During phosphate (Pi) starvation in plants, membrane phospholipid content decreases concomitantly with an increase in non-phosphorus glycolipids. Although several studies have indicated the involvement of phytohormones in various physiological changes upon Pi starvation, the regulation of Pi-starvation induced membrane lipid alteration remains unknown. Previously, we reported the response of type B monogalactosyl diacylglycerol synthase genes (atMGD2 and atMGD3) to Pi starvation, and suggested a role for these genes in galactolipid accumulation during Pi starvation. We now report our investigation of the regulatory mechanism for the response of atMGD2/3 and changes in membrane lipid composition to Pi starvation. Exogenous auxin activated atMGD2/3 expression during Pi starvation, whereas their expression was repressed by cytokinin treatment in the root. Moreover, auxin inhibitors and the axr4 aux1 double mutation in auxin signaling impaired the increase of atMGD2/3 expression during Pi starvation, showing that auxin is required for atMGD2/3 activation. The fact that hormonal effects during Pi starvation were also observed with regard to changes in membrane lipid composition demonstrates that both auxin and cytokinin are indeed involved in the dynamic changes in membrane lipids during Pi starvation. Phosphite is not metabolically available in plants; however, when we supplied phosphite to Pi-starved plants, the Pi-starvation response disappeared with respect to both atMGD2/3 expression and changes in membrane lipids. These results indicate that the observed global change in plant membranes during Pi starvation is not caused by Pi-starvation induced damage in plant cells but rather is strictly regulated by Pi signaling and auxin/cytokinin cross-talk.
Monogalactosyl diacylglycerol (MGDG) and digalactosyl diacylglycerol (DGDG) are predominant components of thylakoid membranes, which account for about 50 and 30 mol% of the total membrane lipid constituents of chloroplasts in higher plants (Block et al., 1983). These galactolipids are synthesized in the envelope membrane of plastids, and were thought to be localized exclusively in plastids (Ohta et al., 2000). Recent studies in Arabidopsis, however, have clarified drastic alterations in the composition of membrane lipid classes under phosphate (Pi)-starvation conditions (reviewed in Dörmann and Benning, 2002). Phospholipids constitute one of the largest phosphorus pools in planta; in response to Pi starvation, phospholipid levels fall concomitantly with a large increase in the non-phosphorus galactolipid DGDG (Essigmann et al., 1998). This DGDG accumulation has been observed in extra-plastidic membrane fractions prepared from Pi-starved Arabidopsis (Härtel et al., 2000). Moreover, in Pi-starved oat (Avena sativa L.), Andersson et al. (2003) showed a large accumulation of DGDG in plasma membranes, which are composed largely of phospholipids and lack galactolipids under nutrient-sufficient conditions. In addition, Jouhet et al. (2004) used Arabidopsis suspension-cultured cells to demonstrate that mitochondria are also a major site of DGDG accumulation under Pi starvation, although these organelles contain hardly any membrane DGDG under Pi-sufficient conditions. These data suggest that DGDG replaces phospholipids as the major membrane component under Pi-starved conditions, and indicate the importance of DGDG in extra-plastidic membrane systems during Pi starvation.
Three Arabidopsis MGDG synthase genes have been cloned and are classified into type A (atMGD1) and type B (atMGD2 and atMGD3) (Awai et al., 2001; Miège et al., 1999). atMGD1 is expressed widely in green tissues, and the enzyme is thought to contribute substantially to the construction of thylakoid membranes. By contrast, atMGD2/3 expression is very low, particularly in photosynthetic tissues under nutrient-sufficient conditions, but is strongly activated upon Pi starvation (Awai et al., 2001; Kobayashi et al., 2004). As for DGDG synthase, two isoforms (DGD1 and DGD2) were identified in Arabidopsis, and both are strongly induced during Pi starvation (Kelly and Dörmann, 2002; Kelly et al., 2003). As DGDG is synthesized by galactosylation of MGDG, it is presumed that these genes function cooperatively during DGDG accumulation under Pi-starved conditions (Benning and Ohta, 2005). During Pi starvation, phospholipid degradation was also observed (Essigmann et al., 1998). Andersson et al. (2004) demonstrated in garden pea (Pisum sativum L.) that a phospholipase D was involved in phospholipid hydrolysis, supplying a substrate for galactolipid synthesis. However, Nakamura et al. (2005) recently reported a novel Arabidopsis phospholipase C, NPC4, that is strongly activated by Pi starvation. They showed that induction of NPC4 expression by Pi starvation coincides with the activation of type B MGD genes. In addition, under Pi-starved conditions, the activity of phospholipase C but not phospholipase D was highly upregulated, particularly in the root, paralleling galactolipid-synthesizing activity (Kobayashi et al., 2004; Nakamura et al., 2005). Because phospholipid degradation is observed concomitantly with glycolipid accumulation, it is likely that Pi-starvation induced conversion from phospholipids to glycolipids is tightly controlled by Pi signaling and/or other factors. However, the regulatory mechanism that activates glycolipid biosynthesis and phospholipid degradation upon Pi starvation remains unknown, although several studies have indicated the involvement of phytohormones in various physiological changes during Pi starvation (Abel et al., 2002; Franco-Zorrilla et al., 2004; López-Bucio et al., 2003; Ticconi and Abel, 2004).
In an effort to understand the mechanism by which MGDG synthesis is induced upon Pi starvation, we analyzed various activation factors including phytohormones for type B MGD genes using MGD promoters coupled with the GUS reporter system, which is an appropriate system to assess minute changes in the expression of MGD genes (Kobayashi et al., 2004). Our results show that auxin is required to activate type B MGD genes, whereas cytokinin represses their induction in the root. These hormonal effects were also observed with regard to changes in membrane lipid composition, indicating the involvement of auxin/cytokinin cross-talk in the Pi-starvation induced conversion of membrane lipids. We also show that signal transduction through a Pi-sensing mechanism regulates not only promoter activity of type B MGD genes but also membrane lipid composition during Pi starvation. Our data demonstrate that the global changes in plant membranes upon Pi starvation are tightly regulated by Pi signaling and auxin/cytokinin cross-talk.
Spatial analysis of type B MGD promoter–GUS activity during Pi starvation
We previously investigated Pi-starvation responsive expression of type B MGD genes using RT-PCR and promoter–GUS analysis. To investigate the histological regulation of type B MGD promoters during Pi starvation, we performed a detailed GUS analysis in Pi-starved type B MGD::GUS transformants (Figure 1). In rosette leaves, GUS activity was observed at the apices of serrated edges, corresponding to hydathodes (Figure 1a). This GUS activity was highly enriched in epithem cells (Figure 1b), although staining was also observed at the base of trichomes and in stipules (Figure 1c,d). In the Pi-starved root, although widespread blue staining was detected throughout the root, GUS activity was most intense in lateral root branches (Figure 1e). These patterns of GUS expression were similar to the auxin-derived GUS activity of auxin-responsive element::GUS (DR5::GUS) constructs (Aloni et al., 2003; Avsian-Kretchmer et al., 2002), implying the involvement of auxin signaling in Pi-starvation induced expression of type B MGD genes. There was no significant difference in the staining pattern between atMGD2::GUS and atMGD3::GUS plants under these conditions, suggesting that the spatial regulation of these genes is the same during Pi starvation. As these staining patterns were not detected under nutrient-sufficient conditions (data not shown), the histological expression of type B MGD genes shown in Figure 1 was specific to Pi starvation.
Effects of phytohormones on the expression of type B MGD genes under Pi-sufficient and Pi-starved conditions
Recent studies have proposed that phytohormones such as auxin, ethylene and cytokinin are involved in several regulatory responses to Pi deficiency in Arabidopsis (Abel et al., 2002; Franco-Zorrilla et al., 2004; López-Bucio et al., 2003; Ticconi and Abel, 2004). Moreover, histochemical analysis in type B MGD::GUS constructs has indicated the possible involvement of auxin in the expression of these genes during Pi starvation. To assess whether these hormones are involved in the expression of type B MGD genes under Pi starvation, Pi-replete and Pi-depleted type B MGD::GUS transformants were treated with 10 μm indole-3-acetic acid (IAA), 1-aminocyclopropane-1-carboxylic acid (ACC) or 6-benzyladenine (BA). Figure 2 shows the quantitative GUS activity driven by the atMGD2 promoter. Under Pi-sufficient conditions, GUS activity was very low in all tissues and did not change in response to IAA, ACC or BA, in agreement with RT-PCR results (Kobayashi et al., 2004). However, upon Pi starvation, these transgenic plants had elevated GUS activity, as previously reported (Kobayashi et al., 2004). When the transformants were treated with IAA under Pi-starved conditions, GUS activity rose approximately threefold in both shoots and roots. By contrast, BA treatment strongly inhibited GUS activity in the root of Pi-starved transformants, resulting in 96% less activity compared with the untreated control, although the BA-treated shoot showed slightly higher GUS activity than the untreated control. Histological staining showed that 10 nm BA was sufficient to repress the expression of both atMGD2 and atMGD3 in Pi-starved roots (data not shown). As the importance of auxin–cytokinin cross-talk has been well-documented in planta, we further investigated GUS activity in the roots of atMGD2::GUS plants treated simultaneously with IAA and BA. Under Pi-replete conditions, GUS activity was not induced. However, under Pi-starved conditions, IAA/BA treatment enhanced GUS activity in the root as high as sixfold compared with BA treatment alone, suggesting that the actions of IAA and BA are antagonistic during Pi-starvation induced atMGD2 expression in the root. Histochemical analysis of atMGD3::GUS plants confirmed this antagonistic effect (data not shown), suggesting that both atMGD2 and atMGD3 are regulated in a similar manner by IAA and BA during Pi starvation. As Pi-deficient responses in the root correlate with ethylene signal transduction (Borch et al., 1999; Ma et al., 2003; Zhang et al., 2003), we evaluated the effect of ethylene on Pi-starvation induced atMGD2 expression. When ACC, an ethylene precursor, was applied to atMGD2::GUS plants under Pi-starved conditions, there was little difference in GUS activity between ACC-treated and untreated plants. Although the involvement of auxin and cytokinin in ethylene biosynthesis has been reported (Vogel et al., 1998; Woeste et al., 1999; Yu and Yang, 1979), our results confirm that ethylene is unlikely to be involved in atMGD2 activation during Pi starvation.
Endogenous auxin requirement for the induction of type B MGDG synthase genes by Pi starvation
Exogenous IAA enhanced Pi-starvation induced atMGD2 expression in both shoots and roots (Figure 2), suggesting that auxin may be involved in the induction of type B MGD genes under Pi-starved conditions. To investigate the effect of endogenous auxin on the promoter activity of type B MGD genes during Pi starvation, auxin transport was blocked by N-(1-naphthyl) phthalamic acid (NPA) – one of the most commonly used polar auxin transport inhibitors. Figure 3(a) shows the effect of NPA on type B MGD::GUS gene expression under Pi-starved conditions. NPA treatment repressed the widespread GUS activity observed in roots under Pi-starved conditions, and staining was restricted to root tips. However, simultaneous addition of IAA and NPA yielded intense GUS staining throughout the Pi-starved root. We confirmed that 0.1 μm of IAA was enough to induce the strong GUS activity in the NPA-treated root (Figure 3a). These results demonstrate that the inhibitory effect of NPA on Pi-starvation induced expression of type B MGD genes in the root is due to the impairment of polar auxin transport. NPA inhibits not only the downward transport from the shoot to the root but also the upward transport from the root tip towards the base (Jones, 1998), and this inhibition of basipetal auxin transport from the root tip causes accumulation of auxin in this area, and induces swelling of the root apex (Ruegger et al., 1997). The swollen root tips of type B MGD::GUS plants, where auxin levels are thought to be increased by NPA treatment, showed strong GUS activity (Figure 3a), adding to the evidence that type B MGD genes are responsive to auxin under Pi-starved conditions.
The effect of NPA was also observed in the aerial tissues of Pi-starved type B MGD::GUS plants. NPA-treated plants showed little GUS staining in the stipules, where intense GUS activity was detected under Pi-starved conditions in the absence of NPA (Figure 3b). Avsian-Kretchmer et al. (2002) showed that NPA impairs auxin-mediated DR5::GUS activity in the shoot apical meristem and stipules, indicating that NPA inhibits auxin transport to the shoot apex, thereby suppressing auxin accumulation in these tissues. Therefore, it is presumed that inhibition of auxin transport by NPA decreases local auxin content in the stipules, thereby reducing the GUS activity of type B MGD::GUS constructs. These results suggest that auxin is necessary for type B MGD gene expression not only in the root but also in the shoot apex.
To exclude secondary effects of NPA on the expression of type B MGD genes, we evaluated the auxin requirement for Pi-starvation induced expression of type B MGD genes using 2,3,5-triiodobenzoic acid (TIBA) and p-chlorophenoxyisobutyric acid (PCIB), which have been used widely to inhibit polar auxin transport and auxin action, respectively. Figure 3(c) shows the quantitative GUS activity of atMGD2::GUS plants grown on Pi-starved media containing TIBA or PCIB. Although GUS activity in shoots was essentially not affected, TIBA and PCIB treatment reduced the GUS activity in transgenic roots by 90% and 75%, respectively, compared with untreated controls. These inhibitory effects were also detected in atMGD3::GUS roots but not in shoots (data not shown), suggesting that the same auxin system regulates atMGD2 and atMGD3 during Pi starvation. The fact that TIBA acts as an inhibitor of polar auxin transport is consistent with our observation that TIBA does not inhibit GUS activity in shoots, a major site of auxin biosynthesis. Oono et al. (2003) reported severe physiological effects of PCIB in roots but not in aerial tissues, suggesting that this compound is less effective on shoots. Thus, PCIB may not inhibit auxin action in shoots with regard to the expression of type B MGD::GUS fusions.
Geldner et al. (2001) demonstrated that commonly used auxin transport inhibitors such as TIBA and NPA inhibit not only PIN1 cycling, which is involved in auxin efflux from the cell, but also trafficking of membrane proteins unrelated to auxin transport. To exclude the possibility of auxin-independent effects of these inhibitors on type B MGD gene expression, we investigated GUS activity in detached roots of type B MGD::GUS plants, in which polar auxin transport from the shoot was completely blocked. Transgenic plants carrying atMGD2::GUS fusions were grown on MS medium for 15 days, separated into shoots and roots, and incubated in Pi-starved liquid medium for another 3 days. In detached roots, GUS activity driven by the atMGD2 promoter was very low even under Pi-starved conditions, whereas detached shoots showed intense induction of atMGD2::GUS expression upon Pi starvation (Figure 3d). However, 10 μm IAA treatment strongly induced GUS activity in detached atMGD2::GUS roots under Pi-starved conditions. Moreover, we observed that 0.1 μm of IAA could induce GUS activity in detached roots although the activity level was low. Histochemical analysis also revealed that type B MGD::GUS expression was elevated by 10 μm IAA treatment in detached roots (Figure 3e). Intense GUS staining was observed in the roots even by treatment with 1.0 μm IAA. Treatment of detached roots with BA, ACC or methyl jasmonate (MeJA) did not induce type B MGD::GUS genes, indicating that the absence of auxin decreased the GUS activity of type B MGD::GUS fusions in Pi-starved detached roots. The analyses using type B MGD::GUS reporter constructs presented here provide strong evidence for auxin involvement in Pi-starvation induced expression of type B MGD genes.
Regulation of membrane lipid alteration by auxin and cytokinin under Pi-starved conditions
The GUS reporter analyses in Figures 2 and 3 demonstrate that auxin and cytokinin are antagonistically involved in the activation of type B MGD genes during Pi starvation. To investigate whether the hormonal regulation of type B MGD genes correlates with membrane lipid alteration during Pi starvation, we characterized lipids extracted from Pi-starved plants treated with BA or TIBA. Figure 4(a) shows the result of one-dimensional thin-layer chromatography (TLC). Both in the shoot and the root, MGDG content appeared constant under all conditions. On the other hand, DGDG content increased in the shoot of Pi-starved plants regardless of exogenous chemical treatment. Although DGDG accumulated substantially in Pi-starved control roots, DGDG levels in BA- or TIBA-treated roots were as low as in Pi-replete plants. The same effect was observed on the DGDG/phosphatidylcholine (PC) molar ratio (Figure 4b). In Pi-starved plants, the DGDG/PC ratio in the root was much higher than that in Pi-replete plants. By contrast, this ratio in the roots of BA- and TIBA-treated plants was as low as in Pi-replete plants. These data are consistent with the results from promoter–GUS analyses of type B MGD genes, showing that not only gene expression but membrane lipid alteration is regulated by auxin and cytokinin.
Reduced galactolipid synthesis in the auxin-response mutant axr4 aux1 during Pi starvation
The pharmacological analyses using auxin inhibitors in Figures 3 and 4 strongly suggest that auxin signaling is involved in galactolipid regulation during Pi starvation. To further support this result, we analyzed the auxin response mutant axr4-2 aux1-7, which is resistant to auxin and shows a great reduction in lateral root formation and gravitropism (Hobbie and Estelle, 1995; Yamamoto and Yamamoto, 1999). Under Pi-sufficient conditions, the transcript levels of MGD2 and MGD3 were very low in both wild-type and the axr4 aux1 double mutant (Figure 5a). Under Pi-starved conditions, however, a great accumulation of MGD2 and MGD3 mRNAs was detected in the wild-type as reported by Awai et al. (2001). Although an increase in MGD2/3 transcripts by Pi starvation was also observed in the axr4 aux1 mutant, the induction level was reduced compared with that in the wild-type, particularly in the root. Together with the results from promoter–GUS analyses, these data demonstrate the importance of auxin signaling for the activation of MGD2/3 expression during Pi starvation. We also investigated DGDG content in the wild-type and axr4 aux1 double mutant (Figure 5b). There was no difference in DGDG content between wild-type and axr4 aux1 under Pi-sufficient conditions. Under Pi-starved conditions, however, the amount of DGDG in the mutant was lower than that in the wild-type. In particular, axr4 aux1 contained 47% less DGDG than wild-type in the root during Pi starvation. As the axr4 aux1 mutant showed the same MGD1 transcript level as the wild-type, the difference between the mutant and the wild-type in DGDG content must be independent of MGD1 expression level. Analyses in axr4 aux1 double mutant demonstrate that in vivo auxin activities regulate the induction of MGD2/3 expression and galactolipid accumulation during Pi starvation.
Inhibition of Pi-starvation inducible changes in membrane lipids by phosphite
We demonstrated that the activation of type B MGD genes during Pi starvation requires auxin, particularly in the root, although exogenous auxin treatment itself could not induce the expression of these genes under Pi-sufficient conditions. These findings suggest that type B MGD gene activation upon Pi starvation requires a signal(s) other than auxin. To test this hypothesis, we used phosphite (Phi, HPO), an inactive analog of the Pi anion, which mimics Pi in signaling pathways, thereby suppressing various Pi-starvation inducible responses (Ticconi et al., 2001; Varadarajan et al., 2002). Because Phi is not an available source of phosphorus for plants, under Pi-deprived conditions plants still suffer Pi starvation even in the presence of Phi. Figure 6(a) shows the effect of Phi on atMGD2::GUS expression under Pi-starved conditions. GUS activity in transgenic plants grown on −Pi/+Phi medium was as low as that in those grown on Pi-replete medium, clearly demonstrating that Phi represses activation of atMGD2 via Pi starvation. Moreover, exogenous IAA treatment of plants grown under −Pi/+Phi conditions showed no detectable effect on GUS activity, both in shoots and roots, indicating that suppression of atMGD2::GUS by Phi treatment was independent of auxin. The same results were obtained by histochemical GUS staining in two type B MGD::GUS transformants (data not shown). Under −Pi/+Phi conditions, plants would be in a state of ‘pseudo Pi sufficiency’− even though starved for available Pi – because Phi mimics Pi signaling (Ticconi et al., 2001; Varadarajan et al., 2002). Therefore, these data suggest that type B MGD gene expression does not depend on the intracellular Pi availability but rather is regulated by a Pi-sensing signaling system.
To unravel whether the membrane lipid alterations are also controlled systematically by Pi signals, we analyzed membrane lipids using Arabidopsis supplemented with Phi. As shown in Figure 6(b), the DGDG content in Pi-starved plants increased threefold compared with that in Pi-replete plants. An increase in sulfoquinovosyl diacylglycerol (SQDG) was also detected in Pi-starved plants. The proportion of membrane phospholipids was drastically reduced in Pi-starved plants, but there was no difference in MGDG content between Pi-replete and Pi-starved plants. These results are consistent with a previous report (Essigmann et al., 1998). However, plants grown on −Pi/+Phi medium had neither increased DGDG or SQDG nor decreased phosphatidylethanolamine (PE) or PC. The application of Phi also affects the fatty acid composition of DGDG. Under Pi-deprived conditions in Arabidopsis, 16:0 fatty acids in DGDG increase, with a concomitant decrease in 18:3, as reported previously (Härtel and Benning, 2000; Kelly et al., 2003). However, there was no difference in the fatty acid composition of DGDG between plants grown under −Pi/+Phi conditions and those grown under Pi-replete conditions (data not shown). These data suggest that Pi-starvation induced membrane lipid alteration is regulated by signaling through a Pi-sensing mechanism. On the other hand, MGDG decreased by 30% in the plants grown under −Pi/+Phi conditions compared with those grown under Pi-replete and Pi-depleted conditions. Ticconi et al. (2001) reported that plants grown under −Pi/+Phi conditions had leaves with light green color and slightly decreased chlorophyll levels, suggesting the reduction of photosynthetic membranes in these plants. Thus, the level of MGDG might be decreased due to a reduction in the amount of photosynthetic membranes.
Under Pi-starved conditions, global changes in membrane lipids have been observed not only in Arabidopsis but also in oat and Acer pseudoplatanus suspension-cultured cells, suggesting that this adaptation mechanism is widespread in planta (Andersson et al., 2003; Jouhet et al., 2003). Although membrane lipid alteration during Pi starvation has been well documented by several studies (Andersson et al., 2003; Essigmann et al., 1998; Härtel et al., 2000; Jouhet et al., 2003, 2004), the regulatory mechanism controlling Pi-starvation induced membrane lipid alteration remains unknown. In Figure 6, we show that the Pi-starvation inducible promoter activity of type B MGD genes is inhibited by Phi application, indicating that activation of these genes by Pi starvation is directly triggered by signal transduction via a Pi-sensing mechanism, not by Pi-starvation induced damage to the plants. Moreover, Pi-starvation induced changes in membrane lipids, such as an increase in DGDG or SQDG and phospholipid degradation, were all repressed by Phi supplementation. This result indicates that Pi-sensing systems regulate overall membrane lipid changes upon Pi starvation. Kelly et al. (2003) demonstrated that DGD2 rather than DGD1 is responsible for the synthesis of DGDG species rich in C16 fatty acids during Pi starvation. In our study, Phi supplementation repressed an increase in the C16:C18 ratio in DGDG upon Pi starvation (data not shown). These data suggest that Pi signaling triggers DGDG synthesis mediated cooperatively by atMGD2/3 and DGD2 (Benning and Ohta, 2005). Moreover, we found that phytohormones are involved in regulating membrane lipid alterations during Pi starvation. The assays showing the inhibition of auxin signaling in Figures 3–5 demonstrate that auxin is indispensable for Pi-starvation inducible expression of type B MGD genes. The auxin requirement was also observed for galactolipid accumulation during Pi starvation. As BA treatment repressed both type B MGD expression and DGDG accumulation in the root during Pi starvation, Pi-starvation inducible changes in membrane lipids must be regulated by auxin/cytokinin signals.
Under Pi-starved conditions, endogenous Pi content decreases simultaneously in leaves and roots (Härtel et al., 2000), indicating Pi deficiency throughout the plant body. Our data show that Pi-starvation induced expression of type B MGD genes is strongly repressed by Phi application, both in the shoot and root (Figure 6a). These data suggest that Pi signaling controls the activation of type B MGD genes throughout the plant. However, Pi-starvation induced expression of type B MGD genes was detected particularly in several restricted regions, as shown in Figure 1. We also show that GUS activity in the root of type B MGD::GUS plants increases in parallel with massive root growth during Pi starvation (data not shown), in which auxin involvement has been proposed (López-Bucio et al., 2003; Nacry et al., 2005). Thus, it is likely that auxin regulates the tissue- and developmental stage-specific activation of type B MGD genes during Pi starvation, whereas a Pi-sensing mechanism appears to be affected by Pi status throughout the plant (Ticconi et al., 2001; Varadarajan et al., 2002).
A series of analyses using the type B MGD::GUS reporter system revealed that the promoter activity of type B MGD genes increased upon IAA treatment under Pi-starved conditions, whereas the promoters were not responsive to IAA under Pi-replete conditions (Figure 2). This result indicates that the auxin-responsiveness of type B MGD genes is enhanced during Pi starvation. López-Bucio et al. (2003) suggested that architectural changes of the root system induced by Pi starvation are related to an increase in auxin sensitivity. It is likely that such an increase in auxin sensitivity upon Pi starvation occurs in the promoters of type B MGD genes. Nacry et al. (2005) demonstrated that, during Pi starvation, IAA concentration increases in the whole primary root and in young lateral roots. In addition, an increase in DR5::GUS activity by Pi starvation was observed in emerged lateral root primordia (López-Bucio et al., 2005; Nacry et al., 2005), where intense type B MGD::GUS staining was detected during Pi starvation (Figure 1). These data suggest that the increase in auxin concentration also contributes to the activation of type B MGD gene expression under Pi-starved conditions.
In contrast to the positive effect of auxin on the promoter activity of type B MGD genes, Pi-starvation induced expression of type B MGD genes was strongly repressed by exogenous BA treatment in the root. A similar effect of cytokinin-mediated repression of Pi-starvation induced gene expression was reported in four Pi-starvation responsive genes, three of which were induced by exogenous auxin during Pi starvation (Martín et al., 2000). Moreover, CRE1, a receptor for cytokinin, is purportedly involved in the regulation of these Pi-starvation responsive genes (Franco-Zorrilla et al., 2002). This idea can be extended to membrane lipid alteration under Pi starvation. Pi starvation results in a reduction in cytokinin levels (Horgan and Wareing, 1980; Salama and Wareing, 1979). Furthermore, CRE1 transcriptional activity is repressed by Pi starvation (Franco-Zorrilla et al., 2002). These data suggest that reduced cytokinin signaling in vivo may be involved in the Pi-starvation response. We demonstrated that simultaneous treatment with IAA and BA decreased the repressive effect of BA on Pi-starvation induced expression of type B MGD genes (Figure 2), indicating the importance of the balance between auxin and cytokinin in this event. As changes in auxin level were also reported during Pi starvation (Nacry et al., 2005), the in vivo balance between auxin and cytokinin may control the promoter activity of type B MGD genes and changes in membrane lipids during Pi starvation.
Although widespread blue staining was detected in the root of Pi-starved type B MGD::GUS plants, high GUS activity was observed in the lateral root-emerging region, suggesting the involvement of type B MGD genes in lateral root growth during Pi starvation. High galactolipid-synthesizing activity as well as substantial DGDG accumulation has also been observed in Pi-starved roots (Härtel et al., 2000; Kobayashi et al., 2004). Moreover, the promoter activity of type B MGD genes in the root increased in parallel with massive root growth during Pi starvation (data not shown). These data suggest that supplementation of galactolipids in plant membranes would support root growth under Pi-starved conditions. As for aerial tissues, although their growth was impaired significantly by Pi starvation, intense expression of type B MGD genes and DGDG accumulation were observed (Figures 2, 5; Härtel et al., 2000), suggesting the importance of Pi-starvation induced galactolipid biosynthesis, even in shoots. Hydathodes have several pores composed of open-ended vessels that function to release water via guttation (Aloni et al., 2003). In addition, this organ is thought to be involved in the retrieval of solutes from the transpiration stream. Indeed, several nutrient transporters, including Pi transporters, are expressed in hydathodes, suggesting that these transporters function in the recovery of solutes released by guttation (Bürkle et al., 2003; Lagarde et al., 1996; Mudge et al., 2002; Shibagaki et al., 2002; Wang et al., 2004). Hydathode epithem cells have a multi-lobed shape that increases the surface area facing the extracellular space. Moreover, epithem cells have a considerable number of mitochondria (Galatis, 1988). Previous studies revealed that Pi starvation causes a significant portion of phospholipids to be replaced by DGDG, not only in plasma membranes but also in mitochondria (Andersson et al., 2003; Jouhet et al., 2004). Given that type B MGD gene promoters are highly activated in hydathode epithem cells under Pi-starved conditions, galactolipid-synthesizing activity might be up-regulated to maintain the characteristic cell structure via DGDG replacement under these conditions.
Our GUS reporter analyses indicate that type B MGD gene promoters are regulated variously by Pi signaling and phytohormones, suggesting transcription factor(s)-mediated regulation at an upstream site. At present, little is known about a consensus DNA sequence(s) associated with the Pi-starvation response. Recently, PHR1 was identified as a MYB transcription factor involved in Pi-starvation induced gene expression, and its consensus binding sequence was established as GNATATNC (Rubio et al., 2001). This motif has been widely identified in the promoters of Pi-starvation inducible genes, including those involved in glycolipid biosynthesis (Franco-Zorrilla et al., 2004). For type B MGD genes, the consensus sequence was found in the 5′ untranslated region (+1 and +52 bp from the transcription initiation site) in atMGD2, but not in atMGD3. As atMGD2 and atMGD3 had similar expression patterns in several experiments under Pi starvation, it is likely that Arabidopsis has another transcription factor(s) that commonly regulates these two type B MGD genes. A recent study on a small ubiquitin-like modifier E3 ligase, AtSIZ1, demonstrated that sumoylation is involved both negatively and positively in various Pi-deficiency responses, including the response mediated by PHR1 (Miura et al., 2005). Moreover, a microRNA miR399 was found to be involved in various Pi-starvation responses in Arabidopsis (Chiou et al., 2006; Fujii et al., 2005). It is possible that multiple regulation mechanisms are present in the Pi-starved response and concertedly involved in various morphological changes and gene expression. In this study, we clarified the regulation of Pi-starvation induced activation of type B MGD genes via Pi signaling and cross-talk of auxin and cytokinin. This regulation also affected membrane lipid alterations during Pi starvation. Further studies are required to identify and characterize transcription factors that control the integrated molecular mechanism of membrane lipid alterations during Pi starvation.
Arabidopsis thaliana (Columbia-0) transgenic lines carrying atMGD2::GUS and atMGD3::GUS have been described in Kobayashi et al. (2004). The axr4-2 aux1-7 double mutant is in the Columbia background. All plants used in this study were germinated and grown on solidified MS medium (Murashige and Skoog, 1962) containing 0.8% agar, 1% sucrose at 23°C under continuous white light unless otherwise indicated. For screening of transgenic plants, 50 μg/ml kanamycin was added to the MS medium.
Growth condition of transgenic plants and treatment with plant growth regulators
For histochemical GUS analyses of type B MGD::GUS transformants, plants were grown on MS medium for 7 days and then transferred to either Pi-replete (1.0 mm) or Pi-depleted (0 mm) medium prepared as described by Härtel et al. (2000). After growing on the Pi-controlled media for another 7 days, plants were used for in vivo GUS assays as described below. To investigate the effects of plant growth regulators on type B MGD gene expression, transgenic plants were grown on MS medium for 7 days. After kanamycin screening, plants were transferred to the Pi-controlled liquid media containing plant growth regulators and then incubated on a rotary shaker for another 7 days. For analyses in detached shoots and roots, plants were grown for 7 days on MS medium and then transferred to fresh medium without kanamycin for 8 days. Shoots and roots were detached and incubated separately in Pi-controlled liquid media with or without hormones for 3 days. In the Phi-supplemented experiment in Figure 6(a), transformants were grown on Pi-deficient medium containing 1 mm Phi (Wako Pure Chemical Industries, Osaka, Japan) for 7 days following growth on MS medium for 7 days. Phi stock solution was prepared according to the method described by Varadarajan et al. (2002). IAA and BA were purchased from Nacalai Tesque (Kyoto, Japan) and Sigma Chemical Co. (St Louis, MO, USA), respectively. ACC was from Wako Pure Chemical Industries, TIBA was from Aldrich Chemical Co. (Milwaukee, WI, USA), NPA was from AccuStandard (New Haven, CT, USA), and PCIB was from ICN Biomedicals (Irvine, CA, USA).
Histochemical GUS staining was carried out as previously described (Kobayashi et al., 2004). Samples were cleared using clearing solution (chloral hydrate, water, glycerol, 8:2:1 v/v). For quantitative GUS analyses, samples frozen in liquid nitrogen were homogenized with GUS extraction buffer (50 mm Pi buffer solution, pH 7.0, 10 mmβ-mercaptoethanol, 10 mm EDTA, 0.1% v/v Triton X-100, and 0.1% w/v sodium N-dodecanoylsalcosinate), and then centrifuged (25 000 g) for 5 min at 4°C to prepare the supernatant as the enzyme mixture. The total protein content of the enzyme mixture was quantified by the method described by Bensadoun and Weinstein (1976) with BSA as a standard. A volume of enzyme mixture corresponding to 10 μg of total protein was brought to 190 μl with GUS extraction buffer, and then the reaction was initiated by mixing the enzyme solution with 10 μl of 20 mm 4-methylumbelliferyl-β-d-glucronide. After incubation at 37°C for 10 min, the reaction was terminated by adding 800 μl 200 mm Na2CO3. The amount of 4-methylumbelliferyl (4-MU) produced in the reaction was determined by measuring 4-MU-specific fluorescence (365 nm excitation, 455 nm emission).
RNA gel blot analysis
Wild-type Arabidopsis (Columbia-0) and the axr4 aux1 mutant were grown on MS medium for 10 days followed by growth on Pi-controlled media for another 10 days. A 2 μg aliquot of total RNA was used for gel blot analysis in each sample. Total RNA isolation, RNA gel electrophoresis and transfer to nylon membranes were carried out according to the methods described by Taki et al. (2005). The membrane was hybridized to α-32P-dCTP-labeled MGD probes (Awai et al., 2001) or ACTIN8 probe (Taki et al., 2005). Hybridized RNA bands were exposed to Image Plate (Fuji Photofilm, Tokyo, Japan) and visualized by an Image Analyzer (Storm; Amersham Biosciences, Piscataway, NJ, USA).
Lipid and fatty acid analyses
For the mutant analysis, plants were prepared as described for the RNA gel blot analysis. For the Phi supplementation experiment, Arabidopsis were also prepared as described above, but grown with or without 1 mm Phi during Pi starvation. For BA (10 μm) or TIBA (50 μm) treatment, plants were grown on MS medium for 15 days before transfer to Pi-controlled media, and then subjected to Pi starvation for 10 days. Samples were collected and frozen immediately in liquid nitrogen, and lipids were then extracted by the method described by Bligh and Dyer (1959). Measurement of membrane lipid content was carried out according to the method described by Dörmann et al. (1995) and Yamaryo et al. (2003). In brief, lipid extracts were separated by silica-gel TLC using a solvent system of acetone:toluene:water (45:15:4 v/v/v). Lipids were visualized with 0.01% w/v primuline in 80% v/v acetone under UV light and isolated individually from the plates. Fatty acid methyl esters were prepared by incubating each lipid in 5% v/v HCl in methanol at 85°C for 2.5 h, and then quantified by gas chromatography (GLC) using myristic acid as an internal standard.
We thank Dr Koichiro Awai for helpful comments on this manuscript. We also thank the Arabidopsis Biological Resource Center (ABRC) for providing axr4-2 aux1-7 seeds. This work was supported by Grants-in-Aid for Scientific Research on Priority Areas (nos 15380049 and 17051009) from the Ministry of Education, Sports, Science and Culture in Japan. KK was supported by a research fellowship for young scientists from the Japan Society for the Promotion of Science. This paper is dedicated to the memory of Professor Ken-ichiro Takamiya, 1943–2005.