https://orcid.org/0000-0002-1429-2089
Abstract
The endogenous circadian (∼24 h) system allows plants to anticipate and adapt to daily environmental changes. Stomatal aperture is one of the many processes under circadian control; stomatal opening and closing occurs under constant conditions, even in the absence of environmental cues. To understand the significance of circadian-mediated anticipation in stomatal opening, we have generated SGC (specifically guard cell) Arabidopsis (Arabidopsis thaliana) plants in which the oscillator gene CIRCADIAN CLOCK ASSOCIATED1 (CCA1) was overexpressed under the control of the guard-cell-specific promoter, GC1. The SGC plants showed a loss of ability to open stomata in anticipation of daily dark-to-light changes and of circadian-mediated stomatal opening in constant light. We observed that under fully watered and mild drought conditions, SGC plants outperform wild type with larger leaf area and biomass. To investigate the molecular basis for circadian control of guard cell aperture, we used large-scale qRT-PCR to compare circadian oscillator gene expression in guard cells compared with the “average” whole-leaf oscillator and examined gene expression and stomatal aperture in several lines of plants with misexpressed CCA1. Our results show that the guard cell oscillator is different from the average plant oscillator. Moreover, the differences in guard cell oscillator function may be important for the correct regulation of photoperiod pathway genes that have previously been reported to control stomatal aperture. We conclude by showing that CONSTANS and FLOWERING LOCUS T, components of the photoperiod pathway that regulate flowering time, also control stomatal aperture in a daylength-dependent manner.
By definition, circadian rhythms are oscillations that persist, even under constant conditions, with a period close to 24 h. In plants, the circadian system regulates processes as diverse as growth, photosynthesis, leaf and petal movements, hormone biosynthesis, shade avoidance, photoprotection, reproductive development, and transcription from approximately one-third of the genome (Greenham and McClung, 2015; Yakir et al., 2007). There is compelling evidence that circadian systems have an important role in regulating metabolism and serve as an adaptation for a rotating world by creating an opportunity for an organism to anticipate and prepare for external environmental changes (Yerushalmi and Green, 2009). Conceptually, a circadian system can be divided into three parts: the oscillator mechanism that generates the rhythms, input pathways by which oscillators are entrained by signals from the environment such as temperature and light changes, and output pathways that convey circadian rhythms to diverse physiological and molecular processes.
The Arabidopsis (Arabidopsis thaliana) oscillator consists of interlocking feedback loops (reviewed in Hsu and Harmer, 2014; Nohales and Kay, 2016). The first loop to be described comprises CIRCADIAN CLOCK ASSOCIATED1 (CCA1), LATE ELONGATED HYPOCOTYL (LHY), TIMING OF CAB EXPRESSION1 (TOC1; also known as PSEUDO-RESPONSE REGULATOR1 [PRR1]), and CCA1 HIKING EXPEDITION (CHE). CCA1 and LHY are homologous myb-related transcription factors that negatively regulate the expression of TOC1; in turn, TOC1 and CHE repress CCA1 and LHY. A second “morning” loop is formed between PRR5, PRR7, and PRR9 and CCA1 and LHY, and a third “evening” loop results from TOC1, EARLY FLOWERING3/4 (ELF3/4), and LUX ARRHYTHMO regulation of PRR7 and PRR9 expression. Positive regulators of the oscillator include members of the REVEILLE family of CCA1 and LHY homologs (Farinas and Mas, 2011; Rawat et al., 2011; Hsu et al., 2013). In addition to transcriptional regulation, there is increasing evidence of other layers of regulation of oscillator components such as chromatin remodeling and protein modification (reviewed in Seo and Mas, 2014). It is likely that almost every cell in the plant has its own oscillator mechanism and the extent to which these oscillators are coupled is being explored (Takahashi et al., 2015; Endo et al., 2014; Fukuda et al., 2007; Yakir et al., 2011; Thain et al., 2000; Wenden et al., 2012). There is also increasing evidence that the model of the Arabidopsis oscillator that has been widely described is an average of the oscillator mechanisms in the whole plant and that different tissues and cells, including stomatal guard cells, may have distinctive variations of the “average” mechanism (Endo et al., 2014; Para et al., 2007; James et al., 2008; Yakir et al., 2011; Takahashi et al., 2015). However, further research is needed to build up a more complete picture of the similarities and differences in oscillator mechanisms in different cells as well as to understand how these variants help fit the oscillator mechanisms to the particular functions of each type of cell.
Stomata allow gas exchange between the interior of the plant and its environment. Opening and closing of the stomata is driven by changes in the turgor pressure of the guard cells. In Arabidopsis and other C3 plants, stomata open early in the morning when ambient temperatures are low. Phosphorylation activates a guard cell plasma membrane H+-ATPase to pump H+ across the membrane, and the resulting hyperpolarization drives an influx of K+ that enters together with Cl− and malate. As the osmotic potential of the solutes inside the cell increases, there is an influx of water that causes a raise in turgor and pore opening. Later in the day, stomata close. The plasma membrane H+-ATPase is inhibited, SLOW ANION CHANNEL1 promotes Cl− efflux, and GUARD CELL OUTWARD RECTIFIER K CHANNEL allows K+ efflux (reviewed in Kim et al., 2010). A multitude of signals regulates stomatal opening, including blue and red light, abscisic acid (ABA), CO2 levels, drought, humidity, and pathogens (Araújo et al., 2011). Guard cell aperture is also regulated by the circadian system (Salomé et al., 2002; Hubbard and Webb, 2011). On an environmental scale, it has been suggested that circadian control of stomata may have large-scale effects on the daily patterns of net ecosystem exchange of CO2 (de Dios et al., 2012; Resco de Dios et al., 2016a). Intriguingly, several key genes in the circadian controlled pathway that regulates photoperiodic flowering have also been shown to affect stomatal aperture. In blue light, the floral integrator FLOWERING LOCUS T (FT) induces stomatal opening by enhancing the phosphorylation status of plasma membrane H+-ATPase (Kinoshita et al., 2011; Kimura et al., 2015; Ando et al., 2013), and GIGANTEA (GI), CONSTANS (CO), and ELF3 regulate FT levels; GI and CO up-regulate FT levels to cause stomatal opening, and ELF3 represses FT, resulting in stomatal closure (Kinoshita et al., 2011; Ando et al., 2013). However, the extent of to which the photoperiodic flowering and guard cell regulatory mechanisms are similar has not been explored.
In this article, we examine the links between the circadian system and stomatal opening. We investigate the significance for plants of the circadian regulation of stomatal aperture and how this impacts plant growth and tolerance to drought stress. We also explore the distinctive features of the guard cell oscillator mechanism and the importance of these features for controlling stomatal aperture and bring evidence for a pathway by which the circadian system may regulate guard cell opening.
RESULTS
Generating Plants in Which Stomatal Opening Is Not under Circadian Control
Our first aim was to examine the physiological importance of circadian regulation of stomatal opening. To generate plants that showed arrhythmic stomatal aperture but normal circadian rhythms in the rest of the plant, we overexpressed YFP and HA-tagged CCA1 using the guard cell-specific promoter GUARD CELL1 (GC1; Yang et al., 2008; GC1::CCA1:HA:YFP). Since overexpression of CCA1 from the Cauliflower mosaic virus 35S promoter affects physiological rhythms at the level of the whole plant (CCA1-ox; Wang and Tobin 1998), we hypothesized that overexpressing CCA1 specifically in guard cells may exclusively affect stomatal aperture rhythms. Although microarray analyses have shown that GC1 may be weakly rhythmic under certain conditions (Michael et al., 2008), it is expressed in wild-type plants at levels that are more than 50-fold higher than CCA1 (Edwards et al., 2006; Yang et al., 2008); therefore, we predicted that GC1-driven CCA1 overexpression should be sufficient to disrupt guard cell rhythmicity (Wang and Tobin, 1998). The GC1::CCA1:HA:YFP construct was used to transform wild-type Arabidopsis to generate CCA1-oxSGC (specifically guard cell) plants. Following selection for the transgene, T1 CCA1-oxSGC plants were grown in long days (LDs; 14 h light:10 h dark) and examined by confocal microscopy to identify lines that expressed high levels of YFP in nuclei of the stomatal guard cells but not in the surrounding cells (Fig. 1A). Seven lines showed guard-cell-specific CCA1 accumulation in the nucleus; two independent transformed lines, CCA1-oxGC line 2 (SGC2) and CCA1-oxGC line 24 (SGC24), were chosen for further study. Figure 1, B and C, show that in wild-type stomata, CCA1 oscillations were low but rhythmic. By contrast, expression of CCA1 in stomatal guard cells of SGC24 plants was high and arrhythmic. Although CCA1 transcript levels were arrhythmic, as previously reported (Endo et al., 2014; Shimizu et al., 2015), high levels of CCA1 transcription from a cell-specific promoter did not completely abolish circadian expression of other oscillator genes (Supplemental Fig. S1)
CCA1 is overexpressed and arrhythmic in stomatal guard cells of CCA1-oxSGC plants. A, SGC24 (a line of CCA1-oxSGC) plants were grown for 2 weeks in LD and then imaged 6 h after lights-on by confocal microscopy. Light field, DAPI, YFP, and YFP + DAPI are shown. B and C, SGC24 and wild-type plants were grown for 4 weeks in LD before being transferred to LL. B, CCA1 levels in stomatal guard cells of SGC24 and the wild type. C, CCA1 levels also shown plotted for the wild type alone. B and C, The average of three independent biological repeats; the highest peak of CCA1/GDPH in the wild type of each repeat was normalized to 1. sd is shown, and the white and hatched bars represent subjective day and night, respectively.
We tested whether the ability of stomata to open and close was affected by the overexpression of CCA1 in guard cells. Wild-type, SGC24, SGC2, and CCA1-ox plants were grown in LD. Figure 2A shows samples taken before lights-on (“dawn”) at ZT-0.5 and 2 h after lights-on at ZT2. The plant material harvested at ZT-0.5 was measured immediately (“anticipation”); samples harvested at ZT2 were either measured immediately (“control”) or treated with an “opening solution” that favors opening or with ABA, to induce closure, as described in “Materials and Methods.” There was no significant difference in stomatal aperture among SGC2, SGC24, CCA1-ox, and wild type in the control or after treatment with opening solution and ABA (Fig. 2A; Supplemental Fig. S2). Thus, the ability of SGC2 and SGC24 plants to open and close their stomata is similar to that of wild-type plants. The slightly wider stomata in CCA1-ox plants (Fig. 2A; Supplemental Fig. S2) may be a result of slight differences in their structure (Supplemental Fig. S3), but we were unable to record sufficient numbers of stomata to quantify structural differences between wild type, CCA1-ox, and SGC stomata. The stomatal indices for wild-type, CCA1-ox, and SGC plants were similar (Supplemental Fig. S4).
SGC plants do not show circadian rhythms of stomatal aperture regulation. A, Wild-type, SGC24, SGC2, and CCA1-ox plants were grown for 4 to 5 weeks in LD before samples were taken at ZT-0.5 (anticipation) or ZT2 (control, opening solution, and ABA) and treated as described in “Materials and Methods,” and stomatal aperture was measured. B to D, Wild-type, SGC24, and SGC2 plants were grown for 4 to 5 weeks in LD and either kept in LD (B and C) or transferred to LL (D). B and D, Stomatal aperture measurements. The white, hatched, and black bars represent light, subjective night, and dark, respectively. C, Leaf temperature measurements; each point represents the average leaf temperature/aluminum foil standard with the highest value normalized to 1, as described in “Materials and Methods,” for a single time point. A 12-point moving average is shown for each genotype. The gray and white regions represent dark and light. ***P < 0.001 (Student’s two-tailed t test).
The timing of stomatal opening was clearly altered in the transgenic lines. At ZT-0.5, wild-type stomata were already open, while SGC2 and SGC24 stomata were still closed (Fig. 2A). As previously shown (Green et al., 2002; Thain et al., 2004; Dodd et al., 2005; Wang and Tobin, 1998), CCA1-ox stomata also showed no dawn anticipation. Analysis of stomatal opening over an entire LD cycle revealed that wild-type stomata start to open before lights-on; by ZT-2 they show significant opening, reaching maximal aperture at ZT-0.5 (Fig. 2B), and there was no significant increase in aperture between ZT-0.5 and ZT2 (P = 0.2, Student’s two-tailed t test). By contrast, SGC2 and SGC24 stomata were still closed at ZT-0.5 and showed no significant opening until ZT2 (P < 0.001 for both SGC2 and SGC24, Student’s two-tailed t test). The maximum stomatal aperture for both SGC2 and SGC24 was similar to wild type but occurred later, at ZT4. Leaf temperature is a useful indicator of transpiration (Negi et al., 2014), and Figure 2C and Supplemental Table S1 show that wild-type leaf temperatures decreased well before lights-on, consistent with increased stomatal opening and transpiration, while temperatures of SGC leaves remained high until lights-on. Together, these results strongly suggest that SGC guard cells are unable to predict daily dark-to-light changes. During the rest of the day, stomata in both SGC2 and SGC24 behaved like wild type and started to close well before lights-off, indicating that additional pathways are regulating closure. Overall, however, SGC2 and SGC24 stomata were more open for longer periods during the light part of the day than were the wild-type controls (Fig. 2B; Supplemental Fig. S5).
We also tested whether circadian regulation of stomatal opening was affected in the SGC plants. Figure 2D shows that under conditions of constant light (LL), wild-type stomata are robustly rhythmic, reaching maximum opening in the subjective morning. Contrasted with wild-type plants, SGC2 and SGC24 showed small, random variations in stomatal aperture throughout the course of the experiment but did not show significant circadian rhythmicity (Fig. 2D).
Overexpression of CCA1 in Guard Cells Does Not Significantly Affect Other Circadian Controlled Physiological Processes
In CCA1-ox plants, the ectopic expression of CCA1 under the control of the Cauliflower mosaic virus 35S promoter affects a wide range of circadian-controlled processes, including hypocotyl elongation, flowering time, leaf movements, and gene expression (Green et al., 2002; Thain et al., 2004; Dodd et al., 2005; Wang and Tobin, 1998); we examined whether these nonstomatal processes were affected in SGC2 and SGC24 plants. Figure 3A shows that in low-light conditions there was little difference in hypocotyl length between SGC2 (2.37 mm ± 0.14 se), SGC24 (2.54 mm ± 0.11 se), and wild-type (2.17 mm ± 0.09 se) seedlings. Consistent with previous reports, CCA1-ox seedlings were taller (5.04 mm ± 0.25 se). In the dark, all the genotypes were a similar height. For the flowering time experiments, CCA1-ox, SGC2, SGC24, and wild-type plants were grown in LD or short days (SDs; 8 h light:16 h dark). As expected, the CCA1-ox plants showed photoperiod-insensitive flowering, with an average of 50.6 ± 0.4 leaves in LD and 51.1 ± 3.08 leaves in SD (Fig. 3B). By contrast, wild-type, SGC24, and SGC2 plants all flowered significantly earlier (20.5 ± 0.39, 20.9 ± 0.38, and 20.5 ± 0.55 leaves, respectively) in LD than in SD (54.5 ± 0.96, 48.8 ± 2.09, and 51.8 ± 1.55 leaves, respectively). SGC plants also showed similar periods of leaf movement rhythms to wild-type plants (wild type 23.68 ± 0.13 se, SGC24 22.85 ± 0.7 se, and SGC2 22.77 ± 1.16 se; Fig. 3C) but higher relative amplitude error values (wild type 0.11, SGC24 0.34, and SGC2 0.51). Relative amplitude error is used to assess individual rhythm robustness, with values close to 0 indicating robust cycling and values at or near 1 indicating a rhythm with an error value as large as the amplitude itself (not statistically significant; Plautz et al., 1997). Together, our results suggest that in the SGC plants, while circadian regulation of stomatal aperture is altered, other circadian-controlled processes in the plant are less affected.
Nonstomatal circadian-controlled processes are less affected in SGC plants. A, CCA1-ox, SGC2, SGC24, and wild-type plants were grown for 5 d under low (25 µE m−2 s−1) white light or in the dark, and hypocotyl lengths were measured (n for each group = 12). B, CCA1-ox, SGC2, SGC24, and wild-type plants were grown in LD or SD, and flowering time was determined by counting the numbers of leaves at bolting (n for each group = 18–20). C, SGC2, SGC24, and wild-type plants were grown for 7 d in LD before being transferred to LL, and leaf movements were measured (n for each group = 6–16). Averages and se are shown.
Plants with Arrhythmic Stomatal Opening Perform as Well or Better Than the Wild Type in Optimal and Mild Drought Stress Conditions
We then examined whether the loss of circadian control of stomatal opening in the SGC plants affected plant growth. After 2 weeks in LD on soil in optimal fully watered conditions, SGC plants were slightly taller than wild type (Supplemental Fig. S6). By 4 weeks, SGC2 and SGC24 plants were significantly larger than wild-type plants (Fig. 4A). The aerial parts of the plants were harvested and several growth parameters measured. The fresh weights of SGC2 (0.8 g ± 0.05 se) and SGC24 (0.95g ± 0.05 se) plants were significantly higher than that of wild-type (0.61 g ± 0.04 se) plants (P < 0.001; Fig. 4B). When dry weight was measured, SGC24 (68 mg ± 3 se) plants weighed significantly more (P < 0.001) and SGC2 (55 mg ± 4 se) plants slightly more (P = 0.081) than the wild-type (48 mg ± 4 se) control (Fig. 4C). We also found that the leaf areas of SGC2 and SGC24 (Fig. 4D) were somewhat greater than those of wild-type plants (P < 0.01 and P = 0.07, respectively). Our results suggest that, under non-drought-stressed conditions, plants with arrhythmic stomatal guard cells may grow better than wild type.
Under optimal conditions SGC plants grow larger than wild type. A to D, SGC2, SGC24, and wild-type plants were grown at 23°C for 4 weeks in LD on soil under optimal watering conditions. B, Fresh weight, C, dry weight, and D, photosynthetic area per plant were measured as described in “Materials and Methods.” B and C, The graphs are averages of two independent biological experiments. D, The results from three independent biological experiments were averaged and normalized to one by the wild type, and the se is shown. Student’s t test, **P < 0.01 (n for each genotype in each repeat 10–12).
Even though the SGC plants showed more biomass yield than wild-type plants under optimal conditions, we considered the possibility that their altered stomatal opening (Fig. 2) may affect their susceptibility to drought stress. Wild-type, SGC2, and SGC24 plants were grown, fully watered, on soil for 1 week after germination, and then either fully watered for a further 3 weeks or given water to 50% or 30% field capacity (FC; see “Materials and Methods”). At the end of the fourth week, the dry weights of the aerial parts of the plants were measured and normalized to 100% for the controls for each line. Figure 5A shows the average results of five independent experiments. While under the mild drought conditions of 50% FC, the growth of wild-type plants was not significantly affected compared with the fully watered wild-type controls; the SGC plants performed significantly less well than the fully watered SGC controls. However, despite the loss of biomass in 50% FC, overall the SGC plants still grew as well (SGC2) or better (SGC24) than the corresponding wild-type plants (Fig. 5B). By contrast, under the more severe, 30% FC, drought conditions, all three lines showed large decreases in biomass (Fig. 5A), and the SGC plants no longer outperformed wild type. Thus, the increased biomass yields we observed in SGC plants only occurred if sufficient water was available; under more severe drought stress conditions, SGC plants, like wild-type plants, grew poorly.
Growth parameters of SGC lines under mild drought stress. SGC2, SGC24, and wild-type plants were grown in fully watered conditions for 1 week then watered to 50% FC, 30% FC, or kept in fully watered conditions (control). A, The dry weights of the plants normalized to 100% by their fully watered controls. The percentage weight loss is calculated by comparison with the fully watered controls for each line. B, The average dry weights were calculated. The average for five biological independent experiments is shown with the se (n for each genotype in each repeat = 10–15). Student’s t test, *P < 0.05, ***P < 0.001.
Stomatal Guard Cells Have a Specialized Version of the Whole-Leaf Oscillator Mechanism
Our next goals were to explore how the circadian oscillator regulates stomatal opening and why the high levels of CCA1 in SGC guard cells caused guard cells to lose circadian anticipation of dawn (Fig. 2, B–D).
We have previously reported that in LL three circadian oscillator genes, TOC1, CCA1, and LHY, show altered expression in stomatal guard cells compared with the average expression in whole leaves (Yakir et al., 2011); to expand on these findings and examine in more detail how the guard cell oscillator may signal to regulate stomatal aperture, we used large-scale qRT-PCR arrays to examine the circadian expression of all the key oscillator genes (Hsu and Harmer, 2014) in wild-type guard cells. As controls for the quality of our guard-cell-enriched samples, we used known guard cell genes, INWARD RECTIFYING K+ CHANNEL1 (KAT1), KAT2 and BLUE LIGHT SIGNALING1 (Nakamura et al., 1995; Pilot et al., 2001; Leonhardt et al., 2004; Tsutsumi et al., 2013). All of the controls were expressed at much higher levels in the guard cell-enriched fractions than in the whole leaf samples (Fig. 6A; Supplemental Fig. S7A).
Expression of most oscillator genes is altered in guard cells. A to E, Wild-type plants were grown for 4 weeks in LD before being transferred to LL. The expression of (A) control genes KAT1, KAT2, and At4g14480, (B) CCA1, PRR7, PRR3, CHE, PRR5, RVE8, BOA, FKF1, and JMJD5, and (E) CO and GI were monitored in large-scale qRT-PCR arrays. E, FT expression was analyzed by small-scale RT-PCR. Expression levels were normalized to reference gene expression. The rest of the normalization results and the other circadian genes analyzed are shown in Supplemental Figure S7. C and D, the maximum expression level at the first peak for each of the rhythmic genes in both guard cells and whole leaves was calculated for all three reference genes (UBQ10, TUB, and GDPH) and averaged. The ratio of average levels in guard cells to average levels in whole leaves is shown together with the sem. C, The results for each gene. D, The allocation of genes into morning, afternoon, and evening categories was based on Mockler et al. (2007), and the average ratios of expression in guard cells to whole leaves was calculated for the genes in each category. Student’s t test, *P < 0.05. F, CO and FT expression in LD was analyzed by small-scale RT-PCR. The averages of two biological repeats and sd were calculated for each sample. The white, hatched, and black bars represent light, subjective night, and dark, respectively.
Our results show that transcript levels of most of the oscillator genes were different in guard cells (Fig. 6B; Supplemental Fig. S7B) compared with whole leaves. Consistent with our previous results (Yakir et al., 2011), the amplitude of CCA1 and LHY was reduced and CHE expression was very low (Fig. 6B; Supplemental Fig. S7B). Other genes with reduced expression include PRR3, PRR7, RVE4, and RVE8. By contrast, a number of genes such as PRR5, BOA, and bHLH69 were up-regulated, while some genes, for example FKF1 and JMJD5, showed similar circadian expression patterns in both guard cells and whole leaves. A potential problem in such studies is finding appropriate reference genes that show constant expression under all the conditions of time, tissue type, and environment used. To minimize the possible issue of reference gene variability affecting our results, we compared the expression of each oscillator gene with three different reference genes (GDPH, TUB, and UBQ10). Supplemental Figure S7 shows that the results from each reference gene tended to be in broad agreement, although the GDPH normalization slightly elevated expression levels of certain genes in guard cells, for example of CCA1 (Fig. 6B; Supplemental Fig. S7B).
To determine whether oscillator genes showing peaks at different times of day are more or less highly expressed in guard cells than whole leaves, we allocated the oscillator genes into morning, afternoon, and evening categories, based on Mockler et al. (2007). Then we averaged the maximum expression level (the first peak) of the experiment in LL for each of the rhythmic genes in both guard cells and whole leaves for all three reference genes (UBQ10, TUB, and GDPH) and the stomata/whole leaf expression ratio calculated. We observed that the morning oscillator genes had lower levels of expression in guard cells than in the whole leaves, in contrast with those expressed later in the day and at night (Fig. 6, C and D), which tended to be expressed more highly in whole-leaf samples.
It is possible that the differences we observed in circadian oscillator gene expression between guard cells and whole leaves in wild-type plants are important for ensuring the accurate circadian regulation of stomatal aperture by affecting the levels of the components of the photoperiodic pathway. Figure 6E shows that in LL, both CO and GI were expressed in guard cells at higher levels than in the whole leaf. The levels of ELF3 in whole leaves and guard cells tended to be similar (Supplemental Fig. S7B). Possibly as a result of the elevated levels of CO and GI, during the third day in LL FT levels were higher in guard cells although they subsequently decrease (Fig. 6E), suggesting that there may be additional regulatory pathways regulating FT in extended free-running conditions. Under diel (light:dark) conditions, expression of both CO and FT were higher in guard cells of wild-type plants compared with whole leaves (Fig. 6F).
Our results also showed a correlation between CCA1 levels and FT peak expression levels. SGC24 plants had higher CCA1 and lower FT expression than wild type, while CCA1-ox plants, which had even higher levels of CCA1 expression than SGC plants had correspondingly lower FT levels (Fig. 7, A and B). The reduced levels of FT at night in SGC plants are consistent with their more closed stomatal around dawn (Fig. 2, A and B). CCA1-ox plants showed similar stomatal opening reactions to SGC plants in anticipation of dawn (Fig. 2A), possibly indicating a minimum threshold level for FT activity. By contrast with wild type, SGC, and CCA1-ox plants, ccal lhy mutants have higher levels of FT during the night (Fig. 7C) and stomata that are more open (Fig. 7D). Together, our results suggest that CCA1 may regulate daily stomatal opening via FT and that this may be a pathway by which the circadian system controls guard cell opening.
CCA1 levels affect FT expression and stomatal aperture. A and B, Wild-type, SGC24, and CCA1-ox plants were grown for 4 weeks in LD. C, cca1 lhy, cca1, and wild-type plants were grown for 3 weeks in LD. The expression of (A–C) CCA1 and FT in guard cells was analyzed by small-scale RT-PCR and normalized to levels of the reference gene GDPH. The averages of three biological repeats and se were calculated for each sample. The white and black bars represent light and dark, respectively. D, Stomatal apertures were measured at lights-on in plants grown for 3 weeks in LD. The averages and se were calculated for each sample.
Regulation of Aperture by FT and CO Is Photoperiod Dependent
Finally, given that previously published work (Ando et al., 2013; Kinoshita et al., 2011) and our results show that components of the circadian/photoperiod pathways are involved in regulating guard cell aperture, we predicted that there may be similarities between stomatal aperture regulation and photoperiod control of flowering time. In Arabidopsis, flowering is induced by the LD conditions in the late spring and summer when CO transcription in the phloem companion cells in the light results in stable CO protein that induces FT expression (Jang et al., 2009; Valverde et al., 2004; An et al., 2004). In winter and early spring, when days are short, CO expression occurs during the dark, and CO protein is rapidly degraded.
We examined whether photoperiod gene regulation in stomatal guard cells of wild-type plants was also day length dependent. Figure 8A shows that in guard cells of SD-grown plants, CO expression levels only started to rise after dark and declined again before the end of the night; as a result, CO levels in SD were low during the light part of the day. In LD-grown plants, by contrast, CO expression was high in the light part of the day, both after dawn and in the “afternoon.” Consistent with the idea that CO induces FT expression in Arabidopsis only when CO is expressed in the light, FT levels were very low in guard cells of SD-grown plants but high in guard cells of LD plants (Fig. 8B). We also observed that the low levels of FT in SD were correlated with a change in stomatal aperture; the pattern of stomatal closure at night and rapid opening just before dawn observed in LD conditions (Fig. 2B) was lost in SD (Fig. 8C). In addition, SD treatments resulted in less stomatal closure at night and greater pre-dawn opening (Fig. 8C; Supplemental Fig. S8). Overall, our results suggest that there are strong parallels between the mechanisms regulating photoperiodic flowering and stomatal aperture.
Photoperiod-dependent regulation of guard cell gene expression and stomatal aperture. A and B, Wild-type plants were grown for 4 weeks in LD or SD. The expression of (A) CO and (B) FT in guard cells was analyzed by RT-PCR and normalized to levels of the reference gene GDPH. The averages of three biological repeats and sem were calculated for each sample. C, Wild-type plants were grown for 4 to 5 weeks in SD. Stomatal apertures were measured as described in “Materials and Methods,” and the averages and se were calculated for each sample. The white and black bars represent light and dark, respectively.
DISCUSSION
The Guard Cell Circadian System May Be Fine-Tuned to Regulate Stomatal Opening
Several studies have shown that the circadian oscillator mechanisms of different organs and tissues differ from that of the average whole-plant oscillator. For example, in roots the morning and evening loops of the oscillator are uncoupled (James et al., 2008). Vascular tissue also appears to have distinctive differences in oscillator mechanism; PRR3 only regulates TOC1 in the vasculature (Para et al., 2007) and, in diel conditions, vascular tissue has inverse gene expression profiles compared to whole leaf and mesophyll, with morning genes being more highly expressed in the mesophyll (Endo et al., 2014). By comparison, shoot apex oscillators appear to be similar to those in the whole plant (Takahashi et al., 2015).
In this article, we have demonstrated that some circadian oscillator components may be differently regulated in guard cells compared with the average whole leaf (Fig. 6; Supplemental Fig. S7). A few genes showed significantly damped expression in guard cells; CHE was arrhythmic and, consistent with previous reports that PRR3 is mostly confined to vascular tissue, PRR3 levels were low (Para et al., 2007). In contrast with mesophyll cells, morning-specific oscillator genes were significantly down-regulated in guard cells compared with whole leaves. A number of other genes, such as PRR5, BOA, and bHLH69 were up-regulated. Although we cannot completely rule out the possible effects of stomatal location on the leaf affecting oscillator performance (most stomata are located on the underside of the leaf and are thus more shaded), we suggest that these differences in the mechanism of the guard cell oscillator may be connected to the distinctive requirements of stomatal functioning, including regulation of stomatal aperture.
Reports that several of the genes involved in the photoperiodic flowering pathway also play key roles in regulating stomatal aperture suggest a pathway by which the circadian system may control stomatal opening. Constitutive open-stomata phenotypes have been reported for mutants overexpressing FT, TWIN SISTER OF FT, and SOC1 (Kimura et al., 2015) as well as for mutants overexpressing GI and CO, which are both known to be FT up-regulators. Conversely, mutations that repress FT cause stomatal closure (Kinoshita et al., 2011; Ando et al., 2013). At the level of the whole plant, CCA1 negatively regulates CO and GI expression (Lu et al., 2012; Mizoguchi et al., 2005). We observed that in free-running conditions, lower levels of CCA1 in guard cells were correlated with elevated CO and GI expression (Fig. 6E), and increased FT expression and a higher degree of stomatal opening. In diel conditions, there was a negative correlation between the levels of CCA1 expression in guard cells and stomatal opening in anticipation of dawn (Fig. 7). High levels of CCA1 prevented anticipation, possibly by repressing FT expression. However, since FT levels peak before stomatal opening, it is likely that other intermediates are also involved in the onset of anticipation. We also showed that after stomata have already opened, high CCA1 levels do not impede their normal behavior. This result contrasts with observations that high levels of TOC1 affect stomatal closure (Legnaioli et al., 2009) and suggests that the circadian oscillator may regulate stomatal aperture by multiple pathways.
The similarity between some of the genes involved in the photoperiodic flowering pathway and stomatal opening pathways also prompted us to examine whether CO and FT may function to regulate stomatal aperture in a daylength-dependent manner. Our results show that they do; only in LD did we observe high CO expression during the light part of the day and FT expression. By contrast, in SD, even though CO levels were high, it was only during the dark period, and FT was not induced. Thus, the circadian oscillator-CO-FT pathway for regulating stomatal aperture, like the pathway for photoperiodic flowering in Arabidopsis, may be restricted to LD. We also observed, as has previously been reported for other species such as Vicia faba and Chrysanthemum (Easlon and Richards, 2009; Schwabe, 1952), that in SD conditions Arabidopsis showed less stomatal closure at night and greater predawn opening (Fig. 8C). This increased stomatal opening may reflect a lack of cost to the plant of having more open stomata during the times of year when water is not expected to be limiting (Christman et al., 2009) or of a requirement for increased nutrient uptake in SD.
Together, our results show that the external coincidence photoperiod pathway that is known to control plant development, such as reproductive development, bud dormancy, and tuberization on a seasonal level (Böhlenius et al., 2006; González-Schain et al., 2012; Suárez-López et al., 2001) also regulates daily responses in Arabidopsis. It has been suggested that promoting stomatal opening and increasing photosynthesis (Ando et al., 2013) prior to flowering can be beneficial to plants; having a similar mechanism for regulating flowering and stomatal aperture in LD may be useful for coordinating metabolic processes and reproduction. In the future, it will be interesting to examine whether this photoperiodic regulation of CO and FT is conserved in the guard cells of day-neutral and short-day flowering plants.
Loss of Circadian Control of Stomatal Opening Affects Plant Growth
Having an appropriately functioning circadian system confers an adaptive advantage to plants (Yerushalmi et al., 2011; Dodd et al., 2005), and there is a strong correlation between robust circadian rhythms and growth vigor (Ni et al., 2009). Recently, mutating RVE genes has been shown to enhance growth of both juvenile and mature Arabidopsis plants (Gray et al., 2017). These findings have led to the suggestion that modifying circadian rhythms may be a means to manipulate crops to develop “improved” plants for agriculture. However, not surprisingly, given the huge range of plant processes that are regulated by the circadian system, misexpression of key circadian oscillator genes at the whole-plant level has pleiotropic effects on plants. For example, CCA1-ox plants are not only arrhythmic, but they also show altered seedling growth, mature plant size, biomass, photosynthesis, pathogen resistance, leaf movements, hormone biosynthesis, and reproductive development (Thain et al., 2004; Wang et al., 2011; Wang and Tobin, 1998; Dodd et al., 2005). Similarly, cca1 lhy double mutants that lack both CCA1 and the closely related LHY have circadian rhythms with very short periods and are affected in nearly every aspect of their growth and development compared with wild type, including height, leaf size, starch content, tolerance to abiotic stress, and photoperiodic flowering (Dong et al., 2011; Graf et al., 2010; Ni et al., 2009; Mizoguchi et al., 2002). Thus, modifications that affect the plant’s entire circadian system are probably rather “blunt objects” with which to manipulate specific aspects of plant metabolism and development. Here, we have shown that it is possible to misexpress CCA1 in guard cells without significantly affecting normal reproductive development and other physiological circadian rhythms. The small differences in leaf movements and seedling height of SGC and wild-type plants (Fig. 3; Supplemental Fig. S4) may be caused by slight leakiness in the GC promoter or intercellular signaling to the epidermal cells (Shimizu et al., 2015) or be an indirect effect of altered stomatal opening.
We observed that under optimal diel conditions, SGC plants grew larger than the corresponding wild type (Fig. 4). This increase in growth may be a result of SGC stomata being more widely open during the light part of the day (Fig. 2). From ZT2 until ZT14, the SGC stomata generally showed significantly more opening than wild-type stomata; only from before ZT0 to ZT2 were wild-type stomata more open (Fig. 2, A and B; Supplemental Fig. S5). A previous report demonstrated that manipulating H+-ATPase levels in guard cells enhanced light-induced stomatal opening and increased growth (Wang et al., 2014). However, when the H+-ATPase was mutated so that stomata were constantly open, even in the dark, the enhanced growth phenotype was lost (Wang et al., 2014). In addition, although predawn stomatal opening has been hypothesized to enhance early-morning photosynthesis and growth in some C3 plants, for example Eucalyptus camaldulensis (Resco de Dios et al., 2016b), a recent report showed that in at least some species grown under low-water stress conditions, predawn opening has little effect on early-morning photosynthesis (Auchincloss et al., 2014). Similarly, a study on Rubus species suggests that night time transpiration may not always be correlated with photosynthetic rate (McNellis and Howard, 2015). We suggest that it is possible that having stomata that are more open in the light may augment growth but that these gains may be offset by increased water loss if stomata are also open in the dark in conditions in which water availability may be limited. Consistent with this idea, our SGC plants that had stomata that were more open in the light but still closed normally in the dark and in response to ABA showed enhanced growth when they were fully watered.
In the field, mild drought stress is often more likely to be encountered than extreme drought conditions. Studies have demonstrated that plants with improved survival under extremely high drought stress generally show comparatively poor biomass yield under optimal or mild drought stress, and it has been suggested that selecting for plants that show improved performance under optimal conditions but still grow well under mild drought might be a more useful strategy for crop improvement (Skirycz et al., 2011). In fully watered conditions, our SGC plants outperformed the wild-type controls, and in FC50%, although they did show a growth reduction, SGC plants still grew as well or better than wild type (Fig. 4). Thus, specific manipulation of guard cell circadian aperture may be useful for generating plants that have higher yields under optimal conditions but still able to cope with mild drought stress. It is important to note that even though the increases in growth we observed in the SGC plants were small, incremental improvements in yield can cumulatively make a large impact.
MATERIALS AND METHODS
Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) ecotype Columbia-0 was used for all experiments. Seeds were imbibed at 4°C for 4 d to optimize germination. For the leaf movement and hypocotyl experiments, plants were grown on Murashige and Skoog (Weigel and Glazebrook, 2002) medium (Duchefa Biochemie) supplemented with 3% Suc. Unless otherwise stated, for all other experiments plants were grown on soil with twice weekly irrigation. For full watering, the pots were kept on trays, the water was added directly to the trays, plants were left for 2 h, and any excess water was drained off. Unless otherwise stated, plants were grown in a 14-h-light (125 µE m−2 s−1):10-h-dark photoperiod (LD) at a constant 23°C. Philips fluorescent lights TLD 18W/29 and TLD18W/33CW provided lighting for plant growth.
Isolation of Guard-Cell-Enriched Fractions and Gene Expression Analysis
Whole-leaf and guard-cell-enriched fraction isolation, RNA preparation, and small-scale qRT-PCR were carried out as previously described (Yakir et al., 2011; Green and Tobin, 1999). For large-scale qRT-PCR, the BioMark multiplex microfluidics system (Fluidigm) was used, according to the manufacturer’s protocols, with cDNA pooled from 4 to 5 independent biological repeats. Calibration curves were checked for each of the primer sets (Supplemental Table S2) on a series of dilutions (1:1, 1:4, 1:16, 1:64 cDNA:water) of cDNA pooled from all the samples.
Stomatal Measurements
To measure stomatal opening, we used a modified version of published protocols (Ando et al., 2013). In brief, a leaf from each plant was blended with 40 mL of water in a Waring blender (Waring Commercial). The blended material was filtered on a 100 μm nylon mesh, the epidermal sheets were collected and mixed with 1 mL of water, and 300 μL of the mixture was transferred to a microscope slide. In Figure 2A, the anticipation data leaf samples were collected 0.5 h before lights-on (ZT-0.5). For the control, opening solution, and ABA treatments, samples were taken 2 h after lights-on (ZT2). The control and anticipation samples were analyzed immediately. For the ABA treatments, after blending and filtering, the plant material was incubated with 700 μL of 5 μm ABA for 90 min at room temperature in the light before imaging. For the opening solution treatment, the epidermal sheets were incubated in 48 mm KCl, 0.1 mm CaCl2, and 5 mm MES, pH 6.1, for 150 min at room temperature in white light. For dark conditions, the slide preparation process was done in low green lighting, and the slides were kept in the dark until the stomatal apertures were measured. In Supplemental Figure S2, the samples were collected at ZT2 and then treated for 90 min with distilled water, opening solution, or ABA. For the circadian experiments, plants were entrained in LD for 4 to 5 weeks before being transferred to LL. Photos of at least 50 stomata from a minimum of five epidermal fragments were taken and analyzed in the ImageJ program. The ratios between the width of the pore and the length of the guard cells were calculated for each stoma as in Legnaioli et al. (2009).
Stomatal indices were measured in 3-week-old plants. To make the negative impression of leaf surfaces, four mature leaves of plants from each genotype were covered with dental impression material (1:1 freshly prepared mixture of the two components of Polysiloxane condensation silicone; Zhermack). After 1 h, the material was removed. A positive impression of the leaf was obtained by covering the Polysiloxane surfaces with clear nail polish (Kagan et al., 1992) that was left to dry overnight. The replica was analyzed with a bright-field microscope (Nikon Eclipse E 200). Stomatal and epidermal pavement cells were counted in an area of leaf encompassing 80 to 120 stomata and the indices calculated as stomata/epidermal cells.
Thermoimaging of Leaf Temperature
Plants were grown for 4 weeks in a 14-h-light/10-h-dark light regime in a 150 µE m−2 s−1 white light at 22°C. After 4 weeks, the plants were moved to the measuring chamber for a 3-d acclimation period at 22°C in 14-h light/10-h dark, 150 µE m−2 s−1 white light provided by 50% red (615 nm) and 50% blue (450 nm) LEDs. After the acclimation period, leaf temperatures were read at 5-min intervals for 72 h using an A305sc thermocamera (FLIR). The light regime and quality during the days of measurement remained the same as in the acclimation period. Data from the thermocamera was analyzed using the FLIR-tools program. In order to account for temperature fluctuations in the measuring chamber, plant temperature was divided by the temperature of an aluminum foil standard placed within the chamber to give relative temperature; the highest value was then normalized to 1. A 12-point moving average was calculated from the relative temperature data.
Transgenic Plants
CCA1::CCA1:HA:YFP (Yakir et al., 2011) was used as a basis to make the GC1::CCA1: HA:YFP construct. The region of 1,140 bp of the GC1 promoter that has been shown to define guard cell expression (Yang et al., 2008) was synthesized by GenScript. The construct was cloned into pMLBART, and plants were transformed and selected as described in Yakir et al. (2011).
Confocal Microscopy
The confocal microscope was used essentially as described in Yakir et al. (2009). Leaf tips from 2-week-old CCA1-oxSGC plants grown in LD on soil were collected 6 h after lights-on. Epidermal sheets were prepared by the blending/filtering method (as for guard-cell-enriched fractions) three times for 15 s each, in ice-cold deionized water, in a Waring laboratory blender (Waring Commercial). The sheets were floated for 20 min in dark, in a solution of 50 ng/mL 4,′6-diamidino-2-phenylindole (DAPI) as a stain for nuclei and subsequently examined using a Leica SP5 confocal microscope. Excitation and emission wavelengths for DAPI-stained cells were 405 nm and 370 to 430 nm; excitation and emission for YFP were 515 nm and 535 to 565 nm.
Electron Microscopy
A slightly modified method of Asakura et al. (2004) was used. In brief, leaves were cut in 5% glutaraldehyde in 0.1 m cacodylate buffer, pH 5.5, vacuum-treated for 15 min, and fixed overnight at 4°C. After three washes in 0.1 m cacodylate buffer, the tissue was postfixed with 2% osmium tetroxide in the presence of 1.5% potassium ferricyanide for 2 h. The fixed samples were dehydrated in ethanol and embedded in Epon resin. Ultrathin sections cut by a LKB Bromma 8800 ultratome were stained with uranyl acetate and lead citrate. Micrographs were taken with Tecnai 12 electron microscope (Phillips) equipped with a Megaview II CCD camera and an Analysis 3.0 Soft Imaging System.
Plant Measurements
Circadian-regulated growth parameters were determined as follows. For hypocotyl measurements, plants were grown under low white light or in the dark, and hypocotyl lengths measured after 5 d. To determine flowering time, plants were grown on soil in LD or SD and the number of leaves counted at bolting. Leaf movements were determined as previously described (Yakir et al., 2009).
To measure size, plants were grown in LD on soil. For fresh weight, the aerial part of the plants was harvested and weighed. For the dry weight measurements, the plants were desiccated at 60°C for 2 d. For the leaf area measurements, plants were harvested and displayed on a white sheet of paper, so that the leaves were nonoverlapping and the leaf photosynthetic area measured using a FluorCam from PSI (Photon System Instruments).
Drought Stress
Plants were grown in fully watered conditions for 1 week then maintained at 100%, 50%, or 30% of field capacity determined by weighing the pots with the plants (FC, the maximum amount of water the soil can hold). After 3 weeks, the aerial parts of the plant were cut and dry weights measured.
Supporting Information
The following supplemental materials are available.
Supplemental Figure S1. LHY expression stomatal guard cells of SGC24 and wild-type plants.
Supplemental Figure S2. Effect of opening solution and ABA on stomatal aperture.
Supplemental Figure S3. Transverse sections through stomata of wild-type, SGC24, and CCA1-ox plants.
Supplemental Figure S4. Stomatal indices are similar in wild-type, SGC24, and CCA1-ox leaves.
Supplemental Figure S5. Comparison of stomatal opening during the light part of the day in wild-type, SGC2, and SGC24 plants.
Supplemental Figure S6. Two-week-old SGC2 plants are slightly taller than the wild type.
Supplemental Figure S7. Circadian expression of canonical oscillator genes in whole leaves and guard cells.
Supplemental Figure S8. Stomatal opening and closure is different in LD and SD.
Supplemental Table S1. Predawn slope analysis (anticipation).
Supplemental Table S2. Primers used for qRT-PCR.
ACKNOWLEDGMENTS
We thank Judy Lieman-Hurwitz for her help with the cloning, Naomi Melamed-Book for confocal imaging, Charles Warren for help and advice about stomatal measurements, and the late Nora Reinhold for her insights and critical reading of the manuscript.
LITERATURE CITED
Author notes
This research was supported by grants from the Deutsche Forschungsgemeinschaft (0308462), the Israel Science Foundation (0398636), the Ministry of Absorption, and the United States-Israel Binational Science Foundation (0378686).
Address correspondence to rgreen@mail.huji.ac.il.
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Rachel M. Green (rgreen@mail.huji.ac.il).
R.M.G. and M.H. conceived and devised the project; M.H., A.T., Y.D., K.G., and E.S. carried out the experiments and analyzed the data; A.A. assisted M.H.; R.M.G. wrote the manuscript; M.H. and Y.D. edited the manuscript.
Articles can be viewed without a subscription.
