Plastid signals that affect photomorphogenesis in Arabidopsis thaliana are dependent on GENOMES UNCOUPLED 1 and cryptochrome 1

Summary

• • 
  When plastids experience dysfunction they emit signals that help coordinate nuclear gene expression with their functional state. One of these signals can remodel a light-signaling network that regulates the expression of nuclear genes that encode particular antenna proteins of photosystem II. These findings led us to test whether plastid signals might impact other light-regulated processes.
• • 
  Photomorphogenesis was monitored in genomes uncoupled 1 (gun1), cryptochrome 1 (cry1), and long hypocotyl 5 (hy5), which have defects in light and plastid signaling, by growing Arabidopsis thaliana seedlings under various light conditions and either treating or not treating them with antibiotics that induce chloroplast dysfunction and trigger plastid signaling.
• • 
  It was found that plastid signals that depend on GUN1 can affect cotyledon opening and expansion, anthocyanin biosynthesis, and hypocotyl elongation. We also found that plastid signals that depend on CRY1 can regulate cotyledon expansion and development.
• • 
  Our findings suggest that plastid signals triggered by plastid dysfunction can broadly affect photomorphogenesis and that plastid and light signaling can promote or antagonize each other, depending on the responses studied. These data suggest that GUN1 and cry 1 help to integrate chloroplast function with photomorphogenesis.

Introduction

Light is one of the most important environmental signals perceived by plants. It is required for chloroplast biogenesis and photosynthesis, regulates metabolism, entrains the circadian clock, and has striking effects on organ growth and development (Jiao et al., 2007). Many photoreceptors and downstream signaling components that transduce these light signals have been identified. Mechanisms for light-signaling networks have been partly defined and are characterized by extensive interactions and redundancies (Casal, 2006; Jiao et al., 2007).

Driving chloroplast biogenesis and function is perhaps one of the most important functions of light-regulated signaling networks because photosynthesis provides the energy required for growth, development, and reproduction. Light regulates chloroplast biogenesis and function by controlling the expression of genes that encode proteins with chloroplast-related functions (Tyagi & Gaur, 2003; Monte et al., 2004; Ohgishi et al., 2004; Toyoshima et al., 2005; Marín-Navarro et al., 2007) and activating a light-dependent enzyme in the chlorophyll biosynthetic pathway (Tanaka & Tanaka, 2007). However, the plastid is not completely subordinate to light conditions. Plastids emit signals that help coordinate nuclear gene expression with the functional state of the plastid. Plastid signals have been reported to coordinate the expression of photosynthesis-associated nuclear genes (PhANGs) and stress-related nuclear genes with plastid function (Strand et al., 2003; Pesaresi et al., 2007; Dietzel et al., 2008; Kim et al., 2008). Plastid signals also contribute to efficient chloroplast biogenesis (Mochizuki et al., 1996; Ruckle et al., 2007), optimize the stoichiometry of photosystems I and II (Dietzel et al., 2008), and trigger stress and programmed cell death responses (Nott et al., 2006; Kim et al., 2008). At least some of these plastid signals can be transduced in the dark (Sullivan & Gray, 1999; Ruckle et al., 2007).

Plastid signals have been reported to affect light responses. Because a subset of mutants with defective plastid-localized proteins exhibit aberrant leaf anatomy, plastid signals are thought to contribute to proper leaf morphogenesis (Rodermel, 2001; Yu et al., 2007). In addition, plastid signals are thought to be required for mesophyll cell division and to promote mesophyll cell elongation when leaves acclimatize to high-intensity light (Tan et al., 2008). Also, an analysis of the chlorophyll-deficient chlorophyll a/b-binding (CAB) protein-underexpressed (cue) mutants provides evidence that plastid signals can have mild inhibitory effects on cotyledon opening and hypocotyl elongation in continuous red and far-red light (Vinti et al., 2005). The long after far-red 6 (laf6) mutant, which has a defect in a plastid-localized ATP-binding-cassette protein, exhibits a striking increase in hypocotyl elongation in far-red light (Møller et al., 2001). The photomorphogenic phenotype of laf6 was proposed to result from accumulation of protoporphyrin IX that might act as a signaling factor (Møller et al., 2001) or defects in iron homeostasis (Xu et al., 2005). Alternatively, Cornah et al. (2003) suggested abnormal tetrapyrrole metabolism in laf6 may lead to reduced levels of phytochromobilin, the tetrapyrrole chromophore of phytochrome, and therefore lower levels of active phytochromes in laf6 relative to wild type.

In seedlings treated with inhibitors of chloroplast biogenesis, two genetically distinct plastid signals repress genes that encode the light harvesting chlorophyll a/b-binding proteins of photosystem II (Lhcb) when plastids experience dysfunction. One of these plastid signals is dependent on GENOMES UNCOUPLED 1 (GUN1), which encodes a plastid-localized pentatricopeptide repeat protein that colocalizes with nucleoids (Koussevitzky et al., 2007). The other plastid signal affects a light signaling network that depends in part on the basic leucine zipper transcription factor long hypocotyl 5 (HY5) acting downstream of the blue light photoreceptor cryptochrome 1 (cry1) (Ruckle et al., 2007). The finding that knocking out both of these signals causes a synergistic upregulation of Lhcb expression implies that the full impact of plastid signals on various processes may have been obscured by genetic complexity during previous analyses of plastid signaling mutants. To test whether the plastid signals that are dependent on GUN1 and light signaling networks contribute to photomorphogenesis, we quantified a variety of commonly studied photomorphogenic processes (Fankhauser & Casal, 2004) in gun1 and light-signaling mutants of Arabidopsis thaliana treated with inhibitors of chloroplast biogenesis in a variety of light qualities and quantities. Our results indicate that plastid signals that are dependent on GUN1 and cry1 can broadly affect photomorphogenesis.

Materials and Methods

Plant material and growth conditions

Plant material and all growth conditions (i.e. growth media, temperature, and light sources) were as described in Ruckle et al. (2007). Briefly, Arabidopsis thaliana (L.) Heynh. ecotype Columbia-0 (Col-0) seedlings were grown on Linsmeier and Skoog media containing 2% sucrose and either 0.5 mm lincomycin, 0.5 mm erythromycin, or no inhibitor of chloroplast biogenesis, as described by Ruckle et al. (2007). For each experiment, seeds were stratified, germination was promoted by irradiating seeds with red light, seedlings were grown in the dark for 23 h as described by Ruckle et al. (2007), and seedlings were then grown for 6 d in the indicated conditions.

Analysis of cotyledon opening and expansion

To quantify cotyledon opening, we placed seedlings in a horizontal position on moist sheets of BioDesignGelWrap (BioDesign Inc. of New York, Carmel NY, USA) alongside a ruler and imaged them using a flatbed scanner at a resolution of 300 dots per inch (dpi). Angles between the cotyledons were quantified using imagej (National Institutes of Health) as recommended by Fankhauser & Casal (2004). For these measurements and all other measurements, statistical significance (P < 0.05) was tested by either one- or two-way anova, as indicated. For anthocyanin measurements, log transformation of the data was necessary to meet the assumptions of the statistical test.

To quantify cotyledon area, we excised cotyledons from each seedling and placed them on moist sheets of BioDesignGelWrap, imaged them, and quantified them as described for cotyledon opening. Fixed and flattened cotyledons were prepared to account for the downward curling of the cotyledons observed in lincomycin-treated wild type in 50 µmol m⁻² s⁻¹ blue light. Separate groups of seedlings were fixed in an ethanol–acetic acid (3 : 1) solution overnight as recommended by Kakiuchi et al. (2007) that also contained 0.01% Coomassie Brilliant Blue R-250 to facilitate the subsequent imaging of the cotyledons. The fixed cotyledons were subsequently washed once with ethanol–acetic acid (3:1) to remove excess stain. The fixed and stained cotyledons were flattened as recommended by Neff & Chory (1998). Briefly, the cotyledons were placed on the sticky side of transparent tape and flattened with forceps. A dry sheet of BioDesignGelWrap was placed on top of these cotyledons, which were then scanned as described for cotyledon opening. We quantified cotyledon areas using imagej, as recommended by Fankhauser & Casal (2004). We calculated the amount of curling by dividing the flattened cotyledon areas by the unflattened areas.

Scanning electron microscopy (SEM) and cross sections

For SEM, samples were fixed in 2.5% glutaraldehyde–2.5% paraformaldehyde with a 0.1 m cacodylate buffer, pH 7.4, and dehydrated. Critical-point-dried samples were osmium coated and examined using the 6400 JEOL scanning electron microscope with accelerating voltage of 10 kV. For cross-sections, tissue was fixed as described for SEM, postfixed in 1% osmium tetroxide in 0.1 m cacodylate buffer, dehydrated in a graded acetone series, and infiltrated and embedded in Poly/Bed 812 resin. Sections of 1µm were cut with a Power Tome XL ultramicrotome (RMC, Boeckeler Instruments, Tucson AZ, USA) and stained with Epoxy Tissue Stain (Electron Microscope Sciences, Hatfield, PA, USA). Cross-sections were visualized with an Axio Imager M1 microscope (Carl Zeiss Inc., Thornwood, NY, USA).

Analysis of SEMs and cross-sections

We measured cell areas from the SEM images using the free-hand tool of imagej, as recommended by Djakovic et al. (2006). The interdigitating lobes of the epidermal pavement cells were quantified as recommended by Djakovic et al. (2006). For this calculation, we measured cell perimeter and area using imagej and, with these values, calculated a form factor, as recommended by Russ (2002). The form factor is equal to (4π)(cell area)/(cell perimeter)²; it describes the amount of convolution at the cell periphery. For example, a circle has a form factor of 1 and the form factor becomes smaller than 1 as convolution, such as interdigitating lobes, increases. We calculated the density of stomata by counting the number of stomata in a micrograph and dividing by the area of each micrograph (120 000 µm²), as recommended by Berger & Altmann (2000). To quantify the wrinkling of the cotyledon surface, we calculated a cotyledon surface factor, which is the length of the abaxial cotyledon surface divided by the diameter of the cotyledon. These values were determined using imagej.

Analysis of anthocyanin levels and hypocotyl lengths

Anthocyanin levels were quantified exactly as recommended by Fankhauser & Casal (2004). We measured hypocotyl lengths by placing seedlings onto moist BioDesignGelWrap. These seedlings were imaged and hypocotyl lengths were quantified as described for cotyledon opening.

Results

Analysis of cotyledon opening and expansion in gun1 and light-signaling mutants

We expected that GUN1 might have a greater impact on photomorphogenesis of the cotyledons in low-fluence rates than in high-fluence rates because: the light-signaling network that represses Lhcb is more important to the repression of Lhcb when fluence rates are increased, and GUN1-dependent plastid signals contribute to the repression of Lhcb regardless of light conditions (Ruckle et al., 2007). To test this hypothesis, we analysed the fluence-rate responses of cotyledon opening and expansion under conditions that either induced or did not induce plastid signaling. We tested these fluence rate responses in blue light because the effects of plastid signals on blue light signaling are better understood than the effects of plastids on the components of the light signaling network that perceive other light qualities (Ruckle et al., 2007). To trigger plastid-to-nucleus signaling, we treated seedlings with lincomycin and erythromycin, which inhibit plastid translation (Mulo et al., 2003). These inhibitors have been shown to block both etioplast and chloroplast biogenesis (Sullivan & Gray, 1999) and to repress PhANG expression in both light and dark (Sullivan & Gray, 1999; Ruckle et al., 2007). Seedlings treated with these inhibitors have nonphotosynthetic plastids rather than chloroplasts, are viable when provided sucrose and become green upon removal of the inhibitor (Sullivan & Gray, 1999; Ruckle et al., 2007). GUN1 has been localized to chloroplasts and is thought to also reside in the nonphotosynthetic plastids found in the aerial parts of lincomycin- and erythromycin-treated seedlings where it contributes to the biosynthesis or transduction of plastid signals (Koussevitzky et al., 2007). We observed that cotyledon opening requires higher fluence rates of blue light when wild-type seedlings were treated with lincomycin compared with untreated wild-type seedlings (Fig. 1 and Supporting Information, Fig. S1). By contrast, cotyledon opening exhibited a similar fluence rate response in treated gun1-1, untreated gun1-1 and untreated wild type (Figs 1, S1). gun1-1 is a leaky allele (Koussevitzky et al., 2007). Similar results were obtained with lincomycin and erythromycin (Fig. S2). Erythromycin and lincomycin both inhibit plastid translation, but by different mechanisms (Mulo et al., 2003). These data indicate that: the observed differences in cotyledon opening are likely caused by reduced GUN1 activity in the mutant and not by resistance to particular inhibitors; and that GUN1-dependent plastid signals can repress cotyledon opening in low-fluence-rate blue light.

Figure 1. Cotyledon opening in gun1-1 and wild-type (Col-0) Arabidopsis thaliana seedlings in different fluence rates of blue light. Col-0 (squares) and gun1-1 mutants (circles) were grown in the presence (closed shapes) or absence (open shapes) of lincomycin in the indicated fluence rates of blue light. For each data point, n > 35. Error bars represent 95% confidence intervals. Statistical significance (P < 0.05) was tested with a two-way anova. *, Significant difference between lincomycin-treated gun1-1 and lincomycin-treated wild type; **, significant difference between lincomycin-treated and untreated wild type.

We observed that cotyledon expansion increased similarly in gun1-1 and wild-type seedlings that were not treated with an inhibitor of chloroplast biogenesis as the fluence rate of blue light increased (Figs S1 and S3a). This relationship between cotyledon expansion and fluence rate of blue light has been reported previously for untreated wild-type seedlings (Jackson & Jenkins, 1995; Ohgishi et al., 2004). By contrast, when seedlings were treated with lincomycin, cotyledon expansion was significantly greater in gun1-1 compared with wild type, but these differences diminished above 5 µmol m⁻² s⁻¹ blue light and there were no significant differences in 50 µmol m⁻² s⁻¹ blue light (Figs S1 and S3a). These data indicate that GUN1-dependent plastid signals can inhibit cotyledon expansion in lincomycin-treated seedlings and that high-fluence-rate blue light can also repress cotyledon expansion when chloroplast biogenesis is blocked regardless of whether GUN1 is active.

In high fluence rates, the cotyledons of wild-type seedlings curled downward (Figs S1 and S6a). To account for this effect, we repeated this fluence response experiment but fixed and flattened cotyledons before measuring cotyledon area (Fig. 2). When wild type seedlings were treated with inhibitors of chloroplast biogenesis in 50 µmol m⁻² s⁻¹ blue light, the flattened cotyledons of lincomycin-treated wild type were 50% larger than cotyledons that were not processed in this manner (Fig. S3b). These data indicate that accounting for the curling of the cotyledons is most important when measuring cotyledon areas in treated seedlings grown in 50 µmol m⁻² s⁻¹ blue light. Cotyledon expansion was greater in gun1-1 than in wild type, regardless of whether chloroplast biogenesis was blocked with lincomycin or erythromycin (Fig. S4a–c). Similar results were obtained with gun1-101 (Fig. S4a–c), a T-DNA allele that appears to be a null (Ruckle et al., 2007).

Figure 2. Flattened cotyledon areas in gun1-1 and wild-type (Col-0) Arabidopsis thaliana seedlings. Col-0 (squares) and gun1-1 mutants (circles) were grown in the presence (closed shapes) or absence (open shapes) of lincomycin in the indicated fluence rates of blue light. Cotyledons were fixed and flattened as described in Materials and Methods section. For each data point, n > 29. Error bars represent 95% confidence intervals. Statistical significance (P < 0.05) was tested with a two-way anova. *, Significant difference between lincomycin-treated gun1-1 and lincomycin-treated wild type; **, significant difference between lincomycin-treated and untreated wild type.

Because plastid signaling triggered by inhibitors of chloroplast biogenesis had been shown to affect the nature of Lhcb gene regulation by cry1 and HY5 (Ruckle et al., 2007) and because cry1 had been shown to affect the photomorphogenesis of the cotyledons in high-fluence-rate blue light (Jackson & Jenkins, 1995; Neff & Chory, 1998; Ohgishi et al., 2004), we tested whether cry1 and HY5 contribute to the repression of cotyledon opening and expansion in high-fluence-rate blue light. We grew gun1-1, cry1 and hy5 mutants in the presence and absence of inhibitors of chloroplast biogenesis in 50 µmol m⁻² s⁻¹ blue light. cry1 and hy5 are T-DNA insertion alleles that are known to be nulls (Ruckle et al., 2007). In 50 µmol m⁻² s⁻¹ blue light, cotyledons were completely unfolded in gun1-1 cry1, the single mutants, and wild type, regardless of whether seedlings were treated with inhibitors of chloroplast biogenesis (Figs 1, 3b, and S1). When plants were not treated with inhibitors of chloroplast biogenesis in this same fluence rate of blue light, no difference was observed in cotyledon areas, except for cotyledons of cry1 hy5, which were two-thirds the size of wild type (Fig. 3a,b).

Figure 3. Cotyledon areas in gun1, cry1, and hy5 mutants of Arabidopsis thaliana. (a) Quantification of cotyledon area measurements in gun1, cry1 and hy5 mutants. The lines indicated were grown on media containing lincomycin (medium tinted bars) or erythromycin (light tinted bars) or containing no inhibitor (dark tinted bars) in 50 µmol m⁻² s⁻¹ blue light. Cotyledons were fixed and flattened, and cotyledon areas were measured as described in the Materials and Methods section. For each line in each condition, n > 33. Error bars represent 95% confidence intervals. Statistical significance (P < 0.05) was tested with a one-way anova. *, Significant difference between single mutants and wild type or a significant difference between double mutants and single mutants grown in the same conditions. (b) Representative seedlings and cotyledons were grown as described in (a). Bar, 2 mm.

When seedlings were treated with inhibitors of chloroplast biogenesis in 50 µmol m⁻² s⁻¹ blue light, the areas of the cotyledons in gun1 or cry1 were indistinguishable from wild type (Fig. 3a,b). By contrast, the areas of the cotyledons increased in the gun1-1 cry1 double mutant by c. 50%, but were still approximately half the size of untreated seedlings (Fig. 3a,b). Because the gun1 cry1 double mutant has larger cotyledons than wild type or the single mutants, we conclude that either the cry1- or the GUN1-dependent plastid signals can compensate for the loss of the other and that these signals are functionally redundant for the repression of cotyledon expansion when chloroplast biogenesis is blocked. In addition, because knocking out the CRY1 gene ameliorates the repressive effect of high-fluence-rate blue light on cotyledon expansion, we conclude that this repression likely depends on a functional cry1 photoreceptor. Cotyledon areas were indistinguishable among hy5, gun1 hy5, and cry1 hy5 double mutants that were treated with inhibitors of chloroplast biogenesis (Fig. 3). In contrast to 50 µmol m⁻² s⁻¹ blue light, we found that gun1-1 had larger cotyledons than wild type when chloroplast biogenesis was blocked in 100 µmol m⁻² s⁻¹ red light (Fig. 4). Moreover, we found that phyB remained a positive regulator of cotyledon expansion regardless of whether seedlings are treated with inhibitors of chloroplast biogenesis (Fig. 4).

Figure 4. Analysis of cotyledon expansion in gun1-1 and phyB. (a) Cotyledon area measurements in gun1-1 and phyB mutants. Arabidopsis thaliana seedlings were grown on media containing lincomycin (light tinted bars) or no inhibitor of chloroplast biogenesis (dark tinted bars) in 100 µmol m⁻² s⁻¹ red light. For each line in each condition, n = 36. Error bars represent 95% confidence intervals. Statistical significance (P < 0.05) was tested with a one-way anova. *, Significant difference between single mutants and wild type or a significant difference between double mutants and single mutants grown in the same conditions. Representative seedlings and cotyledons were grown as described in (a) on media containing lincomycin (b) or no inhibitor of chloroplast biogenesis (c). Bar, 2 mm.

Analysis of epidermal cells in gun1 and cry1 mutants

The light-driven expansion of the cotyledons is largely a consequence of cell expansion rather than cell division (Neff & van Volkenburgh, 1994; Stoynova-Bakalova et al., 2004). To test whether plastid signals might affect cell expansion in the cotyledons, we examined cotyledon surfaces using SEM. The epidermal cells of the cotyledons from untreated wild-type and mutant seedlings all exhibited a similar pattern of interdigitating lobes that are typical of epidermal pavement cells (Fig. 5a). In fact, these pavement cells were comparable in area, shape and stomatal density (Fig. S5a–c). By contrast, when wild-type seedlings were treated with inhibitors of chloroplast biogenesis, the pavement cells lacked the interdigitated lobes of untreated seedlings but had the same surface area and stomatal density as untreated wild type (Figs 5b, S5a–c). The pavement cells of treated gun1-1 and cry1 were half the size of treated wild type and the density of stomata was strikingly increased in treated gun1-1 relative to the wild type (Figs 5b, S5). By contrast to the treated single mutants and wild type, pavement cell area, shape and density of stomata were very similar in treated gun1-1 cry1 and untreated seedlings (Figs 5b, S5a–c). These data indicate that when chloroplast biogenesis is blocked, the development of abnormal epidermal cells is dependent on GUN1 and cry1.

Figure 5. Analysis of epidermal cells in the cotyledons of gun1 and cry1 mutants of Arabidopsis thaliana. (a) Analysis of the epidermis from the untreated cotyledons of wild type (Col-0), gun1 and cry1 mutants. The indicated lines were grown on media that did not contain an inhibitor of chloroplast biogenesis in 50 µmol m⁻² s⁻¹ blue light. The epidermis of the cotyledons was analysed by scanning electron microscopy (SEM). Micrographs were obtained from the center of 10–15 cotyledons for each line. Representative images are shown. (b) Analysis of the epidermis from the cotyledons of lincomycin-treated wild type (Col-0), gun1 and cry1 mutants. The lines indicated were grown on media that contained lincomycin in the same light conditions described in (a). The epidermis was analysed as in (a). Representative micrographs are shown. Bar, 100 µm.

Analysis of anthocyanin levels in gun1 and cry1 mutants

Anthocyanins are induced by photoreceptors and stress. These pigments function in part to protect the chloroplast from high-fluence-rate light-induced photo-oxidative stress and photoinhibition (Gould 2004). Because high-fluence-rate light induces both reactive oxygen species (ROS)-dependent (Nott et al., 2006; Kim et al., 2008) and GUN1-dependent plastid signals (Koussevitzky et al., 2007), we tested whether GUN1-dependent plastid signals also contribute to anthocyanin biosynthesis. We measured anthocyanin levels in seedlings grown in 50 µmol m⁻² s⁻¹ blue light to learn whether plastid signals affect the important role of cry1 and HY5 in the induction of anthocyanin biosynthesis (Ahmad et al., 1995; Jackson & Jenkins, 1995; Shin et al., 2007; Vandenbussche et al., 2007). We found that gun1-1 and gun1-101 contained fourfold less anthocyanin than wild type, but only when chloroplast biogenesis was blocked, and that cry1 and hy5 contained 1.5- to 2.0-fold less anthocyanin than wild type, regardless of whether chloroplast biogenesis was blocked (Fig. 6a,b). Additionally, anthocyanin levels were 1.5- and 3.0-fold lower in the gun1-1 cry1 and gun1-101 hy5 double mutants, respectively, than in the single mutants, but only when chloroplast biogenesis was blocked (Fig. 6a,b). These data indicate that GUN1-dependent plastid signals do not induce anthocyanin biosynthesis by affecting cry1 or HY5 activity and that GUN1-dependent plastid signals have a greater effect on the induction of anthocyanin biosynthesis than cry1 and HY5 when chloroplast biogenesis is blocked in blue light.

Figure 6. Analysis of anthocyanin levels in lincomycin-treated Arabidopsis thaliana seedlings. (a) Analysis of anthocyanin levels in lincomycin-treated and untreated gun1 and cry1 mutants in blue light. Seedlings were grown in 50 µmol m⁻² s⁻¹ blue light in the presence of lincomycin (light tinted bars) or the absence of an inhibitor of chloroplast biogenesis (dark tinted bars). For each line in each condition, n > 4 and there were 20 seedlings per replicate. Error bars represent 95% confidence intervals. (b) Analysis of anthocyanin levels in lincomycin-treated and untreated gun1 and hy5 mutants in blue light. Seedlings were grown and analysed as in (a). Error bars are as described in (a). Statistical significance (P < 0.05) was tested with a one-way anova. *, Significant difference between single mutants and wild type or a significant difference between double mutants and single mutants grown in the same conditions.

We also tested whether plastid signals induce anthocyanins in high-intensity white light because, as stated above, high-intensity light triggers plastid signaling (Nott et al., 2006; Koussevitzky et al., 2007; Kim et al., 2008), plastid signals contribute to successful greening in high-fluence-rate light (Ruckle et al., 2007), and anthocyanin-deficient plants exhibit enhanced sensitivity to photo-oxidative stress (Gould, 2004). We allowed seedlings to germinate in the dark for 23 h and then transferred them to 1000 µmol m⁻² s⁻¹ continuous white light for 6 d. Under these conditions, gun1, cry1, hy5 and the corresponding double mutants are chlorophyll deficient relative to the wild type (Ruckle et al., 2007). We found that anthocyanin levels were indistinguishable in gun1 mutants and wild type, but anthocyanins accumulated to 3- and 50-fold lower concentrations in cry1 and hy5, respectively, compared with wild type. Similarly, anthocyanins accumulate to 30- and 500-fold lower amounts in the gun1 cry1 and gun1 hy5 double mutants, respectively, than in wild type and anthocyanin levels were fivefold lower in the gun1-101 cry1 double mutant than in gun1-1 cry1 (Fig. 7a,b). These data indicate that although cry1 and HY5 are major regulators of anthocyanin biosynthesis in 1000 µmol m⁻² s⁻¹ white light, GUN1 can contribute to anthocyanin biosynthesis when seedlings de-etiolate under these conditions and either cry1 or HY5 is absent.

Figure 7. Analysis of anthocyanin levels during de-etiolation in bright white light. (a) Analysis of anthocyanin levels in gun1 and cry1 mutants during de-etiolation in bright white light. Arabidopsis thaliana seedlings were grown in continuous 1000 µmol m⁻² s⁻¹ white light with no inhibitor of chloroplast biogenesis. For each line in each condition, n > 4 and there were 20 seedlings per replicate. Error bars represent 95% confidence intervals. (b) Analysis of anthocyanin levels in gun1 and hy5 mutants during de-etiolation in continuous bright white light. Seedlings were grown at a different time than the seedlings in (a) and were analysed as described in (a). Error bars are as described in (a). Statistical significance (P < 0.05) was tested with a one-way anova. *, Significant difference between single mutants and wild type or a significant difference between double mutants and single mutants grown in the same conditions.

Analysis of hypocotyl elongation in gun1 mutants

To test whether the effects of plastid signals on photomorphogenesis are restricted to photosynthetic organs or whether they can also affect the growth of nonphotosynthetic organs, we tested the effect of plastid signals on light-regulated elongation of the hypocotyl. Both cry1 and HY5 contribute to the light-dependent inhibition of hypocotyl elongation in untreated seedlings; for cry1, these effects are more striking in blue light (Koornneef et al., 1980; Chory, 1992; Ahmad & Cashmore, 1993; Ang & Deng, 1994). In blue light, gun1 mutants and wild type have similar hypocotyl lengths. However, when seedlings are treated with lincomycin, gun1-101 did have 1.3- to 1.4-fold shorter hypocotyls than wild type in blue light (Figs 3b, 8a). Both cry1 and hy5 have two- to four-fold longer hypocotyls than gun1 and wild type, regardless of whether chloroplast biogenesis is blocked (Figs 3b, 8a). In white light, the hypocotyls of untreated cry1, gun1 and wild type are essentially the same length, and the hypocotyls of untreated hy5 are 1.7-fold longer than cry1, gun1, and wild type. However, when treated with inhibitors of chloroplast biogenesis, cry1 and hy5 have 1.6- to 4.3-fold longer hypocotyls than treated wild type (Figs 8b, S6a). These data indicate that cry1 and HY5 are negative regulators of hypocotyl elongation, regardless of whether seedlings are treated with inhibitors of chloroplast biogenesis, and that cry1 and HY5 have more important roles in regulating hypocotyl elongation when seedlings are treated with inhibitors of chloroplast biogenesis than in untreated seedlings in white light. In the dark, all of the hypocotyl lengths were the same, whether or not seedlings were treated with inhibitors of chloroplast biogenesis (M. E. Ruckle and R. M. Larkin, unpublished). We found that hypocotyls in untreated cry1 hy5 are 1.5 and 4 times as long as hypocotyls in corresponding untreated single mutants in blue and white light, respectively (Figs 3b, 8a,b, S6a). These findings indicate that cry1 and HY5 can only partially compensate for each other in the inhibition of hypocotyl elongation under these conditions.

Figure 8. Analysis of hypocotyl lengths in gun1, cry1 and hy5 mutants of Arabidopsis thaliana. (a) Analysis of hypocotyl length in gun1, cry1 and hy5 mutants grown in blue light. For each line, seedlings were grown in the dark or in 50 µmol m⁻² s⁻¹ blue light in the presence of lincomycin (medium tinted bars) or erythromycin (light tinted bars), or in the absence of an inhibitor of chloroplast biogenesis (dark tinted bars). For each line in each condition, n > 26. For all lines, hypocotyl lengths in a particular line in a particular condition are presented as a percentage of the average hypocotyl length of that line grown on the same media in the dark. Error bars represent 95% confidence intervals. (b) Analysis of hypocotyl length in gun1, cry1 and hy5 mutants grown in white light. For each line, seedlings were grown in the dark or in 125 µmol m⁻² s⁻¹ white light in the presence of lincomycin (medium tinted bars) or erythromycin (light tinted bars), or in the absence of an inhibitor of chloroplast biogenesis (dark gray bars). For each line in each condition, n > 27. Analysis of hypocotyl lengths and error bars are as described in (a). (c) Analysis of hypocotyl length in gun1 and hy5 mutants grown in high-fluence-rate white light. For each line, seedlings were grown in the dark and in 1000 µmol m⁻² s⁻¹ white light with no inhibitor of chloroplast biogenesis. For each line, n > 22. Analysis of hypocotyl lengths and error bars are as described in (a). Statistical significance (P < 0.05) was tested with a one-way ANOVA. *, Significant difference between single mutants and wild type or a significant difference between double mutants and single mutants grown in the same conditions.

The enhanced hypocotyl lengths of treated cry1 and hy5 seedlings in white light lead us to suggest that plastid signals can stimulate hypocotyl elongation. To test this possibility, we inhibited plastid-to-nucleus signaling in cry1 and hy5 backgrounds by preparing gun1-101 cry1 and gun1-101 hy5 double mutants. We observed that hypocotyls were 2- to 3-fold shorter in gun1-101 hy5 relative to hy5 in both light conditions and 1.4-fold shorter in gun1-101 cry1 relative to cry1 in blue light (Figs 3b, 8a,b, S6a). To test whether GUN1 can stimulate hypocotyl elongation under other conditions that trigger GUN1-dependent plastid signaling, we measured the hypocotyl lengths of seedlings grown in continuous high-fluence-rate white light. We observed that high-fluence-rate white light stimulated hypocotyl elongation fourfold in hy5 relative to wild type and gun1-101 and that the longer hypocotyls of hy5 were suppressed in the hy5 gun1-101 double mutant (Figs 8c and S6b).

Discussion

Plastid signals repress the photomorphogenesis of the cotyledons

The cryptochromes are primarily responsible for photomorphogenesis in blue light (Lin & Shalitin, 2003; Ohgishi et al., 2004). Because we found that GUN1-dependent plastid signals inhibit cotyledon opening in 5 µmol m⁻² s⁻¹ blue light and in lower fluence rates of blue light, we suggest that plastid signals likely act downstream of the light signaling network that promotes cotyledon opening in these fluence rates. Cotyledon opening was independent of plastid signals when the fluence rate of blue light was increased. Increased photoreceptor activity in high-fluence-rate blue light (Lin et al., 1998; Neff & Chory, 1998; Ohgishi et al., 2004) may explain these data.

In untreated seedlings, cry1 is a positive regulator of cotyledon expansion in blue light (Jackson & Jenkins, 1995; Neff & Chory, 1998; Ohgishi et al., 2004), but the contribution of cry1 to cotyledon expansion is lower when the fluence rate of blue light is increased (Jackson & Jenkins, 1995). Our analysis of cotyledon areas in cry1, hy5, and cry1 hy5 indicates that cry1 is a positive regulator of cotyledon expansion in 50 µmol m⁻² s⁻¹ blue light and that HY5 can compensate for a loss of cry1 activity in this fluence rate of blue light. Consistent with this interpretation, other photoreceptors have been reported to contribute to cotyledon expansion in high-fluence-rate blue light (Neff & Chory, 1998; Ohgishi et al., 2004) and HY5 appears to act downstream of multiple photoreceptors (Jiao et al., 2007). A similar model likely explains our finding that cry1 hy5 has longer hypocotyls than the corresponding single mutants.

We also found that when chloroplast biogenesis is blocked, cry1 contributes to the repression rather than the promotion of cotyledon expansion (Fig. 9a). These data indicate that the conversion of cry1 from a positive to a negative regulator by plastid signals likely has an impact on the expression of nuclear genes that affect cotyledon expansion and not only on Lhcb genes, as has been previously shown (Ruckle et al., 2007). Plastid signals convert cry1 from a positive to a negative regulator of Lhcb genes by a mechanism that involves converting HY5, which acts downstream of cry1, from a positive to negative regulator of Lhcb genes (Ruckle et al., 2007). Because in contrast to Lhcb, the conversion of cry1 from a positive to a negative regulator of cotyledon expansion does not appear to involve HY5, plastid signals must regulate cotyledon expansion by a mechanism that does not require HY5 or at least also utilizes other factors that can compensate for a loss of HY5 activity. In animal systems, nuclear receptors and Myc can function as positive or negative regulators of transcription depending on which coregulators or transcription factors they bind (Wanzel et al., 2003; Feige & Auwerx, 2007; Kassel & Herrlich, 2007).

Figure 9. Models for the regulation of photomorphogenesis by plastid signals. (a) Model for the regulation of cotyledon expansion by plastid signals. A plastid signal that requires GUN1 for its biosynthesis or transduction (X) (dotted arrows) exits the plastid and triggers an extraplastidic signaling pathway that ultimately represses photosynthesis-associated nuclear genes (PhANGs) expression (Koussevitzky et al., 2007). GUN1-dependent plastid signals also repress photomorphogenesis of the cotyledons. A plastid signal that is genetically distinct from the GUN1-dependent plastid signals (W) converts cry1 from a positive to a negative regulator of cotyledon expansion in high-fluence-rate blue light, as discussed in the text. The same signal may be involved in blocking pavement cell differentiation in the cotyledons. (b) Model for the regulation of anthocyanin biosynthesis by plastid signals. GUN1 and cry1 protect plastids from stress induced by high-intensity light (HL) during de-etiolation (Ruckle et al., 2007). Photoactivated cry1 induces anthocyanin biosynthesis (Ahmad et al., 1995). GUN1, cry1 and HY5 induce anthocyanin biosynthesis, but the GUN1 pathway only appears to be an important regulator of anthocyanin biosynthesis when chloroplasts are stressed or damaged by HL or inhibitors of chloroplast biogenesis (Inhibitors), as discussed in the text. HY5 is not shown but probably acts downstream of cry1 (Jiao et al., 2007). (c,d) Models for the regulation of hypocotyl elongation by GUN1-dependent plastid signals during chloroplast dysfunction. (c) Repression of an inhibitor of hypocotyl elongation by GUN1-dependent plastid signals. In this model, Y is a factor that inhibits hypocotyl elongation by a mechanism that does not involve HY5. When chloroplasts experience dysfunction, GUN1-dependent plastid signals inhibit Y, and cry1 induces Y. (d) Stimulation of an inducer of hypocotyl elongation by GUN1-dependent plastid signals. In this model, Z is a factor that induces hypocotyl elongation and acts downstream of HY5. When chloroplasts experience dysfunction, GUN1-dependent plastid signals stimulate factor Z.

Hormones were previously reported to cause abnormal curling of leaves and cotyledons (Keller & van Volkenburgh, 1997; Hamant et al., 2002; Kakiuchi et al., 2007). Because we found that the cotyledons of wild type but none of the mutants curled downward when chloroplast was blocked in 50 µmol m⁻² s⁻¹ blue light but not in lower fluence rates of blue light, we suggest that plastid signals might affect hormone biosynthesis or responses in 50 µmol m⁻² s⁻¹ blue light. In addition, because we found that the cotyledons of gun1-1 were larger than wild type in 100 µmol m⁻² s⁻¹ red light but not 50 µmol m⁻² s⁻¹ blue light, we conclude that unlike blue light, red light probably does not repress cotyledon expansion when chloroplast biogenesis is blocked. We cannot rule out the possibility that red light might contribute to the inhibition of cotyledon expansion when combined with a distinct light quality, such as blue light. Consistent with this idea, interactions between light-signaling pathways that are triggered by a combination of blue and red light have been reported previously (Casal, 2006).

We found that the pavement cells in the epidermis of cotyledons from wild-type seedlings treated with inhibitors of chloroplast biogenesis did not exhibit a jigsaw-puzzle appearance that is typical of untreated seedlings (Smith, 2003; Panteris & Galatis, 2005). The finding that, in contrast to wild type and the single mutants, the epidermis of treated gun1 cry1 bears a striking resemblance to the epidermis of untreated seedlings suggests that at least two plastid signals, one dependent on GUN1 and the other transduced by cry1, are part of a signaling network that can regulate cell differentiation in the epidermis (Fig. 9a). A role for plastids in the development of the epidermis has been noted previously. For example, similar abnormal epidermal cells have been observed in the pale cress (pac) mutant of Arabidopsis and the defective chloroplasts and leaves-mutable (dcl-m) mutant of tomato, which are both deficient in chloroplast-localized proteins that participate in RNA processing (Reiter et al., 1994; Keddie et al., 1996; Meurer et al., 1998; Tirlapur et al., 1999; Bellaoui et al., 2003).

We expected that differences in pavement cell area (Fig. S5a) would explain gun1 cry1 having larger cotyledons than the single mutants (Fig. 3). However, the observation that the cotyledons of treated wild type have pavement cells of approximately the same size as gun1 cry1 (Fig. S5a) but smaller cotyledons than treated gun1 cry1 (Fig. 3) would seem paradoxical. One possible explanation for these data might lie in the rough appearance of the epidermis in treated wild type (Fig. 5b). Indeed, from a light microscopy analysis of cotyledon sections, we found that the epidermis of treated wild type is extensively invaginated relative to the treated mutants (Fig. S7a,b). These data indicate that measurements of cotyledon area provide a less accurate estimate of cotyledon surface area in treated wild type than in these treated mutants.

One or more GUN1-dependent plastid signals affect anthocyanin biosynthesis and elongation of the hypocotyl

Both cry1 and HY5 had been reported to be important positive regulators of anthocyanin biosynthesis, but whether plastid signals affect anthocyanin biosynthesis has not been previously reported. We induced GUN1-dependent plastid signaling by treating seedlings with either lincomycin or high-fluence-rate white light. Lincomycin provides a complete and uniform block to greening among wild type and all of the mutants tested. By contrast, high-fluence-rate white light inhibits greening most in the double mutants tested, less in the single mutants tested, and least in the wild type (Ruckle et al., 2007). We observed that accumulation of anthocyanins is dependent on GUN1 only when chloroplast biogenesis is severely impaired by lincomycin treatments or by high-fluence-rate light treatments of gun1 cry1 and gun1 hy5 double mutants (Fig. 9b). These findings are consistent with GUN1-dependent plastid signals inducing anthocyanin biosynthesis when plastids experience dysfunction to protect plastids from excess light and to scavenge reactive oxygen species. An alternative interpretation, that gun1 mutants contain fewer anthocyanins than wild type because they experience less plastid dysfunction than wild type during these treatments, is inconsistent with previous reports, which indicate that gun1 mutants experience greater or equal plastid dysfunction compared with wild type (Susek et al., 1993; Mochizuki et al., 1996; Koussevitzky et al., 2007; Ruckle et al., 2007).

Our findings also suggest that GUN1-dependent plastid signals can contribute to the elongation of hypocotyls and that cry1 and HY5 contribute to the repression of hypocotyl elongation regardless of the functional status of the plastid. The finding that plastid signals remodel cry1 signaling that affects Lhcb expression (Ruckle et al., 2007) and cotyledon expansion but not hypocotyl elongation is consistent with the distinct cotyledon and hypocotyl transcriptome responses to light (Ma et al., 2005) and the composition of downstream components of light-signaling pathways varying among distinct organs (Jiao et al., 2007). The plastid signals that affect hypocotyl elongation may originate from plastids in the hypocotyls or in another organ, such as the cotyledon. Lending support to this idea, Tanaka et al. (2002) reported that photoreceptors in cotyledons affect processes in hypocotyls.

The suppression of the long hypocotyl phenotype of hy5 in gun1 hy5 is consistent with two models. (1) Hypocotyls are driven to elongate and hypocotyl elongation is inhibited by a light-signaling network regardless of the functional state of the plastid. In this model, GUN1-dependent plastid signaling, which is triggered by both inhibitors of chloroplast biogenesis and high-fluence-rate white light, stimulate hypocotyl elongation not only in hy5 but also in wild type. In wild type, HY5 would dominate the system, thereby obscuring the stimulatory effect of GUN1 and resulting in an overall reduced rate of hypocotyl elongation. (2) During inhibitor and high light treatments, gun1 hy5 mutants may experience greater plastid dysfunction than hy5, to the extent of being unable to support hypocotyl elongation, for example because of abnormal metabolism. Although the second model is difficult to completely rule out, the first model is more consistent with the available data. For example, we observed that cry1, phyB and hy5 exhibit long hypocotyl phenotypes regardless of whether they are treated with inhibitors of chloroplast biogenesis that induce severe plastid dysfunction. The hypocotyls of cry1 and hy5 treated with lincomycin and erythromycin were longer than untreated controls in white light. Also, alleles of hy1 and hy2, which are deficient in phytochrome signaling, also exhibit elongated hypocotyls when seedlings are photobleached with norflurazon (Mochizuki et al., 2001). Therefore, hypocotyls are not necessarily less driven to elongate when plastids experience dysfunction. Our finding that in the dark, the hypocotyl lengths are the same in wild type and all mutants regardless of whether seedlings are treated with lincomycin indicates that these effects of GUN1-dependent plastid signals on hypocotyl length depend on light.

Connections between plastid function and hypocotyl elongation have been reported previously. For example, phytochrome interacting factor 3 (PIF3) promotes both chloroplast biogenesis (Monte et al., 2004) and hypocotyl elongation in continuous red light (Bauer et al., 2003; Kim et al., 2003; Monte et al., 2004). In addition, hypocotyl lengths are reduced in the chlorophyll-deficient cue mutants (Vinti et al., 2005) but enhanced in laf6 (Møller et al., 2001). The suppression of the long hypocotyl phenotype of hy5 in gun1 hy5 and these other reports are consistent with chloroplast biogenesis and hypocotyl elongation interacting by a complex mechanism. Although a mechanism is not clear at present, we suggest that the current data are consistent with GUN1-dependent plastid signals stimulating hypocotyl elongation by: inhibiting a factor (Y) that represses hypocotyl elongation and acts downstream of cry1 and in a separate pathway from HY5 (Fig. 9c); or by acting downstream of HY5 and stimulating a factor (Z), which induces hypocotyl elongation and is repressed by HY5 (Fig. 9d).

Acknowledgements

We thank Andrea Stavoe, Chris Sinkler, and Stephanie DeMarco for helpful assistance. We thank Ewa Danielewicz and Alicia Pastor at the Center for Advanced Microscopy (Michigan State University) for technical assistance with SEM and light microscopy experiments, Joanne Chory for providing gun1-1 seeds, and Beronda Montgomery-Kaguri and Lyle Burgoon for helpful discussions. This work was supported by DOE grant no. DE-FG02-91ER20021 and NSF grant no. IOB 0517841 to R.M.L.