The Component of the TAC Complex, TCD7, Controls Rice Chloroplast Development at the Early Seedling Stage under Cold Stress

Transcriptionally active chromosome (TAC) is a component of protein-DNA complexes with RNA polymerase activity found in chloroplasts. Although TAC in Arabidopsis thaliana has been extensively investigated, how the rice (Oryza sativa L.) TAC complex functions remains largely unknown. We report the characterization of the mutant thermosensitive chlorophyll-decient7 (tcd7) and the cloning of TCD7. tcd7 mutant seedlings displayed an albino phenotype specically at low temperatures and before the four-leaf stage. We identied TCD7 by map-based cloning followed by transgenic rescue and genome editing tests, showing that TCD7 encodes the putative TAC component FRUCTOKINASE-LIKE 2 (OsFLN2). TCD7 transcripts were highly abundant in green tissues, and the protein localized to chloroplasts. In agreement with the albino phenotype, transcript levels of genes controlling chloroplast development and the establishment of photosynthetic capacity were severely reduced in tcd7 seedlings at low temperatures, but were expressed as in the wild type at high temperatures, implying that TCD7 regulates the PEP pathway and chloroplast development. Moreover, TCD7 interacted with the thioredoxin OsTRXz to form an OsTRXz-TCD7 regulatory module, which might regulate plastid transcription under cold stress. Our results demonstrate that the nucleus-encoded TAC protein TCD7 protects chloroplast development from cold stress via a TRXz-FLN regulatory module.


Introduction
The chloroplast is a semi-autonomous organelle whose genome encodes 100-250 proteins associated with photosynthesis and other physiological and biochemical functions (Wicke et al. 2011). The expression of chloroplast genes relies on two distinct RNA polymerases: a nucleus-encoded phage-type RNA polymerase (NEP) and a plastid-encoded prokaryotic-type RNA polymerase (PEP) (Toyoshima et al. 2005). NEP is mainly responsible for modulating the transcription of plastid housekeeping genes during early chloroplast development, including those encoding PEP core subunits, as well as genes involved in chloroplast functions. PEP activity is itself required for fully active chloroplasts, as it controls the expression of photosynthesis genes (Hajdukiewicz et al. 1997). PEP core proteins can be detected in two chloroplast protein fractions: the soluble RNA polymerase (sRNAP), and the insoluble fraction known as transcriptionally active chromosome (TAC) (Krause et al. 2000).
TAC proteins play essential roles in chloroplast development (Pfalz et al. 2006;Steiner et al., 2011;Pfalz and Pfannschmidt 2013), as revealed by studies in Arabidopsis (Arabidopsis thaliana). For instance, pTAC3 is a chloroplast-localized transcription factor that mediates DNA/RNA binding within the PEP complex (Yagi et al. 2012), while pTAC4 contributes to the formation of the thylakoid membrane (Zhang et al. 2012;Feng et al. 2014). Other sets of TAC proteins, such as pTAC10, pTAC12, pTAC6 and pTAC7, participate in PEP-mediated plastid gene expression and promote chloroplast development (Peter and Åsa, 2015;Chang et al. 2017). pTAC8 (also named PSI-P) is the P subunit of photosystem I (PSI) and in uences the formation of thylakoid grana (Armbruster et al. 2013; Hansson et al. 2003). Likewise, pTAC5 plays an important role during chloroplast development under heat stress (Zhong et al. 2013).
Importantly, the components pTAC12 and pTAC14 coordinate light signaling with chloroplast development, as ptac12 and ptac14 mutants are albino . The TAC fraction is characterized by protein-protein interactions between the subunit pTAC10 and other TACs, such as FE SUPEROXIDE DISMUTASE 2 (FSD2) and FSD3, THIOREDOXIN Z (TRXz), pTAC7 and pTAC14, to regulate chloroplast development (Chang et al. 2017). Arabidopsis FRUCTOKINASE-LIKE PROTEIN2 (FLN2) interacts with FLN1, another member of the fructokinase-like (pfkB) protein family, and forms heterodimers that interacts with TRXz to regulate PEP activity. Notably, FLN2 also interacts with pTAC5, and the n2-2 mutant is albino, although the mutant returns to a more wild-type appearance and accumulates chlorophyll upon exogenous application of 2% sucrose . Although much of the work on TAC components has been carried out in Arabidopsis, whether and how rice (Oryza sativa) TAC genes in uence chloroplast development, especially under cold stress, remains largely unclear. Low temperatures can severely and negatively affect rice production by targeting a range of cellular components (e.g. chloroplasts) and metabolic pathways (e.g. photosynthesis) and thus reducing yield. To date, only a few TAC genes have been reported in rice: OspTAC2, TCM1 and WHITE LEAF AND PANICLE2 (WLP2). OspTAC2 encodes a protein homologous to Arabidopsis TAC2. As in Arabidopsis, the rice ptac2 mutant is albino and seedling lethal (Wang et al. 2016a), Similarly, TCM1 is the rice homolog to Arabidopsis TAC12, and its mutation leads to chloroplast damage in seedlings exposed to cold stress (Lin et al. 2018), WLP2 (also named OsFLN1) is homologous to Arabidopsis FLN1, and its mutation results in seedlings with white leaves and panicles speci cally under heat stress (Lv et al. 2017). To our knowledge, thermo-sensitive chlorophyll-de cient mutants, occurred only in speci c early-leaves, have not been reported in rice yet. Herein, we describe thermo-sensitive chlorophyll-de cient 7 (tcd7), a rice albino mutant with a conditional phenotype that affects leaves before the four-leaf stage and only under cold stress. We cloned TCD7, which encodes a pfkB family protein that is a constituent of the rice TAC complex. In addition, we show that the transcript levels of genes associated with chlorophyll biosynthesis, photosynthesis and chloroplast development are severely impaired in tcd7 mutants before the four-leaf stage during exposure to cold stress. Our work indicates that rice TCD7 plays an important role in chloroplast development and maintenance of PEP activity under speci c conditions.

Results
Phenotypic Characterization of the tcd7 Mutant Seedlings for the wild-type japonica rice variety Jiahua 1 and the tcd7 mutant were grown at four distinct temperatures (20ºC, 24ºC, 28ºC and 32ºC). Seedlings from WT displayed normal green leaves regardless of the growth temperature and leaf-stages, as expected. In sharp contrast, the tcd7 mutant produced albino leaves when grown at 20ºC, up to the fourth leaf. Surprisingly, all subsequent leaves starting with the fth leaf took on a normal appearance, similar to WT seedlings (Fig. 1A). In addition, the albino phenotype seen in the tcd7 mutant was conditional, as we observed white leaves only when seedlings were grown at 20ºC, but not at 24ºC, 28ºC or 32ºC (Fig. 1B-D). We failed to rescue the albino phenotype when tcd7 seedlings were transferred from 20ºC to 32°C, as third and fourth white leaves did not turn green (data not shown). These results indicated that the mutant phenotype is temperature sensitive during the early seedling stage.
In agreement with our visual phenotypic assessment, seedling contents of the pigments chlorophyll a (Chl a), chlorophyll b (Chl b) and carotenoids (Car) before the four-leaf stage were much lower in white tcd7 leaves than in WT seedlings when grown at 20°C ( Fig. 2A), but were similar to WT (Fig. 2B) at the higher growth temperature of 32°C. This result suggested that chlorophyll accumulation in younger tcd7 seedling may be blocked under cold stress. To determine whether the observed reduction of photosynthetic pigments in tcd7 seedlings at 20ºC might be due to altered chloroplast ultrastructure, we observed chloroplasts in seedlings of both genotypes grown at 20ºC or 32ºC by transmission electron microscopy (TEM). WT mesophyll cells contained numerous uniform chloroplasts, regardless of the growth temperature (Fig. 2C, 2D). By contrast, leaves from tcd7 seedlings grown at 20°C were characterized by far fewer grana and abnormal grana structure (Fig. 2E), whereas chloroplasts in leaves from tcd7 mutant seedlings grown at 32ºC had normal morphology, consistent with their lack of a distinct phenotype at this temperature (Fig. 2F). We therefore presume that the abnormal chloroplast structures seen in seedlings grown at 20°C may lead to reduced chlorophyll content in the mutant.

Map-based Cloning of TCD7
To identify TCD7, we generated a mapping population by crossing tcd7 plants with the polymorphic variety Peiai64S (indica). Leaves of F1 hybrid seedlings were a healthy green, indicating that tcd7 is a recessive mutation. In the subsequent F2 generation, the mutant phenotype segregated as a monogenic recessive Mendelian trait, with a wild-type to mutant phenotype ratio of 3:1 (Additional le 1: Table S1, χ2 = 0.21 χ20.05 = 3.84). We then selected 94 F2 seedlings with the mutant phenotype for initial mapping, which placed TCD7 between markers MM3645 and MM3833 on chromosome 3 (Fig. 3A, B). We then increased the size of the mapping population to 624 F2 seedlings with the mutant phenotype for ne mapping, allowing us to narrow the mapping interval containing TCD7 to a 123-kb region between markers ID14867 and RM15419. This interval was covered by four bacterial arti cial chromosomes (BACs AC097276, AC092778, AC097276 and AC109601) and contained six predicted genes (Fig. 3C). We ampli ed all candidate genes by PCR from tcd7 genomic DNA and sequenced all PCR amplicons to detect a polymorphism relative to the wild-type sequence: only LOC_Os03g40550, which encodes the TAC protein most similar to Arabidopsis FLN2 (thus also named OsFLN2 in this study), carried a 9-bp (GTTG CTCTT) deletion in exon 3, 775 bp downstream from the translation start codon (Fig. 3D, 3E). Because of its length, this deletion does not disrupt the open reading frame of OsFLN2/TCD7, but results in a predicted protein lacking three amino acids (Val, Ala, Leu) compared to WT.

Complementation and Genome Editing of TCD7
To assess whether the mutation we identi ed in LOC_Os03g40550 might be responsible for the tcd7 mutant phenotype under cold stress, we used two complementary approaches. First, we tested for rescue of the albino phenotype upon introduction of a functional copy of LOC_Os03g40550 in the tcd7 mutant, driven by the cauli ower mosaic virus (CaMV) 35S promoter. All T0 transgenic plants produced only green leaves when grown at 20ºC (Fig. 4B), whereas independent T0 transgenic plants transformed with the empty vector pCAMBIA1301 retained the mutant phenotype (Fig. 4A). During the subsequent T1 generation, T1 plants with green leaves also carried a copy of the transgene when grown at 20ºC, whereas none of the T1 plants with albino leaves did (Fig. 4C), indicating that the overexpression of LOC_Os03g40550 in the tcd7 mutant background rescued the albino phenotype characteristic of the mutant. Independently, we generated two homozygous transgenic T0 plants by CRISPR/Cas9 genome editing. One edited plant harbored the same 9-bp (CATCAGAAG) deletion detected in TCD7; we designated this line as T0-a. The other edited plant, which we designated T0-b, carried multiple deletions and mutations in TCD7 (Additional le 2: Fig. S1). Importantly, both T0 plants exhibited the same albino phenotype as the tcd7 mutant when grown at 20ºC (Fig. 4D). T0 plants heterozygous at these two edited sites were fully green, indicating that both mutations are recessive, like the original tcd7 mutation. Moreover, the T1 progeny of these heterozygous edited T0 transgenic plants segregated for the albino phenotype when grown at 20°C (Fig. 4E, 4F). Taken together, these results con rm that LOC_Os03g40550 is TCD7.

Analysis of TCD7 Expression and Subcellular Localization of TCD7
We next determined the expression pattern of TCD7 in various tissues in wild-type seedlings by semiquantitative RT-PCR. TCD7 was highly expressed in leaves (second, third and ag leaves), but much less expressed in roots, stems or panicles (Fig. 5A). These results were consistent with the rice gene expression atlas available in the RiceXPro database (Additional le 2: Fig. S2), and were broadly in agreement with a tissue-speci c expression pattern and a vital role in leaf chloroplast development for TCD7.
As TCD7, one of the subunits of the TAC complex, we investigated its subcellular localization. The TargetP 1.1 Server predicted that TCD7 localizes to the chloroplast (http://www.cbs.dtu.dk/services/TargetP/) (Emanuelsson et al. 2000). To test this hypothesis, we fused the rst 149 amino acids of TCD7, including any potential signal targeting sequence, to the N terminus of green uorescent protein (GFP). The resulting clone was driven by the 35S promoter and transiently expressed in tobacco protoplasts. Confocal microscopy revealed co-localization of GFP uorescence and chlorophyll auto-uorescence (Fig.5B), demonstrating that TCD7 is a chloroplast-localized protein.

Characterization of the TCD7 Protein
The TCD7 locus consists of ve exons and six introns (Fig. 3D) and encodes a protein of 589 amino acids with a predicted molecular mass of about 65 kDa. A search of the Pfam database revealed that TCD7 belongs to the transcriptionally active chromosome (TAC) complex and contains a pfkb domain (Additional le 2: Fig. S3). In addition, the tcd7 mutation results in the deletion of three conserved amino acids (Val, Ala, Leu) within the pfkb domain, which is predicted to disrupt the α-helical structure of TCD7, possibly leading to a compromised overall structure (Additional le 2: Fig. S3B). TCD7 is highly conserved across land plants, and was closest to its putative ortholog from the monocot purple false brome (Brachypodium distachyon) (Fig. 6B). Notably, TCD7/OsFLN2 shared only 55% identity with
Although Cab1R (encoding the light harvesting Chla/b-binding protein) and RbcS (encoding the small subunit of Rubisco) showed 50-70% higher expression levels in the tcd7 mutant relative to WT, other photosynthesis-related genes(rbcL, psaA, psbA and LhcpII ) were expressed at levels 30% of WT or below in the tcd7 mutant, suggesting that tcd7 should also hinder photosynthesis (Fig. 7B). Of 16 genes involved in chloroplast development, eight (rpoA, ropB, rps7, rps20, V1, V2, OsV4 and petA) were expressed at low levels at low temperatures (Fig. 7C), which may lead to malformed chloroplasts (Fig.   2E). The PEP subunits rpoA and ropB (Kusumi et al. 2011) were especially downregulated in tcd7, with expression levels only reaching about 20% of WT levels. Notably, the downregulation of gene expression noted above in tcd7 seedlings grown at 20ºC was largely abrogated when the seedlings were grown at 32ºC (Fig. 8), with gene expression levels now more comparable between the WT and the mutant.
To explore the molecular mechanism behind the stage speci city of the albino phenotype exhibited by the tcd7 mutant, we next determined the relative transcript levels of PORA, HEMA1, CAO1, rbcL, psaA, psbA and rpoA in WT and mutant seedlings grown at 32ºC until the fourth-or fth-leaf stage. Although all genes were signi cantly downregulated in the fourth leaves (Fig. 9A, 9B), their relative expression returned to nearly normal levels in the fth leaves (Fig. 9C, 9D). We then turned to the PSI protein PsaA and the PSII protein D1, encoded by psaA and psbA, respectively. Indeed, we had established that expression of psaA and psbA was low in the third (Fig. 7B) and fourth leaves (Fig. 9B) of the tcd7 mutant at 20ºC, but was largely normal in seedlings grown at 32ºC (Fig. 8B) and in the fth mutant leaf from seedlings grown at 20ºC (Fig. 9D). The accumulation patterns of PsaA and D1 were largely congruent with the levels of their corresponding transcripts, with very little protein detected in the fourth leaves of seedlings grown at 20ºC (Fig. 9E), but comparable accumulation in the fth leaves of WT and tcd7 seedlings grown at 32ºC. In conclusion, these results suggest that the altered expression of chlorophyll biosynthesis and chloroplast biogenesis genes may contribute to the albino phenotype seen in the mutant until the fourleaf stage under cold stress.

TCD7 Interacts with OsTRXz in Yeast
Arabidopsis FLN2, which shares 55% identity with TCD7, interacts with TRXz and FLN1 (Arsova et al. 2010. The putative rice ortholog to Arabidopsis TRXz is OsTRXz, encoded by LOC_Os08g29110, while AtFLN1 is most similar to OsFLN1, also named WPL2, which is encoded by LOC_Os01g63220 (Lv et al. 2017). We therefore tested whether TCD7 might interact with OsTRXz and OsFLN1 in a yeast-two hybrid assay. We removed the chloroplast targeting sequences from all proteins during cloning. Unexpectedly, TCD7, interacted only with OsTRXz, and not with OsFLN1, suggesting that TCD7/OsFLN2 and Arabidopsis FLN2 may function via distinct mechanisms (Fig. 10).
A possible biochemical basis for the tcd7 mutant phenotype may stem from a disrupted interaction between TCD7 and its partner OsTRXz. However, the loss of the three amino acids in FLN2tcd7 did not affect the TCD7-OsTRXz interaction (Fig. 10). Structurally, the mutated TCD7, lacking three amino acids in tcd7 mutants, could still interact with OsCITRXz (Fig. 10A), indicating that the tcd7 mutation may affect the function much more than the structure of rice FLN2. tcd7 Mutant Phenotype is not due to Insu cient Energy Supply Arabidopsis n2 mutants display an albino and seedling-lethal phenotype at any growth temperature; however, chlorophyll accumulation can be partially rescued by exogenous supplementation with 2% sucrose, as evidenced by delayed greening . We therefore wondered whether insu cient sucrose or energy supply might underlie the albino phenotype of tcd7 mutant seedlings exposed to cold stress before the fth-leaf stage. However, no amount of exogenous sucrose from 2% to 8% succeeded in turning tcd7 albino seedlings green at the third-or fourth-leaf stage when grown at 22ºC (Fig. 11). On the contrary, higher sucrose concentrations appeared to inhibit seedling growth. We conclude that the albino phenotype is not the result of insu cient energy (sucrose) supply in tcd7 mutant seedlings.

Discussion
In this study, we described the albino mutant tcd7, which is characterized by an absence of chlorophyll accumulation and disrupted chloroplast development up to the fourth leaf when grown at the lower temperature of 20ºC. We cloned TCD7 by map-based cloning and determined that it encodes a chloroplast-localized component of the TAC complex with a pfkB domain. The deletion of three amino acids within the TCD7 pfkB domain resulted in a thermo-sensitive virescent phenotype, and drastically reduced the expression levels of genes associated with chlorophyll biosynthesis, photosynthesis and chloroplast development before the fourth-leaf stage under cold stress (Fig. 7). This study demonstrates that rice TCD7 plays important roles during early chloroplast development under cold stress. functions in the third step. Cold stress thus likely affects all three steps of chloroplast development, via multiple regulatory pathways. During exposure to cold temperatures, the tcd7 mutation severely repressed the expression of genes involved in the second (rpoA, rpoB, OsV4, V1, V2) and third (rbcL, psaA, psbA) steps of chloroplast development (Fig. 7), but not the rst step (e.g. OsRpotp, FtsZ, TSV3). Thus, TCD7 may function during the second step, and later in the third phase as a logical progression of altered chloroplast development. In addition, the expression of all genes that were repressed under cold stress were restored to nearly WT levels when the plants were grown at higher temperatures, providing a molecular basis for the thermosensitivity of the mutant (Fig. 1). It is possible that another protein replaces TCD7 to carry out its function past the fourth leaf, which would explain the normal chloroplast development beyond this stage. The TCD7 paralog PFKB1 (Zhu et al., 2020) might ful ll such a role.

TCD7 is a Component of the TAC Complex Regulating PEP Activity and Photosynthesis under Cold Stress
Transcriptionally active chromosomes (TACs) are involved in replication, transcription, translation, detoxi cation, protein modi cation and plastid metabolism (Pfalz et al. 2006). Arabidopsis mutations in the TAC subunit FLN2 (encoded by At1g69200) lead to an albino, pale-green or seedling-lethal phenotype resulting from defective chloroplast development in a temperature-independent manner (Arsova et al.  (Lin et al. 2018b) are reported to regulate PEP activity in a temperaturedependent manner, although in opposite directions. Indeed, the wpl2 mutant had an albino phenotype under heat stress (Lv et al. 2017), whereas the tcm1 mutant (like the tcd7 mutant) was characterized by an albino phenotype under cold stress (Lin et al. 2018b). In view of the reduced expression levels of rpoA, rpoB and PEP-dependent plastid genes (rbcL, psaA and psbA) (Fig. 7C, Fig. 10B), and the reduced PsaA (PSI) and D1 (PSII) protein abundance (Fig. 9E), we conclude that TCD7 regulates PEP activity and photosynthesis under cold stress.

The rice TRX-FLN Regulatory Module Mediates PEP Activity and Chloroplast Development under Cold Stress
As stated above, our data clearly indicated that TCD7/OsFLN2 is involved in the second step of chloroplast development by modulating PEP activity under cold stress. A hypothetical model for TCD7 function is shown in Fig. 12. PEP is the dominant RNA polymerase in chloroplasts, with a complex and still elusive regulatory mechanism (Kremnev and Strand 2014;Pfalzet al. 2006). In Arabidopsis, TRXz and its two interactors FLN1 and FLN2 form a complex that regulates PEP activity during PEP-dependent transcription (Arsova et al. 2010). The Arabidopsis n1 and n2 mutants have an albino phenotype at all growth temperatures, with FLN1 possibly playing a more important role than FLN2 (Arsova et al. 2010;Gilkerson et al. 2012). In addition, chlorophyll accumulation can be partially restored in n2 mutant seedlings upon provision of exogenous sucrose . Importantly, the expression of neither Arabidopsis FLN1 nor FLN2 was regulated by temperature. By contrast, the rice tcd7 (this study) and wlp2 mutants (Lv et al. 2017) had an albino phenotype only at speci c temperatures, raising the possibility that TCD7 and WLP2 expression might respond to temperature. Notably, the expression of PFKB1, a TCD7 paralog, regulates the expression of PEP-encoding genes irrespective of the temperature and may therefore compensate for the loss of TCD7 when its expression subsides (Zhu et al. 2020).
Multiple lines of evidence suggest that TCD7 and Arabidopsis FLN2 may function via at least partially distinct mechanisms. First, WLP2/OsFLN1 was shown to interact with OsTRXz to regulate PEP activity under cold stress (Lv et al. 2017); Arabidopsis FLN1 can also interact with FLN2 ). In sharp contrast, we found here that TCD7/OsFLN2 did not physically interact with WLP2 (OsFLN1), which is consistent with previous results (Lv et al. 2017). Furthermore, WLP2/OsFLN1 can stabilize OsTRXz to mediate regulation of chloroplast development in the face of heat stress (Lv et al. 2017), whereas TCD7/OsFLN2 appears to stabilize OsTRXz to regulate chloroplast development during exposure to cold (Fig. 10A). Second, adding sucrose to the growth medium did not rescue the albino phenotype of the tcd7 mutant (Fig. 11), unlike for the Arabidopsis n2 mutant. Third, TCD7/OsFLN2 has only 55% identity with Arabidopsis FLN2 (Additional le 2: Fig.S 4), which might be expected to correspond to distinct functions. It is worth noting here that the transcript levels of OsFLN1 and OsFLN2 have been incorrectly assumed to maintain a relative dynamic balance to regulate PEP activity and that OsFLN2 may partially compensate for loss of OsFLN1 function (Lv et al. 2017) due to no mutants of OsFLN2 as tcd7 mutant.
Protecting PEP activity and chloroplast development from cold or heat stress is essential and is mediated via the coordination of the OsTRXz-OsFLN complex. Undoubtedly, additional pathways may also regulate or protect chloroplast development in rice. For instance, the other TAC subunit, TCM1, regulates PEP activity under cold stress, although whether TCM1 can interact with OsTRX remains to be tested. Likewise, the Obg-like GTPase protein TSV3 regulates the biogenesis of the chloroplast ribosome 50S subunit under cold stress (Lin et al. 2018a). Importantly, the putative chloroplast-localized oxidoreductase TSV is also reported to interact with OsTRXz and form a OsTRXz-TSV module for the protection of chloroplast development from cold stress (Sun et al. 2017). One attractive possibility therefore calls upon OsTRXz as a central signaling hub that is absolutely critical for chloroplast development, especially under heat or cold stress, as a function of its interacting partners. Gaining a better understanding of TCD7 function and how TACs interact with OsTRXz to coordinate chloroplast development according to cell type, developmental stage or environmental conditions merits further investigation.

Conclusion
The TCD7 encodes chloroplast-targeted TAC protein which is important for chloroplast development and regulating PEP activity and photosynthesis, and its disruption would lead to a defective chloroplast at early rice seedling stage under cold stress. Our results demonstrate that the TCD7 protein protects chloroplast development from cold stress via a TRXz-FLN regulatory module.

Plant materials and Growth Conditions
We identi ed the thermo-sensitive chlorophyll-de cient mutant tcd7 in our mutant pool in the japonica rice variety Jiahua 1 exposed to 60Co γ-radiation. An F2 mapping population was generated by crossing the tcd7 mutant and the indica variety Peiai64S. The mutant phenotype was easily distinguished from the wild type at the early seeding stage at Hainan island, China (winter season, subtropical climate) and Shanghai, China (early spring season, temperate climate) under local growth conditions. All rice plants were cultivated in paddy elds. Jiahua 1 (wild type, WT) and tcd7 mutant seedlings were also grown in growth incubators (12h light/12 h dark; light intensity 180 μmol m-2 s-1) at a constant temperature of either 20ºC, 24ºC, 28ºC or 32ºC for phenotypic characterization, pigment analysis, and DNA and RNA extraction.

Analysis of Photosynthetic Pigments and Transmission Electron Microscopy
Contents of leaf photosynthetic pigments were determined according to the methods of Arnon (1949) and Wellburn (1994), with slight modi cations. Brie y, fresh leaves of tcd7 and WT seedlings were collected, ground to a powder in liquid nitrogen and incubated in chlorophyll extraction solution (ethanol:acetone:water 5:4:1 [v/v/v]). The absorbance of the extracts was measured at 470 nm, 645 nm and 663 nm with a METASH-UV5100 spectrophotometer. Results are reported as the average of three parallel experiments.
Chloroplast structures were observed from the third leaf of tcd7 and WT seedlings grown at 20ºC (restrictive temperature for the mutant phenotype) and 32ºC (permissive temperature) using transmission electron microscopy (TEM) as previously described ) with minor modi cations. Samples were visualized with a Hitachi765 (Hitachi, Tokyo) transmission electron microscope.

Mapping and Cloning of TCD7
Genomic DNA from fresh rice leaves was extracted by the CTAB method, as described previously (Murray and Thompson 1980). To ne-map the TCD7 locus, 624 individuals with the mutant phenotype were selected from the F2 mapping population and genotyped with simple sequence repeat (SSR) or InDel markers (Additional le1: Table S2). Candidate genes within the nal mapping interval were then PCRampli ed and sequenced (SinoGenoMax, Shanghai, China). Last, the functions and open reading frames for candidate genes were obtained from the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/). Conserved domains were predicted using the online tool SMART (http://smart.embl-heidelberg.de/).

Genetic Complementation of the tcd7 Mutant
Total RNA was extracted from WT seedlings and transcribed into rst-strand cDNAs for use as PCR template. Then, a 1.8-kb cDNA corresponding to LOC_Os03g40550 was PCR-ampli ed with the primers 5 -GGGGTACCATGCACCGAATGGCTTCTCTT-3 (underlined sequences indicate a KpnI site) and 5 -GCGTCGACTCACTCCACATATAAAAGCTCACTC-3 (underlined sequences indicate a SalI site). The resulting PCR amplicon was then digested with KpnI and SalI before being subcloned into the expression vector pCAMBIA1301-CaMV35S to generate the binary vector pCAMBIA1301-CaMV35S-TCD7(cDNA). The pCAMBIA1301-CaMV35S-TCD7(cDNA) construct was then introduced into Agrobacterium (Agrobacterium tumefaciens) strain EHA105 ahead of Agrobacterium-mediated transformation of the tcd7 mutant, as described previously (Hiei et al. 1994), with the temperature used for in vitro plant culture set to 20ºC. The genotype of transgenic seedlings was determined by PCR ampli cation of the hygromycin phosphotransferase (HPT) gene with the primers HPTF (5 -GGAGCATATACGCCCGGAGT-3 ) and HPTR (5 -GTTTATCGGCACTTTGCATCG-3 ) and primers GUSF (5 -GGGATCCATCGCAGCGTAATG-3 ) and GUSR (5 -GCCGACAGCAGCAGTTTCATC-3 ) as selection markers. The resulting T0 transgenic plants were grown in a paddy eld after screening for hygromycin resistance and con rmation by DNA sequencing. After T1 seeds were collected from T0 plants, all T1 transgenic seedlings were grown at 20ºC and used to score the segregation ratio of the mutant phenotype.

Genome editing of TCD7
As an independent validation of LOC_Os03g40550 as the causal gene for the tcd7 mutant, the genomic locus was edited via clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPRassociated nuclease (Cas9) genome editing in WT plants. The targeting guide RNAs (gRNAs) for LOC_Os03g40550 were designed using a CRISPR Primer Designer software (http://www.crispr.dbcls.jp/) (Ma et al., 2015) and are listed in Additional le 1: Table S3. The recognition sequences were cloned between the OsU6 promoter and the gRNA scaffold (from the pYLgRNA-OsU6 vector) in the Cas9 expression backbone vector pYL-CRISPR/Cas9-MH at the BsaI sites, as previously described (Ma et al. 2015). The resulting construct (CRISPR/Cas9 expression) and the empty vector were introduced into Agrobacterium strain EHA105 and used to infect calli from WT plants via Agrobacterium-mediated transformation (Hiei et al. 1994), with the temperature used for in vitro culture set to 20°C. After seeds were collected from the resulting T0 plants, T1 seedlings were grown and used to determine the segregation of genome-edited seedlings and their associated phenotypes at 20ºC.

RT-PCR and Real-time PCR (qPCR) Analysis
Total RNA was extracted from the roots, stems and leaves of WT at the seedling stage, as well as from ag leaves and young panicles at the heading stage, using an RNA Prep Pure Plant kit (Tiangen Co., Beijing, China). The puri ed total RNA was then used for rst-strand cDNA synthesis with the ReverTra Ace kit (ToYoBo, Osaka, Japan) following the manufacturer's instructions. RT-PCR analysis was carried out to assess TCD7 transcript levels. For RT-qPCR of TCD7 and 26 genes associated with chlorophyll biosynthesis (PORA, HEMA1, CAO1, YGL1), photosynthesis (Cab1R, RbcS, rbcL, psaA, psbA, LhcpII) and chloroplast development (RpoTp, rpoA, rpoB, rpoC1, rpoC2, FtsZ, aptA, 23SrRNA, 16SrRNA, rps7, rps20, V1, V2, OsV4, TSV3, petA), total RNA was extracted from the third, fourth and fth leaves of WT and tcd7 seedlings. qPCR analyses were performed using the SYBR Premix Ex TaqTM kit (TaKaRa) on an ABI7500 Realtime PCR System (Applied Biosystems; http://www.appliedbiosystems.com). Relative transcript levels were calculated as described by Livak and Schmittgen (2001). The primers for qPCR were designed according to either Wu et al (2007) and NCBI-published sequences and are listed in Additional le1: Table   S4. The rice ACTIN gene was used as a reference.

Subcellular localization
A cDNA fragment encoding the N-terminal region of TCD7 (amino acids 1-149) was ampli ed using the primer pair 5 -CCATCGATATGCACCGAATGGCTTCTCTT-3 (the ClaI site is underlined) and 5 -GGGGTACCTCTGCCCCTCTTCTTTGGTTC-3 (the KpnI site is underlined). The PCR amplicon was then subcloned into the binary vector pMON530-GFP to generate pMON530-TCD7-GFP. Constructs (pMON530-GFP and pMON530-TCD7-GFP) were transiently transfected into tobacco protoplasts and incubated in the dark at 28ºC for 16 h and then their uorescence was observed with a Leica SP8 confocal laser scanning microscope. Protoplast isolation followed a previously published method (Zhang et al. 2011).
Yeast Two-hybrid Assay and Immunoblot Analysis Yeast two-hybrid techniques were performed as described previously (Gao et al. 2011). The coding regions for LOC_Os03g40550 (TCD7) in WT and tcd7 were PCR-ampli ed and cloned into the bait vector pGBKT7. The coding regions for OsTRXz (LOC_Os08g29110) and OsFLN1/WLP2 (LOC_Os01g63220) were cloned into the prey vector pGADT7. Yeast two-hybrid assays were conducted following the manufacturer's instructions (Clontech). The primers used for vector construction are listed in Additional le 1: Table S5 For immunoblot analyses, chloroplast proteins (D1 and PsaA) were extracted from the fourth and fth leaves of WT and tcd7 seedlings grown at 20ºC or 32ºC. Total proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene di uoride membranes, immunoblotted with corresponding antibodies (anti-D1 or anti-PsaA), and detected using High-sig ECL Western Blotting Substrate (Tanon, China).