Crosstalk between the Circadian Clock and Innate Immunity in Arabidopsis

1Plants are challenged by various pathogens on a daily basis. Accumulating evidence implicates a role of the circadian clock in regulating plant innate immunity. The circadian clock is the internal time measuring machinery important for plant growth and development. However, our understanding of the molecular basis of how the circadian clock controls plant innate immunity is still in its infancy.

2Plants have evolved various mechanisms, some pre-formed and others induced, to ward off pathogen invasion. An example of pre-formed surface structures is the stomate, the natural opening important for photosynthetic gas exchange. This opening can provide a portal for pathogens to enter leaves; however, plants can also control the aperture of stomata to physically limit pathogens,. One type of induced defense is activated when plants recognize pathogen-associated molecular patterns (PAMPs), which are conserved molecules or structures present in groups of related microbes. This defense, also termed PAMP-triggered immunity (PTI), can be highly effective against non-adapted pathogens and provides a basal level of defense even against adapted pathogens,. Another type of induced defense is activated by plant resistance (R) proteins, which specifically recognize secreted pathogen effectors and subsequently activate effector-triggered immunity (ETI). ETI, also termed R gene-mediated resistance, is a stronger and faster elaboration of PTI, and frequently results in hypersensitive cell death at the infection site,,. The small molecule salicylic acid (SA) has been linked to signal transduction in PTI and ETI,,.

3The circadian clock has profound influence on the fitness of organisms,,,,,. The core of the circadian clock is the central oscillator, which in Arabidopsis, is composed of multiple interconnected negative feedback loops that orchestrate biological adjustments independently of external stimuli,. Of these clock components, CIRCADIAN CLOCK ASSOCIATED1 (CCA1) and its close homolog LATE ELONGATED HYPOCOTYL (LHY) are transcription factors that are involved in multiple feedback loops and function synergistically to regulate clock activity,,.

4The role of the circadian clock in controlling plant innate immunity has long been proposed based on circadian-regulation of defense gene expression,,,,. Direct evidence from several research groups has recently emerged to support such a role of the circadian clock. Under free running conditions, wild type Arabidopsis exhibits temporal oscillations in susceptibility to Pseudomonas syringae infection, which are disrupted by overexpression of CCA1. Misexpression of several clock genes, including CCA1, compromises resistance to the bacterial pathogen Pseudomonas syringae and/or to the oomycete pathogen Hyaloperonospora arabidopsidis (Hpa),,. Interestingly, although lhy mutants exhibit similarly shortened circadian period as cca1 mutants, LHY was not shown to play a defense role against Hpa. This raises the question of whether CCA1 is a dual function protein, affecting both the circadian clock and other non-clock related processes, as shown in the case of another central oscillator component GIGANTEA. cca1-conferred disease susceptibility might be attributed to a role of CCA1 in regulating non-clock related processes rather than to its direct involvement in the circadian clock.

5To better understand the role of CCA1 and LHY-mediated circadian clock in defense control, we tested plants misexpressing CCA1 and/or LHY for disease resistance to P. syringae and Hpa. We show that CCA1 and LHY loss-of-function mutants synergistically affect basal resistance and R gene-mediated defense against both pathogens. Disrupting the circadian clock caused by overexpression of CCA1 or LHY also results in severe disease susceptibility to P. syringae. The defense role of CCA1 and LHY against P. syringae is at least partially through circadian control of stomatal aperture but is SA-independent. Furthermore, we found that clock activity is modulated by P. syringae infection or treatment with the elicitor flg22. These data further establish the role of the circadian clock in defense control and for the first time reveal crosstalk between the circadian clock and plant innate immunity.

6To evaluate defense roles of CCA1 and LHY, we constructed the cca1-1lhy-20 mutant via a genetic cross in a Col-0 background that also contains the LUCIFERASE reporter gene driven by the CCA1 promoter (ProCCA1:LUC). The single loss of function mutants, cca1-1 and lhy-20, have shortened circadian periods of ProCCA1:LUC expression in constant light (LL). In LL, we confirmed that cca1-1lhy-20 had a much-shortened period (19.9±0.11 hr), compared with wild type (wt) Col-0 (24.4±0.09 hr) (Figure S1A and ). Although experiments in LL are important for establishing the involvement of the circadian clock in specific phenotypes, such experimental conditions can also be limiting. In entraining conditions (e.g., a 12 hr L/12 hr D cycle; LD), the altered period of clock mutants like cca1-1 and lhy-20 is not seen due to the entraining cycle, which imposes a 24 hr period (Figure 1). The clock remains important in such LD conditions, though, because the clock determines the phase of specific events with respect to as dawn and dusk. Mutants with altered period in LL typically exhibit altered phase in LD, with short period mutants exhibiting a leading (early) phase and long period mutants exhibiting a lagging (late) phase. Moreover, interactions between the endogenous circadian clock and external LD cycles can results in phase differences, sometimes dramatic, when measured in LD versus LL. For example, the phase of maximal hypocotyl elongation during early seedling growth was shifted 8–12 hours between LD and LL conditions,. In their natural environment, plants do not usually encounter LL. Therefore in evaluating the role of the circadian clock on plant defense against pathogens, it is critically important to study plant-pathogen interactions in LD and to consider the potential influence of the circadian clock on the phases of rhythmic events that might influence the plant response to pathogen challenge.

7We show here that in LD the phases of cca1-1 and lhy-20 single mutants were leading with respect to that of wild type Col-0, and that the cca1-1 lhy-20 double mutant exhibited a much earlier phase than either single mutant, consistent with the synergistic contribution of CCA1 and LHY in regulating clock activity (Figure 1 and Figure S1B). Early phase was also reported with other cca1lhy mutants,. In addition, we found that plants overexpressing CCA1 (CCA1ox), which display arrhythmic clock activity in LL, also showed arrhythmic expression of ProCCA1:LUC in LD with an acute peak in response to lights on (Figure 1 and S1B). Low ProCCA1:LUC activity in CCA1ox is consistent with CCA1 being a negative regulator of its own expression. These results emphasize that altered function of the circadian clock can manifest in both LL and LD conditions.

8To test disease resistance of cca1-1 and lhy-20 plants, we performed infection experiments at Zeitgeber Time 1 (Zeitgeber Time is the time relative to dawn; ZT1 is 1 hr after lights on) or ZT13 (1 hr after lights off), two times of day associated with drastic changes of light regime. Plant leaves were pressure-infiltrated with virulent P. syringae pv. maculicola ES4326 strain DG3 (PmaDG3). The infected plants were placed in either LD or LL. Bacterial growth assays at 3 days post infection (3 dpi) revealed no significant difference among Col-0, cca1-1, lhy-20, and cca1-1lhy-20 in either LD or LL (Figure 2 and Figure S2).

9Under natural conditions, P. syringae enters the apoplast of leaves through openings such as stomata and wounds. It is known that stomatal aperture is regulated by the circadian clock,. Therefore, infiltration of bacteria directly into plant tissue might bypass the influence of the circadian clock on stomatal defense. To test this possibility, we spray-infected with PmaDG3 Col-0, cca1-1, lhy-20, and cca1-1lhy-20 at ZT1 and ZT13 in LD. We found that Col-0 supported over 10-fold more bacterial growth with ZT1 infection than with ZT13 infection (Figure 3A and 3B), suggesting that Col-0 is more resistant at night than at dawn when spray-infected. Although we did not observe significant difference in bacterial growth between Col-0 and cca1-1 and lhy-20 single mutants, the double mutant cca1-1lhy-20 showed enhanced susceptibility to PmaDG3 when sprayed at ZT13 (Figure 3A to 3C). Consistent with this result, we found that PmaDG6 (an avirulent strain recognized by the resistance protein RPS2 in Col-0) ) grew significantly more in cca1-1lhy-20 than in Col-0 and the single mutants with ZT13 infection (Figure 3D and 3E). Together these data suggest that CCA1 and LHY share redundant functions to regulate both basal and RPS2-mediated defense against P. syringae.

10To further substantiate the role of CCA1 and LHY in defense regulation, we tested disease resistance of plants overexpressing CCA1 (CCA1ox) or LHY (LHYox), which were shown to have arrhythmic clock activity in LL,. CCA1ox plants also exhibited clock arrhythmicity in LD (Figure 1 and S1B). Disease resistance assays indicate that CCA1ox plants were more susceptible to PmaDG3 than Col-0 with infiltration infection in LD or LL (Figure 2 and S2). CCA1ox plants were also more susceptible than Col-0 to PmaDG3 and to PmaDG6 when spray-infected at ZT1 or ZT13 in LD (Figure 3).

11LHYox plants are in the Landsberg erecta (Ler) background, with which we used P. syringae pv. tomato DC3000 (DC3000) to test disease resistance because this strain induces stronger disease symptoms in our hands than does PmaDG3. Similar to CCA1ox plants, LHYox plants had more bacterial growth than Ler when infiltrated with DC3000 at ZT1 or ZT13 in LD (Figure 4A). In addition, spray-infection at ZT1 or ZT13 in LD also gave similar results (Figure 4B). Together, disruption of the circadian clock by misexpressing CCA1 and/or LHY compromises disease resistance to P. syringae, supporting a direct role of the circadian clock in defense regulation.

12Our data show that cca1-1lhy-20 was more susceptible with spray-infection and CCA1ox and LHYox plants displayed enhanced susceptibility with both spray and infiltration infections. These suggest that both stomata-dependent and -independent defense can be affected by misexpression of either of these two core oscillator genes. Consistent with this notion, a previous study showed that CCA1ox plants had increased CO2 assimilation and stomatal conductance. To further test whether the defense role of CCA1 and LHY is linked to the control of stomatal pore size, we measured plant stomatal aperture at ZT1 and ZT13 in LD. Consistent with Col-0 being more resistant with spray-infection at ZT13 than at ZT1, we found that stomatal aperture of Col-0 was much smaller at ZT13 than at ZT1 (Figure 5A). Compared with Col-0, the cca1-1 and lhy-20 mutants and CCA1ox plants showed similar stomatal aperture at ZT1 but had greater stomatal aperture at ZT13 (Figure 5A). These data suggest that disrupting clock activity mediated by CCA1 and LHY could make plants less responsive to dark-induced stomatal closure at night, thereby enhancing access of P. syringae to the leaf interior.

13To further determine how these mutants respond to P. syringae infection, we measured stomatal aperture in the presence of PmaDG3. PmaDG3 treatment was performed at ZT4 after plants had been exposed to light for four hours to ensure the opening of the stomata (Figure S3). At 1 hr post infection (1 hpi), we observed a 48.1% suppression of stomatal aperture in Col-0, compared with mock treatment (Figure 5C top and Table S1). However, this suppression was much reduced in cca1-1 and lhy-20 and largely blocked in cca1-1lhy-20 and CCA1ox. P. syringae-induced stomatal closure was transient since both mock and PmaDG3-treated leaves showed similar stomatal aperture at 3 hpi (Figure 5C bottom). Although exhibiting similar stomatal aperture at ZT1 and ZT13 (Figure 5B), the LHYox plants also showed reduced suppression of DC3000-induced stomatal closure at 1 hpi (16.9%), compared with Ler control (51.6%) (Figure 5D and Table S1). Hence, these results indicate that disrupting the circadian clock by CCA1 and LHY misexpression impairs plants' capacity of inducing stomatal closure in response to P. syringae.

14CCA1 but not LHY was previously shown to regulate resistance to the oomycete pathogen Hpa. To test whether a contribution of LHY to Hpa resistance could be discerned in the double mutant cca1-1lhy-20, we sprayed seven-day-old seedlings at ZT7 in LD with the virulent strain Hpa Emco5 or the avirulent strain Hpa Emoy2 (recognized by the R protein RPP4 in Col-0). We observed significantly more susceptibility to both Hpa strains in the cca1-1lhy-20 double mutant, compared to Col-0 and the single mutants (Figure 6A and 6B) while the CCA1ox plants were substantially more resistant to Hpa Emco5 (Figure 6A). Our data are broadly in agreement with those previously reported. The reason that we did not observe a significant difference between Col-0 and cca1-1 could be due to the difference in the infection time and/or Hpa strains used - Wang et al inoculated plants with the avirulent strain Hpa Emwa1 at dawn while we used Hpa Emco5 (virulent) and Emoy2 (avirulent) in the afternoon in our experiments. Nevertheless, these data, together with the P. syringae data described earlier, demonstrate that CCA1 and LHY contribute synergistically to basal resistance and R-gene mediated defense against both bacterial and oomycete pathogens. What surprises us is the difference in response to P. syringae (decreased resistance) and Hpa (enhanced resistance) strains observed in CCA1ox plants. We speculate that there are distinct mechanisms that these plants use to defend against the two pathogens.

15Identification of defense-related genes controlled by CCA1 and LHY is critical to gain better understanding of the mechanism of action of CCA1 and LHY in defense regulation. To this end, we analyzed promoters of 571 genes for CCA1-binding site (CBS) and evening element (EE), two cis elements known for CCA1 and LHY binding,,. These 571 genes had been previously selected to construct mini-microarrays, consisting of three groups, selected (337 defense-related genes based on microarray experiments), empirical (127 empirical marker genes for various pathogen responses), and normalization (107 non-defense related genes whose expression levels were relatively stable among experiments with pathogen infection). The online tool POBO was used to analyze up to 3000 bp from the promoter regions of these genes, which do not include the coding sequences of neighboring genes, for an enrichment of CBS or EE motifs. The background for this analysis was generated using pseudo-clusters of 100 promoters of up to 3000 bp in length of randomly sampled Arabidopsis genes (1000 bootstrap replications were used in the sampling). Compared with the background, the CBS motif was found as often as expected by chance in the selected and empirical gene promoters (Figure 7A and 7B) but the motifs were found less frequently in the normalization gene promoters (Figure 7C and Table S2). When compared to the normalization genes, there was a greater than 40% increase of the cluster mean for the CBS motif in both selected and empirical genes. These observations suggest that although defense-related genes (selected and empirical genes) are not particularly enriched with the CBS motif, the non-defense related genes (the normalization genes) are slightly depleted of the motif. The enrichment of the EE motif was more pronounced in both selected and empirical genes, with about 200% increase of the cluster means when compared to the normalization genes (Figure 7D–7F and Table S2). Thus, these results suggest that defense-related genes are preferentially regulated by CCA1 and LHY. However, since the sample size in each group is small, caution should be taken when extrapolating this interpretation to the whole genome level.

16The frequency of CBS or EE motif per promoter region was quantified from the above three sets of genes (Figure S4). Among the genes analyzed, we found that GRP7 (At2g21660; also known as COLD AND CIRCADIAN REGULATED 2 [CCR2]), had the most overrepresentation of the EE motif, with four EE within a 300 bp promoter region. One CBS motif was also found at 1294 bp of the GRP7 promoter. GRP7 is a key constituent of a slave oscillator regulated by the circadian clock, and also has been demonstrated to have roles in regulating floral transition and plant defense,. Expression of GRP7 was previously shown to be circadian regulated with a shortened circadian period in a cca1lhy double mutant and a disrupted pattern in CCA1ox plants,,. However, GRP7 had never been explicitly established as a target gene of CCA1 and LHY. Our northern analysis confirmed circadian expression of GRP7 and showed that such expression was slightly affected by the cca1-1 mutation and became arrhythmic in CCA1ox in LL (Figure S5). We also observed disrupted expression of GRP7 in CCA1ox plants in LD (Figure 8A). Thus, these data further confirm that GRP7 is regulated by CCA1.

17GRP7 was previously demonstrated to regulate stomatal activity. We found that similar to cca1-1lhy-20 and CCA1ox plants, stomatal aperture of grp7-1 was greater than that of Col-0 at ZT13 (Figure 8B). In response to PmaDG3 infection, grp7-1 displayed 14.2% suppression of stomatal aperture whereas Col-0 showed 48.1% suppression at 1 hpi (Figure 8C, S3, and Table S1), suggesting that grp7-1 has reduced responsiveness to PmaDG3 in stomatal closure. We further found that grp7-1 was significantly more susceptible to PmaDG3 than Col-0 when spray-infected at ZT13 in LD (Figure 8D). Together our bioinformatic analysis and experimental evidence indicate that GRP7 is a target of CCA1 and/or LHY that regulates stomatal activity and modulates plant defense.

18SA is a key signaling molecule involved in both basal resistance and R gene-mediated defense. The accelerated cell death 6-1 (acd6-1) mutant shows constitutive defense, high levels of SA, and extremely small size that is sensitized to the change of SA defense,. Thus, acd6-1 has been used as a convenient readout to gauge the effect of some known defense genes in regulating SA-mediated defense,,. To determine whether CCA1 and LHY act through SA, we crossed cca1-1lhy-20 to acd6-1 and obtained homozygous double (acd6-1cca1-1 and acd6-1lhy-20) and triple (acd6-1cca1-1lhy-20) mutants. We found that both double and triple mutants resembled acd6-1, displaying dwarfism and accumulating similar SA levels (Figure 9A and B). However, when spray-infected with PmaDG3 at ZT13, the double mutants were slightly more susceptible while the triple mutant was much more susceptible than acd6-1 (Figure 9C). These results corroborate a synergistic interaction between CCA1 and LHY in clock and defense regulation. They also suggest that the defense role of CCA1 and LHY is largely SA-independent. Consistent with this notion, we found that in the absence of acd6-1, the SA levels are comparable among Col-0, cca1-1, lhy-20, cca1-1lhy-20, and CCA1ox in LD (Figure S6A). In addition, although more susceptible to P. syringae infection, LHYox plants were dwarf, showed spontaneous cell death, and accumulated high levels of SA (Figure 4C, 4D, and S6B). Together, these results indicate that CCA1 and LHY act independently of SA to regulate resistance to P. syringae.

19Our data and those from other groups clearly indicate that plant innate immunity is an output event regulated by the circadian clock. However, it is not known whether this regulatory relationship is reciprocal with defense activation feeding back to affect clock activity. To test this, we infected Col-0 expressing the ProCCA1:LUC reporter with both virulent and avirulent P. syringae strains. Bioluminescence analysis indicated that the period of ProCCA1:LUC was significantly shortened in the presence of the virulent strain PmaDG3 or the avirulent strain PmaDG6 at a high dose (OD = 0.1) (Figure 10 and Table S3). Similarly, infection of Col-0 seedlings expressing ProGRP7:LUC also resulted in period shortening of ProGRP7-controlled luciferase activity (Figure S7 and Table S3). These results suggest that clock activity is modulated by both basal and RPS2-mediated defenses.

20To further investigate which defense signaling pathway(s) are involved in the feedback-regulation of clock activity, we treated Col-0/ProCCA1:LUC seedlings with flg22 or benzo (1,2,3) thiadiazole-7-carbothioic acid (BTH). Flg22 is a 22-aa synthetic peptide from the conserved region of flagellin proteins of P. syringae and elicits plant basal defense in a wide variety of plant species,. BTH is an agonist of SA that efficiently activates SA signaling. We found that flg22 at both doses (1 µM and 10 µM) significantly shortened the period of CCA1 expression. However, BTH treatment (10 µM and 300 µM) did not change CCA1 promoter activity (Figure 11A and Table S3). To further test if SA could affect clock activity, we used a cotyledon movement assay to gauge clock activity in the acd6-1 mutant, which constitutively accumulates high levels of SA,. We found that acd6-1 showed similar period, phase, and amplitude of the rhythm for cotyledon movement to Col-0 (Figure 11B and S8). Taken together, these data indicate that activation of flg22-triggered basal defense but not SA signaling can feedback to regulate clock activity.

21Increasing evidence has implicated a role of the circadian clock in regulating plant innate immunity. Of the components in the central oscillator of the circadian clock, CCA1 is the first shown to affect plant defense against P. syringae and Hpa,. However, its close homolog, LHY, has not been shown such a role, despite the fact that loss-of-function mutants in both genes displayed similarly shortened period. Thus, it was unclear whether plant innate immunity is regulated by the circadian clock mediated by CCA1 or by non-clock related function of CCA1. Here we show that disrupting clock function by misexpression of CCA1 and/or LHY leads to compromised immunity, thus further establishing a direct role of the circadian clock in defense regulation. Our data suggest that one of the mechanisms by which CCA1 and LHY regulate plant innate immunity is through affecting stomatal defense with the downstream target gene GRP7. We further demonstrate that defense activation by P. syringae infection and flg22 treatment shortens circadian period. Thus this study reveals for the first time crosstalk between the circadian clock and plant innate immunity.

22Typical studies of the circadian clock have been performed under constant light (LL) conditions to emphasize the endogenous nature of the clock. In LL, perturbations of the circadian clock typically result in altered period length; for instance, loss of CCA1 or LHY function shortens circadian period. However, plants typically grow in LD cycles in which the environmental cycle entrains even a mutant clock to a 24-hour period. Under such LD conditions, perturbations in the circadian clock can manifest as alterations in phase for reporter gene expression (Figure 1 and S1B and, ) as well as changes in a variety of other traits, including flowering time, metabolism, stomatal activity, gene expression patterns, and defense responses,,,,,,. Thus, the effects of disrupted circadian clock could become apparent under LL and LD conditions.

23Several studies indicate that the circadian clock mediated by CCA1 and LHY regulates plant defense in both LL and LD (, and this study). For instance, Bhardwaj et. al. showed that CCA1ox plants were more susceptible to P. syringe infection than wt in LL. Here we extend this observation by showing that CCA1ox plants had enhanced susceptibility to P. syringae in both LL and LD (Figure 2, 3, and S2). In LD, enhanced susceptibility was also observed in cca1-1lhy-20 and LHYox plants to P. syringae strains (Figure 3 and 4) and in cca1-1lhy-20 to Hpa strains (Figure 6). Consistent with our data, a single cca1 mutant showed compromised resistance to a different Hpa strain and affected expression of some defense-related genes in LD. Together these studies firmly establish that plant innate immunity is an output regulated by the circadian clock under LL and LD conditions.

24While we mainly focus our analyses in this report on defense phenotypes regulated by CCA1 and LHY in LD, we also agree that we should use caution when interpreting our results since the effect of the circadian clock can manifest differently under different light conditions, including both differing daylengths and light intensities. For instance, it is possible that the degree of susceptibility to pathogen infection and the severity of stomatal change in response to dark and P. syringae infection could be different in LD from those in LL in cca1-1lhy-20 (compared with wt). Alternatively, the amplitude, period, and/or phase of defense gene expression could be different in cca1-1lhy-20 (compared with wt) in LD from those in LL. Even different LD conditions could have different effects on clock activity. For example, Michael et al. showed that the set of cycling transcripts increased with the number of different cycling conditions examined. We found that in 12 hr L/12 hr D, expression of GRP7 retained rhythmicity in CCA1ox, compared with that in wt, although the waveform was altered with baseline expression increased (Figure 8A). However, Green et al observed more pronounced alterations in phase of GRP7 expression in CCA1ox (compared with that in wt) in seedlings growing in long or short daylengths (16 hr L/8 hr D or 8 hr L/16 hr D), with maximal transcript accumulation in the dark. Such differences in the patterns of GRP7 transcript abundance could also be due to other reasons besides light conditions. Nonetheless, these observations together with those of Michael et al. emphasize that to better understand the role of the circadian clock in defense control, analyses of defense phenotypes with plants misexpressing CCA1, LHY, and/or other clock genes should be carried out in LL, DD, and different LD conditions for a comprehensive comparison.

25Although encountering pathogens at different times in a day, Arabidopsis plants were suggested to be more resistant in the morning than at night. To support this conclusion, wt plants demonstrated higher resistance and/or defense responses when infiltrated during the day than at night,. We also observed similar results in plants infiltrated with P. syringae in LL or LD (Figure 2, 4A, and S2), thus supporting this conclusion. However, with spray-infection in LD, we observed the opposite phenotype; wt plants were more resistant at night than in the morning (Figure 3, 4B and 8D). During spray-infection, P. syringae initially lands on the leaf surface. Further invasion depends on the success of the bacteria in gaining entry into the host tissue via natural openings, such as stomata,. Consistent with enhanced disease resistance to sprayed P. syringae, plants in the evening have much smaller stomatal pore sizes than in the morning (Figure 5A, 5B, and 8B).

26These two seemly contradicting results actually coalesce to suggest different mechanisms that plants use to defend against pathogens at different times of day, depending on the mode of pathogen invasion. As summarized in Figure 12A, at night plants might rely more on closed stomata to physically restrict pathogen invasion but stomata-independent defense is relatively low. If a pathogen can breach stomatal restriction (i.e. being pressured into host tissue via infiltration in the laboratory) at night, it can be more virulent to the host. However, with stomata widely open during the day, plants apparently compensate for enhanced pathogen access to the leaf interior with enhanced stomata-independent defense that is stronger during the day than at night. This cycling in host resistance means that plants can be more resistant to epiphytic pathogens at night than during the day. But in the presence of apoplastic pathogens, plants can activate stronger defense during the day than at night. Taken together, we conclude that plants rely on distinct mechanisms, involving stomata-dependent and stomata-independent defenses, to respond to pathogen attacks at different times of day.

27Our data suggest that both stomata-dependent and -independent defense can be affected by CCA1, LHY, and its downstream target GRP7. Consistent with such a role of CCA1 and GRP7, these proteins are expressed in guard cells. It is conceivable that CCA1 and/or LHY proteins directly affect the abundance of GRP7 via binding to its promoter at different times of day, which in turn regulates stomatal aperture and thereby stomatal defense (Figure 11B). Since both CCA1 and GRP7 proteins are also found in other cell types besides the guard cells,, it is possible that CCA1/LHY/GRP7 also contribute to stomata-independent defense.

28GRP7 is unlikely to be the only target of CCA1 and LHY to regulate pathogen defense. First, our bioinformatic analysis suggests that a number of defense genes besides GRP7 might be preferentially regulated by CCA1 and LHY (Figure 7). And second, plants overexpressing GRP7 are not more susceptible to P. syringae (J. Alfano and H. Kang, personal communications) while CCA1ox and LHYox plants are more susceptible to P. syringae (this study). Thus, CCA1 and LHY presumably act through multiple downstream target genes to regulate plant defense. Identification of these additional defense genes controlled by CCA1 and LHY should advance our understanding of the mechanisms by which the circadian clock regulates plant defense.

29Rhythmic variation in stomatal aperture is known to be regulated by the circadian clock,,. Besides CCA1 and LHY, other genes encoding components of the central oscillator may also affect stomatal defense. For instance, a mutation in EARLY FLOWERING 3 (ELF3) was recently shown to suppress stomatal closure and disease resistance,. ELF3 might act through the FLOWERING LOCUS T gene, which is highly expressed in stomata of the elf3 mutant and has been shown to affect stomatal activity. In addition, the timing of cab expression1-1 (toc1-1) mutant also shows defects in stomatal aperture,. It is tempting to speculate that ELF3-mediated defense is related to its role in stomatal control and TOC1 could also contribute to plant defense. However, further experiments are necessary to validate these speculations. Nevertheless, these observations suggest that the circadian clock can influence stomatal activity and possibly also stomatal defense via different pathways (Figure 12B).

30Stomata have been proposed as a critical battleground during plant-bacterium interactions,. However, it is not known whether stomatal defense can also restrict the invasion of pathogens with different life styles from those of bacteria. The oomycete pathogen Hpa does not enter host organs through stomata; rather, germinating spores produce hyphae that penetrate between host epidermal cells and extend through the intracellular space in the mesophyll layer. However, near the end of the infection cycle, hyphal tips emerge through the stomata to the exterior of the leaf and then differentiate into spore-bearing structures,. Thus, it is possible that host control of stomatal aperture could influence this stage of the life cycle. Although the role of stomata in defense against Hpa has not been well established, the fact that cca1-1lhy-20 showed enhanced susceptibility to Hpa infection relative to the single mutants and wt suggests such a role of the circadian clock. Interestingly, while conferring enhanced disease susceptibility to P. syringae, CCA1ox heightened resistance to Hpa (Figure 6A and ), suggesting that CCA1ox plants employ different mechanisms to defend against these two pathogens. However, it is not clear whether the enhanced Hpa resistance conferred by CCA1ox is related to the circadian clock or to another function resulting from CCA1 overexpression.

31While regulating multifaceted physiological activities of plants, the circadian clock can also be influenced by external signals, such as changes of light, temperature, hormones, and nutrients,,,,,. Here we show that infection with both virulent and avirulent P. syringae strains shortens circadian period in Arabidopsis (Figure 10 and S7). We further found that such feedback regulation can be recapitulated with flg22 treatment (Figure 11A). Thus, defense activation can also serve as an input signal to regulate clock activity besides being an output of the circadian clock.

32Since flg22-triggered callose deposition and expression of genes involved in flg22 sensing and signal transduction were previously shown to be under circadian clock control, we conclude that the clock-defense crosstalk involves flg22 signaling (Figure 12B). Production of SA is circadian regulated, however, activation of SA defense does not affect clock activity (Figure 11 and S8 and ). Therefore, SA is an output of the circadian clock but does not serve as an input factor. Since our data showed that CCA1 and LHY act largely independently of SA, we speculate that other circadian clock components may act through SA as an output in defense control.

33What would be the advantages for plants to have clock-defense crosstalk? A properly tuned circadian clock enhances growth vigor and confers better survival rate and competitive advantage,,,,,. Regulation of defense by the circadian clock suggests that timing of effective defense against pathogens is crucial for host fitness in the presence of pathogens. However, defense is an energy-costly process intricately connected with plant growth and development. A feedback regulation of the circadian clock by defense activation could be important for the host to balance growth, development, and defense responses, for instance, to redirect the energy from costly disease resistance to primary metabolism. Consistent with this idea, several phytohormones are potential components of the clock-defense circuitry. For instance, auxin regulates clock activity as an input while auxin production and signaling are affected by the circadian clock and thus are clock output events,,. Other hormones, such as abscisic acid, brassinosteroids, cytokinins, and gibberellic acid, have been shown to serve as clock inputs,. Interestingly, cytokinin affects the phase but not the period of the clock,,. However whether these hormones are also on the output pathways of the circadian clock remains to be investigated. On the other hand, ethylene and jasmonic acid production and/or signaling are on the output pathways of the circadian clock,,,, although ethylene does not serve as a clock input in Arabidopsis. The role of jasmonic acid as a clock input is currently unknown. All these phytohormones have been implicated in defense control besides their critical roles in regulating plant growth and development,,. Therefore further investigating the roles of these phytohormones in clock-defense crosstalk should shed light on the molecular mechanisms by which plants employ to regulate growth, development, and responses to pathogen invasion. Such information could potentially lead to a better control of plant growth and resistance to devastating pathogens, ultimately enhancing productivity of plants.

34Unless otherwise indicated, all plants used on this paper are in the Columbia-0 (Col-0) background and were grown in growth chambers with a 12 hr light/12 hr dark cycle, light intensity at 200 µmol m−2 s−1, 60% humidity and 22°C. Single mutants (acd6-1, lhy-20, and grp7-1) and plants overexpressing CCA1 (CCA1ox) or LHY (LHYox) were described previously,,,,. cca1-1 was originally a Wassilewskija allele but was introgressed into Col-0 via five sequential backcrosses. The mutants cca1-1lhy-20, acd6-1cca1-1, acd6-1lhy-20, and acd6-1cca1-1lhy-20 were made by genetic crosses and confirmed with PCR markers corresponding to individual mutations (Table S4 and ). CCA1ox (line #34) and grp7-1 seeds were from Elaine Tobin and James Alfano, respectively.

35P. syringae strains were grown at 28°C with King's B medium (10 g proteose peptone, 1.5 g K2HPO4, 3.2 ml 1 M MgSO4, and 5 g glycerol per liter) containing the appropriate antibiotics for selection. Freshly cultured bacteria were collected, washed once, and resuspended to desired final concentrations in 10 mM MgSO4 for infiltration and spray infections or in sterile water for stomatal aperture measurement and bioluminescence analysis. For infiltration infection, the bacterial solution was pressured into the abaxial side of the fifth to seventh leaves of a plant with a 1 ml needleless syringe. For spray infection, the bacterial solution was mixed with Silwet L-77 (Lehle Seeds) at a final concentration of 0.04% and sprayed onto plants until the leaf surface was evenly wet. Bacterial growth and disease symptoms were analyzed as described previously. Log transformed bacterial growth was used in statistical analysis.

36Hyaloperonospora arabidopsidis (Hpa) strains were propagated and prepared as previously described,. Seven day-old soil-grown seedlings were sprayed with a spore suspension (5×104 spores/ml in water) containing the virulent strain Hpa Emco5 or the avirulent strain Hpa Emoy2. Seven days post inoculation, sporangiophores on both sides of cotyledons were counted to determine the level of resistance. Hpa infections were conducted as blind experiments where plant genotypes were unknown to the experimenters until the completion of the experiments. All bacterial and Hpa infection experiments were repeated at least three times unless otherwise indicated.

37RNA extraction and northern blotting were performed as described. Radioactive probes were made by polymerase chain reaction (PCR) with a specific antisense primer for a gene fragment in the presence of [32P] dCTP. Primers used for making probes were listed in Table S4.

38Stomatal aperture was measured with 25-day-old plants as previously described. Briefly, the fifth to seventh leaves were taken at the indicated times and mounted onto a glass slide at the abaxial side using Telesis 5 silicone adhesive (Premiere Products, Inc., CA). The top layer of the leaf was scratched off with a razor blade. Images of at least three random regions of the bottom layer of the leaf were taken immediately with a camera (Canon Digital Rebel xsi, Japan) connected to an inverted microscope (Olympus Model IMT-2). P. syringae treatment was performed at ZT4 when plants had been exposed to light for 4 hr to ensure that most stomata opened. The fifth to seventh leaves of plants were collected and immersed in PmaDG3 or DC3000 resuspensed in sterile water (108 cfu/ml) or in water as mock treatment. At 1 hpi or 3 hpi, treated leaves were harvested and processed for stomata imaging. At least three leaves per genotype and per time point were taken for stomatal images. The stomatal aperture was determined by the ratio between the width and the length of a stoma, which was measured with the assistance of ImageJ (version 1.45).

39Seedlings expressing the reporter gene LUCIFERASE (LUC) under the control of promoters from CCA1 or GRP7 (At2g21660; also called CCR2), were grown on MS media with 2% sucrose in a 12 hr light/12 hr dark cycle at 22°C for 7–10 days. Seedlings were soaked in P. syringae resuspended in sterile water in the presence of 0.04% Silwet L-77, flg22 (1 µM or 10 µM), or benzo(1,2,3)thiadiazole-7-carbothioic acid (BTH; a SA agonist) (10 µM or 300 µM), and transferred to 96-well plates containing 200 µl of MS media and 30 µl of a 2.5 mM D-luciferin solution. Mock treatments were conducted with sterile water with or without 0.04% Silwet L-77. The seedlings were subsequently transferred to LL at 22°C. LUC activity was detected at 1 hr intervals for 7 days with a TopCount luminometer (Perkin Elmer Life Sciences) and analyzed with MetaMorph image software. Flg22 was purchased from GenScript USA Inc. and BTH was a kind gift from Robert Dietrich (Syngenta).

40For cotyledon movement, surface sterilized Arabidopsis seeds were grown on MS media with 2% sucrose for 6 days in a 12 hr light/12 hr dark cycle at 22°C and were transferred to 24-well cloning plates, one seedling per well. The seedlings were entrained for two more days in the 12 hr light/12 hr cycle at 22°C and were subsequently released into LL at 22°C. Cotyledon movement was recorded with multiple surveillance cameras every 20 min for 7 days and post-run image analysis was performed as described.

41Up to 3000 bp promoter sequences of 571 genes were downloaded from Athena (http://www.bioinformatics2.wsu.edu/cgi-bin/Athena/cgi/analysis_select.pl). These genes were grouped into three sets, selected (337 defense-related gene based on microarray experiments), empirical (127 empirical marker genes for various pathogen responses), and normalization (107 non-defense related genes whose expression levels were relatively stable among experiments with pathogen infection). Promoters of these genes were analyzed for the enrichment of CBS (AA[AC]AATCT) or EE (AAAATATCT) motifs, using the online tool POBO (http://ekhidna.biocenter.helsinki.fi/poxo/pobo/). Pseudo-clusters of 100 promoters of up to 3000 bp in length of Arabidopsis genes, which do not contain the coding sequences of the neighboring genes and were sampled randomly from the entire Arabidopsis genome with 1000 bootstrap replications, were analyzed to generate the background as a control for each motif. The number of the CBS or EE motifs in gene clusters was quantified, using a Perl program.