INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In contrast to C₃ photosynthesis, in which the whole process of primary CO₂ assimilation takes place autonomously within a single cell, C₄ CO₂ assimilation requires metabolic cooperation between bundle sheath cells (BSC) and mesophyll cells (MC) (Edwards and Walker, 1983; Anderson and Deardall, 1991; Suzuki et al., 1999). CO₂ is first fixed by phosphoenolpyruvate carboxylase (PEPCase; EC 4.1.1.31) in MC. The C₄ acid formed diffuses to BSC, where it is decarboxylated and the CO₂ released is refixed by Rubisco (EC 4.1.1.39). The substrate for PEPCase in the initial CO₂ fixation, PEP, is regenerated by pyruvate phosphate dikinase (PPDK, EC 2.7.9.1) in MC (Hatch 1987, 1997; Furbank and Taylor, 1995). The foundation of this metabolic cooperation is the cell-type-specific compartmentation of photosynthetic enzymes and expression of the corresponding enzyme-coding genes.

Cell type specificity of the proteins and their mRNAs has been demonstrated in several monocot and dicot C₄ species (Bauwe, 1984; Martineau and Taylor, 1985; Sheen and Bogorad, 1985, 1986; Langdale et al., 1988a, 1988b; Hudson et al., 1992; Wang et al., 1992; Dengler et al., 1995; Ramsperger et al., 1996; Shu, 1996; Drincovich et al., 1998). The genetic mechanisms that control the cell-type- and tissue-specific expression of genes encoding C₄ enzymes have been studied with both molecular and transgenic approaches in C₄ species (Berry et al., 1985, 1986; Martineau et al., 1989; Sheen, 1991; Matsuoka et al., 1993; Chitty et al., 1994; Bilang and Bogorad, 1996; Furbank et al., 1997; Marshall et al., 1997; Stockhaus et al., 1997; Taylor et al., 1997; Westhoff et al., 1997; Mann, 1999).

Light is one of the most important environmental signals regulating leaf development in plants, including leaf cellular differentiation and photosynthetic gene expression (Tobin and Silverthorne, 1985; Nelson and Langdale, 1992; Fankhauser and Chory, 1997). Light perception and the various light signal transduction pathways that regulate leaf development are currently under intensive study in C₃ plants (Fankhauser and Chory, 1997). In foliage leaves of the C₄ plant maize, light has been shown to play important roles in C₄ cell type differentiation, direct activation of C₄ photosynthetic genes at the transcriptional level, and establishment of cell type specificity of C₄ gene expression (Nelson et al., 1984; Sheen and Bogorad, 1987a, 1987b; Langdale, et al., 1988a; Maroco et al., 1998). However, the role of light in cell type differentiation and cell-type-specific gene expression in C₄ dicot foliage leaves is largely unknown. Previous studies have investigated the pattern of cell-type-specific gene expression in developing foliage leaves of several C₄ dicots (Wang et al., 1992, 1993b; Dengler et al., 1995; Ramsperger et al., 1996; Berry et al., 1997), but the role of light in these processes was not directly addressed.

An important experimental approach in studying the light regulation of plant development is to compare changes in developmental and gene expression patterns between dark- and light-grown foliage organs or whole plants. In the C₄ grass maize, several immature, primordial foliage leaves are present in the embryo and expand under dark-grown conditions. Thus, the role of light in regulating C₄ gene expression and cell type differentiation can be assessed by comparing developing light- and dark-grown leaves (Nelson et al., 1984; Sheen and Bogorad, 1985, 1986, 1987a, 1987b; Langdale et al., 1988a, 1988b). However, in most dicot species, foliage leaves will not initiate or expand in dark-grown seedlings, so the comparative ontogenetic analysis used in C₄ grasses cannot be applied (Dengler et al., 1997). For this reason, studies of the role of light in the development of photosynthetic tissues in C₃ dicots have often exploited post-germination cotyledons, which can develop in both dark- and light-grown conditions and also have simpler developmental patterns than dicot foliage leaves (Tobin and Silverthorne, 1985; Tsukaya et al., 1994; Kretsch et al., 1995; Fankhauser and Chory, 1997).

In cotyledons of the dicot C₄ plant Amaranthus hypochondriacs (Berry et al., 1985, 1986; Wang et al., 1993a), Wang et al. (1993a) found that light is not required for either C₄ cell type differentiation or cell-type-specific gene expression. This finding contrasts with the results from studies of the foliage leaves of the C₄ grass maize, in which light is required for differentiation between BSC and MC (Sheen and Bogorad, 1985; Langdale et al., 1988b). It is not clear whether the discrepancy found between maize foliage leaves and amaranth cotyledons reflects differences between foliage leaves and cotyledons, between monocot leaves and dicot leaves in general, or simply species-specific variation in developmental patterns. More comparative studies of cotyledon development among different C₄ dicot species are needed to address these questions.

In this study, we assessed the role of light in the regulation of cell type differentiation and cell-type-specific expression of C₄ photosynthetic genes in cotyledons of a dicot C₄ plant, Flaveria trinervia. We examined cotyledon anatomical development and temporal and spatial mRNA accumulation of three C₄ photosynthetic genes that encode the small subunit of Rubisco (RbcS), PPDK, and PEPCase in cotyledons of both dark- and light-grown seedlings of F. trinervia from d 0 to d 12 after sowing. We found that light is essential for full morphological differentiation of cotyledon BSC and MC and for cell-type-specific expression of C₄ genes. Cell-type-specific gene expression was established between d 4 and d 7 after germination, coincident with maturation of Kranz anatomy in cotyledons. The light dependence of bundle sheath and mesophyll maturation in F. trinervia is more similar to that of maize than to that of amaranth. We suspect that the difference in light responsiveness and in patterns of C₄ gene expression between the two C₄ dicots F. trinervia and amaranth reflect species-specific differences in embryonic and post-embryonic cotyledon development.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant Material

Seeds were harvested from a self-pollinated Flaveria trinervia plant. Seeds for dark- and light-grown seedlings were soaked in deionized, distilled water for 24 h at 24°C in either regular transparent (light-grown d 0) or light-proof glass beakers (dark-grown d 0). The imbibed seeds were sown in pots with commercial potting soil (PRO-MIX, Premier Horticulture, Red Hill, PA), and covered with 5 mm of vermiculite. The light-grown seedlings were grown in a growth room with 14 h of light/d (approximately 180 µE m² s¹) at 24°C. Twelve pots for dark-grown seedlings were prepared and sealed in an air-fluent dark box and kept in a dark room at 24°C. One pot was randomly sampled each day (every 24 h) after sowing; the pot was opened and the seedlings were fixed in complete darkness.

Tissue Preparation

Whole seedlings were fixed in 4% (w/v) paraformaldehyde for 24 h at 4°C and stored at 4°C in 70% (v/v) ethanol. The cotyledons for in situ hybridization were embedded in paraffin (Paraplast, Oxford Labware, St. Louis) and 6-µm sections were mounted on poly-D-Lys (Sigma, St. Louis)-coated slides. Sections from seedlings of different developmental stages were placed side by side on the same slides for hybridization with the same in situ hybridization probes. Cotyledons used for anatomical observations were dehydrated in an ethanol and acetone series, embedded in Spurr's resin, and sectioned at 2 µm with a diamond knife on an ultramicrotome (MT-7, RMC, Tucson, AZ). The sections were mounted on poly-D-Lys-coated slides and stained in 0.5% (w/v) toluidine blue O in 0.1% (w/v) sodium carbonate (O'Brien and McCully, 1981). Sections were photographed using a microscope (Polyvar, Reichert, Vienna, Austria) and Tmax film (Eastman-Kodak, Rochester, NJ).

mRNA in Situ Hybridization

The DNA templates used for generating sense and antisense RNA probes were as follows: RbcS was from a 400-bp cDNA fragment that covers exons II and III of the RbcS-R1 gene from Flaveria ramosissima; it was cloned in pBlueScript II SK(+) (Shu, 1996), Ppc was a 1.82-kb cDNA fragment from the 3' region of the gene (3 kb) that encodes a C₄ isoform of PEPCase from F. trinervia (Hermans and Westhoff, 1992) Pdk was from a 1.3-kb cDNA fragment of the 5' region of the gene (3.1 kb from F. trinervia (Hermans and Westhoff, 1992; Rosche and Westhoff, 1995). Both Pdk and Ppc cDNA clones were kind gifts from Dr. Peter Westhoff.

Digoxigenin-labeled sense and antisense RNA probes were generated by in vitro transcription using T3 and T7 RNA polymerase (Boehringer Mannheim, Basel). Probe hydrolysis followed Langdale et al. (1988a). Slide pretreatment, prehybridizaton, and hybridization were modified from Langdale et al. (1988a) and Wang et al. (1992). Proteinase K treatment was 20 µg/mL for 20 min at 37°C, and RNase A treatment was 5 µg/mL for 20 min at 37°C. The prehybridization and hybridization solution (1,000 µL) contained 125 µL of 10× in situ salts (3.0 M NaCl, 0.1 M Tris, pH 6.8, 0.1 M sodium phosphate, pH 6.8, and 50 mM EDTA), 500 µL of deionized formamide, 250 µL of 50% (w/v) dextran sulfate, 25 µL of 20 µg/mL tRNA, 60 µL of 5 µg/mL poly(A⁺), and 40 µL of distilled, deionized water. Hybridization was at 50°C overnight. The highest washing stringency was 0.25× SSC at 42°C for 30 min. Immunodetection and colorimetric reaction followed protocols from Boehringer Mannheim. The color substrates nitroblue tetrazolium and X-phosphate were used. Slides were photographed using dark-field microscopy (Labophoto, Nikon, Tokyo) on Ektachrome 64T or 160T (Kodak).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Seedling Morphology, Cotyledon Development, and Cell Type Differentiation

In the light (see "Materials and Methods" for growth conditions used), the radicle of a F. trinervia seedling emerged from the seed coat by d 3 (48 h after sowing in soil). Two cotyledons emerged from the soil surface by d 4 (after 72 h). Newly emerged cotyledons were 2 mm long and yellowish opaque. Greening was first observed on the upper or adaxial side of the cotyledons and then on the lower or abaxial side by d 5 (after 96 h). The cotyledons expanded to 4 mm long by d 7 (after 144 h) and remained green through d 25 before undergoing senescence. The first foliage leaf was visible by d 10 and became fully expanded by d 15. When the seeds were germinated in the dark, radicle protrusion was delayed by 12 h. The 4-d-old seedling had an elongated hypocotyl and yellowish-white cotyledons about 1.5 mm long that were partly enclosed in the seed coat. Beyond d 4, the hypocotyl remained elongated and hooked, and cotyledon expansion was arrested. The whole seedling started to shrink by d 9 and died by d 18. Foliage leaves did not expand in dark-grown seedlings.

Figure 1, A to C, shows cotyledon anatomical development in light-grown seedlings. Kranz anatomy was present in a rudimentary form in the cotyledons of d 0 seedlings (after 24 h of soaking the seeds in distilled, deionized water) (Fig. 1A). Cotyledon vasculature was still forming at this stage, but veins already present had a well-defined bundle sheath layer. The mesophyll was five layers thick and strongly dorsiventral, with a well-defined adaxial palisade layer (layer 1). BSC and MC had numerous storage bodies and lacked differentiated plastids (Fig. 1A); proplastids were present but lacked well-developed thylakoids (V. Pontieri and N.G. Dengler, unpublished results). After 4 d in the light, BSC and MC became somewhat enlarged, storage bodies disappeared, and plastids were visible at the light microscope level. The intercellular space also expanded and some stomata appeared mature (Fig. 1B).

View larger version (197K): [in this window] [in a new window]   Figure 1.   Transverse sections of post-germination cotyledons of light- and dark-grown seedlings of F. trinervia. A, d 0, Light-grown; B, d 4, light-grown; C, d 7, light-grown; D, d 0, dark-grown; E, d 5, dark-grown; F, d 10, dark-grown. Bar = 50 µm. am, Abaxial MC; bs, BSC; i, intercellular space; pm, palisade MC; s, stomata; sm, spongy MC; unlabeled arrows, chloroplasts.

BSC showed a clear differentiation from other MC, with larger, centripetally located chloroplasts. This pattern of plastid localization contrasts with the centrifugal location seen in NADP-malic enzyme-type C₄ grass species (Hattersley and Browning, 1981; Dengler et al., 1996). By 7 d in the light, the Kranz anatomy became mature, with BSC and MC undergoing further enlargement and intercellular space becoming extensive (Fig. 1C). MC not directly adjacent to BSC did not develop plastids visible under the light microscope. Chloroplasts of both cell types had well-developed thylakoid membranes. Chloroplasts in MC formed grana, while these were inconspicuous or lacking in BSC (V. Pontieri and N.G. Dengler, unpublished results).

Figure 1, D and E, shows that in dark-grown seedlings, cotyledon anatomical development was similar to that of the light-grown seedlings between d 0 and d 4. In the dark, development is arrested at this stage, and from d 7 to 10, dark-grown BSC and MC remained small, with small and undifferentiated pro-plastids (Fig. 1F). The cellular expansion characteristic of the maturation phase of Kranz anatomy development in the light did not occur in the dark. There was also very little overall cotyledon expansion in dark-grown seedlings.

While MC expansion was evident in both light- and dark-grown cotyledons, it occured earlier and to a greater extent in the light (Fig. 1). The length to width ratio of palisade-like MC decreased from d 4 to d 7 and was more pronounced in light-grown cotyledons (Fig. 1). The isodiametric form of non-palisade MC was maintained during cell expansion through d 7 (Fig. 1). The dorsiventral mesophyll was moderately developed in the foliage leaves and cotyledons of F. trinervia, and was also evident in both cotyledons and foliage leaves of a C₃ species in the same genus, Flaveria pringlei (Fig. 2F; V. Pontieri and N.G. Dengler, unpublished results). This dorsiventral mesophyll pattern was less conspicuous in the cotyledons and foliage leaves of the C₄ dicot Amaranthus hypochondriacus (Wang et al., 1992, 1993a).

View larger version (72K): [in this window] [in a new window]   Figure 2.   RbcS mRNA accumulation patterns detected by mRNA in situ hybridization to transverse sections of the post-germination cotyledons of light- and dark-grown seedlings of F. trinervia (C4) (A-E) and F. pringlei (C3) (F). A, d 4, Light-grown, RbcS RNA antisense; B, d 7, light-grown, RbcS RNA antisense; C, d 9, light-grown, RbcS RNA antisense; D, d 4, dark-grown, RbcS RNA antisense, the seed coat is visible; E, d 7, dark-grown, RbcS RNA antisense; F, d 9, light-grown, RbcS RNA antisense, a section from F. pringlei (C3). Cotyledon pairs were appressed while embedding in paraffin so that their adaxial sides were adjacent to each other. Transverse sections (6 mm) were taken from a region midway between the apex and base of cotyledons. Bar = 320 µm.

Light Induction of BSC-Specific RbcS mRNA Accumulation

We examined the temporal and spatial accumulation patterns of RbcS mRNA in the cotyledons of both dark- and light-grown seedlings d 0 to d 12 after seed sowing, and the results are shown in Figure 2. The cotyledons of light- and dark-grown seedlings from d 0 to d 3 lacked detectable RbcS mRNA by our in situ hybridization techniques (data not shown).

In light-grown seedlings, accumulation of RbcS mRNAs in BSC was observed at d 4 and the level continued to increase up to d 7 (Fig. 2, A and B). Abundant accumulation of RbcS mRNA was also detected in MC at d 4 (Fig. 2A), but the level continuously decreased and became undetectable in the cotyledons of d 7 and older seedlings (Fig. 2, B and C). At d 4, dark-grown seedlings showed a low level of RbcS mRNA in both BSC and MC (Fig. 2D); the level of RbcS mRNA accumulation remained unchanged by d 6 and was found to decline slightly in both BSC and MC by d 7 (Fig. 2E). RbcS mRNA was not detectable in the cotyledons of d 9 and older seedlings under dark-grown conditions (data not shown). The RbcS mRNA accumulation patterns observed in our in situ hybridization studies fell into a similar temporal scheme to that reported for light- and dark-grown foliage leaves of maize seedlings (Nelson et al., 1984). For comparison, we also examined the pattern of RbcS mRNA accumulation in the cotyledons of a C₃ Flaveria species, F. pringlei. We detected abundant RbcS mRNA accumulation in all cotyledon photosynthetic cell types of light-grown seedlings, with the highest level in adaxial palisade MC (Fig. 2F). This resembles the pattern observed in the dark-grown cotyledons of F. trinervia (Fig. 2, D and E), although the accumulation level was much higher.

The above results suggest that in the cotyledons of F. trinervia, the expression of RbcS genes is developmentally controlled, so that RbcS mRNAs accumulate in both cell types independent of light conditions. However, in the presence of light, RbcS gene expression was either down-regulated or repressed in MC and up-regulated in BSC in d 5 and older cotyledons. The RbcS mRNA accumulation pattern in dark-grown cotyledons was C₃ like in terms of cell specificity and remained unchanged after d 5.

Light Induction of MC-Specific Pdk and Ppc mRNA Accumulation

We also examined the accumulation of Pdk and Ppc mRNA by mRNA in situ hybridization and daily sampling from d 0 to d 12 (Fig. 3). In MC of the light-grown cotyledons, we detected a dorsiventral gradient of mRNA accumulation for both Pdk and Ppc at d 4: high levels of mRNA accumulated in palisade-like MC (layer 1) and very low levels in other MC (layers 2-4) (Figs. 1 and 3, A and D). By d 5, high levels were detected in the MC of layers 2 to 4 (Fig. 3B). The level of mRNA accumulation decreased after d 5 and reached the steady state at d 7. From d 7 to d 12, the mRNA transcripts were detected in all MC except in the abaxial mesophyll (layer 5), which is adjacent to the lower epidermal cells and lacks direct contact with the BSC (Figs. 1C and 3, C and E). The absence of Pdk and Ppc expression in this mesophyll layer has also been observed in the foliage leaves of this species (Shu, 1996). The absence of Ppc mRNA accumulation in MC distal from BSC has also been reported for husk leaves of maize and foliage leaves of Atriplex rosea (Langdale et al., 1988b; Dengler et al., 1995).

View larger version (58K): [in this window] [in a new window]   Figure 3.   Pdk and Ppc mRNA accumulation patterns in F. trinervia cotyledons. A to C and F to H, Pdk mRNA accumulation patterns. D, E, I, and J, Ppc mRNA accumulation patterns. A, d 4, Light-grown, Pdk RNA antisense; B, d 5, light-grown, Pdk RNA antisense; C, d 7, light-grown, Pdk RNA antisense; D, d 4, light-grown, Ppc RNA antisense; E, d 7, light-grown, Ppc RNA antisense; F, d 4, dark-grown, Pdk RNA antisense; G, d 4, dark-grown, Pdk RNA sense probe; H, d 7, dark-grown, Pdk RNA antisense; I, d 7, dark-grown, Ppc RNA antisense; J, d 7, light-grown, Ppc RNA sense. Bar = 320 µm. For details, see Figure 2 legend. Under dark-field microscopy, the abundance of mRNA accumulation corresponds to the color saturation: yellowish red indicates low abundance and dark red indicates high abundance.

In BSC of light-grown cotyledons, low levels of accumulation were observed from d 5 to d 6 for Pdk mRNAs (Fig. 3B), but were undetectable by d 7 (Fig. 3C). A low level of RNA accumulation remained detectable in BSC of the d 7 and older seedlings using both the sense and antisense endogenous Ppc probes (Fig. 3E) at the same high hybridization stringency (Fig. 3, E and J).

We found that the temporal and spatial patterns of Ppc mRNA accumulation were very similar to those of Pdk mRNA, with two exceptions: (a) the steady-state level of Ppc mRNA in the MC of the expanded cotyledons (from d 7 to d 12) was much higher than that of Pdk mRNA (Fig. 3, C and E), which is consistent with the estimates from RNA blotting analysis of mRNA from the maize leaf blade (Sheen and Bogorad, 1987b; Langdale and Kidner, 1994); (b) a low level of accumulation of unknown RNA transcripts was detected in the BSC of light-grown cotyledons with both sense and antisense endogenous Ppc probes, whereas the mRNA accumulation detected with endogenous Pdk gene probes was clearly MC specific in the 7 d and older cotyledons of light-grown seedlings.

In the cotyledons of dark-grown seedlings, no mRNA accumulation for either gene was detected between d 0 and d 3. The level of Pdk and Ppc mRNA accumulation in the palisade mesophyll was higher than in the other mesophyll layers or in BSC between d 4 and d 7 (Fig. 3, F, H, and I).

The accumulation patterns of RbcS, Pdk, and Ppc mRNAs in the expanded light-grown cotyledons of 7 d and older seedlings were essentially the same as those detected in the expanded foliage leaves of this species (Shu, 1996). Studies of both anatomy and photosynthetic physiology have shown that the foliage leaves of F. trinervia are typical C₄ organs (Ku et al., 1991). Although no data on photosynthetic physiology are available for the cotyledons, our results concerning anatomy and gene expression patterns indicate that the expanded cotyledons of light-grown F. trinervia could function as typical C₄ photosynthetic organs and undergo C₄ CO₂ assimilation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Light Is Required for Full Differentiation of Photosynthetic Cell Types in C₄ F. trinervia Cotyledons

Post-germination cotyledon development in F. trinervia appears to include two phases: a light-independent phase, or post-germination I, and a light-dependent phase, or post-germination II. From d 0 to d 4, both light- and dark-grown seedlings shared similar developmental patterns, such as nutrient reserve breakdown, repositioning and differential growth of plastids so that BSC have enlarged, and centripetally placed plastids, while MC had small, peripherally positioned plastids and C₃-like photosynthetic gene expression. Light is therefore apparently not required for these developmental processes. We refer to this phase as post-germination I (Fig. 4A).

View larger version (50K): [in this window] [in a new window]   Figure 4.   Differences in cotyledon and seedling development between two C4 dicotyledonous species, F. trinervia (A) and A. hypochondriacus (B). Embryogenesis, Formation of the embryo proper and endosperm; the two species represent two types of embryogenesis patterns in angiosperms. Post-embryogeny, Absorption of endosperm and seed formation in both species; cotyledon expansion in A, perisperm formation in B; cotyledon cell differentiation and light-induced chloroplast formation take place. Maturation, Nutrient reserve deposition; lipid and protein deposition in cotyledons for both species, starch deposition in cotyledons for A and in perisperm for B. Desiccation, Dehydration and dormancy of seedlings. Germination, From seed imbibition to rupture of testa by radicle. Post-germination I, C3-like development such as mobilization of cotyledon nutrient reserves, resumption of cell type differentiation, and non-cell-type-specific C3 and C4 gene expression. Post-germination II, Full C4 development such as maturation of Kranz anatomy and establishment and maintenance of cell-type-specific C4 gene expression. The two species employ two different subtypes of decarboxylation pathways and show different developmental patterns in light- and dark-grown conditions (shaded). C4 competence, Hypothetical period or developmental window when essential light-induced signals for C4 development and cell-type-specific C4 gene expression are generated.

From d 5 to d 7, light- and dark-grown seedlings showed divergent development patterns. In light-grown seedlings, we observed arrest of hypocotyl elongation, continued BSC and MC expansion and redifferentiation, and high-level, cell-type-specific mRNA accumulation of the three C₄ photosynthetic genes in the cotyledons. In contrast, we observed continuous hypocotyl elongation, arrest of cell expansion and redifferentiation, and low and nonspecific mRNA accumulation for all three genes in dark-grown seedlings. Therefore, light is essential for the full differentiation of cotyledon BSC and MC. We refer to this phase as post-germination II (Fig. 4A).

Our division of post-germination cotyledon development into two phases is based solely on the observation that some development can proceed in both light- and dark-grown seedlings. It is known that regulatory proteins and mRNAs synthesized in the post-embryogeny phase can be stored in various forms in mature and desiccated seeds and be mobilized in the post-germination phases to regulate gene activities (Goldberg et al., 1989; Kretsch et al., 1995). Thus, it is possible that light-independent developmental processes in dark-grown seedlings of post-germination I are actually light dependent and regulated by light-induced signals synthesized during the post-embryogeny phase of seed development.

We found that some veins with morphologically distinctive BSC and associated mesophyll layers were already present in d 0 imbibed embryos, but that fully mature Kranz anatomy, including substantial cell expansion and the formation of extensive intercellular space in the abaxial mesophyll did not develop until d 7. Thus, it is likely that C₄ cell type development is initiated in either late embryogenesis or in the post-embryogeny phase of the parental plant, but that differentiation processes are arrested by nutrient storage and seed desiccation. Further development of Kranz anatomy is resumed during the post-germination I phase and is completed during the post-germination II phase only in light-grown cotyledons.

Based on our observations, the completion of the transition from post-germination I to post-germination II in the cotyledons of F. trinervia requires light-induced signals that contribute to the competence for further C₄ development (Fig. 4). C₄ competence is likely attained around d 5 under light-grown conditions. Previous studies indicate that C₄ competence is also required for foliage leaf development in both monocot and dicot C₄ species, since young leaves that have not obtained C₄ competence always show C₃-like gene expression patterns even in light-grown conditions (Sheen and Bogorad, 1985; Langdale et al., 1988b; Wang et al., 1992, 1993b; Dengler et al., 1995).

Establishment of Cell-Type-Specific C₄ Gene Expression Is Light Dependent

In the present study, we found that light is essential for inducing both high-level and cell-type-specific accumulation of RbcS, Pdk, and Ppc mRNAs in post-germination cotyledons of light-grown seedlings. The two events, up-regulation and the cell-type-specificity of C₄ gene expression, are likely regulated by different light-induced signals. While up-regulation of the three C₄ genes was observed by d 4 (right after seedling emergence), initial mRNA accumulation patterns were not cell type specific. Complete cell-type-specific patterns for RbcS and Pdk genes were only finalized at d 7. The difference of 3 d between the two events indicates that the cell type specificity of C₄ gene expression is likely regulated by light-induced developmental signals or conditions, rather than by direct light action on C₄ gene transcription. A study of maize foliage leaves using biolistic gene transfer methods showed that red light, though important in inducing rapid up-regulation of C₄ genes and chloroplast development, is not sufficient to suppress RbcS gene transcription in improper cell types such as maize epidermal cells (Bilang and Bogorad, 1996).

Recently, transgenic studies using GUS fusion constructs from different sequences at the 5' and 3' ends of Pdk genes, Ppc, and NADP-malic enzyme genes have shown that control of cell type specificity and the level of gene expression use different cis-acting elements (Taylor et al., 1997; Westhoff et al., 1997). The cis-acting elements involved in controlling cell-type-specific expression of the C₄-enzyme-coding genes are not found in the genes that encode enzymes of C₃ isoforms (Sheen, 1991; Chitty et al., 1994; Stockhaus et al., 1994, 1997; Furbank et al., 1997; Marshall et al., 1997). All of the above results suggest that the regulation of C₄ gene expression is 2-fold: regulation of the steady-state level of gene activity, a common theme shared with C₃ plants, and regulation of cell type specificity of gene expression, a unique feature of C₄ genes.

We detected a low level of hybridization signal in cotyledon BSC with the antisense Ppc mRNA probe (Fig. 3E). We also detected a similar level of hybridization signal in both BSC and MC at the same high hybridization stringency with the sense riboprobe of Ppc (Fig. 3J). Therefore, it is likely that the signal detected in BSC with both Ppc sense and antisense probes are background noise, and that the accumulation of Ppc and Pdk mRNAs are MSC specific.

We found that an increase in Pdk and Ppc mRNA levels in MC of light-grown cotyledons coincided with cotyledon greening. High levels of mRNA accumulation for both genes begins in the palisade MC (layer 1) by d 4 and extends to the abaxial side of the cotyledons by d 5. Dorsiventral chlorophyll gradients that cause the dorsiventral cotyledon greening patterns have also been reported in the post-germination cotyledons of a C₃ plant, Cucurbita pepo (Knapp et al., 1988), and have been suggested to be due to differential exposure to blue light along the dorsiventral axis of the cotyledons (Knapp et al., 1988). It remains to be investigated whether the Pdk and Ppc mRNA gradients seen in F. trinervia might have been caused by a blue light gradient.

Cell-Type-Specific C₄ Gene Expression Occurs during Maturation of Morphological Cell Type Differentiation

An important question in understanding C₄ development is whether light induction of C₄ gene expression is coupled with leaf development and leaf cell type differentiation (Nelson and Dengler, 1992; Liu and Dengler, 1994; Berry et al., 1997). Our results show that in F. trinervia, accumulation of mRNA for genes involved in C₄ carbon metabolism occurs as morphological cell type differentiation becomes mature. BSC and MC differentiation proceeded simultaneously in post-germination cotyledons and full differentiation of BSC and MC were observed by d 7, which is also the time when cell-type-specific mRNA accumulation for all three genes takes place. In contrast to our findings for F. trinervia cotyledons, in young foliage leaves of this species the establishment of a MC-specific pattern of Ppc and Pdk expression was conspicuously delayed in relation to the establishment of a BSC-specific pattern of RbcS gene expression. The same delay of Ppc and Pdk expression relative to that of RbcS is seen in the young foliage leaves of maize, amaranth, and Atriplex rosea (Langdale et al., 1988a; Schäffner and Sheen, 1992; Wang et al., 1992; Dengler et al., 1995; Shu, 1996). This distinction between cotyledons and foliage leaves may reflect a difference in the timing of cell proliferation relative to differentiation. In cotyledon development, cell proliferation only occurs during embryogenesis, and post-germination cotyledon cells do not divide (Tsukaya et al., 1994; Kretsch et al., 1995). In contrast, developing foliage leaves undergo considerable cell proliferation, and BSC surrounding the veins may be delimited before cell division within the MC ceases (Dengler et al., 1996).

Our data support the view that signals involved in cell-type-specific regulation of C₄ gene expression are positional relative to bundle sheath or vascular tissues (Langdale and Nelson, 1991; Nelson and Langdale, 1992). It remains unclear whether the up-regulation of Pdk and Ppc and the suppression of RbcS genes in the same MC are regulated by the same positional signals. In maize husk leaves, the absence of Ppc mRNAs and the presence of rbcL and RbcS mRNAs are found to coincide in the MC that do not have direct contact with the BSC (Langdale, 1988b; Langdale and Nelson, 1991). We did not observe the same coincidence in the cotyledons of F. trinervia. No RbcS mRNA accumulation was detected in the fifth layer of MC, where the Pdk and Ppc mRNAs are absent. This observation suggests that up-regulation of Pdk and Ppc and suppression of RbcS genes in MC, though normally coincident, may be controlled by different signals. Results from promoter analyses, mutant studies, and transgenic studies using different C₄ photosynthetic genes in Flaveria sp. also suggest that there is no universal regulation mechanism among different C₄ photosynthetic genes (Langdale and Kidner, 1994; Furbank et al., 1997; Taylor et al., 1997; Westhoff et al., 1997).

F. trinervia and A. hypochondriacus Differ in C₄ Gene Expression Pattern and Cotyledon Development

Our results demonstrate that in F. trinervia, cell-type-specific expression of genes involved in C₄ carbon metabolism in the post-germination cotyledons only takes place in light-grown seedlings, not in dark-grown seedlings. This is true in spite of the existence of clear morphological differentiation between BSC and MC in the cotyledons. This finding emphasizes the fact that light is essential for the establishment and maturation of differential expression of these genes. The light requirement for differential expression of rbcS, PPDK, and PEPCase mRNA in F. trinervia is in marked contrast to the observed light independence of their differential expression in the cotyledons of another C₄ dicot, A. hypochondriacus (Amaranthaceae) (Wang et al., 1993a). Wang et al. (1993a) found that the accumulation of PPDK mRNA and protein and PEPCase protein are MC specific in cotyledons of both dark- and light-grown 2-d-old seedlings. RbcS mRNAs and polypeptides were not cell type specific at d 2, but became BSC specific at d 5 in both dark- and light-grown cotyledons. Thus, light is not required for the establishment of cell-type-specific C₄ gene expression in this species (Wang et al., 1993a). The underlying mechanisms that lead to this difference between the two dicot species is unclear.

It is possible that the apparent difference in light requirement for cell-type-specific C₄ gene expression between the two species is associated with their difference in seedling development and cotyledon cell type differentiation. Previous studies have shown that F. trinervia and A. hypochondriacus have strikingly different patterns of embryogenesis, post-embryogenic development, and post-germination development, and these are summarized in Figure 4. Embryogenesis in angiosperms is classified into six types (Johri et al., 1992): F. trinervia is an Asterad type, whereas A. hypochondriacus is a Chenopodiad type (Misra, 1964; Coimbra and Salema, 1994). Dicot plants are also classified into at least three types based on seed nutrient storage location: cotyledon storage, such as A. hypochondriacus endosperm storage, or perisperm storage. The three types show distinctive temporal patterns of cotyledon development (Johri et al., 1992; Bewley and Black, 1994; Kaplan and Cooke, 1997).

Both Flaveria and Amaranthus spp. have nuclear endosperm that starts to form at the heart stage of embryogenesis. At the post-embryogeny phase, endospermis is gradually absorbed by the growing embryo. In amaranth, a portion of the nucellus converts into a perisperm storage tissue (Fig. 4B). Carbohydrates are stored as starch grains in perisperm plastids, but starch grains are absent from cotyledon plastids (Coimbra and Salema, 1994). In Flaveria, cotyledons and hypocotyl convert to nutrient storage organs during post-embryogenic development, and cotyledon cells are filled with protein bodies, lipids, and plastids with starch grains (Misra, 1964).

Previous studies have shown that the presence of carbohydrates suppresses C₄ gene transcription in isolated maize leaf MC (Sheen, 1990). In developing foliage leaves of A. hypochondriacus, a tight coordination between cell-type-specific C₄ gene expression and the state of carbon metabolism or sink-source transition have been reported (Wang et al., 1993b; Berry et al., 1997). It is plausible that amaranth cotyledons, which lack starch deposition, could have faster photosynthetic development than the Flaveria cotyledons in the post-embryogeny phase, to become more leaf-like and obtain light-induced C₄ competence before seed desiccation (Fig. 4). Further detailed comparative study is needed to confirm this speculation.

The temporal differences in cotyledon development and C₄ gene expression between A. hypochondriacus and F. trinervia are likely an evolutionary adaptation of each species to its germination environment. The great diversity among different plant species in seed structure, germination strategy, and corresponding temporal control of light-regulated genes is well documented (Stebbins, 1974; Johri et al., 1992; Bewley and Black, 1994).

The differences between these two C₄ dicot species in C₄ development of cotyledons provide evidence that independent evolutionary origins of C₄ photosynthetic mechanisms need not arise by the evolution of common developmental control pathways. A third distinctive developmental pattern has been recognized in species of Haloxylon (Chenopodiaceae) that show C₄ photosynthesis in the single large subepidermal BSC/MC ring surrounding assimilatory shoots (Pyankov et al., 1999). The reduced leaves of these plants are nonphotosynthetic, but the cotyledons are C₃ and show no morphological differentiation among the photosynthetic cells at any stage of development (Pyankov et al., 1999).

Evidence of independent evolution of C₄ photosynthetic pathways has been reported and reviewed for grass species (Hattersley and Browning, 1981; Hattersley, 1984; Dengler et al., 1985, 1986; Sinha and Kellogg, 1996). Extensive comparative studies of C₄ gene expression and post-germination cotyledon development across different taxa of C₄ dicots are promising to bring new insight into the evolution of developmental mechanisms in general and the evolution of this complex adaptation in particular.