Plants that are adapted to environments where light is abundant are especially sensitive to competition for light from neighboring vegetation. As a result, these plants initiate a series of changes known as the shade avoidance syndrome, during which plants elongate their stems and petioles at the expense of leaf development. Although the developmental outcomes of exposure to prolonged shade are known, the signaling dynamics during the initial exposure of seedlings to shade is less well studied. Here, we report the development of a new software-based tool, called HyDE (Hypocotyl Determining Engine) to measure hypocotyl lengths of time-resolved image stacks of Arabidopsis wild-type and mutant seedlings. We show that Arabidopsis grows rapidly in response to the shade stimulus, with measurable growth after just 45 min shade exposure. Similar to other mustard species, this growth response occurs in multiple distinct phases, including two phases of rapid growth and one phase of slower growth. Using mutants affected in shade avoidance phenotypes, we demonstrate that most of this early growth requires new auxin biosynthesis via the indole-3-pyruvate pathway. When activity of this pathway is reduced, the first phase of elongation growth is absent, and this is correlated with reduced activity of auxin-regulated genes. Finally, we show that varying shade intensity and duration can affect the shape and magnitude of the growth response, indicating a broad range of the elongation response to shade.
Plants need light to survive and are often in competition with other plants for photosynthetically active wavelengths of light (Franklin and Quail, 2010). As such, plants have evolved sophisticated photoreceptors that are capable of sensing the presence of neighbors by monitoring the ratio of red light (R) (which chlorophyll absorbs as part of the photosynthetically active light spectrum) to far-red light (FR) (which chlorophyll does not absorb, and which is thus reflected and transmitted through leaves) (Kasperbauer, 1987; Ballaréet al., 1990). A high R:FR ratio indicates a sparsely populated environment and an abundance of photosynthetically active radiation. Conversely, a low R:FR ratio (<1) indicates the presence of nearby vegetative neighbors that may soon compete for available light. This low R:FR ratio initiates a suite of responses, termed the shade avoidance syndrome (SAS), that are of physiological and agricultural importance. The SAS has been extensively studied in the reference plant Arabidopsis thaliana (Smith and Whitelam, 1997; Franklin, 2008). Plants grown under shade have a decreased germination rate, increased hypocotyl and petiole elongation, inhibition of both leaf expansion and root elongation, reduced chlorophyll content, a tendency to flower early, reduced fecundity, and an increased susceptibility to herbivory (Smith and Whitelam, 1997; Izaguirre et al., 2006).
The Arabidopsis hypocotyl is an excellent model for studying shade phenotypes due to its simple structure, sensitivity to light, small cell number (approximately 20 cells in each cell file), and reliance on cell expansion versus cell division for elongation growth (Gendreau et al., 1997; Chen et al., 2004). Seedlings that are foraging for light have long hypocotyls, while those growing under bright light (R:FR > 1) have shorter hypocotyls. Hypocotyl length is inversely proportional to the fluence rate of white light. End-point assays quantifying shade avoidance phenotypes are typically performed by measuring hypocotyl length after several days of growth in light supplemented with far-red radiation (e.g. Salter et al., 2003; Sessa et al., 2005; Lorrain et al., 2008). Various studies under these conditions have enabled development of a tentative model whereby the photoreceptors phytochromes B, D and E, perceive the shift in the ratio of R:FR light (Child and Smith, 1987; Smith, 2000). This allows accumulation of at least two phytochrome-interacting basic helix-loop-helix (bHLH) transcription factors, PIF4 (PHYTOCHROME INTERACTING FACTOR 4) and PIF5 (Huq and Quail, 2002; Khanna et al., 2004; Lorrain et al., 2008), which promote growth by modulating gene expression (Lorrain et al., 2008). Of the three phytochromes in the pathway, phyB has the strongest role, while phyD and phyE have relatively minor roles (Franklin et al., 2003). A third bHLH protein, HFR1 (LONG HYPOCOTYL IN FAR-RED LIGHT), whose transcript is rapidly induced upon exposure to shade, constitutes part of a negative feedback mechanism by binding to PIF4 and PIF5, and inhibiting their DNA-binding (and thus growth-promoting) activity (Fairchild et al., 2000; Duek and Fankhauser, 2003; Sessa et al., 2005; Hornitschek et al., 2009). Seedlings with mutations in the photoreceptor phyA showed exaggerated hypocotyl elongation upon transfer to shade after 2 days of growth in continuous white light (Johnson et al., 1994), consistent with phyA signaling through HFR1 (Fairchild et al., 2000). However, the role of phyA in shade avoidance is unclear, as older seedlings do not display this phenotype (Z. Zheng and J. Chory, unpublished data).
New auxin biosynthesis is necessary for shade-induced hypocotyl elongation, as indicated by the discovery of an auxin biosynthesis gene, SHADE AVOIDANCE 3 (SAV3), which encodes the tryptophan aminotransferase TAA1 (TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS 1), which converts tryptophan to indole-3-pyruvic acid. Exposure to shade induces auxin biosynthesis in wild-type plants (producing up to 50% more auxin within 1 h), but not in plants lacking functional SAV3, in which shade avoidance phenotypes are severely reduced (Stepanova et al., 2008; Tao et al., 2008). Treatment with NPA (1-naphthylphylamic acid), a polar auxin transport inhibitor, also attenuates shade-induced hypocotyl elongation (Steindler et al., 1999). Polar auxin transport has thus been shown to mobilize auxin from the leaves to sink tissues, such as the hypocotyl and petiole, consistent with preferential expression of SAV3 in leaf tissue (Tao et al., 2008). Finally, gibberellic acid (GA) promotes the shade response by inducing degradation of DELLA-containing proteins (a five-member family of growth repressors) that can bind to and inhibit PIF transcription factors (Djakovic-Petrovic et al., 2007; Feng et al., 2008; de Lucas et al., 2008). Thus, the shade avoidance response provides an informative platform to study the intersection of light and various hormone signaling pathways as they relate to plant physiology, growth regulation and environmental adaptation (Vandenbussche et al., 2005; Jaillais and Chory, 2010).
Although some of the components that drive the shade avoidance response have been identified, a gap remains between the time at which physiological phenotypes (several days after exposure) and molecular events (usually within 1–4 h) are studied. Physical methods have been used to quantify growth in Vigna sinensis epicotyls (Garcia-Martinez et al., 1987) and Sinapis alba stems (Morgan et al., 1980; Child and Smith, 1987; Casal and Smith, 1989), and have shown that a rapid multi-phasic growth response can occur upon exposure to shade. Similar physical techniques have been used to study etiolated growth in Arabidopsis seedlings undergoing photomorphogenesis in response to monochromatic light (Parks and Spalding, 1999), and revealed subtle differences in growth patterns among photoreceptor mutants. While informative, these studies relied on invasive strategies to record elongation rates, which are difficult (but not impossible) to perform on delicate and morphometrically dynamic Arabidopsis seedlings (Folta and Spalding, 2001; Miller et al., 2007), for which a wealth of experimental tools exist. Recently, non-invasive imaging and feature-detection technologies have enabled the study of real-time growth dynamics in etiolated seedlings (Miller et al., 2007; Wang et al., 2008, 2009), representing a great improvement over previous methods. However, automated image analysis of dark-grown seedlings is different from that of light-grown seedlings, as light-grown petioles can grow independently of the hypocotyl, and existing software exploits features of the petiole/cotyledon junction to analyze dark-grown seedlings (Wang et al., 2009). Thus, new imaging approaches are required for imaging of light-grown seedlings.
Our goal was to assess hypocotyl elongation over short time scales by developing an image-based phenotyping platform. Here, we describe HyDE (Hypocotyl Determining Engine), a new software tool for quantifying hypocotyl length on time-resolved image stacks of single light-grown seedlings. We used this tool to reveal multi-phasic growth patterns of Arabidopsis seedlings in response to shade, and to study the short-term effect of mutations in the shade avoidance response pathway. We defined distinct periods of hypocotyl elongation after exposure to shade, and correlated these periods with accumulation of mRNA for known shade marker genes. Finally, we showed that the magnitude and shape of the shade response could be dynamically altered by adjusting the light regime. As such, our results provide a kinetic framework to further characterize early events and identify novel components of the shade avoidance perception and response pathways.
Results and Discussion
HyDE software can image light-grown hypocotyl dynamics in an automated assay
To establish correlations between shade-regulated molecular events and the phenotypes they are thought to control, we developed an assay that is capable of measuring physiological phenotypes over the same short time scales used to study molecular phenotypes. Our new assay adapts the Phytomorph CCD-camera based imaging system developed by Miller et al. (2007) to image light-grown Arabidopsis hypocotyls in a three-channel LED chamber, simulating high- and low-R:FR light conditions (Figure S1a). To avoid interference from the far-red light used to simulate shade, we utilized an IR filter with a cut-off at 790 nm, slightly longer than used previously (720 nm; Miller et al., 2007), but still able to detect the 880 nm backlight source. To accompany the imaging set-up, we devised a software tool, HyDE, to automatically measure the length of hypocotyls in a time-resolved image series. The software first converts a cropped raw seedling image (Figure S1b) to a binary (foreground/background) format (Figure S1c), and calculates a Euclidean distance transform (EDT; Figure S1d). It identifies local maxima within the EDT, indicating the centroid points of major organs (e.g. cotyledons and the shoot apical meristem). HyDE then constructs a digital hypocotyl based on the center line of the imaged hypocotyl, using the bottom horizontal pixel slice as a reference to determine the horizontal position of the hypocotyl. The hypocotyl should be the only structure present on the lowest portion of the image. HyDE terminates the digital hypocotyl where it intersects with the first local maximum corresponding to the shoot apical meristem, which forms a bulge where the petioles initiate (Figure S1e). This hypocotyl center line is then spline-smoothed, and its length is recorded for all images in the stack (Figure S1f). Our method works best on young (4- to 7-day-old) seedlings, as they lack fully emerged true leaves, which can obscure morphological information at the hypocotyl–petiole junction. Although we have not tested our software on older seedlings, we expect that this 4–7 day window represents the most informative time to capture dynamic information, as older Arabidopsis hypocotyls are less responsive to growth stimuli. Similarly, selected seedlings that grew such that the cotyledons extended perpendicular (±15°) to the plane of imaging, so they would not interfere with hypocotyl length measurements. The software is available for download from http://cactus.salk.edu/hyde.html.
We tested the image-processing algorithm on 5-day-old long-day entrained (16 h fluorescent white light/8 h dark) wild-type Arabidopsis accession Columbia (Col-0) seedlings to assay for growth when released (at zeitgeber time ZT12) into continuous light conditions (LED-simulated daylight). Under these conditions, the hypocotyls of wild-type seedlings did not elongate during the majority of the subjective dark period, instead exhibiting a major growth peak during the middle of the day, beginning at ZT2 and ending at ZT12 (Figure S2). This is consistent with published circadian and diurnal growth data (Nozue et al., 2007). We conclude from these experiments that the HyDE software is capable of measuring hypocotyl length over very short time scales, being able to detect significant length differences over periods as short as 10 min, a resolution that is useful for understanding the growth kinetics of shade avoidance responses.
Arabidopsis thaliana seedlings respond to shade within 1 h, exhibiting a multi-phasic growth pattern that varies with the R:FR ratio
To assay for hypocotyl-specific growth during the shade avoidance response, we manipulated the far-red light intensity to mimic the very earliest ‘shade’ conditions that plants encounter: the threat of shade (reflected FR light) from neighboring vegetation. Based on field studies of soybeans (Glycine max), which indicated that reductions in R:FR may be more prevalent at dusk (Kasperbauer, 1987), we decided to treat the seedlings towards the end of the subjective day under diurnal conditions. Five-day-old Arabidopsis seedlings (grown under fluorescent lights under long-day conditions) were transferred to the imaging platform at ZT12, under high R:FR conditions (R:FR = 2.37). The seedlings were imaged for 2 h to establish a baseline growth pattern, before being challenged with a marked increase in supplementary FR light, decreasing the R:FR ratio to 0.23. After the start of treatment, we imaged seedlings for a further 9.5 h, and hypocotyl lengths were measured using HyDE (Figure 1a). Wild-type seedlings exhibited a lag period of about 45 min after exposure, growing at a basal rate of about 0.1 μm min−1, before rapidly elongating at a rate of 0.45 μm min−1 until approximately 150 min after exposure. The hypocotyl growth then slowed to a rate of 0.2 μm min−1, until about 230 min after exposure, when it accelerated to approximately 0.55 μm min−1 for the remainder of the experiment.
Based on the shape of the curve, we defined four phases of growth under shade: a lag phase (from approximately 30 min before exposure to 45 min after exposure), an initial growth phase (45–150 min after exposure), a slowdown phase (150–230 min after exposure), and a second growth phase (from 230 min after exposure to the end of the assay). These phases are similar to those observed in Sinapis alba primary stems and first internodes (Morgan et al., 1980; Child and Smith, 1987; Casal and Smith, 1988); however, the observed lag time in Arabidopsis is significantly longer reported previously for Sinapis. Additionally, although Sinapis alba grows more quickly in the first phase than the second, Arabidopsis hypocotyls grew at a slightly faster rate during the second growth phase than the first. The results from cowpea (Vigna unguiculata) indicated rapid initiation of growth within 60 min after exposure to supplementary FR, with no reduction in the growth rate until 10 h after the start of the experiment (Garcia-Martinez et al., 1987). These differences in kinetics may reflect the differences in species, tissues or experimental set-up.
To explore whether the severity of shade has a strong effect on growth kinetics, we treated wild-type seedlings as above, but decreased the R:FR to 0.65 (mild shade), 1.2 or 1.6 (non-shade), comparing growth curves from these seedlings to those that were not treated or had the R:FR lowered to 0.23 (full shade). The elongation response was very sensitive, with a measurable increase in the growth rate even during the two least intense treatments (R:FR = 1.6 and 1.2, values that were previously assumed to be at or above the threshold for eliciting a shade avoidance response; Smith, 2000) (Figure 1). Mild shade treatment results in slower growth in both elongation phases (with that in phase IV being slower than that in phase II) when compared to that in full shade, in which seedlings grew faster in phase IV than in phase II (Figure 1). Overall, there was an inverse correlation between the severity of the shade treatment and the magnitude of the elongation responses. A slowdown period was observed for all treatments, during which seedlings treated with 0.23, 0.65 and 1.2 R:FR light grew at approximately the same rate (0.2 μm min−1), but those treated with 1.6 R:FR grew at a slower rate (approximately 0.09 μm min−1), suggesting that this slowdown period does not require a certain amount of growth to occur in order to be observed. Our sensitive hypocotyl measurement method suggests that either the threshold R:FR ratio that elicits a shade response in Arabidopsis hypocotyls is higher than previously thought, or that plants are sensitive to the change in the ratio rather than the absolute ratio itself. For further experiments, we decided to use the most severe treatment (R:FR 0.23), as it yielded the most dramatic results.
The time of treatment has a minimal effect on the quality and magnitude of the response within an 8 h window
Although growth in the absence of shade treatment appeared to be negligible (Figures S2 and S3), we wished to assess whether treating seedlings at various times of the day would affect the magnitude or quality of the resulting growth curve. We subjected wild-type seedlings to the same pre-treatment conditions as before, but applied supplemental FR light 2, 4, or 6 h later than our other assays (at ZT16, ZT18 and ZT20, respectively; Figure 2). The quality of the growth response was nearly identical to that for seedlings treated at ZT14: the seedlings had a sharp initial increase in growth rate approximately 45 min after treatment, which slowed after approximately 150 min, followed by growth at a faster rate after approximately 230 min (Figure 2a). The similarity of these responses is particularly apparent when the data are superimposed such that the subjective time of treatment is the same (0 min, Figure 2b). Because we did not keep the length of shade exposure constant for each assay, we cannot say that growth during phase IV for the seedlings treated at ZT20 is identical to that exhibited by seedlings treated at ZT14, but we observed that the lag period (phase I) is identically timed, the slope and duration of the first growth phase (phase II) are similar, and that these seedlings have a similar slowdown at the same time (Figure 2). Thus, it appears that, if the time of treatment has a strong gating role in regulating hypocotyl growth in the shade, it is not immediately apparent during the 10 h period for which seedlings were treated (at the end of the day), although we cannot rule out the possibility of long-term temporal effects occurring after 12 h of treatment. Our data suggest that factors that affect the qualitative parameters of the shade growth response (e.g. the slowing observed in phase III) are likely to be initiated by the shade treatment itself. Thus, we saw little role for the circadian clock in promoting growth of hypocotyls within our time frame, in contrast to previous studies in which a more prominent role for clock regulation was apparent (Salter et al., 2003). However, we did not perform a full-scale circadian clock assay to directly test this hypothesis, and so we cannot contradict previous reports suggesting a more prominent role for the clock (e.g. Salter et al., 2003).
Auxin is necessary for proper initiation of the elongation growth response, and, together with PIF4 and PIF5, regulates its magnitude
We next tested whether mutations of some of the shade avoidance pathway components would alter the quality or magnitude of the shade growth response. Previous studies that quantified hypocotyl responses to shade in plants deficient in phytochrome A suggested a negative role for this photoreceptor on shade-induced growth (Johnson et al., 1994). We thought that at least one if not all growth phases in a phyA mutant would show higher growth rates, and therefore we imaged null phyA-211 seedlings (Nagatani et al., 1993) using our shade assay. As shown in Figure 3, phyA-211 seedlings showed normal timing and magnitude of the initial elongation and slowdown phases. However, although phase IV is initiated at the same time as in wild-type, phyA-211 seedlings do not reach the same final growth rate as wild-type seedlings, resulting in slightly shorter hypocotyls at the end of the experiment. This is inconsistent with studies suggesting a negative effect of phyA action on the shade elongation response (Johnson et al., 1994). Instead, our conditions (using seedlings older than those used by Johnson et al., 1994) highlight the proposed role of phyA in enhancing the response to low R/FR manifested by reduced phyB Pfr (far-red light absorbing form) levels (Casal, 1996), as opposed to the role phyA plays in mediating the high-irradiance response during de-etiolation.
We observed a more dramatic effect in sav3-2 seedlings, which did not respond at all during the first approximately 4 h of the shade response, lacking growth in both phases II and III. They did show a sharp increase in elongation rate towards the end of the assay at approximately the same time that wild-type seedlings began their second elongation phase; however, this growth rate was only approximately 30% of that seen in wild-type seedlings in the same time period. These results suggest that new auxin biosynthesis by the SAV3/TAA1 pathway plays an important qualitative role in initiating the elongation response, and a quantitative role in maintaining high rates of elongation growth after several hours of shade exposure. We also speculate that the lag time observed directly after treatment (phase I) could be due to biosynthesis and transport rates, as the site of auxin synthesis (the cotyledon margins) is distinct from the hypocotyl (Tao et al., 2008). Regulation of TAA1 expression alone cannot account for the phenotypes observed, as its transcript abundance tends to decrease upon exposure to shade (Tao et al., 2008). Recent results suggest that auxin transport in etiolated hypocotyls is linked to phytochrome activity (Nagashima et al., 2008; Wu et al., 2010); however, the mechanism by which light regulates auxin biosynthesis or transport remains unclear.
pif4 pif5 double mutant seedlings (Fujimori et al., 2004; Lorrain et al., 2008) exhibited all growth phases at the proper time, but the magnitude of both growth phases was slightly less than wild-type growth rates, and this defect was more pronounced for phase IV than phase II growth (Figure 3). This mild phenotype suggests that, although the PIF4 and PIF5 transcription factors have a quantitative role in modulating growth, they either act redundantly with other growth regulators, or transcription has only a minor role in mediating the early elongation response to shade.
Gene expression patterns of shade marker genes correlate with phases of growth
To assess whether gene transcription follows a similar multi-phasic pattern, we prepared cDNA from wild-type plants at time points of the imaging assay reflecting each growth phase (0 and 30 min for the pre- and post-induction lag phase, 60 and 120 min for the first elongation phase, 180 min for the slowdown phase, and 360 min for the second elongation growth phase). We performed quantitative real-time PCR assays for PIL1 (Figure 4a), HFR1 (Figure 4b), ATHB-2 (Figure 4c) and IAA29 (Figure 4d) transcripts, which were previously shown to be strongly induced after 1 h of shade exposure (Steindler et al., 1999; Salter et al., 2003; Tao et al., 2008). As expected, PIL1 expression was highly induced by the 30 min time point, and continued to increase until at least 120 min after exposure to shade. We observed a marked decrease in expression level at the 180 min time point, and a strong increase of expression after 360 min, consistent with phases III and IV of our imaging assays (Figures 1 and 4a). The same general pattern was observed for HFR1, although expression was merely maintained at a constant level through phase III compared to the expression peak during phase II, before increasing again towards the end of the assay (Figure 4b). Although PIL1 and HFR1 have both been shown to be negative regulators of the shade growth phenotype (Sessa et al., 2005; Roig-Villanova et al., 2006; Hornitschek et al., 2009), their expression patterns appeared to track growth. It is important to note that although expression levels fluctuated along the time course consistently with growth rates, shade treatment increased RNA levels at all time points assayed.
We also measured the expression pattern of two genes related to auxin signaling, ATHB-2 and IAA29 (Carabelli et al., 1996; Steindler et al., 1999). The mRNA levels for each gene were induced quickly (ATHB-2 after 30 min, IAA29 after 60 min), and continued to be induced to the 120 min time point (phase II). However, both genes showed reduced expression during phase III, and were not re-induced during phase IV. This is consistent with auxin signaling playing a stronger role towards the beginning of the assay, as indicated by the stronger sav3 phenotype during phase II than phase IV (Figure 3). We speculate that the delayed induction response of IAA29 occurs as a consequence rather than a cause of the auxin-induced growth that occurs during the first elongation phase.
Elongation growth is reversible for at least 3 h of shade treatment
To understand the extent of the elongation response that can be elicited by transient shade, we exposed plants to severe shade for 0.5, 3 or 6 h before returning them to high R:FR light conditions (Figure 5). Even though the shade treatment was reversed prior to the expected time of the initial response (phase II), seedlings still exhibited some elongation growth for at least 15 min before slowing their growth rate to phase I levels when exposed to just 30 min of low R:FR (Figure 5). A similar reversal was seen for the 3 h treatment, in which seedlings did not initiate the second phase of the growth response (phase IV), and instead slowed to phase I growth rates until the end of the experiment. Seedlings exposed to only 6 hours of shade began to deviate from seedlings exposed continuously approximately 1 hour after reversion to high R:FR conditions, when growth slowed to a level greater than the phase I growth rates, indicating that some residual growth remains despite the light reversal. This does not appear to be a lag in reversal of growth, as the elevated rate persists for at least 3 h, whereas the seedlings treated for 0.5 or 3 h reverted to basal growth levels within 1 h after the R:FR was increased (Figure 5). This leads us to speculate that either a growth factor (perhaps auxin) remains at high levels after a longer duration of treatment, or that the cells have committed to an elongation program whereby low R:FR enhances but is not necessary for continued elevated growth rates.
In summary, using temporally resolved imaging assays, we have shown that shade-avoiding hypocotyls exhibit multiple phases of elongation growth, controlled by separate, but possibly integrated mechanisms, one of which is new auxin biosynthesis mediated by SAV3. This response is mounted for a range of R:FR ratios, and the memory of the treatment can be maintained if longer durations of shade are applied. The time of treatment did not have a prominent role in promoting or restricting growth under our shade conditions within a 10 h time frame. Rather, it is likely that the mechanism controlling the dynamic shape of the growth curve is initiated by the treatment itself. Investigation into whether the circadian clock has a strong role in controlling the shade avoidance response at various times of the day (e.g. subjective dawn) would be an interesting subject for future research. Initiating multiple layers of control of elongation in response to the shade stimulus may provide the plant with many decision points before commitment to a shade-avoiding lifestyle. An initial burst of growth in the right direction after perception of a transient shade signal may be just enough to enable the plant to continue to intercept an optimal amount of light for photosynthesis, in which case it need not resort to the detrimental responses seen in plants growing in continuous shade. However, if the signal is more persistent, such shade avoidance growth may be the best strategy to provide a competitive advantage under a less ideal environment. It will be interesting to explore further the mechanisms that control each phase of growth and how a decision is made to commit to a shade-avoiding lifestyle.
Plant growth and light conditions
Arabidopsis thaliana seeds were ethanol-sterilized and suspended in 0.1% agar for stratification prior to plating onto half-strength Linsmaier and Skoog (LS) 0.8% agar plates buffered with MES at pH 5.7. Plates were poured using a plastic mold that occluded half of a circular Petri dish, and the stratified seeds were plated on the ledge formed at the interface between the agar and the mold (such that the apical portions of the seedling grow unobstructed in air, while the basal tissues grow into the agar plate). Plates were grown under 16 h/8 h day/night cycles in growth chambers (Percival Scientific, http://www.percival-scientific.com/) supplemented with fluorescent and incandescent light [approximately 37 μmol m−2 sec−1 blue light (400–600 nm), approximately 30 μmol m−2 sec−1 red light (600–700 nm) and approximately 11 μmol m−2 sec−1 far-red light] for 5 days before being transferred to the imaging platform under LED light [approximately 37 μmol m−2 sec−1 blue light (400–600 nm), approximately 30 μmol m−2 sec−1 red light (600–700 nm) and approximately 11 μmol m−2 sec−1 far-red light] at ZT12 on the 5th day. Plants were then exposed to supplemental FR light by increasing the FR light intensity to reduce the R:FR ratio as indicated in each experiment. For RNA experiments, seedlings were grown on Whatman filter paper on top of half-strength LS agar (0.8%) plates under the same light conditions as for the imaging experiments.
The imaging machinery consisted of two CCD cameras, one Marlin and one Guppy (Allied Vision Technologies, http://www.alliedvisiontec.com/), coupled to macro video zoom lenses (Edmund Optics, NT54-363 for the Guppy, http://www.edmundoptics.com/, Qioptics MVZL for the Marlin, http://www.qioptiq.com/) mounted to a rail on a thick aluminum surface (custom-built). Plates are held in sample mounts for circular or square Petri dishes, tightened by a thumb screw. The lenses were fitted with IR long-pass cut filters (NT54-755, Edmund Optics), which allow only wavelengths longer than 790 nm through. Seedling samples were back-lit with an IR LED back-light emitting at 880 nm (Edmund Optics). Image acquisition was achieved via a FireWire connection to a laptop computer, using AVT SmartView software (Allied Vision Technologies). This software allows imaging up to 15 frames per second for any duration of time; however, in the interest of saving disk space and maximizing the number of genotypes and conditions sampled, we saved images every 5 min (every 240th frame with a frame rate of 0.8 frames per second). Image stacks were cropped using ImageJ (http://rsbweb.nih.gov/ij/) after image acquisition but prior to processing with HyDE software, such that only one seedling hypocotyl was visible (and the hypocotyl was cropped from the very base of the image to avoid interference with the seed coat and agar surface). The imaging platform and analysis software are shown in Figure S1.
The HyDE input consisted of the cropped image stacks as described above, and were converted to binary image format using the matlab image processing toolbox (http://www.mathworks.com/), which automatically determined the threshold grayscale intensity value to assign foreground and background pixels. A coarse midline of the hypocotyl was determined by calculating the midpoint between two boundary pixels in successive horizontal pixel slices, ordered by the vertical position in the image. Large deviations between the current and previous horizontal slices (which typically occur within the petiole/cotyleon junction) was used by the algorithm to define a coarse upper boundary below which to look for the termination point of the hypocotyl. The coarse hypocotyl midline was then refined by finding the maximum Euclidean distance transform value (calculated using the MATLAB image processing toolbox) on these bounded midline points, and terminating the hypocotyl at that point (which should correspond to the hypocotyl/petiole junction). The hypocotyl midline was then smoothed using a 4th-order spline (utilizing three segments), and the distance was measured by summation of 0.1 pixel size increments along the length of the spline. This process was repeated for each image in the stack, and data from images that deviated by more than a defined number of pixels from the previous image were automatically rejected. Whole image stacks were removed from analysis if more than 10 consecutive images were rejected.
RNA extraction and quantitative real-time PCR
Seedlings grown under the indicated conditions were frozen in liquid nitrogen. Three biological replicates were collected per plate, approximately 10 seedlings per replicate. RNA was extracted from each seedling pool using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com/) according to the manufacturer’s instructions. cDNA was synthesized from 2 μg RNA using a Maxima cDNA synthesis kit (Fermentas, http://www.fermentas.com/) according to the manufacturer’s instructions. Ten microliters of each 100-fold diluted cDNA sample were used for quantitative real-time PCR, which was performed using SYBR Green/fluorescein dye on a Bio-Rad iCycler (http://www.bio-rad.com/) according to the manufacturer’s instructions. Gene expression was quantified using the method described by Pfaffl (2001). Primer pairs are listed in Table S1.
We thank Dr Edgar Spalding (Department of Botany, University of Wisconsin) for advice in developing the software, and adapting the imaging platform, Dr Zuyu Zheng (Plant Biology Laboratory, Salk Institute) for information concerning phyA shade phenotypes, and Dr Christian Fankhauser (Center for Integrative Genetics, University of Lausanne) for the pif4 pif5 double mutant line. Drs Yvon Jaillais, Ullas Pedmale, Dmitri Nusinow and Colleen Doherty provided critical feedback on this manuscript. This work was supported by grants from the National Institutes of Health (R01GM52413 to J.C. and R01GM56006 to S.A.K) and the National Science Foundation (IOS-0649389 to J.C.). B.C. was supported by an National Science Foundation Integrative Graduate Education and Research Traineeship (0504645). J.C. is an investigator of the Howard Hughes Medical Institute.