Conserved Organization in the cps Gene Clusters for Expression of Escherichia coli Group 1 K Antigens: Relationship to the Colanic Acid Biosynthesis Locus and the cps Genes from Klebsiella pneumoniae

A conserved region of genes required for translocation and polymerization of group 1 K antigens.Lipid-linked intermediates are polymerized to form high-molecular-weight K30 polysaccharide via the Wzy-dependent polymerization pathway and then translocated through the outer membrane as an unlinked polymer. Surface assembly of capsular group 1 K antigen minimally requires two additional genes from the capsule locus, wza and wzc (17). Wza, Wzb, and Wzc are encoded in the “translocation-surface assembly” region of the K30 cluster and are thought to be an outer membrane lipoprotein, a cytoplasmic phosphatase, and an ATP-binding protein, respectively. Recent work with Acinetobacter johnsonii has shown that a Wzb homologue, Ptp, is capable of dephosphorylating a Wzc homologue, Ptk (20). Similar genes have been described in diverse systems including E. coli K-12 (wza,wzb, and wzc) (44),Klebsiella K2 (orf4, orf5, andorf6) (2), and Erwinia amylovora(amsH, amsI, and amsA) (9), where they are believed to be involved in the production of colanic acid, K2 capsular polysaccharide (CPS), and amylovoran, respectively. Wza and Wzc homologues are found in a variety of bacteria that produce CPS and extracellular polysaccharide (38), leading to the conclusion that these represent a common translocation-surface assembly pathway for cell surface polysaccharides. Although these proteins have been established to function in surface expression of the K30 capsular antigen (17), their precise role in the process has yet to be determined.

The region associated with CPS translocation-surface assembly was examined in detail. Chromosomal DNA was prepared from the strains shown in Table 1. To reduce the expression of CPS, bacteria were grown in Luria broth (33) at 42°C. To isolate chromosomal DNA, bacteria from a 5-ml overnight culture were collected by centrifugation and resuspended in 1.5 ml of a lysis solution (50 mM sodium chloride, 2% sodium dodecyl sulfate, 300 mg of proteinase K per ml). The suspension was incubated at 42°C until clear. Next, 250 μl of 5 M sodium chloride and 200 μl of 10% hexadecyltrimethylammonium bromide in 0.7 M sodium chloride were added, followed by a 30-min incubation at 65°C. The sample was then subjected to two phenol-chloroform-isoamyl alcohol (25:24:1) extractions and precipitated with 5% 5 M sodium chloride and 2 volumes of absolute ethanol. The DNA pellet was washed with 70% ethanol and resuspended in 100 μl of sterile water. Finally, the DNA was treated with RNase and subjected to a chloroform-isoamyl alcohol (24:1) extraction.

The region between wza and wzc was amplified by PCR. PCRs were performed with Pwo or Expand Long DNA polymerase (Boehringer Mannheim) in a Perkin-Elmer GeneAmp PCR System 2400 thermocycler. The primers used for DNA amplification are listed in Table 2 and shown in Fig.1. PCR products identical in size were obtained with primers JD89 and JD109 and chromosomal DNA from strains representing 6 K serotypes of Klebsiella and 12 K serotypes of E. coli (Fig. 2). The sizes of the products were consistent with those predicted from thecps_ECK30 and cps_KPK2sequence data.

• Open in new tab
• Download powerpoint

Fig. 1.

Organization of the cps_ECK30locus, which is located between galF and gnd on the E. coli E69 chromosome (17). Components of the Wzy-dependent polymerization pathway (wzx andwzy) are found in the region dedicated to K30 repeat unit synthesis and polymerization, along with the necessary glycosyltransferases, wbaP, wcaN,wcaO, and wbaZ. Genes involved in the translocation of capsular K30 antigen, wza andwzc, are near the start of the cluster. The functions oforfX, orfY, and orfZ are unknown. Primers used for PCR amplification of selected regions are indicated, and the sequences are given in Table 2.

• Open in new tab
• Download powerpoint

Fig. 2.

PCR analysis of the wza-to-wzcregion from different group 1 K serotypes of E. coli (A) andKlebsiella (B). For reference, K. pneumoniae K20 strain 889/50 is also shown in panel A and E. coli K30 strain E69 is shown in panel B. PCR products identical in size (1.8 kb) were obtained for all of the strains tested, indicating a conservation of gene order between clusters.

For detailed analysis, selected PCR fragments were purified with Qiagen nucleotide removal columns and the region was sequenced at the Guelph Molecular Supercentre (University of Guelph, Guelph, Ontario, Canada). Analyses revealed that the wza-to-wzc regions inE. coli A295b (GenBank accession no. AF118245 ), E75 (GenBank accession no. AF118246 ), N24c (GenBank accession no. AF118247 ), 2151 (GenBank accession no. AF118248 ), and Bi161-42 (GenBank accession no.AF118249 ) had 99.5% identity to E. coli E69 at the nucleotide level and that the equivalent regions incps_KPK2 and cps_ECK30 were 72% identical at the nucleotide level.

The observation that group 1 capsule clusters from E. coliand Klebsiella are organized with a conserved block of translocation-surface assembly genes is reminiscent of the modular structure of gene clusters required for the manufacture of group 2 capsules in E. coli. Group 2 (kps) clusters comprise three regions (51). Region 1 encodes proteins necessary for translocation of polysaccharide through the periplasm and across the outer membrane, as appears to be the case with the initial genes (wza to wzc) of the group 1 cpsclusters. The serotype-specific central region (region 2) is responsible for biosynthesis and polymerization of oligosaccharide repeat units. This is equivalent to the genes downstream ofwzc in the group 1 capsule clusters. Region 3 contains genes encoding the subunits for an ATP-binding cassette transporter that is required for capsule export across the cytoplasmic membrane (8). The region 3 components are not required in Wzy-dependent polysaccharides, such as the group 1 capsules (48, 51). In the group 2 clusters, regions 1 and 3 are conserved in strains that produce structurally distinct capsules and the gene products are functionally interchangeable between these strains (51). Although functional identity has not been formally shown for the translocation-surface assembly region of group 1 capsules, the high degree of conservation seems to indicate that this would be the case. Some conserved feature of the group 1 K antigens or a component in their assembly must therefore be recognized for the translocation-surface assembly processes to be completed.

orfX is a unique component of group 1 K antigencps gene clusters.Although homologues ofwza, wzb, and wzc are found in other systems, including the cps_K-12 cluster for colanic acid biosynthesis, orfX homologues have been found only in cps_ECK30 andcps_KPK2 (where it is designated orf3) (2, 17). In both cases, orfX represents the first gene in the cluster. PCRs were performed with primers designed to amplify the region between orfX (JD95 forKlebsiella and JD99 for E. coli) and a gene upstream of the K-antigen cluster, galF (GALF1) (Table 2; Fig. 1). The sizes of the fragments obtained varied considerably due to polymorphism upstream of cps (see below). However, sequence data showed orfX to be conserved in position and virtually identical at the nucleotide level in all of the E. colistrains (E56b, Bi161-42, A295b, N24c, and E75) andKlebsiella strains (A5054, 889/50, 6613, and 708) investigated (data not shown). In E. coli K30,orfX mutants do not appear to be impaired in translocation-surface assembly of the capsular K antigen (17) and the precise role of orfX remains unclear. However, the high degree of conservation in orfXhomologues would indicate that it is important in the production of group 1 capsules.

Identical cps gene clusters in E. coli K30 and K. pneumoniae K20.There are notable structural similarities between group 1 K antigens in E. coli andK. pneumoniae (structures are available online in the Complex Carbohydrate Structure Database [14a]). In the example of Klebsiella K20 (11) and E. coli K30 (10), the capsular antigen structures are identical. The relationship between these two gene clusters was also examined by PCR and sequencing. Similar to the E. colistrains analyzed above, the wza-to-wzc region in these strains was shown to be 99.5% identical (KlebsiellaK20 GenBank accession no. AF118250 ).

Genes downstream of wza, wzb, and wzcare involved in repeat-unit biosynthesis and polymerization. Genes in this region, in particular, the glycosyltransferases, will vary with the structure of the polymer being produced by any given strain. Among group 1 capsules, this region has been described in detail for E. coli K30 (17) and Klebsiella K2 (2), and although both encode the same classes of gene products, they exhibit only low levels of similarity. PCR amplification of the region from wcaO to wzx with primers JD90 and JD53 (Table 2; Fig. 1) yields products identical in size (3.6 kb) for both E. coli K30 and K. pneumoniae K20 (data not shown). At the nucleotide level, the products are 99% identical over the 550 bp sequenced from each end. The high degree of conservation at the nucleotide level for both the wza-wzcand the wcaO-to-wzx regions may indicate the transfer of cps gene clusters between K. pneumoniae and E. coli (see below).

Conservation of regulatory regions upstream of cps.Gene clusters for bacterial polysaccharides are characteristically preceded by a 39-bp JUMPstart (for “just upstream of many polysaccharide starts”) element (22). The JUMPstart element has also been identified upstream of gene clusters involved in F conjugation pilus assembly and hemolysin toxin secretion (reviewed in reference5). A second element, ops (for “operon polarity suppressor”) is an 8-bp motif located within the JUMPstart sequence (35). Both elements are believed to play a role in transcriptional antitermination of the gene clusters (5, 31, 35). Their role has been described in regulation of E. coli group 2 K-antigen (kps [42, 43]) and LPS O-antigen (rfb [31]) biosynthesis.

The role of the JUMPstart element has been best characterized for the hemolysin operon, where it has been shown to control operon polarity but not transcript stability (5, 35), in concert with a NusG homologue, RfaH (4, 28, 29). The element can function over long distances (2 kb); however, it does so only when present on the nascent transcript (35). In one possible model, the JUMPstart sequence functions by facilitating the formation of a number of stem-loop structures on the mRNA during transcript elongation (31). This region recruits RfaH and potentially other proteins. Binding of RfaH to stem-loop III may inhibit the formation of the other stem-loops, which are thought to induce premature termination when present.

To investigate the conservation of these regulatory elements, regions immediately upstream of the group 1 K-antigen clusters were amplified and sequenced. A conserved JUMPstart element was present in bothE. coli strains (E56b, Bi161-42, A295b, N24c, and E75) andKlebsiella strains (A5054, 889/50, 6613, and 708) (Fig.3A). This finding suggests that group 1 capsule clusters are subject to transcriptional control via antitermination. Interestingly, the cps gene clusters share a feature noted upstream of the O-antigen biosynthesis region ofE. coli O7 (wb*) (31), i.e., the presence of two ops elements.

• Open in new tab
• Download powerpoint

Fig. 3.

(A) JUMPstart sequences from polysaccharide gene clusters directing the synthesis of amylovoran (ams) (27), colanic acid (cps) (45), O7 antigen (wb*) (31), and the E. coliand Klebsiella group 1 capsule clusters examined in this study. Variations from the E. coli K30 sequence are highlighted in boldface. The two ops elements found in theE. coli O7 JUMPstart sequence are underlined. (B) Stem-loop structures which have been identified incps_ECK30 and cps_KPK2. They are located immediately downstream of wzc and may function as transcription terminators.

In the hemolysin system, antitermination is required to avoid extreme operon polarity in a situation where the structural gene for the toxin is separated from those required for toxin maturation and export by a stem-loop structure (5). Analysis of the availablecps_ECK30 and cps_KPK2sequences identified a stem-loop structure in the intergenic region between wzc and the initial gene of the repeat-unit synthesis region (Fig. 3B). This potentially provides a strong transcriptional terminator, thereby allowing differential expression of structural components for capsule translocation and the highly active enzymes involved in polymer synthesis. Equivalent stem-loop structures are also predicted from the nucleotide sequence downstream of thewzc homologue in the gene clusters for amylovoran (9) and colanic acid (44) (data not shown). Differential expression of genes required for translocation-surface assembly and synthesis is known to occur in the E. coligroup 2 kps clusters. In these systems, regions 2 and 3 are organized into one transcriptional unit under the control of the region 3 promoter. Region 2 genes rely on transcriptional antitermination by RfaH to avoid operon polarity problems (42). Thekps region 1 is not regulated via antitermination but has been shown to be thermoregulated (12, 40). In addition, the first gene of region 1, kpsF, plays a poorly understood role in regulation (12). It remains to be established whether the product of orfX plays a similar role in expression of group 1 K antigens.

Group 1 K antigens are known to be regulated by the Rcs (for “regulator of capsule synthesis”) system in both E. coli(24, 25) and Klebsiella (1, 32, 47). This two-component regulatory system is best characterized for colanic acid production in E. coli K-12 (19); however, essentially identical systems operate in E. amylovora(6, 7, 14, 27). It is believed that the RcsC protein senses an environmental signal and, along with a second protein, RcsF, modulates the activity of RcsB through phosphorylation. RcsB can interact with a Lon protease-sensitive protein, RcsA, and upregulatecps transcription. Both RcsA and RcsB have helix-turn-helix DNA-binding motifs and bind to the promoter region of theams cluster for amylovoran production in E. amylovora (27). A potential binding site has also been identified upstream of the colanic acid cluster based on the titration of regulatory proteins by using promoter DNA and sequence homology to the E. amylovora binding site (27, 45). The environmental signal sensed by the Rcs system is uncharacterized but may involve membrane perturbations (13, 18, 37) or osmotic stress (3, 18, 41).

Although group 1 capsules, colanic acid, and amylovoran are all Rcs regulated, there are some important differences in their expression patterns. Most notable is the observation that colanic acid and amylovoran are optimally produced at 20°C and are not manufactured at 37°C. This is not surprising for colanic acid since this polymer is not a virulence determinant (39) and its function may be more important in environments outside the host (18, 41).E. amylovora is a plant pathogen associated with infections at environmental temperatures. However, group 1 capsules are virulence determinants and are produced at 37°C (49). The ability to express cps gene products at 37°C may be due to altered interactions of an RcsA-RcsB dimer with the cps promoter or to the involvement of additional (as yet uncharacterized) regulatory proteins. The Rcs proteins themselves are highly conserved in E. coli K30 and E. coli K-12 (24, 25), suggesting that they are unlikely to determine the different patterns of expression. The regions upstream of cps in E. coli and K. pneumoniae strains with group 1 K antigens lack the published RcsA-RcsB-binding sequences (27). Therefore, although the Rcs system may function to regulate group 1 K antigens, details of this interaction may differ considerably from those for the colanic acid and amylovoran systems.

Analysis of the Klebsiella K2 cps upstream region resulted in the identification of a partial sequence of a putative ς⁵⁴ promoter (2). This is conserved in all theE. coli and K. pneumoniae strains examined here. However, this putative promoter lies downstream of the JUMPstart element and thus could not operate in situations where antitermination by RfaH is required. The precise physiological significance of this “promoter” is therefore unclear.

Possible lateral transfer of group 1 cps genes.The data presented above demonstrates that the K-antigen gene clusters of E. coli and Klebsiella are highly conserved in organization and in cps nucleotide sequence. This is consistent with the possibility of lateral transfer of group 1 capsule gene clusters between these organisms. Further evidence in support of this contention was obtained by the identification of IS elements upstream of the cps gene clusters. In E. coliK30, a partial IS1 element (250 bp) truncates an adjacent gene, orf2. orf2 is also found immediately upstream of thecps_KPK2 cluster (2) but is absent inE. coli K-12. orf2 is not essential for K-antigen production in K. pneumoniae (2). A survey of other strains showed that many of the E. coli cps clusters are flanked by IS sequences (Fig. 4). Of the six E. coli strains examined, five contained IS elements. The type of IS element present is highly variable, with only IS1 being identified in more than one strain (E. coli E69 and E56b). In contrast, only one of the five K. pneumoniae strains had an IS element. The location of the IS sequences seems to be quite highly conserved, with only two of six strains (E. coli A295b and K. pneumoniae 889/50) showing slight variation.

• Open in new tab
• Download powerpoint

Fig. 4.

Diagrammatic representation of the region upstream ofcps in the E. coli and K. pneumoniaestrains investigated. The regions for E. coli E69 (17) and K. pneumoniae Chedid (2) are based on published data. The location and type of IS element identified, if any, is indicated. The putative ς⁵⁴promoters and conserved JUMPstart elements are located between the IS elements (or orf2) and orfX, the first gene of each cluster.

The presence of IS elements adjacent to most of the tested E. coli cps clusters and their absence in all but one of thecps regions examined in Klebsiella strains suggest that IS elements may have mediated an initial transfer of the cluster to E. coli from Klebsiella and could continue to be important for the exchange of genes encoding different capsule types between strains. Support for this hypothesis can be found in an analysis of gnd (6-phosphogluconate dehydrogenase) alleles from a variety of E. coli and Klebsiellaisolates. The cps genes map near his andgnd in E. coli, and Nelson and Selander (34) have suggested that diversity in gndrepresents a surprisingly high degree of recombination. This was attributed to cotransfer with adjacent loci (primarily rfb) whose activities are subject to diversifying selection because of the host immune response. Nelson and Selander also argued for lateral transfer of gnd genes from Klebsiella to E. coli. E. coli strains with the 16 known group 1 K antigens have a limited array of LPS O antigens (23); of the >170 O serotypes, only O9, O9a, O8, O20, and O101 are represented.Klebsiella strains have 77 different K antigens (36) but fewer than 10 structurally distinct O antigens (26, 46). Notably, some O-antigen structures are shared, and lateral transfer of the O3 gene cluster to E. coli has been proposed (46). Collectively, these data suggest that transfer of a large region of DNA including the rfb andcps loci may have occurred. In such a scenario, the extended region between galF and his(orf2-cps-ugd-rfb-gnd) from Klebsiella would replace the “typical” E. coli region (cps[colanic acid]-rfb-gnd-ugd-wzz). This is consistent with our previous analysis of the regions surrounding gnd that confirmed the lack of wzz in E. coli strains with group 1 capsules (15, 16). Such organization also explains why expression of colanic acid and expression of a group 1 capsule are mutually exclusive. In previous work, we proposed that thecps_K-12 and cps_ECK30systems are allelic (25, 50). From the differences in organization (the presence of orfX), altered upstream sequences, and the IS/orf2 region missing in E. coli K-12, these can no longer be considered alleles. Colanic acid is therefore not simply a widespread serotype of group 1 K antigen and should not be included as such.

The simple conclusion that all E. coli group 1 capsules have arisen by lateral transfer of DNA from Klebsiella is complicated by several observations. First, not all E. coligroup 1 capsules have structurally identical counterparts inKlebsiella, although this could reflect further gene transfer and recombination events within cps after the initial transfer, thereby resulting in the production of novel structures. However, of the strains examined here, one E. coli isolate (E75) lacks an IS element immediately upstream ofcps and only one Klebsiella strain (889/50) has an IS element (IS903). These exceptions suggest that the events that resulted in group 1 capsule diversity in E. coliand Klebsiella are more complicated. As noted above, theE. coli K30 and K. pneumoniae K20 (889/50)cps gene clusters appear to be identical, and there are two possible explanations for this. One is that the cluster was mobilized to these strains from a common (unknown) source by a process involving different IS elements. The other is that the gene exchanges betweenKlebsiella and E. coli are possibly not restricted to unidirectional lateral transfer events. There is currently no data to resolve these possibilities.