Functional Characterization of UDP-apiose Synthases from Bryophytes and Green Algae Provides Insight into the Appearance of Apiose-containing Glycans during Plant Evolution*

Apiose is a branched monosaccharide that is present in the cell wall pectic polysaccharides rhamnogalacturonan II and apiogalacturonan and in numerous plant secondary metabolites. These apiose-containing glycans are synthesized using UDP-apiose as the donor. UDP-apiose (UDP-Api) together with UDP-xylose is formed from UDP-glucuronic acid (UDP-GlcA) by UDP-Api synthase (UAS). It was hypothesized that the ability to form Api distinguishes vascular plants from the avascular plants and green algae. UAS from several dicotyledonous plants has been characterized; however, it is not known if avascular plants or green algae produce this enzyme. Here we report the identification and functional characterization of UAS homologs from avascular plants (mosses, liverwort, and hornwort), from streptophyte green algae, and from a monocot (duckweed). The recombinant UAS homologs all form UDP-Api from UDP-glucuronic acid albeit in different amounts. Apiose was detected in aqueous methanolic extracts of these plants. Apiose was detected in duckweed cell walls but not in the walls of the avascular plants and algae. Overexpressing duckweed UAS in the moss Physcomitrella patens led to an increase in the amounts of aqueous methanol-acetonitrile-soluble apiose but did not result in discernible amounts of cell wall-associated apiose. Thus, bryophytes and algae likely lack the glycosyltransferase machinery required to synthesize apiose-containing cell wall glycans. Nevertheless, these plants may have the ability to form apiosylated secondary metabolites. Our data are the first to provide evidence that the ability to form apiose existed prior to the appearance of rhamnogalacturonan II and apiogalacturonan and provide new insights into the evolution of apiose-containing glycans.


Apiose (3-C-[hydroxymethyl]-D-erythrofuranose; Api)
is a branched-chain monosaccharide that is present in many plant secondary metabolites and in the primary cell walls of vascular plants (1). To date, only two cell wall polysaccharides, namely rhamnogalacturonan II (RG-II) and apiogalacturonan (Api-GalA), have been shown to contain Api (2). ApiGalAs may have a limited taxonomic distribution as they have only been detected in the cell walls of seagrasses and duckweeds (3)(4)(5). By contrast, RG-II is present in the primary walls of all vascular plants examined to date (2,6). Apiose links two side chains (A and B) to the galacturonan backbone of RG-II. The apiosyl residues of side chain A in two RG-II molecules are cross-linked by a borate diester to form the RG-II dimer (7)(8)(9)(10). At least 90% of the RG-II in primary walls exists as a dimer (11), and a reduction in the extent of RG-II cross-linking typically results in the formation of abnormal cell walls (12). Plants carrying mutations that affect Api metabolism as well as RG-II structure and cross-linking are dwarfed or fail to develop normally (13)(14)(15)(16).
Early studies of the biosynthesis of the plant flavonoid apiin (apigenin-7-(2-O-apiosylglucoside)) in parsley led Grisebach and Döbereiner (17) to propose that UDP-apiose (UDP-Api) and UDP-xylose (UDP-Xyl) are formed from UDP-glucuronic acid (UDP-GlcA) by UDP-apiose synthase (UAS). Subsequent studies identified UDP-Api in parsley and in Lemna (18). It was then proposed that UDP-Api is the activated nucleotide sugar used by apiosyltransferases that catalyze the incorporation of apiose into ApiGalA and into apiin (3,19). No apiosyltransferase has been purified to homogeneity nor have the genes encoding this glycosyltransferase been identified.
We previously hypothesized that the enzymes responsible for synthesis of UDP-Api are present only in vascular plants (20). The recent availability of the sequenced genome of the moss Physcomitrella patens (21) and publically available transcriptomic data for other avascular land plants and for green algae produced by the 1,000 Plants (1KP) Project (22) allowed us to re-examine this hypothesis.
Here we report the identification and functional characterization of UAS homologs from the monocot Spirodela polyrhiza (a duckweed), four mosses (P. patens, Dicranum scoparium, Hedwigia ciliata, and Sphagnum lescurii), a liverwort (Marchantia paleacea), a hornwort (Megaceros vincentianus), and two green algae (Mougeotia spp. and Netrium digitus). Our results provide evidence that UDP-Api appeared prior to the appearance of wall-associated apiose and that bryophytes and green algae likely synthesize apiose-containing secondary metabolites but lack the biosynthetic machinery required for the synthesis of apiose-containing wall polysaccharides.

Identification of UDP-apiose Synthase Homologs in Avascular Plants and Green Algae-UAS-like homologs with Ͼ70%
amino acid sequence identity to Arabidopsis AXS1/UAS1 were identified in monocots, mosses, liverworts, hornworts, and streptophyte green algae (Table 1) using publically available data from The 1,000 Plants (1KP) Project (22) and the Phytozome genomics portal. No UAS-like homologs were detected in the available transcriptomes of chlorophyte green algae.
An unrooted phylogenetic tree ( Fig. 1) was generated using the amino acid sequences of the UAS-like proteins (Fig. S1) together with other decarboxylases, including Arabidopsis UDP-xylose synthase (UXS), and two bacterial enzymes, a bifunctional UDP-4-keto-pentose/UDP-xylose synthase (RsU4kpxs) from the plant pathogen Ralstonia solanacearum and the C-terminal portion of ArnA that has a UDP-glucuronic acid 4-oxidase-6-decarboxylase activity (23,24). The UASs cluster into a group (III; see Fig. 1) comprising several clades, which are distinct from the clades for UXS (I) and ArnA (II). These data suggest that UAS-like proteins first appeared in the streptophyte lineage of green plants. UAS, UXS, RsU4kpxs, and ArnA are all decarboxylases that contain a conserved N-terminal Gly-X-X-Gly-X-X-Gly motif (supplemental Fig. S1) that is proposed to be involved in NAD ϩ binding as well as the conserved Tyr-X-X-X-Lys motif with an upstream Ser that forms the catalytic site of the short-chain dehydrogenase/ reductase family (25)(26)(27).
In Microbe Formation of UDP-apiose-The coding sequences of the selected UAS homologs were cloned into a modified pET28b E. coli expression vector (28). The UAS-containing plasmids or empty plasmid (negative control) was then individually transformed into E. coli together with a pCDFDuet plasmid containing the UDP-Glc dehydrogenase coding sequence (29) from Bacillus thuringiensis (BtbDH) to ensure the production of UDP-GlcA. Nucleotide sugar-containing extracts from the isopropyl ␤-D-thiogalactoside (IPTG)-induced E. coli cells were shown by hydrophilic interaction liquid chromatography with electrospray mass spectrometry (HILIC-ESI-MS/MS) to contain two product peaks eluting at 11.3 and 12.2 min (Fig. 2). These peaks were not detected in the comparable extract of E. coli cells harboring the empty plasmid (Fig.  2). The ESI mass spectra of both components contained an ion at m/z 535.00 (Fig. 2), which corresponds to [M Ϫ H] Ϫ for a UDP-pentose. MS/MS analysis (Fig. 3) of each product peak gave a fragment ion at m/z 323.00 that is consistent with [UMP Ϫ H] Ϫ . The peak eluting at 12.2 min has the same elution time and MS fragmentation pattern as authentic UDP-Xyl. Proton NMR ( 1 H NMR) analyses confirmed that the UDP-pentose eluting at 11.3 min was UDP-Api. These data suggest that the UAS-like enzymes do synthesize UDP-Api.
Purified Recombinant UAS from S. polyrhiza, Mosses, a Liverwort, a Hornwort, and Green Algae Convert UDP-GlcA to UDP-Api and UDP-Xyl-To obtain additional evidence that the monocot, avascular plant, and green algal UASs form UDP-Api, the recombinant His 6 -tagged proteins were solubilized from E. coli cells and purified using nickel affinity columns. Each recombinant UAS gave one major band on SDS-PAGE with a predicted mass of between 45 and 48 kDa (Fig. 4). Each purified UAS was shown by HILIC-ESI-MS/MS to convert UDP-GlcA to two UDP-pentose products. MS/MS analysis  (Fig. 5) of these product peaks (11.3 and 12.2 min) also gave a fragment ion at m/z 323.00 that is consistent with [UMP Ϫ H] Ϫ . Signals consistent with the presence of UDP-Api and UDP-Xyl were detected in all the 1 H NMR spectra when the recombinant enzyme assays were performed in deuterated buffer (Fig. 6).
S. polyrhiza UAS was the most highly expressed protein and was thus selected for further characterization. Real time 1 H NMR spectroscopic analysis of the products formed when SpUAS reacts with UDP-GlcA ( Fig. 7 and supplemental Table S1) confirmed that UDP-Api is the first product formed. SpUAS produces UDP-Api and UDP-Xyl in a ratio of ϳ1.7: 1.0, which is similar to potato UAS (30). Our studies with SpUAS also confirm that some of the UDP-Api is converted to the apiofuranosyl-1,2-cyclic phosphate during the in vitro reaction (Fig. 7). No degradation of UDP-Xyl is discernible over the course of the reaction.
Real time NMR-based assays provide the opportunity to detect transient intermediates (30). Our real time 1 H NMR data using recombinant SpUAS confirm that UDP-4-keto-xylose is an intermediate formed during the conversion of UDP-GlcA to UDP-Api (Fig. 7). Grisebach and Döbereiner (17) and subsequently Choi et al. (31) proposed that during UAS enzymatic catalysis a ring contraction step occurs through a retroaldol mechanism. We detected no signals indicative of the formation of the proposed enediol intermediate. Nonetheless, if this intermediate is formed it may exist for such a short time or remain secured in the catalytic site of the enzyme and thus be "invisible" to 1 H NMR spectroscopy.
The recombinant enzyme SpUAS is most active in 50 mM Tris-HCl, pH 7.5-8.1, at temperatures between 37 and 42°C (Fig. 8, A and B) and exists in solution as a dimer with a predicted size of 93 kDa (Fig. 8C). SpUAS has a K m of 237 M and K cat /K m of 1.159 M s Ϫ1 , whereas recombinant Arabidopsis AXS1/UAS1 has a reported K m of 7 M and K cat /K m of 0.043 M s Ϫ1 (32). Previous studies have shown that UAS is inhibited by nucleotides and nucleotide sugars, including UDP-GalA (32). Under our assay conditions, UDP-Xyl and NAD(P)H inhibited UAS activity by 20 and 36%, respectively, whereas UDP-GalA reduced activity by 82%. UDP-GalA is not a substrate for the recombinant UAS. Nonetheless, we cannot exclude the possibility that UDP-GalA competes with UDP-GlcA for substrate binding and thereby regulates SpUAS activity in vivo.
Apiose Is Discernible in Aqueous Methanolic Extracts but Not in the Cell Walls of Mosses, Liverworts, and Green Algae-To explore the nature of apiose-containing molecules in the avascular plants and algae, living cultures of the mosses P. patens, Dicranum, Polytrichum, and Sphagnum; the liverworts Conocephalum and Marchantia; and the algae Mougeotia and Netrium were obtained (only the genus was provided by the supplier for those plants without species designation). Fresh tissue was extracted with a series of aqueous and non-aqueous solvents (see "Experimental Procedures"). The residue remaining after these extractions is defined here as the cell wall. Apiose was detected after acid hydrolysis and GC-MS analyses of alditol-acetate derivatives in the aqueous methanol-acetonitrile (ACN)-soluble (methanolic) fractions of all four mosses, both  (Table 2). No additional apiose was detected in the subsequent solvent extracts (Fractions II-IX; see "Experimental Procedures") or in the cell walls generated from these plants ( Table 3). No Api was detected in any fractions from Netrium. By contrast, Api was abundant in the methanolic fraction and cell walls of S. polyrhiza. Such results are not unexpected as the walls of this duckweed are known to contain large amounts ApiGalA (3,33).

liverworts, and Mougeotia
Overexpression of SpUAS in P. patens and Detection of UDP-apiose-We detected small amounts of Api in the methanolic extract of P. patens, but no discernible amounts of Api were present in its cell wall. Moreover, no UDP-Api was detected in the nucleotide sugar-containing extracts of P. patens wild-type gametophyte tissue (Fig. 9B) or any of the other avascular plants and green algae. Thus, we wondered whether UDP-Api is being degraded or rapidly metabolized. We hypothesized that increasing the level of UDP-Api in P. patens gametophytes would generate a pool of activated Api that was sufficient to allow this to be incorporated into glycans of the cell wall. To this end, SpUAS was overexpressed in P. patens. The SpUAS transcript was detected in five independent transgenic lines (Fig. 9A). None of the transformed lines had a visibly altered growth phenotype. The overexpressing lines readily formed detectable amounts of UDP-Api (Fig. 9B). Increased amounts of Api were also present in the methanolic fractions from the overexpressing lines (Fig. 9C). None of the lines contained discernible amounts of Api in their cell walls. Thus, we conclude that UDP-Api is not appreciably degraded and is likely not a limiting factor for the incorporation of Api residues into walls of P. patens gametophyte.

Discussion
Our study is the first to identify functional genes encoding UDP-apiose synthase in avascular plants and green algae. The apiose in these plants was detected in aqueous methanolic extracts. However, we found no discernible amounts of apiose- containing glycans in the walls of any of the avascular plants or the algae. Thus, the Api detected in these plants is likely to be associated with a secondary metabolite. This contrasts with vascular plants where apiose is present in the cell wall polysaccharides RG-II and ApiGalA and in secondary metabolites (1, 34 -36).
Plants, many animals, fungi, Bacteria, and Archaea produce enzymes (UXS) that convert UDP-GlcA to UDP-Xyl (27,(37)(38)(39)(40), whereas UAS forms both UDP-Api and UDP-Xyl in a ratio of ϳ2:1. It is not known whether UXS or UAS is the ancestral gene. To date, no UAS genes have been identified in prokaryotes, whereas UXS is present in many Bacteria and Archaea. Thus, it is likely that UAS first appeared in the plant kingdom, possibly from UXS (40).
The mechanism whereby UAS converts UDP-GlcA into two different UDP-sugars is not known. Plant UXS and UAS and the bacterial ArnA are all enzymes that decarboxylate UDP-GlcA via 4,6-dehydration and a UDP-4-keto-pentose intermediate. X-ray analyses of UXS and ArnA identified domains (supplemental Fig. S1) involved in catalysis and cofactor binding (41,42). Protein sequence alignment (supplemental Fig. S1) shows that residues implicated in nucleotide sugar binding and catalysis are conserved among UXSs from diverse organisms but are different in the UASs. Additionally, regions of unique amino acid insertions are present in UAS but not in UXS (supplemental Fig. S1,  or ArnA (Regions 2-4). These distinct regions may facilitate or necessitate the subsequent ring cleavage and rearrangement. Site-directed mutagenesis of these regions in combination with crystallographic and enzymatic activity studies are required to elucidate the mechanism of UAS activity and determine whether it occurs via an enediol intermediate.
Plants synthesize numerous secondary metabolites. In vascular plants, at least 1,200 of these metabolites have been reported to contain an apiose (1). These apiose-containing compounds include flavonoids, terpenoids, and cyanogenic glucosides (1,35,36,43). Apiosylated secondary metabolites may protect a plant from pathogens and herbivores. Several secondary metabolites, including flavonoids, are known to provide such defense as well as protection against UV radiation and oxidative stress (44 -46). These metabolites are normally water-insoluble. The addition of a glycose such as apiose to these compounds would enhance their water solubility and perhaps facilitate their transport within the plant. Interestingly, P. patens has been reported to lack functional borate exporters (47); thus, the possibility cannot be excluded that apiosides sequester borate, which may become toxic if accumulated to high amounts (48). The ability to form apiose has broadened the library of secondary metabolites available to plants, including avascular bryophytes and algae. Mechanisms to incorporate apiose into the cell wall polysaccharides RG-II and ApiGalA must also have devel-  oped during the evolution of vascular plants from their avascular ancestors. Determining whether the P. patens apiosyltransferases that utilize UDP-Api as a donor have common motifs with the apiosyltransferases involved in secondary metabolite and pectin biosynthesis in vascular plants will provide insight into the evolutionary origins of Api-containing glycans.
The cell walls of several green algae and bryophytes are known to contain homogalacturonan (49 -51). In the cell walls of vascular plants, apiose is attached to the homogalacturonan backbones of RG-II and ApiGalA. Because we could not detect wall-bound apiose even in P. patens lines overexpressing SpUAS, it is probable that the apiosyltransferases required for the synthesis of RG-II and ApiGalA do not exist in bryophytes or green algae. It is notable that phylogenetic analyses of pectin-related gene families in P. patens suggest that this moss lacks homologs of vascular plant xylogalacturonan xylosyltransferases and rhamnogalacturonan I arabinosyltransferases (52). Thus, expanding the repertoire of cell wall pectin structures may have accompanied the transition from avascular to vascular plants.
The identification and functional characterization of UASs from green algae and bryophytes provide a valuable tool to study the role of apiose and apiosylated metabolites in these organisms. The use of comparative genomics and transcript analyses will reveal glycosyltransferases responsible for addition of apiose to secondary metabolites and to wall polysaccharides.

Experimental Procedures
Plant Material and Growth Conditions-Living cultures of the mosses Dicranum, Sphagnum, and Polytrichum; the liverworts Marchantia and Conocephalum; and the algae Mougeotia and Netrium were obtained from Carolina Biological (Burlington, NC), harvested, and kept at Ϫ80°C. P. patens (var.   Table S1.   Phytozome genomic portal were probed for homologs to the amino acid sequence of Arabidopsis AXS1/UAS1 by BLAST analyses. Analysis of the top hits revealed proteins with high sequence identity to Arabidopsis AXS1/UAS1, including several green algae and moss proteins with Ͼ70% amino acid sequence identity, liverwort proteins from Treubia lacunosa (83% identity) and M. paleacea (two incomplete; 65 and 82% identity), and hornwort proteins from Nothoceros aenigmaticus and M. vincentianus with 77 and 76% identity, respectively. The top hit for S. polyrhiza (locus Spipo0G0011100; annotated as "bifunctional polymyxin resistance ArnA protein") has an amino acid sequence identity of 85% to Arabidopsis AXS1/UAS1. The nucleotide sequence cor-

UDP-apiose Synthases of Avascular Plants and Green Algae
OCTOBER 7, 2016 • VOLUME 291 • NUMBER 41 responding to the S. polyrhiza protein was used for primer design and cloning. S. polyrhiza RNA was isolated from 10-day-old fronds. Fronds were collected, vacuum-filtered over nylon mesh, rinsed with deionized water, blotted dry, and then ground to a fine powder in liquid nitrogen using a mortar and pestle. Total RNA was extracted using a Qiagen RNeasy Mini kit with an on-the-column DNase treatment to eliminate genomic DNA contamination. RNA (0.5 g) was then reverse transcribed with an oligo(dT) primer using SuperScript III reverse transcriptase (Life Technologies). A portion of the reverse transcription (RT) reaction (2 l), dNTPs, and 1 unit of Phusion high fidelity DNA polymerase (New England Biolabs, Ipswich, MA) with a 0.2 M concentration of each forward and reverse primer (Integrated DNA Technologies, Coralville, IA; supplemental Table  S2) were used to amplify the SpUAS with the following thermal cycler conditions: one 98°C denaturation cycle for 30 s followed by 25 cycles (each of 8-s denaturation at 98°C, 25-s  annealing at 60°C, and 30-s elongation at 72°C) and finally termination at 4°C. The PCR product was directly cloned into the E. coli expression vector pET28b (Novagen, Darmstadt, Germany) modified to contain an N-terminal His 6 tag followed by a TEV cleavage site (28). RNA was extracted from wild-type and transformed P. patens using 100 mg of 2-week-old gametophyte tissue. Tissue was harvested and immediately frozen in liquid nitrogen, and RNA was extracted and reversed transcribed with oligo(dT). No PpUAS transcript was detected on a 1% (w/v) agarose gel even after a second round of PCR amplification using a 0.2 M concentration of each forward and reverse primer (supplemental Table S2) and an annealing temperature of 56°C. The recently released P. patens Electronic Fluorescent Pictograph browser (54,55) indicates that PpUAS (gene ID Pp1s379_19V6.1) is only expressed at substantial levels in stage S3 sporophyte and archegonia. Thus, a synthetic ORF gene corresponding to PpUAS was obtained (GenScript, Piscataway, NJ).
Genes corresponding to the nucleotide sequences of D. scoparium, H. ciliata, Mougeotia spp., M. paleacea, M. vincentianus, N. digitus, and S. lescurii UASs were obtained from Gen-Script due to the lack of axenic lines. These ORFs were cloned into the modified pET28b expression vector (28) using forward and reverse primers (see supplemental Table S2). The predicted transcript for D. scoparium lacked 8 amino acids at the N-terminal region based on sequence alignment with PpUAS. To obtain the entire D. scoparium ORF, its nucleotide sequence was extended to include the sequence corresponding to the amino acid sequence MTARVLND at the N terminus based on the N-terminal UAS sequence of the moss H. ciliata. For cloning the UAS ORF of M. paleacea, the middle portion of the predicted sequence was constructed to include the amino acid sequence RPLDTIYSNFIDALPVVRYCTDNNKRLIHFSTCE-VYGKTIGCFLPNDSPLRKD based on sequence alignment with a UAS-like homolog in the liverwort T. lacunosa.
Following cloning of the individual UAS genes, the plasmids were sequence-verified (Georgia Genomics Facility, Athens, GA) and termed pET28b-TEV-DsUAS. Transformation of P. patens-The ORF SpUAS was amplified by PCR using forward and reverse primers (supplemental Table S2), directly cloned into pENTR TM /SD/D-TOPO, and Gateway cloned into pTHUbiGate, a P. patens expression vector that has homologous recombination sites at P. patens locus 108 (56) with LR Clonase II (Life Technologies). The expression of SpUAS in pTHUbiGate is driven by a ubiquitin promoter.
For plant transformation, the binary plasmid pTHUbiGate-SpUAS (50 g) was linearized with BsaAI (New England Biolabs) and then precipitated by the addition of ethanol. The precipitate was dissolved in sterile water (1 g/l) and then used to transform P. patens protoplasts (53).
P. patens protoplasts were prepared by treating protonema suspended in aqueous 8.5% (w/v) mannitol for 60 min at room temperature with 2% (w/v) Driselase (Sigma-Aldrich) and shaking at 60 rpm. The suspension was filtered through 40-m nylon mesh, and the released protoplasts were suspended in 8.5% (w/v) mannitol. The suspension was centrifuged (250 ϫ g, 5 min, 22°C), the supernatant was discarded, and the pellet was resuspended in 8.5% mannitol. The filtration and resuspension were repeated two times. The number of protoplasts obtained was then determined using a hemocytometer. Protoplasts were centrifuged, and the protoplasts were suspended to a density of ϳ2 ϫ 10 6 protoplasts/ml in 0.5 M mannitol containing 15 mM MgCl 2 and 0.1% (w/v) MES, pH 5.6 (53). 0.3 ml of the protoplast suspension and 0.3 ml of PEG 8000 (Sigma-Aldrich) were added to 15 l of BsaAI-linearized plasmid DNA, thoroughly mixed, and then kept at room temperature for 10 min. The mixture was heat-shocked for 3 min at 45°C, immediately cooled to room temperature in a water bath, and kept for 10 min. The suspension was then centrifuged, the supernatant was discarded, and the protoplasts were suspended in 5 ml of liquid protoplast regeneration medium (ϭBCDAT plus 8% (w/v) mannitol and 10 mM CaCl 2 ). A portion of this protoplast suspension (1.6 ml) was spread on cellophane layered over protoplast regeneration medium bottom layer (ϭBCDAT plus 6% (w/v) mannitol, 10 mM CaCl 2 , and 1% (w/v) agar) (53). The plates were kept for 5 days in a growth chamber at 22°C.
The cellophane was then transferred onto BCDAT medium containing ampicillin (100 g/ml) and kept for a further 7 days. The cellophane was then transferred onto BCDAT medium lacking antibiotic. After 7 days, the cellophane was transferred to BCDAT medium with ampicillin and kept for an additional 7 days to obtain stable transformants. Transformants were verified by PCR of locus 108 using the appropriate forward and reverse primers (supplemental Table S2) and transcript analysis.

UDP-apiose Synthases of Avascular Plants and Green Algae
The His 6 -tagged proteins were purified using fast flow nickel-Sepharose (GE Healthcare; 2 ml of resin packed in a 15 ϫ 1-cm polypropylene column). Columns were washed and equilibrated with 50 mM sodium phosphate, pH 8.0, containing 300 mM NaCl, and then Fraction S20 was added. Bound His 6 -tagged proteins were eluted with the same buffer containing increasing concentrations of imidazole (10 -250 mM). The active enzymes were eluted in 250 mM imidazole and then dialyzed (6,000 -8,000 molecular weight cutoff; Spectrum Laboratories, Inc.) at 4°C three times for a total of 2 h against 50 mM Tris-HCl, pH 7.6, containing 0.15 M NaCl, 10% (v/v) glycerol, 1 mM DTT, and 10 M NAD ϩ . The dialysates were divided into 150-l aliquots, flash frozen in liquid nitrogen, and stored at Ϫ80°C. Aliquots of purified protein were assayed for activity and run on SDS-PAGE.
SDS-PAGE was performed with 12% (w/w) polyacrylamide gels. Proteins were stained with 0.1% (w/v) Coomassie Brilliant Blue R-250 in aqueous 20% methanol (MeOH) containing 7% (v/v) acetic acid and destained with aqueous 20% methanol containing 7% (v/v) acetic acid. Protein concentrations were determined with the Bradford reagent (58) using bovine serum albumin (BSA) as the standard. The molecular mass of active recombinant SpUAS was estimated by size exclusion chromatography. Purified, recombinant SpUAS was eluted with dialysis buffer as eluent over a Superdex-75 100/300 GL column (23) that had been calibrated with proteins of known molecular mass (Bio-Rad).
Recombinant UAS Enzyme Assays-Unless otherwise indicated, reactions were performed in 50 mM Tris-HCl, pH 7.9 (50 l), containing 1 mM NAD ϩ , 1 mM UDP-GlcA, and 10 g of purified protein. The mixtures were kept at 37°C for up to 45 min, and the reactions were terminated by placing the tubes in boiling water for 2 min followed by the addition of an equal volume of chloroform. The suspensions were vortexed and centrifuged (12,000 ϫ g, 5 min, 22°C), and the aqueous phase was analyzed for nucleotide sugars. 1 H NMR assays (180 l) were performed in D 2 O using 30 g of purified protein.
Characterization of Recombinant SpUAS-SpUAS activity was assayed in different buffers, at different temperatures, and with various additives and nucleotide sugars. For pH studies, purified recombinant SpUAS (10 g) was added to standard reactions (50 l) containing various pH buffers (100 mM), 1 mM NAD ϩ , and 1 mM UDP-GlcA and kept at 37°C for 30 min. Inhibition assays were performed by first supplementing the standard reaction mixtures with various nucleotides and nucleotide sugars, addition of purified protein, and incubation. The amounts of reactants and products were determined by UV spectroscopy and used to calculate enzyme activity as follows. The products from each recombinant enzyme assay were chromatographed over a Q-15 anion exchange column (200 ϫ 1 mm; Amersham Biosciences) by elution with a linear gradient (5 mM to 0.6 M) of ammonium formate over 25 min at a flow rate of 0.25 ml/min using an Agilent (Santa Clara, CA) 1100 Series HPLC equipped with a G1313A autosampler, a G1315B diode array detector, and ChemStation software. Nucleotides and nucleotide sugars were detected by their A 261 nm (for UDP-sugars) and A 259 nm (for NAD ϩ ). The concentrations of reactants and products were determined by comparison of their peak areas with a calibration curve of standard UDP-GlcA (23).
Selected kinetic parameters of recombinant SpUAS (10 g) were determined by varying the concentrations of UDP-GlcA in 50-l reactions consisting of 1 mM NAD ϩ in 50 mM Tris-HCl, pH 7.9. Reactions were kept for 7 min at 37°C, quenched with an equal volume of chloroform, and then vortex mixed. The reaction products in the aqueous phase were separated using a Q-15 anion exchange column as described above, and reaction rates were calculated from the depletion of the UDP-GlcA signal integral normalized to the NAD ϩ signal integral. Values from three independent replicates were used to generate a non-linear regression plot and resultant data using GraphPad Prism Version 6.04.

HILIC-ESI-MS/MS-ESI-MS/MS
analysis was performed on a Shimadzu (Kyoto, Japan) LC-MS-IT-TOF operating in the negative ion mode. Plant nucleotide sugar extracts and in microbe and recombinant enzyme assay products were mixed with 2 ⁄ 3 volume aqueous 95% ACN containing 25 mM ammonium acetate, and an aliquot (10 -20 l) was chromatographed over an Accucore amide-HILIC column (150 ϫ 4.6 mm; Thermo Fisher Scientific, Waltham, MA) eluted at 0.4 ml min Ϫ1 with a linear gradient of aqueous 75% (v/v) acetonitrile containing 40 mM ammonium acetate, pH 4.4, to 50% (v/v) acetonitrile containing 40 mM ammonium acetate, pH 4.4, over 35 min using a Shimadzu LC-30AD HPLC. Mass spectra (mass range, 100 -2,000 m/z) were collected every 1.3 s for 30 min. Second stage MS/MS data were collected by collision-induced dissociation with a collision energy of 35% and a nebulizing helium gas flow of 1.5 ml min Ϫ1 (59).
Nucleotide Sugar Extraction of Plant Tissue-Fresh tissues (100 mg wet weight) were ground in liquid nitrogen using a mortar and pestle and transferred to Eppendorf tubes, and ACN/MeOH/H 2 O (40:40:20, v/v/v; 1 ml) was then added. The tubes were vortexed and rotated for 20 min at 4°C. The tubes were centrifuged (18,000 ϫ g, 5 min, 22°C), and the supernatant was transferred to a clean tube. The solutions were concentrated to ϳ50% of their initial volume using a stream of nitrogen gas, and portions (800 l) were analyzed by HILIC-ESI-MS/MS as stated above.
Fractionation and Cell Wall Polysaccharide Extraction-Fresh plant tissue (0.1-1 g) was suspended in cold ACN/ MeOH/H 2 O (40:40:20, v/v/v; 10 volumes) and ground in a mortar and pestle on ice. The suspension was transferred to a 15-ml Falcon tube and kept for 30 min at 55°C. The suspension was centrifuged (3,000 ϫ g, 5 min, 22°C), and the supernatant (Fraction I; "methanolic extract") was saved. Aqueous 50% (v/v) EtOH (10 volumes) was then added to the pellet, and the suspension was vortexed and kept for 30 min at 55°C. The suspension was centrifuged, and the supernatant (Fraction II) was saved. The procedure was repeated using aqueous 80 and 95% (v/v) EtOH to give soluble Fractions III and IV. The pellet was then suspended in water (5 volumes) and kept for 30 min 55°C. An equal volume of ethyl acetate was added, and the suspension was vortexed and kept for 30 min at 55°C. The suspension was centrifuged, and the top layer (Fraction V) was saved. The lower aqueous layer (Fraction VI) was transferred to a clean borosilicate tube. The pellet was suspended in CHCl 3 /MeOH (1:1, v/v; 5 volumes) and vortexed. After centrifugation, the top aqueous layer (Fraction VII) was saved. The bottom organic layer (Fraction VIII) was also saved. The pellet was suspended in acetone (5 volumes), and after centrifugation the supernatant (Fraction IX) was saved. The final pellet, referred to as cell wall, was allowed to air-dry overnight in a fume hood.
Glycosyl Residue Composition Analysis-The aqueous and organic solvent extracts or cell wall (ϳ1 mg) was supplemented with myo-inositol (10 l of a 5 mM solution) as an internal standard, evaporated to dryness at room temperature using a stream of filtered air (REACTIVAP III, Thermo Fisher), and then hydrolyzed for 2 h at 120°C with 2 M TFA (1 ml). TFA was removed by evaporation under a stream of filtered air (40°C), and the residue was washed with isopropanol (3 ϫ 500 l). The released monosaccharides were then converted into their cor-responding alditol-acetate derivatives according to York et al. (60), and the final residue was dissolved in acetone (100 l).
The alditol-acetate derivatives were analyzed by gas-liquid chromatography (GLC; Agilent 7890A) equipped with a mass selective detector (EI-MS; Agilent 5975C). The sample (1 l) was injected in the splitless mode using an Agilent 7693 autosampler onto a Restek RTx-2330 fused silica column (0.25-mm inner diameter ϫ 30 m, 0.2-m film thickness) with helium as carrier gas at a flow rate of 1.1 ml min Ϫ1 . The oven temperature was held at 80°C for 2 min followed by an increase of 30°C min Ϫ1 to 170°C and then at 4°C min Ϫ1 to 235°C and a hold at 235°C for 20 min. The column was then kept at 250°C for 7 min, cooled to 80°C, and kept at 80°C for 1 min prior to the next injection. The injection port and the transfer line to the EI-MS were kept at 250°C. Alditol-acetate derivatives of authentic apiose, rhamnose, fucose, ribose, arabinose, xylose, mannose, glucose, and galactose (50 g each) were prepared under the same conditions as samples. Monosaccharides were identified based on their retention times and their EI mass spectra. Peak areas, obtained from the total ion chromatogram, were exported to Microsoft Excel and normalized using the amount of sample and the area of the internal standard. The amounts of each monosaccharide in a sample were calculated using the response factors of the monosaccharide standards.