Subcellular localization of rice acyl-CoA-binding proteins (ACBPs) indicates that OsACBP6::GFP is targeted to the peroxisomes



  • Acyl-CoA-binding proteins (ACBPs) show conservation at the acyl-CoA-binding (ACB) domain which facilitates binding to acyl-CoA esters. In Arabidopsis thaliana, six ACBPs participate in development and stress responses. Rice (Oryza sativa) also contains six genes encoding ACBPs. We investigated differences in subcellular localization between monocot rice and eudicot A. thaliana ACBPs.
  • The subcellular localization of the six OsACBPs was achieved via transient expression of green fluorescence protein (GFP) fusions in tobacco (Nicotiana tabacum) epidermal cells, and stable transformation of A. thaliana. As plant ACBPs had not been reported in the peroxisomes, OsACBP6::GFP localization was confirmed by transient expression in rice sheath cells. The function of OsACBP6 was investigated by overexpressing 35S::OsACBP6 in the peroxisomal abc transporter1 (pxa1) mutant defective in peroxisomal fatty acid β-oxidation.
  • As predicted, OsACBP1::GFP and OsACBP2::GFP were localized to the cytosol, and OsACBP4::GFP and OsACBP5::GFP to the endoplasmic reticulum (ER). However, OsACBP3::GFP displayed subcellular multi-localization while OsACBP6::GFP was localized to the peroxisomes. 35S::OsACBP6-OE/pxa1 lines showed recovery in indole-3-butyric acid (IBA) peroxisomal β-oxidation, wound-induced VEGETATIVE STORAGE PROTEIN1 (VSP1) expression and jasmonic acid (JA) accumulation.
  • These findings indicate a role for OsACBP6 in peroxisomal β-oxidation, and suggest that rice ACBPs are involved in lipid degradation in addition to lipid biosynthesis.


Eukaryotic lipid exchange involving intermembrane and intracellular lipid movement occurs via both vesicular and nonvesicular mechanisms (reviewed in Li-Beisson et al., 2010; Toulmay & Prinz, 2011). Lipid transport related to the secretory and endocytic system is achieved by budding and fusion of vesicles (van Meer et al., 2008). Vesicular lipid transfer has also been reported in the chloroplasts (Benning et al., 2006) and a hydrophilic protein, designated VESICLE-INDUCING PROTEIN IN PLASTIDS1 (VIPP1), is known to play a role in the formation of the thylakoid membrane (Kroll et al., 2001). In nonvesicular lipid transfer, the participating proteins include ATP-binding cassette (ABC) transporters (Zolman et al., 2001; Benning, 2009), lipid transfer proteins (LTPs) (Kader, 1996) and ACBPs (reviewed in Xiao & Chye, 2011a). Eukaryotic ABC transporters can shuttle cytosolic compounds to the endoplasmic reticulum (ER), mitochondria, peroxisomes, vacuoles and apoplast (Flügge & van Meer, 2006). In Arabidopsis thaliana, a peroxisomal ABC transporter is encoded by COMATOSE (CTS) (also designated PEROXISOMAL ABC TRANSPORTER1 (PXA1) and PEROXISOME DEFICIENT3 (PED3)) (Zolman et al., 2001; Footitt et al., 2002; Hayashi et al., 2002). The ped3 (alternatively cts or pxa1) mutants show resistance to 2,4-dichlorophenoxybutyrate (2,4-DB) and indole-3-butyric acid (IBA), because they are blocked in β-oxidation and cannot produce bioactive auxins from auxin precursors (Zolman et al., 2001; Footitt et al., 2002; Hayashi et al., 2002). ABC transporters can also shuttle lipids between the ER and the chloroplast (Benning, 2008, 2009). The A. thaliana TRIGALACTOSYLDIACYLGLYCEROL1 (TGD1), TGD2, and TGD3 proteins are components of a putative ABC transporter complex (Benning, 2008, 2009) and tgd mutants disrupted in ‘ER to plastid’ lipid transfer are elevated in the 16C : 18C fatty acid ratio in galactolipids (Xu et al., 2003, 2005, 2008; Awai et al., 2006; Lu et al., 2007). TGD4, localized at the chloroplast outer envelope, transfers lipids from the ER to the chloroplast outer envelope (Xu et al., 2008; Wang et al., 2012).

In plants, LTPs (Kader, 1996) have been reported to be secreted into the culture media of barley (Hordeum vulgare) aleurone cells (Mundy & Rogers, 1986), carrot (Daucus carota) embryogenic cells (Sterk et al., 1991) and grapevine (Vitis vinifera) somatic embryo cells (Coutos-Thevenot et al., 1993). It remains to be determined whether LTPs can shuttle lipids between the endomembranes and organelles (reviewed in Li-Beisson et al., 2010). They are probably involved in cutin synthesis, the defense response, symbiosis, and the protection of plants against environmental stress (Kader, 1996). In comparison to LTPs, ACBPs are more likely candidates for intracellular lipid transport because five of six A. thaliana ACBP family members are targeted within the cell (reviewed in Xiao & Chye, 2011a).

Four classes of plant ACBPs have emerged based on their domain architecture (Meng et al., 2011): class I (small ACBPs) consists of peptides of c. 90 amino acid residues; class II (ankyrin-ACBPs) and class IV (kelch-ACBPs) are multi-domain proteins containing C-terminal ankyrin repeats and kelch motifs, respectively, which enable them to interact with partner proteins (Li & Chye, 2004; Li et al., 2008; Gao et al., 2009, 2010; Du & Chye, 2013; Du et al., 2013a); and class III are large proteins with a C-terminal ACB domain. In A. thaliana, the six ACBPs range from 10 to 73.1 kDa (Leung et al., 2004; Xiao & Chye, 2009). Immunoelectron microscopy revealed that AtACBP1 is associated with the ER, the plasma membrane, vesicles and the cell wall (Chye et al., 1999). AtACBP1 and its ortholog, AtACBP2, were localized to the ER and the plasma membrane by confocal laser microscopy (Li & Chye, 2003) and subcellular fractionation (Chye, 1998; Li & Chye, 2003). They have been proposed to be involved in intermembrane lipid trafficking via vesicles and in cuticle and cutin formation (Chye et al., 1999). Both recombinant AtACBP1 (rAtACBP1) and rAtACBP2 bind acyl-CoA esters (Chye, 1998; Chye et al., 2000; Gao et al., 2009, 2010) and phospholipids (Gao et al., 2009, 2010; Chen et al., 2010; Du et al., 2010). AtACBP3 has been detected in the apoplast, the plasma membrane and the periphery of the ER/Golgi complex in transgenic A. thaliana root cells (Leung et al., 2006; Xiao et al., 2010). It can bind acyl-CoA esters, phosphatidylcholine (PC), and phosphatidylethanolamine (PE) (Leung et al., 2006; Xiao et al., 2010). The remaining three AtACBPs (AtACBP4, AtACBP5 and AtACBP6) are cytosolic proteins (Chen et al., 2008; Xiao et al., 2008b) as confirmed by confocal laser-scanning microscopy, immunoelectron microscopy and subcellular fractionation (Chen et al., 2008; Xiao et al., 2008b). His-tagged rAtACBP4 and rAtACBP5 preferentially bind oleoyl-CoA esters and PC, and have been proposed to function in the cytosolic trafficking of lipids (Xiao et al., 2008b). As rAtACBP6 binds acyl-CoA esters as well as saturated (16:0 and 18:0) and unsaturated (18:1 and 18:2) PC (Engeseth et al., 1996; Chen et al., 2008; Xiao et al., 2008a); it is also a candidate for lipid transport in the cytosol.

Given the important roles AtACBPs play in development and stress responses, we have extended our studies to rice (Oryza sativa), a model monocot and significant food crop in Asia. Although six ACBPs also occur in rice, their distribution across classes I to IV and their expression patterns differ from those in A. thaliana (Meng et al., 2011). An alignment of OsACBPs in comparison to AtACBPs (Supporting Information Fig. S1) shows that OsACBP1, OsACBP2 and OsACBP3 in class I display highest conservation to AtACBP6; OsACBP4 resembles class II members AtACBP1 and AtACBP2; OsACBP5 and AtACBP3 belong to class III; while OsACBP6, AtACBP4 and AtACBP5 are class IV kelch-motif-containing proteins. All rOsACBPs bind [14C]linolenoyl-CoA; rOsACBP1 and rOsACBP4 have also been reported to bind [14C]oleoyl-CoA and [14C]linoleoyl-CoA, respectively (Meng et al., 2011). To investigate the subcellular localization of OsACBPs, OsACBPs::GFP were transiently expressed in tobacco (Nicotiana tabacum) and stably expressed in transgenic A. thaliana. The phospholipid-binding properties of OsACBPs and the potential function of OsACBP6 in peroxisomal β-oxidation were also studied.

Materials and Methods

Plant material and growth conditions

Rice (Oryza sativa L. cv Zhonghua 11) was grown as described in Meng et al. (2011). Arabidopsis thaliana (L.) Heynh. wild-type Columbia (ecotype Col-0) and the pxa1 mutant (CS3950) (Zolman et al., 2001) were purchased from the Arabidopsis Biological Resource Center. Tobacco seeds (Nicotiana tabacum var. Xanthi) were obtained from the Institute of Molecular and Cell Biology (Singapore). Arabidopsis and tobacco seeds were surface-sterilized and grown on Murashige and Skoog (MS) medium (Murashige & Skoog, 1962) with 2% (w/v) sucrose and 0.8% (w/v) agar. Seedlings on medium or in soil were subjected to 16 h : 8 h light : dark cycles at 23°C. In the wound treatment (Weber et al., 1997), 4-wk-old A. thaliana rosette leaves (apical halves) were wounded with a pair of forceps and incubated under light, and RNA was sampled at 0, 0.5, 1, 1.5 and 2 h.

IBA (Sigma I5386) treatment was performed according to Zolman et al. (2001). Arabidopsis thaliana seeds were surface-sterilized and grown on plant nutrient medium with 0.5% (w/v) sucrose (PNS) and 0.6% (w/v) agar (Zolman et al., 2001). IBA was added to the growth medium at 3 or 10 μM. Root lengths of 8-d-old seedlings were measured after IBA treatment.

Construction of plasmids

Total RNA extraction and reverse transcription were performed as previously described (Meng et al., 2011). To investigate the subcellular localization of OsACBPs, 35S::OsACBP::GFP fusions were constructed. A polymerase chain reaction (PCR) fragment encoding the full-length open reading frame (ORF) without the stop codon from each of OsACBP1 (LOC_Os08g06550.1; 0.28 kb), OsACBP2 (LOC_Os06g02490.1; 0.28 kb), OsACBP3 (LOC_Os03g37960.1; 0.47 kb), OsACBP4 (LOC_Os04g58550.1; 1.01 kb), OsACBP5 (LOC_Os03g14000.1; 1.71 kb) and OsACBP6 (LOC_Os03g61930.1; 1.96 kb) was generated using reverse transcription–polymerase chain reaction (RT-PCR) primer pairs ML1147/ML1153, ML1151/ML1152, ML1147/ML1148; ML1149/ML1150, ML1145/ML1146, and ML1020/ML1023, respectively (Supporting information Table S1). The PCR products were cloned into vector pGEM-T Easy (Promega, Madison, WI, USA) to produce plasmids pOS577, pOS578, pOS579, pOS580, pOS581 and pOS489, respectively, which were subsequently confirmed by DNA sequence analysis. Each XbaI-BamHI fragment from pOS577, pOS578, pOS579, pOS580, and pOS581, and the XbaI-XbaI fragment from pOS489 were cloned in-frame in vector pBI-eGFP (Shi et al., 2005) to yield plasmids pOS582, pOS583, pOS584, pOS585, pOS586 and pOS493, respectively.

To generate transgenic A. thaliana overexpressing OsACBP6 in the pxa1 mutant background (OsACBP6-OE/pxa1), the OsACBP6 cDNA encoding the full-length ORF was generated by RT-PCR using primers ML1020/ML1019 (Table S1) and cloned into vector pGEM-T Easy (Promega) to generate plasmid pOS492. The 1.96-kb XbaI-SalI OsACBP6 cDNA from pOS492 was cloned into the XbaI-SalI site of the binary vector pSa13 (Xiao et al., 2008a) to produce pOS619 on which OsACBP6 is driven by the cauliflower mosaic virus (CaMV) 35S promoter.

Transient expression assays

For transient expression of 35S::OsACBP::GFP fusions by agroinfiltration of tobacco (Li & Chye, 2004), images of 30 cells with consistent results were achieved for each OsACBP::GFP construct. Transient expression in A. thaliana by Agrobacterium tumefaciens infiltration was performed following Mclntosh et al. (2004). Two-week-old A. thaliana seedlings expressing 35S::OsACBP6::GFP on MS medium supplemented with 2% (w/v) sucrose and 0.8% (w/v) agar were vacuum-infiltrated for 2 min using the Agrobacterium transformant of plasmid pAT485 (Gao et al., 2010). The plates were placed in a tissue culture room under 16 h : 8 h light : dark cycles at 23°C for 3 d. Leaf epidermal cells were examined using confocal laser-scanning microscopy.

Transient expression in rice sheath cells was conducted according to Wang et al. (2013). Plasmids pOS493 and pAT485 (Gao et al., 2010) were introduced into leaf sheath cells of 7-wk-old rice using a Biolistic PDS-1000/He Particle Delivery System (Bio-Rad, Hercules, CA, USA). Bombardment conditions were set as recommended by the manufacturer (Bio-Rad). Leaf sheath cells were incubated at 28°C in darkness for 48 h and observed by confocal laser-scanning microscopy.

Generation of 35S::OsACBP::GFP and OsACBP6-OE/pxa1 transgenic Arabidopsis thaliana

Wild-type A. thaliana was transformed with plasmids pBI-eGFP, pOS582, pOS583, pOS584, pOS585, pOS586, and pOS493 and the pxa1 mutant was transformed with pOS619 by floral dip (Clough & Bent, 1998). T1 seedlings from primary transformants (T0) were selected on MS medium supplemented with kanamycin (50 μg ml−1). The CaMV35S promoter-specific forward primer and gene-specific reverse primers (Table S1) were used to confirm independent transgenic A. thaliana lines for 35S::OsACBP::GFP by PCR. Derivatives of the pxa1 mutant transformed with pOS619 were verified by RT-PCR using an OsACBP6-specific primer pair, ML1113/ML1114 (Table S1).

Confocal laser-scanning microscopy

For confocal imaging, an inverted confocal laser-scanning microscope (LSM 510 (Figs 1,2,4, S3a–d,h–j) and LSM 710 (Fig. S3e–g); Carl Zeiss; and a high-speed Spinning Disc Confocal Microscope (Fig. 3a–d,f,g; Andor; were used. GFP and DsRed fluorescence was observed following Li & Chye (2004). Samples treated with the MitoTracker Red CMXRos (M7512; Invitrogen) and BODIPY TR ceramide (D7540; Invitrogen) were excited at 543 nm and collected using a BP560–615-nm filter and a 493–598-nm filter, respectively.

For brefeldin A (BFA) (B7450; Invitrogen) treatment and staining with MitoTracker Red CMXRos, roots of 7-d-old A. thaliana seedlings expressing 35S::OsACBP6::GFP were incubated with 100 μM BFA for 90 min and stained with MitoTracker Red CMXRos (0.2 μM) for 15 min. Roots of 3-d-old 35S::OsACBP6::GFP A. thaliana seedlings were stained with BODIPY TR ceramide (5 μM) for 30 min at 4°C and then for 30 min at 37°C followed by complete washing procedures as directed in the user's manual.

Subcellular fractionation of plant protein

Subcellular fractionation was conducted according to Smith et al. (1988). The subcellular fractions were used in western blot analysis and cross-reacting bands were detected with anti-GFP antibodies (A6455; Invitrogen) following Chen et al. (2008).

Quantitative real-time polymerase chain reactions (qRT-PCRs)

VEGETATIVE STORAGE PROTEIN1 (VSP1) (At5g24780) gene-specific primers (Theodoulou et al., 2005) and an internal control, A. thaliana ACTIN2 (At3g18780) (Du et al., 2013a) (Table S1), were used in qRT-PCR. The StepOne Plus (Applied Biosystems, Foster City, CA, USA) Real-time PCR system and FastStart Universal SYBR Green Master (Roche Diagnostics, Basel, Switzerland) reagents were utilized in qRT-PCR: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s followed by 60°C for 1 min. The relative gene expression was determined following Meng et al. (2011).

Measurements of jasmonic acid (JA) content

Harvest of plant samples and JA measurements were conducted according to Pan et al. (2010). Rosette leaves from 4-wk-old A. thaliana plants were collected 0 and 1.5 h post-wound treatment (Theodoulou et al., 2005). Leaves were wounded with a pair of forceps. Leaf samples (50 mg) were ground in liquid nitrogen and extracted with 500 μl of extraction solvent (2-propanol/H2O/concentrated HCl (2 : 1 : 0.002, v/v/v)). After shaking at 100 rpm for 30 min at 4°C, 1 ml of dichloromethane was added to each sample, and further subjected to shaking for 30 min at 4°C. The mixture was then centrifuged at 13 000 g for 5 min at 4°C. The solvent (900 μl) was removed from the lower phase to a new tube, and dried using a nitrogen evaporator. The samples were re-dissolved in 0.1 ml of methanol before high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS) analysis. Separation was achieved on a LUNA C18 5 μm 150 mm × 2 mm column (Phenomenex, Torrance, CA, USA) using a gradient of mobile phases water + 0.1% formic acid and methanol + 0.1% formic acid with a flow rate of 0.2 ml min−1. Quantitative analysis of JA was performed using HPLC-MS/MS (AB Sciex QTrap 3200, Framingham, MA, USA) as described in Pan et al. (2010). Six repeats were carried out for each line tested.

Isothermal titration calorimetry (ITC)

Isothermal titration calorimetry experiments were performed using a MicroCal iTC200 system (GE Healthcare, Piscataway, NJ, USA). (His)6-OsACBP6 protein was purified according to Meng et al. (2011). After refolding, the (His)6-OsACBP6 protein was dialyzed in 100 mM sodium phosphate buffer (pH 7.0). Final concentrations of 4 μM (His)6-OsACBP6, 0.5 mM IBA, 0.5 mM (9S, 13S)-12-oxo-phytodienoic acid (OPDA) (Larodan, Malmö, Sweden) and 0.5 mM linolenoyl-CoA (18:3-CoA) (Avanti Polar Lipids, Alabaster, AL, USA) were used in ITC. Solutions for ITC were degassed under vacuum, with stirring, following the procedure of the manufacturer (GE Healthcare). Experiments were performed at 30°C. Twenty injections were carried out in one experiment. (His)6-OsACBP6 was titrated with 1.8 μl of 0.5 mM IBA, OPDA or 18:3-CoA with a duration of 3.6 s, and a 150-s interval between injections. Nonspecific heat effects were estimated and corrected after saturation following the instructions of the manufacturer (GE Healthcare). Raw data were integrated and analyzed using the origin7 software (OriginLab, Northampton, MA, USA) supplied with the instrument.


Predictions on the subcellular localization of OsACBPs

OsACBP1, OsACBP2 and OsACBP3 were predicted to be cytosolic proteins by psort ( (Table S2). Using Predotar v.1.03 (, OsACBP4 and OsACBP5 were predicted to be localized at the ER with scores of 0.6 and 0.97, respectively. The putative ER localization for OsACBP5 was also supported by psort (score 0.82). However, psort suggested that OsACBP4 is an extracellular protein. By contrast, TargetP1.1 ( indicated that OsACBP3, OsACBP4 and OsACBP5 are associated with the secretory pathway (Table S2). The prediction (ipsort Prediction; of a putative chloroplast transit peptide for OsACBP6 was not upheld by Predotar v.1.03 and TargetP1.1 (Table S2). Tusnády & Simon (2001) ( predicted transmembrane helices in OsACBP4 (amino acids 11–31), OsACBP5 (amino acids 7–31) and OsACBP6 (amino acids 15–32 and 462–479), while ipsort suggested the presence of putative N-terminal signal peptides in OsACBP4 (amino acids 1–30) and OsACBP5 (amino acids 1–30). Using the ‘Target Signal Predictor’ analysis (for Peroxisomal Targeting Signal (PTS) 1 and PTS2 binding sites, available at, a putative PTS2 (e-value of 0.14) was identified in OsACBP6.

Transient expression of OsACBP::GFP fusion proteins in tobacco leaf epidermal cells

To experimentally verify these predictions, each OsACBP was fused in-frame to GFP and the resultant chimeric protein was transiently expressed from the CaMV35S promoter in 6-wk-old tobacco leaf epidermal cells followed by examination by confocal laser-scanning microscopy. A similar strategy had been successfully used to subcellularly localize the AtACBPs (Li & Chye, 2004; Li et al., 2008; Gao et al., 2009, 2010). Vector pBI-eGFP showed a typical nuclear and cytoplasmic localization for GFP (Fig. 1a), while OsACBP1::GFP (Fig. 1b), OsACBP2::GFP (Fig. 1c) and OsACBP3::GFP (Fig. 1d) showed expression patterns resembling that of pBI-eGFP, indicating that they are also targeted to the cytosol. Their apparent nuclear localization, similar to that of the control, had probably arisen from passive diffusion through nuclear pore complexes, similar to diffusion of the 38-kDa AtACBP6::GFP (Chen et al., 2008), their molecular masses being 38.1 kDa for OsACBP1::GFP, 38.2 kDa for OsACBP2::GFP, and 45.6 kDa for OsACBP3::GFP.

Figure 1.

Transient expression of OsACBP::GFPs in tobacco leaf epidermal cells by confocal laser-scanning microscopy. (a) 35S::GFP from pBI-eGFP vector control. (b) OsACBP1::GFP. (c) OsACBP2::GFP. (d) OsACBP3::GFP. (e) OsACBP4::GFP. (f) OsACBP5::GFP. (g) OsACBP6::GFP. (h) Merged image of (g) with chloroplast autofluorescence. Bars, 20 μm. (a–d) Arrowheads indicate nuclei; (e, f) red arrows, the ER surrounding the nuclei; (g, h) yellow arrows, punctate structures; white arrows, chloroplasts. At least 30 cells were photographed in transient expression for each OsACBP::GFP fusion and representative photographs are shown. ACBP, acyl-CoA-binding protein.

GFP fluorescent signals of OsACBP4::GFP (Fig. 1e) and OsACBP5::GFP (Fig. 1f) were localized at the ER surrounding the nuclei in leaf epidermal cells. Randomly distributed punctate signals were observed for OsACBP6::GFP (Fig. 1g). However, they did not overlap with the red chloroplast autofluorescence, excluding expression at the chloroplasts (Fig. 1h). Given such differences in prediction and experimental localization of OsACBP6, further investigations were carried out.

Localization of OsACBP1::GFP and OsACBP2::GFP in the cytosol of transgenic Arabidopsis thaliana

To further confirm the subcellular localization of the six OsACBPs, stably transformed wild-type A. thaliana (Col-0) lines overexpressing each 35S::OsACBP::GFP were generated and verified by PCR using a 35S promoter-specific forward primer and gene-specific reverse primers (Fig. S2). Confocal laser-scanning microscopy indicated that the GFP fluorescent signals from cotyledon cells and plasmolyzed root cells of 35S::OsACBP1::GFP (Fig. 2a–d) and 35S::OsACBP2::GFP (Fig. 2e–h) transformed A. thaliana were localized in the cytosol and shared a common expression pattern with the GFP-vector control (Fig. 2i–l). To confirm the microscopy results, subcellular fractionation was conducted. Western blot analysis using anti-GFP antibodies on subcellular fractions from 35S::OsACBP1::GFP (Fig. 2m) and 35S::OsACBP2::GFP (Fig. 2n) showed cross-reacting bands of 38.1 and 38.2 kDa, respectively, in both total protein and the cytosolic fraction. These results resembled those for the GFP-vector transformed control (Fig. 2o).

Figure 2.

Localization of OsACBP1::GFP, OsACBP2::GFP and GFP in transgenic Arabidopsis thaliana. Cotyledon cells and plasmolyzed primary root cells of 7-d-old plants were observed. (a–d) OsACBP1::GFP. (e–h) OsACBP2::GFP. (i–l) 35S::GFP. Confocal images in cotyledon cells (a, e, i) and root cells (c, g, k) and their corresponding transmitted light images are shown on the right of the confocal images. Bars, 10 μm. (i, k) Arrows indicate nuclei. (m–o) Western blot analysis using anti-GFP antibodies on subcellular fractions of whole plant protein from transgenic Arabidopsis thaliana 35S::OsACBP1::GFP line 1 (m), 35S::OsACBP2::GFP line 1 (n) and 35S::GFP line 1 (o). Lane 1, total whole plant protein; lane 2, crude nuclei; lane 3, large particle fraction (including mitochondria, chloroplasts, lysosomes and peroxisomes); lane 4, microsomal fraction; lane 5, cytosol. Arrows indicate the 38.1-kDa OsACBP1::GFP band (m), the 38.2-kDa OsACBP2::GFP band (n) and the 28-kDa GFP band (o). Bottom panel in each of (m–o) shows an identically loaded gel stained with Coomassie Blue. ACBP, acyl-CoA-binding protein.

OsACBP3::GFP displayed subcellular multi-localization

OsACBP3 phylogenetically belongs to class I, as do OsACBP1 and OsACBP2; OsACBP1 and OsACBP2 share 80% amino acid identity, while OsACBP1 shares 91% amino acid identity with OsACBP3 (Meng et al., 2011). Furthermore, OsACBP3 has an extra C-terminal extension, comprising 64 amino acids, in comparison to OsACBP1 (Meng et al., 2011). This extension contains a putative acetyl-lysine deacetylase domain (predicted by the National Center for Biotechnology Information BLAST (

OsACBP3::GFP resembled the GFP vector control in its cytosolic localization (Fig. 3a–d), but additional fluorescent signals were detected at the intracellular irregular membranous structures (Fig. 3b) and as randomly distributed punctates (Fig. 3d). In western blot analysis using anti-GFP antibodies on the subcellular fractions of 35S::OsACBP3::GFP line 1, a 45.6-kDa OsACBP3::GFP cross-reacting band was weakly detected in crude nuclei (lane 1) and the cytosolic fraction (lane 4), but strong signals were observed in the fraction of large particles (including mitochondria, chloroplasts, lysosomes and peroxisomes) (lane 2), the microsomal fraction (lane 3), and total protein (lane 5) (Fig. 3e). These results from confocal laser-scanning microscopy and western blot analysis suggest that OsACBP3-GFP occurs in multiple subcellular compartments.

Figure 3.

Localization of OsACBP3::GFP, OsACBP4::GFP and OsACBP5::GFP in transgenic Arabidopsis thaliana. Confocal images of root tip and cotyledon cells of 7-d-old transgenic A. thaliana are shown. (a, c) 35S::GFP; (b, d) OsACBP3::GFP. Bars, 10 μm. (b) White arrows indicate irregular membranous structures. (d) Red arrows indicate punctate structures. (e) Western blot analysis using anti-GFP antibodies on subcellular fractions of whole plant protein from transgenic A. thaliana 35S::OsACBP3::GFP line 1. Lane 1, crude nuclei; lane 2, large particle fraction (including mitochondria, chloroplasts, lysosomes and peroxisomes); lane 3, microsomal fraction; lane 4, cytosol; lane 5, total whole plant protein. The arrow indicates the 45.6-kDa OsACBP3::GFP band. Bottom panel, identically loaded gel stained with Coomassie Blue. (f) OsACBP4::GFP expression in the cotyledon cells of 7-d-old transgenic A. thaliana. (g) OsACBP5::GFP expression in the cotyledon cells of 7-d-old transgenic A. thaliana. Bars, 5 μm. ACBP, acyl-CoA-binding protein.

Localization of OsACBP4::GFP and OsACBP5::GFP in the ER

Observations of cortical mesh network patterns for OsACBP4::GFP and OsACBP5::GFP in the cotyledons of 7-d-old 35S::OsACBP4::GFP (Fig. 3f) and 35S::OsACBP5::GFP (Fig. 3g) transgenic A. thaliana seedlings confirmed their ER localization as predicted in silico. In 7-d-old seedlings, OsACBP4::GFP was associated with both tubular and cisternal ER (Fig. 3f), while OsACBP5::GFP was confined to the tubular ER (Fig. 3g).

Localization of OsACBP6::GFP in the peroxisomes

Inconsistency between the results from transient expression of OsACBP6::GFP (Fig. 1g) and computer prediction prompted us to examine the class IV ACBPs from other plant species using psort (Table 1). Class IV orthologs from chlorophytes (Chlamydomonas reinhardtii and Ostreococcus lucimarinus) were predicted to be located in the peroxisomes, while bryophyte (Physcomitrella patens), gymnosperm (Picea sitchensis) and monocot (Oryza sativa, Zea mays and Sorghum bicolor) orthologs were presumed to be targeted to the chloroplast stroma. Orthologs from a lycophyte (Selaginella moellendorffii) and most eudicots, except Vitis vinifera and Ricinus communis, were presumed to be cytosolic proteins. It appeared that class IV ACBPs probably split during the separation of eudicots and monocots. While eudicot class IV A. thaliana ACBPs were destined for the cytosol, monocot ACBPs seemed to be targeted to the chloroplast, similar to mosses and gymnosperms (Table 1).

Table 1. Prediction of subcellular localization of plant class IV acyl-CoA-binding proteins (ACBPs)
 CytoplasmMitochondrial matrix spaceMitochondrial inner membraneChloroplast thylakoid membraneChloroplast stromaChloroplast thylakoid spaceMicrobody (peroxisome)Nucleus
  1. The score represents certainty in the psort Prediction, with the highest score in bold text.

  2. Chlorophytes (C. reinhardtii, Chlamydomonas reinhardtii; O. lucimarinus, Ostreococcus lucimarinus).

  3. Bryophyte (P. patens, Physcomitrella patens).

  4. Lycophyte (S. moellendorffii, Selaginella moellendorffii).

  5. Gymnosperm (P. sitchensis, Picea sitchensis).

  6. Eudicots (A. thaliana, Arabidopsis thaliana; C. papaya, Carica papaya; C. sativus, Cucumis sativus; G. max, Glycine max; M. truncatula, Medicago truncatula; P. trichocarpa, Populus trichocarpa; R. communis, Ricinus communis; V. vinifera, Vitis vinifera).

  7. Monocots (S. bicolor, Sorghum bicolor; O. sativa, Oryza sativa; Z. mays, Zea mays).

  8. Protein sequences were retrieved from;;

  9. a

    When Predalgo was used to predict the ACBPs from the two green algae, Chlamydomonas reinhardtii (CR02G00310) and Ostreococcus lucimarinus (OL07G00010), the results showed that these were assigned to the ‘Other’ category and they were not targeted to the mitochondrion, chloroplast, or secretory pathway.

C. reinhardtii (CR02G00310)a0.4500.100 0.100   0.470  
O. lucimarinus (OL07G00010)a 0.100     0.608  
P. patens (PP00048G00560)   0.200 0.520  0.2480.300
P. patens (PP00068G00210) 0.100 0.631 0.908 0.631  
S. moellendorffii (EFJ14275) 0.450 0.100 0.100  0.168 
P. sitchensis (ACN39866) 0.531 0.621 0.895 0.578  
A. thaliana (AT3G05420) 0.650 0.100 0.100    
A.thaliana (AT5G27630) 0.450 0.360 0.280  0.362 
C. papaya (CP00048G00760) 0.650 0.100 0.100    
C. sativus (Cucsa.091180) 0.450 0.446 0.2800.200   
G. max (GM19G42270) 0.450 0.360 0.280  0.333 
G. max (GM20G37940) 0.450 0.360 0.280  0.342 
G. max (GM03G39640) 0.450 0.447 0.2000.200   
M. truncatula (ABN08040) 0.450 0.4470.136   0.357 
P. trichocarpa (PT13G01720) 0.450 0.360 0.2800.389   
P. trichocarpa (PT00G21960) 0.447 0.329 0.597  0.405 
R. communis (XP_002512412)0.450 0.497 0.2140.280    
V. vinifera (VV14G08600)0.450 0.495 0.211 0.470   
S. bicolor (SB01G002200)0.450  0.393 0.636 0.393  
O. sativa (OS03G61930) 0.100 0.446 0.861 0.446  
Z. mays (GRMZM2G053803)0.450  0.393 0.636 0.393  

On plasmolysis of transgenic A. thaliana primary roots cells transformed with 35S::OsACBP6::GFP, GFP signals were observed subcellularly (Fig. S3a,b), but they did not overlap with the red chloroplast-associated autofluorescence (Fig. 4a–d). To determine the fluorescing organelle, the chemical BFA (Ritzenthaler et al., 2002), the Golgi apparatus-specific dye BODIPY TR ceramide (Ruzin, 1999; Vadaie et al., 2002) and the mitochondrion-specific dye MitoTracker Red (Poot et al., 1996) were applied. The lack of morphological changes after BFA treatment (Fig. S3c,d) eliminated the punctates as the Golgi apparatus. Upon staining of roots with the Golgi apparatus-specific dye (Fig. S3e–g) and the mitochondrion-specific dye (Fig. S3h–j), the OsACBP6::GFP signals were distinct from the Golgi apparatus and mitochondria, respectively (Fig. S3e–j).

Figure 4.

OsACBP6::GFP is localized in the peroxisomes by confocal laser-scanning microscopy. (a–d) OsACBP6::GFP expression in primary root cells (a) and chloroplast autofluorescence (b). (c) Merged images of (a) and (b). (d) Transmitted light image. Bars, 10 μm. (e–h) Transient expression of DsRed::SKL in 35S::OsACBP6::GFP transgenic A. thaliana. (e) OsACBP6::GFP; (f) DsRed::SKL; (g) merged images of (e) and (f). (h) Transmitted light image. Bars, 10 μm. (i–l) Transient expression of OsACBP6::GFP and DsRed::SKL in the rice sheath cells. (i) OsACBP6::GFP; (j) DsRed::SKL; (k) merged images of (i) and (j). (l) Transmitted light image. Bars, 10 μm. (a, e, i) White arrows indicate punctate structures; (b) blue arrows, chloroplasts; (f, j) yellow arrows, peroxisomes; (f) arrowheads, guard cells. ACBP, acyl-CoA-binding protein.

When the peroxisomal matrix marker DsRed::SKL (Gao et al., 2010) was transiently expressed in 35S::OsACBP6::GFP transgenic A. thaliana leaves, the merged images of OsACBP6::GFP with DsRed::SKL signals confirmed their co-localization at the peroxisome matrix (Fig. 4e–h). As the subcellular localization of OsACBP6::GFP was studied using heterologous expression in tobacco and A. thaliana, its localization in rice was further pursued. When OsACBP6::GFP and DsRed::SKL were transiently expressed in rice sheath cells by particle bombardment, they co-localized at the peroxisomes (Fig. 4i–l).

Recovery of IBA metabolism in the pxa1 mutant by the overexpression of peroxisomal OsACBP6

A comparison between the subcellular localizations and acyl-CoA ester binding abilities of AtACBPs and OsACBPs revealed several differences within classes (Table 2). Their interaction with phospholipids also varied (Supporting Information Methods S1, Fig. S4, Table 2). It has been observed that rOsACBP6 binds 18:2-CoA ester, 18:3-CoA ester, PA (18:0, 18:1) and PC (18:0, 18:1 and 18:2) (Meng et al., 2011; Fig. S4). Furthermore, the potential substrates for CTS are eicosenoic acid (20:1), linolenic acid (18:3), IBA, 2,4-DB, and OPDA as well as their CoA esters (Theodoulou et al., 2005). Acyl-CoA esters (C14-CoA, C18-CoA and C24-CoA), rather than the products of free fatty acid hydrolysis, have been demonstrated to stimulate ATPase activity in CTS (Nyathi et al., 2010). As CTS and rOsACBP6 both bind to C18-CoA esters, we were interested to test whether these peroxisomal proteins shared similar functions. To this end, a 35S::OsACBP6 construct (plasmid pOS619) was generated for transformation of the A. thaliana pxa1 mutant (Zolman et al., 2001) and resultant independent lines were confirmed by RT-PCR (Fig. S5).

Table 2. Comparison of Arabidopsis thaliana and rice acyl-CoA-binding proteins (ACBPs)
  Acyl-CoA ester bindingSubcellular localization :(experimentally verified)Phospholipid bindingReference
  1. Class I, small ACBPs; class II, ankyrin-ACBPs; class III, large ACBPs; class IV, kelch-ACBPs.

  2. The results from this study are in bold text.

Class IAtACBP618:2 > 20:4 > 18:1 ≈ 16:0 > 18:3CytosolPC (16:0 18:0 18:1 18:2)

Chen et al. (2008);

Xiao et al. (2009)

OsACBP118:1 > 18:2 > 18:3 > 16:0 Cytosol

PA (18:0 18:1)

PC (18:0 18:1 18:2)

Meng et al. (2011); This study
OsACBP218:3 Cytosol Meng et al. (2011); This study


Irregular membranousstructures


Meng et al. (2011); This study
Class IIAtACBP120:4 > 18:2 > 18:3 > 18:1 > 16:0

Plasma membrane


PA (16:0 18:0 18:1)

PC (18:1 18:2)

Chye (1998); Gao et al. (2009); :Du et al. (2010); :Chen et al. (2010)
AtACBP218:2 > 18:3 > 20:4 ≈ 16:0 > 18:1

Plasma membrane


Lyso PC, PC (18:1 18:2)

Chye et al. (2000);

Li & Chye (2003);

Gao et al. (2009, 2010);

Chen et al. (2010)

OsACBP418:2 > 16:0 > 18:3 ER

PA (16:0 18:0 18:1)

PC (18:0 18:1 18:2)

Meng et al. (2011); This study
Class IIIAtACBP320:4 > 18:3 > 18:2 > 18:1 ≈ 16:0


Endogenous membranes

PE (16:0 18:0 18:1 18:2)

PC (18:0 18:1 18:2)

Xiao et al. (2010)
OsACBP518:3 > 16:0 ER

PA (18:0 18:1)

PC (18:0 18:1 18:2)

Meng et al. (2011); This study
Class IVAtACBP418:1 > 16:0 > 18:2 > 18:3CytosolPC (18:1 18:2)

Leung et al. (2004);

Xiao et al. (2008b, 2009)

AtACBP518:1 > 16:0 > 18:2 > 18:3CytosolPC (18:1 18:2)

Leung et al. (2004);

Xiao et al. (2008b, 2009)

OsACBP618:3 > 18:2 Peroxisomes

PA (18:0 18:1)

PC (18:0 18:1 18:2)

Meng et al. (2011); This study

Zolman et al. (2001) had reported that root elongation in the pxa1 mutant was not inhibited by high concentrations (10 and 30 μM) of IBA and its developmental defects include sucrose dependence and slower growth than the wild type. When OsACBP6 was ectopically expressed in pxa1, its sensitivity to 10 μM IBA was restored in all three OsACBP6-OE/pxa1 lines and they resembled the wild-type 8-d-old seedling in primary root length (Fig. 5). Also, hypocotyl elongation in sucrose-free medium was restored in the OsACBP6-OE/pxa1 lines although their hypocotyls were still shorter than those of the wild type (Fig. S6a). Furthermore, the number of rosette leaves and the diameter of the rosettes of OsACBP6-OE/pxa1 resembled those of the wild type (Fig. S6b,c).

Figure 5.

Transgenic Arabidopsis thaliana OsACBP6-OE/peroxisomal abc transporter1 (pxa1) lines are sensitive to indole-3-butyric acid (IBA). Primary root length measurements of A. thaliana 8-d-old seedlings from the wild type (WT), the pxa1 mutant and OsACBP6-OE/pxa1 lines 21, 26 and 44 grown on plant nutrient medium with 0.5% (w/v) sucrose (PNS) containing various concentrations of IBA are shown. The error bar indicates the SE value (n > 20). * indicates that root length inhibition is more severe than in the pxa1 mutant (< 0.01). ACBP, acyl-CoA-binding protein.

Expression of OsACBP6 in the pxa1 mutant rescued wound-induced VSP1 expression and JA accumulation

The jasmonate precursors, OPDA and dinor-OPDA (dnOPDA), synthesized in the chloroplasts (reviewed in Blée, 1998) are transported to the peroxisomes, and JA is generated after several rounds of peroxisomal β-oxidation (Vick & Zimmerman, 1983). It has been reported that CTS is responsible for transfer of the jasmonate precursor to the peroxisomes (Theodoulou et al., 2005). Also, the basal and wound-induced concentrations of JA and the relative expression of JA-responsive VSP1 in the cts mutant are lower than in the wild type (Theodoulou et al., 2005). To investigate whether OsACBP6 plays a role in the import of JA precursors and VSP1 expression, the pxa1 mutant and its 35S::OsACBP6 transformants were examined before and after wound treatment. Although the relative expression of VSP1 in the pxa1 mutant was not significantly affected 2 h after wounding, its expression was rapidly induced 0.5 h post-wounding in three independent OsACBP6-OE/pxa1 lines (Fig. 6). Its expression in these three lines did not show similar fold increases, but decreases were observed in all these lines 2 h post-wound treatment (Fig. 6).

Figure 6.

Relative expression of VEGETATIVE STORAGE PROTEIN1 (VSP1) following wound treatment. The expression of VSP1 in Arabidopsis thaliana wild type (WT), the peroxisomal abc transporter1 (pxa1) mutant and OsACBP6-OE/pxa1 lines 21, 26 and 44 after wound treatment (0, 0.5, 1, 1.5 and 2 h) as examined by qRT-PCR is shown. The expression levels were normalized to ACTIN. H, value > at 0 h (< 0.05); L, value < at 0 h (< 0.05). Bars represent SE (= 3). ACBP, acyl-CoA-binding protein.

Wound-induced JA production was also examined 1.5 h after wounding. Before treatment (0 h), the basal JA concentration in the pxa1 mutant was higher than in the wild type, while OsACBP6-OE/pxa1 lines displayed concentrations lower than in the wild type (Fig. 7a). However, significant increases in JA content following wounding were observed in OsACBP6-OE/pxa1 line 26 (73.8 ng g−1 FW) and line 44 (73.2 ng g−1 FW), resembling the wild type (72.9 ng g−1 FW) (Fig. 7b). Although OsACBP6-OE/pxa1 line 21 displayed a JA increase (53.7 ng g−1 FW) lower than that in the wild type (Fig. 7b), this was nonetheless higher than in the pxa1 mutant (41.7 ng g−1 FW). This observation is consistent with changes in VSP1 expression in these genotypes following wound treatment (Fig. 6). Taken together, the changes in both VSP1 expression and wound-induced JA accumulation in the OsACBP6-OE/pxa1 lines suggest that JA production was restored upon the overexpression of OsACBP6 in pxa1.

Figure 7.

Jasmonic acid (JA) accumulation following wound treatment. (a) Measurements of JA at 0 h (black bars) and 1.5 h (gray bars) post-wounding in Arabidopsis thaliana wild type (WT), the peroxisomal abc transporter1 (pxa1) mutant and OsACBP6-OE/pxa1 lines 21 (OE-21), 26 (OE-26) and 44 (OE-44). H, JA concentration higher than in the WT; L, JA concentration lower than in the WT; a, JA concentration higher than in the pxa1 mutant (*, < 0.05; **, < 0.01). (b) Increases in JA 1.5 h after wound treatment in the WT (black bar), the pxa1 mutant (open bar) and OE-21, OE-26 and OE-44 (gray bars). Values were obtained after deduction of JA content at time 0. Lowercase letters: a, JA increase higher than in the pxa1 mutant (*, < 0.05; **, < 0.01); b, JA increase lower than in the WT (P < 0.01). Bars represent SE (= 6). ACBP, acyl-CoA-binding protein.

It is not known whether IBA and the JA precursor (OPDA) are transported into the peroxisomes as unmodified forms or as CoA derivatives (reviewed in Hu et al., 2012). Given the recovery of IBA metabolism and JA production in the pxa1 mutant upon transformation with OsACBP6, the in vitro interactions of rOsACBP6 with IBA and OPDA were examined by ITC. Results from ITC did not show binding between rOsACBP6 and IBA (0.5 mM) or OPDA (0.5 mM) (Fig. 8a,b), indicating that the unmodified forms of IBA and OPDA could be discounted as substrates. It has been reported using Lipidex assays that rOsACBP6 binds linolenoyl-CoA (18:3-CoA) (Meng et al., 2011) and this was confirmed in the present study by ITC (KD of 32.25 μM in Fig. 8c).

Figure 8.

Isothermal titration calorimetry profiles of recombinant rice OsACBP6 (rOsACBP6) with indole-3-butyric acid (IBA), (9S, 13S)-12-oxo-phytodienoic acid (OPDA) and 18:3-CoA. Thermograms are shown for the interaction of IBA with rOsACBP6 (a), OPDA with rOsACBP6 (b), and 18:3-CoA with rOsACBP6 (c, upper panel). The lower panel in (c) shows the corresponding binding isotherm with theoretical fits. Upper panel, raw heat signal from 20 injections of 1.8-μl aliquots of a 0.5 mM solution of 18:3-CoA into a cell containing 4 μM rOsACBP6 at 30°C; lower panel, the integrated area (heat) of each injection after background correction. ACBP, acyl-CoA-binding protein.


Possible roles for ACBPs in lipid trafficking have been proposed based on their subcellular localization in A. thaliana (Chen et al., 2008; Xiao et al., 2008a,b, 2010; Gao et al., 2010). In this study, the subcellular localization of rice ACBPs was mapped to various subcellular compartments, similar to A. thaliana ACBPs, including the cytosol (OsACBP1, OsACBP2 and OsACBP3) and the ER (OsACBP4 and OsACBP5). However, unlike A. thaliana ACBPs, a rice ACBP (OsACBP6) was localized to the peroxisomes using a OsACBP6::GFP construct in transgenic A. thaliana and transient expressing rice sheath cells. The localization of OsACBP6 to the peroxisomes demonstrates that the peroxisome is a new organelle with which plant ACBPs can interact and suggests a role for this ACBP in peroxisomal lipid metabolism.

Localization of ACBPs in the cytosol and endomembrane is conserved between Arabidopsis thaliana and rice

The basic functions of ACBPs have been inferred predominantly from findings on class I ACBPs which represent the best-studied group (reviewed in Færgeman et al., 2007). Previous phylogenetic analysis showed that, in plants, class I forms a well-supported clade (Meng et al., 2011); members of class I from Animalia as well as Plantae (eudicots: A. thaliana and Brassica napus) are all cytosolic proteins (Mikkelsen & Knudsen, 1987; Hills et al., 1994; Fyrst et al., 1995; Chen et al., 2008; Yurchenko et al., 2009). The cytosolic localization of rice class I members consisting of 91 amino acids (with the exception of OsACBP3 which contains an additional 64 amino acids) was confirmed in this study to resemble that of its A. thaliana ortholog (AtACBP6; Chen et al., 2008). Our results in rice (a monocot) support an apparent conservation in subcellular localization and suggest that class I ACBPs appear to primarily function in the cytosol.

Interestingly, OsACBP3::GFP was localized to multiple sites subcellularly, making it unique in class I. OsACBP3 also shares high amino acid identity with cytosolic OsACBP1 in its first 91 amino acids inclusive of the ACB domain except for eight mismatches (as underlined in Fig. S1). As the psort prediction for the peptide comprising the first 91 amino acids of OsACBP3 is cytosolic, with a certainty of 0.65, resembling OsACBP1, its noncytosolic localization is probably attributable to its C-terminal extension of 64 amino acids which contains a putative acetyl-lysine deacetylase domain. Histone deacetylase is involved in the removal of acetyl groups from histones (Kuo & Allis, 1998). Lysine acetylation and deacetylation of core histones are important posttranslational events catalyzed by acetyltransferases and deacetylases, respectively (Turner, 1991; Kuo & Allis, 1998; Finkemeier et al., 2011). In A. thaliana, lysine acetylation and deacetylation are also known to occur on cytosolic and organellar proteins of diverse functions, such as cytosolic aldolase and NAD+-dependent malate dehydrogenase, cytochrome c from the mitochondrial respiratory chain, and an integral membrane protein plastid ATP/ADP translocator (Finkemeier et al., 2011). Such multiple sites for deacetylase action may argue for a need for OsACBP3 to be targeted from the cytosol to multiple sites, as detected in the present study.

Although ER-associated OsACBPs (OsACBP4 and OsACBP5) were identified as in A. thaliana, some differences were noted. Unlike AtACBP1 and AtACBP2, ankyrin-containing OsACBP4 was not targeted to the plasma membrane and, unlike AtACBP3, OsACBP5 was not targeted to the apoplast. It remains to be determined whether the functions of OsACBP4 and OsACBP5 overlap with those of their A. thaliana counterparts. AtACBP1 and AtACBP2 are related to heavy metal stress (Xiao et al., 2008a; Gao et al., 2009, 2010), freezing tolerance (Du et al., 2010) and drought tolerance (Du et al., 2013b) in addition to their ability to interact with transcription factors such as the ethylene-responsive element binding protein (AtEBP) as well as enzymes/proteins related to stress (Li & Chye, 2004; Gao et al., 2009, 2010; Du & Chye, 2013; Du et al., 2013a,b). In transgenic A. thaliana, the overexpression of AtACBP3 culminated in an enhanced resistance to Pseudomonas syringae in comparison to the wild type (Xiao & Chye, 2011b).

The subcellular localization of OsACBP4::GFP and OsACBP5::GFP distinguished OsACBP4::GFP localization to the ER cisternae and tubules while OsACBP5::GFP was confined to the tubules. ER shape is known to be associated with specific cellular functions and the cisternae represent the sites for protein translocation while vesicles bud from the tubules (Friedman & Voeltz, 2011). Taken together, the results suggest that OsACBP4 and OsACBP5 may have overlapping functions in the tubules but not the cisternae, and are not redundant in the ER. The expression of OsACBP5, like that of AtACBP3, was induced by salicylic acid and pathogen infection (Meng et al., 2011; Xiao & Chye, 2011b; Meng, 2012) and, as pathogenesis-related (PR) proteins can be secreted from the ER (Jelitto-Van Dooren et al., 1999), OsACBP5 localization at the tubules may be related to defense. Given that the ER is responsible for the synthesis/export of phospholipids (Sparkes et al., 2011) and lipid droplets (Farese & Walther, 2009), OsACBP4 and OsACBP5 may facilitate these processes. The synthesis and mobilization of lipid droplets require rapid lipid exchange (Farese & Walther, 2009), and the high expression of OsACBP5 throughout reproduction (Meng et al., 2011) would be consistent with this proposed role.

Peroxisomal OsACBP6 is unique

While the ER and the plastids have been identified as the sites for plant lipid biosynthesis (Ohlrogge & Browse, 1995), the peroxisomes represent the site for fatty acid β-oxidation (Cooper & Beevers, 1969) and they interact with other organelles to carry out various metabolic processes including substrate exchange (Linka & Esser, 2012). To date, two transporter families (the CTS and the mitochondrial carrier (MC) family) have been localized at the peroxisomes in A. thaliana (Linka & Esser, 2012). CTS imports substrates for β-oxidation, while MC transports cofactors ATP and NAD (Linka & Esser, 2012). Although rice orthologs (Os01g73530 and Os05g01700) of the A. thaliana CTS have been identified (Kaur & Hu, 2011), their substrate preferences and functions remain unknown. As a transporter, CTS comprises two transmembrane domains, for substrate recognition and translocation, as well as two nucleotide-binding domains for the provision of energy (Theodoulou et al., 2006; Hu et al., 2012). OsACBP6 would fit well as a potential peroxisomal lipid transporter because it comprises two transmembrane helices, an ACB domain for substrate recognition and two kelch motifs for potential protein–protein interactions. Other ACBPs in both A. thaliana and rice contain only one helix (classes II and III) or none. Hence, protein architecture can be a criterion that sets OsACBP6 (class IV) apart from the other ACBP family members.

CTS is reported to possess acyl-CoA thioesterase activity to cleave acyl-CoAs during transport into the peroxisomal lumen (De Marcos Lousa et al., 2013). It remains to be determined where the cleavage of acyl-CoAs occurs subcellularly and how acyl-CoAs are transported to the peroxisomes (De Marcos Lousa et al., 2013; Hunt et al., 2014). Furthermore, the protein(s) responsible for the maintenance and regulation of a peroxisomal-CoA pool to support β-oxidation has yet to be identified (Hunt et al., 2014). As ACBPs have been assigned functions as lipid transporters and in the maintenance of intracellular acyl-CoA pools in yeast, animals, Brassica napus and A. thaliana (Mandrup et al., 1993; Rasmussen et al., 1993; Knudsen et al., 1994; Færgeman & Knudsen, 1997; Huang et al., 2005; Yurchenko et al., 2009), OsACBP6 can be considered a potential candidate for involvement in acyl-CoA transport and regulation in the peroxisomes.

The recovery of IBA sensitivity, subsequent to the ectopic expression of OsACBP6 in the pxa1 mutant, supports this hypothesis. Also, the up-regulation of post-wounding JA-responsive VSP1 expression and the significant increase in wound-induced JA in the pxa1 mutant upon overexpression of OsACBP6 support a correlation between OsACBP6 and JA biosynthesis. Park et al. (2013) showed that the amount of post-wounding JA was significantly lower in 8-d-old pxa1 seedlings than in the wild type, consistent with our findings on 4-wk-old pxa1 rosettes (Fig. 7b). The amount of endogenous JA in 4-wk-old pxa1 rosette leaves was greater than in the wild type (Fig. 7a), consistent with previous observations of higher concentrations of JA in dry and developing seeds of the pxa1 and cts mutants in comparison to the wild type (Dave et al., 2011). It has been proposed that JA accumulation in the pxa1 and cts mutants indicates the existence of a CTS-independent route for import of JA precursors (Theodoulou et al., 2005; Dave et al., 2011). It remains to be determined if OsACBP6 is associated with this alternate pathway.

Among the four ACBP classes, members of class IV of A. thaliana and rice show the lowest similarities. Comparison of the predicted subcellular localizations of other plant class IV members clearly indicates a split between the eudicots and monocots. The angiosperms have evolved under a particular set of environmental conditions and have co-evolved with faunal groups (Chaw et al., 2004). Calibration using two fossil nodes following the Li–Tanimura method indicates that monocots branched off from ‘traditional dicots’ (Eudicots, Amborellaceae, Nymphaeaceae, Austrobaileyales, Ceratophyllaceae, Chloranthaceae and Magnoliids (Soltis et al., 2005)) c. 140–150 Myr ago (Chaw et al., 2004) to form a well-supported lineage (Soltis et al., 2005) while core eudicots split 100–115 Myr ago (Chaw et al., 2004). Although there is only one evolutionary origin of extant angiosperms, the relationship among them is not well-understood (Soltis et al., 2005). During the evolution of land plants, the peroxisomal proteome has apparently adapted to meet new metabolic needs, culminating in differences now evident between A. thaliana and rice (Kaur & Hu, 2011; Linka & Esser, 2012). For example, to support seedling establishment, storage reserve utilization is dependent upon specific peroxisomal functions which differ between eudicots and monocots (Linka & Esser, 2012). Hence, given the length of divergence time, monocot class IV ACBPs appear to have specialized. The peroxisomal localization of an ACBP now raises the hypothesis that monocot ACBPs may have functions that are not confined to lipid biosynthesis, but could extend to lipid catabolism.


This work was supported by the Wilson and Amelia Wong Endowment Fund and Collaborative Group Research Award CUHK2/CRF/11G. W.M. and A.-S.H. were supported by postgraduate studentships from the University of Hong Kong. We thank the Arabidopsis Biological Resource Center for provision of pxa1 mutant seeds.