The Clock Gene TOC1 in Shoots, Not Roots, Determines Fitness of Nicotiana attenuata under Drought

The circadian clock gene TOC1 functions in shoots, allowing plants to produce seed when water is limited.

Abstract

The highly conserved core circadian clock component TIMING OF CAB EXPRESSION1 (TOC1) contextualizes environmental stress responses in plants, for example by gating abscisic acid signaling and suppressing thermoresponsive growth. Selective interaction of TOC1 with PHYTOCHROME B under far-red–enriched light suggests a connection between circadian gating of light responses and sensitivity to ABA, an important regulator of growth and stress responses, including under drought. However, the fitness consequences of TOC1 function, particularly in the root, are poorly understood. Here, we used the desert annual, Nicotiana attenuata, to investigate the function of TOC1 in shoots and roots for maintaining fitness under drought, in both field and glasshouse experiments. Despite marked decreases in leaf water loss, TOC1-deficient lines failed to maintain fitness in response to drought stress as measured by total seed capsule production. Restoring TOC1 transcript levels in shoots via micrografting was sufficient to restore wild-type drought responses under field conditions. Microarrays identified a coexpression module in leaves strongly linking red and far-red light signaling to drought responses in a TOC1-dependent manner, but experiments with phytochrome-deficient lines revealed that the effects of TOC1 deficiency under drought cannot be attributed to changes in red/far-red light perception alone. Taken together, these results elucidate the sophisticated, tissue-dependent role of the circadian clock in maintaining fitness in the face of long-term abiotic stresses such as drought.

Plants are sessile, and as a consequence, their fitness depends on the ability to adapt to environmental changes. The circadian clock provides a powerful mechanism for plants to contextualize these adaptive responses to the environment. Both physiological (Green et al., 2002; Izawa, 2012) and genomic evidence (Hofmann, 2012; Lou et al., 2012) indicate the adaptive significance of the plant circadian clock. Much of this evidence is consistent with the importance of the clock in timing responses to abiotic stress (Kant et al., 2008; Mizuno and Yamashino, 2008). TIMING OF CAB EXPRESSION1 (TOC1) is an essential component of the central repressilator loop of the circadian clock in most plant species, repressing and being repressed by LATE ELONGATED HYPOCOTYL (LHY; Alabadí et al., 2001; Pokhilko et al., 2012; Nohales and Kay, 2016). While Arabidopsis (Arabidopsis thaliana) has a paralog to LHY, CIRCADIAN CLOCK-ASSOCIATED1, which displays a partially redundant function to LHY (Mizoguchi et al., 2002; McClung, 2010), other species, including Nicotiana attenuata, lack this paralog (Zdepski et al., 2008; Okada et al., 2009; Takata et al., 2009; Yon et al., 2012). Because of its central role in regulating a variety of endogenous processes, the circadian clock likely helps to maintain plant fitness in the face of stress (Greenham and McClung, 2015; Seo and Mas, 2015).

Drought is a particularly complex stress, for which it has been difficult to link molecular mechanisms to phenotypes and fitness outcomes (Des Marais, 2017). Mechanistically, drought responses are directly connected to the clock through TOC1. In Arabidopsis, TOC1 binds directly to the promoter of the magnesium-protoporphyrin IX chelatase H subunit (ABAR/CHLH/GUN5), controlling its circadian expression, and the time-of-day dependent induction of TOC1 by abscisic acid (ABA) is reciprocally regulated by ABAR (Legnaioli et al., 2009). While the function of ABAR as an ABA receptor has been called into question (Müller and Hansson, 2009), more recent work has shown specific binding of ABA to ABAR, likely through a domain in the C-terminal half of the protein (Wu et al., 2009). The reciprocal gated induction of ABA by TOC1 is regulated by the transcription factor MYB96 (Lee et al., 2016b). The flowering-promoting circadian gene GIGANTEA (GI) provides a connection between ABA-dependent responses related to flowering and the clock by stabilizing the central florigen activator CONSTANS in an ABA- and photoperiod-dependent manner (Riboni et al., 2013). GI in turn is a positive regulator of TOC1 and a member of the clock’s “evening loop,” although the physical evidence for this connection is still lacking (Martin-Tryon et al., 2007).

Variation in circadian clock rhythms among plant tissues has been described in Endo et al. (2014), but the functions of this tissue-level specificity have not been investigated. The circadian clock has been described as tissue-specific, but not tissue-autonomous (James et al., 2008); it has more recently been shown that the plant clock has a hierarchical organization, with the shoot apex clock playing a dominant role in synchronizing distal organ clocks (Takahashi et al., 2015). However, despite the molecular evidence indicating shoot dominance, Arabidopsis roots and shoots have different rhythmic properties and respond independently to light signals via light transmitted through the stem (Lee et al., 2016a; Nimmo, 2018) and roots can be directly entrained by low-intensity light even in antiphase to attached shoots (Bordage et al., 2016). It is unclear how these tissue-independent responses can functionally emerge from hierarchical, tissue-specific circadian clocks. The combination of both root- and shoot-specific responses to drought (Montero-Tavera et al., 2008), and root- and shoot-specific circadian clock variation (James et al., 2008), implies a potential fitness benefit of root- and shoot-specific variation in circadian clock-mediated drought responses. Whether tissue-specific variation in the circadian clock confers an advantage to plants facing drought remains unknown and is one of the motivations for this work.

TOC1 is also directly involved in suppressing thermoresponsive growth in Arabidopsis by inhibiting the phytochrome-interacting factor (PIF4; Zhu et al., 2016). In addition, selective interactions of TOC1 with PHYTOCHROME B (PHYB) under the far-red light conditions that characterize the end-of-day, when TOC1 is expressed, further suggest a connection between circadian gating of light responses and sensitivity to ABA (Yeom et al., 2014). Given that phyB mutants display decreased stomatal sensitivity and impaired transcriptional responses to exogenous ABA (González et al., 2012), we explored the links between TOC1-mediated ABA signaling and light sensing to reveal the ecological significance of these mechanistic interactions.

Despite our deepening molecular understanding of the circadian regulation of drought responses in particular tissues, the significance of these response mechanisms for plant fitness merits more realistic investigation. Rigorous fitness quantification requires the integration of responses on both physiological and developmental time scales in a whole-plant analysis. Allometric, or size-dependent, partitioning of resources between tissue-specific vegetative biomass and final reproductive effort is the gold standard for studying plant fitness responses that occur over developmental time (Weiner, 2004; Younginger et al., 2017). Allometric analyses are essential for understanding life history strategies such as the timing of the switch to reproductive growth from vegetative growth for annual plants (Weiner et al., 2009), and have yet to be used to explore the role of circadian clock components in whole-plant drought responses. Ecologically realistic and sophisticated drought manipulation techniques are also required to study drought responses that occur over both developmental and physiological time scales. To avoid erroneously comparing physiological responses of different genotypes in asynchronous developmental stages, drought treatments must be performed at developmentally matched times (e.g. at time of bolting) rather than at the same absolute time postgermination, following the dictum “stage, not age.” Likewise, erroneous comparisons of genotypes with unequal water status must be avoided (e.g. a drought-stressed genotype and a genotype yet to arrive at a similar stage of drought). Here, we combine appropriate watering techniques for field and glasshouse experiments with techniques for exploring tissue-specific analyses of single genes, to explore the circadian clock’s function in maintaining plant fitness under stressful conditions.

To explore this function, we have utilized the ecological model species N. attenuata, an annual wild tobacco species native to western North America (Baldwin et al., 1994; Kessler and Baldwin, 2001; Dinh et al., 2013). The species’ range is characterized by arid regions of cold and hot deserts (Köppen BWh; see Peel et al., 2007) as well as less xeric regions with both hot and dry summers but wet winters (Köppen Csa, Csb) and cold summer climates at higher elevations (Köppen Dsa, Dsb). Given this range and the many available lines of transgenic plants, N. attenuata provides a useful model species to explore the significance of TOC1 and phytochrome-mediated light signaling for fitness under field conditions.

Here, we explore the consequences of silencing TOC1 on fitness in the face of drought using micrografting and experimentally controlled drought scenarios in both field and glasshouse experiments. As previously reported, TOC1-silenced plants displayed altered physiological responses to drought, including decreased leaf water loss and increased water-use efficiency (Legnaioli et al., 2009; Greenham and McClung, 2015; Seo and Mas, 2015). However, we found that TOC1-silenced plants incurred severe fitness disadvantages under synchronized drought stress and a controlled watering regime. Under field conditions, TOC1 function in shoots was sufficient for wild-type allometric drought responses. Transcriptomic analyses of whole-plant or root-only TOC1-deficient inverted repeat TOC1 (irTOC1) plants revealed a coexpression module in leaves strongly linking red and far-red light signaling to drought responses. Analysis of transgenic lines silenced in phytochrome expression as well as other developmental regulators and circadian clock components revealed that altering red/far-red (R/FR) light perception alone is not sufficient to explain the fitness consequences of TOC1 silencing, or transcriptional responses in the face of drought stress, while ectopic overexpression of LHY alone phenocopied irTOC1’s drought response effects. These results reveal a role for TOC1 in contextualizing developmental as well as physiological responses for maintaining fitness in the face of drought.

RESULTS

N. attenuata TOC1 Is Required to Maintain Fitness under Controlled Drought Conditions

Previously, we identified and characterized the core circadian clock genes, among them TOC1, in N. attenuata (Yon et al., 2012). NaTOC1 displayed high (49% overall) protein sequence similarity to AtTOC1, with both the N-terminal signal receiver domain and the C-terminal CONSTANS, CONSTANS-LIKE, TOC1 motif being conserved, and yeast-two-hybrid assays showed protein interactions between NaTOC1 and NaZTL. NaTOC1-silenced plants also displayed late-flowering phenotypes under long-day (LD) conditions, consistent with the literature (Yon et al., 2012). In agreement with reports in Arabidopsis, we observed shortened periods of transcript accumulation in irTOC1 seedlings from two independently silenced lines under constant light of the marker gene resveratrolo-methyltransferase, which exhibits a circadian-regulated phenotype in EV plants (Supplemental Fig. S1, A and B; Millar et al., 1995; Strayer et al., 2000). The transcript abundance of two other PRR genes was not affected in irTOC1 lines (Supplemental Fig. S1, C and D), and BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) searches against the N. attenuata genome and transcriptome revealed that the sequence used for RNA interference (RNAi) targeting NaTOC1 does not overlap by >18 bp with any region outside of the NaTOC1 coding sequence, making off-target effects of the RNAi construct unlikely, as RNAi sequences have been shown to require a 21-nt match or greater for the formation of secondarysmall interfering RNAs (Schwab and Voinnet, 2010).

We performed a glasshouse experiment comparing irTOC1 and empty vector (EV) plants to measure control of water loss as well as developmental responses to drought. We used a gravimetric watering system to maintain a similar water status for EV and irTOC1 plants, to ensure that both genotypes experienced soil water deficit status simultaneously (Fig. 1A). Plants were treated in the rosette stage simultaneously (Supplemental Fig. S2). irTOC1 plants showed increased water use efficiency (WUE = assimilation rate divided by transpiration; Fig. 1B) and significantly reduced leaf water loss after leaf detachment in response to drought, whereas EV plants did not (Fig. 1C; using Control–Drought comparisons by time point per genotype, generalized least squares model, the “R” package tool “lsmeans” [Lenth, 2016], or Tukey posthoc). Despite this strong evidence of enhanced physiological tolerance of drought, irTOC1 plants subjected to drought stress failed to maintain relative fitness (final seed capsule production) in comparison to well-watered irTOC1 control plants, while EV plants showed no change in their fitness correlates in response to the same standardized drought treatment (Fig. 1D). Despite the expected late-flowering phenotype irTOC1 plants displayed (Fig. 1E), neither genotype elongated or flowered earlier under drought conditions, and the drought event occurred when both genotypes were in the rosette stage of growth (Supplemental Fig. 2, A and B). Because N. attenuata seeds may lie dormant for up to >150 years in between fires (Preston and Baldwin, 1999), extrapolating to Darwinian fitness would require measuring long-term seed viability under realistic seed bank conditions. Seed capsule production has been frequently used as a measure of relative fitness for N. attenuata plants (Karban and Baldwin, 1997; Zavala and Baldwin, 2004; Stitz et al., 2011; Yon et al., 2017) given that seeds per capsule and % viable seeds have been shown to be unaffected by different treatments in populations under native conditions (Baldwin, 1998). In a separate experiment, we did not observe significant effects of an early drought event (similar to Fig. 1) on seed number or seed mass per capsule (Supplemental Fig. S3). Although we did not examine seed viability effects of the drought treatment, we infer that the decrease in capsule number accurately reflects differences in seed set due to drought exposure, given that more extreme manipulations (50% leaf removal and 2 × 500-µg methyl jasmonate application to induce a strong wounding response) failed to induce changes in percentage of viable seeds (Baldwin, 1998). Given these results, we conclude that TOC1 deficiency confers severe fitness disadvantages under drought despite the seemingly advantageous physiological responses of irTOC1 plants.

Figure 1.

Decreases in leaf water loss of irTOC1 do not increase plant fitness in response to drought stress. A, After a period of establishment watering lasting roughly one week, EV and irTOC1 lines were grown using a controlled watering scheme and exposed to a drought period before watering resumed. B–D, Plant physiological responses were measured at the end of the drought, at 15 dpp. irTOC1 plants under drought conditions had increased WUE (B) and lower leaf water loss (C) than EV, but had lower reproductive output than empty vector or (D) control irTOC1 plants after rewatering followed by senescence and dry-down. E, Representative image of irTOC1 developmental delay, bolting 7 d after empty vector plants. ***P < 0.001, **P < 0.01, *P < 0.05, Control–Drought, multiple comparisons of means, Tukey contrasts following a significant effect in an ANOVA. Data are plotted as mean values, and error bars = mean ± se. Per genotype and treatment, 12 biological replicates were used for (A) and (D), and four biological replicates for (B) and (C).

Given the developmental delay observed in irTOC1 plants, we dissected the role of TOC1 in drought responses on both developmental and physiological time scales by generating plants with IrTOC1 roots, and wild-type TOC1 function in shoots, via micrografting (Fragoso et al., 2011). A preliminary experiment confirmed that grafting irTOC1 roots to EV shoots generated plants with TOC1-silenced roots and with wild-type levels of TOC1 in shoots (EV shoot/irTOC1 root [ET]; Supplemental Fig. S4A) that were developmentally indistinguishable from those of plants with grafted EV shoots and EV roots (EV shoot/EV root [EE]; Supplemental Fig. S4, B and C), in contrast to the substantial developmental delay of plants having both roots and shoots silenced in TOC1 expression (irTOC1 shoot/irTOC1 root [TT]; Supplemental Fig. S4C).

To more rigorously examine the fitness consequences of both whole-plant (TT) as well as root-only (ET) silencing of TOC1, we constructed an ecologically appropriate drought scenario for N. attenuata under field conditions (Lytle Preserve, Utah), represented schematically in Figure 2A to illustrate bolting times for each genotype and the timing of water limitation. After an initial establishment phase of three weeks in which plants were watered once per week for 1 h at dusk with a drip irrigation system (liters per hour flow rate), plants were subjected to slowly decreasing water availability by completely shutting off irrigation lines. Given that most N. attenuata lateral roots grow within the first 5 cm of soil under field conditions (Ferrieri et al., 2017), we tracked the rate of dry-down across the plot with soil core measurements from representative locations at depths of 5, 10, 15, and 30 cm over regular time intervals (Fig. 2B). Final root harvesting confirmed that the bulk of field rooting volume was found within the first 15 cm of soil (Supplemental Fig. S5B). Soil moisture levels around drought-treated plants decreased by ∼20%, with the smallest change occurring at 10 cm below the surface (14% decrease to 14.6% soil moisture) and the largest occurring at 30 cm (26% decrease to 11.5% soil moisture). To evaluate whether the drought treatment produced water stress uniformly in all drought-treated plants, we measured leaf water potential at dawn at the end of the experiment. The consistently significant differences between control and drought plants across all three genotypes confirmed that our drought treatment objectives were achieved (Supplemental Fig. S5).

Figure 2.

TOC1 expression in shoots rescues biomass to seed capsule conversions under drought conditions in the field. A, Experimental setups for fitness measures under both field and glasshouse conditions. B, Water status manipulations were monitored using soil moisture and leaf water potential (Supplemental Fig. S5), which indicated that all three genotypes experienced similar drought stress conditions. C, Ultimately, TT plants produced somewhat fewer seed capsules under drought conditions than EE and ET plants. D, TT plants were significantly larger under control conditions in the field, and had somewhat larger roots. E, TT plants failed to increase conversion of biomass to seed capsules under drought conditions, unlike EE and ET plants. Darker bars represent control samples, whereas lighter bars represent drought-treated samples. Dashed lines in (E) represent a 1:1 conversion on a log¹⁰ scale. ***P < 0.001, **P < 0.01, *P < 0.05, pairwise comparisons, multiple comparisons of means, Tukey contrasts following a significant effect in an ANOVA. Data are plotted as mean values, and error bars = mean ± se. Per genotype and treatment, 11–15 biological replicates were used for (C), (D), and (E). DW, dry weight.

We quantified seed capsule production and dry biomass to determine each genotype’s ability to maintain its fitness in the face of decreasing water availability at the end of the growing season. Under drought conditions, EE and ET in fact produced larger amounts of capsules than TT plants by day 54 of the experiment (Fig. 2C). While EE and ET plants maintained similar above- and belowground dry biomass under drought conditions, well-watered TT plants attained significantly greater biomass, roughly twice that of their drought counterparts and of EE and ET plants (Fig. 2D).

The limited effect on fitness correlates and strong effect on final biomass for TT plants under drought implied a change in the relationship between these two parameters. We therefore investigated whether seed capsule production followed an expected allometric pattern (i.e. whether it was dependent on vegetative growth); for annual herbaceous plants like N. attenuata, this pattern is expected to be positive (Weiner et al., 2009). There was a weaker correlation between biomass and seed capsule production for each genotype under control conditions (Fig. 2E, EE R² = 0.10; ET R² = 0.50; TT R² = 0.29). However, EE and ET plants subjected to a controlled dry-down period displayed a stronger relationship between biomass and seed capsules (EE R² = 0.43, ET R² = 0.78). TT plants under water deficit, on the other hand, showed an even weaker correlation than their well-watered counterparts (R² = 0.0067). The allometric relationships between biomass and seed capsule production resolved into statistically significant groups, between drought-treated EE and ET plants (statistical group c) and all control plants as well as drought-treated TT plants (EE and ET control: group bc; TT control: group a; TT drought: group ab).

Plants Silenced in TOC1 in Both Shoots and Roots Show Different Responses in Coexpression Modules under Drought Conditions

To better understand the strong fitness effect of TOC1 silencing under drought conditions, we performed whole-transcriptome microarray analyses on leaves and roots of EE, ET, and TT plants. After grafting, plants were grown for 13 d after potting, after which control plants remained well-watered and drought plants had water withheld for 7 d. Samples were harvested at the end of this week (Fig. 3A). As with all following experiments, plant water status was monitored using leaf angle at midday, after two preliminary experiments. The first experiment showed that midday leaf angles in N. attenuata track the volumetric soil moisture contents and leaf relative water contents (RWC; Supplemental Fig. S6)—measures of plant water availability. The second experiment showed that when leaf angles reached ∼45° to 50° at midday, a strong increase in leaf ABA accumulation and a strong shift in ABA-responsive transcripts were observed (Supplemental Fig. S7). Coexpression module discovery, gene set enrichment analysis (GSEA), and overrepresentation analyses (ORA) were performed within treatment and tissue groups using the “R” package “CEMiTool” (Russo et al., 2018).

Figure 3.

Microarray analysis of plants under control and drought conditions reveals a coexpression module in drought leaves strongly linking drought responses to red and far-red light response. A, Seedlings (10 d postgermination) were grafted to produce three distinct lines: EE plants, TT plants, and ET plants. Seedlings were then transferred to 1-L pots and grown under normal glasshouse conditions for 13 d before water removal. Leaf and root samples from four biological replicates were then harvested for microarray hybridization. B, Principal component analysis–separated samples according to tissues and treatments, and GSEA was performed using the “R” package “CEMiTool” within each treatment and tissue sample set separately. Colors represent higher (red) and lower (blue) activity across each module, whereas the size of each circle corresponds to the absolute NES value. NES values for modules that have an adjusted P value > 0.05 are removed. C, Analysis of coexpression modules revealed one module in drought leaves (M1) having a significant overrepresentation of genes related to R/FR responses, as well as to response to water deprivation; and one module in drought roots (M2) having a significant overrepresentation of genes related to response to water deprivation, water channel activity, and water transport. Both modules were strongly differentially expressed in TT plants relative to EE and ET plants. Bar graphs and colors represent the −log¹⁰ adjusted P value of the enrichment between genes for each GO term. Vertical gray lines represent an adjusted P value of 0.01. Three to four biological replicates per genotype, tissue, and treatment.

Figure 3B shows the results of the GSEA, performed using the algorithm “fgsea” (Sergushichev, 2016). For each circle, Normalized Enrichment Scores (NES) are calculated that represent the likelihood that the genes in each module are randomly distributed throughout a ranked list of genes (in this case, the z-score normalized expression levels) for each genotype (Subramanian et al., 2005). These values are represented by color and size, with red representing higher and blue lower activity across the module, while the size of each circle corresponds to the absolute value of each NES. The NES and P values for each module and genotype within each of the four sets (treatment and tissue groups) can be found in Supplemental Table S1. Modules were selected for those that were differentially expressed in TT in drought leaves and roots relative to both EE and ET genotypes (Fig. 3B). Two modules that fit this criterion were found, and ORA of these modules were enriched in gene ontology (GO) terms with the most significant adjusted P values of the enrichment: One coexpression module in drought-treated leaves, M1, had a strong overrepresentation of genes related to red and far-red light responses, as well as water-deprivation responses, while one coexpression module in drought-treated roots, M2, had a strong overrepresentation of genes involved in water-deprivation responses, water channel activity, and water transport (Fig. 3C).

TOC1-Silenced Drought Responses Are Not Explained by Impaired R/FR Light Perception

Given that the R/FR module is differentially expressed in irTOC1 shoots, as well as the evidence for TOC1 interacting directly with PHYB and PIF3 and PIF4 (Yeom et al., 2014; Soy et al., 2016; Zhu et al., 2016), we asked whether manipulating this module upstream of TOC1 would produce similar drought response phenotypes to those of irTOC1 plants (Fig. 4A). In seedlings, phyB mutants have constitutively elongated hypocotyls under high R/FR as well as under constant R light, while phyA mutants display constitutively elongated hypocotyls under low R/FR as well as under constant FR light (Smith et al., 1997). Thus, PHYA is thought to primarily mediate responses to FR, while PHYB primarily mediates responses to R light (Li et al., 2011). We therefore examined the fitness correlates and transcriptional responses of plants silenced in PHYB1, PHYB2 (partially redundant homologs of NaPHYB, Fragoso et al., 2017), PHYA1, or both PHYB1 and PHYB2 simultaneously. To control for the developmental effects of silencing phytochrome and TOC1 expression that may have affected our fitness measurements, this time we performed drought treatments at the same developmental stage, with each genotype undergoing drought at the onset of elongation (Fig. 4B). Leaf angle was again used as a nondestructive measure to track plant water status (Supplemental Fig. S8C), and plant water status of drought plants was compared to control plants, showing a strong and consistent reduction in RWC of 60% to 80% across all genotypes at sampling time (Supplemental Fig. S8D).

Figure 4.

irTOC1 drought responses are not explained by impaired FR/R light perception. A, Schematic representation of known interactions among TOC1, LHY, PHYB, and PIFs. B, Experimental setup for testing drought responses of phytochrome-deficient lines. Due to the developmental effects of silencing PHYB1 and TOC1, genotypes were subjected to drought treatment at the same developmental stage (beginning of elongation) in three blocks. The earliest elongating genotypes (“Early”) elongated at 16-d dpp, a second group (“Mid”) elongated at 18 dpp, and a third (“Late”) elongated at 23 dpp. C and D, Unlike EV and phytochrome-deficient lines, irTOC1 experienced a drastic reduction of fitness correlates when subjected to drought relative to control EV levels. E–M, Transcriptional analysis of TOC1-associated transcripts under drought. Contrasts were made among Control–Drought treatments for each genotype; where all pairwise comparisons were significant, groupings have been added to emphasize differences across genotypes. ***P < 0.001, **P < 0.01, *P < 0.05, all contrasts relative to EV controls, multiple comparisons of means, Tukey contrasts following a significant effect in an ANOVA. Data are plotted as mean values, and error bars = mean ± se. Four to five biological replicates per genotype and treatment.

All genotypes experienced a significant decrease in total seed capsules and biomass production at the end of the experiment under drought conditions (60 d postpotting [dpp]). However, comparing ratios of seed capsules or biomass of control plants to drought plants revealed a much stronger effect of drought for irTOC1 plants, with a 6- or 5-fold decrease under drought, respectively, indicating that similar results in other experiments did not result from differences in developmental timing. The irPHYB1, irPHYB2, irPHYA, and irPHYB1 × irPHYB2 lines all experienced similar 2-fold decreases under drought, which were also observed in EV plants (Fig. 4, C and D). We examined the transcript abundances of TOC1-linked response genes at TOC1 peaking time in EV plants, 7:00 pm under LD conditions, including GI, ABAR, and NaPIFs, and found similar expression patterns in phytochrome-silenced lines and EV plants, while these genes displayed differential abundance patterns in irTOC1 plants (Fig. 4, E–M).

Impaired Photoperiodic Flowering Control Is Not Sufficient to Explain Drought Responses and Developmental Phenotypes in irTOC1 Plants

Given the lack of irTOC1-like drought response phenotypes of phytochrome-deficient lines, we asked if this phenotype is driven by irTOC1’s developmental delay, or by other disruptions in the circadian clock mechanism. We examined the transcriptional responses and fitness correlates under control and drought conditions of two lines with developmental delays in photoperiodic flowering (irCRYP1a and irFT3). Cryptochromes act as blue and ultraviolet-A light receptors; Plants with cryptochrome 1 (cry1; CRYP1a in N. attenuata) mutant alleles demonstrate late flowering under certain conditions, whereas plants with gain-of-function mutations in cry1 exhibit early flowering and higher transcript abundance of the floral promoters CONSTANS and Flowering Locus T (Exner et al., 2010). We furthermore used lines silenced in, and ectopically overexpressing, LHY, the other central component of the plant circadian clock (irLHY and ovLHY, respectively). We again performed drought treatments in a developmentally standardized stage when plants were at the onset of elongation, splitting these genotypes into three blocks according to the timing of elongation between the earliest elongating genotypes, EV and irLHY (“Early”), a second group consisting of irCRYP1a and irFT3 (“Mid”), and a third group consisting of ovLHY and irTOC1 (“Late,” Fig. 5A) and again used leaf angle as a nondestructive measure of plant water status (Supplemental Fig. S8C), which was confirmed at sampling by RWC measurements (Supplemental Fig. S8D). As this experiment was performed in parallel to the experiments in Figure 4, the same EV and irTOC1 controls were used.

Figure 5.

ovLHY phenocopies irTOC1 drought responses and developmental delay. A, Experimental setup for testing drought responses of phytochrome-deficient lines. Due to the developmental effects of silencing PHYB1 and TOC1, genotypes were subjected to drought treatment at the same developmental stage (beginning of elongation) in three blocks. The earliest elongating genotypes (“Early”) elongated at 16 dpp, a second group (“Mid”) elongated at 18 dpp, and a third (“Late”) elongated at 23 dpp. B and C, irTOC1 and ovLHY experienced a drastic reduction of fitness correlates when subjected to drought, and experienced similar developmental delays. D–L, Transcriptional analysis of TOC1-associated transcripts under drought. ***P < 0.001, **P < 0.01, *P < 0.05, all contrasts relative to EV controls, multiple comparisons of means, Tukey contrasts following a significant effect in an ANOVA. Data are plotted as mean values, and error bars = mean ± se. Three to five biological replicates per genotype and treatment.

Plants from all genotypes experienced a significant decrease in total seed capsules and biomass production at the end of the experiment under drought conditions. While EV, irCRYP1a, irFT3, and irLHY plants showed a similar 2-fold reduction under drought, the decrease in ovLHY was stronger and similar to the drought responses of irTOC1 plants (20- or 8-fold decrease, respectively, Fig. 5, B and C). The transcript abundance of the TOC1-linked response genes GI, ABAR, and NaPIF1 followed similar expression patterns in irTOC1 and ovLHY plants (Fig. 5, E, F, and H). The transcript abundance of other NaPIFs either showed similar drought responses across most genotypes (NaPIF3a and NaPIF3b, Fig. 5, I and J), no response to drought in EV, ovLHY, or irTOC1 (NaPIF7, Fig. 5L), or a significant response in irTOC1 and irCRYP1a plants only (NaPIF4/5, Fig. 5K). These results suggest that impaired photoperiodic flowering mechanisms are not sufficient to phenocopy the drought responses of irTOC1 plants, and that only ovLHY plants exhibit impaired drought responses consistent with those observed in irTOC1 plants.

DISCUSSION

Here we demonstrate that silencing TOC1 incurs a severe fitness disadvantage for N. attenuata under drought conditions in both the glasshouse and the field. In the glasshouse, these fitness disadvantages occur despite improved physiological responses to drought, such as higher WUE (Fig. 1B) and lower rates of leaf water loss in irTOC1 plants (Fig. 1C). We analyzed the performance of micrografted plants under field conditions through an ecologically realistic drought scenario to explore the tissue-specific effects of TOC1 silencing. Unimpaired TOC1 function in shoots was sufficient for both wild-type developmental responses (Supplemental Fig. S4) as well as wild-type allometric relationships between whole-plant biomass and seed capsule production under drought conditions in the field (Fig. 2E), as judged by comparison to EV controls that have a wild-type phenotype (Schwachtje et al., 2008). Whole-transcriptome analysis yielded a coexpression module, suggesting a function for TOC1-mediated R and FR light signaling in drought responses of shoots, but not roots (Fig. 3C). Given that the fitness outcomes were not changed by abrogating R/FR light sensing directly (Fig. 4, C and D), the effect of TOC1 silencing appears to be downstream of phytochromes A, B1, and B2. Further screening of clock- and developmentally shifted lines (irLHY, ovLHY, irFT3, and irCRYP1a) showed that only a line overexpressing LHY displayed the same fitness outcomes and expression pattern of key TOC1-linked response genes as were displayed in irTOC1 plants; because LHY represses TOC1, this may be due to reduced TOC1 transcripts in the ovLHY line. However, crosses of irTOC1 and ovLHY, or a line bearing a double construct, could determine whether the phenotypic effects observed in both lines are additive; if they are, that would imply that the drought responses of irTOC1 and ovLHY, while yielding similar outcomes, do not stem from the same abrogation of clock function via TOC1.

Despite reduced levels of TOC1 at the chosen sampling time, (Fig. 5D), irLHY lines did not display the same drought-induced changes in transcript abundance of key TOC1-linked response genes or share the same fitness outcomes. This may be due to phase and amplitude differences in diurnal TOC1 expression resulting from silencing and overexpression of LHY; previous work has implied a shift in phase during LD conditions for N. attenuata irLHY lines (Joo et al., 2017). It is very likely that the circadian rhythm is shifted in the various lines examined in Figures 4 and 5; indeed, we observe a shift in the rhythm of irTOC1 plants under LD conditions (Supplemental Fig. S1A). How the rhythmicity of the circadian clock is disrupted by drought stress under LD conditions remains unclear. This question cannot be resolved satisfactorily with our single time-point transcript abundance analyses, although kinetic experiments conducted under similar drought conditions with N. attenuata clock-shifted lines (irTOC1, irLHY, and ovLHY) could elucidate the differences between TOC1 transcript abundance dynamics and their resultant effects on drought-related gene expression.

Similar to our findings, impaired TOC1 expression enhances physiological performance under drought stress conditions in Arabidopsis, such as greater seedling dehydration survival through altered ABA signaling (Legnaioli et al., 2009). Despite the fitness effects suggested by these short-term performance tests, the fitness consequences of impaired TOC1 expression in mature plants or throughout development are a novelty from this study, which has not been previously explored in Arabidopsis, particularly in the context of drought responses. More fine-grained fitness analyses of other clock-manipulated plants have shown that, even if circadian clock-impaired plants perform similarly to wild-type varieties under normal conditions, differences arise in a sensitized background. For example, although the fertility of GI-deficient Oryza sativa (osgi) was similar to wild-type under typical transplanting dates in the field, later transplanting dates yielded lower fertility of osgi relative to wild type (Izawa et al., 2011). These developmental effects under field conditions for osgi mutants, as well as the developmental effects of TOC1 silencing in N. attenuata and the clear fitness effects observed after seed set (Fig. 1E), highlight the importance of a more rigorous approach for the analysis of fitness under field conditions. Aside from the dormancy issues stemming from N. attenuata’s long-lived seed bank (up to 100 years or more) that complicate measuring the long-term viability of seeds, seed collection in our field experiments is further complicated by the need to prevent the distribution of ripe seeds from transgenic lines for regulatory compliance. Further tests of the effect of TOC1 silencing and drought treatment on seed viability would require a long-term seedbank experiment buried under natural conditions. Our analysis of seed number and seed mass per capsule showed that both parameters are affected in EE plants under late drought conditions (Supplemental Fig. S3). However, under the two experimental conditions employed in this study, namely “early drought” conditions (wherein all plants underwent drought at EV bolting time, employed in Fig. 1) and under “developmentally timed drought” conditions (wherein each genotype underwent drought at their respective bolting time asynchronously, employed in Figs. 4 and 5), genotypes did not demonstrate differences in these two parameters, and thus seed capsule number likely demonstrates an accurate relative estimate of fitness under these experimental conditions.

Allocation has often been analyzed as reproductive effort, defined as reproductive biomass divided by total biomass. However, reproductive effort is more rigorously analyzed as allometric (size-dependent) relationships of reproductive output (e.g. number of seed capsules or reproductive biomass) regressed against vegetative biomass (Weiner et al., 2009). With an allometric analysis, two key points to consider are plasticity in resource allocation, which can be defined as a change in an allometric trajectory (i.e. the slope of the regression), and optimal allocation theory, which predicts that plants will cluster more closely to the allometric regression (i.e. greater R² values) as they reach their reproductive potential after vegetative growth (Weiner, 2004). Under field conditions, we observed that the allometric trajectory between biomass and seed capsule production was strongly changed under drought conditions for all three genotypes (Fig. 2E). R² values increased under drought conditions for EE and ET plants, implying an earlier transition to reproductive growth than their control counterparts, while TT plants fail to undergo this transition. This plasticity in resource allocation is thus dependent on shoot, but not root TOC1 expression. Interestingly, we observed different responses of seed capsule production (Figs. 2C and 4C) and biomass (Figs. 2D and 4D) under field and glasshouse conditions. This difference may be due in part to variation in rooting volume between field- and glasshouse-grown plants: Field-grown TT plants under control watering displayed twice the root and shoot biomass of EE and ET plants. Differences in blue light between our field and glasshouse setups may have also influenced growth rates, with reduced blue light availability under glasshouse conditions leading to lower stomatal conductance and photosynthetic capacity (Hogewoning et al., 2010).

The allometric analysis from the field experiments, as well as the microarray analysis, point to a shoot-specific role of TOC1 in drought responses. In root-drought samples, the only transcriptional module with clear functional implications for drought (i.e. M2) was differentially expressed between EE/ET and TT plants, implying that a shoot-derived signal may be sufficient for drought responses (Fig. 3B). Although other modules displayed differential expression between EE and ET/TT, these modules had less clearly discernible functional implications for drought responses (Supplemental Table S2). Micrografting of RNAi lines such as irTOC1 does not permit the investigation of shoot-only TOC1 knockdowns, as the RNAi silencing signals travel from shoots to roots (Fragoso et al., 2011). Lines harboring mutant toc1 alleles would allow for the analysis of reciprocal toc1/EV and EV/toc1 grafts; conversely, heterografts of the ovLHY line that phenocopied irTOC1 drought responses (ovLHY/EV and EV/ovLHY) could be used to evaluate the inference regarding the role of shoot TOC1 silencing in the abrogation of drought responses. Before such experiments, crosses between ovLHY and irTOC1 should be used to determine whether the drought-related phenotypes of these lines are additive, which would change the conclusions that could be drawn from reciprocal grafting of ovLHY.

We infer from the results presented here that R/FR-related drought responses are mediated in a TOC1-dependent manner in the shoots of N. attenuata (Figs. 3C and 4 I). TOC1 is known in Arabidopsis to mediate rhythmic growth and gate thermoresponsive growth by direct interactions with PIF3 and PIF4, respectively (Soy et al., 2016; Zhu et al., 2016). TOC1 is also known to directly interact with other PIFs (Yamashino et al., 2003; including PIL5, also known as “PIF1”), which is also implied in N. attenuata by the increased transcript abundance of PIF1 in irTOC1 and ovLHY under drought (Fig. 5H). TOC1-PIF1 may form an analogous gating mechanism to TOC1-PIF3 and TOC1-PIF4 for drought-responsive growth or developmental signals. This proposition could be tested by examining whether NaTOC1 and NaPIF1 interact directly by yeast two-hybrid analysis, as well as by using virus-induced gene silencing of NaPIF1 in wild-type and IrTOC1 backgrounds to evaluate whether virus-induced gene silencing PIF1 irTOC1 plants display wild-type transcript abundances of TOC1-related marker genes. If all irTOC1-background plants displayed similar transcript abundances of these marker genes, a TOC1-PIF1 interaction is unlikely to form a drought-related gating mechanism analogous to TOC1-PIF3 and TOC1-PIF4 in Arabidopsis.

CONCLUSION

By combining careful drought treatments and allometric analyses under field conditions with an RNAi-driven screening of the circadian clock, this work reveals the fitness implications of shoot TOC1 deficiency under water limitation, and shows that TOC1 expression in the root does not contribute to TOC1-dependent fitness responses in biomass to seed capsule conversions. These data provide a functional test, under field conditions, of the hierarchical organization of root and shoot circadian clocks under drought stress. Further transcriptomic analysis and screening of transgenic lines provide evidence that R/FR-related drought responses are mediated in a TOC1-dependent manner in the shoots of N. attenuata downstream of phytochromes and other R/FR light-perception machinery.

MATERIALS AND METHODS

Plant Materials and Constructs

Lines were derived from seeds originally collected from natural populations of Nicotiana attenuata from the Desert Inn Ranch near Santa Clara, Utah (Baldwin et al., 1994). Seed germination and plant growth in the glasshouse were carried out as described by Krügel et al. (2002) and was undertaken in a glasshouse in Jena, Germany. Screening of the EV line (pSOL3NC, line no. A-04-266-3) is described by Bubner et al. (2006). Screening of irTOC1 lines (pSOL8_16844, A-11-205-4) via RNAi is described by Yon et al. (2012), with additional screening described in Supplemental Figure S1, A–C. Screening of ovLHY, irCRYPa, and irFT3 lines is described in Supplemental Table S3; the additional screening of the ovLHY lines is described in Supplemental Figure S1D. BLAST searches against the N. attenuata genome and transcriptome revealed that the sequence used for RNAi targeting NaTOC1 does not overlap by >18 bp with any region outside of the NaTOC1 coding sequence, making off-target effects of the RNAi construct unlikely, as RNAi sequences have been shown to require >21-nt of contiguous exact matches for the formation of secondary small interfering RNAs (Schwab and Voinnet, 2010). The irTOC1-205 line was used for all experiments described in this study. Importation and release of transgenic plants were carried out under Animal and Plant Health Inspection Service (APHIS) import permit nos. 07-341-101n (EV) and 11-350-102m (irTOC1) and release permit no. 16-013-102r. Field growth conditions were described by McGale et al. (2018). Briefly, seedlings were germinated on Gamborg’s B5 media under illumination from fluorescent lights (40-W Plant & Aquarium and 18-W Warm White, both by GE) at ambient temperatures at the field station. One week after germination, seedlings were grafted (see below “Micrografting,” and Fragoso et al., 2011). One to two weeks after grafting, seedlings with four visible leaves were transferred into previously hydrated 50-mm peat pellets (Jiffy 703; www.jiffypot.com) treated with Borax to provide boron, an essential micronutrient (1:100 dilution of a 1.1 g L⁻¹ stock solution) and adapted over two weeks to the field conditions of high light intensity and low relative humidity by keeping seedlings first in shaded, closed translucent plastic 34-quart boxes (Sterilite), then opening the boxes, and subsequently transferring open boxes to partial sunlight in mesh tents (Tatonka). Adapted size-matched seedlings were transplanted into an irrigated field plot at the Lytle Ranch Preserve, Santa Clara, Utah, in April 2016.

Water Limitation Treatments

Glasshouse plants were grown in sand and kept at 18°C to 35°C under 16:8 h of light:dark cycles, with the light period lasting from 6:00 am to 10:00 pm. For the first glasshouse drought experiment (Fig. 1), plants were potted and individual baselines measured of the combined pot, plant, and soil mass; this mass was used as a baseline of 0 g of water. Thereafter, gravimetric water content was measured per pot and per day, and plants were watered with 2× the previous day’s water use. Treated plants were not watered on those days, and allowed to dry-down to 0 g of water in excess of initial water-holding capacity per pot, before watering recommenced. Plants were sampled on the day that 0 g of water was reached.

For the second glasshouse drought experiment (Figs. 4 and 5), plants were watered using table flood irrigation, and treated plants were removed from water for 7–9 d to induce drought stress. Leaves at the +1 nodal position (youngest fully extended leaves) were sampled without the midvein, and whole roots were homogenized for root sampling. All tissue was stored at −80°C before RNA or phytohormone extraction. Plant water status was monitored using leaf angle at midday, after two preliminary experiments. The first showed that midday leaf angles in N. attenuata track the volumetric soil moisture contents and leaf RWCs (Supplemental Fig. S6), which are measures of plant water availability. The second showed that when leaf angles reached ∼45°C to 50°C at midday, a strong increase in leaf ABA accumulation and a strong shift in ABA-responsive transcripts were observed (Supplemental Fig. S7). RWC was calculated using the equation displayed in Supplemental Figure S8B: Cut leaves were weighed immediately to obtain leaf fresh mass, and placed in individual containers filled with distilled water abaxial side down for 2 h, after which turgid mass were obtained. Leaves were subsequently dried in a drying oven for 6 h to obtain dry mass (Turner, 1981).

In the field (Fig. 2), plants were watered using a drip irrigation system approximately once a week, as needed to maintain soil moisture levels, according to soil moisture measurements made in representative locations throughout the field plot using a 53-mm diameter by 40-cm working-length split tube sampler (Fig. 2B; Product no. 04.17, Eijkelkamp Soil & Water). The split tube sampler was used on most days throughout the latter half of the field season to quantify soil water status. Soil core samples were taken from 5, 10, 15, and 30 cm below the surface, weighed, and allowed to dry in closed, aerated boxes outside (air moisture of ∼17% to 20%) for 3–5 d until soil reached a stable weight, and then reweighed. Water-limited plants were disconnected from the drip irrigation system 23 dpp, after an initial establishment phase, and allowed to dry-down naturally until the conclusion of the experiment

Gas Exchange and WUE Calculations

Gas exchange and chlorophyll fluorescence measurements were conducted with plants in the early elongating stage of growth with the first fully extended stem leaves. Measurements were conducted using a LI-6400XT Infrared Gas Analyzer (Li-Cor Bioscience), with an integrated fluorometer in the leaf chamber, and WUE was calculated as assimilation rate divided by transpiration.

Leaf Water Potential

Leaf water potential was measured using a model no. 615 Pressure Chamber Instrument (PMS Instrument). Briefly, the lowest stem leaves of fully mature plants were removed at dawn (5:30 am to 6:30 am) for measurements using the chamber under field conditions near the end of the growing season, when drought treatment water levels were at their lowest.

Micrografting

Seven-day–old seedlings were micrografted as described by Fragoso et al. (2011). EV (E) and irTOC1 (T) genotypes were used as scions or rootstocks yielding EE, TT, and ET grafts (where the first letter refers to the scion genotype and the second letter to the genotype of the rootstock). The average grafting success was 52% under glasshouse conditions (P > 0.05 between genotypes, ANOVA followed by Tukey’s honestly significant difference posthoc) and 59% under field conditions (P < 0.05 among genotypes: 68% EE, 49% ET, and 54% TT, ANOVA followed by Tukey HSD posthoc).

Transcript Abundance

One-hundred fifty milligrams of leaf or 300 mg of root tissue were harvested, and RNA was extracted with TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. Unless otherwise displayed (e.g. transcript abundance kinetics in Supplemental Fig. S1), all tissues for transcript abundance were harvested at the TOC1 peaking time in EV plants, 7:00 pm under LD conditions, 13 h into the light cycle. Total RNA was quantified using a NanoDrop (Thermo Fisher Scientific) and complementary DNA was synthesized from 500 ng of total RNA using RevertAid H Minus reverse transcriptase (Fermentas) and oligo(dT) primer (Fermentas). Reverse transcription quantitative PCR (RT-qPCR) was performed in a Mx3005P PCR cycler (Stratagene) using SYBR GREEN1 kit (Eurogentec). The N. attenuata actin gene homolog NIATv7_g21364 and the N. attenuata elongation factor 1a were used as a standard housekeeping gene for normalization. The sequences of primers used for RT-qPCR are provided in Supplemental Table S4. All RT-qPCR data were normalized using the ΔΔ-Ct method.

ABA Extraction and Quantification

Phytohormone analysis of leaf material was performed on a UPLC-MS/MS (EvoQ Elite Triple quad-MS; Bruker Daltonics) after extraction in precooled acidified methanol and column purification as described by Schäfer et al. (2016). All tissues for phytohormone analysis were harvested at the TOC1 peaking time for EV plants, 7:00 pm, under LD conditions.

Microarray Analysis

Six biological replicates for each treatment and genotype were used for RNA isolation. Total RNA was isolated with TRIzol reagent and labeled cRNA with the Quick Amp Labeling Kit (Agilent Technologies). Each sample was hybridized on single color technology arrays (60-k 60-mer oligonucleotide microarray designed for N. attenuata transcriptome analysis, GEO accession no. GPL13527; http://www.agilent.com). A microarray scanner (model no. G2565BA; Agilent Technologies) and Scan Control Software (Agilent Technologies) were used to obtain intensity of the spots. All microarray data with each probe name were deposited in the NCBI GEO database. The initial normalization and extraction of low-expressed probes was performed using the “R” package “limma” (Ritchie et al., 2015), followed by GSEA and overrepresentation analysis of samples within tissues and treatments using the “R” package “CEMiTools” (Russo et al., 2018).

Statistical Analyses

All data were analyzed using “R” v3.4.2 (R Development Core Team, 2008) and RStudio v1.0.153 (RStudio Team, 2015). Datasets were fit to linear, generalized least squares, or generalized linear models after outlier removal and homoscedasticity and normality tests had been applied before model reduction. Pairwise posthoc comparisons were made using the “R” package tool “lsmeans” (Lenth, 2016) or using Tukey HSD tests after significant results in a two-way ANOVA.

Root Image Analysis

Pictures of roots were taken at the end of the field experiment when harvesting biomass. Pictures were analyzed using the Fiji distribution of the imaging software ImageJ (Schindelin et al., 2012; Schneider et al., 2012). Scale bars were applied using the “Analyze” → “Set Scale” functionality to measure a standardized paper label in each image (3.1 cm × 1.5 cm).

Accession Numbers

Microarray information is available under GEO accession no. GPL13527. Sequence data from this work can be found in GenBank under the following accession numbers: NaTOC1 (LOC109210941), NaLHY (LOC109231023), NaPRR5 (LOC109223559), NaPRR9 (LOC109231058), NaGI (LOC109234610), Flowering Locus T (LOC109243482), resveratrolo-methyltransferase (LOC109207600), NaABAR (LOC109229433), NaRAB18 (LOC109220601), NaPIF1 (LOC109230623), NaPIF3a (LOC109236817), NaPIF3b (LOC109234067), NaPIF4 (LOC109228555), NaPIF7 (LOC109210563), NaPHYB1 (LOC109232144), NaPHYB2 (LOC109216344), NaPHYA (LOC109226699), NaCRYP1a (LOC109214267).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. NaTOC1-silenced plants exhibit clock-silenced phenotypes specific to TOC1.

• Supplemental Figure S2. Synchronized drought did not affect flowering or elongation times of irTOC1 or EV plants.

• Supplemental Figure S3. Effect of irTOC1 in whole plants or roots only and of drought event timing on seed number and seed mass.

• Supplemental Figure S4. EV/irTOC1 micrografting yields plants silenced in TOC1 in roots, but not shoots.

• Supplemental Figure S5. Leaf water potential and root morphology under field conditions.

• Supplemental Figure S6. Relationship between water status and leaf angle under glasshouse drought conditions.

• Supplemental Figure S7. The effects of irTOC1 on transcriptional responses of plants exposed to drought in the glasshouse are dependent on developmental time.

• Supplemental Figure S8. Leaf angle and RWC of drought-stressed lines analyzed in Figures 4 and 5.

• Supplemental Table S1. “CEMiTool” analysis GSEA results.

• Supplemental Table S2. “CEMiTool” ORA: GO annotations and P-value results.

• Supplemental Table S3. Characterization of ovLHY, ifFT3, and irCRYP1a lines.

• Supplemental Table S4. Primer sequences.