coli 075-K-12 hybrid as well as in Salmonella enterica (2). rol appears to be absent in the rjb-deleted strain S0874 and its recA derivative CLM4. Since both pY1214 and pEY5 directed the expression of unimodal 0-specific LPS in CLM4 and bimodal 0-specific LPS in HB101, we conclude that a rol-like gene is not present in these plasmids. It is possible that this gene has not been included in the cloned DNA or, alternatively, that the rol gene in S. flexneri is at The restriction endonuclease maps of recombinant cosmids pY1214 and pEY5 are shown in Fig. 4. To localize the rfbsf3a region in pY1214, various deletion derivatives from this plasmid were obtained by partial digestion of pY1214 DNA with HindIII followed by self-ligation. Deletion plasmids pYS1-2, pYS1-4, pYS1-8, and pYS1-9 lost the ability to direct the synthesis of 0-specific LPS, as revealed by slide agglutination and SDS-PAGE followed by silver staining (Fig. 4a and 5, lanes B, C, E, and F). In contrast, pYS1-5 retained the ability to synthesize the 0-specific LPS in E. coli HB101 and CLM4 ( Fig. 4a and 5, lane D; data not shown). To localize the rfbsf2 region in pEY5, DNA fragments from this plasmid obtained by partial cleavage with C HindIII werepsubcloned in the vector pMAV3. Plasmids pMEY2, pMEY13, pMEY31, pMEY32, and pMEY38 did not express 0-specific LPS, as shown by slide agglutination A.~~and SDS-PAGE ( Fig. 4b and 5, lanes H and J; data not shown), whereas pMEY3 encoded a positive 0-specific phenotype in E. coli HB101 and CLM4 (Fig. 4b and 5, lane I; data not shown). The restriction maps of pYS1-5 and pEY5 showed a common region of approximately 9.5 kb flanked by a PstI and a SphI site, as indicated by the solid arrows in Fig. 4b.
T'he restriction map of this region was also similar to that of the rfb5is, region reported by Macpherson et al. (27). These investigators documented by a hybridization experiment the close relationship at the DNA level among S. flexneri serotypes Y, X, la, lb, 2a, 2b, 3a, 3b, 4a, 4b, 5a, and Sb (27). Therefore, it was assumed that this common region in pYS1-5 and pEY5 should contain most of the rjbsf gene cluster. Single XhoI restriction endonuclease sites were found within the common region in both pYS1-5 and pEY5 (Fig. 4b). These sites were mutated by digestion with XhoI followed by end fillings with the large fragment of DNA polymerase I (Klenow fragment) prior to ligation. Cells containing either pYS1-5 or pEY5 with the mutated XhoI sites lost the ability to express 0-specific antigen as determined by slide agglutination. Therefore, it was confirmed that the XhoI sites in pYS1-5 and pEY5 were contained within rfbsfspecific sequences. Further evidence delineating the rbsf region in these two plasmids was obtained by constructing a 3.1-kb HindIII deletion in pYS1-5 (Fig. 4a,  pYS1-9) and a corresponding 4.2-kb HindIII deletion in pEY5 (Fig. 4b, pMEY2). These deleted plasmids lacked the 1.5-kb PstI-HindIII fragment located to the left of the common region in both parent plasmids (Fig. 4) and did not encode a positive slide agglutination phenotype. A single KpnI site unique to pYS1-5 was found at 0.3 kb to the right of the common region (Fig. 4b). To investigate whether this site was important for expression of 0-specific LPS genes, a mutation was constructed as described above for the XhoI sites. Cells containing the resulting plasmid were not agglutinated by the 0-specific antiserum, indicating that the mutation of the KpnI site affected the expression of an essential gene for the biosynthesis of the 0-specific LPS. Therefore, the cloned flSf3a cluster extended up to at least 0.3 kb to the right of the common SphI site (Fig. 4b). No KY>nI site was found on pEY5 carrying r)f2b. Also, pEY5 lacked a PstI site located at 0.6 kb to the left of the 9.5-kb PstI-SphI common region in pYS1-5 (Fig. 4b). Thus, fbSf3a and rfbSf2a regions showed heterogeneity in the restriction endonuclease sites located at the ends of the common sequences.
Differences were also found in the restriction maps of rflbSf2a regions of strain SF21250 presented in this study (pEY5, Fig. 4b) and that of another S. flexneri 2a strain reported in a previous study (pPM2213 [271). A SmaI site absent in pEY5 was identified in pPM2213 at 0.4 kb to the right of the PstI site delineating the left end of the common region (Fig. 4b, open arrow). Some restriction sites within the common region were found to be present in pPM2213 but in a different order (Fig. 4b, dots), whereas other restriction sites were present in similar positions and in the same relative order in both cases. This indicates the existence of variations in rflsf2. of different strains of S. flexneri 2a.
Since the basic tetrasaccharide structure of the 0-repeating unit in S. flexneri is common to all serotypes except type 6 (12,38) and since the rfb regions in these strains are highly homologous (27), it is not surprising that the physical maps of rfbsf clusters are similar. However, the differences in the restriction endonuclease maps of both the two cloned rfb regions found in this work and the cloned rfb region from S. flexneri 2a reported by Macpherson et al. (27) suggest that rfbsf regions of strains with serotypes other than type 6 are not necessarily identical. The significance of these variations in the genetic structure of rflSf regions remains to be elucidated. It is likely that the rfbs clusters determining the tetrasaccharide 0 repeat have arisen from a common ancestor and that further evolution has taken place independently in strains of each individual type, resulting in the loss of a perfect conservation of the DNA sequence, possibly without alteration of gene functions.
Identification of an 0-acetylation activity encoded by the E. coli K-12 strains. Since the chemical structures and antigenic composition of S. flexneri 0-specific serotypes are known (38), we attempted to infer the structure of the cloned 0-specific LPS by investigating its antigenic properties with type-and group-specific antisera. Table 2 shows that LPS preparations of HB101 cells containing pYS1-5 and pEY5 did not react with type IIor group 7,8-specific antisera. This suggests that glucosylation of the RhaIII (determining serotype 2 specificity) and glucosylation of the RhaI (determining the group 7,8 specificity found in serotype 3a) are indeed not present in the LPS expressed by these plasmids (Fig. 1). The lack of expression of S. flexneri 2a-specific determinants by E. coli K-12 cells carrying pEY5 indicates that the gene determining this modification is not present near the r)Sf2a cluster. Similarly, the lack of expression of group 7,8 antigen in the LPS expressed by pY1214 and pEY5 permitted us to conclude that the genes determining the group 7,8 antigen are not included in the cloned DNA and also cannot be supplied by E. coli K-12 strains.
In contrast, HB1O1(pYS1-5) and HB101(pEY5) were ag-  c A weak positive reaction was found by immunodot blotting of LPS samples extracted from these strains. glutinated by the S. flexneri group 6-specific polyclonal antiserum extensively absorbed with HB101 (Table 2). Recombinant plasmids pYS1-5 and pEY5 transformed into other E. coli K-12 strains such as DHSa and DH1 also conferred a positive agglutination with group 6 antiserum (data not shown). These results were unexpected since group 6 reactivity denotes the 0 acetylation of the RhaIII of the 0-subunit backbone (Fig. 1). The group 6 antiserum was specific since it also recognized epitopes in S. flexneri serotypes lb, 3a, 3b, and 4b possessing similar 0 acetylations in the RhaIII of the 0-repeating unit (Table 2) (38) but did not agglutinate S. flexneri strains representative of the other serotypes (data not shown). These results strongly suggested that the O-acetyl groups are attached to RhaIII, since our group 6 antiserum reacted only with LPS of the S. flexnen serotypes lb, 3a, 3b, and 4b known to possess the group 6 antigen determinant (38), although a definitive proof will require chemical analysis of the LPS.
To demonstrate that reactivity with group 6 antiserum was due to 0 acetylation of the 0-specific polysaccharides, we treated purified LPS with alkali, which causes the release of the O-acetyl groups. De-0-acetylated LPS from HB101 (pYS1-5), HB1O1(pEY5), and S. flexneri 3a strain SF51575 failed to react with the group 6 antiserum as determined by immunodot blot experiments ( Table 2). The presence of O-acetyl groups in the LPS was further confirmed by determining and quantifying the release of acetate by HPLC ( Table 3). The results indicated that LPS purified from HB101(pYS1-5) and HB1O1(pEY5) contained 5 to 10 times more acetyl groups than the LPS from the host HB101 strain. The small amount of acetate released from the HB101 LPS is probably present in the lipid A moiety, since this strain expresses only a lipid A core and lacks any 0-specific LPS. The amounts of acetate detected in LPS from HB101 cells carrying the recombinant plasmids were smaller than that found in the LPS from the wild-type S. flexneri 3a strain SF51575 (Table 3). To investigate the possibility that the cosmids pYS1-5 and pEY5 encode a function involved in 0 acetylation of LPS, we transformed these plasmids by electroporation into S. flexneri serotype Y (strain SF51581). Transformants did not express any S. flexneri type or group antigens other than group 3,4 typical of serotype Y ( Table 2). Plasmid DNA prepared from these strains was transformed back into E. coli HB101, and transformants were now agglutinated with group 6 antiserum (data not shown). These experiments suggest that the gene(s) responsible for 0 acetylation of the 0-specific LPS determined by pYS1-5 and pEY5 is present in the E. coli K-12 chromosome.
Acetylation of recombinant 0-specific LPS is not related to bacteriophage Sf6. 0-acetylation of S. flexneri 3a 0-specific LPS is associated with the presence of the lysogenic bacteriophage Sf6, which carries a gene believed to encode an O-acetyltransferase activity (18,26). This gene has recently been cloned and sequenced by two different groups (9,47). The Sf6 chromosomal integration site maps within the prolac region of the S. flexneri chromosome (36,38). To investigate the possibility that Sf6 bacteriophage accidentally infected HB101 cells used in this study and was integrated in the chromosome, we transformed pYS1-5 and pEY5 into E. coli K-12 strains DHSa, SY327, CC118, 3000X111, and M8820 containing various pro-lac deletions. Transformants from all these strains agglutinated with group 6 antiserum, suggesting that the acetylation function was not encoded in the pro-lac region of the E. coli K-12 chromosome.
Southern blot hybridization analysis with whole Sf6 DNA as a radiolabeled probe demonstrated the presence of the Sf6 sequences only in chromosomal DNA from S. flexneri 3a strain SF51575, whereas the DNAs from S. flexneri Y strain SF51581 and from E. coli HB101 and CLM4 did not reveal any significant homologies with the probe (data not shown). These experiments ruled out the possibility that this bacteriophage or a similar one is responsible for the 0 acetylation of LPS observed in E. coli K-12 strains carrying the recombinant clones. Overall, all these results suggested that the acetylation of S. flexneri 0-specific LPS expressed in E. coli K-12 strains containing pYS1-5 or pEY5 is due to a gene unrelated to that already identified in bacteriophage Sf6.
Early studies have reported that Sf6 has a narrow host range and that this is due mainly to its inability to infect strains of E. coli or S. flexneri carrying 0-specific LPS with structures other than the basic tetrasaccharide unit corresponding to S. flexneri serotype Y (18,26). Since Sf6 lysogens of E. coli and S. flexneri are resistant to infection with this bacteriophage, E. coli HB101(pYS1-5), E. coli HB101(pEY5), S. flexneri Y strain SF51581, and S. flexneri 3a strain SF51575 were tested for sensitivity to Sf6. E. coli HB101(pYS1-5), E. coli HB101(pEY5), and S. flexneri SF51581 were susceptible to Sf6, whereas S. flexneri 3a strain SF51575 was resistant to infection with this bacteriophage. Smaller and more turbid plaques were found in HB101(pYS1-5) and HB101(pEY5) compared with those in S. flexneri SF51581. Thus, our data demonstrate that Sf6 can, although to a lesser extent, infect E. coli K-12 expressing 0-acetylated LPS. This could occur because the acetylation of the 0-specific LPS in E. coli K-12 is less efficient or is not completely specific for rhamnose. A similar situation has been documented for the oafA gene harbored by strains of Salmonella entenca group C2, which is involved in the 0 acetylation of group C2 0-specific LPS but also can acetylate the 0-specific LPS from S. entenca group B but to a lesser extent (19,37).
Localization of an 0-acetylation locus in the sbcB-rjb region of the E. coli K-12 chromosome. During the course of our studies involving group 6 antigen expression in different E. coli K-12 strains, we observed that cells from strain CLM4 containing either pYS1-5 or pEY5 failed to react with group 6 antiserum, although they were agglutinated with S. flexneri polyvalent antiserum. Detection and quantification of acetate released from purified LPS samples of CLM4 containing pYS1-5 and pEY5 indicated only one-to twofold larger amounts of acetate compared with the control values found in LPS extracted from the host strain with no plasmids (Table 3). This demonstrates that strain CLM4 failed to cause a significant 0 acetylation of the 0 antigen determined by the cloned S. flexneri rfb regions carried by these plasmids. Strain CLM4 carries a chromosomal deletion eliminating the sbcB-rfb region. To investigate whether the deleted region in CLM4 is involved in the expression of 0 acetylation, the F'hisl29, which carried part of the deleted region (40), was conjugated into CLM4(pEY5) and CLM4(pYS1-5). Apr his' exconjugants were able to express the group 6 antigen, as determined by slide agglutination. It has been shown by other investigators that F'129 cannot complement mutations in the cps gene cluster located clockwise with respect to the sbcB-his-gnd-rfb chromosomal genes (40). Therefore, the finding that pYS1-5 and pEY5 cannot express LPS with group 6 reactivity in the rfb-deleted strain CLM4 but that this defect can be complemented with the F'129 containing part of the deleted region provides the basis for the demonstration that an 0-acetylation locus is located near the his-rfb region of the E. coli K-12 chromosome. Recent attempts in our laboratory resulted in the isolation of a cosmid clone from E. coli K-12 strain W3110 containing the 0-acetylation gene, which also can complement his-deficient mutants (50). This cosmid clone directs the expression of group 6 antigen in S. flexneri Y strain SF51581 (50).
The finding that cosmids carrying S. flexneri 3a and 2a rfb regions expressed an 0-acetylated LPS in E. coli K-12 demonstrates for the first time the existence of a gene function involved in acetylation of LPS in this strain. A detailed analysis involving the cloning, fine genetic mapping of this gene(s), and characterization of gene product(s) is in progress in our laboratory (50). Our results are consistent with the view that although the cloning of rfb genes in E. coli K-12 is possible (see reference 42 for a review), genes involved in LPS biosynthesis already present in this microorganism may result in differences in the expression of recombinant 0-specific LPS compared with the expression of the 0-specific LPS in the wild-type parent isolate.