Molecular Cloning and Functional Characterization of Components of the Capsule Biosynthesis Complex of Neisseria meningitidis Serogroup A

Background: The isolation of capsular polysaccharides from pathogenic bacteria for vaccine production is cost-intensive. Results: We describe the cloning, recombinant expression, and functional characterization of three enzymes from Neisseria meningitidis serogroup A that facilitate in vitro synthesis of the capsule polymer. Conclusion: The study presents a novel basis for efficient vaccine production. Significance: Economic vaccine production is prerequisite to combat meningococcal diseases. The human pathogen Neisseria meningitidis (Nm) is a leading cause of bacterial meningitis and sepsis globally. A major virulence factor of Nm is the capsular polysaccharide (CPS), which in Nm serogroup A consists of N-acetyl-mannosamine-1-phosphate units linked together by phosphodiester linkages [→6)-α-d-ManNAc-(1→OPO3−→]n. Acetylation in O-3 (to a minor extent in O-4) position results in immunologically active polymer. In the capsule gene cluster (cps) of Nm, region A contains the genetic information for CPSA biosynthesis. Thereby the open reading frames csaA, -B, and -C are thought to encode the UDP-N-acetyl-d-glucosamine-2-epimerase, poly-ManNAc-1-phosphate-transferase, and O-acetyltransferase, respectively. With the aim to use a minimal number of recombinant enzymes to produce immunologically active CPSA, we cloned the genes csaA, csaB, and csaC and functionally characterized the purified recombinant proteins. If recombinant CsaA and CsaB were combined in one reaction tube, priming CPSA-oligosaccharides were efficiently elongated with UDP-GlcNAc as the donor substrate, confirming that CsaA is the functional UDP-N-acetyl-d-glucosamine-2-epimerase and CsaB the functional poly-ManNAc-1-phosphate-transferase. Subsequently, CsaB was shown to transfer ManNAc-1P onto O-6 of the non-reducing end sugar of priming oligosaccharides, to prefer non-O-acetylated over O-acetylated primers, and to efficiently elongate the dimer of ManNAc-1-phosphate. The in vitro synthesized CPSA was purified, O-acetylated with recombinant CsaC, and proven to be identical to the natural CPSA by 1H NMR, 31P NMR, and immunoblotting. If all three enzymes and their substrates were combined in a one-pot reaction, nature identical CPSA was obtained. These data provide the basis for the development of novel vaccine production protocols.

Crucial to the success of vaccination programs in the sub-Saharan meningitis belt is the provision of safe high quality vaccines. MenAfriVac, a conjugate vaccine with CPSA coupled to tetanus toxoid as carrier protein, has been specifically designed to address these needs (6). With a cost of under 50 cents per dose (7), mass vaccination campaigns were possible in Burkina Faso, Mali, and Niger and installed herd immunity (8 -10) protecting not only vaccinated but also non-vaccinated individuals and young children (11).
Recent progress made with the cloning and functional expression of capsule polymerases (CPs) (12)(13)(14)(15) and pioneering studies that demonstrate the suitability of recombinant enzymes for the in vitro production of CPSs (16,17) have opened new perspectives for the economic and safe production of conjugate vaccines. The goal of our study, therefore, was to isolate the minimal number of enzymes needed for in vitro synthesis of immunologically active CPSA and to pioneer protocols for the use of recombinant enzymes in CPSA production chains. As the sugar building block UDP-ManNAc is commercially not available, it was clear from the start that establishing a successful production chain requires the in situ synthesis of UDP-ManNAc from cheap UDP-GlcNAc. Moreover, as the immunogenicity of CPSA depends on O-acetylation (18), an O-acetyltransferase capable to perform this modification was necessary.
The chromosomal locus cps (for capsular polysaccharide synthesis) contains the genetic information for CPS synthesis, modification, and surface transport. The locus is sub-struc-tured into six regions: A-D, DЈ, and E (Fig. 1A). The sequences encoded in region A are serogroup-specific and encode among other things the polymerases responsible for CPS synthesis (19). Regions B and C are highly conserved and encode the proteins necessary for export and assembly of the polysaccharide on the cell surface. In NmA, region A comprises four open reading frames (ORF) csaA, -B, -C, and -D (previously designated sacA-D or mynA-D). Using insertion-mutagenesis Swartley et al. (20) demonstrated that each of these genes is involved in the production of the NmA capsule. In a later study the gene product encoded in csaC was shown to be an acetyltransferase with specificity for the O-3 and O-4 positions in ManNAc (21) (Fig. 1B). Based on their nucleotide and predicted amino acid sequence, csaA was presumed to encode a UDP-N-acetyl-Dglucosamine-2-epimerase and csaB to encode a capsule polymerase (20) (Fig. 1B). Additional evidence that the product of csaB is in fact the NmA-specific capsule polymerase arose from the demonstration that the protein is part of the stealth family comprising exclusively D-hexose-1-phosphate transferases (22).
Here we describe the molecular cloning of the genes csaA, csaB, and csaC from NmA, the production of recombinant proteins, and the characterization of their functional properties. Testing a series of synthetic primer compounds, a ManNAc dimer linked together by phosphodiester linkages and carrying a phosphodiester at the reducing end was found to be the minimal acceptor structure. This artificial primer as well as oligosaccharide primers isolated from natural sources was preferen- tially used by CsaB if presented in non-O-acetylated form. Using a two-step production protocol, O-acetylated CPSA was synthesized by enzyme-catalyzed reactions and purified to homogeneity. Identity with the natural polymer was confirmed by 1 H and 31 P NMR and immunoblotting.

EXPERIMENTAL PROCEDURES
General Cloning-The genomic DNA isolated from Nm strain Z2491 was a kind gift from Dr. Heike Claus (Institute for Hygiene and Microbiology, University of Würzburg). The csaB sequence was codon-optimized for use in Escherichia coli BL21(DE3) using the Gene Designer software package (DNA 2.0) (23) and the codon frequency tables published by Welch et al. (24). The mean codon frequency for each amino acid was calculated from the codon frequency tables FreqA and FreqB (24), and the resulting codon frequency table was used as the template for the in silico generation of csaB co . csaB co flanked 5Ј by a BamHI site and 3Ј by a XhoI site was synthesized from Eurofins MWG Operon. All other csaA-C sequences described herein were amplified by polymerase chain reaction (PCR) using the primers shown in Table 1 and genomic DNA from Nm strain Z2491 or csaB co as template. PCR products were cloned via the restriction sites shown in Table 1 into the corresponding sites of the vector pET22b-Strep (25) driving the expression of recombinant proteins under the control of the T7 promoter. PCR products digested with BglII were cloned into the BamHI site of pET22b-Strep.
Expression and Purification of Recombinant CsaA, CsaB, and CsaC-Freshly transformed E. coli BL21(DE3) were grown at 15°C in PowerBroth medium for 18 h. At an optical density of A 600 ϭ 1.0 protein expression was induced by the addition of 0.1 mM isopropyl-␤-D-1-thiogalactopyranoside and allowed to proceed for a period of 20 h. In test expressions, 0.2 ml of culture-volume were pelleted with 16,000 ϫ g for 1 min. Cell pellets were lysed with 0.1 ml of lysis buffer (50 mM Tris, pH 8.0, 2 mM EDTA, 0.1 mg/ml lysozyme). The lysis was intensified by 3 cycles of sonication (Branson sonifier 450, 100% amplitude) interrupted by 3 min of cooling on ice. Soluble and insoluble fractions were separated by centrifugation (16,000 ϫ g, 30 min, 4°C), and the supernatant was mixed (1:1) with Laemmli buffer and used for PAGE as described below.
SDS-PAGE and Immunoblotting-SDS-PAGE was performed under reducing conditions using 2.5% (v/v) ␤-mercaptoethanol and 1.5% (w/v) SDS. Proteins were stained using Roti-Blue (Carl Roth GmbH) according to the manufacturer's guidelines. For Western blot analysis samples and standard proteins were blotted onto PVDF membranes (Millipore). Histagged proteins were detected with 0.5 g/ml anti-penta-His antibody (Qiagen) and goat anti-mouse IR680 or goat antimouse IR800 antibody (LI-COR) as second antibody. Second antibodies were used in a 1:20,000 dilution.
Preparation of CPSA Oligosaccharides-CPSA oligosaccharide samples with an averaged degree of polymerization (avDP) of 6 and 15, respectively, were generated by acidic hydrolysis of long CPSA chains isolated from bacterial cultures (CPSA n ). Solutions containing 2.5 mg/ml CPSA in sodium acetate buffer (50 mM sodium acetate, pH 4.8) were incubated at 73°C for 6 h, and 2 pool fractions (avDP 6 and 15, respectively) were purified by anionic exchange chromatography (Q-Sepharose column, GE Healthcare) using a sodium chloride gradient. The avDP and the dispersion of saccharide chains was determined by 31 P NMR and high performance anionic exchange chromatography-pulsed amperometric detection (HPAEC-PAD) analysis following an established protocol (26). If used in enzymatic reactions, hydrolyzed CPSA (CPSA hyd ) was dephosphorylated (CPSA hyd -deP) using acid phosphatase (Sigma) according to the manufacturer's guidelines.
Chemical Synthesis of CsaB Acceptors-A short summary of the synthesis of ManNAc and ManNAc derivatives as well as of ManNAc disaccharide units linked together by phosphodiester linkages is provided in the supplemental schemes S1 and S2. A manuscript describing the detailed chemical synthesis and the characterization of these compounds is under preparation. 5 Activity Testing of CsaA/CsaB by Use of a Radioactive Assay System-CsaA/CsaB activity was analyzed using an adaptation of a radioactive incorporation assay previously described for the N-acetylglucosamine-1-phosphate transferase from NmX (17). Briefly, assays were carried out with 5 l of the soluble fractions of bacterial lysates expressing either recombinant CsaB or CsaA (see Fig. 2C) or with purified and epitope-tagged proteins (112 pmol of StrepII-CsaA-His 6; 88 pmol of ⌬69CsaB co -His 6 ) in a total volume of 25 l of assay buffer (50 mM Tris pH 8.0 or various pH for determination of the pH optimum). Divalent cations were added from stock solutions. The reaction was primed with 5 ng of avDP15 and started by the addition of 0.05 mol of UDP-GlcNAc (Calbiochem) containing 0.05 Ci of UDP-[ 14 C]GlcNAc (American Radiolabeled Chemicals). Samples were incubated at 37°C, and 5-l aliquots were spotted onto Whatman 3MM Chr paper after 0, 5, 10, and 30 min. After descending paper chromatography, the chromatographically immobile 14 C-labled CPSA was quantified by scintillation counting.
Physicochemical Analysis of CPSA iv -To produce sufficient CPSA (CPSA iv ) for PAGE and NMR analyses, 0.84 nmol (1.2 M final) of StrepII-CsaA-His 6 and 37.5 pmol (50 nM final) of ⌬69CsaB co -His 6 in reaction buffer (50 mM Tris, pH 8.0, 20 mM MgCl 2 ) were incubated with 5 mM UDP-GlcNAc and 6.8 g of CPSA hyd of avDP6 after de-O-acetylation and dephosphorylation (CPSA hyd(deOAc) -deP). The total reaction volume was adjusted to 750 l. In control 1 shown in Fig. 6B, StrepII-CsaA-His 6 was also used at 50 nM. Reactions as well as control samples (containing reactants as indicated in Fig. 6) were incubated overnight at 37°C. 5 l of each sample were then used for sep-  B, to select the construct most suited for the production of active recombinant CsaB, the wild type and a codon-optimized (CsaB co ) version of the CsaB sequence were cloned (full-length or after N-terminal truncation, ⌬69) to produce proteins with tags on both (StrepII and His 6 ) or only one end (His 6 ), as indicated. Transformed bacteria were lysed, separated into soluble (s) and insoluble (i) fractions, and fractions were separately run on PAGE. After transfer onto nitrocellulose, the blot was developed with an anti-penta-His antibody. m, marker. C, soluble fractions were used to measure CsaB activity in a radioactive incorporation assay. D, Coomassie-stained gel demonstrating the purification result for ⌬69-CsaB co -His 6 (IMAC, immobilized metal ion affinity chromatography; SEC, size exclusion chromatography). aration on high percentage (25%) PAGE and visualized by a combined Alcian blue/silver staining procedure (27).
The residual sample was freeze-dried, solubilized in 0.75 ml of deuterium oxide (D 2 O, 99.9% atom D; Aldrich) to give a concentration of 0.5-1 mg/ml saccharide and used for product characterization by NMR. All the 1 H and 31 P NMR experiments were recorded as previously described (17).
HPLC-anion exchange chromatography (AEC) was performed on a Prominence UFLC-XR (Shimadzu) equipped with a CarboPac PA-100 column (2 ϫ 250 mm, Dionex). Samples were separated as described by Keys et al. (28) with the minor adjustment that H 2 O and 1 M NaCl were used as mobile phases M 1 andM 2 , respectively. 5 l of the samples were loaded for the detection of nucleotides at 280 nm and 50 l for the detection of CPSA at 214 nm. Products were separated using an elution gradient consisting of a Ϫ2 curved gradient from 0 to 30% M 2 over 4 min followed by a linear gradient from 30 to 84% M 2 over 33 min. Enzyme concentrations were used as indicated in Fig. 6. All other reactants were used in the amounts described above.
Analysis of 2-Acetamidoglucal-Assignments of 1 H NMR spectrum were in agreement with those reported in literature (29). 1  In Vitro Synthesis, Purification, and Immunological Analysis of CPSA iv and CPSA iv(OAc) -To generate CPSA iv , 10 nmol of CsaA and 16 nmol of CsaB were incubated overnight at 37°C in reaction buffer (50 mM Tris, pH 8.0, 20 mM MgCl 2 ) with 10 mM UDP-GlcNAc in a total volume of 9 ml. The reaction was primed with 1 g of CPSA hyd(deOAc) -deP of avDP6. Acetylation of 1 mg of CPSA iv was performed for 4 h at 37°C in the presence of 1.2 nmol of CsaC and 14 mM acetyl-CoA (Sigma) in a total volume of 0.5 ml of acetylation buffer (25 mM Tris, pH 7.5, 50 mM NaCl). Both CPSA iv and CPSA iv(OAc) were purified via AEC using a Mono Q HR5/5 column (GE Healthcare) at a flow rate of 1 ml/min and a linear sodium chloride gradient. CPS containing fractions eluting at 540 mM NaCl were pooled, dialyzed (ZelluTrans, Roth, 1 kDa MWCO) against water, and freezedried for further analysis. For dot blot analyses, small aliquots of the purified CPSA iv and CPSA iv(OAc) were spotted onto nitrocellulose (Whatman) and incubated with mAb 932 specifically directed against CPSA OAc (mAb 935 was generated in the laboratory of Prof. Dr. D. Bitter-Suermann, Hannover Medical School, Institute for Medical Microbiology, and was kindly provided for this study) in a 1:10,000 dilution. Dot blots were developed with goat anti-mouse IR800 antibody (LI-COR) in a 1:20,000 dilution.

RESULTS
Cloning and Expression of csaA, csaB, and csaC; Production of Recombinant Proteins-Because previous analyses carried out on these genes provided strong evidence that csaA, csaB and csaC encode the UDP-GlcNAc-epimerase, the poly-ManNAc-1-phosphate-transferase (20), and the O-acetyltransferase (21), respectively, primers were constructed (see "Experimental Pro-cedures") to amplify these ORFs. The genomic DNA isolated from Nm strain Z2491 was used as a template. Obtained PCR products were cloned into the pET22b-Strep vector (25), allowing the expression of recombinant proteins with N-terminal StrepII-and/or C-terminal His 6 tag. After transformation into BL21(DE3) and induction of protein expression (see "Experimental Procedures"), the distribution of recombinant proteins between the soluble (s) and insoluble (i) fraction of bacterial lysates was analyzed by Western blotting against the affinity tags. The recombinant epitope tagged forms of CsaA and CsaC appeared mostly in the soluble fraction and could be purified directly from the bacterial lysates. CsaA was purified by immobilized metal ion affinity chromatography followed by a desalting step and yielded 40 mg of protein/liter expression culture. Although some additional faint bands were visible in Coomassie-stained SDS-PAGE ( Fig. 2A), a protein fraction highly enriched in CsaA was obtained. CsaC was purified following the protocol described by Gudlavalleti et al. (21) and yielded 96 mg of homogenously pure protein/ liter of culture ( Fig. 2A).
Similarly, StrepII-CsaB-His 6 , encoding the putative poly-ManNAc-1-phosphate-transferase, was well expressed, and the excess of the construct appeared in the soluble fraction. However, the major product revealed with the anti-penta-His antibody in Western blot migrated with an apparent molecular mass of 50 kDa, strongly deviating from the calculated molecular mass of 67 kDa (Fig. 2B, left lanes). Because faint signals with molecular masses of Ͼ50 kDa were additionally displayed with the anti-penta-His antibody, we concluded that StrepII-CsaB-His 6 is either prone to N-terminal degradation or translated from an alternative start codon. Consequently, we reinvestigated the NmA genome with bioinformatics techniques. Indeed, two of the used gene prediction softwares (GeNmark and GeNmarkS; Refs. 30 and 31) retrieved an additional ATG (starting with position 183528 of the NmA genome (NC_ 003116.1)). In PRODIGAL (32), the prediction for this second start codon was comparable with the published start codon (base no. 183321; Ref. 33).
To investigate if translation from the alternative ATG leads to a stable protein, the corresponding truncation ⌬69CsaB was cloned with (StrepII-⌬69-CsaB-His 6 ) and without (⌬69-CsaB-His 6 ) the N-terminal StrepII-tag. Test expressions in BL21(DE3) demonstrated the occurrence of proteins of the expected molecular masses, but in repeated experiments the level of expressed protein was significantly lower than for the full-length construct. Moreover, to our surprise, the construct cloned with free N terminus (⌬69-CsaB-His 6 ) was routinely higher expressed than the StrepII-tagged construct (Fig.  2B). Because rare codons that exist in the CsaB sequence may negatively impact protein expression, csaB was codon-optimized using the Gene Designer (DNA 2.0) software (23) and the codon frequency tables published by Welch et al. (24). Expression of the resulting optimized gene (see supplemental data) was tested with the constructs StrepII-CsaB co -His 6 and ⌬69-CsaB co -His 6 . Although increased degradation and concomitantly reduced expression was seen for StrepII-CsaB co -His 6 , ⌬69-CsaB co -His 6 was well expressed, and no degradation was detectable in Western blot with the anti-penta-His antibody (Fig. 2B).

The Capsule Biosynthesis Machinery of NmA
Consecutively, enzymatic activity within the soluble fractions of the bacterial lysates was determined with a radioactive incorporation assay previously developed for the poly-GlcNAc-1-phosphate-transferase from NmX (17). The fractions containing the recombinant CsaB variants were tested in the presence of CsaA and UDP-[ 14 C]GlcNAc. In accordance with the levels of expressed protein (Fig. 2B), StrepII-CsaB-His 6 and ⌬69-CsaB co -His 6 showed identical activity profiles (Fig. 2C). Based on these results the protein variant ⌬69-CsaB co -His 6 was chosen for further experiments. The protein was purified from the soluble fraction of transformed BL21(DE3) by immobilized metal ion affinity chromatography and size exclusion chromatography, yielding 60 mg of highly pure protein from 1 liter of bacterial culture (Fig. 2D).
Optimization of Test Conditions and Characterization of CsaB Substrates-As the donor sugar (UDP-ManNAc) used by CsaB must be produced in situ in the epimerase reaction catalyzed by CsaA, the optimization of test conditions needed the presence of both enzymes. As for CPs of other Nm strains, a hydrolysate of CPSA (CPSA hyd ) was used to prime the reaction in the presence of CsaA and its substrate UDP-[ 14 C]GlcNAc.
Initial studies carried out to evaluate pH and salt conditions showed the best activity values in the presence of 10 -20 mM MgCl 2 and a pH between 8.0 and 8.5. Replacement of Mg 2ϩ by Ca 2ϩ or Mn 2ϩ inactivated the enzyme. Although these results were similar to what we had seen with CsxA (CP of NmX), the CsaA/CsaB reaction, in contrast to the CsxA reaction, did not show sensitivity to DTT (up to 2 mM were tested).
The natural CPSA is O-acetylated in positions 3 and to a minor extent also in position 4 of ManNAc (21,34,35). We, therefore, questioned whether ⌬69-CsaB co -His 6 recognizes and elongates acetylated and non-acetylated CPSA hyd with the same efficiency. Hydrolysis of CPSA was carried out before and after base treatment to obtain O-acetylated (CPSA hyd(OAc) ) and de-O-acetylated (CPSA hyd(deOAc) ) shorter saccharide chains. Knowing that CPSA hydrolysis results in a large distribution of saccharide chain lengths (ranging in size between DP 1 and 70), AEC was used to separate two fractions, the avDP6 (comprising DP1-DP10) and avDP15 (comprising DP10 -DP70). Both fractions were used to prime the enzyme reactions as indicated in the subsequent experimental steps.
To quantitatively assess the enzyme reactions, we adapted a spectrophotometric assay previously designed to analyze the CP from NmB (12). In the multienzyme assay shown in Fig. 3, the ⌬69-CsaB co -His 6 catalyzed product formation is coupled to NADH consumption, which can be continuously monitored at 340 nm. Using this assay the activity of ⌬69-CsaB co -His 6 was determined with CPSA hyd fractions (CPSA hyd(OAc) and CPSA hyd(deOAc) ) of avDP6 and avDP15 (Fig. 4). Moreover, because it was demonstrated that the hydroxyl groups at position 6 (C 6 -OH) on the non-reducing end sugar in CPSA hyd is blocked by phosphomonoesters (36), both fractions were additionally tested after treatment with acid phosphatase (deP)  . Acceptor recognition by CsaB. A, the chemical properties of the primers used to test the acceptor preference of ⌬69-CsaB co -His 6 are displayed. B, ⌬69-CsaB co -His 6 activity was followed using the spectrophotometric assay in the presence of CPSA hyd of avDP6 and avDP15 in either native O-acetylated form (CPSA hyd(OAc) ) or after de-O-acetylation (CPSA hyd(deOAc) ). Because earlier studies showed that the non-reducing ends in CPSA hyd are phosphorylated, samples were additionally tested before and after phosphatase treatment (deP). Samples designated with (deOAc) -deP were the subject of de-acetylation and de-phosphorylation.
to remove this group. Independent of the size of the primers used to start the reaction, the native acetylated oligomers (CPSA hyd(OAc) of avDP6 and avDP15) were found to be poor acceptors, but activity increased considerably after removal of acetyl groups and even further after release of the capping phosphate residue, making CPSA hyd(deOAc) -deP the most efficient acceptors. The size of the priming polymers was not significant for enzymatic activity (Fig. 4; compare avDP6 and avDP15). The obtained results allowed the conclusion that the chain elongation by CsaB proceeds via the non-reducing end and by transfer of ManNAc-1P onto C 6 -OH groups. Furthermore, because CPSA hyd(deOAc) was a better acceptor than CPSA hyd(OAc) , it is likely that O-acetylation takes place after polymer synthesis.
Because no information on the minimal length of the priming acceptor for CsaB could be derived from the CPSA hyd fractions, we used well characterized synthetic compounds to interrogate this question. The compounds synthesized are shown in Fig. 5 and varied not only in length but also with respect to O-acetylation (compounds 2, 4, 6 were 3-O-acetylated) and reducing end modifications (37)(38)(39). In compounds 1, 2, 5, and 6 the reducing ends were occupied by an n-decyl-phosphate ester, whereas a methyl group (OMe) was present in compounds 3 and 4. The ⌬69-CsaB co -His 6 activity did not go beyond background (no acceptor) with compounds 1-4 but, intriguingly, steeply increased with disaccharides carrying an n-decyl-phosphate ester at the reducing end (compounds 5 and 6) (Fig. 5). With the non-O-acetylated compound 5, activity values similar to those obtained with the optimized acceptor CPSA hyd(deOAc) -deP were measured. In line with the above data (Fig. 4), O-acetylation of compound 5 (resulting in compound 6) reduced the quality of the acceptor. Based on these data, the minimal acceptor recognized by ⌬69-CsaB co -His 6 could be defined as the dimer of ManNAc units linked together by phosphodiester linkages. The presence of a phosphodiester at the reducing end seems obligatory, because compounds ending with OMe groups do not work as acceptor substrates.
In Vitro Synthesis of CPSA Chains-To analyze if long CPSA chains can be produced with the recombinant enzymes, test reactions were carried out in the presence of StrepII-CsaA-His 6 , ⌬69-CsaB co -His 6 , and CPSA hyd(deOAc) -deP of avDP6 as the priming oligosaccharide. Reactions were started by the addition of UDP-GlcNAc. Control reactions (Fig. 6A, controls   FIGURE 6. In vitro synthesis of CPSA. A, products synthesized in the CsaA/CsaB reaction in the presence of UDP-GlcNAc and CPSA hyd(deOAc) of avDP6 were analyzed by high percentage PAGE and a combined Alcian blue/silver staining. Long chains were produced in the presence of all reactants (reaction) and, although in small amounts, also in control 1 where no priming oligosaccharides were added. The production of long CPSA chains in control 1 argues for the capacity of CsaB to start the polymerization de novo. B, 31 -4), in which components were omitted as indicated, were carried out in parallel. After overnight incubation, samples were loaded onto high percentage PAGE and developed by Alcian blue/silver staining. CPSA hyd(deOAc) -deP of avDP15 was loaded as the size marker. In the presence of CsaA and CsaB the added oligosaccharide primers were efficiently elongated to long polymer chains (Fig. 6A). 31 P NMR showed the phosphodiester signal, which is characteristic for CPSA as well as the signal that indicates the second reaction product UMP (Fig. 6B,  reaction). Unexpected were the signals observed at Ϫ5.8 ppm and Ϫ9.5 ppm, which indicated the formation of UDP. As these signals were most prominent in control 4 (Fig. 6B), with only StrepII-CsaA-His 6 and UDP-GlcNAc present, we speculated that UDP is a side product of the epimerase reaction. A 1 H NMR analysis carried out with control 4 revealed a signal that, in perfect agreement with published data (29), represented the anomeric proton of 2-acetamidoglucal (Fig. 6C). 13 C, 1 H heteronuclear multiple quantum coherence (see "Experimental Procedures") NMR analyses allowed us for the first time to assign a 13 C spectrum of this intermediate. The complete absence of signals for the anomeric proton of ManNAc (both ␣ and ␤) confirmed that UDP is an intermediate of the CsaA reaction and not produced by hydrolysis of UDP-ManNAc.
Another interesting observation was the de novo start of CPSA chains by ⌬69-CsaB co -His 6 if incubated with equimolar CsaA concentrations (50 nM, see "Experimental Procedures"). The in vitro produced CPSA chains (CPSA iv ) could be doubtlessly identified with both PAGE and 31 P NMR (Fig. 6, A and B,  control 1). However, when CsaA was used in excess of CsaB, none of the methods detected any CPSA signal, and the majority of the donor substrate was converted to UDP and 2-acetamidoglucal (data not shown). Based on these data, it can be concluded that the de novo activity is strongly disadvantaged compared with the elongation of oligosaccharide primers. Although the same type of de novo synthesis was shown for the CP of NmX (17), it came as a surprise in the case of CsaB, as the monosaccharide compounds 1 and 2 were not elongated (see Fig. 5).
To reduce UDP-ManNAc hydrolysis and simultaneously guarantee complete incorporation of ManNAc-1P into the product, the ratio CsaA:CsaB was varied as indicated in Fig.  6D. The reactions were carried out as described above with CPSA hyd(deOAc) -deP of avDP6 as primer. After overnight incubation, products were separated by HPLC and recorded at 214 nm (CPSA iv ) and 280 nm (UDP and UMP). As long as the concentration of CsaB was equal to or higher than the concentration of CsaA, no hydrolysis of UDP-ManNAc was detectable (Fig. 6D), not even if the enzymes were present in only 50 nM concentration and the donor sugar was not completely consumed. These data convincingly show that in vitro CPSA production with high efficiency is possible with the recombinant enzymes.
With the intention to obtain bio-identical CPSA in milligram amounts, the CsaA/CsaB reaction was up-scaled. Long polymers were purified by AEC using a protocol similar to the one described in Fiebig et al. (17). CPSA iv eluted at 540 mM NaCl clearly separated from all other reaction components (Fig. 7A). 1 H NMR spectra, comparatively recorded for CPSA iv and nat-ural CPSA (CPSA n ) treated with alkaline to remove the acetylgroup (CPSA n(deOAc) ), were in congruence (Fig. 7D). 1 mg of the purified CPSA iv was then used for acetylation with recombinant CsaC. To enable acetylation at both O-3 and O-4, the acetyl donor acetyl-CoA was added in a 2-fold molar excess of ManNAc. After a 4-h reaction step, CPSA iv(OAc) was purified by a second AEC step. The single peak that eluted at 540 mM NaCl (Fig. 7B) contained material that in dot blot analysis was recognized by mAb 932, specifically directed against the O-acetylated form of CPSA (Fig. 7C). Based on the signal intensity of the H-2 in 1 H NMR, 88% of the available O-3 groups were acetylated. This value is in perfect agreement with the value obtained for the reference CPSA n (Fig. 7D). However, whereas CPSA n to a minor extent also carried O-4 acetyl groups, this modification was not detectable in CPSA iv (OAc) .
To determine on the analytical scale if acetylated, bio-identical polymer can be produced in a one-step reaction starting with one substrate and just using the polymerase, we incubated CsaB in the presence of 3-O-acetylated UDP-ManNAc (UDP-3OAcManNAc) and analyzed the reaction products by PAGE and 31 P NMR. However, no product signals could be detected by any of the methods (Fig. 8, A and B) and 31 P NMR revealed that UDP-3OAcManNAc was not used by CsaB.
Finally we explored if O-acetylated CPSA can be produced in a one-pot reaction. Therefore, the reaction mixture containing CsaA, CsaB, and UDP-GlcNAc was supplemented with recombinant CsaC and acetyl-CoA. Moreover, control reactions with single compounds missing (see Fig. 8, C and D) were carried out in parallel. After overnight incubation, products were analyzed by Alcian blue/silver-stained high percentage PAGE (Fig. 8C) and immunoblotting with mAb 932 (Fig. 8D). In the presence of all components, a product recognized by mAb 932 was produced (Fig. 8D), indicating that CsaC can acetylate the produced CPSA in situ. The control reactions carried out in this experiment provided clear evidence for the functional nature of CsaA and CsaC being UDP-GlcNAc/UDP-ManNAc epimerase and O-acetyltransferase, respectively. Last, the similarity (size and concentration) of reaction products identified in lanes 1, 4, and 6 ( Fig. 8C) indicated that suitable conditions were established for all enzymes in the one-pot-reaction scheme.

DISCUSSION
Of all pathogenic Nm serogroups, NmA has caused the most disastrous epidemics in sub-Saharan Africa. The prevalence of this pathogen provoked an unprecedented endeavor to develop a highly effective and economic vaccine, MenAfriVac (7). With costs of less than 50 cents per dose, MenAfriVac enabled mass vaccination campaigns in Burkina Faso, Mali, and Niger (8 -10), which installed herd immunity, leading to protection not only for vaccinated but also for non-vaccinated individuals and in particular of young children (11).
All NmA vaccines licensed today are glycoconjugate vaccines coupled to carrier proteins, with CPSA polysaccharides isolated from large scale NmA cultures or oligosaccharides having shorter chain length obtained by acidic hydrolysis (40). To avoid the significant cost and biohazard in association with large scale NmA cultures and pyrogen-free production of polysaccharides, the enzyme-catalyzed in vitro synthesis of CPSA

The Capsule Biosynthesis Machinery of NmA
(CPSA iv ) would provide an attractive alternative. Toward this goal, we describe in this study the molecular cloning and functional expression of the three enzymes (UDP-GlcNAc-2epimerase, CsaA; poly-ManNAc-1-phosphate-transferase, CsaB; O-acetyltransferase, CsaC) that are part of the capsular biosynthesis complex in NmA and represent the minimal number of enzymes needed to produce immunologically active CPSA iv(OAc) in vitro starting from economic precursors. Using the well characterized BL21(DE3) strain as expression host, C-terminal His 6 -tagged and N-terminal StrepII-tagged versions of CsaA and CsaC could be purified in high quality and remarkable quantity (CsaA and CsaC, 40 and 96 mg/liter bacterial culture, respectively). In CsaB a second start codon was identified encoding methionine 70. In the overexpression system the use of this second start codon (construct ⌬69-CsaB-His 6 ) generated a stable protein that could be purified in remarkable amounts (60 mg/liter bacterial culture) after its DNA sequence had been optimized for codon usage in BL21(DE3). Whether this second start codon is actually used in the natural environment remains an open question.
When the CsaA/CsaB reaction was carried out under suboptimal conditions (CsaA concentrations were higher than CsaB concentrations), UDP and 2-acetamidoglucal were formed as side products. A similar finding was made by Sala et al. (41), FIGURE 7. Purification and characterization of in vitro synthesized CPSA. A, in vitro synthesized CPSA (CPSA iv ) was separated from all contaminating reaction products using anion exchange chromatography with the indicated sodium chloride gradient. B, purified CPSA iv after O-acetylation (CPSA iv(OAc) ) was re-purified under the same conditions resulting in a well separated product peak, which in panel C dot blot analyses were recognized by mAb 932. D, corresponding 1 H NMR analysis of the produced CPSA iv in comparison to CPSA n (from natural source). Slight variations in chemical shifts between CPSA n and CPSA iv are due to pH variations resulting from different purification protocols.
who showed that the E. coli UDP-N-acetylglucosamine-2epimerase, if present at high concentrations, releases the two intermediates of the epimerization reaction (UDP and 2-acetamidoglucal) into solution. Because E. coli UDP-N-acetylglucosamine-2-epimerase and CsaA share significant sequence similarity, it is reasonable to believe that CsaA uses the same catalytic mechanism. This CsaA side reaction could be completely suppressed if CsaA concentrations were equal or lower than the CsaB concentrations.
Using a two-step protocol (in test reactions even a one-pot reaction; see Fig. 8) O-acetylated CPSA (CPSA iv(OAc) ) could be produced in vitro in high purity and at medium scale (1 mg). if O-acetylated CPSA could be obtained in a one-step reaction. Analysis of the reaction by high percentage PAGE followed by combined Alcian blue/silver staining did not reveal any product signals (left lane), whereas long CPSA was obtained in the control reaction (right lane). B, 31 P NMR analysis of the samples revealed that UDP-3OAcManNAc (top spectrum) was not consumed by CsaB, whereas product signals were detected in the control (bottom spectrum). C, in vitro synthesis of O-acetylated and non-O-acetylated CPSA using all enzymes CsaA-C in a one-pot reaction. To control product formation, substrates and enzymes were added as indicated. After a 2-h incubation, products were displayed on high percentage PAGE by a combined Alcian blue/silver staining. Long chains were synthesized in all reactions containing CsaA, CsaB, and UDP-GlcNAc. D, only products obtained in the reaction where CsaC and acetyl-CoA were present could be detected with mAb 932 specifically directed against the CPSA OAc .
Similar to the CPSA isolated from natural source (CPSA n ) (42), the CPSA iv(OAc) fraction was of high molecular weight, showed 31 P NMR and 1 H NMR profiles consistent with CPSA n , and was recognized by mAb 932, a standard reagent in the characterization of immunologically active CPSA. Small differences were seen in the acetylation patterns. While the CPSA n reference contained some O-4-acetylation, this modification was not detectable in CPSA iv(OAc) . Although this difference at first sight may suggest the existence of a second enzyme with preference for the 4-O position, this interpretation is highly unlikely since csaC-knockouts are completely devoid of O-acetylation (21). Interesting, rather diverse values exist in literature for the relative occurrence of 3-O-Ac, 4-O-Ac, and free hydroxyl groups in CPSA n (70:0:30 (34), 87:8:5 (35), 40:27:33 (21)). To resolve the question of the importance of O-4-acetylation for the immunogenic quality of in vitro produced CPSA, further experimental work is needed.
In the current study we provide clear evidence that the acceptor quality of CPSA hyd increases after removal of the O-acetyl-groups. Consequently, also, the synthetic ManNAc-1P dimer carrying 3-O-Ac-groups (compound 6) was a less suited primer than the respective compound 5 without Oacetyl-groups. Asking if UDP-activated and 3-O-acetylated ManNAc (UDP-3OAcManNAc) may be a donor substrate for CsaB, we chemically synthesized this compound using a modification of a literature procedure for the preparation of UDP-GlcNAc (43) (see the supplemental scheme 3). Remarkably, no insertion of the modified compound was seen, strongly arguing for a highly selective recognition of the donor substrate by CsaB and further emphasizing the hypothesis that O-acetylation in NmA takes place on the synthesized polysaccharide. Of relevance in this context are the previous demonstrations that O-acetylation of sialic acid residues in the CPSs of NmC and E. coli K1 takes place after the polymer has reached a certain length and could not be detected on the donor substrate (44,45).
In 1999 a study by Ravenscroft et al. (36) demonstrated that acidic hydrolysis of CPSA results in fragments that are capped by phosphate at the non-reducing end. We show in the current study that removal of this capping phosphate steeply increases CsaB activity values (Fig. 4) and thus provide clear evidence that chain elongation proceeds by transfer of ManNAc-1P residues to the non-reducing end of the priming oligosaccharide. In addition, the use of synthetic priming compounds demonstrated that the minimal acceptor for CsaB is the ManNAcphosphate dimer and that the phosphate group at the reducing end can be extended with rather large chemical groups (n-decyl ester in the compounds tested in this study). Particularly this latter finding is of biotechnological relevance because it provides the perspective that CPSA chains can be built with priming oligosaccharides that carry functional groups ready for conjugation to carrier proteins (5).
Similar to the capsule polymerase of NmX (17), CsaB is also capable of initiating polymerization in the absence of any acceptor. Remarkably, this de novo activity was not altered in the presence of the artificial compounds 1-4. More experimental work is needed to interpret these findings, but based on these results it is tempting to speculate that the de novo reaction involves two UDP-ManNAc residues bound to acceptor and donor site in the enzyme.
In summary, we present data in this study that provide a new basis for the development of efficient and economic protocols (even one-pot-reaction protocols) for the synthesis of highly pure and immunologically active CPSA iv(OAc) . This means a large step forward in the combat of epidemics caused by one of the predominant neisserial serogroups NmA.