Protein Phosphatase 2A Holoenzyme Is Targeted to Peroxisomes by Piggybacking and Positively Affects Peroxisomal β-Oxidation1[OPEN]

Protein phosphatase 2A targets peroxisomes and positively affects fatty acid oxidation. The eukaryotic, highly conserved serine (Ser)/threonine-specific protein phosphatase 2A (PP2A) functions as a heterotrimeric complex composed of a catalytic (C), scaffolding (A), and regulatory (B) subunit. In Arabidopsis (Arabidopsis thaliana), five, three, and 17 genes encode different C, A, and B subunits, respectively. We previously found that a B subunit, B′θ, localized to peroxisomes due to its C-terminal targeting signal Ser-Ser-leucine. This work shows that PP2A C2, C5, andA2 subunits interact and colocalize with B′θ in peroxisomes. C and A subunits lack peroxisomal targeting signals, and their peroxisomal import depends on B′θ and appears to occur by piggybacking transport. B′θ knockout mutants were impaired in peroxisomal β-oxidation as shown by developmental arrest of seedlings germinated without sucrose, accumulation of eicosenoic acid, and resistance to protoauxins indole-butyric acid and 2,4-dichlorophenoxybutyric acid. All of these observations strongly substantiate that a full PP2A complex is present in peroxisomes and positively affects β-oxidation of fatty acids and protoauxins.

Reversible phosphorylation of proteins is one of the most common mechanisms used in the regulation of biological processes and requires both a protein kinase and a protein phosphatase. Protein phosphatase 2A (PP2A) is a conserved Ser/Thr protein phosphatase and belongs to one of the major protein phosphatase families in eukaryotes (Farkas et al., 2007;Eichhorn et al., 2009;Uhrig et al., 2013;Lillo et al., 2014). PP2A plays important roles in plant metabolism, development, stress response, and signal transduction. PP2A holoenzymes are heterotrimeric complexes comprising catalytic (C), scaffolding (A), and regulatory (B) subunits, all encoded by multiple genes. The dimer of the C subunit (36 kD) and A subunit (65 kD) makes up the core enzyme of PP2A. This core enzyme associates with a third subunit, B type, which is essential for subcellular localization and substrate specificity. Through selective incorporation of different B subunits, the complexes are recruited to specific subcellular compartments that define the sphere of active function for the various complexes (Janssens and Goris, 2001;Farkas et al., 2007). In Arabidopsis (Arabidopsis thaliana), three genes (A1 [also called Roots curl in naphthylphthalamic acid1]-A3) code for A subunits, 17 genes for B subunits, and five genes for C subunits (Janssens and Goris, 2001;Farkas et al., 2007). Theoretically, up to 255 different PP2A heterotrimers can be formed. B subunits are classified into B (a and b), B9 (a, b, g, d, «, z, h, u, and k), and B99 (a, b, g, d, «, and TONNEAU2/FASS) nonrelated families (Janssens and Goris, 2001;Farkas et al., 2007). The large number of subunits certainly indicates the importance of PP2A in a myriad of processes. For example, the B family members were shown to be important for dephosphorylation and activation of nitrate reductase (Heidari et al., 2011). B9a and B9b from the B9 family were shown to interact with BRASSINOSTEROID INSENSITIVE1, a transcription factor of the brassinosteroid signaling pathway (Tang et al., 2011). Members of the B99 family were found to interact with a key enzyme of isoprenoid synthesis, 3-hydroxy-3methylglutaryl-CoA reductase (Leivar et al., 2011). The B9h subfamily from B9 comprises the close homologs B9g, B9z, B9h, and B9u (Terol et al., 2002;Farkas et al., 2007;Eichhorn et al., 2009), which target the different organelles, nucleus, mitochondria, nucleus and nucleolus, and peroxisomes, respectively (Matre et al., 2009). The most studied member from this subfamily is B9g, which prevents unnecessary defense reactions under low light in Arabidopsis (Trotta et al., 2011a(Trotta et al., , 2011b. In addition, b9g plants display enhanced expression of FLOWERING LOCUS C and flowered later than wild-type plants (Heidari et al., 2013).
Peroxisomes are single-membrane-bound organelles present in all major groups of eukaryotes. They were first discovered as compartments containing hydrogen peroxide-generating oxidases together with catalases that degrade hydrogen peroxide into molecular oxygen and water (De Duve and Baudhuin, 1966;van den Bosch et al., 1992;Kaur et al., 2009). In higher plants, fatty acid b-oxidation takes place in peroxisomes, and all fatty acids can be completely degraded in peroxisomes, whereas in mammalian cells, short-chain fatty acids are transported from peroxisomes to mitochondria for completion of b-oxidation (Mano and Nishimura, 2005). Fatty acid b-oxidation and hydrogen peroxide detoxification are two conserved functions of peroxisomes in all eukaryotes, but specialized functions have also been identified. For example, plant glyoxysomes are specialized peroxisomes in germinating seeds that harbor the glyoxylate cycle, and plant leaf peroxisomes take part in photorespiratory glycolate metabolism and biosynthesis of the plant hormones indole acetic acid, salicylic acid, and jasmonic acid. Peroxisomes are devoid of DNA, and their complete set of matrix proteins are encoded in the nucleus and synthesized in the cytosol before being imported into peroxisomes (Kaur et al., 2009). Peroxisomal matrix proteins contain two types of peroxisome targeting signals, PTS1 and PTS2, by which they are directed to peroxisomes. PTS1 exists in the majority of proteins as a C-terminal conserved tripeptide with the prototype serine-lysine-leucine at C terminus (SKL.). PTS2 is a nonapeptide with RLx5HL as a prototype sequence. Peroxin5 (PEX5) and PEX7 are soluble receptors that recognize proteins with PTS1 and PTS2, respectively, and direct them to the peroxisome membrane (Rehling et al., 1996;Kragler et al., 1998;Woodward and Bartel, 2005;Kaur et al., 2009). There are also examples of peroxisomal proteins with alternative targeting signals and transport by piggybacking through other proteins bearing PTSs (Kaur et al., 2009).
Regulation of peroxisomal metabolism by reversible phosphorylation has not yet been clearly demonstrated, but the presence of protein kinases in peroxisomes has been predicted due to putative targeting signals or implicated from proteome studies (Dammann et al., 2003;Reumann, 2004Reumann, , 2011. Only one protein kinase (calciumdependent protein kinase1) has so far been verified to target peroxisomes (Dammann et al., 2003;Coca and San Segundo, 2010). Besides calcium-dependent protein kinase1, three additional protein kinases were reported to have functional PTS1 domains, but the full-length proteins failed to be imported to peroxisomes (Kaur et al., 2009;Lingner et al., 2011;Sørhagen et al., 2013). No protein phosphatase has so far been identified in peroxisomes except for a B subunit (B9u) of PP2A that targeted peroxisomes when fused N-terminally to enhanced yellow fluorescent protein (EYFP) in transiently transformed cells (Matre et al., 2009). Whereas B9u has a PTS1 serine-serine-leucine (SSL) in the extreme C-terminal end, none of the C and A subunits have putative peroxisomal targeting signals. In this work, the ability of B9u to interact with C and A subunits and target them to peroxisomes was investigated. Reverse genetics was applied to gain insight into the effects of B9u on peroxisomal metabolism and related phenotypes.

B9u Interacts with A2, C2, and C5 Subunits in Peroxisomes
Highly purified peroxisomes were isolated using a Percoll gradient followed by a Suc density gradient (Reumann and Singhal, 2014). In extracts of these peroxisomes, a PP2A C subunit was detected by western blotting using antibodies specific for either methylated or demethylated C subunit ( Fig. 1A; Supplemental Fig. S1A). It cannot be excluded that the antibody toward methylated PP2A would not pick up the PP2A-like enzyme PP4, although much less expressed, but to ensure detection of PP2A, another antibody was also used that specifically picked up (demethylated) PP2A. Purity of the peroxisomal extracts was tested by western-blot analysis using antibodies specific for cytosolic phosphoenolpyruvate carboxylase (PEPC) and the peroxisomal catalase (Fig. 1B). Bimolecular fluorescence complementation (BiFC; Waadt et al., 2008) was used to identify C and A subunits that interact with B9u in peroxisomes. B9u complementary DNA (cDNA) was subcloned in the vectors Venus Yellow N-terminus enhanced yellow fluorescent protein right [pVYNE(R)] and Venus Yellow C-terminus enhanced yellow fluorescent protein right [pVYCE(R)] to give a free C terminus (Supplemental Fig. S1B). C1 to C5 (except C3, which is highly similar to C4), A1, and A3 were subcloned into vectors to give free N-terminal ends, as were A1 to A3 to give a free C-terminal end (Supplemental Fig. S1B). Microscopy analysis of Arabidopsis mesophyll protoplasts transformed with B9u::pVYNE(R) together with   (Leivar et al., 2011). Coexpression of B9uDSSL::VYCE with C2::VYNE (V) or C5::VYNE (VI) shows cytosolic fluorescence in protoplasts. Peroxisomes were labeled with gMDH-CFP (Fulda et al., 2002). The cyan fluorescence color was converted to red. Scale bars = 5 mM. A1::pVYCE or A3::pVYCE revealed a low background fluorescence that remained mostly cytosolic and was probably caused by some yellow fluorescent protein (YFP; Venus variant) self-assembly. Similar observations were made upon cotransformation by B9u::pVYCE(R) with A1::pVYNE(R) or A3::pVYNE(R) (24-and 48-h incubation time). In contrast to such weak background fluorescence, use of B9u::pVYCE(R) with A2::pVYNE(R) revealed strong fluorescence signals, which were observed in organelle-like structures (Fig. 2I).
To determine the identity of these structures, the peroxisomal marker glyoxysomal malate dehydrogenase (gMDH)-cyano fluorescent protein (CFP; Kim and Smith, 1994) was coexpressed with B9u::VYCE and A2::VYNE. The YFP fluorescence from BiFC interaction coincided with CFP fluorescence in the same structures (Fig. 2I). From these results, we conclude that B9u interacts with A2, and they colocalize to peroxisomes.
C subunits of PP2A are divided into two subgroups in Arabidopsis (group 1: C1, C2, and C5; and group 2: C3 and C4; Farkas et al., 2007). Based on gene expression analyses, C2 and C5 appeared to be good candidates for interacting with B9u (Zimmermann et al., 2004;Jen et al., 2006;Supplemental Fig. S2). Coexpression of C2::VYNE and B9u::VYCE(R) (Fig. 2II) and C2::VYCE and B9u::VYNE(R) (Fig. 2III) in protoplasts showed YFP fluorescence that coincided with gMDH-CFP-labeled peroxisomes (Fig. 2, II and III). Coexpression of C5::VYNE and B9u::VYCE(R) also revealed interaction in peroxisomes (Fig. 2IV). No YFP fluorescence was detected for BiFC vectors combining C1 or C4 with B9u. To test the importance of the PTS1 of B9u, truncated B9u (DSSL) was used. When B9uDSSL::VYCE(R) was coexpressed with C2::VYNE or C5::VYNE, fluorescence remained in the cytosol (Fig. 2,V and VI), showing that the B9u PTS1 is necessary to direct the interacting proteins into peroxisomes. We also investigated the BiFC in onion (Allium cepa) epidermal cells after biolistic transformation by B9u::pVYCE(R) and A2::pVYNE(R) and with C2, C5, C1, or C4::pVYNE for 24 to 48 h. Peroxisomal localization was confirmed in the case of B9u interacting with A2 and C5 (Supplemental Fig. S3), whereas no fluorescence was observed with C1, C2, or C4. The lack of signal from B9u::pVYCE(R) combined with C2::pVYNE in onion epidermal cells indicates that the interaction of B9u and C2 is weaker than the interaction between B9u and C5, at least under some conditions. Arabidopsis plants were also stably transformed with a binary vector, and genomic presence of EYFP-B9u was confirmed by PCR. Fluorescence microscopy revealed the presence of EYFP-B9u in leaves in punctate  A, qRT-PCR shows higher levels of B9u transcript in senescent leaves, roots, and flowers than in young and mature leaves. Expression levels are given relative to the expression in mature leaves (set to one). Vertical bars represent SEs. B, Expression levels of B9u mRNA (absolute values) in different tissues. The graph was made using Arabidopsis electronic fluorescent pictograph (eFP) browser data (Winter et al., 2007). C, Schematic representation of B9u encoding gene with position of the T-DNA insertions and TaqMan primers used for analyzing expression. UTR, Untranslated region. structures resembling peroxisomes (Fig. 3I). When the F3 progenies were allowed to grow in liquid onehalf-strength Murashige and Skoog medium, fluorescence was also detected in punctate structures in roots ( Fig. 3II) and was absent in nontransformed roots (Fig. 3III).

Expression Analysis Indicates the Importance of B9u in Specific Tissue and under Stressful Conditions
Using quantitative reverse transcription (qRT)-PCR, B9u expression was found to be higher in roots and reproductive and senescent leaves than in young and mature leaves (Fig. 4A). These results are consistent with the publicly available microarray data (Fig. 4B). It is also seen from the Genevestigator database that expression of B9u is high in dessicated seeds, decreases in stratified seeds, becomes stable throughout early seedling germination steps, and becomes high during senescence stages (Supplemental Fig. S4). B9u expression is up-regulated in response to several plant pathogens, such as virulent Pseudomonas syringae pathovars (Psm and Pst), the nematode Meloidogyne incognita, the oomycete Phytophthora parasitica, and fungi (Phytophthora infestans and Blumeria graminis; Supplemental Figs. S4 and S5). Several microarray experiments that include treatment by the bacterial elicitor (flagellin 22) also show up-regulation of B9u, C2, and C5 expression (Supplemental Figs. S2D and S4). In contrast to its up-regulation due to biotic stress, Figure 5. Evolutionary relationship of PP2A-B9u homologs. B9u is conserved in plants, whereas mammals and yeast (Saccharomyces cerevisiae) contain homologs of the B9 family. Arabidopsis B9 family members are highlighted by a red circle. The homologs that have a conserved peroxisome PTS1 are highlighted by a green triangle, whereas the blue triangle marks a change in SSL. to SSS.. The phylogram was generated by MEGA6 (Tamura et al., 2013) using the neighbor-joining method of Saitou and Nei (1987). The tree is drawn with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method and are in the units of the number of amino acid substitutions per site. Figure 6. Suc dependence assay of b9u mutants. A, Hypocotyl length of seedlings grown for 6 d in the dark (I and III) or 8 h of light/16 h of dark (II and IV) on one-half-strength Linsmeier and Skoog (LS) medium with or without 1% Suc. The homozygous lines b9u-1 (I and II) and b9u-2 (III and IV) were tested. The experiment was repeated three times, and error bars represent SD. B, Representative images showing developmental defects in b9u-1, b9u-2, and pex14 (sugar dependence control) seedlings 6 d after germination in short days without Suc (I). Impaired growth was not seen for medium with Suc (II). After 12 d on Suc-free medium, the differences between mutants and the wild type were still clear, but some mutant seedlings started to recover (III and IV). Scale bars = 5 mm.
B9u expression is down-regulated in response to abiotic stresses (drought, heat, and salt; Supplemental Fig. S4).

B9u Has a Conserved PTS1 within the Tribe Camelineae
Using the BLAST search tool (protein BLAST, National Center for Biotechnology Information), B9u was found to be conserved in plants, whereas most of the other eukaryotes had homologs of the B9 family members, not specifically for B9u ( Fig. 5; Supplemental Fig. S6). Although B9u is conserved in plants, most of its homologs lack the peroxisomal signal SSL. or any other experimentally verified PTS. Two B9u homologs from the same family/tribe (Brassicaceae/Camelineae) have a conserved SSL., one from the same genus Arabidopsis (Arabidopsis lyrata) and the other from the genus Capsella (Capsella rubella; Fig. 5). Eutrema salsugineum is a halophytic species in the Brassicaceae family , and it has a highly similar homolog of Arabidopsis B9u with a conserved C-terminal domain but with a change of SSL. to SSS. (Fig. 5). This mutation most likely will abolish targeting to peroxisomes, as it lowers the prediction score threshold from 0.368 for SSL. to 20.472 for SSS. based on the latest prediction algorithm of plant PTS1 proteins (Lingner et al., 2011).

The b9u Mutant Seedlings Show a Sugar Dependence Phenotype
Oilseed plants, such as Arabidopsis, rely on fatty acid catabolism during the early seedling stage. In mutants that do not properly catabolize fatty acids, growth is hampered unless an alternative energy source is provided (Mano and Nishimura, 2005). Two homozygous transfer DNA (T-DNA) insertion lines, SAIL_300_ B01 (b9u-1; Matre et al., 2009) and SALK_201885 (b9u-2), with inserts in the first coding exon (Fig. 4C), were used to study the effects of B9u inactivation. PEX14 is a peroxisomal membrane protein involved in protein import, and the pex14 mutant was used as a sugar dependence control (Orth et al., 2007;Zhang and Hu, 2010). Hypocotyl length of seedlings grown in continuous darkness (Fig. 6A, I) and in short days (Fig. 6A, II) in the presence and absence of Suc was measured 6 d after germination. On Suc-free medium, hypocotyl elongation was strongly inhibited in pex14 and b9u-1 seedlings as compared to Suc-containing medium, whereas only a small effect was seen for the wild type Columbia (Col-0; Fig. 6A, I and II). The hypocotyl-length phenotype was confirmed by b9u-2 seedlings (Fig. 6A, III and IV). Both b9u-1 and b9u-2 mutants showed seedling development defects on Suc-free medium 6 and 12 d after germination (Fig. 6B, I, III, and IV), whereas this was not the case on Suc-containing medium (Fig. 6B, II). However, some seedlings were able to recover after their photosynthesis had started but still were not comparable with the wild type (Fig. 6B, III and IV). These data indicate that b9u seedlings are impaired in triacylglycerol (TAG) metabolism. Moreover, obtaining the same phenotype for both b9u-1 and b9u-2 lines confirms that the disruption of B9u is the cause of the observed phenotypes.

The b9u Mutant Seedlings Are Impaired in Their Response to Protoauxins
Indole-3-butyric acid (IBA) and 2,4-dichlorophenoxybutyric acid (2,4-DB) are protoauxins that can be transformed by peroxisomal b-oxidation to the bioactive auxins indole-3-acetic acid (IAA) and 2,4-dichlorophenoxyacetic acid, respectively. Mutants impaired in b-oxidation show resistance to the inhibitory effect of these compounds on primary root elongation (Zolman et al., 2001;Mano and Nishimura, 2005). In the presence of low levels of IBA, b9u-1 seedlings showed significant resistance to auxin in comparison with the wild type Col-0 (Fig. 7A). Moreover, b9u-1 seedlings showed resistance to 2,4-DB (Fig. 7B). The pex14 mutant was also resistant to the inhibition of root elongation by IBA and 2,4-DB, consistent with previous reports (Zhang and Hu, 2010). These results support the hypothesis that the b9u mutant seedlings are impaired in transforming protoauxins into active auxins, especially since the seedlings responded as the wild type upon treatment with IAA (Fig. 7C).

The b9u Mutant Seedlings Lack Normal TAG Mobilization
In Arabidopsis, eicosenoic acid (C20:1), which is a component of TAG, is considered as a convenient reporter of TAG breakdown in seeds (Lemieux et al., 1990;Fulda et al., 2004). Normally, C20:1 is rapidly degraded during early seedling growth and is almost absent 4 to 5 d after imbibition of the seeds (Fulda et al., 2004). Mutants blocked in b-oxidation failed to degrade TAG and suffered during early seedling growth (Zolman et al., 2001). The levels of fatty acids in the wild type and b9u-1 seedlings were followed and showed that C20:1 in the wild type was high during the first days, then declined rapidly as expected, whereas in the mutant, the level was lower during the first days, then increased (Fig. 8D). Not only C20:1 but also other long-chain fatty acids were noticeably lower in b9u-1 than in the wild type during the first 2 d of testing (Fig. 8, A-E). The data indicate that b9u-1 cannot mobilize TAG normally and underpin that B9u has an influence on peroxisomal fatty acid b-oxidation.

The b9u Mutant Seedlings Have Unique Phosphorylated Proteins Involved in TAG Metabolism
Very recent studies experimentally identified phosphopeptides related to peroxisomal b-oxidation, glyoxysomal cycle, and PTS1 import proteins and showed that several of the proteins involved in b-oxidation are in fact phosphorylated ( Fig. 9; Supplemental Table S1). Most of the identified proteins are phosphorylated at Ser or Thr residues and, hence, can be potential substrates for PP2A. In an initial attempt to identify peroxisomal targets of PP2A, total protein extracts were prepared from 7-d-old b9u-1 mutant and wild-type Col-0 seedlings grown in Suc-free medium on short days. Tryptic digests were enriched on a titanium dioxide column and subjected to liquid chromatography (LC)-tandem mass spectrometry (MS/MS) analysis on an LTQ-Orbitrap Elite mass spectrometer (Thermo Scientific). The analysis identified nine phosphopeptides from eight enzymes involved in TAG metabolism. Importantly, these phosphopeptides were uniquely detected in the b'u-1 mutant ( Table I). All except one of the peptides were phosphorylated at Ser or Thr residues, thus constituting potential substrates for PP2A. Three of the phosphopeptides were previously unidentified, two of which matched acyl-CoA synthetase (LACS5), which is involved in the first step of peroxisomal b-oxidation, and one of which matched peroxisomal ketoacyl-CoA thiolase (KAT1; Fig. 9; Table I). These two peroxisomal b-oxidation-related proteins show high potential to be PP2A substrates in peroxisomes (Fig. 9).

DISCUSSION
Inspection of stably transformed plants confirmed the presence of B9u in peroxisomes, as was seen in both leaves and roots (Fig. 3). No PP2A subunits had been reported as part of the proteome of isolated peroxisomes when analyzed by mass spectrometry (MS; Fukao et al., 2002;Reumann et al., 2007Reumann et al., , 2009Eubel et al., 2008;Quan et al., 2013), but sensitive immunological methods used in this work showed the presence of the PP2A C subunits in extracts of highly purified peroxisomes (Fig. 1A).
Localization of PP2A C and A subunits to peroxisomes was further confirmed by BiFC analysis of B9u in combination with each of the two other types of subunits. Out of the five C subunits present in Arabidopsis, only C2 and C5 gave positive BiFC results (Fig. 2, II-IV).  (Farkas et al., 2007;Uhrig et al., 2013). Regulation of peroxisomal b-oxidation is a unique function of PP2A that was approved in this study. Several peroxisomal proteins, involved in fatty acid b-oxidation and the glyoxylate cycle, have been shown experimentally to be phosphorylated (Supplemental Table S1), as highlighted in red for the type of phosphorylated amino acid (S, Ser; T, Thr; and Y, Tyr). As highlighted by the red arrows, b-oxidation steps including LACS5 and KAT1 were found to be potential substrates for a peroxisomal PP2A heterotrimer based on the differences between phosphoproteomes of the b'u mutant and the wild type (this work). For more details, see Table I. The protein abbreviations are as follows: ACX, acyl-CoA oxidase; AIM, enoyl-CoA hydratase; CSY, citrate synthase; ICL, isocitrate lyase; MDH, malate dehydrogenase; MFP, multifunctional protein; and MLS, malate synthase. In glyoxysomes, the acetyl-CoA produced by fatty acid b-oxidation is used as a substrate for the glyoxylate cycle, where succinate is produced and transferred to mitochondria to be metabolized for energy production. Moreover, the peroxisomal ATP-binding cassette transporter (PXA1) and the PEX5, PEX13, and PEX14 were found to be phosphorylated (Supplemental Table S1). The pathway of fatty acid b-oxidation and glyoxylate cycle is modified from the online database (Karp et al., 2005).
Interactions of B9u with C5 were more persistent than with C2 under some conditions, indicating that C5 was the more important C subunit in peroxisomes. We also concluded that C2 interacts with B9u but probably under special conditions because they failed to interact in onion epidermal cells.
PP2A C and A subunits are constitutively expressed in many organisms; interestingly, however, the Arabidopsis C5 subunit has very variable expression levels in different tissues and under different conditions. In accordance with the BiFC results, microarray expression data also pointed to C5 as the more important candidate compared with C2 for interacting with B9u (Supplemental Figs. S2, S4, and S5). Of the three A subunits, only A2 targeted peroxisomes (Fig. 2I). Altogether, the experiments strongly support the hypothesis that a full PP2A complex is present in peroxisomes. Peroxisomes are distinguished from other organelles by their capacity to import fully folded proteins. It has been shown for some proteins that, although they do not have a PTS, they are able to enter peroxisomes by being carried or piggybacked by proteins that do possess a PTS (Lee et al., 1997;Kaur et al., 2009). A and C subunits lack endogenous PTSs and are generally found throughout the cell (Koroleva et al., 2005;Michniewicz et al., 2007;Tran et al., 2012). Their import into peroxisomes clearly depended on the B9u PTS1 because truncation of the B9u C terminus abolished their peroxisomal targeting (Fig. 2, V and VI). Taken together, we conclude that A2, C2, and C5 interact with B9u in the cytosol and are imported by a piggybacking process into peroxisomes.
Although B9u is conserved among plants, a peroxisomal targeting signal was only predicted in very close relatives of Arabidopsis (Fig. 5). It is not clear why these plants in particular have evolved a PP2A specialized for peroxisomes. When speculating on a function for PP2A in Arabidopsis peroxisomes, it became clear that expression data as well as features of b9u knockout plants were compatible with PP2A influencing fatty acid b-oxidation. Peroxisomal b-oxidation is activated during senescence, where it is thought to have a role in the turnover of lipids derived from membranes of various organelles (Mano and Nishimura, 2005). The expression of most genes encoding peroxisomal proteins involved in b-oxidation is up-regulated during senescence (Supplemental Fig.  S5). Furthermore, as an oilseed plant, Arabidopsis needs to rapidly metabolize TAGs into sugars for translocation to growing tissues during the early seedling stage. B9u loss-of-function mutants showed a sugar dependence phenotype typically of impaired fatty acid breakdown (Fig. 6). Plant peroxisomes are also the site of some steps in hormone synthesis, which requires a cycle of b-oxidation. It is known that the effect of the protohormones IBA and 2,4-DB on primary root length will be weakened in b-oxidation-defective mutants because of the inability of such mutants to metabolize these compounds into IAA and 2,4dichlorophenoxyacetic acid (Mano and Nishimura, 2005). The b9u mutant seedlings showed resistance to both IBA and 2,4-DB (Fig. 7). The defect in b-oxidation was further confirmed by following the level of C20:1 after germination, which was clearly different in the b9u mutant and the wild type (Fig.  8D). This aligns with the reported defect in peroxisome function in kat2 mutant ( Fig. 9; Supplemental Table S1). In the kat2 mutant, lipid body degradation and TAG mobilization during seedling growth were severely hampered, and it was suggested that the metabolism of peroxisomes would also influence the metabolism in the lipid bodies (Germain et al., 2001).
Transport of fatty acids and protoauxins into peroxisomes has been suggested to occur by the same transporter, PXA1 (Zolman et al., 2001), which has five known Ser/Thr phosphorylation sites (Supplemental Table S1); hence, a possible function of PP2A would be to activate PXA1. Another possibility is that PP2A acts directly on enzymes in the peroxisomal matrix that are Table I. List of identified b-oxidation-related enzymes uniquely detected in the b9u-1 phosphoproteome LC-MS/MS analysis of phosphoproteome differences between b9u-1 and wild-type seedlings grown in Suc-free media. The phosphopeptides related to peroxisomal b-oxidation enzymes were found once in three replicates. The detected peptide and its q value are presented. The phosphorylated residue is shown in boldface for the type of amino acid (S, Ser; T, Thr; and Y, Tyr). AGI, Arabidopsis Genome Initiative; GDSL, glycine, aspartate, serine, and leucine around the active site serine. involved in b-oxidation. This possibility is supported in this study by comparing phosphoproteomes of the b9u-1 mutant and the wild type during seedling growth. Our experiments revealed two peroxisomal matrix proteins (LACS5 and KAT1) involved in b-oxidation that were phosphorylated in the mutant but not in the wild type. Finally, other possibilities, such as PP2A effects on import of enzymes of b-oxidation, cannot be excluded.
Overall, this work shows that Arabidopsis and some close relatives have developed a PP2A B subunit (B9u) with a PTS1 peroxisomal targeting signal. Using Arabidopsis, we have shown that this signal is essential for targeting all three types of PP2A building blocks, C, A, and B subunits, to peroxisomes. Examination of b9u mutant lines strongly supports a function of PP2A in peroxisomal processing of protoauxins and fatty acids. , and At1g13320 [U25062]) were obtained from the Arabidopsis Biological Resource Center. The coding sequences of the genes were cloned by PCR and subcloned via SpeI and XhoI restriction sites into the multicolor BiFC vectors, pVYCE, pVYCE(R), pVYNE, and pVYNE(R) (Waadt et al., 2008), to create protein fusions with the N-or C-terminal fragments of YFP (Venus variant) at the N-terminal end or C-terminal end of the gene (Supplemental Fig. S1B).

Transformation and Microscopy
Arabidopsis (Col-0) wild-type seeds were sown on soil and stratified for 2 d, then transferred to short days (8 h of light) and irrigated once a week with complete Hoagland solution. Plants 3 to 4 weeks old were used for protoplast isolation. For BiFC analysis in Arabidopsis mesophyll protoplasts, plasmids for PP2A subunits and peroxisomal marker proteins were cotransformed into protoplasts using a polyethylene glycol transformation protocol (Yoo et al., 2007). For BiFC analysis in epidermal cells, plasmids were transiently introduced into onion (Allium cepa 'Kepaløk') by a heliumdriven particle accelerator (PDS/1000; Bio-Rad) with all basic adjustments set according to the manufacturer's recommendations. The bombarded onion cells were incubated for 1 to 2 d in the dark at room temperature and then observed under the microscope. Peroxisomal markers used were gMDH-CFP that contains the PTS2 sequence of gMDH from cucumber (Cucumis sativus; Kim and Smith, 1994) and Red fluorescent protein-SKL (Matre et al., 2009). Microscopy was carried out on a Nikon TE-2000U inverted fluorescence microscope equipped with an Exfo X-Cite 120 fluorescence illumination system and filters for CFP (exciter, S436/10; and emitter, S470/30), YFP (exciter, HQ500/20; and emitter, S535/30), and a special red chlorophyll autofluorescence filter set (exciter, HQ630/39; and emitter, HQ680/40; Chroma Technology). Images were captured using a Hamamatsu Orca ER 1394 cooled CCD camera. Volocity II software (Improvision, Coventry, UK) was used to capture 0.5 Z-sections to generate extended focus images. Images were subsequently processed for optimal presentation with Adobe Photoshop version 9.0 (Adobe Systems).

Generation of Transgenic Lines
To generate an overexpressor line of N-terminally fused protein, genespecific primers were used to amplify full-length Arabidopsis cDNA of B9u, and were cloned in the pGEMT-EYFP vector. Subsequently, the available EYFP-fused fragment was excised and subcloned into the 35S promotercontaining binary vector pBA002. The construct was transformed into Agrobacterium tumefaciens nopalin strain from A208 via the freeze-thaw method. Arabidopsis Col-0 was transformed by the floral dip method (Clough and Bent, 1998). Seeds were screened on one-half-strength Murashige and Skoog agar plates containing 10 mg mL 21 phosphinotricin. Phosphinotricin-resistant seedlings were selected 10 to 14 d after germination. Successful transformation was validated by isolation of genomic DNA of the primary transformants and using primers upstream (forward) and downstream (reverse) of the EYFP-B9u insertion sites in the vector.

Peroxisome Isolation and Western-Blot Analysis
Leaves were harvested from 4-to 6-week-old Arabidopsis plants grown in a 12-h photoperiod, and peroxisomes were isolated as described previously . Leaf extracts were obtained by direct grinding of leaf tissue in Laemmli buffer. Leaf extracts or peroxisomes were mixed with loading buffer (Laemmli buffer) and resolved in SDS-PAGE. Western-blot detection of PP2A C subunit was performed with the specified monoclonal antibodies for the methylated and demethylated C-terminal end of PP2A C subunits (Santa Cruz Biotechnology; sc-81603 and sc-80990). Western-blot comparison between isolated peroxisomes and total protein extract was performed with polyclonal anti-Catalase and anti-PEPC (Agrisera AB). Equal amounts of protein from total leaf extract and precipitated peroxisomal protein were used for the comparison. Total protein extraction from 6-week-old leaves was performed using the TCA-acetone method. Proteins from isolated peroxisomal preparations were precipitated by the chloroform-methanol method (Wessel and Flügge, 1984). Protein concentrations were measured using the Bradford protein assay (Bradford, 1976).

Plant Growth
Arabidopsis mutant lines SAIL_300_B01 (Sessions et al., 2002), SALK_201885 (Alonso et al., 2003), SAIL_890_C10 (Sessions et al., 2002), and FLAG_185E04 (Samson et al., 2002), for B9u, were obtained from the European Arabidopsis Stock Centre. Mutant selection was done by PCR using primers for T-DNA insertion lines recommended at the Salk Institute Web site SIGnAL (http://signal.salk.edu/tdnaprimers.2.html). The pex14 mutant (Orth et al., 2007;Zhang and Hu, 2010) seeds were kindly provided by Jianping Hu (Michigan State University). For plant material grown on soil, seeds were sown directly in a regular soil plant mix. Seeds were stratified at 4°C for 2 d and then transferred to standard growth conditions. During germination and growth, plants were placed at 22°C under artificial light in short-day (8 h of light/16 h of dark), 12 h of light/12 h of dark, or long-day (16 h of light/8 h of dark) regimens.

Sugar Dependence and 2,4-DB/IBA Response Assays
For sugar dependence analysis, seeds of the wild-type Col-0 and mutants were sown on one-half-strength LS medium with vitamins (Caisson Labs) with or without 1% (w/v) Suc and stratified in the dark at 4°C for 2 d before being transferred to darkness or short-day conditions (8 h of light/16 h of dark) at 20°C. Six-day-old seedlings were scanned using an EPSON scanner (http://www.epson.com), and hypocotyl length was measured using ImageJ (Schneider et al., 2012; http://rsb.info.nih.gov/ij/). To study the response to protoauxins, 2,4-DB or IBA was added to one-half-strength LS agar medium with 0.5% (w/v) Suc. Seeds were sown and stratified for 2 d, then kept in continuous light for 7 d. Length of the primary root was measured using ImageJ (Zolman et al., 2001;Zhang and Hu, 2010).

qRT-PCR
Total RNA was isolated using the RNeasy Plant Mini Kit and treated with on-column DNaseI digestion (Qiagen). Afterward, RNA was treated in solution with DNaseI (Life Technologies) and precipitated by ammonium acetate (7.5 M) and ethanol. The cDNA was synthesized using the High Capacity cDNA Archive Kit (Applied Biosystems). Real-time PCR reactions were assayed using the ABI 7300 Fast Real-Time PCR System. The reaction volume was 25 mL containing 12.5 mL of TaqMan buffer (Applied Biosystems; includes 6-Carboxyl-X-Rhodamine as a passive reference dye), 8.75 mL of water, 2.5 mL of cDNA, and 1.25 mL of primers. Primers were predesigned TaqMan Gene Expressions for Plant Physiol. Vol. 167, 2015 503 B9u At1g13460 (At02780124; Applied Biosystems identification no.). The TaqMan assay is based on a light signal from each transcript copy formed and allows the comparison of expression levels between the various genes. Standard cycling conditions (2 min at 50°C, 10 min at 95°C, and 40 cycles altering between 15 s at 95°C and 1 min at 60°C) were used. Real-time PCR products were analyzed by Sequence Detection Software version 1.3. The comparative threshold cycle method for relative quantification was used with ACTIN8 At1g49240 (At02270958).

Fatty Acid Analysis
Total fatty acids were extracted (Browse et al., 1986) and quantified by gas chromatography. Heptadecanoic acid methyl ester was obtained from Sigma-Aldrich and served as the internal standard. In brief, the plant materials were treated by 1 mL of 1 M methanolic HCL reagent (dilution 3 M methanolic HCL [Sigma-Aldrich] by methanol + 5% [v/v] 2,2-dimethoxypropane) and heated (80°C) for 1 h in glass tubes sealed with Teflon. After the samples were allowed to cool down, 0.3 and 1 mL of hexane and 0.9% (w/v) sodium chloride, respectively, were added and vigorously vortexed to release fatty acid methyl esters. Centrifugation at 1000g for 30 s was applied to allow transfer of the upper hexane layer for gas chromatography-mass spectrometry/flame ionization detector analysis. An Agilent GC6890 connected to an Agilent 5975 mass selective detector (Agilent Technologies) and equipped with a J&W HP-5MS column (19091S-433; Agilent Technologies) was used. Flame ionization detector quantification using the column type (J&W HP-5, 19091J-413; Agilent Technologies) was applied. The obtained chromatograms were compared with different databases, and the individual fatty acids were monitored based on retention time and quantified by internal standard calibration.

Phosphoproteomics
Total protein from 7-d-old seedlings of the wild type and b9u-1 (grown on one-half-strength LS medium without Suc) was extracted using TCA-acetone precipitation. The precipitate was dissolved in 100 mL of 50 mM NH 4 HCO 3 and 20 mM dithiothreitol and heated for 30 min at 60°C. The sample was cooled to room temperature, and 20 mL of 1 M iodoacetamide was added and incubated for 20 min in the dark at room temperature. The reduced and alkylated sample was treated with trypsin (1:50 trypsin:protein ratio), and the proteins were digested overnight at 37°C. For total protein analysis, the digested sample was made acidic, and a C18 purification step was performed (Rappsilber et al., 2003). For phosphopeptide enrichment, the sample was treated with titanium dioxide (Engholm-Keller and Larsen, 2011). For both total protein and phosphopeptide analyses, the dried peptides were dissolved in 60 mL of 0.1% (v/v) formic acid prior to LC-MS/MS.
Eluted peptides were analyzed on the LTQ-Orbitrap Elite hybrid mass spectrometer operating in positive ion and data-dependent acquisition mode using the following parameters: electrospray voltage, 1.9 kV; collision-induced dissociation fragmentation with normalized collision energy, 35; and automatic gain control target value, 1E6 for Orbitrap MS and 1E3 for MS/MS scans. Each MS scan (mass-to-charge ratio 400:1,600) was acquired at a resolution of 120,000 full width at half maximum, followed by 20 MS/MS scans triggered for intensities above 500, at a maximum ion injection time of 200 ms for MS and 50 ms for MS/MS scans. For phosphopeptide analysis, a multistage activation was used. Neutral loss masses specified for multistage activation fragmentation were 32.67 and 49 D.
Raw data files of spectra obtained were analyzed in the Proteome Discoverer version 1.4 (Thermo Scientific) platform using the SEQUEST HT search engine (Thermo Scientific). Files were searched against the UniProt Arabidopsis protein database using the following search parameters: Enzyme was specified as trypsin with a maximum of three miscleavages allowed; precursor mass tolerance was 20 parts per million, and fragment mass tolerance was 0.5 Daltons. False discovery rate was 0.05. Carbamidomethylated Cys (57.021 Daltons) was set as fixed modification, whereas phosphorylations at Ser/Thr/Tyr as well as Met oxidation were set as variable modifications.

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Identification of PP2A in peroxisomes and constructing BiFC recombinant genes.
Supplemental Figure S2. Clustering and coexpression of Arabidopsis PP2A subunits C and A with B9u.
Supplemental Figure S3. Identification of PP2A complexes in peroxisomes of onion epidermal cells.
Supplemental Figure S4. Gene expression profiles of Arabidopsis PP2A subunits C, A, and B9u in response to various stimuli.
Supplemental Table S1. Summary of phosphorylated peroxisomal proteins involved in b-oxidation, glyoxylate cycle, and PTS1 import.