REDUCED CHLOROPLAST COVERAGE genes from Arabidopsis thaliana help to establish the size of the chloroplast compartment

Significance Mechanisms that determine the cellular volume allocated to organelles are largely unknown. We demonstrate that in the plant Arabidopsis thaliana, a small gene family that encodes proteins of unknown function contributes to a mechanism that establishes the proportion of cellular volume devoted to chloroplasts. We show that this mechanism resides outside of the chloroplast by demonstrating that the protein that makes the greatest contribution to this mechanism resides in the cytoplasm and nucleus and that the trafficking of this protein between the cytoplasm and nucleus may regulate this mechanism. A deeper understanding of this mechanism may lead to the rational manipulation of chloroplast compartment size, which may lead to more efficient photosynthesis and increased yields from important crop plants. Eukaryotic cells require mechanisms to establish the proportion of cellular volume devoted to particular organelles. These mechanisms are poorly understood. From a screen for plastid-to-nucleus signaling mutants in Arabidopsis thaliana, we cloned a mutant allele of a gene that encodes a protein of unknown function that is homologous to two other Arabidopsis genes of unknown function and to FRIENDLY, which was previously shown to promote the normal distribution of mitochondria in Arabidopsis. In contrast to FRIENDLY, these three homologs of FRIENDLY are found only in photosynthetic organisms. Based on these data, we proposed that FRIENDLY expanded into a small gene family to help regulate the energy metabolism of cells that contain both mitochondria and chloroplasts. Indeed, we found that knocking out these genes caused a number of chloroplast phenotypes, including a reduction in the proportion of cellular volume devoted to chloroplasts to 50% of wild type. Thus, we refer to these genes as REDUCED CHLOROPLAST COVERAGE (REC). The size of the chloroplast compartment was reduced most in rec1 mutants. The REC1 protein accumulated in the cytosol and the nucleus. REC1 was excluded from the nucleus when plants were treated with amitrole, which inhibits cell expansion and chloroplast function. We conclude that REC1 is an extraplastidic protein that helps to establish the size of the chloroplast compartment, and that signals derived from cell expansion or chloroplasts may regulate REC1.

Eukaryotic cells require mechanisms to establish the proportion of cellular volume devoted to particular organelles. These mechanisms are poorly understood. From a screen for plastid-to-nucleus signaling mutants in Arabidopsis thaliana, we cloned a mutant allele of a gene that encodes a protein of unknown function that is homologous to two other Arabidopsis genes of unknown function and to FRIENDLY, which was previously shown to promote the normal distribution of mitochondria in Arabidopsis. In contrast to FRIENDLY, these three homologs of FRIENDLY are found only in photosynthetic organisms. Based on these data, we proposed that FRIENDLY expanded into a small gene family to help regulate the energy metabolism of cells that contain both mitochondria and chloroplasts. Indeed, we found that knocking out these genes caused a number of chloroplast phenotypes, including a reduction in the proportion of cellular volume devoted to chloroplasts to 50% of wild type. Thus, we refer to these genes as REDUCED CHLOROPLAST COVERAGE (REC). The size of the chloroplast compartment was reduced most in rec1 mutants. The REC1 protein accumulated in the cytosol and the nucleus. REC1 was excluded from the nucleus when plants were treated with amitrole, which inhibits cell expansion and chloroplast function. We conclude that REC1 is an extraplastidic protein that helps to establish the size of the chloroplast compartment, and that signals derived from cell expansion or chloroplasts may regulate REC1. chloroplast coverage | chloroplast | plastid signaling | chlorophyll | Arabidopsis C hloroplasts drive plant growth, development, and reproduction by converting solar energy into biologically useful forms of energy. Thus, the biogenesis and function of chloroplasts underpin crop yields and, indeed, life on Earth. Chloroplasts develop from proplastids during germination and leaf development (1). After chloroplast biogenesis, chloroplasts divide by binary fission. A number of mutant alleles enhance or reduce the size of individual chloroplasts by attenuating or promoting chloroplast division (2). Regardless of the size of individual chloroplasts, the proportion of cellular volume devoted to all chloroplasts appears indistinguishable from wild type in these mutants (3)(4)(5). Thus, the mechanism that establishes the size of the chloroplast compartment appears distinct from the mechanisms of chloroplast division.
The cell expansion that drives the expansion of leaves also drives the proliferation of chloroplasts. Indeed, the proliferation of chloroplasts is so tightly correlated with cell expansion that the ratio of the size of the chloroplast compartment to the size of mesophyll cells is constant, regardless of cell size (2,6,7). Cell type exerts a major influence over the proportion of cellular volume devoted to the chloroplast. For instance, the size of the chloroplast compartment in mesophyll cells is larger than in epidermal cells. Thus, an extraplastidic mechanism appears to determine the size of the chloroplast compartment (6). However, during the expansion of leaves, chloroplasts are not completely submissive to the cell. Indeed, chloroplast dysfunction inhibits the expansion of leaves (8).
Although mechanisms that establish the proportion of cellular volume devoted to particular organelles are of fundamental importance to biology, these mechanisms remain poorly understood (2). In the particular case of chloroplasts, understanding these mechanisms may lead to significant advances for agriculture. For example, introducing C 4 photosynthesis into rice, a plant that performs C 3 photosynthesis, is one strategy for potentially increasing yields from this important crop (9). C 3 and C 4 leaves are distinct at the metabolic, cellular, and anatomical levels (9,10). One of the conserved features of C 4 leaves is the increase and decrease in the size of the chloroplast compartment in bundle sheath and mesophyll cells, respectively, relative to C 3 leaves (10). The engineering of C 4 photosynthesis in important C 3 crops, such as rice, is thought to depend on the ability to rationally manipulate the size of the chloroplast compartment (11).
We performed a screen for plastid-to-nucleus signaling mutants in Arabidopsis thaliana (12). From this screen, we obtained one mutant allele of a gene that encodes a protein of unknown function.

Significance
Mechanisms that determine the cellular volume allocated to organelles are largely unknown. We demonstrate that in the plant Arabidopsis thaliana, a small gene family that encodes proteins of unknown function contributes to a mechanism that establishes the proportion of cellular volume devoted to chloroplasts. We show that this mechanism resides outside of the chloroplast by demonstrating that the protein that makes the greatest contribution to this mechanism resides in the cytoplasm and nucleus and that the trafficking of this protein between the cytoplasm and nucleus may regulate this mechanism. A deeper understanding of this mechanism may lead to the rational manipulation of chloroplast compartment size, which may lead to more efficient photosynthesis and increased yields from important crop plants.
This gene is homologous to two Arabidopsis genes that encode proteins of unknown function and to FRIENDLY. FRIENDLY and its orthologs promote the normal distribution of mitochondria in Arabidopsis and in nonphotosynthetic organisms (13). However, these three Arabidopsis homologs of FRIENDLY are found only in photosynthetic organisms. Based on these data, we thought that FRIENDLY may have expanded into a small gene family to help manage the energy metabolism of cells that contain both chloroplasts and mitochondria. We tested this idea by examining the phenotypes of mutants in which one, two, three, or all four of these genes are knocked out. We found that these mutants exhibited a number of chloroplast phenotypes, including a smaller chloroplast compartment relative to wild type. Thus, we named these genes REDUCED CHLOROPLAST COVERAGE (REC). We also found that the protein that contributes most to establishing the size of the chloroplast compartment, REC1, localizes to both the nucleus and the cytosol, and we provide evidence that signals derived from dysfunctional chloroplasts or the inhibition of cell expansion may regulate the nucleocytoplasmic partitioning of REC1.

Results
Cloning and Characterizing rec1-1. We obtained one mutant allele of REC1 that we named rec1-1 from a screen for genomes uncoupled (gun) mutants (12). gun alleles disrupt the plastid-to-nucleus signaling that down-regulates photosynthesis-associated nuclear gene (PhANG) expression when chloroplast biogenesis is blocked. Thus, gun mutants express higher levels of PhANGs than wild type when chloroplast biogenesis is blocked (8). In addition to a gun phenotype, rec1-1 greens after seedlings are grown in far-red light and then transferred to white light ( Fig. 1A and SI Appendix, Fig. S1A). When wild-type seedlings are grown in far-red light, they overaccumulate the chlorophyll precursor protochlorophyllide. In white light, these elevated levels of protochlorophyllide block chloroplast biogenesis by inducing increases in the levels of singlet oxygen (14,15). We mapped this greening phenotype to a 70-kb interval near the top of chromosome 1 (SI Appendix, Fig. S1B). We sequenced seven genes in this interval. We found a C-to-T transition that causes a nonsense mutation in the codon that encodes Q1415 of At1g01320 (SI Appendix, Fig. S1 C and D). At1g01320 encodes a protein of unknown function with a calculated mass of 200 kDa. This protein contains three tetratricopeptide repeats (TPRs) and 106 residues near the carboxyl terminus that are 29% proline (SI Appendix, Fig. S1D). TPRs contribute to the protein-protein interactions that underpin diverse biological processes (16).
We obtained three publicly available T-DNA insertion mutants with insertions in At1g01320. We named these T-DNA insertion alleles rec1-2, rec1-3, and rec1-4. We found reduced levels of mRNA transcribed from At1g01320 in all of these mutants (SI Appendix, Fig. S2). We also analyzed these mutants by immunoblotting with affinity-purified antibodies raised against fragments of the protein encoded by At1g01320 that spanned from P1419 to F1673 (SI Appendix, Fig. S3A). When whole-leaf extracts were analyzed by immunoblotting, these antibodies recognized a single band in an extract that was prepared from wild type (SI Appendix, Fig. S3B). Based on the mobility of this protein in a 5% SDS gel (SI Appendix, Fig. S3D), we estimated that its mass is 240 kDa. This 240-kDa protein does not accumulate to detectable levels in rec1-1, rec1-2, rec1-3, or rec1-4 (SI Appendix, Fig. S3 B and C).
We found that, similar to the rec1-1 allele, the three T-DNA insertion alleles inhibited the far-red block of greening (Fig. 1A). We found that the chlorophyll levels in rec1-1 were 71% of wild type. In contrast, we found that the chlorophyll levels of rec1-2, rec1-3, and rec1-4 were 57-58% of wild type (Fig. 1B) and 81-82% of rec1-1 (P = 0.001-0.006). We conclude that rec1-1 attenuated the accumulation of chlorophyll and that rec1-2, rec1-3, and rec1-4 are null alleles. These data are consistent with rec1-1 expressing a truncated protein that promotes the accumulation of chlorophyll. We were not able to detect such a truncated protein because our antibodies bind residues that are carboxyl-terminal to Q1415.
To test whether the three T-DNA insertion mutants are gun mutants, we grew them on media that contained either norflurazon or lincomycin. Norflurazon and lincomycin specifically block chloroplast biogenesis by distinct mechanisms and severely down-regulate the expression of PhANGs, such as the genes encoding the light harvesting chlorophyll a/b-binding proteins of photosystem II (Lhcb) and the small subunit of RuBisCO (RbcS) (8). We found that rec1-1 accumulated 3.2-to 4.3-fold more Lhcb1.2 than wild type and that rec1-2, rec1-3, and rec1-4 accumulated significantly more Lhcb1.2 mRNA than wild type (SI Appendix, Fig. S4). However, rec1-2, rec1-3, and rec1-4 accumulated significantly less Lhcb1.2 mRNA than rec1-1 (P < 0.0001-0.009) (SI Appendix, Fig. S4). We observed a similar trend with RbcS1A expression (SI Appendix, Fig. S4). In contrast, gun1-101 (12), a null allele of a relatively well studied GUN gene (8), induced 13-and 35-fold increases in Lhcb1.2 expression when chloroplast biogenesis was blocked with norflurazon and lincomycin treatments, respectively (SI Appendix, Fig. S4). These data provide evidence that norflurazon treatments activate a repressive plastid-to-nucleus signaling mechanism that is not activated by lincomycin treatments or that an inductive mechanism present in lincomycin-treated seedlings is absent in norflurazon-treated seedlings.
In untreated seedlings, Lhcb1.2 mRNA accumulated to significantly lower levels in rec1-2, rec1-3, and rec1-4 than in wild type (SI Appendix, Fig. S4), which is consistent with the chlorophyll-deficient phenotypes of these mutants (Fig. 1B). The levels of RbcS1A mRNA were not significantly different in the untreated wild type and rec1 mutants (SI Appendix, Fig. S4). Based on these data, we conclude that these rec1 alleles specifically disrupted the plastid-regulated expression of these PhANGs and that At1g01320 is REC1.
Our characterization of these rec1 alleles indicates that rec1-1 behaves as a loss-of-function allele when the far-red block of greening and the chlorophyll accumulation phenotypes are scored and that rec1-1 enhances the levels of PhANG expression relative to other mutant alleles of REC1 when the gun phenotype is scored. A truncated REC1 protein that is partially active may underpin these phenotypes. There is precedence for mutant alleles causing both loss-of-function and gain-of-function phenotypes (17,18).  The far-red block of greening phenotype of four rec1 mutants. A far-red block of greening experiment was performed as described in SI Appendix, Materials and Methods. The proportions of seedlings that greened are indicated (n = 188-234; numbers were pooled from two biological replicates). The asterisk indicates a statistically significant difference relative to wild type (Col-0) (P < 0.0001). (B) Chlorophyll phenotypes of rec1 mutants. Plants were grown on soil for 24 d. Six biological replicates were analyzed for wild type (Col-0) and each mutant. Error bars indicate 95% confidence intervals. The asterisk indicates a statistically significant difference relative to wild type (Col-0) (P < 0.0001).
Characterization of the REC Gene Family Mutants. The REC1 protein is homologous to three other Arabidopsis proteins that are encoded by At4g28080, At1g15290, and FRIENDLY (At3g52140) (SI Appendix, Fig. S5). The proteins encoded by At4g28080 and At1g15290 have no known function. Loss-of-function alleles of FRIENDLY and its orthologs in Dictyostelium discoideum, Saccharomyces cerevisiae, and Drosophila melanogaster cause mitochondrial clustering (19). FRIENDLY and orthologous proteins were reported to perform a variety of functions, such as binding and regulating atypical protein kinase C (20), binding mRNAs that encode mitochondrial proteins and promoting the biogenesis of mitochondria (21), and promoting intermitochondrial associations before mitochondrial fusion (13). The calculated masses of the proteins that are encoded by At4g28080, At1g15290, and FRIENDLY are 200 kDa, 180 kDa, and 150 kDa, respectively. We named At4g28080 and At1g15290 REC2 and REC3, respectively, based on their derived amino acid sequence similarity to REC1. REC1 is more similar to REC2 and REC3 than to FRIENDLY (SI Appendix, Fig. S5). Homologs of REC1, REC2, REC3, and FRIENDLY are present in other plant species (SI Appendix, Fig. S6). In contrast, eukaryotes that do not perform photosynthesis contain only orthologs of FRIENDLY, usually as a single-copy gene (SI Appendix, Fig. S6).
To test whether REC2, REC3, and FRIENDLY contribute to the same or different processes as REC1, we obtained T-DNA insertion alleles for each of these genes. We found that T-DNA insertions reduced the levels of mRNA transcribed from these genes (SI Appendix, Fig. S2). In REC3 mutants, the low levels of mRNA are not expected to encode functional proteins because the insertions are in exons. In REC2 and FRIENDLY mutants, the reduced levels of mRNA might express truncated proteins that are not functional. Indeed, three independent T-DNA insertions into REC2 caused chlorophyll to accumulate at significantly lower levels than in wild type, and these reduced levels of chlorophyll were not significantly different from each other (P = 0.3-0.6) (SI Appendix, Fig. S7A). We conclude that these three alleles are nulls. Based on the robust mitochondrial clustering phenotype of the T-DNA insertion allele of FRIENDLY (SI Appendix, Fig. S7B), we conclude that this allele is either a null or a strong loss-of-function allele.
To determine the full impact of these four genes on chloroplastrelated processes, we prepared double, triple, and quadruple mutants with rec1-3, rec2, rec3-1, and friendly in all possible combinations. We found that rec2, rec3-1, rec3-2, and friendly exhibited a far-red block of greening that was more similar to wild type than to rec1-1 (SI Appendix, Fig. S8A). We found that after 3 d of growth in far-red light, protochlorophyllide accumulated to significantly lower levels in rec1-3 than in wild type, protochlorophyllide accumulated to significantly higher levels in friendly than in wild type, and protochlorophyllide levels were not significantly different from wild type in rec2 and rec3-1 (SI Appendix, Fig. S8B). Significantly less protochlorophyllide accumulated in double and triple mutants containing rec1-3 and the quadruple mutants relative to the relevant single, double, and triple mutants (P < 0.0001-0.005). Significantly more protochlorophyllide accumulated in rec3-1 friendly, rec1-3 rec3-1 friendly, rec1-3 rec2 friendly, and rec1-3 rec2 rec3-1 friendly than in rec3-1, rec1-3 rec3-1, rec1-3 rec2, and rec1-3 rec2 rec3-1, respectively (P = 0.0003-0.02). We conclude that mutant alleles of REC1 inhibit the far-red block of greening by reducing the levels of protochlorophyllide.
Attenuating phytochrome A signaling inhibits the far-red block of greening (15). Perhaps the simplest way to test for defects in phytochrome A signaling is to measure hypocotyl lengths in farred light. We found that, in the dark, the lengths of the rec1-3 hypocotyls were not significantly different from wild type but that those of rec1-3 rec2 rec3-1 friendly were 11% shorter (SI Appendix, Fig. S8C). However, we found that in 3 μmol·m −2 ·s −1 far-red light, the hypocotyls of rec1-3 and rec1-3 rec2 rec3-1 friendly were not significantly different from wild type (P = 0.4-0.5) (SI Appendix, Fig. S8D). We conclude that the inhibition of the far-red block of greening in rec1 mutants is probably not caused by defects in phytochrome A signaling.
We found that when chloroplast biogenesis was blocked with a lincomycin treatment, Lhcb1.2 expression in rec2 and rec3-1 was 69% and 68% of the levels observed in wild type, respectively. In friendly, RbcS1A expression was increased 1.5-fold but Lhcb1.2 expression was not significantly different from wild type (SI Appendix, Fig. S8E). Double and triple mutants containing rec1-3 and friendly tended to exhibit gun phenotypes, as did the quadruple mutant, whereas the double and triple mutants containing rec2 and rec3-1 generally did not exhibit gun phenotypes (SI Appendix, Fig. S8E). The most robust gun phenotype was observed for rec1-3 rec2 friendly, which accumulated 3.3-and 2.2fold more mRNA from Lhcb1.2 and RbcS1A, respectively, than wild type (SI Appendix, Fig. S8E). Thus, this gene family makes significant but minor contributions to the plastid regulation of these PhANGs relative to GUN1 (SI Appendix, Fig. S4).
The chloroplast ultrastructure in many of the mutants resembled wild type (SI Appendix, Fig. S10). The chloroplast ultrastructures of chlorophyll-deficient mutants typically resemble wild type (22,23). The most severely chlorophyll-deficient mutants (Fig. 2) tended to have long and thin chloroplasts relative to wild type (SI Appendix, Fig. S11). Similar phenotypes were not reported for tetrapyrrole biosynthesis mutants, such as the severely chlorophyll-deficient gun4 hy1 and gun5 hy1 double mutants (22). Holes were observed in the chloroplasts of rec1-3 rec2 rec3-1 that were surrounded by double membranes (SI Appendix, Fig. S12). These holes probably resulted from the sectioning of cytosolic protrusions into the chloroplasts that do not disrupt the chloroplast double membrane. Consistent with this interpretation, a mitochondrion was found in one of these holes (SI Appendix, Fig. S12).
The grana thylakoids of the mutants appeared similar to wild type, but rec1 rec2 rec3-1 contained fewer grana thylakoids than wild type (SI Appendix, Fig. S13). A reduction in the number of grana thylakoids is not commonly observed in Arabidopsis mutants unless the chlorophyll deficiencies are severe (22)(23)(24). The thylakoid membranes appeared swollen in four friendly mutants (SI Appendix, Fig. S14) and similar to wild type in the other friendly mutants (SI Appendix, Fig. S13). Swelling of the thylakoid membranes was not reported previously for friendly mutants (13,25). These data are consistent with complex interactions among these alleles affecting the ability of the thylakoid membranes to withstand osmotic pressure (26).
To further characterize the chloroplast defects of these mutants, we imaged the chloroplasts in live cells from the abaxial epidermal cells to the cortical region of the spongy mesophyll cells in leaves using confocal laser scanning microscopy. We found a number of differences relative to wild type. These differences included a tendency of the chloroplasts to localize along the anticlinal walls of mesophyll cells, enhanced chlorophyll fluorescence in both the mesophyll and epidermal cells, and fewer chloroplasts in the mesophyll cells in particular triple mutants and the quadruple mutant ( Fig. 3A and SI Appendix, Fig. S15). The chloroplasts of the mesophyll cells residing along the anticlinal walls and the increases in chlorophyll fluorescence are consistent with increased sensitivity to light (27,28). To test whether a potentially enhanced sensitivity to light led to a rise in the levels of reactive oxygen species, we stained leaves with 3,3′-diaminobenzidine (DAB) and nitrotetrazolium blue (NBT) to detect hydrogen peroxide and superoxide, respectively. These mutants appear to accumulate variable levels of reactive oxygen species (SI Appendix, Fig. S16).
The chlorophyll biosynthesis mutants cch and gun4-1 (29, 30) were analyzed to test whether chlorophyll deficiencies might cause similar phenotypes. cch and gun4-1 were previously reported to accumulate only 30-40% of the chlorophyll found in wild type (30). Similar to the REC gene family mutants, the chloroplasts of cch and gun4-1 appeared along the anticlinal walls in mesophyll cells (Fig. 3A), which is consistent with the enhanced sensitivities of chlorophyll-deficient mutants to light (31). cch exhibited enhanced chlorophyll fluorescence relative to gun4-1 and wild type (SI Appendix, Fig. S15). Thus, chlorophyll deficiencies were not correlated with increases in chlorophyll fluorescence.
The distribution of chloroplasts within the mesophyll cells in many of these mutants appeared to differ from wild type (Fig. 3A). To quantify these differences, we calculated a fractal dimension (D F ) and a lacunarity parameter (Λ) for each image (32). Calculating D F allowed us to compare the geometric complexity of the distribution of chloroplasts in wild type and each mutant (32). Calculating Λ allowed us to compare the heterogeneity (e.g., the gaps) in the distribution of chloroplasts in wild type and each mutant (32). Significant differences in the value of the D F were observed in nine of these mutants (Fig. 3B). The values for D F were not significantly different from wild type in rec1-3, rec1-3 rec3-1, and mutants containing both rec3-1 and friendly, namely rec3-1 friendly, rec1-3 rec3-1 friendly, and rec2 rec3-1 friendly (Fig. 3B). In contrast to other mutants containing both rec3-1 and friendly, the D F of rec1-3 rec2 rec3-1 friendly was significantly less than wild type (Fig. 3B). This difference may reflect the decrease in the number of the chloroplasts in rec1-3 rec2 rec3-1 friendly relative to other mutants that contain both rec3-1 and friendly (Fig. 3A). Chlorophyll deficiencies do not explain the differences in geometric complexities of these images because although the D F values of the images from cch were significantly less than wild type, the D F values of the images from gun4-1 were not significantly different from wild type (Fig. 3B). Λ was significantly reduced in all of the mutants except rec2 and rec1-3 rec3-1 (Fig. 3C). Although Λ was significantly less in gun4-1 than in wild type, Λ was not significantly different between cch and wild type (Fig. 3C). Based on these data, we conclude that significant differences in the gaps of the chloroplast networks of the rec and friendly mutants were not caused by chlorophyll deficiencies. Our analysis of D F and Λ provides evidence that the rec alleles, friendly, and combinations of the rec and friendly alleles affect the distribution of chloroplasts. Differences in the numbers, sizes, and shapes of these chloroplasts may account for some of these differences in D F and Λ.
Mutants with abnormal distributions of chloroplasts often exhibit abnormal chloroplast movements because of deficiencies in the machinery that helps chloroplasts track along the cytoskeleton. Abnormal chloroplast movements are readily detected with chloroplast photorelocation experiments (27). We found that chloroplast photorelocation was indistinguishable between rec1-3 rec2 rec3-1 friendly and wild type (SI Appendix, Fig. S17 and Movies S1 and S2). We also imaged stromules by targeting yellow fluorescent protein (YFP) to the plastids because the number of stromules is reduced when the association of stromules and the cytoskeleton is disrupted (33). We observed no significant differences in the appearance or number of stromules in wild type, rec1-3, and rec1-3 rec2 rec3-1 (P = 0.1-0.5) (SI Appendix, Fig. S18).
To quantify the reduced chloroplast compartment size phenotypes that we observed in live cells from the cortical region of the spongy mesophyll, we fixed leaf sections with glutaraldehyde and quantified the plan areas of fixed mesophyll cells and their chloroplasts (3). In the mutants, the size of the chloroplast compartment in the glutaraldehyde-fixed cells appeared reduced relative to wild type, especially in particular triple mutants and the quadruple mutant (Fig. 4A). The number of chloroplasts was correlated with the mesophyll cell plan area in each mutant and wild type (SI Appendix, Fig. S19). This correlation was previously demonstrated for wild type (2,3,7). We found that in many of the mutants, the number of chloroplasts per cell plan area was reduced relative to wild type. The greatest reductions in the number of chloroplasts per cell plan area were 44-62%, observed in rec1-3 rec2, rec1-3 rec2 friendly, rec1-3 rec2 rec3-1, and rec1-3 rec2 rec3-1 friendly (Fig. 4B and SI Appendix, Fig. S19). The reductions in the number of chloroplasts per cell plan area exhibited by rec1-3 rec3-1 and rec1-3 rec3-1 friendly (24-25%) were significantly greater than in the relevant single mutants (P < 0.0001-0.03) (Fig. 4B). Without rec1-3, combinations of rec2, rec3-1, and friendly did not significantly decrease the number of chloroplasts per cell plan area. In most instances, friendly did not significantly affect the number of chloroplasts per cell plan area (P = 0.0714-0.9198). However, a significant increase in the number of chloroplasts per cell plan area was observed in rec2 rec3-1 friendly relative to rec2 rec3-1 (P = 0.0003). Based on these data, we conclude that rec1-3 attenuates the number of chloroplasts per cell plan area more than the other mutant alleles tested and that friendly promotes the number of chloroplasts per cell plan area in a specific genetic context. The mutants with no or small (15-25%) reductions in the number of chloroplasts per cell plan area also exhibited small (7-14%) reductions in chloroplast plan area relative to wild type (Fig. 4C). The largest reduction in chloroplast plan area was 22% observed in rec2 rec3-1 friendly. However, rec2 rec3-1 friendly was unusual relative to the other mutants tested in that it also exhibited a 17% increase in the number of chloroplasts per cell plan area relative to wild type (Fig. 4B). In contrast to the other mutants, the three most chlorophyll-deficient REC gene family mutants (rec1-3 rec2, rec1-3 rec2 rec3-1, and rec1-3 rec2 rec3-1  friendly; Fig. 2) exhibited 31-46% increases in their chloroplast plan areas relative to wild type (Fig. 4C). The increases in chloroplast plan area observed in rec1-3 rec2, rec1-3 rec2 rec3-1, and rec1-3 rec2 rec3-1 friendly are small compared with the 500% to ∼2,000% enlargements in chloroplast plan area that were reported for chloroplast division mutants (2,5,34). These findings are consistent with the idea that rec and friendly alleles disrupt mechanisms that regulate chloroplast division rather than those that perform chloroplast division. Numerous signals appear to regulate chloroplast division (2).
Consistent with the observed large reductions in the number of chloroplasts per cell plan area and either no increases or small increases in chloroplast plan area, we found that chloroplast coverage (total chloroplast plan area per cell plan area) was significantly reduced in many of the mutants compared with wild type (Fig. 4D and SI Appendix, Fig. S20). Moreover, we observed a significant interaction between the presence of rec1-3 and the number of mutant alleles (P < 0.00001). In the presence of rec1-3, chloroplast coverage fell as the number of mutant alleles increased (Fig. 4D). The greatest reduction in chloroplast coverage (50%) was observed in the quadruple mutant (Fig. 4D).
In contrast, when plants contained REC1, adding further mutant alleles did not significantly reduce chloroplast coverage, regardless of the number of mutant alleles present (Fig. 4D).
We conclude that both the number of chloroplasts per cell plan area and chloroplast plan area are probably controlled by a number of mechanisms that regulate chloroplast division, and that these signals include light signaling and signals induced by chlorophyll deficiencies. Chlorophyll deficiencies do not completely explain these phenotypes because, although the chloroplast plan areas were larger than wild type in cop1-4, det1-1, phyA phyB, rec1-3 rec2, rec1-3 rec2 rec3-1, and rec1-3 rec2 rec3-1 friendly, the chloroplast plan areas were smaller than wild type in cch and gun4-1. Additionally, we conclude that, in general, altering light-regulated development and significantly attenuating the accumulation of chlorophyll do not affect chloroplast coverage in Arabidopsis mesophyll cells. We also conclude that chlorophyll deficiencies do not explain the reduced chloroplast coverage phenotypes of the rec mutants.
To test whether the levels of the REC1 protein limit the size of the chloroplast compartment, we used the 35S promoter from cauliflower mosaic virus to drive the overexpression of REC1 in sgs3 rdr6-11-a double mutant that is resistant to transgeneinduced gene silencing (37). We found three independent transgenic lines that accumulated more REC1 protein than sgs3 rdr6-11 (SI Appendix, Fig. S28). Chloroplast coverage increased from 11% to 17% in the transgenic lines that accumulated elevated levels of REC1 ( Fig. 5A and SI Appendix, Fig. S29). In lines 40 and 66, the increase in chloroplast coverage appeared to result mostly from an increase in the number of chloroplasts per cell plan area without a significant change in the plan areas of individual chloroplasts (Fig.  5B). However, plots of total chloroplast plan area and chloroplast number versus cell plan area provide evidence for heterogeneity in the organellar basis for the increase in chloroplast coverage for line 40 (SI Appendix, Fig. S29), possibly resulting from the attenuation of chloroplast division in some of the larger cells from line 40. In line 44, the increase in chloroplast coverage appeared to result from an increase in the plan areas of individual chloroplasts and no change in the number of chloroplasts per cell (Fig. 5C and  SI Appendix, Fig. S29), possibly resulting from the misregulation of chloroplast division in this particular transgenic plant. Subcellular Distribution of REC1. To gain insight into the mechanism that underpins the biological functions of REC1, we determined the subcellular distribution of REC1 by fusing YFP to the carboxyl terminus of REC1, transiently expressing the REC1-YFP fusion protein in tobacco leaves and monitoring the subcellular distribution of REC1-YFP by confocal laser scanning microscopy. To test whether REC1-YFP was transiently expressed as a full-length protein, we analyzed leaf sections expressing REC1-YFP by immunoblotting with anti-GFP antibodies. The immunoblotting analysis indicates that REC1-YFP accumulates as a single band (SI Appendix, Fig. S30A) and that the mass of REC1-YFP is greater than the mass of native REC1 (SI Appendix, Figs. S3D and S30B).
Live-cell confocal microscopy analyses with settings that distinguish GFP, YFP, and chlorophylls (38) show REC1-YFP in the nucleus and the cytosol (Fig. 6A). The nucleus is recognizable by the appearance of the nucleolus (Fig. 6A, arrow) and by the localization of ssGFP-HDEL, an endoplasmic reticulum marker that defines the nuclear envelope (39) (Fig. 6B). We found that REC1-YFP is excluded from chloroplasts ( Fig. 6 C and D).
To test whether the functional state of the chloroplast might affect the subcellular distribution of REC1, we infiltrated one leaf of a tobacco plant with amitrole, an herbicide that causes chlorophyll deficiencies (40). We found that amitrole caused observable chlorophyll deficiencies in leaves that developed from shoot apical meristems after infiltrations (Fig. 6I). We suggest that amitrole moved from the sites of infiltrations to the shoot apical meristem, where it attenuated the chloroplast biogenesis that occurred during leaf development. In these chlorophyll-deficient leaves, chloroplasts were significantly smaller than in untreated leaves (Fig. 6 C and G). In these chlorophyll-deficient leaves, REC1-YFP localized to the cytosol and was excluded from the nucleus (Fig. 6 E-H).
We tested whether REC1 was enriched in nuclear and cytosolic fractions prepared from Arabidopsis seedlings. The enrichment of histone H3 and UDP glucose pyrophosphorylase in the nuclear and cytosolic fractions, respectively, indicates that nuclei and cytosols were enriched in these preparations (Fig. 7A). Using affinitypurified anti-REC1 antibodies (SI Appendix, Fig. S3), we found that REC1 was enriched in these nuclear and cytosolic fractions (Fig. 7A). Although these anti-REC1 antibodies recognized multiple bands in the nuclear and cytosolic fractions, they recognized only one band in the whole-seedling extracts prepared using denaturing conditions (Fig. 7A and SI Appendix, Fig. S3B). We conclude that the affinity-purified anti-REC1 antibodies recognize multiple bands in the nuclear and cytosolic fractions, because partial proteolysis of REC1 occurred during the purification of the nuclear and cytosolic fractions. Next, we tested whether herbicides affect the levels of REC1 in purified nuclei. We found that REC1 was present at similar levels in nuclei regardless of whether nuclei were purified from untreated seedlings or seedlings that lacked functional chloroplasts because chloroplast biogenesis was blocked with a norflurazon treatment (Fig. 7 A and B and SI Appendix,  Fig. S31A). When chloroplast biogenesis was blocked with an amitrole treatment, much longer exposures of immunoblots than used to detect REC1 in the nuclear fractions from norflurazontreated and untreated seedlings provided evidence that only trace quantities of REC1 accumulated in nuclei purified from amitroletreated seedlings (Fig. 7B and SI Appendix, Fig. S31A). Methyl viologen induces a rise in the levels of chloroplastic reactive oxygen species that causes chloroplast dysfunction and activates the plastid-to-nucleus signaling that induces the expression of nuclear genes, such as the nuclear genes that encode cytosolic ascorbate peroxidase (41). To test whether chloroplastic reactive oxygen species affect the nucleocytoplasmic partitioning of REC1, we transferred 5-d-old seedlings to a growth medium that contained 1 μM methyl viologen and allowed the seedlings to grow for an additional 3 d. Higher levels of cytosolic ascorbate peroxidase and lower levels of chlorophyll accumulated in the methyl viologentreated seedlings than in untreated seedlings (SI Appendix, Fig.  S31 B-D). We conclude that this methyl viologen treatment induced chloroplast dysfunction and the expression of cytosolic ascorbate peroxidase, probably by inducing a rise in the levels of chloroplastic reactive oxygen species. We found that REC1 accumulated to similar levels in the nuclei of methyl viologentreated and untreated seedlings (Fig. 7C). In addition to inhibiting chloroplast function, amitrole inhibits root elongation (40), which depends on cell division and cell elongation (42). To test whether inhibiting cell expansion might drive REC1 out of the nucleus, we tested whether REC1 accumulates in the nuclei of the Arabidopsis null mutant rhd3-7 (43). rhd3-7 and other RHD3 mutants are smaller than wild type because cell expansion is inhibited in these mutants (SI Appendix, Fig. S31A) (43,44). We found that REC1 accumulated to similar levels in the nuclei of rhd3-7 and wild type (Fig. 7D). We conclude that amitrole treatments specifically exclude REC1 from the nucleus.

Discussion
Our phenotypic characterizations provide evidence that the proteins encoded by the REC gene family link increases in mesophyll cell size to increases in chloroplast compartment size. Other than A  the proteins encoded by the REC gene family, there are no candidates for proteins that contribute to this mechanism (2,6,45). Cells of the shoot apical meristem contain fewer than 10 proplastids. After chloroplast biogenesis and during the mesophyll cell expansion that occurs during leaf development, there is a large increase in the number of chloroplasts in mesophyll cells (45). The mesophyll cells from rec1-3 rec2 rec3-1 friendly contained 22 ± 8 chloroplasts (SI Appendix, Figs. S19 and S22A). Thus, the mechanism that establishes the size of the chloroplast compartment is partially active in rec1-3 rec2 rec3-1 friendly. A few models are consistent with these data. One possibility is that the proteins encoded by the REC gene family directly contribute to the mechanism that establishes the size of the chloroplast compartment and that this mechanism retains partial activity in rec1-3 rec2 rec3-1 friendly. Another possibility is that the proteins encoded by the REC gene family regulate a mechanism that establishes the size of the chloroplast compartment. A third possibility is that the proteins encoded by the REC gene family contribute to a mechanism that helps to establish the size of the chloroplast compartment, and a distinct mechanism that does not require the REC genes independently contributes to the size of the chloroplast compartment.
Further study of the REC genes may lead to the ability to rationally manipulate the size of the chloroplast compartment and to the engineering of C 4 photosynthesis into C 3 plants, which is anticipated to increase yields from important crops, such as rice (9,11). The contribution of the REC genes to the distribution of chloroplasts in mesophyll cells may also contribute to the engineering of C 4 photosynthesis into C 3 plants, because the distribution of chloroplasts within mesophyll cells differs between C 3 and C 4 plants (11). The rec and friendly mutants appear deficient in a mechanism that is distinct from the mechanisms that are deficient in other chloroplast distribution mutants because in contrast to the normal photorelocation observed in rec1-3 rec2 rec3-1 friendly, the photorelocation of chloroplasts is abnormal in other chloroplast distribution mutants (27).
In contrast to the rec2, rec3, and friendly alleles that we tested, rec1-3 reduces the accumulation of chlorophyll. This observation raised the possibility that the chlorophyll deficiencies of the rec1 mutants might prevent chloroplasts from keeping pace with a mechanism that establishes the size of the chloroplast compartment. Indeed, in crumpled leaf (crl) and clumped chloroplasts 1 (clmp1), chloroplasts do not keep pace with this mechanism because chloroplasts are not equally distributed between daughter cells during cell division (46,47). Our finding that inducing an increase in the levels of chlorophyll does not affect chloroplast coverage in rec1-3 rec2 rec3-1 friendly and that chloroplast coverage is not significantly different from wild type in the chlorophyll-deficient mutants cch, gun4-1, cop1-4, det1-1, and phyA phyB indicates that chlorophyll deficiencies do not prevent chloroplasts from keeping pace with the mechanism that establishes the size of the chloroplast compartment. Chlorophyll deficiency is a common phenotype in Arabidopsis. Thus, disrupting any number of chloroplast functions may indirectly affect the thylakoid membranes. Indeed, disrupting mechanisms that drive chloroplast division and mechanisms that potentially regulate chloroplast division appears to indirectly affect the normal development of the thylakoid membranes (34,48,49) and the accumulation of chlorophyll (47,50,51).
The screening of 3,500 ethylmethanesulfonate (EMS) mutants for plants with aberrant morphologies in the chloroplast compartment, such as abnormal chloroplast coverage (3), was not sufficient to saturate this screen (52). Nonetheless, screening ∼2,000 EMS-mutagenized Arabidopsis plants is usually sufficient for isolating specific mutants (53). Indeed, the screen of Pyke and Leech (3) yielded more than 10 different genes that, in most instances, encode proteins that directly contribute to chloroplast division (2). An additional screen of 10,000 EMS mutants for plants with enlarged chloroplasts yielded only one additional gene that directly contributes to chloroplast division (54). Additionally, a screen of 5,200 Arabidopsis mutants that were homozygous for T-DNA insertions in nuclear genes that mostly encode chloroplast proteins for mutants with aberrant morphologies in the chloroplast compartment, such as abnormal chloroplast coverage, yielded one chloroplast division mutant and no chloroplast coverage mutants (46). These data and our finding that chloroplast coverage is not different from wild type in cch, gun4-1, gun1-101, cop1-4, det1-1, and phyA phyB indicate that a reduction in the size of the chloroplast compartment is not a common phenotype. We suggest that the REC genes may directly contribute to a mechanism that links increases in the size of the chloroplast compartment to increases in the size of mesophyll cells because reduced chloroplast coverage is not a common phenotype and because overexpressing REC1 induces significant increases in chloroplast coverage. Our finding that the REC1 protein resides in the cytosol and nucleus further supports the idea that REC1 helps to establish the size of the chloroplast compartment because this mechanism is expected to reside in an extraplastidic compartment.
A less than twofold increase in chloroplast coverage was observed in the fully expanded leaves and mature green fruits of the high pigment 1 (hp-1) mutant of tomato (55). HP-1 encodes UVdamaged DNA-binding protein 1 (DDB1) (56). DDB1 interacts with DET1, a master repressor of light-regulated development that contributes to ubiquitin-proteasome-mediated protein degradation (35  that defects in light-regulated development do not necessarily affect the size of the chloroplast compartment in mesophyll cells. The biochemical functions of REC1, REC2, and REC3 are not known. FRIENDLY appears to perform a variety of functions (13,20,21). The indistinguishable photorelocation of chloroplasts that we observed in wild type and rec1-3 rec2 rec3-1 friendly and the localization of REC1 to both the cytosol and the nucleus conflict with the idea that REC proteins directly contribute to the tracking of chloroplasts along the cytoskeleton. Similarly, mitochondrial clustering in friendly does not result from a cytoskeletal or motor defect (13). The enhanced chlorophyll fluorescence, chloroplasts aligning along the anticlinal walls, abnormal chloroplast ultrastructure, and chlorophyll deficiencies in the rec and friendly mutants indicate that these mutant alleles induce chloroplast dysfunction. REC1, REC2, REC3, and FRIENDLY may contribute to processes in the cytosol and nucleus that promote the function of chloroplasts and mitochondria. Consistent with this interpretation, FRIENDLY was localized to the cytosol, and mutations in FRIENDLY and the Drosophila ortholog of FRIENDLY induce metabolic dysfunction (13,57).
The finding that REC1 is excluded from the nucleus when chloroplast biogenesis is blocked or attenuated with an amitrole treatment is consistent with a plastid signaling mechanism that is activated by amitrole treatments regulating the subcellular distribution of REC1. This inference follows because proteins larger than ∼40 kDa are actively transported across the nuclear envelope (58) and because the nucleocytoplasmic partitioning of proteins contributes to a number of processes, including diverse signaling mechanisms (35,58). The finding that, in contrast to amitrole treatments, norflurazon and methyl viologen treatments do not exclude REC1 from the nucleus is consistent with a plastid signaling mechanism that is activated by amitrole treatments but not activated by norflurazon and methyl viologen treatments regulating the subcellular distribution of REC1. Consistent with this interpretation, plastid-to-nucleus signaling mechanisms appear numerous (8).
Another possibility is that an extraplastidic mechanism that is inhibited by amitrole treatments might affect the subcellular distribution of REC1 because in contrast to norflurazon and methyl viologen (8,59), amitrole is not a specific inhibitor of chloroplast function. Although amitrole inhibits plastidic processes, such as carotenoid biosynthesis and histidine biosynthesis, amitrole is more effective at inhibiting root elongation, which depends on cell division and cell elongation (40). Our findings that the REC genes link increases in cell size to increases in chloroplast compartment size and that amitrole specifically affects the nucleocytoplasmic partitioning of REC1 are consistent with the inhibition of cell expansion regulating the subcellular distribution of REC1. If the inhibition of cell expansion excludes REC1 from the nucleus, rhd3-7 may not affect the nucleocytoplasmic partitioning of REC1 because rhd3-7 does not sufficiently inhibit cell expansion.
Amitrole treatments may inhibit a process in the nucleus by excluding REC1 from the nucleus. Alternatively, amitrole-treated plants may attempt to stimulate a cytosolic process by driving REC1 into the cytosol. Another possibility is that REC1 performs the same function or different functions in the nucleus and the cytosol.
In summary, we provide evidence that REC1, REC2, REC3, and FRIENDLY promote the proper morphology, function, and size of the chloroplast compartment. These functions may depend on the trafficking of REC1 between the nucleus and the cytosol. Signals activated by the inhibition of cell expansion or chloroplast dysfunction may regulate this mechanism. Understanding the biochemical function of REC1 and the nucleocytoplasmic partitioning of REC1 may lead to the rational manipulation of the size of the chloroplast compartment and higher yields from important crops.

Materials and Methods
Detailed information on the materials and methods used in this study is provided in SI Appendix.
Plant Materials and Growth Conditions. All mutants used in this study were derived from A. thaliana ecotype Columbia-0 (Col-0). Plants were grown on soil or on Linsmaier and Skoog (LS) growth medium in either broadspectrum white light or far-red light in controlled-environment chambers. The far-red block of greening experiments were adapted from Barnes et al. (15).
Analysis of Chlorophylls. Chlorophylls were extracted using N,N′-dimethylformamide and quantified by spectrophotometry.
Analysis of Chloroplasts by Microscopy. Chloroplasts were imaged by confocal laser scanning microscopy. Chlorophyll fluorescence, D F , and Λ were calculated from maximum-intensity projection images built from Z-stack images. The chloroplast number per cell plan area, chloroplast plan area, and chloroplast coverage were quantified as described previously (3).
Construction, Expression, and Imaging of the REC1-YFP Fusion Gene. The ORF that encodes YFP was fused in-frame and downstream of an ORF that encodes the full-length REC1. Transient expression of REC1-YFP in Nicotiana tabacum was performed by infiltrating amitrole-treated or untreated leaves with an Agrobacterium tumefaciens strain harboring the REC1-YFP fusion gene. The imaging of YFP fluorescence was performed using confocal laser scanning confocal microscopes.
Analysis of Whole-Seedling Extracts, Nuclear Fractions, and Cytosolic Fractions. Seedlings were grown on LS medium containing no sucrose or on the same medium containing 1% sucrose and an herbicide. Whole-seedling extracts were prepared by boiling frozen and powdered seedlings in SDS/PAGE sample buffer and then clarifying these extracts by centrifugation at 16,000 × g for 10 min. Nuclei were purified from seedlings using Percoll step gradients. Cytosolic fractions were the supernatants that were obtained by lysing purified protoplasts and then clarifying these lysates by centrifugation at 21,000 × g for 15 min. Equal amounts of protein were analyzed by immunoblotting.