Heterosis of Arabidopsis hybrids between C24 and Col is associated with increased photosynthesis capacity

Abstract

Arabidopsis thaliana shows hybrid vigor (heterosis) in progeny of crosses between Columbia-0 and C24 accessions. Hybrid vigor was evident as early as mature seeds and in seedlings 3 d after sowing (DAS). At 3 DAS, genes encoding chloroplast-located proteins were significantly overrepresented (187) among the 724 genes that have greater than midparent values of expression in the hybrid. Many of these genes are involved in chlorophyll biosynthesis and photosynthesis. The rate of photosynthesis was constant per unit leaf area in parents and hybrids. Larger cell sizes in the hybrids were associated with more chloroplasts per cell, more total chlorophyll, and more photosynthesis. The increased transcription of the chloroplast-targeted genes was restricted to the 3–7 DAS period. At 10 DAS, only 118 genes had expression levels different from the expected midparent value in the hybrid, and only 12 of these genes were differentially expressed at 3 DAS. The early increase in activity of genes involved in photosynthesis and the associated phenomena of increases in cell size and number through development, leading to larger leaf areas of all leaves in the hybrid, suggest a central role for increased photosynthesis in the production of the heterotic biomass. In support of this correlation, we found that an inhibitor of photosynthesis eliminated heterosis and that higher light intensities enhanced both photosynthesis and heterosis. In hybrids with low-level heterosis (Landsberg erecta x Columbia-0), chloroplast-targeted genes were not up-regulated and leaf areas were only marginally increased.

In many species, hybrid progeny of selected parental lines have enhanced performance relative to their parents. This phenomenon is known as hybrid vigor or heterosis (1–3), and is important in production agriculture of major crops such as maize, rice, sorghum, sunflower, and canola. Genetic models to explain the increased yield of hybrids consider the interaction of alleles at many loci generating altered expression levels and patterns of gene expression (2, 4). Recently, Ni et al. (5) suggested that changes in the expression of circadian regulator genes may be important in heterosis by leading to downstream changes in gene expression.

Whole-genome transcriptomes of parents and hybrids have been analyzed in a number of species (reviewed in refs. 3 and 6). In maize, a number of modes of gene action—additive, dominance, and overdominance—occur at loci in the hybrids generated from crosses between B73 and Mo17 inbred lines (7). The superhybrid rice LYP9 shows altered gene expression relative to parental cultivars in several different tissue types and developmental stages (8). Genes with altered expression are overrepresented in processes such as energy metabolism and transport [gene ontology (GO) analysis (8)], but no consistent pattern of altered gene expression has emerged, with many genes having expression levels up- or down-regulated in different tissues. In addition to whole-genome transcriptome analyses, epigenetic states such as DNA methylation, small RNA production, and histone modification have been found to be altered in hybrids (9, 10).

Hybrids between the Arabidopsis accessions Columbia-0 (Col) and C24 display increased vegetative biomass over that of the parents at 10 DAS (1), indicating that the mechanisms underlying heterosis begin early in the life cycle of the hybrids. We have examined early developmental stages in the C24 x Col hybrid system. Our results at 3–7 d after sowing (DAS) show up-regulation of genes encoding chloroplast-targeted proteins in the cotyledons and the earliest leaves of the hybrid. The transcription increases are associated with increases in cell size in the palisade mesophyll cells and with increased numbers of chloroplasts per cell. Later leaves also show increased cell size, but the enhanced transcription of the chloroplast-targeted genes is not maintained. These later leaves of the hybrids show increased numbers of cells as well as increased cell size, and both of these factors contribute to the larger leaves of the hybrids. The larger leaves of the hybrid plant with their additional photosynthetic capacity may play a central role in increasing biomass throughout development.

Results

Previous reports on heterosis in Arabidopsis and other species (11) suggested that late flowering could contribute to increased vegetative biomass in hybrids. In crosses between the Arabidopsis accessions Col and C24, flowering times differ significantly between the parental lines (Col, 18.7 ± 0.2 d; C24, 28.1 ± 0.5 d) and the hybrids (C24 x Col, 44.5 ± 0.7 d; Col x C24, 43.1 ± 0.7 d). The flowering times are controlled by the combination of the alleles at the FRIGIDA (FRI) and FLOWERING LOCUS C (FLC) loci (12); FLC encodes a repressor of flowering, with its activity regulated by FRI (13). Late flowering occurs in plants with high FLC activity. In the hybrids, C24 contributes an active FRI allele and Col contributes a strong FLC allele (14), resulting in high FLC activity and consequent late flowering.

To avoid any growth effects of late flowering, we modified flowering time by vernalization or, in other experiments, by using nonfunctional alleles of FLC in the cross to produce the hybrid. In both strategies, flowering time in the hybrids was similar to the parental flowering times and later (>21 DAS) than the period of our analyses (Dataset S1, Table S1). Rosette diameters of the earlier- and late-flowering hybrids at 10 and 14 DAS were comparable and ∼20% greater than the corresponding rosette diameters of the parents (Tables 1–3 and Fig. S1), indicating that flowering time does not influence heterosis in the stages of development before the transition to flowering. Our analyses address the preflowering vegetative developmental stage.

Table 1.

Rosette diameter of F₁ hybrids and parents: Nonvernalized growth

	Col (fri/fri, FLC/FLC)	C24 (FRI/FRI, FLC*/FLC*)	F1-ColxC24 (fri/FRI, FLC/FLC*)	F1-C24xCol (FRI/fri, FLC*/FLC)
10 DAS	1.00 ± 0.002 (n = 125)	1.05 ± 0.002 (n = 115)	1.26 ± 0.002 (n = 124)	1.24 ± 0.002 (n = 104)
14 DAS	1.00 ± 0.001 (n = 111)	0.99 ± 0.001 (n = 100)	1.15 ± 0.001 (n = 115)	1.19 ± 0.002 (n = 102)

n, number of plants. An asterisk indicates the weak FLC allele from C24.

Table 3.

Rosette diameter of F₁ hybrids and parents: flc mutants

	flc-20xC24 (FRI/FRI, flc/FLC*)	F1-flc-20xCol (FRI/fri, flc/FLC)	F1-flc-20xflc-b (FRI/fri, flc/flc)	F1-flc-20xflc-3 (FRI/fri, flc/flc)
10 DAS	1.00 ± 0.001 (n = 118)	1.20 ± 0.016 (n = 103)	1.21 ± 0.018 (n = 59)	1.21 ± 0.016 (n = 65)
14 DAS	1.00 ± 0.001 (n = 118)	1.22 ± 0.013 (n = 102)	1.20 ± 0.012 (n = 65)	1.21 ± 0.014 (n = 72)

flc-b (SALK-003346) and flc-3 mutants are in Col. flc-20 mutant is in C24. n, number of plants.

*Weak FLC allele from C24.

Table 2.

Rosette diameter of F₁ hybrids and parents: 6-wk vernalization

	Col-6wkv (fri/fri, FLC/FLC)	C24-6wkv (FRI/FRI, FLC*/FLC*)	F1-ColxC24-6wkv (fri/FRI, FLC/FLC*)	F1-C24xCol-6wkv (FRI/fri, FLC*/FLC)
10 DAS	1.00 ± 0.020 (n = 39)	1.01 ± 0.028 (n = 38)	1.20 ± 0.028 (n = 36)	1.24 ± 0.026 (n = 38)
14 DAS	1.00 ± 0.022 (n = 39)	1.03 ± 0.031 (n = 38)	1.17 ± 0.027 (n = 36)	1.21 ± 0.027 (n = 38)

6wkv, 6-wk vernalization. Underlining indicates an FLC allele not expressed because of absence of FRI or vernalization. n, number of plants.

*Weak FLC allele from C24.

Heterosis in Early Development.

In the period from germination through to 10-DAS seedlings, there are a series of different heterotic responses. The hybrids with C24 as a female parent have larger seeds and cotyledons than the parents or the reciprocal hybrid (Dataset S1, Table S2); Meyer et al. (1) reported a similar result. At 3 DAS, both reciprocal hybrids are larger and developmentally more advanced than the parental seedlings (Fig. S2) as a consequence of earlier germination. We chose 4-d parental seedlings and 3-d hybrid seedlings to analyze gene expression because they were at the same developmental stage, with the cotyledons just emerged (Fig. S2). In the transcriptome analysis of the C24 x Col hybrid, we used a whole-genome tiling array (Affymetrix) that contained intergenic regions, transposable elements, genic regions, and their flanking sequences. Most of the transcripts that showed expression in the hybrids different from the expected midparent values (MPVs) were from protein-coding sequences. In the C24 x Col hybrid, there were 724 up-regulated genes and 329 down-regulated genes with equal to or greater than 1.5-fold difference in expression level between the hybrid and the MPV (Dataset S1, Table S3; see SI Materials and Methods for detailed analysis methods). Of the 724 up-regulated genes, 698 had expression levels in the hybrid greater than the higher parental value. Only 26 of the up-regulated genes had expression levels differing between the parents, whereas 113 of the 329 down-regulated genes had different parental levels of expression (Fig. 1A).

Fig. 1.

Analysis of differentially expressed genes. Venn diagrams summarizing the numbers of differentially expressed genes at 3/4 DAS (A) or 10 DAS (B). (C) Venn diagram summarizing the differentially expressed genes in the F₁ relative to the MPV in common between 3/4 and 10 DAS. Blue and red letters indicate the number of down- and up-regulated genes, respectively. The black circles show the number of genes differentially expressed between Col and C24 at 3/4 or 10 DAS. (C) The blue and green circles show the number of genes differentially expressed between the F₁ and MPV at 3/4 and 10 DAS, respectively. Categories overrepresented in up-regulated genes at 3/4 DAS in the chloroplast cellular component (D) and biological process (E). GO annotations are listed in Dataset S1, Tables S5 and S6.

Using quantitative real-time (q)PCR, we verified the expression values obtained from the tiling array for 50 of the differentially expressed genes (Dataset S1, Table S4; 46 of the 50 genes had P values less than 0.05). These genes, classified by gene ontology software (SI Materials and Methods), were represented in a range of cellular locations and metabolic functions. The up-regulated genes that located products to the chloroplast were significantly overrepresented; 187 (26.5%) of the products of the up-regulated genes were targeted to the chloroplast compartment compared with the expected number of 55 (7.6%) (Materials and Methods, Fig. 1D, and Dataset S1, Table S5). The gene products were located in the thylakoid membrane, chloroplast stroma, chloroplast envelope, and photosynthetic membranes (Fig. 1D and Dataset S1, Table S6). The gene products were involved in a number of biological processes in the chloroplast compartment, especially those related to photosynthesis and to tetrapyrrole and chlorophyll biosynthesis (Fig. 1E and Dataset S1, Table S6). None of the down-regulated genes were in the chloroplast-located category, and were involved in other metabolic functions including responses to biotic and abiotic challenges (Fig. 1D, Fig. S3, and Dataset S1, Tables S5 and S6).

Of the 16 genes encoding enzymes involved in tetrapyrrole and chlorophyll biosynthetic pathways, 14 were in the substantially up-regulated group (Fig. 2) in the hybrids; the expression of 9 of these genes was verified by qPCR (Dataset S1, Table S4). The reciprocal hybrid with Col as the female parent had overexpression of these same genes (Fig. S4). Although the 3-d hybrid seedlings were developmentally equivalent to 4-d parental seedlings, we measured the expression levels in the parents at 3 and 5 DAS to confirm that the up-regulation of the chlorophyll biosynthesis genes was not due to differences in the age of the hybrid and parents. We found that the MPVs in both samples were lower than the expression levels in the hybrid (Fig. S5).

Fig. 2.

Up-regulation of the genes involved in tetrapyrrole and chlorophyllide a biosynthetic pathways in the F₁ hybrid at 3/4 DAS. The genes up-regulated in the F₁ hybrid relative to the parental lines at 3/4 DAS are shown in red. Expression was validated by qPCR (Dataset S1, Table S4).

Transcript Analysis of 10-D Plants.

At 10 DAS, the developmental stage of the plants as measured by rosette leaf number was the same in the parents and reciprocal hybrids (rosette leaves, 6.8 ± 0.16 for C24 x Col hybrid; 6.6 ± 0.18 for C24; 6.6 ± 0.16 for Col). RNA extracted from the aerial tissues of these plants was hybridized to the cDNA ATH1 array (SI Materials and Methods). In the hybrids, of 118 genes with 1.5-fold or greater difference from the MPV, 95 were up-regulated and 23 were down-regulated (Fig. 1B). Among the up-regulated genes, 63 involved loci that were differentially expressed between the two parents. All 23 of the down-regulated genes had expression levels differing between the parents (Fig. 1B and Dataset S1, Table S7). Expression levels of 15 of the genes with expression levels different from the MPV level and 7 genes expressed at MPV level were examined by qPCR, and all were consistent with the microarray data (Dataset S1, Table S8).

Only 12 genes differed from the MPV at both 3 and 10 DAS (Dataset S1, Table S9). Eleven of the 12 were up-regulated and 1 was down-regulated at both times (Fig. 1C). None of these genes were targeted to chloroplasts (GO analysis). In the 10-DAS suite of genes that were not expressed at MPV levels, there was no overrepresentation of chloroplast-related genes. The other genes that were not expressed at MPV level at 10 DAS were related to the environmental or biotic stress categories of metabolic pathways (GO; Fig. S3 and Dataset S1, Tables S5 and S6). In the period between 3 and 10 DAS, of 23 chloroplast-targeted genes that were up-regulated at 3 DAS, 18 were up-regulated at 6 and 7 DAS in the first and second leaves but not in the cotyledons, which were beginning to senesce, or in whole plants (Fig. S6 and Dataset S1, Table S10). At 9 and 10 DAS, in leaves 3 and 4, these genes did not show any expression level higher than the MPV expression level (Fig. 3 and Dataset S1, Table S10). By 10 DAS the cotyledons had senesced, and leaves 1 and 2 no longer showed any up-regulation of the monitored genes.

Fig. 3.

Up-regulation of genes involved in chlorophyll biosynthesis or photosynthesis in the cotyledon and early leaves of the F₁ hybrid at 6/7 and 9/10 DAS. Expression levels of genes relative to the IPP2 gene were calculated using a comparative quantification analysis. Data presented are the average and SE from three or four independent experiments with two biological and three experimental replications. RNAs for qPCR were isolated from 15 cotyledons at 6 or 7 DAS, 15 first and second rosette leaves (1-2 RL) at 6, 7, 9, or 10 DAS, and 15 third and fourth rosette leaves (3-4 RL) at 9 or 10 DAS. The bar graph was made from Dataset S1, Table S10. *P < 0.05; **P < 0.01.

Cell Size and Number Are both Involved in Forming the Larger Leaves of C24 x Col Hybrids.

At the mature seed stage, seed size and dry weight of the hybrid seed with C24 as the female parent were greater than in the seed of the reciprocal hybrid or of the parents (Dataset S1, Table S2). At this mature seed size, the hybrid with C24 as the female parent had cotyledons 35% greater in area than the MPV, and the reciprocal hybrid with Col as the female parent had a cotyledon area only 13% greater than the MPV (Dataset S1, Table S2). The increased cotyledon size in the hybrids could be due to an increased number of cells, an increased size of the cells, or a combination of these two possibilities. In the adaxial layer of palisade mesophyll cells, the C24 parent had more cells per unit area than the reciprocal hybrids or Col (Dataset S1, Table S2), indicating that the larger size of the cotyledon in the hybrid must be due to an increased number of cells and not to an increase in cell size (Dataset S1, Table S2).

At later stages of development (4 and 7 DAS), the cotyledon area of the hybrid was greater than that of the parents (Fig. 4A), but the number of cells per unit area in the hybrid cotyledon had decreased, indicating that cell size had increased (Fig. 4B). The basis of increased cotyledon area changes from cell number in the mature seed stage to a combination of cell size and cell number in the early seedling stage (Fig. 4B). In seedlings at 10 DAS, leaves 1 and 2 show increased cell size to be the major basis of increased leaf area (Fig. 4B). At 14 and 21 DAS, the third and fourth leaves show increased cell size, but increased cell number also contributes to the increased area. Similarly, at 21 DAS, leaves 5 and 6 show the larger leaf size in the hybrid is due to an increase in both size and number of cells (Fig. 4B).

Fig. 4.

Leaf area and cell size in F₁ hybrids. (A) Leaf area relative to Col. (B) Number of cells in a given unit area. The number has been divided by the number of cells per unit area in Col. Data presented are the average and SE from five independent experiments with more than 10 cotyledons at 0 (mature seed), 4, and 7 DAS and more than 10 first and second leaves at 10 and 14 DAS, third and fourth leaves at 14 and 21 DAS, and fifth and sixth leaves at 21 DAS.

Photosynthetic Capacity Is Increased in the Leaves in the C24 x Col Hybrid.

At 3 DAS, transcription levels of many genes with chloroplast-located functions are up-regulated in the hybrid seedling. The palisade mesophyll cells of the cotyledons are larger than the corresponding cells of the cotyledons in the parents. The up-regulated transcription of the chloroplast-targeted genes and the increased size of palisade mesophyll cells also occur in leaves 1 and 2 of the hybrid seedling. In leaves 3 and 4 there is increased cell size but no increased transcription activity of the chloroplast-targeting genes. Leaves 5 and 6 of the hybrid are larger than the corresponding leaves of the parents, but there is no transcriptional up-regulation of the chloroplast-targeting genes (Figs. 3 and 4B). Pyke and Leech (15) showed that in Arabidopsis the number of chloroplasts in palisade mesophyll cells is positively correlated with the size of the cells. We confirmed this correlation in the C24 x Col hybrid in the palisade cells of 7-DAS cotyledons and in the first and second leaves at 10 and 14 DAS (Fig. S7). Because the total leaf area is larger, the hybrid leaf has a greater number of chloroplasts than the parent leaves.

The rate of photosynthesis per unit area is the same in the hybrids as in the parents (Dataset S1, Tables S11 A and B) in both saturating light intensity (1,000 μmol photons⋅m⁻²⋅s⁻¹) and in a light intensity (200 μmol photons⋅m⁻²⋅s⁻¹) similar to the growth conditions of our experiments. We examined vertical sections of cotyledons at 7 d and leaves at 7, 10, and 14 DAS and found the cell-layer construction is the same in the parents and the hybrids in all of these stages. In all leaves, there was one palisade mesophyll layer and approximately three spongy mesophyll layers. At 7 DAS, Col had a relative vertical thickness of 1.00 ± 0.03; C24 had a relative value of 0.80 ± 0.01; and C24 x Col had 0.87 ± 0.02. At 14 DAS, Col was 1.00 ± 0.03; C24 was 0.95 ± 0.04; and C24 x Col was 1.07 ± 0.01. The cotyledons had a different cellular structure, with large cells that were not arranged as uniformly in a palisade layer as in the leaves, but there was no difference in thickness between the hybrids and parents, consistent with the unchanged photosynthesis per unit area.

The chlorophyll content per gram fresh weight is also similar in the hybrid and parents (Dataset S1, Table S11C), but the total chlorophyll content of the hybrid is greater than that of the parents because of the increased size and number of cells resulting in an increased leaf area (Dataset S1, Table S11B). These findings indicate a greater photosynthate production in the hybrid relative to the parents, leading to increased biomass (Dataset S1, Table S11D).

We tested the implied direct relationship between photosynthesis and biomass by treating seedlings at 3 DAS with norflurazon, an inhibitor of phytoene desaturase (Fig. 5 A and B) (16). The treated seedlings show an inhibition of chlorophyll production, and growth rate was reduced (Fig. 5 A and B). The area of the bleached cotyledons in the hybrid at 8 DAS is larger than the cotyledon areas in the parents (Fig. 5A), probably as a result of earlier heterosis before inhibitor application. The hybrid did not demonstrate any heterosis at 21 d (Fig. 5 B and C).

Fig. 5.

Loss of heterosis following treatment with norflurazon. (A) Cotyledons at 8 DAS with 0.5 μM norflurazon. Seeds were germinated on MS medium (SI Materials and Methods) for 3 d and plants were transferred to MS medium with 0.5 or 1.0 μM norflurazon. (B) Phenotypes at 21 DAS. (C) Leaf area of first and second leaves at 21 DAS following treatment with norflurazon. Data presented are the average and SE from three independent experiments with more than 10 plants.

Columbia-0 x Landsberg erecta Hybrids Have only a Low-Level Heterosis.

F₁ (first generation) hybrids between the Arabidopsis accessions Col and Landsberg erecta (Ler) have a low level of hybrid vigor; rosette diameters in the hybrids at 14 DAS are 9% greater than the rosette diameters of the parents. At 7 DAS the cotyledon area is ∼25% greater than the parents, contrasting with the 80% increase in area in the cotyledons of the C24 x Col hybrids. At 14 DAS, leaf area of the Col x Ler hybrid was the same as the Col parent (Dataset S1, Table S12). The genes of the chlorophyll biosynthesis pathway do not show a deviation from MPV at 4 DAS (Fig. S8), contrasting with the C24 x Col hybrid, and the photosynthetic rate per unit area was the same as the parents. There was no overall increase in photosynthesis (Dataset S1, Table S12). These data from the minimally heterotic hybrid Col x Ler contrast with the corresponding data of the highly heterotic hybrids formed by C24 x Col crosses.

Discussion

The two Arabidopsis ecotypes C24 and Col have very similar genomes, differing only in SNPs and small deletions and insertions (17). Despite the near identity of the genomes, their F₁ hybrid shows marked heterosis, generating more than 200% greater biomass at maturity. Hybrid vigor of the mature plant biomass was probably augmented by growth resulting from the longer period of vegetative development due to the extreme late flowering of the hybrid. For this reason, we concentrated our analyses of the hybrid to the early period of development before any transition to the reproductive developmental phase. We generated an early-flowering hybrid by vernalization treatment, which brought flowering forward from 45 DAS to 28 DAS or, alternatively, by introducing early-flowering alleles of the FLC gene to both parents, which again moved flowering forward to 24 DAS. We showed that up to 21 DAS there was no detectable difference in plants destined to be late- or early-flowering.

We identified a number of phases of hybrid vigor in the first 21-DAS period of growth. In the mature seed, the hybrid with C24 as maternal parent showed hybrid vigor in both embryo and seed size. The increased cotyledon area was due to cell number and not cell size, but increase in cell size became a factor at 3 DAS. At 3 DAS more than 1,000 genes, some 4% of the genes in the genome, were either up- or down-regulated relative to the MPV, mostly up-regulated, and among the up-regulated genes there was an overrepresentation of genes concerned with metabolic activities in the chloroplast. Many of these (187) up-regulated genes are involved in chlorophyll biosynthesis, tetrapyrrole biosynthesis, and the process of photosynthesis; only 7%, or ∼50 genes, are expected among the 724 up-regulated genes. Up-regulation of these chloroplast-targeted genes extended only to 7 DAS and applied only to the cotyledon and to the first two leaves of the developing plant. After 7 DAS, leaves 3, 4, 5, and 6 did not have this pattern of up-regulated chloroplast-targeted genes. At 10 DAS, when leaves 6 and 7 were formed, only 300 genes showed expression at a level other than the MPV and these, predominantly, were up-regulated. Only 12 of these genes were among genes with altered expression level in the 3–7 DAS period. These transcription analyses provided an observation of different gene expression responses at different times of development.

During the 3–7 DAS period, the up-regulated suite of genes, other than the chloroplast-targeted genes, included genes involved in stress responses both to abiotic and biotic stimuli and genes involved in basic metabolic processes. In the up-regulated genes at 10 DAS, there were also genes that were annotated as being involved in responses to abiotic and biotic stimuli, but these were different from those in the 3–7 DAS suite of genes. In the 10-DAS transcriptome, many of the genes expressed at other than MPV showed levels of expression differing between the parents. This finding parallels the finding in C24 x Ler hybrids that 24-nt siRNAs are not produced at MPV levels at loci that differ between the parents in terms of siRNA frequencies (10). In the C24 x Col hybrid, the up-regulated genes at 3 DAS did not show any difference in expression between parents.

Although we found stability in leaf anatomy of the hybrid in relation to the parents, we did find transcription-level responses in the hybrid, and leaf and cotyledon areas were greater than those in the parent plants. Cell numbers in the cotyledons and the early leaves were greater than cell numbers in the comparable parental structures. In addition, and particularly in the cotyledons and earliest leaves, cell sizes of the palisade mesophyll cells were larger in the hybrid than in the parents. The hybrid plants had larger leaves than either parent and more chloroplasts in the leaves of the hybrid relative to the corresponding leaves of the parents (15). The larger mesophyll cells of the hybrid had more chloroplasts than the smaller mesophyll cells of the parents, and the increased number of cells in the leaves also resulted in an increased number of chloroplasts in the hybrid leaves. In turn, this contributed to a greater chlorophyll content of the hybrid leaves and, although we determined the rate of photosynthesis per unit area was not altered in the hybrids, the overall quantity of photosynthate production was greater in the larger hybrid leaves.

We do not have any direct evidence as to the relationship between increased photosynthetic activity and increased cell size and cell number; however, earlier reports in Arabidopsis show there is a correlation of increased chlorophyll a/b (CAB) binding polypeptide gene (Lhcb) activity with increased cell size (18). It is possible that the increased activity of the CAB proteins in the photosystem II complex drives the increase in cell size in the green cells of the leaves.

We explored the correlation of photosynthesis and biomass production in three other ways. As expected, increased light intensity increased photosynthetic activity and increased the level of heterosis (1). We used norflurazon, an inhibitor of photosynthesis, to inhibit phytoene desaturase and hence chlorophyll synthesis (16). This inhibitor prevented chlorophyll production in the developing seedlings and slowed growth markedly. The heterotic response of the hybrid was completely removed. Both of these treatments support a positive correlation between photosynthate production and the magnitude of heterotic biomass production. The data from the Col x Ler hybrid with low-level heterosis add further support for the importance of photosynthesis.

In the hybrids it is possible that the exposure of the developing cotyledons and early leaves to light in the germinating seed stimulates the up-regulation of the chloroplast-targeted and photosynthesis-related genes, resulting in increased photosynthetic activity. Increased photosynthesis can result in increased cell size, the consequence is increased chlorophyll content and increased photosynthate. Although our analysis was restricted to vegetative hybrid vigor, the correlation of increased photosynthetic activity with increased biomass production in the hybrid may also drive the heterosis we have noted in the reproductive structures of the plant.

In other species, the heterosis of grain yield sought in agricultural crops is accompanied by vegetative heterosis, and may be dependent on increased photosynthetic activity of the larger leaves of the hybrid. The positive correlation between photosynthetic activity and heterosis is not an unexpected association, but to our knowledge it has not been previously suggested as being directly involved in hybrid vigor. Increased photosynthesis may well be a general factor in hybrid vigor in plants.

Materials and Methods

Experimental design, analysis, and plant material are described in detail in SI Materials and Methods. RNA was extracted from seedlings at various stages of development and hybridized to Affymetrix arrays. Photosynthetic carbon dioxide fixation rate was examined by the light-dependent consumption of CO₂. Chloroplasts in leaf cells were counted using Nomarski optics, and cell area was measured using ImageJ software (National Institutes of Health).