Purification, cloning, and expression of a cytidine 5'-monophosphate N-acetylneuraminic acid synthetase from Haemophilus ducreyi.

An N-acetylneuraminic acid cytidylyltransferase (EC 2.7.7.43) (CMP-NeuAc synthetase) was isolated from a Haemophilus ducreyi strain 35000 cell lysate and partially characterized. The enzyme catalyzes the reaction of CTP and NeuAc to form CMP-NeuAc, which is the nucleotide sugar donor used by sialyltransferases. Previous studies have shown that the outer membrane lipooligosaccharides of H. ducreyi contain terminal sialic acid attached to N-acetyllactosamine and that this modification is likely important to its pathogenesis. Therefore, to investigate the role of sialic acid in H. ducreyi pathogenesis, the gene encoding the CMP-NeuAc synthetase was cloned using degenerate oligonucleotide probes derived from NH2-terminal sequence data, and the nucleotide sequence was determined. The derived amino acid sequence of the CMP-NeuAc synthetase gene has homology to other CMP-NeuAc synthetases and to a lesser extent to CMP-2-keto-3-deoxy-D-manno-octulosonic acid synthetases. The gene was cloned into a T7 expression vector, the protein expressed in Escherichia coli, and purified to apparent homogeneity by anion exchange, Green 19 dye, and hydrophobic interaction chromatography. The final step yielded 20 mg of pure protein/liter of culture. The protein has a predicted molecular mass of 25440.6 Da, which was confirmed by electrospray mass spectrometry (Mexpt = 25439.9 +/- 1.4 Da). The enzyme appears to exist as a dimer by size exclusion chromatography. In contrast to other bacterial CMP-NeuAc synthetases, the H. ducreyi enzyme exhibited a different substrate specificity, being capable of also using N-glycolylneuraminic acid as a substrate.

Sialic acid can be synthesized in bacteria by condensation of N-acetylmannosamine with either phosphoenolpyruvate (Ia) or pyruvate (Ib). The first reaction is catalyzed by NeuAc synthetase, which has been partially purified from N. meningitidis (27). This enzyme differs from the mammalian enzyme which synthesizes NeuAc-9-P (from N-acetylmannosamine-6-P and phosphoenolpyruvate), which is then dephosphorylated by a phosphorylase. Recently, it was demonstrated that NeuAc synthesis in E. coli K1 occurs only through NeuAc aldolase (Ib) (28). CMP-NeuAc synthetases catalyze the second step (II) in the pathway, forming the nucleotide-sugar donor used by sialyltransferases. These enzymes have been purified and characterized from several bacterial and animal sources (29 -33). A LOS-specific sialyltransferase has been detected in extracts of Neisseria sp. and partially characterized (34). To date, no bacterial LOS-specific sialyltransferase has been purified to homogeneity or cloned, although a recent report characterizing a sialyltransferase deficient mutant may soon lead to cloning of the sialyltransferase gene (35). In contrast, bacterial polysialyltransferases, the enzymes that synthesize polysialic acid capsules, have been cloned (36) as have several animal sialyltransferases (37). As a group, these enzymes may be attractive targets for therapeutic intervention considering the role of sialic acid containing molecules as virulence and/or adhesion factors. In this report we detail the isolation of the CMP-NeuAc synthetase from H. ducreyi, the cloning of its gene, and expression of the enzyme in E. coli using a T7lac promoter system. Moreover, sequence analysis of CMP-NeuAc and CMP-KDO synthetases reveal a number of residues and domains conserved throughout both enzyme classes.
Purification of CMP-NeuAc Synthetase from H. ducreyi Cell Lysate-The bacterial suspension (3 g of wet weight cells in 20 ml of buffer A) was adjusted to 2 mM EDTA and 0.2 mg/ml lysozyme and was stirred for 20 min at room temperature. Unlysed cells and cellular debris were removed by centrifugation at 13,000 ϫ g for 10 min. The supernatant was applied to two 5-ml EconoPac Q cartridges (Bio-Rad) connected in series and equilibrated in buffer A. The enzyme was eluted with 0.2 M NaCl in buffer A. The active fraction (11 ml) was applied to a 5-ml Cibacron Blue 3GA-Agarose dye column also equilibrated with buffer A. The column breakthrough (20 ml), which contained most of the enzyme activity, was concentrated to 0.5 ml with two Centricon-30 concentrators (Amicon) and applied to two Bio-Sil TSK-125 (Bio-Rad; 7.5 ϫ 600 mm each) size exclusion columns connected in series and equilibrated with 20 mM KH 2 PO 4 , 0.5 mM EDTA, 20% (v/v) glycerol, pH 7.0. The active fractions (4.5 ml) were pooled, 0.5 ml of 0.5 M Tris-HCl, pH 8.5, was added to increase the pH, and the sample applied to a MonoQ HR 5/5 column (Pharmacia Biotech Inc.; 5 ϫ 50 mm) equilibrated with 50 mM Tris-HCl, 20% (v/v) glycerol, pH 8.0. The enzyme was eluted with a KCl gradient (0.5 ml/min, 50 -200 mM KCl in 40 min) in the same buffer. The active fractions (3 ml) were pooled, concentrated to 0.4 ml with a Centricon-30 concentrator, and adjusted to 1.5 M (NH 4 ) 2 SO 4 . The sample was applied to a Hydropore-5-HIC column (Rainin; 4.6 ϫ 100 mm), equilibrated with 0.1 M KH 2 PO 4 , 0.5 mM EDTA, 1.2 M (NH 4 ) 2 SO 4 , pH 7.0 (buffer B), and eluted with a gradient of 0.1 M KH 2 PO 4 , 0.5 mM EDTA, pH 7.0 (buffer C) (0.5 ml/min, 1.2-0 M (NH 4 ) 2 SO 4 in 40 min). The active fractions (3 ml) were readjusted to 1.2 M (NH 4 ) 2 SO 4 and rechromatographed on the HIC column using a shallower gradient (0.4 ml/ min, 1.125-0.6 M (NH 4 ) 2 SO 4 in 52.5 min). The active fractions (2.4 ml) were pooled, concentrated, and desalted with a Centricon-30 concentrator to a volume of 0.5 ml and applied to a C4 reverse phase HPLC column (Vydac; 2.1 ϫ 150 mm). The C4 column was equilibrated with 0.1% trifluoroacetic acid, and the proteins were eluted with a gradient of 0.08% trifluoroacetic acid, 70% CH 3 CN over 70 min at a flow rate of 0.2 ml/min.
Amino Acid Sequencing-The three major peaks eluted from the HPLC C4 column were each subjected to NH 2 -terminal sequence analysis using an Applied Biosystems 470A gas-phase sequencer with an on-line ABI Model 130A phenylthiohydantoin analyzer. For the fraction identified as containing the CMP-NeuAc synthetase, 43 cycles were obtained in which the amino acid residue could be assigned with confidence. The first cycle produced 4 pmol of methionine with the yield dropping to 0.1 pmol phenylalanine by the 43rd cycle. Based on typical conversion yields (30 -40%), approximately 10 -13 pmol (0.25-0.33 g) of CMP-NeuAc synthetase was loaded on the sequencer.
Screening of the H. ducreyi Genomic DNA Library-The NH 2 -terminal amino acid sequence was reverse-translated employing the Compu-Gene program (38). Two degenerate oligonucleotides were constructed (see Table I). Oligonucleotide 1, a 23-mer containing a single inosine at position 15 and a degeneracy of 108, corresponds to the reverse translation of amino acids 1-8. Oligonucleotide 2 corresponds to the reverse translation of amino acids 14 -20 and is a 20-mer with a single inosine and a degeneracy of 48. Both oligonucleotides were prepared by the Protein Chemistry Facility at Washington University School of Medicine. A DASHII library of H. ducreyi 35000 DNA (39) was plated, plaques were transferred to Hybond-N, and DNA was immobilized by UV cross-linking (Stratalinker) employing standard methodologies. Oligonucleotide 1 was end-labeled with ␥-32 P using T4 polynucleotide kinase, and hybridization was performed overnight in 5 ϫ SSPE (1 ϫ SSPE ϭ 0.18 M NaCl, 0.01 M sodium phosphate, and 1 mM EDTA, pH 7.7), 5 ϫ Denhardt's solution, 0.5% SDS, and 20 g/ml Salmon testes DNA at 30°C. Following hybridization, the membrane was washed three times at 35°C in 2 ϫ SSPE-0.1% SDS. The membrane was then wrapped in Saran Wrap and autoradiographed. Phage plaques, which gave a positive hybridization signal, were picked and rescreened. Phage DNA was isolated from 10 ml of liquid lysates and characterized by restriction analysis.
Cloning and DNA Manipulations-The low copy number vector pWSK30 was employed for the initial subcloning (40). For subsequent studies, the DNA fragment containing CMP-NeuAc synthetase was amplified by 20 rounds of PCR using primers 3 and 4 (see Table I). The PCR product was cloned with the TA cloning kit into pCRII according to the manufacturer's instructions (Invitrogen). Plasmid vectors pT7-7 (41) and pET24b (Novagen) were employed for the preparation of the T7 expression constructs. Plasmid DNA was purified by using the Qiawell-8 plasmid kit (Qiagen, Chatsworth, CA).
DNA sequence was determined by the dideoxychain termination method employing Sequenase according to the manufacturer's directions (U. S. Biochemicals Corp.). Lasergene software (DNASTAR, Madison, WI) was employed for contig assembly and sequence analysis.
Protein Sequence Alignment-Protein sequence data bases (Gen-Bank, Release 92.0, Dec. 1995; SWISS-PROT, Release 32.0, Dec. 1995; PIR, Release 45.0, June 1995) were searched using the search program BLAST (42), and protein sequences were aligned with the PILEUP program from the Wisconsin Sequence Analysis Package, version 8.0 (Genetics Computer Group, Inc., Madison, WI). The similarity matrix was calculated using the scoring method in Ref. 43. In pairwise comparisons, the smaller of the two sequences was taken as the denominator.
Protein Expression-E. coli BL21(DE3) harboring the pET24 expres-sion construct containing the CMP-NeuAc synthetase (designated pMVT1) was grown at 30°C in LB broth containing 50 g/ml kanamycin. After reaching an A 600 between 0.3 and 0.8, the culture was induced with 0.1 mM isopropyl ␤-D-thiogalactopyranoside for 10 h at 30°C. The cells were harvested and frozen at Ϫ80°C until needed.
Purification of Recombinant CMP-NeuAc Synthetase-The cell paste (11 g) from a 3-liter culture was suspended in 40 ml of 50 mM Tris-HCl, 0.1 M NaCl, pH 7.5 (buffer D), and the cells were lysed by two passages through a French press (approximately 20,000 psi). The lysate was sonicated (four 30-s pulses at 25 watts with 2-min pauses) to shear nucleic acids and reduce its viscosity. The lysate was then centrifuged at 15,000 ϫ g for 90 min to remove cellular debris. The supernatant (39 ml) was applied to a DEAE-cellulose column (2.5 ϫ 13.5 cm) equilibrated with buffer D. After washing the column with 120 ml of buffer D, the enzyme was eluted with a gradient of 0.1-1.0 M NaCl of 270 ml. The active fractions were pooled (120 ml) and adjusted to 50% (NH 4 ) 2 SO 4 . The precipitate was pelleted, redissolved to 10 ml, and desalted into buffer D using a 15-ml Swift desalting column (Pierce). The sample was applied to a Reactive Green 19-Agarose dye column (2.5 ϫ 14 cm) equilibrated with buffer D. The column was washed with 120 ml of buffer D, and most of the enzyme was then eluted with a step gradient of 50 mM Tris-HCl, 0.5 M NaCl, pH 7.5. The sample (25 ml) was adjusted to 50% (NH 4 ) 2 SO 4 , and the precipitate was pelleted. The pellet was redissolved to 4.8 ml with buffer C, and then saturated (NH 4 ) 2 SO 4 (1.2 ml) was slowly added. The sample was centrifuged at 12,000 ϫ g for 30 min to remove particulates before loading on the HIC column equilibrated with buffer B. The enzyme was eluted with a gradient of buffer C (1 ml/min, 1.2-0.6 M (NH 4 ) 2 SO 4 in 40 min). Four separate runs were required to purify the entire sample. Fractions judged to be pure by SDS-PAGE were pooled (13.5 ml) and adjusted to 65% (NH 4 ) 2 SO 4 for storage as a suspension at 4°C.
CMP-NeuAc Synthetase Assay-CMP-NeuAc synthetase activity was assayed by the following procedure: 10 l of enzyme diluted in 0.1 mg/ml bovine serum albumin (BSA) was added to 90 l of assay mix and incubated at 37°C for 10 min. The reaction was stopped by adding 400 l of cold water and freezing the sample in dry ice. The final concentrations of components in the assay mixture were 1 mM CTP, 2 mM NeuAc, 20 mM MgCl 2 , 0.2 M MOPS, pH 7.1, 10 g/ml BSA, and 4 -10 milliunits/ml (nmol/min/ml) CMP-NeuAc synthetase. The assay solutions were thawed directly before analysis. A 20-l loop was completely filled by a 100-l injection on a Rainin (Woburn, MA) HPLC system. A nucleotide analysis column (Vydac; 4.6 ϫ 50 mm) was used to resolve CMP-NeuAc from the substrate CTP and any small amounts of CMP or CDP present. Solvent A was water and solvent B was 0.5 M ammonium formate (adjusted to pH 3.5 with formic acid). The column was operated at a flow rate of 2 ml/min, and the separation was complete in 2 min followed by a 1.5-min reequilibration cycle (Fig. 1). Quantitation of the CMP-NeuAc peak area was done using a standard curve of CMP (44). Detection was at 280 nm ( max of CMP and CMP-NeuAc at pH 3.5) using an Applied Biosystems 1000S diode array detector and was linear from 10 to 10,000 pmol. A unit is defined as the amount of enzyme that catalyzes the formation of 1 mol of CMP-NeuAc/min under the conditions of the assay.
For calculation of the K m for CTP and NeuAc, the enzyme was diluted with 0.2 M NaCl, 1 mM MgCl 2 , 1 mg/ml BSA to 1 and 4 g/ml, respectively. The diluted enzyme (35 l) was added to 315 l of assay mix preheated to 37°C and incubated at 37°C for up to 10.5 min. Each substrate was measured with the other held constant at a saturating concentration. NeuAc was varied from 0.25-8 mM with CTP kept at 1 mM, and CTP was varied from 0.025-0.8 mM with NeuAc at 10 mM. The final concentrations of the other components of the assay mixture were 20 mM MgCl 2 , 20 mM NaCl, 0.2 M MOPS, pH 7.1, 0.1 mg/ml BSA, and 0.1 or 0.4 g/ml CMP-NeuAc synthetase. Aliquots (100 l) were removed from the assays at 3.5 and 7 min and frozen in dry ice along with the final time point at 10.5 min. CMP-NeuAc product for each time point was quantitated as above, except that a 50-l loop was filled with a 100-l injection. A straight line was fit to the three time points, and the slope was taken as the initial velocity. The K m for CTP and NeuAc were determined by fitting the data directly to the equation v ϭ V max [S]/(K m ϩ [S]) using the computer program KinetAsyst (IntelliKinetics, State College, PA).
Protein Determination-Protein was determined using the Bio-Rad DC Protein Assay, which is a modification of the Lowry method (45), according to the manufacturer's instructions. BSA was used as a standard.
Size Exclusion Chromatography-To estimate the native molecular weight, purified enzyme (25-200 g) was applied to two Bio-Silect SEC 250 -5 columns (Bio-Rad; 7.8 ϫ 300 mm each) connected in series and equilibrated with 0.1 M KH 2 PO 4 , 0.15 M NaCl, pH 7.0. Elution was at 1 ml/min. The column was calibrated with molecular weight standards from Bio-Rad (thyroglobulin, 670000; IgG, 158000; ovalbumin, 44000; and myoglobin, 17500) and Sigma (albumin, 66000; carbonic anhydrase, 29000; cytochrome c, 12400; and aprotinin, 6500), both run separately from each other and the CMP-NeuAc synthetase. The CMP-NeuAc synthetase was exchanged into the column buffer with a 15-ml Swift desalting column, desalted with water using a Microcon-10 microconcentrator (Amicon), or resuspended from an ammonium sulfate pellet with the column buffer prior to chromatography.
Electrospray Mass Spectrometry-The CMP-NeuAc synthetase was desalted by gradient elution from a Vydac C4 reverse phase HPLC column or by repeated concentration and dilution with water using a Microcon-10 microconcentrator. A 5 l aliquot (Ϸ5 g) of this solution was injected into a stream of H 2 O/CH 3 CN (1:1) (with or without 1% acetic acid) at a flow rate of 20 l/min, which was coupled to the electrospray ionization source of a Fison Platform quadrupole mass spectrometer (Fison, Manchester UK). The mass spectrometer was run in the positive-ion mode and scanned from m/z 600-2000 in 8.1-s intervals. Horse heart myoglobin (average M r ϭ 16951.5) was used as a reference compound and external calibrant, and a typical mass accuracy of Ϯ 0.01% was obtained.

RESULTS AND DISCUSSION
The CMP-NeuAc synthetase from the H. ducreyi cell lysate was purified to homogeneity using a procedure consisting of anion exchange, dye, size exclusion, hydrophobic interaction, and reverse phase chromatography (Fig. 2). The final step using reverse phase chromatography resolved three major peaks, which were sequenced by Edman degradation. The CMP-NeuAc synthetase was identified by homology to other known CMP-NeuAc synthetases.
The relatively crude preparation of enzyme eluted from the first anion exchange column was stable to storage as a 65% ammonium sulfate suspension or 20% glycerol solution at 4°C or as a 50% glycerol solution at Ϫ20 and Ϫ80°C for at least several months. This material was used for initial characterization of the enzyme's properties. The enzyme has a rather broad pH optimum from 8 to 9.5, which is similar to other CMP-NeuAc synthetases (30,31). The enzyme has a requirement for Mg 2ϩ , with 20 mM giving the most activity under the assay conditions used. Mn 2ϩ is only 40% as effective at 20 mM, whereas Ca 2ϩ gave no reaction. The divalent cation requirements of other bacterial CMP-NeuAc synthetases are similar, whereas enzymes from animal sources are capable of using Ca 2ϩ and other divalent cations in addition to Mg 2ϩ and Mn 2ϩ (30,32). Sulfhydryl reagents are necessary for or greatly stimulate the enzyme activity of some other CMP-NeuAc synthetases (30, 31). However, dithiothreitol was found to have little effect on enzyme activity and so was not included in purifica- FIG. 1. HPLC chromatogram of an enzyme assay using a Vydac nucleotide analysis column. The product, CMP-NeuAc, was separated from CDP and CTP in under 2 min using a gradient of NH 4 HCO 2 / HCO 2 H. CMP (none present) elutes before CMP-NeuAc and is also baseline resolved. tion buffers or enzyme assays. Interestingly, N-glycolylneuraminic acid was also a substrate for the H. ducreyi CMP-NeuAc synthetase. N-Glycolylneuraminic acid can be used as a substrate by CMP-NeuAc synthetases from animal tissues but not by the enzymes from E. coli or N. meningitidis (31). The monosaccharide, KDO, was not a substrate.
In order to clone the gene for the H. ducreyi CMP-NeuAc synthetase, the amino-terminal sequence was reverse translated, and two degenerate oligonucleotides were constructed (Table I). Approximately 1200 clones from a DASHII library of strain 35000 DNA were screened with oligonucleotide 1, and seven hybridization-positive clones were identified. After plaque purification and rescreening, DNA was prepared from each clone and characterized by restriction analysis and by Southern hybridization. Two of the clones were identical, and the additional five clones had overlapping restriction maps. One clone was chosen for further characterization. H. ducreyi DNA fragments of 5.5 and 8 kb were identified in a NotI digest and cloned into NotI-digested pWSK30, and the mixture was transformed into E. coli DH5␣. Clones containing either the 5.5-or the 8-kb fragment were identified. Southern analysis using oligonucleotides 1 and 2 indicated that the H. ducreyi CMP-NeuAc synthetase gene was localized to the 8-kb fragment. A clone containing the 8-kb NotI fragment, designated pRSM1627, was saved. Oligonucleotide 1 was used as a sequencing primer and sequence was determined with Sequenase. Additional primers were constructed, and the complete DNA sequence of the H. ducreyi CMP-NeuAc synthetase gene (neuA) was determined in both directions. The position and direction of transcription of the H. ducreyi CMP-NeuAc synthetase gene in pRSM1627 was determined by PCR analysis (Fig. 3).
An open reading frame of 684 base pairs encoding a 228-

FIG. 2. Purification of the CMP-NeuAc synthetase from a H. ducreyi cell lysate.
The fractions containing enzyme activity are shaded. A, anion exchange chromatography of active fractions from the size exclusion column. B, hydrophobic interaction chromatography of active fractions from A. C, rechromatography of active fractions from B on the HIC column using a shallower gradient. D, the fractions containing enzyme activity from C were pooled, desalted, and loaded on a Vydac C4 reverse phase column. After Edman sequencing, the CMP-NeuAc synthetase was identified as peak 2 (indicated by an arrow) by its homology to other CMP-NeuAc synthetases. The DNA sequence of the H. ducreyi neuA gene was determined on both strands of pRSM1627. The initial sequence was determined with the degenerate oligonucleotide 1 as a sequencing primer, and the remainder of the sequence was determined by walking using primers synthesized as needed using the previously determined sequence. Amino acids 1-43 correspond to the sequence determined by Edman degradation. Underlined nucleotides correspond to an inverted repeat characteristic of a rho-independent transcriptional terminator.

TABLE I
Oligonucleotides employed in this study R ϭ A/G, Y ϭ C/T, H ϭ T/C/A, and I is inosine. The sequences in bold are the NdeI site in oligonucleotide 3 and the BamHI site in oligonucleotide 4.

5Ј-CGGGATCCTGGCAATTTCTTTCATCGTTAT
residue protein was identified (Fig. 4). An inverted repeat characteristic of a rho-independent transcriptional terminator was identified 63 nucleotides downstream of the termination codon. Putative promoter sequences were not identified. A T7 expression system was employed to generate large quantities of the H. ducreyi CMP-NeuAc synthetase. The gene was first amplified by PCR and recloned by TA cloning. The 5Ј PCR primer contained an NdeI site (Table I, oligonucleotide 3), and the 3Ј primer contained a BamHI site (Table I, oligonucleotide 4). The sequence of the clone in the TA vector was again determined to verify that there were no PCR errors. The CMP-NeuAc synthetase gene was initially cloned as a NdeI-BamHI fragment into pT7-7, which had been digested with the same enzymes. The construct was readily transformed into E. coli strains lacking a source of T7 polymerase; however, no transformants were obtained with competent BL21(DE3) cells, suggesting that the synthetase gene product was being expressed and was toxic to the cells. Transformants were obtained after cloning the gene as a NdeI-BamHI fragment into the more tightly regulated T7lac pET24 vector. Initial expression experiments at 37°C and inducing with 1 mM isopropyl ␤-D-thiogalactopyranoside showed the synthetase to be the most abundant protein in cell lysates and enzyme activity levels in a crude lysate to be several hundred times higher than from H. ducreyi lysates. However, a large portion of the enzyme (approximately 75%) was found in the insoluble cell pellet. Lower temperatures (20 -30°C) and isopropyl ␤-D-thiogalactopyranoside concentration (0.1 mM) favored more soluble enzyme, although insoluble CMP-NeuAc synthetase was still found in the lysate pellet under the best conditions identified.
The enzyme was the major component of the soluble cell lysate (Fig. 5). This greatly simplified the purification and allowed a homogeneous preparation of enzyme to be obtained in only three steps. For this, successive anion exchange, Green 19 dye, and hydrophobic interaction chromatography steps were used, and 20 mg of pure protein per liter of culture was obtained. In the final purification step a single, large peak was eluted from an HIC column using a shallow gradient (23.5 column volumes over 40 min) (Fig. 6). A summary of the purification of the recombinant enzyme is given in Table II. The enzyme ran as a single band by SDS-PAGE with a molecular mass of 25.6 kDa. The enzyme also appeared homogeneous by reverse phase chromatography. The calculated average mass of 25440.6 Da was confirmed by electrospray mass spectrometry with an accuracy to within 1.4 Da (Fig. 7). By size exclusion chromatography the enzyme's molecular mass was calculated as 47-74 kDa in 0.1 M KH 2 PO 4 , 0.15 M NaCl, pH 7.0 buffer and thus appears to exist as a dimer or possibly larger species. The concentration of enzyme affected the retention time, with higher concentrations shifting the peak to earlier retention times (higher molecular mass). Whether the enzyme was resuspended from an ammonium sulfate suspension or desalted with the column buffer or water, most of the enzyme eluted as a higher molecular weight species with little evidence for the presence of a monomeric form. However, if the enzyme was left at 4°C overnight in water or column buffer, both dimer and monomer (26 -28 kDa) were detected by size exclusion chromatography. Incubation overnight at 4°C in 0.2 M NaCl (no phosphate buffer) resulted in very little of the monomeric species. The retention time of the monomer showed little dependence on enzyme concentration. The rat liver enzyme was identified as a dimer by size exclusion chromatography (32). The E. coli enzyme is thought to be active as a monomer but did form dimers and larger aggregates in some buffers (47). The N. meningitidis enzyme also possibly exists as a dimer based on size exclusion chromatography results (29). The apparent K m for CTP and NeuAc were found to be 0.035 mM and 0.26 mM (mean of three experiments), respectively, at pH 7.1. The K m for NeuAc is similar to that measured for the N. meningitidis enzyme, but the K m for CTP is much lower than values obtained for previously described activities (see Ref. 30 for review). A possible reason for such a large difference is that most other CMP-NeuAc synthetases have been assayed at a higher pH where activity is greater, whereas the H. ducreyi enzyme was assayed at a more physiological pH.
There are now five sequences in the sequence data bases identified as CMP-NeuAc synthetases. A sixth sequence from Campylobacter coli is not identified as such but is highly similar (ptmB, U25992; identified as being involved in the posttranslational modification of flagellin). All six sequences are   (Table III). These three enzymes all have nearly the same molecular weight, whereas the E. coli and Streptococcus agalactiae enzymes are nearly twice as large.
The H. ducreyi CMP-NeuAc synthetase is 32% identical with the E. coli and S. agalactiae enzymes. Taken as a whole, the CMP-NeuAc synthetases have 32-65% identity and 50 -79% similarity with each other. The CMP-NeuAc synthetases also have homology to the four known CMP-KDO synthetase sequences (11-18% identity, 32-39% similarity). Although the specificity of these enzymes for their sugar substrate is high, they catalyze a very similar reaction in which CMP is transferred from CTP to either NeuAc or KDO; both of which are ␣-keto acid sugars (48 -50). These proteins do not contain the HiGH motif that is involved in nucleotide binding in the recently elucidated cytidylyltransferase superfamily (51). The highest homology is found in the NH 2 -terminal region. The two previously noted regions of highest similarity between the E. coli NeuA and KdsB proteins largely holds true for the larger set of sequences as well (residues 6 -10 and 46 -51 in H. ducreyi NeuA) (47). There are only a few residues conserved throughout in these two enzyme classes, and these are likely to be involved in catalysis and binding of the common substrate MgCTP (Fig. 8). A conserved arginine and lysine are found nine residues apart in the NH 2 terminus along with a third arginine or lysine in between. Chemical modification and site-directed mutagenesis of the E. coli CMP-NeuAc synthetase suggests that arginine and lysine are important for enzyme activity (52,53). In addition, there is almost a complete lack of negatively charged amino acids in this NH 2 -terminal region. There is also a third region of high similarity with a conserved glutamine, proceeded by several hydrophobic residues, and a conserved proline four residues apart (residues 98 -107 in H. ducreyi NeuA). The residues in between are conserved within each class but differ between the two classes ((P/V)TS and GDE for CMP-NeuAc and CMP-KDO synthetases, respectively). The NH 2 -terminal regions of CMP-NeuAc and CMP-KDO synthetases also share some homology with other nucleotidyltransferases that catalyze the general reaction: where NTP is a nucleotide triphosphate. The sequence alignment of several representative sugar nucleotidyltransferases is shown in Fig. 9. The conserved domains are aligned with a consensus sequence of ADP-glucose synthetases that is be- FIG. 7. Electrospray mass spectrum of the recombinant CMP-NeuAc synthetase. The mass of the enzyme (25439.9 Ϯ 1.4 Da) was within experimental error of the predicted molecular mass (25440.6 Da). In addition to the intact protein, two fragments (16440.1 ϩ 8997.7 ϭ 25437.8 Da) were detected. These species arose from a gas-phase fragmentation between residues Ile 150 and Pro 151 and not chemical hydrolysis, i.e., no addition of H 2 O was found in the NH 2 -terminal fragment, and the percentage of fragmentation was highly dependent on the cone voltage. a The abbreviations used are the same as in Fig. 8. b The percentage of identical residues is given for each pair of proteins from the sequence alignment shown in Fig. 8. The percentage of identical plus similar residues is given in parentheses. As a class, the CMP-NeuAc synthetases are 32-65% identical and share 50 -79% similarity. Likewise, the CMP-KDO synthetases are 37-65% identical and share 55-79% similarity. The two classes are 11-18% identical and 32-39% similar.
c The values for the E. coli K1 NeuA and the S. agalactiae CpsF sequence comparison are for residues 1-261. When the entire proteins were compared, the values were lowered to 27% identity and 44% similarity due to less homology in their carboxyl-terminal region. FIG. 9. Sequence alignment of conserved region of some nucleotidyltransferases. Representative nucleotidyltransferases from E. coli and Salmonella typhimurium are aligned with a region in ADP-glucose synthetase sequences that is thought to be involved in binding phosphorylated compounds. The residues found in the corresponding position in the nine CMP-NeuAc and CMP-KDO synthetases are indicated below the alignment. An X indicates more than four different residues at that position. The nucleotide sugar product of each enzyme is indicated to the right. lieved to be involved in phosphate binding (54). This domain, along with a second, were shown to be highly conserved in glucose-1-phosphate nucleotidyltransferases (55). The second domain is not found in CMP-NeuAc, CMP-KDO, or UDP-Nacetylglucosamine synthetases. Based on the conserved positive charges and high sequence homology in the NH 2 terminus, chemical modifications and site-directed mutagenesis experiments with the E. coli enzyme, and homology between CMP-NeuAc synthetases and other nucleotidyltransferases, it is quite possible that the NH 2 -terminal region is involved in binding CTP. There are several other conserved residues and domains within the CMP-NeuAc and CMP-KDO synthetases that are not shared between the classes and may be involved in sugar substrate specificity.
Regarding the role of CMP-NeuAc synthetase in H. ducreyi, it has been recently reported that many strains contain terminal lactosamine and sialylated lactosamine as part of their LOS (12,13). These epitopes are also a component of human blood group antigens and have important roles in molecular and cellular recognition (16,18). Understanding the function of these sugars in the interaction of H. ducreyi with host cells should lead to a better understanding of the pathogenesis of this organism. Recent studies have shown that H. ducreyi can adhere to and invade various human cells (56 -59), and LOS appears to play an important role in this process. Sialic acid may offer the bacteria more protection from the host immune defenses, but exposed terminal galactose may be necessary for the initial infection as has been suggested for N. gonorrhoeae LOS (60). An isogenic mutant of H. ducreyi incapable of producing sialylated LOS will enable a better understanding of the role of sialic acid and the terminal galactose in adhesion to and invasion of host cells and susceptibility of the organism to host defenses. Such a mutant is currently under construction.
The importance of sialylated molecules in general and as virulence factors in pathogenic bacteria in particular make understanding the biosynthesis of sialic acid containing molecules an area of great interest. Further study will hopefully better define the role of sialic acid and LOS in the pathogenesis of H. ducreyi. With large quantities of recombinant CMP-NeuAc synthetase now available, more detailed characterization of this enzyme can be accomplished, including screening conditions for crystallographic analysis. Along with the information gained from conserved residues and domains in the nine sequences of CMP-NeuAc and CMP-KDO synthetases now available, site-directed mutagenesis and active site labeling with reactive substrate analogs will hopefully lead to a better understanding of the residues necessary for substrate binding and catalysis in CMP-NeuAc synthetases.