WaaA of the Hyperthermophilic Bacterium Aquifex aeolicus Is a Monofunctional 3-Deoxy-d-manno-oct-2-ulosonic Acid Transferase Involved in Lipopolysaccharide Biosynthesis*

The hyperthermophile Aquifex aeolicus belongs to the deepest branch in the bacterial genealogy. Although it has long been recognized that this unique Gram-negative bacterium carries genes for different steps of lipopolysaccharide (LPS) formation, data on the LPS itself or detailed knowledge of the LPS pathway beyond the first committed steps of lipid A and 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo) synthesis are still lacking. We now report the functional characterization of the thermostable Kdo transferase WaaA from A. aeolicus and provide evidence that the enzyme is monofunctional. Compositional analysis and mass spectrometry of purified A. aeolicus LPS, showing the incorporation of a single Kdo residue as an integral component of the LPS, implicated a monofunctional Kdo transferase in LPS biosynthesis of A. aeolicus. Further, heterologous expression of the A. aeolicus waaA gene in a newly constructed Escherichia coli ΔwaaA suppressor strain resulted in synthesis of lipid IVA precursors substituted with one Kdo sugar. When highly purified WaaA of A. aeolicus was subjected to in vitro assays using mass spectrometry for detection of the reaction products, the enzyme was found to catalyze the transfer of only a single Kdo residue from CMP-Kdo to differently modified lipid A acceptors. The Kdo transferase was capable of utilizing a broad spectrum of acceptor substrates, whereas surface plasmon resonance studies indicated a high selectivity for the donor substrate.

Lipopolysaccharide (LPS) 7 is the major constituent of the outer leaflet of the outer membrane (OM) of virtually all Gramnegative bacteria. LPS is a unique amphiphilic molecule composed of a hydrophilic heteropolysaccharide and a covalently bound lipid moiety, lipid A, which anchors the molecule in the OM. The polysaccharide component of many wild-type bacteria can be subdivided into a highly variable O-specific polysaccharide and a structurally less heterogeneous outer and inner core oligosaccharide (1). The 8-carbon sugar 3-deoxy-Dmanno-oct-2-ulosonic acid (Kdo) links the lipid A to the carbohydrate domain of LPS and is the only conserved structural element found in all inner core regions investigated to date (2).
The ubiquitous nature of Kdo within LPS structures and its essential role in maintaining OM integrity and viability of the majority of Gram-negative bacteria has prompted detailed studies into its biosynthesis. The Kdo pathway is initiated by the enzyme D-arabinose-5-phosphate isomerase, which catalyzes the interconversion of D-ribulose 5-phosphate and D-arabinose 5-phosphate (3). The Kdo-8-phosphate synthase KdsA subsequently condenses D-arabinose 5-phosphate with phosphoenolpyruvate to form Kdo 8-phosphate (4), followed by hydrolysis of Kdo 8-phosphate to Kdo and inorganic phosphate by the Kdo-8-phosphate phosphatase KdsC (5) and activation of Kdo to CMP-Kdo by the CMP-Kdo synthetase KdsB, before finally Kdo is transferred from CMP-Kdo to the lipid A moiety by the glycosyltransferase WaaA (6). In Escherichia coli, the Kdo-dependent late acyltransferases LpxL and LpxM subsequently transfer the fatty acids laurate and myristate, respectively, to Kdo 2 -lipid IV A to generate the characteristic acyloxyacyl units of hexaacylated Kdo 2 -lipid A (7).
It has long been recognized that Kdo transferases are unusual glycosyltransferases. WaaA is bifunctional in bacteria with LPS that contains an ␣-(234)-linked Kdo disaccharide in the inner core region, such as E. coli (6), Klebsiella pneumoniae (8), Legio-nella pneumophila (9), Acinetobacter baumannii, and Acinetobacter haemolyticus (10). In E. coli, CMP-Kdo is utilized for the transfer of Kdo to the tetraacylated lipid A precursor lipid IV A , resulting in an ␣-(236)-linkage between the distal glucosamine (GlcN) of the lipid A backbone and the first Kdo residue and an ␣-(234)-linkage between a second Kdo residue and the first one. Thus, WaaA is capable of catalyzing the formation of two different glycosidic bonds, tolerating acceptor molecules with varying extents of acylation but strictly depending on the 4Ј-phosphate group of the tetraacyldisaccharide 1,4Ј-bisphosphate intermediate (6). In chlamydiae, however, which express an LPS composed of a Kdo trisaccharide with an unusual ␣-(238)-linkage between the second and a third Kdo residue (11), the Kdo transferases were shown to display at least trifunctional activity (12). The LPS of Chlamydophila psittaci consists of up to four Kdo residues of the structure ␣-Kdo-(234)-[␣-Kdo-(238)]-␣-Kdo-(234)-␣-Kdo (13), and heterologous expression of the waaA gene in E. coli was found to be sufficient for synthesis of the complete chlamydial Kdo structure (12). Finally, the Kdo transferases of Haemophilus influenzae and Bordetella pertussis were shown to be monofunctional (14,15), consistent with the presence of a single phosphorylated Kdo residue in their respective LPS (16,17).
On the basis of phylogenetic analyses of 16 S ribosomal RNA sequences, members of the family Aquificaceae with growth-temperature maxima near 95°C are thought to represent the deepest branching species of the kingdom Bacteria (18). The cells are Gram-negative with a rather complex cell envelope of a surface protein layer, murein, and an OM (19). Previous studies provided the first direct evidence for the presence of smooth form LPS in Aquifex pyrophilus (20). Furthermore, KdsA and the UDP-(3-O-(R-3-hydroxymyristoyl))-N-acetylglucosamine deacetylase (LpxC) of A. aeolicus, a close relative of A. pyrophilus, have been characterized in detail, and it was demonstrated that these enzymes catalyze the first committed steps in Kdo and lipid A formation, respectively (21)(22)(23). Moreover, a number of genes presumably encoding different steps of LPS biosynthesis have been identified on the A. aeolicus genome, including putative kdsB and waaA orthologues for Kdo activation and subsequent incorporation of the sugar into LPS (24). However, the number of Kdo residues transferred by WaaA of A. aeolicus remains unknown. We herein provide evidence that the A. aeolicus enzyme is a strictly monofunctional Kdo transferase through the characterization of its enzymatic activity and the chemical analysis of the native A. aeolicus LPS.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-All strains and plasmids used in the present study are described in Table 1. Standard cultures of E. coli and Salmonella enterica sv. Typhimurium strains were routinely grown aerobically with shaking (250 rpm) at 37°C in LB medium (10 g/liter Bacto-tryptone, 5 g/liter yeast extract) supplemented with 10 g/liter NaCl to avoid colanic acid production in E. coli Kdo pathway mutants (KPM designation) (25), whereas BL21-SI212 and BL21-SI216 were propagated under non-inducing conditions in LBON (LB without NaCl). When necessary, kanamycin (30 g/ml), chloramphenicol (30 g/ml), ampicillin (100 g/ml), or carbenicillin (50 g/ml) was added to the media. DNA Methods-Standard recombinant DNA methods were used for nucleic acid preparation and analysis (26). Primer sequences are listed in supplemental Table S1. To both restore gene function and construct gene deletions, the phage Red recombinase procedure was used essentially as described (27), except plasmids pKD46 and pCP20 were cured at 37°C to accommodate the temperature sensitive phenotype of lipid IV A -expressing KPM22 derivatives (28). For generation of a kdsD insertion in KPM22, primers ECOyrbG and ECOkdsC were used to amplify the kdsD gene of E. coli K-12 wild-type strain BW30270, followed by transformation of KPM22/pKD46 with the resulting PCR product and selection for restored LPS biosynthesis on McConkey agar plates to yield strain KPM50. The ⌬waaC ⌬waaA strain KPM56 was obtained by successive deletion of the waaC gene of KPM50 and waaA of the resulting ⌬waaC strain KPM53, for which the primer pairs ECO-waaCH1/ECOwaaCH and ECOwaaAH1/ECOwaaAH2 with pKD4 as template were used to construct the insert cassettes targeting waaC and waaA, respectively.
To clone kdsB (aq_692) of A. aeolicus into pT7-7 and waaA genes of different origin into pET-16b, genomic DNAs were used as templates to amplify the kdsB gene with primer pair AAE5NdeIkdsB/AAE3BamHIkdsB and the waaA genes of A. aeolicus (aq_326) and E. coli K-12 with primers AAE5NdeIwaaA/AAE3BamHIwaaA and ECO5NdeIwaaA/ ECO3BamHIwaaA, respectively, whereas the plasmids pCB23 (14) and pUM140 (29) served as templates for amplification of the waaA genes of H. influenzae I69 (primers HIN5NdeIwaaA/HIN3BamHIwaaA) and C. psittaci 6BC (primers CPS5NdeIwaaA/CPS3BamHIwaaA), respectively. The purified amplification products were digested with NdeI and BamHI and ligated into the NdeI and BamHI sites of the expression vectors. The ligation mixtures were used to transform E. coli XL1-Blue cells. Plasmid DNA isolated from several of the respective transformants and identified by restriction analysis to contain the PCR products was subjected to DNA sequencing to confirm the sequences of the desired genes. Plasmids pUM210 with waaA CPS , pUM211 (waaA HIN ), pUM212 (waaA ECO ), and pUM216 (waaA AAE ) were subsequently used for transformation of KPM56 to obtain the strains KPM122, KPM123, KPM124, and KPM125, respectively. For overexpression of the histidine-tagged Kdo transferases of A. aeolicus and E. coli, plasmids pUM216 and pUM212 were introduced into BL21-SI to yield BL21-SI216 and BL21-SI212, respectively. Strain BL21-RIL01 was generated by transformation of BL21-CodonPlus(DE3)-RIL cells with pT7-kdsB to overexpress KdsB of A. aeolicus.
Overexpression and Purification of KdsB from A. aeolicus-To induce expression of KdsB AAE in BL21-RIL01 cells, isopropyl-␤-D-thiogalactoside was added to a final concentration of 1 mM when the culture reached the mid-exponential growth phase at an A 600 of 0.6 -0.8. The cells were harvested 3 h postinduction by centrifugation (9000 ϫ g, 4°C, 20 min), washed with phosphate-buffered saline, and resuspended in buffer A (20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 5 mM 2-mercaptoethanol) at 0.1 g of wet biomass/ml. The suspension was incubated with lysozyme (0.2 mg/ml), DNase I (0.1 mg/ml), and RNase A (0.1 mg/ml) at 20°C for 30 min with continuous stirring and placed on ice for 15 min, and phenylmethylsulfonyl fluoride was added to a final concentration of 1 mM. The cells were broken by passage three times through a French pressure cell at 20,000 p.s.i., followed by centrifugation of the homogenate to remove cellular debris (27,000 ϫ g, 4°C, 30 min). The supernatant was heated to 85°C in a boiling water bath and incubated at 85°C for an additional 15 min with continuous swirling. The suspension was allowed to cool to 22°C and placed on ice for 15 min, and the precipitated protein was removed by centrifugation (27,000 ϫ g, 4°C, 30 min). The supernatant was loaded onto a DEAE-Sepharose Fast Flow column (GE Healthcare) pre-equilibrated with buffer A. The column was eluted with a linear gradient from 0 to 0.2 M NaCl in the same buffer. Fractions containing KdsB AAE were pooled, and solid (NH 4 ) 2 SO 4 was added to a final concentration of 20% (w/v). The sample was subsequently applied to a phenyl-Sepharose CL-4B column (GE Healthcare) pre-equilibrated with 20% (NH 4 ) 2 SO 4 (w/v) in buffer A. A reverse gradient from 20% (w/v) to 0% (NH 4 ) 2 SO 4 in buffer A was used to elute KdsB AAE . The fractions containing KdsB AAE were combined, dialyzed against buffer A, and concentrated by chromatography on a HiTrap Q-Sepharose HP (5 ml) column (GE Healthcare) using a linear gradient from 0 to 0.5 M NaCl in buffer A. For final KdsB AAE polishing, the pooled fractions were directly loaded onto a size exclusion HiLoad 26/100 Superdex 200 pg column (GE Healthcare) and eluted with buffer B (20 mM Tris-HCl, pH 7.5, 0.1 M NaCl, 5 mM MgCl 2 , 2 mM 2-mercaptoethanol). The final preparation was concentrated using Centricon Plus-20 concentrators (Amicon), dialyzed against buffer B, and stored at 4°C.
Overexpression and Purification of KdsB from E. coli-The synthetase KdsB ECO has been previously obtained by immobilized metal affinity chromatography, using Corynebacterium glutamicum R163/pJKB72 as the expression host (30). Here we have upgraded the published procedure to purify KdsB ECO to apparent homogeneity. The C. glutamicum R163/pJKB72 culture was grown to an A 600 of 0.6 -0.8, followed by expression of KdsB ECO in the presence of 0.2 mM isopropyl-␤-D-thiogalactoside for an additional 16 h. The cells were harvested by centrifugation (9000 ϫ g, 4°C, 20 min), washed with phosphate-buffered saline, resuspended at 0.3 g of wet biomass/ml of buffer C (50 mM NaH 2 PO 4 , pH 8.0, 300 mM NaCl, 5 mM 2-mercaptoethanol) supplemented with 10 mM imidazole, and incubated with continuous stirring at 37°C for 45 min with lysozyme (1 mg/ml), lipase (1 mg/ml), DNase I (0.1 mg/ml), and RNase A (0.1 mg/ml). Then phenylmethylsulfonyl fluoride was added to a final concentration of 1 mM before the chilled cells were disrupted by passage three times through a French pressure cell at 20,000 p.s.i. After removal of cellular debris by centrifugation (27,000 ϫ g, 4°C, 30 min), the supernatant was loaded onto a nickel-Sepharose HP column (GE Healthcare) pre-equilibrated with 10 mM imidazole in buffer C. KdsB ECO was eluted with 250 mM imidazole in buffer C, following an extensive washing step using 80 mM imidazole in the same buffer. The fractions containing KdsB ECO were dialyzed against buffer A and subjected to ion exchange chromatography on HiTrap Q-Sepharose HP and gel filtration on HiLoad 26/100 Superdex 200 pg as described above for KdsB AAE purification. The final concentrated protein sample was dialyzed against buffer B and stored at Ϫ80°C.
Overexpression and Purification of WaaA from A. aeolicus-The conditions used to express soluble WaaA AAE in BL21-SI216 were adapted from a previously published protocol for recovery of active heterologous protein from E. coli cells grown at decreased temperature and acidic pH values (31). Strain BL21-SI216 was grown to the mid-exponential growth phase (A 600 of 0.6 -0.8) at 20°C with vigorous shaking (250 rpm) in a medium containing 10 g/liter peptone and 20 g/liter yeast extract. At this point, the medium was supplemented with NaCl (120 mM), glucose (1%), glycerol (0.4%), and potassium phosphate (100 mM), pH 4.5, to induce WaaA expression along with acidification of the medium. The cells were allowed to grow at 20°C for an additional 18 h before they were sedimented (9000 ϫ g, 4°C, 20 min), washed with phosphate-buffered saline, and resuspended at 0.3 g of wet biomass/ml of buffer D (50 mM Tris-HCl, pH 8.7, 10% glycerol, 0.1% Triton X-100, 5 mM 2-mercaptoethanol). After incubation of the suspension with lysozyme (0.2 mg/ml) at 20°C for 30 min and continuous stirring, the cells were chilled on ice and supplemented with DNase I (0.1 mg/ml), RNase A (0.1 mg/ml), and Complete protease inhibitor mixture (EDTA-free) according to the recommendations of the manufacturer (Roche Applied Science) and disintegrated by passage four times through a French pressure cell at 20,000 p.s.i. The cellular debris was removed by centrifugation (27,000 ϫ g, 4°C, 30 min), and the supernatant was loaded onto a DEAE-Sepharose Fast Flow column pre-equilibrated with buffer D. Under these conditions, WaaA AAE was eluted in the flow-through, whereas impurities found to interact with WaaA AAE in subsequent purification steps remained bound on the column. The eluate was mixed with an equal volume of a solution containing 2 M NaCl, 10% glycerol, 0.1% Triton X-100, 5 mM 2-mercaptoethanol, and 20 mM imidazole, followed by incubation of the mixture on ice for 2 h with stirring, centrifugation (27,000 ϫ g, 4°C, 30 min), and filtration through a 0.22-m membrane. The supernatant was subjected to immobilized metal affinity chromatography on a nickel-Sepharose HP column pre-equilibrated with buffer E (25 mM Tris-HCl, pH 8.7, 10% glycerol, 0.1% Triton X-100, 5 mM 2-mercaptoethanol) containing 1 M NaCl and 10 mM imidazole. The column was extensively washed applying a stepwise gradient of 20, 80, and 100 mM imidazole before WaaA AAE was eluted with 500 mM imidazole in buffer E supplemented with 1 M NaCl. The fractions containing WaaA AAE were combined, dialyzed against 0.1 m NaCl in buffer E, heated to 80°C in a boiling water bath, and incubated at 80°C for an additional 20 min with continuous swirling. The solution was chilled on ice for 15 min, followed by sedimentation of the denatured protein (27,000 ϫ g, 4°C, 30 min) and affinity chromatography of the cleared supernatant on a HiTrap Heparin-Sepharose HP (5 ml) column (GE Healthcare) pre-equilibrated with the above buffer. The column was then washed with 0.5 M NaCl in buffer E, and WaaA AAE was eluted with 0.9 M NaCl in the same buffer. Finally, the WaaA AAE pool was diluted with buffer E to decrease the NaCl concentration to 0.1 M and loaded once more onto a HiTrap heparin-Sepharose HP (1 ml) column. The washing and elution steps were performed essentially as described above, except buffer E was used with 50 mM octyl-␤-D-glucopyranoside instead of 0.1% Triton X-100. The combined WaaA AAE peak fractions were dialyzed against 25 mM Tris-HCl, pH 8.7, 0.1 M NaCl, 10% glycerol, 50 mM octyl-␤-D-glucopyranoside, and 5 mM 2-mercaptoethanol and stored at 4°C.
Overexpression and Purification of WaaA from E. coli-The conditions used for overexpression of the Kdo transferase from E. coli in BL21-SI212 cells as well as purification of the enzyme were essentially the same as described above for WaaA AAE , except buffers D and E contained 1% Triton X-100, heating of the WaaA ECO solution at 80°C was omitted, 1.5 M NaCl was necessary to elute the protein from the HiTrap Heparin-Sepharose HP (5 ml) column, and final exchange of the detergents was not performed due to a strong propensity of the protein to aggregate in octyl-␤-D-glucopyranoside-containing buffers.
KdsB Activity Assay-The activities of the CMP-Kdo synthetases from A. aeolicus and E. coli were monitored by colorimetric measurement of CMP-Kdo formation essentially as reported previously (34), except KdsB AAE was assayed in 100 mM glycine-NaOH buffer, pH 9.0, at the assayed temperature, at 50°C for 30 s.
In Vitro Analysis of WaaA Activity-The in vitro reactions were performed under optimized conditions in a total volume of 100 l containing 100 mM HEPES, pH 7.5, at the assayed temperature, 10% glycerol, 10 mM MgCl 2 , 2 mM Kdo, 5 mM CTP, 3.2 mM Triton X-100, 17 milliunits of either KdsB AAE or KdsB ECO , 75 M either WaaA AAE or WaaA ECO , and a synthetic lipid A acceptor. The tetraacyl-1,4Ј-bisphosphate lipid A precursor 406 used at a concentration of 107 M, the tetraacyl-1monophosphate precursor 405 (113 M), the hexaacyl-1,4Јbisphosphate lipid A compound 506 (83 M), the 1-carboxymethyl (CM) derivative CM-506 of hexaacyl lipid A (84 M), and the tetraacyl-1,4Ј-CM compound Bis-CM-406 (110 M) have been synthesized as described (35)(36)(37)(38)(39). Tetraacylated lipid A of A. pyrophilus was applied at a concentration of 91 M. To assay WaaA AAE , the reactions were routinely incubated in thin walled PCR tubes at 60°C for 1 h using the Gene-Amp PCR system 9700 thermocycler (PerkinElmer Life Sci-ences), whereas the WaaA ECO reactions were incubated at 37°C. Triton X-100 in the reactions was depleted by subsequent treatment of the mixtures with Calbiosorb Adsorbent, as recommended by the supplier (Calbiochem), followed by extensive dialysis against H 2 O, lyophilization, and analysis of the reaction products by electrospray-ionization Fourier-transformed ion cyclotron mass spectrometry (ESI FT-ICR MS).
Thermostability Measurements-Purified enzyme was either separately preincubated at 70, 80, 90, or 99°C for 1 h or constantly maintained at 80°C for 16 h prior to in vitro analysis of WaaA activity using the standard assay.
LPS Isolation and Analysis-The LPS profiles of proteinase K-digested whole cell lysates were investigated on silverstained 13% SDS-polyacrylamide gels according to the method of Hitchcock and Brown (40). The LPS of E. coli strains was extracted from the dried biomass of stationary phase 2-liter cultures using either the original phenol/chloroform/light petroleum (PCP) procedure (41) or a modified PCP protocol for strains lacking the K-12 core oligosaccharide (42). For isolation and purification of A. aeolicus LPS, the dried biomass was extracted using hot phenol/water supplemented with 2% sodium N-lauroylsarcosine (w/v), as described previously (20). Oligosaccharides were obtained by hydrolysis of the LPS in 0.1 M HCl at 100°C for 90 min, followed by removal of lipid A using ultracentrifugation (105,000 ϫ g, 4°C, 4 h). The carbohydrate portion (supernatant) was purified by gel permeation chromatography on a Biogel P4 column using a pyridine/acetic acid/water (4 ml/10 ml/986 ml) mixture as eluent. Combined GLC/MS analyses were performed to identify the sugar components of LPS and the oligosaccharide fraction either after weak (0.5 M HCl in methanol, 85°C, 45 min, followed by peracetylation) or strong methanolysis (2 M HCl in methanol, 85°C, 20 h, followed by peracetylation) or after hydrolysis with 4 M trifluoroacetic acid (100°C, 4 h) and reduction with NaBH 4 or NaBD 4 to confirm the presence of GalA (43). Neutral sugars were analyzed quantitatively by GLC following hydrolysis of the samples in 2 M trifluoroacetic acid at 120°C for 2 h (43). The content of Kdo was determined photometrically using the thiobarbituric acid assay as published (44). The preparation of O-deacylated lipid A from A. pyrophilus has been described previously (20). The A. aeolicus VF5 biomass was purchased from Harald Huber (Institute of Microbiology, University of Regensburg, Germany).
Electrospray-Ionization Fourier-transformed Ion Cyclotron Mass Spectrometry-Mass spectra were recorded in the negative ion mode using an APEX II and a hybrid Apex Qe Instrument (Bruker Daltonics) equipped with a 7-Tesla actively shielded magnet. Details on sample preparation and MS characteristics of LPS have been published (45). The mass spectra were charge-deconvoluted, and the mass numbers given refer to the monoisotopic peaks of the neutral molecules.
Surface Plasmon Resonance Binding Studies-Surface plasmon resonance (SPR) experiments were conducted in a BIAcore 3000 instrument (Biacore AB). A CM5 sensor chip, 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC), N-hydroxysuccinimide, 1 M ethanolamine-HCl, pH 8.5, and 10 mM sodium acetate, pH 4.0, were purchased from Biacore. The Bistris propane running buffer consisted of a 50 mM Bistris propane solution in double-distilled water containing 150 mM NaCl and 10 mM MgCl 2 . The pH was adjusted to 7.4 at 25°C with a solution of 17.5% HCl. All buffers were filtered through a 0.22-m membrane and degassed prior to use. All experiments were carried out at 25°C.
The Kdo transferase of A. aeolicus was covalently immobilized to the carboxyl-methylated dextran surface of the CM5 sensor chip according to the amine coupling protocol of the manufacturer (Biacore AB). The surface was activated at a flow rate of 5 l/min with a 7-min pulse of N-hydroxysuccinimide/ EDC, followed by WaaA AAE immobilization (20 M in 10 mM sodium acetate, pH 4.0) and subsequent neutralization of unreacted N-hydroxysuccinimide-esters with a 7-min pulse of 1 M ethanolamine-HCl, pH 8.5. A total of 7406 resonance units (RU) of the Kdo transferase was immobilized to one flow cell of the sensor chip, whereas 8954 RU of neutravidin (17 M in 10 mM sodium acetate buffer, pH 4.0) was immobilized by the same procedure to a reference flow cell to mimic the environment of the WaaA AAE -containing cell.
To investigate the interaction of WaaA AAE with CMP, a concentrated solution of CMP (200 mM in Bistris propane running buffer) was serially diluted, yielding concentrations in the range between 39 M and 20 mM. Samples of 10 l were injected over the sensor chip at a flow rate of 10 l/min. CMP dissociated very fast from the surface so that further regeneration steps were not necessary. A 10-l pulse of buffer was injected between ligand injections. The titration was repeated under the same experimental conditions to ensure reproducibility. The resulting sensorgrams were processed and evaluated with BIAevaluation software version 3.2 (Biacore AB) and the Scrubber software package (BioLogic Software), following a procedure based on the double referencing method (46). The response from the neutravidin-coated flow cell was first subtracted from the response in the flow cell containing WaaA AAE . Second, the averaged response of three buffer injections was subtracted to obtain the final processed sensorgrams. The equilibrium dissociation constant was determined by steady state analysis of the sensorgrams, using non-linear square curve fitting to the equation, [AB] eq ϭ AB max (1/(1 ϩ K D /[A])), with A and AB being the analyte (CMP) and the analyte-ligand complex (CMP-WaaA AAE ), respectively, AB max the maximum surface binding capacity (RU), and K D the equilibrium dissociation constant (M). Further, SPR binding studies were performed to investigate the interaction of AMP, TMP, UMP, or GMP with WaaA AAE , employing the same methodology as described for evaluation of the affinity between CMP and the Kdo transferase. Finally, SPR was used to examine the binding of Kdo to WaaA AAE , for which a concentrated solution of Kdo (200 mM) in Bistris propane running buffer was serially diluted to obtain concentrations ranging from 625 M to 20 mM and processed as described above. (21), LpxC (23), and presumably other LPSrelated functions (24) indicates the ability of A. aeolicus to synthesize LPS and incorporate Kdo into LPS, data on the LPS itself, including the number of Kdo residues to predict WaaA functionality, is still lacking. Analysis by SDS-PAGE revealed a ladder-like appearance of the LPS preparation, suggesting the synthesis of smooth form LPS in A. aeolicus (supplemental Fig.  S1, lane 7). In contrast to the LPS of A. pyrophilus with a decidedly non-modal O-specific polysaccharide banding (supplemental Fig. S1, lane 8), the LPS of A. aeolicus showed the characteristic pattern of modal chain length distribution of the supposed O-polysaccharide, with a significant LPS fraction apparently being not ligated with O-specific polysaccharide. Electrospray-ionization Fourier-transformed ion cyclotron mass spectrometry (ESI FT-ICR MS) of the lipid A identified a molecular mass of 1917.33 u (data not shown), consistent with the previously published pentaacylated structure of A. pyrophilus lipid A (20). In contrast to A. pyrophilus (20), investigation of the native LPS from A. aeolicus by MS revealed a remarkably homogeneous pattern of the unligated LPS population with three prominent molecular peaks (supplemental Fig. S2A). Assignment of the masses suggested a chemical composition of lipid A, one Kdo, two heptose, one hexose, and three hexosuronic acid residues for peak M1 at 3211.67 u (calculated mass 3211.614 u), whereas peak M2 at 4161.03 u corresponded to a species containing in addition one hexose, one N-acetyl hexosamine, and four deoxyhexose moieties (calculated mass 4160.978 u). The peak M3 at 4323.10 u apparently originated from a molecule with another hexose attached (calculated mass 4323.030 u) (supplemental Table S2 Table S2). Compositional analysis of the LPS and the oligosaccharide fraction isolated after hydrolysis of the LPS identified Rha, Man, Glc, Gal, GalA, Hep, and Kdo (supplemental Table S3), confirming that the polysaccharide chains of the LPS of A. aeolicus and A. pyrophilus (20) were of a quite different nature. Noteworthy, thiobarbituric acid-reactive material could be liberated only upon strong hydrolysis of the LPS and the oligosaccharide fraction but not using mild acid treatment, indicating the lack of acid-labile side chain Kdo substitutions (44). Thus, the application of harsh hydrolysis conditions enabled the determination of an average molar ratio of the common inner core constituents Kdo and Hep of 1.0:1.8 for the LPS and 1.0:1.7 for the LPS-derived oligosaccharide, with Kdo being probably a constituent of the main core oligosaccharide chain. Taken together, the results suggested that the LPS of A. aeolicus contains a single Kdo residue, and therefore, it was assumed that WaaA of A. aeolicus is a monofunctional Kdo transferase.

Characterization of the LPS from A. aeolicus-Although the presence of KdsA
Expression of WaaA from A. aeolicus in E. coli-Preliminary in vitro experiments revealed that, although WaaA of A. aeolicus is thermophilic (see below), it is still capable of catalyzing Kdo transfer at a temperature of 37°C (data not shown). In order to investigate the in vivo activity of the A. aeolicus enzyme as well as for characterization of other heterologous Kdo transferases in E. coli, the strain KPM56, a kdsD ϩ ⌬waaC ⌬waaA derivative of KPM22, was constructed. Strain KPM22, defective in the D-arabinose-5-phosphate isomerases KdsD and GutQ, is a nonconditional E. coli K-12 Kdo pathway null mutant that lacks Kdo and is viable despite predominantly elaborating the lipid A precursor lipid IV A (28). The lethal ⌬Kdo phenotype of KPM22 was shown recently to be suppressed by an Arg to Cys substitution at position 134 of the inner membrane protein YhjD, a gain-of-function mutation that converts the normally essential Kdo pathway genes into nonessential genes (28,42). Thus, the presence of the yhjD400 suppressor allele in KPM22 enabled the derivation of the target KPM56 in three steps. First, the Kdo pathway of KPM22 was reconstituted by insertion of the kdsD wild-type gene, which yielded strain KPM50 with restored K-12 LPS biosynthesis (Fig. 1, lane 3). Second, to facilitate subsequent analyses of the LPS from strains expressing heterologous Kdo transferases, the LPS pathway following attachment of the Kdo residues to lipid A was interrupted in KPM50 by deleting the waaC gene for heptosyltransferase I, which resulted in strain KPM53 expressing LPS of the Re chemotype (Kdo 2 -lipid A) (Fig. 1, lane 4). Finally, KPM53 served as the host for the waaA knock-out to yield strain KPM56, which was indeed viable and capable of maintaining the normally lethal ⌬waaA mutation. Like KPM22 (Fig. 1, lane 2) (28), silverstained SDS-polyacrylamide gels detected no LPS bands for KPM56 (Fig. 1, lane 5). Analysis of the KPM56 LPS by ESI FT-ICR MS, however, revealed two prominent LPS-related peaks with molecular masses of 1404.85 and 1527.86 u, consistent with the structures of the tetraacyl-1,4Ј-bisphosphate LPS precursor lipid IV A (calculated mass 1404.854 u) and lipid IV A modified with a phosphoethanolamine (P-EtN) group (calculated mass 1527.863 u), respectively ( Fig. 2A). Although SDS-PAGE analysis of the LPS isolated from control strains KPM124 and KPM122 readily demonstrated complementation of the ⌬waaA mutation by Kdo transferases from E. coli and C. psittaci (Fig. 1, lanes 8 and 9), no LPS bands could be observed for KPM125 and the KPM123 control strain carrying the Kdo transferases of A. aeolicus and H. influenzae (Fig. 1, lanes 6 and  7), respectively. However, when the LPS samples of A. aeolicus and H. influenzae were examined using MS, we could not only confirm the monofunctional activity of WaaA from H. influenzae by detecting a nearly exclusive peak of Kdo-lipid IV A in KPM123 (1624.91 u) (supplemental Fig. S3A and Table S4) but also could detect peaks at 1624.91 and 1747.92 u in the KPM125 sample, which indicated the presence of only a single Kdo residue attached to lipid IV A and P-EtN-modified lipid IV A , respectively (Fig. 2B). Although substantial amounts of the lipid IV A precursors remained nonglycosylated, the results suggested that the Kdo transferase of A. aeolicus is monofunctional in E. coli. Strikingly, a common feature of KPM123 and KPM125 was the lack of the secondary lauroyl and myristoyl chains on Kdo-lipid IV A derivatives. To verify that their parental strain KPM56 still contained functional copies of the late acyltransferases LpxL and LpxM, we analyzed the LPS of KPM124 by ESI FT-ICR MS. As shown in supplemental Fig.  S3B and Table S4, complementation of the chromosomal waaA deletion with the homologous, bifunctional Kdo transferase resulted in expression of mainly hexaacylated lipid A species, each carrying the secondary acyl chains and two Kdo residues. Although MS revealed a rather heterogeneous mixture of LPS molecules with acyl chains of various lengths differing in mass each by 14.02 u, the results indicated the capability of LpxL and LpxM to form the characteristic acyloxyacyl units of lipid A in KPM56-derived strains, provided that lipid IV A was glycosylated with two Kdo residues. Thus, our data indicating lipid IV A precursors remaining underacylated with only a single Kdo residue attached corroborate previous in vitro (47,48) and in vivo (28) findings of an extremely high specificity of LpxL and LpxM for the Kdo 2 -lipid IV A substrate.
Overexpression and Purification of WaaA from A. aeolicus-Our initial attempts to express the Kdo transferase of A. aeolicus at high levels in T7-promoter based expression hosts were unsuccessful due to sequestration of the protein within insoluble inclusion bodies. Soluble and catalytically active WaaA AAE could only be obtained in sufficient quantities when WaaA AAE expression was performed in BL21-SI216, which allowed tightly regulated protein production at moderate levels. The formation of inclusion bodies was nearly avoided when the growth rate of the induced cells was reduced under suboptimal conditions in acidic medium at 20°C. Under these conditions, the maximal ratio of recombinant to cellular protein was achieved by expression of the Kdo transferase for a minimum of 18 h.
Purification to apparent homogeneity of recombinant WaaA AAE was accomplished in three chromatographic steps using anion exchange chromatography on DEAE-Sepharose, immobilized metal affinity chromatography on nickel-Sepharose, and affinity chromatography on heparin-Sepharose. Like the Kdo transferase of E. coli (6), WaaA of A. aeolicus displayed reversible affinity for the heparin resin, and therefore, it was used for purification of the enzyme. Taking advantage of the enzyme's tolerance to high temperatures (see below), residual host cell protein could be easily removed by heat treatment of the samples. The Kdo transferase of A. aeolicus as analyzed by SDS-PAGE and verified by immunoblotting using an anti-histidine tag antibody appeared as four bands with apparent relative molecular masses of about 42, 83, 122, and 185 kDa (Fig. 3, A and B). The apparent relative molecular mass of the 42 kDa band was in good agreement with the calculated mass of the subunit of 40.7 kDa fused to a 2.5-kDa N-terminal histidine tag, whereas the other three bands migrated with apparent relative molecular masses about 2, 3, and 4 times the mass of the subunit, suggesting the capability of WaaA AAE to form remarkably stable oligomers that were To go further into the question of reliability of the herein developed methods for WaaA AAE overexpression and purification, as well as to address the question of an equivalent control for subsequent in vitro assays, we attempted to express the Kdo transferase of E. coli in BL21-SI212 cells and purify the enzyme essentially as outlined for the Kdo transferase of A. aeolicus. Using identical conditions for expression and minor modifications of the basic purification scheme (for details, see "Experimental Procedures"), we could purify the Kdo transferase of E. coli to apparent homogeneity as well. The subunit as shown by SDS-PAGE under denaturing conditions migrated as a single band with an apparent relative molecular mass of about 45 kDa (Fig. 3C), which was marginally different from the calculated mass of 49.8 kDa but in accord with the relative mobility previously observed for WaaA ECO in SDS-PAGE (6).
In Vitro Analysis of WaaA AAE Activity-To ascertain whether the Kdo transferase of A. aeolicus was monofunctional, the WaaA-catalyzed transfer of Kdo from CMP-Kdo to synthetic lipid acceptors was assayed in vitro at 60°C, using the herein developed novel method for detection of WaaA reaction products by MS (for details, see "Experimental Procedures"). Since CMP-Kdo is known to be highly unstable (49), a CMP-Kdo-generating system consisting of CTP, Kdo, Mg 2ϩ , and CMP-Kdo synthetase is required to determine WaaA activity in vitro (6). For this purpose and like the Kdo transferases of A. aeolicus and E. coli, we have purified the homologous CMP-Kdo synthetases to apparent homogeneity as described under "Experimental Procedures" (data not shown).
As demonstrated by ESI FT-ICR MS, the Kdo transferase of A. aeolicus was capable of converting the synthetic tetraacyl-1,4Ј-bisphosphate lipid A precursor 406 with a molecular mass of 1404.86 u (Fig. 4A) into a molecule with a molecular mass of 1624.92 u, consistent with the structure of 406 being substituted with a single Kdo residue (calculated mass 1624.912 u) (Fig. 4B). Further, a minor batch-dependent triacylated byproduct of 406 synthesis with a mass of 1318.61 u (Fig. 4A) was converted into a product with a mass of 1538.67 u (Fig. 4B), indicating the attachment of only a single Kdo residue to the artificial acceptor. No products were obtained in reactions when either WaaA AAE or KdsB AAE was omitted (not shown).  pyrophilus lacking the ester-bound octadecanoic acid (1651.12 u) was incubated with WaaA AAE , only a marginal fraction of the acceptor was glycosylated with a single Kdo (1871.18 u) (Fig. 5), indicating that tetraacylated lipid A with GalA residues in place of the phosphate groups at positions 1 and 4Ј is a poor substrate for the Kdo transferase. Collectively, the results showed that WaaA of A. aeolicus is a monofunctional Kdo transferase also in vitro.   AUGUST 14, 2009 • VOLUME 284 • NUMBER 33

Studies of the Interaction between WaaA AAE and Donor
Substrates-Due to the high instability of CMP-Kdo (49), interaction of CMP and Kdo with the Kdo transferase of A. aeolicus was investigated separately using SPR technology (50). WaaA AAE (ligand) was immobilized to the surface of a CM5 sensor chip, and different concentrations of CMP or Kdo (analytes) were injected to the surface, as described under "Experimental Procedures." Binding of the analyte to the ligand was monitored as an increase of the response (RU) versus time (sensorgram). When CMP was injected to the surface with WaaA AAE , an increase of the response was detected, which suggested binding of CMP to the Kdo transferase. The resulting sensorgrams were characteristic of a binding event with fast on and off rates, and thus, only steady state analysis of the sensorgrams was possible (Fig. 6A). The corresponding binding isotherm was fitted to a 1:1 model, and a K D of 401 Ϯ 25 M was determined (Fig. 6B). When Kdo was delivered to the surface coated with WaaA AAE , we could not measure any increase of the response, indicating no binding of Kdo to the Kdo transferase under the experimental conditions used (data not shown). Injection of AMP, TMP, UMP, and GMP detected much lower signals compared with the response caused by binding of CMP to WaaA AAE , and plot of the R eq versus NMP concentration resulted in all cases in linear graphs characteristic of nonspecific electrostatic interactions between the ligand and the analyte (supplemental Fig. S10). Taken together, the data indicated that CMP but not Kdo alone and other NMPs are specifically recognized by the Kdo transferase of A. aeolicus.
The Effect of High Temperature on WaaA AAE Activity-The hyperthermophilic origin of WaaA AAE prompted us to examine the effect of temperature on the WaaA AAE -catalyzed conversion of compound 406 into Kdo-406 as a function of time. Pretreatment of the enzyme at 70 and 80°C for 1 h prior to the determination of residual activity at 60°C did not affect the ability of WaaA AAE to perform transfer of a single Kdo residue to the lipid A acceptor, whereas preincubation of the enzyme at 90 and 99°C for 1 h resulted in gradual loss of its activity (supplemental Fig. S11). Remarkably, we could not observe any detrimental effect on Kdo glycosylation of compound 406 after long term storage of the Kdo transferase at 80°C for 16 h (Fig.  7). Thus, the extraordinary thermostability demonstrated by the recombinant Kdo transferase of A. aeolicus allowed us to conclude that the enzyme was produced and isolated in its native state.

DISCUSSION
This study describes the characterization of WaaA from A. aeolicus and represents the first report on a Kdo transferase from an extreme hyperthermophile. The results presented herein strongly support the conclusion that WaaA AAE belongs to the group of monofunctional Kdo transferases of LPS biosynthesis.
A. aeolicus Synthesizes LPS Containing a Single Kdo Residue-Except for Helicobacter pylori and Francisella novicida, the LPSs of which possess a single Kdo in the inner core region while expressing bifunctional Kdo transferases (51,52), the number of Kdo residues in the core oligosaccharide of LPS normally predicts the functionality of WaaA. Our data on the initial characterization of the A. aeolicus LPS indicate that the molecule contains a single Kdo residue and, therefore, implicates a monofunctional Kdo transferase in LPS biosynthesis of A. aeolicus. In fact, we provide direct evidence for the existence of LPS in A. aeolicus, thus confirming the results of previous investigations, which predicted the presence of LPS in A. aeolicus on the basis of genome analysis (24), biochemical and structural characterization of KdsA (21,22) and LpxC (23,53), or detection of LPS and a unique lipid A structure in A. pyrophilus, a close relative of A. aeolicus (20). Since lipid A is considered to be a valuable chemotaxonomic and phylogenetic marker (54), the close relation of A. aeolicus and A. pyrophilus is further supported by our ESI FT-ICR MS experiments that suggest that the structure of A. aeolicus lipid A is identical to that of A. pyrophilus (20). However, we observed a striking differences between the LPS from A. aeolicus and A. pyrophilus by SDS-PAGE and ESI FT-ICR MS as well as a different sugar composition in the LPS of A. aeolicus, indicating significant strain- A. aeolicus waaA Encodes a Monofunctional Kdo Transferase That Can Complement a ⌬waaA Mutation in E. coli-In vivo examination of Kdo transferases of different origin has usually been performed in E. coli to gain insight into the structure of the resulting LPS. In consideration of the fact that viability of wild-type E. coli cells is normally dependent upon the presence of a functional waaA copy (55), several strategies have been developed to reduce or even avoid LPS complexity stemming from co-expression of the heterologous and the host-specific Kdo transferase, such as cloning of Kdo transferases in an E. coli Re mutant (56,57), plasmid-encoded expression of WaaA in an E. coli waaA null mutant by displacement under nonpermissive conditions of a functional waaA ECO copy provided in trans (9,15,55), or homologous recombination of WaaA expression cassettes into the waaA locus of a recBC sbcBC E. coli strain (12,14).
Here we have constructed the ⌬waaA suppressor strain KPM56 as a novel host for expression and in vivo analysis of exclusively plasmidborne Kdo transferases. Although our results show that WaaA from A. aeolicus is monofunctional in KPM125, the enzyme apparently does not reach its full catalytic activity in the mesophilic E. coli host. Our in vitro data demonstrating conclusively the conversion of chemically synthesized compound 406 into Kdo-406 do not give us reason to assume that its natural counterpart lipid IV A represents a poor substrate of WaaA AAE but suggest a diminished capability of the thermophilic enzyme to glycosylate the lipid A precursor at a suboptimal temperature of 37°C. Although thermophilic enzymes usually develop optimal activity at high temperatures close to the optimal growth temperatures of their hosts, they are thought to adopt structures of high rigidity at mesophilic temperatures below 40°C, which in turn may cause a decrease in catalytic efficiency (58,59). The successful expression of WaaA from H. influenzae in KPM123, leading to predominant formation of Kdo-lipid IV A , indicates that KPM56 is a suitable host also for characterization of monofunctional Kdo transferases. This is emphasized especially in view of the fact that investigation of monofunctional Kdo transferases has been met with restrictions under conditions of customarily used E. coli ⌬waaA complementation systems, such as the need to co-express the Kdo kinase KdkA from H. influenzae in order to compensate for the lethal phenotype associated with the substitution of waaA ECO for waaA of H. influenzae (14). In view of Kdo-lipid IV A being a very poor substrate of LpxL in vitro (60) and, as we show now, also in vivo, with LpxM apparently being unable to transfer a myristate chain to Kdo-lipid IV A in the absence of laurate, the underacylated and underglycosylated precursor is expected to accumulate in large toxic amounts in the inner membrane due to the high selectivity of the ATP binding cassette transporter MsbA for hexaacylated LPS/lipid A substrates (61,62). E. coli KPM123, however, is viable despite predominantly elaborating Kdo-lipid IV A . Therefore, we propose that the YhjD R134C suppressor protein, which has been suggested to participate in transmembrane transport of lipid IV A in E. coli KPM22 (42), can also transit Kdo-lipid IV A molecules across the inner membrane of KPM22-derived strains, restoring viability by supplementing the insufficient amounts of Kdo-lipid IV A being translocated by MsbA alone.
WaaA AAE Is a Thermostable Kdo Transferase-Kdo transferases have been described to share a low degree of similarity at the primary amino acid sequence level (10,29). Comparison of the deduced amino acid sequence of 353 residues of the A. aeolicus waaA gene with the predicted sequences of other Kdo transferase genes identified a total of 81 amino acid residues within the Aquifex enzyme that are either identical or similar to amino acids being conserved across Kdo transferases of mesophilic origin (supplemental Fig. S12). Unlike its mesophilic E. coli orthologue, WaaA of A. aeolicus, as herein demonstrated by SDS-PAGE, is able to form highly stable oligomers of up to quadruple the mass of its subunit, suggesting strong intersubunit interactions that resist dissociation when exposed to denaturants, such as SDS. Moreover, we have shown that WaaA of A. aeolicus is a thermostable Kdo transferase, the stability of which falls into a range of the thermostability exhibited by other proteins isolated from hyperthermophilic bacteria (63). An increasing body of evidence shows that thermophilic enzymes are generally more robust than their mesophilic counterparts, including a higher resistance to chemical denaturation (59,64,65). Although it has become widely accepted that stabilization of thermophilic enzymes is not a result of a single type of reaction but rather occurs cumulatively at all stages of structure formation, oligomerization via stabilizing forces, such as hydrophobic interactions, hydrogen bond formation, and/or surface ion pair networks, is thought to be a major mechanism of stabilizing the proteins (58,59). Together with the detection of WaaA AAE dimers and trimers using analytical gel filtration chromatography on a calibrated Superdex 200 pg column under native conditions (data not shown), the results give reason to assume that the thermophilic Kdo transferase features the intrinsic propensity to adopt extremely stable dimeric, trimeric, or even tetrameric forms in solution, which in turn could contribute to stabilization of the protein.
WaaA AAE Exhibits Relaxed Specificity for Acceptor Substrates-Our in vitro data provide further evidence that the Kdo transferase of A. aeolicus is indeed monofunctional. The results allow us to conclude that the enzyme shares some significant properties with its mesophilic orthologues. WaaA of A. aeolicus resembles Kdo transferases of E. coli (6), A. baumannii, A. haemolyticus (10), or L. pneumophila (9) in its ability to utilize a fairly broad spectrum of acceptor substrates for glycosylation. Under our standard reaction conditions to assay Kdo transferase activity in vitro, WaaA AAE was neither selective for a defined acylation pattern of the synthetic acceptor molecules tested nor dependent upon the ␤-(136)-linked 2,3-diamino-2,3-dideoxy-D-glucopyranose disaccharide backbone of Aquifex lipid A. The latter conclusion is supported by the finding that the presence of a ␤-(136)-linked GlcN disaccharide backbone in all chemically synthesized lipid A derivatives investigated in the present study did not detrimentally affect WaaA AAE activity. Phosphorylation of the 4Ј-OH group of the lipid A precursor has been suggested to be an absolute prerequisite for the activity of distal LPS enzymes, including Kdo transferases (6,9). To the best of our knowledge, WaaA of A. aeolicus as shown by the partial conversion of the tetraacyl-1monophosphate precursor 405 into Kdo-405 is the first Kdo transferase that does not strictly depend on a negatively charged phosphate group at position 4Ј of the lipid A intermediate. Previous studies with CM derivatives of compound 506 indicated that the function of some mesophilic Kdo transferases is already abolished when the 4Ј-phosphate group of the acceptor is substituted for the less negatively charged CM moiety (66). The activity of WaaA AAE was not impaired by using either CM-506 or Bis-CM-406 as acceptor substrate, suggesting that the thermophilic Kdo transferase can tolerate a decrease in the negative net charge not only at position 4Ј but also at positions 4Ј and 1 of the lipid A carbohydrate backbone. Although we currently cannot respond to the question of whether the acceptor substrate promiscuity of WaaA AAE plays a role in biosynthesis of the A. aeolicus LPS in vivo, our in vitro results are consistent with the proposal for the existence of common although still unknown WaaA recognition motifs within the acceptor molecules (9). Although the enzyme apparently did not display any preference for most of the differently substituted lipid A acceptors tested, we cannot rule out differences in the rate of Kdo incorporation. Although ESI FT-ICR MS coupled to WaaA in vitro reactions currently does not allow the determination of kinetic values, it could be applied here for the first time to determine the simultaneous and direct monitoring of multiple WaaA substrate and product species with a high degree of specificity. Thus, MS may represent a valuable alternative to customarily used techniques that rely on multistep synthesis and purification of radioactively labeled WaaA substrates (9,10,14,55,67) and/or immunological detection of lipid A acceptors or Kdo epitopes of the glycosylated reaction products (9,10).
GalA Substitution of Lipid A Inhibits Kdo Transfer-The ability of WaaA AAE to transfer a single Kdo residue to various lipid A acceptors in vitro is virtually abolished when authentic tetraacylated Aquifex lipid A is used as a substrate. The most straightforward explanation for this finding is that the presence of two GalA residues replacing the phosphate groups at positions 4Ј and 1 of the 2,3-diamino-2,3-dideoxy-D-glucopyranose disaccharide backbone prevents the molecule from efficient glycosylation, suggesting that the acceptor binding site of WaaA AAE cannot accommodate more than a disaccharide unit of the acceptor substrate. This in turn argues for a Kdo transfer to a phosphorylated lipid A intermediate prior to GalA attachment in vivo, an assumption that is indirectly supported by the existence of an LpxK orthologue for incorporation of a phosphate moiety at O-4Ј of the lipid A backbone in A. aeolicus (24). Thus, we suspect that lipid A processing, leading to mature Kdo-glycosylated lipid A in A. aeolicus, is similar to that of Rhizobium leguminosarum, F. tularensis, or H. pylori, the lipid A of which has been shown to lack one or both phosphate groups (68 -70). In these bacteria, the biosynthesis of lipid A diverges from that of E. coli and other Gram-negative bacteria after the first seven enzymes of the constitutive E. coli lipid A pathway have generated the intermediate Kdo 2 -lipid IV A in the cytoplasmic compartment (1, 7). The lipid A molecules are subsequently completed at the periplasmic face of the inner membrane, which includes the removal of the 1-and/or 4Ј-phosphate groups by LpxE and/or LpxF as a prerequisite for additional species-specific lipid A modifications (71)(72)(73)(74)(75). Except for the UDP-2,3-diacylglucosamine hydrolase LpxH, all of the other six orthologues constituting the conserved Kdolipid IV A pathway have been discovered in A. aeolicus, namely LpxA, LpxC, LpxD, LpxB, LpxK, and WaaA (24). However, studies to shed light on the entire process of lipid A formation in A. aeolicus are hindered by the fact that no significant similarities to genes for possibly distantly related lipid A-modifying enzymes have so far been identified in the genome of the hyperthermophilic bacterium.
WaaA AAE Is Highly Specific for the Donor Substrate-Like all mesophilic Kdo transferases, WaaA AAE displayed an absolute dependence upon the presence of CTP, Kdo, and KdsB AAE (data not shown). Two lines of evidence support the conclusion that CMP-Kdo, although extremely labile at moderate temperatures (6,49), is the donor for Kdo transfer in the hyperthermophilic A. aeolicus. First, at pH 7.5 and a temperature of 80°C, KdsB of A. aeolicus was found to be highly specific for CTP and Kdo, with ATP and GTP being unable to replace CTP. UTP could act as a substitute for CTP, albeit to a much lesser extent. 8 Second, the Kdo transferase of A. aeolicus as herein monitored in real time by SPR clearly displayed affinity for CMP but not for AMP, TMP, UMP, or GMP. Although the equilibrium dissociation constant of the reaction indicated that the strength of CMP binding to WaaA AAE is moderate under the conditions examined, the results suggest that utilization of CMP-Kdo for Kdo transfer in A. aeolicus is not exclusively defined by the specificity of the CMP-Kdo synthetase for CTP and Kdo but also by a high specificity of the Kdo transferase itself, at least for the nucleotide moiety of the donor substrate. However, our SPR measurements failed to detect any affinity of WaaA AAE for Kdo. It has been shown for other glycosyltransferases that the nucleotide and sugar moieties of the donor substrate play different roles in the interaction between the enzyme and the activated nucleotide sugar, with the nucleotide moiety being crucial to recognition of the substrate and the sugar being instead involved in determining the specificity of the enzyme (76). This supports the idea that CMP may guide CMP-Kdo to the donor substrate binding site of the Kdo transferase. The lack of bind-ing of Kdo to the enzyme currently favors the hypothesis that WaaA AAE is strictly selective for the anomeric configuration of Kdo. Taking into account that Kdo in solution exists predominantly in the ␣-pyranose form, with a minor fraction of about 2% being ␤-pyranosidic (77), as well as that the latter Kdo anomer was demonstrated to be the preferred KdsB substrate for synthesis of CMP-Kdo containing Kdo in the ␤-configuration (78), we believe that Kdo must be ␤but not ␣-configured to bind to the Kdo transferase of A. aeolicus.
In summary, we have characterized the Kdo transferase from the hyperthermophilic bacterium A. aeolicus and shown that the enzyme adds a single Kdo residue to lipid A. The work presented here suggests that the Kdo transferase of the phylogenetically ancient A. aeolicus is presumably part of the conserved constitutive Kdo-lipid IV A pathway before the phosphorylated lipid A moiety is eventually processed to yield the phosphateless lipid A of Aquifex LPS.