Fatty acid β-oxidation is an essential process in many aspects of plant development, and storage oil in the form of triacylglycerol (TAG) is an important food source for humans and animals, for biofuel and for industrial feedstocks. In this study we characterize the effects of a small molecule, diphenyl methylphosphonate, on oil mobilization in Arabidopsis thaliana.
Confocal laser scanning microscopy, transmission electron microscopy and quantitative lipid profiling were used to examine the effects of diphenyl methylphosphonate treatment on seedlings.
Diphenyl methylphosphonate causes peroxisome clustering around oil bodies but does not affect morphology of other cellular organelles. We show that this molecule blocks the breakdown of pre-existing oil bodies resulting in retention of TAG and accumulation of acyl CoAs. The biochemical and phenotypic effects are consistent with a block in the early part of the β-oxidation pathway.
Diphenyl methylphosphonate appears to be a fairly specific inhibitor of TAG mobilization in plants and whilst further work is required to identify the molecular target of the compound it should prove a useful tool to interrogate and manipulate these pathways in a controlled and reproducible manner.
Beta-oxidation is the major metabolic pathway by which fatty acids and fatty acid-like molecules are metabolized. In plants and yeasts, unlike mammals, β-oxidation is wholly compartmentalized in peroxisomes which contain a complete sequence of metabolic reactions to convert both saturated and unsaturated fatty acids to acetyl coenzyme A (acetyl-CoA) as well as metabolism of hormone precursors such as indole butyric acid (IBA) and oxo phytodienoic acid (OPDA) (Goepfert & Poirier, 2007; Graham, 2008; Wiszniewski et al., 2009; Hu et al., 2012). In oil seeds, reserve lipids are stored primarily as triacylglycerols (TAGs) in oil bodies of the cotyledons and/or endosperm that are formed from the endoplasmic reticulum during seed development (Murphy, 2012).
Upon seed germination, TAG hydrolysis is carried out by oil body-located TAG lipases such as SDP1 and SDPL (Eastmond, 2006; Kelly et al., 2011), Fig. 1. Substrates for β-oxidation are transported into peroxisomes by the ABC transporter CTS/PXA1/PED3. Recent biochemical characterization of the CTS protein has shown that it possesses an intrinsic thioesterase activity that cleaves acyl-CoAs during the transport cycle (De Marcos Lousa et al., 2013), so presumably fatty acids are activated by cytosolic acyl CoA synthetases before transport and reactivated within the peroxisome by acyl CoA synthetases LACS 6 and 7 (Fulda et al., 2004) which require ATP supplied by the peroxisome ATP transporters PNC1 and PNC2 (Arai et al., 2008; Linka et al., 2008). LACS 6 and 7 are specific for fatty acids and it is not known if other β-oxidation substrates are also accepted as CoA thioesters and cleaved upon transport. A peroxisomal protein PXN that transports a range of substrates including NAD+ and CoA has been identified (Agrimi et al., 2012; Bernhardt et al., 2012).
Once within peroxisomes, acyl-CoAs are oxidized by acyl-CoA oxidases (ACXs) 1-4 which exhibit overlapping chain length specificity (Graham, 2008, and references cited therein) to produce a Δ2-trans enoyl-CoA. The Δ2-trans enoyl-CoA is the substrate for the multifunctional protein (MFP) which has 2-trans enoyl hydratase and 1-3-hydroxyacyl-CoA dehydrogenase activities. Arabidopsis contains two MFP genes: MFP1(AIM1), expressed at low level in seedlings (Richmond & Bleecker, 1999) and MFP2 which encodes the major MFP activity in seedlings (Rylott et al., 2006). The resulting 3-ketoacyl-CoA, is cleaved by 3-ketoacyl thiolase (PED1/KAT2) to liberate acetyl CoA, and an acyl CoA two carbons shorter than the original molecule (Hayashi et al., 1998; Germain et al., 2001). The acetyl CoA produced by the thiolase reaction can enter the glyoxylate cycle and play a gluconeogenic role (Kunze et al., 2006) or be converted to citrate by peroxisomal citrate synthase and exported to mitochondria for metabolism by the tricarboxylic acid cycle (TCA cycle) (Pracharoenwattana et al., 2005).
Thus β-oxidation requires a concerted set of enzyme activities localized in peroxisomes, where disruption of any one of which can impact on oil mobilization. The genetic approach has been very valuable in identifying additional roles for β-oxidation beyond germination (Baker et al., 2006), in dormancy breaking (Footitt et al., 2002), auxin responses (Zolman et al., 2001b), biosynthesis of jasmonates (Afitlhile et al., 2005; Theodoulou et al., 2005), fertility (Footitt et al., 2007) and floral development (Richmond & Bleecker, 1999). Not all of these roles are yet fully understood.
An alternative, complementary, approach to genetics is to use inhibitors of specific enzyme steps. Whilst mutants are defective in specific steps, they may still exhibit pleiotropic effects. The MONODEHYDROASCORBATE REDUCTASE 4 (sdp2) mutant has a block in TAG mobilization because the SDP1 lipase is sensitive to inactivation by H2O2 which accumulates in this mutant (Eastmond, 2007) and the chy1 mutant, defective in peroxisomal valine catabolism, is also defective in β-oxidation due to accumulation of toxic methacrylyl-CoA (Zolman et al., 2001a). Inhibitors have the advantage of controlled application with respect to time and concentration, and can be applied to specific tissues or structures. Additionally, where multiple genes encode overlapping activities such as with the ACX family, multiple knockouts may be required to see a phenotype whereas application of a single inhibitor may inhibit all family members to a greater or lesser degree. A number of inhibitors of mammalian β-oxidation have been described (Schulz, 1983, 1987; Youssef et al., 1994) but there do not appear to be any available inhibitors to study this process in plants, which differ from mammals in having an exclusively peroxisomal β-oxidation system.
In this study, we report on the identification and characterization of a small molecule which blocks TAG mobilization in Arabidopsis seedlings, providing a valuable tool for the interrogation and manipulation of oil breakdown. This could be particularly useful in the investigation of tissue or developmental specific roles of oil breakdown, or in plant species where genetic analysis is not as facile as model plants like Arabidopsis.
Materials and Methods
Plant materials and growth conditions
Arabidopsis thaliana (L.) Heynh. lines contained the following constructs, 35S:: GFP-MFP2 (Cutler et al., 2000); 35S::CSY3-GFP (Pracharoenwattana et al., 2005); 35S::ST-GFP (Saint-Jore et al., 2002); 35S::GFP-HDEL (Batoko et al., 2000); 35S::GFP-FABD2 (Sheahan et al., 2004); Oleosin-GFP (Wahlroos et al., 2003); 35S::ATPβ-GFP (Logan & Leaver, 2000).
Seeds were sterilized, stratified in darkness for 48 h at 4°C, and sown on ½ Murashige and Skoog (MS) media (Duchefa, Haarlem, Netherlands) 0.8% (w/v) plant agar for vertical growth, or 0.3% (w/v) plant agar for 24-well microplate growth. Diphenyl methylphosphonate (TCI Europe nv Zwijndrecht, Belgium), 25 mM stock solution in DMSO, was diluted as indicated with hand hot ½MS media. Sucrose was added to 20 mM when required. Light (16 h d−1) seedlings were grown for 6 d (unless otherwise stated) at 23°C. Transplant assay: Stratified seeds were grown for 7 d on ½MS containing 0.1% (v/v) DMSO or 25 μM diphenyl methylphosphonate then transferred to ½MS containing 0.1% (v/v) DMSO or 25 μM diphenyl methylphosphonate for a further 7 d. IBA assay was as described (Dietrich et al., 2009).
For confocal microscopy an upright laser scanning microscope (LSM 510; Zeiss, Jena, Germany) with a ×40 or ×63 oil immersion objective lens was used for imaging. All images were scanned under identical conditions (laser power, photomultiplier gain, pinhole diameter and zoom) in relation to the relevant controls. GFP was imaged with the 488 nm line of an argon ion laser with a 505–530 band pass filter and Nile Red (Dietrich et al., 2009) with a 543 nm helium neon laser and 560–615 nm band pass filter. Post-acquisition image processing was done using the LSM 5 browser (Zeiss) and Adobe Photoshop 9.0 software (Adobe Systems, Mountain View, CA, USA).
Electron microscopy: Short sections of hypocotyls were fixed in 1% (v/v) glutaraldehyde and 1% (v/v) paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 6.9, washed 3× in buffer and post fixed in 2% (w/v) aqueous osmium tetroxide for 90 min (Faso et al., 2009). Samples were washed four times in water and subsequently block stained overnight in 1% aqueous uranyl acetate. Samples were dehydrated in acetone and embedded in TAAB low viscosity resin (TABB, Reading, UK) and sectioned with a RMC PowerTome XL ultra-microtome. Post-staining was carried out in lead citrate for 5 to 10 min and sections were observed with a Hitachi H-7650 transmission electron microscope.
Fatty acid and acyl-CoA profiling
Stratified seeds were plated onto ½MS media (0.8% (w/v) plant agar) containing 25 μM diphenyl methylphosphonate or 0.1% (v/v) DMSO and 20 mM sucrose. After 5 d in the light, hypocotyls and cotyledons were harvested for fatty acid profiling. Fatty acids and acyl-CoAs were profiled from the same extracts as described in Larson & Graham (2001) with the modifications described in Larson et al. (2002). Lipid extraction and neutral lipid analysis by liquid chromatography/tandem mass spectrometry (LC/MS/MS) were performed as described (Burgal et al., 2008).
Results and Discussion
Diphenyl methylphosphonate treated seedlings retain oil bodies but other organelles are unaffected
A small scale confocal laser scanning microscopy (CLSM)-based screen for compounds that altered peroxisome morphology and/or fluorescence intensity in an Arabidopsis line that expresses a peroxisomal targeted GFP reporter was carried out (Brown et al., 2011). The 70 compounds used in the small scale screen originated from a primary screen for compounds that affected hypocotyl gravitropism (Surpin et al., 2005) thus were known to be bioavailable and bioactive in plants. This report describes the detailed characterization of diphenyl methyl phosphonate which was identified as causing peroxisome clustering.
Control 6-d-old light grown seedlings have punctate peroxisomes that are distributed throughout the cytoplasm of hypocotyl cells, (Fig. 2a,c). When 25 μM diphenyl methylphosphonate is present in the medium, clustering of peroxisomes around spherical bodies of varying size is observed (Fig. 2d–f). These bodies stain with the lipophilic dye Nile Red (Fig. 2b,e,g) and contain oleosin, as revealed by oleosin-GFP fluorescence (Fig. 2h), identifying them as oil bodies. Co-localization of the peroxisomal GFP reporter and the Nile Red stained oil bodies shows that the peroxisomes cluster around the oil bodies (Fig. 2f,g). Transmission electron microscopy shows this dramatic accumulation of multiple large oil bodies in hypocotyl cells in the presence of diphenyl methylphosponate (Fig. 2i) and in the presence of diphenylmethyl phosphonate and 20 mM sucrose (Fig. 2j). Hypocotyl cells from control seedlings grown in the presence of sucrose but without diphenylmethyl phosphonate showed a large vacuole with just a thin layer of cytoplasm and normal cellular organelles (Fig. 2k). The effect of the compound was compared in the presence and absence of sucrose to facilitate subsequent comparisons with mutants disrupted in β-oxidation that are dependent upon sucrose for post-germinative growth.
Only oil bodies and peroxisomes are affected by diphenyl methylphosphonate treatment. When seedlings expressing ST-GFP (Golgi marker; Fig. 3a,b), GFP-FABD2 (actin marker; Fig. 3c,d) GFP-HDEL (ER marker Fig. 3e,f) or ATPβ-GFP (mitochondrial marker; Fig. 3g,h) were grown for 6 d on 25 μM diphenyl methyl phosphonate there was no obvious difference in appearance between control (Fig. 3a,c,e,f) and treated seedlings (Fig. 3b,d,g,h). Germination was not affected by the compound (data not shown), however, seedlings grown for 4 wk in the presence of 25 μM diphenyl methylphosphonate were small pale and stunted (Fig. 3j,k) compared to control (Fig. 3i) seedlings.
Hypocotyl growth in the dark is compromised in the absence of sucrose but IBA conversion to IAA is not affected
When seedlings were grown in the dark in the presence of diphenyl methylphosphonate a dose-dependent inhibition of hypocotyl growth was observed which was completely rescued by sucrose at concentrations of up to 5 μM and partially rescued even at 50 μM diphenyl methylphosphonate (Fig. 4a). Hypocotyl growth in the dark depends upon energy and carbon supplied by β-oxidation and many mutants defective in β-oxidation show a similar sucrose rescue phenotype (Pinfield-Wells et al., 2005). Beta-oxidation is also required to convert IBA to indole acetic acid (IAA) (Fig. 1) which results in stunting of roots and hypocotyl (Zolman et al., 2000). Seedlings were grown in the dark on sucrose in the presence of different concentrations of IBA and in the presence or absence of diphenyl methylphosphonate (Fig. 4b). Treated and untreated seedlings at the same IBA concentration showed the same percentage of hypocotyl shortening compared to the zero IBA treated and untreated controls, which were set to 100% (Fig. 4b). IBA and fatty acids share some steps in β-oxidation (Fig. 1) such as transport by CTS/PXA1, cleavage by KAT2/PED1 and the need for ATP and CoA supplied by membrane transporters, but other steps are distinct. These include an as yet unidentified acyl CoA synthetase(s), probably the acyl CoA oxidase/dehydrogenase IBR3 and IBA-specific enoyl CoA hydratases ECH2 and IBR10 (reviewed in Hu et al., 2012). IBA metabolism is arguably a very sensitive test of steps in β-oxidation that are shared with fatty acids since the pxa1-1 mutation is dominant for IBA resistance but recessive for sucrose dependence of hypocotyl elongation (Zolman et al., 2001b). Thus the diphenyl methylphosphonate target would appear to be specific to fatty acid degradation.
Organophosphorus phosphonate containing compounds inhibit a wide range of enzymes of the serine hydrolase class that possess a His/Asp/Ser catalytic triad, including acetyl cholinesterase, serine proteases and lipases, where a reaction between the active site serine and the phosphonate gives a covalent complex that mimics the transition state. Different inhibitors show different specificities (Oskolkova & Hermetter, 2002; Salisbury & Ellman, 2006). The clustering of peroxisomes around retained oil bodies seen in Fig. 2 is reminiscent of the phenotype of the sdp1 and sdp2 mutants and SDP1 is inhibited by the inhibitor E600 (diethyl p-nitrophenyl phosphate) and diisopropyl fluorophosphate (Eastmond, 2006).
Quantitative lipid analysis of treated seedlings reveals retention of TAG and accumulation of acyl CoAs
To investigate the block in oil body mobilization, a comprehensive analysis of the lipid composition of 5-d-old hypocotyl and cotyledon tissue from seedlings grown in the presence or absence of diphenyl methylphosphonate was carried out. Since oil bodies are retained in the hypocotyls of treated seedlings in the presence of sucrose (Fig. 2), seedlings were grown in the presence of 20 mM sucrose in order to facilitate comparison with a large data set of fatty acid, acyl-CoA and TAG profiles of a range of β-oxidation mutants (Hernandez et al., 2012).
Figure 5(a) shows an overview of lipids in control and diphenyl methylphosphonate treated seedlings, compared with dry seed. Galactolipids, the major lipids of chloroplast membranes, are absent in dry seed but no difference is seen between control and treated seedlings. Diacylglycerols (DAGs) are slightly elevated in treated compared to control seedlings, but the most striking change is in the level of TAGs. As expected, 5-d-old control seedlings had very low TAG levels (7% of dry seed) but diphenyl methylphosphonate-treated seedlings retained 64% of the TAG seen in dry seed. In treated seedlings the profile of acyl-CoAs show elevated levels of 20:0 and especially 20:1 compared to control (Fig. 5b), similar to a number of β-oxidation mutants (e.g. pxa1(cts), lacs6/lacs7 acx1/acx2 and kat2) but less like the TAG lipase sdp1 mutant which shows a significantly reduced level of acyl-CoAs in 2-d-old seedlings compared to the corresponding wild type (Eastmond, 2006) and a very slightly increased level of 20:1 CoA in 5-d-old sdp1 compared to wild type (Hernandez et al., 2012).
A quantitative analysis of TAG species in control and treated seedlings and dry seed was carried out (Fig. 5c) and shows that breakdown of all TAG species is compromised by diphenyl methylphosphonate, but some species are more compromised than others and some even show slightly increased levels in the treated seedlings compared to dry seed. This is particularly true for 18:3–20:1–18:3, 20:1–18:2–18:3 and to a lesser extent for 18:3–18:2–18:3, 20:1–20:1–18:3, 18:3–18:2–22:1 and 18:3–18:3:18:3. This is reflected in the higher levels of 18:3n318:2n6c and 20:1n9 in the treated seedlings (Fig. 5d) and is similar but not so extreme as seen in the pxa1 mutant, but clearly different to the acx1/acx2 double mutant, which shows increased levels of TAG after 5 d (Hernandez et al., 2012). Further, interference with CoA biosynthesis can be excluded since such mutants have a reduced acyl-CoA pool (Rubio et al., 2006). By contrast, mutants which block β-oxidation directly or indirectly downstream of the TAG lipase accumulate acyl-CoAs and oil bodies are retained, suggesting a feedback inhibition of lipolysis (Graham, 2008). Inhibition of processes other than β-oxidation could impact the acyl-CoA pool, for example inhibition of incorporation of acyl-CoAs into membrane lipids or cuticular wax, resulting in feedback inhibition of the TAG lipase. However this is unlikely in this instance as total monogalactosyl diacyl glycerol, the most abundant chloroplast lipid, is unchanged in diphenyl methylphosphonate treated seedlings relative to control (Fig. 5) and the level of MGDG with two 18:3 acyl chains is increased relative to MGDG with one 16:3 and one 18:3 in the treated samples (not shown), suggesting membrane lipids are acting as a sink for excess 18 carbon acyl-CoAs. Indeed these acyl-CoAs do not show increased levels in treated vs control seedlings (Fig. 5b).
Diphenyl methylphosphonate treatment results in irreversible inhibition of oil body breakdown, not induction of oil body synthesis
To determine whether the inhibitor blocked the mobilization of existing oil bodies or promoted their de novo synthesis, seedlings were grown in the absence (0.1% DMSO; Fig. 6a) or presence of 25 μM diphenyl methylphosphonate (Fig. 6b). After 7 d, when oil bodies had been completely metabolized in the control and were retained in the treated seedlings, the seedlings were transplanted to media without (Fig. 6c,f) or with (Fig. 6d,e) 25 μM diphenyl methylphosphonate. Transplantation of 7-d-old seedlings onto the compound did not induce the formation of oil bodies or lipid droplets (Fig. 6d). Oil bodies were retained in hypocotyls of treated seedlings after 7 d on chemical-free media, suggesting an irreversible inhibition of TAG breakdown (Fig. 6f). In contrast to the pxa1 mutant where post-germinative recycling of fatty acids into TAG occurs in seedlings (Hernandez et al., 2012) and leaves (Slocombe et al., 2009), the mode of action of the compound is not through promoting TAG synthesis.
In summary we have identified a compound, diphenyl methyl phosphonate, which appears to be a reasonably specific inhibitor of oil body mobilization during post-germinative growth. The phenotypes show a number of similarities with the sdp1 mutant and the SDP1 lipase has been shown to be inhibited by diethyl p-nitrophenyl phosphate, but the TAG profiles differ and show more similarity to the pxa1 mutant. We considered whether diphenyl methylphosphonate might inhibit the thioesterase activity of the CTS protein but in vitro assays on insect expressed protein showed no inhibition of the activity and neither was the activity of the peroxisomal acyl CoA thioesterase ACH2 which has a Asp/Gln/Ser catalytic triad (Tilton et al., 2004) affected (data not shown). It is possible that in vivo there is partial inhibition of more than one target. While further analysis is required to delimit the precise mode of action this compound will be a useful tool for investigating aspects of lipid breakdown. It has the advantage that it can be applied at different concentrations at any developmental stage allowing early effects of inhibition to be studied, or to any tissue allowing investigation of tissue specific effects. It should be useful in the study of lipid breakdown in less genetically tractable plant species.
This work was funded by The Leverhulme Trust (RF/2/2005/0378 to A.B.) and BBSRC (BB/E013740/1 to A.B. and S.L.W.). A.B. gratefully acknowledges Prof. Julia-Bailey-Serres, Prof. Natasha Raikhel, Dr Marci Surpin and Dr David Carter UC Riverside for their support and generosity in the initiation of this project. Also, Prof. Patrick Hussey (University of Durham, UK) for GFP-FABD2, Dr David Logan (University of Angers, France) for ATPβ-GFP, Prof. Steven Smith (University of Western Australia) for CSY3-GFP and Prof. John Browse (Washington State University, USA) for ACH2. The authors thank Drs Yvonne Nyathi, Carine De Marcos Lousa and Thomas Lanyon-Hogg for testing the effects of diphenyl methyl phosphonate on ATPase and thioesterase activities of CTS and ACH2.