Morphology and biochemical processes can be markedly affected when plants are exposed to a Pi-depleted growth environment. Morphological changes, such as growth retardation of shoots and primary roots and elongation of lateral roots and root hairs, have been documented extensively in Arabidopsis plants transferred to Pi-depleted conditions (Bates and Lynch, 1996; Raghothama, 1999; Ticconi and Abel, 2004; Desnos, 2008; Péret et al., 2011; Sato and Miura, 2011). When sesame seedlings were grown under Pi-depleted conditions, morphological changes in the lateral roots seemed to be more obvious than those observed in Arabidopsis (Figure 1A). We measured fresh weight per seedling under Pi-sufficient or -depleted growth conditions for shoots and roots, respectively (Figure 1B). It should be noted that conflicting effects of Pi depletion have been reported for Arabidopsis. Whereas Kobayashi et al. (2009a) reported that the fresh weight of shoots and that of roots both decreased when plants were grown under Pi-depleted conditions (0 mM Pi in agar medium containing 1% sucrose), Versaw and Harrison (2002) observed that the fresh weight of shoots decreased but that of roots was unchanged under other Pi-depleted conditions (0.2 mM Pi in agar medium without sucrose). Sucrose in the growth medium enhances the sensitivity of plants to Pi depletion (Lei and Liu, 2011). Given that we used the same growth medium as Kobayashi et al. (2009a), we compared the sesame phenotype with their Arabidopsis data.

In sesame, fresh weight of shoots was drastically decreased upon transfer to Pi-depleted medium, similar to the effects shown by Kobayashi et al. (2009a) for Arabidopsis. In contrast, the fresh weight of roots remained unchanged upon transfer to Pi-depleted medium (Figure 1B). These data showed that, when sesame and Arabidopsis are compared, the morphological changes caused by low-Pi stress are similar in shoots but different in roots. The data further suggested that the significant elongation of lateral roots in sesame plants grown under Pi-depleted conditions might offer an advantage for survival under conditions of low Pi availability.

Regarding the lipid compositions of cellular membranes under Pi-depleted conditions, Arabidopsis is known to degrade phospholipids and supply Pi needed for essential biological processes. A simultaneous increase in the concentration of the galactolipid DGDG compensates for the loss of phospholipids from membranes. In both shoots and roots of sesame plants grown under Pi-depleted conditions, the mol% (relative to total membrane lipids) of DGDG and sulfoquinovosyldiacylglycerol (SQDG) increased and that of phospholipids (phosphatidylglycerol, phosphatidylethanolamine, phosphatidylinositol, and PC) decreased (Figures 2A, B). In shoots, most of the changes were comparable between Arabidopsis and sesame (Figure 2A). However, Pi depletion decreased the phospholipid mol% (relative to total membrane lipids) by 32% (from 79 to 47%) in sesame roots (Figure 2B). This reduction is slightly larger than the 24% reduction (from 88 to 64%) observed in Arabidopsis roots (Kobayashi et al., 2009a). In sesame roots grown under Pi-sufficient conditions, the molar ratio of galactolipids MGDG and DGDG in the total membrane lipids (20%) was much higher than that in Arabidopsis roots (9%; Kobayashi et al., 2009a). This difference might result from differences in the abundance of plastids in sesame roots under both of the Pi conditions tested. Exposure of Arabidopsis to Pi-depleted conditions increases the ratio of DGDG to total acyl lipids in roots but not that of MGDG to total acyl lipids (Kobayashi et al., 2009a). Under Pi-depleted conditions, however, not only were mol% of DGDG and MGDG in sesame roots increased relative to the respective mol% under Pi-sufficient conditions, but also the total molar ratio of MGDG and DGDG (combined) reached 53% of the total membrane lipids, which is 20% higher than that measured in Arabidopsis roots (Kobayashi et al., 2009a).

Phosphate depletion caused a marked decrease in the abundances of phospholipids in sesame roots. These data suggested that, under Pi-depleted conditions, MGDG synthesis in sesame roots was more strongly up-regulated than in Arabidopsis roots and that the increased supply of Pi from the accelerated degradation of phospholipids might render sesame plants tolerant to low Pi availability by enabling them to maintain root growth under Pi-depleted conditions.

The effects of Pi depletion on the fatty acid composition of MGDG and DGDG in sesame are shown in Figures 2C–F. Given that MGDG contains no 16:3 fatty acid, which is a signature of “16:3 plants,” sesame should clearly be categorized as one of the “18:3 plants,” which only possess eukaryotic pathways for lipid synthesis (Figures 2C, D; Heinz and Roughan, 1983; Li-Beisson et al., 2013). Fatty acid compositions of DGDG under both of the Pi conditions tested were similar in sesame (Figures 2E, F). Given that DGDG is mainly synthesized from diacylglycerol derived from eukaryotic pathways under both Pi-sufficient and -depleted conditions, the fatty acid composition of DGDG does not differ substantially between Arabidopsis and sesame, although they are distinctly categorized as “16:3 plants” and “18:3 plants,” respectively (Figures 2E, F; Kobayashi et al., 2009a).

We isolated two genes for sesame MGDG synthase, namely SeMGD1 and SeMGD2 (accession numbers AB841066 and AB841067, respectively). SeMGD1 is a 516-amino acid residue protein that is predicted by ChloroP to contain a chloroplast transit peptide (Emanuelsson et al., 1999) and is categorized as a type-A MGDG synthase on the basis of its similarity to other type-A MGDG synthases (Figure 3). SeMGD2 is a 475-residue protein that is categorized as a type-B MGDG synthase. The failure to identify a putative chloroplast transit peptide in the SeMGD2 sequence suggests that, like AtMGD2 and AtMGD3, SeMGD2 localizes to the outer envelope membrane of chloroplasts. Comparison of the amino acid sequences of SeMGD1 and SeMGD2 with other MGDG synthases (Figure 3) indicated that SeMGD1 and SeMGD2 are highly similar to AtMGD1 (81.3%) and AtMGD2 (79.6%), respectively. We could not find the third gene for MGDG synthase by screening a sesame cDNA library. In Arabidopsis, AtMGD2 and AtMGD3 are both involved in MGDG synthesis under Pi-depleted conditions, but AtMGD3 is rather predominantly involved in MGDG synthesis under Pi-depleted conditions (Kobayashi et al., 2009a). Moreover, differences in expression patterns in seedlings suggested that AtMGD2 and AtMGD3 have distinct tissue-specific roles in development, although the mgd2 mutant did not show an apparent phenotype compared with the WT (Awai et al., 2001; Kobayashi et al., 2004). It is possible that sesame might have another type-B MGDG synthase that has another function, given that the cDNA library used was generated from cotyledons of seedlings grown under Pi-sufficient conditions.

In Arabidopsis, AtMGD1 is constitutively expressed, whereas the transcript abundance of AtMGD2 and AtMGD3 transcripts are markedly increased after transfer to Pi-depleted conditions (Awai et al., 2001; Kobayashi et al., 2004). However, it is not known whether type-B MGDG synthases from other plant species are transcriptionally up-regulated under Pi-depleted conditions. We analyzed the abundance of SeMGD1 and SeMGD2 mRNAs under Pi-sufficient and Pi-depleted growth conditions using quantitative RT-PCR (Figure 4). In sesame, the observed increases in levels of the SeMGD2 transcript under Pi-depleted conditions both in shoots and roots (Figures 4C, D) showed that the transcriptional up-regulation of genes that encode type-B MGDG synthases under Pi-depleted conditions is conserved in plants other than Arabidopsis. Similar to what was observed for AtMGD1, SeMGD1 expression in shoots remained unchanged under Pi-sufficient and Pi-depleted conditions (Figure 4A), but that in roots under Pi-depleted conditions was higher than that under Pi-sufficient conditions (Figure 4B). AtMGD1 is constitutively expressed in various organs, and levels of AtMGD1 transcript are not affected by the Pi concentration in the growth medium (Awai et al., 2001; Kobayashi et al., 2004). Thus, the regulation of MGDG synthesis under Pi-depleted conditions seems to differ between Arabidopsis and sesame.

Cucumis sativus MGD1 (CsMGD1) was the first MGDG synthase for which the gene was isolated and identified in higher plants (Shimojima et al., 1997). The enzymatic properties of CsMGD1 have been well characterized (Yamaryo et al., 2003, 2006). To verify whether SeMGDs could be regulated by cellular redox status, we measured SeMGD activity under both oxidizing and reducing conditions (Figures 5A, B). The specific activity of SeMGD1 was much higher than that of SeMGD2, and the MGDG synthase activity of each SeMGD1 and SeMGD2 was regulated by redox status. The inactivation of both SeMGD1 and SeMGD2 by oxidization and their activation by reduction suggests that the activities of both type-A and type-B MGD might be also regulated by cellular redox status, as has been shown for CsMGD1 (Yamaryo et al., 2006).

When comparing various sequences of MGDG synthases, we found five highly conserved cysteine residues at positions 277, 291, 383, 418, and 420 of CsMGD1 (Figure 3). Given that our results showed that both types of SeMGDs are regulated by redox state, these five cysteines may be important for enzyme activation.

CsMGD1, spinach MGD1, and AtMGD1 are each activated in vitro by the inclusion of either phosphatidic acid (PA) or phosphatidylglycerol (Covès et al., 1988; Ohta et al., 1995; Dubots et al., 2010). Extensive biological analyses using Arabidopsis concluded that the activation of MGD1 by PA is essential and indispensable for MGDG synthesis in chloroplasts (Dubots et al., 2010). Although the PA-binding residue in the mature protein of AtMGD1 was not determined, the important residues for full activation by PA were previously shown to be P189, H251, W287, and E456 (Dubots et al., 2010; Figure 3). The conservation of all four of these residues in SeMGD1 and SeMGD2, as well as the other MGDG synthases compared in Figure 3, suggests that both SeMGD1 and SeMGD2 might be activated by PA. All published research on the activation by lipids was performed using type-A MGDG synthases. We analyzed activation of type-A and type-B SeMGDs by PA using partially purified recombinant enzymes expressed in Escherichia coli. Intriguingly, both SeMGDs were activated by PA (Figure 6), but again the specific activity of SeMGD1 was much higher than that of SeMGD2. It should be noted that the specific activity of desalted SeMGD1 and SeMGD2 in Figures 5A, B (~6.0 × 10⁴ and ~0.7 × 10² nmol min^-1 mg^-1 protein, activity of oxidized SeMGD1 and SeMGD2 + DTT, respectively) was higher than that of the non-desalted proteins in Figures 6A, B (~0.2 × 10⁴ and ~0.4 × 10² nmol min^-1 mg^-1 protein, activity of SeMGD1 and SeMGD2 at PA = 0 mol%, respectively). The presence of salts could also explain the difference in PA concentrations required for the maximum activation of both types of SeMGD (10 mol% for SeMGD1, 18 mol% for SeMGD2) were substantially higher than those required for AtMGD1 (0.6 mol%; Dubots et al., 2010).

Sesame seedlings (S. indicum L. strain 4294) were grown on Murashige and Skoog medium solidified with 0.8% (w/v) agar containing 1% (w/v) sucrose for 5 days at 28°C under continuous white light, transferred to Pi-sufficient (1.0 mM) or Pi-depleted (0 mM) medium, and grown for another 14 days (Härtel et al., 2000).

Total lipids were extracted and separated by two-dimensional thin layer chromatography as described by Kobayashi et al. (2007). Lipids isolated from silica gel plates were methylated, and fatty acid methyl esters were quantified by gas chromatography using pentadecanoic acid as an internal standard (Kobayashi et al., 2006).

Total RNA was isolated from sesame cotyledons using an RNA isolation kit (Stratagene). A sesame cDNA library was generated using a cDNA synthesis kit, ZAP-cDNA synthesis kit, and ZAP-cDNA Gigapack III gold cloning kit (Stratagene).

Consensus sequences of type-A MGDG synthase (nucleotide number 540–1378 of AtMGD1 ORF, NM 119327) and type-B MGDG synthase (nucleotide number 527–1184 of AtMGD2 ORF, NM 122048) were selected based on the nucleotide sequence alignment of the sequences encoding MGDG synthases from C. sativus (U62622), Nicotiana tabacum (AB047476), O. sativa (AK243359, AB112060, NM 001068022), Spinacia oleracea (AJ249607), and G. max (AB047475). Degenerate primers for nested RT-PCR were constructed; for type-A MGDG synthases: forward 5′-tttntggncngancanacnccntggcc-3′ and reverse 5′-cnttnccngcntcntgnccngcnatgta-3′ for the first PCR and forward 5′-ccngatatnatnatcagtgtncatcc-3′ and reverse 5′-gnccngccttngtnatnatncantcaca-3′ for the second PCR; for type-B MGDG synthases: forward 5′-gngtncanccnntnatgcaacanattcc-3′ and reverse 5′-gcnccattntnnacnacatanggnacatt-3′ for the first PCR and forward 5′-ganctnannacntgncancctacntggtt-3′ and reverse 5′-tgnnatnatgcantcacangcncccat-3′ for the second PCR. These PCR fragments were used to screen the sesame cDNA library. The full-length SeMGD1 cDNA was obtained by 5′-RACE using the 5′-Full RACE Core Set (TaKaRa), and that of SeMGD2 cDNA was obtained by 5′-RACE and 3′-RACE using the RNA PCR kit (AMV) Ver. 3.0 (TaKaRa).

Total RNA was isolated from three independent sesame samples of shoot and root using the SV Total RNA isolation system (Promega). Reverse transcription was performed using the PrimeScript RT reagent kit (TaKaRa). cDNA was amplified using SYBR PreMix Ex Taq (TaKaRa). Signal detection and quantification were performed in duplicate using the Thermal Cycler Dice Real Time System (TaKaRa). Quantification of SeMGD1 and SeMGD2 transcripts by quantitative PCR was carried out using the SiACT (JP631637) and SiUBQ6 (JP631638) transcript levels for normalization, respectively (Wei et al., 2013). Expression levels were obtained from at least three replicates. The following gene-specific primers were used:

SeMGD1_fw: 5′ GGTCTATGGCCTTCCCGTACGG 3′

SeMGD1_rv: 5′ TGACTGCAATTTGCTGGCTAGC 3′

SeMGD2_fw: 5′ CTAGATGGGCTCGAAGAATCTC 3′

SeMGD2_rv: 5′ GATAATCAGTTGTCCGATTGGC 3′

SiACT_fw: 5′ GACGAGCTCTGCTGTAGAGAAG 3′

SiACT_rv: 5′ CCACCACTGAGAACAATATTGC 3′

SiUBQ6_fw: 5′ GATAGGCACTACTGTGGTAAGTG 3′

SiUBQ6_rv: 5′ GGAACAGCAATGGTGTCAGCAAC 3′.

SeMGD1 and SeMGD2 were amplified by PCR using a gene-specific primer set (forward primer 5′-gccatatggttgatgccggggagaataa-3′ and reverse primer 5′-gcgaattcgtagttgtacaatactgaggtact-3′ for SeMGD1; forward primer 5′-gccatatgatgttcagaagctctacatttat-3′ and reverse primer 5′-gcgaattcgtgtaaattaggctcgagaagga-3′ for SeMGD2) and were then each cloned separately into the pET24a vector. Vectors were transformed into competent cells of E. coli strain BL21(DE3). Transformed cells were grown in Luria–Bertani medium at 37°C for 16 h and diluted 10-fold using medium to grow for a further 2 h. Protein expression was induced using 1 mM isopropyl thiogalactopyranoside for 3 h at 37°C for SeMGD1 and at 20°C for SeMGD2. Cells were collected by centrifugation at 3000 × g for 15 min and suspended in lysis buffer (0.1 M MOPS–NaOH, pH 7.8).

Transformed E. coli were resuspended in lysis buffer and disrupted by sonication (Ultrasonic disrupter UD-201, TOMY). Cell lysates were centrifuged at 5000 × g for 3 min at 4°C, and the resulting supernatant was centrifuged again at 125,000 × g for 30 min at 4°C. Each supernatant was applied to a nickel-charged resin (Ni-NTA Agarose, Qiagen GmbH), Recombinant MGDG synthase was allowed to bind to the resin at 4°C for 30 min, and the resin was washed three times with a three-bed volume of lysis buffer that contained 0.2 M NaCl and 10 mM imidazole. Each His-tagged MGDG synthase was eluted with lysis buffer containing 0.2 M NaCl, 10% (v/v) glycerol, and 200 mM imidazole to yield a pure protein. In all experiments, protein concentrations were determined using bovine serum albumin as the standard (Bensadoun and Weinstein, 1976).

MGDG synthase activity was measured by determining the amount of [4,5-³H]galactose incorporated into the lipid fraction (Yamaryo et al., 2003). Briefly, after pre-incubation of SeMGD in 190 μL of assay mixture (6.4 mM diacylglycerol in 0.01% (w/v) Tween 20, 10 mM dithiothreitol, 10 mM sodium acetate, and 0.1 M MOPS–NaOH pH 7.8) at 30°C, 10 μL of UDP-[4,5-³H]galactose (5.1 mM, 29 Bq nmol^-1) was added to initiate the reaction, which was allowed to proceed at 30°C for 30 min. The reaction product was quantified using an image analyzer after thin layer chromatography or measured using a liquid scintillation counter.

Lipids were removed from the purified SeMGD preparation (1 mg mL^-1) by stirring at 4°C for 30 min in lysis buffer that contained 24 mM lauryl dimethylamino oxide. MGDG synthase activity was then assessed after the addition of PA using the method described above. PA concentration was calculated as the mol% of total micelle concentration in the assay mixture (Maréchal et al., 1994).

Oxidation and reduction of the SeMGD isoforms were performed according to Yamaryo et al. (2006) with slight modifications. Briefly, the purified MGD was treated with 50 μM CuCl₂ for 1 h on ice. The oxidized enzyme was then subjected to gel filtration (ProbeQuant G-50 Micro Column, GE Healthcare) to remove Cu²⁺ ions and salts and was then used in assays to measure the activity of oxidized MGDG synthase. Reduced MGD was prepared by incubation with 1 mM dithiothreitol at 30°C for 30 min.