Functional expression of Pseudomonas aeruginosa GDP-4-keto- 6-deoxy- D -mannose reductase which synthesizes GDP-rhamnose Minna MaÈki 1 , Nina JaÈ rvinen 1 , Jarkko RaÈ binaÈ 1 , Christophe Roos 2 , Hannu Maaheimo 3 , Pirkko Mattila 2 and Risto Renkonen 1 1 Department of Bacteriology and Immunology, Haartman Institute and Biomedicum, University of Helsinki, Finland; 2 MediCel, Helsinki, Finland; 3 VTT Biotechnology, Espoo, Finland; 4 HUCH Laboratory Diagnostics, Helsinki University Central Hospital, Helsinki, Finland Pseudomonas a eruginosa is an opportunis tic Gram-negative bacterium that c auses severe infections in a number of hosts from plants to mammals . A-band lipopolysaccharide of P. aeruginosa contains D -rhamnosylated O-antigen. The synthesis of GDP- D -rhamnose, the D -rhamnose donor in D -rhamnosylation, starts from GDP- D -mannose. It is ®rst converted by the GDP-mannose-4,6-dehydratase (GMD) into GDP-4-keto-6-deoxy- D -mannose, and t hen redu ced to GDP- D -rhamnose by GDP-4-keto-6-deoxy- D -mannose reductase (RMD). Here, we describe the enzymatic c har- acterization of P. aeruginosa RMD expressed in Sacchar- omyces cerevisiae. Previous success in functional expression of bacterial gm d genes i n S. cerevisiae allowedustoconvert GDP- D -mannose into GDP-4-keto-6-deoxy- D -mannose. Thus, coexpression of the Helicobacter pylori gmd and P. aeruginosa rmd genes resulted in conversion of the 4-keto-6-deoxy intermediate into GDP-deoxyhexose. This synthesized GDP-deoxyhexose was con®rmed t o be GDP- rhamnose by HPLC, matrix-assisted laser desorption/ion- ization time-of-¯ight MS, and ®nally NMR spectroscopy. The functional expression of P. aeruginosa RMD in S. cerevisiae will provide a tool for generating GDP-rham- nose f or in vitro rhamnosylation of g lycoprotein a nd glyco- peptides. Keywords: A-band O-antigen; GDP-4-keto-6-deoxy- D -mannose r eductase(RMD);GDP-rhamnose; Pseudomonas aeruginosa. Pseudomonas a eruginosa is an opportunis tic Gram-negative bacterium t hat c an cause infections in immunocompro- mised patients including th ose with severe burn wounds, cystic ®brosis and c ancer. The lipopolysaccharides that are cell surface m olecules and virulence fac tors of P. aeruginosa are both endotoxic and protective against serum-mediated lysis. The latter phenomenon is mainly due to the highly heterogeneous O-antigen (O-polysaccharide). P. aeruginosa synthesizes concomitantly two chemically distinct variants of lipopolysaccharide, designated A and B bands [1]. The A-band O-antigen is a homopolymer consisting of D -rhamnose s ugar residues arranged as repeating trisaccha- ride units ()3 D -Rhaa1±2 D -Rhaa1±3 D -Rhaa1-) n [2]. In con- trast, the B-band O-antigen is a heteropolymer composed of repeating disaccharide to pentasaccharide units of many different monosaccharides [3]. Rhamnose is a deoxyhexose sugar found widely in bacteria and plants, but not in mammals. Of the two isomers, L and D , the former is much more common. The L -isomer is found from the core oligosaccharide, B-band O-antigen and rhamnolipids of P. aeruginosa, while the D -isomer is found from the A-band O-antigen [1,4]. The route of synthesis of GDP- D -rhamnose, the precursor of D -rhamnosylated glycans, was proposed in the 1960s [5]. It starts from GDP- D -mannose, which is ®rst converted into GDP-4-keto-6-deoxy- D -mannose by GDP-mannose-4,6- dehydratase (GMD) (Fig . 1). This 4-keto-6-deoxy interme- diate can be reduced to the GDP-monodeoxyhexose, GDP- L -fucose, GDP- D -rhamnose, GDP-deoxy- D -talose or GDP-deoxy- D -altrose, by separate enzymes (in the case of GDP- L -fucose th e 3,5-epimerization occurs before reduction). It is also an intermediate in the conversion of GDP- D -mannose into the GDP-dideoxyhexose, GDP-col- itose [6], and the GDP-dideoxy amino sugar, GDP- D -per- osamine [ 7]. In the pathway o f GDP- D -rhamnose synthesis, the GDP-4-keto-6-deoxy- D -mannose reductase (RMD) is responsible for the targeted reduction of the 4-keto group, with NADH and NADPH as hydride donors [8]. The genetics of O-antigen biosynthesis in P. aeruginosa has been extensively studied by Rocchetta et al.[1].On the basis of mutagenesis analyses, it was proposed that the function of the rmd gene could be the conversion of the 4-keto-6-deoxy intermediate into GDP- D -rhamnose [9]. However, the rmd gene has not been expressed and therefore its enzymatic properties have not been characterized. The aims of this study were to provide proof of the function of the P. aeruginosa rmd geneandtosynthesize GDP-rhamnose for further glycobiological use. We have previously shown that Saccharomyces cerevisiae is an ideal host for expressing enzymes needed for the synthesis of Correspondence to R. Renkonen, Department of Bacteriology and Immunology, Haartman Institute and Biomedicum, PO Box63, FIN-00014 University of Helsinki, Helsinki, Finland. Fax: + 359 9 19125110, Tel.: + 359 9 19125111, E-mail: Risto.Renkonen@Helsinki.Fi Abbreviations: GMD, GDP-mannose-4,6-dehydratase; RMD, GDP- 4-keto-6-deoxy- D -mannose reductase; MALDI-TOF, matrix-assisted laser desorption/ionization time-of-¯ight. (Received 4 October 2001, accepted 19 November 2001) Eur. J. Biochem. 269, 593±601 (2002) Ó FEBS 2002 deoxyhexose sugar nucleotides [10] because the glycosyla- tion is largely restricted to mann osylation and yeast is not known to have deoxyhexose metabolism of its own [11,12]. Therefore, the cytoplasm of yeast cells is a relative ly rich sourc e of GD P- D -mannose, the starting material for the rhamnose p athway, without any competing e ndogenous enzymatic activity. As detection and identi®cation of various deoxyhexoses is not straightforward, the lysates of yeast transformants were analyzed by HPLC, matrix- assisted laser desorption/ionization time-of-¯ight ( MAL- DI-TOF) MS and 1 H NMR to verify the structure of the synthesized GDP-rhamnose. EXPERIMENTAL PROCEDURES Strains and culture conditions The bacterial and yeast strains and plasmids used in this study are listed in Table 1. P. aeruginosa and Escherichia coli were grown at 37 °C, and t he media used for bacterial culture and maintenance were King's broth [13] and Luria broth [14], respectively. S. cerevisiae strains were grown at 30 °C. For the S. cerevisiae host strains, the medium for culture and maintenance was YPAD medium [14], and for strains harboring pESC-LEU plasmid or its derivatives, the medium was synthetic dextrose minimal medium (SD dropout medium) lacking leucine [14]. Synthetic galactose minimal medium (SG dropout medium) lacking leucine [14] was used for the fusion protein inductions. When appro- priate, antibiotic concentrations for plasmid propagation were 50 lgámL )1 kanamycin and 100 lgámL )1 ampicillin. Recombinant DNA techniques Chromosomal DNA was isolated from P. aeruginosa ATCC 27853 using a QIAamp Tissue kit (Qiagen, Hilden, Germany). The rmd gene was a mpli®ed from chromosomal DNA u sing the primer set RMDF 5¢-GAAGATCTTA ACTCAGCGTCTGTTCGTC (creating a BglII site) and RMDR 5¢-GGTTAATTAATCAGATAAAAGGCCCG CTT (creating a PacI site). The PCR product was ®rst cloned into the pCR-XL-TOPO vector using the TOPO XL Cloning kit (Invitrogen, Carlsbad, CA, USA). The rmd gene was digested out and subcloned into the BglII/PacIsites of the pESC-LEU (Stratagene, La Jolla, CA, USA) and the pHP1 (N. Ja È rvinen, M. Ma È ki, J. Ra È bina È , C. Roos, P. Mattila & R. Renkonen, unpublished) vectors in-frame with an N-terminal FLAG epitope, yielding the corre- sponding plasmids pRHA1 and pRHA2. pHP1 is a derivative of the pESC-LEU vector containing the H. pylori gmd gene cloned in-frame with an N-terminal c-Myc epitope Table 1. Bacterial and yeast strains and plasmids. Strain/plasmid Description Reference Strain P. aeruginosa ATCC 27853 ATCC E. coli TOP10 F ± mcrA D(mrr-hsdRMS-mcrBC) /80lacZDM15 DlacX74 deoR recA1 araD139 D(ara-leu)7697 galU galK rpsL(Str R ) endA1 nupG Invitrogen S. cerevisiae YPH501 ura3±52 lys2±801 amber ade2±101 ochre trp1-D63 his3-D200 leu2-D1, mating type a/a Stratagene Plasmid pCR-XL-TOPO E. coli cloning vector Invitrogen pESC-LEU S. cerevisiae expression vector Stratagene pHP1 pESC-LEU derivate containing the H. pylori gmd gene as a 1143-bp fragment under GAL10 promoter N. Ja È rvinen et al. (unpublished) pRHA1 pESC-LEU derivate containing the P. aeruginosa rmd gene as a 915-bp fragment under GAL1 promoter This study pRHA2 pESC-LEU derivate containing the H. pylori gmd gene under GAL10 promoter and the P. aeruginosa rmd gene under GAL1 promoter This study Fig. 1. Biosynthetic pathways of the deoxyhexoses from the common 4-keto-6-deoxy intermediate. GDP- D -mannose is ®rst converted into the 4-keto-6-deoxy intermediate, which is then reduced to dierent deoxyhexoses. The enzymes involved in the GDP- D -fucose and GDP- D -rhamnose pathways have been characterized, whereas the reaction steps and enzymes involved in the GDP-deoxy- D -talose and GDP- deoxy- D -altrose pathways have not been identi®ed. 594 M. Ma È ki et al.(Eur. J. Biochem. 269) Ó FEBS 2002 (N. Ja È rvinen, M. Ma È ki, J. Ra È bina È ,C.Roos,P.Mattila& R. Renkonen, unpublished). The recombinant plasmids were sequenced on an automated A BI 3100 sequencer (PE Biosystems). pESC-LEU, pHP1, pRHA1 and pRHA2 vectors were transformed into S. cerevisiae host strains by the lithium acetate method following the instructions of the manufacturer (Stratagene). The transformants were selected on the SD dropout plates lacking leucine. Protein expression and analysis The GAL1 and GAL10 promoters of the pESC-LEU expression vector are repressed by dextro se and induced by galactose. In the expression experiments, the yeast strains were ®rst grown in the SD dropout medium overnight at 30 °C. After centrifugation, overnight cultures were inocu- lated into the same volume of the SG dropout medium and then incubated for 24 h at 3 0 °C. The yeast cells w ere harvested and ly sed with Y -PER Yeast Protein Extract ion reagent (2.5 mL for 1 g cell paste; Pierce, Rockford, IL, USA) supplemented with 5 m M MgCl 2 ,200l M GDP- D -mannose (Sigma, St Louis, MO, USA), 200 l M NADP + (Calbiochem, San Diego, CA, USA) and 200 l M NADPH (Calbiochem). The suspensions were agitated gently for 20 min at room temperature and the cell debris removed by centrifugation at 13 000 g for 10 min. The cell lysates were assayed for protein expression as well as enzyme activity. Expression of the fusion proteins was analyzed by Western blot u sing antibodies (Invitrogen) against the c-Myc and FLAG epitopes. Chemiluminescence (ECL, Amersham Pharmacia Biotech, Amersham, Bucks, UK) was used to detect the antibodies. Enzymatic reactions and preparation of nucleotide sugar samples The yeast lysates with or without GMD a nd/or RMD were incubated at 3 7 °C for 1 h in the presence of GDP- D -man- nose, NADPH, NADP + ,andMgCl 2 , after which 250 lL of the reaction mixture was subjected to puri®cation before HPLC analysis. Macromolecules were removed using PD-10 columns (Amersham Pharmacia Biotech, Uppsala, Sweden). The macromolecular fraction w as discarded, and the micromolecular fraction was collected. After being dried in a vacuum centrifuge and being redissolved, samples were treatedfor30minat37°C with 50 U alkaline phosphatase (Finnzymes, Espoo, Finland), which removed the phos- phate groups from the nucleotides but left the nucleoside diphosphate sugars intact. The reaction mixtures were diluted with 10 m M NH 4 HCO 3 and applied to Bond Elut columns (Varian, Harbor City, CA, USA) packed with 2 mL D EAE-Sepharose Fast Flow (Amersham Pharma- cia). The anion-exchange columns were washed with 10 m M and 50 m M NH 4 HCO 3 , and the nucleotide sugars were then eluted with 250 m M NH 4 HCO 3 .Afterbeingdriedand redissolved in water several times, nucleotide sugars were analyzed by HPLC. HPLC methods Nucleotide sugars were analyzed by ion-pair reversed-phase HPLC on a Supelcosil LC-18 column (0.46 ´ 25 cm; Supelco Inc., Bellafonte, PA, USA) a t a ¯ow rate of 1mLámin )1 . Isocratic 10 m M triethylammonium acetate buffer (pH 6.0) was used for 5 min, then a linear gradient of 0±3% acetonitrile in triethylammonium acetate buffer over 25 min. The ef¯uent was monitored with a UV detector at 254 nm. Size-exclusion HPLC on a Superdex Peptide HR 10/30 column (Amersham Pharmacia Biotech) was performed at a ¯ow rate of 1 mLámin )1 using 50 m M NH 4 HCO 3 ,andthe ef¯uent was monitored at 254 nm. The amount of GDP- rhamnose in both HPLC methods was calculated from the peak areas by reference to an external standard (GDP- L -fucose; Calbiochem). The samples containing GDP- sugars were collected from HPLC runs for structural analysis with MALDI-TOF MS and NMR. Maldi-tof ms MALDI-TOF MS was performed with a Bi¯ex mass spectrometer (Bruker Daltonics, Leipzig, Germany). Analysis was performed in the negative-ion linear delayed- extraction mode, using 2,4,6-trihydroxyacetonephenone (Fluka Chemica) as a matrix [15]. External calibration was performed with this matrix dimer and sialyl Lewis X b-methylglycoside (Toronto Research Chemicals, Toronto, Ontario, Canada). NMR experiments An 11-nmol sample of GDP-rhamnose was dissolved in 300 lLD 2 O (Aldrich) and freeze-dried. The sample was thendissolvedin40lLD 2 O. All NMR experiments w ere carried out a t 35 °C on a 500-MHz Varian Inova spectrometer equipped with a nanoprobe. The 1D 1 H- NMR spectrum was recorded using a modi®cation of the weft sequence for water suppression [16]. A total of 4096 transients were acquired with a spectral width of 6100 Hz. For the DQF COSY spectrum [17], a total of 4000 ´ 256 complex data points were acquired, 256 transients per increment. Before Fourier transformation, the data matrix was multiplied by a cosine function in both dimensions. The 1 H chemical shifts were referenced to external 3-(trimethyl- silyl)propionic-2,2,3,3-d4 acid (d  0). Sequence analysis The tools used for homology searches were mostly BLAST [18], FASTA [19], the Smith±Waterman implementation and other programs available in the GCG package (Wisconsin Package, version 10.0; Genetics Computer Group, Madison, WI, USA). DNA sequences were aligned with the pro gram PILEUP (Wisconsin Package) or ClustalW (version 1.7) [20], using an identity matrix, a gap weight of 8, and a gap length weight of 0.1. Amino-acid sequences were aligned with the same programs using a Blosum32 protein w eight matrix, a gap weight of 12, and a gap length weight of 0.5. The DNA alignments were checked by eye using the GENEDOC program [21] and corrected to avoid alignments with disrupted reading frames. Trees were constructed from the data using maximum parsimony using programs from the PHYLIP [22] package and the GCG implementation of PAUP * (Wisconsin Package). H euristic se arches were utilized in parsimony analyses because o f the great number of taxa examined. Branch swapping was done by tree bisection±reconnection. Ó FEBS 2002 Biosynthesis of GDP-rhamnose (Eur. J. Biochem. 269) 595 Bootstrap analyses (unshown) of 1000 replicates were performed to examine the relative support of each relation- ship in the resultant topologies. GeneDoc an d TreeView [23] were used to prepare illustrations of the alignments and the trees. RESULTS Cloning of P. aeruginosa rmd into a S. cerevisiae expression vector We have previously cloned the gmd and wcaG genes of E. coli and the gmd and wbcJ genes of H. pylori into pESC- LEU, a yeast expression vector with two separate multiple cloning sites. With these double constructs, we generated S. ce revisiae strains that converted yeast endogenous GDP- D -mannose into GDP- L -fucose ([7]; N. Ja È rvinen,M.Ma È ki, J. Ra È bina È , C. Roos, P. Mattila & R. Renkonen, unpub- lished). In the present study, we aimed to produce GDP- rhamnose from the 4-keto-6-deoxy intermediate synthesized by the H. pylori GMD enzyme, and t herefore we identi®ed, cloned, and expressed the P. aeruginosa rmd gene together with the H. pylori gmd gene. TheA-bandgeneclusterofP. aeruginosa (EMBL/ GenBank/DDBJ accession number AE004958) containing the putative rmd gene was obtained from the database. The size of the putative rmd gene was 915 bp, and the starting codon was TTG, not ATG as suggested by the study of Rocchetta et al.[9].Thermd gene was ampli®ed from P. aeruginosa ATCC 27853 chromosomal DNA and cloned into the expression vectors, pESC-LEU and pHP1 in-frame with an N-terminal FLAG epitope , yielding pRHA1 and pRHA2, respectively (Fig. 2). The sequenced plasmids pHP1, pRHA1 and pRHA2 were subsequently transformed into the expression strain S. ce revisiae YPH501. Expression of GMD and RMD proteins in S. cerevisiae The H. pylori gmd and P. aeruginosa rmd genes were expressed under g alactose-inducible promoters of the pESC-LEU vector tagged with the c-Myc and FLAG epitopes, respectively (Fig. 2). After induction, expression of GMD and RMD proteins was analyzed in the yeast lysates by Western immunoblots using antibodies against the c-Myc and FLAG epitopes. As shown in Fig. 3, the presence of the 44-kDa GMD could be detected in the lysates of S. ce revisiae YPH501(pHP1) and YPH501(pRHA2). The de no vo expressed FLAG-tagged putative RMD protein was present in the cell lysates of YPH501(pRHA1) and YPH501(pRHA2) (Fig. 3). The size o f this protein wa s 34 k Da, which corresponded to the calculated molecular mass of RMD (33.9 kDa). No relevant bands could be detected from the lysate of the YPH501(pESC-LEU) strain used as a vector control (Fig. 3). Characterization of enzymatic activities To show that GMD and RMD were functionally active, we performed a thorough analysis of sugar nucleotides formed in reactions of yeast lysates with exogenously added GDP- D -mannose. The s ugar nucleotides from yeast lysates were analyzed by HPLC, MALDI-TOF MS and 1 HNMR. The ion-pair reversed-phase HPLC analysis (Fig. 4 ) showed that the vector control S. cerev isiae YPH501 (pESC-LEU) gave only a peak with the same retention time as the GDP- D -mannose standard at 17.4 min and some uncharacterized peaks from yeast cells (Fig. 4A). The peaks from the yeast strain YPH501(pRHA1) expressin g RMD were similar to the vector control (Fig. 4B). In contrast, YPH501(pHP1) expressing GMD gave a smaller GDP- D -mannose peak and a novel peak at 21.2 min (Fig. 4C). As no standard was a vailable for this intermediate product, we isolated it from HPLC and tried to analyze its mass by MALDI-TOF MS. However, we could not analyze it with con®dence, possibly b ecause 4-keto-6-deoxysugars are known to be labile. The presence of the 4-keto-6-deoxy intermediate in the reaction mixture was also studied by chemical reduction with NaBH 4 . As expected, two new peaks were detected by HPLC (not shown). The minor peak co-migrated with the GDP- D -rhamnose standard, and the retention time of the major peak was near that of the 4-keto intermediate product. The latter is probably GDP-deoxy- D -talose, but we were not able to con®rm this because of the lack of a standard for this nucleotide sugar. Fig. 2. Schematic drawing of the pRHA2 plasmid. The pRHA2 plas- mid is a derivative of the yeast expression vector, pESC-LEU. The H. pylori gmd gene was s ubclon ed under the GAL1 promoter in-frame with the c-Myc epitope, and the P. aeruginosa rmd ge ne was subclon ed under the GAL10 promoter in-frame with the FLAG epitope. Fig. 3. Western blots of the expression of GMD and RMD in S. cere- visiae YPH501 detected with c-Myc antibody (A) and FLAG antibody (B). Lane 1, YPH501(pESC-LEU), the vector control; lane 2, YPH501(pHP1); lane 3, YPH501(pRHA1); lane 4, YPH501(pRHA2). The pre sence of 44-kDa H. pylori GMD was detected in lanes 2 and 4, and the presence of 34-kD a P. aeruginosa RMD in lanes 3 and 4. 596 M. Ma È ki et al.(Eur. J. Biochem. 269) Ó FEBS 2002 In HPLC analysis of the double-construct strain YPH501(pRHA2) expressing both GMD and RMD, the 4-keto-6-deoxy intermediate was not detected, but a novel peak appeared at 19.2 min (Fig. 4D). Once again, there was no commercially available s tandard for GDP- D -rhamnose, but we could show that the molecule with a retention time of 19.2 min gave a single peak at m/z 588.05 on MALDI-TOF MS analysis, which is the mass of GDP-deoxyhexoses (calculated m/z for [ M-H] ± is 588.08). This new GDP- deoxyhexose peak, together with the putative 4-keto- 6-deoxy intermediate peak, was also seen when the lysates of YPH501(pHP1), expressing GMD, and YPH501 (pRHA1), expressing RMD, were mixed together (Fig. 4E). Formation of the putative GDP-rhamnose product was further increased in this reaction, as compared with the double-construct strain YPH501(pRHA2). As the retention time of the reaction product was different from the GDP- L -fucose standard in ion-pair reversed-phase HPLC (19.2 and 21.6 min, respectively), it clearly represented a novel GDP-deoxyhexose, and we therefore analyzed it further by NMR (see below). Preparative synthesis and puri®cation of GDP-rhamnose The ®nal reaction product was puri®ed from the reaction mixture to con®rm its structure and con®guration. In the large-scale puri®cation of GDP-rhamnose for NMR ana- lysis, new sample preparation techniques developed in our laboratory for nucleotide sugars were used (J. Ra È bina È ,M. Ma È ki,N.Ja È rvinen, E. Saulahti & R. Renkonen, unpub- lished results). Sh ortly, after the puri®cation on Envi-Carb graphite columns (Supelco), DEAE-Sepharose anion- exchange chromatography and r eversed-phase HPLC iden- tical with the analytical runs (see above) were performed. After further puri®cation by size-exclusion HPLC,  11 nmol sugar nucleotide was pooled from several HPLC runs. The yield of GDP-rhamnose after the puri®cation steps was determined from HPLC of the product (see Experi- mental procedures). As calculated from the GDP- D -man- nose added t o the cell e xtracts, 3±4% of the substrate was converted into GDP-rhamnose in the independent experi- ments with the double-construct strain, S. cerevisiae YPH501(pRHA2). When the yeast lysates of strains YPH501(pHP1), expressing GMD, and YPH501(pRHA1), expressing RMD, were mixed and incubated together with GDP- D -mannose, the yield of GDP-rhamnose was 9%. NMR analysis The 1 H-NMR spectrum (Fig. 5) of the puri®ed 19.2 min peak from the HPLC pro®le (Fig. 4) was assigned, and the proton±proton coupling constants 3 J H,H (Table 2) were determined from a DQF COSY spectrum (not shown). The NMR results established the structure as GDP-rhamnose, propably the D -isomer. The coupling constants between the ring protons of the rhamnosyl unit were characteristic of a manno- con®guration and clearly distinguished the struc- ture from GDP- L -fucose. The H6 signal at 1.270 p.p.m. was on the region of a methyl group, and had the intensity of three protons. T he large g eminal coupling t ypical of a hydroxymethyl group was not observed. The chemical shifts obtained were similar to those published by Kneidinger et al. [8] for GDP- D -rhamnose and different from those reported for GDP- L -fucose [24]. Conservation of RMD protein homologs among different bacteria After the enzymatic function had been con®rmed, th e RMD protein of P. aeruginosa was u sed as a probe to ®nd more putative RMD sequences from the databases based on the primary sequence similarity. Relatively high homologies were found with the Aneurinibacillus thermoaerophilus RMD sequence (EMBL/GenBank/DDBJ accession num- ber AF317224) as well as with three other bacterial ORFs of Thiobacillus ferrooxidans (the TIGR accession number gnl|TIGR|t_ferrooxidans_6147), Mycobacterium tuberculo- sis (EMBL/GenBank/DDBJ accession number AL123456) and Xylella fastidiosa (EMBL/GenBank/DDBJ accession number AE003849). Signi®cant similarities were also found with GDP-mannose dehydratase (EC 4.2.1.47), dTDP- glucose dehydratase (EC 4.2.1.46) and UDP-glucose epi- merase (EC 5.1.3.2) protein families. Therefore, we aligned the selected gene sequences from the four enzyme families t o evaluate the distance inter se. As a measure of distance w e used the mutation rate, and, to visualize the result, we used standard phylogenetic tools. The analysis showed that, while the different genes clustered according to their Fig. 4. HPLC analysis of the products of enzymatic reactions catalyzed by H. pylori GMD and P. aeruginosa RMD. GDP- D -mannose was incubated with t he lysates o f S. cere visiae YPH501 recombinant strains. (A) YPH501(pESC-LEU), the vector control; (B) YPH501 (pRHA1) e xpressing RMD ; (C) YPH501(pHP1) expressing GMD; (D) YHP501(pRHA2) coexpressing GMD and RMD; (E) YPH501(pHP1) and YPH501(pRHA1), mixture of the lysates of singularly expressed GMD and RMD. Peaks: M  GDP- D -man- nose; R  GDP-rhamnose; K  GDP-4-keto-6-deoxy- D -mannose. Ó FEBS 2002 Biosynthesis of GDP-rhamnose (Eur. J. Biochem. 269) 597 proposed function (Fig. 6), their relatedness within the group was not much higher than between the groups. This was further analyzed by aligning the sequences from the GDP-mannose reductase group (EC 1.1.1.187): the seq- uences of P. aeruginosa and A. thermoaerophilus with proven RMD activities, and the ORFs of T. ferrooxidans, M. tuberculosis and X. fastidiosa with putative RMD activity. As can be deduced from the alignment (Fig. 7), the sequences did not have any major stretches of similarity, but rath er short patterns, most of which are also found in the other three enzyme families (unshown). The ®rst draft of the human genome (http://www. celera.com) was also probed with the P. aeruginosa RMD sequence, but no RMD analogue was found. DISCUSSION In this paper, we describe the molecular identi®cation of the P. aeruginosa rmd gene and the enzymatic characterization of the corresponding recombinant enzyme expressed in S. ce revisiae YPH501. Using the yeast expression system, we have had previous success in converting yeast endoge- nous GDP- D -mannose into GDP- L -fucose by functionally active E. coli GMD and GMER enzymes [10]. Now, we also have a yeast expression system for synthesizing GDP- rhamnose. Our results indicate that the yeast lysate can convert exogenously added GDP- D -mannose into GDP- rhamnose when P. aeruginosa RMD is e xpressed together with H. pylori GMD. Because the level of yeast endogenous GDP- D -mannose is probably not high enough for abundant GDP-rhamnose production, we added exogenous GDP- D -mannose to the reaction mixture. It is likely that there are Fig. 5. Structure of GDP- D -rhamnose (A) and expansion and assignments of 500-MHz 1 H-NMR spectrum of GDP-rhamnose at 35 °C (B). The signals arising from a small fraction of impurities present in the sample are marked with asterisks. In addition to the signals shown in this expansion , H8 resonance of the guanine unit was observed at 8.123 p.p.m. Table 2 . 1 H chemical s hifts and coupling constants of GDP- a- D -rhamnose. Chemical shifts were measured at 35 °C with reference to external 3-(trimethylsilyl)propionic-2,2,3,3,-d4 a cid. 3 J H, H+1 val- ues from ®rst-order analysis of the DQF COSY spectrum. ND, Not determined. Residue Proton Chemical shift ( p.p.m.) 3 J H, H+1 (Hz) 3 J H,P (Hz) Rhamnose 1 5.446 < 2 6 2 4.051 2.6 ± 3 3.886 9.3 ± 4 3.435 10.0 ± 5 3.923 6.3 ± 6 1.270 ± ± Ribose 1 5.942 ND ND 2 4.808 ND ± 3 4.529 ND ± 4 4.359 ND ± 5, 5¢ 4.22 ND ± Guanine 8 8.123 ± ± 598 M. Ma È ki et al.(Eur. J. Biochem. 269) Ó FEBS 2002 enzymes other than H. pylori GMD in the crude yeast lysate, such as mannosyltransferases [25,26], which also compete for GDP- D -mannose. From the HPLC pro®les, we calculated that most of the exogenously added GDP- D -mannose is converted into something other than GDP- rhamnose (data not shown). However, puri®cation of the GMD and RMD e nzymes and optimization of the reaction conditions would probably lead to increased GDP-rham- nose yield. The biosynthesis of GDP- D -rhamnose, which acts as a nucleotide sugar donor for D -rhamnosylation, received increased interest after D -rhamnose was shown to b e an essential extracellular and cell wall component of several pathogenic bacteria [27]. P. aeruginosa is commonly isolated Fig. 6. Grouping of four enzyme families UDP-glucose-4-epimerases a , dDTP-glucose-4,6-dehydratases b , GDP-4-keto-6-deoxy- D -mannose reductases c and GDP-mannose-4,6-dehydratases d on the basis of their sequence similarity. The scale bar indicates a ÔdistanceÕ in numb er of mutations per site and the asterisks indicate the functionally characterized enzymes. EMBL/GenBank/DDBJ and TIGR accession numbers of the used gene sequences: M. tuberculosis H37Rv (Z95436 a , Z95390 b , AL123456 c , AL021926 d ); P. aeruginosa PAO1 (gnl|TIGR|PAGP_287 a , AE004929 b , AE004958 c , U18320 d ); E. coli K12 (X06226 a , AE000294 b , U38473 d ); X. fastidiosa 9a5c (AE003906 a,c , AE003849 d ); A. thermoaerophilus L420-91 T (AF317224 c , AF317224 d ); T. ferrooxidans ATCC 23270 (gnl|TIGR|t_ferrooxidans_6147 c ); S. enterica serovar typhimurium (X56793 b ); H. pylori J99 (AE001443 d ). Fig. 7. Alignment of the known and putative RMDs from P. aeruginosa, A. thermoaerophilus, T. ferrooxidans, M. tu bercul osis and X. fastidiosa. The alignment emphasizes the c onserved motifs, and the parentheses mark those conserved in the three other enzyme families (Fig. 6) as we ll. Ó FEBS 2002 Biosynthesis of GDP-rhamnose (Eur. J. Biochem. 269) 599 from specimens obtained from lungs of patients with cystic ®brosis, and the respiratory P. aeruginosa isolates express mainly D -rhamnosylated A-band lipopolysaccharide [1]. Interestingly, the same O-polysaccharide structure has been isolated from the other opportunistic patho gens, Burkholderia cepacia and Stenotrophomonas maltophilia, that are also linked to this congenital monogenic disease with severe pulmonary manifestations [28±30]. Research groups studying the relevance of rhamnosyla- tion of various bacteria would bene®t from availability of the building blocks required for synthesis of rhamnosylated molecules. However, before these molecules can be synthe- sized in vitro, the activated sugar nucleotides, GDP- D -rhamnose and dTDP- L -rhamnose, and the corresponding rhamnosyltransferases that catalyze the speci®c glycosidic linkages are needed. Two enzymes responsible for RMD activity have recently been characterized from the nonpathogenic bacterium A. thermoaerophilus [8]. A. thermoaerophilus is a Gram- positive bacterium, and the D -rhamnose residues are found in the extracellular S-layer. A. thermoaerophilus GMD has been proposed to be bifunctional, acting as a GDP- mannose-4,6-dehydratase and a reductase. The latter activ- ity was relatively weak compared with A. thermoaerophilus RMD acting only as a reductase. In our analysis, we could not show the bifunctionality of the H. pylori GMD enzyme, which suggests that the speci®city of GMD enzymes varies between bacterial species. Currently, very little is known about bacterial rhamno- syltransf era ses. L -Rhamnosyltransferases, which use dTDP- L -rhamnose as a donor, have been reported in several bacteria [31±33], whereas putative D -rhamnosyltransferases, which use GDP- D -rhamnose as a donor, have only been identi®ed in P. aeruginosa [1] . It has been shown by muta- genesis studies that these three putative D -rhamnosyltransfe- rases participate in the synthesis of the A-band O-antigen. If rhamnosylation is con®rmed to be essential for the viability or virulence of pathogenic bacteria, the enzymes involved in the biosynthesis of rhamnosylated glycans could be ideal targets for antibacterial chemotherapy. Human patients lack rhamnosylation and thus would probably not suffer if enzymes involved in rhamnosylation were inhibited. ACKNOWLEDGEMENTS The work was supported in part by Research Grants f rom the Academy of Finland,Technology Development Centre (TEKES), Helsinki, and the Sigrid Juselius F oundation and a grant from the Helsinki University Central Hospital Fund . We thank Dr Jari Helin and Leena Penttila È for the MALDI-TOF M S analysis. Sirkka-Liisa Kaur anen and T uula Kallioinen are thanked for skilled technical assistance with the molecular biology, and Jonna-Mari Ma È ki for invaluable help with the ®gures. REFERENCES 1. Rocchetta, H.L., Burrows, L.L. & Lam, J.S. (1999) Genetics of O-antigen biosynthesis in Pseudomonas aeruginosa. Microbiol. Mol. Biol. Rev. 63, 523±553. 2. Arsenault, T.L., Huges, D.W., MacLean, D.B., Szarek, W.A., Kropinski, A.M. & Lam, J.S. (1991) Structural studies on the polysaccharide p ortion of ÔA-bandÕ lipopolysaccharide from a mutant (AK14O1) of Pseudomonas aeruginosa strain PAO1. Can. J. Chem. 69, 1273±1280. 3. Knirel, Y.A. & Kochetkov, N.K. (1994) The structure of lipo- polysaccharides of gram-negative bacteria. III. The structure of O-antigens: a review. Biochemistry 59, 1325±1382. 4. Sadovskaya, I., Brisson, J.R., Thibault, P., Richards, J.C., Lam, J.S. & Altman, E. (2000) Structural characterization of the outer core and the O-chain linkage region of lipopolysaccharide from Pseudomonas aeruginosa serotype O5. Eur. J. Biochem. 267, 1640± 1650. 5. Barber, G.A. (1969) The synthesis of guanosine 5¢-diphosphate D -rhamnose. Biochemistry 8, 3692±3695. 6. Elbein, A.D. & H.E. (1965) The biosynthesis of cell wall lipopolysaccharide in Escherichia coli. J. Biol. Chem. 240, 1926± 1931. 7. Albermann, C. & Piepersberg, W. (2001) Expression and identi- ®cation of the RfbE protein from Vibrio cholerae O1 and its use for the enzymatic synthesis of GDP- D -perosamine. Glycobiology 11, 655±661. 8. Kneidinger, B., Graninger, M ., Adam, G., Puchberger, M., Kosma, P., Zayni, S. & Messner, P. (2001) Identi®cation of two GDP-6-deoxy-D -lyxo-4-hexulose reductases synthesizing GDP- D-rhamnose in Aneurinibacillus thermoaerophilus L420-91 T *. J. Biol. Chem. 276, 5577±5583. 9. Rocchetta, H.L., Pacan, J.C. & Lam, J.S. ( 1998) Synthesis of the A-band polysaccharide s ugar D -rhamnose requires Rmd and WbpW: identi®cation of multiple AlgA homologues, WbpW and ORF488 in Pseudomonas aeruginosa. Mol. Microbiol. 29, 1419± 1434 (erratum appears in Mol. Microbiol. 31, 397±398). 10. Mattila, P., Ra È bina È , J., Hortling, S., Helin, J. & Renkonen, R. (2000) Functional expression of Escherichia coli enzymes synthe- sizing GDP- L -fucose from inherent GDP- D -mannose in Sacchar- omyces cerevisiae. Glycobiology 10, 1041±1047. 11. Hashimoto, H., Sakakibara, A., Yamasaki, M. & Yoda, K. (1997) Saccharomyces cerevisiae VIG9 encodes GDP-mannose pyro- phosphorylase, which is essential for protein glycosylation. J. Biol. Chem. 272, 16308±16314. 12. Romanos, M.A., Scorer, C.A. & Clare, J.J. (1992) Foreign gene expression in yeast: a review. Yeast 8, 423±488. 13. King, E.O., Ward, M.K. & Raney, D.E. (1954) Two simple media for t he demonstration of phycocyanin and ¯uorescin. J. Lab. Clin. Med. 44, 301±307. 14. Sambrook, J. & Russell, D.W. (2001) Molecular Cloning. A Lab- oratory M anual, 3rd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. 15. Nyman, T.A., Kalkkinen, N., Tolo, H. & Helin, J. (1998) Struc- tural characterisation of N-linked and O-linked oligosaccharides derived from interferon-alpha2b and interferon-alpha14c pro- duced by Sendai-virus-induced human peripheral blood leuko- cytes. Eur. J. Biochem. 25 3, 485±493. 16. Ha Ê rd, K., van Zadelho, G., Moonen, P., Kamerling, J.P. & Vliegenthart, J.F.G. (1992) The Asn-linked carbohydrate chains of human Tamm-Horsfall glycoprotein of one male. Novel sulfated and novel N-acetylgalactosamine-containing N-link ed carbo- hydrate chains. Eur. J. Biochem. 209, 895±915. 17. Rance, M., Sùrensen, O.W., Bodenhausen, G., Wagner, G., Ernst, R.R. & Wu È thrich, K. (1983) Improved spec tral resolution in COSY 1 H NMR spectra o f proteins v ia double quantum ®ltering. Biochem. Biophys. Res. Commun. 117, 479±485. 18. Altschul, S.F., Madden, T.L., Scha È er, A.A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D.J. (1997) Gapped BLAST and PSI- BLAST: a new generation of protein database searc h programs. Nucleic Acids Res. 25, 3389±3402. 19. Pearson, W.R. (1990) Rapid and sensitive sequence com- parison with FASTP and FASTA. Methods Enzymol. 183, 63±98. 20. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment throu gh se qu ence weighting, position speci®c g ap 600 M. Ma È ki et al.(Eur. J. Biochem. 269) Ó FEBS 2002 penalties and weight matrix choice. Nucleic Acids Res. 22, 4673±4680. 21. Nicholas, K.B. & Nicholas, H.B.J. (1997) GeneDoc: a tool for editing and annotating multiple sequence alignments, available at http://www.psc.edu/biomed/genedoc 22. Felsentstein, J. (1993) Phylip, Phylogeny Inference Package. University of Washington, Seattle, version 3.5c, available at http://evolution.genetics.washington.edu/phylip/phylip.html 23. Page, R.D. (1996) TreeView: an application to display phyloge- netic trees on personal computers. Comput. Appl. Biosci. 12, 357±358. 24. Adelhorst,K.&Whitesides,G.M.(1993)Large-scalesynthesisof b- L -fucopyranosyl phosphate and the preparation of GDP- b- L -fucose. Carbohyd. Res. 242, 69±76. 25. Kojima,H.,Hashimoto,H.&Yoda,K.(1999)Interactionamong the sub unit s of Golgi m embrane mannosyltransferase c omplexes of the y east Saccharomyces cerevisiae. Biosci. Biotechnol. Biochem. 63, 1970±1976. 26. Todorow, Z., Spang, A., Carmack, E., Yates, J. & Schekman, R. (2000) Active recycling of yeast Golgi mannosyltransferase com- plexes through the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA 97, 13643±13648. 27. Giraud, M F. & Naismith, J.H. (2000) The rhamnose pathway. Curr. Opin. Struct. Biol. 10, 687±696. 28. Cerantola, S. & Montrozier, H. (1997) Structural elucidation of two poly saccha rides present in th e lipopolysaccharide of a clinical isolate of Burkholderia cepacia. Eur. J. Biochem. 246, 360±366. 29. Winn, A.M. & Wilkinson, S.G. (1998) The O7 antigen of Steno- trophomonas maltophilia is a linear D -rhamnan with a trisaccharide repeating unit that is also present in polym ers for s ome Pseudo- monas and Burkholderia species. FEMS Microbiol. Lett. 166, 57±61. 30. LiPuma, J.J. (2000) Expanding microbiology of pulmonary infectionincystic®brosis.Pediatr. Infect. Dis. J. 19, 473±474. 31. Ochsner, U.A., Koch, A.K., Fiechter, A. & Reiser, J. (1994) Iso- lation and characterization of a regulatory gene aecting rha- mnolipid biosurfactant synthesis in Pseudomonas aeruginosa. J. Bacteriol. 176, 2044±2054. 32. Reeves, P.R., Hobbs, M., Valvano, M.A., Skurnik, M., Whit®eld, C., Coplin, D., Kido, N., Klena, J., Maskell, D., Raetz, C.R. & Rick, P.D. (1996) Bacterial polysaccharide synthesis and gene nomenclature. Trends Microbiol. 4, 495±503. 33. Eckstein, T.M., Silbaq, F.S., Chatterjee, D., Kelly, N.J., Brennan, P.J. & Belisle, J.T. (1998) Identi®cation and recombinant expres- sion of a Mycobacterium avium rhamnosyltransferase gene (rtfA) involved in glycopeptidolipid biosynthesis. J. Bacteriol. 180, 5567± 5573. Ó FEBS 2002 Biosynthesis of GDP-rhamnose (Eur. J. Biochem. 269) 601 . Functional expression of Pseudomonas aeruginosa GDP-4-keto- 6-deoxy- D -mannose reductase which synthesizes GDP-rhamnose Minna MaÈki 1 ,. been characterized. The aims of this study were to provide proof of the function of the P. aeruginosa rmd geneandtosynthesize GDP-rhamnose for further glycobiological