Plant apyrases are nucleoside triphosphate (NTP) diphosphohydrolases (NTPDases) and have been implicated in an array of functions within the plant including the regulation of extracellular ATP. Arabidopsis encodes a family of seven membrane bound apyrases (AtAPY1–7) that comprise three distinct clades, all of which contain the five conserved apyrase domains. With the exception of AtAPY1 and AtAPY2, the biochemical and the sub-cellular characterization of the other members are currently unavailable. In this research, we have shown all seven Arabidopsis apyrases localize to internal membranes comprising the cis-Golgi, endoplasmic reticulum (ER) and endosome, indicating an endo-apyrase classification for the entire family. In addition, all members, with the exception of AtAPY7, can function as endo-apyrases by complementing a yeast double mutant (Δynd1Δgda1) which lacks apyrase activity. Interestingly, complementation of the mutant yeast using well characterized human apyrases could only be accomplished by using a functional ER endo-apyrase (NTPDase6), but not the ecto-apyrase (NTPDase1). Furthermore, the substrate specificity analysis for the Arabidopsis apyrases AtAPY1–6 indicated that each member has a distinct set of preferred substrates covering various NDPs (nucleoside diphosphates) and NTPs. Combining the biochemical analysis and sub-cellular localization of the Arabidopsis apyrases family, the data suggest their possible roles in regulating endomembrane NDP/NMP (nucleoside monophosphate) homoeostasis.
The apyrase class of enzymes (EC 18.104.22.168) is nucleoside triphosphate (NTP) diphosphohydrolases (NTPDases) that belong to the guanosine diphosphatase 1 (GDA1) - CD39 nucleoside phosphatase superfamily and contain five apyrase conserved regions (ACRs). They are active against both nucleoside tri- and nucleoside di-phosphates (NTPs, NDPs), converting them into nucleoside monophosphates (NMPs). Apyrases have been identified in an array of species, including plants, mammals, insects, fungi and bacteria . The NTPDase activity requires divalent cations (Mg2+, Ca2+, Mn2+) and is distinct from the ATPases due to their broader substrate activities and insensitivities to F-type, P-type and V-type ATPase inhibitors .
In mammals, apyrases were initially characterized as having cell surface ATPase activity (ecto-apyrase). The human apyrases are the most extensively characterized family and comprise cell surface localized ecto-apyrases (NTPDases 1, 2, 3 and 8) and endo-apyrases which are associated with the endoplasmic reticulum (ER), Golgi and intracellular vesicles (NTPDase 4–7) . The plasma membrane localized apyrases are mainly involved in the regulation of extracellular ATP to prevent desensitization of purine receptors . In contrast, the intracellular localized ER/Golgi human endo-apyrases are involved in the conversion of NDP to NMP to both drive lumenal glycosylation reactions and produce co-substrates (NMP) for the membrane localized nt sugar antiporters . Saccharomyces cerevisiae (yeast) encodes two apyrases, namely guanosine diphosphatase (GDA1) and yeast nucleoside diphosphatase 1 (YND1) [5,6]. The yeast GDA1 protein is an NDPase (nucleoside diphosphatase) with preferential activity against GDP . YND1 has a broader substrate specificity and can readily hydrolyse both NDPs and NTPs, although with a preference for GDP . The functions of these two yeast apyrases are somewhat redundant as YND1 can partially complement glycosylation defects when expressed in the Δgda1 background. Interestingly, yeast cells (Δynd1Δgda1) lacking both apyrases are still viable .
In plants, the involvement of extracellular ATP as a potential signalling molecule has been proposed for a number of years . A number of studies have demonstrated that plant cells release significant quantities of ATP into their extracellular matrix when they are mechanically stimulated , wounded , during growth  and during stomatal opening . Recently, with the characterization of a plasma membrane localized ATP receptor kinase  a role for plant apyrases in the regulation of extracellular ATP has been strengthened. In the reference plant Arabidopsis thaliana, a total of seven NTPDases have been identified based on the presence of the ACRs . Among the seven members, APYRASE 1 (AtAPY1 At3g04080) and APYRASE 2 (AtAPY2: At5g18280) have been the most extensively investigated. Both AtAPY1 and AtAPY2 have been shown to play numerous physiological roles in pollen development, vegetative growth and stomata opening/closure [12,15,16]. Collectively, these responses were attributed to defects in ecto-nt signalling responses. However, previously both AtAPY1 and AtAPY2 have been identified in plant Golgi proteomes  and their localizations confirmed by fluorescent protein tagging [18,19]. In addition, knocking out either AtAPY1 or AtAPY2 affects latent lumenal UDPase/GDPase activity in microsomal preparations from Arabidopsis which resulted in a minor change to the galactose content of their cell walls . Furthermore, the conditional suppression of AtAPY1 in the atapy2 background resulted in structural changes to the cell wall . These data provide strong evidence to support the hypothesis that AtAPY1 and AtAPY2 function as plant endo-apyrases and are necessary for lumenal glycosylation. However, this functionally defined role as an endo-apyrase would not necessarily preclude a role as regulators of ecto-ATP/ADP concentration via a secretary mechanism, as has been recently argued  based on data showing that immunochemical  and genetic  suppression of AtAPY1 and AtAPY2 results in an increase in extracellular ATP.
Aside from AtAPY1 and AtAPY2, a further five apyrase members are encoded by Arabidopsis (AtAPY3–7), although their biochemical and physiological functions remain elusive. Some initial characterization of AtAPY6 and AtAPY7 has been undertaken, with double knockout (dKO) plants (atapy6atapy7) resulting in late anther dehiscence, exine deformation and low male fertility . These structural changes to the pollen cell wall, in combination with an internal localization for AtAPY6-tagged lines, further support roles as endo-apyrases involved in polysaccharide biosynthesis . Consequently, in an effort to resolve the functional roles of the Arabidopsis apyrase family, we sought to systematically investigate their sub-cellular localizations and determine their substrate specificities and relate these findings to functional roles in the context of the well characterized apyrase family members from humans.
Cloning procedures for heterologous protein expression
The Arabidopsis apyrase family members AtAPY3 (At1g14240), AtAPY4 (At1g14230), AtAPY5 (At1g14250), AtAPY6 (At2g02970) and AtAPY7 (At4g19180) were cloned from a mixed organ Arabidopsis cDNA library using primers designed based on sequences in The Arabidopsis Information Resource (TAIR)  (Supplementary Table S1). PCR products were recombined into pDONR™/Zeo by BP reaction (Life Technologies) and verified by sequencing. The genes AtAPY1 (At3g04080) and AtAPY2 (At5g18280) were previously cloned using a similar approach . For transient subcellular localizations, the AtAPY1 to 7 pDONR™/Zeo constructs were recombined into the N-terminal YFP and C-terminal YFP Gateway® compatible pBullet vectors  by LR reactions (Life Technologies). The human apyrase cDNA sequences were obtained from the Mammalian Gene Collection  and comprised ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1) (BC047664.1), ENTPD6 (BC025980.2) and ENTPD7 (BC122857.1). These sequences were codon optimized for yeast expression (Supplementary Figure S1), synthesized (GenScript), recombined into the pDONR™/Zeo vector by BP reaction (Life Technologies) and verified by sequencing. The AtAPY7 sequence was codon optimized (Supplementary Figure S1) and synthesized (GenScript) for yeast expression. For yeast complementation assays, the pDONR™/Zeo constructs were recombined into a pDR-Leu Gateway® yeast expression vector .
Chromosomal deletion of the GDA1 locus from Saccharomyces cerevisiae
The chromosomal GDA1 locus (YEL042W) was replaced with orotidine 5′-phosphate decarboxylase (URA3) by homologous recombination as previously described . Genomic DNA was extracted from the S. cerevisiae wild-type strain BY4741 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0) using YeaStar™ Genomic DNA Kit (Zymoresearch) and used as template. The yeast GDA1 gene was cloned by PCR using the pGDA1 primers and inserted into pENTR™/D-TOPO® (Life Technologies) to be used as the templates to create the knockout cassette border sequence. A further round of PCR was undertaken using the pGDA1-R primers to create a BamHI site used to replace GDA1 ORF with auxotrophic marker URA3 ORF. The resultant product was cloned into pENTR™/D-TOPO®. The URA3 gene was amplified by PCR from the vector pRS416-GPD  and restriction sites BamHI and MscI were added by PCR using the URA3 primers. The URA3 PCR product was digested with BamHI and MscI and ligated into the pENTR/D-TOPO-GDA backbone to create the knockout cassette pGDA–URA3–tGDA. To knockout the chromosomal GDA1 gene, the Saccharomyces cerevisiae strain Δynd1 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0, ynd1Δ0) obtained from the Yeast Knockout Collection (Thermo Scientific) was transformed with the linearized vector pGDA–URA3–tGDA using Frozen-EZ Yeast Transformation II Kit™ (Zymoresearch). The transformants were selected on solid medium containing yeast nitrogen base (YNB) without amino acids (Becton, Dickinson and Company) supplemented with 2% (w/v) glucose and 1× CSM-Ura (Sunrise Science Products). Genomic DNA was isolated from candidate transformants using YeaStar™ Genomic DNA Kit (Zymoresearch). The integrity of the GDA1 locus was examined by PCR using the following primer sets: left border using primers GDA-L; middle using primers GDA-M and right border using primers GDA3-R. The presence of the inserted URA3 sequence was examined by PCR using the primers URA3–ORF. Primers are detailed in Supplementary Table S1.
Yeast transformation and complementation assay
The Δgda1Δynd1 dKO yeast strain was transformed with the various plasmids using the EZ-YEAST™ transformation kit (MP Biomedicals) and selected on solid media containing YNB without amino acids (Becton, Dickinson and Company) supplemented with 2% (w/v) glucose and 1× CSM-Leu-Ura (Sunrise Science Products). Complementation was assessed by growing single colonies overnight at 30°C in liquid media as described above. Liquid cultures were serially diluted and spotted on solid selection media as outlined above.
Monosaccharide analysis of the yeast cell wall
Sample extraction and preparation procedures were undertaken according to previously described methods employing trifluoroacetic acid (TFA) hydrolysis . Cultures (50 ml) were grown until D=1.0–1.2 and cells harvested by centrifugation at 2000 g for 5 min. Cells were disrupted in 0.5 ml of 10 mM Tris/HCl (pH 8) using glass beads and a vortex. Cell walls were collected by centrifugation (3800 g for 5 min) and washed in cold distilled water and dried in a vacuum concentrator. Cell wall pellets were hydrolysed with 1 ml of 2 M TFA at 100°C for 4 h. Samples were lyophilized and re-suspended in 1 ml of water prior to analysis. Monosaccharide composition was performed using high performance anion exchange chromatography on a Dionex ICS 3000 equipped with a pulse amperometric detector, as previously described . The monosaccharide composition of yeast samples was calculated by linear regression from a five point standard curve comprising glucose, mannose and glucosamine loaded before, during and after the sample set.
RNA extraction and RT-PCR
Total RNA was isolated from yeast strains using YeaStar™ RNA Kit (Zymo Research). Approximately 300 ng of total RNA was treated with DNase (Invitrogen) and used as template for cDNA synthesis by SuperScript III Reverse Transcriptase (Invitrogen). PCR was undertaken using Taq 2X Master Mix (New England Biolabs Inc.) using conditions as provided by the manufacturer. The RT-PCR of apyrase transcripts was undertaken with attB1 and attB2 primers (Supplementary Table S1). The yeast UBC6 gene (ubiquitin-conjugating enzyme) was used as a control.
Total protein was isolated from overnight yeast cultures as previously described . The protein was quantified by Bradford (Thermo Scientific) . A total of 5 μg of total protein was re-suspended in 0.2 M Tris/HCl, pH 6.5, 8% (w/v) SDS, 8% (v/v) 2-mercaptoethanol, 40% (v/v) glycerol and 0.04% (w/v) Bromophenol Blue and boiled for 5 min. Samples were subjected to SDS/PAGE [10% gel (w/v)] and blotted on to PVDF membrane. Heterologous expressed proteins were detected using the Universal antibody (UNI) against the Gateway® attB2 site , followed by incubation with a secondary antibody and detection by chemiluminescence using the Protein Detector™ LumiGLO® Western Blotting Kit (KPL Inc.).
Yeast microsomal preparations
The yeast membranes were isolated from the complemented Δgda1Δynd1 dKO by initial disruption with glass beads in 400 μl of chilled extraction buffer [20 mM Tris/HCl, 10 mM MgCl2, 1 mM EDTA, 5% (v/v) glycerol, 1 mM DTT, 1 mM PMSF and 1× Roche Complete Protease Inhibitor Cocktail]. The cells were centrifuged at 5000 g for 10 min at 4°C and supernatants collected. The supernatant was centrifuged at 50000 g for 1 h at 4°C and the resultant membrane pellet was re-suspended in 10 mM Tris buffer (pH 7.5) for the NTPDase assay.
Measurement of NTPDase activity
A total of 50 μg of microsomal protein was incubated in 500 μl of reaction buffer [3 mM NDP or NTP or NMP (Sigma–Aldrich), 3 mM MnSO4, 30 mM Tris/MES, pH 6.5 and 0.03% (v/v) Triton X-100] for 1 h at room temperature. The released phosphate was measured using the Malachite Green Phosphate Assay (ScienCell Research Laboratories) with slight modifications, namely that 100 μl of reagent A and 100 μl of reagent B were each added to the 50 μl of solution. The incubation times were undertaken according to the protocol.
Plasmid DNA was isolated using QIAprep Spin Miniprep Kit (Qiagen). Particle bombardments for transient localizations were conducted according to previous methods . Essentially, 0.6 μg of plasmid DNA was added to 25 μl of microcarrier/glycerol solution containing 400 μg of microcarriers (1 μm gold, Bio-Rad), followed by 25 μl of 2.5 M CaCl2 and 10 μl of 0.1 M spermidine. The solution was mixed for 10 min at 3000 rpm and supernatant removed. The pellet was washed with 100% (v/v) ethanol and re-suspended in 20 μl of 100% (v/v) ethanol, loaded on to a macrocarrier and air dried. The macrocarrier was placed on to the hepta adapter macrocarrier holder (leaving the other six empty). Fresh epidermal peels from yellow onions or whole Arabidopsis rosettes harvested from 6- to 8-week-old plants were bombarded under vacuum (710 mmHg) at a target distance of 6 cm and a helium pressure of 1100 psi (1 psi ≈ 6.9 kPa). Arabidopsis rosettes were bombarded on 1% (w/v) agar plates containing half strength Murashige and Skoog basal salt mixture. After bombardment, plant materials were kept on plates overnight in the dark until imaging by confocal microscopy .
Phylogenetic analysis and informatics
The Arabidopsis apyrase protein sequences were obtained from TAIR . The human and plant apyrase protein sequences were obtained from GenBank  whereas yeast sequences were from the Saccharomyces Genome Database . Phylogenetic trees were created using MEGA6 , with sequences aligned using MUSCLE (using UPGMB and default parameters), phylogenetic reconstruction was undertaken using maximum likelihood with 1000 Bootstrap Replications. Protein domains were obtained from InterProScan  and predicted transmembrane helices from TMHMM . Protein features were visualized using DOG (Domain Graph, version 1.0) .
The following sequences have been deposited at GenBank: AtAPY1/At3g04080 (JQ937231); AtAPY2/At5g18280 (JQ937238); AtAPY3/At1g14240 (JF830008); AtAPY4/At1g14230 (JF830009); AtAPY5/At1g14250 (JF830010); AtAPY6/At2g02970 (JF830011); AtAPY7/At4g19180 (JQ965809).
The Apyrase family of Arabidopsis thaliana
A total of seven loci have been identified in the Arabidopsis genome that contain the apyrase domain, namely AtAPY1 (At3g04080), AtAPY2 (At5g18280), AtAPY3 (At1g14240), AtAPY4 (At1g14230), AtAPY5 (At1g14250), AtAPY6 (At2g02970) and AtAPY7 (At4g19180). A recent phylogenetic analysis of several hundred plant apyrases indicated that they fall into three major clades . The seven member Arabidopsis apyrase family contains representatives in each clade and are clustered into the AtAPY1-2 clade I (GDA1-like), the AtAPY3–6 (clade II) and AtAPY7 in clade III (Figure 1A).
The apyrase family of Arabidopsis thaliana
Since the eight Homo sapiens (human) apyrase genes (NTPDase1–8) have been characterized both biochemically and genetically  and the two apyrase enzymes from yeast, (GDA1 and YND1) have been extensively characterized , we undertook a phylogenetic analysis with several plant apyrases, the human apyrase family and the two yeast enzymes (Figure 1B). The clade I (GDA-like) Arabidopsis members (AtAPY1 and AtAPY2) form a distinct clade with the other characterized plant apyrases, human apyrases and the yeast GDA1 enzyme (Figure 1B). Although NTPDase6, GDA1, AtAPY1 and AtAPY2 appear to have a substrate preference for NDPs [5,18,37], a number of the plant apyrases in this clade have been associated with exhibiting NTPase activity, namely StAPY3  and PsAPY2 . In humans, NTPase activity is associated with the secreted ecto-apyrase clade members (NTPDase1–3 and NTPDase8) and all display type IV-A membrane protein topology (Supplementary Figure S2). In contrast, plant members of the GDA-like clade are typical type II membrane proteins (Figure 1A). The Arabidopsis AtAPY7 is only weakly associated with the human ecto-apyrase clade, it has a similar membrane topology and can also be classed as a type IV-A membrane protein (Figure 1A).
The final group of Arabidopsis apyrase members forms an independent cluster (clade II) comprising AtAPY3–6. The apyrase members AtAPY3, AtAPY4 and AtAPY5 are recurrent tandem gene duplications on chromosome 1. All three contain a single putative N-terminal transmembrane domain typical of type II membrane proteins. In contrast, AtAPY6 (clade II) would appear to be type IV-A membrane protein (Figure 1A).
Disruption of the yeast apyrases GDA1 and YND1 affects the yeast cell wall composition
The Δynd1Δgda1 dKO was previously created by crossing the Δynd1::URA3 haploid (XGY4) with the Δgda1::LEU2 haploid (G2–11). The Δynd1Δgda1 dKO (KAI1) showed slow growth compared with the single mutants and the wild-type on yeast, peptone, adenine, dextrose (YPAD) plates at 30°C . However, the laboratory that created this Δynd1Δgda1 dKO has lost the original strain. We sought to recreate the Δynd1Δgda1 dKO by replacing the GDA1 ORF with the URA3 ORF by a heterologous exchange in the Δynd1 single mutant background [25,40]. The Δynd1Δgda1 dKO was verified using primer sets designed to assess the presence of the GDA3 ORF (Figure 2A). Only the URA3 ORF was detected by PCR in the Δynd1Δgda1 dKO indicating that the URA3 ORF had successfully replaced the GDA1 ORF in the Δynd1 single mutant (Figure 2A).
Generation and cell wall analysis of Δgda1Δynd1 yeast dKO
The yeast cell wall contains β(1→3)-D-glucan, β(1→6)-D-glucan, chitin and mannoproteins which are mostly synthesized at the plasma membrane . Cell wall mannoproteins are synthesized in the ER/Golgi lumen and are dependent on the delivery of GDP-mannose from the cytosol, a process that is driven by the co-transport of GMP generated by apyrases (GND1/YND1) in the ER/Golgi lumen . We analysed the monosaccharide composition of a TFA hydrolysed insoluble fraction extracted from the Δynd1Δgda1 dKO after growth in modified YNB medium after 2 days at 30°C. The composition of mannose in this insoluble fraction was ~50% less than that found in wild-type (BY4741) cells (Figure 2B). The Δynd1Δgda1 dKO also contained less cell wall material including non-lumenal derived polymers, namely β(1→3)-D-glucan, β(1→6)-D-glucan and N-acetylglucosamine from chitin (Figure 2C), highlighting the importance of the mannoprotein component in the construction of the yeast cell wall. Finally, similar to previously reported results , the newly generated Δynd1Δgda1 dKO also exhibits very slow growth on modified YNB medium (result not shown).
Yeasts lacking endogenous apyrase activity are complemented by human endo-apyrases
The human apyrases represent a biochemically well characterized family of enzymes with varying sub-cellular locations and activities. The members of the human apyrase family comprise the ER lumenal GDA-like NDPases (e.g. NTPDase6), an intracellular membrane-associated clade with NDPase/NTPase activity (e.g. NTPDase7) and the ecto-apyrase group with NTPase activities (e.g. NTPDase1). We were interested in assessing the ability of these defined classes of apyrases to complement the Δynd1Δgda1 dKO generated above. The GDA-like human NTDPase6, with a substrate specificity for NDPs, was able to recover the growth phenotype observed in the Δynd1Δgda1 dKO (Figure 3A). Neither the NTPDase1 (ecto-apyrase) nor the NTPDase7 (intracellular with reported NDPase/NTPase activities) were able to complement the growth phenotype exhibited by the Δynd1Δgda1 dKO (Figure 3A). NTPDase1 has a mixed NTP/NDP substrate specificity , whereas NTPDase7 has a preference for NTPs . The presence of the yeast codon optimized human apyrase gene transcripts was verified by RT-PCR (Supplementary Figure S3). Overall, these results indicate that subcellular context as well as substrate specificity is necessary for complementation of this yeast Δynd1Δgda1 dKO.
Complementation of the Δgda1Δynd1 yeast dKO strain
Yeasts lacking endogenous apyrases can be complemented by Arabidopsis apyrases
In order to assess the in vivo activities of the Arabidopsis apyrase family, we performed a complementation assay in the Δynd1Δgda1 dKO, as described above. Previously, the Arabidopsis clade I apyrase members AtAPY1 and AtAPY2 were shown to act as lumenal NDPases through the independent complementation of the glycosylation phenotype associated with the Δgda1 mutant, as well as the hygromycin sensitivities of the Δynd1 mutant-related defects in the cell wall . When these clade I Arabidopsis apyrases were expressed in the Δynd1Δgda1 dKO, both AtAPY1 and AtAPY2 were able to complement the growth phenotype when compared with yeast harbouring the empty vector (pDR-Leu; Figure 3A). These results support our previous findings that both AtAPY 1 and AtAPY2 are able to function as internal Golgi lumenal NDPases. The heterologous expression of the clade II Arabidopsis apyrase members (AtAPY3–6) in the Δynd1Δgda1 dKO revealed that AtAPY4, AtAPY5 and AtAPY6 were all able to complement the growth defect phenotype of the Δynd1Δgda1 dKO (Figure 3A), demonstrating these enzymes are also able to function as internal Golgi lumenal NDPases. In contrast, AtAPY3 exhibited relatively weak complementation compared with other members of this clade (Figure 3A). The clade III Arabidopsis apyrase AtAPY7 was unable to complement the growth phenotype of the Δynd1Δgda1 dKO (Figure 3A).
An analysis of the monosaccharide composition of the insoluble cell wall fractions from the complemented Δynd1Δgda1 dKO further supported a role for the Arabidopsis apyrases as lumenal NDPases (Figure 3B). The proportion of mannose in cell wall extracts significantly increased in all the complemented strains with the AtAPY5 construct resulting in near wild-type levels (Figure 2B). The AtAPY3 construct was the least able to recover cell wall mannose, reflecting the reduced growth phenotype. The AtAPY4 construct resulted in only a marginal increase in cell wall mannose compared with AtAPY3, but was very capable of complementing the growth phenotype (Figure 3A). The ability to recover mannose in cell wall extracts of the Δynd1Δgda1 dKO probably reflects the activity of each apyrase with respect to the substrate GDP (derived from lumenal GDP-mannose). Cell extracts were not analysed from cells harbouring the AtAPY7 construct as it exhibited no complementation of the growth phenotype.
To ensure the Arabidopsis apyrases were being adequately expressed in the Δynd1Δgda1 dKO, we analysed protein extracts by immunoblotting. Evidence for the expression of constructs containing AtAPY1–5 in the Δynd1Δgda1 dKO was apparent (Supplementary Figure S4). A faint band was detected for AtAPY6 when 25 μg of microsomal protein was analysed by immunoblotting and indicated some full-length product, however no evidence for the AtAPY7 protein could be obtained. Previously we had observed processing of the AtAPY1 construct when expressed in the Δgda1 . In this instance, it is possible that the C-terminal is processed from AtAPY6 and AtAPY7. As a consequence, we undertook RT-PCR analysis to verify the presence of all apyrase transcripts in the Δynd1Δgda1 dKO. Evidence for the presence of all transcripts was apparent for all constructs (Supplementary Figure S3).
The Arabidopsis AtAPY1–6 exhibit apyrase-like activities
Microsomal preparations from the seven Arabidopsis apyrase members expressed in the Δynd1Δgda1 dKO were used to measure latent NTPDase activity via inorganic phosphate release using Malachite Green . The GDA-like clade I members AtAPY1 and AtAPY2 exhibited a clear preference towards the nt substrates UDP (0.7 μmol Pi h−1 μg−1) and UDP/GDP (1–3 μmol Pi h−1 μg−1) respectively (Figure 4), supporting previous reports indicating they both function as UDP/GDPases [18,19]. The clade II member AtAPY3 has a strong preference toward NTPs (8–12 μmol Pi h−1 μg−1) but also has significant activities toward ADP and GDP with 4–6 μmol Pi h−1 μg−1 (Figure 4). In contrast, other members of the clade II apyrase family displayed an array of substrate preferences. No significant NTPase or NDPase activity could be detected for AtAPY4 except a slight affinity for CTP, whereas AtAPY5 demonstrated the highest level of NDP activity measured in our assay, ranging from 10 to 18 μmol Pi h−1 μg−1 (Figure 4). AtAPY6 appears to have a broad range of substrate activities toward all NTP and NDP substrates analysed, with values from 0.5 to 2.5 μmol Pi h−1 μg−1 (Figure 4). Finally, the single clade III representative AtAPY7 displayed no detectable NTPase or NDPase activity under our experimental conditions (result not shown), although the presence of the protein could not be confirmed. In summary, the AtAPY1–6 Arabidopsis enzymes all exhibit classic apyrase-like NTPase and/or NDPases activities, with an absence of NMP activity.
Specific activity of the Arabidopsis apyrase enzymes AtAPY1—6
Subcellular localization of the Arabidopsis apyrase family
Previously, two members of the Arabidopsis apyrase family (AtAPY1 and AtAPY2) implicated as ecto-apyrases were shown to localize to Golgi membranes [17–19]. In an effort to resolve the subcellular distribution of Arabidopsis apyrases and subsequently their potential roles within the cell, we sought to transiently co-localize all seven members using N- and C-terminal YFP fusions. As previously observed, both AtAPY1 and AtAPY2 localize to the cis-Golgi when either the N- or the C-terminal YFP construct was used (Figure 5). Similarly cis-Golgi localization results were identified for AtAPY4, AtAPY5 and AtAPY7 using either the N- or the C-terminal YFP constructs (Figure 5). These data indicate that AtAPY1, 2, 4, 5 and 7 are probably cis-Golgi resident proteins. Neither the AtAPY3 nor the AtAPY6 constructs significantly overlapped with the cis-Golgi marker (Figure 5). The AtAPY3 C-terminal YFP construct resulted in an internal punctate signal with minimal cis-Golgi marker overlap. Further analysis with a trans-Golgi marker vesicle transport through t-SNARE interaction 12 (CFP–VTI12) and an endosomal marker Ras-related in brain F2a (RabF2a) (CFP–RabF2a) indicted that AtAPY3 probably localizes to the endosome (Figure 6). Since the C-terminal AtAPY6 construct produced a diffuse web-like structure, we co-localized this construct using an ER marker (Figure 6). This resulted in a significant signal overlap indicating co-localization with the ER marker. Identical results were obtained by transient localizing in Arabidopsis rosette leaves by particle bombardments (Supplementary Figure S5). The outcome of these localization experiments is summarized in Table 1.
Sub-cellular localization of the Arabidopsis apyrase family
Sub-cellular localization of AtAPY3 and AtAPY6
|AGI||Name||Onion (C-YFP)||Onion (N-YFP)||Arabidopsis (C-YFP)||SUBA* (MS)||SUBA† (FP)||Inferred location|
|AGI||Name||Onion (C-YFP)||Onion (N-YFP)||Arabidopsis (C-YFP)||SUBA* (MS)||SUBA† (FP)||Inferred location|
The apyrase family of Arabidopsis would appear to be representative of plant species with members present in each phylogenetic clade . Based on their sub-cellular distributions, the seven members of the Arabidopsis apyrase family are endo-apyrases. Their sub-cellular distributions are remarkably similar to the human endo-apyrase members with the majority localized to the Golgi apparatus, a single ER-localized candidate and a single member localized to an intracellular vesicle (Figure 7). Our biochemical analysis indicated a wide range of substrate preference for most members of the family, providing evidence of functional diversity.
Schematic diagram summarizing the sub-cellular localization, putative topology and major specific activity of the Arabidopsis apyrase family
The yeast and human apyrase families
The recreation of the Δynd1Δgda1 dKO enabled an investigation of the cell wall of yeast lacking any substantive lumenal apyrase activity. The results indicated that only the mannose content derived from mannoproteins was being substantially affected, with some compensation by D-glucan and chitin occurring. Similar observations have been made with gda1 single mutant in Candida albicans where chitin levels were found to increase in their cell walls . The reduced mannose content of the wall also appeared to affect the total amount of cell wall material, supporting an integrated process for the biosynthesis and construction of the yeast cell wall . The elimination of apyrase activity in yeast did not completely prevent the production of cell wall-derived mannose, indicating that the transport of GDP-mannose was still able to occur, probably at a reduced rate, without the counter substrate GMP. This is in contrast with deletion of VRG4, the Golgi resident GDP-mannose transporter from yeast, which is lethal , as is the VIG9 null mutant, the GDP-mannose pyrophosphorylase essential for the biosynthesis of GDP-mannose . Thus, it is possible that lumenal GDP (or some other molecule) is able to be utilized as a counter substrate to enable the delivery of some GDP-mannose into the Golgi lumen.
The regeneration of the Δynd1Δgda1 dKO enabled us to examine the function of human apyrase family members. We selected the well-characterized ecto-apyrase NTPDase1 to assess the complementation of the mutant yeast strain with a secreted apyrase. Although complementation of the yeast dKO mutant with human apyrases was not as strong as our results with Arabidopsis apyrases, there was clear complementation by Golgi localized human apyrase NTPDase6. Minimal complementation was observed for the vesicle-localized human apyrase NTPDase7 or the ecto-apyrase NTPDase1 (Figure 3). These results indicate that sub-cellular localization and biochemical function are important components of endo-apyrase yeast complementation assays.
Clade I Arabidopsis apyrases: AtAPY1 and AtAPY2
The Arabidopsis apyrases AtAPY1 and AtAPY2 are related to yeast GDA1-like (clade I) and are the most extensively characterized plant apyrases. Although they have been implicated to function at the plasma membrane as ATPases and ADPases regulating ecto-ATP/ADP concentrations [16,48], earlier evidence provides a distinct functional role in Arabidopsis for both AtAPY1 and AtAPY2, namely as endo-apyrases residing in the Golgi lumen with UDPase and GDPase activities [18,19]. We have now demonstrated that both enzymes exhibit a clear substrate preference for UDP, as would be expected for apyrases responsible for the turnover of UDP after glycosylation reactions within the Golgi lumen.
Several plant apyrases associated with the GDA1-like clade have been implicated as ecto-apyrases through their association with NTPase activity and apoplastic localizations [38,39]. However, since the human ecto-apyrase NTPDase1 appears to require glycosylation for NTDPase activity  progression through the secretory pathway to provide glycan maturity and NTDPase function may be required. Whether this is also a feature of plant apyrases is currently unknown, but the fact that AtAPY1 and AtrAPY2 are both able to complement endo-apyrase activity in yeast, localize to the cis-Golgi and possess NDPase activities in vitro, would indicate a central role for these enzymes as Arabidopsis endo-apyrases involved in the conversion of NDPs to NMPs as an important component of endomembrane glycosylation, as previously discussed [18,19].
Clade II Arabidopsis apyrases: AtAPY3, AtAPY4 and AtAPY5
The Arabidopsis clade II apyrase members have a diverse topology, with AtAPY3–5 exhibiting a single putative N-terminal transmembrane domain, whereas AtAPY6 appears to possess both an N- and a C- terminal transmembrane domain. The mixed topology for this clade is not unique to Arabidopsis, with examples in both Glycine max and Vitis vinifera to name a few . AtAPY3, AtAPY4 and AtAPY5 occur as recurrent tandem duplications and share 68% identity; all three are expressed during Arabidopsis development with AtAPY3 predominately in the roots and both AtAPY4/AtAPY5 in the vegetative rosette . Based on this information, it may be possible to speculate that these enzymes undertake similar functions at different developmental stages. However, the biochemical and localization analyses would support a more varied functional role in Arabidopsis.
The AtAPY3–YFP construct localized to small punctate structures with minimal overlap to the cis-Golgi or trans-Golgi markers, however there was considerable overlap with the late endosomal marker, RabF2a [51,52]. This non-Golgi localization was recently confirmed in Nicotiana benthamiana . The enzyme exhibited a clear substrate preference for NTPs and was unable to successfully complement the yeast Δgda1Δynd1 dKO, which could have been due to poor expression or high protein turnover (Supplementary Figure S4). Among the reported intracellular human NTPDases (NTPDase4–7), only NTPDase7 displays a strong preference for NTPs and is also reported to localize to internal vesicles . However, no specific functional role for NTPDase 7 has been reported . Yeast complementation involving NTPDase7 also resulted in poor growth of the Δgda1Δynd1 dKO. Although similar results were observed for AtAPY3, it is possible that sub-cellular localization played an important role in these experiments. Given its NTP preference and sub-cellular localization, a possible role in intra-cellular signalling through an involvement with the GTP-binding/GTPase regulatory networks  or NTP secretion  would be conceivable.
In contrast, both AtAPY4-YFP and YFP-AtAPY5 localized to the cis-Golgi and their coding regions were able to complement the yeast Δgda1Δynd1 dKO. AtAPY5 exhibited the highest specific activities for NDPs of all the Arabidopsis apyrases, which resulted in the high mannose yield from cell wall extracts of complemented yeast strain. The biochemical analysis of AtAPY4 resulted in the lowest NDPase activates measured, exhibiting a substrate preference for CTP. However, even with this reduced NDPase activity, its localization to the Golgi lumen probably assisted in the positive complementation phenotype in yeast Δgda1Δynd1 dKO. Overall, these results suggest in planta endo-apyrase roles for these enzymes with functional roles related to their NTPDase activities which could constitute lumenal NDPase activity during specific aspects of vegetative growth. Whether AtAPY4 also functions as a lumenal NTPase requires further investigations.
The clade II Arabidopsis apyrase: AtAPY6
Similar to AtAPY6, all four human ecto-apyrases are reported to contain both N- and C-terminal transmembrane domains. Based on these structural characteristics, we initially considered AtAPY6 a potential ecto-apyrase. However, AtAPY6–YFP constructs were localized to the ER, biochemical assays indicate broad NDP/NTP substrate preferences and its heterologous expression in yeast complemented the Δgda1Δynd1 dKO, resulting in a high amount of mannose recovered from cell wall extracts. In contrast, the human ecto-apyrase NTPDase1 was unable to successfully complement the yeast mutant strain when compared with results from the human GDA1-like apyrase, NTPDase6.
Similar to other members of clade II, AtAPY6 exhibits a defined expression pattern during Arabidopsis development, namely mature pollen . A recent analysis of AtAPY6 confirmed its high expression in mature pollen and an analysis of atapy6 mutants indicated a minor role in pollen development associated with abnormal exine patterning . The localization of AtAPY6 to the ER and its broad substrate specificity is unique among the Arabidopsis apyrase family. The only other ER-localized apyrase is the human NTPDase5, which is thought to remove the inhibiting effects of UDP and support the efficient re-glucosylation of proteins enabling correct folding of glycoproteins . These observations in combination with the biochemical and molecular data would support an endo-apyrase role for AtAPY6. Finally with its ER localization, it may have a role in supporting glucosylation of nascent N-glycans  through the turnover of UDP; inhibition of which is most evident in maturing pollen .
Clade III Arabidopsis apyrase: AtAPY7
The final member of the seven apyrase-like proteins encoded by Arabidopsis and the only member of clade III is AtAPY7. Although localization of the AtAPY7–YFP construct seems to indicate cis-Golgi localization, the construct was unable to complement the Δgda1Δynd1 dKO. Furthermore, biochemical analysis of microsomal fractions showed no NTPDase activity. Although it is possible that the heterologous expression in yeast was unsuccessful, we were able to detect AtAPY7 transcripts in the transformed yeast cells. The AtAPY7 protein sequence contains the well characterized five ACRs, indicating it is a member of the apyrase family .
A recent molecular analysis of AtAPY7 determined that it was ubiquitously expressed in a range of Arabidopsis tissues and developmental stages. An analysis of atapy7 mutants also indicated minor aberrations to the pollen exine as observed in atapy6 mutants. Interestingly, dKO mutants lacking both AtAPY6 and AtAPY7 produced relatively normal plants but with low male fertility from collapsed pollen which further resulted in reduced seed set . Given the likelihood that AtAPY7 does not appear to function as a typical apyrase, it is difficult to explain the synergistic effects observed in atapy6atapy7. The expression pattern for AtAPY7 would not indicate a specific role in pollen development, indicating its function could be associated with an important lumenal process which is in demand during pollen maturation. A more detailed analysis of AtAPY7 function needs to be undertaken to determine its role in pollen development.
Overall our results would indicate that, at least for the reference plant Arabidopsis, all members of the apyrase family are localized internally within the endomembrane system. Although there is evidence that individual apyrase members from other plant species can be secreted, we saw no evidence for this with Arabidopsis apyrase family members. Indeed, biochemical evidence and yeast complementation experiments would suggest an overall preference for NDPs over NTPs. These findings, along with a number of recent studies [18,19,58], do not necessarily preclude the existence of an ecto-apyrase in Arabidopsis, however they do indicate that further investigations are required. Interestingly, many of the characterized plant ecto-apyrases are encoded by legume species and appear to have roles in host–pathogen interactions such as nodulation . Since the extracellular space is where symbiotic interactions initially occur, it is possible that apyrases from legume species have evolved to undertake ecto-apyrase functions associated with these interactions. Given Arabidopsis is incapable of forming such nitrogen fixing associations, it is conceivable that this function never evolved in certain plant lineages. However, further work is necessary to determine whether this is the case.
Tsan-Yu Chiu, Dominique Loqué and Joshua Heazlewood designed the experiment. Tsan-Yu Chiu, Jeemeng Lao and Bianca Manalansan performed the experiments. Tsan-Yu Chiu, Joshua Heazlewood, Dominique Loqué and Stanley Roux analysed the data. Tsan-Yu Chiu and Joshua Heazlewood wrote the manuscript.
The vector pRS416-GPD was kindly provided by Dr Arlen Johnson (University of Texas, Austin). We would also like to thank Huu Tran and Dr Suzan Yilmaz (Joint BioEnergy Institute) for advice and assistance with the assays.
This work was supported by the Office of Science, Office of Biological and Environmental Research, of the U.S. Department of Energy [grant number DE-AC02-05CH11231]; and the Australian Research Council Future Fellowship [grant number JLH FT130101165].
apyrase conserved regions
ectonucleoside triphosphate diphosphohydrolase
nucleoside triphosphate diphosphohydrolase
Ras-related in brain
The Arabidopsis Information Resource
orotidine 5'-phosphate decarboxylase; VTI12, vesicle transport through t-SNARE interaction 12; YNB, yeast nitrogen base; YND1, yeast nucleoside diphosphatase