Comparative Analysis of Two UDP-glucose Dehydrogenases in Pseudomonas aeruginosa PAO1*

UDP-glucose dehydrogenase (UGDH) catalyzes a two-step NAD+-dependent oxidation of UDP-glucose to produce UDP-glucuronic acid, which is a common substrate for the biosynthesis of exopolysaccharide. Searching the Pseudomonas aeruginosa PAO1 genome data base for a UGDH has helped identify two open reading frames, PA2022 and PA3559, which may encode a UGDH. To elucidate their enzymatic identity, the two genes were cloned and overexpressed in Escherichia coli, and the recombinant proteins were purified. Both the gene products are active as dimers and are capable of utilizing UDP-glucose as a substrate to generate UDP-glucuronic acid. The Km values of PA2022 and PA3559 for UDP-glucose are ∼0.1 and 0.4 mm, whereas the Km values for NAD+ are 0.5 and 2.0 mm, respectively. Compared with PA3559, PA2022 exhibits broader substrate specificity, utilizing TDP-glucose and UDP-N-acetylglucosamine with one-third the velocity of that with UDP-glucose. The PA2022 mutant and PA2022-PA3559 double mutant, but not the PA3559 mutant, are more susceptible to chloramphenicol, cefotaxime, and ampicillin. The PA3559 mutant, however, shows a reduced resistance to polymyxin B compared with wild type PAO1. Finally, real time PCR analysis indicates that PA3559 is expressed primarily in low concentrations of Mg2+, which contrasts with the constitutive expression of PA2022. Although both the enzymes catalyze the same reaction, their enzymatic properties and gene expression profiles indicate that they play distinct physiological roles in P. aeruginosa, as reflected by different phenotypes displayed by the mutants.

UDP-glucose dehydrogenase (UGDH) 2 (EC 1.1.1.22) converts UDP-glucose (UDP-Glc) to UDP-glucuronic acid (UDP-GlcUA) accompanied by the reduction of two molecules of NAD ϩ . The importance of UDP-GlcUA is readily apparent considering the downstream polymers that utilize this compound or its descendents, such as UDP-xylose, UDP-arabinose, and UDP-galacturonic acid (1). The cellular functions of UGDH have been investigated in a number of organisms, and these studies demonstrate its importance in detoxification, polysaccharide biosynthesis, and embryonic development (2). In the liver, glucuronide conjugation of xeno-and endobiotic compounds by UDP-glucuronosyltransferases is an important way to aid in solubilizing these metabolites (3). In plants, both UDP-GlcUA and its derivative, UDP-xylose, are essential to synthesizing cell wall polysaccharides such as hemicellulose (4). Animals require UDP-GlcUA to synthesize major structural polysaccharides such as hyaluronic acid (HA), heparan sulfate, and chondroitin sulfate (5). An inability to synthesize these structural polysaccharides may partly explain why a UGDH deficiency disrupts the Wingless signal pathway in Drosophila melanogaster (6 -8) and arrests mesoderm and endoderm migration during gastrulation in mice (9).
In many pathogenic bacteria, UDP-GlcUA is necessary for the synthesis of exo-polysaccharides (EPS) and lipopolysaccharides (LPS). The formation of these polysaccharides is critical to bacterial virulence (10,11) because it enables the bacteria to evade attacks by a host immune system, such as phagocytosis (12,13). Some pathogens, including group A and C streptococci, can even synthesize nonimmunogenic EPS such as HA to prevent antibody-mediated killing (14). Recent studies demonstrate that UGDH mutation in the pathogenic fungus Cryptococcus neoformans alters the cell integrity and nucleotide sugar pool. The cells also become temperature-sensitive and fail to grow in an animal model (15).
Pseudomonas aeruginosa, a Gram-negative bacterium, is noted for its environmental versatility, resistance to antibiotics, and ability to cause life-threatening infections in cystic fibrosis, burn, and immunocompromised patients (16). Epidemiological data indicate that P. aeruginosa is the third leading cause of nosocomial infection (17). The significance of EPS in P. aeruginosa virulence is well documented in the cystic fibrosis strains, in which the synthesis of a large quantity of alginate has frequently been observed. However, many other pathogenic P. aeruginosa strains, such as PAO1, do not normally produce alginate. The precise structure and composition of the EPS and the genes required for EPS synthesis in these P. aeruginosa strains remain unclear. Because UDP-GlcUA is one of the most commonly found sugar donors for bacterial EPS synthesis, this study attempts to identify the UGDH encoding gene in P. aeruginosa PAO1 and to investigate the biochemistry and functional roles of this enzyme in this important pathogen. This study first demonstrates that at least two P. aeruginosa genes can encode a product with UGDH activity, each with a distinct expression pattern and enzyme kinetic parameters. The presence of two types of UGDH motivated the construction of deletion mutants for each of the genes, and this study analyzes and reports on the functional roles of these genes in several biological properties of P. aeruginosa PAO1. Table 1 lists the bacterial strains and plasmids used in this study. Escherichia coli and P. aeruginosa were cultured in standard Luria-Bertani broth (LB: 10 g/liter tryptone, 5 g/liter yeast extract, 10 g/liter sodium chloride) or on LB agar. The basal medium employed in the determination of minimal inhibition concentration (MIC) of antibiotics and in RNA extraction for quantitative PCR contained 30 mM glucose, 40 mM K 2 HPO 4 , 22 mM KH 2 PO 4 , 7 mM (NH 4 ) 2 SO 4 , and either 2.0 or 0.02 mM MgSO 4 for Mg 2ϩ -sufficient or Mg 2ϩ -deficient conditions, respectively. The following antibiotic concentrations were used to select recombinant E. coli: tetracycline, 10 g/ml; ampicillin, 100 g/ml; kanamycin, 25 g/ml. The concentration of tetracycline for selecting recombinant P. aeruginosa was 50 g/ml. Unless indicated oth-erwise, all chemicals were obtained from Sigma and were of the highest purity available. Table 2 lists the oligonucleotide primers used in this study.

Bacterial Strains and Growth Conditions, Plasmids, Oligonucleotide Primers, and Chemicals-
Recombinant DNA Manipulation-Recombinant DNA experiments were carried out by standard procedures as described (18). Restriction endonucleases and DNA-modifying enzymes were purchased from either New England Biolabs (Beverly, MA) or MBI Fermentas (Hanover, MD) and were used according to supplier recommendations. The PA2022 and PA3559 overexpression clones were constructed using PCR amplification of the coding region of each respective gene. Specific primer sets and PAO1 genomic DNA served as the template in the reaction. The nucleotide sequence of PA3559 and PA2022 was retrieved from the Pseudomonas Genome Project. The forward and reverse primers for PA2022 were HY1001 and HY1002, and those for PA3559 were RJ004 and RJ003, containing BamHI and XhoI restriction endonuclease sites, respectively. The PCR product was cloned into the pET30a expression vector at corresponding restriction sites to be in-frame with the His 6 tag at the N terminus of the protein. The constructs were then introduced into E. coli NovaBlue (DE3) by the rubidium chloride method, and the transformants were selected on LB agar containing 25 g/ml kanamycin. The DNA was sequenced to verify the correctness of the cloned gene and the reading frame fusion with the His 6 tag.  Overexpression and Purification of His 6 -tagged PA2022 and PA3559 Products-The production of PA2022 and PA3559 in E. coli was carried out by growing the recombinant strains in LB broth supplemented with 25 g/ml kanamycin and 10 g/ml tetracycline. The cells were first cultivated by shaking at 37°C until the A 600 ϭ 0.5. Isopropyl thio-␤-D-galactopyranoside was subsequently added at a final concentration of 0.2 mM and growth continued for 5 h at 24°C under shaking. Cells were harvested by centrifugation at 4,000 ϫ g for 10 min, resuspended in binding buffer (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 5 mM imidazole), and lysed by sonication on ice with a macroprobe sonicator (model XL2020, Misonix) at a power setting of 4 for 15 min. After centrifugation at 5000 rpm at 4°C for 20 min, the supernatant was applied to a nickel-Sepharose column (GE Healthcare), and the chromatography was performed as recommended by the manufacturer. The eluting buffer consisted of 500 mM imidazole. The purity of the protein was then evaluated by SDS-PAGE.
Ion Pair HPLC Analysis of Nucleotide Sugars-This study used a Waters HPLC system (Waters 2796 Bioseparations Module) composed of an autosampler and a UV detector (Waters 2996 photodiode array detector) to analyze the amount of nucleotide sugars in cells. Separation of nucleotide sugars, NAD ϩ , and NADH was performed at 34°C on a Bondapak C18 column (10-m particle size, 3.9 ϫ 300 mm) and a guard column cartridge (5-m particle size, 4.6 ϫ 20 mm, Atlantis, Waters). The method was performed as described previously (19,20). Two buffers were used for the HPLC analysis. Buffer A was 100 mM KH 2 PO 4 /K 2 HPO 4 plus 8 mM tetrabutylammonium-bisulfate hydrogen sulfate used as the pairing reagent, pH 5.3. Buffer B was 70% buffer A plus 30% methanol, pH 5.9. Concentrate KOH or phosphoric acid was used for adjusting the buffer pH. All solutions were freshly prepared, filtered through a 0.2-m filter (Millipore), and degassed before use. The sequence of chromatographic gradient was as follows: 100% buffer A for 2.5 min, 0 -40% buffer B for 14 min, 40 -100% buffer B for 1 min, 100% buffer B for 6 min, 100 -100% buffer B for 1 min, followed by an equilibration phase of 100% buffer A for 8 min. The flow rate was 1.0 ml/min.
Enzymatic Activity and Kinetic Characterization-UGDH activity was measured by monitoring the change in absorbance at 340 nm, which accompanies the reduction of NAD ϩ to NADH. The assay was performed at room temperature in 0.1 M Tris-HCl buffer, pH 7.5, containing 5 mM UDP-Glc, 1 mM dithiothreitol (DTT), 1 mM MgCl 2 , and 5 mM NAD ϩ . One unit of UGDH activity is defined as the amount of enzyme required to produce 2 mol of NADH per min at room temperature. Prior to HPLC analysis, the reaction mixture was mixed thoroughly with an equal volume of methanol, and the protein precipitate was removed by centrifugation at 20,000 ϫ g for 30 min. The K m and V max values for UDP-Glc and NAD ϩ were determined independently using standard assay conditions. Kinetic parameters for UDP-Glc as the substrate were measured by holding NAD ϩ constant (5 mM) and varying UDP-Glc from 0.02 to 5 mM. Similarly, NAD ϩ kinetic measurements were made by holding UDP-Glc constant (5 mM) and varying NAD ϩ from 0.02 to 5 mM. GraphPad Prism4 software plotted the data, and K m and V max were calculated by fitting the data to the equation where h is the Hill coefficient. Substrate specificity was measured by substituting UDP-Glc with the same concentration of ADP-Glc, TDP-Glc, GDP-Man, UDP-Gal, or UDP-GlcNAc in the standard reaction. To determine the K i value of UDP-GlcUA, the K m value of UDP-Glc at a variety of concentrations of UDP-GlcUA was first measured. The results were then fitted onto a curve where X is the concentration of UDP-GlcUA, and Y is the observed K m value. Linear regression was used to determine the x axis intercept, which equals the negative K i value. The measurement of K i for NADH was carried out similarly in the presence of variable concentrations of the compound.
Molecular Mass Determination-The molecular masses of PA2022 and PA3559 in aqueous solution were determined using a Superdex-200 10/300 GL column (GE Healthcare) on a fast protein liquid chromatograph in 100 mM Tris-HCl, pH 7.5, containing 300 mM NaCl, at a flow rate of 0.5 ml/min. Size determination was made by comparing molecular mass standards (GE Healthcare) chromatographed under the same conditions. The molecular mass standards used were as follows: ferritin, 440 kDa; aldolase, 158 kDa; albumin, 67 kDa; RNase A, 15 kDa. A calibration curve was plotted as the log (molecular mass) versus K av , where K av ϭ (V e Ϫ V 0 )/(V t Ϫ V 0 ), giving correlation values above 0.99. The term V e is the elution volume for the protein; V 0 is the column void volume, and V t is the column total bed volume. The molecular masses of PA2022 and PA3559 were subsequently determined from the calibration curve.
Construction of PA2022 and PA3559 Deletion Strains-Constructing the PA3559 gene-specific deletion mutants in P. aeruginosa involved obtaining DNA fragments of ϳ1 kb in size both upstream and downstream of the targeted sequences to be deleted by PCR amplification of genomic DNA of P. aeruginosa PAO1. Primers pmrE05 and pmrE03 were used for amplifying upstream fragments, and primers pmrE04 and pmrE06 (Table  2) were used for the downstream region. The amplified fragments were then cloned in tandem into the suicide vector pEX18Tc. For PA2022 mutant construction, a 4.1-kb DNA segment comprising the PA2022 gene was first cloned into the PstI and EcoRI sites of the suicide vector pSUP202. The next step was to insert a kanamycin resistance gene in the unique BamHI site of the PA2022 gene. In either case, the resulting gene replacement plasmids were transformed individually into E. coli SM10 and subsequently mobilized into P. aeruginosa PAO1 or its derivatives via conjugation. The transconjugants were selected on LB agar containing tetracycline (50 g/ml) and ampicillin (100 g/ml), which was used to counterselect donor E. coli. The surviving colonies were then propagated in LB broth containing 5% sucrose, which selected against cells containing the sacB gene harbored on pEX18Tc. These colonies were also tested for tetracycline sensitivity to ensure plasmid loss. The resultant mutants were confirmed by PCR amplification and Southern hybridization with a probe specific to each of the genes.
Preparation of the Cell-free Extracts for UGDH Activity Determination-The bacteria grown overnight were refreshed in 50 ml of LB broth for 2 h and collected by centrifugation. The pellet was resuspended in 30 ml of buffer containing 100 mM Tris-HCl, 20 mM NaCl, 1 mM MgCl 2 , 1 mM DTT, 1 mM phen-ylmethylsulfonyl fluoride, pH 7.5, disrupted by a brief ultrasonication, and centrifuged for 20 min at 14,000 ϫ g to remove the debris. The clarified cell lysates were dialyzed overnight against 100 mM Tris-HCl, 1 mM MgCl 2 , 1 mM DTT, pH 7.5, prior to determination of the UGDH activity. The cell lysate (ϳ15 g of protein) in this specific assay was first incubated with NAD ϩ to determine the endogenous NAD ϩ reduction rate. UDP-Glc was then added to the reaction mixture, and the difference in the NAD ϩ reduction rate was measured. Here the definition of 1 unit is the activity that produces 1 mol of NADH per min.
Extraction of Intracellular Nucleotide Sugars-Bacterial cells from the 0.5-ml overnight culture were collected by centrifugation. The pellet was briefly washed with phosphate-buffered saline, and 185 l of a cold 0.5 M perchloric acid solution was added to the pellet and mixed vigorously. After incubation on ice for 2 min, the samples were centrifuged at 10,000 ϫ g at 4°C, for 5 min. The supernatants containing the soluble molecules were neutralized with 42 l of ice-cold 2.5 M KOH in 1.5 M K 2 HPO 4 and incubated on ice for 2 min. The neutralized samples were centrifuged again and filtered with 0.2-m filters (Millipore, Billerica, MA) to remove potassium perchlorate precipitate. The samples were stored at Ϫ70°C until analysis.
Susceptibility to Antibiotics-Antibiotic-impregnated disks (BBL TM , Oxoid or Difco) were used to measure the sensitivity of the bacterial strains to antibiotics. For the disk diffusion assay, overnight aerobically grown cultures in LB were diluted to an A 600 of 0.1, and 100-l aliquots were spread uniformly on an LB plate. The disk loads were as follows: ampicillin, 10 g; polymyxin B, 300 IU; ceftazidime, 30 g; cefotaxime, 30 g; chloramphenicol, 30 g; gentamicin, 10 g; imipenem, 10 g; kanamycin, 30 g; lincomycin, 2 g; and tetracycline, 30 g. Comparing the zone of inhibition to a standard sensitivity chart (BD Biosciences) revealed the resistance to an antibiotic. The MIC of each antibiotic was determined by the microdilution method in multiple-well plates. The MIC was defined as the lowest concentration of antibiotic that prevented growth of the testing bacterial strains after 18 h of incubation at 37°C.
Bacterial Motility Assays-For swimming motility assays (21), tryptone swim plates (1% tryptone, 0.5% NaCl, 0.3% agar) were inoculated with a sterile toothpick and incubated for 12-14 h at 25°C. Motility was then assessed qualitatively by examining the circular turbid zone formed by the bacterial cells migrating away from the point of inoculation. For the twitching motility assay, cells were stab-inoculated with a toothpick through a thin (ϳ3 mm) LB agar layer (1% agar) to the bottom of the Petri dish. After incubation from 24 to 48 h at 37°C, a hazy zone of growth at the interface between the agar and the polystyrene surface was revealed by crystal violet staining (1% (w/v) solution) (22).
LPS Analysis-The hot phenol-water method (23) and the proteinase K digestion method (24) were utilized to extract LPS. The LPS preparations were separated on a 12.5% SDSpolyacrylamide gel and silver-stained according to a previously published method (25).
Reverse Transcription and Real Time-PCR-Total RNA was isolated by using the High Pure RNA isolation kit (Roche Applied Science). Two micrograms of total RNA was reverse-transcribed in a 20-l reaction mixture containing 50 units of Moloney murine leukemia virus reverse transcriptase, 5 M DTT (both from Invitrogen), 40 units of RNaseOUT recombinant ribonuclease inhibitor (Promega) 0.5 M of random hexanucleotide primers (GE Healthcare), and 500 M of dNTP mixture. The reverse transcription reaction was carried out at 50°C for 60 min. Heating the reaction mixture at 70°C for 15 min was subsequently performed to terminate the reaction, and the cDNA was stored at Ϫ20°C.
All real time PCR analyses were performed on an ABI PRIZM 7000 instrument. Each reaction included a 20-l reaction mixture containing 0.1 M of each primer, 10 l of 2ϫ SYBR Green PCR master mix (Applied Biosystem, including AmpliTaq Gold DNA polymerase with buffer, dNTPs mix, SYBR Green I dye, Rox dye, and 10 mM MgCl 2 ), and 1 l of the template cDNA. The typical amplification program included activation of the enzyme at 94°C for 10 min, followed by 40 cycles of denaturation at 94°C for 15 s, and annealing and extension at 60°C for 1 min. The C T (cycle threshold) value for each gene was determined by automated threshold analysis function of the ABI instrument and normalized to C T(G3PDH) to obtain dC T (ϭ C T(G3PDH) Ϫ C T(test) ). The difference in gene expression at different Mg 2ϩ concentrations was indicated with the ddC T values, which were calculated as dC T(low Mg 2ϩ ) Ϫ dC T(high Mg 2ϩ ) . The difference of n between two C T or dC T values indicates a 2 n -fold difference in amount of the target sequence between the two cDNA samples being compared.
Biofilm Assay-The kinetics of biofilm formation in a static system was measured as described by O'Toole and Kolter (26). Briefly, LB was inoculated with an overnight culture of each strain at a 1/100 dilution, and 100-l aliquots of the diluted bacterial suspension were added to a 96-well polystyrene microtiter dish. The microtiter dish was incubated at 37°C in a humidified chamber. The liquid culture was removed by aspiration, and the wells were rinsed thoroughly two times with phosphate-buffered saline at each time point. Samples were stained by adding 100 l of 1% crystal violet solution to each well and allowing them to sit for 15 min. After staining, the wells were washed three times with water, and any remaining crystal violet in the wells was dissolved in 100 l of 95% ethanol, and the absorbance at 595 nm was determined. Alternatively, the biofilm-forming capability of these strains was evaluated by using a polydimethylsiloxane flow cell connected to a bacterial culture system, which allowed the culture media flowing through the flow cells at uni-direction. Before perfusion, 10 ml of mid-log phase (A 600 ϭ 0.5) P. aeruginosa was injected into the microchannels by a polypropylene syringe. The flow cells were incubated for 1 h at 25°C to allow initial bacteria attachment. A peristaltic pump then passed a 1:10 dilution of LB broth through the flow cell at a flow rate of 10 ml/min. After 24 h, the biofilms formed in the flow cell were labeled with SYBR Green solution for 5 min and then examined under a Zeiss LSM 510 laser-scanning confocal microscope. Optical sections with a step size of 3 m in the z axis were collected and reconstructed into three-dimensional cross-sections.

Identification of UGDH Homologs in P. aeruginosa PAO1
Genome-Because the Streptococcus pyogenes UGDH (SpUGDH) molecular structure was solved previously (27), the protein sequence was used as a template to search for UGDH homologs in P. aeruginosa PAO1. Four genes, PA2022, PA3159, PA3540, and PA3559, were capable of encoding a protein with a sequence identity ranging from 22 to 25% compared with that of SpUGDH. Among them, PA3540 (algD) and PA3159 (wbpA) have been characterized previously as GDP-Man dehydrogenase (GMDH) (28) and UDP-GlcNAc 6-dehydrogenase encoding gene, respectively (29), but PA2022 and PA3559 have not been investigated. Alignment of the putative protein sequences of these genes indicates that, in comparison with PA3540 and PA3159, PA2022 and PA3559 resemble SpUGDH with a BLASTP e value of 2 ϫ 10 Ϫ23 and 7 ϫ 10 Ϫ22 , respectively. PA3559 and PA2022 are closely related to each other and share 52% identity and 67% similarity. Both proteins contain a NAD ϩ dinucleotide-binding domain in the N-terminal regions and a nucleotide sugar-binding domain in the C-terminal regions (Fig. 1). The signature of the NAD ϩ -binding domain, i.e. three glycines in the GXGXXG "fingerprint" of the Rossmann fold (30), was strictly conserved among these putative nucleotide sugar dehydrogenases. In addition, these two proteins also contain Cys-260 in SpUGDH and its flanking sequence GGXCXXXD, which is known to participate in catalysis. Finally, like the other UGDH, PA2022 and PA3559 both contain an Arg at a position corresponding to Arg-244 in SpUGDH. This Arg is thought to be one of the determinants of substrate specificity in nucleotide sugar dehydrogenases (27). This contrasts with GMDH and UDP-GlcNAc dehydrogenase, in which that position contained a Lys and a Phe, respectively (Fig. 1). Based on the highly conserved sequence features, this study hypothesizes that both PA2022 and PA3559 encode a nucleotide sugar dehydrogenase, and it is most likely a UGDH.
Oligomeric Structure of PA2022 and PA3559 Gene Products-The oligomeric structure of nucleotide sugar dehydrogenases varies considerably in different organisms. The P. aeruginosa UDP-GlcNAc dehydrogenase (WbpA) is a trimer (29), whereas SpUGDH is a monomer (27) in solutions. On the other hand, E. coli UGDH (32) and P. aeruginosa GMDH display a dimeric and a tetrameric form, respectively (33). Gel filtration chromatography reveals that the native conformation of PA2022 and PA3559 is ϳ104 and 117 kDa, respectively (Fig. 2). Because the predicted mass of PA2022 and PA3559 is 54.5 and 57.4 kDa, respectively, both proteins are probably present in a dimeric form in native conditions, as also found for the E. coli UGDH.
Characterization of Enzymatic Activity in PA2022 and PA3559-Determining whether PA2022 and PA3559 exhibit UGDH activity was evaluated by incubating the enzyme with UDP-Glc and analyzing the reaction product via HPLC. Commercially available nucleotide sugars served as the references. The recombinant PA2022 and PA3559 were first produced in E. coli as a His 6 -tagged protein and purified to near homogeneity as estimated by SDS-PAGE. Once purified, DTT and MgCl 2 were immediately added to the enzyme preparation, to a final concentration of 1 mM each, to maintain the enzyme activity. The conversion of UDP-Glc into UDP-GlcUA and NAD ϩ into NADH by either PA2022 or PA3559 could be detected after incubating the enzyme in standard assay conditions for 30 min (Fig. 3). Thus, the HPLC assay demonstrates that both PA2022 and PA3559 are UGDH. The specificity of these two UGDHs to several other nucleotide sugars was further investigated by monitoring the rate of NADH synthesis in the reaction at 340 nm using a spectrophotometer. Using TDP-Glc, UDP-Gal, and UDP-GlcNAc as a substrate for PA2022 yielded a relative rate of 38.51, 6.37, and 34.95%, respectively, to the UDP-Glc rate ( Table 3). PA3559 did not react with TDP-Glc and showed relatively low activity toward UDP-Gal and UDP-GlcNAc compared with PA2022. Therefore, PA3559 seems to possess higher substrate specificity than PA2022.
Determination of Kinetic Parameters of PA2022 and PA3559-The kinetic parameters of the UGDHs were determined by measuring NADH absorbance at 340 nm to detect the initial velocity of the reaction. The data were fitted to a modified  Table 4. Our result shows that PA2022 could utilize lower con-centrations of both UDP-Glc and NAD ϩ more efficiently than PA3559 as demonstrated by the smaller K m values to these substrates (Table 4). In addition, the Hill plot of the UDP-Glc rate data revealed a Hill coefficient of 1.20 and 1.67 for PA2022 and PA3559, respectively. These numbers indicate that PA3559 exhibits slight allosteric effects for UDP-Glc kinetics. In addition, the initial reaction rate of PA3559 displays a clear lag when UDP-Glc was used as the substrate. This result contrasts with the PA2022 result, which reaches its maximal rate rapidly reflecting the distinct kinetic behaviors between these two enzymes (data not shown).
Effects of Reaction Product and Divalent Ions on the Enzyme Activity-Both NADH and UDP-GlcUA were used to investigate product inhibition effects on the reaction carried out by the two enzymes. The presence of UDP-GlcUA, the immediate product of the reaction catalyzed by UGDH, caused moderate inhibitions on the activity of both enzymes. The K i values of UDP-GlcUA for PA2022 and PA3559 are 0.061 and 0.714 mM, respectively. The other reaction product, NADH, exhibits a relatively strong inhibition on both the enzyme activities. Compared with PA3559, PA2022 is more sensitive to NADH feedback inhibition and can be strongly affected by the compound even at 0.05 mM ( Table 5). The K i value of NADH, however, could not be determined accurately. Presumably, this is

TABLE 3 Substrate specificity of PA2022 and PA3559
The specific activity of PA2022 and PA3559 used in the study was 0.52 and 0.23 units/mg, respectively.

Substrate
Relative Two UGDHs of P. aeruginosa JUNE 15, 2007 • VOLUME 282 • NUMBER 24 because of its strong absorption at 340 nm, which makes the analysis less accurate. These results suggest that the synthesis of UDP-GlcUA in P. aeruginosa is under tightly controlled and depends on the cellular NADH concentrations. Several divalent ions, including Mn 2ϩ , Ca 2ϩ , SO 4 2Ϫ , and chelator EDTA, were tested, but none of them showed significant activation or inhibition effects on the activity of both enzymes ( Table 5).
Characterization of PA2022 and PA3559 Mutants of P. aeruginosa PAO1-Three gene-specific deletion strains in PAO1, RJ101 (⌬PA2022), RJ102 (⌬PA3559), and RJ103 (⌬PA2022, ⌬PA3559), were constructed using the allelic exchange strategy to determine the functional roles of the two UGDHs in P. aeruginosa physiology. The correctness of the deletions was confirmed by PCR using the genomic DNA extracted from the strains as a template (supplemental Fig. S1). Real time quantitative PCR also verified the absence of corresponding mRNA in the mutant stains. The UGDH activity in the crude cell extract of these mutants was then determined and compared with UGDH activity in the parental strain. Our results revealed that under standard growth and assay conditions (see "Experimental Procedures"), the UGDH activity in the double deletion mutant RJ103 was clearly reduced, yielding a specific activity of 6.7 Ϯ 0.4 units/mg of protein compared with 13.7 Ϯ 0.7 units/mg of protein in the wild type PAO1. The enzyme activity in RJ101 and RJ102 also fell moderately to 9.4 Ϯ 0.6 and 11.7 Ϯ 0.9 units/mg of protein, respectively. The low UGDH activity remaining in the double deletion mutant RJ103 suggests that other nucleotide sugar dehydrogenases may also function as UGDH.

Nucleotide Sugar Profiling of PAO1 and UGDH Mutant
Strains-Whether the cellular nucleotide sugar pool is affected in the UGDH mutants was unclear. Presumably, the accumulation of UDP-Glc and the absence of UDP-GlcUA would be observed in the UGDH mutants, in particular the double deletion mutant RJ103. Nevertheless, reverse-phase ion pair HPLC analysis of the cell extract indicates that UDP-Glc (retention time ϭ 8.63 min) seems to remain at a constant level in all the strains. On the other hand, UDP-GlcUA could not be detected in any of the extracts, suggesting that the compound turnover rate is rapid in the bacterial cells. The metabolite profile of RJ101 is similar to that of RJ103, showing an additional peak compared with wild type PAO1. The extra peak eluted at a time point very close to UDP-Gal, suggesting that any extra UDP-Glc in RJ101 and RJ103 was partially converted to UDP-Gal (supplemental Fig. S3). Finally, the total uronic acid content amount in the EPS in each strain was examined. Contrary to expectations that the uronic acid content would be reduced because of the UGDH deficiency, all UGDH mutant strains synthesized a higher quantity of uronic acid than the wild type PAO1 (Table 6). A possible explanation for this finding is that GlcUA accounts for only a minor fraction of the uronic acid pool in P. aeruginosa cells, and the UDP-Glc that accumulated because of UGDH deficiency is re-directed to pathways for the synthesis of other uronic acids such as ManUA or GlcNAcUA.
Expression Levels of PA2022 and PA3559 in Response to Low Concentrations of Mg 2ϩ -In Salmonella enterica serovar Typhimurium, the two-component system response regulator PmrA can activate UGDH gene expression when the cells are propagated in low Mg 2ϩ concentrations. To test whether the expression of PA2022 and PA3559 can also be activated in low Mg 2ϩ , a real time quantitative PCR experiment was performed to measure the expression level of these two genes. Our experimental data show that PA2022 was constitutively expressed regardless of the concentration of Mg 2ϩ in the medium (Fig. 4). In contrast, PA3559 expression was activated in response to low Mg 2ϩ with a 32-fold increase compared with high Mg 2ϩ conditions (2.0 mM). This result strongly supports the hypothesis that PA2022 and PA3559 are regulated differently and respond to different environmental cues.
Susceptibility of the UGDH Mutants to Antibiotics-P. aeruginosa is notorious for its natural resistance to many antibiotics partly because of the permeability barrier afforded by its cell surface components. As UGDH normally plays an important role in LPS and EPS biosynthesis, it would be interesting to know if there is any difference in antibiotic susceptibility between the UGDH-deficient mutants and the wild type PAO1. An antibiotic sensitivity assay was performed by the standard disk diffusion assay. As Table 7 demonstrates, the susceptibility

TABLE 5 Effects of reaction products and ions on PA2022 and PA3559 activity
Both UDP-Glc and NAD ϩ concentrations used in the assays, with the exceptions in UDP-GlcUA and NADH inhibition studies, were 5.0 mM. a The reaction rate was measured at 5.0 mM UDP-Glc and 1.0 mM NAD ϩ and compared with that at the same condition without NADH. b The reaction rate was determined at 0.5 mM UDP-Glc and 5.0 mM NAD ϩ and compared with that at the same condition without UDP-GlcUA.

Two UGDHs of P. aeruginosa
of PA2022 and PA3559 mutants to polymyxin B, lincomycin, ampicillin, and kanamycin showed no significant differences from the wild type PAO1 when tested on LB plates. Interestingly, the mutants RJ101 and RJ103 were more susceptible to chloramphenicol and cefotaxime but not RJ102. Testing the MIC of each of the antibiotics to these bacterial strains further verified the findings. Two different Mg 2ϩ concentrations were also tested to differentiate the possible roles of the two UGDH in antibiotic resistance. Our results indicate that RJ101 and RJ103 are indeed more susceptible than wild type PAO1 and RJ102 to ampicillin, chloramphenicol, and cefotaxime in either high (2.0 mM) or low concentrations of Mg 2ϩ (0.02 mM) medium, although high concentrations produce a more pronounced effect (Table 8). On the other hand, RJ102 and RJ103 were more sensitive to polymyxin B when grown in low Mg 2ϩ conditions, but not RJ101, implying that PA3559 is responsible for the polymyxin B resistance.
Effects of UGDH Mutations on the Bacterial Biofilm Formation-Another mechanism producing drug resistance in P. aeruginosa is that the bacterium tends to colonize on solid surfaces, forming a biofilm that renders cells impervious to therapeutic concentrations of antibiotics. The biofilm formation capability of these UGDH mutant strains was examined by using the microtiter dish assay. As illustrated in Fig. 5, the parental strains PAO1, RJ101, and RJ102 showed similar levels of biofilm formation. The double mutant, RJ103, produced substantially smaller amounts of biofilm in microtiter dish wells than the other strains. All of the strains displayed similar growth rates indicating that the deficiency in biofilm formation in RJ103 was not the result of slow growth. Biofilm formation dynamics were also measured using nonagitated 96-well polystyrene microtiter dishes and in flow cells over a period of 3 days. The RJ103 strain apparently produced as much biofilm as PAO1 after an 8-h incubation, but the quantity of biofilm produced by this double UGDH mutant dropped at a faster rate than the other tested strains thereafter (Fig. 5). Confocal microscopy revealed that the biofilm formed by RJ103 appeared to be thinner and than that for wild type PAO1 (Fig. 6).
Effects of UGDH on Bacterial Motility-Bacterial motility has been shown to closely associate with biofilm formation. For instance, flagella participate in the initial stages of P. aeruginosa biofilm development. Type IV pilus-mediated twitching motility is also involved in biofilm formation (19), particularly when microcolonies grow to mature mushroom-like structures on a glass surface (14). Determining whether the UGDH mutant strains had a defect in either swimming or twitching motility might explain the    -The presence of two UGDH genes in P. aeruginosa PAO1 is most likely advantageous to the bacterium when confronting environmental changes. This finding prompted investigation of whether other Pseudomonas species also possess multiple copies of UGDH genes. Using PA2022 and PA3559 as templates to search homologous genes in the genome of 11 Pseudomonas species revealed that all of the tested genomes, including P. aeruginosa, P. fluorescens, P. putida, and P. syringae, harbored multiple copies of UGDH homologous gene (Table 9). Phylogenetic tree analysis indicates that these UGDH-like genes can be divided into four clades as follows: one is very similar to PA2022, another group resembles PA3559, one group contains UDP-GlcNAc dehydrogenase encoding genes, and the rest are more closely related to GMDH. We would like to note that PA2022-like genes only occur in P. aeruginosa, whereas PA3559 orthologs are widely distributed in all Pseudomonas sp. (Fig. 7). In addition, PA2022 is closely linked to the galU gene that encodes UDP-Glc pyrophosphorylase, an enzyme that synthesizes UDP-Glc. In comparison, no UGDH-like gene can be found near the galU and gor, which encodes a glutathione reductase, in other Pseudomonas species. On the other hand, all PA3559-like genes in Pseudomonas species, except for P. putida, are adjacent to the 7-gene polymyxin resistance (pmr) gene cluster. These genes are designated PA3552-3558 in PAO1 and have been annotated to encode enzymes for polysaccharide synthesis (34). Phylogenetic analysis suggests that PA2022, a unique member of UGDH gene family only found in P. aeruginosa, may be evolved from PA3559 by gene duplication to adapt to environmental changes. The distinct biochemical properties and gene expression profiles of PA2022 and PA3559 support this notion.

DISCUSSION
This study presents evidence showing that both PA2022 and PA3559 encode a product with UGDH activity. Nevertheless, their biochemical properties as well as gene expression profiles are rather distinct. The fact that the PA2022 is the more efficient UGDH and is also constitutively expressed leads to the hypothesis that it is a housekeeping enzyme, responsible for converting most UDP-Glc to UDP-GlcUA in regular growth conditions. On the other hand, PA3559 is expressed primarily under low Mg 2ϩ stresses. This finding, together with its relatively large K m value, implies that the PA3559 serves as a backup UGDH that functions only when the bacterium encounters unusual circumstances. These circumstances would include low concentrations of Mg 2ϩ inside the phagosomes of a macrophage and whenever a large amount of UDP-Glc is available. In this regard, PA3559 behaves more like the Salmonella UGDH gene whose expression is also activated in low Mg 2ϩ . The UGDH is important in LPS and EPS modification and renders the bacteria resistant to antimicrobial peptides and phagocytes.
The three-dimensional molecular structure of SpUGDH was solved and reported by Campbell et al. (27). Subsequently FIGURE 5. Kinetics of biofilm formation in P. aeruginosa PAO1 and UGDH mutants. The parental and mutant strains were grown in static culture in microtiter dishes. Biofilm formation kinetics was analyzed at different time points. The data are the means from two independent experiments, with each sample assayed in triplicate. FIGURE 6. Micrographs of the biofilm formed by the wild type PAO1 and the double mutant RJ103. The tested bacteria were grown in a flow cell culture system and labeled by SYBR Green. Photographs were taken after 24 h of incubation using a laser-scanning confocal microscope at ϫ100 magnification with scans taken through the z axis to obtain 3-m optical sections. The panels above and to the right of each top-down micrograph show side view micrographs of axis reconstructions. Snook et al. (33) determined the structure of a related enzyme, GMDH of P. aeruginosa PAO1. Using these structures as the template, the PA2022 and PA3559 structure predictions were performed at the Swiss-Model, and Fig.  8 shows the results. The structure model did not include C-terminal 100 amino acids because the high diversity of this region failed to fit proper templates. Overall, the structures of the two gene products resemble P. aeruginosa GMDH more than SpUGDH. The NAD ϩ and UDP-Glc binding sites and catalytic sites of PA2022 and PA3559 were identified by comparing highly conserved SpUGDH domains and are indicated in Fig. 8, A and B. Both proteins contain a NAD ϩ dinucleotide-binding domain in the N terminus and a nucleotide sugar-binding domain in the C-terminal regions, which are interconnected with a long (33 residues) ␣-helix. Superimposing PA2022 and PA3559 reveals that both main structures are highly similar except for some flexible loops. Cys-266 in PA2022 and PA3559 is at a position equivalent to the catalytic nucleophile Cys-260 in SpUGDH. The distance between the UDP-Glc-binding site and the catalytic site is too great for the reaction occurring in a single protein molecule. Therefore, this study proposes that the enzymes become active only when they form domain-swapped dimers that place the N-terminal domain of one monomer in close association with the C-terminal domain of the adjacent monomer. However, this speculation must be tested experimentally.
An intriguing question is the fate of UDP-GlcUA in P. aeruginosa PAO1. Many cystic fibrosis P. aeruginosa strains produce exopolysaccharide alignate, which does not rely on UGDH activity. However, it is not clear whether these strains could also produce another type of EPS containing GlcUA or its derivative. Similarly, the composition of EPS in noncystic fibrosis clinical P. aeruginosa isolate is also not clear. Some early studies demonstrated that certain P. aeruginosa strains produce a polysaccharide that could be degraded by hyaluronidase (35,36). Another study found that HA accounts for as much as 7% of the biofilm slime materials in a P. aeruginosa strain (37). HA is a polysaccharide consisting of alternative GlcUA and GlcNAc residues. The UDP-GlcUA synthesized by PA2022 and PA3559 might therefore contribute to HA synthesis in P. aeruginosa PAO1, because PAO1 also has a GlcNAc synthesis pathway according to the Kyoto Encyclopedia of Genes and Genomes data base analysis. Consistent with this notion, data from the current study show that RJ103, a PA2022-PA3559 double mutant, produces thinner biofilm than the wild type PAO1. Taken together, this information strongly suggests that the functional role of UGDH in P. aeruginosa at least partly involves  HA synthesis, which also contributes to biofilm formation. In addition, HA enables bacteria to evade the immune system of a host because of its nonimmunogenic nature. In most bacteria, there can be an overlap between EPS and LPS synthesis gene functions (37). Examining the LPS composition of UGDH mutant strains and the wild type PAO1 addresses this question and shows that they are indistinguishable on a silver-stained SDS-polyacrylamide gel.
This study provides direct evidence showing that UGDH is associated with antibiotic resistance in P. aeruginosa. In addition, PA2022 and PA3559 appear to play a distinct role in the phenomenon. The strains RJ101 and RJ103, in which PA2022 is deleted, are more susceptible to chloramphenicol and cefotaxime (Table 8) in both high and low Mg 2ϩ conditions. This susceptibility reflects on the housekeeping role of PA2022. Because these antibiotics are distinct in their chemical structure and mode of action, changes in susceptibility to these antibiotics are most likely because of a nonspecific mechanism such as alterations in bacterial membrane permeability that modulate antibiotic entry into the cell. On the other hand, deleting PA3559 in P. aeruginosa caused a decrease in resistance levels to polymyxin B when propagated in low Mg 2ϩ medium. Presumably, this effect is because of the higher expression levels of PA3559 in the growth conditions. Thus, PA3559 is likely to behave like its orthologous gene in the S. enterica serovar Typhimurium. The Salmonella UGDH gene participates in the pathway for synthesizing 4-amino-4-deoxy-L-arabinose, which affects lipid-A modification and produces polymyxin resistance (38). Nevertheless, other possibilities cannot be ruled out. For instance, the enzyme may modify or metabolize antibiotics directly, although we are not aware of any related finding being reported before.
Because of its critical role in EPS synthesis, UGDH has been considered as a crucial drug target (27). The presence of multiple UGDHs in Pseudomonas species makes this idea seem less attractive, however. Nevertheless, as a deficiency in UGDH could produce higher drug susceptibility in P. aeruginosa, a UGDH inhibitor may be used in combination with a clinically available antibiotic, as exemplified in the ␤-lactam drugs/␤lactamase inhibitor. Further investigation on how UGDH affects drug susceptibility in bacteria may reveal new implications for future drug development.