RCB-mediated chlorophagy caused by oversupply of nitrogen suppresses phosphate-starvation stress in plants

Abstract

Inorganic phosphate (Pi) and nitrogen (N) are essential nutrients for plant growth. We found that a five-fold oversupply of nitrate rescues Arabidopsis (Arabidopsis thaliana) plants from Pi-starvation stress. Analyses of transgenic plants that overexpressed GFP-AUTOPHAGY8 showed that an oversupply of nitrate induced autophagy flux under Pi-depleted conditions. Expression of DIN6 and DIN10, the carbon (C) starvation-responsive genes, was upregulated when nitrate was oversupplied under Pi starvation, which suggested that the plants recognized the oversupply of nitrate as C starvation stress because of the reduction in the C/N ratio. Indeed, formation of Rubisco-containing bodies (RCBs), which contain chloroplast stroma and are induced by C starvation, was enhanced when nitrate was oversupplied under Pi starvation. Moreover, autophagy-deficient mutants did not release Pi (unlike wild-type plants), exhibited no RCB accumulation inside vacuoles, and were hypersensitive to Pi starvation, indicating that RCB-mediated chlorophagy is involved in Pi starvation tolerance. Thus, our results showed that the Arabidopsis response to Pi starvation is closely linked with N and C availability and that autophagy is a key factor that controls plant growth under Pi starvation.

Disturbance of the carbon/nitrogen ratio induces partial chloroplast degradation via autophagy under phosphate starvation and rescues phosphate starvation stress.

Introduction

Phosphorous is one of the essential macronutrients for plant growth, because it is a critical component of many biomolecules and metabolites, such as phospholipids, nucleotides, and ATP. Plants can take up phosphorous as inorganic phosphate (Pi), but Pi is limited in soils. Metals form insoluble complexes with phosphate and microbial activity converts Pi into organic phosphate, which is not available to plants (Raghothama, 1999), either of which can lead to phosphate deficiency. Phosphate deficiency inhibits crop productivity and affects about 70% of arable land (Estelle and Somerville, 1987). Pi fertilizer is produced from phosphate rock, a non-renewable resource (Blanco, 2011). Plants have several responses to Pi starvation, including membrane lipid remodeling, organic acid exudation, and promotion of symbiotic arbuscular mycorrhizal fungi (Härtel and Benning, 2000; Dörmann and Benning, 2002; López-Arredondo et al., 2014).

Nitrogen (N) is also an essential macronutrient for plant growth and is used to produce many biomolecules and metabolites, such as nucleic acids, amino acids, proteins, and chlorophyll (Crawford and Forde, 2002; Peng et al., 2007). The most common N species absorbed and used by plants are nitrate (${\text{NO}}_3^{–}$) and ammonium (${\text{NH}}_4^{+}$). Many plants, including Arabidopsis (Arabidopsis thaliana), do not grow well with ${\text{NH}}_4^{+}$ as their sole source of N (Wilkinson and Crawford, 1993).

Autophagy is an intracellular degradation process for bulk protein and organelles. Plant autophagy has an important role in nutrient recycling under starvation (Yoshimoto et al., 2004; Izumi et al., 2010; Sankaranarayanan and Samuel, 2015; Shinozaki et al., 2020). The substrate of autophagy can be either specifically selected (selective autophagy) or randomly acquired (non-selective autophagy). The chloroplast is one example of a target for selective autophagy (Izumi et al., 2017; Nakamura et al., 2018). Photodamage, such as that caused by ultraviolet B and strong light, induces the degradation of the whole chloroplast via activation of autophagy (Izumi et al., 2017; Nakamura et al., 2018). In contrast, partial chloroplast degradation can also occur by autophagy. For example, carbon (C) starvation or senescence induces Rubisco-containing body (RCB) formation (Chiba et al., 2003; Ishida et al., 2008). RCB is a type of autophagic body that contains chloroplast stroma (Chiba et al., 2003; Ishida et al., 2008).

The mechanism of the response to N starvation is related to that of Pi starvation (Kant et al., 2011; Park et al., 2014) and the mechanism of the response to Pi starvation is related to that of N starvation (Yoshitake et al., 2017). Moreover, an excess supply of N induces expression of the gene encoding the Pi transporter in the root under Pi-depleted conditions (Cerutti and Delatorre, 2013; Maeda et al., 2018). Although these studies suggest that the mechanism of the response to Pi starvation and that of N starvation are closely linked to one other, the regulatory mechanism responsible for this connection is not known.

In this study, we show that an oversupply of nitrate, consisting of a five-fold higher concentration of nitrate in the growth medium than is present in normal growth medium, can rescue plants from Pi starvation. Our results suggested that an increase in N availability reduced the C/N ratio and induced the formation of autophagic bodies under Pi-depleted conditions. Furthermore, the induction of autophagy by an excess supply of nitrate increased the Pi content in the cell, and plants under these conditions showed a healthy phenotype.

Results

An oversupply of nitrate saves plants from Pi-starvation stress

We usually grow Arabidopsis plants on normal growth medium (1-mM Pi and 4.5-mM N; Pi1/N4.5; Figure 1, A). To clarify the growth effect of the N and Pi concentrations in the growth medium, we grew plants on three additional media: Pi-starved (0-mM Pi and 4.5-mM N; Pi0/N4.5; Figure 1, B), nitrate-oversupplied (1-mM Pi and 22.5-mM N; Pi1/N22.5; Figure 1, C), and Pi-starved and nitrate-oversupplied (0-mM Pi and 22.5-mM N; Pi0/N22.5) media (Figure 1, D). To exclude the possibility that an oversupply of nitrate caused contamination of Pi in the medium, we analyzed the Pi content in these media and confirmed that Pi levels in the nitrate-oversupplied media were low enough so as not to block the Pi-starvation response (Shimojima et al., 2015; Supplemental Figure S1). When wild-type plants were grown on Pi-starved medium (Pi0/N4.5), they were small when compared with the plants on Pi1/N4.5 medium (Figure 1, A, B, and E). When grown on Pi1/N22.5 medium, plants grew larger than those on Pi1/N4.5 medium (Figure 1, A, C, and E). However, unexpectedly, when plants were grown on Pi0/N22.5 medium, they showed a healthier phenotype and their fresh weight was greater than those grown on Pi0/N4.5 medium (Figure 1, B, D, and E). As anthocyanin accumulation is often observed in seedlings grown under Pi starvation (Ticconi and Abel, 2004), we also compared the anthocyanin content of these plants (Figure 1, F). The anthocyanin content in the shoots of plants grown on Pi0/N4.5 medium was about five-fold higher than that on Pi1/N4.5 medium (Figure 1, F). However, a five-fold oversupply of nitrate repressed the anthocyanin accumulation under Pi starvation (Figure 1, F). Given that the anthocyanin content in the shoots on Pi0/N22.5 medium was less than that on Pi0/N4.5 medium, an oversupply of nitrate most likely suppressed the accumulation of anthocyanin under Pi starvation (Figure 1, F).

Figure 1.

Effect of oversupply of nitrate on A. thaliana under Pi starvation. (A–D) Growth phenotypes of 20-d-old A. thaliana seedlings grown on (A) Pi1/N4.5, (B) Pi0/N4.5, (C) Pi1/N22.5, and (D) Pi0/N22.5 media. Bars = 1 cm. (E) Shoot fresh weight of A. thaliana seedlings grown as in (A–D). (F) Anthocyanin content of shoots from A. thaliana seedlings grown as in (A–D). (G–I) Expression of (G) PHT1;1, (H) At4, and (I) MGD3 in the shoots of A. thaliana seedlings grown as in (A–D). (J) Pi contents of shoots from A. thaliana seedlings grown as in (A–D). Data represent the mean ± sd from three independent experiments. Different letters indicate significant differences at P < 0.05 (Tukey’s test).

We also analyzed the expression of phosphate starvation response (PSR) genes (Figure 1, G–I). Although the expression levels of Phosphate transporter 1;1 (PHT1;1), Arabidopsis mt4 (At4), and Monogalactosyldiacylglycerol synthase 3 (MGD3) were increased under Pi starvation as previously reported (Burleigh and Harrison, 1999; Mudge et al., 2002; Kobayashi et al., 2004), expression of these genes was suppressed by five-fold nitrate (Figure 1, G–I). It should be noted that five-fold nitrate itself had no effect on the expression of these genes (Figure 1, G–I). As the suppression of gene expression could be due to an increase in Pi availability in the seedlings, we measured the Pi content in their shoots (Figure 1, J). The Pi content in plants grown on Pi0/N4.5 medium was markedly lower than that on Pi1/N4.5 medium as previously reported (Narise et al., 2010; Figure 1, J). However, the Pi content in the shoots of plants grown on Pi0/N22.5 medium was significantly higher than that on Pi0/N4.5 medium (Figure 1, J). It is noteworthy that, because Pi0/N22.5 medium is deficient in Pi (Supplemental Figure S1), the enhancement of Pi uptake from roots is certainly not responsible for the increase in the Pi content of the seedlings. Thus, these results suggested that the Pi content in the seedlings on Pi0/N22.5 medium was elevated by enhanced remobilization of Pi in the cell, followed by the suppression of the Pi-starvation response.

An oversupply of nitrate inhibits lipid remodeling

Membrane lipid remodeling is one of the systems induced by Pi starvation for Pi remobilization. Under the Pi0/N4.5 condition, phospholipids that contain Pi, such as phosphatidylcholine (PC) and phosphatidylethanolamine (PE), are replaced by the galactolipid digalactosyldiacylglycerol (DGDG), which does not contain Pi (Härtel and Benning, 2000; Dörmann and Benning, 2002). In plastids, phosphatidylglycerol (PG) is also replaced by sulfoquinovosyldiacylglycerol (SQDG; Härtel and Benning, 2000; Dörmann and Benning, 2002; Yu et al., 2002). Phospholipid degradation during this lipid remodeling supplies Pi to the cell and enables plants to grow better even under Pi starvation. Thus, we analyzed the membrane glycerolipid content in the shoots (Figure 2). Regarding the membrane lipid composition, we confirmed that the amounts of DGDG and SQDG were increased and the amounts of PC, PE, and PG were decreased on Pi0/N4.5 medium (Figure 2, A). Neither galactolipid accumulation nor phospholipid degradation was enhanced by five-fold nitrate under the 0-mM Pi condition (Figure 2, A). On a fresh-weight basis, although the amounts of all lipids on Pi1/N22.5 medium were higher than on the other media, an increase in DGDG and a decrease in phospholipids were not observed on Pi0/N22.5 medium (Figure 2, B). These results clearly indicated that there was no correlation between membrane lipid remodeling and an increase in Pi content in plants grown on Pi0/N22.5 medium.

Figure 2.

Membrane lipids in the shoots. (A, B) Membrane lipid (A) composition and (B) content of 20-d-old A. thaliana seedlings grown under Pi1/N4.5, Pi0/N4.5, Pi1/N22.5, and Pi0/N22.5 conditions. Data represent the mean ± sd from three independent experiments. Different letters indicate significant differences at P < 0.05 (Tukey’s test). PI, phosphatidylinositol.

Oversupply of nitrate induces autophagy flux

To confirm whether vacuolar degradation, especially autophagy, contributed to plant growth under the Pi0/N22.5 condition, we observed autophagic bodies using autophagy-monitoring GFP-AUTOPHAGY8a (GFP-ATG8a) plants. Concanamycin A (ConcA), a vacuolar H⁺-ATPase inhibitor, blocks vacuolar hydrolysis, leading to the accumulation of autophagic bodies inside vacuolar lumens (Tamura et al., 2003; Yoshimoto et al., 2004). Thus, ConcA treatment allowed us to analyze autophagy flux (Shin et al., 2014). We analyzed the number of autophagic bodies labeled with GFP-ATG8a in the vacuole by laser-scanning confocal microscopy (LSCM) after a 20-h incubation with or without ConcA (Figure 3 and Supplemental Figure S2). In mesophyll cells from plants grown on Pi0/N4.5 medium, the number of autophagic bodies was higher than that on Pi1/N4.5 and Pi1/N22.5 media in the presence of ConcA (Figure 3, A–D). These results suggested that Pi starvation induces autophagy flux. Moreover, the number of autophagic bodies in mesophyll cells from plants grown on Pi0/N22.5 medium was more than that on Pi0/N4.5 medium (Figure 3, A, C, and E). Surprisingly, in mesophyll cells grown on Pi0/N22.5 medium, autophagic bodies were observed even without ConcA because autophagy flux was extremely induced under Pi0/N22.5 conditions (Figure 3, A and Supplemental Figure S2). These results indicated that autophagy flux occurs at a high level in the presence of an oversupply of nitrate under Pi-depleted conditions. Additionally, we observed autophagosomes using GFP-ATG8a plants (ProUBQ10::GFP-ATG8a; Supplemental Figure S3). The number of autophagosomes grown on Pi0/N4.5 was more than that on Pi1/N4.5. This result indicated that Pi starvation induces autophagy flux. Furthermore, the number of autophagosomes in mesophyll cells under Pi0/N22.5 conditions was more than that under Pi0/N4.5 conditions. These results supported our hypothesis that the vacuole degradation system contributed to plant growth under the Pi0/N22.5 condition. However, this enhanced induction of autophagy flux could be caused by nitrate toxicity, as general gardening knowledge indicates that high concentrations of nitrate can to some extent disrupt plant growth. To exclude this possibility, we carried out a GFP-ATG8 processing assay (Chung et al., 2010; Klionsky et al., 2012), which is used to measure autophagy flux. The free GFP/GFP-ATG8a ratio under the Pi1/N90 condition was comparable to that under the Pi1/N4.5 condition (Supplemental Figure S4), confirming that the exogenous nitrate level alone did not affect autophagy flux under these conditions.

Figure 3.

Effect of Pi starvation and oversupply of nitrate on the formation of autophagic bodies. (A) The number of autophagic bodies in the field of view (134.95 × 134.95 µm) based on LSCM visualization of GFP-ATG8a fusion protein in living leaf cells of A. thaliana grown on the indicated media. The leaves were incubated for 20 h in MES-NaOH (pH 5.5) with cocanamycin A (ConcA, white) and without ConcA (DMSO, gray). Data represent the mean ± sd from nine independent experiments. Different letters indicate significant differences at P < 0.05 (Tukey’s test). (B–E) LSCM visualization of GFP-ATG8a fusion protein in living leaf cells of A. thaliana grown on (B) Pi1/N4.5, (C) Pi0/N4.5, (D) Pi1/N22.5, and (E) Pi0/N22.5 media. The leaves were observed after incubation for 20 h in MES-NaOH (pH 5.5) with 1-µM ConcA. Bars = 10 µm. Arrowheads: autophagic bodies.

Oversupply of nitrate did not enhance endoplasmic reticulum degradation by autophagy under Pi starvation

Pi starvation induces autophagy mediated by endoplasmic reticulum (ER) stress in Arabidopsis roots, which is a consequence of Pi sensing rather than a means of Pi remobilization (Naumann et al., 2019). To clarify whether the induced autophagy observed in the mesophyll cells in plants grown under our experimental conditions is ER stress-mediated autophagy, we analyzed the expression of the ER stress response genes CNX1 and BiP3 (Figure 4, A and B). The expression of ER stress response genes under the Pi0/N4.5 condition was higher than that under the Pi1/N4.5 condition, suggesting that Pi starvation induced ER stress (Figure 4, A and B). An oversupply of nitrate alone (Pi1/N22.5) did not significantly induce the expression of these genes when compared with Pi1/N4.5. Moreover, an oversupply of nitrate under Pi starvation (Pi0/N22.5) did significantly induce the expression of BiP3 when compared with that in cells grown on Pi0/N4.5 (Figure 4, A and B). These results suggested that an oversupply of nitrate alone did not induce ER stress, but an oversupply of nitrate under Pi starvation markedly induced ER stress. Next, we observed ER behaviors using transgenic plants expressing ER-targeted GFP protein (ER-gk; Nelson et al., 2007) with or without ConcA to observe ER degradation by autophagy (Figure 4, C–G and Supplemental Figure S5). When leaves were treated with ConcA, the number of GFP vesicles in the vacuole, which correspond to ER in autophagic bodies, of plants grown on Pi0/N4.5 medium was more than that on Pi1/N4.5 (Figure 4, C). Compared with the number of GFP vesicles in plants grown on Pi1/N4.5, an oversupply of nitrate alone (Pi1/N22.5) did not increase the number of GFP vesicles (Figure 4, C). However, the number of GFP vesicles when nitrate was oversupplied under Pi starvation (Pi0/N22.5) was similar to that on Pi1/N4.5 (Figure 4, C). These results also suggested that an oversupply of nitrate did not enhance ER stress-mediated autophagy, whereas Pi starvation induced autophagic ER degradation in leaf mesophyll cells as observed in root cells previously (Naumann et al., 2019).

Figure 4.

Effect of Pi starvation and oversupply of nitrate on ER degradation. (A, B) Expression of (A) CNX1 and (B) BiP3 in the shoots of A. thaliana grown on the indicated media. Data represent the mean ± sd from three independent experiments. (C) The number of GFP vesicles in the field of view (134.95 × 134.95 µm) based on LSCM visualization of ER-targeted GFP fusion protein in living leaf cells of A. thaliana grown on the indicated media. The leaves were incubated for 20 h in MES-NaOH (pH 5.5) with ConcA (ConcA, white) and without ConcA (DMSO, black). Data represent the mean ± sd from nine independent experiments. (D–G) LSCM visualization of ER-targeted GFP fusion protein in living leaf cells of A. thaliana grown on (D) Pi1/N4.5, (E) Pi0/N4.5, (F) Pi1/N22.5, and (G) Pi0/N22.5 media. The leaves were incubated for 20 h in MES-NaOH (pH 5.5) with ConcA. Bars = 10 µm. Arrowheads: GFP vesicles. Different letters indicate significant differences at P < 0.05 (Tukey’s test).

Oversupply of nitrate disrupts the C/N balance and enhances RCB-mediated autophagy

The ratio of C to N in the growth medium is important for the regulation of plant growth and is referred to as the “C/N balance” (Martin et al., 2002; Aoyama et al., 2014). Under elevated C conditions with limited N availability, plant growth is remarkably retarded (Aoyama et al., 2014). Moreover, expression of several genes induced by Pi starvation is also induced by sucrose supplementation (Lejay et al., 2003). These studies suggested that there are interactions between Pi-related responses and C-related responses in plants. Thus, we analyzed the effect of the sucrose concentration in the growth medium (Figure 5). Wild-type plants were grown on Pi1/N4.5 medium, Pi0/N4.5 medium, Pi1/N22.5 medium, and Pi0/N22.5 medium containing 1% (w/v) sucrose (the typical amount) or 2% (w/v) sucrose for 10 d after an initial 10-d growth period on Murashige and Skoog (MS) medium (with 1% w/v sucrose; Figure 5, A–H). Although the fresh weight of plants grown under the 2% (w/v) sucrose condition was similar to that of plants grown under the 1% (w/v) sucrose condition (Figure 5, I), the anthocyanin content increased in response to higher sucrose levels both on Pi0/N4.5 medium and Pi0/N22.5 medium (Figure 5, J). Moreover, exogenously supplied sucrose induced the expression of At4 (Figure 5, K). These results indicated that an exogenous supply of sucrose resulted in a higher C/N ratio and enhanced the response to Pi starvation (Figure 6).

Figure 5.

Analysis of the effect of the sucrose concentration in the growth medium on A. thaliana plants. (A–H) Growth phenotypes of 20-d-old A. thaliana seedlings grown on (A, B) Pi1/N4.5, (C, D) Pi0/N4.5, (E, F) Pi1/N22.5, and (G, H) Pi0/N22.5 media with (A, C, E, and G) 1% (w/v) sucrose or (B, D, F, and H) 2% (w/v) sucrose. Bars = 1 cm. (I) Shoot fresh weight of A. thaliana seedlings grown as in (A–H). (J) Anthocyanin content of shoots from A. thaliana seedlings grown as in (A–H). (K) Expression of At4 in the shoots of A. thaliana seedlings grown as in (A–H). Data represent the mean ± sd from three independent experiments. Different letters indicate significant differences at P < 0.05 (Tukey’s test).

Figure 6.

A proposed model for the relationship between C/N balance and suppression of Pi starvation response. An oversupply of nitrate reduces the C/N ratio, which enhances autophagy. Autophagy induction elevates the intracellular Pi concentration and supports plant growth under Pi starvation. An exogenous supply of sucrose increases the C/N ratio, and thus autophagy is not induced and the response to Pi starvation is enhanced.

An oversupply of nitrate might decrease the C/N ratio, followed by induction of the C-starvation response. To clarify this effect, we analyzed the expression of Dark induced 6 (DIN6) and DIN10, which are induced by C starvation in plant cells (Fujiki et al., 2001; Figure 7, A and B). Although Pi availability had only a small effect on the expression of DIN6 and DIN10, their expression was significantly higher when nitrate was oversupplied (Figure 7, A and B). This suggested that the C-starvation response was induced when the N concentration in the growth medium was increased, even under conditions in which C was sufficiently supplied.

Figure 7.

Effect of oversupply of nitrate on RCB formation under Pi-depleted conditions. (A, B) Expression of (A) DIN6 and (B) DIN10 in the shoots of A. thaliana plants. Data represent the mean ± sd from three independent experiments. Different letters indicate significant differences at P < 0.05 (Tukey’s test). (C) The number of RCBs in the field of view (134.95 × 134.95 µm) based on LSCM visualization of CT-GFP fusion protein in living leaf cells of A. thaliana grown on the indicated media. The leaves were incubated for 20 h in MES-NaOH (pH 5.5) with ConcA (ConcA, white) and without ConcA (DMSO, gray). Data represent the mean ± sd from nine independent experiments. (D–G) LSCM visualization of CT-GFP fusion protein in living leaf cells of A. thaliana grown on (D) Pi1/N4.5, (E) Pi0/N4.5, (F) Pi1/N22.5, and (G) Pi0/N22.5 media. The leaves (the first or second true leaves) were incubated for 20 h in MES-NaOH (pH 5.5) with ConcA. GFP fluorescence (green) and Chlorophyll a autofluorescence (magenta) are shown. Bars = 10 µm. Arrowheads: RCBs. Different letters indicate significant differences at P < 0.05 (Tukey’s test).

The RCB pathway is induced by C starvation (Ishida et al., 2008). To clarify whether C starvation induced by an oversupply of nitrate activates the RCB system, we observed RCB behaviors using transgenic plants expressing GFP targeted to the chloroplast stroma (CT-GFP; Pro35S::CT-GFP) and mRFP targeted to the chloroplast stroma (RBCS2B-mRFP; ProRBCS2B::RBCS2B-mRFP) after a 20-h incubation with or without ConcA (Figure 7, C–F and Supplemental Figs. S6–S8). In leaves from plants grown on Pi1/N4.5 and Pi1/N22.5 medium, CT-GFP and RBCS2B-mRFP fluorescence overlapped with chlorophyll fluorescence, indicating that the RCB pathway was not induced under these conditions (Figure 7, C, D, and F and Supplemental Figure S6, A and C). However, the number of GFP vesicles under the Pi0/N4.5 condition was slightly more than that under the Pi1/N4.5 condition (Figure 7, C–F). In plants grown on Pi0/N22.5 medium, we observed many GFP vesicles, which were not observed in CT-GFP/atg5-1, suggesting that the GFP vesicles were RCBs (Figure 7, C and F and Supplemental Figure S9). In contrast, without ConcA, GFP vesicles in CT-GFP plants and mRFP vesicles in RBCS2B-mRFP plants were rarely observed (Supplemental Figures S7 and S8). Thus, Pi starvation alone does not fully activate the RCB pathway, but the combination of a low C/N ratio and Pi starvation massively induces RCB-mediated chlorophagy.

Pi is not released into the cell in autophagy mutants atg5-1 and atg7-2

Next, we analyzed the autophagy-deficient mutants atg5-1 and atg7-2 to determine whether autophagy contributes to an increase in the Pi content in the seedlings and to the suppression of the Pi-starvation response observed for plants grown on Pi0/N22.5 medium. The autophagy mutants showed a severe phenotype when grown on Pi0/N22.5 medium (Figure 8, E, F, K, and L). The shoot fresh weight of autophagy mutant plants on Pi0/N22.5 medium was comparable to that on Pi0/N4.5 medium (Figure 8, M). These results suggested that autophagy is crucial for this mechanism. We also analyzed the Pi content in the shoots (Figure 8, N). Among autophagy mutant plants, the Pi content was similar for plants grown on Pi0/N22.5 or Pi0/N4.5 medium. These results clearly indicated that under Pi starvation, an oversupply of N enhanced autophagy, which increased the Pi content in the cell and suppressed the Pi starvation response.

Figure 8.

Effect of oversupply of nitrate on atg mutants under Pi-depleted conditions. (A–L) Growth phenotype of 20-d-old (A, D, G, and J) wild-type, (B, E, H, and K) atg5-1, and (C, F, I, and L) atg7-2 A. thaliana seedlings grown on (A–C) Pi1/N4.5, (D–F) Pi0/N4.5, (G–I) Pi1/N22.5, and (J–L) Pi0/N22.5 media. Bars = 1 cm. (M) Shoot fresh weight of plants grown as in (A–L). (N) Pi content of shoots from plants grown as in (A–L). Data represent the mean ± sd from three independent experiments. Different letters indicate a significant difference at P < 0.05 (Tukey’s test).

Discussion

It is widely supposed that autophagy is induced by nutrient starvation. However, we found that autophagy was induced by an oversupply of nitrate under Pi starvation (22.5-mM N in 0-mM Pi medium). It has been suggested that the C/N balance is crucial for plant growth (Martin et al., 2002; Aoyama et al., 2014). Given that C starvation induces autophagy (Izumi et al., 2010), an excess of N might reduce the C/N ratio in the seedlings and induce the C-starvation response (Figure 7, A and B). RCB-mediated chlorophagy, which was induced by a low C/N ratio, released Pi into the cell and suppressed the Pi starvation response under the Pi0/N22.5 condition (Figure 8). Indeed, when the C/N ratio was elevated by additional sucrose in the Pi0/N22.5 medium, in which autophagy was probably suppressed, the expression of At4, one of the PSR genes, was upregulated (Figure 5, K). Thus, we concluded that the Pi starvation response was not always correlated with the concentration of Pi alone in the growth medium because the efficiency of intracellular Pi recycling via lipid remodeling and/or autophagy is affected by the C and N concentrations in the growth medium.

Our results showed that RCB-mediated chlorophagy was not induced under Pi1/N22.5 conditions even though the expression of DIN6 and DIN10 was enhanced (Figure 7). DCMU, an inhibitor of photosynthetic electron transport, was previously shown to induce RCB-mediated chlorophagy (Izumi et al., 2010). Although the authors proposed that C starvation induces RCB-mediated chlorophagy, this result could also suggest that photosynthetic electron transport may have some function to suppress RCB-mediated chlorophagy. In our experimental conditions, we measured the number of RCBs in Pi1/N22.5 plants under light conditions, and thus photosynthetic electron transport was active. Therefore, RCB-mediated chlorophagy could not be induced even under Pi1/N22.5 conditions, under which C starvation was mimicked (Figure 7).

Pi starvation induces phospholipid degradation and galactolipid synthesis, and thus the increased levels of DGDG and SQDG substitute for the decreased amounts of PC and PG, respectively (Härtel and Benning, 2000; Dörmann and Benning, 2002; Yu et al., 2002). However, in the shoots grown on Pi0/N22.5 medium, the ratio of DGDG was decreased, but the ratio of PG was increased in total membrane lipids when compared with shoots grown on Pi1/N4.5 medium (Figure 2, A). Although an increase in the PG content was not observed on a fresh-weight basis, the DGDG content was lower than that under Pi1/N4.5 (Figure 2, B). Under the Pi0/N22.5 condition, the plants presumably sensed C starvation (Figure 7, A and B). Regarding the relationship between membrane lipid composition and C supply, a previous study showed that sucrose supplementation enhances the expression of MGD3, which is followed by a significant increase in DGDG content in the microsomal lipids of plants (Murakawa et al. 2014). Although the expression of MGD3 under the Pi0/N22.5 condition was not significantly lower than that under the Pi1/N4.5 condition (Figure 1, I), a decrease in DGDG might be caused by the slight downregulation of MGD3. From a physiological analysis using transgenic Arabidopsis plants overexpressing MGD3, it was suggested that an increase in DGDG is correlated with the growth enhancement of plants grown under sucrose supplementation (Murakawa et al., 2014). Thus, DGDG is likely to have an essential role other than the substitution of PC under Pi starvation, and the C/N balance might be one of the factors regulating the expression of MGD3 and the DGDG content.

Induction of autophagy during Pi starvation has been reported by several groups in algae, tobacco BY-2 cells, and Arabidopsis roots, although the speed of induction was reduced relative to that under N starvation in some organisms (Tasaki et al., 2014; Shemi et al., 2016; Couso et al., 2018; Naumann et al., 2019). It was proposed that Pi starvation in Arabidopsis roots induces ER stress-dependent autophagy, which is a consequence of Pi sensing rather than a means of Pi remobilization (Naumann et al., 2019). The ER is degraded by autophagy during ER stress in Arabidopsis (Liu et al., 2012). To determine whether ER stress-dependent autophagy is involved in the autophagy observed in our experimental conditions, we analyzed the membrane lipid composition. If ER stress-dependent autophagy was induced, the amount of PC, the most abundant lipid in the ER, should have been decreased. However, the amount of PC under the Pi0/N22.5 condition was higher than that under Pi0/N4.5 and was comparable to that under Pi1/N4.5 (Figure 2, A and B). Additionally, the number of ER-derived vesicles in the vacuolar lumen in cells grown on Pi0/N22.5 medium was similar to that in cells grown on Pi0/N4.5 medium (Figure 4, C), suggesting that ER stress-mediated autophagy was not enhanced by an oversupply of nitrate under Pi starvation. Thus, we concluded that RCB-mediated chlorophagy, which was induced by a reduced C/N ratio, is the predominant autophagy when nitrate was oversupplied under Pi starvation and is responsible for the suppression of the Pi starvation phenotype.

Another cargo for the autophagy process in these cells was thought to be chloroplast contents, as our data indicated that mesophyll cells recognize the Pi0/N22.5 condition as C starvation, which induces chlorophagy (Figure 7, A and B). The degradation of chloroplasts via autophagy, namely chlorophagy, consists of two pathways: the whole chloroplast pathway and the RCB pathway (Otegui, 2018; Izumi et al., 2019; Zhuang and Jiang, 2019). Whole-organelle chlorophagy degrades entire chloroplasts; in contrast, RCB-mediated chlorophagy degrades a small portion of stroma proteins but not thylakoid membrane proteins (Minamikawa et al., 2001; Chiba et al., 2003). If whole-organelle chlorophagy is induced under the Pi0/N22.5 condition, the monogalactosyldiacylglycerol (MGDG) content should be reduced, because whole-organelle chlorophagy degrades chloroplast membranes including thylakoid membranes in which the most abundant lipid is MGDG (Block et al., 1983). Our results did not, however, show a marked decrease in the MGDG content under the Pi0/N22.5 condition, suggesting that the autophagy observed under these conditions was not whole-organelle chlorophagy. RCB formation is induced by C starvation (Izumi et al., 2010). This observation is consistent with our hypothesis that RCB-mediated chlorophagy was enhanced under the Pi0/N22.5 condition. Indeed, our results with transgenic plants expressing CT-GFP and RBCS2B-mRFP showed that activation of autophagy via the RCB pathway was the major autophagy pathway involved in the increased amount of Pi under the Pi0/N22.5 condition. Although this increase in the amount of Pi might be derived from degradation of plastid nucleoids, further experiments are required to prove this hypothesis.

Ribophagy, the degradation of mature ribosomes, is another possible pathway for the autophagy process under the Pi0/N22.5 condition (Kraft et al., 2008). In Zea mays, granules that contain a high amount of RNA are incorporated into young vacuoles in the primary root meristem (Niki et al., 2014). In Arabidopsis, mutants lacking RNS2, one of the genes that encodes RNase T2, display an increased formation of autophagosomes containing ribosomes and RNA (Floyd et al., 2015). These results suggest that an oversupply of N might also induce ribophagy under Pi-depleted conditions, which should be analyzed in future work.

Materials and methods

Plasmid construction and plant transformation

For transgenic plants expressing GFP-ATG8a driven by the ACTIN2 promoter, the ACTIN2 (At3g18780) promoter and the ATG8a coding sequence (At4g21980.1) were amplified from Arabidopsis (A. thaliana) genomic DNA and cDNA, respectively, by PCR using forward and reverse primers (Supplemental Table S1). Gel-purified PCR products were cloned into the pEZS-CL vector (S. Cutler and D. Ehrhardt, Carnegie Institution for Science, Stanford, CA, USA), from which the endogenous promoter had been removed by Sac I and Nco I digestion. The ProACT::GFP::ATG8a fragment was digested with Not I, followed by blunt-end treatment. The fragment was ligated into the pCAMBIA vector (Cambia, Canberra, Australia) digested with Sma I. The pCAMBIA vector was introduced into Arabidopsis by the floral dip method (Clough and Bent, 1998).

For transgenic plants expressing GFP-ATG8a driven by the UBIQUITIN10 promoter, the ATG8a coding sequence was amplified from Arabidopsis cDNA by PCR using forward and reverse primers containing an attB1 and attB2 site, respectively, at the 5′-end. These primers are listed in Supplemental Table S1. Gel-purified PCR products were cloned into pDONR221 vector (Thermo Fisher Scientific) using BP-clonase II (Thermo Fisher Scientific) according to the manufacturer’s instructions. Recombinant entry clones were amplified in Escherichia coli DH5α cells (TOYOBO) and were verified by restriction enzyme digestion and sequencing. The ATG8a coding region was then transferred to pUBN-GFP-Dest vector in a LR recombination reaction and then introduced into Arabidopsis by the floral dip method (Clough and Bent, 1998; Grefen et al., 2010).

Plant material and growth conditions

We used Columbia-0 as wild-type Arabidopsis. The seeds of T-DNA knockout mutants (atg5-1 [SAIL_129B07] and atg7-2 [GABI_655B06]) and transgenic plants (Pro35S::GFP-ATG8a, ER-gk, Pro35S::CT-GFP, and ProRBCS::RBCS-mRFP) were previously described (Köhler et al., 1997; Thompson et al., 2005; Holzinger et al., 2007; Nelson et al., 2007; Hofius et al., 2009; Ono et al., 2013). The atg mutant containing Pro35S::CT-GFP was obtained previously (Izumi et al., 2017). Seeds were surface sterilized and incubated at 4°C in the dark on MS medium containing 0.8% (w/v) INAgar (Ina Food Industry Co., Ltd.) supplemented with 1% or 2% (w/v) sucrose (Murashige and Skoog, 1962). Then, plants were incubated at 22°C under continuous white light (40–50 µmol m⁻² s⁻¹) for all growth conditions. After plants were grown on MS plates for 10 d, they were transferred and grown for another 10 d on Pi 1 mM N 4.5 mM, Pi 0 mM N 4.5 mM, Pi 1 mM N 22.5 mM, Pi 0 mM N 22.5 mM, Pi 1 mM N 0 mM, or Pi 1 mM N 90 mM medium supplemented with 1% (w/v) sucrose and 0.8% (w/v) INAgar. For the Pi 1 mM N 4.5 mM and Pi 0 mM N 4.5 mM media, we adjusted the N concentration with ${\text{NO}}_3^{–}$ as described (Estelle and Somerville, 1987; Gaude et al., 2007).

Measurement of anthocyanin content

The anthocyanin content of 20-d-old seedlings was measured as described previously (Ticconi et al., 2001). Briefly, seedlings were homogenized with extraction buffer (18:1:81 [v/v/v] propanol/HCl/water) and boiled for 3 min. Next, samples were centrifuged for 10 min (25,000 × g). The absorbance corresponding to the amount of anthocyanins was determined by A₅₃₅ – A₆₅₀ and was reported relative to the fresh weight of the sample.

RNA extraction and RT-qPCR

Total RNA was isolated from 20-d-old seedlings and cDNA was synthesized as described (Yoshitake et al., 2017). UBQ10 mRNA was used as an internal control, because UBQ10 mRNA levels in leaves are not affected by the Pi or N concentration (Sun and Callis, 1997; Bari, 2006). Gene-specific primers are listed in Supplemental Table S1.

Measurement of Pi content

The Pi content of 20-d-old seedlings was measured as described previously (Chiou et al., 2006). Briefly, seedlings were homogenized with extraction buffer (10-mM Tris, 1.0-mM EDTA, 100-mM NaCl, 1.0-mM β-mercaptoethanol; pH 8.0). After centrifugation (25,000 × g) for 10 min, the supernatant was mixed with 1% (v/v) glacial acetic acid and then incubated at 42°C for 30 min. After centrifugation (25,000 × g) for 5 min, the supernatant was mixed with assay solution (0.35% w/v NH₄MoO₄, 0.86-N H₂SO₄, and 1.4% w/v ascorbic acid) and then incubated at 42°C for 30 min, after which the absorbance at 820 nm was measured. The standard curve was generated with H₃PO₄.

Lipid analysis

Total lipids were extracted using the Bligh and Dyer method (Bligh and Dyer, 1959). Each lipid extract was separated by two-dimensional silica gel thin-layer chromatography with the solvent system 115:80:8 (v/v/v) chloroform/methanol/7-N ammonia for the first dimension and 170:25:15:3 (v/v/v/v) chloroform/methanol/acetic acid/sterile water for the second dimension. Separated lipids were then hydrolyzed and methylated with 5% (w/v) hydrogen chloride methanol solution (Wako), and fatty acid methyl esters were quantified by gas chromatography (GC) using pentadecanoic acid as an internal standard (Kobayashi et al., 2006; Murakawa et al., 2014). GC was connected to a flame ionization detector (GC-2014, Shimadzu; column, ULBON HR-SS-10, 0.25-mm I.D., 25 m, SHINWA CHEMICAL INDUSTRIES).

Image analysis with LSCM

First and second true leaves from 20-d-old plants were observed with LSCM. LSCM was performed with an inverted LSM800 system equipped with a C-apochromat LD63 × water-immersion objective lens (numerical aperture = 1.15; Carl Zeiss). GFP was excited with a 488-nm laser (emission, 500–535 nm), RFP was excited with a 561-nm laser (emission, 600–650 nm) and chlorophyll autofluorescence was excited with a 633-nm laser (emission, 670–750 nm).

To visualize autophagic bodies, ER-derived vesicles and RCBs, leaves were cut from the plants and incubated in a solution consisting of MES-NaOH (pH 5.5) with or without 1-µM ConcA for 20 h before observation.

GFP-ATG8 processing assay

Total protein was extracted from the first and second true leaves as described previously (Yoshimoto et al., 2004). Western blot analysis was performed using anti-GFP (3H9-100; Chromotek) and anti-rat IgG (cat. no. 112-035-003; Jackson ImmunoResearch).

Statistical analysis

Statistical analysis in this study was performed with R software. Tukey’s test was used to compare multiple samples as indicated in the figure legends.

Accession numbers

Genetic information in this article can be obtained from the Arabidopsis Information Resource (TAIR) under the following accession numbers: PHT1;1, At5g43350; At4, At5g03545; MGD3, At2g11810; ATG8a, At4g21980; CNX1, At5g61790; BIP3, At1g09080; DIN6, At3g47340; DIN10, At5g20250.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Pi concentration in each medium.

Supplemental Figure S2. Localization of GFP-ATG8a fusion protein without ConcA treatment.

Supplemental Figure S3. Effect of Pi starvation and oversupply of nitrate on autophagosome formation.

Supplemental Figure S4. GFP-ATG8 processing assay.

Supplemental Figure S5. Localization of ER-targeted GFP fusion protein without ConcA treatment.

Supplemental Figure S6. Effect of oversupply of nitrate on RCB formation under Pi-depleted conditions without ConcA treatment.

Supplemental Figure S7. Localization of CT-GFP fusion protein without ConcA treatment.

Supplemental Figure S8. Localization of RBCS2B-RFP fusion protein without ConcA treatment.

Supplemental Figure S9. Effect of core autophagy protein on RCB formation.

Supplemental Table S1. Primers used in this study

Funding

This work was supported by a Grants-in-Aid for Scientific Research on Innovative Areas [Nos. 17H06417 to M.S., 19H05713 to K.Y., 20H04916 to M.I., and 20H05352 to S.N.); a Grant-in-Aid for Scientific Research (C) [No. 15K07335 to M.S.] from the Ministry of Education, Culture, Sports, Science and Technology of Japan; by the Sasakawa Scientific Research Grant [2020-4032 to Y.Y.] from the Japan Science Society; and by the program of Open Innovation Platform with Enterprise, Research Institute and Academia (OPERA) of the Japan Science and Technology Agency.

Conflict of interest statement. None declared.

 

Y.Y., S.N., M.I., K.Y., and M.S. designed the experiments and wrote the manuscript. Y.Y., S.N., and D.S. performed the experiments. Y.Y., H.O., and M.S. analyzed the data.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys) is: Mie Shimojima (shimojima.m.aa@m.titech.ac.jp).