Varying Atmospheric CO2 Mediates the Cold-Induced CBF-Dependent Signaling Pathway and Freezing Tolerance in Arabidopsis

Changes in the stomatal aperture in response to CO2 levels allow plants to manage water usage, optimize CO2 uptake and adjust to environmental stimuli. The current study reports that sub-ambient CO2 up-regulated the low temperature induction of the C-repeat Binding Factor (CBF)-dependent cold signaling pathway in Arabidopsis (Arabidopsis thaliana) and the opposite occurred in response to supra-ambient CO2. Accordingly, cold induction of various downstream cold-responsive genes was modified by CO2 treatments and expression changes were either partially or fully CBF-dependent. Changes in electrolyte leakage during freezing tests were correlated with CO2′s effects on CBF expression. Cold treatments were also performed on Arabidopsis mutants with altered stomatal responses to CO2, i.e., high leaf temperature 1-2 (ht1-2, CO2 hypersensitive) and β-carbonic anhydrase 1 and 4 (ca1ca4, CO2 insensitive). The cold-induced expression of CBF and downstream CBF target genes plus freezing tolerance of ht1-2 was consistently less than that for Col-0, suggesting that HT1 is a positive modulator of cold signaling. The ca1ca4 mutant had diminished CBF expression during cold treatment but the downstream expression of cold-responsive genes was either similar to or greater than that of Col-0. This finding suggested that βCA1/4 modulates the expression of certain cold-responsive genes in a CBF-independent manner. Stomatal conductance measurements demonstrated that low temperatures overrode low CO2-induced stomatal opening and this process was delayed in the cold tolerant mutant, ca1ca4, compared to the cold sensitive mutant, ht1-2. The similar stomatal responses were evident from freezing tolerant line, Ox-CBF, overexpression of CBF3, compared to wild-type ecotype Ws-2. Together, these results indicate that CO2 signaling in stomata and CBF-mediated cold signaling work coordinately in Arabidopsis to manage abiotic stress.


Introduction
Cold stress (freezing or chilling temperatures) is one of the significant environmental stresses that can severely impact crop yields and limit plant species' geographical distribution. above mimicked stomatal responses to high and low CO 2 conditions, respectively, and we hypothesized that these mutants could be used as genetic tools to understand molecular responses to cold treatments.
Here, we report that varying CO 2 levels in the air altered the low-temperature expression of three CBF genes and selected downstream cold-responsive genes. Our findings were further supported using mutants with impaired stomatal responses to CO 2 . In addition, changes in the expression of cold-responsive genes by CO 2 were positively correlated with freezing tolerance as measured by electrolyte leakage.

Varying Atmospheric CO 2 Modulates the CBF-Mediated Cold Signaling Pathway
Initially, we examined the effects of supra-ambient, ambient, and sub-ambient CO 2 (800, 400, and 80 µmol mol −1 , respectively) on changes in expression of CBF1, CBF2, and CBF3 in wild-type Arabidopsis (Col-0) in response to 2 h of exposure to 4 • C. The expression of all three CBF genes was most significant when plants were exposed to sub-ambient CO 2 . The expression levels decreased proportionally (p ≤ 0.01) in response to supra-ambient CO 2 ( Figure 1A). Later, we observed the effect of varying atmospheric CO 2 on downstream CBF-dependent target genes in the cold signaling pathway, i.e., COR47 (RD17), COR78 (RD29A/LTI78), ERD10 (LTI45/LTI29), KIN1, KIN2 (COR6.6), COR15a, and COR15b ( Figure 1B,C). The low temperature-dependent expression for COR47, COR78, and ERD10 was enhanced (p ≤ 0.01) at sub-ambient compared to ambient CO 2 , but the expression of these same three genes did not differ (p > 0.5) between the ambient and supra-ambient CO 2 treatments (partially CO 2 dependent, Figure 1B). The low temperature-induced expression of KIN1, KIN2, COR15a, and COR15b was greatest (p ≤ 0.01) at sub-ambient and least at supra-ambient CO 2 (fully CO 2 dependent, Figure 1C). respectively, and we hypothesized that these mutants could be used as genetic tools to understand molecular responses to cold treatments.
Here, we report that varying CO2 levels in the air altered the low-temperature expression of three CBF genes and selected downstream cold-responsive genes. Our findings were further supported using mutants with impaired stomatal responses to CO2. In addition, changes in the expression of cold-responsive genes by CO2 were positively correlated with freezing tolerance as measured by electrolyte leakage.

Varying Atmospheric CO2 Modulates the CBF-Mediated Cold Signaling Pathway
Initially, we examined the effects of supra-ambient, ambient, and sub-ambient CO2 (800, 400, and 80 μmol mol −1 , respectively) on changes in expression of CBF1, CBF2, and CBF3 in wild-type Arabidopsis (Col-0) in response to 2 h of exposure to 4 °C. The expression of all three CBF genes was most significant when plants were exposed to sub-ambient CO2. The expression levels decreased proportionally (p ≤ 0.01) in response to supra-ambient CO2 ( Figure 1A). Later, we observed the effect of varying atmospheric CO2 on downstream CBF-dependent target genes in the cold signaling pathway, i.e., COR47 (RD17), COR78 (RD29A/LTI78), ERD10 (LTI45/LTI29), KIN1, KIN2 (COR6.6), COR15a, and COR15b ( Figure 1B,C). The low temperature-dependent expression for COR47, COR78, and ERD10 was enhanced (p ≤ 0.01) at sub-ambient compared to ambient CO2, but the expression of these same three genes did not differ (p > 0.5) between the ambient and supra-ambient CO2 treatments (partially CO2 dependent, Figure 1B). The low temperature-induced expression of KIN1, KIN2, COR15a, and COR15b was greatest (p ≤ 0.01) at sub-ambient and least at supra-ambient CO2 (fully CO2 dependent, Figure 1C). The low temperature expression of CBF and downstream CBF-dependent genes in Arabidopsis is modified by air levels of CO2. Individual 3.5 week-old Arabidopsis plants (Col-0) were exposed to 4 °C temperature for 2 h with sub-ambient, ambient, or supra-ambient CO2, (80, 400 and 800 μmol mol −1 , respectively). Transcript abundance of three CBF genes (A), and seven downstream CBF target genes (B and C) were determined by q-PCR. The expression of measured transcripts was normalized to ACT2. Relative values are means (×1000) of three biological replicates and two technical Figure 1. The low temperature expression of CBF and downstream CBF-dependent genes in Arabidopsis is modified by air levels of CO 2 . Individual 3.5 week-old Arabidopsis plants (Col-0) were exposed to 4 • C temperature for 2 h with sub-ambient, ambient, or supra-ambient CO 2 , (80, 400 and 800 µmol mol −1 , respectively). Transcript abundance of three CBF genes (A), and seven downstream CBF target genes (B and C) were determined by q-PCR. The expression of measured transcripts was normalized to ACT2. Relative values are means (×1000) of three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using the one-way analysis of variance procedure (ANOVA) and Tukey HSD.

Varying Atmospheric CO 2 Modifies the Freezing Tolerance of Arabidopsis
The electrolyte leakage was measured using two ecotypes of Arabidopsis, Col-0 and Ws-2, to determine the effects of changing CO 2 concentration on CBF-mediated cold signaling altered freezing tolerance (Figure 2A). In addition, a plant that over-expresses CBF3 (Ox-CBF) in the Ws-2 background [37], which constitutively activates downstream cold signaling cascades (Figure 2A), was included for the comparison. At −2 • C, 3.5-week-old Col-0 plants displayed 17% electrolyte leakage at sub-ambient CO 2 (p ≤ 0.01) and this increased to 26 and 30% leakage at ambient and supra-ambient CO 2 , respectively ( Figure 2B). The response of electrolyte leakage by Col-0 plants to changing CO 2 was similar at −2 and −4 • C, but the total amount of electrolyte leakage was greater at the lower temperature ( Figure 2B,C). Interestingly, no significant differences in electrolyte leakage in response to CO 2 at −2 • C were observed using the Ws-2 ecotype (p > 0.5). However, at −4 • C, the electrolyte leakage was 2.7-fold greater (p ≤ 0.01) at supra-ambient than at sub-ambient CO 2 ( Figure 2B,C). When the temperature was lowered to −6 • C, the CO 2 effects on electrolyte leakage did not differ between Col-0 and Ws-2 (p > 0.05), which indicated that both ecotypes were approaching maximal electrolyte leakage ( Figure 2D). Interestingly, the freezing-tolerant Ox-CBF plant displayed somewhat unexpected results. Electrolyte leakage from Ox-CBF remained unaffected by the change in CO 2 levels (p > 0.5) at −2 and −4 • C ( Figure 2B,C). However, significant differences (p ≤ 0.01) in electrolyte leakage were observed at supra-ambient CO 2 treatments at −6 • C compared to sub-ambient and ambient CO 2 treatments. In addition, electrolyte leakage was more than 50% less in the Ox-CBF line compared to Ws-2 ( Figure 2D), irrespective of CO 2 treatment. This indicated a possible role for CBF-mediated signaling in CO 2 -dependent responses of cold-induced electrolyte leakage. The above results suggest that the changes in CO 2 level in air modified the freezing tolerance of Arabidopsis via CBF-mediated cold signaling (Figures 1 and 2 replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using the one-way analysis of variance procedure (ANOVA) and Tukey HSD.

Varying Atmospheric CO2 Modifies the Freezing Tolerance of Arabidopsis
The electrolyte leakage was measured using two ecotypes of Arabidopsis, Col-0 and Ws-2, to determine the effects of changing CO2 concentration on CBF-mediated cold signaling altered freezing tolerance (Figure 2A). In addition, a plant that over-expresses CBF3 (Ox-CBF) in the Ws-2 background [37], which constitutively activates downstream cold signaling cascades (Figure 2A), was included for the comparison. At −2 °C, 3.5-week-old Col-0 plants displayed 17% electrolyte leakage at subambient CO2 (p ≤ 0.01) and this increased to 26 and 30% leakage at ambient and supra-ambient CO2, respectively ( Figure 2B). The response of electrolyte leakage by Col-0 plants to changing CO2 was similar at −2 and −4 °C, but the total amount of electrolyte leakage was greater at the lower temperature ( Figure 2B,C). Interestingly, no significant differences in electrolyte leakage in response to CO2 at −2 °C were observed using the Ws-2 ecotype (p > 0.5). However, at −4 °C, the electrolyte leakage was 2.7-fold greater (p ≤ 0.01) at supra-ambient than at sub-ambient CO2 ( Figure 2B,C). When the temperature was lowered to −6 °C, the CO2 effects on electrolyte leakage did not differ between Col-0 and Ws-2 (p > 0.05), which indicated that both ecotypes were approaching maximal electrolyte leakage ( Figure 2D). Interestingly, the freezing-tolerant Ox-CBF plant displayed somewhat unexpected results. Electrolyte leakage from Ox-CBF remained unaffected by the change in CO2 levels (p > 0.5) at −2 and −4 °C ( Figure 2B,C). However, significant differences (p ≤ 0.01) in electrolyte leakage were observed at supra-ambient CO2 treatments at −6 °C compared to sub-ambient and ambient CO2 treatments. In addition, electrolyte leakage was more than 50% less in the Ox-CBF line compared to Ws-2 ( Figure 2D), irrespective of CO2 treatment. This indicated a possible role for CBFmediated signaling in CO2-dependent responses of cold-induced electrolyte leakage. The above results suggest that the changes in CO2 level in air modified the freezing tolerance of Arabidopsis via CBF-mediated cold signaling (Figures 1 and 2). Electrolyte leakage due to freezing temperatures in Arabidopsis was altered in response to CO2 treatments. Individual 3.5 week-old wild type Arabidopsis plants ecotypes (ecotypes Col-0 and Ws-2) and the CBF3 over-expression (Ox-CBF) plant (A) were initially incubated at 4 °C for 3 h at subambient, ambient or supra-ambient CO2 (80, 400 and 800 μmol mol −1 , respectively). The chamber temperature was then lowered at 1 °C h −1 and plants were sampled for electrolyte leakage at −2 (B), −4 (C), and −6 °C (D). Electrolyte leakage into deionized water was measured as changes in conductivity and the percentages shown are mean values from three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at Figure 2. Electrolyte leakage due to freezing temperatures in Arabidopsis was altered in response to CO 2 treatments. Individual 3.5 week-old wild type Arabidopsis plants ecotypes (ecotypes Col-0 and Ws-2) and the CBF3 over-expression (Ox-CBF) plant (A) were initially incubated at 4 • C for 3 h at sub-ambient, ambient or supra-ambient CO 2 (80, 400 and 800 µmol mol −1 , respectively). The chamber temperature was then lowered at 1 • C h −1 and plants were sampled for electrolyte leakage at −2 (B), −4 (C), and −6 • C (D). Electrolyte leakage into deionized water was measured as changes in conductivity and the percentages shown are mean values from three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using Two-way ANOVA and Tukey HSD. White scale bars in (A): 0.5 cm.

Arabidopsis Mutants Defective in Stomatal Responses to CO 2 Exhibit Altered CBF-Mediated Cold Signaling
Considering the observations above, we further investigated whether stomatal response mutants mimicking supra-and sub-ambient CO 2 conditions would show similar changes in CO 2 control through CBF-mediated cold signaling. For these experiments, we used the ht1-2 mutant, hypersensitive to CO 2 , and the ca1ca4 double mutant, whose stomata are insensitive to CO 2 and remain open regardless of CO 2 levels ( Figure 3). All CBFs and CBF target genes tested here were significantly upregulated in Col-0 and in the ht1-2, and ca1ca4 mutants in response to cold treatment (i.e., 0 h vs. 2 h) ( Figure S2). These measurements were performed using 3.5-week-old plants exposed to supra-ambient, ambient, and sub-ambient CO 2 (800, 400, and 80 µmol mol −1 ) and maintained at 22 • C (0 h) or exposed to 4 • C for 2 h. The cold induced expression levels of CBF genes significantly decreased in response to increasing CO 2 levels in Col-0 ( Figure 3A). This CO 2 -dependent, cold stress-triggered change in CBF expression was abolished in the ht1-2 mutant, which closes stomata when exposed to sub-ambient CO 2 concentrations ( Figure 3A). In contrast, cold-induced CBF expression at supra-ambient CO 2 conditions was greater in the ca1ca4 mutant, which does not close stomata under supra-ambient CO 2, compared to Col-0 ( Figure 3A). This result indicates the involvement of molecular components (i.e., βCA1/4 and HT1) in CO 2 control of stomatal movement. The changes in the atmospheric CO 2 conditions did not have a significant effect on the basal expression of the CBF genes measured at room temperature ( Figure S3).  under varying CO2 levels (800, 400, and 80 μmol mol −1 ) was altered in two Arabidopsis CO2 stomatal response mutants along with wildtype control, Col-0. Transcript abundance was determined using Arabidopsis (Col-0) plants and two stomatal response mutants that were hypersensitive or insensitive to CO2 (i.e., ht1-2 and ca1ca4, respectively). Individual 3.5-week-old plants grown at 22 °C were exposed to 4 °C temperature for 2 h under ambient CO2 conditions. Relative values are means (×1000) of three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using two-way ANOVA and Tukey HSD. This experiment was performed independently from the one in Figure 1; therefore, the data values are not the same, but a similar trend was observed.

Defects in Stomatal Responses to CO2 Alter Freezing Tolerance in Arabidopsis
To confirm that the CO2-dependent modulation of CBF-mediated cold signaling in Arabidopsis was positively correlated with freezing tolerance, we examined electrolyte leakage at −2, −4, and −6 °C under ambient CO2 condition in the two CO2 stomatal response mutants ( Figure 4E). Electrolyte leakage from the ht1-2 mutant was about 2-fold greater (p ≤ 0.01) at −2 and −4 °C than that of Col-0 ( Figure 4A,B), which is consistent with reduced expression of both CBF and downstream CBF target genes in this mutant (see Figure 3A). At −6 °C, however, electrolyte leakage from the Col-0 ecotype and the ht1-2 mutant were similar (p > 0.05) ( Figure 4C). Unlike ht1-2, electrolyte leakage from the ca1ca4 double mutant was similar to that of Col-0 at both −2 and −4 °C (p > 0.05), but freezing tolerance at −6 °C was enhanced for the double mutant compared to Col-0 (p ≤ 0.01) ( Figure 4C). Values of EL50 (a freezing temperature that produces 50% cell damage) were −5, −4, and −6 °C for the Col-0, ht1-2 and ca1ca4 mutants, respectively. This indicates that the ht1-2 (more closed) and the ca1ca4 (more open) stomatal mutants were less and more freezing-tolerant than Col-0, respectively. As expected, the Ox-CBF line had the lowest electrolyte leakage at any freezing temperatures ( Figure 4A-C).
We further examined CO2 modulation of freezing tolerance using the CO2 response mutants, ht1-2 and ca1ca4 ( Figure 4D,E). The percentage of ion leakage at sub-ambient, ambient, and supraambient CO2 levels was higher in the ht1-2 mutant compared to the Col-0 at sub-ambient as well as ambient CO2 conditions indicating that low CO2-dependent freezing tolerance was abolished in the under varying CO 2 levels (800, 400, and 80 µmol mol −1 ) was altered in two Arabidopsis CO 2 stomatal response mutants along with wildtype control, Col-0. Transcript abundance was determined using Arabidopsis (Col-0) plants and two stomatal response mutants that were hypersensitive or insensitive to CO 2 (i.e., ht1-2 and ca1ca4, respectively). Individual 3.5-week-old plants grown at 22 • C were exposed to 4 • C temperature for 2 h under ambient CO 2 conditions. Relative values are means (×1000) of three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using two-way ANOVA and Tukey HSD. This experiment was performed independently from the one in Figure 1; therefore, the data values are not the same, but a similar trend was observed.
Next, we explored the link between the regulation of stomatal aperture and the expression of downstream CBF target genes. Interestingly, the two groups of genes, i.e., either partially ( Figure 1B) or fully ( Figure 1C) CO 2 -dependent, responded differently in the ht1-2 and ca1ca4 mutants under ambient CO 2 condition ( Figure S2B,C). The cold-induced expression of KIN1, KIN2, COR15a, and COR15b that were fully CO 2 -dependent was decreased in the ht1-2 mutant relative to Col-0 ( Figure S2B,C), while that of COR47, COR78, and ERD10, that were partially CO 2 -dependent ( Figure 1B), was greater (p ≤ 0.05) in the ca1ca4 mutant than in Col-0 ( Figure S2B). Enhanced expression of the partially CO 2 dependent CBF target genes and reduced expression of the fully CO 2 -dependent CBF target genes in the ca1ca4 mutant compared to their expression in the Col-0 wild type was observed ( Figure S2B). Furthermore, we studied the low-temperature expression of cold-regulated genes that are independent of CBF regulation, i.e., ARABIDOPSIS THALIANA HOMEOBOX PROTEIN 2 (ATHB2), RD22, and CHALCONE SYNTHASE (CHS). The cold-induced expression of ATHB2 and RD22 in the Col-0 plants were not significantly induced by sub-ambient CO 2 treatment. However, the expression levels of CHS were significantly higher in the ca1ca4 mutant, and sub-ambient CO 2 -treated Col-0 compared to ambient CO 2 -treated Col-0 plants (p < 0.05 to p < 0.01) ( Figure S1).
We further explored CO 2 s effect on cold-induced expression of CBF target genes, i.e., KIN1, COR15b, and COR78. In general, the cold-induced expression of these CBF target genes decreased in response to rising CO 2 levels in Col-0 and two stomatal response mutants, ht1-2 and ca1ca4 ( Figure 3B). However, this CO 2 -dependent, cold stress-triggered CBF target gene expression was reduced in the ht1-2 mutant compared to Col-0. In contrast, the cold-induced KIN1 expression at supra-ambient CO 2 conditions was slightly greater in the ca1ca4 mutant compared to Col-0, except COR78 ( Figure 3B).

Defects in Stomatal Responses to CO 2 Alter Freezing Tolerance in Arabidopsis
To confirm that the CO 2 -dependent modulation of CBF-mediated cold signaling in Arabidopsis was positively correlated with freezing tolerance, we examined electrolyte leakage at −2, −4, and −6 • C under ambient CO 2 condition in the two CO 2 stomatal response mutants ( Figure 4E). Electrolyte leakage from the ht1-2 mutant was about 2-fold greater (p ≤ 0.01) at −2 and −4 • C than that of Col-0 ( Figure 4A,B), which is consistent with reduced expression of both CBF and downstream CBF target genes in this mutant (see Figure 3A). At −6 • C, however, electrolyte leakage from the Col-0 ecotype and the ht1-2 mutant were similar (p > 0.05) ( Figure 4C). Unlike ht1-2, electrolyte leakage from the ca1ca4 double mutant was similar to that of Col-0 at both −2 and −4 • C (p > 0.05), but freezing tolerance at −6 • C was enhanced for the double mutant compared to Col-0 (p ≤ 0.01) ( Figure 4C). Values of EL 50 (a freezing temperature that produces 50% cell damage) were −5, −4, and −6 • C for the Col-0, ht1-2 and ca1ca4 mutants, respectively. This indicates that the ht1-2 (more closed) and the ca1ca4 (more open) stomatal mutants were less and more freezing-tolerant than Col-0, respectively. As expected, the Ox-CBF line had the lowest electrolyte leakage at any freezing temperatures ( Figure 4A-C).

Relationship between CO2 and Cold Signaling in Stomatal Movement
To further understand the association of CO2-induced stomatal movement and the CBFmediated cold response, we examined time-course changes in stomatal conductance (gs) using Ox-CBF and its corresponding wild-type plant, Ws-2. CO2 stomatal response mutants ht1-2 and ca1ca4 were also studied for their stomatal response along with the corresponding wild ecotype Col-0. Plants were initially pre-treated at supra-ambient CO2 for 20 min followed by sub-ambient CO2 for 40 min. These CO2 treatments sequentially induced stomata to close and then open. Plants were then . Defects in stomatal responses to CO 2 alter freezing tolerance in Arabidopsis. Electrolyte leakage was analyzed using Arabidopsis stomatal response mutant (ht1-2, ca1ca4) ant their wild-type ecotype, Col-0, as well as a CBF3 over-expression plant (Ox-CBF) and its wild-type ecotype, Ws-2. Individual 3.5-week-old plants (E) were initially incubated at 4 • C for 3 h at ambient CO 2 . The chamber temperature was then lowered at 1 • C h −1 and plants were sampled for electrolyte leakage at −2 (A), −4 (B), and −6 • C (C). Electrolyte leakage under varying CO 2 levels at −4 • C (D) was examined. Electrolyte leakage into deionized water was measured as changes of conductivity and the percentages shown are mean values from three biological replicates and two technical replicates (error bars indicate SEM). Bars labeled with different lowercase letters differed at p ≤ 0.05 and differences were determined using One-way (A-C), Two-way ANOVA (D) and Tukey HSD (A-D).
We further examined CO 2 modulation of freezing tolerance using the CO 2 response mutants, ht1-2 and ca1ca4 ( Figure 4D,E). The percentage of ion leakage at sub-ambient, ambient, and supra-ambient CO 2 levels was higher in the ht1-2 mutant compared to the Col-0 at sub-ambient as well as ambient CO 2 conditions indicating that low CO 2 -dependent freezing tolerance was abolished in the CO 2 hypersensitive mutant ( Figure 4D). The opposite result was observed in the ca1ca4 mutant in that percentage of ion leakage at sub-ambient, ambient, and supra-ambient CO 2 levels was lower in the ca1ca4 mutant compared to the Col-0 at supra-ambient as well as ambient CO 2 conditions, indicating that CO 2 -dependent freezing tolerance was retained in the CO 2 insensitive mutant ( Figure 4D).

Relationship between CO 2 and Cold Signaling in Stomatal Movement
To further understand the association of CO 2 -induced stomatal movement and the CBF-mediated cold response, we examined time-course changes in stomatal conductance (g s ) using Ox-CBF and its corresponding wild-type plant, Ws-2. CO 2 stomatal response mutants ht1-2 and ca1ca4 were also studied for their stomatal response along with the corresponding wild ecotype Col-0. Plants were initially pre-treated at supra-ambient CO 2 for 20 min followed by sub-ambient CO 2 for 40 min. These CO 2 treatments sequentially induced stomata to close and then open. Plants were then subjected to 4 • C cold treatment while maintaining the sub-ambient CO 2 treatment. For comparison, stomatal conductance was continuously measured with control plants exposed to 22 • C air temperatures. Whereas the conductance measurements indicated that stomata of Ws-2 wild type were almost completely closed after 30 min of a 4 • C treatment, conductance of Ox-CBF decreased by about 50% after 30 min of exposure to 4 • C ( Figure 5A). When stomatal conductance g s values were normalized, and slopes were calculated for stomatal aperture response to low-temperature treatment, the rates of stomatal closure were about double for Ws-2 compared to Ox-CBF ( Figure 5B). Interestingly, when Ox-CBF, a freezing tolerant plant, was kept at sub-ambient CO 2 at 22 • C, the stomata of Ox-CBF continued to open rapidly over the 30 min of sub-ambient CO 2 exposure. In contrast, the stomata of Ws-2 opened slowly under this treatment ( Figure 5A and Figure S4). The CO 2 stomatal response mutants ht1-2 and ca1ca4 displayed different stomatal responses compared to their corresponding ecotype Col-0. The hypertensive mutant closed its aperture early and insensitive ca1ca4 closed later than Col-0 at 4 • C. Col-0, ht1-2, and ca1ca4 all closed their stomata after 8-16 min at 4 • C ( Figure 5C,D).

Discussion
The function of stomata is to optimize gas exchange and water relations between the plant and the surrounding atmosphere. Changes in stomatal aperture establish a balance between two vital

Discussion
The function of stomata is to optimize gas exchange and water relations between the plant and the surrounding atmosphere. Changes in stomatal aperture establish a balance between two vital processes-CO 2 uptake from the air and water loss from the leaf. Therefore, stomatal movement is an intricately regulated process that involves simultaneously sensing CO 2 levels in the air, the water potential of the leaf and various other environmental parameters. In the current study, we demonstrated that varying atmospheric CO 2 altered cold stress signaling in Arabidopsis and that this partly involved the CO 2 signaling pathway associated with stomatal movement. Supra-ambient CO 2 decreased the cold-induced expression of three CBF genes, and this presumably diminished the expression of downstream cold responsive genes, KIN1, KIN2, COR15a, and COR5b. Contrary to the expected results, supra-ambient CO 2 did not suppress the cold induction of COR47, COR78, and ERD10 genes; however, sub-ambient CO 2 enhanced CBF's cold-induced expression of all CBF target genes. Therefore, the expression of CBF-dependent downstream cold-responsive genes in Arabidopsis was either fully or partly modulated by CO 2 .
It was previously suggested that freezing injury primarily results from dehydration [38] and that freezing tolerance involves resistance to dehydration [39][40][41]. The partially CO 2 -dependent CBF target genes, COR47, COR78, and ERD10, belong to a gene family that encodes late embryogenesis abundant (LEA) or LEA-related proteins and are involved in mitigating dehydration stress [19,[42][43][44][45][46]. LEA proteins enhance freezing tolerance by alleviating damage from dehydration associated with freezing injury [19]. In addition, the fully CO 2 -dependent CBF target genes identified in this study, KIN1, KIN2, COR15a, and COR15b, are known to function as antifreeze proteins that enhance the cryostability of the plasma membrane [47,48].
Varying CO 2 levels during low temperature treatment altered freezing tolerance of the Col-0 ecotype as measured by electrolyte leakage at −2 and −4 • C. This was not significant when electrolyte leakage was measured at −6 • C, at which ion leakage was more than 65%. Electrolyte leakage was enhanced by supra-ambient CO 2 and the opposite response was observed for low CO 2 treatments ( Figure 2B,C). Therefore, changes in freezing tolerance in the Col-0 ecotype were consistent with the observed changes of CBF expression and concomitant changes in the expression of downstream cold responsive genes in response to varying CO 2 levels described above. Elevated CO 2 was found to modulate the xylem sap pH, which increases the ABA concentration, affecting several COR gene products [2]. However, the precise mechanism by which CO 2 modulates the low-temperature induction of different groups of CBF-dependent target genes is not clear. The enhanced freezing tolerance observed in this study may be associated with the functions associated with the CBF genes studied [1]. The changes in the atmospheric CO 2 levels did not have a significant effect on the basal expression of the CBF genes measured at room temperature ( Figure S3). This suggests that the CO 2 concentrations in air altered the freezing tolerance of Arabidopsis Col-0 by mediating the expression of CBF and various CBF target genes.
Low-temperature experiments were performed using CO 2 stomatal response mutants in the Col-0 genetic background (i.e., the ht1-2 mutant, hypersensitive to CO 2 and the ca1ca4 double mutant, insensitive to CO 2 ) to understand the involvement of CO 2 -triggered stomatal movement in the CBF cold signaling pathway in Arabidopsis. Interestingly, CBF expression in response to cold treatment was greater in Col-0 than in either stomatal response mutant. In Arabidopsis, the HT1 gene product has been reported to control stomatal movement in response to CO 2 [36]. The finding that stomatal closure in the ht1-2 mutant at ambient CO 2 conditions resulted in decreased CBF expression relative to Col-0 ( Figure S2) is consistent with the observed effects of supra-ambient CO 2 on CBF expression in Col-0 ( Figure 1A). CO 2 modulation of cold-induced CBF expression was also evident in the CO 2 response mutants, ht1-2 and ca1ca4. The increase in cold-induced CBF expression at a sub-ambient CO 2 level in the wild-type Col-0 ( Figure 1A) was abolished in the CO 2 hypersensitive mutant, ht1-2 ( Figure 3A), while the decrease in cold-induced CBF expression at a supra-ambient CO 2 level ( Figure 1A) was slightly impaired in the CO 2 insensitive mutant, ca1ca4, compared to Col-0 ( Figure 3A). Given that sub-ambient CO 2 triggered stomatal opening and the observed cold-induced expression of CBF ( Figure 1A), our findings suggest that enhanced stomatal opening, as occurs in the ca1ca4 double mutant, did not fully implement the CBF-dependent cold signaling pathway in Arabidopsis. Furthermore, the suppression of cold-induced CBF expression in both stomatal response mutants implied that βCA1/4 and HT1 function upstream of CBF during cold signaling. The HTl protein kinase functions as a major negative regulator in the high CO 2 stomatal closure pathway. However, these mutants preserve their response to ABA, suggesting that HT1 acts upstream of the convergence of the CO 2 -and ABA-induced pathways [36]. Among CBF-independent cold signaling pathways, the ABA-dependent cold signaling pathway is important and regulates about~10% of cold responsive genes [15,49].
Furthermore, the expression analysis of seven CBF target genes in the ht1-2 and ca1ca4 stomatal response mutants also suggests the involvement of both CBF-dependent and independent responses. Based on the reduced expression of CBF genes shown in both the mutants ( Figure S2A), it was hypothesized that the cold-dependent expression of all seven CBF target genes would be lower in the ht1-2 and ca1ca4 double mutant compared to Col-0. The expression of KIN1, KIN2, COR15a and COR15b was reduced, which was consistent with the change in CBF expression in the ht1-2 mutant ( Figure S2C). However, the cold-induced expression of COR47, COR78, and ERD10 remained unchanged in this mutant relative to Col-0 under ambient CO 2 ( Figure S2B). On the contrary, in ca1ca4 mutant relative to Col-0, the expression of COR47, COR78 and ERD10 in response to low temperature was enhanced, while the expression of KIN1, KIN2, COR15a and COR15b in response to low temperature remained similar ( Figure S2B,C). One possible explanation is that βCA1/4 may mediate suppression of the cold-induced expression of COR47, COR78 and ERD10 in Col-0 and de-repression (or activation) of the expression of KIN1, KIN2, COR15a and COR15b via a CBF-independent cold signaling pathway. To explore the possibility of CBF-independent regulation that can be modified by CO 2 , we examined the expression of several CBF-independent cold-responsive genes upon cold treatment in the ca1ca4 mutant and compared to sub-and ambient CO 2 treated Col-0 plants. Interestingly, the low temperature increased the expression of CHS both in the ca1ca4 double mutant and in Col-0 in sub-ambient CO 2 , comparable to CHS expression changes in the Col-0 plants treated with ambient CO 2 . It was previously reported that a mutation in an Arabidopsis CHS gene (tt4) decreased anthocyanin biosynthesis [50] and this reduced the capacity of the plant for cold acclimation [51]. Conversely, plants with enhanced levels of anthocyanin and flavonoids had improved cold tolerance [52]. Based on these results, we propose that in Arabidopsis, there exists CO 2 regulation of both CBF-dependent and -independent cold signaling. In general, CO 2 modulation of cold induced expression of CBF target genes in the CO 2 hypersensitive mutant, ht1-2, was reduced compared to Col-0 while the opposite was the case in the CO 2 insensitive mutant, ca1ca4 ( Figure 3B).
Electrolyte leakage analysis in the two stomatal response mutants supports a role for stomatal movement in cold signaling, which may directly or indirectly affect cold-induced gene expression. The ht1-2 mutant results are consistent with the changes to the CBF-dependent cold signaling data discussed above and showed that electrolyte leakage increased when measured at −2 and −4 • C. Conversely, no differences were observed for electrolyte leakage between the ca1ca4 double mutant and Col-0 when freezing tolerance was measured at −2 and −4 • C. However, freezing tolerance of the ca1ca4 mutant exceeded that of Col-0 when measured at −6 • C. The latter finding may be partly due to the enhanced expression of cold responsive genes, such as COR47, COR78, ERD10, and CHS in ca1ca4 relative to Col-0 that was observed in Figure S2B. These observations were reflected in the EL 50 values, which indicated that the ca1ca4 mutant was more cold-tolerant and the ht1-2 mutant was more cold-sensitive than Col-0. These results demonstrate that cold-induced expression of CBFs and their target genes were enhanced, and resulting freezing tolerance was improved when plants were subjected to sub-ambient CO 2 . However, a stomatal response mutant that mimicked low-CO 2 treatment and possessed opened stomata did not enhance CBF-dependent cold-induced expression relative to Col-0. Additional experiments were performed to further examine CO 2 modulation of freezing-induced ion leakage using these stomatal response mutants, which demonstrated that increase in freezing tolerance at sub-ambient CO 2 condition in the Col-0 was abolished in the CO 2 hypersensitive mutant, ht1-2, while the opposite, decrease in freezing tolerance at supra-ambient CO 2 condition, was impaired in the CO 2 insensitive mutant, ca1ca4 ( Figure 4D).
To further understand the relationship between CO 2 -induced stomatal movement and CBF-mediated cold response, we measured g s on the Ox-CBF line and two CO 2 stomatal response mutants along with their corresponding wild-type plants ( Figure 5A-D and Figure S4). The conductance data demonstrated that low temperature (4 • C) induced stomatal closure even though Arabidopsis plants were treated with sub-ambient CO 2 . This finding demonstrates that rapid cold signaling in response to low temperature was capable of overriding the CO 2 signaling involved in stomatal movement. Low temperature diminishes plant survivability; therefore, it is likely that severe stress would induce stomatal closure even when sub-ambient CO 2 levels signal stomatal opening. However, it is important to note that measurements of g s using Ox-CBF, i.e., the freezing tolerant line, showed that stomata were about half opened 30 min after 4 • C cold treatment was initiated. Furthermore, low temperature-induced stomatal closure rates were delayed in a cold tolerant stomatal response mutant, ca1ca4, and accelerated in a cold sensitive mutant, ht1-2, compared to Col-0 ( Figure 5C and D). Stomatal conductance of the Ox-CBF (Overexpressed CBF3) mutant demonstrated a significant response in stomatal opening to sub-ambient CO 2 levels at 22 • C when compared to other plants, reflecting CO 2 s effect on enhanced freezing tolerance observed in the mutant at low temperatures (−6 • C) ( Figure 2D).

Conclusions
In summary, this study demonstrated that the induced gene expression due to varying CO 2 and low temperatures had a more significant impact on cold tolerance (electrolyte leakage) than the actual stomatal aperture. A general model for interactions of regulatory pathways for stomatal movement, CO 2 , and cold-induced gene expression is summarized in Figure 6. It is proposed that two β-carbonic anhydrases, i.e., βCA1/4, and a protein kinase, HT1, mediate CO 2 s effects on low temperature-induced CBF expression in Arabidopsis. This alters the expression of downstream cold-responsive genes. The model suggests that in response to supra-ambient CO 2 , βCA1/4 suppresses HT1, which decreases both CBF-mediated cold signaling and freezing tolerance. Conversely, in response to sub-ambient CO 2 , suppression of HT1 by βCA1/4 is reduced, enhancing CBF-mediated freezing tolerance.
We also propose the existence of βCA1/4-dependent repression and de-repression of downstream cold-responsive genes that likely occur in a CBF-independent manner. Interactions among genes upstream of CBF (e.g., ICE1/SCRM and SPCH) and the CO 2 signaling pathway in Arabidopsis are currently unknown. However, it is interesting to note that ABA-dependent cold signaling also occurs in plants and that ABA levels increase in plants exposed to low temperatures [15]. ABA is also an essential factor in stomatal closure [49]. Unraveling the role of ABA signaling in cold response may provide additional links between CO 2 and cold signaling pathways. Our findings demonstrated in the current study will add another layer of the complex regulation of the plant's response to environmental stimuli. Studying stomatal responses to CO 2 under cold stress will also help us to understand plant cold-responsive mechanisms under climate change conditions. This information can be translated into essential crops to enhance freezing tolerance and productivity. genes. The model suggests that in response to supra-ambient CO2, βCA1/4 suppresses HT1, which decreases both CBF-mediated cold signaling and freezing tolerance. Conversely, in response to subambient CO2, suppression of HT1 by βCA1/4 is reduced, enhancing CBF-mediated freezing tolerance. Figure 6. A schematic diagram for CO2-mediated CBF-dependent cold signaling and freezing tolerance in Arabidopsis. Black arrows indicate a proposed CO2 modulated cold signaling pathway (center). Grey arrows represent signaling pathways involved in stomatal movement (left) and freezing tolerance (right). This model proposes that CO2-modulated CBF-dependent cold signaling is mediated by βCA1/4 and HT1. In response to supra-ambient CO2, βCA1/4 suppresses HT1, which decreases both CBF-mediated cold signaling and freezing tolerance, while suppression of HT1 by βCA1/4 is reduced in response to sub-ambient CO2, and this enhances CBF-mediated freezing tolerance. In addition, βCA1/4 mediate suppression of CBF-independent expression of downstream cold-responsive genes. A potential new connection based on our preliminary results is shown as a dotted line. The question mark indicates that the gene responsible for the indicated pathway has not yet been definitely identified.
We also propose the existence of βCA1/4-dependent repression and de-repression of downstream cold-responsive genes that likely occur in a CBF-independent manner. Interactions among genes upstream of CBF (e.g., ICE1/SCRM and SPCH) and the CO2 signaling pathway in Arabidopsis are currently unknown. However, it is interesting to note that ABA-dependent cold signaling also occurs in plants and that ABA levels increase in plants exposed to low temperatures [15]. ABA is also an essential factor in stomatal closure [49]. Unraveling the role of ABA signaling in cold response may provide additional links between CO2 and cold signaling pathways. Our findings demonstrated in the current study will add another layer of the complex regulation of the plant's response to environmental stimuli. Studying stomatal responses to CO2 under cold stress will also help us to understand plant cold-responsive mechanisms under climate change conditions. This information can be translated into essential crops to enhance freezing tolerance and productivity.

Plant Materials and Growth Conditions
Arabidopsis thaliana L. Heynh wild type and mutant plants were grown in controlled environment chambers (model M-15, EGC corp., Chagrin Falls, OH, USA) under ambient 400 ppm CO2 and day/night temperatures of 22 °C/18 °C. Two wild type (Col-0 and Ws-2) and two mutant Figure 6. A schematic diagram for CO 2 -mediated CBF-dependent cold signaling and freezing tolerance in Arabidopsis. Black arrows indicate a proposed CO 2 modulated cold signaling pathway (center). Grey arrows represent signaling pathways involved in stomatal movement (left) and freezing tolerance (right). This model proposes that CO 2 -modulated CBF-dependent cold signaling is mediated by βCA1/4 and HT1. In response to supra-ambient CO 2 , βCA1/4 suppresses HT1, which decreases both CBF-mediated cold signaling and freezing tolerance, while suppression of HT1 by βCA1/4 is reduced in response to sub-ambient CO 2 , and this enhances CBF-mediated freezing tolerance. In addition, βCA1/4 mediate suppression of CBF-independent expression of downstream cold-responsive genes. A potential new connection based on our preliminary results is shown as a dotted line. The question mark indicates that the gene responsible for the indicated pathway has not yet been definitely identified.  Table S1. The photoperiod was 16 h light and 8 h dark with a photosynthetic photon flux density of 110 ± 10 µmol m −2 s −1 . Relative humidity was not controlled but was 70 to 80% during the light period . Individual 10 day-old seedlings were transplanted to 1.5 × 1.5 inch pots filled with 85% peat mixture (MetroMix, Morgantown, PA, USA) and vermiculite (Grace Construction Products, Cambridge, MA, USA). Pots were irrigated from the bottom with water and received one quarter strength nutrient solution once or twice weekly.

Low Temperature and CO 2 Treatments
Four to twelve pots containing a single, 3.5-week old plant were placed in a 45 × 37 × 12 cm deep incubation chamber that was clamped shut with a clear acrylic plastic lid. An adhesive foam seal (Armaflex Insulation Tape (9.1 × 3 × 50 mm), Armacell LLC., Mebane, NC, USA) was placed between the top of the incubation chamber and the lid to minimize air exchange. Illumination (115 ± 10 µmol m −2 s −1 PPFD) was provided by a bank of fluorescent lamps (model 48T10/CW, 110W, VHO, General Electric, Buffalo, NY, USA) located 1 m above the upper surface of the incubation chamber. Control plants were enclosed in the sealed incubation chamber at 22 • C and kept in the same controlled environment chamber used for plant growth. Modified air of known CO 2 concentrations and temperature was pumped through the entrance and exit holes in the sealed lid of the incubation chamber at a flow rate of 3 to 4 l min −1 . The CO 2 concentration of the air entering and exiting the incubation chamber was monitored with an infrared gas analyzer (model WMA-3, PP systems, Haverhill, MA, USA). Supra-ambient CO 2 was obtained by slowly injecting 1% CO 2 into the airstream entering the incubation chamber. Standard gases were from Airgas (Hyattsville, MD, USA), and the other contents of the cylinder were approximately 79% N 2 and 20% O 2. Sub-ambient CO 2 was obtained by scrubbing the air entering the incubation chamber with soda-lime and then injecting known amounts of CO 2 from a standard gas cylinder via a micro metering valve. The three CO 2 treatments were 80, 400, and 800 ± 10 ppm (µmole CO 2 mole −1 air). Low temperature treatments were initiated by placing the incubation chamber, gas cylinder, CO 2 scrubber and infrared gas analyzer in a walk-in cold room maintained at 4 • C. After an equilibration period to attain desired CO 2 levels and treatment temperatures, plants were placed in the incubation chamber and the lid was rapidly sealed. The time to attain the desired CO 2 concentration within the incubation chamber after sealing the plants inside was about 5 to 10 min for a small chamber (for transcript analysis) and 15 to 20 min for a big incubation chamber (for freezing tolerance assay). Illumination was as described above and temperatures within the incubation chamber were measured with a fine wire thermocouple. Temperature and CO 2 treatments were maintained for 2h before the incubation chamber was quickly opened and rosette leaves were collected into liquid N 2 for transcript analysis and stored at −80 • C until analysis.

Transcript Analysis
Frozen rosette leaves were ground to a fine powder under liquid N 2 using a sterile mortar and pestle. Total RNA was extracted from 30 mg fresh weight of frozen tissue and cDNA synthesis was performed as previously described [53,54]. Up to 5 µl of diluted cDNA was used in qPCR reactions contained in a total volume of 20 µl. The gene-specific primers for reference genes Actin 2 (ACT2) and Transcription Elongation Factor (TEF) and for CBF1-3, COR78, COR15a, and ERD10 from Arabidopsis were previously described [55]. All other primer pairs were designed using the Primer 3 program and primer sequences used for qPCR are listed in Table S2. qPCR was performed as previously described [54] and relative transcript abundance was measured by normalizing the expression level of each gene of interest using the expression of reference gene ACT2. ACT2 and TEF showed the same expression pattern, but % PCR efficiency of ACT2 was greater than TEF (97 vs. 94% respectively); therefore, ACT2 data were used for normalization. The presence of a single PCR product was further verified by dissociation analysis. Three independent experiments were performed with three to four plants per treatment. Significant differences were determined using a one-way or two-way ANOVA procedure in combination with Tukey HSD (StatView 5.0, Mountain View, CA, USA).

Electrolyte Leakage Freeze Test
Electrolyte leakage tests were performed on excised leaves as previously described [56,57] and freezing tolerance was determined by the method of Sukumara and Weiser [58]. Four or more excised Arabidopsis leaves from 4-week old plants were placed in 13 × 100 mm glass culture tubes (Corning Inc., Corning, NY, USA), which were maintained at 4 • C for 3 h in a low-temperature chamber (LHU-113 Temperature & Humidity Cabinet, ESPEC Corp., Osaka, Japan). Freezing treatments were initiated by adding a few ice chips to each tube. Following a 3 h equilibration period, the chamber temperature was lowered manually by 1 • C h −1 . Samples were withdrawn from the temperature chamber after plants were exposed to −2, −4 and −6 • C for 1h in descending order. The harvested samples were immediately stored on ice until the end of the experiment and were then thawed overnight in a cold room at 4 • C. To estimate electrolyte leakage, 2 mL of distilled water was added to each tube, and the samples were gently shaken in a water bath (Model G76D, New Brunswick Scientific Co., Inc., Edison, NJ, USA) at room temperature for 3 h. The resulting solution's electrical conductivity was measured using a conductivity meter (Model 35, YSI, Yellow Springs, OH, USA). A value for 100% leakage was obtained by freezing the tested samples at −80 • C for 1 h before extraction. Levels of CO 2 within the temperature chamber were controlled during the freezing treatment, as described above, using access ports located on the chamber's side. The air within the chamber also was circulated with an internal fan to ensure proper mixing. Significant differences were determined using a one-way or two-way ANOVA procedure in combination with Tukey HSD (StatView 5.0, Mountain View, CA, USA).

Stomatal Conductance Determinations
Stomatal conductance (gs) was measured using a portable Photosynthesis System (model 6400, Open System 4.0, Li-Cor Inc., Lincoln, NE, USA) using a 0.85-dm 3 natural sun-lit cuvette. Experiments were performed with single, attached leaves of wild type Arabidopsis, i.e., ecotypes Col-0, Ws-2, and related mutants. All measurements were performed in growth chambers matching those used for plant growth, and irradiance was provided by fluorescent bulbs as described above. Treatments were initiated 3 h after the start of the light period with plants initially maintained at 22 • C, 110 µmol m −2 s −1 PPFD, and at 400 ppm CO 2 . Plants were first exposed to 800 ppm CO 2 for 20 min before being switched to low CO 2 (80 ppm) for 40 min. Half of the plants were then transferred to 4 • C for 30 min, and the remaining plants served as ambient temperature controls. Data were collected continuously from pre-treatment to the end of the experiment. Conductance values (gs) were calculated by the Photosynthesis System and were collected using five to eight plants from two independent experiments.

Conflicts of Interest:
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.