Seeds of Arabidopsis [Arabidopsis thaliana (L.) Heynh.] containing the T-DNA insertion allele ztl-3 (SALK_035701) in the Columbia-0 (Col-0) background (Jarillo et al., 2001; Kim et al., 2003) were obtained from the Salk collection¹ at the Arabidopsis Biological Resource Centre (ABRC), Ohio State University, Ohio, United States via the Nottingham Arabidopsis Stock Centre (NASC), Nottingham, United Kingdom. Seed of the ost1-3/snrk2.6/srk2e mutant (SALK_008068; Yoshida et al., 2002) was provided by Dr. Kazuko Yamaguchi-Shinozaki (University of Tokyo, Japan) to the Israelsson-Nordström laboratory. The Webb laboratory received seed of the ztl-4, fkf1-2 and lkp2-1 triple mutant (Baudry et al., 2010) from Dr. Steve Kay (Keck School of Medicine, University of Southern California, United States) and of the prr5-11 mutant (Nakamichi et al., 2005) from Dr. Takeshi Mizuno (Nagoya University, Japan). The ztl-1 mutant in the C24 background, the ztl-21 mutant in the Wassilewskija-2 (Ws-2) background and prr5-1 (SALK_006280) have all been described previously (Eriksson et al., 2003; Kevei et al., 2006; Kiba et al., 2007).

The ztl-3 and ost1-3 double mutant in the Col-0 background were generated using ztl-3 as the pollen recipient and ost1-3 as the pollen donor. The ztl-3, prr5-1 and ost1-3 triple mutant, and combinations thereof, were generated using ztl-3; prr5-1 (Norén et al., 2016) as the pollen recipient and ost1-3 as the pollen donor.

To complement the loss-of-function mutant ztl-3, we generated transgenic ztl-3 plants expressing ZTL under the control of the 35S CaMV constitutive promoter. Agrobacterium tumefaciens GV3101 (pMP90RK-pSoup) was transformed with the p35S::HA::ZTL construct in pGreenII 0229. The construct was introduced into ztl-3 mutant Arabidopsis plants by floral dipping (Bechtold et al., 1993). The pGreenII 0229 construct contains the bialaphos resistance gene (BAR) that confers resistance to glufosinate-ammonium (Hellens et al., 2000). Seeds from dipped plants were sterilised and plated on full-strength Murashige and Skoog (MS) medium supplemented with vitamins (Duchefa, BH Haarlem, Netherlands), 3% w/v sucrose (SIGMA-Aldrich, Saint Louis, MO, United States) and 0.8% agar (E1674, Duchefa), pH 5.7, plus 10 mg/l glufosinate-ammonium (SIGMA-Aldrich). Resistant plants were allowed to self-fertilise and next-generation seeds were again screened on glufosinate-ammonium. Expression of the p35S::HA::ZTL construct was confirmed by Western blotting (not shown), prior to phenotyping.

Radicle emergence assays were conducted using seeds sown on plates containing half-strength MS medium with 0.8% agar, pH 5.7. Seeds of the different genotypes were grown and harvested at the same time under the same conditions and stored for least 3 months after harvest as seed age and storage affect germination responses. Seeds were surface-sterilised by consecutive washes with 15% hypochlorite, 70% ethanol with 0.1% tween and 95% ethanol before plating. Plated seeds (60 seeds per genotype per replicate; Supplementary Figure S1) were stratified at 4°C for 2 days and transferred to a growth chamber (LD 16:8 at 22°C; light intensity during day: 150 μmol m⁻² s⁻¹) to initiate germination. Germination was determined by the appearance of the testa, endosperm rupture and radicle protrusion (Wu et al., 2012). Radicle emergence was scored every 12 h, starting at 24 h, and calculated as a percentage of the plated seeds.

In vitro-cultivated, rooted cuttings of PttZTL1,2 RNA interference (RNAi) lines and wild-type (WT) Populus tremula L. × P. tremuloides Michx. cv. T89 (Nilsson et al., 1992; Populus; Ptt) plants were potted in a 3:1 mix of fertilised peat and perlite, and established under long day cycles consisting of light:dark (LD) 18:6 at constant temperature (18°C) and 80% relative humidity for 4 weeks; light intensity during the day was 200 μmol m-² s⁻¹ (Osram Powerstar HQI-T 400 W/D lamps; Osram, München, Tyskland). After this point, temperature, humidity and irradiance during the day were maintained but the photoperiod was shortened to LD 15:9 at 18°C, keeping the time of dawn unchanged.

A sub-set of lines (1, 3, 4, 5 and 7) with variable levels of ZTL1,2 downregulation were selected and samples for analysis taken twice a week subsequent to the shift from LD 18:6 to LD 15:9. Growth cessation, which refers to the elongation of shoots ceasing, was scored according to a predefined scale (score 2), as previously described (Ibáñez et al., 2010).

In Arabidopsis, sensitivity to ABA, which induces stomatal closure, increases in the late afternoon (Correia and Pereira, 1995). This timing correlates with high levels of ZTL (Kim et al., 2007). We therefore investigated the sensitivity of ztl mutants to ABA. As defects in the different protein domains affect different aspects of ZTL function (Kevei et al., 2006; Kim et al., 2007), we measured sensitivity to ABA in ztl-3 (complete loss-of-function mutant), ztl-1 (Kelch domain mutant) and ztl-21 (LOV domain mutant). Although ABA evoked stomatal closure in ztl mutants, the response was significantly greater in wild-type (WT) plants (Figures 1A–C).

We measured stomatal conductance in ztl-3 and ost1-3 leaves using a handheld SC-1 leaf porometer (Decagon). Stomatal conductance levels were higher in ztl-3 and ost1-3 than in WT controls (Figure 1D). ost1-3 mutants have the same stomatal density as WT plants (Engineer et al., 2014; Jalakas et al., 2018). Stomatal density was similar in WT (208 stomata/mm²) and ztl-3 plants (210 stomata/mm²; Student’s t-test p = 0.727, n = 3 biological replicates, each containing 71–93 measurements/genotype). Thus, the increased water loss of ztl-3 and ost1-3 mutants was not caused by an increase in the number of stomata. The higher stomatal conductance in ztl-3 (Figure 1D) observed is consistent with their reduced response to ABA (Figures 1A–C). Overall, the ABA sensitivity and stomatal conductance phenotypes of ztl-3 resembled the strong ost1-3 phenotype.

A detached-leaf assay revealed that ztl-3 and ost1-3 leaves have significantly higher rates of water loss than WT leaves at ZT 8–9 (Figure 1E).

To further determine the role of ZTL in the drought response, we measured the water loss phenotype of ztl-3 plants overexpressing ZTL under the control of the 35S CaMV promoter (p35S::HA::ZTL). There were no differences in the rates of water loss between WT plants and three independent T₃ lines overexpressing ZTL in the ztl-3 background (Figure 1F); thus, ZTL could rescue the leaf water loss phenotype of ztl-3. This indicated that ZTL and OST1 are both needed for stomatal closure under stress.

We also analysed seed germination time (determined by radicle emergence), a developmental phase transition, to establish whether ZTL and OST1 showed shared effects across biological responses (Penfield and Hall, 2009). Both ztl-3 and ost1-3 mutants showed a similar significant delay in radicle emergence compared to WT seeds (Supplementary Figure S1).

Circadian components and ABA-signalling pathways are largely conserved between Arabidopsis and Populus (Ibáñez et al., 2010; Kozarewa et al., 2010; Cai et al., 2017; Yu et al., 2017; Tylewicz et al., 2018; Rigoulot et al., 2019). We hypothesised that ZTL’s roles in regulating the circadian clock and physiological responses to drought and ABA would also be conserved. The role of ZTL was examined in transgenic Populus trees in which expression of both PttZTL1 and PttZTL2 was downregulated by RNAi (PttZTL1,2 RNAi lines). RNAi line 5, which had a strong reduction in PttZTL (Figures 2A,B; Supplementary Figure S2), leading to earlier growth cessation (Supplementary Figure S3), and RNAi line 7, which also showed a strong, significant reduction of PttZTL transcript (Figures 2A,B; Supplementary Figure S2), were selected for further analysis. Microarray expression profiles from short period (lhy-10 RNAi) and WT Populus trees (Edwards et al., 2018) were examined for comparison (Supplementary Figure S4), together with an RT-qPCR analysis of time-series data from WT trees (Supplementary Figure S5). These data show PttZTL1,2 expression appears disrupted in lhy-10 trees in LD cycles (Supplementary Figure S4); thus, disruption of PttLHY 1 and PttLHY2, expressed in the morning clock loop, apparently affects expression of PttZTL1 and 2, an evening gene, in Populus as in Arabidopsis. PttZTL expression is rhythmic in WT Populus leaves (Supplementary Figure S5).

Clock function was analysed in PttZTL1,2 RNAi lines 5 and 7. Their circadian periods were measured in detached leaves using delayed fluorescence from photosystem II (Gould et al., 2009; Johansson et al., 2015). The normalised fluorescence traces (Best Y) showed both RNAi lines had significantly longer circadian periods than WT trees (Figure 2A; Table 1). The associated relative amplitude error (RAE) values indicate the level of rhythmicity associated with an individual leaf (Figure 2B); values ≥0.6 (dotted line) indicate arrhythmia. These results are in line with the downregulation of PttZTL1,2 in RNAi lines 5 and 7 (Supplementary Figures S2A,B). Downregulation of PttZTL1,2 expression resulted in a slower running but still strongly rhythmic circadian clock, as shown by the longer periods and low RAE values, indicating that the phenotypes of PttZTL1,2 RNAi trees resembled those of Arabidopsis ztl mutants (Kevei et al., 2006).

Thus, these lines provided suitable material for further studies comparing ZTL-dependent function in ABA-related stress responses across plant species.

To investigate whether ABA-induced stomatal closure was conserved between Arabidopsis and Populus, we confirmed downregulation of ZTL at ZT 8–9 in Populus (i.e. at dusk, when ZTL levels peak and sensitivity to ABA are high in WT Arabidopsis) using an additional reference gene (Ef1a; Figure 3A). This confirmed that PttZTL1 and PttZTL2 were reduced to at least ~40% of WT levels in both lines at ZT 8–9. Further, we measured stomatal aperture in leaves from Populus trees grown in vitro (Figure 3B). Stomatal closure in response to ABA was reduced in RNAi lines 5 and 7, relative to WT plants. In addition, stomatal conductance in leaves from both PttZTL1,2 RNAi lines was significantly higher than that of WT leaves (Figure 3C). These results resembled those from Arabidopsis ztl mutants (Figures 1A–D).

Next, we investigated water loss using detached-leaf assays. Both PttZTL1,2 RNAi lines 5 and 7 showed significantly higher rates of water loss than WT plants (Figure 3D), similar to the Arabidopsis ztl-3 mutant (Figures 1E,F). Stomatal density measurements revealed leaves from PttZTL1,2 RNAi line 5 (135 stomata/mm²) differed significantly from WT leaves (141 stomata/mm²; Student’s t-test; p = 0.04; n = 3 biological replicates, each containing 120 to 144 measurements); however, there was no significant difference in stomatal density between WT leaves and leaves from PttZTL1,2 RNAi line 7 (141 stomata/mm²; Student’s t-test p = 0.89; n = 3 biological replicates, each containing 120 to 144 measurements). These results indicated the increased water loss resulted from greater stomatal openness, not changed stomatal density. The lower sensitivity to ABA, increased stomatal aperture and conductance, and higher water loss observed in both PttZTL1,2 RNAi lines (Figure 3) indicated they phenocopied Arabidopsis ztl mutants (Figure 1), supporting a conserved role for ZTL between the two species.

Given stomatal closure in response to ABA was impaired in ztl mutants (Figures 1A–C) and the similar changes in stomatal regulation in Arabidopsis and Populus trees with reduced ZTL function (Figures 2, 3), we evaluated whether ZTL modulated expression of early and late ABA-signalling components and ABA-responsive genes in Arabidopsis. We analysed expression of key ABA response and signalling genes in ztl-3 and WT plants treated with ABA for 3 h (Figure 4). PYL5 was selected as a representative of an ABA reception gene, ABI2, HAB1, OST1, ABI5, ABF3 and ABF4 as examples of early and progressing ABA-signalling genes, and RAB18 and RD29A as late ABA-responsive genes (Santiago et al., 2009; Raghavendra et al., 2010; Gonzalez-Guzman et al., 2012).

Analysis by two-way ANOVA indicated there were no treatment (T), genotype (G) and treatment × genotype (T × G) effects on PYL5 expression (Figure 4A). Analysis of expression of ABI2 and HAB1 (Figures 4B,C), two negative regulators of ABA signalling (Saez et al., 2004, 2006), revealed that genotype had a significant effect on ABI2, with a significant reduction in transcript level in ztl-3 (Figure 4B), but not on HAB1, which only showed a treatment effect (Figure 4C). Both treatment and genotype had significant effects on OST1, expression of which was significantly reduced in ztl-3 relative to wild-type plants (Figure 4D). Expression of the later ABA-signalling gene ABI5 was severely diminished in response to ABA in ztl-3 plants, with significant effects of both treatment and genotype, as well as a treatment × genotype interaction (Figure 4E). Only treatment affected ABF3 expression (Figure 4F), but both treatment and genotype affected expression of ABF4 (Figure 4G). Expression of another ABA-responsive gene, RAD29A, showed significant treatment and genotype effects (Figure 4H). Both treatment and genotype had significant effects on expression of RAB18 (Figure 4I), and these factors interacted to produce a further significant effect. All these data show that ZTL was required for ABA sensitivity and significantly promoted the expression of major ABA-signalling components (Table 2).

Given the function of ZTL in stomata closure and in ABA-induced gene expression, we tested if it has the capacity to interact with OST1 when expressed in plant cells. Co-immunoprecipitation (Co-IP) assays following expression of tagged ZTL and OST1 proteins in protoplasts revealed an interaction between ZTL and OST1 (Figure 5A). The ratios of signal from Co-IP bands to input, measured from three experiments, showed a strong signal from HA-OST1, when it was coexpressed with Myc-OST1 (Figure 5B). Co-IP of the known interactors, ZTL and TOC1, showed a similar result (Supplementary Figures S6A,B; Johansson et al., 2011).

The effect of ZTL on OST1 stability was investigated using mesophyll protoplasts obtained from the triple mutant ztl-4, fkf1-2 and lkp2-1 (Supplementary Figure S7). This mutant lacks all three members of the ZTL F-box protein family [ZTL, FLAVIN-BINDING, KELCH REPEAT, F BOX 1 (FKF1) and LOV KELCH PROTEIN 2 (LKP2)] (Baudry et al., 2010). ZTL-like activities were diminished in this background, as expected. No ZTL-dependent degradation of OST1 was detected in protoplasts transfected with OST1 and ZTL, regardless of whether the proteasome inhibitor MG132 was present (Supplementary Figures S7A,B, S8). The effect of proteasome inhibition was also tested using TOC1, and in contrast, coexpression of TOC1 and ZTL led to lower TOC1 expression at time 0 (Supplementary Figures S7C,D). Although the effect on TOC1 in blue light was not entirely consistent with the current model, wherein ZTL acts to degrade TOC1 in the dark (Más et al., 2003), the lack of activity of the ZTL family members FKF1 and LKP2 in the triple mutant protoplasts affected TOC1 expression (Supplementary Figures S7C,D), consistent with the results of Baudry et al. (2010). OST1 stability, on the other hand, showed no effect of ZTL coexpression or proteasome inhibition (Figures 6A,B), which supports our conclusion that ZTL does not regulate OST1 protein levels. OST1 stability was not negatively affected in protoplast suspension cultures, regardless of the presence of ZTL, light or CHX (Supplementary Figure S8).

The genetic data suggested that ZTL and OST1 both severely and similarly affected stomatal closure in response to ABA (Figure 1E). In order to determine if additional factors were involved, we investigated the role of PRR5, a ZTL substrate that affects ABA metabolism and induced responses (Yang et al., 2021). We adjusted ABA treatment time to match the circadian period of each genotype (ztl-3: 28 h; prr5-11: 23 h and WT: 24 h) and scored the stomatal ratio at Circadian Time (CT) 8–9 for each genotype to confirm that the phenotypic change resulted from ABA treatment rather than circadian mistiming. Under these conditions, the stomata of ztl-3 mutants failed to close following ABA treatment, while stomata of prr5-11 closed more readily (Figure 7A). The responses of both mutants to ABA differed significantly from WT plants, albeit in opposite directions, consistent with the changes in endogenous period.

Given that ZTL mediates proteasomal degradation and several ztl mutants show alterations in PRR5 protein stability and other alterations that may affect PRR5 function (Kiba et al., 2007; Fujiwara et al., 2008), we hypothesised that altered ZTL-dependent regulation, for instance stabilisation of PRR5 in the ztl-3 mutant, might increase stomatal opening. Consistent with this, PRR5 appeared to be required for normal stomatal responses. Although prr5-1 mutants remained sensitive to ABA and showed stomatal closure in response to ABA treatment, in the absence of exogenous ABA, their stomata were more open than those of WT plants (Figure 7A). This may result from an increase in ZTL levels in prr5-1 in the absence of ABA. We found a clear interaction between PRR5 and OST1 in plant cells (Figure 7B).

To understand how OST1, ZTL and PRR5 interacted to regulate stomata (i.e. responsiveness to ABA), we performed water loss assays using detached leaves from a series of single, double and triple mutants carrying different combinations of the ztl-3, ost1-3 and prr5-1 alleles to determine their physiological responses to drought (Figure 6A). The ztl-3 and ost1-3 single mutants had a similar phenotype that differed from that of WT plants, consistent with previous data implicating both ZTL and OST1 in stomatal regulation; moreover, each single mutant differed significantly from the double mutant, ztl-3 and ost1-3. The increased severity of the loss-of-function phenotype in the ztl-3 and ost1-3 double mutant indicated both ZTL and OST1 were required for stomatal control and potentially acted additively to each other.

Leaves from prr5-1 single mutants showed rates of water loss similar to WT with slight deviations (Figures 6A,B). The double mutant ztl-3 and prr5-1 behaved as WT (Figure 6B). The double mutant prr5-1 and ost1-3 behaved like ost1-3 (Figure 6A). The ztl-3 and ost1-3 double mutant had the strongest water loss phenotype, indicating that retaining PRR5 in the absence of OST1 and ZTL partially blocked regulation of stomatal aperture by ABA. The ztl-3, prr5-1 and ost1-3 triple mutant showed a significantly lower rate of water loss than the ztl-3 and ost1-3 double mutant between 90 and 150 min, and again at 180 min, which is consistent with the suggestion that PRR5 is involved in ABA signalling in the absence of ZTL and OST1.

These results taken together suggest that inhibition of the ABA-signalling pathway by PRR5 is opposed by ZTL and OST1. The ztl-3, prr5-1 and ost1-3 triple mutant had a significantly stronger water loss phenotype than the ost1-3 single mutant, indicating that ZTL can oppose the inhibitory effect of PRR5 in the absence of OST1. By contrast, the prr5-1 and ost1-3 double mutant and the ost1-3 and ztl-3 single mutants all showed very similar phenotypes that were less extreme than the ztl-3, prr5-1 and ost1-3 triple mutant (Figure 6A), indicating the importance of the interactions between ZTL, OST1 and PRR5.

In Arabidopsis, light and the circadian clock act via ZTL to regulate the daily pattern of stomatal opening (Somers et al., 1998; Salomé et al., 2002; Dodd et al., 2004). Many metabolic processes and enzymes associated with photosynthesis are under circadian control (Dodd et al., 2014). We show that ZTL was also required for circadian clock function (Figure 2; Table 1) in Populus, and ZTL regulated stomatal closure in both Arabidopsis and Populus (Figures 1, 3). In Arabidopsis, both OST1- and ABA-dependent gene expression and integration of ABA-signalling require ZTL (Figure 4; Table 2). ZTL regulated the physiological responses of guard cells to ABA and drought by acting in synergy with OST1 (Figures 1, 5, 6).

OST1 is a central component of ABA signalling and stomatal closure. The increased rate of water loss from Arabidopsis ztl mutants as well as their impaired stomatal regulation in response to ABA (Figure 1) suggests a requirement of ZTL for normal OST1 function. The similar phenotypes observed in Populus trees with reduced ZTL function (Figure 3) indicates a similar mechanism operates in Populus.

We showed that ZTL bound directly to OST1 in plant cells. Loss-of-function prr5-11 mutants showed an opposite stomatal phenotype to ztl-3, a null mutant (Somers et al., 2004) that does not produce mRNA or protein (Figure 7A). The water loss phenotype of the prr5-1 mutant (a low level of water loss comparable to WT) was as expected, given the prr5-11 mutant showed high levels of stomatal closure in the presence of ABA (Figure 7A). In contrast, the stomata of the ztl-3 and ost1-3 single mutants remained more open in response to ABA (Figure 7A), which matched the high levels of water loss shown by these mutants (Figures 6A,B). The genetic data confirmed that a triple mutant that combined the loss-of-function prr5-1 allele with the ztl-3 and ost1-3 alleles had a lower rate of water loss than a ztl-3 and ost1-3 double mutant but did not completely phenocopy it (Figure 6A). This may result from TOC1 accumulation, in the triple mutant as TOssC1 is a substrate of ZTL that affects ABA responses (Legnaioli et al., 2009); moreover, overexpression of TOC1 slightly increases stomatal apertures and reduces water use efficiency (Legnaioli et al., 2009; Simon et al., 2020).

We hypothesised that an increased level of PRR5 in the absence of ZTL would exacerbate the water loss phenotype. Thus, we carried out physiological, biochemical and genetic assays to probe PRR5 function in stomatal closure, rate of water loss and interaction with OST1 (Figures 6, 7). Direct interactions between ZTL and OST1 (Figure 5), and between PRR5 and OST1 (Figure 7B), occurred in plant cells. In addition, the level of PRR5 affected stomatal aperture (Figure 7A). Removing both OST1 and ZTL function (ztl-3 and ost1-3 double mutant) produced a strong water loss phenotype (Figure 6A); however, a triple mutant (ztl-1, prr5-1 and ost1-3) showed a less extreme water loss phenotype (Figure 6A), indicating that, in the absence of ZTL and OST1, PRR5 partially blocked the ABA pathways controlling stomatal aperture. The interactions between OST1, ZTL and PRR5 are thus essential for ABA-signalling and water regulation under stressful conditions. ABA treatment may produce a change in circadian period which is dependent on genotype and thus, the effect of ZTL on ABA-signalling gene expression may result from a phase shift (Legnaioli et al., 2009; Liu et al., 2013), although, in our hands, treatment with 20 μM ABA did not significantly alter the phase in the first 24 h (Supplementary Figure S9) or period of WT and ztl-21 seedlings (Supplementary Figure S9; Supplementary Tables 1, 2).

These findings suggest that both ZTL and PRR5 interact with OST1 to regulate stomata and ABA-dependent responses. As the loss of PRR5 did not influence the ost1-3 water loss phenotype (Figure 6A), PRR5 appears to act upstream of OST1. In contrast, the loss of PRR5 strongly influenced the water loss phenotype of ztl-3 mutants (Figure 6B). PRR5 acting downstream of ZTL but upstream of OST1 would also explain the phenotype of the ztl-3, prr5-1 and ost1-3 triple mutant. These physical and genetic interactions provide a biochemical framework linking ZTL directly to ABA-regulated components. ZTL thus acts with OST1 as a circadian short-cut to help phosphorylate and/or degrade PRR5 enabling osmotic regulation of guard cells and connecting the clock with responses essential for water conservation.

There is a significant overlap between ABA and cold signalling pathways and control by the circadian clock (Eriksson and Webb, 2011); for instance, the circadian MYB-transcription factors CCA1 and LHY contribute to cold responses in both Arabidopsis and Populus sp. (Espinoza et al., 2010; Ibáñez et al., 2010; Dong et al., 2011). The circadian clock regulates ABA signalling via transcriptional changes induced by TOC1 (Legnaioli et al., 2009; Huang et al., 2012). Mutations at the circadian clock-associated EARLY BIRD/NFX1-LIKE 2 locus also increase resistance to salt and drought stress (Lisso et al., 2006), as well as inducing hypersensitive responses to ABA (Lisso et al., 2012). Genome-wide analysis of Arabidopsis recently showed that LHY controlled expression of ABA biosynthesis and receptor genes, as well as other aspects of signalling (Adams et al., 2018). We investigated several of these genes, including ABI2, OST1, ABI5, ABF4 and RD29A, and found ZTL affected their expression (Figure 4).

The receptors regulating ABA responses have increased in number since plants first colonised dry land (Umezawa et al., 2010) in response to the need to manage water status and detect and respond to heat and drought stresses. Such stresses are largely managed by controlling stomata. Stomata provide a means of CO₂ entry, thus enabling photosynthesis, and also of controlling water loss through transpiration. They thus are critical regulators of plant growth.

Stomatal conductance is a crucial trait affecting water status and photosynthetic capacity that directly impacts biomass accumulation of trees. Stomatal conductance is under diurnal regulation in field-grown Eucalyptus sp., and thus likely controlled by the clock (Resco de Dios et al., 2013). We found previously that Eucalyptus sp. exhibit robust circadian rhythms under constant conditions (Johansson et al., 2015), and our present data suggest that ZTL acts in similar ways in both Arabidopsis and Populus to link the clock to stomatal control. Thus, the important roles of ZTL and OST1 in controlling stomata is conserved across species.

Our work highlights plants’ dependence on the circadian clock to respond to drought stress. Studies in Populus balsamifera (Pb) suggest that PbZTL2 is under local climatic selection, as are ABA-related signalling components, PbGIs and additional clock-associated genes (Keller et al., 2017). In addition, in Populus trichocarpa, PtPRR5, PtPRR7 and other clock genes are associated with biomass, phenology and physiological traits in Genome-Wide Association Studies (GWAS) and appear to have undergone selection (McKown et al., 2014); for example, PtPRR5 is part of an adaptive introgression of genes on Chromosome 15 transferred from P. balsamifera into P. trichocarpa (Suarez-Gonzalez et al., 2016).

OST1 is a kinase and thus a target for both phosphorylation and ubiquitination (Kim et al., 2013). The partnership between ZTL and OST1 underlies changes in ubiquitination and/or phosphorylation. Their associations with PRR5 may facilitate localisation of PRR5 to the nucleus and its interactions with, for instance, ABI5. The circadian clock PRR proteins are expressed sequentially between dawn and dusk in the order PRR9, PRR7, PRR5 and TOC1/PRR1. This may provide a set of ‘cogs’ enabling interactions with OST1, with or without ZTL, to be integrated into circadian clock and ABA-signalling pathways and respond to abiotic stresses occurring at different times across the day. Other circadian genes including PRR7 influence ABA signalling (Liu et al., 2013). Future efforts to resolve the underlying mechanisms controlling stomatal regulation and stress tolerance will benefit from consideration of the interaction between OST1, ZTL and the PRRs. Further studies in both Arabidopsis and Populus will determine the detailed mechanisms controlling diel stomatal closure and ABA-signalling responses.