Transport of cytokinins mediated by purine transporters of the PUP family expressed in phloem, hydathodes, and pollen of Arabidopsis


For correspondence (fax +49 7071 29 3287; e-mail


Nucleobases and derivatives like cytokinins and caffeine are translocated in the plant vascular system. Transport studies in cultured Arabidopsis cells indicate that adenine and cytokinin are transported by a common H+-coupled high-affinity purine transport system. Transport properties are similar to that of Arabidopsis purine transporters AtPUP1 and 2. When expressed in yeast, AtPUP1 and 2 mediate energy-dependent high-affinity adenine uptake, whereas AtPUP3 activity was not detectable. Similar to the results from cell cultures, purine permeases (PUP) mediated uptake of adenine can be inhibited by cytokinins, indicating that cytokinins are transport substrates. Direct measurements demonstrate that AtPUP1 is capable of mediating uptake of radiolabeled trans-zeatin. Cytokinin uptake is strongly inhibited by adenine and isopentenyladenine but is poorly inhibited by 6-chloropurine. A number of physiological cytokinins including trans- and cis-zeatin are also efficient competitors for AtPUP2-mediated adenine uptake, suggesting that AtPUP2 is also able to mediate cytokinin transport. Furthermore, AtPUP1 mediates transport of caffeine and ribosylated purine derivatives in yeast. Promoter–reporter gene studies point towards AtPUP1 expression in the epithem of hydathodes and the stigma surface of siliques, suggesting a role in retrieval of cytokinins from xylem sap to prevent loss during guttation. The AtPUP2 promoter drives GUS reporter gene activity in the phloem of Arabidopsis leaves, indicating a role in long-distance transport of adenine and cytokinins. Promoter activity of AtPUP3 was only found in pollen. In summary, three closely related PUPs are differentially expressed in Arabidopsis and at least two PUPs have properties similar to the adenine and cytokinin transport system identified in Arabidopsis cell cultures.


As a result of their immobile nature, plants require highly efficient mechanisms for adaptation to rapidly changing environmental conditions and for communication between the distal organs of the plant. Long-distance signaling is preferentially mediated by chemical signals as opposed to electrical signaling used by animals. It is therefore not surprising that besides classical hormones, such as steroids, oligopeptides, and eicosanoid-like compounds, plants have developed a specific set of phytohormones. Most phytohormones are synthesized by a few conversions from normal metabolic intermediates. The phytohormone cytokinin is generated by prenylation of nucleosides or nucleotides by cytokinin synthase (Astot et al., 2000; Kakimoto, 2001; Takei et al., 2001). Cytokinins not only control cell division, but affect many physiological and developmental functions, e.g. leaf senescence, nutrient mobilization and biomass distribution, apical dominance, formation and activity of shoot apical meristems, floral transition, breaking of bud dormancy, and seed germination (Mok and Mok, 2001). Root apical meristems seem to represent major sites of the synthesis of free cytokinins, however, young leaves and embryos can also serve as sources. Cytokinins are transported from roots to shoots via the xylem, additionally reflux occurs in the phloem, supporting the hypothesis that they serve as mobile signaling molecules (Beveridge et al., 1997; Emery et al., 2000; Weiler and Ziegler, 1981). Thus, it is obvious that multiple cellular importers and exporters are required to allow efficient mobilization and targeted translocation. However, in contrast to the mechanisms of polar transport of the hormone auxin, basically nothing is known about the mechanisms of cytokinin transport (Swarup et al., 2000). When cell cultures are exposed to free cytokinin bases, these are rapidly taken up and inactivated by glycosylation and stored in the vacuole (Fußeder et al., 1989). Cis- and trans-zeatin O-glucosyl and O-xylosyl transferase genes have been identified, and will help towards a better understanding of phytohormone metabolism (Martin et al., 1999a,b, 2001).

At elevated atmospheric CO2, plants undergo an accelerated life cycle leading to increased nitrogen requirements, uptake and mobilization, and to increased leaf expansion (Ludewig and Sonnewald, 2000). One cause of the higher growth rates could be increased production and transport of cytokinins as compared to ambient conditions. In fact, increased concentrations of cytokinins were found in xylem exudate of plants grown in elevated CO2 (Yong et al., 2000). The lack of a direct correlation between the increases of cytokinin concentration in xylem sap and leaf contents indicates a complex interaction between transport and metabolism along the translocation path.

Recently, the enzymes responsible for cytokinin metabolism, i.e. cytokinin oxidase and corresponding genes, have been identified (Bilyeu et al., 2001; Houba-Hérin et al., 1999). Ectopic overexpression of several Arabidopsis cytokinin oxidase genes had a dramatic effect on the shoot to root ratios and on the leaf cell number (Werner et al., 2001). The presence of signal sequences together with yeast expression data may indicate that some of the enzymes are secreted. Besides a role in cytokinin degradation, cytokinin oxidases thus may also control cytokinin concentrations available to the plasma membrane cytokinin receptors. Candidates for plasma membrane receptors belonging to the class of two component histidine kinase systems (Inoue et al., 2001; Kakimoto, 1996; Suzuki et al., 2001; Ueguchi et al., 2001) have been identified. On the other hand, the presence of potential intracellular cytokinin receptors may suggest that besides extracellular signal perception, the uptake of cytokinins into cells may also be important (Brault et al., 1997; Kulaeva et al., 1995, 1998). Furthermore, if a significant portion of cytokinin is catabolized extracellularly, transporters are required for the recycling of adenine. Recently, nuclear response regulators and transcriptional repressors have also been identified, providing a deep insight into the cytokinin-mediated signal transduction cascade (Hwang and Sheen, 2001).

In agreement with the above findings, transport seems to be highly controlled, as a local supply of cytokinins by microorganisms on senescing leaves leads to a local delay of senescence. Furthermore, the grafting of wild-type shoots onto root stocks of transgenic tobacco overproducing cytokinins as a result of ectopic expression of the Agrobacterium tumefaciens ipt gene, did not lead to the release of lateral shoot buds from suppression or to accelerated senescence (Faiss et al., 1997). Also, localized induction of ipt gene expression had only local effects, leading to a paracrine hypothesis of cytokinin action. Thus, the major step, which is least understood and still undefined at the molecular level, is the role of cytokinin and cytokinin catabolite transporters. Surprisingly, no reference was found concerning uptake measurements using radiolabeled cytokinins aiming at a characterization of plant cytokinin transport.

To identify potential cytokinin transporters, an indirect approach was chosen based on the structural similarity of cytokinins and purine bases and the necessity of the co-existence of adenine and cytokinin transport systems, if metabolism can occur in the apoplasm. A purine transport-deficient yeast mutant fcy2 was used for suppression cloning of plant transporter genes, enabling the identification and characterization of a new superfamily of small hydrophobic polytopic membrane proteins purine permeases (PUP) mediating high-affinity transport of nucleobases (Gillissen et al., 2000). Competition studies showed that AtPUP1 was also able to recognize cytokinins.

The present study describes an adenine uptake system found in cultured Arabidopsis cells that is inhibited by cytokinins and thus shares properties similar to that of AtPUP1. The ability of AtPUP1 to transport cytokinins was proven by direct uptake studies using radiolabeled trans-zeatin in yeast expressing AtPUP1. Analysis of expression using promoter–reporter fusion indicates that AtPUP1 may play a role in the retrieval of nucleobase derivatives including xylem-delivered cytokinins in the epithem of hydathodes and at the stigma surface of siliques. Two homologs of AtPUP1 were cloned and expressed in yeast. AtPUP2 mediated adenine uptake and recognized trans- and cis-zeatin, isopentenyladenine, kinetin, and benzylaminopurine as substrates. In contrast, AtPUP3 did not show detectable activity in yeast. The AtPUP2 promoter drives expression of the GUS reporter gene in the phloem of Arabidopsis leaves, suggesting a role in phloem loading and transport of adenine and cytokinins, whereas AtPUP3 promoter activity is restricted to pollen.


Uptake of adenine by Arabidopsis cell cultures

To determine the properties of adenine uptake by Arabidopsis cell cultures, 14C-adenine transport was measured. Uptake was linear for at least 3 min and sensitive to the protonophore CCCP, suggesting a secondary active uptake mechanism (Figure 1a). Competition with a 10-fold excess of unlabeled adenine, isopentenyladenine, and trans-zeatin strongly reduced the uptake rate, indicating that adenine and cytokinins are taken up by a common transport system (Figure 1b). Trans-zeatin riboside had almost no inhibitory effect, similar as for AtPUP1 (Gillissen et al., 2000). Expression of the adenine transporter AtPUP1 in the cell culture was confirmed by RT-PCR (Figure 1c).

Figure 1.

Uptake of adenine by Arabidopsis suspension cell culture.

(a) Time and energy dependence. Arabidopsis suspension cells were assayed for 14C-adenine uptake at 20 µm substrate concentration and pH 5.7 in the presence (○) or absence (□) of 100 µm carbonyl cyanide m-chlorophenyl-hydrazone (CCCP).

(b) Competition studies with adenine and cytokinins. Uptake of 14C-adenine was determined at 20 µm adenine in the presence of a 10-fold excess (200 µm) of unlabeled competitors. Values represent the mean ± SD of three independent experiments; WW, wet weight; iPA, isopentenyladenine; ZR, trans-zeatin riboside.

(c) Analysis of AtPUP1 expression by RT-PCR. RNA from suspension cells was converted to cDNA by reverse transcription. A 524 bp fragment of AtPUP1 was amplified by 30 PCR cycles.

Transport of cytokinins by AtPUP1

Energy dependence and specificity indicated that a purine transporter of the PUP family may be responsible for the uptake systems found in cell cultures. Additionally, competition studies in yeast have indicated that AtPUP1 is able to recognize cytokinins. However, competition does not prove that a substance is actually transported (Gillissen et al., 2000). To increase the sensitivity of the uptake system and in order to prove cytokinin transport directly, AtPUP1 was expressed in yeast under the control of a modified strong yeast H+-ATPase PMA1 promoter. For this purpose, the cDNA encoding AtPUP1 was subcloned into pDR195 (Rentsch et al., 1995). Expression of AtPUP1 in the yeast mutant MG887-1 deficient for adenine uptake revealed significantly better growth on a medium containing adenine as the sole N-source as compared to expression under control of the PGK promoter in the original vector pFL61 (Figure 2). This result indicates that a stronger expression leads to an increased number of AtPUP1 molecules at the plasma membrane and thus to higher transport rates for adenine.

Figure 2.

Functional complementation of the yeast adenine transport mutant MG887-1 by AtPUP1 under control of two different promoters.

MG887-1 yeast was transformed with AtPUP1 expressed under control of a PMA1 promoter fragment in pDR195, from the PGK promoter in pFL61 or with the empty pDR195 vector as control. Yeast strains were plated on minimal medium containing adenine as the sole nitrogen source (1 mg ml−1) and grown for 6 (left panel) and 12 (right panel) days.

The ability of AtPUP1 to transport cytokinin was analyzed by direct uptake measurements of radiolabeled trans-zeatin into strain MGG887-1 transformed with AtPUP1. A typical time course of trans-3H-zeatin uptake is shown in Figure 3(a). The results demonstrate an uptake of trans-zeatin by AtPUP1 at a concentration of 100 µm, however, uptake was linear for only a short time (approximately 60 sec). In contrast, almost no uptake was detectable in yeast transformed with the empty pDR195 vector. The tritium label of radiolabeled trans-zeatin is located on the purine ring of trans-zeatin (Figure S1a). To exclude the possibility that uptake is caused by radiochemical impurities like 3H-adenine, HPLC analysis was performed, revealing that more than 98% of detected radioactivity had the same retention time as trans-zeatin (Figure S1b), indicating that the measured uptake corresponds to transport of trans-zeatin by AtPUP1. Using only the initial linear phase, a crude determination of the affinity indicates a Km value of approximately 40 µm similar to the Ki of trans-zeatin and kinetin for adenine uptake by AtPUP1 determined previously (Gillissen et al., 2000). To investigate the substrate specificity of AtPUP1, competition experiments with other cytokinins like isopentenyladenine and kinetin, and other different structural and metabolism-related substances like 6-chloropurine, adenine, allantoin, and sucrose as a control were performed. The uptake of trans-zeatin was strongly inhibited by isopentenyladenine and kinetin (Figure 3b) and, as expected, also by adenine (data not shown). In contrast, sucrose had no effect on the uptake of trans-zeatin (data not shown), whereas allantoin and 6-chloropurine had only weak inhibitory effects. The results indicate that AtPUP1 can transport a variety of physiologically relevant cytokinins. Lack of allantoin transport activity is consistent with the finding that allantoin uptake is mediated by ureide permeases (UPS) heterocycle transporters (Desimone et al., 2002).

Figure 3.

AtPUP1-mediated uptake of trans-zeatin in yeast.

(a) Time dependence. The uptake of trans-zeatin (nmol mg−1 dry weight (DW)) was determined radioactively (trans-3H-zeatin, specific activity: 87 Bq µl−1) in the MG887-1 yeast strain transformed with AtPUP1 in pDR195 (□) or with empty vector pDR195 (○) at 100 µmtrans-zeatin and pH 5.0.

(b) Substrate specificity of AtPUP1. Uptake of trans-zeatin by AtPUP1 in pDR195 was determined by competition with a 10-fold excess (1 mm) of allantoin (▴), 6-chloropurine (○), kinetin (●), and isopentenyladenine (◊), respectively. Trans-zeatin uptake without competitor is shown by (□). Background uptake rates in MG887-1 transformed with the empty pDR195 vectors were subtracted. Results represent the mean ± SD of three independent experiments.

Transport of caffeine by AtPUP1

It has been suggested that caffeine has a cytokinin-like activity in different plant tissues (Vitória and Mazzafera, 1997). Caffeine inhibited adenine uptake by AtPUP1 (Gillissen et al., 2000). Caffeine has inhibitory and toxic effects on the growth of Saccharomyces cerevisiae (Bard et al., 1980; Hannan and Nasim, 1977). To test whether AtPUP1 is able to mediate the transport of caffeine, the growth inhibitory effect of caffeine on the growth of MG887-1 transformed with AtPUP1 was analyzed in the presence of ammonium as a nitrogen source. Yeast expressing AtPUP1 was significantly more sensitive to caffeine at concentrations of 0.1 and 0.2%, resulting in decreased colony size as compared to yeast transformed with the empty vector (Figure 4). The observed inhibition of growth is most probably caused by higher caffeine uptake, indicating the ability of AtPUP1 to recognize caffeine as a substrate, which is transported into the yeast cell.

Figure 4.

AtPUP1 confers hypersensitivity to caffeine.

MG887-1 yeast strains expressing AtPUP1 (right panel) and the vector control (left panel) were grown for 6 days on minimal medium containing caffeine at concentrations of 0, 0.1, and 0.2% (magnification in all panels 4×).

Transport of nucleosides by AtPUP1

As adenosine and ribosides of cytokinin are also transported in plants, it was investigated whether AtPUP1 transports adenosine. As Saccharomyces cerevisiae does not possess an efficient adenosine transport system, adenine auxotrophic strains cannot grow on adenosine as the sole purine source (Detke, 1998; Mäser et al., 1999). In contrast to controls, the DM734-238D yeast strain expressing AtPUP1 in pDR195 was able to grow on adenosine-containing media, proving that AtPUP1 can transport ribosides such as adenosine (Figure 5). This finding is in agreement with the weak inhibitory effect of adenosine on adenine uptake (Gillissen et al., 2000).

Figure 5.

Functional complementation by AtPUP1 of the adenine auxotrophic DM734–284D yeast strain deficient in adenosine uptake.

Yeast strains expressing AtPUP1 (left panel) or containing pDR195 vector (right panel) were plated on minimal medium supplemented with 150 µm adenosine as the sole purine source and grown for 3 days.

Isolation of AtPUP2 and 3 cDNAs

Database searches and phylogenetic analyses were used to classify 20 paralogs from the completed Arabidopsis genome (Figure S2). Within the PUP family, AtPUP2 and 3 are the closest relatives of AtPUP1. These two proteins show 64 and 58% identity to AtPUP1 at the amino acid level, respectively (Figure S3). AtPUP1 and AtPUP3 are located on chromosome I, whereas AtPUP2 maps to chromosome II. To study the function of these new proteins, full-length cDNAs were amplified by RT-PCR. AtPUP2 cDNA contains an open reading frame of 1041 bp encoding a protein of 347 amino acids with a calculated molecular mass of 38.1 kDa. The cDNA of AtPUP3 contains an open reading frame of 1053 bp encoding a protein with 351 amino acids with a calculated molecular mass of 38.9 kDa. The hydropathy patterns generated by THMM1.0 (Sonnhammer et al., 1998) suggest the presence of 10 transmembrane-spanning domains with N- and C-termini being located in the cytosol (Figure S4).

Functional analysis of AtPUP2 and 3 in yeast

To analyze the functionality of AtPUP2 and 3 in yeast, both cDNAs were cloned into the yeast expression vector pDR195. The capacity to transport adenine was tested by direct uptake measurements in the purine uptake-deficient yeast strain MG887-1 transformed with AtPUP2 or AtPUP3 in pDR195. AtPUP2-mediated linear uptake for 14C-adenine (Figure 6a). Uptake was low, but significantly higher when compared to controls transformed with the empty vector, demonstrating that AtPUP2 is a functional adenine transporter. However, transport rates for adenine mediated by AtPUP2 were significantly lower compared to AtPUP1, potentially because of less efficient targeting of AtPUP2 to the plasma membrane. Adenine uptake was pH dependent (Figure 6b) with highest uptake rates at pH 3.0. Using the Michaelis–Menten equation and non-linear regression analysis, a Km value of 22.6 ± 1.0 µm was determined at pH 3.5 (Figure 6c). AtPUP2 was not able to complement the growth deficiency of MG887-1 when tested on media containing adenine as the sole nitrogen source (data not shown). This is not surprising as the assay is not very sensitive. Adenine is a poor nitrogen source, therefore metabolization of adenine is rate limiting. The ability of AtPUP2 to recognize cytokinins as a substrate was tested by competition of 14C-adenine uptake in the presence of an excess of unlabeled cytokinins (Figure 6d). 14C-adenine uptake was strongly inhibited by isopentenyladenine, kinetin, benzylaminopurine, trans- and cis-zeatin, and adenine. Significant inhibition was also observed with trans-zeatin riboside.

Figure 6.

Kinetics of AtPUP2-mediated adenine uptake.

(a) Time course of adenine uptake. MG887-1/ura3 yeast transformed with AtPUP1 in pDR195 (□), AtPUP2 in pDR195 (●) or empty pDR195vector (○) were assayed for 14C-adenine uptake at 100 µm substrate concentration at pH 3.5. DW, dry weight.

(b) pH dependence of adenine uptake. Adenine uptake rates of MG887-1/ura3 yeast expressing AtPUP2 in pDR195 measured at different pH values, at 100 µm substrate concentration. Background uptake rates (empty vector) were subtracted.

(c) AtPUP2-mediated adenine uptake at different substrate concentrations. Experiments were performed at pH 3.5. Background uptake rates (empty vector) were subtracted.

(d) Substrate specificity of AtPUP2. Uptake of adenine by AtPUP2 in pDR195 was determined at 60 µm adenine and pH 3.5 in presence of 10-fold excess (600 µm) of unlabeled competitors. Values represent the mean ± SD of three independent experiments; BAP, benzylaminopurine; iPA, isopentenyladenine; ZR, trans-zeatin riboside.

In contrast, AtPUP3 did not show any transport activity in yeast. The MG887-1 yeast strain expressing AtPUP3 did not take up adenine and was not able to grow on adenine as the sole N-source. In addition, AtPUP2 and 3 did not complement the growth deficiency of the adenine auxotrophic mutant DM734-238D on adenosine as the sole purine source. Differences in activity have been observed in cases of other transporters and can often be attributed to inefficient targeting of the transporters to the plasma membrane of yeast. However, the lack of complementation does not exclude that PUP3 has a different substrate or function.

Expression pattern of three PUPs in Arabidopsis

Expression of the three PUP genes was examined histochemically using promoter–GUS fusions and RT-PCR with gene-specific primers. For histochemical analysis, two promoter fragments of AtPUP1 (0.8 and 1.9 kb), two promoter fragments of AtPUP2 (0.9 and 1.9 kb), and a 0.7 kb promoter fragment (region upstream of the translation start up to the next upstream ORF) of AtPUP3 were fused to the GUS reporter gene in the binary vector pCB308 and introduced into Arabidopsis plants. RT-PCR analysis was carried out on total RNA isolated from different tissues of mature Arabidopsis plants grown in soil.

Thirty-seven out of 40 AtPUP1-GUS T1 plants analyzed showed qualitatively the same expression pattern. Ten independent transformants per construct were analyzed histochemically for the localization of GUS activity in more detail. Although significantly stronger expression was observed in transgenic plants containing the longer promoter region, the overall expression pattern was identical for both constructs. In 3-week-old plants, GUS activity was detectable at the leaf margins in hydathodes (Figure 7a). Higher magnification of hydathodes from transgenic plants revealed GUS activity in epithem cells (Figure 7b–d) located between the xylem endings and the epidermis, cells proposed to be directly involved in the retrieval of solutes from the xylem sap (Wilson et al., 1991). No staining was detected in the epidermis surrounding hydathodes. In mature plants, expression of AtPUP1 was found at the stigma surface of siliques, potentially fulfilling a function similar to that in hydathodes (Figure 7e). In 2-week-old seedlings grown on MS medium, GUS activity was also observed in cotyledons and sometimes in patches of mature leaves (Figure 7f,g). Consistent with histochemical analysis, the RT-PCR experiments showed high-expression levels of AtPUP1 in hydathode-enriched fractions and intercostal fields. No signal was obtained in roots as shown by both reporter gene expression RT-PCR (Figure 8). In contrast to the GUS analysis, RT-PCR analysis, however, provided evidence that the AtPUP1 gene is also expressed in other tissues such as petioles, stems, major veins, and flowers (Figure 8). These discrepancies may be caused by differences in the status of the plants used for analysis (see patches in Figure 7g), the absence of additional promoter elements or differences in mRNA stability.

Figure 7.

Analysis of AtPUP1 expression by promoter–GUS fusions. Arabidopsis plants transformed with promoter–GUS fusion constructs were stained with 1 mm X-Gluc for 24 h, de-stained and documented. Transgenic Arabidopsis plants harboring the 1.9 kb AtPUP1 promoter fragment driving the β-glucuronidase gene.

(a) Whole plant, (b, c) close-up of hydathode with epithem and epidermis, (d) section of hydathode, (e) close-up of a developing silique, (f) cotyledon, (g) mature leaf showing patchy staining (arrows). Ep, epidermis; Et, epithem; H, hydathode; St, stigma surface; T, trichome; X, xylem.

Figure 8.

Analysis of AtPUP1–3 expression by RT-PCR. RNA from different tissues of mature Arabidopsis plants were converted to cDNA by reverse transcription. A 524 bp fragment of AtPUP1 and a 619 bp fragment of AtPUP2 were amplified by 30 PCR cycles. To obtain the 515 bp AtPUP3 fragment, 35 PCR cycles were carried out. A 78 bp ubiquitin fragment was amplified simultaneously by 25 PCR cycles as control. hef, hydathode-enriched fraction.

In AtPUP2-GUS plants, reporter gene activity was found in more than 14 independent T1 lines and the expression pattern was indistinguishable for both constructs containing promoters differing in length. In 3- and 5-week-old plants, GUS activity was detectable in the vascular system of source leaves (Figure 9a(i)). In older plants, strong staining was also present in the vascular tissue of cauline leaves (Figure 9a(ii)). No GUS activity was observed in stems and roots. Longitudinal- and cross-sections revealed that GUS staining is limited to the phloem (Figure 9a(iii–v)), consistent with a potential role of AtPUP2 in long-distance transport of adenine and cytokinins. The results obtained by RT-PCR are consistent with the expression pattern of the AtPUP2 promoters in the vascular tissue because AtPUP2 transcript was detectable in major vein fractions and petioles (Figure 8). The weak signal obtained in the hydathode-enriched fraction is probably derived from contamination with other leaf tissue (veins). Both methods did not detect expression in stems. In addition, RT-PCR showed AtPUP2 expression in flowers and roots, a pattern not observed in case of the GUS analysis, indicating potential differences in the status of the plants, the absence of additional promoter elements or differences in mRNA stability.

Figure 9.

Analysis of AtPUP2 and 3 expression by promoter–GUS fusions. Arabidopsis plants transformed with promoter–GUS fusion constructs were stained with 1 mm X-Gluc for 24 h, de-stained and documented.

(a) Transgenic Arabidopsis plants harboring the 1.9 kb AtPUP2 promoter fragment driving the β-glucuronidase gene. (i) Whole 3-week-old plant showing staining of the vascular system in source leaves. (ii) Staining of the vascular tissue in a cauline leaf from a mature plant. (iii) Longitudinal section through a source leaf petiole. (iv) Cross-section of the vascular system of a source leaf. (v) Larger magnification. P, phloem; X, xylem.

(b) Transgenic Arabidopsis plants harboring the 0.7 kb AtPUP3 promoter fragment driving the β-glucuronidase gene. (i) Flower showing staining in anthers, (ii) anther, (iii) magnification of pollen.

For AtPUP3-GUS plants, the activity of the reporter gene was detected in 12 out of 13 independent T1 lines. GUS staining was restricted to pollen (Figure 9b(i–iii)), indicating a potential role of AtPUP3 in transport of purine derivatives during pollen germination and tube elongation as found in Petunia pollen (Kamboj and Jackson, 1987). The same expression pattern has been obtained by RT-PCR (Figure 8), confirming that the short promoter fragment contains all necessary elements.


A common transport system for adenine and cytokinins

The presence of purines in phloem sap and cytokinins in phloem and xylem exudate strongly suggests the existence of carriers controlling long-distance transport of these substances (Beck and Wagner, 1994; Kluge and Ziegler, 1964; Weiler and Ziegler, 1981; Yong et al., 2000; Ziegler and Kluge, 1962). In our report, transport studies in cell cultures indicate that adenine and cytokinins use the same transport systems. At least one set of transporters is required to be capable of cellular import in planta. A second set of transporters must be postulated for cellular export (Figure 10). For example, cytokinins produced in the root have to enter the xylem vessels via cellular exporters, probably from xylem parenchyma cells, with functions comparable to that of the outwardly rectifying potassium channels (Gaymard et al., 1998). At the other end of the xylem, i.e. in leaves, cellular importers are required both for the uptake of cytokinins into leaf cells and for the cellular uptake and recycling of the cytokinin degradation product adenine. Furthermore, phloem loading with cytokinins, e.g. in leaves, similar to sucrose loading, requires cellular export systems and subsequently uptake carriers for the phloem cells.

Figure 10.

Hypothetical model of cytokinin signal perception and transport.

(a) Cellular model for cytokinin metabolism and transport. The extracellular level of cytokinins is controlled by the activity of cytokinin oxidase (Ckx) and cytokinin transporter (PUP). Cytokinins can either bind to the plasma membrane receptor (CRE1) eliciting a hormone-specific signaling cascade or can be taken into the cell by a transporter, e.g. PUP1. Adenine derived from cleavage of cytokinin in the apoplasm can be retrieved into the cell by PUP1 as well. Alternatively, cytokinins that have been taken up by PUP1 may bind to an intracellular receptor, controling the same or a different signaling cascade.

(b) Whole plant model for cytokinin transport. Both cellular importers and exporters are required in different cell types of the vascular system and the parenchymatous tissues surrounding them for long-distance transport of cytokinins in xylem and phloem. AtPUP1 localized in the epithem may serve as a retrieval system to preserve purines and purine derivatives including cytokinins from the guttation sap. AtPUP2 localized in the phloem of leaves, functions as a cellular import system, and thus may be involved in loading of the phloem with purines and cytokinins. SE, sieve elements; P, parenchyma, PP, phloem parenchyma; X, xylem; XP, xylem parenchyma.

The competition of adenine uptake by cytokinins in both cell cultures and yeast expressing Arabidopsis PUP purine permeases, demonstrated here, may indicate that cytokinins are substrates for transport. However, competition is not direct proof of the actual transport of a compound. Thus, direct uptake studies using radiolabeled cytokinins were carried out with yeast expressing AtPUP1. AtPUP1 mediates the transport of both cytokinins and adenine when expressed in yeast. AtPUP1 is assumed to function as a cellular uptake system at the plasma membrane by means of proton co-transport, because uptake is energy dependent, occurs against a concentration gradient, is stimulated by acidification, and is inhibited by protonophores (Gillissen et al., 2000). When determined in yeast, the affinity of AtPUP1 for adenine was similar to the Km for trans-zeatin in the low micromolar range. At first sight, the affinity range may seem relatively high, considering that trans-zeatin serves as a phytohormone. However, the apparent affinities of O-glycosylating enzymes for trans-zeatin range from 2 to 28 µm (Dixon et al., 1989; Turner et al., 1987), the Km of cytokinin oxidase measured in plant extracts was found between 0.1 and 31 µm, and that of the cloned enzyme was 14 µm (Bilyeu et al., 2001; Galuszka et al., 1999). Thus, the affinities of enzymes involved in cytokinin metabolism and AtPUP1 are in a comparable range, further supporting the hypothesis that also in planta, PUPs can function in cytokinin transport. As AtPUP1 is also able to mediate adenosine transport, it may contribute to cellular uptake of adenosine, although with very low efficiency. However, in analogy to the PUP's broad selectivity, high-affinity nucleoside transporters such as ENT1,At may be candidates for such a function (Möhlmann et al., 2001). Although the activity of AtPUP2 was relatively low, the overall properties regarding affinity and substrate specificity were similar to those of AtPUP1. Various cytokinin analogs including cis- and trans-zeatin, isopentenyladenine, and kinetin strongly competed for adenine uptake mediated by AtPUP1. In contrast, no activity was detected for the closely related AtPUP3, potentially resulting in ineffective plasma membrane targeting in the heterologous expression system. Thus, in principle, it is possible that the three proteins fulfil similar functions in the plant.

Different roles of the PUP transporters

As a first hint for the actual physiological function, a promoter analysis was used to determine the cellular expression pattern of the three transporters. Expression of the PUP1 promoter was confined to the epithem cells of hydathodes and to the stigma surface of siliques. A similar expression pattern had been observed for the inward K+ channel AKT1 (Lagarde et al., 1996). Hydathodes mediate guttation and may be involved in the retrieval solutes from the transpiration stream. With respect to phylogenesis, hydathodes are related to nectaries (Vogel, 1998). A typical hydathode consists of a particular parenchymatous tissue, the epithem, and an epidermis with modified stomata serving as water pores. The morphological features of epithem cells are large numbers of mitochondria and a lobed surface resulting in an increase of surface area facing the extracellular space, suggesting involvement not only in the active secretion of solutes, but also in the selective absorption and retrieval of both inorganic and organic solutes (Galatis, 1988; Höhn, 1951). In agreement with the findings of Wilson et al. (1991), the epithem is able to take up 14C-labeled aspartate fed to the leaves via the transpiration system. Similarly, nectaries, although thought to serve mainly as secretory organs, are also able to take up radiolabeled amino acids (Ziegler and Lüttge, 1959). Therefore, the expression of AtPUP1 in the epithem of hydathodes may suggest a role in retrieval of purines and cytokinins from xylem sap. Epithem cells accumulate sulforhodamine G, indicating the presence of highly active H+-ATPases and respective H+-symporters for retrieval of ions like potassium and other important organic compounds (Wilson et al., 1991). Consistent with a potential role in the retrieval, circumstantial evidence suggests that AtPUP1 functions as a H+-co-transporter, fulfilling the requirements of a retrieval system. As AtPUP1 is able to transport another adenine analog caffeine, which is translocated in the xylem of coffee plants, a PUP1 ortholog may serve a similar function in retrieval (Mazzafera and Gonçalves, 1999). Interestingly, caffeine has also been implicated in cytokinin-like functions (Vitória and Mazzafera, 1997). PUP1 expression was also detected in cotyledons, consistent with the finding that at least in Ricinus communis the cotyledons take up adenine rapidly (Kombrink and Beevers, 1983). The presence of mRNA as detected by RT-PCR in other tissues and the presence of patchy GUS staining in other tissues may indicate that the gene can be induced under certain conditions or that potentially additional elements in the gene contribute to regulation.

AtPUP2 promoter expression was found in the vascular system of leaves, i.e. in the phloem of leaves, pointing to a potential role in phloem loading of adenine and cytokinins. Also in this case, RT-PCR analysis indicates that the gene may be expressed in additional tissues. The expression pattern of AtPUP3 was limited to pollen, in which it might play a role in the transport of purine derivatives during pollen germination and tube elongation (Kamboj and Jackson, 1987). It is obvious from the above discussion that multiple transporters are required for long-distance transport (Figure 10b). Thus, it is not surprising that the PUPs constitute a large, however extremely diverse, gene family, in which the most distant members share less than 20% similarity at the amino acid level. Further studies are required to obtain an overview of the transport properties, the expression pattern, and regulation of various other members of the PUP family.

Interestingly, a member of the PUP family from tomato was found as an expressed sequence tag in Agrobacterium-induced tumors (GenBank accession number BG131703). Similar to the situation in animals, plant tumors are also highly vascularized (Aloni et al., 1995). Agrobacteria transform plant cells to tumor cells by overproducing auxins and cytokinins. The balance of cytokinins and auxins is not only critical for callus or tumor formation but also for the development of the vascular system. Despite local cytokinin overproduction, infected plants do not show systemic symptoms of cytokinin action. Further studies are required to determine whether PUPs expressed in tumors have a function in channeling cytokinins, in retrieving them or in protecting the plant from systemic effects of the infection. Similar mechanisms may also operate in the restriction of cytokinin movement in fungal infection of leaves and may explain why local overproduction of cytokinins does not lead to systemic loss of apical dominance (Faiss et al., 1997).

In summary, transport studies in Arabidopsis cell cultures indicate that adenine and cytokinins are transported by a common system, which has properties similar to heterologously expressed PUP transporters. Transport studies in yeast in combination with promoter reporter gene studies indicate that different members of the PUP family are involved in the long-distance transport of adenine, cytokinins, and alkaloids. Whereas AtPUP2 may play a role in phloem transport, AtPUP1 may be involved in retrieval of nucleobases and derivatives to prevent secretion in hydathodes and at the stigma surface of siliques and AtPUP3 in supply of pollen. These studies provide a basis for a more direct analysis of the physiological function of PUPs by studying the effect of specific inhibition of transport by PTGS by overexpression or by using insertional mutants.

Experimental procedures

Yeast strains and growth conditions

Yeast strains used were MG887-1 (Mat a, fcy2, ura3) (Gillissen et al., 2000) and DM734–284D (Mat a, ade8–18, ade2–1, arg4–16, leu2–27, trp1–1, lys2, ura3, gal7) (DM Yeast Genetic Stock Center, Berkeley, USA). Yeast cells were cultured in YPD or minimal media. The minimal medium was supplemented with 7.4 mm adenine as the sole nitrogen source for growth analysis of MG887-1. For growth experiments with DM734–284D, 5 g l−1 ammonium sulfate and 150 µm of the purine source were added to the minimal medium. To study caffeine toxicity in MG887-1, the minimal medium was supplemented with caffeine (0.1 and 0.2%) and 5 g l−1 ammonium sulfate.

DNA work

To generate pDR195/AtPUP1, NotI-restricted full-length cDNA encoding AtPUP1 was isolated from the pFL61/AtPUP1 (Gillissen et al., 2000) and ligated into pDR195 (Rentsch et al., 1995). Orientation of inserts was tested by restriction analysis.

The cDNA of AtPUP2 was obtained by RT-PCR (RETROscript Kit, Ambion, Austin, USA) with RNA from Arabidopsis thaliana roots (ecotype Col-0) as a template. For the first-strand synthesis, random decamer primers were used. The following PCR was performed with gene-specific AtPUP2 primers (5′-GAGAGGATCCGAAGATGAAGATGAAGACAG-3′ and 5′-GAGAGGATCCGCTTTTAAGCTA-CATAATCAG-3′). The amplified AtPUP2 cDNA was ligated with pCRII-TOPO (Invitrogen, Groningen, Netherlands) and subcloned into the yeast expression vector pDR195, previously cut with BamHI and NotI (Rentsch et al., 1995). The cDNA encoding AtPUP3 was amplified by RT-PCR from RNA of A. thaliana (ecotype Col-0) flowers. Oligo dT primers were used for the first-strand synthesis. The following PCR was carried out using primers: 5′-TATCTTGGATCCAGACAAGAATGGTGAAGGCTCTTG-3′ and 5′-ATATGCGGATCCGAGTTTTAACACTCACTTAATGGGTC-3′. The obtained PCR product was digested with BamHI and ligated into BamHI-digested pDR195 (Rentsch et al., 1995). The cDNAs encoding AtPUP2 and 3 were sequenced, and no differences to the genomic sequence were found except the presence of an intron. The pDR195 plasmids containing the AtPUP1-, AtPUP2- or AtPUP3-coding sequence were introduced into MG887-1 and DM734–284D yeast strains following a modified method previously described (Dohmen et al., 1991). Colonies able to grow on plates lacking uracil were used for further growth analysis and uptake experiments.

The promoter regions of AtPUP1–3 were isolated by PCR with genomic DNA from A. thaliana ecotype Col-0 as a template. The 0.8 kb promoter fragment of AtPUP1 was obtained using Pfu polymerase (Stratagene, La Jolla, USA), the F1a (5′-GAGATCTAGACCTAGCGGGGTAGAAACCTTG-3′) and R1 (5′-GAGAGGATCCTTCTTCTTGCTGCTTGCTG-3′) primers. The 1.9 kb promoter fragment of AtPUP1 was isolated using Taq polymerase (Roche Molecular Biochemicals, Germany), F1b (5′-GAGATCTAGAGTCATCAAAGATTTCCTAAAC-3′) and R primers. The obtained DNA promoter fragments were subcloned into pCRII-TOPO, afterwards excised with BamHI and XbaI and fused to the GUS gene located on the binary vector pCB308 (Xiang et al., 1999). The 0.9 kb promoter fragment of AtPUP2 was amplified using the Pfu polymerase, the F2a (5′-GAGAACTAGTGGAAACTCATAACTCCGATG-3′) and R2 (5′-GAGAGGATCCTTCTTCTTCTTGCTATAACC-3′) primers. The 1.9 kb fragment of AtPUP2 was isolated using the Taq polymerase, the F2a primer (5′-GAGAACTAGTGATGGAACGTCTACGAACAC-3′) and R2 primers. The 0.7 kb promoter fragment of AtPUP3 was amplified using the Pfu polymerase, the F3 (5′-GAGAACTAGTGTTCTTGTGTATAAGTAATG-3′) and R3 (5′-GAGAGGATCCTTGTCTGATTTCTTGTGG-3′) primers. The PCR products were cut with SpeI and BamHI and ligated into pCB308. Sequence analysis of the inserts revealed five point mutations in the longer promoter fragment of AtPUP1 that did not affect the qualitative expression pattern. Transformation of A. tumefaciens PGV2260 using the resulting pCB308 derivatives was performed as described by Deblaire et al. (1985).

Plant transformation and histochemical analysis of GUS activity

Arabidopsis thaliana L. Heynh. ecotype Col-0 plants were transformed via vacuum infiltration (Clough and Bent, 1998). Transgenic Arabidopsis plants were selected with BASTA herbicide (Aventis Crop Science, Frankfurt, Germany). Histochemical assays for β-glucuronidase activity in transgenic plants were performed as described in Martin et al. (1992). Tissues were cut into 2 mm × 5 mm pieces, incubated in GUS staining solution containing 100 mm sodium phosphate (pH 7.0), 10 mm EDTA, 3 mm K4[Fe(CN)6], 0.5 mm K3[Fe(CN)6], 0.1% (v/v) Triton X-100, and 2 mm 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid (X-gluc) for 3–24 h at 37°C. Slight vacuum was applied before incubation to facilitate substrate infiltration. For resin sections, X-gluc stained tissues were fixed in 4% glutaraldehyde, 50 mm sodium phosphate (pH 7.0) overnight at 4°C. Fixed tissues were dehydrated in EtOH and embedded in LR White resin (London Resin Company Ltd, Berkshire, UK). Embedded material was cut into 1.5–5 µm sections with glass knives using an ultramicrotome and observed by bright field, phase contrast, and dark field microscopy. For the fresh sections, tissues were embedded in 5% low melting agarose in 50 mm sodium phosphate buffer pH 7.0. After solidification, agar blocks were cut, and fresh sections (75–150 µm) were made with razor blades using a vibratome. Sections were collected in ice-cold water and incubated in GUS staining solution described above, but without 0.5 mm K3[Fe(CN)6]. For chlorophyll-containing tissues, fresh sections were cleared in 70% EtOH.

RT-PCR analysis

RNA for RT-PCR analysis was extracted from suspension cells and mature soil-grown Arabidodpsis plants. Major veins, the regions around leaf edges containing hydathodes (hydathode-enriched fraction), and intercostal fields were excised from leaves, and immediately frozen in liquid nitrogen. Aliquots of 2 µg RNA were used as a template for the first strand synthesis, using RETROscript Kit (Ambion, Austin, USA) according to the manufacturer's instructions. An aliquot of 2 µl of the first strand cDNA was used for PCR with gene-specific primers. To avoid amplification of genomic DNA, reverse primers were spanned splice sites. The 524 bp AtPUP1 fragment was amplified by 30 PCR cycles using primers: 5′-CTAACAACGCGGAAAACAAGC-3′ and 5′-CTCTTGCTATCACCTTAAAATCTC-3′. The 619 bp AtPUP2 transcript was obtained by 30 PCR cycles with specific primers: 5′-TATCTTGGTACCAAAGGATCTGGTTTCCAAGC-3′ and 5′-TCCTGCTATCACCTTGAAATCG-3′. The 515 bp AtPUP3 fragment was amplified by 35 PCR cycles with primers: 5′-ACAATGTGGGTGATAGTACAAG-3′ and 5′-CTTTGGTAAGGCCTTGAAAATC-3′. To ensure that equal amounts of cDNA were added to each PCR reaction, a cDNA fragment of the constitutively expressed ubiquitin gene was amplified simultaneously by 25 PCR cycles using primers: 5′-GAATCCACCCTCCACTTGGTC-3′ and 5′-CGTCTTTCCCGTTAGGGTTTT.

Transport measurements into Arabidopsis suspension cells

The suspension cell culture from Arabidopsis thaliana ecotype Landsberg erecta (May and Leaver, 1993) was a gift from Mike Bevan (The Jones Innes Centre, Norwich, UK). Cells were maintained as described by Fuerst et al. (1996). For the uptake experiments, cells were harvested 4 days after dilution into new medium by centrifugation at 18 g for 3 min, and were re-suspended in fresh cytokinin-free medium containing 10 mm potassium phosphate buffer (pH 5.7) at a concentration of 0.25 ml packed cells per 1 ml of suspension. To start the reaction, 850 µl of this suspension was mixed with 50 µl medium containing 14C-adenine (final concentration 13 Bq µl−1) and the unlabeled analog (final concentration 20 µm). Samples of 170 µl were removed after 1–4 min, filtered on glass-fiber filters, and washed twice with 10 ml medium. Radioactivity on the filters was determined by liquid scintillation spectrometry (Wallac, Turku, Finland). The protonophore CCCP was added together with the radiolabeled mix. Competition experiments were performed with a 10-fold excess of the respective competitor resulting in a final concentration of 200 µm.

Transport measurements into yeast cells

For adenine uptake studies, yeast cells were harvested at OD600 of 0.6, washed in water and re-suspended in 100 mm sodium citrate buffer (pH 3.5) containing 1% glucose to a final OD600 of 12. An aliquot of 100 µl of the cell suspension was pre-incubated for 2 min at 30°C. To start the reaction, a 100 µl buffer containing 13 Bq µl−1 14C-labeled adenine (Amersham), 1% glucose, and the unlabeled analog, as indicated, was added. Samples of 50 µl were removed after 30, 60, 120 and 180 sec, transferred to 4 ml ice-cold water, filtered on glass-fiber filters, and then washed with 8 ml water before radioactivity was determined. For analysis of the pH dependence, cells were washed in water, and then re-suspended in 100 mm sodium phosphate buffer at different pH. For trans-zeatin uptake assays, yeast cells were harvested at OD600 of 0.6, washed, and re-suspended in 100 mm sodium phosphate buffer (pH 5.0) to a final OD600 of 10. An aliquot of 500 µl of the cell suspension was pre-incubated for 3 min at 30°C, supplemented with glucose (final concentration 10 mm) and incubated for a further 2 min. To start the reaction, a 39.5 µl buffer containing labeled trans-2-3H-zeatin (final concentration 87.3 Bq µl−1; OlchemIm, Olomouc, Czech Republic) and unlabeled analogs was added (final concentration 100 µm). Samples of 100 µl were removed after 20, 40, 60, and 90 sec, transferred to 4 ml ice-cold 5 mm adenine solution, filtered on glass-fiber filters, and washed with 8 ml adenine (5 mm). Transport measurements were repeated independently and represent the mean of at least three experiments. For analysis of purity, 100 µl of a radioactive sample (1 : 200 dilution) was loaded (using an isocratic gradient: methanol (0.2 ml min−1) and 50 mm acetic acid/ammonia (0.6 ml min−1)) on a HPLC column (LUNA 5 µm C18(2), Phenomenex, Germany). The chromatograms were recorded by UV detection (diode array, Kontron, Germany) or radioactivity (LB507B, Berthold, Germany). Retention times of trans-zeatin and adenine were 13.91 and 4.63 min, respectively.


We are grateful to Bettina Stadelhofer (Analytics Facility ZMBP) and Bettina Million for their excellent technical assistance and we would like to thank Thomas Schmülling (Free University Berlin) and Felicity de Courcy for their critical reading of the manuscript. We gratefully acknowledge the support of the Deutsche Forschungsgemeinschaft (SPP ‘CO2 and transport’ and SFB 446).

Supplementary Material

The following material is available from

Figure S1. Structure and radiochemical purity of trans-3H-zeatin.
(a) Structural formula of trans-3H-zeatin.
(b) HPLC profile of trans-3H-zeatin using a radioactivity detector. The retention times of adenine (4.63 min) and trans-zeatin (13.91 min) are indicated.

Figure S2. Phylogenetic tree of Arabidopsis PUP family. Maximum parsimony analysis was performed using paup version 4.0b8a (Swofford, 1998) with all DNA characters unweighted and gaps scored as missing characters. Heuristic tree searches were executed using 100 random sequence additions and the tree bisection re-connection branch swapping algorithm with random sequence analysis. The complete alignment was based on 364 sites. A total of 313 sites were phylogenetically informative. The AtPUP paralogs were re-named and grouped into four clades. The alignment underlying the phylogenetic tree is available at

Figure S3. Alignment of AtPUP1–3. The deduced amino acid sequences were aligned by using the megalign program (DNASTAR, Madison, WI). Identical amino acids are shaded.

Figure S4. Prediction of 10 putative membrane spanning regions in the amino acid sequence of AtPUP1–3. Hydropathy plots were performed by using THMM1.0 (Sonnhammer et al., 1998). N- and C-termini were predicted to be cytosolic.