Previous Article | Next Article ![]()
Journal of Bacteriology, January 2005, p. 758-764, Vol. 187, No. 2
0021-9193/05/$08.00+0 doi:10.1128/JB.187.2.758-764.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.
TEDA School of Biological Sciences and Biotechnology,1 Tianjin State Laboratory of Microbial Functional Genomics, TEDA College, Nankai University,2 Tianjin Biochip Corporation, TEDA, Tianjin,3 National Institute for Communicable Disease Control and Prevention, Chinese Center for Disease Control and Prevention, Beijing, People's Republic of China,6 N. D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Russia,4 School of Molecular and Microbial Biosciences, University of Sydney, Sydney, Australia5
Received 31 May 2004/ Accepted 4 October 2004
|
|
|---|
|
|
|---|
The O antigen (O-specific polysaccharide), comprising repeats of an O unit of generally two to seven sugars, is the outer variable part of lipopolysaccharide (LPS) and is a major target of the immune system and bacteriophages. Each strain expresses only a particular O-antigen form, and the variation is thought to offer a selective advantage in the niche occupied (46). Bacteria belonging to the species E. coli and Shigella spp. are closely related (38, 45). E. coli has 166 O-antigen forms, and Shigella has 33 O-antigen forms, 13 of which are present in both organisms. Rough mutants lacking O antigen are serum sensitive or impaired in virulence (43), but there is little direct evidence for the role of O-antigen specificity or variety in pathogenicity.
The differences among the many forms of O antigen are due mainly to genetic variation in the O-antigen gene cluster (46). In E. coli, Shigella, and Salmonella enterica, the O-antigen gene clusters are located between housekeeping genes galF and gnd and contain genes for the synthesis of nucleotide sugars specific to O antigen, sugar transferase genes to make the O unit, and genes for O-unit processing, including those for the polymerization and translocation of O antigen. In E. coli, a conserved 39-bp JUMPStart sequence is located in the intergenic region between the galF gene and the O-antigen gene cluster (40). The sequences of sugar transferase genes and O-unit-processing genes are normally specific to a particular O antigen. Specific PCR methods based on O-antigen-specific genes have been proposed for molecular typing of many E. coli and Shigella O serogroups (21, 25, 26, 28, 54, 57-59).
Structural studies of the O polysaccharide. The E. coli O145 type strain (G1100) from the Institute of Medical and Veterinary Science, Adelaide, Australia, was grown to late log phase in 10 liters of Luria-Bertani medium with a 16-liter fermentor (BIOSTAT C-10; B. Braun Biotech International, Melsungen, Germany) under constant aeration at 37°C and pH 7.0. Bacterial cells were washed and dried as described by Robbins and Uchida (47). The LPS (0.54 g) was isolated from dried cells (6.5 g) by the phenol-water method (60) and purified by precipitation of nucleic acids and proteins with CCl3CO2H as described previously (63).
The LPS (80 mg) was hydrolyzed with aqueous 2% acetic acid at 100°C for 75 min, and a lipid precipitate was removed by centrifugation at 13,000 x g for 20 min. The water-soluble carbohydrate portion was fractionated by gel permeation chromatography on a column (56 by 2.6 cm) of Sephadex G-50 (S) in 0.05 M pyridinium acetate buffer (pH 4.5) with monitoring by a Knauer differential refractometer to give trisaccharide 1 (18.2 mg) and a higher oligosaccharide fraction (27.5 mg) but no polysaccharide. Trisaccharide 1 resulted from depolymerization of the O polysaccharide by the glycosidic linkage of N-acetylneuraminic acid (Neu5Ac) and corresponds to the repeating unit of the O polysaccharide (see below).
For nuclear magnetic resonance (NMR) spectroscopic studies, samples were deuterium exchanged by freeze-drying twice from D2O and then examined as solutions in 99.96% D2O at 50°C on a Bruker DRX-500 spectrometer with internal acetone (
H 2.225,
C 31.45) as a reference. The 13C-NMR spectrum of trisaccharide 1 (Fig. 1) contained signals for three anomeric carbons at
97.3, 97.8, and 103.0; one carboxyl group (C-1 of Neu) at
177.6; one methylene group (C-3 of Neu) at
40.6; one methyl group (C-6 of a 6-deoxyhexose) at
16.8; two hydroxymethyl groups (C-6 of a hexose and C-9 of Neu) at
62.0 and 64.6; 10 oxygen-bearing carbons at
68.4 to 79.7; three nitrogen-bearing carbons at
51.7, 54.3, and 57.0; two N-acetyl groups at
23.7 (CH3), 174.7, and 175.2 (both C
O); and one N-acetimidoyl group at
20.5 (CH3) and 167.7 (C
N). Accordingly, the 1H-NMR spectrum of trisaccharide 1 contained major signals for two anomeric protons at
4.56 and 5.17, one methylene group (H-3 of Neu) at
1.89 (axial proton) and
2.41 (equatorial proton), one methyl group (HP6 of a 6-deoxyhexose) at
1.19, two N-acetyl groups at
1.98 and 2.06, and one N-acetimidoyl group at
2.30.
![]() View larger version (17K): [in a new window] |
FIG. 1. 13C-NMR spectrum of trisaccharide 1 obtained by mild acid degradation of the E. coli O145 LPS. Arabic numerals refer to carbons in sugar residues. F, FucN; G, GlcN; N, Neu.
|
3 Hz showed that GlcN and FucN are
-linked, and hence, Neu occupies the reducing end of the trisaccharide. A rotating-frame nuclear Overhauser effect spectrometry experiment (mixing time, 100 ms) demonstrated the modes of glycosylation and the sequence of the monosaccharides by correlations between FucN H-1/GlcN H-3 and GlcN H-1/Neu H-4 at
5.17/
3.72 and
4.56/
4.10, respectively. The structure of trisaccharide 1 was finally confirmed by the measurement of the molecular mass of 698.1 Da with negative-ion electrospray ionization mass spectrometry.
Comparison of the NMR chemical shifts of and structural data for trisaccharide 1 with a trisaccharide (designated trisaccharide 2) isolated from the O polysaccharide of S. enterica subsp. arizonae O21 (56) showed their close structural similarity. The only difference between them is that Neu is inserted at position 4 in trisaccharide 1 rather than at position 7, as in trisaccharide 2. The 1H- and 13C-NMR chemical shifts of FucN were essentially the same, and therefore, in both trisaccharides, the N-acetimidoyl group is linked to FucN to form 2-acetamidoylamino-2,6-dideoxy-L-galactose (FucNAm) (replacement of the N-acetimidoyl group with an N-acetyl group would cause significant changes in the NMR parameters; e.g., see Table S1 in the supplemental material). These data together showed that trisaccharide 1 has the structure
-L-FucpNAm-(1
3)-ß-D-GlcpNAc-(1
4)-Neu5Ac 1 and that trisaccharide 2 has the structure
-L-FucpNAm-(1
3)-ß-D-GlcpNAc-(1
7)-Neu5Ac 2.
A high-molecular-mass polysaccharide (polysaccharide I [PSI], 58 mg) was obtained by O-deacylation of the LPS (80 mg) by treatment with aqueous 12% ammonia (4 ml) at 37°C for 16 h followed by gel chromatography on Sephadex G-50 (S). PSI consisted of oligosaccharide repeating units of three types containing different fucosamine derivatives, namely, (i) that with the free amino group (FucN), (ii) an N-acetimidoyl derivative (FucNAm), and (iii) an N-acetyl derivative, 2-acetylamino-2,6-dideoxy-L-galactose (FucNAc). The last derivative was evidently derived from L-fucosacetamidine by alkaline hydrolysis (36). Further alkaline treatment of PSI with aqueous 12% ammonia (4 ml) at a higher temperature (50°C, 16 h) fully converted FucNAm into FucNAc to give PSII (23 mg). Acid hydrolysis of PSII with 2 M CF3CO2H at 120°C for 2 h followed by analysis on a Biotronik LC-2000 amino acid analyzer (Chromex UAX8 cation exchanger, 0.7 M sodium citrate buffer, pH 5.28, 67°C) produced GlcN and FucN (from 2-acetamido-2-deoxy-D-glucose [GlcNAc] and FucNAc) in the ratio
1:1. Similar hydrolysis of trisaccharide 1 released GlcN smoothly from GlcNAc but only trace amounts of FucN from FucNAm.
PSI and PSII were studied by NMR spectroscopy as described above for trisaccharide 1 (see Table S1 in the supplemental material for 1H- and 13C-NMR chemical shifts). The observed differences between the 13C-NMR chemical shifts of C-3 and C-2 of FucNAm in trisaccharide 1 (
69.1 and 54.3, respectively) and in PSI (
74.7 and 52.4, respectively) are characteristic of substitution of this residue in the polysaccharide at position 3 (39). The H-3e chemical shift of
2.89 to 2.93 in PSI and PSII demonstrated the
-linkage of Neu5Ac [compare published H-3e chemical shifts (16)
2.72 ± 0.05 for
-Neu5Ac and
2.32 ± 0.08 H-3e for ß-Neu5Ac].
These data elucidated the structure of the O polysaccharide of E. coli O145, which is closely related to those of S. enterica serovar Touera O48 and S. enterica subsp. arizonae O21 (Fig. 2). Most likely, all of the O polysaccharides have the same O unit shown in Fig. 2 and the distinctions between them are incorporated at the stages of (i) polymerization with formation of either 1
4- or 1
7-linkage between the O units and (ii) O acetylation in the S. enterica O polysaccharides, which is a commonly occurring, often nonstoichiometric postpolymerization modification.
![]() View larger version (19K): [in a new window] |
FIG. 2. O-polysaccharide structures of E. coli O145 S. enterica serovar Toucra O48 (29a), and S. enterica subsp. arizonae O21 (59). OAc, acetyl. FucNAm is 2-acetimidoylamino-2,6-dideoxygalactose. In the original paper on S. enterica serovar Toucra O48 (29a), FucNA was misidentified as FucNAc.
|
, which was purchased from Beijing Dingguo Biotechnology Development Center (Beijing, People's Republic of China). Sequencing was carried out using an ABI 3730 automated DNA sequencer. Sequence data were assembled with the Staden package (52). A sequence of 16,932 bases, which covers the DNA from the JUMPStart site to the start of the gnd gene, was obtained. E. coli O145 O-antigen gene cluster. Fifteen open reading frames (ORFs) (not including the gnd gene) were found with the program Artemis (49) (Fig. 3). BLAST and PSI-BLAST were used for searching the GenBank database (4). The program BlockMaker was used for searching conserved regions in protein sequences (32). The protein domain database Pfam was searched by using program HMMER (8). The program TMHMM 2.0 was used for identification of potential transmembrane segments in protein sequences (37). The program CLUSTAL W was used for performing sequence alignment (55). All the ORFs were putatively identified based on homology comparisons by using available databases and found to be involved in O-antigen synthesis (Table 1). The ORFs overlapped slightly or had small intergenic regions as commonly found in E. coli O-antigen gene clusters, except that a 131-bp noncoding region is present between ORFs 4 and 5. Although a similarity search showed no information for this noncoding sequence, we can propose a recombination occurrence because of its abnormal length.
|
View larger version (8K): [in a new window] |
FIG. 3. O-antigen gene cluster of E. coli O145. All the genes are transcribed in the direction from galF to gnd. The shaded areas represent significant intergenic regions.
|
|
View this table: [in a new window] |
TABLE 1. Characteristics of the ORFs in the E. coli O145 O-antigen gene cluster
|
Genes for biosynthesis of CMP-Neu5Ac. ORFs 1, 2, 3, and 4 shared, respectively, 59, 70, 51, and 60% identity to NnaD (NeuD), NnaB (NeuB), NnaC (NeuA), and NnaA (NeuC), encoded by the E. coli K1 capsule gene cluster (GenBank entries AAC43301, AAC43302, AAA24210, and AAA24211, respectively) (6). NnaA, NnaB, and NnaC of E. coli K1 synthesize CMP-Neu5Ac when they are cloned into E. coli K-12 on a plasmid (6), and NnaD has a role in the synthesis of CMP-Neu5Ac by interacting with NnaB (17, 18). NnaA is the GlcNAc 2-epimerase that converts GlcNAc to ManNAc. NnaB condenses the latter and phosphoenolpyruvate to form NeuNAc. NnaC, the CMP-NeuNAc synthetase, activates the sugar before it is linked to oligosaccharide segment. ORFs 1, 2, 3, and 4 also shared, respectively, 74, 81, 56, and 63% identity to WckD (NnaD), NnaB, NnaC, and NnaA, encoded by the O-antigen gene cluster of E. coli O104 (57). In E. coli O104, the four genes were proposed to synthesize CMP-Neu5Ac. Therefore, ORFs 1, 2, 3, and 4 in the E. coli O145 O-antigen gene cluster are proposed to encode the enzymes for CMP-Neu5Ac synthesis and are named nnaD, nnaB, nnaC, and nnaA, respectively.
Genes for biosynthesis of L-FucNAm. ORFs 11, 12, and 13 showed 81, 57, and 70% identity to FnlA (WbjB), FnlB (WbjC), and FnlC (WbjD), respectively, of the Pseudomonas aeruginosa O11 O-antigen gene cluster (GenBank entries AF72954, AAF72955, and AAF72956) (19); 79, 43, and 50% identity to FnlA (Cap5E), FnlB (Cap5F), and FnlC (Cap5G), respectively, of the Staphylococcus aureus type 5 capsule gene cluster (GenBank entries AAC46088, AAC46089, and AAC46090); and 88, 71, and 89% identity to FnlA, FnlB, and FnlC, respectively, of the E. coli O26 O-antigen gene cluster (21). FnlA, FnlB, and FnlC are enzymes of the UDP-L-FucNAc biosynthesis pathway (21, 35). Therefore, ORFs 11, 12, and 13 are proposed to encode the enzymes for synthesis of UDP-L-FucNAc, as an intermediate in synthesis of the L-FucNAm in the O antigen of E. coli O145 and are named fnlA, fnlB, and fnlC, respectively.
ORF 8 shared 41% identity or 66% similarity with WbpG of P. aeruginosa O5 (GenBank accession number AAG06538) and shared the motif SGGLDSS with homologues of WbpG. In P. aeruginosa O5, WbpG is an aminotransferase forming the C-3 acetiminido group on the first sugar residue of the O unit (13, 48). We propose that ORF 8 is an aminotransferase gene related to the amination of L-FucNAc to synthesize L-FucNAm, and we name it wbuX.
O-unit-processing genes The only two ORFs encoding predicted membrane proteins are ORFs 5 and 6. ORFs 5 had 12 predicted transmembrane segments, which is a typical topology for Wzx, and belonged to the Pfam family (Pfam accession number PF01943) (E value = 0.02), members of which are flippases of surface oligosaccharides in bacteria. ORF 5 also shares 23% identity or 45% similarity with Wzx of Shigella flexneri (GenBank accession number CAA50771). It is clear that ORF 5 is the expected O-unit flippase gene (wzx) and is named accordingly. ORF 6 had 10 predicted transmembrane segments and a large periplasmic loop of 52 amino acid residues, which is a typical topology for Wzy, and shares 22% identity or 45% similarity with the putative Wzy of the Vibrio cholerae O22 O-antigen gene cluster (GenBank accession number BAA33635). It is clear that ORF 6 is the expected O-unit polymerase gene (wzy), and we name it accordingly.
Putative glycosyltransferase genes. Three sugars are present in the O antigen of E. coli O145. WecA, encoded in the enterobacterial common antigen gene cluster, is presumably responsible for adding the first sugar GlcNAc onto the lipid acceptor UndP in the assembly of the O units of E. coli (1), while genes in the O-antigen gene cluster encode the remaining glycosyltransferases for synthesis of O units. We expected two genes for the transfer of L-FucNAm and Neu5Ac.
ORF 14 showed 71% identity to WbuB (GenBank accession number AAN60464), a putative L-FucNAc transferase present in E. coli O26 and O172 (21), and was in the glycosyltransferase family 1 (pfam00534; E value = 2 x e5). It is proposed that ORF 14 is the gene for the transfer of L-FucNAm in E. coli O145, and we name it wbuB.
ORF 7 could not be assigned a function by homology, but the only remaining function (putative functions of ORFs 9, 10, and 15 are discussed below) is for the NeuNAc transferase. We propose that ORF 7 is probably the NeuNAc transferase gene in E. coli O145 and name it wbuW.
A proposed novel ammonia tunnel. ORFs 9 and 10 shared 45 and 52% identity with the deduced proteins of hisH and hisF in the B band (wbp) O-antigen gene cluster of P. aeruginosa O5, respectively (13). The two genes were expressed (13), but inactivation did not affect O-antigen synthesis (48). E. coli has a separate his operon with functional hisH and hisF genes downstream of the O-antigen gene cluster (14). In the case of P. aeruginosa O5, the same PAO1 strain that was used for analysis of the O-antigen gene cluster (13) later had its genome sequenced (53). The presence of hisF and hisH genes in the PAO1 O-antigen gene cluster was confirmed, and these were referred to as hisF2 and hisH2. The chromosome was shown to include in addition a full set of his genes. There is no reason to doubt that the E. coli O145 strain also has a full his operon, as has been found in other E. coli strains. It is most unlikely that the O-antigen-associated genes have a role in histidine synthesis. The genes are of typical length, hisF being 791, 755, 776, and 770 bp in the E. coli O145 and P. aeruginosa PAO1 O-antigen gene clusters, the E. coli K-12 his operon, and the P. aeruginosa PAO1 hisFAHB operon, respectively, while the hisH genes are 614, 608, 590, and 614 bp, respectively. The hisH and hisF genes are generally linked even in bacteria in which the his pathway genes are absent from one operon (2). However, in prokaryotes, hisA is generally located between them (48). The finding of homologues of hisH and hisF without a homologue of hisA in two O-antigen gene clusters suggests that it is not a coincidence. However, the substantial difference in sequence indicates an independent origin, or alternatively, that this arrangement is of long standing. In both E. coli and P. aeruginosa, the O-antigen gene cluster-associated his genes show no sign of mutational damage and were shown to be expressed in the latter case. It is also interesting that these gene clusters have related genes, wbpG and wbuX, adjacent to the hisF and hisH homologues that are proposed to be transferases for the amino component of the N-acetimidoyl component found in one of their sugars. HisF and HisH have been shown to form a complex (2), and recently it has been proposed that HisH, which acts as a glutaminase, passes the ammonia to HisF, where it is conducted through an ammonia tunnel to the active site of HisF, where it is used to amidate N'-[(5-phos-phoribulosyl)formimino]-5'-aminoimidazole-4-carboxamide-ribonucleotide, leading to cleavage, with one product being a precursor of histidine (5, 20). It seems very likely that the hisH homologues in the two cases described above act as glutaminases and, with the hisF homologue, conduct the ammonia to WbpG or WbuX. The report that the Pseudomonas hisF and hisH homologues are not required for O-antigen synthesis may be because growth was carried out in the presence of ammonium ions, as this is usual in laboratory culture, but the conditions were not reported (48). In the case of histidine synthesis, hisH is not required if ammonium ions are available (9, 34), but hisF is required. This requirement is to be expected, as the active site for amidation is on HisF, whereas we propose that the HisF homologues act only as a tunnel to convey ammonia derived from the HisH homologue to another protein, which is presumed to have the amino transferase activity. There is no experimental evidence for this proposal, but the distribution pattern of the three genes makes it highly probable. The hisF and hisH homologues, ORFs 9 and 10, were temporarily named wbuY and wbuZ.
Gene remnant. ORF 15 shared 68% identity with WbuC of E. coli O26, which was proposed to be a gene remnant in its O-antigen gene cluster (21). As ORF 15 (399 bp) was much smaller than normal O-antigen genes (approximately 1 kb), it was highly likely that ORF 15 was no longer functional, and we named it wbuC.
Identification of E. coli O145-specific genes. Primer pairs were designed based on the O-unit processing genes wzx and wzy, which are normally specific to different O antigens (Table 2). Two primer pairs for each gene were used to screen DNA pools consisting of E. coli and Shigella type strains of the 186 different O serogroups described in a previous study (27). The chromosomal DNA prepared from each of the E. coli and Shigella type strains to represent the broadest range of O-antigen forms was examined by PCR amplification of the mdh gene (coding for malate dehydrogenase) with primers wl-101 (5'-TTC ATC CTA AAC TCC TTA TT) and wl-102 (5'-TAA TCG CAG GGG AAA GCA GG) (44) to confirm their high quality for PCR assay. A total of 13 pools of DNA were made, each containing DNA from 12 to 19 strains, based on a similar approach of previous studies (27), except that control pool 13 is the same as pool 7 but lacks E. coli O145. Pools were screened by PCR using primer pairs based on wzx and wzy genes, respectively, of E. coli O145 (Table 2). The PCR protocol was as follows: 30 cycles of denaturation at 95°C for 30 s, annealing at 59°C for 45 s, and extension at 72°C for 1 min. PCRs were carried out in a total volume of 25 µl. All four primer pairs based on wzx and wzy produced bands of the correct sizes with the pool containing E. coli O145 chromosomal DNA, and no bands were detected with any other pools. The four primer pairs based on wzx and wzy, respectively, were further used with 48 E. coli clinical isolates representing different O antigens, of which seven are E. coli O145 (see Table S2 in the supplemental material) (strains were kindly provided by Lothar Beutin, Division of Microbial Toxins, Robert Koch Institute, Berlin, Germany, and James R. Johnson, Medical Service, Veterans Affairs Medical Center, Minneapolis, Minn.). A double-blind test was performed with the following conditions: each strain was cultured in Luria-Bertani medium at 200 rpm in 37°C for 12 h, and 3-ml cultures were centrifuged at 5,000 x g for 5 min. The pellet containing E. coli O145 was mixed with 100 µl of Milli-Q water, boiled at 100°C for 15 min, and centrifuged at 12,000 x g for 8 min. One microliter of supernatant was used as the template in the PCR assay, which was performed as described above. All E. coli O145 isolates were specifically detected, while none of the non-O145 strains produced any band. Therefore, all four primer pairs are highly specific to E. coli O145.
|
View this table: [in a new window] |
TABLE 2. PCR specificity test with E. coli O145 wzx and wzy genesa
|
This work was supported by the Chinese National Science Fund for Distinguished Young Scholars (grant 30125001), the NSFC General Program (grants 30270029, 30370339, and 30370023), the NSFC International Cooperation Program (grant 30125001), the 863 Program (grant 2002AA2Z2051), the Cooperation Research Fund for Nankai University and Tianjin University from the Chinese Ministry of Education, the Science and Technology Committee of Tianjin City (grant 013181711 to L.F. and L.W.), and the Russian Foundation for Basic Research (grant 03-04-39020 to S.N.S.).
Supplemental material for this article may be found at http://jb.asm.org/. ![]()
|
|
|---|
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»