Recycling of pyridoxine (vitamin B6) by PUP1 in Arabidopsis


For correspondence (e-mail


Vitamin B6 is a cofactor for more than 140 essential enzymatic reactions and was recently proposed as a potent antioxidant, playing a role in the photoprotection of plants. De novo biosynthesis of the vitamin has been described relatively recently and is derived from simple sugar precursors as well as glutamine. In addition, the vitamin can be taken up from exogenous sources in a broad range of organisms, including plants. However, specific transporters have been identified only in yeast. Here we assess the ability of the family of Arabidopsis purine permeases (PUPs) to transport vitamin B6. Several members of the family complement the growth phenotype of a Saccharomyces cerevisiae mutant strain impaired in both de novo biosynthesis of vitamin B6 as well as its uptake. The strongest activity was observed with PUP1 and was confirmed by direct measurement of uptake in yeast as well as in planta, defining PUP1 as a high affinity transporter for pyridoxine. At the tissue level the protein is localised to hydathodes and here we use confocal microscopy to illustrate that at the cellular level it is targeted to the plasma membrane. Interestingly, we observe alterations in pyridoxine recycling from the guttation sap upon overexpression of PUP1 and in a pup1 mutant, consistent with the role of the protein in retrieval of pyridoxine. Furthermore, combining the pup1 mutant with a vitamin B6 de novo biosynthesis mutant (pdx1.3) corroborates that PUP1 is involved in the uptake of the vitamin.


Vitamin B6 is an essential enzymatic cofactor for all organisms. De novo biosynthesis of the vitamin occurs only in bacteria, fungi and plants. Auxotrophic organisms, including most animals, thus have to take it up from their diet or from their extracellular environment. Vitamin B6 uptake activities have been observed in various bacterial species, yeast and Homo sapiens, regardless of if they are auxotrophic or prototrophic (Mulligan and Snell, 1976; Shane and Snell, 1976; Yagi et al., 1996; Stolz and Vielreicher, 2003; Stolz et al., 2005; Said et al., 2008). However, such activities have not been investigated in plants. The term vitamin B6 refers to pyridoxine (PN), pyridoxal (PL), pyridoxamine (PM) and their respective 5′ phosphorylated forms (Figure 1). At the chemical structure level, these compounds share a pyridine ring but differ in the substitution at the 4′ position being an alcohol [PN and pyridoxine 5′-phosphate (PNP)], an aldehyde [PL and pyridoxal 5′-phosphate (PLP)] or an amine [PM and pyridoxamine 5′-phosphate (PMP)]. PLP is the predominant cofactor form of vitamin B6 and is, with a few exceptions (mainly certain members of the γ-division of proteobacteria), de novo biosynthesized by a protein complex made of PDX1 and PDX2 (Strohmeier et al., 2006; Zein et al., 2006). PDX1 and PDX2 function as a glutamine amidotransferase and biosynthesize PLP directly from glutamine, glyceraldehyde 3-phosphate and ribose 5-phosphate (Burns et al., 2005; Raschle et al., 2005, 2007; Tambasco-Studart et al., 2005). The genome of Arabidopsis contains three PDX1 homologues named PDX1.1 (At2g38230), PDX1.2 (At3g16050) and PDX1.3 (At5g01410; Tambasco-Studart et al., 2005). However, only PDX1.1 and PDX1.3 code for catalytically active vitamin B6 biosynthetic enzymes. On the other hand, only one gene encoding PDX2 (At5g60540) was identified in Arabidopsis, the corresponding mutant of which is embryo lethal (Tambasco-Studart et al., 2005). All these biosynthetic mutants can be rescued by supplementation with PN (Chen and Xiong, 2005; Tambasco-Studart et al., 2005, 2007; Wagner et al., 2006).

Figure 1.

Depiction of B6 vitamers as well as selected purines and pyrimidines. The predominant forms of vitamin B6 are illustrated in the upper and middle rows. The lower row illustrates 4-deoxypyridoxine (4-dPN), a potent antagonist of vitamin B6, as well as cytosine (a pyrimidine) and adenine (a purine). The accepted numbering of the atoms is shown for PLP. PL, pyridoxal; PLP, pyridoxal 5′-phosphate; PM, pyridoxamine; PMP, pyridoxamine 5′-phosphate; PN, pyridoxine; PNP, pyridoxine 5′-phosphate.

In addition to acting as a cofactor, vitamin B6 has been implicated in protection against oxidative stress (Ehrenshaft et al., 1999; Osmani et al., 1999; Bilski et al., 2000). Indeed, supplementation of plant, yeast or mammalian cell cultures with exogenous vitamin B6 leads to increased cell resistance to oxidative damage (Jain and Lim, 2001; Danon et al., 2005; Endo et al., 2007). In Arabidopsis, the decrease in total vitamin B6 content observed in the pdx1.3 mutant is concomitant with an alteration of the maximum quantum efficiency of photosystem II as well as with a decrease in the level of the D1 protein (Titiz et al., 2006). Additionally, chlorophyll contents are decreased in this mutant as a consequence of altered chlorophyll synthase activity (Havaux et al., 2009). Interestingly, the diminution of the D1 protein level observed in the pdx1.3 mutant reverts to wild-type levels in the presence of exogenous PN (Titiz et al., 2006). In addition to the rescue of the pdx mutant developmental phenotypes with exogenous PN, these observations suggest that the latter can be taken up from exogenous sources, implying that there is an uptake system in place in plants.

The only vitamin B6 transporters identified to date are the yeast transporters, Tpn1p and Bsu1 from Saccharomyces cerevisiae and Schizosaccharomyces pombe, respectively (Stolz and Vielreicher, 2003; Stolz et al., 2005). While no strong homology has been observed in other organisms, Tpn1p displays 29% identity with Fcy2, the yeast transporter for adenine, guanine and cytosine (Weber et al., 1990; Stolz and Vielreicher, 2003). Indeed, mutation of the latter allowed the identification of the Arabidopsis nucleobase transporter PUP1 (for purine permease) by growth phenotype complementation (Gillissen et al., 2000). PUP1 directs a high affinity transport system for adenine, cytosine and cytokinins, likely to be mediated by proton symport (Gillissen et al., 2000; Bürkle et al., 2003). It is annotated as a member of 21 phylogenetically related PUP genes found by in silico analysis of the Arabidopsis genome (Gillissen et al., 2000; Jelesko, 2012; Table S1). Previous findings have shown that PUP1 is expressed in leaves and is localised to the epithem cells of the hydathodes, in particular (Gillissen et al., 2000; Bürkle et al., 2003).

The present study reports on the ability of Arabidopsis PUPs to transport vitamin B6. Several members of the family complement the growth phenotype of a Saccharomyces cerevisiae mutant strain impaired in both de novo biosynthesis of vitamin B6 as well as its uptake, with PUP1 having the strongest activity. We demonstrate through uptake studies in yeast as well as Arabidopsis that PUP1 is a high affinity transporter for PN, an activity that appears to be compatible with our estimation of the concentration of vitamin B6 compounds in plant cells. Confocal microscopy studies indicate that PUP1 is localised at the plasma membrane of Arabidopsis cells. We have isolated a pup1 mutant and by combining it with pdx1.3 furthermore corroborate that PUP1 participates in exogenous PN uptake in planta. Most strikingly, we observe alterations in the PN content of the guttation sap upon overexpression of PUP1 and in the pup1 mutant. Given the confinement of PUP1 to the hydathodes, these data are consistent with the role of PUP1 in the retrieval of PN from the guttation sap, which can later be used as a source of this essential compound.


Screening for putative Arabidopsis vitamin B6 transporters

No gene with strong homology to the yeast vitamin B6 transporters, TPN1 or bsu1+, could be found in the Arabidopsis genome. Therefore, we screened for putative vitamin B6 transporters by performing a complementation study of the MVY30 S. cerevisiae mutant strain (snz1-sno1∆::his5+ tpn1∆::kanMX4) employing an Arabidopsis cDNA library (Minet et al., 1992). The latter yeast strain has the SNZ1 and SNO1 genes deleted (the yeast homologues of PDX1 and PDX2, respectively) and cannot biosynthesize vitamin B6 de novo in thiamine-containing medium (Rodriquez-Navarro et al., 2002). Moreover, the deletion of TPN1 prevents the uptake of exogenous vitamin B6 (Stolz and Vielreicher, 2003). It was reported that supplementation with a PN concentration of at least 2 μm restores growth of the strain, presumably either through passive diffusion or by transport through non-specific proteins (Stolz and Vielreicher, 2003). Thus, transformation of the yeast strain with an Arabidopsis cDNA library and screening for complementation on medium with PN concentrations below 2 μm should allow the identification of Arabidopsis transporters that can import vitamin B6. This approach, however, did not reveal an Arabidopsis cDNA encoding a protein that can transport the vitamin, i.e. no transformants could be identified on low PN-containing medium.

As an alternative strategy, we sought to select known transporters that could have the ability to transport vitamin B6 in Arabidopsis. As mentioned above Tpn1p is 29% identical to the yeast nucleobase transporter Fcy2. A complementation study of a mutant of the latter permitted the identification of the Arabidopsis PUP family of proteins (Gillissen et al., 2000). While the name of the latter family derives from their ability to transport purines, they also appear to transport pyrimidine derivatives, e.g. cytosine (Gillissen et al., 2000). As vitamin B6 possesses chemical structure homology with pyrimidines (Figure 1), we decided to investigate the ability of the PUP family to transport vitamin B6. Progress in the annotation of the Arabidopsis genome since the work of Gillissen et al. (2000) and a more recent report (Jelesko, 2012) has now led to the establishment of a complete list of the PUP family. An updated list of the 21 PUP annotations that constitute the family is thus presented in Table S1. A phylogenetic analysis of PUP cDNA and amino acid sequences establishes a classification comprising four different subfamilies (Figure S1a,b). PUP9, putatively encoding a protein of 45 amino acid residues in length (Table S1) and therefore not compatible with the predicted structure of a functional transporter was omitted from this analysis.

Evidence for vitamin B6 import by members of the PUP family of proteins

In order to test the ability of members of the PUP protein family to complement the growth phenotype of the MVY30 S. cerevisiae strain, 13 of the PUP cDNAs were reverse transcribed from Arabidopsis (ecotype Columbia) leaf mRNA by reverse transcription polymerase chain reaction (RT-PCR) and cloned into the yeast/bacteria shuttle vector pDR195 (Rentsch et al., 1995; a list of primers used is given in Table S2). Notably, members of all subfamilies are represented in the complementation analysis (see Figure S1a,b). Each amplified member of the PUP family, with the empty vector as a control, was transformed into the MVY30 strain. Transformants were then tested for their ability to grow on minimal medium that contained selective concentrations of B6 vitamers (Figure 2). While all transformants grew in the presence of 2 μm PN, only expression of PUP1, 2, 3 or 4 restored growth of the MVY30 strain on medium that contained 0.2 μm PN after 2 days of incubation, whereas only PUP1 complemented the growth phenotype on 0.02 μm PN within the same time frame (Figure 2). Several members of the PUP family appeared to at least partially complement the growth phenotype on 0.02 μm PL-containing medium, with PUP1, 2, 4 and 14 being the most significant, as the growth was considerably stronger than the control (Figure 2). Only PUP1, 2, 4 and 10 could complement the growth of MVY30 on PM-containing medium and necessitated a concentration of 2 μm to observe a comparable level of growth. No growth was observed with the phosphorylated vitamers. Notably, in all cases the strongest level of complementation was observed with PUP1 (Figure 2). It is also worth mentioning that PUP1 from ecotype C24 (from which the Minet library is derived) did not complement the MVY30 yeast strain as strongly as that from the Columbia ecotype. We have noted that there are several nucleotide substitutions in PUP1 between the C24 and the Columbia ecotypes, which may account for the difference in specificity and may explain why we did not retrieve it in our screen.

Figure 2.

Several members of the Arabidopsis purine permease (PUP) family complement a vitamin B6 transport deficient yeast strain. Saccharomyces cerevisiae strain MVY30 (snz1-sno1∆::his5+ tpn1∆::kanMX4) deficient in vitamin B6 biosynthesis and transport, transformed with either the empty vector (C) or various members of the PUP family (as indicated). The three rows represent suspensions (optical density (OD) at 600 nm of 0.2, 0.02 and 0.002, from top to bottom). Pictures were taken 2 days after incubation at 30°C.

Biochemical characteristics of vitamin B6 uptake by PUP1 in yeast

We next performed uptake assays with [3H]-labelled PN, which is available commercially. We selected PUP1, PUP2, PUP3 and PUP4 for this analysis, in addition to using the empty vector as a negative control (Figure 3a). Employing this assay, uptake of PN could only be detected in the strains that expressed PUP1, while weaker activity could be observed with PUP2. No significant uptake activity could be measured for the strains that expressed PUP3 or PUP4 over that of the control strain (Figure 3a). As PUP1 demonstrated the highest transport activity, we focused on this transporter for further analyses. PUP1 displays an optimum pH of 4 for PN uptake (Figure 3b). As this could imply a proton-symport transport mechanism as has been shown for adenine and cytosine uptake by PUP1 (Gillissen et al., 2000), we also tested sensitivity to the protonophores carbonylcyanide-4-trifluoromethoxyphenylhydrazone (FCCP) and carbonyl cyanide 3-chlorophenylhydrazone (CCCP; Figure 3c). Indeed, the activity of PUP1 for transport of PN was reduced to either 3 or 44% in the presence of the respective protonophores. Activity is also reduced in the absence of glucose (to 66%). As the latter directly activates the plasma membrane proton ATPase, leading to an increased proton motive force across the plasma membrane, these combined findings could suggest that PN uptake occurs by proton symport (Figure 3c). Interestingly, PUP1 has an estimated KM value of 102 ± 40 μm for PN under these conditions (Figure 3a inset), which is in the same range as that reported for adenine (30 μm), cytosine (20 μm) or trans-zeatin (40 μm; Gillissen et al., 2000; Bürkle et al., 2003). We also measured uptake of [3H]-PN in competition with a 10-fold excess of PL, PM, 4-deoxy PN, adenine, cytosine or trans-zeatin (Figure 3d). Whereas PL reduced PN uptake by 77%, PM only reduced it by 24%. This supports the complementation assays and may indicate that PUP1 can additionally transport both PM and PL. Adenine, cytosine and trans-zeatin reduced uptake activity to a similar extent as PN and PL, indicating that all of these substrates are likely to be carried by PUP1 to a similar degree. It is intriguing that the vitamin B6 antagonist, 4-deoxy PN, is the strongest inhibitor of PN uptake among the compounds tested.

Figure 3.

Purine permease (PUP)1 shows the strongest import of vitamin B6 in yeast by an apparent proton-symport mechanism and competes with cytokinins. (a) Time course of pyridoxine (PN; 100 μm) uptake at pH 4.0 in Saccharomyces cerevisiae MVY30 expressing either PUP1, 2, 3 or 4 from Arabidopsis. For the Michaelis–Menten plot (inset), the strain expressing PUP1 was incubated with various substrate concentrations (from 0.01 to 1.0 mM) and the uptake velocity (kobs) was determined for 2 min at pH 4. (b) Determination of the optimum pH of PUP1 for PN transport in yeast monitored at 20-sec intervals for 2 min in different phosphate/citrate buffers with a substrate concentration of 100 μm and pH values as indicated. (c) Energy requirement of PUP1 in yeast. Uptake experiments were performed as described in (b) at pH 4 and in the presence of 1% (w/v) of d-glucose (+Glucose) or in its absence. The protonophores carbonyl cyanide 3-chlorophenylhydrazone (CCCP) and carbonylcyanide-4-trifluoromethoxyphenylhydrazone (FCCP) were added to a final concentration of 100 μM 2 min before the labelled substrate. (d) Competition studies with B6 vitamers, 4-deoxypyridoxine (4-deoxyPN), adenine, cytosine or zeatin. PN uptake activity was determined at a total PN concentration of 200 μm in the presence of a 10-fold excess of competitor (2 mm). Activities in the presence of inhibitors were expressed as a percentage compared with the activity measured with PN alone (% activity). In all cases, error bars represent the standard deviation of three independent measurements.

Subcellular localisation of PUP1 in Arabidopsis

As PUP1 was the most efficient PUP protein for vitamin B6 uptake in yeast, we analysed the subcellular localisation of a PUP1:YFP (yellow fluorescent protein) fusion protein in Arabidopsis by fluorescence confocal microscopy. For this analysis, PUP1 was cloned upstream of the YFP sequence in the pB7YWG2.0 vector, where expression is under control of the cauliflower mosaic virus 35S promoter (Karimi et al., 2002). The constructs employed were initially transiently expressed in Arabidopsis protoplasts that had been transformed by the polyethylene glycol-mediated method (Jin et al., 2001). While fluorescence of YFP alone was visible in the cytosol as would be expected, PUP1:YFP fluorescence was observed at the periphery of the cell, which suggested localisation to the plasma membrane (Figure S2). We also analysed expression of the same constructs as above in stable Arabidopsis lines (i.e. PUP1:YFP in addition to those that expressed YFP alone as a control). Cotyledons and roots of five independent lines homozygous for the PUP1:YFP transgene that displayed YFP fluorescence were analysed by confocal microscopy 3 days after germination. In all lines, PUP1:YFP fluorescence was visible as a thin line that surrounded upper epidermal cells of cotyledons as well as the guard cells of stomata (Figure 4a). Fluorescence at the cellular periphery was also observed in root epidermal cells (Figure 4b). Moreover, in both types of tissues, no fluorescence indicative of internal organelle localisation could be observed (Figure 4a,b). Indeed, plasmolysis of the cells corroborated the latter statement with fluorescence being maintained at the membrane periphery of the shrunken cell (Figure 4c,d). Furthermore, PUP1:YFP fluorescence was visible as fluorescent fibrous structures that linked the plasma membrane to the cell wall likely to be Hechtian strands, (Figure 4d). This feature has been reported in previous studies of plasmolysed plant cells that expressed plasma membrane proteins fused to a fluorescent protein (Sun et al., 2005; Speth et al., 2009).

Figure 4.

Purine permease (PUP)1 localises to the plasma membrane of Arabidopsis epidermal cells. (a) Confocal sections of the upper epidermis of cotyledons from 3-day-old seedlings expressing either the PUP1–YFP (yellow fluorescent protein) fusion protein (top panel) or YFP alone (bottom panel). Auto-fluorescence of chlorophyll (left panel) and YFP fluorescence (middle panel) were recorded and overlaid (right panel). The scale bars represent 50 μm. White arrows indicate the plasma membrane of epidermal cells or stomatal guard cells. (b) Confocal sections of root epidermis from 3-day-old Arabidopsis seedlings expressing either the PUP1–YFP fusion protein (top panel) or the YFP (bottom panel). The fluorescence signals of the FM4–64 dye (plasma membrane specific, left panel) and the YFP (middle panel) were recorded and superimposed (Merge). The corresponding differential interference contrast (DIC) pictures are shown in the right panel. The scale bars represent 10 μm. The cytosolic localisation of the YFP alone is indicated by white arrows. (c) As in (a) except that the DIC channel pictures (left panel) and YFP fluorescence (middle panel) were recorded and overlaid (right panel), and cotyledons were plasmolysed by placing them in 1 m potassium nitrate for 10 min prior to microscopic analysis. (d) As in (b) except that the samples were plasmolysed for 5 min in 1 m potassium nitrate prior to observation. White arrows indicate the fluorescent fibrous structures linking the plasma membrane to the cell wall, likely to be Hechtian strands, in plasmolysed seedlings expressing the PUP1–YFP fusion protein.

Evidence for PUP1 mediated PN uptake in planta

In order to investigate the effect of PUP1 inactivation in planta, we obtained seeds of a putative pup1 Arabidopsis T-DNA insertion mutant line (SALK_110605) from the European Arabidopsis Stock Centre (NASC). Plants homozygous for the pup1 mutation could be isolated from the pool of seeds given and quantitative real-time RT-PCR confirmed the impaired expression of PUP1 (736-fold decrease; Figure 5a). No obvious developmental phenotype was observed during the complete growth cycle of pup1 plants compared with wild-type (ecotype Columbia) under our standard culture conditions. The absence of a visible phenotype is not surprising, as functional redundancies between PUP1 and the other PUPs are expected to occur in the plant.

Figure 5.

The restoration of chlorophyll content by pyridoxine (PN) supplementation is impaired in a pdx1.3 pup1 double mutant. (a) Cartoon of purine permease 1 (PUP1) organisation. The PUP1 exons and single intron are displayed by black boxes and raised lines, respectively. The numbers reflect the base pair number relative to the start codon (labelled 1). The T-DNA insertion site (inverted triangle) and the positions of hybridisation sites for primers (P1 and P2) used in quantitative RT-PCR amplification are indicated. PUP1 expression was analysed in leaves of 3-week-old wild-type and pup1 mutant plants. The reactions were normalised using ubiquitin-conjugation enzyme (At5g25760) as a control. Error bars represent the standard error of three biological replicates with two technical repeats. (b) Phenotypic effect of PN supplementation on pdx1.3, pup1 and pdx1.3 pup1 mutant plants. Plants were watered normally for 3 weeks and then separated into two batches. One batch was watered from then on with 500 μm PN (+PN) and the other with water (–PN). Pictures were taken 3 days after PN supplementation. (c) Chlorophyll content of mutant and wild-type plants. Total chlorophyll, chlorophyll a (Chla) and chlorophyll b (Chlb) contents were determined by absorption spectrophotometry. Black and grey bars correspond to –PN and +PN watering conditions, respectively. Error bars represent the standard deviation of three biological replicates. The asterisks indicate significant decreases compared with the wild-type, according to Student's t-test with < 0.05. (d) Modification of chlorophyll content induced by PN supplementation. Changes in chlorophyll contents between –PN and +PN watering conditions for each Arabidopsis line were expressed in per cent. Black bars: chlorophyll a; Light grey bars: chlorophyll b; Dark grey bars: total chlorophyll. Error bars represent the standard error of three biological replicates. Student's t-test was applied to evaluate significant differences between pdx1.3 and pdx1.3 pup1 mutants (*: P-value < 0.05).

In order to investigate the relationship between PUP1 and the transport of vitamin B6, we took advantage of the fact that deficiency in the vitamin such as in the Arabidopsis de novo biosynthetic mutant, pdx1.3, leads to a slightly pale phenotype under moderate light conditions due to increased sensitivity to photo-oxidative stress and a decreased chlorophyll content (Titiz et al., 2006; Havaux et al., 2009). This phenotype can be rescued by supplementation of the growth medium with 5 μm PN, a situation that suggested transport of the vitamin from the medium to leaf cells (Titiz et al., 2006). To study the effect of the pup1 mutation on the rescue of the pdx1.3 mutant phenotype, we crossed both mutant lines to generate the pdx1.3 pup1 double mutant line. Double mutant lines in the Columbia genetic background were verified by PCR analysis of genomic DNA. When cultivated under moderate light intensity, pdx1.3 and pdx1.3 pup1 plants are paler than wild-type plants (Figure 5b). Indeed, total chlorophyll content in pdx1.3 mutant plants is reduced by 23% compared with the wild-type (Figure 5c), in accordance with previously published data (Titiz et al., 2006; Havaux et al., 2009). Conversely, pdx1.3 pup1 double mutant lines display a decrease of 15% of total chlorophyll content relative to wild-type (Figure 5c). In both cases, the decrease in chlorophyll content is more pronounced for chlorophyll b than for chlorophyll a (Figure 5c). Interestingly, while application of exogenous PN reduces the paleness of pdx1.3 to approach the wild-type phenotype, this rescue seems to be less complete in the pdx1.3 pup1 double mutant (Figure 5b). This finding is corroborated by measurement of the chlorophyll content, where the chlorophyll levels in pdx1.3 are statistically closer to that of the wild-type compared with that of pdx1.3 pup1 after supplementation with PN (Figure 5c). Indeed, the percentage increase in chlorophyll content is lower in the pdx1.3 pup1 double mutant (6% increase in total chlorophyll) compared with that of the pdx1.3 mutant plants (16% increase in total chlorophyll; Figure 5d).

PUP1-mediated PN uptake in Arabidopsis protoplasts

To further probe the vitamin B6 transport activity of PUP1 in planta, we measured [3H]-PN uptake in mesophyll protoplasts isolated from Arabidopsis plants with stable expression of PUP1 fused to the YFP as well as from plants with stable expression of YFP alone. After 7 min of incubation with the radioactive substrate, the amount of PN taken up in PUP1 over-expressing cells was 14 times higher than in control cells (Figure 6a). The viability rate, average diameter (23.4 ± 0.5 μm, 23.5 ± 0.5 μm for YFP and PUP1:YFP, respectively) and size repartition (Figure 6b) were similar for both protoplast preparations, showing that the observed increase results from PUP1 activity. Furthermore, kinetic analysis of PN uptake by PUP1 revealed a KM of 78 ± 14 μm (Figure 6c), in the same order of magnitude as the KM value determined in yeast, and a maximum velocity of 130 pmol PN per min per 105 cells.

Figure 6.

Purine permease (PUP)1 mediates pyridoxine (PN) uptake in Arabidopsis. (a) Pyridoxine uptake (90 μm total) by protoplasts at pH 5.6 expressing either PUP1–YFP or YFP alone as a control. Error bars represent the standard deviation of three biological replicates. (b) Analysis of protoplast size distribution. The numbers of cells within the indicated size ranges (in μm cell diameter) were determined in populations of 244 (YFP, light grey) or 212 (PUP1–YFP, dark grey) protoplasts. Average diameters were 23.4 ± 0.46 and 23.45 ± 0.51 for YFP and PUP1–YFP expressing protoplasts, respectively. (c) Estimation of the KM for [3H]-PN uptake. The linearity of the uptake reaction between 1 and 3 min was confirmed for each concentration of PN. Error bars represent the standard deviation of three biological replicates.

In order to compare these data with the physiological concentration of vitamin B6 in Arabidopsis mesophyll cells, we estimated the B6 vitamer contents in wild-type mesophyll protoplasts by High-Performance Liquid Chromatography (HPLC) (Figure S3). The concentration of PLP and PMP (4.5 μm and 3.2 μm, respectively) under these conditions was higher than those of PM, PL or PN (0.5, 1 and 0.3 μm, respectively), with the total concentration of the five vitamers measured being approximately 10 μm.

HPLC analysis of the vitamin B6 vitamer contents in Arabidopsis guttation sap

PUP1 was previously reported to be expressed in the epithem cells of hydathodes, according to histochemical analyses of a GUS-promoter fusion in Arabidopsis (Bürkle et al., 2003). These cells are located at the interface between the xylem endings and the leaf epidermis and have been proposed to function in both the secretion and the retrieval of solutes from the guttation sap (Höhn, 1951; Galatis, 1988). The fact that PUP1 seems to act as a proton-coupled symporter is in favour of a function in the retrieval more than in the export of purines, pyrimidines and cytokinins from the xylem sap. Thus, a model in which PUP1 conserves these compounds from the xylem sap preventing loss through guttation has been proposed (Bürkle et al., 2003).

To assess if this protein is involved in vitamin B6 recycling following a similar process, we analysed the vitamin B6 contents in the guttation sap of wild-type and pup1 mutant plants, as well as that of the pup1 mutant line complemented with PUP1 protein under the control of the cauliflower mosaic virus 35S promoter (Figure 7). While the phosphorylated vitamers could not be detected in our analysis, the estimated concentrations of PM, PL or PN in the wild-type sap were found to be 1.7, 3.7 and 3 nm, respectively (Figure 7a). This concentration is similar to the PN concentrations determined in three species of monocot in a previous study, ranging from 0.6 to 60 nm (Goatley and Lewis, 1965). Intriguingly, the PN and PL contents in the sap of the pup1 mutant were significantly increased (by 64 and 253%, respectively) compared with the wild-type (Figure 7b). On the other hand, the PN content dropped back to wild-type levels while the PL concentration was significantly lower than that of wild-type in the guttation sap of the pup1 mutant complemented with PUP1 (Figure 7b). These observations provide strong evidence that PUP1 is involved in the retrieval of PN and PL during guttation.

Figure 7.

The inactivation of purine permease 1 (PUP1) leads to an increase of pyridoxal and pyridoxine concentrations in the guttation sap of Arabidopsis plants. (a) HPLC analysis of the guttation sap (750–950 μl) collected from wild-type (Col-0), the pup1 mutant or pup1 complemented with PUP1 under the control of the CaMV 35S promoter (pup1/35S PUP1). The profiles shown are from a single experiment but representative of five biological replicates. The peaks corresponding to the B6 vitamers were identified with the use of commercial standards (Stds). The amount of each of the standard compounds injected was 5 pmol. (b) Relative contents of PM, PL and PN in the guttation sap of pup1 and pup1/35S-PUP1 plants compared with the wild-type reference (Col-0). The vitamer concentrations in pup1 and pup1/35S-PUP1 samples were expressed as a percentage compared with wild-type. Error bars represent the standard error of five biological replicates. The asterisks indicate significant differences compared with the wild-type, according to a Student's t-test with < 0.01. 4-dPN, 4-deoxypyridoxine; PL, pyridoxal; PLP, pyridoxal 5′-phosphate; PM, pyridoxamine; PMP, pyridoxamine 5′-phosphate; PN, pyridoxine; PNP, pyridoxine 5′-phosphate.


To date there had been no report on the transport of vitamin B6 in plants. Yet, the previous observation that the impaired root phenotype as well as the plastidic D1 protein steady-state level in Arabidopsis pdx1.3 mutant plants can be rescued by vitamin B6 supplementation, indicated that both long distance and subcellular transport occurs in plants (Titiz et al., 2006). However, no gene with strong homology to identified transporters, e.g. Tpn1p or Bsu1+ from yeast, is present in the Arabidopsis genome. We chose to study the Arabidopsis PUP family of proteins given their ability to transport pyrimidines as well as purines (Gillissen et al., 2000; Bürkle et al., 2003). Our yeast growth complementation experiments suggest that certain members of the PUP family may indeed increase PN, PL, and PM transport. As we observed, the strongest complementation of the MVY30 yeast mutant strain with PUP1, we focused on its characterization in this study.

PUP1 has been described as a high affinity transporter for adenine, cytosine and trans-zeatin (Gillissen et al., 2000; Bürkle et al., 2003). Transport activity is sensitive to protonophores and to a lack of glucose, suggesting that PUP1 is a proton-coupled transporter. The increase of transport activity concomitant with a decrease of pH also argues in favour of this hypothesis (Bürkle et al., 2003). In line with this, PN transport activity as measured here with PUP1 in yeast has an optimum pH of 4 and appears to involve proton symport, suggesting that PN transport occurs via the same mechanism. Importantly, the ability to transport PN as reflected in the almost identical KM of 102 and 78 μm measured in yeast and in plants, respectively, is very similar to that (ca. 30 μm) measured for adenine, cytosine or trans-zeatin in yeast (Gillissen et al., 2000; Bürkle et al., 2003). Therefore, the substrate range for PUP1 comprising adenine, cytosine and the cytokinins can now be extended to PN. The competition experiments in yeast reveal a preference of transport for the B6 vitamers or its derivatives in the order 4-dPN, PN, PL, PM but PUP1 does not seem to transport the phosphorylated vitamers.

The root apical meristem is the major site for cytokinin biosynthesis, yet this phytohormone affects several processes taking place in aerial organs of the plant (Mok and Mok, 2001). Therefore, cytokinins have to be transported from roots to shoots via the xylem (Weiler and Ziegler, 1981; Emery et al., 2000). Contrary to cytokinin metabolism, no evidence for organ specificity of vitamin B6 biosynthesis has been reported. The associated genes (PDX1.1, PDX1.3 and PDX2) are expressed in all plant organs according to quantitative PCR and protein expression analyses (Titiz et al., 2006). However, the long distance transport of vitamin B6 from roots to shoots [as occurs when mutant plants (e.g. pdx1.3) are supplemented with a solution of PN] or even if they take up the vitamin from the soil as has been previously suggested (Titiz et al., 2006), could follow the same mechanism as proposed for cytokinins. PUP1 appears to play a role in this process, as rescue of the chlorophyll content of the pdx1.3 pup1 double mutant line by PN supplementation is significantly lower than the rescue of pdx1.3 single mutant plants. However, this rescue is not totally abolished suggesting that secondary transport systems coexist with PUP1 for uptake of the vitamin by leaf cells. Candidates for this activity are PUP2, 3, 4, 14 and 10 (see Figure 1).

The lack of observation of a strong phenotype in pup1 lines presumably reflects redundancy within the PUP family of proteins. For example, the expression of PUP2 has been reported to partially overlap with that of PUP1, being also expressed in leaves and hydathodes, albeit weaker, but is strongly expressed in the phloem (Bürkle et al., 2003). As adenine transport activities are important in supplying pollen and germinating seeds with high ATP concentrations, an associated phenotype in the pup1 mutant would be anticipated. However, in this study, we did not observe such a phenotype in the pup1 mutant analysed and therefore does not support this hypothesis. Whereas the local function of cytokinins is thought to be mediated through de novo biosynthesis, salvage between different forms as well as translocation and long distance transport is also thought to play a role (Sakakibara, 2006; Cedzich et al., 2008). Root to shoot translocation is thought to be through the xylem, while a reflux mechanism through the phloem is also thought to operate (Cedzich et al., 2008). Thus, multiple cellular importers and exporters would be expected to exist. Whether there is a ubiquitous transporter or tissue specific distribution has not yet been addressed. The multiple members of the PUP family could fulfill such functions given their diverse tissue distribution and expression levels (Bürkle et al., 2003; Jelesko, 2012).

On the other hand, our study on the guttation sap of Arabidopsis provides clear evidence for the involvement of PUP1 in the recycling or retrieval of PN (and PL). Guttation refers to the loss of water and its associated dissolved materials from leaves. Hydathodes are the main points of guttation fluid production, as they represent low resistance routes to the flow of fluid from the trachery endings to the outside through the apoplastic spaces between the epithem and mesophyll cell layers. It has previously been demonstrated that the guttation fluid of rye, wheat and barley seedlings comprises inorganic compounds, sugars and amino acids (Goatley and Lewis, 1965). Intriguingly, several of the B vitamins were also found in this fluid, among them vitamin B6. Here we demonstrate that the non-phosphorylated B6 vitamers, PN, PL and PM are found in the guttation fluid of Arabidopsis (Figure 7). Hydathodes have been shown to be involved in the secretion of compounds, e.g. boron exudation as observed in the Sahara barley cultivar (Sutton et al., 2007), and export of amino acids as demonstrated with the Arabidopsis GLUTAMINE DUMPER 1 mutant (Pilot et al., 2004). However, they are also thought to be involved in selective absorption and retrieval of solutes, analogous to the kidney (Pilot et al., 2004). PUP1 has been localised to the epithem cells of hydathodes (Bürkle et al., 2003). In this study, we observed a considerable increase in the level of PN and PL in the guttation fluid of the pup1 mutant, while the levels of the compounds were restored to that of the wild-type in the complemented mutant (Figure 7). These data support a mechanism whereby PUP1 serves to import (i.e. retrieve) these compounds from the guttation sap. We consider this as another fine example of ‘waste not, want not’, conferring plants with the ability to recycle valuable compounds turning them into a resource.

Experimental Procedures

Plasmid constructs for PUP expression in yeast

The constructs used for expression in S. cerevisiae were obtained by cloning the corresponding PUP cDNAs into pDR195 (Rentsch et al., 1995). The cDNAs of PUP1, 4, 6, 7, 10, 11 and 12 were amplified upon isolation of the corresponding cDNA by RT-PCR from RNA (100 mg) isolated from Arabidopsis rosette leaves (ecotype Columbia) using the NucleoSpin RNA Plant kit (Macherey-Nagel, cDNA was synthesized from 1 μg of total RNA using Superscript Reverse Transcription Enzyme II (Life Technologies, using oligo-dT primers. cDNA was amplified by PCR using the primers listed in Table S2 and subcloned into the pCR2.1-TOPO plasmid (Life Technologies). The cDNAs of PUP14, 16, 17 and 18 were amplified by PCR from the following clones ordered at the European Arabidopsis Resource Center (NASC, BX814406 (PUP14), BX815869 (PUP16), DQ446369 (PUP17), C105208 (PUP18). The constructs obtained were digested with appropriate restriction enzymes and ligated into pDR195. The plasmids pDR195:PUP2 and pDR195:PUP3 were as described in Bürkle et al. (2003). All constructs were verified by sequencing (Microsynth,

Yeast complementation

S. cerevisiae strain MVY30 (MATa leu2-3, 112 his3-11, 15 trp1-1 ade2-1 ura3-1 can1-100 (W303-1A), snz1-sno1∆::his5+ tpn1∆::kanMX4; Stolz and Vielreicher, 2003) was generated by transformation of W303–1A snz1-sno1∆::his5+ with a tpn1∆::kanMX4 disruption cassette that was amplified from a deletion strain (Euroscarf, This strain was grown on YPD (2% d-glucose, 2% peptone, 1% yeast extract) plates for 48 h prior to transformation. Fifty micro litre of cells were resuspended in 1 ml of sterile water, washed and resuspended in 50 μl of 100 mm LiAc prior to addition of 50 μg of salmon sperm DNA (Sigma-Aldrich, and 5 μg of plasmid DNA. Three hundred μl of 40% polyethylene glycol 3350/100 mm LiAc were mixed into the suspension followed by heat shock at 42°C for 20 min. The cells were plated on synthetic defined (SD) medium minus uracil (ForMedium, and transformants became visible after 72 h at 30°C.

For the complementation assays, the strains were cultivated overnight at 30°C in 5 ml of SD medium minus uracil. Cells were washed and the OD600 of each cell suspension was adjusted to 0.2, 0.02 and 0.002 units by dilution with sterile water. Yeast suspensions were spotted with a replica-plating tool (Sigma-Aldrich) on SD medium minus uracil containing either PN, PM, PMP (Sigma-Aldrich), or PLP, PL (Acros Organics, PNP was produced by phosphorylation of PN with Bacillus subtilis PdxK (Park et al., 2004). At least three independent experiments were performed with the use of the non-phosphorylated vitamers whereas the phosphorylated vitamers were tested once.

Uptake experiments in yeast

S. cerevisiae MVY30 cells that expressed PUP1, PUP2, PUP3, PUP4 or carrying the empty vector were grown in medium that contained 0.67% yeast nitrogen base without amino acids (BD Biosciences,, 2% glucose, 20 mg l−1 adenine, 20 mg l−1 histidine, 20 mg l−1 tryptophan and 30 mg l−1 leucine to OD600 of 0.8–1.5. The cells were washed and resuspended to OD600 2.5–10 in phosphate/citrate buffer. Standard uptake assays were performed at pH 4.0 and at 30°C. Tests were started by addition of 5.25 μl of a solution of PN, which consisted of 0.25 μl [3H]-PN (20 Ci mmol−1, 1 mCi mL−1, ARC Inc., and 5 μl unlabelled 10 mm PN, producing a final concentration of 100 μm. Six aliquots of 60 μl were removed at intervals of 20–30 sec, diluted with 5 ml of water, filtered on glass fibre filters (Pall Type A/E, 1 μm pore size) and washed with 5 ml water. The filters were placed in scintillation vials, mixed with 3 ml of Rotiszint Eco Plus and counted (Perkin Elmer TriCarb liquid scintillation counter). The optimum pH of PUP1 was determined with phosphate/citrate buffers adjusted to the pH values indicated. Competition experiments were performed at a substrate concentration of 200 μm PN with 2 mm of putative competitors. Competitors were added 120 sec before the labelled substrate. The KM value was determined with substrate solutions containing a constant amount of labelled PN and various concentrations of unlabelled PN. The protonophores carbonyl cyanide m-chlorophenylhydrazone (CCCP) or carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP) were used at a final concentration of 100 μm. The cellular concentrations of PN were calculated from the ratio of the radioactivity on the filters to that of an unfiltered aliquot of cells.

Uptake experiments in Arabidopsis protoplasts

Arabidopsis mesophyll protoplasts were isolated as described above and PN uptake was measured according to (Fettke et al., 2011) with some modifications: 6.8 × 105 cells were incubated for 7 min in a total volume of 1.1 ml of W5 solution (1.5 mm MES, pH 5.6 containing 154 mm sodium chloride, 125 mm calcium chloride, 5 mm potassium chloride and 5 mm glucose) with the addition of 90 μm of PN and 9.25 kBq [3H]-PN (20 Ci mmol−1, 37 MBq mL−1, ARC Inc., For kinetic analysis, 4.4 × 105 cells were incubated for 3 min with a constant amount of [3H]-PN, as mentioned above and varying total PN concentrations. Viability tests were carried out with fluorescein diacetate (Sigma-Aldrich) as described by (Jones and Senft, 1985) using an Eclipse 80i fluorescence microscope (Nikon Instruments Europe B.V., equipped with a Green Fluorescent Protein filter set (excitation: 457/487 nm, emission: 502/537). Image analysis and diameter measurements were performed with the ImageJ software (

HPLC analysis of the Arabidopsis guttation sap

Three-week-old Arabidopsis plants, cultivated under long-day conditions (i.e. a 16-h photoperiod at 100–120 μmol photons m−2 sec−1 at 22°C and 8-h in the dark at 18°C), were placed in plastic trays with water containers 10 h after the beginning of the light period. The trays were covered with plastic lids in order to maintain a high level of humidity during 14 h. Trays were uncovered just after the end of the dark period and guttation droplets were collected with a hypodermic syringe. The collected fluid (250–1050 μl) was freeze-dried for 48 h in darkness and resuspended in 65 μl of 50 mm ammonium acetate, pH 4. Fifty micro litre samples were analysed by HPLC as described in the supporting methods. Vitamer quantification was carried out using the linear range of a standard curve constructed with known amounts of commercial standards.

Accession numbers

Ubiquitin-conjugation enzyme, At5g25760; PDX1.3, At5g 01410; PUP1, At1g28230; PUP2, At2g33750; PUP3, At1g2 8220; PUP4, At1g30840; PUP6, At4g18190; PUP7, At4g 18197; PUP10, At4g18210; PUP11, At1g44750; PUP12, At5 g41160; PUP14, At1g19770; PUP16, At1g09860; PUP17, At1g57943; PUP18, At1g57990. Germplasm information for the pdx1.3 (SALK_086418) and pup1 (SALK_110605) T-DNA insertion lines used in this study can be found in the European Arabidopsis Stock Centre under accession number N586418 and N610605, respectively.


The generous support of the Swiss National Science Foundation (SNF) grant PP00A_119186 to T.B.F. is gratefully acknowledged as well as the support of the University of Geneva and financial support from the Ernst & Lucie Schmidheiny and Marc Birkigt Foundations. We thank the European Arabidopsis Stock Centre (NASC) for seeds of line SALK_110605 and cDNAs of the indicated PUP members. We are also indebted to Christian Megies, Florence Ares-Orpel and Mireille De Meyer-Fague for technical support.