|Home | About | Journals | Submit | Contact Us | Français|
Escherichia coli K1 is responsible for 80% of E. coli neonatal meningitis and is a common pathogen in urinary tract infections. Bacteria of this serotype are encapsulated with the α(2-8)-polysialic acid NeuNAc(α2-8), common to several bacterial pathogens. The gene cluster encoding the pathway for synthesis of this polymer is organized into three regions: (i) kpsSCUDEF, (ii) neuDBACES, and (iii) kpsMT. The K1 polysialyltransferase, NeuS, cannot synthesize polysialic acid de novo without other products of the gene cluster. Membranes isolated from strains having the entire K1 gene cluster can synthesize polysialic acid de novo. We designed a series of plasmid constructs containing fragments of regions 1 and 2 in two compatible vectors to determine the minimum number of gene products required for de novo synthesis of the polysialic acid from CMP-NeuNAc in K1 E. coli. We measured the ability of the various combinations of region 1 and 2 fragments to restore polysialyltransferase activity in vitro in the absence of exogenously added polysaccharide acceptor. The products of region 2 genes neuDBACES alone were not sufficient to support de novo synthesis of polysialic acid in vitro. Only membrane fractions harboring NeuES and KpsCS could form sialic polymer in the absence of exogenous acceptor at the concentrations formed by wild-type E. coli K1 membranes. Membrane fractions harboring NeuES and KpsC together could form small quantities of the sialic polymer de novo.
Polysialic acids are important virulence factors in bacterial pathogenesis. These capsular polysaccharides are associated with the pathogens Moraxella sp. non-liquefaciens, Pasteurella haemolytica A2, Neisseria meningitidis, and Escherichia coli (1, 3, 8). E. coli produces two types of polysialic acid: NeuNAc(α2-8) of K1 (17) and the alternating structure, NeuNAc(α2-8)NeuNAc(α2-9) of K92 (9). Both K1 and K92 capsules belong to the so-called group II capsules. Group II capsule gene clusters are organized into three regions (15, 38). While regions 1 and 3 are genetically conserved and functionally interchangeable among all group II capsular types (24), region 2 is usually capsule specific. However, region 2 sequences in K1 and K92 are practically identical and functionally interchangeable. Although the K1 and K92 polysialyltransferases synthesize polysialic acid with different structures, these enzymes are 82% identical and 91% similar at the amino acid level (34).
We have shown that K92 polysialyltransferase expressed in a host free of other capsule cluster genes is sufficient to elongate an existing polymer in vitro but cannot synthesize the polysialic chain de novo without other products of the gene cluster (16). On the other hand, if polysialyltransferase was accompanied by all other products of gene cluster in the same host, it was able to synthesize the full-length polysialic acid. Thus, polysialyltransferase needs some acceptor to start the chain, which may be formed by the presence of the other products of the gene cluster. Neither the naturally occurring acceptor molecule nor the gene products necessary for the polysaccharide production de novo have been described.
The purpose of this study was to determine which cluster gene products help polysialyltransferase NeuS to assemble the full-length sialic acid polymer de novo. We measured the ability of the various combinations of region 1 and 2 fragments to restore polysialyltransferase activity in vitro in the absence of exogenously added polysaccharide acceptor. By using this approach in a host free of other capsule cluster genes, we established the minimal set of genes necessary for the effective production of the full-length polysialic acid.
The strains used in this study are described in Table Table1.1. Bacteria were grown at 37°C on Luria-Bertani (LB) agar plates or broth, supplemented with a suitable antibiotic.
To express pACYC Duet-1-based vectors in the ΔnanA background, we introduced a T7 RNA polymerase-encoding gene into the EV80 chromosome (36). We used the λDE3 lysogenization kit (Novagen) according to the instructions provided by the manufacturer. We transformed the resulting lysogens with pWN660 (pACYC Duet-1 K92 neuS) and compared the levels of sialyltransferase activity in membrane fractions. For future work we have chosen the EV80(DE3) lysogen, whose membranes showed the same level of sialyltransferase activity as HMS174(DE3) pWN660 membranes.
We used a Primus 96 PCR system (MWG Biotech, Inc.) for all DNA amplification reactions. DNA fragments for TA TOPO cloning were amplified with Expand high-fidelity enzyme mix (Roche). The reaction mixture consisted of 20 ng of template DNA, 5 μl of Expand high-fidelity buffer (with 15 mM MgCl2), 5 μl of deoxynucleoside triphosphate solution (2 mM each), 15 pmol of each primer, 0.75 μl (2.5 U) of Expand high-fidelity enzyme mix, and PCR-grade water to a final volume of 50 μl. We used the thermal program recommended by the manufacturer of the polymerase. Briefly, an initial step at 94°C for 2 min was followed by 10 cycles of 15 s at 94°C, 30 s at 45°C, and 6 min 20 s at 68°C, then by 20 cycles of 15 s at 94°C, 30 s at 45°C, and 6 min 20 s plus 5 s for each successive cycle at 68°C, and final elongation at 68°C for 7 min.
KOD XL DNA polymerase (Novagen), which produces the blunt-ended DNA fragments, was employed in all the other amplifications. The reaction mixture was assembled according to the manufacturer. Briefly, the reaction mixture consisted of 5 μl of PCR buffer for KOD XL DNA polymerase, 5 μl of deoxynucleoside triphosphate solution (2 mM each), 20 ng of template DNA, 15 pmol of each primer, 1 μl (2.5 U) of KOD XL DNA polymerase, and PCR-grade water to a final volume of 50 μl. An initial step of the thermal profile at 94°C for 45 s was followed by 30 cycles of 15 s at 95°C, 30 s at 43°C, and 1 min 30 s at 72°C.
All primers used in this study are presented in Table Table2.2. Primers were custom synthesized by Invitrogen and the CBER core facility. We used two cosmids as templates for PCR amplification and sequencing of E. coli K1 and K92 genes: (i) pSR23 (26), which carries the entire K1 gene cluster in pHC79 (14), and (ii) pGB20 (24), which carries the entire K92 gene cluster in pHC79 (14).
Plasmid DNA was purified with Bio-Rad DNA purification kits. puReTaq Ready-to-Go PCR beads (Amersham) were routinely used for screening of transformants by PCR. Confirmatory sequencing of the region between neuCand neuS genes was performed by ACGT, Inc. We used the cosmid pSR23(26) as template for K1 neuE sequencing and the cosmid pGB20 (24) forK92.
Alignment of K1 and K92 neuE DNA sequences, as well as of predicted amino acid sequences, was performed with the ClustalW program (13).
A description of plasmids used in this study is presented in Table Table1.1. Some additional information is provided here for the following constructs. pWN639 was constructed by digestion of pWN623 with SmaI/SacI to delete kpsSC genes. The remaining part of pWN623 was treated with mung bean nuclease (NEB) and religated; thus, the resulting construct was K1 kpsEDU in pACYC Duet-1. To construct pWN640, pWN623 was digested with NcoI to delete the kpsED genes and religated, yielding a pACYC Duet-1 vector harboring K1 kpsUCS. To construct pWN642, an XhoI/SmaI DNA fragment containing full-length kpsS was recloned from pWN640 into XhoI/EcoRV sites of pACYC Duet-1.
pET-46 Ek/LIC-based plasmids containing the neuE gene were constructed by amplification of K1 neuE with pSR23 as template. The primers (Table (Table2)2) were designed as recommended by the manufacturer. PCR products were cloned into the Ek/LIC cloning site of the pET-46 Ek/LIC (Novagen) expression vector by the method recommended by the manufacturer. All neuE constructs were cloned in frame with C-terminal S-tag. Two neuE constructs, pWN658 and pWN663, had an additional N-terminal His6 tag.
Membrane fractions were prepared as described by McGowen et al. (16) with minor modifications. Briefly, cells were grown at 37°C in 1 liter of LB broth supplemented with suitable antibiotic. We used 50 μg/ml ampicillin in experiments with pSR23-, pGB20-, pSR426-, and pET46 Ek/LIC-based vectors, 50 μg/ml kanamycin in experiments with pCR XL TOPO-based vectors, or 20 μg/ml chloramphenicol in experiments with pACYC Duet-1-based vectors. In coexpression experiments in which two antibiotics were necessary, we reduced the concentration of each individual antibiotic by 25%. Bacterial cultures were grown until the A600 reached 0.5 to 0.6 and were induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 2 to 3 h. The cells were harvested by centrifugation, resuspended in “Cryo” buffer (50 mM Tris-HCl, 25 mM MgCl2, 150 mM KCl, 1.4% glycerol, 1.4% d-sorbitol, pH 8.0), and lysed in a French pressure cell. Cell debris was precipitated at 10,700 × g for 15 min. The cleared lysate was ultracentrifuged at 100,000 × g for 1 h, and the membrane pellet was resuspended with a glass tissue grinder in 0.5 to 1.0 ml of the Cryo buffer. Membrane fractions not used on the day of preparation were stored at −80°C.
The sialyltransferase assays were performed as described elsewhere (16), with the following modification of the assay mixture: 0.25 nmol [14C]CMP-NeuNAc (0.15 μCi), membrane preparation (0.5 to 0.7 mg protein), and 4 μg (10 μl of a 2-mg/ml solution) of colominic acid were mixed in buffer A (50 mM Tris-HCl, pH 8.0, 25 mM MgCl2) to a final volume 50 μl. The mixture was incubated at 37°C for 1 h, and the reaction was terminated by spotting 30-μl aliquots on Whatman 3M paper and developing as described elsewhere (16). Incorporation of the radiolabel into chromatographically immobile product was normalized per milligram of membrane protein.
Colominic acid solution was replaced with buffer in the assay above to measure polysaccharide production de novo.
Protein concentrations of the membrane preparations were determined with the BCA protein assay reagent (Pierce).
The expression plasmid pWV231 contains the endo-N-acetylneuraminidase gene of the K1-5 phage (25). This plasmid was constructed by PCR amplification of the endoneuraminidase gene using phage DNA as template and ligation of the PCR fragment into pTrcHis2TOPO (Table (Table1).1). The resulting plasmid, pWV231, encodes endo-N-acetylneuraminidase that is hexahistidine tagged at the carboxyl terminus. Bacterial cultures (TOP10 cells) (Table (Table1)1) harboring this plasmid were grown at 37°C in 200 ml of LB broth with 50 μg/ml ampicillin until the A600 reached 0.5 to 0.6 and were induced with 1 mM IPTG for 3 h. Cells were harvested by centrifugation, resuspended in 50 mM Tris-HCl, 100 mM NaCl, pH 8.0, and lysed in a French pressure cell. Cell debris was removed by centrifugation at 10,700 × g for 15 min. A 1-ml suspension of Superflow Ni-agarose beads (QIAGEN) was added to the cleared lysate (about 10 ml), and the mixture was tumbled at 4°C for 30 min. Ni-agarose was recovered by centrifugation (at 1,000× g for 2 min) and washed twice with 10 ml of 50 mM Tris-HCl, 100 mM NaCl, pH 8.0, and endo-N-acetylneuraminidase bound to Ni-agarose was eluted with 1 ml 50 mM Tris-HCl, 100 mM NaCl, 250 mM imidazole, pH 8.0. Enzyme stored at 4°C was active for 2 weeks.
Reaction mixtures for gel filtration were prepared as follows: 0.5 nmol [14C]CMP-NeuNAc (0.15 μCi) and the appropriate membrane preparation (1.0 to 1.2 mg protein) were mixed in buffer A (50 mM Tris-HCl, 25 mM MgCl2, pH 8.0), in a 100-μl final volume. The mixture was incubated at 37°C for 1 h, and the reaction was terminated by addition of 0.5 ml of phenol. All components were thoroughly mixed by vortexing, after which the phenol and aqueous fractions were separated by centrifugation. The aqueous fraction was extracted twice with 0.5 ml chloroform and lyophilized. The residue was dissolved in 100 μl of H2O and loaded onto a 1.5- by 50-cm column of Sephacryl S-300 (Amersham) equilibrated in 100 mM ammonium acetate, pH 7.5, and developed at a flow rate of 0.4 ml/min. If necessary, we treated the samples with purified endo-N-acetylneuraminidase (1 h at 37°C) before loading. 14C label was quantitated by liquid scintillation counting (Anoroc Scientific Envirosafe).
EV36 (positive control) and RS3036 E. coli cells (Table (Table1)1) were grown at 37°C in 1 liter of M9 minimal medium with 20 μg/ml streptomycin. Bacterial cultures were grown until the A600 reached 1.0 and processed for polysaccharide purification as described below.
EV80(DE3):NeuES-KpsCS-NeuA (pWN646 + pWN656) and EV80(DE3):NeuES-KpsC-NeuA (pWN646 + pWN674) cells were grown at 37°C in 1 liter of M9 minimal medium supplemented with 20 mg/liter free sialic acid, 25 μg/ml kanamycin, and 10 μg/ml chloramphenicol. Bacterial cultures were grown until the A600 reached 0.5 to 0.6 and were induced with 1 mM IPTG for 3 h prior to processing for polysaccharide purification.
Polysaccharides were purified as described elsewhere (30) with the following modifications. E. coli EV36 cultures were used as the source of extracellular wild-type polysaccharide. Cells were removed by centrifugation, and polysaccharide was precipitated from the cleared medium overnight with 0.1% cetavlon (Sigma Chemical Co.). Polysaccharide was purified from the cetavlon paste as described below for intracellular polysaccharide.
Intracellular polysaccharide was purified from RS3036, EV80(DE3):pWN646 + pWN656, and EV80(DE3):pWN646 + pWN674. Cells were harvested by centrifugation, resuspended in 1 mM ammonium acetate buffer, pH 7.5, and lysed in a French pressure cell. Cell debris was precipitated at 10,700 × g for 15 min. The cleared lysate was extracted with an equal volume of phenol, and the aqueous phase was washed twice with an equal volume of chloroform. The aqueous phase was adjusted to 0.1% cetavlon and stored in the cold overnight. The cetavlon pellet was dissolved in 1 M CaCl2 and adjusted to 25% ethanol to precipitate nucleic acids. The nucleic acids were removed by centrifugation (3,000 × g, 10 min), and the cleared supernatant was adjusted to 85% ethanol to precipitate the polysaccharide. Polysaccharide was collected by centrifugation (3,000 × g, 10 min), dissolved in 1 mM ammonium acetate, pH 7.5, and incubated with 2 μg/ml DNase (Sigma Chemical Co.) and 2 μg/ml RNase A (Sigma Chemical Co.) on ice for 1 h, and the suspension was extracted once more with an equal volume of phenol. The aqueous phase was extracted with chloroform and dialyzed against 1 liter of 1 mM ammonium acetate, pH 7.5, overnight. The dialysate was ultracentrifuged at 100,000 × g for 1 h to remove lipopolysaccharide and lyophilized.
Samples for polyacrylamide gel electrophoresis (PAGE) were prepared as described for gel filtration. Lyophilized extracts were dissolved in water and divided into two equal fractions; one was incubated with the purified endo-N-acetylneuraminidase (25) for 1 h at 37°C, and the remaining fraction was left as a control.
Polyacrylamide gels were prepared as described by Pelkonen et al. (20). Samples were prepared in 0.2 M sucrose, 0.089 M Tris-borate, 2 mM EDTA, pH 8.3, in a 20-μl total volume. The following mixture of dyes was used as molecular weight (MW) markers (16). Trypan blue corresponds to a degree of polymerization of polysialic acid (DP) of 100, for xylene cyanol it is DP52, for bromophenol blue it is DP19, for bromcresol purple it is DP11.5, and for phenol red it is DP4. Samples (20 μl) were electrophoresed at room temperature at 100 V for 2 h.
After electrophoresis the gel was incubated for 1 h in 15 ml of gel drying solution (15% glycerol, 45% ethanol) and dried in a gel dryer (model 583; Bio-Rad) at 80°C for 2 h. Dried gels were placed in a storage phosphor screen cassette (Molecular Dynamics) overnight, and the image of the gel was scanned with a PhosphorImager 445 SI (Molecular Dynamics).
Polysaccharide isolated from 1-liter cultures of EV80(DE3):(NeuES-KpsCS-NeuA) and EV80(DE3):(NeuES-KpsC-NeuA) was dissolved in 0.5 ml H2O. A 10-μl aliquot of each sample was used for electrophoresis. PAGE gels were loaded with an amount of polysaccharide equivalent to 15 μg of free sialic acid for EV36, RS3036, and EV80(DE3):NeuES-KpsCS-NeuA. The EV80(DE3):NeuES-KpsC-NeuA sample contained the equivalent of 1 μg of free sialic acid, because of the small amount of polysaccharide produced. After electrophoresis, the gel was stained for 1 h in 0.5% Alcian Blue 8GX (Sigma) aqueous solution, pH 5.5.
The total concentration of sialic acid in purified polysaccharide samples was determined by the resorcinol method (7).
RS3036 polysaccharide (5 mg) was exchanged three times with D2O, and 1H and 13C nuclear magnetic resonance (NMR) spectra were recorded as described previously (16).
Membrane fractions of cells harboring pET-46 Ek/LIC plasmids were subjected to sodium dodecyl sulfate (SDS) electrophoresis on a 4-to-12% gradient NuPAGE gel (Invitrogen) according to the manufacturer's instructions. We loaded 0.8 μg total protein per lane if recombinant protein was expressed from the vector ribosome binding site (RBS) (pWN658 and pWN663). We loaded 5 μg total protein per lane if the recombinant protein was expressed from the relatively weak native RBS (pWN649, pWN654, pWN657, and pWN659) and for negative controls. Separated proteins were transferred to a PROTRAN nitrocellulose membrane (BioScience), and the nitrocellulose membrane was blocked overnight at room temperature with 5% (wt/vol) bovine serum albumin. S-protein-horseradish peroxidase (HRP) conjugate (Novagen) was used at a 1:5,000 dilution. The HRP conjugate-treated membrane was developed with ECL plus Western blotting detection reagents (Amersham) and visualized with an Image Station 440 (Kodak). We used membrane fractions of an uninduced strain harboring pWN649 and membranes of the host strain as negative controls. Perfect Protein Western markers (Novagen) were used as the MW standards.
The revised K1 neuE nucleotide sequence was submitted to GenBank under accession no. AY937259.
Polysialyltransferase (NeuS) catalyzes the transfer of NeuNAc from CMP-NeuNAc to the nonreducing end of a polysialic acid acceptor. In this study we used paper chromatography to detect de novo production of the polysialic acid. Although this method did not distinguish sialic polymers of various lengths, the chromatographic conditions used allowed separation of 14C-labeled sialic polymer from the [14C]CMP-NeuNAc and free sialic acid. Oligosaccharides equivalent to hexamer or shorter are mobile in this paper chromatography system. The products of the de novo assay reactions are digestable with endoneuraminidase. No radioactive products between the position of polymer (origin) and position of CMP-NeuNAc were detected by phosphorimaging (results not shown).
Since NeuS is a membrane-bound enzyme and could not be solubilized in an active form, we used membrane fractions in all assays. The ability of membranes harboring NeuS to incorporate [14C]NeuNAc into chromatographically immobile material in the absence of exogenously added polysaccharide acceptor was operationally defined as de novo polysaccharide production. To compare the level of de novo polysaccharide production achieved by polysialyltransferase NeuS alone and in the presence of other cluster products, we used E. coli TOP10 as a background strain. TOP10 is a derivative of E. coli K-12, whose complete genome has been sequenced and is therefore known to be devoid of genes involved in capsule biosynthesis. No de novo polysaccharide production was observed in NeuS-containing membranes (pWN609) lacking other products of the gene cluster (Table (Table3).3). However, membranes having all of the products of the gene cluster (pSR23) showed de novo polysaccharide production in vitro (Table (Table33).
We expressed region 2 of E. coli K1 in TOP10 and assayed membranes prepared from this construct to determine whether region 2 is sufficient for the de novo polysaccharide production. The data in Table Table33 imply that the products of K1 region 2 alone (pSR426) are not sufficient for polysialic acid production in vitro in a “capsule-free” TOP10 background. Additional gene products are required from either region 1 or 3, or both, for de novo synthesis. We used the E. coli K5 mutant MS108 as a background strain to determine whether any region 2 products other than NeuS are involved in the in vitro initiation reaction. E. coli K5 possesses a group II capsular gene cluster with regions 1 and 3, which are functionally interchangeable between E. coli K5 and K1 (24). However, K5 region 2 genes are responsible for synthesis of an alternating polysaccharide of glucuronic acid and N-acetylglucosamine and, therefore, do not have any sialic acid-related genes. Strain MS108 is a kfiA (region 2) mutant, lacking N-acetylglucosamine transferase activity, and is acapsular.
Membranes isolated from MS108 cells harboring the plasmid pSR426, which contains all K1 region 2 genes, catalyzed de novo polysaccharide production from CMP-NeuNAc (Table (Table4).4). However, membranes prepared from MS108 containing only NeuS (pWN609) did not support polymerization in the absence of exogenous acceptor. This result suggests that some other region 2-encoded products are required for the polysialic acid production. With the exception of NeuE, functions have been assigned to all K1 region 2 products. Presently, NeuD (4), NeuB (32), NeuC (29), and NeuA (31) are identified as members of the CMP-NeuNAc synthesis pathway. Consequently, these proteins are unlikely to participate in polysaccharide polymerization. The function of NeuE is not known. In addition, neuE sequences in the data bank (GenBank, University of Wisconsin E. coli K1 Genome Project [http://www.genome.wisc.edu/]) have multiple discrepancies, preventing the reliable prediction of the length of encoded product. Thus, it was possible that the activity of full-length NeuE had never been tested.
We constructed a plasmid which contains 100 bp of the neuC 3′ terminus, full-length neuS, and the entire region between these two open reading frames (ORFs) (pWN646) to check if NeuE participates in the production of polysialic acid de novo (Fig. (Fig.1).1). The data in Table Table44 demonstrate that NeuES proteins encoded by pWN646 are sufficient to observe polymerization of sialic acid by MS108 membrane fractions in the absence of exogenously added polysialic acid acceptor.
The data described above in Table Table44 suggest that the products of neuES genes (pWN646) are essential for the initiation of polysialic acid synthesis in an MS108 background. However, NeuES proteins expressed alone in a TOP10 background were unable to support polysialic acid synthesis in the absence of exogenous acceptor (Table (Table3).3). Thus, the gene products of region 1 and/or region 3 are also required to observe production of polysialic acid in vitro. Our next objective was to determine which proteins in addition to NeuES are necessary for de novo polysialic acid synthesis. The region 3 genes kpsM and kpsT encode components of an ABC transporter (18, 19) and, therefore, were unlikely candidates. For this reason we focused our attention on region 1. We designed constructs containing fragments of region 1 based on a pACYC184-derived pACYC Duet-1 vector (Fig. (Fig.1).1). This vector is compatible with the pUC-derived pCR XL TOPO, which we used to clone K1 region 2 fragments. Since pACYC Duet-1 contains the T7 promoter, we chose the K-12 derivative HMS174(DE3) strain, which lacks the genes involved in capsule production but encodes the requisite T7 RNA polymerase.
Plasmid pWN623, which contains most of region 1 but lacks kpsF, was able to complement NeuES (pWN646) for polysialic acid synthesis (Table (Table5).5). When we compared KpsEDU (pWN639) and KpsUCS (pWN640), only KpsUCS (pWN640) supported the production of polymer in the presence of NeuES. Coexpression of NeuES (pWN646) with either KpsCS (pWN648), KpsC (pWN673), or KpsS (pWN642) (Table (Table5)5) demonstrated that KpsCS (pWN648) was the most effective combination of region 1-encoded proteins necessary to support polysialic acid synthesis in vitro. A very low level of chromatographically immobile material was observed at the chromatogram origin when KpsC (pWN673) was coexpressed with NeuES (pWN646). KpsS expressed alone (pWN642) did not complement polysaccharide synthesis activity of NeuES (pWN646) (Table (Table55).
We attempted to complement the chromosomal kpsS deletion of strain RS3036 (Table (Table1)1) with the plasmid pWN642 to demonstrate that KpsS in pWN642 is active and that the inability to support polysaccharide production is not due to some cloning mistake. As shown in Table Table6,6, increased in vitro polysaccharide synthesis was observed using membranes obtained from strain RS3036 harboring plasmid pWN642. Interestingly, membranes of strain RS3036 incorporated a low level of [14C]NeuAc into polymer in the absence of an exogenously added polysialic acid acceptor. Similarly, a low level of incorporation was observed in the experiments described above with membranes of “capsule-free” strain HMS174(DE3) harboring NeuES-KpsC. Note that RS3036 membranes contained all the K1 cluster-encoded proteins except KpsS (Table (Table66).
Membranes derived from strain RS3036 (K1 Δ kpsS) were able to produce small amounts of polysaccharide de novo in the absence of exogenous polysialic acid acceptor. In order to determine whether the kpsS deletion affects the structure or size of the polysialic acid product, we isolated and characterized the polysaccharide assembled in vivo by K1 ΔkpsS strain RS3036. Analysis of the purified polysaccharide produced in vivo by this kpsS negative mutant showed that the 1H and 13C spectra are characteristic of an α(2-8)NeuNAc polymer and not short oligosaccharides, suggesting that the low level of incorporation is not due to low-MW product containing short fragments. The size of purified RS3036 intracellular polysialic acid was confirmed by PAGE (data not shown) (25).
The product formed from CMP-sialic acid by membranes harboring NeuES and KpsCS proteins was characterized by gel filtration and polyacrylamide gel electrophoresis. Although the paper chromatography assay is simple and reliable, the method does not distinguish sialic polymers of various lengths. The size of the polysaccharides, assembled from [14C]CMP-NeuNAc by the combination of NeuES-KpsCS and NeuES-KpsC proteins, was determined by gel filtration on Sephacryl S-300. After incubation with [14C]CMP-NeuNAc, membranes harboring NeuES-KpsCS were extracted with phenol and the extract was loaded on the column. The elution profile of this reaction mixture consisted of two peaks, one eluting at the void volume of the column and the other eluting at the position of CMP-NeuNAc (Fig. (Fig.2A).2A). The void volume peak of the NeuES-KpsCS sample (Fig. (Fig.2B)2B) was completely eliminated after incubation of the phenol extract with the K1-5 phage endo-N-acetylneuraminidase (25) prior to gel filtration. Thus, the large radioactive product was composed of α(2-8)polysialic acid. Similar results were obtained with the NeuES-KpsC reaction mixture. The only difference observed between the NeuES-KpsCS and NeuES-KpsC reaction mixtures was in the amount of radioactive product eluted in the void volume (Fig. (Fig.2A).2A). In the presence of KpsS, about 25% of total CMP-sialic acid was converted into the low-MW polymer and only 2 to 4% of total radioactivity was converted in the absence of KpsS protein. The absence of KpsS protein affects the total amount of polysaccharide produced by the membranes, but not its size.
The radioactive polymers produced by membranes harboring NeuES-KpsCS or NeuES-KpsC proteins were also characterized by PAGE. Membrane fractions were extracted with phenol after 1 h of incubation with [14C]CMP-NeuNAc and loaded on a PAGE gel. The results in Fig. Fig.3A3A (lanes 1 and 3) demonstrate that membranes harboring both NeuES-KpsCS and NeuES-KpsC assembled a large radioactive product, since most of the 14C label was retained in the wells. Incubation of the extracts with endo-N-acetylneuraminidase prior to PAGE almost completely eliminated the radioactive polymer in both samples (Fig. (Fig.3A,3A, lanes 2 and 4).
The results described above demonstrate that membrane fractions harboring only four K1 capsular cluster-encoded proteins, NeuES-KpsCS can assemble polysialic acid de novo from CMP-NeuNAc. We wanted to determine whether this in vitro result could also be observed in vivo. To investigate whether NeuES-KpsCS could also form polysialic acid in vivo we constructed a strain containing neuES-kpsCS plus neuA, which could convert free sialic acid from the medium into CMP-NeuNAc (31). We designed the plasmid pWN656, which contains K1 kpsCS and K1 neuA, and coexpressed it with pWN646, which encodes NeuES. We expressed these proteins in EV80(DE3). EV80 (36) is a ΔnanA K-12 derivative which is unable to catabolize the free sialic acid as a result of the N-acetylneuraminate lyase (aldolase) deletion. We introduced the DE3 lysogen in the EV80 background to facilitate expression of the T7 vectors.
We purified intracellular polysialic acid from EV80(DE3) cells harboring NeuES (pWN646) together with KpsCS-NeuA (pWN656) after growth in the presence of free sialic acid in growth medium. Analysis of the purified polysialic acid by PAGE suggested that the polysaccharide produced by NeuES-KpsCS(NeuA) (Fig. (Fig.3B,3B, lane 1) is similar in size to extracellular polysaccharide purified from the growth medium of the wild-type strain, EV36 (36) (Fig. (Fig.3B,3B, lane 5). Both polymers are high MW and labile to endo-N-acetylneuraminidase (Fig. (Fig.3B,3B, lanes 2 and 6).
To investigate the effect of KpsS protein on polysaccharide production in vivo, we designed a pACYC Duet-1-based vector (pWN674) which contains K1 kpsC and neuA genes and coexpressed it with the neuES-containing vector (pWN646) in an EV80(DE3) background. When grown under similar conditions as EV80(DE3) cells harboring NeuES-KpsCS(NeuA), NeuES-KpsC(NeuA)-harboring cells were able to produce a small amount of the long polymer (Fig. (Fig.3B,3B, lane 3). The quantity of the polymer produced in the absence of KpsS protein was 15 times less than in the presence of KpsS. KpsS has a similar effect on polysialic acid production in vivo and in vitro: it changes the quantity but not the length of polysialic acid chains. Figure Figure3B3B (lane 4) also demonstrates that polysaccharide produced in the absence of KpsS was labile to endo-N-acetylneuraminidase.
We sequenced the region between and overlapping neuC and neuS to determine the starting point of neuE. An ORF of 1,233 bp was observed which overlaps neuC by 92 bp and neuS by 8 bp. This ORF was contained on the plasmid pWN646 (Fig. (Fig.1),1), which was shown to support production of polysialic acid de novo when expressed in E. coli strain MS108 (Table (Table2).2). Closer analysis of the K1 neuE sequence revealed five possible start sites (Fig. (Fig.4).4). There are four in-frame ATG sites at positions bp 1, 61, 88, and 685 and one GTG site at 103 bp from the beginning of the ORF. We were unable to detect a strong RBS within 20 nucleotides of any of these start codons and therefore could not predict the translation start for the NeuE protein.
In order to determine the minimum functional size of the neuE gene, we designed a series of pET-46 Ek/LIC-based neuE plasmids, differing by the position of the translation start codon (Fig. (Fig.4).4). Note that the transcription start is located on the vector and is identical for the all of constructs described below. This fact suggests that the initiation rate of transcription in all these constructs should be similar.
Four plasmids (pWN649, pWN654, pWN657, and pWN659) were constructed such that the first of potential NeuE start codons were more than 100 bp downstream from the RBS of the expression vector. Therefore, protein expression should be initiated primarily from the RBS and start codon present in the K1 insert DNA. All plasmids have a C-terminal S-tag sequence to facilitate detection of the expressed gene product. We also constructed two plasmids (pWN658 and pWN663) that contained chimeric ORFs optimally positioned adjacent to the RBS and start codon of the expression vector (Fig. (Fig.44).
Each of these NeuE constructs was coexpressed with K1 NeuS-KpsCS (pWN662) in the HMS174(DE3) background to determine the level of de novo polysaccharide production resulting from their expression. As shown in Table Table7,7, all of the plasmids with translation start sites beginning at or before the ATG at position 88 supported the synthesis of polysialic acid de novo without exogenous acceptor. On the other hand, plasmids in which neuE lacks ATG 88 (pWN659 and pWN663) did not impart polysaccharide production activity, even when neuE was expressed from the strong RBS of an expression vector. These results indicate that the shortest neuE ORF that encodes functional protein starts at the ATG at position 88.
Although the ATG at position 88 may be the start of the shortest functional ORF, it is not necessarily the native start site for neuE. The plasmid pWN649 (Fig. (Fig.4)4) was specifically designed to include all potential translation start sites of the neuE ORF and about 200 bp of native DNA upstream of the first ATG. We theorized that this construct would allow the expression of all naturally occurring isoforms of NeuE. By comparing the level of expression and the size of the product(s) expressed from pWN649 and other pET-46 Ek/LIC-based neuE constructs, we expected to estimate the size of native NeuE. We used Western blot analysis to determine the size of the all proteins expressed in vivo from pET-46 Ek/LIC-based neuE constructs. We detected S-tag-labeled chimeras with S-protein-HRP conjugate. All NeuE chimeras were incorporated into the membrane fraction, as expected, since NeuE has a predicted transmembrane domain (27). As an internal size reference we used a double-tagged NeuE fragment encoded by pWN658 (Fig. (Fig.5,5, lane 1). The predicted mass of this recombinant protein is 49.2 kDa. Note that if the NeuE chimera encoded by pWN649 were expressed from the ATG at position 1, this protein would have a mass of about 51 kDa, which is slightly higher than the mass of the reference protein (49.2 kDa).
Bands of the same size migrating at positions corresponding to a mass slightly less than the 49.2 kDa of the reference protein were observed for membrane fractions of cells harboring pWN649 and pWN654 (Fig. (Fig.5,5, lanes 2 and 3). pWN654 contains a neuE ORF whose start codon at position 1 is followed by a stop codon, thus making the ATG at position 61 the first potential transcription start. This observation suggests that translation of the naturally occurring NeuE most likely begins at the ATG at position 61.
A very faint band of a size similar to pWN649 and pWN654 was observed when the ATG at position 61 was followed by a stop codon (pWN657), indicating a dramatic drop in the level of expression (expected mass for this protein is 47.5 kDa) (Fig. (Fig.5,5, lane 4). This finding suggests that the ATG at position 61 is the likely start point for native NeuE translation. The low level of protein expression from pWN657 may be sufficient to support formation of polysialic acid de novo when pWN657 is coexpressed with K1 NeuS-KpsCS (pWN662) (Table (Table77).
No product was detected when a stop codon was placed at position 88 (replacing the ATG) in front of the GTG start at position 103 (pWN659). This construct did not support formation of polysialic acid de novo when coexpressed with K1 NeuS-KpsCS (pWN662) (Table (Table7).7). Expression of the ORF beginning with the ATG at position 685 under control of the pET46 Ek/LIC RBS yielded a bright band migrating in the expected low-MW position (pWN663) (Fig. (Fig.5,5, lane 6). Membranes harboring this product together with NeuS-KpsCS could not synthesize polysialic acid de novo. Thus, even if this lower MW NeuE were expressed in vivo, it is unlikely that it participates in the polysialic acid synthesis initiation reaction.
Since the K1 and K92 gene clusters are very similar, we tested the functional equivalence of the K1 and K92 NeuE. K92 NeuE was able to support polysaccharide production activity of K1 or K92 NeuS equally well. Similarly, K1 NeuE was able to support polysaccharide production activity of K92 NeuS in the absence of acceptor when expressed in a KpsCS-positive background (results not shown).
The polysialic acid synthesis pathway in E. coli consists of three steps: (i) formation of the precursor, CMP-NeuNAc, (ii) polymerization of sialic acid, and (iii) export of the polymer to the cell surface. The polysaccharide is polymerized by the transfer of sialic acid from CMP-NeuNAc to the nonreducing end of the growing chain in a linkage-specific fashion. Polymerization is believed to be initiated by transfer to an endogenous acceptor molecule and would therefore imply two types of sialyltransferase reactions, (i) one in which NeuNAc is transferred from CMP-NeuNAc to a growing chain of polysialic acid and (ii) a second in which NeuNAc is transferred from CMP-NeuNAc to an endogenous acceptor. The nature of this acceptor remains unknown.
Although there are 14 genes in the E. coli K1/K92 gene cluster, only one gene, neuS, is known to encode a sialyltransferase. The neuS-encoded polysialyltransferase can only catalyze the elongation reaction in the absence of the other proteins encoded by the gene cluster (16, 27). In this study we identified the minimum combination of gene products required to observe the overall synthesis of polysialic acid de novo from CMP-sialic acid in vitro. Our results show that there are three genes necessary to observe de novo polysialic acid synthesis in E. coli K1 in vitro, the neuES of region 2 and kpsC of region 1. KpsS increases polysaccharide production by several fold (about 10 to 15 times) over that observed with neuES and kpsC. We think that it is unlikely the difference is due to the level of KpsC expression, since the constructs only differ by the presence or absence of a complete kpsS gene. Nevertheless, it is conceivable that KpsS could influence the stability of expressed KpsC protein. The size and repeating structure of the polysialic acid produced in membranes or cells lacking kpsS do not appear to be affected by the absence of this gene. Gel filtration and PAGE experiments showed that in the presence of [14C]CMP-NeuNAc three proteins (NeuES-KpsC) were able to assemble high-molecular-weight polymer in vitro. Intracellular polysaccharide isolated from a kpsS deletion mutant was structurally similar to wild-type K1 polysaccharide as determined by 1H and 13C NMR. We conclude that kpsC plays some role in de novo polysaccharide synthesis since region 2 gene products were not observed to polymerize sialic acid in the absence of kpsC.
Strains carrying neuES and kpsC or neuES and kpsCS also produce polysialic acid in vivo when supplied with a source of CMP-NeuNAc. We coexpressed NeuES-KpsCS proteins together with CMP-NeuNAc synthetase NeuA (31) and supplied free sialic acid in the medium. We purified polysaccharides produced in vivo during equivalent time periods by NeuES-KpsC and NeuES-KpsCS under these growth conditions. Polysaccharides isolated from these constructs were similar to each other in size and very similar to K1 wild-type extracellular polysaccharide. The only difference observed between NeuES-KpsC and NeuES-KpsCS samples was the decreased amount of polysaccharide produced in the absence of KpsS. Based on these experiments, the minimum combination of K1 capsule cluster genes required for de novo synthesis of polysialic acid from CMP-NeuNAc in vivo and in vitro is the same.
One of the interesting aspects of this study was the determination of the size of the neuE ORF and the conclusion that the NeuE protein plays some role in the initiation of polysialic acid synthesis. An acapsular phenotype has been reported for a neuE deletion mutant (35), implying that the product of this gene participates in capsule production. Since the original report of the neuE ORF sequence by Steenbergen et al. (27), other neuE sequences have been submitted to the data bank, suggesting that this gene might encode a larger protein than was presumed initially. Our results showed that the shorter version neuE did not support a polysialic acid synthesis initiation in vitro in a region 1- and 3-positive background. To determine the size of the neuE ORF we sequenced the region between and overlapping neuC and neuS. A large ORF was observed which overlaps both neuC and neuS genes. The fact that the plasmid containing the longer ORF could support de novo polysialic acid synthesis in the appropriate background suggested not only that the neuE ORF is longer than it was assumed before, but also that the product of this gene plays a role in de novo polysialic acid synthesis. Our results indicate that NeuE is the only region 2-encoded protein other than NeuS needed for synthesis of polysialic acid from CMP-NeuNAc. Our results established the minimal effective size of the neuE gene. We demonstrated that translation of an active protein could begin at ATG 88; however, the ATG 61 of the neuE ORF is most likely the native translation start in vivo.
In earlier reports KpsC and KpsS were suggested to a play role in phosphatidyl-KDO substitution of group II E. coli capsular polysaccharide (10, 11, 21-23). kpsC and kpsS mutants accumulate intracellular polysaccharide (23, 33), which in K5 lacks a phosphatidyl-KDO moiety at the reducing end of the chain (23). Because kpsC or kpsS mutations did not prevent production of K5 polysaccharide in vivo, it was proposed that phosphatidyl-KDO substitution occurs after initiation of synthesis (10, 38). In the absence of the exogenous acceptor we were unable to detect [14C]NeuNAc incorporation by the membranes harboring K1 NeuES KpsS in vitro, but a low level of synthesis by the membranes harboring K1 NeuES KpsC was detected. Thus, in our hands KpsC appears to be essential for de novo synthesis of polysialic acid.
In E. coli K5, KpsC and KpsS proteins have been shown to be associated with the inner membrane (21). Deletion of kpsS or kpsC in E. coli K5 results in failure of membrane targeting of both K5 glycosyltransferases KfiA and KfiC (21). We do not know how far we can draw parallels between K1 and K5 mechanisms of polysaccharide synthesis. The membrane localization of K1 and K92 polysialyltransferases does not depend on the presence of the other capsule proteins. However, the fact that K1 proteins NeuES were able to initiate polysialic acid synthesis without exogenous acceptor in the MS108 (K5) background suggests that the K5 and K1 KpsCS have similar functions in both strains.
The major focus of this report was to describe the genetic requirement for the polymerization reaction catalyzed by the polysialyltransferase. We have clearly demonstrated that membranes isolated from E. coli carrying neuES and kpsC are sufficient to observe synthesis of polysaccharide from the polysialyltransferase substrate CMP-NeuNAc. Exactly what reaction occurs prior to polymerization will not be known until an endogenous acceptor can be isolated. In preliminary experiments we have extracted a hydrophobic substance that can act as an acceptor for TOP10 membranes harboring NeuS. This hydrophobic substance has been extracted from a neuS neuB-negative K1 strain and is currently being characterized (J. Vionnet and W. F. Vann, unpublished results). Nevertheless, we have learned from the present experiments that NeuE and KpsC function along with the polysialyltransferase to form a polymer. Whether these two proteins function in the formation of endogenous acceptor or support the transferase reaction as part of a complex is not clear at this point. These experiments lead one to ask whether NeuE, KpsC, and KpsS interact with the polysialyltransferase or whether either of these participates directly in the catalytic reaction.
One cannot deduce the function of these proteins from sequence homology since homologs do not have established functions. Interestingly, the translated Neisseria meningitidis gene (NMB0065) shares homologous amino acid sequences with the NeuE gene (5, 6, 12) (Table (Table8).8). This gene or its homologs are present in sialic acid-producing N. meningitidis (serogroups B, C, Y, and W135) but are not present in serogroup A, which does not produce a sialic acid-containing polysaccharide (6). Transposon insertion into the N. meningitidis serogroup C 0065 gene resulted in an unencapsulated phenotype (12). It would be interesting to know if NMB0065 and NeuE play similar roles in polysaccharide production. While there are not many genes with significant homology to neuE, there is a larger group of the gram-negative bacteria (encapsulated and acapsular) possessing “kpsCS tandem,” a conserved pair of genes with significant homology to kpsC and kpsS and in the same direction of the transcription (Table (Table8).8). No definitive function has been assigned to any of the homologs of neuE, kpsC, and kpsS. In the future it would be interesting to investigate if the kpsCS of other organisms play a similar role in capsule polysaccharide initiation as in K1/K92 E.coli.
KpsC and KpsS were analyzed using the NCBI BLAST software and shown to share homologous sequences with two families of glycosyltransferases. It is conceivable that these proteins may be involved in the formation of acceptor molecules. Both proteins have been suggested (22, 23) to play a role in the addition of phosphatidyl KDO to the reducing end of E. coli K5 polysaccharide. Thus, one could speculate that transfer of KDO to a phospholipid by KpsC would yield a putative acceptor molecule. This would in turn allow the polysialyltransferase to begin de novo synthesis. If the interaction of the polysialyltransferase with this phospholipid acceptor were less than ideal, a low level of polymer synthesis would be expected. The role of KpsS could then be to improve this interaction and facilitate normal de novo polysialic acid synthesis. Whether this scenario actually occurs awaits further experimentation.
We thank Richard Silver at the University of Rochester, Rochester, N.Y., and Ian Roberts at Manchester University, United Kingdom, for providing plasmids and strains used in this study. We thank Eric Vimr for helpful discussions and Justine Vionnet for technical assistance.