A peroxisomal carrier delivers NAD+ and contributes to optimal fatty acid degradation during storage oil mobilization


(fax: +49 211 81 137 06; e-mail Nicole.Linka@uni-duesseldorf.de).


The existence of a transport protein that imports cytosolic NAD+ into peroxisomes has been controversially discussed for decades. Nevertheless, the biosynthesis of NAD+ in the cytosol necessitates the import of NAD+ into peroxisomes for numerous reduction/oxidation (redox) reactions. However, a gene encoding such a transport system has not yet been identified in any eukaryotic organism. Here, we describe the peroxisomal NAD+ carrier in Arabidopsis. Our candidate gene At2g39970 encodes for a member of the mitochondrial carrier family. We confirmed its peroxisomal localization using fluorescence microscopy. For a long time At2g39970 was assumed to represent the peroxisomal ATP transporter. In this study, we could show that the recombinant protein mediated the transport of NAD+. Hence, At2g39970 was named PXN for peroxisomal NAD+ carrier. The loss of PXN in Arabidopsis causes defects in NAD+-dependent β-oxidation during seedling establishment. The breakdown of fatty acid released from storage oil was delayed, which led to the retention of oil bodies in pxn mutant seedlings. Based on our results, we propose that PXN delivers NAD+ for optimal fatty acid degradation during storage oil mobilization.


Oil-seed plants, such as Arabidopsis, accumulate storage oil in form of triacylglycerol (TAG) in their seeds to ensure the survival of the next generation. Upon germination, TAG is hydrolysed and the released fatty acids are broken down via β-oxidation to fuel seedling development (Graham, 2008; Quettier and Eastmond, 2009). The degradation of fatty acids takes place in peroxisomes and is essential for this metabolic process. A defect in β-oxidation led to plants that are unable to mobilize their storage oil. Hence, seedling establishment is compromised. In the last decade enzymes associated with β-oxidation have been extensively studied by molecular genetic analysis (Graham, 2008; Quettier and Eastmond, 2009).

In the course of the fatty acid oxidation, NAD+ is reduced to NADH by the peroxisomal multifunctional protein (MFP) (Richmond and Bleecker, 1999; Rylott et al., 2006). To maintain the flux through this metabolic pathway, the NADH produced needs to be regenerated in peroxisomes (Graham, 2008). The peroxisomal malate dehydrogenase (PMDH) re-oxidizes NADH to NAD+ via the reversible reduction of oxaloacetate to malate (Mettler and Beevers, 1980; Pracharoenwattana et al., 2007, 2010). The resultant dicarbonic acid is exported to the cytosol, where it is converted to oxaloacetate to re-enter the peroxisomes. In addition, the co-factor NAD+ is required for the monodehydroascorbate reductase (MDAR) to detoxify hydrogen peroxide that is generated by β-oxidation (Eastmond, 2007). These three peroxiosmal enzymes are essential for this process, because a loss of MFP, PMDH or MDAR activities led to Arabidopsis plants that are compromised in storage oil breakdown (Rylott et al., 2006; Eastmond, 2007; Pracharoenwattana et al., 2007, 2010).

To date, how the oxidative fatty acid breakdown is provided with NAD+ is still to be resolved. As NAD+ is synthesized in the cytosol (Noctor et al., 2006; Hashida et al., 2009), the following possibilities come into consideration: (i) peroxisomes come ‘pre-packed’ with NAD+ when they are derived from the ER; (ii) peroxisomal NAD-dependent enzymes bind their co-factor in the cytosol, and both are then targeted to peroxisomes; or (iii) NAD+ is taken up from the cytosol into peroxisomes via a specific transport system (Rottensteiner and Theodoulou, 2006). Recently, two members of the mitochondrial carrier family (MCF) have been identified to import NAD+ across the plastid and mitochondrial inner membrane from Arabidopsis, named NDT1 and NDT2, respectively (Palmieri et al., 2009). Do plant peroxisomes also possess such an NAD+ import system?

The transport of NAD+ across the peroxisomal membrane has been controversially discussed for several decades. Uptake assays with isolated rat peroxisomes indicated the presence of transport proteins mediating NAD+ diffusion (Van Veldhoven et al., 1983, 1987). Contradictory observations claimed that the peroxisomal membrane is impermeable to NAD+ (Donaldson, 1981; van Roermund et al., 1995; Antonenkov et al., 2004). For instance, externally provided NAD+ could not sustain the activities of NAD-dependent enzymes in intact peroxisomes purified from rat and Ricinus communis (castor bean; Donaldson, 1981; Antonenkov et al., 2004). Secondly, a genetic study in Saccharomyces cerevisiae also refuted the existence of an NAD+ import system (van Roermund et al., 1995). The disruption of the peroxisomal malate dehydrogenase (Mdh3p) in yeast inhibits β-oxidation, because NAD+ cannot be regenerated for this pathway. This implies that cytosolic NAD+ is unable to enter yeast peroxisomes, which would restore the loss of the Mdh3p and ensure the action of fatty acid breakdown (van Roermund et al., 1995). Yeast possesses two NAD+ carrier proteins, Ndt1p and Ndt2p, but they are both localized in the mitochondrial inner membrane (Todisco et al., 2006).

In 2001 an abundant protein with an apparent molecular mass of 38 kDa has been detected in a purified peroxisomal membrane fraction of etiolated Cucurbita sp. (pumpkin) cotyledons (Fukao et al., 2001). The obtained peptide showed high sequence similarity to the Arabidopsis protein At2g39970, which encodes for an MCF member (Palmieri et al., 2011). Because of its phylogenetic proximity to mitochondrial ATP/ADP carriers, the At2g39970 protein was assumed to represent the peroxisomal ATP transport protein (Fukao et al., 2001). This hypothesis was rejected, because this transporter was unable to complement the growth phenotype of a yeast mutant deficient in the endogenous peroxisomal ATP transporter (Linka et al., 2008). In addition, the same study identified two other MCF members as peroxisomal ATP transporters (Linka et al., 2008), which raised the question: what is the function of the At2g39970 protein?

In this study we showed that plant peroxisomes possess an NAD+ import system. Uptake studies using recombinant protein revealed that the peroxisomal membrane protein encoded by At2g39970 catalyses the NAD+ import. To test the hypothesis, that this carrier provides β-oxidation with NAD+, we analysed the corresponding Arabidopsis knock-out mutants. We observed that the degradation of storage oil-derived fatty acids is delayed in these transgenic lines. The partial resistance against the root growth inhibitor 2,4-dichlorophenoxybutyric acid (2,4-DB) also supports a limited conversion of this compound via β-oxidation. Our study indicates that during the early stages of seedling germination the peroxisomal NAD+ carrier is required for optimising flux through β-oxidation.


Membrane protein encoded by At2g39970 is targeted to peroxisomes

We confirmed the peroxisomal localization of the At2g39970 protein in plants, which was indicated by several proteomic studies (Eubel et al., 2008; Reumann et al., 2009). At2g39970 fused at the C terminus to the yellow fluorescent protein (YFP) and a peroxisomal marker consisting of cyan fluorescent protein (CFP) tagged with the peroxisomal targeting signal 1 (PTS1) were transiently co-expressed in tobacco leaves mediated via Agrobacterium infiltration. After co-infiltration, tobacco protoplasts were isolated and analysed by confocal fluorescence microscopy. The merged images of the protoplast revealed a similar fluorescence pattern for the expression of At2g39970-YFP and CFP-PTS1, indicating that At2g39970 was targeted to plant peroxisomes (Figure 1).

Figure 1.

 At2g39970-YFP protein is targeted to peroxisomes in Nicotiana tabacum (tobacco) protoplasts. Confocal microscopic images were taken from protoplasts of 4-week-old tobacco leaves co-expressing At2g39970-YFP and the peroxisomal marker CFP-PTS1. Scale bars: 10 μm.

Peroxisomal transport protein catalyses NAD+ uptake

At2g39970 is related in sequence with mitochondrial and plastidial NAD+ transporter from Arabidopsis and yeast (Figure 2) (Todisco et al., 2006; Palmieri et al., 2009). To assess whether At2g39970 exhibits NAD+ transport function, we expressed the At2g39970 protein in a cell-free wheat germ expression (WGE) system (Figure 3a). The translation reaction was performed in the presence of detergents and lipid vesicles to ensure that the newly produced membrane protein remained soluble and was functionally integrated into the lipid bilayer (Nozawa et al., 2007). The identity of the histidine (His)-tagged At2g39970 protein (At2g39970-His), with a calculated molecular mass of 38.3 kDa, was validated by immunoblot analysis using anti-His antibody (Figure 3b).

Figure 2.

 Amino acid sequence alignment represents At2g39970 and its closely related peroxisomal ATP transporters, and NAD+ carriers from mitochondria and plastids of Arabidopsis and yeast: PNC1 (At3g05290), PNC2 (At5g27520), NDT1 (At2g47490), NDT2 (At1g25380), Ant1p (YPR128c), Ndt1p (YIL006w) and Ndt2p (YEL006w). Black shading indicates identical amino acid residues in all eight sequences, whereas grey shading indicates similar amino acid residues in at least five sequences. The mitochondrial energy transfer signature domains are boxed, and the six predicted transmembrane-spanning domains are underlined. Arrows mark the region of the hydrophilic loop between the third and fourth transmembrane domain that used for an At2g39970-specific antibody.

Figure 3.

In vitro expressed At2g39970-His protein exhibited NAD+ uptake activities.
(a) Coomassie-stained SDS-PAGE indicates proteins of the wheatgerm extract (lane 1, negative control) and At2g39970-His protein expressed by wheatgerm extract (lane 2).
(b) Immunoblot analysis verifies the expression of At2g39970-His protein by the wheatgerm extract. The expressed At2g39970-His protein was detected with anti-HIS antibody (lane 2). The arrowhead indicates the expressed At2g39970-His protein.
(c) Time-dependent uptake of radiolabelled [α-32P]NAD+ (125 μm) in the presence (filled circles) or absence of NAD+ (open circles) as internal substrate (30 mm). Recombinant At2g39970 protein was reconstituted into liposomes. Error bars represent the standard error of the arithmetic mean from three independent uptake experiments.
(d) Internal substrate dependence on NAD+ import activities. Liposomes were preloaded with various substrates (20 mm) and reconstituted with recombinant At2g39970-His protein. Uptake experiments were initiated with 125 μm [α-32P]NAD+. The initial velocity of time-dependent NAD+ uptake was determined. Relative uptake activities were compared with the NAD+/NAD+ counter-exchange experiment, which was set to 100%. The data represent arithmetic means ± SEs of three independent experiments.

The recombinant protein obtained was reconstituted into liposomes for transport studies. The uptake of radioactive labelled [α-32P]NAD+ (125 μm) was measured in the presence or absence of 30 mm NAD+ inside the liposomes. The reconstituted At2g39970 protein mediated high NAD+ uptake rates when liposomes were preloaded with NAD+ (Figure 3c, filled symbols). The NAD+ homo-exchange followed first-order rate kinetics, with an equilibrium plateau (Vmax) of 1.3 ± 0.3 μmol NAD+ mg protein−1 and an initial rate of 0.03 ± 0.002 μmol NAD+ mg protein−1. In comparison, only a marginal uptake of NAD+ was observed when vesicles did not contain NAD+ as a counter-exchange substrate (Figure 3c open symbols). Thus the At2g39970 carrier imports NAD+ in an antiport mechanism. Negligible accumulation rates were observed when the protein translation reaction was reconstituted in the absence of At2g39970, or when boiled recombinant protein was used as a negative control (Figure S1a,b). On the basis of its transport function, At2g39970 was named PXN for peroxisomal NAD+ carrier.

Peroxisomal NAD+ carrier prefers NAD+ and adenine nucleotides as substrates

To discover additional substrates for PXN, vesicles were preloaded with saturating concentrations (i.e. 20 mm) of various potential counter-exchange substrates and the initial rates of [α-32P]NAD+ uptake (125 μm) were determined (Figure 3d). Relative to the NAD+/NAD+ homo-exchange experiments, the highest NAD+ uptake rates were observed when liposomes were preloaded with AMP (90 ± 8%).

Among other tested nucleotides, the initial NAD+ uptake rates for ADP as internal substrate were 56% (±18%) of the NAD+/NAD+ exchange rate. In contrast, ATP, GTP and GDP were non-suitable substrates for the NAD+ import (Figure 3d). Notably, NADH did serve as a suitable transport substrate for PXN (69 ± 8%), whereas negligible NAD+ uptake rates were measured when vesicles contained NADP+, NADPH or FAD (<10% of the NAD+/NAD+ homo-exchange, compared with non-preloaded proteoliposomes). The intermediate of NAD+ biosynthesis, nicotinate adenine dinucleotide (NaAD), led to an active NAD+ import (71 ± 8%), whereas nicotinamide mononucleotide (NMN) and nicotinamide (Nam), both metabolites of the NAD+ salvage pathway, were not transported by PXN. Based on our substrate specificity study, PXN favours the NAD+in/AMPout exchange in particular (Figure 3d). The plastidial and mitochondrial localized NAD+ carriers, in contrast, prefer both ADP and AMP in exchange for NAD+ (Table S1) (Todisco et al., 2006; Palmieri et al., 2009).

The kinetic constants, such as the Michaelis constants (KM) and the inhibitory constants (Ki), reflect the affinity of PXN towards a putative substrate. If KM or Ki values for a particular substrate are low, the substrate is a prime candidate for the import of NAD+ via PXN in vivo. To determine the apparent KM value of PXN for NAD+, the initial rates of [α-32P]NAD+ uptake into NAD+-containing liposomes (30 mm) were measured at different external NAD+ concentrations (Table 1). PXN exhibits an apparent KM value for NAD+ of 246 ± 64 μm. Compared with the NAD+ transporting carriers from Arabidopsis and yeast (Todisco et al., 2006; Palmieri et al., 2009), PXN exhibits a similar affinity for NAD+ (Table S2).

Table 1.   Kinetic constants of the recombinant PXN for various metabolites
  1. The Michaelis constant (KM) for NAD+ was determined using various external [α-32P]NAD+ concentrations (10–2 mm). The competitive inhibition constant (Ki) of [α-32P]NAD+ uptake (125 μm) was assayed with increasing inhibitor concentrations (16–8 mm). All proteoliposomes were preloaded with 30 mm NAD+. Data are summarized as the arithmetic means ± SEs of three independent experiments.

KMNAD+246 ± 64
KiNADH150 ± 1
KiAMP356 ± 1
KiADP251 ± 1
KiNaAD536 ± 1

To identify the counter-exchange substrate for NAD+, the Ki values of PXN were assayed via the inhibition of the NAD+/NAD+ homo-exchange in the presence of increasing concentrations of NADH, AMP, ADP and NaAD, respectively (Table 1). The low Ki values of 251 ± 2 and 357 ± 1 μm for ADP and AMP, respectively, confirmed that PXN accepts both adenine nucleotides as potent substrates for the import of NAD+. The highest affinity, however, was observed for NADH. The Ki value of 150 ± 1 μm for NADH was 0.5-fold lower than the KM value for NAD+ (246 μm). In contrast, NaAD had only an inhibitory effect on the NAD+/NAD+ exchange at high concentrations, indicating its low affinity for the PXN binding site. Thereby, NaAD can be excluded as an in vivo substrate for PXN.

Characterization of transgenic plants lacking the peroxisomal NAD+ carrier

To investigate the role of PXN in plants, we established three independent homozygous T-DNA insertion lines (Figure S2a). Each T-DNA was inserted in an exon region of the peroxisomal NAD+ carrier, as verified via PCR analysis on genomic DNA (Figure S2b). DNA sequencing of the T-DNA specific PCR product confirmed the predicted position of the T-DNA location. The mutations in the homozygous pxn plants led to a disruption of the mRNA expression based on RT-PCR analysis, which failed to detect PXN full-length transcript in either of the T-DNA insertion lines (Figure S2c).

A PXN-specific antibody directed against the hydrophilic loop between the third and fourth transmembrane domain was raised. This structural feature, which is distinctive for PXN (Figure 1), was expressed as His-tagged peptide in Escherichia coli, purified and used as an antigenic epitope (Figure S3). The resulting polyclonal antiserum was specific for PXN and did not cross-react with recombinant PNC1 and NDT1 proteins (Figure S4). In total membrane fractions from 5-day-old etiolated seedlings of wild-type plants, the PXN-specific antibody recognized a 36-kDa protein that corresponds with the predicted molecular mass of PXN in the wild-type membrane fraction (Figure S5a). In the mutant plants pxn-1 and pxn-3, a matching band was absent, indicating that both alleles are loss-of-function mutants. By contrast, pxn-2 seedlings contain hardly detectable levels of PXN protein. The T-DNA insertion in the latter allele is located in the first exon of the PXN gene, and PXN mRNA downstream of the T-DNA insertion point is present in this mutant (Figure S5b), thereby predicting a PXN protein lacking the first six amino acid residues. Secondly, the size of the mutated PXN protein is equal to that of the wild type, implying that the chimeric transcript containing T-DNA border sequences does not expand the protein sequence. We cannot exclude that the mutated PXN protein might be functional. Nevertheless, the expression of the PXN protein is drastically reduced in the mutant line. Thus the pxn-2 plant represents a strong knock-down mutant.

Mutant plants do not exhibit an obvious sucrose-dependent phenotype

To elucidate the role of PXN in plants, we gathered information about its developmental expression using the publicly available microarray database Arabidopsis eFP Browser (Winter et al., 2007). PXN is ubiquitously expressed in all tissues, but dry mature seeds store high levels of PXN mRNA, indicating a possible involvement in early developmental stages upon germination (Nakabayashi et al., 2005). Based on gene expression, we hypothesized that PXN provides the NAD+-dependent fatty acid breakdown with NAD+ during storage oil mobilization. It was shown that the exogenous provision of sucrose rescues the arrested seedling growth phenotype of Arabidopsis β-oxidation mutants (Graham, 2008; Quettier and Eastmond, 2009). To study whether pxn mutant plants exhibit such a sucrose-dependent phenotype, we assayed the growth of 6-day-old seedlings grown on agar plates either with or without sucrose under short-day conditions (Figure S6) and in the dark (Figure S7). None of the mutants showed a discernible sucrose-dependent phenotype under both conditions. The pxn seedlings develop normally like the wild type in the absence of sucrose. To quantify seedling growth, we measured root length as well as hypocotyl length (Figures S6 and S7). The growth rates of the three mutant alleles were not consistently reduced compared with the wild type when grown in the absence of sucrose. In sum, storage oil mobilization was not affected seriously enough to cause apparent growth phenotypes in these mutant alleles.

Seedlings deficient in peroxisomal NAD+ carrier are delayed in storage oil mobilization

Although pxn mutants do not exhibit an obvious seedling growth phenotype, we determined whether an impaired NAD+ import by PXN has an effect on storage oil mobilization. To this end, we measured the fatty acid level in seeds and seedlings when grown on agar plates in the absence of sucrose. The content of the suitable marker eicosenoic acid (C20:1) was determined to monitor TAG breakdown in Arabidopsis (Lemieux et al., 1990).

In wild-type seedlings, the level of eicosenoic acid declined by 77% at 4 days after imbibition (DAI), until it was completely depleted by six DAI (Figure 4). In 4-day-old mutant plants, the eicosenoic level was only reduced by approximately 7% (Figure 4). At 6 DAI 87% of the seed-stored fatty acid was degraded in the pxn seedlings. This observation indicates that the pxn mutants were able to mobilize storage oil, but because of the loss of PXN the rate of fatty acid breakdown was considerably delayed.

Figure 4.

 Mutant deficient in PXN are compromised in storage oil mobilization. Seed-oil specific eicosenoic acid content was measured in seeds and seedlings established in the absence of sucrose.
(a) Relative eicosenoic acid content from seedlings that were grown vertically on half-strength MS agar plates under short-day conditions.
(b) Relative eicosenoic acid content from seedlings that were grown vertically on half-strength MS agar plates in darkness. Data plotted are means ± SEs of measurements on three batches of 10-mg seeds or 20-mg seedlings, respectively. ***P < 0.001; extremely significant.

Examination of 4-day-old etiolated seedlings stained with the lipophilic dye, Nile Red, revealed that oil bodies were still retained in the hypocotyl of pxn-3 seedlings (Figure 5, middle panel), as well as for the other mutant alleles (Figure S8). The same observation was found for the Arabidopsis pxa1-1 mutant (Zolman et al., 2001), in which the lack of the peroxisomal ABC transporter PXA1 leads to a complete block of fatty acid breakdown (Figure 5, lower panel). In contrast, the oil bodies almost disappeared in wild-type cotyledons at 4 DAI (Figure 5, upper panel). Taken together, the retention of oil bodies indicates that the impaired peroxisomal β-oxidation in pxn seedlings has a negative impact on lipolysis during early seedling growth.

Figure 5.

 Oil bodies were retained in the pxn-3 mutant. Light microscopic images of hypocotyl cells from 4-day-old wild-type and pxn-3 seedlings that were grown on agar plates containing half-strength MS plus 1% (w/v) sucrose and stained with the lipophilic dye Nile Red. As a positive control for the presence of oil bodies in hypocotyls, the pxa1-1 seedlings were used. Scale bar: 5 μm.

To exclude the possibility that impaired fatty acid oxidation is a consequence of a developmental defect prior to and during seed germination, we compared the germination efficiencies of pxn seeds with that of the wild type. Figure S9a showed no significant alteration in the germination rates of the wild type and mutant lines. Exogenous sucrose, which inhibits seed germination, also had no additive effect in any of the mutant lines disrupted in the peroxisomal NAD import (Figure S9b).

pxn seedlings are less sensitive to 2,4-DB

The response to 2,4-DB on early seedling growth is an alternative assay to display a defect in fatty acid degradation. In the wild type, β-oxidation converts 2,4-DB to the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D), which severely inhibits root elongation (Estelle and Somerville, 1987). Mutants lacking functional β-oxidation are 2,4-DB-resistant (Hayashi et al., 1998; Zolman et al., 2000). Thus, we analysed primary root elongation in wild-type and pxn seedlings that were grown on agar plates containing either 2,4-DB or 2,4-D. As shown in Figure 6, the pxn seedlings are more resistant to 2,4-DB than are the wild type. The primary root growth was significantly less affected in the presence of 2,4-DB. However, all mutant lines behave like the wild type in their response to the 2,4-D herbicide: root elongation was markedly inhibited by the herbicide. The partial resistance phenotype indicates that peroxisomal NAD+ import mediated by PXN is required for optimal fatty acid oxidation.

Figure 6.

pxn mutants are less sensitive to 2,4-dichlorophenoxybutyric acid (2,4-DB).
(a) Effect of 2,4-DB and 2,4-dichlorophenoxyacetic acid (2,4-D) on the growth of the wild type and pxn mutants. Plants were grown vertically on half-strength MS agar plates under short-day conditions in the presence of 0.8 μm 2,4-DB or 0.23 μm 2,4-D, respectively. As a negative control seedlings were grown in the presence of 0.001% (v/v) ethanol, as 2,4-DB and 2,4-D were dissolved in ethanol. Photographs of representative seedlings were taken 6 days after imbibition (6 DAI). Scale bars: 1 cm.
(b) Effect of 2,4-DB and 2,4-D on primary root length from wild-type and pxn mutant seedlings. Each value represents the mean root length with standard error of three replicates (n = 35 seedlings). ***P < 0.001 (extremely significant).


This work addresses the fundamental question of how plant peroxisomes protect their supply of NAD+. NAD+ is an essential co-factor for numerous redox reactions. The fact that NAD+ is synthesized in the cytosol necessitates NAD+ uptake into peroxisomes (Noctor et al., 2006; Hashida et al., 2009). Over decades the existence of a peroxisomal NAD+ transporter was controversially discussed in eukaryotic organisms (Donaldson, 1981; Van Veldhoven et al., 1983; van Roermund et al., 1995; Antonenkov et al., 2004). Our study proves that plant peroxisomes possess a protein-mediated NAD+ import system. Here, we characterized an MCF member encoded by At2g39970 that resides in the peroxisomal membrane (Figure 1) and catalyses the import of NAD+ (Figure 3c). This carrier, called PXN, is able to transport NAD+ with a high affinity (KM) value of 246 ± 64 μm (Table 1). Our knowledge about intracellular levels of free NAD+ in plant tissues is limited (Noctor et al., 2006). Most reports determined the total concentration of both free and bound NAD+ from various plant extracts, because most of the pyridine nucleotides are bound to proteins, presumably dehydrogenases (Igamberdiev and Gardestrom, 2003; Queval and Noctor, 2007; Wang and Pichersky, 2007; Quettier et al., 2008). In the cytosol of spinach leaves Heineke et al. (1991) calculated a free NAD+ concentration of 600 μm, estimated from the concentrations of metabolites and enzyme activities. Under normal conditions the cytosolic pyrimidine nucleotide pool is predominantly in the oxidized state of various organs from diverse plant species (Igamberdiev and Gardestrom, 2003; Queval and Noctor, 2007; Wang and Pichersky, 2007; Quettier et al., 2008). Thus, it is quite likely that the KM value for NAD+ is lower than the cytosolic free NAD+ levels in the cytosol of Arabidopsis, thereby allowing the import of NAD+ from the cytosol into peroxisomes via PXN.

In our experimental set-up, transport of NAD+ mediated by PXN occurs in antiport mode, meaning that NAD+ is taken up in a strict exchange with a counter substrate (Figure 3c). PXN exhibited the highest NAD+ uptake rates against AMP (Figure 3d). Furthermore, the Ki values of AMP and NAD+ are comparable, indicating that both substrates bind to the PXN binding site with equal affinities (Table 1). Thus, we suggest that AMP is the most likely substrate of PXN for NAD+ import. Under the hypothesis that PXN provides NAD+-dependent enzymes with its co-factor, the question arises of how PXN achieves a net influx of NAD+ to build up a peroxisomal NAD+ pool, if it acts as an antiporter?

The yeast orthologue Ndt1p catalyses, in addition to the exchange of NAD+ with AMP, the unidirectional transport of NAD+, and thus contributes to setting up the mitochondrial NAD+ pool (Todisco et al., 2006; Agrimi et al., 2011). In plants, the plastidial and mitochondrial NAD+ carriers NDT1 and NDT2, respectively, operate just as anti-port carriers, like PXN (Palmieri et al., 2009). The NAD+ uptake into plastids by NDT1 is enabled through the efflux of ADP or AMP: both counter-exchange substrates can be withdrawn from the adenylates pool, because they are synthesized in the stroma (Palmieri et al., 2009). For a net NAD+ influx into plant mitochondria, the NDT2-mediated NAD+/AMP exchange is coupled to the mitochondrial adenine nucleotide carrier ADNT1: this transporter takes up cytosolic AMP in exchange with ATP that is synthesized in the mitochondrial matrix (Palmieri et al., 2008).

Based on our in vitro uptake studies, PXN favours mediating the NAD+/AMP anti-port. To allow a net NAD+ transport into plant peroxisomes via PXN, we postulated two scenarios that could balance the loss of AMP. On one hand, the unidirectional AMP import could refill the peroxisomal adenine nucleotide pool (Figure 7a). Up to now, such an adenylate uniporter has been only identified for the plastidial membrane (Leroch et al., 2005; Kirchberger et al., 2008). Alternatively, a member of the Arabidopsis Nudix hydrolase family, NUDT7, predicted to be peroxisomal (Reumann et al., 2004), might provide efflux substrates for PXN. This enzyme preferentially hydrolyses NADH to NMNH and AMP (Ge et al., 2007; Ge and Xia, 2008). Stoichiometrically, a net import of one NAD+ molecule in the peroxisomal matrix results when PXN imports two molecules of NAD+ against one molecule of AMP and NMNH, with both derived from the NUDT7-catalysed hydrolysis of one NADH molecule (Figure 7b). As yet, we cannot test whether PXN accepts NMNH as a counter-exchange substrate, because this compound is not commercially available. Although the performed uptake assays could not detect any NAD+ import activities into non-preloaded proteoliposomes, we cannot exclude that PXN might import in vivo NAD+ through a uniport mechanism (Figure 7c).

Figure 7.

 Schematic transport mechanism for net NAD+ import into plant peroxisomes.
(a) Two-transporter model: PXN-mediated NAD+/AMP exchange is coupled to an unknown adenine nucleotide carrier that catalyses a unidirectional AMP uniport.
(b) Three-substrate model: PXN imports two molecules NAD+ against one molecule of AMP and NMNH, both derived from the NUDT7-catalysed hydrolysis of one NADH molecule.
(c) Uniport model: PXN might import in vivo NAD+ in a uniport mechanism. (1) Reduction of NAD+ to NADH by peroxisomal metabolism. (2) Peroxisomal nudix hydrolase NUDT7.

The role of any NAD+ carrier in supplying redox reactions with NAD+ has only been investigated for yeast mitochondria (Todisco et al., 2006). The corresponding double mutant lacking both mitochondrial NAD+ carriers, Ndt1p and Ndt2p, displays a severe delay in growth on non-fermentable carbon sources, such as lactate, pyruvate, acetate or ethanol (Todisco et al., 2006). The loss of NAD+ import caused by the NAD+-fuelled tricarboxylic acid cycle is inhibited, and these carbon skeletons cannot be used for respiratory growth (Todisco et al., 2006). Phenotypic analysis of the Arabidopsis mutant lacking the mitochondrial or plastidic NAD+ carrier, respectively, has not been investigated so far (Palmieri et al., 2009). Our studies have described the impact of the NAD+ transport system for plant peroxisomes during storage oil mobilisation. We showed that the fatty acid breakdown during storage oil mobilisation is impaired when PXN is absent. Although the pxn seedlings were able to mobilize enough fatty acids to become established (Figure S6), rate of β-oxidation was slowed, as shown by the high levels of C20:1 (Figure 4) and the retention of oil bodies in the hypocotyl cells (Figure 5). Considering the delay in fatty acid breakdown, however, the lack of seedling growth phenotype was not surprising. A mutation in the acyl-CoA oxidase 2 gene led to Arabidopsis seedlings that accumulate up to threefold more C20:1 lipids than the wild type, but did not show significant differences in early post-germinative growth (Pinfield-Wells et al., 2005). From the observed intermediate phenotype we assume the following possible roles for PXN during seed oil mobilization.

  • (i) PXN builds up the peroxisomal NAD+ pool, but an alternative route exists to provide peroxisomes with cytosolic NAD+. The peroxisomal ATP transport proteins PNC1 and PNC2 from Arabidopsis might be putative candidates, because adenine nucleotides and nicotine adenine dinucleotides are structurally related. Whereas the plastidial and mitochondrial ATP carrier are highly specific for adenine nucleotides, the affinity to NAD+ has not been investigated for any peroxisomal ATP transporter from Arabidopsis, yeast and human (Palmieri et al., 2001; Visser et al., 2002; Linka et al., 2008). In this context Haferkamp et al. (2004) have discovered a bacterial carrier from the obligate endosymbiont Protochlamydia amoebophila that accepts both NAD+ and ATP as transport substrates, but it belongs to a different transporter family than the PNCs.
  • (ii) PXN replenishes the peroxisomal NAD+ pool instead of setting it up. Peroxisomes might contain catalytic levels of NAD+ for their metabolism (Pracharoenwattana et al., 2007, 2010). NAD+ could be already present in pre-peroxisomes when they derived from the endoplasmic reticulum (ER) (Tabak et al., 2008). On the other hand, peroxisomal NAD-dependent enzymes could import their co-factor in cytosol, where they are synthesized, and then be translocated as holoenzymes into peroxisomes, as shown for the peroxisomal FAD-dependent acyl-CoA oxidase (Titorenko et al., 2002). In contrast to plastids and mitochondria, the peroxisomal protein import machinery forms a large pore that facilitates the membrane passage of fully folded and assembled proteins (Meinecke et al., 2010). Both prospects explain why peroxisomal NAD+ is available in the absence of PXN, which enables TAG mobilization at early stages of seedling development until the seedling becomes photoautotrophic. PXN is then required to maintain a certain NAD+ level in plant peroxisomes to ensure the optimal operation of NAD+-dependent processes.
  • (iii) PXN mediates a unique NAD+/NADH exchange. PXN is unique in terms of its capability to exchange in vitro NAD+ with its reduced form, NADH (Figure 3d; Table 1). The other NAD+ carriers characterized from Arabidopsis and yeast (Todisco et al., 2006; Palmieri et al., 2009) are unable to transport NAD+ in exchange with NADH (Table S3). This NAD+/NADH exchange transfers reducing equivalents across the peroxisomal membrane, like the malate/oxaloacetate shuttle (Mettler and Beevers, 1980). Both systems would be redundant, explaining why the fatty acid degradation still proceeds in the pxn mutants. Interestingly, a similar phenotype was described for an Arabidopsis mutant defective in the redox shuttle mediated by both PMDHs. Although seedling establishment was severely impaired, the double mutant was still able to break down fatty acids, but in a highly decelerated mode (Pracharoenwattana et al., 2007). It was assumed that peroxisomal oxidoreductases, such as hydroxypyruvate reductase (HPR), allow the sufficient regeneration of NAD+ to support β-oxidation (Pracharoenwattana et al., 2010). However, the absence of HPR in the pmdh1 pmdh2 mutant did not cause a complete block in TAG breakdown, implying the presence of further NAD-oxidizing enzymes or the possibility to directly exchange NAD/NADH across the peroxisomal membrane (Pracharoenwattana et al., 2010). If NADH is a counter-exchange substrate for NAD+, it needs to be available for PXN in the peroxisomal matrix under physiological conditions; otherwise, the relevance of NADH transport can be excluded. The cellular NADH levels in seedlings are up to 10-fold lower than NAD+ (Wang and Pichersky, 2007; Quettier et al., 2008). However, the information about the intercellular concentrations is rare. In the yeast Pichia pastoris, a peroxisome-targeted probe that senses redox-related metabolites, such as glutathione, NAD(P)H and reactive oxygen species, indicates that the peroxisomal matrix is more reductive than the cytosol under induced fatty acid oxidation: the in vivo visualization of the peroxisomal redox state was performed when the yeast cells metabolized oleic acid as the sole carbon source (Yano et al., 2010). Transferring this observation to seedlings, elevated NADH levels in the peroxisomes during storage oil mobilization could favour the NAD+ import in exchange with NADH.

In sum, we discovered the peroxisomal NAD+ carrier that imports cytosolic NAD+ for β-oxidation. This transport protein would not have been identified in a forward-genetic screen for a conditional seedling-lethal phenotype, germination deficiency, or 2,4-DB or indole-3-butyric acid (IBA) resistance (Hayashi et al., 1998; Zolman et al., 2000; Footitt et al., 2002; Eastmond, 2006). For the future, we are interested to elucidate if, in addition to β-oxidation, other NAD+-dependent processes in peroxisomes, such as photorespiration, are affected by impaired NAD+ import. To date we can state that the pxn plants that were grown during this work under ambient CO2 conditions did not show a typical photorespiratory phenotype; however, a detailed metabolic analysis remains to be conducted.

Experimental Procedures


Chemicals were purchased from Sigma-Aldrich (http://www.sigmaaldrich.com). Reagents and enzymes for recombinant DNA techniques were obtained from Invitrogen (http://www.invitrogen.com), New England Biolabs (http://www.neb.com), Qiagen (http://www.qiagen.com), Fermentas (http://www.fermentas.com) and Promega (http://www.promega.com). The primers used in this study are listed in Table S2.

DNA and protein sequences

Sequences were retrieved from the Aramemnon database (http://aramemnon.uni-koeln.de) and Saccharomyces Genome Database (SGD) under the following accession numbers: At2g39970 (PXN), At3g05290 (PNC1), At5g27520 (PNC2), At2g47490 (NDT1), At1g25380 (NDT2), YPR128C (Ant1p), YIL006W (ScNDT1p) and YEL006W (ScNDT2p). The sequences were aligned via clustalw (http://www.ebi.ac.uk/Tools/msa/clustalw2). The alignment layout was performed with genedoc (http://www.nrbsc.org/gfx/genedoc).

Cloning procedures and protein biochemistry

Cloning steps were performed using standard molecular techniques (Sambrook et al., 1989). DNA sequences were verified by DNA sequencing (GATC Biotech, http://www.gatc-biotech.com). SDS-PAGE and immunoblot analysis was conducted as described in Sambrook et al. (1989). The following antibody combinations were used for immunodetection: penta-His antibody (Qiagen)/alkaline phosphatase (AP)-conjugated anti-mouse IgG (Promega) and anti-PXN-specific serum/AP-conjugated anti-rabbit IgG (Promega). To estimate molecular masses, the PageRuler Prestained Protein Ladder (New England Biolabs) was used. Protein concentrations were determined using a bichinchonic acid protein assay (ThermoFisher Scientific, http://www.thermofisher.com).

Plant material and growth conditions

Arabidopsis mutants were obtained from the ABRC at the Ohio State University: pxn-1 (GABI_046D01), pxn-2 (GABI_830A06), pxn-3 (SAIL_636F12) and pxa1-1 (CS3950, Zolman et al., 2001). Seeds were surface sterilized, stratified for 4 days at 4°C and germinated on 0.8% (w/v) agar-solidified half-strength MS medium (Duchefa, http://www.duchefa.com) supplemented with 1% (w/v) sucrose. Unless stated otherwise, plants were incubated in a 16-h-light/8-h-dark cycle (22/18°C) in growth chambers (100 μmol m−2 sec−1 light intensity). Then 14-day-old seedlings were transferred to soil for further growth under the same conditions.

Seed germination experiments were conducted on half-strength MS agar plates plus ± sucrose, and after 4 days stratification seeds were incubated under long-day conditions. For the analysis of seedling growth, plants were grown on half-strength MS agar plates supplemented with or without sucrose in constant darkness or in short-day conditions (8-h light/16-h dark cycles). To study the response to 2,4-DB and 2,4-D, seeds were grown on half-strength MS agar plates supplemented with or without 0.8 μm 2,4-DB or 0.23 μm 2,4-D (added from stocks dissolved in 100% (v/v) ethanol) under short-day conditions. After 6 days, seedlings were photographed and roots or hypocotyls were measured using imagej (http://rsbweb.nih.gov/ij). For fatty acid analysis, seeds were grown on half-strength MS agar in constant darkness or under short-day conditions. For the isolation of plant membrane isolation and lipid staining, seeds were grown on half-strength MS agar in constant darkness.

Statistical analysis

Unpaired Student’s t-tests were performed using prism 5.0d (GraphPad, http://www.graphpad.com/prism/prism.htm) to investigate the statistical significance observed between the responses of the wild type and pxn mutants. The change in question was termed ‘extremely significant’ if the P value was < 0.1% (P < 0.001), and is indicated by three asterisks.

Transient expression of YFP fusions in tobacco protoplast

For cloning the YFP-PXN fusion protein, the PXN coding sequences were amplified using P21/P22 and cloned via SmaI/BamHI for YFP fusion into pNL3 vector. pNL3 and the peroxisomal marker CFP-PTS1 were generated in Linka et al. (2008). Agrobacterium strain GV3101 (Koncz and Schell, 1986) was transformed with vectors encoding for either YFP-PXN or CFP-PTS1, respectively. Tobacco leaves were co-infiltrated (Romeis et al., 2001). Three days after co-infiltration protoplasts were isolated (Sheen, 1991; Siemens and Sacristan, 1995) and analysed with a Zeiss Laser scanning microscope 510 Meta (Zeiss, http://www.zeiss.com), using either a YFP filter (excitation, 514 nm; emission, 527 nm) or a CFP filter (excitation, 433 nm; emission, 475 nm). Digital images were processed using the LSM image browser 4.2 (Zeiss).

Cell-free protein expression using wheatgerm extract

Recombinant protein of PXN, NDT1, NDT2 and PNC1 were expressed in a cell-free wheat germ extract (WGE) system. For C-terminal His-tag fusion of the target proteins, a histidine linker (NL230/NL231) was introduced via BamHI/XbaI into pEU3a vector (CellFree Sciences, http://www.cfsciences.com). Coding sequences of the designated proteins were obtained by PCR using the following primer pairs: NL33/NL34 (PXN), KB37/KB38 (NDT1), KB40/KB41 (NDT2) and NL78/NL79 (PNC1). PCR products were inserted via NcoI/BamHI into the modified pEU3a-C-His vector.

Templates for in vitro transcription were generated from the pEU3a-based constructs via PCR using P73/P74. The resultant PCR products were purified using QIAquick Gel Extraction Kit (Qiagen). The in vitro translation was performed as described in Nozawa et al. (2007) using the lipid bilayer method in the presence of 1% (w/v) l-α-phosphatidylcholine vesicles and 0.04% (w/v) Brij-35. Recombinant proteins were desalted using Sephadex G-25 columns (GE Healthcare, http://www.gelifesciences.com) with 10 mm tricine-KOH, pH 7.6.

Reconstitution of transport activities into liposomes

Expressed proteins were reconstituted into liposomes by the freeze–thaw procedure (Kasahara and Hinkle, 1976). Uptake assays were started by the addition of radiolabelled [α-32P]NAD+ (Perkin Elmer, http://www.perkinelmer.com). For each measurement point, 150 μL of proteoliposomes was passed over Dowex AG1-X8 columns (Acetate form, 100–200 mesh, equilibrated with 150 mm sorbitol). The non-incorporated substrate was removed by binding to the anion-exchange columns. The flow-through containing the proteoliposomes was collected in scintillation vials filled with 10 ml of deionized water. The imported radiolabelled NAD+ was quantified by a liquid scintillation counter.

The time-dependent uptake data were fitted using nonlinear regression analysis based on one-phase exponential association. Kinetic constants were determined by measuring the initial velocity of each experiment. The Michaelis constant (KM) for NAD has been analysed with eight different external NAD+ concentrations (10 and 2 mm) and determined from the fitted substrate–velocity curve. The competitive inhibition constant (Ki) was assayed at a single substrate concentration with varying concentrations of a competitive inhibitor (16–8 mm). Ki values were calculated by nonlinear regression using a one-site competition-binding model. Analysis of all kinetic data was performed using GraphPad prism 5.0d.

Isolation of T-DNA insertion lines

PCR-based screening was used to identify homozygous T-DNA insertion lines for PXN using gene-specific primers (NL29/NL30 for pxn-1, NL319/NL320 for pxn-2 and NL291/NL292 for pxn-3) and primers for T-DNA/gene junction (NL29/P52 for pxn-1, NL320/P52 for pxn-2 and NL292/P49 for pxn-3). The position of the T-DNAs in the PXN gene was verified by DNA sequencing. As a positive control, amplification of ACTIN2 (ACT2, At3g18780) was performed using P33/P34.

Reverse transcriptase-polymerase chain reaction (RT-PCR)

RNA was extracted according to the method described by Chomczynski and Sacchi (1987), and was subjected to cDNA synthesis using Superscript II reverse transcriptase (Invitrogen). The presence of the PXN transcript was determined by PCR using P48/P49. DNA sequencing of T-DNA-specific PCR products confirmed the exact T-DNA position in the PXN gene. For the amplification of the truncated PXN transcript in the pxn-2 T-DNA insertion line, the primer pair P26/P49 was used. As a control for cDNA quality, ACT7 cDNA was amplified using P67/P68.

Fatty acid analysis

Fatty acids were analysed by gas chromatography - electron impact - time of flight - mass spectrometry (GC-EI-TOF-MS), following conversion to methyl esters after Browse et al. (1986). The fatty acid 17:0 was used as an internal standard to enable quantification.

Visualization of oil bodies

Lipid bodies of 4-day-old etiolated seedlings were stained with Nile Red, as previously described by Linka et al. (2008). Images were recorded with a Zeiss laser scanning microscope 510 Meta (Zeiss) (excitation, 488 nm; emission, 530–600 nm). Digital images were processed using the LSM image browser 4.2 (Zeiss).

Generation of PXN-specific polyclonal antibody

The hydrophobic loop between the third and fourth transmembrane domain of the PXN protein (123–201 aa) was amplified by PCR using NL43/NL44 and cloned into pET-16b (Merck Biosciences, http://www.merck.com) via XhoI/BamHI for N-terminal His-tag fusion. After transformation into BL21-CodonPlus (DE3)-RIL (Agilent Technologies, http://www.agilent.com), protein synthesis was induced with isopropyl-β-d-1-thiogalactopyranoside. Inclusion bodies were purified according to Schroers et al. (1997). Expression of the His-Loop fusion peptide was confirmed by immunoblot analysis using anti-His antibody (Qiagen). Recombinant His-tagged PXN-antigen was purified by using Ni-NTA agarose (Qiagen) and subjected for generation of polyclonal antiserum in rabbits (Pineda Antikörper Service, http://www.pineda-abservice.de).

Plant membrane isolation of etiolated seedlings

Membranes from 6-day-old etiolated seedlings were extracted in 100 mm tricine-KOH, pH 7.6, 4 mm EDTA, 4 mm DTT, 0.5% (v/v) β-mercaptoethanol, 6 mm ascorbate, 0.1% (w/v) polyvinylpolypyrrolidone, 0.5% (v/v) protease inhibitor cocktail for plant tissue (Sigma-Aldrich) and 0.1% (w/v) BSA (two volume extraction buffer per g fresh weight). Cell debris was removed by short centrifugation at 700 g. The supernatant was centrifuged at 10 000 g for 10 min and transferred to a new centrifuge tube. The membrane fraction was collected by ultracentrifugation (100 000 g for 45 min) and resuspended in 10 mm HEPES-KOH, pH 7.6, and 0.8 mm MgCl2. The membrane pellet was analysed by SDS-PAGE and immunoblot using PXN-specific antibody.

Accession Numbers

At2g39970 (PXN), At3g05290 (PNC1), At5g27520 (PNC2), At2g47490 (NDT1), At1g25380 NDT2), YPR128C (Ant1p), YIL006W (ScNDT1p), YEL006W (ScNDT2p), GABI_046D01 (pxn-1), GABI_830A06 (pxn-2), SAIL_636F12 (pxn-3), and CS3950 (pxa1-1).


This work was supported by the DFG-grant 1781/1-1 (to KB, NL and APMW). We thank Florian Hahn and Friederike Philipp for their help with plant experiments, Kirsten Abel for technical assistance and Elisabeth Klemp and Katrin L. Weber for their support with GC-MS analysis. We thank Johannes M. Herrmann for providing the COX2 antiserum.