Peroxisomal ATP Import Is Essential for Seedling Development in Arabidopsis thaliana

Abstract

Several recent proteomic studies of plant peroxisomes indicate that the peroxisomal matrix harbors multiple ATP-dependent enzymes and chaperones. However, it is unknown whether plant peroxisomes are able to produce ATP by substrate-level phosphorylation or whether external ATP fuels the energy-dependent reactions within peroxisomes. The existence of transport proteins that supply plant peroxisomes with energy for fatty acid oxidation and other ATP-dependent processes has not previously been demonstrated. Here, we describe two Arabidopsis thaliana genes that encode peroxisomal adenine nucleotide carriers, PNC1 and PNC2. Both proteins, when fused to enhanced yellow fluorescent protein, are targeted to peroxisomes. Complementation of a yeast mutant deficient in peroxisomal ATP import and in vitro transport assays using recombinant transporter proteins revealed that PNC1 and PNC2 catalyze the counterexchange of ATP with ADP or AMP. Transgenic Arabidopsis lines repressing both PNC genes were generated using ethanol-inducible RNA interference. A detailed analysis of these plants showed that an impaired peroxisomal ATP import inhibits fatty acid breakdown during early seedling growth and other β-oxidation reactions, such as auxin biosynthesis. We show conclusively that PNC1 and PNC2 are essential for supplying peroxisomes with ATP, indicating that no other ATP generating systems exist inside plant peroxisomes.

The β-oxidation of fatty acids, a process that exclusively occurs within peroxisomes in plants and yeast, plays an important role in storage oil mobilization to support seedling establishment of oilseed plants, such as Arabidopsis thaliana (Graham and Eastmond, 2002; Baker et al., 2006; Graham, 2008). Upon germination, fatty acids are released from storage oil triacylglycerol (TAG) by lipolysis, degraded via β-oxidation in specialized peroxisomes, termed glyoxysomes, and subsequently converted to sucrose, which drives growth and development until seedlings become photoautotrophic (Graham and Eastmond, 2002; Baker et al., 2006; Graham, 2008). Before the fatty acids can enter β-oxidation, they are imported into peroxisomes by a peroxisomal ATP binding cassette (ABC) transporter, variously known as CTS (COMATOSE), At PXA1 (Arabidopsis peroxisomal ABC transporter), or PED3 (peroxisomal defective 3) and hereafter referred to as CTS (Zolman et al., 2001; Footitt et al., 2002; Hayashi et al., 2002). Subsequently, the imported fatty acids are activated by esterification to CoA. This ATP-dependent reaction within peroxisomes is catalyzed by long-chain acyl-CoA synthetases 6 and 7 (LACS6 and LACS7, respectively), which are named according to their substrate specificity for long-chain fatty acids, which are significant components of seed storage oil in Arabidopsis (Fulda et al., 2002, 2004).

In Saccharomyces cerevisiae, two mechanisms exist for import and activation of fatty acids, depending on chain length (Hettema et al., 1996). Long-chain fatty acids (C16 and C18) are converted to acyl-CoA esters in the cytosol prior to transport by the heterodimeric peroxisomal ABC transporter, Pxa1p/Pxa2 (Hettema et al., 1996). By contrast, short- and medium-chain fatty acids (≤C14) that enter the peroxisomes by passive diffusion or by an unknown transport protein are activated within peroxisomes (Hettema et al., 1996). The possibility cannot be excluded, though, that CTS imports the corresponding CoA derivatives, as is the case for the yeast Pxa1p/Pxa2p heterodimer (Hettema et al., 1996; Verleur et al., 1997), implicating a cytosolic activation of the fatty acids, catalyzed by a hitherto unknown enzyme. The actual substrates transported by CTS in Arabidopsis have not yet been experimentally determined (Theodoulou et al., 2006). However, the sucrose-dependent seedling growth phenotype of the lacs6 lacs7 double knockout mutant demonstrated that peroxisomal activation is essential for lipid mobilization to provide energy for early seedling growth (Fulda et al., 2004). The lacs6 lacs7 mutant is impaired in the degradation of fatty acids, leading to growth arrest shortly after germination (Fulda et al., 2004).

Besides fatty acid mobilization, β-oxidation is also involved in generation of signaling molecules, such as the phytohormones auxin and fatty acid–derived jasmonic acid (JA) (Zolman et al., 2000; Schaller et al., 2004; Delker et al., 2007). By analogy to fatty acids released from storage oil, the precursors of these signaling molecules require CoA esterification before they can enter β-oxidation (Baker et al., 2006; Goepfert and Poirier, 2007). While the enzymes responsible for ATP-dependent activation of natural auxin (indole butyric acid [IBA]) and proherbicide 2,4-dichlorophenoxybutyric acid (2,4-DB) are currently unknown, several enzymes belonging to the acyl-activating enzyme (AAE) family have been implicated in jasmonate biosynthesis (Schneider et al., 2005; Koo et al., 2006; Kienow et al., 2008). Moreover, several as yet uncharacterized members of the large AAE family carry a putative peroxisome targeting signal (PTS) and thus might be good candidates to activate the additional β-oxidation substrates within peroxisomes (Shockey et al., 2002, 2003).

In the case where activation of fatty acids or other substrates takes place within peroxisomes, the question arises as to how these ATP-dependent reactions are supplied with ATP. It is currently unknown whether plant peroxisomes are able to produce ATP by substrate-level phosphorylation or whether they depend on external ATP to supply energy-dependent reactions within peroxisomes. So far, transport proteins that supply plant peroxisomes with energy for fatty acid oxidation have not been characterized. However, in bakers' yeast, a peroxisomal adenine nucleotide transporter, ANT1, that is required for the ATP-dependent activation of medium-chain fatty acids inside peroxisomes has been characterized (Palmieri et al., 2001).

ATP transport proteins play an important role in the distribution of the primary agent coupling endergonic and exergonic reactions in every cellular compartment (Winkler and Neuhaus, 1999). In Arabidopsis and other plants, various adenine nucleotide carriers have been identified at the molecular level. The mitochondrial ADP/ATP carrier mediates the export of ATP that is synthesized in the mitochondrion to provide energy for cellular metabolism (Heimpel et al., 2001; Haferkamp et al., 2002). The plastidial ATP/ADP transporter (nucleotide transporter) is involved in ATP uptake by both chloroplasts and heterotrophic plastids, to enable the nocturnal ATP supply required for chlorophyll biosynthesis (Reiser et al., 2004; Reinhold et al., 2007), as well as by heterotrophic plastids to drive starch biosynthesis (Batz et al., 1992; Tjaden et al., 1998). Yet another ATP/ADP antiporter located in the endoplasmic reticulum (ER) membrane provides energy by importing ATP into the ER for the accumulation of ER-related storage lipids and proteins (Leroch et al., 2008).

In this study, we identified two novel peroxisomal adenine nucleotide carrier proteins (PNC1 and PNC2) from Arabidopsis. Colocalization studies demonstrated that these proteins are targeted to peroxisomes. Yeast complementation and in vitro ATP uptake assays showed that both PNC1 and PNC2 catalyze the counterexchange of ATP with AMP. Using an inducible RNA interference (RNAi) repression strategy, we further established several transgenic Arabidopsis lines with reduced expression levels of both PNC1 and PNC2. Our results showed that import of ATP into peroxisomes that is catalyzed by PNC1 and PNC2 is essential for activation of fatty acids during seedling germination and plays a role in other β-oxidation reactions in peroxisomes, such as auxin metabolism. Analysis of PNC1 and PNC2 repression lines further indicates that no other ATP generating systems exist inside plant peroxisomes and that ATP import is the only way to supply the peroxisomal matrix with ATP.

RESULTS

Identification of Genes Encoding Putative Peroxisomal ATP Carriers in Arabidopsis

To identify candidate genes encoding peroxisomal ATP carriers, the Arabidopsis genome was searched using the sequence of the S. cerevisiae peroxisomal ATP carrier, ANT1 (Palmieri et al., 2001), and the BLAST search algorithm (Altschul et al., 1990). The three candidate genes showing highest similarity to ANT1 were selected for further analysis.

At2g39970 encodes a protein of 331 amino acids with a calculated molecular mass of 36.2 kD that shares 21% identity with ANT1. The At3g05290 gene product consists of 322 amino acid residues, has a molecular mass of 35.4 kD, and is 23% identical to the yeast carrier. The 321–amino acid protein encoded by At5g27520 has a predicted molecular mass of 35.2 kD and shows 22% identity to the yeast protein. A protein alignment revealed a strong similarity between At3g05290 and At5g27520 (86% identity) (Figure 1). Both proteins display a weaker similarity to the corresponding protein of the third Arabidopsis candidate, At2g39970 (22% identity).

Figure 1.

Amino Acid Sequence Alignment of the Three Arabidopsis Peroxisomal ATP Carrier Candidate Genes, At3g05290, At5g27520, and At2g39970, with Rice, Yeast, and Human Homologs.

Black shading indicates identical amino acid residues in all six sequences, whereas gray shading indicates identical amino acid residues in at least four sequences. The mitochondrial energy transfer signature domains are boxed, and the six predicted transmembrane-spanning domains are underlined. The arrows flank the region that was used for the RNAi approach to silence expression of PNC1 and PNC2 simultaneously.

All three Arabidopsis candidate genes belong to the so-called mitochondrial carrier family (MCF) (Picault et al., 2004; Haferkamp, 2007). They possess characteristic sequence features typical for MCF members: three repeats of a conserved mitochondrial energy transfer signature, identified as PS50920 by the PROSITE program (http://au.expasy.org/prosite), and six predicted transmembrane-spanning domains (http://aramemnon.botanik.uni-koeln.de) (Figure 1). The MCF is widespread in microorganisms, animals, and humans, and several members of the MCF have been described as carriers with various functions and subcellular localizations (Picault et al., 2004; Bedhomme et al., 2005; Bouvier et al., 2006; Palmieri et al., 2006; Haferkamp, 2007; Weber and Fischer, 2007).

One of the candidate genes, At2g39970, is annotated as peroxisomal membrane protein 38 kD (PMP38) and was previously proposed to represent a peroxisomal ATP carrier (Fukao et al., 2001). However, biochemical data investigating its transport function or mutants deficient in this protein have not yet been reported.

At3g05290 and At5g27520 Complement a Yeast Mutant That Is Deficient in the Peroxisomal ATP Transporter

To investigate whether the three Arabidopsis candidate genes identified in silico indeed encode peroxisomal ATP carriers, we attempted genetic complementation of a yeast mutant deficient in ANT1 function (Palmieri et al., 2001). This knockout strain, called ant1Δ, is unable to grow on media containing medium-chain fatty acids, such as lauric acid (C12), as the sole carbon source (Palmieri et al., 2001). In Figure 2, we show that heterologous expression of At3g05290 or At5g27520 restored the capability of ant1Δ to grow efficiently on lauric acid, similar to mutant yeast cells expressing the endogenous ANT1 gene and the wild-type strain transformed with an empty vector (positive control). By contrast, the knockout yeast strain expressing At2g39970 failed to form colonies, as did the ant1Δ mutant transformed with the empty vector (negative control). These data indicate that the proteins encoded by At3g05290 and At5g27520 are able to mediate ATP import into yeast peroxisomes, which is required for growth on lauric acid.

Figure 2.

Complementation of a Yeast Mutant Deficient in Peroxisomal ATP Uptake (ant1Δ).

The ant1Δ yeast mutant transformed with either empty vector (pDR195) or with the ORFs of At2g39970, At3g05290, At5g27520, and Sc ANT1 cloned into pDR195 was tested for growth on agar plates containing medium with lauric acid as the sole carbon source. As an additional control, the wild-type strain transformed with the pDR195 vector was also used for the growth assays. The plates were photographed after incubation at 30°C for 6 d.

At2g39970, At3g05290, and At5g27520 Are Targeted to Peroxisomes in Yeast

To assess whether the gene products of the Arabidopsis peroxisomal ATP carrier candidates were targeted to peroxisomes in yeast, their subcellular localization was analyzed using fluorescent protein tagging in combination with fluorescence microscopy. The open reading frames (ORFs) of At2g39970, At3g05290, and At5g27520 were fused in frame to the N or C terminus of enhanced yellow fluorescent protein (EYFP) and transformed into the yeast mutant ant1Δ. The EYFP fusion proteins were coexpressed with a peroxisomal-targeted fluorescent marker that consisted of the cyan fluorescent protein (CFP) tagged with the peroxisomal target signal 1 (PTS1) at its C terminus (Gould et al., 1989). Figure 3A shows that fluorescent signals colocalized in yeast cells expressing both CFP-PTS1 and EYFP fusions exhibited, forming a punctuate pattern. These data indicate that all three candidate proteins are targeted to peroxisomes in yeast. Importantly, the EYFP-tagged proteins encoded by At3g05290 and At5g27520 complemented the mutant growth phenotype (see Supplemental Figure 1 online), demonstrating that these fluorescently tagged plant proteins were functional in yeast peroxisomes. Jointly, complementation and subcellular localization data strongly suggest that At3g05290 and At5g27520 represent peroxisomal ATP carriers and thus were assigned the acronyms PNC1 and PNC2, for Arabidopsis peroxisomal adenine nucleotide carrier 1 and 2, respectively. Although the third candidate, At2g39970, is targeted to peroxisomes in yeast, it does not complement the phenotype of the ant1Δ mutant. It is thus unlikely that it functions as a peroxisomal ATP transporter.

Figure 3.

Colocalization Studies of EYFP Fusion Proteins and CFP-PTS1 in the ant1Δ Yeast Mutant and in Tobacco Epidermal Cells.

(A) The ORFs of PNC1 and PNC2 were fused at their C termini to EYFP, whereas the coding sequence of At2g39970 was fused at its N terminus to EYFP. The EYFP fusion proteins were coexpressed with the peroxisomal marker CFP-PTS1 under the control of a constitutive promoter. The subcellular localization was analyzed in the ant1Δ knockout yeast mutant by fluorescence microscopy. The merged images of CFP and EYFP fluorescence are also shown.

(B) PNC1 and PNC2 fused to EYFP were transiently coexpressed with CFP-PTS1 in tobacco epidermal cells. Images were taken 3 d after Agrobacterium infiltration. The colocalization is demonstrated in the merged images of the PNC1/2-EYFP fusions with those of CFP-PTS1.

PNC1 and PNC2 Are Targeted to Peroxisomes in Tobacco Epidermal Cells

To assess whether PNC1 and PNC2 are targeted to peroxisomes in plants as well, young tobacco (Nicotiana benthamiana) leaves were infiltrated with Agrobacterium tumefaciens that were cotransformed with the corresponding PNC1/2-EYFP construct and a plasmid carrying the peroxisomal-targeted marker CFP-PTS1. Three days after infiltration, tobacco epidermal cells were analyzed in vivo by fluorescence microscopy (Figure 3B). The peroxisomes were visualized as blue fluorescent dots by the peroxisomal CFP-PTS1 marker construct. The EYFP and CFP signals colocalized, indicating that PNC1-EYFP and PNC2-EYFP are targeted to peroxisomes not only in yeast, but also in plant cells.

PNC1 and PNC2 Catalyze an ATP/AMP Exchange

While genetic complementation of a yeast knockout mutant deficient in its peroxisomal ATP carrier by PNC1 and PNC2 strongly indicated that the plant proteins encode peroxisomal ATP importers, this experiment does not provide direct proof of protein function. To substantiate their transport properties, we analyzed whether recombinant PNC1 and PNC2 reconstituted into liposomes were able to transport ATP in vitro (Flugge and Weber, 1994; Weber et al., 1995). The peroxisomal ATP transport proteins from Arabidopsis and the yeast ANT1 were expressed in the yeast knockout mutant ant1Δ (Palmieri et al., 2001). Membrane proteins from the corresponding transgenic yeast cells were extracted, directly reconstituted into liposomes, and assayed for ATP import activity. The previously characterized ANT1 catalyzes the transport of ATP, ADP, or AMP in a strict counterexchange mode (Palmieri et al., 2001). Thus, the uptake of radioactively labeled [α-³²P]ATP into proteoliposomes was measured in the presence or absence of a suitable counterexchange substrate. Given that the mitochondrial ATP/ADP carriers from yeast are highly specific for ATP and ADP, respectively (Heimpel et al., 2001; Haferkamp et al., 2002), AMP was selected as the counterexchange substrate to avoid possible background from mitochondrial ATP carrier activity. Using this experimental setup, no significant ATP uptake into liposomes reconstituted with membrane proteins from ant1Δ control cells containing the empty vector was measured (Figure 4A). By contrast, liposomes reconstituted with membranes from the yeast mutant expressing either PNC1 or PNC2 showed a substantial ATP/AMP counterexchange activity (Figures 4B and 4C, closed symbols). In the absence of AMP, no ATP uptake was observed (Figures 4B and 4C, open symbols). The transport rates obtained by both Arabidopsis proteins were comparable to those observed for ANT1 (Figure 4D), further emphasizing that both PNC1 and PNC2 are able to catalyze an ATP/AMP exchange, like their putative yeast ortholog.

Figure 4.

Time Kinetics of ATP Uptake into Proteoliposomes Harboring Recombinant PNC1, PNC2, and ANT1 Proteins.

Liposomes were reconstituted with the membrane fraction from ant1Δ mutant yeast containing the corresponding expression constructs for PNC1 (B), PNC2 (C), and ANT1 (D) or with the empty vector as a control (A). The uptake of radioactively labeled [α-³²P]ATP (0.2 mM) was measured into proteoliposomes preloaded internally with 30 mM AMP (closed symbols) or in the absence of AMP as a counterexchange substrate (open symbols). ATP transport was terminated by loading the proteoliposomes onto anion-exchange columns. Radioactivity of the column flow-through in the sample was quantified by liquid scintillation counting. The means of a representative experiment with three technical replicates ± se are shown.

Since we could not directly test whether yeast-expressed Arabidopsis PNC1 and PNC2 are able to catalyze the exchange of ADP with ATP, due to the background activity of the yeast mitochondrial ATP/ADP carrier, we also expressed PNC1 and PNC2 in Escherichia coli, which does not show endogenous ATP carrier activities (see Supplemental Figure 2A online). Uptake of radiolabeled [α-³²P]ADP into proteoliposomes preloaded with either ATP, ADP, or AMP was measured for reconstituted recombinant proteins purified from E. coli. ADP was efficiently taken up in the presence of the internal substrates ATP, ADP, or AMP, but not in their absence (see Supplemental Figure 2B online). Thus, PNC1 and PNC2 represent nucleotide carriers with specificity for ATP, ADP, and AMP in a strict counterexchange mode, like the yeast ANT1 transporter.

PNC1 and PNC2 Are Highly Specific for ATP, ADP, and AMP

A detailed knowledge of substrate affinity and specificity is required to understand the function of a transport protein in vivo. To determine the apparent K_m values of PNC1 and PNC2 for ATP, the initial rate of [α-³²P]ATP uptake into AMP-preloaded proteoliposomes was measured at different external ATP concentrations. PNC1 and PNC2 exhibit an apparent K_m value for ATP of 83 μM (± 15 se, n = 3) and 150 μM (± 36 se, n = 3), respectively, whereas ANT1 has an apparent K_m value of 54 μM (± 13 se, n = 3) for ATP (Table 1). To assess the substrate specificity of the ATP carriers, the inhibition constant (K_i) was determined. For this, the import of radiolabeled ATP into proteoliposomes preloaded with AMP was measured in the presence of nonlabeled ADP, AMP, and inorganic ortho-phosphate (P_i). For PNC1, PNC2, and ANT1, both ADP and AMP competitively inhibited the uptake of ATP, whereas P_i exerted no significant inhibitory effect. In addition, the obtained apparent K_i values shown in Table 1 exhibit similar K_i/K_m ratios for all three recombinant proteins. The estimated K_i values of ADP are twofold to threefold higher than their corresponding K_m values for ATP. In the case of the apparent K_i value for AMP, the half-maximal inhibition of ATP transport was observed at even higher concentrations.

Table 1.

Apparent K_m (ATP) and K_i Values of Recombinant PNC1, PNC2, and ANT1 for Various Metabolites

Substrate	PNC1	PNC2	ANT1
	Km (μM)	Km (μM)	Km (μM)
ATP	83 (±15)	150 (±36)	54 (±13)
	Ki (μM)	Ki (μM)	Ki (μM)
ADP	211 (±52)	254 (±65)	171 (±51)
AMP	266 (±66)	427 (±136)	199 (±63)
Pi	>100,000	>100,000	>100,000

All experiments were performed with liposomes preloaded with 30 mM AMP. The data represent the mean values ± se from three experiments.

Expression Pattern of PNC1 and PNC2

As a first step to elucidate the role of PNC1 and PNC2 in planta, their expression patterns were investigated by searching the publicly available microarray databases using the Arabidopsis eFP Browser (Winter et al., 2007). Both genes are ubiquitously expressed in Arabidopsis throughout all developmental stages (see Supplemental Figure 3 online). Notably, PNC2 is highly expressed in tissues where β-oxidation plays an important role (Baker et al., 2006), such as flowers, stamens, pollen, developing seeds, and senescent leaves. In addition, the Arabidopsis coresponse database CSB.DB was searched for genes that are coexpressed with PNC1 and PNC2 (Steinhauser et al., 2004). Most of the genes coexpressed with PNC1 and PNC2 are part of the glyoxylate cycle and gluconeogenesis, processes known to be primarily associated with β-oxidation (see Supplemental Table 1 online). Importantly, one of the highest-ranking coexpressed genes (Spearman rank correlation coefficient > 0.7) for both PNC1 and PNC2 is the long-chain acyl-CoA synthetase LACS7 that plays an important role in the activation of fatty acids during seedling establishment (Fulda et al., 2004).

Generation of Transgenic Arabidopsis Plants Displaying Impaired Peroxisomal ATP Import

To elucidate the impact of peroxisomal ATP transporters in vivo, T-DNA insertion lines for PNC1 (SAIL_303H02, referred to as pnc1-1) and PNC2 (SALK_014579, referred to as pnc2-1) were isolated from the publicly available collections (see Supplemental Figure 4A online). No abberant phenotype was observed under normal growth conditions in homozygous pnc1-1 and pnc1-2 plants. Importantly, compared with the wild type, seedling development was not delayed in either of the homozygous insertion lines, indicating that storage oil mobilization was not compromised (see Supplemental Figure 4B online). However, in the pnc1-1 mutant, the T-DNA is located in the last exon of the PNC1 gene, leading to the expression of a truncated transcript that encodes a protein lacking the last 19 amino acids of the C terminus. We expressed recombinant PNC1Δ19C protein in the yeast knockout ant1Δ as described above and found that the truncated transport protein retains modest protein-mediated ATP uptake activity when reconstituted into liposomes (see Supplemental Figure 5 online). Thus, the Arabidopsis T-DNA insertion line pnc1-1 does not represent a loss-of-function mutant, although the T-DNA had inserted into an exon of the corresponding gene.

To produce Arabidopsis lines with impaired ATP import into peroxisomes mediated by PNC1 and PNC2, plants repressing the expression of both transporter genes were generated using an RNAi approach (Wesley et al., 2001) with an ethanol-inducible promoter AlcA (Roslan et al., 2001; Sweetman et al., 2002). A 480-bp cDNA fragment of PNC1 was introduced into an intron-spliced hairpin RNA construct (see Supplemental Figure 6 online). The selected cDNA fragment of PNC1 is highly similar (86% identity) to the corresponding region of PNC2, thus enabling simultaneous silencing of both genes from a single construct. The ethanol-inducible RNAi construct (AlcA_pro:PNC1RNAi) was introduced into Arabidopsis plants by Agrobacterium-mediated transformation. Several independent transgenic lines (T1 generation) were obtained after kanamycin selection and integration of the T-DNA was confirmed by genomic PCR. After selfing, plants of the T2 generation were screened for homozygous iPNC1 iPNC2 lines by segregation analysis of the nptII resistance gene, further propagated, and named iPNC12.

To determine the degree of simultaneous repression of PNC1 and PNC2 mRNA in these transgenic lines, we assessed PNC1 and PNC2 steady state transcript levels using RT-PCR. RNA from independent iPNC1/2 lines grown in the presence or absence of ethanol was isolated from 5-d-old seedlings. A strong simultaneous suppression of PNC1 and PNC2 was observed in two independent RNAi lines after ethanol induction, consistent with their high nucleotide sequence identity, whereas the expression of the third gene, At2g39970, was not affected (Figure 5A). The results obtained by RT-PCR for iPNC1/2 lines 1 and 2 were independently verified by quantitative RT-PCR (qRT-PCR) using two biological replicates (Figure 5B). PNC1 and PNC2 are ∼15- and 11-fold downregulated after ethanol induction in iPNC1/2 line 1 when compared with empty vector control plants. An even stronger reduction of the relative PNC mRNA amounts was observed in the second iPNC1/2 line. The relative transcript levels of both PNC genes showed a slight repression (twofold and fourfold reduction in lines 1 and 2, respectively) prior to induction when compared with control plants. Roslan et al. (2001) described previously that growth on Murashige and Skoog (MS) agar plates might trigger endogenous ethanol production by anaerobic fermentation due to low oxygen conditions. This could explain the slight decrease of target genes in iPNC1/2 lines prior to ethanol application relative to the control. Importantly, the transgenic lines showed growth defects only after ethanol induction (Figure 6), in parallel with the strong simultaneous repression of both PNC1 and PNC2 under these conditions (Figure 5B).

Figure 5.

Analysis of the PNC1 and PNC2 Transcripts in Arabidopsis Seedlings of PNC1/2 RNAi and Control Plants.

Seedlings of two representative PNC1/2 RNAi lines and seedlings transformed with the empty vector (control) were grown for 5 d on half-strength MS agar plates in the presence (+) of 0.2% (v/v) ethanol or in the absence (−) of ethanol.

(A) For the RT-PCR analysis, gene-specific primers were used to amplify PNC1, PNC2, At2g39970, and ACT7 from total RNA. RNA gel electrophoresis was performed to verify that equal amounts of total RNA were used for the cDNA synthesis by staining with ethidium bromide.

(B) Relative transcript levels of PNC1 and PNC2 were determined by qRT-PCR using gene-specific primers binding at the exon-exon junction and the 3′ untranslated region. The reference gene, TIP41-like gene (At4g34270), was used for normalization of gene transcript levels among all samples. The data represent the mean normalized expression (MNE) values ± se of three technical replicates from two independent experiments.

Figure 6.

Postgerminative Seedling Growth of Arabidopsis PNC1/2 RNAi Plants.

(A) Seedlings of two representative PNC1/2 RNAi lines and seedlings transformed with the empty vector (control) were grown for 15 d vertically on half-strength MS agar plates supplemented with 1% (w/v) sucrose (+Suc) or lacking sucrose (−Suc) in the presence of 0.2% (v/v) ethanol (+EtOH) or in the absence of ethanol (−EtOH). Seedlings were rearranged on fresh plates and photographed. Bars = 1 cm.

(B) Effect of exogenous sucrose on hypocotyl growth in the dark. Seedlings of two representative PNC1/2 RNAi lines, wild type, and plants transformed with empty vector (control) were plated on half-strength MS agar plates supplemented with 0.2% (v/v) ethanol in the absence (−Suc) or presence of 0.8% (w/v) sucrose (+Suc). Following 24 h of exposure to light to permit germination, plates were incubated for 5 d in the dark. Values are means ± se of measurements from 20 to 30 hypocotyls.

The PNC1/2 RNAi Lines Are Compromised in Seedling Establishment

The role of PNC1 and PNC2 in β-oxidation during postgerminative growth was investigated by examining seedling establishment of homozygous iPNC1/2 plants. Mutants defective in import (cts-1), activation (lacs6/7), and degradation of fatty acids (β-oxidation mutants, such as 3-ketoacyl-CoA thiolase mutant [kat2-1]/peroxisome defective 1 [ped1]) are unable to mobilize seed storage oil and seedling growth is arrested (Hayashi et al., 1998, 2002; Germain et al., 2001; Footitt et al., 2002; Fulda et al., 2004). However, the presence of an alternative carbon source, such as sucrose, can rescue the phenotype.

In our study, seeds of several independent iPNC1/2 lines were grown on agar plates containing 0.2% (v/v) ethanol under controlled growth conditions. To assess whether sucrose was required for seedling establishment, the seeds were placed on media either with or without sucrose. A defect in seedling growth upon ethanol-induction of RNAi expression was reproducibly observed in the absence of sucrose (Figure 6A). Growth and development were arrested in iPNC1/2 seedlings compared with control plants carrying the empty vector. In contrast with other known mutants that exhibit the sucrose-dependent phenotype described above, iPNC1/2 treated with ethanol did develop seedlings on plates with sucrose but grew significantly slower than the control plants (delayed seedling growth). In the absence of the inducer ethanol, the growth rates and phenotypes of iPNC1/2 were indistinguishable from the wild type (Figure 6A). Similarly, dark-grown hypocotyls of PNC1/2 RNAi seedlings were much shorter than those of wild-type or empty vector controls when grown in the absence of sucrose, indicative of their inability to mobilize storage oil for hypocotyl elongation (Penfield et al., 2004). Sucrose increased the elongation of RNAi hypocotyls but did not restore them to the same length as the wild type, indicating other functions for PNC1 and 2, in addition to storage oil mobilization (Figure 6B).

Although some mutants in which β-oxidation is severely impaired exhibit a defect in germination (Pinfield-Wells et al., 2005; Pracharoenwattana et al., 2005; Footitt et al., 2006), germination efficiency was not affected in the iPNC1/2 repression lines as shown for the kat2-1 and cts-1 mutant (see Supplemental Figure 7 online). Two independent iPNC1/2 lines, line 1 and line 2, that showed a severe ethanol-inducible phenotype were selected for further studies.

Transgenic Lines with an Impaired ATP Import into Peroxisomes Accumulate Fatty Acids and Fatty Acyl-CoA and Retain Oil Bodies

In wild-type plants, TAG stored in oil bodies is degraded via lipolysis and peroxisomal β-oxidation and converted to sucrose to fuel postgerminative seedling establishment (Cooper and Beevers, 1969). To examine whether the arrested seedling growth of the PNC1/2 RNAi lines is linked to impaired lipid breakdown, levels of fatty acids and acyl-CoA esters were measured in 5-d-old seedlings of two independent iPNC1/2 lines treated with ethanol, and the presence of oil bodies in seedling tissues was investigated by microscopy. The total content of fatty acids was increased eightfold in 5-d-old iPNC1/2 seedlings compared with seedlings transformed with the empty vector (control). As shown in Figure 7A, C16:0, C18:0, C18:1. C18:2, C18:3, C20:1, C20:2, and C20:3 fatty acids accumulated in the repression lines. In particular, the content of eicosenoic acid (C20:1), which is considered to be diagnostic of TAG (Lemieux et al., 1990), was 10 to 20 times higher in iPNC1/2 seedlings than in control plants. Consistent with the accumulation of fatty acids, a strong increase in the quantity of long/very-long-chain acyl-CoA esters (C20:0, C20:1, and C22:1) was observed in iPNC1/2 seedlings in comparison with the controls (Figure 7B). In the absence of ethanol, little difference in either the level of fatty acids or acyl-CoAs was observed in iPNC1/2 seedlings compared with the empty vector control (see Supplemental Figure 8 online), indicating that ethanol-induced silencing of PNC1 and PNC2 gene expression causes the accumulation of fatty acids and acyl-CoA esters.

Figure 7.

Fatty Acid and Acyl-CoA Profile and Visualization of Hypocotyl Oil Bodies in 5-d-Old Arabidopsis Seedlings of PNC1/2 RNAi and Control Plants.

(A) and (B) Seedlings of PNC1/2 RNAi line 1 (gray bars) and line 2 (black bars) and seedlings transformed with the empty vector control (white bars) were grown for 5 d on half-strength MS agar plates supplemented with 0.5% (w/v) sucrose in the presence of 0.2% (v/v) ethanol. The values are means ± se of the mean ([A] three replicates of 50 seedlings; [B] five replicates of 10 to 25 seedlings).

(C) Seedlings of a representative PNC1/2 RNAi line (line 1; A and B) and the wild type (Col-0; C and D) were grown for 5 d on half-strength MS agar plates supplemented with 0.5% (w/v) sucrose in the presence of 0.2% (v/v) ethanol. Hypocotyl cells were visualized by microscopy. A and C, bright-field images; B and D, Nile Red fluorescence. Bars = 35 μm.

Examination of seedlings stained with the lipophilic dye, Nile Red, revealed that, while oil bodies were largely absent in wild-type hypocotyls 5 d after germination, the RNAi seedlings exhibited a marked retention of oil bodies (Figure 7C). Taken together, the retention of oil bodies and the accumulation of fatty acids and acyl CoAs indicate that both lipolysis and β-oxidation are impaired in iPNC1/2 seedlings.

The PNC1/2 RNAi Lines Are Resistant to Root Growth Inhibition in the Presence of IBA and 2,4-DB

The requirement of peroxisomal ATP import for β-oxidation of compounds other than fatty acids was assessed by screening the response of pnc1/2 RNAi lines to IBA and 2,4-DB. In wild-type plants, both compounds are activated with CoA in an ATP-dependent manner and then converted by one cycle of β-oxidation to the active auxin indole-3-acetic acid and the herbicide 2,4-D, respectively, which severely inhibit root and hypocotyl growth (Hayashi et al., 1998; Zolman et al., 2000; Rashotte et al., 2003). Primary root elongation was analyzed in PNC1/2 RNAi seedlings grown on agar plates supplemented with 0.5% (w/v) sucrose, 0.2% (v/v) ethanol, and either 30 μM IBA or 0.8 μM 2,4-DB. As shown in Figure 8A, the PNC1/2 RNAi seedlings exhibited a moderate level of resistance to IBA and 2,4-DB compared with wild-type seedlings and empty vector controls. We also investigated auxin responses of dark-grown hypocotyls from seedlings of two PNC1/2 RNAi lines, empty vector controls, and the wild type. Although hypocotyls are somewhat less sensitive to the application of exogenous auxins than roots (Zolman et al., 2000; Rashotte et al., 2003), both IBA and 2,4-DB inhibited elongation of wild-type and control hypocotyls. By contrast, the hypocotyls of PNC1/2 RNAi seedlings exhibited resistance to these compounds, consistent with a block in their β-oxidation (Figure 8B). These results indicate that the activation of IBA and 2,4-DB is dependent on peroxisomal ATP import catalyzed by PNC1/2.

Figure 8.

Root and Hypocotyl Growth Response to IBA and 2,4-DB.

(A) Effect of IBA and 2,4-DB on primary root elongation. Seedlings of two representative PNC1/2 RNAi lines, wild-type seedlings, and control plants carrying the empty vector (control) were grown for 5 d on half-strength MS agar plates supplemented with 0.5% (w/v) sucrose and 0.2% (v/v) ethanol in the presence of 0.8 μM 2,4-DB (black bars) or 30 μM IBA (gray bars). Relative growth is expressed as a percentage of the length of the primary root in the absence of 2,4-DB and IBA. The values are means ± se of the mean of measurements from three replicates of 15 to 20 seedlings.

(B) Effect of IBA and 2,4-DB on elongation of hypocotyls in the dark. Seedlings of two representative PNC1/2 RNAi lines and of the wild type were allowed to germinate in the light for 24 h and then grown for 5 d in the dark on half-strength MS agar plates supplemented with 0.8% (w/v) sucrose and 0.2% (v/v) ethanol in the presence of 0.8 μM 2,4-DB (black bars) or 30 μM IBA (gray bars). Relative growth is expressed as a percentage of the hypocotyl length in the absence of 2,4-DB and IBA. The values are means ± se of the mean (three replicates of 20 to 30 seedlings).

DISCUSSION

β-Oxidation plays an essential role in seedling establishment to support growth and development (Baker et al., 2006; Graham, 2008). Fatty acids released from TAG are degraded via peroxisomal β-oxidation and converted to sucrose during postgerminative growth to supply metabolic energy and carbon skeletons. However, before entering β-oxidation, fatty acids are imported into peroxisomes by CTS (Zolman et al., 2001; Footitt et al., 2002; Hayashi et al., 2002) and further activated with CoA by LACS6 and LACS7 (Fulda et al., 2004). Both steps are required for postgerminative storage oil breakdown to enable normal seedling establishment (Footitt et al., 2002; Hayashi et al., 2002; Fulda et al., 2004).

In this study, we investigated whether the initial ATP-dependent activation reaction of β-oxidation depends on the import of external ATP into peroxisomes by specific transport proteins during postgerminative growth. We identified two novel peroxisomal ATP transporters from Arabidopsis using a candidate gene approach. Based on sequence similarity to previously characterized ATP transporters from yeast peroxisomes (Palmieri et al., 2001; van Roermund et al., 2001), three members of the Arabidopsis mitochondrial carrier family encoded by the genes At2g39970, At3g05290, and At5g27520 were identified. Complementation studies of the yeast mutant deficient in peroxisomal ATP uptake revealed that At3g05290 and At5g27520 could be involved in ATP import into yeast peroxisomes. Such a transport function is required to metabolize medium-chain fatty acids and restore the yeast growth phenotype (Figure 2). Thus, both proteins (PNC1 and PNC2) are peroxisomal nucleotide carriers in Arabidopsis.

Interestingly, the third candidate, which is encoded by At2g39970 and annotated as peroxisomal membrane protein PMP38, did not complement the yeast growth phenotype, although colocalization experiments using fluorescence microscopy confirmed the peroxisomal localization in yeast (Figure 3). The peroxisomal localization of Arabidopsis PMP38 in pumpkin (Cucurbita sp) has already been demonstrated by cell fractionation and immunocytochemical analysis in an earlier study (Fukao et al., 2001). Because the abundance of the PMP38 protein was coregulated with that of the fatty acid β-oxidation enzymes during postgerminative growth, this transporter was proposed to be involved in fatty acid degradation as a putative peroxisomal ATP carrier (Fukao et al., 2001). However, PMP38 most likely has a different substrate specificity, and our results strongly indicate that PMP38 does not function in transport of ATP across the peroxisomal membrane. Further analysis to determine the transport function or in vivo role of mutants deficient in this protein has yet to be done.

Our in vitro ATP uptake studies showed that recombinant reconstituted PNC1 and PNC2 catalyze the strict counterexchange of ATP with AMP (Figure 4). This is consistent with the physiological function of the peroxisomal carrier. ATP is imported into peroxisomes, where it is converted to AMP by fatty acid activation, and AMP is then subsequently exported to the cytosol (Rottensteiner and Theodoulou, 2006). Like their putative yeast ortholog ANT1, both Arabidopsis transport proteins accept ADP as an additional substrate. Among the diverse group of eukaryotic adenylate transporters, only the Brittle1 homolog from potato (Solanum tuberosum) (Leroch et al., 2005) shares the unique substrate specificity for AMP with the peroxisomal ATP carriers. All other ATP transporters from mitochondria, plastids, and ER are highly specific for ATP and ADP and do not transport AMP (Heimpel et al., 2001; Haferkamp et al., 2002; Leroch et al., 2008). In the physiological context, counterexchange of ATP with AMP across the peroxisomal membrane is appropriate because during β-oxidation LACS proteins consume ATP and produce AMP (Fulda et al., 2004). Since several protein kinases and ATP-dependent chaperones have recently been identified by proteomics in the peroxisomal matrix (Reumann et al., 2004, 2007), which requires ATP and generates ADP, an ATP/ADP counterexchange activity is needed. The determination of inhibition constants (K_i) revealed that peroxisomal ATP transporters from Arabidopsis and S. cerevisiae exhibit a two to three times higher apparent K_i for AMP than for ATP (Table 1), indicating that AMP only inhibits the import of ATP into peroxisomes at high concentrations of AMP. Especially in sink tissues, such as cotyledons during postgerminative growth, the cellular ATP level is low and the adenosine monophosphate kinase (adenylate kinase) provides additional ATP by converting two ADP molecules into ATP and AMP, resulting in a low ATP/AMP ratio (Stitt et al., 1982; Geigenberger and Stitt, 2000; Igamberdiev and Kleczkowski, 2006). Under these conditions, the import of ATP into peroxisomes is favored due to the kinetic properties of the PNC proteins.

Since PNC1 and PNC2 exhibit similar biochemical transport properties and tissue expression profiles (see Supplemental Figure 3 online), it is likely that these peroxisomal transporters are functionally redundant. To investigate the in vivo role of ATP import into peroxisomes, we initiated a strategy to construct a double knockout for both PNC genes. As the pnc1-1 T-DNA insertion line did not lead to a dysfunctional PNC1 protein and a second allele is not available in any of the publicly accessible collections, we generated ethanol-inducible pnc1/2 RNAi plants to suppress the peroxisomal ATP import activity in Arabidopsis. The repression of both PNC genes after ethanol induction was confirmed at the transcript level in these RNAi lines. The physiological and biochemical analyses of iPNC1/2 plants strongly indicate that the storage oil breakdown in peroxisomes during postgerminative growth requires an external supply of ATP, suggesting that PNC1/2 activity is critical for mobilizing fatty acids through β-oxidation.

The iPNC1/2 plants are defective in seedling growth and development when induced with ethanol (Figure 6). Supply of sucrose restored seedling growth (Figure 6), which is also the case for several other known mutants affected in the import of fatty acids into peroxisomes (cts-1), activation of fatty acids (lacs6/7), and in fatty acid β-oxidation, such as the kat2/ped1 mutant, which is deficient in thiolase activity (Hayashi et al., 1998, 2002; Germain et al., 2001; Footitt et al., 2002; Fulda et al., 2004). Intriguingly, however, the observation that sucrose could not completely rescue the arrested growth phenotype of the iPNC1/2 seedlings in the presence of ethanol (Figure 6) indicates a pleiotropic effect of impaired peroxisomal ATP import beyond activation of fatty acids. We assume that hitherto unknown ATP-dependent processes within peroxisomes, besides the activation of fatty acids, require ATP and depend on external ATP import into peroxisomes mediated by PNC1 and PNC2 (Reumann et al., 2004, 2007).

A common feature of β-oxidation mutants, including the cts-1 mutant and the lacs6 lacs7 double mutant, is impaired fatty acid breakdown, associated with retention of storage oil–specific fatty acids in the cotyledons and hypocotyls (Baker et al., 2006; Graham, 2008). After ethanol induction, iPNC1/2 seedlings also accumulated fatty acids, such as C18:2, C18:3, and C20:1 (Lemieux et al., 1990; Millar and Kunst, 1999). This indicates that ATP import is required for complete TAG breakdown to support seedling establishment. A similar fatty acid profile as observed in the iPNC1/2 lines was reported for seedlings defective in the peroxisomal ABC transporter CTS and in fatty acid activating enzymes LACS6/7 (Footitt et al., 2002; Fulda et al., 2004). Together with the results from the cts-1 and lacs6 lacs7 mutants, it can be concluded that storage oil–derived fatty acid breakdown is compromised in the PNC1/2 RNAi lines, causing the defect in seedling establishment. Interestingly, the marked increase in corresponding acyl-CoA esters in the iPNC1/2 seedling is also observed in the cts-1 and lacs6 lacs7 mutants. It is likely that the CoA pool serves as a sink for nonesterified fatty acids released by lipolysis of TAG in these mutants. Because an accumulation of free fatty acids in the cytosol is potentially deleterious, they have to be diverted to other pools. While long-chain fatty acids can be incorporated into membrane lipids, very-long-chain fatty acids are converted to CoA esters by an unknown cytosolic or microsomal acyl-CoA synthetase (Larson et al., 2002). The Arabidopsis genome encodes a large family of acyl-activating enzymes with diverse substrate specificity and subcellular localization (Shockey et al., 2003; Kienow et al., 2008). Although not supported by experimental evidence to date, it is likely that one of the currently uncharacterized AAE members is involved in the CoA transfer.

At first sight, the retention of oil bodies and accumulation of TAG-derived fatty acyl-CoAs appears counterintuitive. However, this phenotype, shared by other β-oxidation mutants, is consistent with a combination of limited lipolysis and a defect in β-oxidation. It has been proposed that accumulation of acyl-CoAs in mutant seedlings leads to a block in storage oil mobilization from the oil body, perhaps by inhibition of lipolysis (Germain et al., 2001; Graham et al., 2002; Graham, 2008). In agreement with this, activity of neutral TAG lipase from castor bean (Ricinus communis) is inhibited by oleoyl-CoA (Hills et al., 1989).

Mutants in which β-oxidation is severely impaired, such as kat2/ped1, cts-1, and the peroxisomal citrate synthase double mutant, csy2 csy3, exhibit a defect not only in postgerminative seedling establishment, but also in germination (Hayashi et al., 1998; Footitt et al., 2002, 2006; Pinfield-Wells et al., 2005; Pracharoenwattana et al., 2005). Interestingly, the germination rate of PNC1/2 RNAi seeds was not compromised (see Supplemental Figure 7 online). This indicates that the unknown function of β-oxidation, which is required for germination, is not significantly affected in PNC1/2 RNAi seeds, possibly reflecting the incomplete suppression of these genes in the transgenic lines. In this context, it should be noted that the lacs6 lacs7 double mutant also germinates, albeit more slowly than the wild type (Fulda et al., 2004; Footitt et al., 2006).

The iPNC1/2 plants are partially resistant to the inhibitory effect of IBA and 2,4-DB on elongation of primary roots and dark-grown hypocotyls, suggesting that the activation of the corresponding precursors and hence their subsequent β-oxidation is dependent on peroxisomal ATP import. However, the degree of resistance is lower than that observed in known β-oxidation mutants, including cts-1, kat2/ped1, and acx3 (Hayashi et al., 1998; Zolman et al., 2000, 2001; Footitt et al., 2002; Rylott et al., 2003). A residual ATP import activity in the iPNC1/2 lines might be sufficient to channel low levels of IBA and 2,4-DB into β-oxidation for conversion to indole-3-acetic acid and 2,4-D, resulting in partial inhibition of root elongation. Root growth of the lacs6 lacs7 double mutant is inhibited in the presence of IBA and 2,4-DB, indicating that LACS6 and LACS7 are only specific for long-chain fatty acids (Fulda et al., 2004). Thus, an unknown acyl-activating enzyme must catalyze the activation of the IBA and 2,4-DB precursor. To date, it was not known whether the protein(s) catalyzing IBA/2,4-DB activation is (are) localized inside the peroxisomes or whether CTS imports the activated precursor into peroxisomes. Our results suggest that the activation depends on peroxisomal ATP import, implying that the activation reaction takes place within peroxisomes and not in the surrounding cytosol.

Taken together, the data presented here are consistent with the hypothesis that the diverse substrates of the β-oxidation, such as fatty acids, IBA, and 2,4-DB, are transported across the peroxisomal membrane by CTS as free acids and then activated by ATP-dependent reactions inside peroxisomes, which are supplied with external ATP by PNC1/2. PNC1 and PNC2 represent the exclusive route for peroxisomal ATP supply because Arabidopsis plants defective in peroxisomal ATP import are compromised in storage oil mobilization and resistance to root growth inhibitors as a consequence of a limiting supply of ATP within peroxisomes. However, it remains a formal possibility that at least some β-oxidation substrates are imported by CTS as CoA esters and require an intraperoxisomal ATP supply because they are deesterified at some stage during the β-oxidation spiral. This is consistent with the identification of three peroxisomal acyl-activating enzymes of the AAE family, which are involved in JA biosynthesis (Schneider et al., 2005; Theodoulou et al., 2005; Koo et al., 2006; Kienow et al., 2008). The JA precursor 12-oxophytodienoic acid is imported into the peroxisome by two routes, one of which requires CTS (Theodoulou et al., 2005), where it is reduced and further converted to JA by three consecutive rounds of β-oxidation (Schaller et al., 2004; Baker et al., 2006; Delker et al., 2007). The observation that the peroxisomal AAE enzymes exhibit specificity for different JA biosynthetic intermediates (12-oxophytodienoic acid, OPC:8, OPC:6, and OPCC:4) indicates that the CoA moiety is removed at several alternate steps in the pathway. Accordingly, peroxisomes of all organisms contain a range of thioesterases, one of which has been shown to play an essential role in the breakdown of short straight-chain and branched-chain fatty acids in yeast (Maeda et al., 2006). Two peroxisomal thioesterases have been reported in Arabidopsis, but their physiological function is not yet clear (Tilton et al., 2004). It has been suggested that removal of the CoA moiety may play a role in the regulation of β-oxidation and prevent sequestration of free CoA due to buildup of unusual CoA-esterified species that are inefficiently metabolized (Hunt and Alexson, 2008). Thus, the requirement of several physiological processes for peroxisomal ATP import could also reflect the need to reactivate intermediates to enter β-oxidation.

Knowledge of metabolite transport across the peroxisomal membrane is scarce. Multiple peroxisomal pathways are connected with metabolic pathways in other cellular compartments, implying that a large variety of small molecules must be transported across the peroxisomal membrane. The characterization of PNC1 and PNC2 reported here is a step toward obtaining a mechanistic and molecular understanding of peroxisomal transporters in plants. The PNC1/2 RNAi lines established in this work provide an opportunity to further investigate whether, in addition to the initial activation of β-oxidation substrates, yet to be defined reactions demand ATP in plant peroxisomes.

METHODS

Materials

Chemicals were purchased from Sigma-Aldrich, and anion exchange resin Dowex AG1-X8 Resin was supplied by Bio-Rad Laboratories. Radiochemical [α-³²P]ATP was provided by GE Healthcare. Reagents and enzymes for recombinant DNA techniques were obtained from Invitrogen, New England Biolabs, and Promega.

Plant Material and Growth Conditions

Wild-type Arabidopsis thaliana (ecotype Col-0) and the Arabidopsis T-DNA insertion lines, SAIL_303H02 (pnc1-1) and SALK_014579 (pnc2-1), were obtained from the ABRC at the Ohio State University. Seeds of Arabidopsis plants were surface-sterilized, stratified for 4 d at 4°C, and germinated on 0.8% (w/v) agar-solidified half-strength MS medium supplemented with 1% (w/v) sucrose. Unless stated otherwise, plants were grown at 22°C (70% humidity) in a 16-h-light/8-h-dark cycle in growth chambers (100 μmol m⁻² s⁻¹ light intensity). Selection of the transgenic Arabidopsis plants was performed with kanamycin at a concentration of 50 μg/mL. Selected plants were transferred to soil for further growth in a chamber under the same conditions. For the analysis of postgerminative seedling growth, plants were grown on half-strength MS agar plates supplemented or not with 1% (w/v) sucrose in the presence or absence of 0.2% (v/v) ethanol at 16-h-light/8-h-dark cycles and 100 μmol m⁻² s⁻¹ light intensity. To study the response to 2,4-DB and IBA and for lipid analysis, seeds were plated on half-strength MS agar plates supplemented with 0.5% (w/v) sucrose ± 30 μM IBA or 0.8 μM 2,4-DB (added from stocks dissolved in ethanol) in the presence of 0.2% (v/v) ethanol at constant illumination (150 μmol m⁻² s⁻¹). After 5 d, seedlings were photographed and roots measured using MetaMorph software (Molecular Devices). For measuring the hypocotyl growth in the dark, seeds were plated on half-strength MS agar supplemented with 0.2% (v/v) ethanol. Sucrose (0.8% [w/v]), 2,4-DB (final concentration 0.8 μM), and/or IBA (final concentration 30 μM) were added as indicated. Following 2 d of stratification and 24 h of exposure to light to induce germination, plates were kept in the dark for 5 d. Hypocotyls were photographed and measured using MetaMorph software (Molecular Devices). For subcellular colocalization studies, Nicotiana benthamiana was grown on soil in a greenhouse under a light regime of 8 h of darkness and 16 h of light.

Phylogenetic Analyses

The sequences were retrieved from the GenBank database of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov). The BLAST program (Altschul et al., 1990) was used to search for putative peroxisomal ATP transport proteins in Arabidopsis based on sequence similarity to the previously identified ANT1 from Saccharomyces cerevisiae (Palmieri et al., 2001). The amino acid sequences were aligned by the CLUSTALW program (http://www.ebi.ac.uk/Tools/clustalw) using default settings. The alignment layout was performed with the GeneDoc program (http://www.nrbsc.org/gfx/genedoc/index.html).

DNA Sequencing and Cloning Procedures

The clones U13126 and U18363 containing the full-length ORF cDNA of At3g05290 and At5g27520 were obtained from the ABRC. The corresponding clone for At2g39970 (pda.00682) was provided by Riken BioResource Center.

For general cloning, DNA sequences were amplified by PCR using proofreading polymerase (Platinum Pfx polymerase; Invitrogen). The PCR primers designed for this study are listed in Supplemental Table 2 online. The PCR products were subcloned into pGEM-T Easy (Promega), and the sequences were verified by sequencing (Research Technology Support Facility, Michigan State University). Subsequent cloning steps were performed using standard molecular techniques (Sambrook et al., 1989).

Yeast Complementation Studies

The ant1Δ knockout mutant deficient in the peroxisomal ATP transporter ANT1 was purchased from Open Biosystems. For complementation studies, the ORFs of At2g39970, At3g05290, and At5g27520 were amplified from the cDNA clones by PCR as described above. The following gene-specific primers were used: NL16/NL17 for At2g39970, NL64/NL65 for At3g05290, and NL97/NL106 for At5g27520. The PCR fragments were cloned via XhoI and BamHI into yeast expression vector pDR195 under the control of a constitutive promoter of the plasma membrane H⁺-ATPase (PMA1_pro) from S. cerevisiae (Rentsch et al., 1995). In addition, the coding sequence of ANT1 was PCR-amplified from genomic DNA of the lab yeast strain INVSc1 (Invitrogen) using the primers NL55 and NL25, and a yeast expression construct was obtained according to the cloning strategy described above.

The ant1Δ strain was cultivated using standard YPD media according to Sambrook et al. (1989) and transformed with the expression constructs using the lithium chloride method (Schiestl and Gietz, 1989). The transformants were selected on synthetic complete minimal medium lacking uracil (SC-U). For yeast complementation studies, the transformed knockout mutants were analyzed on SC-U agar plates using lauric acid as the sole carbon source (Palmieri et al., 2001).

Subcellular Localization of YFP Fusions in Yeast

For localization studies in yeast, a pDR195-based expression vector was generated containing EYFP for C-terminal fusion of the gene of interest. The EYFP sequence was obtained from the pEYFP-C1 vector (BD Biosciences Clontech) by PCR using primers NL260/NL261, subcloned into pBluescript II SK (Stratagene), and then introduced into pDR195. The resulting construct was named pNL9. The ORFs of At3g05290 and At5g27520 were amplified without stop codons by PCR using the primer pairs NL62/NL63 and NL93/NL94, respectively, and inserted into pNL9, leading to proteins that are tagged at the C terminus with EYFP. In the case of At2g39970, the ORF was cloned into the pEFYP-C1 vector via BamHI and XhoI restriction sites, which were introduced by PCR using primers NL7 and NL8, respectively. The EYFP-At2g39970 fusion was inserted into pDR195 using blunt-end ligation. To generate the yeast expression vector harboring the peroxisomal marker, a Ser-Lys-Leu motif, known as PTS1, and a stop codon were added to the end of the CFP sequence from the pECFP-N1 vector (BD Biosciences Clontech) by PCR using primers NL254/NL255. The CFP-PTS1 sequence was then cloned into yeast expression vector pACTII (BD Biosciences Clontech) under the control of a constitutive yeast alcohol dehydrogenase promoter (ADH_pro).

The resulting EYFP fusion constructs and the peroxisomal marker were cotransformed into the ant1Δ mutant using the lithium chloride method (Schiestl and Gietz, 1989). The transformed yeast cells were selected on SC medium in the absence of uracil and leucine. The expression of the fusion proteins in yeast was analyzed by epifluorescence microscopy.

Transient Expression of YFP Fusions in Tobacco Epidermal Leaf Cells

The vector pNL3 derived from pUC18 (New England Biolabs) was designed by introducing the cauliflower mosaic virus 35S promoter (35S_pro), the ORF of EFYP for fusion to the C terminus of ORFs At3g05290 and At5g27520, and the polyadenylation site of the ribulose-1,5-bis-phosphate carboxylase/oxygenase small subunit (rbcs_ter) from Pisum sativum. As described for yeast EYFP fusion constructs, the ORFs of At3g05290 and At5g27520 were cloned via KpnI into pNL3. The resulting expression cassettes were then inserted via EcoRI and a blunt-end site into the binary plant vector pCAMBIA3301 (CAMBIA). In the case of At2g39970, the ORF was PCR-amplified using the primer set NL7/NL8 and cloned into pEFYP-C1 via BamHI and XhoI. The sequence encoding the EYFP-At2g39970 fusion protein was inserted into pCAMBIA3301 using the blunt-end site. To construct the plant vector harboring the peroxisomal marker, the EYFP sequence of the pNL3 vector was exchanged with CFP-PTS1, which was amplified from the pECFP-N1 vector (BD Biosciences Clontech) and modified by PCR using NL254 and NL255 primers to introduce the PTS1 signal and a stop codon. The expression cassette of the peroxisomal-targeted marker was subsequently cloned into the plant vector pPZP211 (Hajdukiewicz et al., 1994).

Agrobacterium tumefaciens strain GV3101 (Koncz and Schell, 1986) containing the expression cassette of EYFP fusions was coinfiltrated with those transformed with the peroxisomal marker construct into leaves of N. benthamiana (Romeis et al., 2001). After 72 h, the epidermal cell layers from the tobacco leaves were peeled and analyzed for transient coexpression of fusion proteins by fluorescence microscopy.

Fluorescence Microscopy

Coexpression of the EYFP fusion proteins and the peroxisomal marker (CFP-PTS1) in yeast or tobacco epidermal cells was examined using a Zeiss Axio Imager microscope (Carl Zeiss) with either a YFP filter (excitation, 500 ± 12 nm; emission, 542 ± 13.5 nm) or a CFP filter (excitation, 438 ± 12 nm; emission, 483 ± 16 nm). Digital images were processed using the AxioVision software (Carl Zeiss).

Reconstitution of Transport Activities into Liposomes

To reconstitute recombinant proteins of PNC1, PNC2, and S. cerevisiae ANT1, the corresponding ORFs were expressed in ant1Δ using the pDR195-based constructs designed in this study for yeast complementation assays. Fifty milliliters of SC-U medium containing 2% (w/v) glucose was inoculated with an overnight culture of ant1Δ harboring the various yeast expression constructs and cultured to an OD₆₀₀ of 0.4. The yeast cells were grown for 6 h aerobically at 30°C. Control cultures with an empty vector were processed in parallel. Harvest and enrichment of yeast membranes with heterologously expressed PNC1, PNC2, and ANT1 proteins were achieved according to Bouvier et al. (2006).

Recombinant PNC1 and PNC2 proteins were also obtained by expression in the Escherichia coli strain BL21-CodonPlus (Stratagene). The gene-specific primers NL64/65 and NL97/106 were used to amplify and clone the corresponding ORFs into the pET16b expression vector (Merck Biosciences, using the XhoI and BamHI restriction sites. Protein production was induced with isopropyl-β-d-1-thiogalactopyranoside, and transformed cells were harvested after 2 h. Inclusion bodies were purified as described by Schroers et al. (1997). The expression of the N-terminal His-tagged proteins was confirmed by SDS gel blot analysis using a monoclonal anti-penta-His antibody (Qiagen).

The yeast membrane fractions or E. coli inclusion bodies were reconstituted into liposome suspension by the freeze-thaw procedure (Kasahara and Hinkle, 1976). The liposomes had been prepared from acetone-washed l-α-phosphatidylcholine (PC) by sonication (5 min on ice, Branson Sonicator 250, output 2, duty cycle 30%) to a final concentration of 2% (w/v) in 100 mM Tricine-KOH, pH 7.8, 20 mM potassium gluconate, and 30 mM AMP (unless stated otherwise). For uptake in the absence of internal substrate, phospholipids were resuspended in 100 mM Tricine-KOH, pH 7.8, and 50 mM potassium gluconate. After thawing, the proteoliposomes were pulsed 15 times to yield unilamellar vesicles. Unincorporated counterexchange substrates were removed by passing the proteoliposomes over PD10 gel filtration columns (GE Healthcare) that were equilibrated with 100 mM sodium acetate, 40 mM potassium gluconate, and 10 mM Tricine-KOH, pH 7.2.

The uptake assay was started by adding radiolabeled [α³²P]ATP or [α³²P]ADP, which was synthesized from [α³²P]ATP according to Tjaden et al. (1998) to the proteoliposomes. To terminate the reaction, 150 μL of proteoliposomes for each measurement point was passed over Dowex AG1-X8 columns (Chloride form, 100 to 200 mesh, equilibrated with 150 mM sodium acetate). Radiolabeled substrate that was not taken up into liposomes was removed by binding to the anion-exchange column. The eluted proteoliposomes were collected in scintillation vials filled with 10 mL deionized water. The imported radiolabeled ATP or ADP was quantified by a liquid scintillation counter (Tri-Carb 2100TR; Packard). Kinetic constants were determined by measuring the initial velocity of each experiment. The Michaelis-Menten constant (K_m) has been analyzed with seven external ATP concentrations, ranging from 0.01 to 1 mM, and competitive compounds of ATP transport were assessed by the inhibitor constant K_i, as described by Dixon (1953). Nonlinear regression analysis based on the Marquardt algorithm was performed with Grafit software version 3.0 to fit all enzyme kinetic data (Erithacus Software).

Gene Expression Data

Tissue-specific gene expression patterns of PNC1 and PNC2 in Arabidopsis were retrieved from the Arabidopsis eFP Browser (Winter et al., 2007).

Isolation of T-DNA Insertion Lines

T-DNA insertion lines for PNC1 (SAIL_303H02, referred to as pnc1-1) and PNC2 (SALK_014579, referred to as pnc2-1) were obtained from the ABRC. To identify homozygous T-DNA insertion lines, genomic DNA was isolated and genotyped using gene-specific primer pairs (NL268/NL269 for pnc1-1 and NL273/NL274 for pnc2-1) and a primer pair for T-DNA/gene junction (P48/NL268 for pnc1-1 and P46/NL274 for pnc2-1). The position of the T-DNA in the PNC genes was verified by sequencing. As a positive control, amplification of the actin gene ACT2 (At3g18780) was performed with the primer pair P33/P34. Independent homozygous pnc1-1 and pnc2-1 plants were further propagated.

Generation of Arabidopsis RNAi Lines

To construct a double-stranded RNA hairpin structure, a 480-bp cDNA fragment of PNC1 that showed 86% similarity to PNC2 on the nucleotide level was amplified from the ABRC cDNA clone U13126 by PCR using two different sets of primers (NL66/NL67 and NL68/NL69) (see Supplemental Figure 6 online). One PCR product was inserted into the pHannibal vector (AJ311872) (Wesley et al., 2001) in the sense orientation via EcoRI and XhoI restriction downstream of the pyruvate orthophosphate dikinase (PDK) intron from E. coli, whereas the other was introduced in the antisense orientation using XbaI and BamHI sites upstream of the PDK intron. The resulting intron-spliced hairpin RNA cassette was subcloned into the pACN vector (AF169416) between the ethanol-inducible alcohol dehydrogenase promoter (AlcA_pro) from Aspergillus nidulans and the nopaline synthase terminator (nos_ter) (Caddick et al., 1998). For in planta transformation, the AlcA_pro:PNC1RNAi-nos_ter construct was introduced using the HindIII site into the pBIN19-derived plant transformation vector (Caddick et al., 1998), which carries the cassette for the expression of the ALCR transcription factor (35S_pro:ALCR-nos_ter) required to activate the AlcA promoter in the presence of ethanol. Wild-type Arabidopsis plants were transformed by an Agrobacterium-mediated procedure (Clough and Bent, 1998) and were selected using kanamycin resistance. T-DNA insertion was confirmed in several transgenic lines (T1 generation) by genomic PCR using the primer pair P13/P30. T2 plants obtained after selfing were screened for homozygous lines by segregation analysis of the nptII resistance gene and further propagated.

RT-PCR and qRT-PCR

Total RNA was extracted from wild-type and transgenic Arabidopsis plants generated in this study using the guanidinium thiocyanate-phenol-chloroform extraction method (Chomczynski and Sacchi, 1987) and subjected after DNase treatment to cDNA synthesis using Superscript II RNase H⁻ reverse transcriptase (Invitrogen). The transcript level of PNC1, PNC2, and At2g39970 was analyzed by PCR using gene-specific primer sets binding to the corresponding 5′ and 3′ untranslated regions of the PNC1 (primer set NL84/NL85), PNC2 (NL115/NL116), and At2g39970 (NL48/49) cDNAs. For the amplification of the truncated PNC1 transcript in the pnc1-1 T-DNA insertion line, the primer set NL84/NL325 was used. As a control for the cDNA quality, a cDNA fragment of an actin gene (ACT7, At5g09810) was amplified using the primers ML167 and ML168. PCR conditions used were as follows: 94°C for 2 min, followed by 35 cycles of 94°C for 30 s, 58°C for 45 s, 72°C for 90 s, and a final extension of 72°C for 3 min. For relative quantification of the transcript level, qRT-PCR was performed with the DNA Engine Thermal Cycler Detection system (Bio-Rad Laboratories) and MESA GREEN qPCR MasterMix for SYBR Assay (Eurogentec). The following primer pairs were used for gene-specific amplification: NL326/NL85 for PNC1 and NL326/NL116 for PNC2. For all samples, cDNA levels were normalized against those of the Type 2A phosphatase interacting protein 41 (TIP41)-like gene (At4g34270) amplified with the primer pair P71/P72 (Czechowski et al., 2005). The qRT-PCR primers were designed with PerlPrimer (Marshall, 2004).

Lipid Analysis

Fatty acids were analyzed by gas chromatography following conversion to methyl esters (FAMEs). The method of Browse et al. (1986) was employed with the following modification: FAMEs were extracted with 1 mL hexane, the solvent was removed under a stream of N₂, and FAMEs were resuspended in 50 μL hexane. The fatty acid 17:0 was used as an internal standard to permit quantification. Fatty acyl CoAs were measured as described by Larson and Graham (2001).

Visualization of Oil Bodies

Seedlings were grown as described above and stained for 5 min in a 1 μg/mL solution of Nile Red (Molecular Probes). Images were recorded with a Zeiss Axiophot fluorescence microscope (Carl Zeiss) using a 450- to 490-nm filter. Corresponding bright-field images were also recorded.

Accession Numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: Sc ANT1 (NP_0154531), At2g39970 (NP_181526), At3g05290 (PNC1, NP_566251), At5g27520 (PNC2, NP_198104), Os05g32630 (NP_001055452), and PMP34 (NP_006349).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure 1. Complementation of ant1Δ with the EYFP-Tagged Arabidopsis Proteins.

• Supplemental Figure 2. Expression of PNC1 and PNC2 in E. coli and ADP Uptake into Liposomes Reconstituted with Recombinant PNC Proteins.

• Supplemental Figure 3. Tissue-Specific Expression Pattern of PNC1 and PNC2 in Arabidopsis.

• Supplemental Figure 4. Identification of Arabidopsis T-DNA Insertion Lines for PNC1 and PNC2.

• Supplemental Figure 5. Time Kinetics of ATP Uptake into Liposomes Reconstituted with Recombinant PNC1 and Truncated PNC1Δ19C Protein.

• Supplemental Figure 6. Hairpin RNA Expression Construct.

• Supplemental Figure 7. Germination Phenotype of PNC1/2 RNAi Lines.

• Supplemental Figure 8. Fatty Acid and Acyl-CoA Profile.

• Supplemental Table 1. Genes That Are Coexpressed with PNC1 and PNC2.

• Supplemental Table 2. Gene-Specific Primers Used for Cloning in This Study.