At the onset of leaf senescence, an important physiological process at the final stage of leaf development, a leaf is reallocated from a sink to a source organ, from which macromolecules synthesized during the sink phase are degraded and are subsequently redistributed into the upper tissues or reproductive organs. Various factors such as transcriptional regulations, phytohormones, reactive oxygen species (ROS), and environmental factors influence leaf senescence progression (Lim et al., 2007; Woo et al., 2013). The major targets of the degradation processes have been demonstrated to reside in chloroplasts: macromolecules in chlo roplasts, such as protein, lipid, nucleic acid and pigment, undergo degradation (Krupinska, 2006; Hörtensteiner, 2013; Kato and Sakamoto, 2013; Sakamoto and Takami, 2014). Furthermore, autophagy has been demonstrated to incorporate chloroplasts during senescence (Ishida and Yoshimoto, 2008; Wada et al., 2009). These degradation processes contribute to the efficient recycling and redistribution of nitrogen compounds.

Among the degraded compounds in senescing leaves, chlorophyll (Chl) is the most abundant pigment. Higher plants contain both Chl a and Chl b in the light harvesting antennae and only Chl a in the reaction center of Photosystem I (PSI) and PSII. Chl is necessary for photosynthesis because the excitation energy received by Chl drives the water splitting reaction and subsequent electron transport. In contrast, an excited state of free Chl is toxic because it generates ROS ultimately leading to cell death. Chl degradation is therefore implicated as a safe means of detoxification (Tanaka and Tanaka, 2011; Hörtensteiner, 2013). Conversion of Chl b to Chl a mediated by Chl b reductase (CBR) is an initial step for Chl degradation (Ito et al., 1996; Rüdiger, 2002; Kusaba et al., 2007) because only Chl a undergoes the subsequent degradation pathway that involves (i) de-chelation of Chl a (pheophytin), (ii) removal of phyto-tail (pheophorbide a), and (iii) oxygenation of rings, leading to the production of non-toxic primary fluorescent catabolites. Mutations that impair any of the reactions above give rise to a ‘stay-green’ phenotype in which leaves implement senescence but retain Chl (Kusaba et al., 2013). This phenotype, designated as cosmetic stay-green, is apparent in many plant species such as Arabidopsis (Oh et al., 2004; Ren et al., 2007), rice (Kusaba et al., 2007; Park et al., 2007), pea (Armstead et al., 2007; Sato et al., 2007), tomato (Cheung et al., 1993; Akhtar et al., 1999), and soybean (Luquez and Guiamét, 2002; Fang et al., 2014).

Reportedly, Chl degradation takes place not only in senescing leaves but also during embryogenesis and cotyledon development. De-greening of developing embryos is usually observed in Arabidopsis and other species (Chao et al., 1995; Nakajima et al., 2012). Some enzymes for Chl degradation pathway appear to be up-regulated at these stages. Although the physiological implications of Chl degradation during embryogenesis and cotyledon development remain unclear, it is likely that Chl are catabolized properly during embryo maturation. Supporting the importance of Chl degradation in embryo, Nakajima et al. (2012) reported that the Arabidopsis mutant lacking CBR, Non-Yellow Coloring 1 (NYC1) and “NYC1-Like” (NOL1) caused the impairment of seed maturation and storability. This observation strongly implies that Chl accumulated in seeds are liable to cause oxidative damage to photosystems or peroxidation of chloroplast membranes, by which seeds do not show lethality but show reduced fitness.

In this study, we hypothesize that one target for the damage caused by impaired Chl degradation is chloroplast membranes. Previously, we demonstrated that VIPP1 plays a crucially important role in protecting photosynthetic membranes (Zhang et al., 2012; Zhang and Sakamoto, 2013, 2015). VIPP1 was originally identified as being abundant in envelopes (Li et al., 1994). Based on the seedling-lethal phenotype in vipp1 mutant lacking vesicle formation, VIPP1 was suggested to play a role in thylakoid formation through vesicle fusion (Kroll et al., 2001; Westphal et al., 2001). However, its precise function remains elusive. VIPP1 forms an extremely large homo-complex (>2 MDa) that is associated with inner envelope and thylakoid membranes. The disassembly/reassembly of the VIPP1 complex, often designated as VIPP1 functional particle, takes place according to the membrane damage (Aseeva et al., 2004; Liu et al., 2005, 2007). It is particularly interesting that we recently demonstrated that overexpression of VIPP1 prevents chloroplast membranes from being damaged against heat shock and hypotonic stress. Therefore, it is possible to assume that VIPP1 overexpression improves seed defects in Arabidopsis nyc1 mutant. Our results demonstrate that this is indeed the case, corroborating our hypothesis that the membrane integrity can be disturbed by Chl remaining in seeds. Our results also corroborate that VIPP1 functions in preventing chloroplast membranes from oxidative damage.

The aberrant cotyledon phenotype described in this study (Figure 1) has been briefly reported by Nakajima et al. (2012), in which the authors rather focused on the effect of impaired Chl degradation on seed storability. They noted that nyc1/nol double mutant displays the aberrant cotyledon that looked similar to what is reported in this study, showing plant-to-plant variability from white to pale yellow colors. However, this phenotype has not been investigated further. Here, we demonstrated that it is nyc1, but not nol, from which the aberrant phenotype derives: it is also the case in which in Arabidopsis, nyc1 shows a cosmetic stay-green phenotype during leaf senescence, whereas nol does not (Horie et al., 2009). These findings together suggest that unlike rice, NYC1 plays a major role in Chl degradation not only of senescing leaves but also of seed maturation. Furthermore, we demonstrate that light intensity affects the appearance of aberrant cotyledons (Figure 2), which partly explains why the penetrance of the cotyledon color phenotype varied. Chl acts as a potential photosensitizer when not integrated into the photosystems (Hörtensteiner, 2004). It is therefore likely that Chl degradation in seed tissues must be tightly regulated to maximize fitness. Although, no difference in germination rate was detected in our study (Figure 2), retention of Chl in the mature seeds has been implicated as correlated with the lower seed germination rate (Jalink et al., 1998). We detected only the aberrant cotyledon, not lowered germination rate, probably because we used seeds that had been stored only for a short period (3 months).

We reasoned that our result of trans-complementation should be confirmed by some other means because we used only one allele of nyc1. A stay-green transgenic line (termed BCG) showing the phenotype similar to nyc1 was reported by Sakuraba et al. (2012). Instead of knockout in NYC1, BCG overproduces Chl b oxygenase (CAO) that engenders increased Chl b levels. As a consequence, BCG was shown to exhibit a prolonged stay-green in senescing leaves. CAO consists of three domains (respectively designated as A, B, and C), and A domain negatively regulates CAO accumulation in vivo. To circumvent this negative regulation, BCG expressed CAO that contains only B and C domains: because BC-CAO is tagged with GFP, the resulting transgene (and transgenic line) was designated as BCG (Sakuraba et al., 2010). We found that under our growth condition (70 μmol photon m^–2 s^–1), BCG cotyledons show white or pale yellow as seen in nyc1 (Figure 6). Interestingly, introduction of VIPP1-GFP into BCG recovers the aberrant cotyledon phenotype (Figure 6). VIPP1-GFP/BCG apparently retained the stay-green phenotype in senescing leaves. Moreover, the envelope damage assessed by the balloon-like chloroplasts in cotyledons was also recovered by VIPP1-GFP (Figure 6). Results from VIPP1-GFP/BCG were consistent with VIPP1-GFP/nyc1, leading us to confirm that VIPP1 trans-complemented aberrant cotyledon development because of the disturbance in Chl degradation.

The similar phenotype between nyc1 and BCG implies tight regulation of Chl levels during cotyledon development. In contrast to the cotyledon, however, true leaves of nyc1 and BCG develop normally because of their distinct differentiation paths in these two organs. Chloroplasts in true leaves originate from proplastids of apical meristem cells, where robust thylakoid membranes are formed in a light-dependent manner. Chloroplasts in cotyledons are differentiated upon germination from pre-existing plastids, which contain thylakoid membranes developed during embryogenesis. Together, our data enable us to conclude that Chl degradation is necessary for the initial stage of plant life: seedling development.

Given the light-dependent chlorotic phenotype, we inferred that overproduction of Chl (and perhaps its degradation compounds detected as high fluorescence, Figure 4) tend to accumulate excitation energy that is not transferred to photosystems and ultimately engender production of toxic molecules such as ROS. Under this scenario, it was speculated that ROS accumulates more in nyc1 than Col cotyledons. However, histochemical detection of hydrogen peroxide showed no significant difference (Supplementary Figure S1). Perhaps subtle and transient difference in ROS accumulation is sufficient to result in aberrant cotyledons. Sakuraba et al. (2010) has shown that in BGC plants, true leaves accumulate high ROS in a light-dependent manner, which suggests that excess Chl may lead to photo-oxidative stress. Reportedly, oxidative stress is one of the factors that affect the seed storability (Sattler et al., 2004). Another remaining question is whether excess Chl can associate with envelopes or in its vicinity, which subsequently affects envelope membrane integrity. We infer that it is unlikely, because Nakajima et al. (2012) reports that not only Chl but also LHCII exist in nyc1 cotyledons. Although the physiological status of plastids in aberrant cotyledons is unclear, excess Chl appeared to form LHC, similarly to those in senescing leaves. A possibility that VIPP1 scavenges ROS directly is also unlikely: no such implication has been suggested.

Numerous mutants showing white cotyledons have been reported in Arabidopsis (e.g., sco1, cyo1, sco2, sig2). For example, cyo1 exhibits white cotyledons; CYO1 has been shown to have disulfide reductase activity (Shimada et al., 2007). Preferential expression of CYO1 in young seedlings implicates that, although the precise substrate of CYO1 remains unclear, redox regulation is crucial in cotyledons. Unlike nyc1, however, the appearance of white cotyledons in cyo1 is independent of the light intensity and does not result from photobleaching. It has been suggested that ROS generation in chloroplasts renders membranes sensitive to oxidative stress damage (Mahajan and Tuteja, 2005). Chloroplast membrane stability has been compromised when membrane lipid peroxidation has resulted from ROS (Anjum et al., 2011; Lata et al., 2011). Therefore, it seems reasonable to assume that one target damaged by Chl over-accumulation in the cotyledons of nyc1 is the chloroplast membrane.

Supporting this assumption, results of this study show that chloroplast envelopes in green cotyledons of nyc1 sustained membrane damage, as evidenced by balloon-like chloroplasts (Figure 5). Direct observation of chloroplasts in unfixed leaf tissues enables us to infer that chloroplast swelling is indicative of hypotonic emphasize that disturbs envelope membrane potential. Originally, we used this method to ascertain the important role of VIPP1 in envelope integrity: vipp1-kd shows the typical balloon-like chloroplasts (Zhang et al., 2012). VIPP1 is found specifically in photosynthetic organisms and is a multifunctional protein originated from bacterial ancestral protein PspA. It has a C-terminal extension that does not exist in PspA, has no transmembrane helix, and associates with membranes by forming large complexes (Zhang and Sakamoto, 2015). The large homo complex consists of VIPP1 function particles that mutually associate, thereby forming various dot-like, rod-like, and lattice-like structures. Various functions such as vesicle formation, maintenance of photosystems, protein import across thylakoid membranes, envelope maintenance, and membrane fusion have been inferred. Along with the lethal phenotype of vipp1-ko, there observations suggest that VIPP1 plays a fundamental role in maintaining chloroplast membranes. Trans-complementation of nyc1 with VIPP1 overexpression also suggests an important role of VIPP1 in chloroplast maintenance.

We drew two conclusions from our current study. First, one target for photo-oxidative damage caused by impaired Chl degradation is chloroplast membranes, particularly during seedling formation. VIPP1 apparently has little effect on stay-green phenotype in nyc1 because membranes undergo extensive destruction during leaf senescence. In addition, the possibility that VIPP1 assists in degrading Chl is excluded because none of VIPP1-GFP/nyc1 or VIPP1-GFP/BCG recovered stay-green during leaf senescence. A second conclusion is that VIPP1 can prevent chloroplasts from receiving damage against oxidative stress. In this regard, trans-complementation of nyc1 by overexpressing VIPP1 is related to tolerance against heat shock in VIPP1-GFP/Col, as found from our earlier study. The present study therefore reinforces our notion that VIPP1 is important to protect chloroplast membranes against environmental stresses.

Stay-green is a key plant trait with wide usage for managing crop production. Functional stay-green can increase biomass and grain yield of crop through delayed senescence of the photosynthetic organs (Kusaba et al., 2013). Although cosmetic stay-green is unlikely to contribute to high yields, it is useful for high-quality food production (e.g., evergreen leaves, grasses). The shortcoming of Chl retention inside seeds appears to accompany a low germination rate (Parcy et al., 1997; Nakajima et al., 2012), which can be expected to hinder the practical application of stay-green trait. In this paper, we presented the possibility that VIPP1 overcomes such weakness when its overexpression is combined with stay-green.

WS, MK, AT designed the work; LZ, WS, AT prepared the materials; LZ and WS performed the experiments; LZ, WS, MK, AT analyzed the data; WS and LZ wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.