Protein kinase SnRK2 . 4 is a key regulator of aquaporins and root hydraulics in Arabidopsis

Soil water uptake by roots is a key component of plant water homeostasis contributing to plant growth and survival under ever-changing environmental conditions. The water transport capacity of roots (root hydraulic conductivity; L p r ) is mostly contributed by ﬁnely regulated Plasma membrane Intrinsic Protein (PIP) aquaporins. In this study, we used natural variation of Arabidopsis for the identiﬁcation of quantitative trait loci (QTLs) contributing to L p r . Using recombinant lines from a biparental cross (Cvi-0 x Col-0), we show that the gene encoding class 2 Sucrose-Non-Fermenting Protein kinase 2.4 (SnRK2.4) in Col-0 contributes to > 30% of L p r by enhancing aquaporin-dependent water transport. At variance with the inactive and possibly unstable Cvi-0 SnRK2.4 form, the Col-0 form interacts with and phosphorylates the prototypal PIP2;1 aquaporin at Ser121 and stimulates its water transport activity upon coexpression in Xenopus oocytes and yeast cells. Activation of PIP2;1 by Col-0 SnRK2.4 in yeast also requires its protein kinase activity and can be counteracted by clade A Protein Phosphatases 2C. SnRK2.4 shows all hallmarks to be part of core abscisic acid (ABA) signaling modules. Yet, long-term ( > 3 h) inhibition of L p r by ABA possibly involves a SnRK2.4-independent inhibition of PIP2;1. SnRK2.4 also promotes stomatal aperture and ABA-induced inhibition of primary root growth. The study identiﬁes a key component of L p r and sheds new light on the functional overlap and speciﬁcity of SnRK2.4 with respect to other ABA-dependent or independent SnRK2s.


INTRODUCTION
Plants have evolved numerous physio-morphological traits to maintain their water status and face adverse environmental conditions, including water deficit in the soil (drought) or the atmosphere (evaporative demand) (Zhu, 2016).These traits, which allow a continuous adjustment of water delivery throughout the plant for growth and transpiration, include stomatal opening and closing, cell osmotic adjustment, regulation of tissue water permeability (hydraulics) and, on the longer term, alteration of root system growth and architecture, and leaf area.
A critical role of roots in this context is uptake of soil water (Maurel & Nacry, 2020;Scharwies & Dinneny, 2019).This first requires radial transport of water across concentric root cell layers prior to loading into xylem vessels and axial transport to the plant aerial parts (Boursiac et al., 2022).Radial water transport occurs along cell wall continuities (apoplastic path) or from cell to cell, through plasmodesmadata (symplastic path), or across membranes (transcellular path).The latter path is dominated by the function and regulation of aquaporins of the Plasma membrane Intrinsic Protein (PIP) subclass (Chaumont & Tyerman, 2014;Maurel et al., 2015).Fine-tuning of their opening and closing and regulated expression at the plasma membrane provides multiple potentialities for rapid and reversible adjustment of root hydraulic conductivity (Lp r ) in response to numerous Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 1 environmental stimuli (e.g.water deficit, flooding, nutrient availability) or endogenous signals related to abiotic stress or plant defense (Chaumont & Tyerman, 2014;Maurel et al., 2015).For instance, the plant hormone abscisic acid (ABA), which accumulates during water deficit, salinity, heat, and cold stresses acts on aquaporin expression in numerous plant species and causes changes in whole root and root cell hydraulic conductivity (Fan et al., 2015;Hose et al., 2000;Mahdieh & Mostajeran, 2009;Olaetxea et al., 2015).Effects of ABA were stimulatory in most studies although ABAdependent inhibitions of root hydraulics could be observed, depending on plant species and hormone concentration (discussed in Mahdieh & Mostajeran, 2009).In Arabidopsis thaliana guard cells, phosphorylation-dependent activation of AtPIP2;1 by ABA mediates a dual hydraulic and signaling function of this hormone, thereby favoring stomata closure and water conservation (Grondin et al., 2015;Rodrigues et al., 2017).
ABA signaling is mediated by core functional modules comprising ABA-binding PYR/PYL/RCAR proteins acting in concert with clade A Protein Phosphatases 2C (PP2C) (Cutler et al., 2010;Raghavendra et al., 2010).In association with ABA, this complex releases the PP2C-dependent deactivation of class 2 Sucrose-Non-Fermenting Protein kinases (SnRK2s), including SnRK2.2,SnRK2.3 and SnRK2.6 (also known as OST1), which in turn activate various targets such as AREB transcription factors (Yoshida et al., 2010), membrane transport proteins (Brandt et al., 2012;Grondin et al., 2015) or miRNA processing components (Yan et al., 2017).While an ABA signaling pathway was initially reconstituted in vitro (Fujii et al., 2009), recent co-expression studies in yeast allowed to assemble a variety of core ABA signaling modules acting in concert with various SnRK2s to activate in an ABA-dependent or independent manner AREB transcription factors, which in turn enhance expression of a luciferase reporter (Ruschhaupt et al., 2019).
The SnRK2 gene family comprises ten members (SnRK2.1 to 10) in Arabidopsis (Kulik et al., 2011), further classified into three groups (I-III).SnRK2 protein kinases are key regulators of plant response to water deficit, as induced by drought or salt stress.Group II (SnRK2.7 and 8) and group III (SnRK2.2, 3 and 6) members are known to be responsive to ABA, showing weak and strong activation, respectively (Boudsocq et al., 2004).However, all SnRK2s can be activated by osmotic stress and, in addition, group I SnRK2s (SnRK2.1, 4 and 10) mediate responses to oxidative treatments (Boudsocq et al., 2004;McLoughlin et al., 2012;Kulik et al., 2012).Recent studies showed that group I SnRK2s interact with VARICOSE, an mRNA decapping activator, and VARICOSE RELATED to regulate mRNA decay of numerous osmotic and salt stress-responsive genes thereby controlling root development and response to salt (Kawa et al., 2020;Soma et al., 2017).SnRK2s are themselves phosphorylated by osmotic-stress-activated Raf-like protein kinases which act as SnRK2 activators (Katsuta et al., 2020;Lin et al., 2020;Soma et al., 2020;Takahashi et al., 2020).
Natural genetic variation offers a great opportunity to discover adaptive mechanisms that fine-tune plant response to their environment and to identify genetic variants that can be used in plant improvement.As for numerous other traits, Arabidopsis exhibits marked withinspecies variation for drought sensitivity, phytohormone accumulation and/or signaling, and root hydraulics (Weigel, 2012).In the present work, we pursued studies on natural variation for root water permeability (root hydraulic conductivity, Lp r ) in Arabidopsis (Shahzad et al., 2016;Sutka et al., 2011;Tang et al., 2018).To identify genetic factors relevant to this trait, we explored recombinant inbred lines (RILs) developed by crossing two accessions, Col-0 and Cvi-0.Multiple quantitative trait loci (QTLs) controlling Lp r under standard growth conditions were detected in this population.Alternative genotypes harboring either a Col-0 or Cvi-0 allele at one QTL displayed remarkable (>30%) variation for Lp r .Positional cloning of the corresponding causative gene identified SnRK2.4.The overall work unravels SnRK2.4 as a major control of root water transport through modulation of aquaporin function.

RESULTS
Identification of Lp r QTLs in a Cvi-0 3 Col-0 population.
A recombinant inbred line (RIL) population obtained by crossing the Col-0 and Cvi-0 accessions was grown under standard conditions and phenotyped for root hydraulics (Lp r ; root hydraulic conductance: Lo), plant growth (shoot and root dry weight) and development (bolting time).Plant growth and development traits were mapped to check for possible pleiotropic effects on root hydraulics.As previously reported for a Bur-0 x Col-0 RIL population (Shahzad et al., 2016), the Cvi-0 x Col-0 population showed transgressive segregation for all traits (Figure S1).Estimated heritability (h 2 ) varied from 0.24 to 0.39 (Table S1).Linkage mapping using Lp r data identified three QTLs (Figure 1A), on the top of chromosome (Chr) 1 (Lprt18), bottom of Chr 2 (Lprb28), and bottom of Chr 5 (Lprb58).The two major Lp r QTLs, Lprt18 and Lprb58, are positively contributed by Col-0 and Cvi-0 alleles, respectively.They were each confirmed using nearly isogenic lines of Heterogeneous Inbred Family (HIF)-type, fixed for either Col-0 or Cvi-0 allele: HIF046 for Lprt18 and HIF062 for Lprb58 (Figure 1B; Figure S2).HIF046-Col-0 showed a higher Lp r than HIF046-Cvi-0 by 33.2% (Figure 1B), whereas HIF062-Col-0 showed a lower Lp r than HIF062-Cvi-0 by 28.9% (Figure S2).While Lprb58 will be the object of future studies, we focused on Lprt18 in the present work.Upon treatment with sodium azide (NaN 3 ), a pharmacological inhibitor of aquaporins (Tournaire-Roux et al., 2003) (Figure 1B).The overall data show that the Col-0 allele at Lprt18 confers an enhanced Lp r due to a higher aquaporin functionality.
SnRK2.4 underlies the natural variation of Lp r conferred by Lprt18.
HIF046 was heterozygous for a region of more than 5 Mb (Figure 2A).To reduce the QTL interval, fine mapping was performed by selecting recombinant HIFs (rHIFs) from a selfed heterozygous individual of HIF046 and by extensively geno-and pheno-typing the rHIFs in a progeny testing procedure.In total, 12 rHIFs helped to restrain the QTL to a ~50 kb region harboring 14 genes (Figures 2A,B).DNA sequence comparison of the restrained QTL interval between Col-0 and Cvi-0 (The 1001 Genomes Consortium, 2016) revealed the existence of numerous polymorphisms between the two accessions (Figure 2B).However, the only polymorphism that causes a change in protein sequence was present in the coding DNA sequence (CDS) of At1g10940 (SnRK2.4).A single nucleotide deletion in the CDS of Cvi-0 SnRK2.4 leads, with respect to Col-0, to a frameshift and insertion of a premature stop codon that results in partial truncation of the protein kinase domain (Figure 2B; Figure S3).This polymorphism was also confirmed in the alternative genotypes of HIF046 (Figure S3).
Based on this dramatic change in Cvi-0, we considered SnRK2.4 as a potential candidate gene for Lprt18.2C); this phenotype is primarily due to a reduction in aquaporin activity, as revealed by a NaN 3 treatment.
To validate that SnRK2.4 underlies Lprt18, we performed two distinct complementation tests.For quantitative complementation, advanced rHIFs segregating for Col-0 or Cvi-0 alleles within a region of about 50 kb around SnRK2.4 were crossed to either snrk2.4-1or its wild-type background (Col-0), and the resulting 4 hybrid allelic combinations were phenotyped for Lp r .A significant interaction between the QTL allele and the SnRK2.4allele was found, with higher Lp r conferred by the Col-0 allele (Figure 2D).Molecular complementation of snrk2.4-1was also performed by introducing genomic regions carrying Col-0 or Cvi-0 SnRK2.4 alleles using Agrobacteriummediated transformation.Expression of the SnRK2.4transgenes was analyzed using qRT-PCR in independent transgenic lines (Figure S4a).Overall, the introduction of Col-0 SnRK2.4 but not Cvi-0 SnRK2.4 was able to complement the reduced Lp r phenotype of snrk2.4-1(Figure 2E; Figure S4b).These complementation tests establish that variation at SnRK2.4 explains the Lp r phenotype segregating with Lprt18.Cvi-0 SnRK2.4 appears as a non-functional allele, probably due to the insertion of a premature stop codon in its CDS (Figure S3).SnRK2.4 does not alter the expression of PIP aquaporins in the plant but can phosphorylate their first cytoplasmic loop.
To address the mode of action of SnRK2.4 on Lp r and aquaporin activity, we investigated various regulatory levels of aquaporin expression.Comparative qRT-PCR and ELISA assays did not reveal any difference between Col-0 and snrk2.4-1or snrk2.4-2plants in the abundance of each of the 13 PIP mRNAs nor in protein accumulation of members of the PIP1 (PIP1;1 to PIP1;4) or PIP2 (PIP2;1 to PIP2;3) subfamilies, respectively (Figures S5a,b).PIP2;1 has previously been used as a prototypal isoform for studying the modes of PIP regulation (Li et al., 2011;Prak et al., 2008;Qing et al., 2016;Wudick et al., 2015).Introduction of a 35S::RFP-PIP2;1 construct in Col-0 and snrk2.4-1yielded similar fluorescence cellular patterns (Figures S5c,d) and similar fluorescence intensity at the cell surface (Figure S5e) between genotypes, indicating that SnRK2.4 does not interfere with the main localization and abundance of RFP-PIP2;1 at the plasma membrane.This lack of effects of SnRK2.4 on PIP mRNA/protein abundance and sub-cellular localization prompted us to investigate the direct impact of SnRK2.4 on aquaporin phosphorylation, and aquaporin gating thereof (Maurel et al., 2015).Recombinant GST-tagged allelic forms of Col-0 and Cvi-0 SnRK2.4 were produced in Escherichia coli (Figure S6) and tested for their ability to phosphorylate PIP2;1 peptides in vitro.We opted for peptides corresponding to the first cytoplasmic loop (loop B) or the C-terminus domain of the protein, since these regions harbor well-described phosphorylation sites at Ser-121 and Ser-280/Ser-283, respectively (Grondin et al., 2015;Prak et al., 2008).Whereas C-terminal phosphorylation sites were not recognized, a strong phosphorylation activity was observed, specifically when Col-0 SnRK2.4 was incubated with the PIP2;1 loop B peptide (Figure 3).These results conform to the substrate specificity of SnRK2.6 (OST1), a SnRK2.4homolog previously shown to phosphorylate PIP2;1 at Ser121 (Grondin et al., 2015).They also support the idea that the Cvi-0 SnRK2.4 allelic form is functionally inactive (Figure 3).
To further establish the activating role of SnRK2.4 and the contribution of its kinase activity, we used a yeast expression system and measured water transport using stopped-flow spectrophotometry.In this system, expression of PIP2;1 increased the osmotic water permeability (P f ) of isolated spheroplasts by about 10-fold with respect to a vector-transformed strain (Figure 5A).Co-expression of Col-0 SnRK2.4 with PIP2;1 further enhanced P f by 40% without altering the overall expression level of PIP2;1 (ELISA assays on whole yeast extracts (in a.u.AE SE, n = 17): PIP2;1: 38.37 AE 1.86; PIP2;1 + SnRK2.4:40.35 AE 1.79).Consistent with oocyte expression experiments, these data show that Col-0 SnRK2.4 activates PIP2;1 water transport activity.Furthermore, we used growth assays in yeast to monitor the capacity of PIP2;1 to transport hydrogen peroxide (H 2 O 2 ).The enhanced sensitivity of yeast cells to 0.2 mM H 2 O 2 observed upon expression of PIP2;1 was used as an easy, semi-quantitative readout of its activity (Figure 5B).In addition, GFP-SnRK2.4fusions indicated that the Col-0 and Cvi-0 SnRK2.4 isoforms showed similar expression levels, mostly in the yeast cell cytoplasm (Figure S7).Yeast cells co-expressing Col-0 SnRK2.4 but not Cvi-0 SnRK2.4 with PIP2;1 displayed relatively higher sensitivity to H 2 O 2 than PIP2;1 expressing cells.This assay confirms that Col-0 SnRK2.4 enhances PIP2;1 activity.Furthermore, the effects of Col-0 SnRK2.4 on stimulating PIP2;1 activity were abolished by a point mutation (D123A) in its catalytic site (Figure 5B).They also were counteracted by co-expression with the ABI1 PP2C but not its null D177A mutant (Figure 5B).Similar to ABI1, all 8 other members of the clade A PP2C family, except AHG1, were able to counteract the enhanced sensitivity to H 2 O 2 of yeast cells coexpressing Col-0 SnRK2.4 with PIP2;1 (Figure S8).The overall data are interpreted to mean that the SnRK2.4 protein kinase activity, which can be inhibited upon coexpression with PP2Cs (Ruschhaupt et al., 2019), is required for PIP2;1 activation in yeast.
We conclude that SnRK2.4 can activate PIP2;1 likely through phosphorylation of loop B. The corresponding site is well conserved in PIPs and plays a general role in gating (T€ ornroth-Horsefield et al., 2006) thereby controlling water  In all experiments, cells were injected, as indicated, with 1 ng YFPc-PIP2;1 cRNA and/or 2.5 ng YFPn-SnRK2.4 (Col-0 or Cvi-0) cRNA.
Ó 2023 The Authors.Effects of exogenous ABA on Lp r and role of PIP2;1.
SnRK2.4 is commonly described as an osmotic and saltstress-induced protein kinase (Kawa et al., 2020;McLoughlin et al., 2012;Soma et al., 2017).Although responsiveness The expression of the catalytically active Col-0 allele and PIP2;1 is marked by the plus sign.Yeast expressing the Cvi-0 allele, ABI1, and functional inactive forms of SnRK2.4 and ABI1 with the mutation D123A and D177A in the catalytic domain, respectively, are indicated.Yeast cultures were grown to comparable cell densities and serial tenfold dilutions were spotted on solidified medium in the presence and absence (control) of 0.2 mM H 2 O 2 .Growth of three independent transformants per combination was assessed after 2 days at 30°C.
Ó 2023 The Authors. of SnRK2.4 to ABA had not been anticipated by earlier activation studies (Boudsocq et al., 2004), recent data based on yeast expression (Ruschhaupt et al., 2019) indicate that SnRK2.4 may be involved in ABA-dependent signaling.These observations prompted us to characterize the responsiveness of Arabidopsis root hydraulics to both salt and ABA.In initial experiments with HIF046, a significant reduction of Lp r was induced after treatment with a high exogenous ABA concentration (1 lM for 3 h) of HIF046-Col-0 (À30.7 AE 3.3%) but not HIF046-Cvi-0 (À12.1 AE 6.2%) plants (Figure S9).By contrast, HIF046-Col-0 and HIF046-Cvi-0 exhibited a similar inhibition of Lp r by salt (150 mM NaCl), by À65.5 AE 2.7% and À 68.7 AE 1.7%, respectively (Figure S9).We therefore investigated in closer detail the effects of ABA on Col-0 and snrk2.4-1roots.Exogenous application of ABA resulted in a progressive inhibition of Lp r in Col-0 which was visible after 3 h (À28%) and reached À51% after 7 h (Figure 6A).By contrast, Lp r of snrk2.4-1plants or snrk2.4-1plants complemented with Cvi-0 SnRK2.4 remained unchanged during the first 3 h of ABA treatment (Figure 6A; Figure S10).After this time point, snrk2.4-1plants showed a tendency to progressive inhibition of Lp r , with absolute values similar to those recorded in Col-0 (Figure 6A).These data suggest that the pool of aquaporins that are activated by SnRK2.4 in roots under resting conditions is targeted during early inhibition of Lp r by exogenous ABA.On a longer-term (>3 h), the hormone targets additional aquaporin pools active in both Col-0 and snrk2.4plants.To determine the PIP isoforms contributing to Lp r and targeted by ABA-induced inhibition, we used a quintuple pip mutant (pip2;1 pip2;2 pip2;4 pip2;6 pip2;7: Quint) and the same genotype but complemented using a PIP2;1 construct (Comp).When grown under control conditions, Quint plants showed, with respect to Col-0, a dramatically reduced Lp r (À65%).In addition, Quint Lp r was insensitive to inhibition after plant exposure to 1 lM ABA for 5 h.Under resting conditions, the Comp genotype showed a partial restauration of Lp r that allowed, with respect to Quint, to monitor PIP2;1 activity.The Lp r of Comp was significantly reduced after 5 h in the presence of ABA, demonstrating that ABA inhibits PIP2;1 activity.
When inspecting intact leaves, we found that, under our growth conditions, snrk2.4 mutants exhibited >30% reduction in stomatal conductance as compared to Col-0 (Figure 7B; Figure S11g).This defect could be compensated by the expression of the Col-0 SnRK2.4 but not the Cvi-0 SnRK2.4 allele.The same pattern of stomatal conductance was observed in both adaxial and abaxial sides of the leaves (Figure 7B; Figure S11g).

DISCUSSION
In the present work, we developed a quantitative genetics approach in a biparental (Cvi-0 x Col-0) segregating population to search for regulators of root hydraulics.In spite of recent advances in the field (Shahzad et al., 2016;Tang et al., 2018), this approach remains challenging due to the technical difficulty of phenotyping Lp r at high throughput and to the strong dependence of this trait on the environmental conditions.Cloning of one of the major QTLs revealed that the SnRK2.4 protein kinase, which occurs in the two parents as two allelic forms with dramatically different functionalities, functions as an activator of aquaporins and contributes to up to 30% of constitutive Lp r in Col-0.
Under resting conditions, aquaporins control 50-80% of Arabidopsis Lp r depending upon natural accessions (Sutka et al., 2011), with for instance 5 PIP isoforms being responsible for 65% of Lp r in Col-0 (Figure 6).Here, we found that SnRK2.4 contributes to >30% of overall Lp r of Col-0, acting on its azide-dependent component.This identifies this protein kinase as a potent regulator of root azidesensitive aquaporins.Phosphorylation of PIPs is one of the key post-translational modifications that correlates well with Arabidopsis Lp r and rosette hydraulic conductivity under numerous physiological stresses (Di Pietro et al., 2013;Prado et al., 2019).In addition, transgenic complementation has shown that multiple phosphorylation sites in the C-terminal domain of PIP2;1 govern its water transport activity in Arabidopsis leaves (Prado et al., 2013(Prado et al., , 2019;;Qing et al., 2016).In stomata, ABA-and flagellininduced closure critically depends on phosphorylation of PIP2;1 at Ser121 (Grondin et al., 2015;Rodrigues et al., 2017), SnRK2.6 being the most likely candidate for mediating these effects.The capacity of SnRK2.4 to directly regulate aquaporins in roots was addressed along the same lines, taking PIP2;1 as a prototypic PIP isoform.BiFC assay revealed that Col-0 SnRK2.4 can directly interact with PIP2;1.The protein kinase also phosphorylated PIP2;1derived peptides at Ser121 in vitro, and stimulated its water and H 2 O 2 transport activity upon coexpression in oocytes and yeast cells.Mutational study in yeast suggested that protein kinase activity of Col-0 SnRK2.4 is required for its enhancing effects on PIP2;1.This is in agreement with the activating role of loop B (Ser121 in PIP2;1) phosphorylation, as pointed out by molecular dynamics studies (T€ ornroth-Horsefield et al., 2006) and H 2 O 2 transport assays in transgenic wheat (Lu et al., 2022).Contrastingly, Cvi-0 SnRK2.4 failed in the in vitro phosphorylation, BiFC, and yeast growth assays, probably because it lacks part of a C-terminal regulatory domain and it is poorly expressed or instable in oocytes.More precisely, the Cvi-0 SnRK2.4 variant lacks the osmotic regulatory domain DI (amino acid residues 283-302) that stabilizes the catalytically essential aC helix of protein kinases, by aligning with aC within a hydrophobic pocket, in the amino terminal lobe of the protein (Yunta et al., 2011).Deletion of this domain abrogated the protein kinase activity of SnRK2.6/OST1(Belin et al., 2006).Overall, results from the two heterologous expression systems, Xenopus oocytes, and yeast cells, are consistent with in planta data showing that, in contrast to Col-0 SnRK2.4,Cvi-0 SnRK2.4 cannot complement the constitutive Lp r defects of snrk2.4.We propose that deletion of the DI domain in Cvi-0 SnRK2.4 represents a major causal polymorphism, although others may exist since SnRK2.4shows a 2-fold lower expression in Cvi-0 than Col-0 (Kawakatsu et al., 2016).
Despite the emerging functional importance of PIP phosphorylation at loop B (Grondin et al., 2015;Lu et al., 2022;Rodrigues et al., 2017; T€ ornroth-Horsefield  , 2006;Yaneff et al., 2016), all previous phosphoproteomic studies except a recent one in wheat (Lu et al., 2022) have failed to detect this phosphorylation in planta (Di Pietro et al., 2013;Van Wilder et al., 2008;Whiteman et al., 2008).Unfortunately, and at variance with the study by Lu et al. (2022), neither phosphoproteomics nor Phos-tag detection allowed us to detect loop B phosphorylation and thereby compare its abundance in Col-0 and snrk2.4.Yet, combination of our findings in vitro (Figure 3), in heterologous systems (Figures 4 and 5), and in planta (Figures 2C-E) clearly highlights the important role of Col-0 SnRK2.4 to increase Lp r under resting conditions.Therefore, we propose that the protein kinase mediates phosphorylation of PIPs at the well-conserved loop B phosphorylation site.SnRK2.4 is a well-described member of the SnRK2 family.These protein kinases have long been identified as central players in plant hormonal and environmental stress signaling (Kulik et al., 2011).In complement to biochemical evidence for ABA-dependent activation, double and triple knock-out mutants for group II and III members have established their positive role in osmotic and ABA signaling (Fujii & Zhu, 2009;Fujita et al., 2009;Mizoguchi et al., 2010;Nakashima et al., 2009).At variance, group I members such as SnRK2.4 and SnRK2.10 are thought to be essentially involved in plant responses to osmotic and oxidative stresses (Kulik et al., 2012;McLoughlin et al., 2012).Here, we show that SnRK2.4 plays a genuine role in plants under resting conditions and exerts a key control of root hydraulics.
Recent studies in yeast (Ruschhaupt et al., 2019) indicate that similar to SnRK2.6,SnRK2.4 and its group I homologs have all hallmark to be part of core ABA signaling modules.Canonical group II and III SnRK2s show negative regulation by clade A PP2Cs that act as ABA coreceptors (Umezawa et al., 2009;Vlad et al., 2009).One of these, ABI1, was shown to also interact with SnRK2.4, to dephosphorylate Ser158, thereby inhibiting its activity (Kawakatsu et al., 2016).Functional analysis in yeast further revealed the inhibitory action of various clade A PP2Cs on group I SnRK2s including SnRK2.4 (Ruschhaupt et al., 2019).Whereas the readout was transcriptional stimulation of reporter expression, we analyzed in this study PIP2;1 stimulation by SnRK2.4 and inhibition of the response by PP2Cs.We found that PP2C expression without SnRK2.4coexpression did not alter H 2 O 2 transport by PIP2;1, consistent with a direct regulatory role of PP2Cs on SnRK2.4 activity (Krzywinska et al., 2016).
These observations prompted us to investigate a possible contribution of SnRK2.4 to ABA-dependent regulation of Lp r .In guard cells, SnRK2s including SnRK2.6 are thought to mediate an ABA-dependent activation of PIP2;1 by ligand-receptor-mediated sequestering of the inhibitory PP2Cs (Grondin et al., 2015;Rodrigues et al., 2017).In contrast to stomata, Arabidopsis roots showed an inhibition of water permeability in response to high ABA concentration (1 lM) (Figure 6).Thus, in the presence of the hormone, ABA-dependent SnRK2s present in roots would act on molecular components that negatively target root aquaporins including PIP2;1.These would in turn dominate the direct positive effects of SnRK2.4 on their activity, which was revealed under resting conditions.ABA-induced induction of PP2Cs such as the Highly ABA-Induced (HAI) members, that act as effective inhibitors of ABA signaling (Tischer et al., 2017), might provide such down-regulatory mechanism.In such a scheme, the insensitivity of snrk2.4Lp r to short term (<3 h) inhibition by exogenous ABA may be explained by a signaling defect.Alternatively, SnRK2.4 and ABA may independently target the same pool of root aquaporins, and the insensitivity of snrk2.4Lp r to ABA may be due to a simple lack of activated aquaporins.Nevertheless, in the long term (>3 h), ABA seems to target aquaporins (including PIP2;1) in a SnRK2.4-independentmanner.
The Lp r phenotype of snrk2.4plants at resting conditions also challenges the classical view that SnRK2.4 is specifically activated in response to osmotic and salt stresses (Boudsocq et al., 2004).In plants under resting conditions, SnRK2.4 indeed shows a residual activation by as yet unknown factors, potentially the low endogenous levels of ABA.In support of this, yeast expression showed SnRK2.4 to be very ABA-sensitively regulated by the receptor complex RCAR11 and ABI1, compared to SnRK2.6 that was approximately a hundred times less sensitive (Ruschhaupt et al., 2019).The recent finding that Raf-like kinases act upstream of SnRK2s (Katsuta et al., 2020;Lin et al., 2020;Soma et al., 2020;Takahashi et al., 2020) as integrators of osmotic stress and ABA responses in plants opens interesting avenues to address the modes and amplitude of SnRK2.4 activation in plants under resting or stressing conditions.
While initially focused on root hydraulics, our work revealed that SnRK2.4 plays a more general role in regulating and integrating water transport throughout the whole plant.In addition to defects in ABA-dependent primary root growth inhibition, snrk2.4plants showed under resting conditions and with respect to Col-0, a significantly reduced stomatal conductance.This defect could be due to both a general hydraulic limitation and a genuine dysfunction of stomata.
In conclusion, the present work identifies SnRK2.4 as a key component of root functions and plant water transport that exists in alternative forms in Arabidopsis.Adding to the proposed role of SnRK2.6 in ABA-dependent regulation of guard cell aquaporins (Grondin et al., 2015, Rodrigues et al., 2017), this work raises important questions about the functional overlap and specificity of SnRK2.4 with respect to other ABA-dependent or independent SnRK2s.
Seeds were surface sterilized in 90% ethanol with 0.35% sodium hypochlorite for 7 min, and rinsed with 96% ethanol thrice.Surface sterilized seeds were sown on 0.5 9 MS agar vertical plates [2.2 g L À1 MS (Sigma, Saint Quentin Fallavier, France), 1% sucrose (Euromedex, Souffel Weyersheim, France), 0.05% MES (Euromedex), and 0.7% agar (Sigma), pH 5.7 adjusted using KOH], and incubated for at least 2 days at 4°C in the dark for stratification.Seedlings were grown on these plates for 10 days in a growth chamber at 70% relative humidity (RH) and 20°C, with cycles of 16 h of light (250 lE m À2 sec À1 ).They were transferred to a hydroponic medium as described previously (Javot et al., 2003).The plants were further grown for 12-13 days under the same growth conditions and phenotyped for Lp r .

Root hydraulic measurements
Lp r measurements were made on root systems of 22-to 23-dayold plants grown in hydroponic medium, treated or not with ABA.In brief, Arabidopsis detopped root systems were sealed at hypocotyl using a combination of plastic and metallic seals with a lowviscosity dental paste (President Light, Coltene, Switzerland) and inserted into a pressure chamber containing 50 ml hydroponic medium with or without ABA in a falcon tube.Lp r data were obtained as described previously (Shahzad et al., 2016).Pressure chamber measurements were performed throughout the day, indistinctly of the treatment.We also checked that Lp r values were independent of the time of day and, therefore, combined individual measurements according to treatments.For aquaporin inhibition experiments, excised roots were incubated in the presence of 1 mM NaN 3 and Lp r was derived from continuous measurement of water flow (J v ) at 320 kPa until a new steady state J v value was reached after 30 min.
Fixed HIFs were generated as described (Tuinstra et al., 1997) from segregating F7 RIL 8RV046 and 8RV062 and used for confirmation of Lprt18 and Lprb58, respectively.A heterozygous 8HV046 individual in the Lprt18 region was self-fertilized to develop a fine-mapping population, and a total of 152 individuals recombined within the candidate interval were identified.These recombinant HIFs (rHIFs) were further genotyped with additional markers between c1_02212 and msat1.10markers in order to locate the recombination breakpoint (primer sequences of markers are provided in Table S2) (Figure 2A).Twelve rHIFs were phenotypically tested in a 'progeny testing' process to discern if genotypic segregation was linked with segregation of Lp r phenotype (Loudet et al., 2008).

Quantitative and molecular complementation
For quantitative complementation, advanced rHIFs (arHIFs) were developed as described (Loudet et al., 2008) by crossing two closely related rHIFs (rHIF046-64 and rHIF046-99) differing for a small interval of 50 kb within Lprt18.Plants carrying one of the two allelic forms, arHIF046-Cvi-0 or arHIF046-Col-0, were used for making independent reciprocal crosses with snrk2.4-1and Col-0 plants.F1 plants originating from these crosses were phenotyped for Lp r , and each individual plant was genotyped using the IND1.03647marker to ensure that it was a real F1 with the expected allelic combinations.For molecular complementation, a 3752 bp genomic region (Chr1: 3655576-3 659 328, corresponding to Col-0 TAIR10 sequence) harboring SnRK2.4 was PCRamplified from Cvi-0 or Col-0 using iProof TM High-Fidelity PCR Kit (Bio-Rad, Marnes-La-Coquette, France).Primer sequences are provided in Table S2.The SnRK2.4 amplicons were cloned in a plant expression vector (pGreen 0179) and introduced into snrk2.4-1plants using the floral dip method (Clough & Bent, 1998).For each construct, three independent homozygous transgenic lines were selected in T2 generation on hygromycin B (30 mg L À1 ) and were phenotyped for Lp r .

Quantitative real-time (qRT)-PCR
Total RNA was isolated using an SV Total RNA Isolation System (Promega, Charbonni eres-Les-Bains, France) from roots of Col-0 and snrk2.4plants grown in hydroponics.Two lg of total RNA were used for first strand cDNA synthesis using M-MLV Reverse Transcriptase, RNase H Minus, Point Mutant (Promega), and Oligo (dT) 15 Primer (Promega) in a final volume of 25 ll, according to the manufacturer's instructions.qRT-PCRs were performed in 384-well plates with a LightCycler 480 Real-Time PCR System (Roche Diagnostics, Meylan, France) using 0.1 ll first strand cDNA as template.cDNA amplification was monitored using SYBR Green at an annealing temperature of 57°C as described previously (Shahzad et al., 2010).Transcript abundance was expressed relative to EF1a (At1g07920) transcripts in a Col-0 sample used as calibrator and, in all samples, relative to the average of UBC21 (At5g25760) and EF1a transcripts used as reference genes.The primer sequences used are described in Table S2.

ELISA assays
For experiments aimed at quantifying the overall abundance of PIP1 and PIP2 aquaporins, plants of Col-0 and snrk2.4 were cultured in hydroponics under standard conditions.Total proteins from the roots were purified as previously described (Boursiac et al., 2005).The ELISA assays were performed as described Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), doi: 10.1111/tpj.16494(Santoni et al., 2003).0.5 lg of protein extracts were loaded in triplicate in a carbonate buffer (30 mM Na 2 CO 3 , 60 mM NaHCO 3 , at pH9.5) on ELISA plates (Nunc Maxisorpâ, eBioscience SAS, France).The anti-PIP1 and anti-PIP2 antibodies used for the detection of PIPs were as described (Li et al., 2015).
Bimolecular fluorescence complementation and water permeability (P f ) measurements in Xenopus oocytes The PIP2;1 coding sequence was cloned in fusion with YFP C and YFP N in a 35S-SPYCE and 35S-SPYNE vector, respectively (Waadt et al., 2008).The constructs were then PCR amplified and cloned into a modified pGEMHE Xenopus laevis oocyte expression vector (Fisher Scientific, Illkirch Graffenstaden, France).A corresponding YFP N -SnRK2.4construct was obtained by substituting an Apa1-EcoR1 fragment in the former plasmid with the SnRK2.4coding sequence.cRNA production, expression in oocytes, and P f measurements were performed as described (Maurel et al., 1993) and fluorescence quantification was performed as explained in the previous sections.To determine the expression level of the BiFC constructs, oocytes extracts were analyzed by Western blot using an anti-YFPn antibody (abm, Vancouver, Canada).In brief, oocytes were ground in a RIPA buffer (abm), incubated for 25 min on ice, and centrifuged for 15 min at 7500 g.The supernatant was centrifuged again for 2 min at 10000 g and 9 ll of the resulting supernatant was mixed with 3 ll of Laemmli 4X prior to SDS-PAGE and immunoblotting.

Sub-cellular localization of PIP2;1
An RFP-PIP2;1 fusion protein under the control of a 35S promoter was obtained by cloning in a pB7WGR2 plant expression vector (Bellati et al., 2016) and introduced into Col-0 and snrk2.4-1genotypes using the floral dip method (Clough & Bent, 1998).For both Col-0 and snrk2.4-1genotypes, five independent transgenic lines were characterized.Plants expressing RFP protein fusions were observed using a Leica SP8 confocal laser-scanning microscope with a 40 9 NA.1.1 objective and a 561 nm laser.Light was collected between 580 and 620 nm.In the case of oocytes expressing constructs fused to YFP C or YFP N , a 209/0.3Dry objective and the 488 nm line of the argon were used.Fluorescence emission was collected between 520 and 560 nm.For quantitative measurement of PIP2;1 abundance (in plants) and split YFP experiments (in oocytes), the laser power, pinhole and gain settings of the confocal microscope were identical among different treatments.Calculation of mean pixel intensity was obtained for a set of defined regions of interest (ROI) after background subtraction (ImageJ).

Yeast expression and growth assays
The expression constructs for SnRK2.4 and PP2Cs for yeast and yeast transformation were reported earlier (Ruschhaupt et al., 2019).The cloning of the SnRK2.4alleles and PIP2;1 of Arabidopsis was performed accordingly.Briefly, cDNAs were cloned into level I (LI_BpiI) vector of the Golden Gate Cloning system (Chiasson et al., 2019).All LI vectors were verified for correctness by DNA sequence analysis.PIP expression constructs were assembled in the level II vector (LII) pYEAST_LII_2-3_CEN_HIS with ADH1 promoter and TDH1 terminator (tTDH1) for PIP2;1, pYEAST_LII_3-4_CEN_TRP with TDH3 promoter (pTDH3) and tTDH1 for SnRK2.4 variants, and pYEAST_LII_1-2_CEN_LEU with pTDH3 and tTDH1 for expression of PP2Cs.Yeast transformation of strain BMA64-1A (EUROSCARF, Uni-Frankfurt, Germany) and colony selection were performed according to Amberg et al. (2005).Freshly transformed yeast colonies were inoculated in 0.5 ml supplemented synthetic dextrose (SD) medium containing 2% glucose and incubated at 30°C in a Thermoshake-THO500 gyratory shaker (Gerhardt, K€ onigswinter, Germany) at 200 rpm for 16-18 h.After overnight growth, 4.5 mL SD medium was added and the culture was allowed to reach OD 600 between 0.6 and 0.8.Cells were sedimented (1500 g, 5 min) and the cell density was adjusted to OD 600 of 0.1.Ten ll of the cell suspension were spotted as a 10-fold dilution series on freshly prepared SD agar plates containing either no or 0.2 mM H 2 O 2 and incubated at 30°C for 2 days.

Stopped-flow spectrophotometry on isolated spheroplasts
Cells were grown in 25 ml of YPD medium [0.5% (w/v) yeast extract, 1% (w/v) peptone, 2% (w/v) glucose], with orbital shaking, at 28 °C and harvested in the exponential phase (OD600 % 1) by centrifugation (18 000 g) for 5 min at 4 °C, and washed once with H 2 O and once with 1,2 M sorbitol.Cells are resuspended in 2.5 ml of digestion buffer (1.2 M Sorbitol, 100 mM Sodium Phosphate (pH 6,5)) containing 20 mM DTT and 500 U of Lyticase and incubated at 30 °C.When the digestion was complete, the spheroplasts were centrifuged for 4 min at 1300 g (4 °C), and carefully washed twice in 5 ml of digestion buffer (-DTT).The pellet was finally carefully resuspended in 2 ml of digestion buffer (-DTT) and stored in ice for 10 min (4 °C) before stopped-flow experiments.The spheroplast preparations were homogeneous, spheroplasts were spherical in shape, and their average size was determined under light microscopy.Kinetics of spheroplast volume adjustment were followed by 90°light scattering at k ex = 515 nm.Measurements were performed at 15°C, in a SFM3 stopped-flow spectrophotometer (Biologic, Claix, France) essentially as previously described (Maurel et al., 1997).Briefly, spheroplasts were diluted 10-fold into the digestion buffer.They were then loaded in the stopped-flow device and mixed (dead time <3 ms) with an equal volume of the digestion buffer but with a concentration of 1.8 M sorbitol.The traces from more than 10 individual stopped-flow acquisitions were averaged and the curves were fitted to single exponential equations to determine an exponential rate constant k exp .The osmotic water permeability coefficient (P f ) was computed from the light scattering time course and the size of membrane vesicles according to the following equation: P f = k exp Á V o / A Á V w Á C out where V o is the initial mean spheroplast volume, A is the mean spheroplast surface, V w is the molar volume of water, and C out is the external osmolality.

Root growth assays
Seedlings were germinated on a ½ MS medium, transferred after 3 days onto the same medium in the absence or presence of 25 lM ABA, and further grown vertically for 7 days.Primary root length (PRL) was measured using ImageJ.

Stomatal assays
Stomatal conductance was measured using an AP4 leaf porometer on the abaxial and adaxial sides of rosette leaves of 22-day-old plants cultured in hydroponics.

Statistical analyses
Unless otherwise stated, the statistical significance of data at P < 0.05 was assessed using either Student's t-test (see *) or oneway ANOVA (see letters).The number of biological replicates (n) is indicated in figures or figure legends.

Figure 1 .
Figure 1.The Lprt18 QTL contributes to the natural variation of Lp r .(A) QTL map for root hydraulics (Lp r , L 0 ), root and shoot growth (DW r , DW s , and DW s /DW r ), and bolting time obtained by multiple QTL mapping in a Cvi-0 x Col-0 RIL population.A LOD threshold of 2.5 was obtained from permutation.Positive and negative LOD scores represent a positive contribution of Cvi-0 and Col-0 parental alleles, respectively.The Lp r QTLs are indicated using dotted rectangles.(B) Lp r of HIF046-Cvi-0 and HIF046-Col-0 lines in the absence or presence of NaN 3 .Mean values (AESE) from the number of individual plants indicated on the top.

Figure 2 .
Figure 2. SnRK2.4 underlies the root hydraulic Lprt18 QTL.(A) Genetic structure of recombinant heterogeneous inbred families (rHIFs) used to fine map Lprt18.Heterozygous and homozygous regions are represented by gray and black bars, respectively.Genetic markers are listed at the top of the map.Phenotypic results for segregation (+) or lack of segregation (À) of Lp r pheno- type (similar to original HIF046) are indicated for each rHIF.(B) Enlarged genomic region between SNP1.03630 and SNP1.03680 markers.The genes annotated in the fine-mapped candidate interval and the nucleotidic polymorphisms between Cvi-0 and Col-0 are indicated.Red, orange, green, blue, and black ticks represent adenine, thymine, guanine, cytosine, and indel, respectively.The indel polymorphism at At1g10940 (SnRK2.4)leading to a stop codon in the Cvi-0 coding DNA sequence is indicated by a black triangle.(C) Lp r of Col-0 and snrk2.4plants in the absence or presence of NaN 3 .Data (mean AE SE) from the indicated number (n) of plants.(D) Quantitative complementation of Lprt18.The graph shows the Lp r of F1 allelic combinations obtained after crosses of either arHIF046-Col-0 or arHIF046-Cvi-0 with Col-0 or snrk2.4-1.The Lprt18 x snrk2.4genotype interaction term is significant at P < 0.05 (Two-way ANOVA) (n = 29-30).(E) Lp r of Col-0, snrk2.4-1,and snrk2.4-1lines expressing Col-0 or Cvi-0 SnRK2.4 alleles.Same conventions as in (C).
Figure 2. SnRK2.4 underlies the root hydraulic Lprt18 QTL.(A) Genetic structure of recombinant heterogeneous inbred families (rHIFs) used to fine map Lprt18.Heterozygous and homozygous regions are represented by gray and black bars, respectively.Genetic markers are listed at the top of the map.Phenotypic results for segregation (+) or lack of segregation (À) of Lp r pheno- type (similar to original HIF046) are indicated for each rHIF.(B) Enlarged genomic region between SNP1.03630 and SNP1.03680 markers.The genes annotated in the fine-mapped candidate interval and the nucleotidic polymorphisms between Cvi-0 and Col-0 are indicated.Red, orange, green, blue, and black ticks represent adenine, thymine, guanine, cytosine, and indel, respectively.The indel polymorphism at At1g10940 (SnRK2.4)leading to a stop codon in the Cvi-0 coding DNA sequence is indicated by a black triangle.(C) Lp r of Col-0 and snrk2.4plants in the absence or presence of NaN 3 .Data (mean AE SE) from the indicated number (n) of plants.(D) Quantitative complementation of Lprt18.The graph shows the Lp r of F1 allelic combinations obtained after crosses of either arHIF046-Col-0 or arHIF046-Cvi-0 with Col-0 or snrk2.4-1.The Lprt18 x snrk2.4genotype interaction term is significant at P < 0.05 (Two-way ANOVA) (n = 29-30).(E) Lp r of Col-0, snrk2.4-1,and snrk2.4-1lines expressing Col-0 or Cvi-0 SnRK2.4 alleles.Same conventions as in (C).

Figure 5 .
Figure 5.The protein kinase activity of SnRK2.4 is required to stimulate PIP2;1 activity upon co-expression in yeast.(A) Yeast spheroplast water permeability (P f ) was determined by stopped-flow spectrophotometry (see Materials and methods).The figure shows the relative P f values (means AE SE) of control yeasts (EV) or yeasts co-expressing SnRK2.4Col-0 and PIP2;1 (red) with respect to yeasts expressing PIP2;1 alone (gray).Measurements were made in triplicate for every biological repeat (single yeast culture), the number of such repeats being shown in the figure.(B)Growth assay of yeast lines expressing different SnRK2.4 alleles, PIP2;1, and the PP2C ABI1 for enhanced sensitivity towards H 2 O 2 .The expression of the catalytically active Col-0 allele and PIP2;1 is marked by the plus sign.Yeast expressing the Cvi-0 allele, ABI1, and functional inactive forms of SnRK2.4 and ABI1 with the mutation D123A and D177A in the catalytic domain, respectively, are indicated.Yeast cultures were grown to comparable cell densities and serial tenfold dilutions were spotted on solidified medium in the presence and absence (control) of 0.2 mM H 2 O 2 .Growth of three independent transformants per combination was assessed after 2 days at 30°C.

Figure 6 .
Figure 6.Effects of exogenous ABA on Lp r of various snrk2.4and pip genotypes.(A) Time-dependent effects of exogenous ABA application (1 lM at t = 0) on Lp r of Col-0 and snrk2.4-1plants.Cumulated data (mean AE SE) from 6 independent experiments with n = 12-40 plants.Different letters (ANOVA) indicate statistical differences between kinetic points for the same genotype (Col-0: red; snrk2.4-1:black).(B) Effects of ABA (1 lM, 5 h) on Lp r of Col-0, a quintuple pip mutant (Quint) and the same genotype complemented with a GFP-PIP2;1 fusion (Comp).Data (mean AE SE) from 4 independent experiments and the number of individual plants indicated on the top.The three genotypes showed distinct Lp r values in the absence of ABA (ANOVA; see different letters).A genotype-by genotype t-test analysis indicated that ABA inhibits Lp r in Col-0 and Comp (*) but not Quint (not significant: n.s.).Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), doi: 10.1111/tpj.16494

Ó
2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), doi: 10.1111/tpj.16494