The autophagic degradation of chloroplasts via rubisco-containing bodies is specifically linked to leaf carbon status but not nitrogen status in Arabidopsis.

Autophagy is an intracellular process facilitating the vacuolar degradation of cytoplasmic components and is important for nutrient recycling during starvation. We previously demonstrated that chloroplasts can be partially mobilized to the vacuole by autophagy via spherical bodies named Rubisco-containing bodies (RCBs). Although chloroplasts contain approximately 80% of total leaf nitrogen and represent a major carbon and nitrogen source for new growth, the relationship between leaf nutrient status and RCB production remains unclear. We examined the effects of nutrient factors on the appearance of RCBs in leaves of transgenic Arabidopsis (Arabidopsis thaliana) expressing stroma-targeted fluorescent proteins. In excised leaves, the appearance of RCBs was suppressed by the presence of metabolic sugars, which were added externally or were produced during photosynthesis in the light. The light-mediated suppression was relieved by the inhibition of photosynthesis. During a diurnal cycle, RCB production was suppressed in leaves excised at the end of the day with high starch content. Starchless mutants phosphoglucomutase and ADP-Glc pyrophosphorylase1 produced a large number of RCBs, while starch-excess mutants starch-excess1 and maltose-excess1 produced fewer RCBs. In nitrogen-limited plants, as leaf carbohydrates were accumulated, RCB production was suppressed. We propose that there exists a close relationship between the degradation of chloroplast proteins via RCBs and leaf carbon but not nitrogen status in autophagy. We also found that the appearance of non-RCB-type autophagic bodies was not suppressed in the light and somewhat responded to nitrogen in excised leaves, unlike RCBs. These results imply that the degradation of chloroplast proteins via RCBs is specifically controlled in autophagy.

Autophagy is the major pathway by which proteins and organelles are transported for degradation in the vacuoles of yeast and plants or the lysosomes of animals (for detailed mechanisms, see reviews by Ohsumi, 2001;Levine and Klionsky, 2004;Bassham et al., 2006;Bassham, 2009). In these systems, a portion of the cytoplasm, including entire organelles, is engulfed in a doublemembrane vesicle called an autophagosome and delivered to the vacuole/lysosome. The outer membrane of the autophagosome then fuses with the vacuolar/ lysosomal membrane, and the inner membrane structure called the autophagic body is degraded within the vacuole/lysosome by resident hydrolytic enzymes. A recent genome-wide search confirmed that Autophagyrelated genes (ATGs) are well conserved across yeast, plants, and animals (Meijer et al., 2007). Using ATG knockout mutants and a monitoring system with an autophagy marker, GFP-ATG8, numerous studies have demonstrated the presence of the autophagy system in plants and its importance in several biological processes (Yoshimoto et al., 2004Liu et al., 2005;Suzuki et al., 2005;Xiong et al., 2005Xiong et al., , 2007Fujiki et al., 2007;Patel and Dinesh-Kumar, 2008;Phillips et al., 2008;Hofius et al., 2009). Plant autophagy is thought to play an important role in nutrient recycling under starvation, similar to its role in starvation previously noted in yeast and animals Bassham et al., 2006). For example, the creation of autophagosomes is induced by nitrogen, carbon, or both combined starvation in plant heterotrophic tissues such as roots (Yoshimoto et al., 2004;Xiong et al., 2005), hypocotyls Phillips et al., 2008), and suspension-cultured cells (Chen et al., 1994;Aubert et al., 1996;Moriyasu and Ohsumi, 1996;Rose et al., 2006). Arabidopsis (Arabidopsis thaliana) atg mutants show accelerated leaf senescence and cannot survive and respond to nutrient resupply after severe carbon or nitrogen starvation (Doelling et al., 2002;Hanaoka et al., 2002;Xiong et al., 2005;Phillips et al., 2008).
The majority of plant nitrogen and other nutrients are distributed to leaves in the vegetative growth stage (Schulze et al., 1994;Makino et al., 1997). In C 3 plants, 75% to 80% of total leaf nitrogen is distributed to chloroplasts, primarily as photosynthetic proteins such as Rubisco (Makino and Osmond, 1991;Makino et al., 2003). During senescence and suboptimal environmental conditions, Rubisco and most stromal proteins are degraded, and the released nitrogen is remobilized to growing organs and finally stored in seeds (Friedrich and Huffaker, 1980;Mae et al., 1983). Chloroplast proteins can be degraded under carbonlimited conditions caused by darkness (Wittenbach, 1978), with their carbon used as a substrate for respiration. Therefore, considering the role of autophagy in nutrient recycling, much attention should be paid to the study of chloroplast degradation. ATG transcript abundances are elevated during starvation-induced leaf senescence (Yoshimoto et al., 2004;van der Graaff et al., 2006;Chung et al., 2009). However, previous reports have not analyzed the effects of nutrient status on the appearance of autophagosomes or autophagic bodies in leaves. Recently, we observed the accumulation of autophagic bodies and Rubisco-containing bodies (RCBs), a kind of autophagic body containing chloroplast stroma, using fluorescent markers in the vacuole of excised leaves treated with concanamycin A, which suppresses vacuolar lytic activity (Ishida et al., 2008). Autophagy of chloroplast components could be observed during senescence of individually darkened leaves (Wada et al., 2009). However, it is not clear how the production of autophagic bodies and RCBs is affected by nutrient status in leaves.
In this study, we aimed to demonstrate the relationship between the nutrient status of Arabidopsis leaves and chloroplast degradation via RCBs. We examined the effects of nutrient conditions during leaf incubation, leaf carbohydrate contents over a diurnal cycle, mutations affecting starch metabolism, and nitrogen limitation on the appearance of RCBs. All analyses showed that carbon status is a major factor controlling the production of RCBs, while nitrogen status is less important in the nutrient response of autophagy in leaves. We also found that the production response of RCBs and non-RCB-type autophagic bodies containing cytoplasmic components other than chloroplasts to nutrient conditions was not always the same in excised leaves. This suggests that there is a mechanism specifically controlling RCB production in plant autophagy.

Responses of RCB Production to Carbon But Not Nitrogen Supply in Excised Leaves
RCBs could be visualized in the vacuolar lumen with laser-scanning confocal microscopy (LSCM) when leaves of chloroplast stroma-targeted GFPexpressing Arabidopsis (named CT-GFP) were excised and incubated in darkness in "incubation buffer," containing concanamycin A, an inhibitor of vacuolar H + -ATPase that suppresses vacuolar lytic activity (Fig. 1A;Ishida et al., 2008). RCBs were rarely seen when attached leaves were excised and immediately observed during natural senescence (Ishida et al., 2008). RCBs were also rarely observed even if excised leaves were incubated in darkness without concanamycin A for 20 h, suggesting that they are rapidly degraded in the vacuolar lumen under natural cell conditions (Ishida et al., 2008). Thus, inhibitor treatments for suppressing vacuolar lytic activity are essential for quantitative examination of RCB production. When excised leaves were incubated in incubation buffer, RCBs were detected more in mature and early senescent leaves than in young leaves, but RCBs were rarely seen by the presence of Murashige and Skoog (MS) medium containing Suc, even in early senescent leaves (Ishida et al., 2008). These previous data suggested that the production of RCBs was influenced by nutrient conditions during incubation in excised leaves. First, we quantitatively analyzed the effects of external nutrient supply and light conditions during incubation on the appearance of RCBs (Fig. 1E). Around 50 RCBs were detected in the field of view (188 mm 3 188 mm each) in leaves incubated in nutrient-free medium in darkness (dark MES; Fig. 1A). The addition of MS medium containing Suc (dark MS + Suc) greatly suppressed the appearance of RCBs. This suppression of RCBs was not relieved by the depletion of nitrogen (dark MS-N + Suc) but by the depletion of Suc (dark MS-N; dark MS). Furthermore, RCB appearance was also suppressed by the sole addition of Suc (dark + Suc; Fig. 1B), Glc (dark + Glc), or Fru (dark + Fru). Mannitol did not affect the appearance of RCBs (dark + Mann); thus, it is unlikely that osmotic stress affects RCB appearance.
The appearance of RCBs was suppressed by irradiation (light MES; Fig. 1C). The light-mediated suppression of RCB appearance was relieved by the addition of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU), an inhibitor of photosynthetic electron transport (light + DCMU; Fig. 1D), indicating the possibility that the appearance of RCBs was suppressed by photosynthesis, or the products of photosynthesis, rather than by light-mediated signals. Changes in starch and metabolic sugar contents were determined in excised leaves incubated either in darkness or in the light (Fig. 2). Although starch and metabolic sugars were decreased and almost consumed after 20 h of incubation in darkness (dark MES) or in the light with DCMU (light + DCMU), carbohydrate levels were increased during incubation in the light (light MES). These results suggested that RCB production was affected by photosynthetically accumulated internal carbohydrates in the light as well as externally added sugars in darkness.
Differential RCB accumulation might be due to differences in the stability of RCBs or the maker, GFP, in the vacuolar lumen rather than differential RCB production between the various treatments used here. To clarify this possibility, we used transgenic Arabidopsis expressing vacuole-targeted GFP, which is a GFP fusion of N-terminal signal peptide and C-terminal signal propeptide of pumpkin (Cucurbita sp.) 2S albumin (SP-GFP-2SC; Tamura et al., 2003), as an indicator for the stability of the GFP signal in the vacuole (Supplemental Fig. S1). When leaves of SP-GFP-2SC were excised and immediately observed with LSCM, the GFP signal was not observed in the vacuole but in the endoplasmic reticulum and Golgi complex, as reported previously (Supplemental Fig.  S1A; Tamura et al., 2003). It was also reported that GFP was rapidly degraded by vacuolar proteases in the light but that the degradation was suppressed by the addition of concanamycin A or E-64d (Tamura et al., 2003). When leaves of SP-GFP-2SC plants were incubated in the presence of concanamycin A and E-64d, GFP was observed in the vacuole irrespective of the treatment conditions used, darkness, darkness + MS medium containing Suc, or in the light, where the degree of RCB appearance was greatly different ( Fig. 1;  Supplemental Fig. S1, B-D). This suggests that the treatments used did not have significant effects on vacuolar lytic activity. If RCBs were degraded in leaves of CT-GFP plants under conditions where RCB appearance was significantly suppressed (Fig. 1), stromatargeted GFP, which was mobilized from chloroplasts via RCBs into the vacuole, should be distributed uniformly as SP-GFP-2SC in the vacuolar lumen, since GFP would not be degraded in the presence of protease inhibitors (Supplemental Fig. S1, B-D). However, we could not observe uniformly distributed GFP in vacuolar lumen in the leaves of CT-GFP plants under those conditions (Fig. 1). These results suggest that the appearance of RCBs in various incubation conditions was mainly due to differences in RCB production.

RCB Production Is Specifically Controlled during Autophagy of Mesophyll Cells
Previous studies have shown that autophagic bodies accumulate under nitrogen, carbon, or nitrogen and carbon combined starvation in heterotrophic tissues Figure 1. Effects of incubation conditions on the appearance of Rubisco-containing bodies. A to D, Visualization of RCBs by stroma-targeted GFP in various incubation conditions. Fourth leaves of stromatargeted GFP expressing Arabidopsis at 30 d after sowing (5 d after bolting) were excised and incubated in 10 mM MES-NaOH (pH 5.5) with 1 mM concanamycin A and 100 mM E-64d at 23°C for 20 h in darkness (A), with 3% (w/v) Suc in darkness (B), in the light (C), or with DCMU in the light (D). GFP appears green, and chlorophyll fluorescence appears red. In merged images, overlap of GFP and chlorophyll fluorescence appears yellow. In B, C, and D, merged images are shown. Bars = 50 mm. E, The number of RCBs in various incubation conditions. Excised leaves were incubated as described in A to D in darkness (blue columns) or in the light (yellow columns) and with various constituents: MES, no other constituent; MS, MS medium; MS-N, nitrogen-free MS medium; +Suc, 3% (w/v) Suc; +Glc, 3% (w/v) Glc; +Fru, 3% (w/v) Fru; +Mann, 3% (w/v) mannitol; +DCMU, 10 mM DCMU. Leaves from four independent plants were incubated, eight quadrangular regions (188 mm 3 188 mm each) per leaf were monitored by LSCM, and the number of accumulated RCBs was counted (RCB number). Each image was considered as an independent data point and subjected to statistical analysis by Tukey's test. Data represent means 6 SE (n = 32). Columns with the same letter were not significantly different (P # 0.01). (Yoshimoto et al., 2004;Xiong et al., 2005;Phillips et al., 2008). RCBs are a kind of autophagic body that specifically contain the stromal portion of chloroplasts. However, the pattern of RCB accumulation was not similar to autophagic body accumulation shown in these previous studies (Fig.  1). Therefore, we next compared RCB production with autophagic bodies containing cytoplasmic components other than the stroma (non-RCB-type autophagic bodies) in excised leaves under various nutrient conditions.
RCBs and entire autophagic bodies can be visualized at the same time using transgenic plants expressing stroma-targeted DsRed and the GFP-ATG8 fusion (Fig. 3, A and B;Ishida et al., 2008). In these transgenic plants (ecotype Wassilewskija), the number of RCBs and non-RCB-type autophagic bodies detected in the field of view (188 mm 3 188 mm each) was 41 6 2.7 and 20 6 1.7 (n = 64) under control conditions (dark MES; Fig. 3C), respectively. The appearance of RCBs was significantly suppressed by the presence of Suc or in the light, although this was relieved by the presence of DCMU (Fig. 3C), similar to the CT-GFP plants (ecotype Columbia; Fig. 1). The pattern of appearance of non-RCB-type autophagic bodies was not always the same as the pattern of RCB appearance (Fig. 3C). In darkness, the appearance of non-RCB-type autophagic bodies was suppressed by the presence of Suc, similar to RCBs (dark MS + Suc, dark MS-N + Suc, dark + Suc, and light + Suc). This suggested that exogenous excess sugars suppress entire autophagy. In Suc-free conditions, the appearance of non-RCB-type autophagic bodies was also suppressed by the presence of MS medium (dark MS) or nitrogen (dark + N), while RCB production was unaffected. These data indicated that nitrogen also affects the appearance of non-RCB-type autophagic bodies to some extent in leaves, as reported previously (Yoshimoto et al., 2004;Xiong et al., 2005;Phillips et al., 2008), but the effect of nitrogen was small compared with the effect of Suc. The appearance of non-RCB-type autophagic bodies was not suppressed in the light (light MES), while RCB production was consistently and markedly suppressed. This specific suppression of RCB production related to photosynthesis suggests that RCB production was more sensitive to increases in leaf carbohydrate contents than autophagosome creation.

Effects of Timing of Leaf Excision in a Diurnal Cycle on the Appearance of RCBs
To examine the effects of physiological changes in leaf carbohydrate contents on the appearance of RCBs, we investigated RCB production during a diurnal cycle. Fourth rosette leaves of CT-GFP plants were excised at three time points: the end of the regular night cycle, the end of a prolonged night, and the end of the day, and carbohydrate contents and RCB number were determined (Fig. 4A). As expected, starch levels were significantly higher at the end of the day compared with the end of the night, with most of the starch used up at the end of prolonged night (Fig. 4B). Among soluble sugars, Suc was significantly lower at the end of prolonged night, although no significant changes were detected in Glc or Fru levels. Recorded sugar levels were similar to those in previous reports (Gibon et al., 2004;Bläsing et al., 2005;Smith and Stitt, 2007). Leaves were excised and incubated in incubation buffer for 10 h in darkness. The number of RCBs was low in leaves excised at the end of the day, when carbohydrate levels were highest, and high in leaves excised at the end of prolonged night, when carbohydrate levels were at a minimum. Particularly, starch contents seemed to be a primary factor affecting RCB production.

Effects of Mutations Affecting Leaf Starch Content on the Appearance of RCBs
To provide further evidence of the close relationship between RCB production and leaf carbohydrate contents, we examined RCB production in starch content genetic mutants, specifically starchless phosphoglucomutase (pgm-1) and ADP-Glc pyrophosphorylase1 (adg1-1) and starch-excess starch-excess1 (sex1-1) and maltose-excess1 (mex1-3). The mutant pgm-1 lacks chloroplast phosphoglucomutase (Caspar et al., 1985), and adg1-1 lacks the activity of ADP-Glc pyrophosphorylase (Lin et al., 1988). Neither mutant can accumulate starch during the day. The mutant sex1-1 lacks glucan water dikinase 1 and is reduced in its capacity to degrade starch (Caspar et al., 1991;Yu et al., 2001), while mex1-3 lacks the chloroplast maltose transporter and contains both high levels of starch and very high  (Niittylä et al., 2004). These starchrelated mutants grow like wild-type plants in continuous light; however, their growth is progressively impaired as the duration of the night period is increased (Caspar et al., 1985(Caspar et al., , 1991Lin et al., 1988). We first set up starch-related mutants expressing stromatargeted GFP and visualized the chloroplast stroma of each mutant (Supplemental Fig. S2). In mex1-3, we observed both normal and abnormally shaped chloroplasts, which exhibited swollen chloroplasts, as revealed by the fluorescence signals of stroma-targeted GFP (Supplemental Fig. S2, D and E). This "swollen chloroplast" phenotype has been noted previously in mex1 mutant cells observed by transmission electron microscopy (Stettler et al., 2009).
In continuous-light growth conditions, both the wild type and the mutants all took about 21 d to start bolting. Mutants pgm-1, adg1-1, and sex1-1 were almost the same size and had the same leaf number as the wild type at 5 d after bolting (Fig. 5A). Only mex1-3 was smaller than the wild type; however, leaf number was almost the same as in the wild type (Fig. 5A). Carbohydrate contents were determined in the fourth leaves at 5 d after bolting (Fig. 5B). As expected, starch levels were significantly higher in starch-excess mutants than in the wild type. No starch was detected in starchless mutants. Soluble sugar levels were similar in wild-type and mutant plants. Leaves from each set of plants were excised and incubated in incubation buffer. RCBs were also observed in leaves of each of Figure 3. Effects of incubation conditions on the appearance of RCBs and autophagic bodies. A and B, Visualization of both RCBs and autophagic bodies. Excised leaves from both GFP-ATG8-and stromatargeted DsRed-expressing plants at 25 d after sowing (5 d after bolting) were incubated at 23°C for 20 h in 10 mM MES-NaOH (pH 5.5) with 1 mM concanamycin A and 100 mM E-64d in darkness (A) or in the light (B). GFP appears green and DsRed appears red. In merged images, overlap of GFP and DsRed appears yellow. Bars = 10 mm. C, The appearance of RCBs and non-RCB-type autophagic bodies in various incubation conditions. Excised leaves from GFP-ATG8-and stroma-targeted DsRed-expressing plants were incubated as described in A and B in darkness or in the light with various constituents: MES, no other constituent; MS, MS medium; MS-N, nitrogen-free MS medium; +Suc, 3% (w/v) Suc; +N, nitrogen nutrition of MS medium; +DCMU, 10 mM DCMU. Leaves from six independent plants were incubated, eight quadrangular regions (188 mm 3 188 mm each) per leaf were monitored by LSCM, and the number of accumulated RCBs (red columns) and non-RCB-type autophagic bodies not containing DsRed (green columns) were counted. Each image was considered as an independent data point and subjected to statistical analysis by Tukey's test. Each value is shown as a percentage of the level in "dark MES." Data represent means 6 SE (n = 64). Values with the same letter were not significantly different in RCBs (red values) and non-RCB-type autophagic bodies (green values; P # 0.05).
the starchless and starch-excess mutants, although the number of RCBs was significantly higher in leaves of starchless mutants and lower in starch-excess mutants than in the wild type (Fig. 6).
Next, we analyzed RCB production throughout the leaf life span in the wild type and starch-related mutants under long-day growth conditions with a 14-h photoperiod, because the growth rates of each mutant differed under long-day conditions. Here, wild-type plants took 25 d to bolt. A further 2 to 3 d were required in mex1-3, 3 to 4 d in sex1-1, and 5 to 6 d in pgm1-1 and adg1-1. In wild-type and mutant plants, leaf area of the fourth leaf rapidly increased before bolting, and leaf expansion had almost finished as bolting commenced (Supplemental Fig. S3). We quantified the appearance of RCBs in fourth leaves from 5 d before bolting, during active leaf expansion, to 10 d after bolting, the senescent stage, in each plant (Fig.  7A). In the wild type, the number of RCBs increased in expanding, mature and early senescent leaves, as reported previously (Ishida et al., 2008), but decreased at the later senescent stage. In starchless mutants, the appearance of RCBs was significantly higher than in the wild type in expanding and mature leaves but declined during leaf senescence faster than in the wild type. In leaves of starchless mutants, the transcription levels of senescence-associated genes, SAG12 and SAG13, were higher than in the wild type (Fig. 7B). This implied rapid leaf senescence in starchless mutants, leading to a rapid decline of RCB production. In starch-excess mutants, RCB appearance was always lower than in wild-type plants during the period examined. In senescent leaves, the appearance of RCBs was significantly lower in starch-excess mutants than in the wild type. The maximum number of RCBs during the leaf life span was significantly higher in starchless mutants and lower in starch-excess mutants compared with the wild type (Fig. 8A). Starch could not be detected in the leaves of starchless mutants, and significantly higher starch contents were detected in leaves of starch-excess mutants than in the wild type when the number of RCBs after incubation was maximum in each plant (Fig. 8B), while soluble sugar levels were higher in starch-related mutants than in the wild type (Fig. 8C). The starch-mediated suppression of RCB production in starch-related mutants was also found in the long-day growth condition used here.

RCB Production Is Attenuated during Nitrogen-Limited Senescence
Carbon or nitrogen limitation accelerates leaf senescence. We previously reported that RCBs can be observed in the vacuole without inhibitor treatment in individually darkened leaves, where senescence was induced by carbon limitation (Wada et al., 2009). Here, we examined the appearance of RCBs during leaf senescence accelerated by nitrogen limitation at the whole plant level.
Nitrogen-limited plants were smaller than nitrogensufficient plants at 5 d after the imposition of treatments (Fig. 9A). In nitrogen-limited plant leaves, the contents of chlorophyll, nitrogen, soluble protein, and Rubisco rapidly decreased (Fig. 9B). In spite of the accelerated degradation of Rubisco and soluble protein, RCBs were not observed without concanamycin A treatment in leaves of nitrogen-limited plants (data not shown). Even if excised leaves were incubated with incubation buffer containing concanamycin A, significantly fewer RCBs were noted in the leaves of nitrogenlimited plants compared with nitrogen-sufficient plants from 1 d after the start of the treatment (Fig.  10A). Five days after the start of the treatment, few RCBs could be noted in leaves of nitrogen-limited plants even in the presence of concanamycin A (Fig.  10A). In nitrogen-limited plants, carbohydrates, starch, Glc, and Fru were rapidly accumulated, and Suc was also increased (Fig. 10B). When attached leaves of nitrogen-limited plants were individually darkened to impede the accumulation of carbohydrates, few RCBs and faint GFP signals were observed in the vacuole, similar to individually darkened leaves of nitrogensufficient plants (data not shown). This result supported our conclusion that RCB production is primarily mediated by carbohydrate accumulation rather than by nitrogen limitation in individual leaves.

DISCUSSION
Recent works have demonstrated the importance of autophagy in nutrient recycling during both nitrogen and carbon starvation in plants (Doelling et al., 2002;Hanaoka et al., 2002;Yoshimoto et al., 2004;Xiong et al., 2005;Phillips et al., 2008). However, it has not been revealed how autophagy of chloroplast proteins, which are a major source of carbon and nitrogen for new growth, is regulated by nutrient status. In this study, we examined the appearance of RCBs, autophagic bodies containing chloroplast proteins, under various conditions that affect the nutrient status of leaves. Our data indicate that the autophagy of chloroplasts via RCBs is closely and specifically linked to leaf carbon but not to nitrogen status in individual leaves.

Physiological Roles of RCB and Autophagy in Leaves
The data presented here suggest one possible role of RCBs, which is energy supply under temporal carbon starvation. In the light, chloroplasts provide energy for cellular processes and growth via photosynthesis. In darkness or insufficient light conditions, which cause energy limitation, photosynthetic carbon fixation is reduced and the proteins may be used as an energy source via autophagic recycling. In unicellular organisms, degradation of unnecessary cellular components is known to be an important role of autophagy. For example, when methylotrophic yeasts such as Pichia pastoris and Hansenula polymorpha are grown in the presence of methanol as their sole carbon source, they increased their peroxisome numbers, as this organelle is essential for metabolizing methanol. The leaf carbohydrate contents in the wild type and starch-related mutants. The contents of starch (left) and Suc, Glc, and Fru (right) were determined in leaves of stroma-targeted GFP transgenic plants of background wild type, pgm-1, adg1-1, sex1-1, or mex1-3 at 26 d after sowing. Data represent means 6 SE (n = 3). Statistical analysis was performed by Tukey's test. Columns with the same letter were not significantly different (P # 0.05).
When they are transferred to a Glc-or ethanol-rich environment, the excess peroxisomes are selectively degraded via autophagy for adaptation to the environmental change (Dunn et al., 2005;Sakai et al., 2006). This mechanism is called pexophagy. RCBs may be a similarly selective autophagy system for the degradation of the chloroplast component. In pexophagy systems, whole peroxisomes are transported to the vacuole. Chloroplasts are usually essential for carbon assimilation in plants, while many peroxisomes are not required for growth on most carbon sources other than methanol in yeast. Whole chloroplast degradation would impair the resumption of photosynthesis when light conditions improve. Partial degradation via RCBs is possibly effective for maintaining some basal functions of chloroplasts with a supply of needed energy. Therefore, one possible role of RCBs is supplying stromal proteins and chloroplast envelope as a carbon source under carbon-limited conditions without the loss of whole chloroplasts. In fact, free amino acid levels increase during extended darkness in Arabidopsis leaves, possibly by protein degradation (Usadel et al., 2008). Several Arabidopsis atg mutants exhibit accelerated leaf senescence and could not survive under severe carbon limitation with ex-tended darkness (Doelling et al., 2002;Hanaoka et al., 2002;Xiong et al., 2005;Phillips et al., 2008). RCBs may contribute to adaptation and survival under extended darkness by supplying respiratory carbon.
In plants, carbon starvation by darkness accelerates leaf senescence in some cases. Weaver and Amasino (2001) provided an experimental model of leaf senescence in Arabidopsis; senescence was induced when individual leaves were darkened but not when whole plants were darkened. Dark-induced senescence was highly localized (i.e. senescence was locally accelerated in darkened but not in illuminated areas in the same leaf). Our recent work demonstrates that chloroplast autophagy occurs in the senescence of individually darkened leaves (Wada et al., 2009). Together with this report, our data here strongly indicate that RCBs also have an important role in chloroplast degradation during accelerated senescence by darkness in Arabidopsis. During senescence of individually darkened leaves, transportation of whole chloroplasts by ATG4 gene-dependent autophagy was also observed (Wada et al., 2009). This implied that maintenance of basal chloroplast functions was no longer necessary in darkened sectors, and it may be more advantageous for plant growth that nutrients were remobilized to organs under more direct illumination. In fact, in individual leaves, dark-induced senescence was not reversed by a return to light; conversely, whole darkened plants exhibited delayed senescence relative to nontreated plants (Weaver and Amasino, 2001). Thus, two roles of RCBs, supply of temporal energy source under plant carbon limitation and effective degradation of chloroplasts under leaf carbon limitation, may exist. RCBs that possibly contribute to the effective degradation of chloroplast in darkened leaves cooperated with whole chloroplast degradation. Alternatively, because autophagic machinery may not have sufficient size capacity for the transportation of normal chloroplasts, shrinkage of chloroplast size by RCBs may be a prerequisite for the degradation of whole chloroplasts.
The progression of leaf senescence is regulated by several interacting environmental signals, with nutrient status being one of the most important (Lim et al., 2003;Yoshida, 2003). Sugar is considered as an important internal signal; it has been indicated that leaf senescence is induced not only by sugar starvation caused by darkness but also by sugar accumulation (Weaver and Amasino, 2001;Pourtau et al., 2006;Wingler et al., 2006Wingler et al., , 2009van Doorn, 2008). Leaf carbon limitation can be an important factor controlling senescence in natural conditions. For example, in a plant canopy, upper leaves can shade older leaves, which can be a factor accelerating loss of nitrogen and shortening a leaf life span in shaded leaves (Oikawa et al., 2006). Our study suggests that RCBs can function during senescence under natural conditions depending on leaf carbon status, particularly promoted by carbon limitation with insufficient light.

RCB Production and Carbohydrate Metabolism in Starch-Related Mutants
The appearance of RCBs tended to be higher in leaves of starchless mutants and lower in starch-excess mutants compared with the wild type in both continuous light and day/night conditions . These results suggest that starch accumulation directly influences RCB production. Leaf starch content also seemed to be a key factor of RCB production over a diurnal cycle (Fig. 4). Accumulation of starch granules may alter the tension of the chloroplast envelope. Two mechanosensory proteins, MSL2 and MSL3, which have been identified in the plastid envelope, control plastid shape and size (Haswell and Meyerowitz, 2006). Envelope tension seems to be a signal that regulates RCB production. However, the appearance of RCBs was also suppressed by light during leaf incubation in starchless mutants similar to wild-type plants (data not shown). Additionally, in nitrogenlimited plants, the suppression of RCB appearance could be observed from 1 d after treatment, while starch accumulation was detected later (Fig. 10). Glc and Fru accumulated from 1 d after the start of the treatment (Fig. 10). Thus, we cannot attribute the suppression of RCB production simply as results of leaf starch accumulation. Rather, we can understand the changes of RCB production in starch-related mutants by referring to a diurnal metabolism of leaf carbohydrates. Carbon is assimilated during the day, with a part of the photoassimilate retained as starch in the chloroplast; however, at night, this starch is degraded to support leaf respiration, allowing diurnal maintenance of Suc levels in Arabidopsis leaves (Gibon et al., 2004;Smith and Stitt, 2007). In starchless mutants, soluble sugars accumulated to higher levels than in the wild type during the day, but they rapidly decreased and were almost completely consumed by the middle of the night (Caspar et al., 1985;Gibon et al., 2004). Similarly, during incubation of excised leaves in darkness, reduced starch levels appeared to Figure 7. Effects of mutations affecting leaf starch content on the appearance of RCBs throughout leaf life span under a 14-h photoperiod. A, Changes of the appearance of RCBs in leaves of starch-related mutants under a 14-h photoperiod. Leaves of stroma-targeted GFP transgenic plants of background wild type (black squares), pgm-1 (black circles), adg1-1 (white circles), sex1-1 (black triangles), or mex1-3 (white triangles) were excised 5 d before bolting and 0, 5, and 10 d after bolting. Leaves were incubated at 23°C for 20 h in 10 mM MES-NaOH (pH 5.5) with 1 mM concanamycin A and 100 mM E-64d in darkness. Leaves from six independent plants were incubated, eight quadrangular regions (188 mm 3 188 mm each) per leaf were monitored by LSCM, and the number of accumulated RCBs was counted at each stage. Data represent means 6 SE (n = 48). B, The transcription levels of SAG13, SAG12, and RBCS2B as marker genes for leaf senescence. Total RNA from third and fourth leaves of stroma-targeted GFP transgenic plants of background wild type and each mutant during 10 d after bolting was isolated and subjected to semiquantitative reverse transcription-PCR using gene-specific primers. PCR products were separated by electrophoresis, stained with SYBR Green I, and quantified by a fluorescence image analyzer. 18S ribosomal RNA was used as an internal control.
allow more rapid consumption of sugars in starchless mutants, possibly resulting in increased production of . During a diurnal cycle in starch-and maltose-excess mutants, soluble sugar levels were similar or slightly lower than wild-type levels during the night, although the supply of soluble sugars was partially defective by lower starch degradation capacity (Caspar et al., 1991;Trethewey and ap Rees, 1994;Zeeman and ap Rees, 1999;Niittylä et al., 2004). During the extended night, there was a more than 30% decrease in starch content in sex1 mutant leaves, while most of the starch was already consumed at the start of the extended night in the wild type (Caspar et al., 1991). During leaf incubation in darkness, soluble sugars may have been supplied for a longer period by degradation of starch in starch-excess mutants than in wild-type plants, leading to lower production of . We propose that available carbohydrate contents may be an important factor controlling RCB production in leaves, although its sensing mechanisms are largely unknown. Starch contents are greatly different from soluble sugar contents (Figs. 4, 5, and 8), indicating that starch is the main available carbohydrate in darkness.
Recently, Stettler et al. (2009) reported that the accumulation of maltose and malto-oligosaccharides in mex1 mutants causes chloroplast dysfunction, which triggers chloroplast degradation, possibly by autophagy. In our study, the appearance of RCBs was low in leaves of mex1 plants (Figs. 6-8). However, we occasionally could observe aggregated or round structures exhibiting chlorophyll fluorescence in dark-incubated leaves of mex1 plants expressing stroma-targeted GFP (data not shown). Vacuole-located chloroplasts transported by ATG4 gene-dependent autophagy also exhibited chlorophyll fluorescence without stroma-targeted fluorescent proteins (Wada et al., 2009). In mex1 plants, only whole chloroplast autophagy, not RCBs, may be induced. When naturally senescent leaves were subjected to 20 h of darkness with concanamycin A, autophagy of only RCBs, but not whole chloroplasts, was observed ( Fig. 1A; Ishida et al., 2008). Therefore, the results in mex1 plants also suggest that the partial mobilization of chloroplast components via RCBs and autophagy of whole chloroplasts may be differentially regulated depending upon the cellular status.

RCB Production Is Specifically Controlled in Autophagy
The effects of incubation conditions on the appearance of RCBs and non-RCB-type autophagic bodies were not always the same (Fig. 3), indicating that there exists a novel mechanism specifically controlling RCB production in the progression of autophagy. In yeast and animal cells, autophagy is induced by the starvation of either nitrogen or carbon sources (Takeshige et al., 1992;Mizushima, 2005). Starvation-induced autophagy has also been confirmed directly using a  Figure  7A. B and C, The leaf carbohydrate contents when the appearance of RCBs was maximum. The contents of starch (B) and Suc, Glc, and Fru (C) were determined in fourth leaves at 4 to 6 h into the photoperiod at 5 d after bolting in wild-type and mex1-3 plants or at bolting in pgm-1, adg1-1, and sex1-1, when the appearance of RCBs after leaf incubation was maximum in Figure 7A. fluorescent marker, GFP-ATG8, in roots, hypocotyls, and cultured plant cells (Yoshimoto et al., 2004;Xiong et al., 2005, Phillips et al., 2008. In our study, we grew plants autotrophically and observed autophagy in leaves. Specific features of RCB induction may reflect differences in heterotrophs and autotrophs and also in heterotrophic and autotrophic tissues in plants. In nature, plants have to get carbohydrates by photosynthesis and assimilate inorganic nitrogen into organic compounds. Chloroplasts have a central role in both photosynthesis and nitrogen assimilation, and around 80% of total leaf nitrogen is distributed to chloroplasts. Therefore, although the underpinning mechanisms are still  unknown, it seems reasonable to consider that autophagy of chloroplast proteins must be specifically controlled. In yeast, some selective autophagy pathways have been identified (e.g. cytoplasm-to-vacuole targeting pathway, mitophagy, pexophagy, and also specific ATG genes required for respective pathways were identified; for review, see Kraft et al., 2009;Nakatogawa et al., 2009). Our study is suggestive of the existence of novel plant ATG genes specifically required for the autophagic degradation of chloroplasts, and identification of such genes may be effective for elucidating the regulatory mechanism of RCB production and specific features of plant autophagy.

Image Analysis with LSCM
LSCM was performed with a Nikon C1si system equipped with a CFI Plan Apo VC603 water-immersion objective (numerical aperture = 1.20; Nikon) as described previously in detail (Ishida et al., 2008). The signals of GFP and DsRed were obtained using the unmixing program of Nikon C1si software from fluorescence spectra between 500 and 650 nm.

Detection of RCBs and Autophagic Bodies
Fourth rosette leaves were excised from four to six independent plants expressing CT-GFP 4 to 6 h into the photoperiod. Incubation buffer containing 10 mM MES-NaOH (pH 5.5), 1 mM concanamycin A, and 100 mM E-64d was infiltrated into excised leaves, which were then incubated for 20 h at 23°C in darkness. When effects of nutrient factors during leaf incubation on RCB production were examined, full-strength MS medium and various constituents were added to the incubation buffer, or light (50 mmol quanta m 22 s 21 ) was irradiated. The concentration of Suc, Glc, Fru, and mannitol added was 3% (w/v), and DCMU was 10 mM. In incubation with nitrogen-free MS medium (MS-N), MS-N medium was prepared by removal of NH 4 NO 3 and replacing the KNO 3 with KCl. When diurnal effects were examined, leaves were excised at three time points representing the end of the regular night, the end of the prolonged night, and the end of the day, and incubation time was 10 h as described in Figure 4A. After incubation, each leaf was divided into four parts, and two quadrangular regions (188 mm 3 188 mm each) in each section were monitored by LSCM and images were obtained. The number of accumulated RCBs in the images was counted. Each image was considered an independent data point and subjected to statistical analysis.
The analysis of production of RCBs and autophagosomes in plants expressing stroma-targeted DsRed and the GFP-ATG8 fusion was performed in the same way, except that leaves from six independent plants at 25 d after sowing (5 d after bolting) were used. The number of RCBs and non-RCB-type autophagic bodies that do not exhibit DsRed was counted in each image.

Sugar Analysis
Leaves for carbohydrate quantification were rapidly frozen in liquid N 2 and then stored at 280°C until analysis. Frozen leaves were homogenized in a liquid N 2 -chilled tube with a pestle. The resulting powder was immediately extracted four times with 80% (v/v) ethanol at 80°C (Nakano et al., 1995), and the extractant was pooled and concentrated by vacuum evaporation. The residue was dissolved in distilled water, and the concentrations of Suc, Glc, and Fru were determined using an F-kit for a-Glc/Suc/Fru (Roche Diagnostics).
The 80% ethanol-insoluble fraction was dried overnight at room temperature. Starch in the ethanol-insoluble fraction was extracted with 0.6 mL of dimethyl sulfoxide and 0.15 mL of 8 M HCl at 60°C for 30 min, and the pH was neutralized to between 4.0 and 5.0 with 0.1 mL of 2 M Na-acetate buffer (pH 4.5) and 8 M NaOH. The solution was made up to its final volume of 1.0 mL with distilled water. The starch concentration was determined using an F-kit for starch (Roche Diagnostics).

Quantification of Chlorophyll, Nitrogen, Soluble Proteins, and Rubisco Protein
Frozen fourth rosette leaves were homogenized in an ice-chilled mortar and pestle in 50 mM Na-phosphate buffer (pH 7.5) containing 2 mM iodoleacetic acid, 0.8% (v/v) 2-mercaptoethanol, and 5% (v/v) glycerol (Makino et al., 1994). Chlorophyll was determined by the method of Arnon (1949). Soluble protein content was measured in supernatant of part of the homogenate according to Bradford (1976) using a Bio-Rad protein assay with bovine serum albumin as the standard. Total leaf nitrogen was determined from part of the homogenate with Nessler's reagent after Kjeldahl digestion. Triton X-100 (0.1% final concentration) was added to the remaining homogenate. After centrifugation, the supernatant was mixed with an equal volume of SDS sample buffer containing of 200 mM Tris-HCl (pH 8.5), 2% (w/v) SDS, 0.7 M 2-mercaptoethanol, and 20% (v/v) glycerol, boiled for 3 min, and then subjected to SDS-PAGE. The Rubisco content was determined spectrophotometrically by formamide extraction of the Coomasie Brilliant Blue R-250-stained bands corresponding to the large and small subunits of Rubisco separated by SDS-PAGE. Calibration curves were made with bovine serum albumin.

Supplemental Data
The following materials are available in the online version of this article.  Figure S1. Effects of incubation conditions on the accumulation of vacuole-targeted GFP in vacuolar lumens.
Supplemental Figure S2. Visualization of stroma-targeted GFP in leaves of starch-related mutants.
Supplemental Figure S3. Changes of leaf area in starch-related mutants under a 14-h photoperiod.