The biosynthesis of the branched-chain sugar d-apiose in plants: functional cloning and characterization of a UDP-d-apiose/UDP-d-xylose synthase from Arabidopsis

Identification of a candidate gene for UDP-d-apiose/UDP-d-xylose synthase

As UDP-d-apiose/UDP-d-xylose synthase and UDP-d-glucuronate decarboxylase (synonymous with UDP-d-xylose synthase; EC 4.1.1.35) both use UDP-d-glucuronate as a substrate and catalyze identical reactions in the formation of UDP-d-xylose, we hypothesized that UDP-d-apiose/UDP-d-xylose synthases would show sequence similarities to UDP-d-glucuronate decarboxylases, for which genes have recently been cloned from the pathogenic fungus Cryptococcus neoformans (Bar-Peled et al., 2001), rat (Moriarity et al., 2002), pea (Kobayashi et al., 2002), and Arabidopsis (Harper and Bar-Peled, 2002). By searching the Arabidopsis genome sequence with these coding regions, we identified two genes named AXS1 (At2g27860) and AXS2 (At1g08200), encoding proteins with 29% amino acid sequence identity to the UDP-d-xylose synthase UXS3 from Arabidopsis (Figure 3). Both AXS1 and AXS2 have predicted sizes of 43.6 kDa, which closely match that of UDP-d-apiose/UDP-d-xylose synthase from parsley (44 kDa based on SDS–PAGE; Matern and Grisebach, 1977). For this reason, we considered AXS1 and AXS2 promising candidate genes for UDP-d-apiose/UDP-d-xylose synthases in Arabidopsis.

AXS1 and AXS2 both have eight introns and are located on chromosome 2 and chromosome 1, respectively. Both genes are predicted to encode 389 amino acid proteins belonging to the NAD⁺-dependent epimerase/dehydratase protein family. A highly conserved GxxGxxG motif (Wierenga et al., 1986) involved in binding of NAD⁺, is located at positions Gly²⁴–Gly³⁰ (Figure 3). Also, a conserved YxxxK sequence believed to function in catalysis (Thoden et al., 1997) is located at positions Tyr¹⁸⁵–Lys¹⁸⁹ in both the proteins (Figure 3). Hydropathy plots on the derived amino acid sequences (Kyte and Doolittle, 1982) did not predict any transmembrane domains or signal peptides, suggesting that AXS1 and AXS2 are soluble cytoplasmic proteins (data not shown). This is in agreement with the properties of UDP-d-apiose/UDP-d-xylose synthases from parsley and duckweed (Matern and Grisebach, 1977; Mendicino and Abou-Issa, 1974) and the soluble nature of recombinant AXS1 expressed in Escherichia coli (see below). The deduced amino acid sequence of AXS1 shares 96% sequence identity to AXS2 and 84% identity to a putative protein in rice (accession number BAB85329). tblastn searches of dbEST with the derived protein sequence of AXS1 identified homologs in Lotus japonicus, Brassica campestris, poplar, iceplant, cotton, alfalfa, potato, soybean, tomato, sorghum, maize, and wheat (data not shown). Outside the plant kingdom, AXS1 and AXS2 are most similar to the UDP-d-glucuronate dehydrogenase domain of ArnA (34% identity, see Figure 3), an enzyme involved in the biosynthesis of UDP-4-amino-4-deoxy-l-arabinose. This nucleotide sugar is a precursor in the biosynthesis of lipopolysaccharides in E. coli (Breazeale et al., 2002).

Expression of AXS1 in E. coli

To determine the function of the AXS1 gene product, a full-length cDNA was isolated and the coding region was cloned into the pET28b expression vector, utilizing a C-terminal histidine tag. Total protein from crude extracts of induced E. coli BL21 (DE3) were analyzed by SDS–PAGE, indicating the presence of a protein of approximately 46 kDa that was not observed in crude extracts of E. coli transformed with the empty vector (Figure 4a). This size is in agreement with the molecular mass predicted from the amino acid sequence of the recombinant AXS1 protein, including a histidine tag (45.8 kDa). To determine whether an antibody raised against UDP-d-apiose/UDP-d-xylose synthase from parsley (Gardiner et al., 1980) recognized the recombinant protein, crude extracts were analyzed by immunoblotting. Results from this experiment showed that the antiserum recognized the 46-kDa protein, but did not react with protein from E. coli transformed with the empty vector (Figure 4b). This supported our hypothesis that AXS1 encodes a UDP-d-apiose/UDP-d-xylose synthase.

AXS1 has UDP-d-apiose/UDP-d-xylose synthase activity

To assay AXS1 for nucleotide sugar interconversion activities, crude extracts from E. coli were incubated with UDP-d-[¹⁴C]-glucuronate, and reaction products were analyzed by thin layer chromatography (TLC) after hydrolysis to monosaccharides. Incubation with extracts containing the recombinant AXS1 protein resulted in two products, with R_f values typical of d-apiose and d-xylose, respectively, as determined by authentic standards (Figure 5a, lane 3). Extracts from E. coli transformed with the empty vector did not convert the UDP-d-[¹⁴C]-glucuronate substrate into any detectable products (Figure 5a, lane 1).

To verify the identity of the reaction products by an independent method, the assay was repeated with unlabeled UDP-d-glucuronate and affinity-purified AXS1 protein. After acid-catalyzed hydrolysis of the reaction products, monosaccharides were separated and quantitated by gas-liquid chromatography (GLC) of alditol acetates. This experiment indicated the formation of products with retention times identical to those of authentic apiitol and xylitol pentaacetate (Figure 5, panels b and c). Boiled enzyme preparations did not exhibit any detectable activity (Figure 5d). Incubation of purified AXS1 with UDP-d-[¹⁴C] galactose, UDP-d-[¹⁴C]-glucose, or UDP-d-[¹⁴C]-xylose did not yield any detectable products (data not shown).

To verify that the reaction products from the AXS1 assays with radiolabeled substrate represented nucleotide sugars rather than monosaccharides or sugar phosphates, the reaction products were purified using anion-exchange chromatography (Bonin and Reiter, 2000; Burget et al., 2003). Ionic strengths required to elute the three sugar types were determined by loading and eluting d-xylose, d-apiose, α-d-xylose-1-phosphate, and a mixture of unlabeled and radiolabeled UDP-d-Xylose. d-xylose and d-apiose eluted at 1 mm potassium phosphate (pH 6.0), whereas α-d-xylose-1-phosphate eluted at 50 mm and UDP-d-xylose eluted at 100 mm of this buffer. Once these conditions were established, the reaction products from the assays with AXS1 were loaded onto an anion exchange column and eluted with two column volumes each of 10, 60, 120, and 200 mm potassium phosphate (pH 6.0). More than 85% of the radioactivity eluted with 120 mm of this buffer, indicating that the reaction products represented nucleotide sugars.

pH optimum, enzyme stability, and effect of co-factors

To examine the influence of pH on UDP-d-apiose/UDP-d-xylose synthase activity, enzyme preparations were assayed in 100 mm potassium phosphate buffer (pH 6.5–7.5) and 100 mm Tris–HCl (pH 7.5–9.0) using UDP-d-[¹⁴C] glucuronate as a substrate. The pH optimum of AXS1 was determined to be 8.0 (Figure 6a).

Storage of purified AXS1 at 0 or 25°C overnight resulted in 40 and 50% loss of activity, respectively, whereas storage at −20°C in the presence of 50% glycerol caused a twofold decrease in the activity over 1 week. The enzyme was also sensitive to freezing in the presence of 50% glycerol at −80°C (74% loss of activity), and was completely inactivated by freezing in the absence of cryoprotectant. For these reasons, only freshly purified AXS1 was used to characterize the enzyme further.

The effect of oxidized and reduced forms of pyridine nucleotide co-factors (NAD⁺, NADH, NADP⁺, and NADPH) on the ability of AXS1 to convert UDP-d-[¹⁴C]-glucuronate to UDP-d-[¹⁴C]-apiose and UDP-d-[¹⁴C]-xylose was also determined. Results of this analysis showed that purified recombinant AXS1 was active in the absence of added co-factors, although addition of 2 mm NAD⁺ stimulated the enzyme activity by 8.5 folds. This suggests that some of the affinity-purified enzyme contained tightly-bound NAD⁺, leading to a low but detectable enzymatic activity. Addition of 2 mm NADH, NADP⁺, or NADPH to the assay had no effect. However, pre-incubating purified AXS1 with 2 mm NADH at 25°C overnight resulted in total loss of enzyme activity, which could not be rescued by addition of exogenous NAD⁺ to the assay. This suggested that enzyme-associated NAD⁺ could be slowly replaced by exogenous NADH, which would prevent the enzyme from catalyzing the first step in the reaction sequence, i.e. the oxidation of C-4 of the glucuronate moiety.

Addition of up to 1 mm DTT has been reported to have a stabilizing effect on UDP-d-apiose/UDP-d-xylose synthase from parsley and duckweed (Baron et al., 1972; Mendicino and Abou-Issa, 1974; Wellmann and Grisebach, 1971). However, up to 2 mm DTT did not have any effect on the recombinant Arabidopsis enzyme. Addition of EDTA to 5 mm final concentration had no effect, indicating that divalent cations are not required for AXS1 activity.

Effect of nucleotides and nucleotide sugars on AXS1 activity

UDP-d-glucose, UDP-d-galactose, and UDP-d-xylose did not serve as substrates for AXS1, and addition of up to 2 mm UDP-d-glucose or UDP-d-galactose had no inhibitory effect on AXS1 activity. UDP and UDP-d-xylose inhibited AXS1 at concentrations higher than 370 μm (Figure 6b), which is equal to a 109-fold excess over the substrate UDP-d-glucuronate. However, addition of 3.7 and 7.4 μm UDP-d-galacturonate (equal to 1.1- and 2.2-fold substrate concentration) reduced AXS1 activity to 82 and 31%, respectively (Figure 6b), suggesting that the concentration of UDP-d-galacturonate might regulate the activity of AXS1 in vivo.

AXS1 is a dimer

To determine whether AXS1 is a monomeric or multimeric enzyme, the molecular weight of the recombinant AXS1 was analyzed by size-exclusion chromatography. A calibration curve with protein standards was used to determine the molecular weight of the recombinant AXS1. The elution volume for purified AXS1 and AXS1 in crude extracts corresponded to a molecular weight of 84 kDa, suggesting that the recombinant AXS1 exists as a dimer (predicted molecular weight of 91.6 kDa).

Kinetic parameters of recombinant AXS1

The kinetics of AXS1 were analyzed using 3 μg purified AXS1 in the presence of 2 mm NAD⁺ and various concentrations of UDP-d-[¹⁴C]-glucuronate. The data were analyzed both by a computerized non-linear regression fit and a Lineweaver-Burk plot (Figure 6c). The K_m value was determined to be 7 μm for both plots (R² = 0.96), indicating a high affinity for the substrate. This value is comparable to those of the enzymes from parsley and duckweed (2 and 5 μm, respectively; see Grisebach, 1980). The V_max value for the conversion of UDP-d-glucuronate to UDP-d-apiose plus UDP-d-xylose was 3.1 nmol min⁻¹ mg⁻¹ protein, giving a k_cat value of 0.3 min⁻¹, calculated based on the molecular weight of the dimeric protein. The apiose/xylose ratio was determined to be 0.6, based on data obtained by both TLC and GLC. As this ratio has been reported to increase in a potassium phosphate buffer system (Grisebach, 1980; Matern and Grisebach, 1977), we performed the enzyme assay under these conditions, but did not observe any change in the relative amounts of apiose and xylose.

NH₄⁺ has been reported to influence the activities of the UDP-d-apiose/UDP-d-xylose synthases from parsley and duckweed by inhibiting the formation of UDP-d-apiose and stimulating the biosynthesis of UDP-d-xylose (Baron et al., 1972; Wellmann and Grisebach, 1971). To test if NH₄⁺ has the same effect on the recombinant Arabidopsis enzyme, AXS1 was assayed in the presence of 2 mm up to 200 mm (NH₄)₂SO₄. We found that increasing concentrations of (NH₄)₂SO₄ inhibited the activity of AXS1 (Figure 6d) without changing the apiose/xylose ratio. Similar effects were obtained by adding NH₄Cl, KCl, and MgCl₂ (data not shown).

Expression pattern of the AXS1 gene in A. thaliana

The expression pattern of AXS1 was analyzed by an AXS1::GUS construct transformed into Arabidopsis. All plant organs were analyzed for GUS staining. Results from this experiment showed that AXS1::GUS is expressed throughout the Arabidopsis plant (Figure 7a–d). The GUS staining data were in agreement with RT-PCR results using AXS1-specific oligonucleotides, which indicated that the AXS1 transcript could be amplified from total RNA extracted from all tissues analyzed (roots, rosette leaves, stems, flowers, and siliques; Figure 7e).

Plant material and growth conditions

Plants were grown in an environmental chamber at 23°C and 60–70% humidity under continuous fluorescent light (60–70 μmol m⁻² sec⁻¹). Arabidopsis plants of the ecotype Columbia were used for transformation and the isolation of DNA and RNA.

Plasmid constructs and cloning

A cDNA encoding the AXS1 protein was amplified by RT-PCR from total leaf RNA using the oligonucleotides 5′-ATTTGCCCATGGCGAATGGAGCTAATAGAGTGGATCTCGACG-3′ and 5′-CGTTAAGAATTCGCGGAAGCCACTGGTTTGGATGTTGCCTTCTTCAC-3′ (NcoI and EcoRI sites engineered into the primers are underlined). After cleavage with NcoI and EcoRI, the cDNA was cloned in frame with the polyhistidine tag into the E. coli expression vector pET28b (Novagen, Madison, WI, USA), giving the construct pET28b-AXS1 in E. coli host strain BL21 (DE3).

For gene-expression studies, 2 kb of DNA directly upstream of the AXS1 coding region were PCR-amplified from genomic DNA using the oligonucleotides 5′-ATGCTGAATTCAAAGCAACGAGAAATTAGCATCTTTTTTTCATC-3′ and 5′-GCCTAAAGATCTTCCATTTTTTTTCTCAGATCAGAAAAAATAATC-3′ (EcoRI and BglII sites are underlined) and cloned in frame with the gusA reporter gene in pCAMBIA1301 (CAMBIA, Canberra, Australia), giving pCAMBIA1301-AXS1. All PCR reactions were performed with Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA, USA), and the sequences of all constructs were verified using the ABI Prism Big Dye Terminator Cycle Sequence Reaction Kit (Perkin-Elmer Applied Biosystems, Foster City, CA, USA) and an ABI Prism 377 DNA sequencer. A. thaliana was transformed using Agrobacterium tumefaciens strain GV3101 (pMP90) essentially as described by Bechtold et al. (1993).

Expression of AXS1 in E. coli and purification of the AXS1 protein

Luria-Bertani medium containing 50 μg ml⁻¹ kanamycin was inoculated with E. coli BL21 (DE3) transformed with pET28b-AXS1 and shaken at 260 r.p.m. and 37°C until an OD₆₀₀ of 0.6–1.0 was attained. Gene expression was induced by addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 1 mm and further incubation for 3 h. Cultures were cooled on ice for 5 min, and cells were harvested by centrifugation at 5000 g for 5 min at 4°C. Cell pellets were stored at −80°C. To prepare a crude protein extract from recombinant E. coli, frozen cells obtained from a 50-ml culture were re-suspended in 1 ml of 100 mm Tris–HCl (pH 8.0), followed by the addition of 1× Complete protease inhibitor (Boehringer Mannheim, Indianapolis, IN, USA) and 1 mg ml⁻¹ lysozyme (Sigma, St Louis, MO, USA). After incubation at 4°C for 30 min, cells were lyzed by sonication. Measurements of protein concentration were performed with a bicinchoninic acid protein quantitation kit (Sigma). Purification of the native recombinant AXS1 protein from crude extracts was carried out with Ni-NTA agarose columns according to the manufacturer's protocol (Qiagen, Valencia, CA, USA).

SDS–PAGE and Western analysis

SDS–PAGE was performed in 10% polyacrylamide gels, and proteins were stained with 0.1% (w/v) Coomassie Brilliant Blue R-250 in methanol/acetic acid/water (9:2:9, v/v/v), and de-stained in the same solution without the dye. Western blotting was performed using a semi-dry electroblotter, nitrocellulose membranes from MSI Micron Separations (Westborough, MA, USA), and transfer buffer composed of 25 mm Tris–HCl (pH 8.3), 192 mm glycine, and 20% (v/v) methanol. For immunologic detection, membranes were incubated overnight at 4°C with a primary antibody against UDP-d-apiose/UDP-d-xylose synthase from parsley (Gardiner et al., 1980). This was followed by a 3-h incubation at room temperature with alkaline phosphatase-conjugated goat antirabbit-IgG (Sigma). Both incubations were carried out with 1:2000 dilutions of the respective antibodies in 50 mm Tris–HCl (pH 7.4), 500 mm NaCl, 0.5% (v/v) Tween-20, and 5% (v/v) horse serum. Alkaline phosphatase was detected colorimetrically using 0.03% (w/v) 4-nitro blue tetrazolium chloride and 0.02% (w/v) 5-bromo-4-chloro-3-indolyl-phosphate (p-toluidine salt) in 100 mm Tris (pH 9.5), 0.1 m NaCl, and 5 mm MgCl₂.

Assay of AXS1 enzyme activity

Unless stated otherwise, purified recombinant AXS1 (3 μg protein) was incubated at 25°C overnight in 50 μl of 100 mm Tris–HCl (pH 8.0), 2 mm NAD⁺ (Sigma) in the presence of 25 nCi of UDP-d-[¹⁴C]-glucuronate (300 mCi mmol⁻¹), UDP-d-[¹⁴C]-galactose (300 mCi mmol⁻¹), UDP-d-[¹⁴C]-glucose (300 mCi mmol⁻¹), or UDP-d-[¹⁴C]-xylose (238 mCi mmol⁻¹; American Radiolabeled Chemicals, St Louis, MO, USA). To assay the effect of various nicotinamide dinucleotide co-factors, NAD⁺ was replaced by NADH, NADP⁺, or NADPH. The effects of all other additives were assayed in the presence of 2 mm NAD⁺. Substitution of enzyme-bound co-factor was tried by incubation with 2 mm NADH at 25°C for 10 h. Nucleotide sugars were hydrolyzed by the addition of trifluoroacetic acid to 1.6 m final concentration, and incubation at 95°C for 30 min. Samples were dried under vacuum and re-suspended in 15 μl of 80% (v/v) ethanol. Resultant monosaccharides were separated by TLC on silica TLC plates (Aldrich, Milwaukee, WI, USA) in a 6:2:1 (v/v/v) mixture of 1-propanol, saturated ammonia solution, and water. Radioactivity was visualized by phosphorimaging (Bio-Rad Laboratories, Hercules, CA, USA) and quantified using Molecular Analyst Software (Bio-Rad Laboratories, Hercules, CA, USA). Authentic sugar standards run in parallel were stained with aniline-hydrogen phthalate (Fry, 1988). In assays with unlabeled substrate, 20 μg AXS1 was incubated overnight at 25°C with 50 mm UDP-d-glucuronate and 3 mm NAD⁺ in 250 μl 100 mm Tris–HCl (pH 8.0). All assays were repeated at least once. Free apiose was obtained by hydrolyzing 1,2:3,5-di-O-isopropylidene-α-d-apiose (Sigma) at 80°C for 30 min in 2 m trifluoroacetic acid followed by drying under vacuum and dissolving the residue in water.

For kinetic evaluation, linearity with respect to protein concentration (2–5 μg) and time (2–45 min) was first established. The concentration of UDP-d-[¹⁴C]-glucuronate was then independently varied while keeping the concentrations of all other reactants constant (3 μg purified AXS1, 2 mm NAD⁺ in 50 μl of 100 mm Tris–HCl (pH 8.0)). The incubation time was 30 min. Computerized non-linear regression fits and double reciprocal plots were constructed using GraphPad Prism Version 3.03, and the equation of the best-fit was determined. To examine the influence of pH on AXS1 activity, purified enzyme preparations were assayed using 25 nCi UDP-d-[¹⁴C]-glucuronate and 2 mm NAD⁺ in 100 mm Tris–HCl or 100 mm potassium phosphate at 25°C for 30 min.

Gas-liquid chromatography

Nucleotide sugars were hydrolyzed with trifluoroacetic acid, and monosaccharides were converted to alditol acetates as described by Kindel and Cheng (1990). Separation and quantitation of alditol acetates was carried out by GLC on a 30-m SP2330 column (Supelco, Bellefonte, PA, USA) with an inner diameter of 0.25 mm. One microliter of sample was injected under splitless conditions at an oven temperature of 140°C. After 1 min at 140°C, the oven temperature was raised to 170°C at 25°C min⁻¹, held at 170°C for 40 min, then raised to 240°C at 2°C min⁻¹, and held at 240°C for 15 min. Helium was used as a carrier gas at a pressure of 125 kPa. The injector and detector temperatures were 240 and 250°C, respectively.

Anion-exchange chromatography

Four separate Superclean LC-SAX solid-phase columns (Supelco, Bellefonte, PA, USA) were used to determine the ionic strengths required to elute 200 μg of d-xylose, 200 μg of d-apiose, 200 μg of α-d-xylose-1-phosphate, and a mixture of 200 μg of unlabeled UDP-d-xylose and 100 nCi of UDP-d-[¹⁴C]-xylose, respectively. Nucleotide sugars were identified by liquid scintillation counting and the monosaccharides and sugar phosphates were identified by total carbohydrate assays of the eluted fractions (Fry, 1988).

Size-exclusion chromatography

Purified AXS1 (70 μg) and AXS1 in crude extracts (400 μg total protein) were fractionated at 4°C on a 40 cm × 1 cm Sepharose CL-6B (Sigma) column equilibrated with 50 mm Tris–HCl (pH 7.5), 100 mm KCl. Four hundred microliter fractions were collected. Protein concentrations in each fraction were measured by determining the absorbance at 280 nm. The molecular-weight determination of AXS1 was made by comparing the ratio V_e/V₀ (V_e is the elution volume and V₀ is the void volume) to the V_e/V₀ of 160–800 μg of protein standards (β-amylase from sweet potato, 200 kDa; yeast alcohol dehydrogenase, 150 kDa; bovine serum albumin, 66 kDa; bovine carbonic anhydrase, 29 kDa; horse cytochrome c, 12.4 kDa; Sigma) dissolved in 80 μl equilibration buffer containing 5% (v/v) glycerol. Blue dextran (2000 kDa, 160 μg in 80 μl) was used to measure V₀. SDS–PAGE was used to verify fractions containing AXS1.

Histochemical β-glucuronidase assays

Arabidopsis plants transformed with pCAMBIA1301-AXS1 were selected on 0.8% (w/v) agar plates containing half-strength MS medium (Sigma), 3% (w/v) sucrose, 50 μg ml⁻¹ hygromycin B (Calbiochem, La Jolla, CA, USA), and 500 μg ml⁻¹ vancomycin (Sigma). The T₂ generation was analyzed for GUS staining using a protocol adapted from Jefferson et al. (1987). To determine the expression pattern of AXS1::GUS, more than 15 individual transformants were analyzed.

RT-PCR

To determine the presence of AXS1 mRNA in various organs, the AXS1 transcript was amplified by RT-PCR using total RNA from various plant organs. RNA was purified with the RNeasy plant mini kit from Qiagen. The reverse transcription reactions were performed in a volume of 70 μl with 2 μg of RNA, 150 U M-MLV reverse transcriptase, 0.1 mm oligo (dT)_12–18, and reverse transcriptase core reagents (Gibco-BRL, Rockville, MD, USA), according to the manufacturer's protocol. As a control, reverse transcriptase was omitted from the reaction mixture. PCR on reverse transcription reactions was performed in a volume of 50 μl using Pfu Turbo DNA polymerase, PCR core reagents (Stratagene, La Jolla, CA, USA), and the oligonucleotides 5′-ATCTCGACGGGAAACCGATACAA-3′ and 5′-GTTTGGATGTTGCCTTCTTCACA-3′. Restriction enzyme digests with PvuII were used to distinguish the AXS1-specific PCR products (which are cleaved into two fragments) from potential AXS2-specific products, which do not contain a PvuII cleavage site. In all cases, complete cleavage of the PCR product was observed indicating the absence of AXS2-derived amplification products.