DELLA protein function in growth responses to canopy signals

Authors


(fax +31 30 2518366; e-mail r.pierik@uu.nl).

Summary

Plants can sense neighbour competitors through light-quality signals and respond with shade-avoidance responses. These include increased shoot elongation, which enhances light capture and thus competitive power. Such plant–plant interactions therefore profoundly affect plant development in crowded populations. Shade-avoidance responses are tightly coordinated by interactions between light signals and hormones, with essential roles for the phytochrome B photoreceptor [sensing the red:far red (R:FR) ratio] and the hormone gibberellin (GA). The family of growth-suppressing DELLA proteins are targets for GA signalling and are proposed to integrate signals from other hormones. However, the importance of these regulators has not been studied in the ecologically relevant, complex realm of plant canopies. Here we show that DELLA abundance is regulated during growth responses to neighbours in dense Arabidopsis stands. This occurs in a R:FR-dependent manner in petioles, depends on GA, and matches the induction kinetics of petiole elongation. Similar interactions were observed in the growth response of seedling hypocotyls and are general for a second canopy signal, reduced blue light. Enhanced DELLA stability in the gai mutant inhibits shade-avoidance responses, indicating that DELLA proteins constrain shade-avoidance. However, using multiple DELLA knockout mutants, we show that the observed DELLA breakdown is not sufficient to induce shade-avoidance in petioles, but plays a more central role in hypocotyls. These data provide novel information on the regulation of shade-avoidance under ecologically important conditions, defining the importance of DELLA proteins and GA and unravelling the existence of GA- and DELLA-independent mechanisms.

Introduction

Dense natural vegetation is characterized by strong interactions between plants. A well-studied example is light-mediated neighbour detection, which induces shade-avoidance responses during competition for light. These responses include enhanced shoot elongation and upward tilting and extension of leaves, allowing plants to reach the light. The prime trigger for these responses is the reduced ratio of red to far-red radiation (R:FR), sensed by the phytochrome family of photoreceptors (particularly PHYB) and resulting from the selective absorbance of red light by neighbouring plants (Franklin and Whitelam, 2005; Smith and Whitelam, 1997). Blue-light photon fluence rates, sensed by blue-light receptors (cryptochromes, phototropins), are also reduced in canopies, and this has also been shown to induce shade-avoidance features (Ballaré, 1999; Pierik et al., 2004b). These light signals interact with various hormones to control cell elongation, which drives the elongation responses to neighbours (Halliday and Fankhauser, 2003; Vandenbussche et al., 2005). Such multiple interactions require proper integration. A group of growth-suppressing proteins, called DELLA proteins, have been identified as gibberellin signal transduction components (Alvey and Harberd, 2005; Dill and Sun, 2001; Silverstone et al., 2001), and were recently shown to integrate the action of several plant hormones (Achard et al., 2003, 2006; Fu and Harberd, 2003). However, these growth regulators have not been investigated to date in the ecologically important situation of plant competition with neighbours and the related growth responses to neighbour-derived canopy signals.

Gibberellins (GA) are a key growth factor in plants, and GA is essential for shade-avoidance (Halliday and Fankhauser, 2003; López-Juez et al., 1995; Pierik et al., 2004a). In Arabidopsis, low-R:FR treatment appears to enhance both GA biosynthesis (Hisamatsu et al., 2005) and responsiveness (Reed et al., 1996). Thus, a low R:FR ratio promotes the expression of GA-related genes (Devlin et al., 2003), and the constitutively elongated phenotype of phytochrome mutants (phyb and hy1) is suppressed by GA deficiency and GA insensitivity (Peng and Harberd, 1997).

Downstream elements in GA signalling are the DELLA family of growth-repressing proteins (Alvey and Harberd, 2005; Dill and Sun, 2001; Silverstone et al., 2001). The DELLA family in Arabidopsis consists of five members: GAI (GA-insensitive), RGA (repressor of ga1) and RGA-like (RGL1, RGL2, RGL3; Fleet and Sun, 2005). The DELLA proteins are localized in the nucleus where they suppress the expression of GA-responsive genes. In the presence of GA, however, DELLA proteins are targeted for breakdown (Alvey and Harberd, 2005). This was recently shown to occur by binding of GA to its receptor (GID1 in rice and GID1a, GID1b and GID1c in Arabidopsis), which then interacts with an SCF E3 ubiquitin ligase complex to allow ubiquitination and subsequent DELLA breakdown (Nakajima et al., 2006; Ueguchi-Tanaka et al., 2005). Interestingly, DELLA protein stability and GA-mediated DELLA breakdown are affected by several other signals, such as the hormones auxin and ethylene (Achard et al., 2003), which are both essential for shade-avoidance (Morelli and Ruberti, 2000; Pierik et al., 2004b). This has led to the understanding that DELLA proteins may act as molecular integrators of various growth-regulating signalling routes (Alvey and Harberd, 2005).

In the present study, we used DELLA knockout and gain-of-function mutants and a GFP reporter for the DELLA protein RGA (GFP–RGA) to study GA and DELLA involvement in shade-avoidance. We show that RGA protein abundance is regulated during plant competition in dense Arabidopsis stands. Low R:FR-induced petiole elongation required the presence of GA, and the kinetics of DELLA breakdown matched the induction kinetics of enhanced petiole elongation. These interactions with DELLA occur in seedling hypocotyls as well as petioles of more mature rosettes, and occur in response to two different canopy light signals (low R:FR and low blue light). Enhanced DELLA stability leads to reduced shade-avoidance responses, indicating the importance of DELLA breakdown in allowing these growth responses to occur. However, these shade-avoidance responses can also be induced in DELLA knockout mutants. This indicates that DELLA proteins play a permissive role in R:FR responses, but that important signal transduction processes that act to regulate shade-avoidance, at least in leaf petioles, are GA- and DELLA-independent. Taken together, our data provide novel insights into the regulation of plant growth responses to the ecologically important light signals derived from competing neighbours.

Results

Neighbours induce petiole elongation and DELLA breakdown

Growth responses to neighbours were studied in dense Arabidopsis canopies (10 000 plants per m2). After approximately 18 days of canopy development, enhanced petiole elongation was visible in crowded plants (Figure 1a), which was accompanied by upward movement of leaves (Figure 1c). This confirms that Arabidopsis shows essential shade-avoidance traits when competing for light with con-specific neighbours (Figure 1e). Shade-avoidance responses are triggered by a reduced R:FR ratio, but also by reduced blue-light levels (Ballaré, 1999; Pierik et al., 2004b), and both signals were present in the Arabidopsis canopies (Figure 1b). Confocal imaging of canopy-grown pRGA:GFP–RGA plants showed that the abundance of the DELLA protein RGA decreased in parallel with increased petiole elongation (Figure 1d), which probably indicates enhanced GA action. To more precisely identify the mechanism underpinning these responses at the whole-plant level in canopies, we investigated the regulation of elongation responses of isolated plants to both low-R:FR and low-blue-light signals

Figure 1.

 Shade-avoidance in Arabidopsis canopies (10 000 plants per m2).
(a) Petiole elongation.
(b) Spectral composition of horizontally propagated light at t = 21 days for individually grown plants versus plants in a canopy (10 versus 0.1 μmol m−2 sec−1 blue light, 1.95 versus 0.13 R:FR ratio, and 64 versus 2 μmol m−2 sec−1 photosynthetically active radiation).
(c) Cross-section through a canopy at t = 24 days with the double front row of edge plants taken away.
(d) DELLA abundance indicated by the GFP–RGA signal of pRGA:GFPRGA petioles (red fluorescence shows chloroplasts).
(e) Isolated plant (top) and canopy-grown plant (bottom), the latter showing the classic shade-avoidance phenotype with enhanced petiole length and reduced leaf lamina investment.

DELLA proteins in low-R:FR-induced petiole elongation

Pronounced petiole elongation responses occurred in the wild-type, Ler, during 24 h low-R:FR treatments (Figure 2a). However, no growth stimulation was observed upon low-blue-light treatment for 24 h (data not shown). The response to low R:FR was diminished by GA deficiency in Ler plants treated with the GA biosynthesis inhibitor paclobutrazol and in the GA-deficient ga1-3 mutant (Figure 2a). Subsequent GA addition restored the response to low R:FR and control growth rates, whereas GA addition to normal wild-type Ler plants did not affect petiole elongation (Figure 2a). This suggests saturation by endogenous GA under control light conditions, and thus that enhanced GA biosynthesis only would not be sufficient to induce shade-avoidance. GA presence and function are thus essential for low-R:FR-induced petiole elongation in Arabidopsis. In agreement with this, phytochrome-mediated shade-avoidance is reduced in the gai mutant, which has reduced GA sensitivity due to enhanced stability of the DELLA protein gai (Figure 2b) (Peng and Harberd, 1997). Although this mutant displays constitutive suppression of petiole elongation under both light conditions, its response to low R:FR is reduced to approximately half of that in Ler (55% growth stimulation in Ler versus 29% stimulation in gai). This was also confirmed by a dedicated two-way ANOVA on these two genotypes which identified a significant (= 0.03) genotype by environment interaction. The fact that the R:FR response was not totally absent in this gai mutant probably indicates the action of other DELLA proteins, such as RGA. Thus, enhanced DELLA stability inhibits low-R:FR-induced petiole elongation.

Figure 2.

 Petiole responses to 24 h low-R:FR treatment.
(a) GA deficiency inhibits petiole elongation and its response to low R:FR, and can be rescued by GA.
(b) The DELLA gain-of-function gai mutant shows a reduced response to low R:FR, whereas the response in various DELLA knockouts is unaltered. wt, wild-type; g.o.f., DELLA gain-of-function mutant; K.O., DELLA knockout mutants.
(c) DELLA knockouts can counteract and overcome the lack of response in GA-deficient plants.
(d) The RGA–GFP signal from pRGA:GFPRGA petioles is reduced in low R:FR; this is prevented by GA deficiency (PAC, paclobutrazol) and rescued by additional GA3.
(a–c) Data are means and SE (n = 10). No overlapping letters indicate a significant difference (< 0.05).

Next, we studied whether (i) DELLA breakdown is a major regulatory step in the signalling cascade that directly induces the processes that induce shade-avoidance, or (ii) DELLA proteins are a constraint on shade-avoidance that must be degraded, but without this breakdown itself further inducing the shade-avoidance response. To study the roles of individual DELLA proteins, knockouts of four DELLA proteins (gait6, rga24, rgl1-1, rgl2-1) were investigated. Hypothesis 1 predicts constitutively high elongation in the knockouts with little or no response to canopy signals, whereas, in hypothesis 2, DELLA knockouts are not necessarily elongated and can still normally respond to the canopy signals. Figure 2(b) shows that these single and multiple DELLA knockouts do not show constitutively elongated petioles, indicating that low DELLA abundance alone does not lead to shade-avoidance. Furthermore, these DELLA knockouts show normal petiole elongation responses to low R:FR (Figure 2b). DELLA knockouts thus have very mild phenotypes under conditions with normal GA content (Figure 2b) (King et al., 2001). As DELLA proteins are highly abundant under GA-deficient conditions, where they are not targeted by GA for breakdown, GA-deficient plants show strong suppression of growth and shade-avoidance (i.e. paclobutrazol treatment and ga1-3 mutant, Figure 2a). Interestingly, we show here that DELLA knockouts can rescue the far-red response in GA-deficient plants (Figure 2c), as the full low-R:FR response occurred in the GA-deficient ga1-3 gait6 rga24 triple mutant. This indicates that GAI and RGA are the major DELLA proteins that inhibit low-R:FR-induced petiole elongation when present at sufficiently high levels. Accordingly, the rgl1-1 and rgl2-1 knockouts could not restore the response in ga1-3 plants, and the paclobutrazol-treated quadruple gait6 rga24 rgl1 rgl2 knockout mutant behaved similarly to gait6 rga24 (Figure 2c). These DELLA interactions are GA-specific as illustrated by the fact that RGA abundance is much reduced upon low-R:FR treatment and this reduction in abundance is prevented by GA deficiency (paclobutrazol treatment) and restored by adding back GA (Figure 2d). In conclusion, DELLA proteins are broken down in a GA-dependent manner under low R:FR (as indicated by regulation of the RGA–GFP signal), and this is important for shade-avoidance (indicated by the reduced response in gai). However, DELLA breakdown alone is certainly not sufficient for enhanced elongation growth of petioles, as DELLA knockouts do not have constitutively shade-avoiding petioles. This is strongly supported by the dramatic response to low R:FR in petioles of the multiple DELLA knockouts under GA-deficient conditions (Figure 2d).

Timing of RGA breakdown matches that of low-R:FR-induced petiole elongation

We studied the kinetics of low-R:FR-induced petiole elongation and RGA abundance to determine whether the timing of DELLA breakdown matches the elongation response. Petiole elongation rates increased approximately 2 h after the start of low-R:FR treatment, reached a maximum after 3 h, and remained approximately constant until at least 6 h (Figure 3a). In accordance with this, the RGA protein disappeared after 2 h and remained undetectable during this same growth phase (Figure 3b). These data confirm that the kinetics of enhanced petiole elongation and RGA breakdown are similar, supporting the notion that DELLA proteins are negative regulators of shade-avoidance and thus are broken down to allow enhanced petiole elongation to occur.

Figure 3.

 Kinetics of petiole elongation rates (a) (mean ± SE, n = 6) and GFP–RGA signal (b) of pRGA:GFPRGA petioles under normal and low-R:FR conditions.

DELLA regulation during low-R:FR-induced hypocotyl elongation

Not only fully grown plants but also very young seedlings can be exposed to neighbour detection signals. We therefore tested hypocotyl responses to low R:FR to investigate whether DELLA involvement in shade-avoidance is general throughout plant development. The elongation response to low R:FR was absent in GA-deficient seedlings, but could be rescued by adding back GA (Figure 4a). Low-R:FR-induced hypocotyl elongation was absent in the GA-insensitive gai mutant (Figure 4b), which had wild-type hypocotyl lengths under control light conditions. Enhanced DELLA stability thus did not affect hypocotyl length under control light conditions but did prevent low-R:FR-induced hypocotyl elongation, indicating that the reduced length of gai under low R:FR does not result from reduced general growth, but rather from a reduced response to low R:FR. Accordingly, low R:FR led to disappearance of the RGA–GFP signal, which was prevented by paclobutrazol treatment and could be restored by adding back GA (Figure 4d). Losing both DELLA proteins RGA and GAI restored control hypocotyl elongation in GA-deficient seedlings to wild-type levels under control light conditions, as described previously (King et al., 2001). The hypocotyl response to low R:FR, however, was only partly restored by multiple DELLA knockouts in a GA-deficient background (Figure 4c). In contrast to petioles, hypocotyls of the quadruple DELLA knockout were constitutively elongated, accounting for approximately half of the low-R:FR-induced hypocotyl elongation response (Figure 4b). DELLA proteins may thus not only constrain low-R:FR-induced hypocotyl elongation, but, unlike petioles, DELLA absence in hypocotyls may even induce part of the shade-avoidance response. However, this cannot account for the full response to low R:FR.

Figure 4.

 Hypocotyl responses to 5 days of low-R:FR treatment.
(a) GA deficiency inhibits the hypocotyl elongation response to low R:FR and can be rescued by GA.
(b) The DELLA gain-of-function gai mutant shows a reduced response to low R:FR, whereas the response in various DELLA knockouts persists. wt, wild-type; g.o.f., DELLA gain-of-function mutant; K.O., DELLA knockout mutants.
(c) DELLA knockouts can counteract and overcome the stunted hypocotyl phenotype of GA-deficient seedlings.
(d) The RGA–GFP signal from pRGA:GFPRGA hypocotyls is reduced under low R:FR; this is prevented by GA deficiency (PAC, paclobutrazol) and rescued by additional GA3.
(a–c) Data are means and SE (n = 30-50). No overlapping letters indicate a significant difference (P < 0.05).

Low blue-light photon fluence rates induce strong hypocotyl elongation in a GA-dependent manner

As reduced blue-light photon fluence rates are another neighbour detection signal in canopies, we tested whether the R:FR data can be reproduced for blue light. The response of hypocotyl elongation to low blue light was much stronger than the response to low R:FR but was similarly GA-dependent, as GA deficiency almost completely abolished the response (Figure 5a). As for low R:FR, low-blue-light-induced elongation could be rescued in GA-deficient plants by adding back GA (Figure 5a). Enhanced DELLA stability in the gai mutant did not affect control growth, but did strongly suppress the hypocotyl elongation response to low blue light (Figure 5b). This is very similar to the low-R:FR-induced elongation response. DELLA knockouts could rescue low-blue-light-induced shade-avoidance in GA-deficient seedlings (Figure 5c). This confirms that the involvement of GA in shade-avoidance is through its regulation of DELLA proteins. In accordance with this, the low-blue-light-induced disappearance of the RGA–GFP signal was prevented by inhibition of GA biosynthesis with paclobutrazol (Figure 5d). The elongated hypocotyls of quadruple DELLA knockouts under control light conditions (Figure 5b) suggests that DELLA reduction in hypocotyls may induce downstream processes leading to shade-avoidance. However, as for low R:FR, this cannot account for the entire low-blue-light-induced hypocotyl elongation response, indicating that other processes must also be operating to induce the strong elongation response to low blue light.

Figure 5.

 Hypocotyl responses to 5 days of low-blue-light treatment.
(a) GA deficiency inhibits the hypocotyl elongation response to low R:FR and can be rescued by GA.
(b) The DELLA gain of function gai mutant shows a reduced response to low blue light, whereas the response in various DELLA knockouts is unaltered. wt, wild-type; g.o.f., DELLA gain-of-function mutant; K.O., DELLA knockout mutants.
(c) DELLA knockouts can counteract and overcome the stunted hypocotyl phenotype of GA-deficient seedlings and rescue the response to low blue light.
(d) The RGA–GFP signal from pRGA:GFPRGA hypocotyls is reduced under low blue light; this is prevented by GA deficiency (PAC, paclobutrazol) and rescued by additional GA3.
(a–c) Data are means and SE (n = 30-50). No overlapping letters indicate a significant difference (< 0.05).

Discussion

DELLA proteins are essential elements in GA signal transduction and integrate signals from other signalling routes as well (Alvey and Harberd, 2005). Here we show that enhanced elongation during plant–plant interactions in dense canopies is accompanied by breakdown of the DELLA protein RGA. The induction of increased petiole elongation of isolated plants under low R:FR was also paralleled by RGA disappearance. Low R:FR strongly stimulated petiole elongation within 2 h, whereas low blue light, a more recently identified neighbour detection cue, did not induce a substantial increase of petiole elongation within 24 h. This suggests that neighbour detection in Arabidopsis canopies occurs primarily through phytochrome-mediated perception of the R:FR ratio. However, shade-avoidance responses also occur in seedling hypocotyls, and at this developmental stage low blue light was very effective, even more so than low R:FR. The relative contributions of the different canopy light signals thus depend on the plant developmental stage.

GA regulation of low-R:FR-induced elongation in petioles and hypocotyls

GA deficiency strongly reduced growth responses to low R:FR, confirming earlier work in cucumber (López-Juez et al., 1995), Arabidopsis (Peng and Harberd, 1997) and tobacco (Pierik et al., 2004a). Adding GA to plants with normal endogenous GA did not induce petiole elongation under any light conditions, suggesting saturation by the endogenous GA levels. This rules out the possibility that changes in GA biosynthesis account for the enhanced growth under low R:FR, and implies the existence of a separate growth control mechanism. Most probably this would involve enhanced GA responsiveness, which might for example be regulated at the level of DELLA protein stability (King et al., 2001), leading to enhanced DELLA breakdown under low R:FR. Tentatively, a similar effect could be obtained through enhanced GA receptor (GID1) abundance, which was recently shown to enhance internode length in rice (Ueguchi-Tanaka et al., 2005). We therefore determined whether the R:FR ratio controls the expression of the genes encoding the GID1 GA receptor family in Arabidopsis (Nakajima et al., 2006), but we found no effect on any of the three Arabidopsis GA receptor genes (GID1a, GID1b and GID1c) in a 24 h time series (data not shown). Likewise, expression of the RGA and GAI genes appeared not to be affected by the R:FR ratio. Alternatively, low R:FR might enhance GA responsiveness downstream of the GA receptors and DELLA proteins, for example at the level of the GAMYB transcription factors (Achard et al., 2004).

In contrast to petioles, GA addition to hypocotyls did give a length increase comparable to that for low-R:FR treatment, suggesting that a low-R:FR-induced increase of GA biosynthesis (Hisamatsu et al., 2005) may be sufficient for enhanced hypocotyl elongation. However, it is unlikely that regulation occurs only at the level of GA biosynthesis, as there is still an elongation response to low R:FR, albeit reduced, in plants with GA levels that are experimentally fixed (paclobutrazol-treated plants with GA added back) and thus not modulated by the light signal.

Towards a working model for interaction of GA and DELLA proteins in shade-avoidance

Loss-of-function mutations in the DELLA loci counteract many aspects of GA deficiency. The gait6 and rga24 mutations resulted in partial resistance to GA deficiency of petiole and hypocotyl responses to low R:FR and low blue light, respectively. GA-deficient plants even showed wild-type elongation responses when both these DELLA proteins were knocked out, the exception being the low R:FR response in hypocotyls, which could only be partly restored by the gait6 rga24 double knockout. These data confirm that the lack of shade-avoidance under GA-deficient conditions is caused by the lack of DELLA breakdown by GA. When this DELLA restraint is relieved, in this case by DELLA knockouts, GA deficiency can be overcome. This indicates that the relief of restraint model for DELLA action in GA signal transduction (Harberd, 2003) also applies to an environmental signalling context, such as light-mediated shade-avoidance. This is further illustrated by the finding that GA deficiency prevents the elongation response as well as the light-induced breakdown of the DELLA protein RGA, and both can be rescued by adding back GA to GA-deficient plants. Furthermore, the lack of DELLA breakdown in the gai mutant, carrying a deletion in the DELLA motif thereby making the GAI protein unrecognizable for the GA signal, prevents the elongation responses. Thus DELLA breakdown occurs during light-mediated shade-avoidance, and this seems to be functionally important for the elongation responses to occur. However, although regulated DELLA breakdown allows for shade-avoidance, this relief of restraint is in itself not sufficient to induce shade-avoidance responses in petioles of mature plants. This is indicated by the fact that the petioles of DELLA knockouts are not constitutively elongated. Seedlings of the quadruple DELLA knockout, on the other hand, do show an elongated hypocotyl phenotype, albeit not to the full extent of low-blue-light or even low-R:FR treatment. This may suggest that, in seedling hypocotyls, low DELLA abundance can induce part of the shade-avoidance response. It may be hypothesized that the varying importance of DELLA proteins in mediating the growth responses of these two organs could be related to the much shorter treatment duration in petiole experiments (1 day) compared to those on hypocotyls (5 days). Therefore, we also tested low-R:FR-induced petiole elongation over a 3 day period in Ler and the quadruple DELLA knockout, and this gave identical results to the standard 1 day treatments, ruling out the possibility that treatment duration could cause the differences between these organs. In short, DELLA proteins impose a restraint on elongation growth responses to canopy light signals, and this restraint is relieved by a light-signal-induced increase in GA action and subsequent DELLA breakdown. In addition, part of the hypocotyl elongation response may also be regulated directly by DELLA breakdown itself, as indicated by the constitutively elongated hypocotyls in the quadruple DELLA knockout.

The often close to wild-type shade-avoidance responses to low R:FR and low blue light in the combined GA-deficient DELLA knockout mutants indicate the existence of GA- and DELLA-independent routes for shade-avoidance. Thus the environmental signal, i.e. low R:FR or low blue light, in addition to inducing GA-mediated DELLA degradation, activates additional signalling route(s) that stimulate cell elongation. It will be an interesting future challenge to elucidate whether other hormones involved in shade-avoidance, such as ethylene and auxin, have DELLA-dependent as well as DELLA-independent modes of action to control shade-avoidance responses. Elucidating putative DELLA-independent modes of action of these hormones could provide important data for understanding DELLA-independent routes for photomorphogenic responses, such as shade-avoidance.

Taken together, we show here that regulation of DELLA abundance does not only occur at the seedling stage, but also in mature plants, and this can even be extrapolated to the complex realm of interacting plants in dense canopies. The elegant mechanism of growth regulation by DELLA proteins had not previously been studied in such a complex, yet ecologically relevant, environment. The data obtained through this novel approach confirm that DELLA proteins are important to permit growth responses to neighbours, but also show that GA-mediated regulation of DELLA proteins only is not sufficient to induce petiole elongation upon exposure to low R:FR. Rather, DELLA breakdown is probably required to allow for shade-avoidance to occur, but does not mediate the growth enhancement in petioles. In hypocotyls, DELLA breakdown itself may already mediate part of the elongation response to neighbour cues (low blue light and low R:FR ratio). However, in hypocotyls, GA- and DELLA-independent modes of regulation are also required to explain the observed shade-avoidance responses. By studying growth regulation and DELLA proteins under ecologically relevant conditions, we have thus provided novel insights into the extent to which shade-avoidance is regulated by DELLA and GA, and have unravelled the existence of alternative pathways for growth regulation.

Experimental procedures

Plant material and growth

All mutant and transgenic lines used in this study are in the Arabidopsis thaliana (L.) Landsberg erecta (Ler) background. GA involvement in shade-avoidance was investigated using the GA-deficient ga1-3 (Sun et al., 1992) and GA-insensitive gai (Koornneef et al., 1985) mutants. The roles of four of the five Arabidopsis DELLA proteins were studied using the gait6 (Peng and Harberd, 1993), rga24 (Dill and Sun, 2001), rgl1-1 (Wen and Chang, 2002) and rgl2-1 (Lee et al., 2002) single mutants, and the rga24 gait6 (Dill and Sun, 2001) double and rga24 gait6 rgl1-1 rgl2-1 (Achard et al., 2006) quadruple knockout lines in the Ler background. These DELLA knockouts were also studied in the GA-deficient (ga1-3 background) mutants rga24 ga1-3 (Silverstone et al., 1997), gait6 ga1-3, gait6 rga24 ga1-3 (Dill and Sun, 2001), rgl1 ga1-3 and rgl2 ga1-3 (Lee et al., 2002). Abundance of the DELLA protein RGA was studied in transgenic plants (pRGA:GFP–RGA) expressing the GFPRGA fusion protein. Seeds were generously donated by Dr N.P. Harberd, John Innes Centre, Norwich, UK (gai, gait6, rga24, gait6 rga24, rga24 gait6 rgl1-1 rgl2-1 and pRGA:GFP–RGA), Dr J. Peng, Institute of Molecular and Cell Biology, Singapore (rgl1-1, rgl2-1, rgl1-1 ga1-3 and rgl2-1 ga1-3) and Dr T.P. Sun, Duke University, Durham, NC, USA (gait6 ga1-3, rga24 ga1-3, gait6 rga24 ga1-3).

For petiole elongation studies, seeds were sown on filter paper soaked with water or a 100 μm solution of GA3 (GA-deficient lines), stratified for 4 days at 4°C, and then germinated for 4 days in a growth chamber under standard growth conditions of 20°C, 9 h light (200 μmol m−2 sec−1 photosynthetically active radiation, Philips Master HPI 400 W)/15 h dark (http://www.lighting.philips.com). Seedlings were then transferred to pots [70 ml with one individual for light and GA treatments, and 280 ml with 49 individuals (7 × 7 plants in a square pattern, 1 cm distance between plants) for canopy studies], with a 1:2 potting soil/perlite substrate mixture and additional nutrients (Millenaar et al., 2005). Plants were used for light and GA experiments 35 days after sowing. The resulting densities were 250 plants per m2 for individually potted plants and 10 000 plants per m2 for canopy-grown plants. For hypocotyl experiments, surface-sterilized seeds were sown on agar (5.5 g l−1) with low nutrients (0.22 g l−1 MS), stratified, and then transferred to a growth chamber. After 3 days, seedlings were transferred to the various light-quality compartments.

Treatments and growth measurements

Canopy studies were performed under standard growth conditions, and the central nine plants from each canopy were measured (the other plants minimized edge effects). Light quality was measured using a Licor 1800 spectroradiometer (Licor; http://www.licor.com) with a remote cosine receptor attached. The receptor was held horizontally at approximately 1 cm above the soil in the canopies to record the quality of light reflected from neighbouring plants.

GA biosynthesis was inhibited by adding 20 ml of 50 μm paclobutrazol solution to the pots 7 days prior to the light treatments, or by adding paclobutrazol (1 μm final concentration) to the agar plates. Paclobutrazol gave typical GA deficiency symptoms, such as dwarfing and dark-green leaves. This could be rescued by GA treatments, which took place by daily spraying with 50 μm GA3 solution (controls were sprayed with water) or by adding GA3 to the plates (10 μm final GA concentration). Light-quality manipulations took place in a white light background (Philips Master HPI-T Plus 400 W and Philips Plusline Pro 150 W). The R:FR ratio was reduced from 1.2 to 0.25 by supplemental far-red light (730 nm LED; Shinkoh Electronics Co. Ltd; http://www.shinkohelecs.com). Blue-light photon fluence rates (400-500 nm) were reduced from 26 to < 1 μmol m−2 sec−1 using two layers of Lee 010 medium yellow filter (Lee Filters; http://www.leefilters.com). Photosynthetically active radiation for all light treatments was maintained at 140 μmol m−2 sec−1. One day before the experiment, plants were transferred to the relevant growth cabinet for acclimatization. Light treatments started on the subsequent day at 10 am and lasted 24 h. Petiole lengths were measured using a digital caliper at t = 0 and 24 h to calculate growth as the length increment over 24 h. Petiole elongation rates were also determined by linear displacement transducers to determine petiole elongation kinetics under normal and low-R:FR conditions (Benschop et al., 2005).

For hypocotyl studies, seedlings were transferred to the light treatments 7 days after sowing. Hypocotyl lengths were measured after 5 days of light treatment using a custom-built image analysis system [Sony CCD camera (http://www.sony.com) (type XC-77CE) combined with KS400 software (Carl Zeiss Vision; http://www.zeiss.com/)]. Two-way analysis of variance (ANOVA), with Tukey B post hoc comparisons (SPSS version 12.0.1; SPSS Inc., Chicago, IL, USA), revealed whether there were significant differences between genotypes and treatments. Experiments were independently repeated at least three times.

DELLA abundance

The abundance of the DELLA protein RGA was studied using the pRGA:GFP–RGA reporter line (Achard et al., 2003, 2006; Silverstone et al., 2001). Fluorescence was detected using an inverted confocal laser scanning microscope (Zeiss LSM Pascal, 40× C-apochromat objective). The excitation wavelength was 488 nm, a 505–530 nm bandpath filter was used for GFP emission, and a 560 nm long-pass filter was used to visualize red fluorescence by chloroplasts. Images depict 149.5 μm Z-stacks of petioles and hypocotyls. The GFP–RGA fluorescence of petioles and hypocotyls was determined at the same time points as the length measurements. The basal, middle and uppermost regions of hypocotyls and petioles were studied, but only images from the basal region (abaxial in petioles with the mid-rib horizontally approximately halfway up the image) are shown as this is where most of the cell elongation occurs (Gendreau et al., 1997; Kozuka et al., 2005). Confocal images are representative selections from at least nine replicates of at least three independent experiments. It was recently shown by Western blotting with an anti-GFP antibody that the RGA–GFP signal visualized through confocal imaging gives a reliable estimation of the abundance of this fusion protein (Achard et al., 2006).

Acknowledgements

We thank Dr A.J.M. Peeters for helpful comments and discussion, and F. Kindt and M. Terlou for technical assistance with microscopic imaging. This research was funded by the Netherlands Organisation for Scientific Research (PIONIER grant number 80074470 to L.A.C.J.V. and VENI grant number 86306001 to R.P.) and NUFFIC (T.D.-P.).

Ancillary