Functional expression and characterization of Arabidopsis ABCB, AUX 1 and PIN auxin transporters in Schizosaccharomyces pombe

Summary

Heterologous expression systems based on tobacco BY-2 cells, Arabidopsis cell cultures, Xenopus oocytes, Saccharomyces cerevisiae, and human HeLa cells have been used to express and characterize PIN, ABCB (PGP), and AUX/LAX auxin transporters from Arabidopsis. However, no single system has been identified that can be used for effective comparative analyses of these proteins. We have developed an accessible Schizosaccharomyces pombe system for comparative studies of plant transport proteins. The system includes knockout mutants in all ABC and putative auxin transport genes and Gateway^®-compatible expression vectors for functional analysis and subcellular localization of recombinant proteins. We expressed Arabidopsis ABCB1 and ABCB19 in mam1pdr1 host lines under the inducible nmt41 promoter. ABCB19 showed a higher ³H-IAA export activity than ABCB1. Arabidopsis PIN proteins were expressed in a mutant lacking the auxin effluxer like 1 (AEL1) gene. PIN1 showed higher activity than PIN2 with similar protein expression levels. Expression of AUX1 in a permease-deficient vat3 mutant resulted in increased net auxin uptake activity. Finally, ABCB4 expressed in mam1pdr1 displayed a concentration-dependent reversal of ³H-IAA transport that is consistent with its observed activity in planta. Structural modelling suggests that ABCB4 has three substrate interaction sites rather than the two found in ABCB19, thus providing a rationale for the observed substrate activation. Taken together, these results suggest that the S. pombe system described here can be employed for comparative analyses and subsequent structural characterizations of plant transport proteins.

Introduction

Polar transport of the phytohormone auxin underlies the generation of auxin gradients and localized auxin concentrations which control multiple plant developmental processes (Benjamins and Scheres, 2008). Directional auxin flow is motivated by chemiosmotic and pH gradients, a proton symport uptake system, and both passive and ATP-dependent efflux systems (Rubery and Sheldrake, 1974; Raven, 1975;Li et al., 2005; Petrasek et al., 2006; Yang et al., 2006). The principal auxin indole-3-acetic acid (IAA) enters cells via lipophilic diffusion and anionic uptake mediated by the presumptive proton symporters of the AUXIN RESISTANT (AUX/LAX; TC 2.A.18) family (Bennett et al., 1996; Swarup et al., 2005; Yang et al., 2006). A reversible active transport mechanism may also contribute to uptake in some cells (Santelia et al., 2005; Terasaka et al., 2005). Auxin exits cells via the PIN-FORMED auxin efflux carrier proteins (PINs; TC 2.A.69) and some members of the ABCB transporter family (TC 3.A.1) (Noh et al., 2001; Geisler et al., 2005; Bouchard et al., 2006; Blakeslee et al., 2007; Cho et al., 2007; Lewis et al., 2007; Wu et al., 2007).

Molecular genetic approaches have been extensively applied to characterise these complementary auxin transport systems in planta, but the difficulty of expressing the proteins in heterologous systems has precluded their extensive biochemical characterisation. AUX1 has been shown to mediate high-affinity auxin uptake in in Xenopus oocytes and human HeLa cells (Yang et al., 2006; Blakeslee et al., 2007), but these systems are not accessible to a wide range of plant researchers. The lack of cell walls, the necessity of viral transfection, and difficulty in performing kinetic assays in immobilised cells also limit the utility of the HeLa system for this purpose. Recombinant PIN proteins have been variously expressed in plant cells, yeasts, and HeLa cells, but no one system has been appropriate for analysis of all PIN proteins in an environment independent of other plant factors (Chen et al., 1998; Luschnig et al., 1998; Petrasek et al., 2006; Blakeslee et al., 2007; Titapiwatanakun et al., 2008). Similarly, Arabidopsis ABCB1, ABCB19 and ABCB4 confer auxin export in various heterologous systems, but ABCB19 cannot be successfully expressed in Saccharomyces cerevisiae and ABCB4 exhibits 1-naphthylphthalamic acid (NPA)-reversible auxin uptake activity not observed in plant cell expression systems (Noh et al., 2001; Geisler et al., 2005; Terasaka et al., 2005; Cho et al., 2007; Titapiwatanakun et al., 2008). As a common and accessible heterologous assay system for all auxin transporters has not been developed, the relative biochemical transport activities of these transporters and their interactions with regulatory proteins has been difficult to assess. Plant cell expression systems could provide an ideal environment for these studies, however, their limitation is that direct transport may not be distinguishable from indirect facilitation of endogenous auxin transport activity (Titapiwatanakun et al., 2008). This is especially the case when a clear mechanistic rationale for the transport activity is unknown, as is the case with PIN proteins. Further, to date, expression and characterisation of AUX/LAX uptake transporters has not been successful in plant cell systems.

Two specific issues limit the utility of S. cerevisiae and mammalian cell cultures for characterisation of recombinant plant proteins: membrane composition and glycosylation. The activity of ABCB19 and PIN1 is highly dependent on sterol-enriched microdomains (Willemsen et al., 2003; Titapiwatanakun et al., 2008). Normal functionality of other transporters including plasma membrane ATPases may also be regulated by the presence of membrane microdomains (Mongrand et al., 2004; Borner et al., 2005; Titapiwatanakun et al., 2008). Such microdomains are found in animal cells, but contain cholesterol and sphingosine, whereas plant and yeast membranes contain other sterols, dihydrosphingosine and hydrosphingosine (Worrall et al., 2003). As such, yeasts generally provide a better environment for heterologous expression of plant transporters, as, like plants, yeasts have cell walls and, their primary membrane sterol (ergosterol) is more similar to the major plant sterol (β-sitosterol).

Saccharomyces cerevisiae has sterol-enriched membrane domains that function in regulating the stability of transport proteins (Grossmann et al., 2007, 2008). However, in S. cerevisiae, sterols are relatively uniformly distributed in small plasma membrane patches rather than polarised domains during the vegetative life cycle (Wachtler and Balasubramanian, 2006; Grossmann et al., 2007). Schizosaccharomyces pombe has polarised, sterol-enriched plasma membrane domains that persist throughout the vegetative life cycle (Wachtler et al., 2003; Grossmann et al., 2007). This difference appears to be a factor in the recent success in functional expression of Arabidopsis PIN1 in S. pombe (Titapiwatanakun et al., 2008).

The other primary factor limiting the use of S. cerevisiae as a common heterologous expression system for plant transporters is the S. cerevisiae protein glycosylation mechanism, which can produce hyperglycosylated, mislocalised plant membrane proteins (Noh et al., 2001; Geisler et al., 2005). Removal of glycosylation enzymes does not appear to be sufficient to allow for successful expression of glycoprotein transporters in S. cerevisiae, as the use of mutants deficient in protein glycosylation has not resulted in expression of active ABCB19 or PIN1 in our hands. However, in S. pombe, the addition of a galactose subunit onto N-glycans prevents the excessive addition of mannose subunits to N-glycans as is seen in S. cerevisiae (Gemmill and Trimble, 1999).

These considerations led us to adapt an S. pombe system for studies of plant transport proteins. The system is attractive because available growth, transformation, and gene knockout protocols are adaptable to multiple environments. Unlike mammalian and insect cell systems, little or no specialized equipment is required for implementation. Unlike adherent cell culture systems, the S. pombe system is amenable to kinetic and growth inhibition studies. Finally, yields of functional recombinant proteins from S. pombe are generally higher than from S. cerevisiae, thus providing an avenue to subsequent structural determinations.

Here we show that all of the major known auxin transporters from Arabidopsis can be successfully expressed and characterised in S. pombe. The results presented here are a comparative analysis of PIN, ABCB, and AUX1 auxin transporters in a single heterologous system where othologous genes can be interrupted selectively. Knockout lines were developed for all S. pombe ABC transporter genes, a PIN-like auxin efflux like (AEL1) transporter, and the AUX1-like amino acid permease (AVT3). Expression vectors were evaluated and then modified to simplify cloning and localisation of recombinant plant proteins. The utility of the system was demonstrated by a detailed analysis of ABCB4 transport activity that had not been possible in other systems (Terasaka et al., 2005; Cho et al., 2007). These results also demonstrate that differences in membrane composition and glycosylation in S. pombe allow for successful expression of the PIN1 and ABCB19 auxin exporters which cannot be functionally expressed in S. cerevisiae.

Results and discussions

Creation of Schizosaccharomyces pombe ABC knockout lines for ³H-IAA transport assays

Schizosaccharomyces pombe has only 11 ABC transporters compared to 32 in S. cerevisiae, and all 11 ABC transporter genes have been disrupted and the proteins subcellularly localised (Iwaki et al., 2006). Two of these proteins, BFR1 and PMD1, confer drug resistance and are localised to the plasma membrane (PM) (Iwaki et al., 2006). Two other ABC transporters, MAM1 and PDR1, are primarily localised to endosomes and the ER, but also exhibit some PM localisation (Iwaki et al., 2006). We constructed ABC mutant lines to be used for auxin transport assays of plant ABCB transporters by analysing both their sensitivity to auxins and their ³H-IAA transport activity. Mutants that exhibited higher sensitivity to auxins or enhanced ³H-IAA retention were chosen for use in auxin transport assays of auxin efflux transporters, and mutants showing lower sensitivity and/or reduced ³H-IAA retention were selected for assays of import activity. Filipin staining showed that the polarized sterol-enriched domains found in wild type cells are also present in the ABC transporter knockout lines (Figure 1a).

Figure 1.  Construction of S. pombe and Gateway compatible vectors for heterologous expression of plant ABCB transporters. (a) Polar localisation of of sterols detected by filipin staining in mam1pdr1 cells. Bar = 10 μm. (b) ³H-IAA transport assays in all 11 ABC transporter mutants of S. pombe. (c) Double mutant mam1pdr1 showed higher ³H-IAA accumulation. Error bars, SD(n = 4). Asterisks indicate P < 0.05. (d) Gateway compatible vectors were constructed for rapid insertion of genes of interest, and epitope tags (3XHA and YFP) for subsequent subcellular localization of expressed proteins.

Across a wide range of concentrations, S. pombe ABC mutants were relatively insensitive to IAA and the artificial auxin 1-naphthylene acetic acid (NAA), although mam1 and abc3 exhibited slightly more sensitivity than wild type. abc4 and pmd1 were slightly more IAA tolerant, bfr1 and mam1 were slightly more NAA sensitive, and abc1 and abc4 were slightly more NAA resistant (data not shown). The accumulation of ³H-IAA in the mitochondrial and vacuolar ABCB transporter mutants abc2, abc3, bfr1 and mdl1 (Iwaki et al., 2006) was significantly lower than in the wild type, apparently a result of decreased subcellular ³H-IAA compartmentalization (P < 0.001, Figure 1b). As mam1 (P = 0.015) and pdr1 (P < 0.001) showed higher ³H-IAA accumulations, a mam1pdr1 double mutant was constructed as a host for assays of ABCB transporters and it accumulated higher ³H-IAA levels as expected (P < 0.002, Figure 1c) without any obvious alteration in growth. However, double, triple, and quadruple interruptions of genes encoding vacuolar ABC1, ABC2, ABC3 and ABC4 proteins resulted in altered growth that limited use of these strains as hosts for transport studies.

Construction of expression vectors

Analysis of plasmids optimized for single copy cDNA integration in S. pombe indicated that expression of Arabidopsis auxin transport protein genes was more optimal with the pREP41 plasmid (Basi et al., 1993) than the convenient pDUAL Gateway^®–compatible vectors (Matsuyama et al., 2004, 2006). The nmt1-41 inducible promoter in the pREP41 vector provides for moderate expression of recombinant membrane proteins at levels sufficient to characterise their transport activity without inducing overexpression problems such as protein aggregation and was previously used to successfully express Arabidopsis PIN1 (Titapiwatanakun et al., 2008). A Gateway^®-compatible pREP41GW expression vector was created by inserting a Gateway^® cloning cassette into pREP41 between the nmt1 promoter and nmt1 terminator (Figure 1d). A pREP41HAGW vector for N-terminal fusion of a 3X hemagglutinin epitope (3XHA) was then constructed for assessment of protein expression using Western blotting or immunocytochemistry, and pREP41NYFPGW was created to provide an N-terminal fusion of yellow fluorescent protein (YFP) for direct detection of protein expression or localization using fluorescence microscopy.

³H-IAA transport assays of recombinant PIN proteins in Schizosaccharomyces pombe

Previously, Arabidopsis PIN1 was shown to catalyse auxin efflux when expressed in an S. pombe mutant lacking the AEL1 auxin effluxer-like gene (Titapiwatanakun et al., 2008). Arabidopsis PIN2 and PIN7 had already been functionally expressed in other non-plant heterologous systems (Chen et al., 1998; Luschnig et al., 1998; Petrasek et al., 2006; Blakeslee et al., 2007), but comparisons of efflux activity of the three proteins were difficult to make across expression systems. We expressed PIN1, PIN2 and PIN7 in ael1 at similar levels of abundance (by Western blotting) and found that all of the recombinant PINs exhibited transport activity. Recombinant PIN1 and PIN7 (approximately 25% reduction in ³H-IAA accumulation) expressed in ael1 exhibited higher transport activity than PIN2 (approximately 17% reduction in ³H-IAA accumulation), but all were greater than empty vector controls (P < 0.03, Figure 2a,c,f). However, when PIN1 and PIN2 were expressed in the mam1pdr1 ABC transporter deficient background under the same induction conditions, PIN2 (approximately 25% reduction in ³H-IAA accumulation) exhibited a slight greater transport activity than PIN1 (approximately 20% reduction in ³H-IAA accumulation) (P < 0.02, Figure 2b and Figure S1). This result is consistent with previous reports that auxin transport increases when PIN1 is co-expressed with some plasma membrane ABC transporters (Blakeslee et al., 2007; Titapiwatanakun et al., 2008). In planta, ABCB19 appears to enhance PIN1 transport activity by recruiting PIN1 to sterol-enriched membrane domains (Blilou et al., 2005; Titapiwatanakun et al., 2008), and the S. pombe ABC transporters MAM1 and PDR1 may play a similar role with recombinant PIN proteins, as PIN1 catalysed less auxin export than PIN2 in the mam1pdr1 background (Figure S1).

Figure 2.  Comparative auxin transport assays of PIN proteins in S. pombe ael1. (a) ³H-IAA transport assay of PIN1. (b) ³H-IAA transport assay of PIN2. (c) ³H-IAA transport assay of PIN7. Reduced ³H-IAA accumulation indicates export activity of PINs. (d) Inhibitory effect of IAA, 2,4-D and NPA on ³H-IAA export of PINs. (e) ³H-benzoic acid accumulation in S. pombe expressing PINs. (f) Western blots indicate comparable PIN1 and PIN2 expression in S. pombe. Error bars, SD (n = 8). Asterisks indicate P < 0.05.

The IAA export activity of cells expressing PIN1, PIN2 and PIN7 was inhibited by 5 μm 1-naphthylphthalamic acid (NPA) compared to solvent (1% final concentration of ethanol) controls (P > 0.28, Figure 2d), which is consistent with results observed in other heterologous systems (Petrasek et al., 2006; Blakeslee et al., 2007; Titapiwatanakun et al., 2008). A 10× concentration of unlabelled IAA (500 nm) or 2, 4-dichlorophenoxyacetic acid (2,4-D) competitively eliminated net ³H-IAA export mediated by PIN2 and PIN7 (P > 0.05, Figure 2d). The ³H-IAA export activity of PIN1 was competitively inhibited by 10× IAA (P = 0.04), but was only slightly inhibited by 2, 4-D (P = 0.02, Figure 2d). These results suggest that 2,4-D is a substrate of PIN2 and PIN7 but not PIN1, consistent with a lack of 2,4-D polar transport activity along the axis from the shoot to root apices defined by PIN1 and ABCB19 (Jacobs et al., 1966; Ito and Gray, 2006; Blakeslee et al., 2007). When assayed with 50 nm³H-benzoic acid, only PIN1 showed some benzoic acid export activity (P = 0.037). No clear ³H-benzoic acid transport was observed in cells expressing PIN2 (P = 0.96) and PIN7 (P = 0.46), which suggested higher auxin transport specificity of PINs when expressed in S. pombe (Figure 2e) than was seen in mammalian and baker’s yeast cells (Petrasek et al., 2006; Blakeslee et al., 2007). The increased substrate specificity seen with PINs expressed in S. pombe, may reflect oligimerisation effects, as higher molecular weight (approximately 140 KDa) PIN2 signals were observed in Western blots of non-reducing SDS–PAGE gels derived from Arabidopsis and S. pombe that were not observed with PIN1 or any of the PIN proteins expressed in S. cerevisiae or HeLa cells (not shown).

Activity of recombinant Arabidopsis ABCB1 and ABCB19 in Schizosaccharomyces pombe

More than 130 ABC proteins are present in the Arabidopsis genome and 29 of them are now classified in the ABCB subfamily (Verrier et al., 2008). Of these, 21 members encode full-length P-glycoprotein transporters, but only four Arabidopsis transporters (ABCB1, ABCB4, ABCB14 and ABCB19) have been functionally characterized in heterologous systems to date (reviewed in Titapiwatanakun and Murphy, 2008; Lee et al., 2008). Arabidopsis ABCB14 is a malate importer and regulates stomatal responses to CO₂ (Lee et al., 2008). ABCB1 and ABCB19 are clearly auxin exporters while the directionality of auxin transport mediated by ABCB4 remains somewhat controversial (reviewed in Titapiwatanakun and Murphy, 2008). However, comparative analyses of the Arabidopsis ABCB transporters and more detailed screens of their substrate specificity have been limited by mistargetting of ABCB19 when it is expressed in S. cerevisiae (Noh et al., 2001; Geisler et al., 2005) and by the difficulty of performing large scale screens in vaccinia-transfected adherent HeLa cells where ABCB19 can be successfully expressed (Blakeslee et al., 2007; Rojas-Pierce et al., 2007; Titapiwatanakun et al., 2008).

Recombinant ABCB1 and ABCB19 were expressed in the double mutant mam1pdr1 under the control of the inducible nmt41 promoter. Consistent with the documented efflux activity of ABCB1 and ABCB19 (Geisler et al., 2005; Petrasek et al., 2006; Blakeslee et al., 2007), S. pombe cells expressing ABCB1 and ABCB19 accumulated less ³H-IAA compared to empty vector controls (P < 0.037 for ABCB1 and P < 0.005 for ABCB19, Figure 3a,b). ABCB19 expression conferred approximately 30% reduction in ³H-IAA net accumulation, which is the highest apparent IAA export activity among all of the ABCB and PIN proteins tested in both ael1 and mam1pdr1 backgrounds. ABCB1 reduced net auxin accumulation approximately 17%, and showed an export activity similar to or lower than that catalyzed by PIN1, PIN2, PIN7.

Figure 3.  Comparative auxin transport assays of ABCB1 and ABCB19 in S. pombe. (a) ³H-IAA transport assays of ABCB1 and ABCB19. Reduced ³H-IAA accumulation indicates export activity of ABCBs. (b) Western blot using anti-ABCB19 shows ABCB19 expression in S. pombe. Lane 1-2, empty vector; lane 3-4, ABCB1, lane 5-6, ABCB19. (c) inhibitory effects of IAA, 2,4-D and NPA on ³H-IAA export of ABCBs. (d) ³H-benzoic acid accumulation in S. pombe expressing ABCBs. (e) Co-expression of ABCB19-PIN1 in S. pombe. Error bars, SD (n = 8). Asterisks indicate P < 0.05.

³H-IAA export activity of cells expressing recombinant ABCB1 and ABCB19 was inhibited by a 10× concentration of cold IAA and 2,4-D (P > 0.07 for IAA, P > 0.8 for 2,4-D, Figure 3c). Although 2,4-D is poorly transported in planta (Jacobs et al., 1966; Brown and Phillips, 1982; Ito and Gray, 2006) and by ABCB19 and ABCB1 in heterologous systems (Titapiwatanakun et al., 2008), it does appear to inhibit auxin efflux, suggesting a rationale for its particularly strong herbicidal effects observed in apical tissues where ABCB1 and ABCB19 are abundant (Jacobs et al., 1966; Wernicke et al., 1986; Dudler and Hertig, 1992; Geisler et al., 2005; Blakeslee et al., 2007). Five micromolar NPA inhibited ³H-IAA export mediated by ABCB1 and ABCB19 by 97 and 88%, respectively (P > 0.29, Figure 3c), which is a greater level of inhibition than was seen in HeLa cells (Geisler et al., 2005; Bouchard et al., 2006). Finally, unlike what was previously seen with assays in HeLa cells, (Titapiwatanakun et al., 2008), ABCB1 and ABCB19 did not exhibit clear export activity in comparison with empty vector control when assayed with 50 nm³H-benzoic acid (Figure 3d).

Co-expression of recombinant ABCB19 and PIN1

To test if the S. pombe heterologous system can also be used to co-express and analyse recombinant plant proteins, we co-transfected ael1 with equal amounts of pREP41-ABCB19 and pREP41-PIN1 recombinant plasmids. As was seen in HeLa cells (Blakeslee et al., 2007), expression of PIN1 and ABCB19 in ael1 conferred a 37% reduction in ³H-IAA levels, which was higher export activity than was seen in cells expressing ABCB19 or PIN1 alone (Figure 3e).

Recombinant ABCB2 does not catalyze IAA transport

We had some concern that all plant ABCB transporters expressed in S. pombe might confer some level of increased auxin transport. ABCB2 exhibits sequence similarity to both ABCB1/19 and the malate transporter ABCB14 and was previously reported to have had no effect on auxin transport in HeLa cells where it has been overexpressed (Titapiwatanakun et al., 2008). We expressed full length ABCB2 in S. pombe but no change in IAA transport activity was observed (P > 0.3, Figure 4a). However, ABCB2 does not transport ³H-benzoic acid (P = 0.88, Figure 4b), as was the case when ABCB2 was expressed in HeLa cells (Titapiwatanakun et al., 2008). This suggests that ABCB2 exhibits greater substrate specificity in the cellular environment of S. pombe. It is also possible that expression of ABCB2 in HeLa cells activates the same organic anion transporter activity that is activated by PIN2 expression (Petrasek et al., 2006; Blakeslee et al., 2007) and that this same activation is not seen in S. pombe. The relative substrate specificity of the ABCB isoforms examined here should not be surprising, as the ABCB14 guard cell malate transporter exhibits sequence homology to the ABCB1 and ABCB19 auxin transporters, but also exhibits a high degree of specificity for malate (Lee et al., 2008).

Figure 4.  ABCB2 is not an auxin transporter in S. pombe. (a) ³H-IAA transport assay of ABCB2. (b) ³H-benzoic acid transport assay of ABCB2. Error bars, SD (n = 8).

Recombinant AUX1 exhibits strong auxin import activity in Schizosaccharomyces pombe

AVT3 and AVT4 are AUX1 homologs in S. cerevisiae that exhibit plasma membrane localization and have been implicated in IAA uptake (Prusty et al., 2004). The homologous gene in S. pombe is SPAC3H1.09c (sequence similarity to AUX1 approximately 5%, Figure S2). We have named it AVT3 after its closest homolog in S. cerevisiae. At the plant root apoplastic pH of 5.6, avt3 accumulates lower ³H-IAA levels than wild type S. pombe cells (P < 0.05, Figure 5a).

Figure 5.  Uptake of ³H-IAA by AUX1 in S. pombe avt3. (a) Auxin transport assays in S. pombe avt3 showed less ³H-IAA accumulation than wild type (WT). (b) ³H-IAA transport assays of AUX1. (c) Inhibitory effects of IAA and 2,4-D, 1-NAA and NPA on ³H-IAA import activity of AUX1. (d) ³H-benzoic acid accumulation in S. pombe avt3 expressing AUX1. Error bars, SD (n = 8). Asterisks indicate P < 0.05.

AUX1 has been characterized in planta, in Xenopus oocytes, and in mammalian HeLa cells (Bennett et al., 1996; Yang et al., 2006; Carrier et al., 2008), but has not been successfully expressed in more tractable systems. AUX1 expression in avt3 resulted in a 40% increase of ³H-IAA accumulation compared to empty vector controls (P < 0.002, Figure 5b). Consistent with previous reports (Yang et al., 2006), the import activity in S. pombe cells expressing AUX1 can be inhibited by IAA and 2,4-D but not by NAA and NPA (P = 0.034 by IAA, P < 0.001 by 2,4-D, Figure 5c). However, NPA increased the ³H-IAA level in all cells including empty vector controls (P < 0.01, Figure 5c), presumably due to its inhibitory effect on IAA export mediated by ABC transporters and AEL1. Cells expressing AUX1 showed only a small increase in ³H-benzoic acid transport uptake (P = 0.04, Figure 5d), suggesting that AUX1 has very weak benzoic acid import activity. These results indicate that the substrate specificity of AUX (IAA, 2,4-D, but not NAA) differs from that of PIN1 and ABCB 1/19 (IAA and NAA, but not 2,4-D) and PIN2/7 (IAA and NAA, weak 2,4-D). Comparative and mutational analyses of the AUX/LAX family of transporters can now be more readily performed by multiple investigators now that activity has been demonstrated in a more amenable system.

Recombinant ABCB4 catalyzes concentration-dependent IAA transport in Schizosaccharomyces pombe

Arabidopsis ABCB4 functions in light, sucrose, and temperature-dependent primary root/root hair root elongation and negatively regulates root gravitropism through redistribution of auxin out of the lateral root cap (Diana et al., 2005; Santelia et al., 2005; Terasaka et al., 2005; Cho et al., 2007; Lewis et al., 2007). Although ABCB4 has been functionally characterized in plant and non-plant heterologous systems, the auxin transport directionality of ABCB4 is still controversial. Previous assays of ABCB4 function in root hair growth and 1-NAA transport assays in BY-2 tobacco cells indicated that ABCB4 functions as an auxin exporter (Cho et al., 2007). However, uptake mediated by ABCB4 would be difficult to detect in these assays, as the transport substrate used was the lipophilic artificial auxin NAA, which is so much more readily taken up by plant cells that it was used in differential studies to initially characterize the AUX1 auxin uptake transporter (Bennett et al., 1996). Further, when expressed in HeLa cells, ABCB4 mediated net uptake of ³H-IAA when the radiotracer was supplied in low nanomolar concentrations (Terasaka et al., 2005). However, the apparent direction of transport was reversed by co-expression with PIN2 or treatment with NPA (Terasaka et al., 2005; Blakeslee et al., 2007). These results suggest that interactions of ABCB4 with other cellular components may modulate its auxin transport activity. As ABCB4 exhibits more hydrophobic solubilization characteristics and is found in sterol-enriched detergent resistant membranes (Borner et al., 2005; Titapiwatanakun and Murphy, 2008), the activity of the protein could be modulated by the presence of sterol-enriched microdomains, proteins associated with those domains, or auxin itself.

ABCB4 was introduced into mam1pdr1 cells, and its expression was confirmed by Western blot (Figure 6a, inset). mam1pdr1 cells expressing ABCB4 exhibited clearly different ³H-IAA transport kinetics from those expressing ABCB1 and ABCB19 (Figure 6a) and cells accumulated approximately 30% more ³H-IAA than mam1pdr1 cells carrying empty vector after 6 min incubation in medium with 50 nm³H-IAA (P < 0.001, Figure 6a). However, after this time point, cells expressing ABCB4 suddenly and reproducibly reversed to export of ³H-IAA and exhibited lower ³H-IAA accumulations than vector controls after 8 min (Figure 6a). At this time point and thereafter, ³H-IAA accumulations were approximately 30% lower than in vector controls (P < 0.005), which is comparable to the export activity of ABCB19 and greater than the activity seen when recombinant ABCB1 and PIN2 were assayed. These results suggest that, unlike constitutive exporters such as PIN2 and ABCB1/19, ABCB4 export is substrate activated and that ABCB4 functions in substrate uptake until a threshold concentration is reached. Such activity is consistent with the conditional nature of Arabidopsis abcb4 mutant phenotypes which vary with conditions that also vary internal auxin accumulations (Gray et al., 1998;Moore et al., 2003; Santelia et al., 2005; Terasaka et al., 2005; Moon et al., 2007). In all auxin accumulation assays, the initial rate of accumulation from 0 to 4 min was much greater than at subsequent timepoints. A slight depression in the rate of accumulation could sometimes be observed at the transition point, although standard deviations in the transitional values never altered the significance of observed differences in accumulation (Figure 6a, also in Figure 2a–c).

Figure 6.  Concentration dependent IAA transport in S. pombe expressing ABCB4. (a) ³H-IAA transport assays of ABCB4. Inset, Western blots using anti-ABCB4 show ABCB4 expression in S. pombe. (b) ³H-IAA transport assays of ABCB4 with 2X cold IAA added. (c) ³H-benzoic acid transport assays in S. pombe expressing ABCB4. (d) Inhibitory effects of IAA, NAA, 2,4-D and NPA on ³H-IAA import of ABCB4 at 6 mins. (e) Inhibitory effects of IAA, NAA, 2,4-D and NPA on ³H-IAA export of ABCB4 at 12 mins. Error bars, SD (n = 8). Asterisks indicate P < 0.05.

To determine if external high IAA concentrations could activate ABCB4 activity, the 50 nm³H-IAA in the assay medium was supplemented with 100 nm cold IAA. Under these conditions, accumulation of ³H-IAA increased to a level 50% higher than that in vector controls in only 4 min (P = 0.0056) but then rapidly reversed to endpoint ³H-IAA levels similar to vector controls (P > 0.5, Figure 6b). No net endpoint ³H-IAA export activity of ABCB4 was seen with this 2× competition of external cold IAA, suggesting that, as previously proposed, ABCB4 functions primarily in retention of auxin in the root elongation zone and reversible import of auxin in mature epidermal and cortical cells where auxin levels are very low (Terasaka et al., 2005; Lewis et al., 2007; Cho et al., 2007; Peer and Murphy, 2007). These results demonstrate a differential import and export activity of ABCB4 which is consistent with its role in auxin reflux from lateral root cap and its role in regulating auxin homeostasis in epidermal and cortical cells in the elongation and differentiation zone (Titapiwatanakun and Murphy, 2008). Further, this activity is consistent with both the ABCB4 function in root hair elongation shown by Cho et al. (2007) and recent models of auxin movement into root hair cells proposed by Jones et al. (2009), which lacked a control mechanism for initial uptake of auxin from the trichoblasts themselves.

Inhibitor assays of ABCB4 were performed with 10× cold IAA, NAA, 2,4-D and 5 μm NPA with incubations of 6 min for ³H-IAA import activity (Figure 6d) and of 12 min for export activity (Figure 6e). After a 6 min incubation with 50 nm³H-IAA, mam1pdr1 cells expressing ABCB4 accumulated 27% more ³H-IAA than vector controls (P = 0.01, Figure 6d). However, addition of 500 nm cold IAA or 1-NAA had little or no effect on ³H-IAA influx (both still significant, P = 0.026 and P < 0.001, respectively) while addition of 500 nm 2,4-D reduced ³H-IAA levels in cells expressing ABCB4 to vector control levels (P = 0.69). As had been seen in HeLa cells (Terasaka et al., 2005), 5 μm NPA not only inhibited the import activity of ABCB4 but also reversed ABCB4 activity to net export in 6 min (P < 0.001, Figure 6d). In the 12-min incubations, NPA did not inhibit ³H-IAA export of ABCB4 at this timepoint since ³H-IAA levels in cells expressing ABCB4 were approximately 19% lower than vector control (P < 0.001). This is consistent with a previous report that NPA also could not inhibit export activity of ABCB4 in BY-2 cells (Cho et al., 2007). Competition with 10× IAA (P = 0.57), NAA (P = 0.52) and 2,4-D (P = 0.16) all inhibited ³H-IAA export activity of ABCB4 (Figure 6e). These results suggest that auxin export mediated by ABCB4 is more specific than auxin uptake and confirm that the auxin efflux activity of ABCB4 is optimal when auxin is exported into tissues with lower auxin concentrations. These results are also consistent with the proposed function of ABCB4 in mature root tissues (Peer and Murphy, 2007).

As was the case when ABCB4 was expressed in HeLa cells (Titapiwatanakun et al., 2008), ABCB4 expression in S. pombe reduced ³H-benzoic acid levels by approximately 40% compared to empty vector controls after a 12 min incubation at 30°C (P < 0.001, Figure 6c). The benzoic acid export activity of ABCB4 is even higher than its IAA export activity. These results suggest that ABCB4 is an IAA exporter but also exports similar phenolic acids and has a broader substrate profile than ABCB1 and ABCB19.

Computational modelling provides a rationale for ABCB4 reversibility

We developed computational models of ABCB4 and ABCB19 (see Methods Modelling ABCB4 and ABCB19 structures) based on the crystal structure of Sav1866, a bacterial multidrug resistance ABC transporter (Dawson and Locher, 2006). Structural comparisons indicate that ABCB4 and ABCB19 share a common architecture (Figure 7a). Both transporters are composed of two modules – each consisting of an amino-terminal transmembrane domain (TMD) containing 6 transmembrane helices (TMH 1–6) and two intracellular loops (ICL I and II) followed by a carboxy-terminal nucleotide binding domain (NBD). A linker domain connects NBD1 with TMD2 (Figure 7a).

Figure 7.  Computational modelling and substrate docking provide a rationale for ABCB4 reversibility. (a) Comparison of the Arabidopsis ABCB19 and ABCB4 models. In AtABCB4 model, TM4 (pale green) shifts down off the membrane plane. Two coiled-coil domains (highlighted in oval) are not found in ABCB exporters. (b) Two putative IAA binding sites in ABCB19 TMDs. (c) Three IAA binding sites identified in ABCB4. (d) Nonspecific IAA docking in ABCB14 malate transporter.

Detailed sequence and structural comparisons identified some differences between ABCB4 and ABCB19. First, the coils software program (http://www.ch.embnet.org/software/COILS_form.html) (Lupas et al., 1991) predicts N-terminal coiled-coil structures in ABCB4 (Figure 7a, highlighted in top oval) that are also found in the Arabidopsis ABCB14 guard cell malate importer (Lee et al., 2008), the Coptis japonica CjMDR1 putative berberine importer (Shitan et al., 2003), and ABCB21, a highly similar Arabidopsis ABCB protein (Verrier et al., 2008). No such N-terminal coiled-coil domain was found in the ABCB19 and ABCB1 exporters. An interaction of a protein or other inhibitory molecule with this N-terminal coiled-coil domain may move the positions of the TMHs and change the transport properties of ABCB4 and other plant importers. Second, the hydrophobic region of TMH4 in ABCB4 is shifted below the membrane plane in the models. The hydrophobicity of this domain would likely embed the structure in the membrane and change the distance and interaction between ICL2 and NBD2, an important feature of Sav1866 exporter structure (Dawson and Locher, 2006; Zolnerciks et al., 2007). These adjustments could alter the TMD arrangement in ABCB4 sufficiently to alter the direction of transport. Third, the linker domain of ABCB4, connecting NBD1 and TMD2, contains another coiled-coil structure (Figure 7a, highlighted in the bottom oval), which is not found in the plant ABC importers CjMDR1, ABCB14 and ABCB21 or the ABCB1 and ABCB19 exporters (Figure 7a), suggesting that this unique feature could also regulate ABCB4 activity. The switch between binding of the N-terminal coiled-coil to a binding protein or the linker coiled-coil domain could provide an explanation for the changeable directionality observed in ABCB4 that is not detected in other plant importers (Santelia et al., 2005; Cho et al., 2007; Lee et al., 2008).

We developed probability-based models of IAA binding in the TMDs of ABCB4 and ABCB19 (see methods) and compared the predicted interactions with those seen in the crystal structure of the TIR1 auxin receptor (Tan et al., 2007; Hayashi et al., 2008) to determine the consistency of the docking models. To validate the specificity of the modelled docking interactions, the probability of IAA interactions with the TMD of the ABCB14 malate importer were evaluated. After 15–20 docking runs, IAA was primarily docked into two binding sites in the TMDs of PGP19 (Figure 7b). One site is surrounded by the transmembrane helices (TMH) 1, 2 and 3, and the other is located at the opening between TMH 7, 8 and 9. In addition to the two IAA binding sites in the same positions, a unique third IAA binding site was identified in ABCB4 (Figure 7d). The third IAA binding site in ABCB4 is formed by TMH5/8 where the IAA is partially inserted into the TMDs with the remaining portion of the molecule extending into the cytosol (Figure 7d). In contrast, docking models place IAA molecules in a low probability dispersed pattern within the TMD pathway of ABCB14 malate importer (Figure 7c), indicating that IAA is a poor substrate for this protein. The third binding site of the ABCB4 might function as a regulatory site, in which an interaction with an IAA molecule activates export activity.

The fission yeast S. pombe heterologous expression system described here provides a common expression host environment for analysis of all known auxin transporters. With cell walls, more plant-like N-glycosylation, and sterol-rich membrane domains, S. pombe can be used for functional expression of plant membrane transporters that are not active when expressed in budding yeast S. cerevisiae or mammalian cells. Gene knockouts are readily achieved in this system and are increasingly available as a result of community efforts such as PombeNet (http://www-rcf.usc.edu/~forsburg/index.html#res). As the expression of recombinant transporters is controlled by an inducible nmt1-41promoter that is suppressed in the presence of thiamine (Basi et al., 1993; Forsburg, 1993), under suppressive conditions stable recombinant S. pombe lines can be stored without harm from exogenous gene product. Overexpression of recombinant proteins during cell line storage is prevented and retransformation is not required for each transport assay. The moderate expression provided by the nmt1-41 promoter also prevents overexpression artefacts in the system, making it a convenient tool for biochemical characterisation of plant transporters. As purification of quantities of active proteins from S. pombe in some membrane proteins has proven to be more successful than from S. cerevisiae (Lundstrom, 2007), the system is also a good platform for subsequent efforts to produce proteins of high enough quality to be used for X-ray crystallography. Detailed protocols for use of the system are available on our lab website (http://www.hort.purdue.edu/hort/research/murphy/mainpage.htm) and materials are available through the Arabidopsis Biological Resource Centre.

The expression and characterisation of all three groups of auxin transporters shown illustrates the value of this system. PIN1, PIN2 and PIN7 were successfully expressed in S. pombe with similar expression under same inducible conditions levels and showed comparable IAA export activity. IAA export activity of PIN proteins seen in both ael1 and mam1pdr1 suggests that PINs function as direct auxin efflux carriers, although secondary activation of other transport proteins cannot be ruled out. However, no apparent activation of endogenous organic anion transporters was seen in this system as was the case with expression in HeLa cells (Petrasek et al., 2006).

ABCB19 and some other Arabidopsis ABCB proteins cannot be successfully expressed in bakers’ yeast. Here we have been able to directly compare the transport activity of ABCB auxin transports with PINs and AUX1. ABCB19 showed highest transport activity in this system, which is consistent with its primary role in long distance polar auxin transport rather than the localized vectorial redirection of auxin mediated primarily by PIN proteins (Wu et al., 2007; Blakeslee et al., 2007; Titapiwatanakun et al., 2008; Mravec et al., 2008). AUX1, functionally characterized only in Xenopus oocytes (Yang et al., 2006), was also successfully expressed and exhibited import selectivity consistent with previous reports (Yang et al., 2006).

Perhaps the most striking result of the experiments presented here is the clarification of the activity of ABCB4. ABCB4 expressed in S. pombe showed substrate activation of IAA export as well as IAA import activity in initial loading. When evaluated in the context of the substrate binding structural model presented here, we can suggest that ABCB4 acts as an auxin importer under low substrate concentrations and as an auxin exporter under high substrate concentrations as a result of auxin binding in the third substrate binding site. Thus, the directionality of ABCB4 may be moderated both by its substrates and other factors such as protein and membrane interactions. Such a model is consistent with its apparent physiological role in planta. Site-directed mutagenesis studies of the protein in S. pombe and in planta now underway will allow this model to be tested.

Experimental procedures

Strains, media and materials

Escherichia coli strain TOP10 Blue (Invitrogen, http://www.invitrogen.com) was used for all cloning procedures. S. pombe gene disruption mutants were constructed using wild-type strain YF016 (h⁻leu1-32 ura4-C190T ade7:: ura4) grown in rich medium (YES) and synthetic Edinburgh minimal medium (EMM) as described in Iwaki et al.(2006). All chemicals were from Sigma (http://www.sigmaaldrich.com) unless otherwise specified.

Filipin staining and microscopy

Schizosaccharomyces pombe cells were grown to OD₆₀₀ approximately 2.0, washed in 50 nm phosphate buffer (pH 5.5), stained with 5 μg ml⁻¹ filipin for 5 min, washed, and visualised with a Nikon Eclipse800 epifluorescence microscope (ex 330–380) equipped with a SpotRT Slider CCD camera (Diagnostic Instrument, Inc, http://www.promega.com).

Construction of Schizosaccharomyces pombe mutants

Construction of ABC mutants and their disruption plasmid (URA4 as selectable marker) were as described in Iwaki et al. (2006). The ael1 mutant was constructed as in Titapiwatanakun et al. (2008). vat3 was purchased from http://pombe.bioneer.co.kr/.

Expression and Gate way vector constructs

pREP41-ABCB1 was created by excising the ABCB1 ORF from pTM1-ABCB1 (Geisler et al., 2005) with SmaI/BamHI and ligation into the BamHI and Klenow-filled NdeI sites in pREP41. pREP41-ABCB19 was created by ligating an NcoI/blunt-ABCB19-BamHI fragment (Blakeslee et al., 2007) into NdeI/blunt-pREP41- BamHI. pREP41-PIN7 and pREP41-PIN2 were created by ligating fragments of NcoI/blunt-PIN7-Sal I and NcoI/blunt-PIN2-Sal I (Petrasek et al., 2006) into NdeI/blunt-REP41-SalI. pREP41-AUX1 was created by PCR amplification of AUX1 (Blakeslee et al., 2007) and ligation into pGEM-T Easy vector (Promega) followed by NdeI-AUX1-XmaI excision and ligation into pREP41 at Nde I/Xma I sites. pREP41GW was created by PCR amplifying the Gateway^® cassette using the primers GW5′ (CCATATTAATACAAGTTTGTACAAAAA-AGCTGAAC) and GW3′ (ACCCTCGAGCACCACTTTGTACAAGAAAG), digesting the product with AseI/Xho I and ligating it into the NdeI/Sal I sites in pREP41. pREP41HAGW was created in the same way using the primers GW3′ and 3xHA5′-AseI (GAATATTAATATGTACCCATACGATGTTCCTGACTATG), and pREP41YFPGW, was created using the primers GW3′ and YFPGW 5′ (GAATATTAATATGGGCAAGGGCGAGGAGCT). ABCB4–pREP41GW was created by PCR amplification of ABCB4 (Terasaka et al., 2005) with the primers BP-ABCB4-5′ (GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGCTTCAGAGAGCGGCTT) and BP-ABCB4-3′ (GGGGACCACTTTGTACAAGAAAGCTGGGTTCAAGAAGCCGCGGTTAGAT), ligation into the pDONR entry vector, and insertion into pREP41GW with an LR reaction (Invitrogen).

Membrane preparation for Western blot

For Western blots, microsomal fractions from S. pombe of ABCBs were prepared by growing S. pombe cells to OD₆₀₀ 0.25 in 50–100 ml EMM. Cells were collected by centrifugation and washed once with water. Cells were disrupted at 4°C with 50 μl bead buffer (20 mm Tris-HCl, pH 7.4, 0.3 m sorbitol, 10 mm MgCl_2, 1% BSA, 15 mm EGTA, 1 mm DTT, 1 mm PMSF, 20 mg L⁻¹ leupeptin, 200 mg L⁻¹ pepstatin, 40 mg L⁻¹ aprotinin) and 200 ml glass beads by 10 cycles of 30 sec vortexing + 30 sec on ice. One millilitre homogenisation buffer (20 mm Tris-HCl, pH 7.4, 0.3 m sorbitol, 10 mm MgCl_2, 0.2% BSA, 5 mm EDTA, 1 mm DTT, 0.1 mm PMSF, 0.2 mg L⁻¹ leupeptin, 2 mg L⁻¹ pepstetin, 0.1 mg L⁻¹ aprotinin) was added and the suspension was centrifuged at 1000 g for 5 min. This extraction was repeated two more times and supernatants were pooled, then centrifuged at 4°C, 100 000 g for 1 h. The pellet was resuspended in 200 μl storage buffer (20 mm K-HEPES pH 7.4, 0.3 m sorbitol, 1× Sigma protein inhibitor cocktail, 1% methyl-β-cyclodextrin).

Microsome fraction from S. pombe for Western blots of PINs was prepared in the same procedure as mentioned above, except the cell disruption buffer was altered [50 mm Tris pH 7.8; 5% (v/v) glycerol; 1.5% (w/v) insoluble polyvinyl-polypyrrolidone; 150 mm KCl; 50 mm NaF; 20 mm beta-glycerolphosphate; 0.5% (v/v) solubilized casein, 10 mm benzamidine; 1% BSA, 5 mm EDTA, 15 mm EGTA, 1 mm DTT, 10 mm PMSF, 2 0mg L⁻¹ leupeptin, 200 mg L⁻¹ pepstetin, 40 mg L⁻¹ aprotinin], the homogenisation buffer pH was 7.8, and the storage buffer was replaced with that used previously (Petrasek et al., 2006). Protein concentration was determined using amido black. 30-μg total protein each sample was loaded for SDS–PAGE and Western blot.

³H-IAA transport assay

Schizosaccharomyces pombe cells were grown to OD₆₀₀ approximately 2.0 in EMM containing 15-μm thiamine. The thiamine was removed by washing twice with EMM, and cells were transferred to fresh EMM and incubated for 19 h (final OD₆₀₀ = 1–2) to induce the expression of proteins. Cells were spun at 6000 g for 30 sec. Pellets were washed once and resuspended in EMM (pH 4.5) to OD₆₀₀ = 2. Cell samples were kept at 4°C in all following steps except where noted. One microlitre of 10× diluted ³H-IAA (0.1 μCi) (specific activity 20 Ci mmol⁻¹, American Radiolabelled Chemicals) was added into 100 μl cells and incubated at 30°C for 0, 2, 4, 6, and 8 min. Cells were washed twice with EMM (pH 4.5) and resuspended in 0.5 ml EMM (pH 4.5). Two hundred and fifty microlitre aliquots were taken and retained radioactivity was quantified by scintillation counting. All transport assays were performed with four transformants in at least three independent experiments.

Modelling ABCB4 and ABCB19 structures

The sequence alignment of ABCB4 and 19 with Sav1866 was performed in Multalin (Corpet, 1988): http://bioinfo.genotoul.fr/multalin/multalin.html. Based on the sequence alignment and the crystal structure of Sav1866, ABCB4 and 19 models were computed with Modeller structural modeling software http://salilab.org/modeller (Sali and Blundell, 1993; Eswar et al., 2000). Computational simulation of ligand binding to ABCB TMDs was performed using MEDock (Maximum-Entropy based Docking) with 20 runs in each docking (http://medock.csbb.ntu.edu.tw) (Chang et al., 2005). Each docking run generates a binding confomer.

Acknowledgements

This work was supported by the Department of Energy, Basic Energy Sciences, grant no. DE-FG02-06ER15804 to ASM. We would like to thank Dr Kaoru Takegawa for providing the pREP41 vector and S. pombe ABC transporter mutants and Lauren Evans for help in growing S. pombe and performing auxin transport experiments.