SlRCM1, which encodes tomato Lutescent1, is required for chlorophyll synthesis and chloroplast development in fruits

In plants, chloroplasts are the sites at which photosynthesis occurs, and an increased abundance of chloroplasts increases the nutritional quality of plants and the resultant color of fruits. However, the molecular mechanisms underlying chlorophyll synthesis and chloroplast development in tomato fruits remain unknown. In this study, we isolated a chlorophyll-deficient mutant, reduced chlorophyll mutant 1 (rcm1), by ethylmethanesulfonate mutagenesis; this mutant produced yellowish fruits with altered chloroplast development. MutMap revealed that Solyc08g005010 is the causal gene underlying the rcm1 mutant phenotype. A single-nucleotide base substitution in the second exon of SlRCM1 results in premature termination of its translated protein. SlRCM1 encodes a chloroplast-targeted metalloendopeptidase that is orthologous to the BCM1 protein of Arabidopsis and the stay-green G protein of soybean (Glycine max L. Merr.). Notably, the yellowish phenotype of the lutescent1 mutant can be restored with the allele of SlRCM1 from wild-type tomato. In contrast, knockout of SlRCM1 by the CRISPR/Cas9 system in Alisa Craig yielded yellowish fruits at the mature green stage, as was the case for lutescent1. Amino acid sequence alignment and functional complementation assays showed that SlRCM1 is indeed Lutescent1. These findings provide new insights into the regulation of chloroplast development in tomato fruits.


Introduction
Fruit development is a complex and highly coordinated process that involves a series of specific physiological and biochemical changes 1,2 . During tomato fruit development, chloroplasts serve as sites of photosynthesis and carbohydrate accumulation and can be transformed into chromoplasts for carotenoid formation 3 . Chloroplast development in tomato plants is directly proportional to fruit development and nutrient accumulation. Both the synthesis and degradation of chloroplasts in plants are in a stable state of dynamic equilibrium. Therefore, it is of enormous importance to study the molecular mechanism underlying chloroplast development in the process of fruit ripening and development.
Chloroplast development has been demonstrated to be regulated by multiple transcription factors, among which GLK2 is an influential transcription factor that regulates this process. Tomato fruits with full-length transcripts of GLK2 mRNA exhibit a dark green shoulder that can promote photosynthesis and accumulate more nutrients, while glk2 mutation eliminates this green shoulder 4 . GLK2 overexpression resulted in dramatic upregulated expression of the Solyc08g005010 gene, indicating a regulatory network involving chloroplast development 5 . The TKN2 and TKN4 genes in tomato modulate the gradient of chloroplast development in fruits by regulating GLK2 expression 6 . Overexpression of the APRR2-like gene in tomato increased the content of chlorophyll in fruits 7 . SlBBX20 in tomato regulates chloroplast development by modulating the expression of the SlCAB1B,SlCAB6A and SlCHL27 genes 8 . SlBEL11 encodes a transcription factor that negatively regulates chloroplast development and chlorophyll synthesis in tomato fruits. The SlBEL11 protein can directly bind to the promoters of genes involved in chloroplast development and chlorophyll synthesis, such as TKN2 ,CAB and POR, and downregulate the expression of these types of genes 9 . In plant growth and development, auxin signaling affects chlorophyll synthesis and chloroplast development. The auxin response factor ARF10 directly binds to the promoter of GLK1 and activates its expression, thereby promoting chloroplast development and sugar accumulation 10 . SlARF6A directly binds to the promoters of the GLK1, CAB1, CAB2 and RbcS genes and promotes their expression to modulate tomato fruit chloroplast development 11 . ARF2A, an auxin signal component, participates in fruit ripening and chloroplast development. In Arabidopsis, the ARF2 gene is involved in the degradation of chlorophyll in leaves 12 . The chlorophyll content of ARF2A-OX transgenic fruit was significantly lower than that of wild-type fruit at 42 days post-anthesis 13 .
Chlorophyll synthesis in plants is a complex process involving 15 enzymes encoded by 27 genes 14 . Mgchelatase catalyzes the binding of Mg 2+ to protoporphyrin IX, which represents the first step in chlorophyll synthesis 15 , and Mg-protoporphyrin IX is further methylated by Mg-protoporphyrin IX methyltransferase, followed by four subsequent catalytic reactions to produce chlorophyll 16 . Chlorophyllide a oxygenase (CAO) catalyzes the conversion of chlorophyll a to chlorophyll b and plays an important role in the balance of chlorophyll a and chlorophyll b. Overexpression of CAO in tobacco promotes the synthesis of chlorophyll b and decreases the ratio of chlorophyll a and chlorophyll b 17 . GUN4, a porphyrin-binding protein, enhances the activity of Mgchelatase by binding to protoporphyrin, which is the substrate of the Mg-chelatase reaction 18 . Overexpression of GUN4 can significantly increase the chlorophyll content in tobacco leaves 19 . SGR1, encoding Mg dechelatase, promotes chloroplast degradation during plant maturation 20 . Mutation of SGR1 in Chinese cabbage leads to increased chlorophyll concentrations and a stay-green phenotype 21 . In Arabidopsis, the bcm1 mutant exhibits a pale-green leaf phenotype due to reduced chlorophyll contents. BCM1 interacts with GUN4 to enhance Mgchelatase activity. BCM1 can also interact with SGR1 to destabilize the SGR1 protein 22 . GluTR is a glutamyl tRNA reductase involved in porphyrin and chlorophyll biosynthesis 23 . Interactions between BCM1 and GluTR affect the synthesis of 5-aminolevulinic acid, a key precursor in the biosynthesis of porphyrin during chlorophyll synthesis 22 . The product of CHLM is magnesium protoporphyrin IX methyltransferase that converts Mgprotoporphyrin IX to Mg-protoporphyrin IX methylester during chlorophyll synthesis 24 . BCM1 interacts with CHLM to promote the formation of a MgCH-GUN4-CHLM enzyme complex 22 . The stay-green G gene is a homologous gene of BCM1 in soybean; this gene positively regulates chlorophyll synthesis in the soybean seed coat 25 . Therefore, BCM1 plays an important and conserved role in chlorophyll synthesis in different crop species.
In tomato, the Lutescent2 (L2) gene, which encodes a zinc metalloprotease, has been characterized, and this gene has been shown to regulate chloroplast development and fruit maturity 26 . Similarly, the number of chloroplasts per cell in immature green fruit pericarp tissue of lutes-cent1 (l1) mutant was reduced significantly compared with that of the wild type. In addition, fully expanded leaflets of the l1 mutant exhibited a more dramatic yellowish phenotype 26 . It was previously reported that the chlorophyll content of the fruits of the l1 mutant was reduced, leading to a senescent phenotype for l1 27 . Chloroplast development is hindered in l1 28 , although the causative gene underlying chloroplast defects and chlorophyll reduction in l1 has not yet been identified.
In this study, we obtained a reduced chlorophyll mutant (rcm1) of tomato; this mutant is an l1 allelic mutant with altered chloroplast development and was generated via ethylmethanesulfonate (EMS) mutagenesis. We discovered via BSA + DNA-Seq and MutMap that the rcm1 mutant carries a single-nucleotide polymorphism (T → A), resulting in premature termination of the SlRCM1 protein. Sequence analysis and functional characterization showed that the SlRCM1 gene, located at the Lutescent1 locus, encodes an ortholog of the BCM1 protein of Arabidopsis and the stay-green G protein of soybean (Glycine max L. Merr.). SlRCM1 regulates chlorophyll synthesis and chloroplast development in fruits at the mature green and red ripe stages, while its ortholog regulates chlorophyll synthesis in the leaves of Arabidopsis and seeds of soybean 22,25 . These findings highlight the molecular mechanisms underlying chlorophyll synthesis and chloroplast development in tomato fruits.

Results
The tomato rcm1 mutant exhibits altered chloroplast development Seeds of the tomato cultivar Ligeer 87-5 were treated with 1% EMS to obtain novel mutants. A reduced chlorophyll mutant (rcm1) was isolated from the EMSmutagenized population (Fig. 1a). The fruits of the rcm1 mutant exhibited reduced amounts of chlorophyll relative to those of the wild type. The fruits at the mature green (MG) and breaker (BR) stages of the rcm1 mutant showed a sharp decrease in chlorophyll relative to those of the wild type. However, the fruits of the rcm1 mutant could still turn red at the red ripe (RR) stage (Fig. 1b). Furthermore, the thylakoid membranes were collapsed in the Fig. 1 Phenotypes of wild-type and rcm1 mutant plants. a Whole-plant morphology of the reduced-chlorophyll mutant (rcm1) and the wild type (Ligeer 87-5). Scale bars, 15 cm. b Phenotypes of Wild Type (WT) and rcm1 tomato fruits at four developmental stages. MG, mature green stage; BR, breaker stage; YR, yellow ripe stage; RR, red ripe stage. Scale bars, 2 cm. c Chloroplast ultrastructure of mature green fruits of wild-type and rcm1 mutant plants via transmission electron microscopy. P, plastoglobuli; TGS, thylakoid granum stacks. Scale bars, 10 μm (left) and 0.5 μm (right). d Chlorophyll content of fruit pericarps from wild type and rcm1 at the MG, BR, YR and RR stages. The data are presented as the means ± SDs (n = 6). e Carotenoid contents of fruit pericarps from wild type and rcm1 at the MG, BR, YR and RR stages. The data are presented as the means ± SDs (n = 6). f Phenotypes of the pistils of the wild type and rcm1 mutant. Scale bars, 2 mm. g Morphology of the leaves of the rcm1 mutant and the wild type. Scale bars, 2 cm. h Chloroplast ultrastructure of leaves of wild-type and rcm1 plants using transmission electron microscopy. SG, starch grains; TGS, thylakoid granum stacks. Scale bars, 0.5 μm. i Chlorophyll content of leaves from wild-type and rcm1 plants. The data are presented as the means ± SDs (n = 6). The asterisks indicate statistically significant differences according to t-tests: **, P-value < 0.01. nd, not detected chloroplasts of fruits of the rcm1 mutant at the mature green stage (Fig. 1c). The chloroplasts in the rcm1 fruits were smaller than those of the wild type (Fig. 1c). Subsequently, the chlorophyll content in the rcm1 mutant fruits was significantly lower than that in the wild-type fruits at the MG, BR and yellow red (YR) stages, and no chlorophyll was detected at the RR stage (Fig. 1d). The carotenoid content in the rcm1 mutant fruits was also significantly lower than that in the wild-type fruits at all four developmental stages (Fig. 1e). The a*/b* colorimetric values (Fig. S1a), which are proportional to the lycopene, and total soluble solids contents (Fig. S1b) of red ripe fruits of the rcm1 mutant decreased relative to those of the wild type. Fruit ripening of the rcm1 mutant was delayed by approximately 6 days compared to that of the wild-type (Fig. S2a), and the ethylene production rate of the rcm1 mutant fruits was lower than that of the wildtype fruits (Fig. S2b).
To fully evaluate the phenotype of the rcm1 mutant, the leaves and pistils were compared between the rcm1 mutant and wild type. A lack of chlorophyll accumulation was observed in the developing pistils and leaves of rcm1 relative to the wild type ( Fig. 1f, g). Transmission electron microscopy revealed impaired development of thylakoid membranes in the chloroplasts of the rcm1 mutant compared to those of the wild type (Fig. 1h). Accordingly, the chlorophyll content in the rcm1 mutant leaves was significantly lower than that in the wild-type leaves (Fig.  1i). Compared to that of the wild type, the maximum photochemical efficiency (Fv/Fm) of leaves of the rcm1 mutant decreased significantly (Fig. S3a, b). Additionally, the quantum efficiency of PSII (Y(II)) of the leaves of the rcm1 mutant was significantly impaired compared with that of the wild type (Fig. S3c). Additionally, pollen vitality in rcm1 was lower than that in Ligeer 87-5 (WT); the percentage of malformed pollen in rcm1 was greater than that in the wild type (Fig. S4).

Cloning of SlRCM1
To genetically characterize the yellowish fruit phenotype of the rcm1 mutant, we generated an F 2 population consisting of 307 individuals by crossing Ligeer 87-5 with the rcm1 mutant. The F 1 generation of the cross between Ligeer 87-5 and rcm1 displayed a normal phenotype similar to that of Ligeer 87-5, which suggested that the gene underlying the reduced chlorophyll phenotype of the rcm1 mutant is a recessive gene. In the F 2 population, the ratio between the number of individuals with a normal phenotype (237) and the number of individuals with a chlorophyll-deficient phenotype (70) was approximately 3:1 (χ 2 = 0.68, χ 2 0.05 = 3.84), indicating that the chlorophyll-deficient phenotype was controlled by a single gene (Table S1).
Next, we performed BSA and MutMap analyses to isolate the candidate genes. We sequenced two bulk populations comprising 25 individuals with green fruit or yellowish fruit at the mature green stage. Each library was sequenced at a depth of a ×25 genome equivalent. The generated reads were mapped to the tomato reference genome (M82), and allele frequency differences of 40,124 SNPs from the two pools were calculated and mapped across the 12 chromosomes in tomato to form a Manhattan plot (Fig. 2c). The confidence threshold exceeded 95% only at the beginning of chromosome 8, between SL2.50ch08_1 and SL2.50ch08_1010000 (Table S2). Generally, single-base mutations are frequently generated by EMS mutagenesis 29,30 . There were six SNPs between the two pools between SL2.50ch08_1 and SL2.50ch08_1010000 (Table S3). Genetic analysis of the segregating populations indicated that the phenotype of the rcm1 mutant is likely controlled by a single locus. Since M82 (the genotype of the reference genome) develops normal chloroplasts in its fruits, SNPs in the recessive pool with an allele frequency of 1 and different from the reference genome (M82) were scored, and only one SNP (A → T) at SL2.50ch08_16268 out of all SNPs in the coding regions was identified.
Furthermore, no InDels or other structural variations were detected via 25 equivalent genome sequencing events of the WT and rcm1 mutants. We analyzed the allele frequency of the SNP at SL2.50ch08_16268 in both pools with wild-type and mutant phenotypes. The SNP at SL2.50ch08_16268 in M82 was A, whereas in the mutant pool, it was T (100%) ( Table 1). Furthermore, the SNP at SL2.50ch08_16268 in 30 F 2 progeny with yellowish fruit was T (100%), which was confirmed via PCR and Sanger sequencing. This SNP (SL2.50ch08_16268) occurs in the second exon of Solyc08g005010 and was further verified by Sanger sequencing. Because the transcriptional direction of Solyc08g005010 is opposite to that of the genome, the nucleotide of SL2.50ch08_16268 in the coding strand of Solyc08g005010 in M82 and Ligeer 87-5 was T, but it was A in the rcm1 mutant ( Fig. 2d, e). The substitution of T (wild type) to A (rcm1 mutant) resulted in the conversion of Tyr (wild type) to a stop codon (rcm1 mutant) (Fig. 2e), indicating that the translation of the SlRCM1 protein was terminated prematurely in the rcm1 mutant. The predicted protein encoded by the SlRCM1 allele from the rcm1 mutant did not include any functional domain (Fig. S5a). Therefore, we inferred that the SlRCM1 allele in the rcm1 mutant is null. Furthermore, the expression level of the SlRCM1 allele in the rcm1 mutant was equivalent to that in the wild type (Fig. S5b). Therefore, we considered Solyc08g005010 to be the candidate SlRCM1 gene, which encodes a CAAX-type endopeptidase.
The yellowish phenotype of fruits of rcm1 was complemented by SlRCM1 from Ligeer 87-5 To verify whether SlRCM1 is the causal gene for the chlorophyll-deficient phenotype of the rcm1 mutant, SlRCM1 was overexpressed (OE) under the control of the CaMV35S promoter in the rcm1 mutant (Fig. 3a). Nineteen independent transgenic lines were obtained. The expression levels of the SlRCM1 gene in the OE-11 (#11) and OE-18 (#18) lines in the T 1 generation significantly increased relative to those of the rcm1 mutant. The normal-green phenotypes of the fruits of the OE-11 and OE-18 lines were restored at the mature green stage (Fig.  3b). To further verify the biological function of SlRCM1, a transformation construct (Pro SlRCM1 (Ligeer 87-5) :: CDS SlRCM1 (Ligeer 87-5) ) was prepared by inserting the SlRCM1 CDS from Ligeer 87-5 into a pHELLSGATE8 vector under the native promoter of the transgene (Fig.  3c). We introduced this construct into the rcm1 mutant by Agrobacterium-mediated transformation. Eleven independent transgenic plants were identified. The phenotype of the rcm1 transgenic lines was restored to that of the  wild type (normal green fruit) (Fig. 3d). Overall, these results confirmed that SlRCM1 is the correct candidate gene.

SlRCM1 regulates chloroplast development in tomato
To further verify that SlRCM1 regulates chloroplast development in tomato fruits, we overexpressed the SlRCM1 gene in S. lycopersicum (L.) cv. Alisa Craig (AC) by Agrobacterium-mediated transformation. Compared to those of the wild type, the fruits of the OE-1 and OE-2 lines appeared dark green at the mature green stage (Fig.  4a). The number of thylakoids and thylakoid granum stacks in mature green fruits of the OE lines was higher than that of the wild type (Fig. 4b). Interestingly, the locular material surrounding the seeds remained green, and chlorophyll could still be detected at the red ripe stage in the overexpression lines (Fig. 4a). Indeed, the expression levels of the SlRCM1 gene in the OE-1 and OE-2 lines were significantly higher than those in the wild-type (AC) line (Fig. 4c). The chlorophyll content in the MG-and RR-stage fruits of OE-1 and OE-2 lines increased significantly relative to that in the wild type (Fig.  4e, f). In addition, the carotenoid content in the MG-and RR-stage fruits of the OE-1 and OE-2 lines increased significantly relative to that of the wild type (Fig. 4g, h). Chlorophyll in the RR-stage fruit of AC was not detected. Furthermore, we used CRISPR/Cas9 (CR) to edit the first exon of SlRCM1 in the AC background (Fig. 4a, d). We determined the type of mutation in the T 1 generation using PCR and Sanger sequencing. The SlRCM1 knockout line CR-1 contained a 2-bp deletion in SlRCM1, and the knockout line CR-2 contained a 1-bp insertion in SlRCM1 (Fig. 4d). Both the CR-1 and CR-2 lines exhibited a reduced chlorophyll phenotype at and after the mature Overexpressing SlRCM1 in rcm1 rescued the phenotype, as was the case for Ligeer 87-5. The relative expression of SlRCM1 in mature green fruits of SlRCM1-overexpressing lines and untransformed control (rcm1) plants was quantified. Scale bars, 2 cm. c Schematic structure of the expression vector with SlRCM1 driven by its native 2944-bp promoter from Ligeer 87-5. The two arrows represent the primers used to analyze the transcript level of SlRCM1. Fw: Forward primer; Rv: Reverse primer. d The phenotype of the fruits of the transgenic lines could be restored to that of Ligeer 87-5. The relative expression of SlRCM1 in mature green fruits was quantified in the transgenic lines and rcm1. Scale bars, 2 cm. The ACTIN gene (Solyc11g005330) was used as the internal control. The data are presented as the means ± SDs (n = 3) and SlRCM1-CRISPR (CR) lines at the RR stage. The data are presented as the means ± SDs (n = 6). The asterisks indicate statistically significant differences according to t-tests: **, P-value < 0.01. nd, not detected green stage (Fig. 4a). Furthermore, the development of thylakoid membranes was impaired in the mature green fruits of the CR lines compared to that of the wild type (Fig. 4b). The chlorophyll content in the MG-stage fruits of CR-1 and CR-2 plants decreased significantly relative to that in the wild-type plants (Fig. 4e). These results indicate that SlRCM1 is responsible for chlorophyll synthesis and chloroplast development in tomato fruits.

SlRCM1 is Lutescent1
A spontaneous lutescent1 mutant (l1), LA3717, was derived from Ailsa Craig (AC). The l1 mutation has been shown to dramatically affect tomato fruit development. l1 mutants have a low chlorophyll content in their fruits, especially under high-light and dark conditions, which enhances the rate of chlorophyll loss 26 . To date, the gene underlying the l1 mutant has not yet been cloned. The fruits of SlRCM1 knockout lines (CR) generated using the CRISPR/Cas9 system exhibited the same yellowish phenotype as did the fruits of the l1 mutant in the Alisa Craig (AC) background described in the Tomato Genetics Resource Center (TGRC; https://tgrc.ucdavis.edu) (Fig. 5a). The yellowish phenotype of the fruits of the l1 mutant is purportedly controlled by a single gene located at the beginning of chromosome 8 26,31 . We used SlRCM1-specific primers to amplify full-length gDNA and the 5-kb promoter of SlRCM1 from AC, Ligeer 87-5 and l1 (Table S4). A base deletion led to premature termination of the SlRCM1 protein in l1 (Fig. 5b, c). Furthermore, the expression level of the SlRCM1 allele in the l1 mutant was equivalent to that in AC (Fig. S5c). SlRCM1 was subsequently overexpressed under the control of the CaMV35S promoter in the l1 mutant. Nine independent transgenic lines were identified. The expression level of the SlRCM1 gene in the 2 nd (#2) and 6 th (#6) transgenic lines was significantly higher than that in the control (l1) (Fig. 5d). Functional complementation of SlRCM1 in l1 restored normal plant growth and development (Fig. 5e). Taken together, these results indicate that SlRCM1 may be a causal gene underlying the yellowish phenotype of the l1 mutant.

SlRCM1 expression and subcellular localization
The gDNA of the SlRCM1 gene is 6,299 bp in length and consists of nine exons and eight introns. The SlRCM1 gene encodes a protein comprising 376 amino acids. Alignment of amino acid sequences (Fig. S6) and phylogenetic analysis (Fig. 6a) showed that the SlRCM1 protein is highly conserved among soybean, potato, pepper, tobacco, Arabidopsis, rice, maize, jute and Cephalotus follicularis, suggesting that the SlRCM1 protein may play an essential role in plants. Tomato SlRCM1 may function in chloroplast development and chlorophyll synthesis through a conserved biological process, similar to that in soybean and Arabidopsis thaliana.
To investigate the expression pattern of SlRCM1, quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed as described previously 32 . SlRCM1 expression was detected in all tissues, and relatively high expression levels were detected in the leaves, flowers and immature green fruits (Fig. 6b). Furthermore, we used GUS staining to study the expression of SlRCM1 in several tissues. The GUS staining revealed that SlRCM1 was expressed in the stems, leaves, flowers and fruits (Fig. 6c).
To determine the SlRCM1 subcellular localization, we constructed a SlRCM1::GFP fusion protein. Through Agrobacterium-mediated infiltration, the SlRCM1::GFP fusion protein was transiently expressed in tobacco. Under a Nikon e600 fluorescence microscope (Nikon, Tokyo, Japan), the autofluorescence of the chloroplast appeared red 33 . The SlRCM1::GFP fluorescence overlapped with the red fluorescence, indicating that the SlRCM1 protein is localized in the chloroplast (Fig. 6d).

The expression level of SlRCM1 is directly regulated by the transcription factor SlARF2A
It was previously reported that ARF proteins regulate gene expression by binding to the TGTCTC cis-element of the target gene promoter 34 . Yeast one-hybrid (Y1H) and dual luciferase experiments indicated that SlARF2A binds to cis-elements of the SlRCM1 promoter. The TGTCTC cis-element is located -245 to -239 bp upstream of the start codon of SlRCM1 (Fig. 7a). Yeast cells that were cotransformed with pGADT7-SlARF2A and pAbAi-SlRCM1-Pro were able to grow on 20 ng/mL aureobasidin A (ABA) SD/-Leu media, but the negative control cotransformed with pGADT7 and pAbAi-SlRCM1-Pro did not grow (Fig. 7b), indicating that the ARF2A protein could directly bind to the promoter of SlRCM1.
To further indicate that the SlARF2A protein interacted with the promoter of SlRCM1 in plants, we constructed a luciferase (LUC) reporter harboring a promoter sequence from -1,478 bp to the start codon (ATG) derived from SlRCM1 (Fig. 7c). Tobacco leaves were coinfiltrated with Agrobacterium tumefaciens (GV3101) strains containing the indicated effector constructs containing SlARF2A and the reporter constructs containing the SlRCM1 promoter (Fig. 7d). Taken together, these results indicate that SlARF2A physically binds to the SlRCM1 promoter to downregulate its expression.

Discussion
Chlorophyll is essential for plants to absorb, transfer and convert light energy to bioenergy and plays a vital role in photosynthesis to promote plant growth and development 35 . Here, we cloned the SlRCM1 gene, which controls chlorophyll synthesis and chloroplast development in tomato fruits. A single-nucleotide replacement resulting in a premature termination mutation in SlRCM1 impaired chloroplast development in the rcm1 mutant. Increasing the chlorophyll content in tomato fruits contributes to improved nutrition 36 . Consistently, the soluble solids content of red ripe fruits of SlRCM1-overexpressing (OE) lines was significantly higher than that of the wild type, whereas the knockout lines (CR) showed the opposite effect (Fig. S7). In addition, the a*/b* colorimetric value of the ripe red fruits of the SlRCM1-overexpressing lines (OE) was significantly higher than that of the wild type, while the knockout lines (CR) showed the opposite effect (Fig. S8). Therefore, characterization of the SlRCM1 gene may provide insights into chlorophyll synthesis and chloroplast development in tomato fruits.
The rcm1 mutant was obtained from an EMSmutagenized population. EMS mutagenesis is widely used in the construction of mutant libraries and in plant functional genomics 37,38 . For instance, the hst1 mutant generated from EMS mutagenesis was crossed with the wild type to map the OsRR22 gene 39 . TILLING analysis 40 of an EMS-mutagenized population of Arabidopsis showed that EMS mutagenesis produces a large number of single-base substitutions within hotspot segments 41 . In this study, a reduced-chlorophyll mutant, rcm1, was acquired by EMS mutagenesis and crossed with wild type to map the SlRCM1 gene responsible for the yellowishfruit-producing mutant. Due to the close genetic background between the EMS-induced mutant and its wild type, crossing a mutant with its wild type is a fast and effective approach to isolate a gene 42,43 . Two SNPs in the recessive pool with an allele frequency of 1 were found between SL2.50ch08_1 and SL2.50ch08_1010000 (Table Fig. 5 Lutescent1 is the allele of SlRCM1. a Phenotypes of fruits from the CR-1 line, lutescent1 mutant (l1) and AC at different developmental stages. IMG, immature green stage; MG, mature green stage; BR, breaker stage; YR, yellow ripe stage; RR, red ripe stage. The l1 is a mutant in the AC background. Scale bars, 3 cm. b Nucleotide sequence alignment of SlRCM1 fragments in Ligeer 87-5, Ailsa Craig (AC) and l1. The other bases are consistent among Ligeer 87-5, AC and l1. The black star indicates the base at SL2.50ch08_16268 in the first exon of the SlRCM1 gene. c Amino acid sequence alignment of the SlRCM1 protein of Ligeer 87-5, AC and l1. d Relative expression of SlRCM1 in mature green fruits of the SlRCM1overexpressing lines and the control (l1). The ACTIN gene (Solyc11g005330) was used as an internal control. The data are presented as the means ± SDs (n = 3). e Complementation of the l1 mutant using SlRCM1. Phenotypes of mature green fruits from l1-and SlRCM1-overexpressing lines were imaged and are shown. Scale bars, 3 cm S2). Since the SNP at SL2.50ch08_ 394612 was located in an intergenic region, we speculated that it is not a causal SNP for the reduced chlorophyll phenotype of rcm1. The SNP (A → T) in the second exon of Solyc08g005010 resulted in a premature stop codon (Fig. 2e). Furthermore, the rcm1 mutant phenotype was rescued by overexpression of the SlRCM1 gene in the rcm1 mutant, which confirmed that SlRCM1 was responsible for the mutant phenotype (Fig. 3). MutMap analysis could therefore be a feasible approach to genetically identify genes in EMSmutagenized mutants.
SlRCM1 has highly conserved functions in both chloroplast development and chlorophyll synthesis. There are two homologous genes of SlRCM1 in Arabidopsis: BCM1 and BCM2 22 . SlRCM1 was found to share 78% amino acid sequence identity with BCM1 and 84% identity with BCM2 (Fig. S1). Previous studies have shown that BCM1 and BCM2 regulate chlorophyll synthesis and chloroplast development in Arabidopsis 22 . Yeast two-hybrid (Y2H) assays, bimolecular fluorescence complementation (BiFC) assays, coimmunoprecipitation (Co-IP) assays and enzyme activity-measuring experiments in Arabidopsis demonstrated that BCM1 interacts with GUN4 to stimulate Mg-chelatase activity and optimize chlorophyll synthesis 22 . BCM1 also interacts with SGR to prevent chlorophyll degradation 22 . SlRCM1's ortholog in soybean is the stay-green G gene and regulates chlorophyll synthesis and chloroplast development in the seed coat of soybean 25 . However, biological divergence has shown that SlRCM1 regulates chlorophyll synthesis and chloroplast development in fruits, while its ortholog regulates chlorophyll synthesis in the seed coat of soybean 25 and in the leaves of Arabidopsis 22 .
It has been reported that SlRCM1 orthologs modulate chlorophyll synthesis in the leaves of Arabidopsis and in the seed coat of soybean 22,25 . BCM1 overexpression did not alter the chlorophyll content in Arabidopsis thaliana, but chlorophyll accumulation was inhibited only in the leaves of the bcm1 mutant 22 . In the present study, fruits of the tomato SlRCM1 knockout lines showed a yellowish phenotype due to impaired chloroplasts. The chlorophyll content in the MG-stage fruits of the SlRCM1-overexpressing lines increased significantly. Moreover, chlorophyll was also detected in the RR-stage fruits of the SlRCM1 overexpression lines (Fig. 4). Taken together, these results showed that SlRCM1 has a diversified and strong effect on chlorophyll synthesis and chloroplast development in fruits.
In summary, we have demonstrated that SlRCM1, a chloroplast-targeted ortholog of the BCM1 protein in Arabidopsis and stay-green G protein of soybean, participates in chlorophyll synthesis and chloroplast development in tomato fruits. SlRCM1 modulates the number of thylakoids and the structure of thylakoid membranes in chloroplasts. SlRCM1 was identified as the causal gene at the Lutescent1 locus. An understanding of SlRCM1 provides insights into the molecular mechanism underlying fruit development and target genes for genetic improvement in horticultural crop species.

Plant materials and mutant screening
A chlorophyll-deficient mutant (rcm1) was derived from EMS mutagenesis of the processed tomato inbred line Ligeer 87-5. For EMS mutagenesis, Ligeer 87-5 seeds were immersed in 1% EMS solution and shaken in a shaker for 12 h. The seeds were then rinsed with running water for 10 min and germinated in an artificial incubator at 30°C. Mutants were screened in the M 2 generation 44 . We constructed an F 2 mapping population by crossing Ligeer 87-5 with rcm1. The l1 mutant (LA3717) in the AC background was obtained from the TGRC (https://tgrc. ucdavis.edu). For the overexpression constructs, SlRCM1 was amplified and cloned into a pHELLSGATE8 vector driven by the CaMV35S promoter and its native promoter 45 . The overexpression vector was introduced into The complete open reading frame (ORF) of SlARF2A was cloned into pGADT7 (AD) to generate a pGADT7-SlARF2A effector construct. The reporter vector was introduced into yeast strain Y1Hgold together with the effector vector. A combination of the reporter vector and the empty vector pGADT7 was used as a negative control. The yeast cells were cultured on SD-Leu media supplemented with various concentrations of ABA (0, 10, and 20 ng/ml). c Schematic representation of constructs used for the dual luciferase assay. The SlARF2A ORF was cloned into pGreenII 62-SK to generate a pGreenII 62-SK-SlARF2A effector construct. The SlRCM1 promoter was fused to pGreenII 0800-LUC to create a pGreen II 0800-SlRCM1-Pro reporter construct. d Relative LUC/RLU ratio. The vectors of both the effectors and reporters were transformed into Agrobacterium cells, which were then used to infiltrate tobacco leaves. LUC, firefly luciferase activity; RLU, Renilla luciferase activity. The data are presented as the means ± SDs (n = 6). The asterisks indicate statistically significant differences according to t-tests: **, P < 0.01 Solanum lycopersicum cv. AC and the rcm1 and l1 mutants through Agrobacterium-mediated transformation. SlRCM1 knockout mutants in the AC background were generated using the CRISPR/Cas9 system 46 . The 2934-bp promoter of the SlRCM1 gene was amplified from AC and cloned into a pV3P vector driving β-glucosidase (GUS), yielding a Pro SlRCM1 ::GUS vector, which was subsequently transformed into AC. The tomato tissues were quickly frozen in liquid nitrogen and stored at −80°C. The tomato plants used in this study were grown in a greenhouse, and the primers used in this experiment are listed in Table S4.

Transmission electron microscopy
The fruits of Ligeer 87-5, the rcm1 mutant, AC and the transgenic lines were isolated and fixed in 0.05 M cacodylate buffer consisting of 2% glutaraldehyde and dehydrated in ethanol. After embedding in Spurr resin, ultrathin sections of the samples were obtained using a Leica EMUC6 ultramicrotome. A Hitachi H-7650 transmission electron microscope was used to observe the ultrastructure of the plastids.

Determination of the chlorophyll content
For chlorophyll extraction, fruits of Ligeer 87-5, the rcm1 mutant, AC and the transgenic lines at different developmental stages were placed in 10 mL of 80% (v/v) acetone in the dark until the tissues became white. Their absorbance was subsequently measured at 646, 663 and 470 nm, and the chlorophyll content was measured and calculated as previously reported 47 .

BSA and DNA-seq
An F 2 population comprising 307 individuals was derived from a cross between Ligeer 87-5 and the rcm1 mutant. In this population, equal amounts of DNA were pooled from 25 plants with a wild-type phenotype and 25 plants with a mutant phenotype. Approximately 25× genome sequences for each pool were generated using the Illumina HiSeq X Ten platform 48 . Due to the close genetic background of the M82 and Ligeer 87-5 processed tomato genotypes, the M82 genome version SL 2.50 (http:// solgenomics.net) was used as the reference genome to facilitate mining of the causal SNPs.

Subcellular localization
The SlRCM1 coding sequence without the stop codon was amplified from the cDNA of AC and then cloned into a pHBT vector driven by the CaMV35S promoter, yielding CaMV35S::SlRCM1-GFP. This vector was subsequently transformed into Agrobacterium tumefaciens strain GV3101, which was then injected into the leaves of Nicotiana benthamiana (N. benthamiana) as previously described 49 . After 48 h of incubation at 25°C, the fluorescence of GFP and RFP in the tobacco leaves was observed using Leica Confocal software. Chloroplasts that exhibit red fluorescence were used as positive controls. The primers used in this experiment are listed in Table S4.

Quantitative reverse transcription PCR
Total RNA was extracted from frozen tissues using TRIzol reagent (Vazyme, Nanjing, China). RNA was then reverse transcribed into cDNA using a first-strand cDNA Synthesis Kit (Vazyme). The product length ranged from 80 to 200 bp. The qRT-PCR was used to determine the transcript levels of genes in 96-well plates with a Roche LightCycler ® 480 system according to the manufacturer's protocol 32 . The expression of the ACTIN gene (Solyc11g005330) was used as an internal control. The primers used were designed using Primer Premier 5 (Table S4).

Colorimeter-based evaluation of red ripe fruits
The a* and b* values of ripe red fruits were measured using a CM-5 colorimeter. Three independent points were determined for each fruit assay, and six fruits were measured per line.

Measurement of total soluble solids
The total soluble solids (Brix) of red ripe fruits were measured using a digital refractometer (PR100, Atago Co., Ltd.). All sample assays were performed for three technical replicates and six biological replicates.

Ethylene assays
Tomato fruits were collected at 37, 43 and 49 days after flowering. The determination of ethylene content in the tomato fruits was based on previous methods 50 . All the samples consisted of three technical replicates and three biological replicates.

Pollen viability assays
The pollen of blooming flowers was soaked in I-KI solution (1% KI and 0.5% I 2 ) for 2 min. The stained pollen was then examined under a low-magnification microscope. The viable pollen was stained dark blue by I-KI.

Chlorophyll fluorescence measurements
After the leaves of Ligeer 87-5, the rcm1 mutant, AC and L1-CR were conditioned in the dark for 30 min, an imaging pulse amplitude modulated chlorophyll fluorimeter (IMAG-MAXI; Heinz Walz, Effeltrich, Germany) was used to determine chlorophyll fluorescence. The maximal photochemical efficiency of PSII (Fv/Fm) and the quantum efficiency of PSII (Y(II)) were calculated and determined according to previously reported methods 51 .

Yeast one-hybrid assays
The SlARF2A ORF was cloned into a pGADT7 vector to yield a prey construct. The SlRCM1 promoter was inserted into a pAbAi vector to yield a bait construct. The pAbAi bait vector was then used to transform yeast strain Y1HGold, which was integrated into the yeast genome to generate reporter strains. The prey vector was introduced into the reporter strains and grown for three days on SD/-Leu-Ura media. The positive yeast strains were selected and diluted in double-distilled water to an OD600 of 0.1, and 2 μL of the suspension was spotted onto SD/-Leu media with or without ABA (0, 10 and 20 ng/mL), followed by 3 days of incubation at 30°C. pGADT7 and pAbAi-PSY1-Pro served as negative controls. The primers used in this experiment are listed in Table S4.

Dual luciferase transactivation assays
To generate an effector construct, the full-length ARF2A ORF was inserted into a pGreenII 62-SK vector. Similarly, the promoter from SlRCM1 was inserted into pGreenII 0800-LUC to yield a reporter construct. The constructs and the pSoup helper plasmid were simultaneously introduced into Agrobacterium tumefaciens (GV3101). Tobacco leaves were infiltrated with the Agrobacterium strains and harvested three days later. The firefly LUC and Renilla luciferase (RLU) activities were quantified using a dual-luciferase reporter assay system. The transactivation activities were expressed as the ratio of LUC to RLU activity. The primers used in this experiment are listed in Table S4.