Seeds of A. thaliana accessions Col-0 and Landsberg erecta (Ler-0), and mutants aba2-1 [Col-0] (Koornneef et al., 1982), abi1-2 [Col-0] (Gosti et al., 1999), and abi1-1 [Ler-0] (Koornneef et al., 1984) were sown on autoclaved river sand under ambient CO₂ conditions (450 ppm). Two weeks later, seedlings were transferred to 60-ml pots containing a sand/potting soil mixture that was autoclaved twice for 20 min. Then, they were placed in high (800 ppm), ambient (450 ppm), or low (150 ppm) atmospheric CO₂ conditions, where they remained for the rest of the experiment, except in the experiments shown in Figures 3D,E and Supplementary Figure S3, for which the seedlings continued growing for 2 more weeks under the ambient CO₂ condition and were only after inoculation with the Pst bacteria or treatment with ABA placed under the different CO₂ conditions for the remainder of the experiment. The technical specifications of the CO₂-controlled growth chambers used in this study have been described in detail by Temme et al. (2015). Plants grew at a 10-h day at 20°C and 14-h night at 18°C cycle (350 μmol m^-2 s^-1) with 70% relative humidity. Plants were watered every other day and received half-strength Hoagland solution (Hoagland and Arnon, 1938) twice a week. Plants were treated when 4 weeks old in all experiments. For dry weight measurements, 10 rosettes per time point were put separately in a paper bag and dried for 3 days at 60°C.

Pst (Whalen et al., 1991) and Pst cor^- (strain DB29 of Pst, which is a cmaA cfaA double mutant; Brooks et al., 2004) were grown on KB medium (King et al., 1954) supplemented with 50 μg ml^-1 rifampicin. To prepare inoculum, bacteria were streaked from rifampicin-selective KB agar plates and subsequently cultured in liquid KB medium in a shaker at 220 rpm at 28°C for 24 h. Bacteria were collected by centrifugation for 10 min at 1,500 × g, and resuspended in 10 mM MgSO₄. For dip inoculation, the bacterial inoculum was diluted to a final concentration of 5 × 10⁷ cfu ml^-1 of 10 mM MgSO₄ containing 0.015% (v/v) Silwet L-77 (Van Meeuwen Chemicals, Weesp, Netherlands). For pressure infiltration, the bacterial suspension was adjusted to a concentration of 5 × 10⁶ cfu ml^-1 (disease assay) or 1 × 10⁸ cfu ml^-1 (ABA measurement). The abaxial side of leaves was pressure infiltrated with a needleless syringe.

To determine bacterial growth in planta, leaf disks of infected plants were harvested, weighed, surface sterilized in 70% ethanol for 8 s, and washed with water immediately after. Subsequently, 200 μl of 10 mM MgSO₄ was added to the leaf disks after which they were ground thoroughly. Aliquots of 10 μl of different dilutions were plated onto KB plates containing 25 μg ml^-1 rifampicin. After 48 h of incubation at room temperature, bacterial colonies were counted and growth of the bacteria was calculated after log-transformation of the cfu data. Eight biological replicates were included for each time point.

Stomatal aperture and density were measured by a modified protocol of dental resin impressions (Geisler et al., 2000). Two components of Present Light Body (Coltène, Altstatten, Switzerland) were mixed thoroughly (v/v, 1:1) and the abaxial side of the leaves was softly pressed onto the dental resin immediately after harvesting. Leaves were removed 10 min later when the mixture had hardened. Transparent nail polish was applied to the dental resin molds to create casts, which were fixed on microscope slides with Anutex modeling wax (Kemdent, Purton, Swindon, Wiltshire, UK) for further observation.

Stomata were examined using an Olympus microscope and Analysis D Olympus Software on the pictures taken. Stomatal aperture was determined by measuring the width and length of the stomata. At least six leaves were harvested for each treatment and 20–30 observations were recorded from each leaf.

For ABA quantification, 60–250 mg leaf material was harvested 24 h after treatments and ground to a fine powder using liquid nitrogen. ABA was extracted as described (Scala et al., 2013). Briefly, the samples were homogenized in 0.5 ml of 70% methanol using a Precellys24 tissue homogenizer (Bertin Technologies, Berlin) by shaking at 6,000 rpm for 40 s. Subsequently, the homogenates were centrifuged at 10,000 ×g for 20 min at 4°C. The supernatants of two extraction steps were pooled together. ABA was quantified by liquid chromatography-mass spectrometry (LC-MS) analysis on a Varian 320 Triple Quad LC-MS/MS. Endogenous ABA levels were quantified by comparing the integrated surface area from each sample with its corresponding internal standard.

To measure stomatal responsiveness to exogenously applied ABA, leaves were dipped in a solution of 15 μM ABA in 0.015% (v/v) Silwet L-77.

Numerous studies have been conducted to assess the effect of high CO₂ levels on plant performance, including plant growth, stomatal behavior, and disease resistance. However, only limited information is available on the effects of low CO₂ levels on the plant. Here, we studied the effects of three different CO₂ levels on Arabidopsis plants in the absence and presence of pathogens: high (800 ppm), ambient (450 ppm), and low (150 ppm) levels of CO₂. Plants were cultivated under ambient CO₂ until they were 2 weeks old, after which they were placed under the three respective CO₂ conditions. We noticed that plants that had grown under the low CO₂ condition for an additional 2 weeks had smaller rosette sizes compared to plants grown under high and ambient CO₂ conditions (Figure 1A). Also the dry weight of the rosettes was significantly lower in the low CO₂ condition (Figure 1B). In contrast, there were no effects on rosette growth under high CO₂ conditions, which was rather unexpected since most previous studies have reported an increase in biomass (Bowes, 1991; Leakey et al., 2009). However, our experimental conditions may not have been optimal for stimulated growth by elevated CO₂ (Temme et al., 2015) and moreover, the Col-0 accession that we used may respond differently to high CO₂ than other plant species (Li et al., 2006; Leakey et al., 2009; Temme et al., 2015).

Stomatal density and aperture were investigated under the three CO₂ conditions as well. At high atmospheric CO₂, stomatal density was not influenced, but a significant decrease in stomatal aperture was found (Figures 1C,D). At low atmospheric CO₂, an increase in both stomatal density and stomatal aperture was detected. These results are in line with previous studies that found that the inverse relationship between atmospheric CO₂ and stomatal behavior was more evident under sub-ambient CO₂ conditions than under elevated CO₂ conditions (Royer, 2001), which is a phenomenon that is referred to as the CO₂ ‘ceiling’ phenomenon. The major effects on plant growth and stomatal behavior observed at especially the low CO₂ level prompted us to introduce a pathogen into the system that naturally enters through stomata in order to study the effects of CO₂ on plant immunity.

To explore whether the differential stomatal behavior at the three tested atmospheric CO₂ levels affects stomatal defense responses, we examined the stomatal responsiveness of Arabidopsis plants to infection by the bacterial leaf pathogen Pst. Previously, it was shown that at ambient CO₂ the stomata close within 1 to 2 h after dip inoculation with Pst, and reopen again at 3 to 4 h due to the action of the virulence factor COR (Melotto et al., 2006). Our results under ambient conditions are in line with this finding, since we found that the stomata closed within 1 h after dip inoculation with wild-type Pst or the COR-deficient mutant Pst cor^- (Figure 2). Subsequently, at 4 h after inoculation, the stomatal aperture was significantly greater in leaves infected by the wild-type Pst strain than by the Pst cor^- mutant strain. It should be noted that for proper determination of the COR effect one should compare the mutant infection with the wild-type infection, and not with the mock treatment at 4 hpi. The circadian rhythm influences the stomatal aperture (closing in the afternoon), which may explain the somewhat more closed stomata in the mock situation at 4 hpi. Therefore, in comparison to mock, stomata of Pst cor^--infected leaves were no longer statistically significantly more closed at 4 hpi and moreover, the stomata of wild-type Pst-challenged leaves were not statistically significantly more open, although a trend, likely due to COR action, was visible.

High CO₂-grown plants contained stomata that were generally more closed (Figures 1D, 2). Still, the stomata initially closed further when the leaves were attacked by Pst and near significant closure was also induced by Pst cor^- infection. Similar to the ambient condition, under high CO₂ the stomata were subsequently reopened by the wild type in comparison to the mutant bacteria. This suggests that stomatal responsiveness to PAMP-triggered closure and COR-triggered opening is intact at the high CO₂ level. Under the low CO₂ condition, stomata were opened more widely (Figures 1D, 2), but still they closed within 1 h after inoculation with Pst wild type or Pst cor^- mutant, which is comparable to the ambient and high CO₂ conditions. In contrast, at 4 h after inoculation, the stomata of both the Pst- and the Pst cor^--challenged leaves remained closed under the low CO₂ condition. These data show that plants grown under high and low CO₂ conditions initially respond to Pst infection by closing their stomata, despite their original differences in stomatal aperture. However, the subsequent COR-mediated stomatal reopening occurs only under high and ambient CO₂ conditions, whereas it is blocked under the low CO₂ condition.

The resistance of Arabidopsis plants to Pst infection under the different atmospheric CO₂ conditions was tested by determining the growth of Pst in plants cultivated at different CO₂ levels. Initially, at 4 h after dip inoculation, plants grown at high CO₂ levels contained significantly less Pst than plants grown at low CO₂ (Figure 3A), which coincided with the lower stomatal density and aperture in leaves of high CO₂-grown plants, thereby allowing fewer bacteria to enter the leaves (Figures 1C,D, 2). However, at 4 days after inoculation, the Pst bacterial titer in high CO₂-grown plants was significantly higher compared to that in ambient and low CO₂-grown plants (Figure 3A). In this particular experiment, the bacterial titer in low CO₂-grown plants showed a trend of reduced amounts compared to ambient-grown plants; in other experiments the difference between the two treatments was often found to be statistically significant (Figures 3D,E, 5B and Supplementary Figure S2). Moreover, less-severe chlorotic disease symptoms on plants grown at low CO₂ compared to plants grown at ambient and high CO₂ were detected (Figure 3B). Given the different plant growth rates when cultivated at different CO₂ levels, it is possible that differences in the weight per leaf area caused the detected differences in Pst titer when depicted per leaf area. To rule out this possibility, the bacterial titer was also determined per gram of leaf tissue. The same trend, namely enhanced multiplication of Pst under high, and reduced amplification under low CO₂ conditions was found (Figure 3C).

The role of COR in successful infection by Pst through facilitation of stomatal reopening, but also by suppression of SA-mediated defense signaling and disease symptom development, has been well established (Mittal and Davis, 1995; Brooks et al., 2005). The in planta growth of the Pst cor^- mutant strain was on average, over all three tested atmospheric CO₂ conditions, significantly lower than that of the Pst wild-type strain, as was reported previously (Figure 3C) (Melotto et al., 2006). However, while the bacterial titer of wild-type Pst was significantly higher in high CO₂-grown plants and lower in low CO₂-grown plants, growth of the mutant Pst cor^- was severely limited under all three CO₂ conditions, reaching the same low bacterial titer as that of wild-type Pst in low CO₂-grown plants (Figure 3C). The statistically significant interaction of atmospheric CO₂ with the in planta growth difference between Pst and Pst cor^- suggests that atmospheric CO₂ regulates the plant’s sensitivity to COR, leading to enhanced responsiveness at high CO₂ and impaired responsiveness at low CO₂. This differential responsiveness to COR could play a role in the observed differences in resistance levels to Pst under the three tested CO₂ conditions.

Whether the inhibition of COR-mediated stomatal reopening under the low CO₂ condition could explain the observed high level of resistance to Pst was tested by assaying bacterial growth after pressure infiltration of the Pst bacteria into the leaves. However, due to the small leaf size of low CO₂-cultivated plants, a different experimental set-up had to be employed in which all plants were grown at ambient CO₂ until they were inoculated with Pst (when the plants were 4 weeks old), after which they were placed under either the high, ambient, or low CO₂ condition for the remainder of the experiment. First, we tested in this set-up the effect of different CO₂ levels on Pst growth after dip inoculation and found that the bacterial growth was affected similarly to the original set-up in which plants had experienced the different CO₂ conditions already 2 weeks preceding the dip inoculation (Figures 3A,D). Also upon pressure infiltration of Pst into the leaves, the effects of CO₂ on bacterial growth were very much alike: high CO₂ caused enhanced susceptibility whereas low CO₂ caused reduced susceptibility (Figure 3E). This suggests that under the low CO₂ condition not only COR-mediated stomatal reopening is affected, but also the suppression of post-invasive defense responses is reduced. On the other hand, under the high CO₂ condition post-invasive defense appears stronger downregulated.

To gain more insight into how the differential responsiveness to COR observed under different atmospheric CO₂ levels may alter plant immunity, we assessed the role of the hormone ABA, a known regulator of stomatal aperture, in Pst-triggered stomatal closure and subsequent reopening (Melotto et al., 2006). First, we studied the effect of ABA on stomatal behavior under our ambient CO₂ condition. It has previously been shown that in non-induced situations the stomata of the Arabidopsis ABA-deficient mutant aba2-1 and the ABA-insensitive mutant abi1-1 can be more open than that of wild-type plants, but this effect is not always evident (Figures 4, 5A, and Supplementary Figure S1) (Merlot et al., 2002; Melotto et al., 2006), which may be related to difference in stomatal aperture of wild-type plants at different times of the day on which the experiments were executed (Somers et al., 1998; Dodd et al., 2004). Nevertheless, in all experiments challenge with Pst triggered initially stomatal closure in wild type as well as aba2-1 and abi1-1 plants within 1 h after dip inoculation (Figure 4 and Supplementary Figure S1a). Moreover, while stomata subsequently reopened in a COR-dependent manner in wild-type plants as detected at 4 h after inoculation, stomata did not reopen in the aba2-1 and abi1-1 mutants when treated with either Pst or Pst cor^- (Figure 4 and Supplementary Figure S1a). These findings are in contrast with those described by Melotto et al. (2006) who showed that neither the ABA-deficient mutant aba3-1, nor the ABA-insensitive mutant ost1-2, responded with stomatal closure to infection by Pst or treatment with flg22, the active epitope of bacterial flagellin. However, in accordance with our results, Montillet et al. (2013) reported that mutants aba2-1 and ost1-2 did close their stomata upon flg22 treatment. These results suggest the existence of an ABA-independent pathway in guard cells involving pathogen-induced stomatal closure. Furthermore, our data reveal a role for ABA signaling in COR-mediated stomatal reopening.

Based on our finding that under ambient CO₂ conditions ABA mutants show a stomatal response pattern to Pst infection that is similar to that of wild-type Arabidopsis plants grown under low CO₂ conditions, we hypothesized that there could be a role for ABA signaling in atmospheric CO₂-altered disease resistance to Pst. To test this, we first measured stomatal aperture of the mutant aba2-1 when cultivated at different levels of atmospheric CO₂ without Pst infection. As expected, stomata of wild-type Col-0 plants were more closed at high CO₂ and more opened at low CO₂ (Figure 5A). However, stomata of the aba2-1 mutant did not change their aperture under different CO₂ conditions. In fact, their stomata were relatively open under all three CO₂ conditions, to the same high level as that of low CO₂-grown wild-type plants (Figure 5A). This statistical interaction between ABA and CO₂ levels suggests that ABA can be involved in stomatal responsiveness to different atmospheric CO₂ conditions.

Subsequently, we tested whether atmospheric CO₂ can alter disease resistance to Pst in ABA mutants. Under ambient and high CO₂ conditions, both aba2-1 and abi1-1 exhibited reduced in planta growth of Pst compared with wild-type plants (Figure 5B and Supplementary Figure S2), supporting a negative role for ABA signaling in the defense response against Pst, as shown previously (Melotto et al., 2006; De Torres-Zabala et al., 2007). More importantly, under all three CO₂ conditions Pst growth in both ABA mutants was as low as in the wild-type plants grown at low CO₂ (Figure 5B and Supplementary Figure S2). Together, these results suggest that ABA signaling plays an important role in atmospheric CO₂-regulated plant defense responses against Pst.

Previously, ABA has been reported to accumulate upon Pst infection (De Torres-Zabala et al., 2007). Moreover, enrichment in atmospheric CO₂ can also change ABA levels or ABA signaling, although variable effects in Arabidopsis have been described (Li et al., 2006; Teng et al., 2006). We assayed the ABA content in leaves infected with Pst under different atmospheric CO₂ conditions. In the absence of Pst, ABA accumulation under the low CO₂ condition was significantly reduced compared to that under the high CO₂ condition, but no statistically significant differences with the ambient CO₂ condition were detected (Figure 5C). In Pst-challenged leaves the ABA concentrations rose significantly compared with mock-treated leaves under all three CO₂ conditions. However, in the low CO₂-grown plants the ABA levels were significantly lower than those in the ambient and high CO₂-grown plants upon infection by Pst. These results suggest that reduced ABA levels in low CO₂-grown plants may be responsible for enhanced resistance to Pst.

Previous studies, focusing mainly on the effects of elevated atmospheric CO₂, showed that growth of various plant species was promoted by high CO₂ levels and inhibited by low CO₂ levels (Leakey et al., 2009; Eastburn et al., 2010; Temme et al., 2015). In our study, we show that the low CO₂ condition significantly reduced growth of Arabidopsis and caused the stomata to be opened more widely than under ambient CO₂ (Figure 1). Reduced ABA content and sensitivity in low CO₂-grown plants (Figure 5C and Supplementary Figure S3) may be causal for the enhanced stomatal opening phenotype, and although a reduced plant stature has been reported for ABA mutants as well (Chatfield et al., 2000; LeNoble et al., 2004), the growth reduction at low CO₂ is more likely caused by reduced photosynthesis. We found that Arabidopsis grown at high CO₂ displayed a reduced opening of their stomata, as has been reported previously (Araújo et al., 2011). However, we found no reduction in stomatal density, nor an increase in rosette dry weight under the high CO₂ condition. No or small effects of high CO₂ on growth enhancement has previously been reported (Li et al., 2006; Temme et al., 2015). In our experimental set-up light intensity and nutrient constraints may have limited the stimulating effect of high CO₂ on plant growth. Our findings on stomatal density are in line with the previously described CO₂ ‘ceiling’ phenomenon, which refers to reaching a maximum stomatal density at a CO₂ level of 400 ppm and that stomata respond more strongly to sub-ambient than to elevated CO₂ concentrations (Kürschner et al., 1997; Royer, 2001).

Besides CO₂, ABA determines the stomatal aperture. Here, we provide evidence for a role of ABA signaling in the regulation of stomatal aperture by different CO₂ levels in un-infected plants. Unlike the wild type, the stomatal aperture of the ABA biosynthesis mutant aba2-1 was not influenced by the CO₂ conditions (Figure 5A). Moreover, the ABA signaling mutant abi1-1 was unresponsive to high CO₂-induced stomatal closure, albeit sensitivity regarding low CO₂-induced opening was observed (Supplementary Figure S1). In general, our data corroborate previous findings on the interrelationship of ABA with elevated CO₂-regulated signaling in guard cells (Leymarie et al., 1999; Nishimura et al., 2010; Xue et al., 2011; McLachlan et al., 2014).

Activation of stomatal closure has been demonstrated to be an essential pre-invasive defense response to foliar pathogens in various plant species (Melotto et al., 2006; Lee et al., 2013; Li et al., 2013; Du et al., 2014). The ABA-deficient mutant aba3-1 was previously shown to be compromised in its ability to close its stomata in response to Pst infection, suggesting a requirement for ABA biosynthesis in Pst-induced stomatal closure (Melotto et al., 2006). However, in our experiments, at all CO₂ levels tested, both wild-type plants and the ABA mutants aba2-1 and abi1-1 responded to Pst infection with closure of their stomata (Figure 4 and Supplementary Figure S1). This indicates that the Pst-induced stomatal closure occurs at least partly independently of ABA and that atmospheric CO₂ does not influence this mechanism. This is in line with a recent finding that an ABA-independent oxylipin pathway is responsible for flg22- and Pst-induced stomatal closure (Montillet et al., 2013). Moreover, a genetic screen of Arabidopsis mutants that are impaired in Pst-induced stomatal closure generated multiple mutants that still exhibited ABA-induced stomatal closure (Zeng et al., 2011). In addition, Lim et al. (2014) demonstrated that ABA hyposensitive PP2CA1 overexpressors closed their stomata in response to Pst infection or flg22 treatment. Taken together, these results support the notion that besides ABA signaling, additional mechanisms that are independent of ABA and independent of atmospheric CO₂ play a role in the stomatal closure response upon Pst infection.

The phytotoxin COR that is produced by Pst induces stomatal opening to stimulate infection. We found that under both ambient and high CO₂ conditions, initial Pst-induced stomatal closure was followed by COR-dependent stomatal reopening (Figure 2). Interestingly, while stomata in low CO₂-grown plants still responded to Pst with closing within 1 hpi, they did not reopen at 4 hpi, resulting in a stomatal aperture very similar to that of Pst cor^--infected plants (Figure 2). Thus, if COR production by Pst is not affected by the low CO₂ level, sensitivity to COR in terms of stomata reopening seems compromised under the low CO₂ condition.

Previous reports demonstrated that under ambient CO₂ conditions, COR and ABA signaling can influence each other’s activity either negatively or positively. For instance, ABA-induced stomatal closure is inhibited by COR (Melotto et al., 2006; Zheng et al., 2012), while both COR and ABA repress SA-regulated defense signaling (Brooks et al., 2005; De Torres-Zabala et al., 2007). Also, ABA and COR both activate gene expression of three NAC transcription factors that suppress Pst-induced SA biosynthesis and stomatal closure (Zheng et al., 2012). Here, we show that the mutants aba2-1 and abi1-1 closed their stomata upon inoculation with Pst and were unable to reopen them in response to COR production by Pst at 4 hpi under all three CO₂ conditions tested (Figure 4 and Supplementary Figure S1). This pointed to an unexpected role for ABA signaling in COR-mediated stomatal reopening, which is independent of CO₂ levels. As a follow-up experiment, expression of the COR-inducible, ABA-dependent NAC transcription factor genes ANAC019, ANAC055, and ANAC072 (Zheng et al., 2012) could be assessed in aba2-1 and abi1-1; their reduced expression may contribute to the ABA-dependency of COR-mediated stomatal reopening. It has been demonstrated that reactive oxygen species (ROS) act as essential second messengers in Arabidopsis guard cells, functioning in CO₂- and ABA-induced stomatal closure (Pei et al., 2000; Mustilli et al., 2002; Chater et al., 2015). Interestingly, it has recently been reported that COR inhibited ROS production in guard cells, thereby aiding the inhibition of stomatal closure (Toum et al., 2016). Whether ROS production may act as a point of convergence between CO₂, ABA, and COR signaling and in doing so determines the outcome of COR responsiveness in ABA mutants and under different CO₂ conditions is an important question.

In accordance with the blocked COR-induced stomatal reopening, low CO₂-grown plants exhibited significantly reduced growth of Pst at 4 dpi compared with ambient CO₂-grown plants (Figures 3C–E, 5B and Supplementary Figure S2). However, this decrease in susceptibility to Pst was apparent both in dip-inoculated and pressure-infiltrated leaves (Figures 3D,E), suggesting that the CO₂ effect on Pst infection is beyond the interference of CO₂ with stomatal defenses. The in planta growth of Pst under the low CO₂ condition was arrested to the same level as that of Pst cor^- under low, ambient or high CO₂ conditions (Figure 3C). This demonstrates that impairment of COR-mediated defense suppression that is apparent under low CO₂ conditions severely reduces the virulence of Pst. It is known that SA plays an essential role in the defense response of Arabidopsis against Pst. By acting as a structural JA mimic, COR triggers a signaling cascade that counteracts SA-dependent defenses, thus promoting susceptibility to Pst infection (Zheng et al., 2012). It is possible that under low CO₂ conditions the function of COR is impaired, which alleviates the suppression of downstream SA signaling, resulting in enhanced resistance to Pst.

The inability of the mutants aba2-1 and abi1-1 to respond with stomatal reopening to COR, was associated with enhanced resistance to Pst (Figure 5B and Supplementary Figure S2), indicating the important role of ABA signaling in suppression of defenses by COR. The link with defective responsiveness to COR and enhanced resistance has been shown previously, albeit the role of ABA herein was in closing instead of reopening of the stomata (Lim et al., 2014). Interestingly, the aba2-1 and abi1-1 mutants were under all three CO₂ conditions as resistant to Pst as wild-type plants grown at low CO₂ (Figure 5B and Supplementary Figure S2) and Pst and Pst cor^- grew to a similar level in the aba2-1 mutant (data not shown). Thus, the resistance phenotype of low CO₂-grown wild-type plants resembles that of the ABA mutants. Remarkably, also plant growth is inhibited by both low CO₂ and ABA-related mutations (Figures 1A,B) (Chatfield et al., 2000; LeNoble et al., 2004). Moreover, in low CO₂-grown plants the ABA levels were induced to a lower extent by Pst infection than in ambient or high CO₂-grown plants and a trend of reduced ABA content was already visible in the non-infected situation (Figure 5C). In addition, responsiveness to ABA was affected under low CO₂, shown by a significantly smaller effect on stomatal closure induced by exogenous application of ABA than under the high and ambient CO₂ conditions (Supplementary Figure S3). Altogether, these results suggest that the enhanced resistance to Pst that is evident under the low CO₂ condition is related to a decrease in ABA content and signaling.

In addition to a role in COR-triggered stomatal reopening that we demonstrated, ABA is also known to suppress SA defense signaling, possibly in part by activation of the three above-mentioned NAC transcription factors (De Torres-Zabala et al., 2007; Zheng et al., 2012). Modulation of ABA signaling by atmospheric CO₂ may affect expression of SA-mediated defense responses. Not only did we show that the ABA mutants aba2-1 and abi1-1 were more resistant to Pst infection, we also demonstrated that the ABA hypersensitive mutant abi1-2 is more susceptible to Pst (Supplementary Figure S4). At high CO₂ levels, wild-type Arabidopsis plants were also more susceptible to Pst (Figures 3, 5B and Supplementary Figure S2). We did not detect a significant increase in ABA content under high CO₂, although a trend was visible (Figure 5C). Under high CO₂ conditions, stomatal aperture was decreased, causing fewer Pst bacteria to enter the leaves, but still at 4 dpi higher bacterial titers were measured, and also upon pressure infiltration enhanced Pst in planta growth was detected (Figures 1C, 3). It is unclear how high CO₂ interferes with plant defense responsiveness. Besides an effect on plant defense, the enhanced Pst growth may be caused by a favorable endophytic environment for the bacteria in terms of nutrition and water availability in high CO₂-grown plants (Lake and Wade, 2009; Pangga et al., 2011), but this enriched condition was unlikely established within the 4 days time frame in which the plants were transferred from ambient to high CO₂ condition, which was used for some of the experiments (Figures 5D,E). Moreover, in contrast to our findings, tomato plants were reported to exhibit reduced infection by Pst at elevated CO₂ levels (Li et al., 2014). Thus, enrichment of the endophytic environment alone unlikely explains the full effect of high CO₂ on enhanced Pst growth in Arabidopsis that we demonstrated. Possibly, the difference in genetic make-up between Arabidopsis and tomato plants can explain the difference in effect of high CO₂ on susceptibility to Pst.