Mutation of SlARC6 leads to tissue-specific defects in chloroplast development in tomato

The proliferation and development of chloroplasts are important for maintaining the normal chloroplast population in plant tissues. Most studies have focused on chloroplast maintenance in leaves. In this study, we identified a spontaneous mutation in a tomato mutant named suffulta (su), in which the stems appeared albinic while the leaves remained normal. Map-based cloning showed that Su encodes a DnaJ heat shock protein that is a homolog of the Arabidopsis gene AtARC6, which is involved in chloroplast division. Knockdown and knockout of SlARC6 in wild-type tomato inhibit chloroplast division, indicating the conserved function of SlARC6. In su mutants, most mesophyll cells contain only one or two giant chloroplasts, while no chloroplasts are visible in 60% of stem cells, resulting in the albinic phenotype. Compared with mature tissues, the meristem of su mutants suggested that chloroplasts could partially divide in meristematic cells, suggesting the existence of an alternative mechanism in those dividing cells. Interestingly, the adaxial petiole cells of su mutants contain more chloroplasts than the abaxial cells. In addition, prolonged lighting can partially rescue the albinic phenotypes in su mutants, implying that light may promote SlACR6-independent chloroplast development. Our results verify the role of SlACR6 in chloroplast division in tomato and uncover the tissue-specific regulation of chloroplast development.


Introduction
Chloroplasts are an important organelle where plants absorb solar energy and produce sugars 1 . The number, size, and morphology of chloroplasts directly affect leaf color and photosynthesis intensity.
Chloroplast division and proliferation are important for maintaining the chloroplast population. In Arabidopsis, a number of mutants that are defective in the accumulation and replication of chloroplasts (arc) have been identified, in which chloroplast number, size, and shape are severely affected [2][3][4][5] . Similar to their original microbial ancestors, chloroplasts replicate by binary fission in plants, which is driven by ring-like dynamic division machinery located at the middle of the organelle [6][7][8][9] . In plants, the contractile component of the division machinery is composed of the FtsZ ring (Z ring), which is located in the inner membrane of the chloroplast, and tubulin-like heteropolymer-forming proteins (FtsZ1, FtsZ2, and DRP5B), which are located in the outer membrane of the chloroplast [10][11][12][13][14][15][16][17][18] . ARC6 and PARC6 (paralog of ARC6) encode chloroplast-targeted proteins that assemble and stabilize the Z ring by directly interacting with FtsZ2 and FtsZ1 [19][20][21][22] . ARC6 is closely related to Ftn2, a prokaryotic cell division protein. PARC6 is unique in vascular plants. It is possible that PARC6 was duplicated from ARC6 after the separation between nonvascular and vascular plants 23 . In Arabidopsis arc6 mutants, a mesophyll cell usually contains only two giant chloroplasts 2,21 . The chloroplast number in parc6 mutants is tenfold less than that in the wild type (WT), while PARC6 overexpression often inhibits chloroplast division by repressing FtsZ assembly [23][24][25][26] . ARC6 and PARC6 can recruit PLASTID DIVISION1 (PDV1) and PDV2, both of which are located in the outer envelope membrane (OEM) 23,27 . PDV1 and PDV2 then further recruit dynamin-related protein 5B (DRP5B/ARC5) to the OEM of the mid-chloroplast 28 . The Z ring is confined to the mid-chloroplast, and the formation of the Z ring requires the chloroplast Min system, which includes ARC3, MinD1, and MinE1 9,24,27,[29][30][31][32] . Multiple chloroplast division site 1 (MCD1) is another plant-specific protein that is required for Z ring positioning 33,34 . It was previously shown that MCD1 interacted with ARC6 in the stroma and interacted with FtsZ2 in an ARC6-dependent manner 33 .
Much of our understanding of chloroplast division has been derived from studies on Arabidopsis leaves. However, fossil records revealed that most ancient vascular plants had only the axis without leaves, suggesting that chloroplasts emerged much earlier than leaves 35,36 . Thus, knowledge of chloroplasts derived from other tissues can contribute to our understanding of chloroplast division and physiology. In this study, we analyzed a tomato mutant named su, which is a naturally spontaneous mutant collected by the TGRC Tomato Genetics Resource Center (http://tgrc.ucdavis.edu). A prominent phenotype of su mutants is albinic stems with visually normal leaves. Bulked segregant analysis (BSA), mapbased cloning, and functional verification by virusinduced gene silencing (VIGS) and CRISPR all showed that mutation of SlARC6 led to the differential albinic phenotype. Further observations indicated that the chloroplasts in the stem of su mutants almost disappeared, while they fused into the giant abnormal plastids in leaves, suggesting the differential effect of SlARC6 mutation on stems and leaves. The defective chloroplasts in the stem of su mutants could partially be rescued by prolonged light exposure, suggesting that light signaling could regulate chloroplast development.

Phenotypic analyses of the suffulta (su) mutant
In the natural lines collected by the TGRC (http://tgrc. ucdavis.edu/), we characterized a mutant named su (LA0628), in which the spontaneous mutation causes the phenotype of albinic stems. We first examined the color changes of stems and leaves over different developmental stages in su mutants and WT (Ailsa Craig) plants ( Fig.  1A-L, Supplementary Fig. 1). In 1-week-old seedlings, su mutants exhibited pale hypocotyls, while the color of cotyledons and young leaves stayed similar to those in WT (Fig. 1A, B, G, H). At 2-3 weeks, the newly formed tissues, including the stems, petioles, and rachis, in su mutants were still albinic, but the leaves appeared WT-like  Fig. 1), suggesting that the defect of albinic stems is not a developmental stage-associated phenotype.
To further assess whether the albinic phenotype is related to chlorophyll, we measured the chlorophyll content in 3-week-old tomato seedlings. Our results showed that the levels of multiple chlorophylls, including Chla, Chlb, Car, and total Chl, in su stems were significantly lower than those in WT stems, but such a difference was not detected in the leaves (Fig. 1M, N). The pale color and the defective chlorophyll content prompted us to hypothesize that the albinic phenotype in su mutants might derive from the absence of chlorophyll precursors or dysfunctional chloroplasts.

Fine mapping of Su
To identify the mutation, we developed an F 2 population by crossing su mutants to WT. All F 1 plants exhibited normal color in both stems and leaves, indicating that the albinic phenotype was caused by a recessive mutation. In the F 2 population, the separation ratio between normal stems and albinic stems was~3:1, consistent with Mendel's law of single-gene inheritance (Supplementary Table  1). Using a next-generation sequencing-based BSA approach, we identified an associated locus on the long arm of chromosome 4 ( Fig. 2A). We next generated seven InDel molecular markers (M17, M24, M28, M220, M221, M223, and M31) in the candidate interval of 2 Mb between 63.5 and 65.5 Mb of chromosome 4. Using these InDel markers among 456 F 2 plants, we delimited the su mutation to the region between markers M28 and M220, at which 11 and 3 recombination events were detected, respectively ( Fig. 2B Table 3). We next amplified and sequenced the genomic sequences (including all introns and exons) of the 19 candidate genes. A comparative sequence alignment showed that there was a missense mutation with a C-T at 718 bp downstream of the predicted translation initiation site in ORF8 of su mutants, and the mutation site was heterozygous in F1 plants ( Fig. 2E; Supplementary Fig. 2). According to the annotated tomato genome (ITAG release 2.4), ORF8 encodes a heat shock protein of 819 amino acids. The N-terminal region of ORF8 contains a putative DnaJ domain and a chloroplast-targeting signal. The C-terminal region contains a transmembrane domain (TMD). The mutation in su mutants led to a premature stop codon (TGA) in the truncated protein with a deletion of the conserved TMD domain (Fig. 2E). Phylogenetic analyses showed that the homolog of ORF8 in Arabidopsis is AtARC6, which was reported to function in chloroplast division ( Supplementary Fig. 3) 37 . Together with the physiological phenotypes, we speculated that the mutation in su mutants could cause defective division and development of chloroplasts.

Functional verification of Su mutation
To further test ORF8 function in tomato, we knocked down ORF8 by VIGS. To this end, a fragment of ORF8 was inserted into the pTRV2 vector for infection. We used phytoene desaturase (PDS) in the same vector as the positive control, and empty pTRV2 was used as the negative control. The WT-like appearance of the negative control ( Fig. 3A-C) and photobleaching phenotype in the positive control ( Fig. 3D-F) indicated the effectiveness of VIGS of PDS. As expected, ORF8 silencing showed albinic stems without a leaf phenotype, which was similar to that in su mutants ( Fig. 3G-I). Consistent with the phenotype, the expression level of ORF8 was markedly decreased in both leaves and stems ( Supplementary Fig. 4).
To examine whether ORF8 functions in chloroplast division, we isolated mesophyll protoplasts from leaves and the epidermis of stems. Under microscopy, we observed many protoplasts with only one (or occasionally a number of) giant chloroplast in the ORF8 VIGS-silenced plants, which was in sharp contrast with multiple round chloroplasts in the protoplasts isolated from the negative control ( Fig. 3J, K). To further verify this result, we constructed orf8 mutants using the CRISPR/Cas9 technique (Fig. 4). Two homozygous CRISPR lines were obtained: line 2 with a G insertion and line 23 with a G deletion (Fig.  4A). Similar to the su mutants, the ORF8 CRISPR lines showed albinic stems (Fig. 4A, C). These results indicate that ORF8 is indeed the target gene.

Tissue specificity of chloroplast development in tomato
Loss of function of ARC6 in Arabidopsis was shown to disrupt the stabilization of the Z ring during chloroplast division 21,22 . However, it is unclear why the albinic phenotype was only observed in the stem of su mutants. Interestingly, in the tissue sections prepared by the vibratome, we indeed observed a remarkable alteration of chloroplast morphology in the leaves of su mutants. The mesophyll cells of the su mutants usually contained only one or two giant chloroplasts, which was apparently different from the multiple round-shaped chloroplasts in the WT mesophyll cells (Fig. 5A-F). In contrast to the case of the leaves, the cross-sections of the su stems showed that the chloroplasts almost all disappeared (Fig. 5G-J). Occasionally, a number of cells containing one or two giant chloroplasts were observed in stem cells, and those chloroplasts were~396 μm 2 (area) in size, which is 40 times that of the WT (Fig. 5K, L; Supplementary Fig. 5). To further confirm this observation, we performed highresolution imaging by transmission electron microscopy (TEM). The overall morphology and the number of chloroplasts in su mutants were consistent with the observation of the live tissue sections (Fig. 5M, P, S, V). In WT, the ultrastructure of chloroplasts appeared uniform, with clearly visible stacks of thylakoids (Fig. 5N, Q, T, W). However, the ultrastructure of chloroplasts in su mutants was fairly diverse under TEM (Fig. 5O, R, U, X). The lamellae comprising grana thylakoids and stroma thylakoids were sparse in su mutants (Fig. 5O, R, U, X). These results suggest that the absence of chloroplasts, as well as the sparse thylakoids and stromal thylakoids in the giant chloroplasts, contribute to the albinic stem phenotype in su mutants.

The role of SlARC6 in chloroplast division in meristematic cells
One prominent difference between the leaf and the stem is the cell division pattern during postembryonic growth. Leaf growth involves continuous cell division and cell expansion, while stem cells mostly undergo cell expansion. To understand whether cell division is the major reason for the distinct chloroplast phenotype, we examined meristematic cells. The typical meristem dome consists of three cell layers, in which most cells contain several immature vacuoles 38 . Chloroplasts are derived from proplastids, which feature two envelope membranes and limited stromal thylakoids 3,39 . In the meristem, proplastids and chloroplasts were found in almost all cells in both WT and su mutants, but some chloroplasts of su mutants also showed abnormal morphology (Fig. 6A-D). It is possible that chloroplast division and cell division are tightly associated. We next observed the lower part of the meristem where cell division is as rare as in the stem (as shown in Fig. 6G). The cells in this area featured large but not fully expanded vacuoles (Fig. 6E, F). Compared with WT cells, many cells in this area of su mutants contained abnormal chloroplasts (Fig. 6E, F). Compared with those in the meristem, most chloroplasts in mature stems disappeared, suggesting that the chloroplasts may divide concomitantly with cell division but do not divide when there is only cell expansion in the stem.

Light can partially rescue the albinic phenotype
In the WT, the adaxial and abaxial surfaces of the petiole seemed to have no noticeable difference in color (Fig. 7A, B). However, in su mutants, compared with the adaxial surface, the abaxial surface appeared to be yellowish (Fig. 7D, E). We then observed the chloroplasts in the fresh tissue sections of the petiole. Under differential interference contrast (DIC) microscopy, the epidermal cells of the petiole in WT were full of chlorophyll, so the cells appeared to be opaque, whereas the epidermal cells in su mutants seemed to be transparent due to the lack of chloroplasts (Fig. 7C, F). Interestingly, the adaxial surface of su mutants was more transparent than the adaxial side. Under the microscope, the chloroplast number within the cells on the adaxial side was significantly higher than those on the abaxial surface in su petioles (Fig. 7C, F). This distinction could derive from the different light received by adaxial and abaxial sides.
Light plays an important role in chloroplast development and chlorophyll biosynthesis 40,41 . To assess the role of light in the albinic phenotype, we cultured both WT and su mutants under conditions of 10 h day/14 h night or 16 h day/8 h night for 20 days. The WT and su mutants grown under longer lighting had greener leaves and stems than those grown on the 10 h day/14 h night (Fig. 7G-J, O-S). The stems of su mutants seemed to have the most significant change in color (Fig. 7G-J). In the freshly sectioned tissues, we observed a substantial rise in chloroplast-containing cells in su stems after prolonged light exposure, with cell chloroplast-containing cells elevated from~50% under 10 h day/14 h night conditions tõ 90% under 16 h day/8 h night conditions (Fig. 7G-N, U). In the protoplasts isolated from the epidermal cells of stems, we found that the proportion of chloroplastcontaining cells increased from 38 to 52% with prolonged light exposure, confirming that light could promote the formation of chloroplasts. To further verify this, we isolated protoplasts from the stems of su mutants grown and then exposed the protoplasts to light for 5 h. Our quantification indicated that the proportion of chloroplast-containing protoplasts increased from 38 to 48% (Fig. 7U). In addition, we quantified the average number of chloroplasts within the protoplasts of su mutants. The results showed that up to 99% of the su protoplasts contained one or two chloroplasts at 10 h day/14 h night, while~27% of the protoplasts contained three or more chloroplasts at 16 h day/8 h night (Fig. 7O-S, V). This result indicates that light can promote chloroplast development in su mutants.

Discussion
Chloroplasts are biological factories where plants transform solar energy into organic substances for plant growth and development. Thus, it is important to maintain the appropriate number and physiology of chloroplasts. In Arabidopsis, a number of mutants named ARCs were reported to have defective chloroplast division. In Arabidopsis mutants of ARC family members, larger chloroplasts in mesophyll cells were observed 2,4,5 . In Arabidopsis, arc6 mutants have abnormal chloroplasts in mesophyll cells but no other dramatic changes in whole plants 21 . Here, we found that the mutation of SlACR6 in tomato leads to the albinic phenotype in stems. A similar phenotype was previously reported in three su mutants that had slightly paler leaves and albinic stems 42 . Interestingly, the results here demonstrated phenotypic variability among the three accessions, implying that su mutations with different backgrounds could affect the phenotype 42 .
In higher plants, chloroplasts divide and replicate via a contractile division complex including the FtsZ ring (Z ring), a protein complex located on the stromal surface of the inner envelope membrane, and the ARC5 ring, a protein complex located on the cytosolic surface of the OEM [10][11][12][13][14][15][16][17][18] . ARC6 has evolved from a prokaryotic cell division protein and bears three conserved domains: the N-terminal region, which protrudes into the stroma and directly interacts with FtsZ2; the C-terminal region, which extends into the intermembrane space and interacts with PDV2; and a TMD 27 . The ARC6-FstZ1 interaction is required for the Z ring to localize to the stromal surface of the inner envelope membrane 23,24,27 . In su mutants, the absence of the TM domain and the C-terminal region of SlACR6 may result in failed localization on the inner membrane of chloroplasts, which also prevents the localization of the Z ring mid-chloroplast. However, the number of chloroplasts was dramatically different in the leaves and stems of su mutants. In mature mesophyll cells, chloroplasts inherited from mother cells propagate through chloroplast division. Previous observations in Arabidopsis arc6 mutants showed that the cells in the shoot apical meristem, leaf primordium, and mature leaves all contain two larger plastids, suggesting that plastids could divide during cell division 2,21,37 . In the leaves of tomato su mutants, we found a similar phenotype but rarely observed any chloroplasts in stems, suggesting that chloroplast development was entirely blocked in expanding cells. Based on these findings, we speculated that the tissue specificity of plastid development is likely not caused by differential ACR6 functions but instead by the distinct cell division patterns between leaves and stems. The mutation of ACR6 provides a good example to gain insight into the spatial and tissue-specific regulation of chloroplast development.
In Arabidopsis, overexpression of PDV1 and PDV2 can increase the number of chloroplasts. Interestingly, expression of PDV1 and PDV2 can be promoted by exogenous cytokinin treatment or overexpression of cytokininresponsive transcription factor 2 43,44 . In addition to cytokinin, GA-deficient mutants of Arabidopsis (ga1-3) and Oryza sativa (d18-AD) both exhibited reduced chloroplast division and decreased expression of FtsZ2, ARC6, DRP5B, and PDV 45 . As GA and cytokinin coordinate to regulate the development of stems and leaves from the shoot apical meristem, the varied distribution of these two hormones in different tissues could be involved in the tissue specificity of chloroplast division. In addition to hormones, FHY3, a key regulator of far-red light signaling, was reported to activate the expression of ARC5, and the large chloroplast phenotype in fhy5 mutants could be rescued by expression of ARC5 46,47 . This result suggests that light could also be a key regulator of chloroplast division. The partial complementation of su mutants by extended lighting provides further evidence that light is involved in ARC-regulated chloroplast division.

Plant materials and growth condition
Spontaneous su mutants LA0628 and LA1589 and Ailsa Craig (AC, accession number LA2838A) were provided by the Tomato Genetics Resource Center (http://tgrc.ucdavis. edu). The mapping population of Su was constructed by crossing the su mutant and LA1589. In addition, the F2 population was obtained from the F1 selfing. All plants were cultured in a greenhouse at 26°C.

Determine of chlorophyll content
A 0.2 g sample was placed in a 2 mL centrifuge tube with liquid nitrogen and then ground into powder.
Chlorophyll was extracted in methanol, and the absorbance was detected using a microplate fluorometer at 666, 653, and 470 nm. The chlorophyll contents were calculated using the following formulas: Chl a (mg/g FW

Bulked segregant analysis
Individuals (>30) exhibiting the su mutant-like phenotype and individuals (>30) exhibiting the WT phenotype were collected from the F 2 population generated by crossing the su mutant and LA1589. Their genomic DNA was extracted by CTAB. Two micrograms of DNA was mixed to construct two samples (mutant-like and WT samples). Genome sequencing of the two samples was performed by HiSeqXten-PE150 (Novogene, Beijing) with a depth of 30× coverage of the tomato genome. The candidate region was analyzed according to the method in ref. 49 .

Map-based cloning
InDel molecular markers were designed in the candidate region according to the resequenced DNA and were polymorphic between the su mutant and LA1589 50 (Supplementary Table 3). First, seven markers were analyzed using 456 individuals exhibiting a su mutant-like phenotype of the F 2 population. The candidate region was determined according to the recombinants. Then, six markers in the candidate region were analyzed using 515 individuals exhibiting a su mutant-like phenotype. Finally, the candidate gene was delimited in a smaller interval.

Virus-induced gene silencing
A fragment of ARC6 was designed using the VIGS tool (https://vigs.solgenomics.net) and inserted into the pTRV2 vector (named TRV2-ARC6). A PDS gene was also inserted into the pTRV2 vector and used as the positive control. pTRV1, pTRV2, pTRV2-PDS, and pTRV2-ARC6 vectors were transferred into Agrobacterium tumefaciens (GV3010) and then grown on LB plates with 100 μg/mL kanamycin and 50 μg/mL rifampicin. When tomato plants had two fully expanded cotyledons, we injected the cotyledons with an A. tumefaciens suspension. The details for preparing the A. tumefaciens suspension and injection are the same as those in ref. 51 . The plants were cultured at 24°C in a growth chamber. After 20 days, the phenotype was recorded.

Construction of knockout plants
To generate the CRISPR/Cas9-Slorf8 construct, we inserted two target sites of Slorf8 (http://skl.scau.edu.cn/ targetdesign) into the pTX vector using the Clone Express II One Step Cloning Kit (Vazyme Biotech C112-01/02). The construct was introduced into tomato cv. Micro-Tom by Agrobacterium (A. tumefaciens) mediated transformation. Homozygous transgenic plants were used for phenotypic characterization.

Vibratome sectioning
Agarose (5%) was boiled in a microwave oven, and the agarose solution was cooled to~65°C before being poured into a tube. Subsequently, a specimen was suspended in the agarose solution, and the air bubbles were aspirated. The tube was put at 4°C for hardening. Specimen blocks were cut into 80 µm sections by a vibratome. The vibratome sections were observed by DIC microscopy

Transmission electron microscopy
Fresh samples were cut into 1 mm × 3 mm strips and fixed in 2.5% glutaraldehyde overnight at 4°C. The glutaraldehyde was removed, and the samples were washed three times with phosphate buffer (0.1 M, pH 7.0). The samples were fixed again in 1% (wt/vol) osmium tetroxide for 1.5 h at room temperature. Osmium tetroxide was removed, and the samples were washed three times with phosphate buffer (0.1 M, pH 7.0). After that, the samples were dehydrated by using successive treatments in 50, 70, 80, 90, 95, and 100% (all vol/vol) ethanol for 20 min and 100% (all vol/vol) acetone for 20 min. Then, the samples were embedded in Spurr's low-viscosity resin. After polymerization for 24 h at 70°C, the ultrathin sections were cut on an ultramicrotome with a diamond knife and picked up on copper grids. Before observation, the samples were stained with uranyl acetate and lead citrate. Images were captured using a Hitachi-7650 transmission electron microscope.