Role of Hydraulic Signal and ABA in Decrease of Leaf Stomatal and Mesophyll Conductance in Soil Drought-Stressed Tomato

Drought reduces leaf stomatal conductance (gs) and mesophyll conductance (gm). Both hydraulic signals and chemical signals (mainly abscisic acid, ABA) are involved in regulating gs. However, it remains unclear what role the endogenous ABA plays in gm under decreasing soil moisture. In this study, the responses of gs and gm to ABA were investigated under progressive soil drying conditions and their impacts on net photosynthesis (An) and intrinsic water use efficiency (WUEi) were also analyzed. Experimental tomato plants were cultivated in pots in an environment-controlled greenhouse. Reductions of gs and gm induced a 68–78% decline of An under drought conditions. While soil water potential (Ψsoil) was over −1.01 MPa, gs reduced as leaf water potential (Ψleaf) decreased, but ABA and gm kept unchanged, which indicating gs was more sensitive to drought than gm. During Ψsoil reduction from −1.01 to −1.44 MPa, Ψleaf still kept decreasing, and both gs and gm decreased concurrently following to the sustained increases of ABA content in shoot sap. The gm was positively correlated to gs during a drying process. Compared to gs or gm, WUEi was strongly correlated with gm/gs. WUEi improved within Ψsoil range between −0.83 and −1.15 MPa. In summary, gs showed a higher sensitivity to drought than gm. Under moderate and severe drought at Ψsoil ≤ −1.01 MPa, furthermore from hydraulic signals, ABA was also involved in this co-ordination reductions of gs and gm and thereby regulated An and WUEi.


INTRODUCTION
Soil water scarcity is one of the major environmental constraints to the plant physiological processes and yield (Easlon and Richards, 2009;Olsovska et al., 2016). To achieve high plant water-use efficiency under a drier environment in the future, it is essential to improve crop photosynthesis and productivity with a given unit of water . For C 3 plants, leaf photosynthesis is strongly limited by three factors, i.e., stomatal conductance (g s ), mesophyll diffusion conductance to CO 2 (g m ), and biochemical photosynthetic capacity (Grassi and Magnani, 2005;Cano et al., 2013). g s and g m determine the diffusion of CO 2 from ambient air of leaf to sub-stomatal cavities and from the sub-stomatal cavities to chloroplast stroma, respectively (Flexas et al., 2002;Niinemets et al., 2009). Recent studies have shown that both g s and g m were the main limitations for maximum photosynthesis under drought conditions (Tosens et al., 2016;Wang et al., 2018). Therefore, revealing the mechanisms underlying the decreases of g s and g m in response to drought is necessary for enhancing our understanding of plant adaptation to water limitation.
Different regulatory mechanisms such as chemical messengers like abscisic acid (ABA), electrical signals, and hydraulic signals have been identified in the control of stomatal movement (Dodd, 2005;Ache et al., 2010;Tombesi et al., 2015;Huber et al., 2019). Despite the large list of candidates in regulating guard cells, ABA and hydraulic signals have gained most of the attention in regulating stomatal aperture. ABA is a phytohormone that has been involved in different strategies of plants to avoid excessive water loss, and many reports demonstrated its important role in stomatal control (Wilkinson and Davies, 2002;Assmann and Jegla, 2016). The decrease of g s in response to drought has been generally modulated by the accumulation of leaf ABA in a wide number of plant species including soybean, grapevine and tomato (Liu et al., 2005;Tombesi et al., 2015;Yan et al., 2017). However, stomata closed with a wide range of variations of leaf hydraulic signals, such as leaf water potential (Ψ leaf ), possibly due to differences of experimental plant materials and the intensity of applied drought under investigation. For example, g s decreased with decreasing Ψ leaf during leaf dehydration (Kim et al., 2012;Wang et al., 2018). On the contrary, other studies showed that stomata closed with little change in Ψ leaf under moderate soil drying, but both parameters decreased under severe drought (Tardieu, 1998;Yan et al., 2017), or g s decreased as Ψ leaf increased under mild soil drying but then no significant relationship existed between both variables with continued soil drying (Kudoyarova et al., 2007). It is difficult to explore the response of Ψ leaf and g s under a single soil water condition. Progressive soil drying, representing a natural process of soil water loss, could help us explore the dynamic responses of g s to Ψ leaf during drying process.
Leaf mesophyll conductance to CO 2 (g m ) has been recognized to be finite, variable, and rapid acclimation to varying environmental conditions. Although a reduction in g m response to soil drought has been reported in many studies, the mechanisms underlying this reduction have not been elucidated substantially (Flexas et al., 2002;Théroux-Rancourt et al., 2014;Sorrentino et al., 2016). Recent studies on hydraulic signals suggested that the parallel decreases in g s and g m were caused by leaf hydraulic vulnerability as a result of decrease in Ψ leaf (Wang et al., 2018). Similarly, g m was strongly correlated with leaf hydraulic conductance (K leaf ), as the ratio of transpiration rate to the water potential driving force across the leaf (K leaf = transpiration/ Ψ leaf ), across species under light-saturated conditions . This correlation between g m and leaf hydraulic signals might be due to CO 2 partially shared common diffusion pathways with H 2 O through mesophyll tissues (Ferrio et al., 2012). These studies confirmed that leaf hydraulic signals played an essential role in controlling g m in response to drought. However, the effects of chemical ABA signal on g m are not consistent. Vrabl et al. (2009) did not observe any reduction in g m when applied exogenous ABA in Helianthus annuus plants. In line with this, Flexas et al. (2013) found that g m was highly insensitive to endogenous ABA among ABAinsensitive and ABA-hypersensitive genotypes or to exogenous ABA application in Arabidopsis thaliana. However, several studies yielded contrasting results. For instance, Mizokami et al. (2015) compared the responses of g m to leaf ABA in wild type and ABA-deficient mutant of Nicotiana plumbaginifolia and confirmed that the increase in leaf ABA concentration was crucial for the decrease in g m under drought conditions. Still, g m reduced effectively in response to ABA in a short term in three of the four species in Sorrentino et al. (2016). Recently, Mizokami et al. (2018) examined the responses of g m to high CO 2 and ABA application and revealed that g m was able to respond to high ABA levels, which was intrinsically different from the response to the elevated CO 2 . These contrasting results possibly due to species differences or the experimental approaches utilized to modify ABA, e.g., the exogenous ABA concentration or the applying period. In brief, it has been largely demonstrated that hydraulic signals play an important role in regulating g m , while the role of ABA on g m is still not unequivocal. Therefore, a deep understanding about the mechanisms of g m response to endogenous ABA under progressive soil drying conditions awaits further investigation.
Leaf intrinsic water use efficiency (WUE i ), expressed as the ratio of net photosynthetic rate (A n ) to g s at leaf level, can explain instantaneous responses to environmental factors (Flexas et al., 2016;Qiu et al., 2019). Improving WUE i need increase A n and decrease g s simultaneously. Using A n /g s to explain the changes of WUE i would be too coarse due to the decrease in g s inevitably affect CO 2 uptake and thereby limit A n . g m determines the CO 2 concentration at the carboxylation site in the chloroplast, increasing g m would increase A n without increasing water loss. Therefore, g m might play a role in improving WUE i . Despite all of the negative impacts of drought stress on leaf gas exchange, many studies reported that drought was beneficial to improve WUE i (Liu et al., 2005;Xue et al., 2016). However, the reasons of this improvement of WUE i have not been elucidated clearly. Evidences have suggested that g m /g s played a key role on increasing WUE i in response to water limitation (Flexas et al., 2016;Han et al., 2016). Revealing the exact responses of g m /g s or WUE i to stressed signals especially ABA under progressive soil drought would be of great interest in the selection of varieties with high yield in breeding and strong adaptability under varied environmental conditions.
In this study, relationships between g s , g m , and Ψ leaf or ABA were examined in tomato seedlings under progressive soil drying conditions. The objectives of this study were (i) to evaluate the effects of limiting factors of g s and g m on A n in tomato plants during progressive soil drying, (ii) to investigate the responses of g s and g m to drought signals (Ψ leaf and ABA) under increasing drought stress, and (iii) to reveal the effects of g s /g m on WUE i in tomato seedlings during the progressive soil drying.
Water treatments (including well-watered and progressive drought-stressed treatments) were conducted at the 27 day after transplanting (DAT). For the well-watered treatment, RSWC was maintained within the range of 70-82% θ FC throughout the experiment. Plants remained well-watered acted as a control group (CK). For the drought-stressed treatment (withholding water), RSWC decreased from 82.90% θ FC to 37.27% θ FC from 27 to 33 DAT. On each day of the drying period (28-33 DAT), the relevant experimental indexes were measured and collected for the two treatments.

Leaf Gas Exchange and Chlorophyll Fluorescence Measurements
Leaf gas exchange and chlorophyll fluorescence were measured simultaneously using an open gas exchange system Li-Cor 6400 photosynthesis system (Li-Cor Inc., Lincoln, NE, USA) equipped with an integrated leaf fluorometer chamber (Li-Cor 6400-40) from 9:00 to 14:00 h. All measurements were recorded on the same fully expanded leaves (the 6th or 7th leaves from the base of the plant) during 28-33 DAT, using two or six replicate plants for CK and water stressed treatment, respectively. During the measurements, the PPFD was kept at 1500 µmol m −2 s −1 , the sample CO 2 concentration was maintained at 400 µmol mol −1 with a CO 2 cylinder. Relative humidity was kept at 55%. Leaf gas exchange, chlorophyll fluorescence and leaf temperature were recorded when A n was stabilized on these conditions (usually 20 min after clamping the leaf). After that, A-C i response curves were conducted. During the measurements, the PPFD was kept as constant of 1500 µmol m −2 s −1 , sample CO 2 concentration was adjusted in a series of: 400, 300, 200, 150, 100, 50, 400, 400, 600, 800, 1000, 1200, 1400, 1600 µmol mol −1 .
The intrinsic water use efficiency (WUE i , µmol CO 2 mol −1 H 2 O) was calculated as the ratio of net photosynthetic rate divided by stomatal conductance: where A n is net photosynthesis rate (µmol CO 2 m −2 s −1 ), g s is stomatal conductance (mol H 2 O m −2 s −1 ). The actual photochemical efficiency of photosystem II (Φ PSII ) was determined by measuring steady-state fluorescence (F s ) and maximum fluorescence (F ′ m ) during a light-saturating pulse of ca. 8000 mmol m −2 s −1 : The electron transport rate (J f ) was then calculated as: where PPFD was maintained at 1500 µmol m −2 s −1 on both the well-watered and water-stressed leaves. α represents the leaf absorptance and β reflects the partitioning of absorbed quanta between photosystems II and I. α and β were assumed to be 0.84 and 0.5 in the study, respectively (Laisk and Loreto, 1996;Flexas et al., 2002).

Estimation of g m by Gas Exchange and Chlorophyll Fluorescence Method
where C i represents intercellular CO 2 concentration (µmol CO 2 mol −1 ), R d represents the light mitochondrial respiration (µmol CO 2 mol −1 ), which was calculated as 1/2 of the dark respiration Xiong et al. (2018), Γ * is the chloroplast CO 2 compensation point (µmol CO 2 mol −1 ), a leaf temperaturedependent parameter, and estimated as: where c is the scaling constant (dimensionless), H a is the energies of activation (KJ mol −1 ), and R is the molar gas constant (8.314 J K −1 mol −1 ). At the leaf temperature of 25 • C, c and H a in S. lycopersicum were equal to 12.7 and 23.2 (KJ mol −1 ), respectively (Hermida-Carrera et al., 2016). T k is the leaf absolute assay temperature (K), which was recorded by the LI-6400 system and corrected to Kelvin temperature.
Given the potential errors in estimation made by the variable J method, sensitivity analyses were conducted to determine the effect of ±20% error of R d , Γ * , J f , and C i on calculation of g m .

Photosynthetic Limitation Analysis
The relative photosynthesis limitations of A n resulting from g s (l s ), g m (l m ), and biochemical photosynthetic capacity (l b ) (l s + l m + l b = 1) was determined using the method of Grassi and Magnani (2005), as follows: where g sc is the stomatal conductance to CO 2 (mol CO 2 m −2 s −1 ), g sc = g s /1.6, g t is the total conductance to CO 2 from the leaf surface CO 2 to chloroplast (1/g t = 1/g sc + 1/g m ). According to the Farquhar model (Farquhar, 1980), ∂A/∂C c can be calculated as follows: where K c and K o are the Rubisco Michaelis-Menten constants for CO 2 and O 2 , both of them were temperature-dependent and calculated as Equation (5). Specific values of these parameters in Equation (5) were obtained from Sharkey et al. (2007). O is the atmospheric O 2 concentration (210 mmol mol −1 ). V cmax is the maximum carboxylation capacity (µmol m −2 s −1 ). V cmax was calculated from the A/C i curve fitting method (Long and Bernacchi, 2003).

Soil and Leaf Water Potential Measurement and Shoot Sap Collection
Leaf water potential (Ψ leaf ) was measured on the same leaves as the measurement of gas exchange. Soil samples at the 10-12 cm under soil surface were collected to measure soil water potential (Ψ soil ). Both Ψ leaf and Ψ soil were measured by the WP4C Dewpoint Potentiometer (Meter Group Inc., Pullman, WA, USA) with two or six repetitions for CK and water stressed treatment. Meanwhile, the shoot part (including stem and leaf) was put into the Model 3115 pressure chamber (Plant Moisture Equipment, Santa Barbara, CA, USA). Pressure was increased gradually until sap outflowed at the cut surface. After discarding the first 1-2 drops, nearly 2 ml of sap was collected into centrifuge tube frozen in liquid nitrogen and then stored at −80 • C for ABA analysis.

ABA Determination
The concentration of ABA was determined as previously described by Li et al. (2020). Briefly, sap ABA concentration was measured with a high-performance liquid chromatographytandem mass spectrometry (Agilent Technologies Inc., Santa FIGURE 1 | Dynamics of RSWC and Ψ soil in the well-watered (CK) and drought-stressed tomato seedlings during 27-33 DAT. Mean values and SD were presented (n = 6). ns indicated no significant difference and ** indicated significant difference at P < 0.01 level between drought and well-watered treatment.
Clara, CA, USA), quantitated as the methods of isotope internal standard.

Statistical Analysis
All statistical analyses were performed using SPSS 16.0 (IBM Corp., Armonk, NY, United States). The significance of differences between mean values was assessed by One-way analysis of variance (ANOVA) according to Dennett's test at P < 0.05 level. Regressions were fitted by linear or non-linear models, and the model with higher regression coefficient (r 2 ) was selected. Regression lines was shown when P < 0.05. All graphics and regressions were performed in Origin-Pro 2017 (Origin Lab, Northampton, MA, USA).

Dynamic of Soil Water Status
Relative soil water content (RSWC) and Ψ soil of the well-watered pots were maintained at an average of 75.13% and −0.43 MPa, indicating no water stress occurred during the experiment. By withholding irrigation from 27 to 33 DAT during the progressive drying process, RSWC in the drought treatment decreased gradually from 82.90 to 37.27% and Ψ soil decreased by 1.04 MPa correspondingly. Interestingly, significant reduction of both RSWC and Ψ soil occurred simultaneously at 29 DAT (Figure 1).

Effects of Drought on Ψ leaf and ABA
In the well-watered treatment, Ψ leaf maintained at an average of −0.72 MPa from 27 to 33 DAT. Along with decreasing Ψ soil in the pots, Ψ leaf of the drought-stressed tomato seedlings kept almost constant until Ψ soil reached to −0.71 MPa (Figure 2A). However, ABA did not statistically increase within the range of Ψ soil from −0.42 to −0.83 MPa, indicating that compared to Ψ leaf , chemical FIGURE 2 | Leaf water potential (n = 6) (A) and shoot sap ABA concentration (n = 3) (B) in response to progressive soil water potential decrease. Colorful labels indicated significant difference at P < 0.001 level between well-watered and drought treatment.
signal, ABA showed a delayed response in face to mild soil drying.
As soil further drying, ABA increased exponentially with Ψ soil decreasing from −1.01 to −1.44 MPa (Figure 2B). It should be noteworthy that ABA in the drought-stressed plants increased up to an average of 97.86 ng ml −1 at the end of experiment, resulting in an around 300 times higher than the well-watered treatment.

Quantitative Analysis of Photosynthetic Limitation in Response to Soil Drying
The relative contributions of all limiting factors (l s , l m , l b ) to photosynthetic capacity can be divided into three stages (Figure 3). Firstly, l b contributed to around an average of 51.46% limitation when Ψ soil was >-0.71 MPa, suggesting that photosynthetic biochemistry was the main factor under no water stressed condition. Secondly, with Ψ soil decreasing from −0.83 to −1.15 MPa, l b declined, whereas both l s and l m increased, but l s was higher than l m , which contributed solely to an almost 50.30% reduction in A n , indicating that g s was the main limiting factor to photosynthetic capacity under mild and moderate drought. Thirdly, with Ψ soil decreasing to −1.44 MPa, l m contributed to 41.99% reduction in photosynthesis, followed by l s (36.93%) and l b (21.08%), showing that g m was the most important limiting factor to photosynthetic capacity under the severe drought condition.
Ψ leaf and ABA in the Regulation of g s , g m , g t , and A n As compared to g s in CK, g s in the water-stressed tomato seedlings increased firstly with Ψ leaf decreasing from −0.72 to −0.95 MPa and then decreased with Ψ leaf decreasing from −1.05 to −1.63 MPa ( Figure 4A). However, g m kept unchanged within the range of Ψ leaf from −0.72 to −1.05 MPa and decreased significantly when Ψ leaf was <-1.28 MPa (Figure 4C). The output of ANOVA showed that drought had significant effect on the slopes of the regression lines between g s and g m to Ψ leaf (Supplementary Figure 1). In addition, under mild and moderate drought, the ratio of g s reduction was higher than g m during 30-32 DAT (Supplementary Figure 2). These results indicated that g s was more sensitive to mild and moderate drought stress than g m . In summary, there was a significant FIGURE 4 | Effects of leaf water potential, ABA concentration on stomatal conductance (g s ) (A,B), mesophyll conductance (g m ) (C,D), total conductance (g t ) (E,F), and net photosynthesis (A n ) (G,H). Colorful labels indicated significant difference between the well-watered (CK) and drought treatments at P < 0.01 level.
Frontiers in Plant Science | www.frontiersin.org 6 April 2021 | Volume 12 | Article 653186 TABLE 1 | Correlation matrix between studied parameters including intrinsic water use efficiency (WUE i ), net photosynthesis (A n ), mesophyll conductance (g m ), stomatal conductance (g s ) and the ratio (g m /g s ), abscisic acid (ABA), and leaf water potential (Ψ leaf ). * and **mean statistically significant relationship according to the Pearson correlation analysis at P < 0.05 and P < 0.01.

FIGURE 5 |
The relationship between the stomatal conductance to H 2 O (g s , mol H 2 O m −2 s −1 ) and mesophyll conductance to CO 2 (g m , mol CO 2 m −2 s −1 ) in the leaves under progressive drought. Data were fitted by a linear regression with r 2 = 0.59 at P < 0.01 level.
We also investigated the relationship between ABA and g s or g m (Figures 4B,D). g s changed with no significant increasing ABA during 28-29 DAT. As soil further dried, g s continued decreasing and g m started to decrease with significant increase in ABA (Figures 4B,D). g m was closely related to g s during drying (r 2 =0.59, P < 0.01) (Figure 5). Notably, g m and ABA changed concurrently at the threshold of Ψ soil = −1.01 MPa (Figures 2B,  4D). In summary, ABA was negatively related to g m (r = −0.64, P < 0.01) and g s (r = −0.55, P < 0.01) ( Table 1). These results indicated that the decline of g s was regulated by Ψ leaf in the early stage of drought, whereas under moderate or severe drought, g s and g m were controlled by both Ψ leaf and ABA. Drought significantly affected A n and g t during 30-33 DAT. When Ψ leaf decreased to −1.05 MPa or ABA increased to 2.04 ng ml −1 , A n and g t declined by 40.18 and 45.13%, respectively ( Figures 4E-H). As soil further dried, i.e., Ψ leaf decreasing from −1.28 to −1.63 MPa, A n and g t reduced by 62.84-88.94% and 74.33-92.92% in the drought-stressed plants as compared with the well-watered plants, respectively (Figures 4E,G).

g m /g s and WUE i in Response to Ψ leaf and ABA Under Progressive Soil Drying
The dynamics of g m /g s in response to Ψ leaf and ABA during progressive soil drying were presented in Figure 6. Higher g m /g s was observed as Ψ leaf decreased from −1.05 to −1.33 MPa or as ABA increased from 2.04 to 31.23 ng ml −1 (Figures 6A,B), indicating that g s declined more than g m under mild or moderate drought. However, no significant difference of g m /g s between CK and the intense water stress with Ψ leaf = −1.63 MPa was found. WUE i in response to these signals changed in the same way as g m /g s (Figures 6C,D), it increased firstly and then decreased. In addition, WUE i was positively related to g m /g s with a logarithmic relationship (r 2 = 0.62, P < 0.001) during the progressive soil drying (Figure 6E), indicating that WUE i was strongly correlated to g m /g s .

Sensitivity Analyses of Parameters in the Estimation g m
10% variation of R d and J f did not affect g m significantly, whereas Γ * has a significantly effect on g m in well-watered plants ( Table 2). As compared to g m in the well-watered plants, g m in the drought treatment was unaffected by the 20% underestimation of J f , showing that g m in the drought treatment was less sensitive to J f than in the well-watered plants. Variation of C i resulted in an overestimation of g m in well-watered plants, whereas g m in drought treatment was unaffected by overestimation of C i . These results indicated that overestimation of C i had a slighter effect on calculation of g m than underestimation in the current study.

Effects of g s and g m on A n Under Soil Drought
Efficient CO 2 fixation is important for plant acclimation to environmental factors. In the present study, the total diffusion conductance of CO 2 (g t ) and A n declined synchronously under drought (Figures 4E-H). The total diffusion conductance of CO 2 mainly includes g s and g m (Grassi and Magnani, 2005). Many authors have reported that CO 2 diffusion from sub-stomatal cavities to chloroplasts is a significant factor determining photosynthetic capacity in C 3 plants such as tomato (Han et al., 2016;Du et al., 2019;Xu et al., 2019). Our analysis showed that l s and l m increased as soil drying proceeded and contributed to an almost 68-78% reduction in A n when Ψ soil was <-0.83 MPa (Figure 3). Our results, as well as those of previous studies (Niinemets et al., 2009;Wang et al., 2018), confirmed the significance of g s and g m on assimilation rate under various drought conditions. It should be acknowledged that, many authors have highlighted the effects of leaf anatomical traits on g m , such as cell thickness, cell packing and area of chloroplasts FIGURE 6 | Correlation between ratio of mesophyll conductance to stomatal conductance (g m /g s ) or intrinsic water use efficiency (WUE i ) and leaf water potential (A,C) or shoot sap ABA (B,D) under progressive drought. The relationship between g m /g s and WUE i was presented with a non-linear regression at P < 0.001 level (E). Colorful labels indicated significant difference between the well-watered (CK) and drought treatments at P < 0.01 level.
exposed to the intercellular air spaces (S c /S) across many species including tomato (Tomas et al., 2013;Muir et al., 2014). This effect was a result of plants acclimation to the long-term stressed environmental factors lasting for weeks. However, rapid response of g m to stress could occur within minutes response to elevating CO 2 (Mizokami et al., 2018) or hours response to application of ABA (Sorrentino et al., 2016). Perhaps this meant that different mechanisms of g m determination existed under short and long term drought conditions. Therefore, to minimize the effects of leaf anatomy on g m , we focused on the responses of g m to drought stress and the involvement of ABA in a short water stress cycle.

Response of g s to Ψ leaf and ABA Under Soil Drought
We found that g s generally decreased as Ψ leaf decreased (Figures 2A, 4A), suggesting that Ψ leaf might induce stomatal closure at the early stage of drought. The mechanisms of this hydraulic regulation remain unclear, but the reduction in Ψ leaf has been tightly associated with decreasing leaf hydraulic signals (leaf turgor or K leaf ) in understanding the closure of stomata (Ripullone et al., 2007;Wang et al., 2018). On the one hand, evidences have suggested that decline of leaf turgor could explain the decrease in g s within no change of ABA (Rodriguez-Dominguez et al., 2016;Huber et al., 2019), possibly due to the decrease of elastic modulus and the activity of anion channel in guard cell during leaf dehydration (Ache et al., 2010;Saito and Terashima, 2010). On the other hand, progressive drop of plant water potential might decrease xylem pressure and increase the likelihood of embolism and hydraulic failure (Martorell et al., 2014;Tombesi et al., 2015). Responding to the future unpredictable soil water availability, stomata closed to prevent water loss and avoid xylem cavitation. Here, the increase of shoot sap ABA concentration was statistically insignificant, which implied that stomatal closure was not initiated by ABA with Ψ soil not approaching to −1.01 MPa (Figures 2B, 4B). Indeed, the delayed increase in leaf ABA in the present study was consistent with the recent findings that leaf ABA did not increase until after stomata closed, which was different from the actions of leaf turgor subjected to drought stress (Huber et al., 2019). However, as soil drought proceeded, g s continued decreasing with significant changes in both ABA and Ψ leaf , suggesting that Ψ leaf was not solely controlling g s , but chemical ABA was also involved in the reduction of g s . A similar variation between ABA and g s was also reported by Tombesi et al. (2015), who indicated that ABA played a crucial role in maintaining stomatal closure under long and severe drought. However, it should be noteworthy that our data need to be further interpreted, as shoot sap ABA 2 | Sensitivity analyses of the effects of ±20% error of light mitochondrial respiration (R d ), chloroplast CO 2 compensation point (Γ * ), electron transport rate (J f ), and intercellular CO 2 concentration (C i ) on calculation of g m in well-watered and severe drought tomato at Ψ soil = −1.44 MPa as compared with the original value of g m . 0.146 ± 0.005 ** 0.013 ± 0.002 ns C i -20% 0.433 ± 0.025 ** 0.020 ± 0.003 * Γ *-10% 0.168 ± 0.009 ** 0.013 ± 0.002 ns C i -10% 0.270 ± 0.011 ** 0.017 ± 0.003 ns Γ * +10% 0.238 ± 0.015 ** 0.014 ± 0.002 ns C i +10% 0.155 ± 0.005 ** 0.013 ± 0.002 ns Γ * +20% 0.301 ± 0.011 ** 0.014 ± 0.002 ns C i +20% 0.127 ± 0.004 ** 0.011 ± 0.002 ns Data were mean ± SD (n = 6). ns indicated no significant difference and ** indicated significant difference at P < 0.01 level between drought and well-watered treatment.
was collected in the pressurized stem and leaf tissues instead of in localized guard cells.

Response of g m to Ψ leaf and ABA Under Soil Drought
The variable J method (Harley et al., 1992), as the most commonly and easily accessible approach, was used to determine g m during the dry-down stage. To obtain precise calculation of g m , the highest possible accuracy of gas exchange and chlorophyll fluorescence were required during the process of measurement. As reported previously, the decrease in g m under drought was likely to associate with an overestimation of C i due to stomatal closure (Pons et al., 2009). However, the sensitivity analyses showed that an overestimation of C i did not induce g m decline in drought-stressed plants ( Table 2). Thus, overestimation of C i was unlikely to have a significant effect on g m in this study, might due to the influence of other environmental variations was ruled out under controlled environment. Therefore, it is reasonable to conclude that the reduction in g m during drought was mostly attributed to the decline of g m per se rather than the overestimation of C i . Compared to the response of g s , g m in the drought-stressed seedlings remained almost constant with Ψ leaf not decrease to −1.28 MPa (Figures 4A,C), indicating that g m was less sensitive to the decrease in Ψ leaf than g s at the beginning of soil drought. This result was in agreement with an earlier study conducted by Théroux-Rancourt et al. (2014), who found that g m only responded to more negative Ψ leaf or more severe soil drought, e.g., Ψ soil < −1.01 MPa in the present study. Hydraulic compartmentalization of the mesophyll cell from the transpiration stream may account for this delayed response of g m to Ψ leaf (Zwieniecki et al., 2007;Théroux-Rancourt et al., 2014). This delayed response of g m under the mild soil drought might be beneficial for mesophyll cells to be buffered against little variation in leaf water status and allow plants to maintain a greater A n ( Figure 4G).
However, as soil drought proceeded, g m declined as Ψ leaf continued decreasing (Figures 4C,D). Based on literature surveys, the causes of this decrease in g m may be influenced by three main factors: mesophyll structure, membrane permeability, and biochemical enzymes activity (Flexas et al., 2008;Evans et al., 2009;Sorrentino et al., 2016). Mesophyll structural properties may not be involved in this rapid reduction of g m under the short-term drought. Instead, it is well-established that the K leafinduced reduction in g m was associated with the decrease in mesophyll density or membrane permeability under drought conditions (Aasamaa et al., 2005;Xiong et al., 2018). Water moves through leaf mesophyll tissues via apoplastic, symplastic and vapor phase pathways, which shared a part of pathways of CO 2 diffusion (Xiong and Nadal, 2020). The decline in hydraulic conductance under drought usually leads to reductions in water supply to the leaves, therefore affecting mesophyll cells water relations and functions. Although the effect of K leaf on g m was not investigated in this study, we observed a strong and positive relationship between Ψ leaf and g m (r 2 = 0.77, P < 0.01) (Figure 4C), because K leaf was strongly influenced by Ψ leaf under drought stress (Wang et al., 2018). Therefore, the decline in Ψ leaf might contribute to this decrease in g m , as CO 2 diffusion and liquid water shared partly common pathways within leaves .
Most notably, rapid reduction of g m occurred following with increase of ABA when Ψ soil was below −1.01 MPa in the current study. Fast fluctuations in g m have also been recorded in response to ABA application (Sorrentino et al., 2016;Mizokami et al., 2018). The concurrent responses between g m and ABA with Ψ soil decreasing from −1.01 to −1.44 MPa was not a mere coincidence. This might suggest that Ψ leaf was not the only factor influencing g s and g m under drought, other signals (ABA) could be involved in this reduction. Though mechanisms for the effect of ABA on g m remain unclear, the results from both Sorrentino et al. (2016) and the current studies indicated that the reduction in g m was most likely regulated by biochemical components due to the rapid reduction of g m to ABA (Flexas et al., 2008;Kaldenhoff et al., 2008;Xiong et al., 2018). Evidences have indicated two candidates are likely to play this biochemical role: carbonic anhydrase and aquaporins. CO 2 molecules passing from sub-stomatal cavities to chloroplasts diffuse through the gas phase among intercellular air spaces and the liquid phase from the cell wall to stroma. Carbonic anhydrase (CA) plays a key role on the conversion of gaseous CO 2 to aqueous carbonic acid (H 2 CO 3 ) (Flexas et al., 2008). Higher ABA accumulation was likely to change the extracellular pH and decrease the activity of H + -ATP-ase, an important ion transporter in plant cell plasma membrane, thus affect the CA activity (Hayat et al., 2001;Sukhov et al., 2017). Aquaporins (AQPs) are pore-forming integral membrane proteins that transport of water, CO 2 and other small neutral molecules across the plasma membrane (Flexas et al., 2006;Kaldenhoff, 2012). A higher abundance of AQPs increased the cellular CO 2 uptake rates several folds. Expressions of plant AQPs could be influenced by drought stress and ABA (Kapilan et al., 2018). Additionally, an indirect role of ABA on decreasing K leaf might also be involved in regulating g m , due to the ability of ABA on inactivation bundle sheath aquaporins such as the plasma membrane intrinsic proteins (PIPs) (Shatil-Cohen et al., 2011;Pantin et al., 2013). Based on these, we considered that the reduction in g m was not attributed solely to hydraulic regulation, ABA seemed to maintain the decrease in g m under moderate or severe soil drought, e.g., Ψ soil < −1.01 MPa in the present study. The regulation of g m is complex, and regulated by many factors, including hydraulic or chemical signaling and mesophyll structure. It is still unclear the mechanism of g m response to ABA under stress, further analysis of the expressions of carbonic anhydrase and cooporin protein in membrane may elucidate the biochemical mechanisms underlying this response. Notably, g s and g m decreased as ABA significantly increased (Figures 4B,D). Pooling all the data, a strong and positive relationship between both variables was observed in Table 1. In addition, 59% of the variation in g m can be explained by g s (Figure 5). Coupled changes between g s and g m was also found in response to drought (Perez-Martin et al., 2009;Han et al., 2016;Olsovska et al., 2016) or ABA application (Mizokami et al., 2018). Therefore, it seems that drought regulated g m in order to match the variation of g s , thereby optimization balance between CO 2 uptake and water loss. However, the role of g s on regulating g m response to ABA is still debated by many scientists (Sorrentino et al., 2016;Mizokami et al., 2018), further detail investigations are needed to address this issue.
Variability of WUE i Under Drought Depends on g m /g s In this study, g m /g s and WUE i increased concurrently with Ψ soil in the range of −0.83 to −1.15 MPa with a strong correlation ( Figure 6E). Our results showed that WUE i was closely related to g m /g s compared to the correlation between WUE i and g m or g s ( Table 1). This result was consistent with Han et al. (2016) who also found WUE i and g m /g s were closely correlated compared to the correlation between WUE i and g s or g m . These suggested that variations in WUE i were much more sensitive to changes of g m /g s . Stomata controls the water loss and mesophyll determines the photosynthesis, thus it would be better that using g m /g s instead of A n /g s explained the variations of WUE i . Interestingly, this improvement of WUE i were coupled with increase in ABA. This might due to g s reduced more in response to ABA than g m under moderate drought. Though the mechanisms of ABA improving WUE i remain largely unknown, it is likely to be one of the most promising strategies to improve WUE i by means of decoding of the ABA signaling pathway or manipulating the expression of ABA-related genes on stomatal conductance or CA activity (Flexas et al., 2016;Cardoso et al., 2020). Nonetheless, such improvement of WUE i controlled by ABA could only be beneficial for maintaining water status under short-term drought during Ψ soil reduction from −0.83 to −1.15 MPa, not for long and serious drought ( Figure 6D). This was beacuse the increase in WUE i at leaf scale may not always mean an improvement of WUE at the whole plant scale under serious soil drought, as the closure of stomata restricts CO 2 uptake and hence diminish plant productivity (Xue et al., 2016).

CONCLUSION
The limitation of g s and g m increased along with progressive soil drying and diffusive conductance to CO 2 from ambient air to chloroplasts was the crucial constraints to photosynthesis under drought conditions. The decrease in Ψ leaf triggered stomata closure at the onset of drought. As soil drying proceeded, g s and g m declined synchronously. Both hydraulic and ABA signals were involved in this consistent decrease under moderate and severe drought. WUE i improved as g m /g s increased under mild and moderated drought due to a larger reduction of g s to ABA than g m . Manipulation of ABA levels might be a promising approach to improve plant water use efficiency for breeding project. For future research, examining the influence of stomatal closure on g m response to ABA will give further detailed insight on working of g m to ABA.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
AD and YG planned and designed the experiments. SL and JL performed the experiments and analyzed the data. SL wrote the draft manuscript. AD, YG, HL, and RQ revised the manuscript. All authors read and approved the final manuscript.