A pot experiment was conducted in climate-controlled greenhouses at the Faculty of Science, University of Copenhagen, Taastrup, Denmark. Seeds of isogenic tomato cv. Ailsa Craig (AC) and an ABA-deficient tomato mutant (flacca; Solanum lycopersicum) were provided by the Lancaster Environment Centre (Lancaster University, United Kingdom). The ABA-deficient mutant is impaired in the oxidation of ABA-aldehyde to ABA, thus possessing significantly lower endogenous ABA concentrations than AC (Sagi et al., 2002). All plants were grown in 4-L pots filled with 2.6 kg of peat material (Plugg-och Såjord-Dry matter ca. 110 kg m⁻³, organic matter >95%, pH 5.5–6.5 and EC 1.5–2.5 mS cm⁻¹). Four weeks after sowing, perlite was covered on the soil surface to minimize evaporation and fertilizers as NH₄NO₃ (2.8 g) and H₂KPO₄ (3.5 g) per pot were added together with irrigation water to avoid any nutrient deficiency.

From sowing, the plants were grown in two greenhouse cells with CO₂ concentrations of 400 ppm (ambient CO₂, a[CO₂]) and 800 ppm (elevated CO₂, e[CO₂]), respectively. During the experiment, the actual daily average [CO₂] was 420.2 and 804.2 ppm, respectively. The [CO₂] in the cells was sustained by pure CO₂ emission from a bottle tank and distributed evenly by the internal ventilation system. The [CO₂] in the cells was monitored every 6 s by a CO₂ Transmitter Series GMT220 (Vaisala Group, Helsinki, Finland). The climate conditions in the two glasshouse cells were set at: 20/16 ± 2°C day/night air temperature, 60 ± 2% relative humidity, 16 h photoperiod, and > 500 μmol m⁻² s⁻¹ photosynthetically active radiation (PAR) supplied by sunlight plus LED lamps (Philips GreenPower LED toplighting, Denmark). The vapor pressure deficit in the greenhouse cells ranged from 0.8 to 1.0 kPa.

After seedling establishment, the pots were constantly irrigated to 90% of the pot holding capacity. The soil drying treatment started at March 6 and March 15, 2018 for AC tomato and flacca, respectively (5 weeks after sowing). In each cell and genotype, at the onset of drought stress, four well-watered plants were harvested as the initial control; during progressive soil drying, four plants were well watered to 95% of the pot water holding capacity as control plants, and other 16 tomato plants were subjected to progressive soil drying by withholding irrigation from the pots until transpiration rate (E) decreased to ~10% of the control plants. The drought-stressed plants were harvested four times at different soil water statuses; and for each cell and each genotype at each harvest, four plants were harvested.

Before imposing drought stress, the maximum instantaneous transpiration rate (E) of AC was 32.2 and 41.0% lower under a[CO₂] and e[CO₂], respectively, compared to flacca. Moreover, the E_max of AC under e[CO₂] was 13.1% lower than those under a[CO₂]. During progressive soil drying, E of AC under e[CO₂] started to decline at a lower FTSW threshold (C_E) than that under a[CO₂] (i.e., 0.46 vs. 0.62). While in flacca, there was no notable difference in E_max and C_E between the two [CO₂] environments. In addition, the C_E of AC was higher than that of flacca under both a[CO₂] and e[CO₂] growth environments (i.e., 0.62 vs. 0.37 and 0.46 vs. 0.32, respectively; Figures 1A,B; Table 1). At the last harvest, in well-watered AC, depression on E by e[CO₂] became less significant (Supplementary Figure S1A).

Although e[CO₂] decreased E in AC, it increased plant daily water consumption in both genotypes when FTSW ranged from 1.0 to ca. 0.4 (plants maintained the maximum E in this range of FTSW). However, when FTSW dropped below 0.4, there was no difference in daily water consumption of both genotypes (Figures 1C,D). Plant water use efficiency (WUE_plant) was obtained from the slopes of the linear relationship between cumulative water consumption and cumulative dry weight. AC plants grown under e[CO₂] had significantly greater WUE_plant than those under a[CO₂] (p = 0.007), whereas no significant difference (p = 0.054) in WUE_plant for flacca was found between two [CO₂] environments, though a clear tendency of increasing WUE_plant when grown at e[CO₂] was observed for both genotypes (Figures 1E,F).

At the final harvest, drought stress decreased leaf water potential (Ψ_leaf), root water potential (Ψ_root), and leaf relative water content (RWC) in both AC and flacca, whereas e[CO₂] had no influence on the two variables under both watering conditions. In addition, e[CO₂] significantly increased osmotic adjustment (OA) in both AC and flacca, and AC had lower OA than flacca under both [CO₂] environments (Figure 2).

When FTSW ranged from 1.0 to ca. 0.4, as the same with E, plants maintained the maximum K_leaf and K_root, and AC had higher K_{leaf max} and K_{root max} than flacca especially under a[CO₂]. Under e[CO₂], AC had 30.29 and 29.73% lower K_{leaf max} and K_{root max} than a[CO₂]-grown plants, respectively, whereas those effects were attenuated in flacca (19.56 and 14.55% lower K_{leaf max} and K_{root max} resulted by e[CO₂]). During the course of progressive soil drying, in both genotypes, the FTSW thresholds (C_kl and C_kr) did not respond to e[CO₂] (Figure 3). At the last harvest, in well-watered AC, depression on K_leaf and K_root by e[CO₂] was still observed (Supplementary Figures S1B,C).

The ABA concentration in xylem sap ([ABA]_xylem) increased exponentially with the declining of FTSW in AC plants grown under both [CO₂] environments, while the increase was more pronounced under e[CO₂] than under a[CO₂] and resulting in a greater [ABA]_xylem in the e[CO₂] plants compared to a[CO₂] plants during progressive soil drying. For flacca, [ABA]_xylem increased during soil drying and was identical between the two [CO₂] environments, and remained significantly lower than those of the AC plants (Figure 4).

In leaves of well-watered AC plants, transcripts of four PIPs (PIP1;3, PIP2;1, PIP2;8, and PIP2;9) responded to e[CO₂] with a 2–9-fold downregulation of expression levels, and those effects were absent in flacca. Moreover, transcripts of those same four PIPs were 2–15-fold downregulated by drought under a[CO₂], and an interaction between [CO₂] and drought stress was observed, showing that those downregulation became less significant under e[CO₂]. In flacca, drought also downregulated the transcript levels of PIP1;3, PIP2;8, and PIP2;9 with more than 2-fold change under both [CO₂] environments. When comparing leaf PIP transcript levels between the two genotypes, AC showed higher expression levels of PIP2;1 and PIP2;4 than flacca under a[CO₂]. In addition, within five PIPs, PIP2;8 and PIP2;9 were dramatically influenced by drought in both AC and flacca (Figure 5; Table 2).

Generally, PIP expression levels in AC roots were less sensitive to e[CO₂] than the same PIPs in leaves. In well-watered AC plants, PIP2;1, PIP2;8, and PIP2;9 showed indications of being 2–3-fold downregulated in response to e[CO₂]. In flacca, all five PIPs responded to e[CO₂], but downregulations of PIP1;3, PIP2;1, and PIP2;8 expressions were only significant in stressed plants. In both genotypes, 2–8-fold downregulation of PIP1;3 and PIP2;4 were observed in response to drought under both [CO₂] environments; whereas a slight upregulation of PIP2;1 transcript level was observed in both AC and flacca. Interestingly, in AC plants, an interaction between [CO₂] and drought stress on expression levels of PIP2;8 and PIP2;9 also existed, showing attenuated drought response under e[CO₂]. When comparing root PIP expression levels between the two genotypes, the only expression of PIP2;9 was found to be significantly different (Figure 6; Table 2).

In leaves of AC plants, the expression levels of OST1 were slightly upregulated by e[CO₂], and there was an interaction between [CO₂] and drought stress, being more significant in stressed plants. Under drought stress, OST1 transcripts were 2–5-fold upregulated in both AC and flacca. Moreover, there was no difference in OST1 transcripts between the two genotypes (Figure 7).

PCA plots of K_leaf, K_leaf, and [ABA]_xylem and gene expression of all leaf/root PIPs and OST1 for AC and flacca were depicted in Figure 8. Overall, PCA showed clear group formation related to the [CO₂] and watering treatments, and PC1 and PC2 axes explained 78.1 and 77.4% of the variation for AC and flacca, respectively. In both genotypes under both [CO₂] (Figures 8A,B), the well water treatments were clustered in the same direction as the vector for the following parameters related to K_leaf, K_leaf, and a majority of PIPs; the drought stress treatments were clustered in the same direction as the vector for OST1 and ABA. However, in AC, e[CO₂] showed a strong tendency with root PIP2;1, and ABA presented the smallest angle to OST1; in flacca, leaf PIP2;4 was opposed by most PIPs and hydraulic conductance, and no covariation of ABA and OST1 was observed. In addition, under well-water conditions, a[CO₂] and e[CO₂] treatments are separated on each side of the plot only in AC. In Figure 8C, PCA of all the parameters in two genotypes revealed that AC and flacca observations clustered toward the top and the bottom of the plot, respectively, and separation along PC2 confirmed the different responses in two genotypes regarding to ABA, OST1, leaf PIP2;4, root PIP2;8 and PIP2;9.

There is a common understanding that e[CO₂] can ameliorate the negative impacts of drought stress and enhance plant drought tolerance due to curtailed transpiration and enhanced carbon accumulation, hereby leading to an increase in WUE_plant (Robredo et al., 2007; van der Kooi et al., 2016; Li et al., 2020). Consistent with this, here we found that under mild drought stress, AC plants grown under e[CO₂] had lower E and greater WUE_plant despite of an increased daily water consumption at the whole plant level (Figure 1). An increase in plant water consumption could be ascribed to a larger leaf area when grown under e[CO₂], which has previously been reported by many researchers (Manea and Leishman, 2015; Temme et al., 2019). In flacca, it has been found that opened stomata resulted in faster photosynthetic induction at a cost of reduced WUE, which is the case in the present study (Kaiser et al., 2020). During progressive soil drying, it is notable that the FTSW threshold at which E started to decrease was retarded by e[CO₂] only in AC but not in flacca, affirming our earlier findings that e[CO₂] could delay stomatal response to drought in tomato plants in an ABA-dependent way (Wei et al., 2020). Kaiser et al. (2020) also reported that compared with wild-type, flacca lacked the ability to respond to CO₂ or vapor pressure deficit in terms of g_s. At last harvest, although Ψ_leaf and Ψ_root did not respond to e[CO₂], slight effects of e[CO₂] on RWC and OA were found in both AC and flacca (Figure 2). A high OA might be due to increased accumulation of solutes, which is considered to be associated with drought resistance (Jensen et al., 1996), and the accumulation of compatible solutes also could protect specific cellular functions and maintain leaf turgor (Johnson et al., 2002; Liu and Stützel, 2002). Therefore, plants grown under e[CO₂] might possess improved drought resistance due to better water-holding capacity in leaves.

Plant water balance under drought stress is controlled by a fine-tuned coordination between transpirational water loss and plant hydraulic conductance. As shown in Figure 1 and Figure 3, both AC and flacca plants maintained relatively high E, K_leaf and K_root until FTSW declined to a threshold value (ca. 0.4). Thereby, under mild drought stress conditions, plants still possessed favorable leaf and root hydraulic properties, which could be able to minimize the risk of excessive dehydration via a coordinated balance between leaf transpiration and water transport from root to leaf, consistent with the previous findings (Brodribb and Holbrook, 2004; Creek et al., 2018). Moreover, stomatal properties (e.g., stomatal size and density) are responsible for leaf water leak even when stomata are fully closed, which have a great impact on plant water use strategies (Machado et al., 2021). Dramatic higher stomatal density in flacca than in AC has been reported by Fang et al. (2019), which might explain the significant low K_root accompanied by still relatively high E and K_leaf in stressed flacca, representing an extensive hydraulic failure and dysfunction in the regulation of the plant water balance.

Although several studies reported that e[CO₂] could decrease plant hydraulic conductance (Bunce, 1996; Robredo et al., 2007; Domec et al., 2009; Hao et al., 2018; Fang et al., 2019), there is no consensus about the roles of K_leaf and K_root in maintaining plant water status when combining e[CO₂] and drought stress. As illustrated in Figure 3, under mild drought stress conditions (i.e., FTSW ranged from 1.0–0.4), e[CO₂] decreased both K_leaf and K_root, in two genotypes, though those depression was attenuated in flacca. However, at the last harvest, well-water flacca lacked the ability to respond to e[CO₂] in terms of hydraulic conductance (Supplementary Figure S1), consistent with our the previous study (Fang et al., 2019). It should be noted that a slight increasing trend in [ABA]_xylem of flacca was still observed during progressive soil drying, which might contribute to the impact of e[CO₂] on hydraulic conductance. Those contrasting results raised the possibility that different hydraulic responses existed under different intensities of drought stress. Overall, it is plausible that ABA is involved in the e[CO₂]-induced changes in plant hydraulic conductance.

Under severe drought stress, i.e., FTSW < 0.4, K_leaf and K_root decreased sharply in both AC and flacca as well as E (Figure 3). Although it is widely believed that ABA could alter plant hydraulic properties (Parent et al., 2009; Rosales et al., 2019), our results demonstrated that hydraulic response to severe drought could be ABA-independent and might be attributed to decreased leaf turgor as reported in our previous study (Wei et al., 2020). Moreover, as shown in Figure 4, the increase of [ABA]_xylem in AC during progressive soil drying was more pronounced under e[CO₂]. Although a study has shown that rapid ABA biosynthesis in angiosperms occurred in leaves rather than in root (Sussmilch et al., 2018), recently by using reciprocal grafting technique on AC and flacca, Li et al. (2018) found that long-distance ABA transport could affect foliar ABA concentrations under salinity but not control conditions. Moreover, when soil drying got severe, the release of root ABA into xylem could be attenuated due to the reduced water flow in roots (Liang et al., 1997), and decreased hydraulic conductance also might exert a drag on the delivery rates of ABA. Those results provide a clue concerning mechanisms of ABA metabolism and transport under e[CO₂].

Several studies have demonstrated that the regulation of plant hydraulic conductance by ABA under drought stress is associated with modulating aquaporin activities (Parent et al., 2009; Dayer et al., 2017; Veselov et al., 2018), and e[CO₂] could decrease the abundance of PIP1 and PIP2 proteins in both leaves and roots (Zaghdoud et al., 2013). In the present study, AC plants had higher transcript levels of genes encoding two and one out of five PIPs in leaves and roots, respectively, compared with those in flacca (Figures 5, 6). High transpiration demand could result in transcriptional upregulation of specific root PIPs, thus maintaining a favorable plant water status (Sakurai-Ishikawa et al., 2011). Therefore, discordance between high transpiration and low PIP levels in flacca might disturb plant hydraulic homeostasis, leading to desiccation of the leaves, which was the case in the present study. In addition, under well-watered conditions, e[CO₂] had a more pronounced influence on leaf PIPs than root PIPs (Figures 5, 6), indicating that leaf PIPs might be more sensitive to the change in [CO₂]. Our previous study has demonstrated that decreases in K_leaf and K_root of AC tomato plants under e[CO₂] were associated with downregulation of PIPs, which was not the case for ABA-deficient mutant (Fang et al., 2019). However, here we found that the root PIPs of flacca still responded to e[CO₂] though those effects were only shown in stressed plants, which was consistent with slight changes in hydraulic conductance of flacca (Figures 3, 6; Supplementary Figure S1; Table 2). In summary, e[CO₂] could modulate plant hydraulic conductance via the regulation of leaf and root PIP transcript levels, where ABA had an obligate role but was dependent on the watering conditions.

At the end of drought treatment, there were dramatic decreases in E, Ψ_leaf, Ψ_root, RWC, K_leaf, and K_root, which accompanied by changes in transcriptional regulation of PIPs in both genotypes and [CO₂] environments (Figures 1–3, 5, 6). The PCA in Figures 8A,B also confirmed that e[CO₂] affected hydraulic conductance and the gene expression of PIPs only under well-watered but not drought-stressed conditions. Interestingly, the regulation of leaf PIP expression levels by drought was more significant compared with those in roots. As the ability of plants to conserve water during drought stress involves timely and sufficient downregulation of specific AQPs (Zupin et al., 2017), leaf PIPs might be more sensitive in their response to severe drought stress. In roots of both AC and flacca, the transcript levels of four out of five PIPs were downregulated by drought stress (Figure 6), which is in accordance with the previous studies (Shekoofa and Sinclair, 2018). Interestingly, in AC plants grown under e[CO₂], the expression of PIP2;8 and PIP2;9 were not affected by drought, indicating that e[CO₂] might disturb the ABA-mediated response of some root PIPs to drought stress, which merits further investigations. It should be noted that the expression of PIP2;1 in roots was upregulated by drought in both AC and flacca (Figure 6), which represented a different expression pattern when compared to other root PIPs, indicating that PIP2;1 of tomato plants might play a crucial role in root water uptake especially under drought stress. In tobacco, NtPIP2;1 showed remarkable water transportability, though it was downregulated by drought stress in response to decreased K_root (Mahdieh et al., 2008; Secchi et al., 2016). Root PIPs and root anatomical properties were both correlated with hydraulic traits, and there is evidence that the contribution of root PIPs to K_root was enhanced by drought stress (up to 85%; Grondin et al., 2016). Therefore, a large genetic diversity for AQPs expression and root anatomy among various species might exit, and more research is necessary to underpin the multiple functions related to hydraulic traits. It is noteworthy to mention that here we only studied the leaf and root PIP transcript abundance under severe drought, and the previous report has shown that the expression patterns of PIPs were varied under different water stress intensities (Galmés et al., 2007). Therefore, further investigations to reveal the coordination between hydraulic conductance and PIP activities under progressive soil drying are needed.

OST1 is a crucial component in the CO₂ signaling pathway, and it may accelerate CO₂ permeability under high [CO₂] conditions (Wang et al., 2015). Here, we found that a significant effect of e[CO₂] on the expression of OST1 only existed in AC, suggesting that the function of OST1 in CO₂ transport might require ABA involvement. Recent reports showed that OST1 and PIP2;1 could function together to accelerate both water transport and CO₂ transport in guard cells (Grondin et al., 2015; Wang et al., 2015), though here we did not found an obvious correlation between leaf transcripts of PIP2;1 and OST1. In addition, there was no difference in expression of OST1 between AC and flacca under either treatment, though flacca plants had dramatically lower Ψ_leaf and RWC under both well-watered and drought-stressed conditions. In Figure 8, the results of PCA highlight a significant difference in drought response of two genotypes regarding to the covariation of ABA and OST1. Nevertheless, OST1 showed an upregulation of transcript levels under drought stress in both AC and flacca under either [CO₂] environments (Figure 7), suggesting that factors rather than ABA (e.g., osmotic potential, Yoshida et al., 2006) could directly regulate the OST1-mediated drought response.

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.