Mendel’s green cotyledon trait in pea (Pisum sativum) has been widely used to demonstrate the principle of inheritance and the single recessive gene responsible for the non-yellowing trait has intrigued scientists for many years. In 2006, Armstead and colleagues finally identified the gene controlling Mendel’s green cotyledon trait (Armstead et al., 2006), and it is now named STAYGREEN1 (also termed STAY-GREEN1; SGR1) or NONYELLOWING1 (NYE1) in Arabidopsis (Cha et al., 2002; Park et al., 2007; Ren et al., 2007). At about the same time, different groups also independently identified SGR1 homologs in Arabidopsis thaliana (Ren et al., 2007) and rice (Oryza sativa; Jiang et al., 2007; Park et al., 2007), and subsequent work isolated and characterized SGR1 homologs in various plant species, including tomato (Solanum lycopersicum; Barry et al., 2008), bell pepper (Capsicum annuum; Barry et al., 2008), tall fescue (Festuca arundinacea; Wei et al., 2011), Medicago truncatula (Zhou et al., 2011), and soybean (Glycine max; Fang et al., 2014) (Fig. 1A).

Early analysis indicated that SGR1 might function in chlorophyll (Chl) degradation, as: (i) sgr1 knockout mutants exhibit a common stay-green phenotype in several plant species, (ii) sgr1 mutants display a type C nonfunctional (cosmetic) stay-green phenotype, which affects Chl degradation, but not other aspects of senescence (Thomas and Howarth, 2000), and (iii) rice SGR1 physically interacts with light-harvesting complex subunits of photosystem II (LHCII) in vitro and in vivo (Park et al., 2007). Indeed, Park et al. (2007) examined whether SGR1 has enzymatic properties typical of Chl catabolic enzymes (CCEs) in the chloroplast, but found that SGR does not bind to Chl nor does it have chlorophyllase activity. This indicates that SGR1 does not function as a CCE in the chloroplast.

The function of other SGR homologs is another missing piece of the puzzle in the study of SGR1 (Fig. 1B). Previous phylogenetic analysis classified the SGR family of higher plants into two groups (Barry et al., 2008; H?rtensteiner, 2009; Sakuraba et al., 2014b). One group comprises the genuine SGR subfamily, and mutations in these SGR genes cause a type C nonfunctional stay-green phenotype, as described above. In the other group, the SGR-LIKE (SGRL) subfamily, the C-terminal sequence of SGRL proteins differs considerably from that of the SGR subfamily proteins (H?rtensteiner, 2009). To our knowledge, however, all higher plants contain at least one SGRL, indicating that SGRL has essential functions in green plants, probably acting in Chl breakdown. Furthermore, some plant species contain two or more SGRs; for example, Arabidopsis has SGR1 and SGR2, in addition to one SGRL (Barry et al., 2008; Sakuraba et al., 2014b). The physiological roles of these SGR homologs had remained unclear. However, after 2010, several studies have addressed the physiological and molecular roles of SGRs and revealed their functions.

Many studies have examined the functions of SGR1 and CCEs in dark-induced senescence (Kusaba et al., 2007; Meguro et al., 2011; Park et al., 2007; Pru?insk? et al., 2003; Sakuraba et al., 2012; Schelbert et al., 2009). Although SGR1 interacts with the subunits of LHCII in vivo (Park et al., 2007), how SGR activates Chl degradation during leaf senescence has remained unclear. Work in Arabidopsis and rice has identified six plastid-localized CCEs: Non-yellowing coloring1 (NYC1) (Horie et al., 2009; Kusaba et al., 2007) and NYC1-Like (NOL) (Sato et al., 2009) encoding two different Chl b reductases, 7-Hydroxymethyl chlorophyll a reductase (HCAR) (Meguro et al., 2011), Pheophytin pheophorbide hydrolase (PPH) (Morita et al., 2009; Schelbert et al., 2009), Pheophorbide a oxygenase (PAO) (Pru?insk? et al., 2003; Tanaka et al., 2003), and Red chlorophyll catabolite reductase (RCCR) (Pru?insk? et al., 2007). SGR1 physically interacts with all six CCEs (Sakuraba et al., 2012; 2013). Moreover, SGR1 and six CCEs specifically interact with LHCII proteins, but not with other photosystem components, such as LHCI or photosystem core proteins. These results indicate that SGR1, CCEs, and LHCII form dynamic, multi-protein complexes for Chl degradation during senescence. These multi-protein complexes may also be required for Chl detoxification in non-senescing green tissues. Colored intermediates of Chl breakdown, pheophorbide a (Pheide a) and red Chl catabolite (RCC), are potentially phototoxic, because deficiency of PAO or RCCR in pre-senescent leaves leads to accelerated cell death in Arabidopsis (Hirashima et al., 2009; Mach et al., 2001; Pru?insk? et al., 2003; 2007), maize (Gray et al., 1997; 2002), and rice (Tang et al., 2011). Thus, the SGR1-CCE-LHCII complex likely allows the metabolic channeling of potentially phototoxic Chl breakdown intermediates to minimize the risk of photodynamism caused by Chl breakdown intermediates during natural or dark-induced senescence (Fig. 2), and SGR1 has an important role in recruiting CCEs to LHCII (Sakuraba et al., 2012).

Chl breakdown also occurs in response to several abiotic and biotic stresses, in addition to senescence (Lim et al., 2007). Arabidopsis SGR1 and SGRL overexpressors exhibit an early leaf yellowing phenotype, while sgr1/nye1-1 and sgrl mutants display a stay-green phenotype under abiotic stress conditions, including high salinity, osmotic stress (mannitol), and drought stress conditions (Sakuraba et al., 2014a; 2014b; 2014c, Fig. 3). Furthermore, one of the Arabidopsis sgr1 alleles, termed noc1, develops less severe disease symptoms (leaf chlorosis and/or necrosis) than wild type after infection with the bacterial pathogen Pseudomonas syringae pv. tomato (Pst) DC3000 or the fungal pathogen Alternaria brassicicola (Mecey et al., 2011). Thus, SGR1, SGRL, and CCEs also function in leaf yellowing induced by biotic and abiotic stress.

SGRs and CCEs may also form a multi-protein complex for Chl degradation under biotic and abiotic stress conditions, similar to their interactions during senescence. Interestingly, transcript levels of SGR1, NYC1, PPH, and PAO dramatically increase during senescence, but transcripts of SGRL and NOL are more abundant in pre-senescent leaves during vegetative growth (Sakuraba et al., 2014b). Furthermore, SGRL interacts more strongly with NOL than with other CCEs (Sakuraba et al., 2014c). Thus, the composition of the SGRL-CCE complexes likely differs, with SGRL and NOL as the main components of the SGRL-CCE complex during stress-induced leaf yellowing and SGR1 and NYC1 as the main components during senescence (Fig. 2). The transcripts of HCAR and RCCR remain at nearly constant levels throughout development, including during natural senescence (Sakuraba et al., 2013), indicating that HCAR and RCCR participate in the SGR-CCE complex during both stress- and senescence-induced leaf yellowing (Fig. 2).

As described above, different plant species have different numbers of SGR homologs. For example, rice and barley (Hordeum vulgare) have just one SGR homolog, but Arabidopsis and soybean have two SGR homologs, SGR1 and SGR2 (Sakuraba et al., 2014b). Recent discoveries have shed light on the differing physiological functions and relationships of these pairs of SGR homologs. In soybean, the leaves and seed embryos of d1 d2 double mutants show a significant stay-green phenotype (Chao et al., 1995; Guiamet et al., 1991) and recent work showed that D1 and D2 encode GmSGR1 and GmSGR2, respectively (Fang et al., 2014). The GmSGR1 and GmSGR2 homologs resulted from one of the two whole-genome duplication events that occurred in soybean 13 million years ago. Interestingly, d1 leaves showed a much weaker stay-green phenotype than the d1 d2 leaves, and the d1 seeds turned yellow normally, indicating that the two GmSGRs function redundantly.

In contrast with the two GmSGRs, the two Arabidopsis SGRs have different functions during leaf yellowing. An Arabidopsis SGR2 overexpressor line shows a stay-green phenotype and sgr2 null mutants exhibit a leaf yellowing phenotype under dark-induced senescence and abiotic stress conditions (Sakuraba et al., 2014b; Fig. 3). By contrast, SGR1 overexpressor shows early leaf yellowing and sgr1-1/nye1-1 mutants exhibit a stay-green phenotype, opposite phenotypes to those of SGR2 overex- pressors and sgr2 mutants. SGR2 has a much weaker physical interaction with CCEs compared with the interaction of SGR1 with CCEs. Because SGR2 also can interact with LHCII, an excess of SGR2 may interrupt the formation of the SGR1-CCEs-LHCII complex. Furthermore, SGR1 and SGR2 can interact with each other to form homo- and heterodimers. These two interesting characteristics of SGR2, (i) its lower capacity to interact with CCEs and (ii) its ability to form heterodimers with SGR1, may cause its negative effect on Chl degradation during leaf senescence. Arabidopsis SGR1 and SGR2 likely acquired their antagonistic functions to balance Chl catabolism in senescing chloroplasts, thus preventing inappropriate Chl degradation (Fig. 2). Indeed, rice SGR1 and SGRL overexpressors showed a premature leaf-yellowing phenotype (Jiang et al., 2011; Rong et al., 2013) and Arabidopsis SGR1 and SGRL overexpressors showed a normal phenotype during vegetative growth (Sakuraba et al., 2012, 2014c). These findings suggest that this difference is partly caused by the existence of SGR2 in Arabidopsis. Thus, SGR homologs are not always positive regulators of Chl degradation, and this depends on the independent evolutionary processes that occurred in each species.

Proper seed maturation requires prompt degradation of Chl because any Chl remaining in mature seeds has severe, negative effects on seed storability and longevity (Clerkx et al., 2003; Johnson-Flanagan and Spencer, 1994). Although Arabidopsis sgr1 (SALK_070891) and sgr2 (SALK_003830) mutants produce normal yellow seeds, sgr1 sgr2 double mutants produce green seeds, indicating that Arabidopsis SGR1 and SGR2 function redundantly in seed degreening (Delmas et al., 2013). Knockout mutants of the abscisic acid (ABA) signaling-associated transcription factor ABA INSENSITIVE3 (ABI3) also produce green seeds (Clerkx et al., 2003; Koornneef et al., 1989) and electrophoretic mobility shift assays revealed that ABI3 directly binds to the promoters of SGR1 and SGR2 (Delmas et al., 2013). Furthermore, ectopic expression of SGR1 and SGR2 in abi3 knockout mutants rescues the phenotype from green to yellow seeds, indicating that SGR1 and SGR2 have pivotal roles in seed degreening during maturation. Arabidopsis nyc1 mutant seeds also contain green pigments (Nakajima et al., 2012), indicating that NYC1 also functions in seed degreening. Zhang et al. (2014) recently reported that SGR1 and SGR2 also affect the biosynthesis of tocopherol, a major component of Arabidopsis seeds (Sattler et al., 2004), during seed development; for example, the seeds of SGR1-overexpressing plants had significantly lower tocopherol contents. Thus, these results indicate that SGR1 and SGR2 have important roles in balancing Chl degradation and tocopherol biosynthesis during seed maturation.

Some aspects of seed degreening remain unclear. The seeds of pph mutants do not show a green phenotype, although the pph leaves show a stay-green phenotype (Shelbert et al., 2009). We also confirmed that two other mutants of CCEs, hcar and pao, have a stay-green phenotype but do not produce green seeds (unpublished). Furthermore, some CCE genes, such as HCAR, NOL, and RCCR have lower expression levels during seed maturation than during other developmental stages. Thus, it is not yet clear whether these CCEs function in seed degreening in Arabidopsis. Also, SGR2 acts as a negative regulator of leaf yellowing (Sakuraba et al., 2014b), but it promotes seed degreening. Thus, these results strongly suggest that Chl degradation mechanisms differ during seed degreening and leaf senescence.

Ripening of tomato fruits involves accumulation of carotenoids, including lycopene and β-carotene, along with a concomitant decrease in Chl levels (Fraser et al., 1994). A recent study indicated that tomato SGR1 (Solanum lycopersicum SGR1; SlSGR1) balances these two major events of fruit ripening. In SlSGR1 RNAi fruits, three kinds of carotenoids (β-carotene, lycopene, and phytoene) accumulate to higher levels than in non-transgenic tomato fruits (Luo et al., 2013), indicating that SlSGR1 affects the regulation of carotenoid accumulation in ripening tomato fruits. Furthermore, SlSGR1 physically interacts with the carotenoid biosynthesis enzyme phytoene synthase 1 (PSY1), which has an important role in tomato fruit nutrition (Fraser et al., 1994; 2004). Based on these results, the authors proposed that the physical interaction of SlSGR1 and PSY1 leads to a decrease in PSY1 activity and thus causes reduced lycopene accumulation. This study showed the physical relationship between SGR1 and PSY1 in tomato fruits, prompting the intriguing question of whether SGR1 interacts with PSY1 or other carotenoid biosynthesis enzymes to regulate carotenoid accumulation in other tissues, such as pre-senescent leaves, or in other plant species. During leaf senescence, the level of carotenoids declines as Chl is degraded (Biswal, 1995), indicating the activation of carotenoid degradation pathways and inhibition of carotenoid biosynthesis. Thus, SGR1 may have pivotal functions in the regulation of carotenoid accumulation during leaf senescence.

All SGR family proteins in higher plants are predicted to localize to chloroplasts (Barry et al., 2008), indicating that they likely function in plastids, most likely in Chl degradation. However, in addition to its role in senescing chloroplasts, Medicago truncatula SGR (MtSGR) also functions in nodule senescence in roots (Zhou et al., 2011). MtSGR expression is higher in senescing nodules than any other organs, including senescent leaves. Furthermore, the nodules of Medicago sgr mutants (termed NF2089) show significant down-regulation of several nodule senescence-associated genes, indicating that MtSGR affects nodule senescence in legumes. Nodule cells contain other plastid types, such as leucoplasts, and it would be an intriguing idea (and likely) that SGRs do not only occur in chloroplasts, but generally occur in plastids where they execute their diverse functions. Thus, their role in fruit carotenoid biosynthesis likely involves chromoplasts. Functional analyses of MtSGR in nodule senescence of Medicago and SlSGR1 in fruit ripening of tomato will expand our understanding of the potential roles of SGRs in physiological processes beyond Chl degradation.

Although most studies of SGR have examined its biochemical and physiological functions, recent work in Arabidopsis and rice has identified several transcription factors that directly promote SGR expression. As described above, ABI3 binds to the canonical B3 domain-binding RY motif (CATGCA) in the promoters of SGR1 and SGR2 during seed degreening (Delmas et al., 2013). Recent work showed that Phytochrome Interacting Factor 4 (PIF4) and PIF5 strongly activate SGR1 expression during dark-induced senescence (Sakuraba et al., 2014b). In the PIF4/PIF5-dependent leaf senescence cascade, ABI5 and ENHANCED EM LEVEL (EEL), direct targets downstream of PIF4/PIF5, bind to the ABA-Responsive Element (ABRE) motif (YACGT) in the SGR1 promoter (Sakuraba et al., 2014d). Notably, ABI3, ABI5, and EEL are ABA signaling-associated transcription factors (Giraudat et al., 1992; Jakoby et al., 2002), and ABRE-Binding Factor 2/3/4 (ABF2/3/4) bZIPs also induced SGR1 expression (Fujita et al., 2009), indicating that ABA signaling has an important role in the activation of SGR1 expression and Chl degradation. Interestingly, NYC1 is also directly activated by ABA-signaling transcription factors, including ABI5 and EEL (Sakuraba et al., 2014d) during dark-induced senescence and ABF4 during seed degreening (Nakajima et al., 2012). Furthermore, the OsNAP transcription factor, which is an ABA signaling-associated gene, directly activates the transcription of both SGR1 and NYC1 in rice (Liang et al., 2014). Thus, several ABA-signaling transcription factors likely function to co-activate SGR1 and NYC1, and this close transcriptional relationship between SGR1 and NYC1 is important for Chl degradation during both leaf senescence and seed degreening.

Recent functional studies of SGR1 and its homologs revealed that SGR1 can interact with various chloroplast proteins, including LHCII proteins, all known CCEs, and PSY1 (Luo et al., 2013; Sakuraba et al., 2012; 2013). However, the amino acids in SGR that affect these physical interactions with other proteins remain to be identified. Park et al. found that V99 is important for OsSGR1 function because the rice sgr mutant has a V99M missense mutation, but this amino acid substitution does not affect the interaction of OsSGR1 with LHCII (Park et al., 2007). More detailed analyses, such as the combination of site-directed mutagenesis and protein-protein interaction assays, will reveal more about the biochemical function of SGR.

SGR1 could interact with other, unknown proteins. The most intriguing candidate for a potential SGR1-interactor is metal-chelating substance (MCS), which is considered to be required for the removal of the Mg atom of Chl a during Chl breakdown, but remains to be identified (H?rtensteiner and Krautler, 2011). Because all of the six known CCEs physically interact with SGR1 (Sakuraba et al., 2012; 2013), MCS likely also interacts with SGR1 and other CCEs as an essential factor in the multi-protein complexes. Another candidate SGR1-interactor is Thylakoid Formation 1 (THF1), a homolog of psb29 identified in the cyanobacterium Synechocystis sp PCC6803 (Wang et al., 2004). THF1 is closely associated with Chl breakdown because both Arabidopsis and rice thf1 mutants exhibit a stay-green phenotype (Huang et al., 2013; Yamatani et al., 2013). Furthermore, THF1 physically interacts with all LHCII subunit proteins (Lhcb1 to Lhcb6; Huang et al., 2013). Thus, it is possible that THF1 is near SGR1 and CCEs in LHCII, possibly by physical interactions, during Chl breakdown. The subunits of chloroplast proteases are also candidate SGR-interacting proteins. Chl and photosystem apoproteins should be degraded at almost the same stage, and chloroplast proteases likely degrade apoproteins. It is well known that chloroplast-localized FtsH protease participates in the degradation of D1 protein, a subunit of photosystem II, under stress conditions (Kato and Sakamoto, 2009). Glutamyl endopeptidase, a chloroplast-localized serine protease, is proposed to be involved in the degradation of Lhcb1 (Forsberg et al., 2005). Furthermore, the transcripts of several subunits of Clp protease are most abundant in the chloroplasts of senescing leaves (Olinares et al., 2011). This indicates that Clp protease also has an important role in degradation of chloroplast proteins, possibly including CCEs, during leaf senescence. Indeed, Arabidopsis Clp protease is involved in destabilization of the Chl biosynthetic enzyme chlorophyllide a oxygenase (Nakagawara et al., 2007). Collectively, further proteomic analyses, such as mass spectrometric analysis of pull-down fractions, using epitope-tagged SGR1 protein, will provide good approaches to find as-yet-unidentified SGR1-interacting proteins.