CAM guard cell anion channel activity follows transcript abundance and is suppressed by apoplastic malate.

* Plants utilizing crassulacean acid metabolism (CAM) concentrate CO2 around RuBisCO while reducing transpirational water loss associated with photosynthesis. Unlike stomata of C3 and C4 species, CAM stomata open at night for the mesophyll to fix CO2 into malate (Mal) and store it in the vacuole. CAM plants decarboxylate Mal in the light, generating high CO2 concentrations within the leaf behind closed stomata for refixation by RuBisCO. * CO2 may contribute to stomatal closure but additional mechanisms, plausibly including Mal activation of anion channels, ensure closure in the light. * In the CAM species Kalanchoë fedtschenkoi, we found that guard cell anion channel activity, recorded under voltage clamp, follows KfSLAC1 and KfALMT12 transcript abundance, declining to near-zero by the end of the light period. Unexpectedly, however, we found that extracellular Mal inhibited the anion current of Kalanchoë guard cells, both in wild-type and RNAi mutants with impaired Mal metabolism. * We conclude that the diurnal cycle of anion channel gene transcription, rather than the physiological signal of Mal release, is a key factor in the inverted CAM stomatal cycle.


Introduction
Stomata are pores that facilitate gas exchange across the largely impermeable cuticle of leaves and stems. They open and close in response to exogenous and endogenous signals and thereby control the exchange of gases, most importantly water vapour and CO 2 , between the interior of the leaf and the atmosphere (Jezek & Blatt, 2017). Stomata exert major controls on the water and carbon cycles of the world and can limit photosynthetic rates by 50% or more when demand exceeds water supply (Lawson & Blatt, 2014;Franks et al., 2017). Transpirational water loss through stomata is a key factor in the challenges of fresh water availability and crop production that are expected to unfold over the next 20-30 years (Franks et al., 2017). Their activity is a critical determinant of water use efficiency (WUE), often defined as the amount of dry matter produced per unit of water transpired (Leakey et al., 2019). Indeed, a very large body of data relates stomata, transpiration and carbon assimilation (Hetherington & Woodward, 2003;Lawson & Blatt, 2014), highlighting the trade-off between CO 2 availability and stomatal transpiration.
Guard cells surround the stomatal pore and regulate its aperture. They coordinate membrane transport through a complex network of intracellular signals to regulate solute flux, mainly of K + , Cl À and malate (Mal), and drive guard cell turgor (Hetherington & Woodward, 2003;Jezek & Blatt, 2017). Our deep knowledge of these processes has made the guard cell one of the best known plant cell models for membrane transport, signalling and homeostasis. Several well defined signals, including light and CO 2 , affect ion transport to drive changes in water flux and alter cell volume, turgor and stomatal aperture. In most plants, light promotes stomatal opening by activating H + -ATPases through a blue light-dependent signal cascade (Inoue & Kinoshita, 2017) leading to hyperpolarisation of the plasma membrane. These changes in voltage engage inward-rectifying K + channels and H + -coupled K + and Cl À symporters to facilitate osmotic solute uptake (Jezek & Blatt, 2017) while suppressing anion channels at the plasma membrane (Roelfsema & Hedrich, 2005;Marten et al., 2007;Ando & Kinoshita, 2018). Stomatal closure in the dark and at high partial pressures of CO 2 (pCO 2 ) is signalled through both Ca 2+ -independent and Ca 2+ -dependent pathways that inactivate inward-rectifying K + channels and activate outward-rectifying K + and Cl À channels, including the SLAC1 and ALMT12 anion channels in Arabidopsis, to promote net osmotic solute loss (Blatt, 1990;Lemtiri-Chlieh & MacRobbie, 1994;Brearley et al., 1997;Grabov & Blatt, 1998Marten et al., 2007;Chen et al., 2010;Jezek & Blatt, 2017).
Stomata of plants exhibiting crassulacean acid metabolism (CAM) differ in their regulation from this norm. At the heart of CAM is a syndrome of metabolic and physiological adaptations that facilitate CO 2 concentration around RuBisCO while circumventing much of the water loss associated with gas exchange for photosynthesis (Hartwell et al., 2016;Yang et al., 2017). Compared with C 3 plants, CAM plants yield up to a 16-fold increase in WUE with little cost in assimilation (Borland et al., 2009;Hartwell et al., 2016), characteristics that underpin their adaptation to arid environments. They achieve these savings by initially fixing atmospheric CO 2 in the dark into Mal for storage in the mesophyll vacuole; they release and decarboxylate the stored Mal in the light for refixation by RuBisCO. CAM avoids futile cycling by temporal separation and optimisation of primary and secondary CO 2 fixation via transcriptional, translational and post-translational control of key metabolic enzymes and transporters in the mesophyll (Hartwell, 2006;Abraham et al., 2016;Boxall et al., 2017;Yang et al., 2017).
Significantly, CAM depends on a cycle of stomatal movement that is inverted relative to that of most other plants: CAM stomata open at night and close during the day, facilitating high pCO 2 within the leaf to favour its fixation by RuBisCO (Von Caemmerer & Griffiths, 2009;Owen & Griffiths, 2013). How CAM stomata achieve this inverted diurnal cycle remains a puzzle. During the daylight, CO 2 released with Mal breakdown is thought to promote stomatal closure, however experimental manipulations of pCO 2 in Kalancho€ e have suggested that stomatal closure in the daylight phase of the CAM cycle is mediated through means additional to that of high internal pCO 2 (Von Caemmerer & Griffiths, 2009). Furthermore, preventing nocturnal CO 2 fixation to malate, either by supplying leaves with pure nitrogen or CO 2 -free air, or using transgenic mutants that lack key CAM enzymes, shows that CAM stomata continue to close at least partially in the light period, even in the absence of an internal supply of CO 2 normally afforded by the decarboxylation of malate in the mesophyll (Borland et al., 1999;Von Caemmerer & Griffiths, 2009;Dever et al., 2015;Boxall et al., 2017).
One plausible factor promoting stomatal closure could be the activation of plasma membrane anion channels by daytime Mal loss from the mesophyll to the leaf apoplast. Such a mechanism would accord with evidence that the anion channels in guard cells of some plants are enhanced by extracellular Mal (Hedrich & Marten, 1993;Hedrich et al., 1994;Wang & Blatt, 2011;Mumm et al., 2013), which may thus be an important factor in promoting anion efflux from the guard cells and stomatal closure. However, until now, this mechanism has not been examined in the guard cells of a CAM plant. Here we report that, in guard cells of the CAM species Kalancho€ e fedtschenkoi, the anion current is not enhanced, but is suppressed by extracellular Mal. We find that the relative anion current amplitude during the diurnal cycle follows the pattern of transcript levels, declining to a minimum at the end of the daylight period. Furthermore, we show that CAM loss-of-function RNAi mutants of mitochondrial NAD malic enzyme (NAD-ME) and pyruvate orthophosphate dikinase (PPDK), which are altered in nocturnal Mal accumulation and its decarboxylation in the light (Dever et al., 2015), exhibit anion currents similar to the wild-type plants. These, and additional findings, led us to conclude that Mal release from the mesophyll cannot serve as a metabolic signal in CAM stomatal control and that the diurnal cycle in gene transcription and turnover of the anion channels is vital for the temporal coordination of the CAM stomatal cycle.

Plant growth and tissue preparation
Kalancho€ e fedtschenkoi Hamet et Perrier plants were propagated clonally on a mixture of John Innes no. 3 potting soil and perlite (70 : 30) at 22°C and 60% relative humidity under 200 lmol m À2 s À1 white light in a 16 h : 8 h, light : dark cycle. NAD-ME and rPPDK1 RNAi mutant lines of Kalancho€ e fedtschenkoi were generated as described previously (Dever et al., 2015) and grown under the same conditions. For all experiments, epidermal tissue was taken from leaves at positions 6-8 leaf pairs below the apex corresponding to tissue fully matured and undergoing obligatory CAM (Dever et al., 2015;Boxall et al., 2017). Representative examples of gas exchange from these leaves of wild-type, NAD-ME and rPPDK1 RNAi mutant lines of Kalancho€ e fedtschenkoi are found in Supporting Information Fig. S2 of Dever et al. (2015). Epidermal peels were prepared from these leaves and mounted, much as described previously (Blatt, 1987a;Wang & Blatt, 2011). All operations were carried out on an Axiovert S100TV microscope (Zeiss, http://www.zeiss.com/) fitted with Nomarski differential interference contrast optics. Peels were preincubated for 1 h in 50 mM KCl and 5 mM 2-(N-morpholino)ethanesulphonic acid (MES) titrated to pH 6.1 with KOH.
Measurements were conducted at 22°C in continuous flowing solutions controlled by a gravity-fed system at a rate of 20 chamber volumes min À1 . The standard perfusion medium contained 15 mM tetraethylammonium chloride (TEA-Cl), 15 mM CsCl and 5 mM MES, titrated with Ca(OH) 2 to a pH 6.1 ([Ca 2+ ] = 1.2 mM). Malate (Mal) and other compounds were added and removed as indicated during the course of the experiments. Thus, each guard cell served as its own control with tests against possible 'rundown' of the current over the course of each impalement. Surface areas and volumes of impaled guard cells were calculated assuming a spheroid geometry using HENRY IV EP software (Glasgow University, http://www.psrg.org.uk). Chemicals were from Sigma Aldrich (Poole, UK) unless otherwise specified.

Electrophysiology and current analysis
Voltage clamp recordings were obtained using double-barrelled microelectrodes coated with paraffin to reduce capacitance and filled with 100 mM CsCl (pH 7.5) to block K + channel currents (Grabov et al., 1997;Chen et al., 2010). Microelectrodes were connected to amplifier headstages via 1 M KCl/Ag-AgCl halfcells, and a 1 M KCl agar bridge served as the reference electrode. Membrane voltage was typically clamped at +40 mV for 10 s to activate the anion current before stepping the voltage to values between +40 and À200 mV, much as described previously (Grabov et al., 1997). Current and voltage were recorded at 2 kHz after filtering the current with an 8-pole Bessel filter set to New Phytologist (2020) 227: 1847-1857 © 2020 University of Glasgow New Phytologist © 2020 New Phytologist Trust www.newphytologist.com

Research
New Phytologist a cut-off frequency of 0.5 kHz. Currents were analysed using HENRY IV EP software and fitted by nonlinear least-squares using SIGMAPLOT v.11.1 (Systat, London, UK) as described before (Chen et al., 2010;Wang & Blatt, 2011). Instantaneous and steady-state currents were determined at 20 ms and 10 s, respectively, into clamp steps following channel activation. Data are presented as means AE SE of n observations and differences validated post hoc by ANOVA or Student's t-test.

Epidermal peel sampling for RNA isolation
Before sampling for leaf epidermal peels, plants were entrained for 7 d in a Snijders Microclima MC-1000 growth cabinet set to a 12 h : 12 h, light : dark (450 µmoles m À2 s À1 ), 25°C : 15°C, 60% humidity : 70% humidity cycle. Epidermal peel samples were isolated from leaf pairs 6-8, which perform full CAM in the wild-type (Boxall et al., 2020). Peels were collected from individual clonal plants every 4 h over a 12 h : 12 h, light : dark cycle (12 : 12 LD), starting at 00:00 h when the lights came on. Epidermal peels were detached from the leaves using a 'snap and peel action' that destroys >95% of the pavement cells and were immediately frozen in liquid nitrogen. Six leaves were peeled for each sample to obtain both upper and lower epidermides. Epidermal samples were stored at À80°C until use.

Total RNA isolation and RT-qPCR
Total RNA was isolated from 100 mg of frozen, ground leaf epidermal tissue using the Qiagen RNeasy kit (Qiagen, Germany) following the manufacturer's protocol with the addition of 13.5 µl 50 mg ml À1 PEG 20 000 to the 450 µl RLC buffer used for each extraction. cDNA was synthesised from the total RNA using the Qiagen Quantitect RT kit according to the manufacturer's instructions (Qiagen, Germany). The resulting cDNA was diluted 1 : 4 with molecular biology grade water before use in RT-qPCR. Transcript levels were determined using the SensiFAST SYBR No Rox kit (Bioline, London UK) in an Agilent MX3005P qPCR System Cycler. The results for each target gene transcript of interest were normalised to the reference gene THIOESTERASE/THIOL ESTER DEHYDRASE-ISOMERASE SUPERFAMILY PROTEIN (TEDI; Kaladp0068s0118.1; Arabidopsis orthologue AT2G30720.1). Gene expression in a pool of RNA generated from epidermal peels from LP6 samples collected every 4 h over a 12 h : 12 h, light : dark cycle was set to 1. Primer pairs for RT-qPCR analyses were as follows: SLAC1 Kal-adp0050s0214.1 (SLAC1F GTACTTCTCGTCGCTAAAGGG; SLAC1R ACTCAAGTCTCCATTTCAGCG) and ALMT12 Kaladp0091s0013.1 (ALMT12F GTGACATCGAAATCAATG TGAC; ALMT12R AAATGGGAAGGAGCCTGTTTC).

Kalancho€ e exhibits a characteristic SLAC-like anion current
Stomata of Kalancho€ e are comparable in size with those of Vicia and Nicotiana, making them attractive targets for microelectrode impalement once isolated in epidermal peels (Fig. 1). For these studies, we peeled the mature leaves of Kalancho€ e fedtschenkoi, taken from leaves positioned at pairs six to eight below the shoot apex. At these positions, the leaves are known to be fully developed in CAM metabolism; they retain full CAM activity in isolation and show no appreciable plasticity in this activity (Borland et al., 2009;Dever et al., 2015;Boxall et al., 2017). Representative examples of gas exchange from these leaves, and from leaves at the same positions in the NAD-ME and rPPDK1 RNAi mutant lines of Kalancho€ e fedtschenkoi employed below, are found in Fig. S2 of Dever et al. (2015). In every case, we found the guard cells to be easily impaled with multibarrelled microelectrodes when peels were mounted as described before for impalements (Blatt, 1987a;Grabov et al., 1997;Chen et al., 2010).
We used double-barrelled microelectrodes for two-electrode voltage clamp recording, pulled to yield tips equivalent to those described previously with tip resistances greater than 200 MΩ when filled with 200 mM K + acetate and measured in 10 mM KCl solution (Blatt, 1987b;Blatt & Armstrong, 1993). To analyse the Kalancho€ e anion channels, the electrodes were filled with 100 mM CsCl. Additionally, all measurements were carried out under continuous superfusion of epidermal peels with a bathing solution of 5 mM Ca 2+ -MES buffer, pH 6.1, with additions of 15 mM CsCl and 15 mM tetraethylammonium chloride (TEA-Cl). The combination of the Cs + electrolyte with Cs + and TEA + in the bath ensured that all background K + channel current was blocked in Nicotiana and Arabidopsis (Grabov et al., 1997;Chen et al., 2010), and this was found to hold true also for Kalancho€ e guard cells. The characteristics of Kalancho€ e guard cell K + channels will be the subject of a later publication. All measurements reported here were carried out by adding and removing compounds by superfusion in the bath during the course of recordings. Thus, each cell served as its own control for each of the various experimental treatments. On continuous superfusion with 15 mM CsCl and 15 mM TEA-Cl (30 mM Cl À ) in the bath, clamping the membrane to voltages of +30 to +40mV yielded a slow-activating current that deactivated over periods of several seconds on stepping to voltages negative of 0 mV. Fig. 2 shows both the instantaneous and steady-state currents recorded following 10-s steps to +40 mV to activate the current. A significant feature of the instantaneous current was its modest dependence on voltage (Fig. 2a) which relaxed to a steep dependence in the steady state (Fig. 2b,c). The current also showed kinetics in deactivation with halftimes in the range of 1-3 s (Fig. 2b). Fitting the steady-state currents to a Boltzmann function (see Fig. 2, legend) indicated a voltage dependence consistent with a channel gating charge of c. 1 and a mid-point voltage near 0 mV (Fig. 2d). All of these characteristics were similar to those reported previously for SLAC-like anion currents of Vicia, Nicotiana and for SLAC1 in Arabidopsis (Schroeder & Keller, 1992;Schwartz et al., 1995;Grabov et al., 1997;Roelfsema et al., 2004;Geiger et al., 2009;Chen et al., 2010).
Analysis of the current relaxations on increasing Clin the bath indicated a negative displacement of the apparent reversal voltage as expected for an anion permeant channel (Fig. 2e), albeit with a sub-Nernstian dependence concentration suggesting that Clis not the only permeant ion. Many anion channels are equally, or more permeant to NO 3 - (Schmidt & Schroeder, 1994). Because the dominant current clearly was carried by anion passage out of the cell, we loaded guard cells by including CsNO 3 in the microelectrodes. In this case, we observed no difference in instantaneous current and only a modest shift to more negative voltages in the reversal voltage of the steady-state current (Fig. S1). This difference is less than reported for Vicia guard cells (Schmidt & Schroeder, 1994;Grabov et al., 1997) and may reflect a lower relative conductance for NO 3 -. Additions of 20 µM of the anion channel blocker 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) to the bath during recordings eliminated the current (see Fig. 2), and the current recovered on washing out the blocker from the superfusion medium (see Fig. S2) so precluding current rundown. Thus, we concluded that the current was carried by Cl --and NO 3 --permeable anion channels at the plasma membrane.
Kalancho€ e anion channel activity follows transcript abundance during the daylight period The diurnal CAM cycle is commonly divided between four phases (see Fig. 3a). Phase I corresponds to the period of nocturnal stomatal opening when CO 2 is fixed by PEPC to form malic acid, which is then transported into the mesophyll vacuole. Phase II marks a shift to RuBisCO activity just after dawn when, for a brief period, CO 2 is fixed by both PEPC and RuBisCO. Phase III marks a long phase of malic acid decarboxylation and CO 2 concentration behind closed stomata that promotes its refixation These currents relaxed to the mean steady-state curves (c) that showed little inward current, an apparent activation at voltages positive of À60 mV, and reversal voltages positive of 0 mV. We analysed the mean steady-state IV curves using a Boltzmann function with an offset of the form: by RuBisCO. Finally, Phase IV marks the end of the light period when stomata typically reopen (Luttge, 2004;Davis et al., 2014). We recorded anion currents at intervals over 10 h spanning the light period into the beginning of the dark period, that is through Phase III to the beginning of Phase I. Current amplitudes quantified at À120 mV from over 50 independent experiments (Fig. 3b) showed a clear trend and substantial anion current early in Phase III with a mean amplitude of À197 AE 17 µA cm À2 at À120 mV, declining over the daytime to a mean current amplitude at the beginning of Phase IV of À51 AE 6 µA cm À2 at the same voltage. Statistical differences were evident in measurements between time points in the first 6 h, those at 8 h and measurements taken thereafter up to 12 h into the diurnal cycle. At the beginning of Phase I the anion current was generally too small to resolve. Throughout the period of measurements, the currents showed little evidence of a change in reversal voltage, consistent with a primary effect on the number of channels. Joint fittings yielded common values for the gating parameters V ½ and d between data sets; only the maximum ensemble conductance G max varied and, like the instantaneous current (Fig. 3c), G max decreased over the light period. We also challenged guard cells with 20 µM NPPB at intervals to confirm the identity of the current. In every case, the current was inhibited by the anion channel blocker (Figs 3d,  S2). We cannot rule out more subtle alterations in gating control. Nevertheless, the lack of change in the key gating parameters of V ½ , d and the apparent reversal voltage, and the constant sensitivity to the anion channel blocker NPPB, all suggested that the intrinsic origin of the current did not vary over the diurnal cycle. The consistency in V ½ and d, and the sensitivity to NPPB also discounted any significant changes in distribution between any underlying subpopulations of ion channels over the diurnal period. Thus, we concluded that the primary effect of daytime period was on the mean current amplitude.
To assess the possible association of the current amplitude with gene expression, we examined the transcript abundance of the two main anion channel genes, KfSLAC1 and KfALMT12 (Boxall et al., 2020). RNA was isolated using separated epidermal peels from leaf pairs 6-8 of wild-type K. fedtschenkoi to enrich for intact guard cells from CAM-performing leaves. We also examined the transcript abundance of KfSLAC1 and KfALMT12 in epidermal peels of the Kalancho€ e rNAD-ME1 and rPPDK1 RNAi mutant lines (Dever et al., 2015;Boxall et al., 2017) with suppressed activities of mitochondrial NAD malic enzyme (NAD-ME) and pyruvate orthophosphate dikinase (PPDK). These enzymes play key roles in Mal decarboxylation in the light period during CAM. Mitochondrial NAD-ME releases CO 2 and pyruvate from malate, and PPDK converts pyruvate to PEP, which is subsequently recycled through gluconeogenesis to starch. These mutants were previously shown to reduce Mal cycling by roughly 70% (Dever et al., 2015). We reasoned that the current might track the relative transcript abundance if the current amplitude reflected a diurnal variation in anion channel transcription and translation. We also anticipated that a similar overall pattern in anion channel transcript abundance might be evident, even if Mal cycling was reduced in the rNAD-ME1 and rPPDK1 RNAi mutant lines.
(d) Fig. 3 The Kalancho€ e anion current declines throughout the daylight period of the CAM cycle. (a) The Kalancho€ e CAM cycle is commonly divided between Phases I-IV as illustrated (refer to text for a description). Rc refers to the principle voltage clamp recording period. (b) Recordings were carried out throughout Phases III and IV to the beginning of Phase I and quantified as the mean instantaneous current recorded in 30 mM Clat À120 mV from IV curves such as shown in (c) following a 10-s activation at +40 mV. Data are full scatter of measurements from more than 50 independent experiments (closed circles) in which each data point represents a single experiment and the mean of at least three measurements from one stomatal guard cell. Also shown are the means (open squares) binned at 1-h intervals with n > 5 independent experiments within each bin. Each data point represents the means of at least three measurements from one stomatal guard cell. The solid curve is a nonlinear least-squares fitting of the current values to a sigmoidal function and is included for visual reference. Letters above indicate statistical differences (P < 0.05). (c) Instantaneous IV curves presented as means of more than 13 separate recordings at times between 4-6, 6-8 and 8-10 h. Data are cross-referenced by symbol to representative current traces in (d). (d) Anion current sensitivity to 20 lM of the anion channel inhibitor 5-nitro-2-(3-phenylpropyl-amino)benzoic acid (NPPB) showed block independent of the daytime period. Here measurements were carried out (top to bottom) at times between 4-6 (closed squares), 6-8 (closed triangles) and 8-10 h (filled circles; see (a, c) above) and symbols cross-referenced to the IV curves in (c). Scale, 400 pA (vertical), 1 s (horizontal). Representative data for NPPB washout are included in Supporting Information Fig. S2.
Ó 2020 University of Glasgow New Phytologist Ó 2020 New Phytologist Trust New Phytologist (2020) 227: 1847-1857 www.newphytologist.com In the wild-type, we found that KfSLAC1 transcript abundance peaked around dawn and declined steeply throughout the light period, reaching a minimum after 8-12 h of light (Fig. 4, above). This temporal pattern was broadly followed in both rNAD-ME1 and rPPDK1 mutant lines, although the overall level of KfSLAC1 transcript was marginally reduced in the rPPDK1 and rNAD-ME1 mutants. KfALMT12 transcript levels also declined to a minimum 8 h into the light period in the wild-type, and rose in the dark to a daily peak at 20:00 h, 4 h before dawn (Fig. 4,  below). KfALMT12 transcript levels were roughly 20-50% higher in both rNAD-ME1 and rPPDK1 mutants relative to the wild-type, although the differences were not substantial except at the 16 and 20 h time points (Fig. 4, below). A direct comparison of transcript abundance and current amplitudes between the three genotypes is not meaningful for several reasons, which we return to at a later point in this paper. Furthermore, we can anticipate some lag time between transcription and the appearance of the functional channels at their target membrane. Nonetheless, it is clear that, in wild-type Kalancho€ e and in the two metabolically impaired mutants, the transcript abundance of the two main anion channel genes broadly anticipated the variation in the anion current activity recorded in vivo.

Kalancho€ e guard cell anion channels are suppressed by extracellular malate
Activation of guard cell anion channels is a vital component of the mechanism driving stomatal closure and, in several species, has been reported to be favoured by extracellular Mal (Hedrich & Marten, 1993;Hedrich et al., 1994;Wang & Blatt, 2011;Mumm et al., 2013). Because Phase III, which occupies the majority of the light period, is associated with substantial Mal release from the mesophyll vacuole for decarboxylation (Luttge, 2004;Davis et al., 2014), we reasoned that some Mal might be lost to the apoplast as the organic acid anion. If so, it could serve to enhance the anion channel activity of the Kalancho€ e guard cells, thereby promoting stomatal closure.
To test this idea, we challenged the guard cells, adding 0.1-10 mM Mal outside by superfusion in the bath during experiments, while recording the anion current under voltage clamp. Because impalements could generally be held for periods of 30-40 min only, trials were also carried out over two different time periods within Phase III of the CAM cycle. In every case (Figs 5, S2, S3), we found that adding Mal to the bath suppressed the anion current, notably at millimolar concentrations, while at 0.1 mM Mal we observed no appreciable change in the anion current. We also observed the current to recover when Mal was washed out of the bath (see Fig. S2), so discounting a general rundown of the current during recordings. The reduction in the anion current and conductance with Mal also argues against a major permeability of the channels to Mal. The effects of millimolar Mal were most evident in the first half of Phase III. Later in Phase III the current was much reduced, making quantification of any additional reductions in current with Mal difficult to assess. Regardless, it was clear that low millimolar Mal led to a suppression of the anion current, not its enhancement. Thus, we concluded that, in itself, Mal was not effective as a signal to enhance the guard cell anion current.
To validate this conclusion, we also examined the current in the rNAD-ME1 and rPPDK1 RNAi mutant lines (Dever et al., 2015). We reasoned that one mechanism favouring stomatal opening in these mutant lines might be the reduced activity of the anion current. However, voltage clamp analysis of the guard cells showed anion currents in both mutant lines that were statistically equivalent to those observed in guard cells from wild-type plants; like the wild-type plants, the currents of the two RNAi lines showed a similar decline over the light period of the CAM cycle (Fig. 6), consistent with the pattern of transcript abundance (Fig. 3). Thus, we concluded that changes in extracellular Mal Fig. 4 Transcripts for the principal Kalancho€ e guard cell anion channel genes KfSLAC1 and KfALMT12 decline throughout the daylight period to a minimum during Phase IV. Data are the means AE SE of six technical replicates for each of three independent experiments are for KfSLAC1 (a) and KfALMT12 (b) measured using RNA taken from wild-type Kalancho€ e (closed circles) and from the rNAD-ME1 (upright triangles) and rPPDK1 mutant lines (inverted triangles). Data are plotted as a function time with the dark period indicated by the grey shading. Note that the 0 and 24 h time points are the same. Within each genotype, for KfSLAC1, data at 8, 12 and 16 h are significantly less than at the 0 (24), and 4 h time points; for KfALMT12, data at 16 and 20 h are significantly greater than at the 0 (24), 4, 8 and 12 h time points; in each case, for both transcripts, data at 0 (24) h is significantly greater than at 8 h (P < 0.05).

Research
New Phytologist could not explain the elevated anion current during the daylight period of the CAM cycle and we return to this point at a later point in this paper.

Discussion
CAM depends fundamentally on a diurnal cycle of stomatal movements that is inverted relative to that of C 3 and C 4 plants.
CAM plants achieve substantial gains in WUE through nocturnal stomatal opening and CO 2 capture when the driving force for transpiration is low, and through CO 2 release behind closed stomata during the day which elevates pCO 2 in the leaf to facilitate its fixation by RuBisCO in the light (Ting, 1985;Luttge, 2004;Davis et al., 2014). In C 3 species, there is strong evidence demonstrating that elevated pCO 2 within the leaf promotes stomatal closure (Lawson & Blatt, 2014;Jezek & Blatt, 2017;Zhang et al., 2018), but several studies have indicated that pCO 2 alone cannot explain CAM stomatal behaviour. Notably, Von Caemmerer & Griffiths (2009) observed that Kalancho€ e stomata opened when pCO 2 was reduced in the dark and at the end of the light period, but they failed to respond to a decreased pCO 2 in the light, even when pCO 2 in the leaf was reduced experimentally.
We have asked whether anion channel activity follows transcript abundance for the orthologues of the two major anion channels associated with stomatal closure in Arabidopsis (Jezek & Blatt, 2017). Mal release to the apoplast from the mesophyll during the daylight period might act also as a signal between the mesophyll and guard cells for stomatal closure to help elevate pCO 2 within the leaf. A critical question in this case is whether the activity of the anion channels might be enhanced by extracellular Mal, as has been reported in several non-CAM species. Our analysis for the guard cells of Kalancho€ e fedtschenkoi yielded four key observations. We found that: (1) the anion current activity exhibited a pattern high activity early in the light that declined towards the end of the light period, consistent with the diurnal cycle of stomatal movements; (2) this variation in current activity followed the diurnal cycle of transcript abundance for the major (3) manipulating Mal accumulation in the dark and decarboxylation in the light with genetic mutants showed no substantial effect on the anion current; and, finally, (4) experimental additions of Mal in the apoplast, even at low millimolar concentrations, led to an inhibition in the anion current, not its enhancement. These findings demonstrated an unexpected sensitivity to external Mal in Kalancho€ e guard cells that was opposite to that found for anion channels in non-CAM guard cells. They indicated that a mechanism of anion channel stimulation by apoplastic Mal cannot explain the diurnal cycle of stomatal closure in Kalancho€ e and that the variation in stomatal movements is likely to depend on the diurnal cycle of anion channel transcription and turnover.
Could apoplastic Mal signal stomatal closure in Kalancho€ e?
There is good reason to suspect that the apoplast of CAM leaves might exhibit diurnal variations in apoplastic Mal content although, to our knowledge, there exist no direct measurements of Mal in the CAM leaf apoplast. Several anion channels of plant membranes, including those of CAM plants, are conductive for Mal (Keller et al., 1989;Smith et al., 1990;Iwaskaki et al., 1992;Cheffings et al., 1997;Kohler & Raschke, 2000;Hafke et al., 2003;Meyer et al., 2010;Medeiros et al., 2016;Eisenach et al., 2017) and there is evidence that Mal can be lost to the apoplast, at least from the guard cells of Vicia and Commelina during stomatal closure (Dittrich & Raschke, 1977;Van Kirk & Raschke, 1977, 1978. Micromolar to low millimolar concentrations of Mal have been identified in the leaf apoplast of several non-CAM, dicotyledonous species (Gabriel & Kesselmeier, 1999;Lopez-Millan et al., 2000;Hedrich et al., 2001;O'Leary et al., 2016). Furthermore, we know that Mal is the most abundant metabolite that undergoes a substantial diurnal rhythm in the CAM mesophyll (Chen et al., 2002). Thus, if Mal is found in the Kalancho€ e leaf apoplast, we anticipated that it might well be elevated during the light period by loss across the mesophyll plasma membrane when Mal is released from the vacuole for decarboxylation.
The anion channels of guard cells of Vicia, Nicotiana and Arabidopsis have been the focus of many studies over the past 3 decades. All three species show very similar channel characteristics at the plasma membrane that divide between dominant, slowly activating (S-type) and transient, rapidly activating (Rtype) currents (Keller et al., 1989;Schroeder & Keller, 1992;Grabov et al., 1997;Chen et al., 2010;Wang & Blatt, 2011). In Arabidopsis, these currents are associated with the SLAC1 and ALMT12 ion channel genes, respectively, that provide major pathways for anion efflux and promote solute loss for stomatal closure (Jezek & Blatt, 2017).
Our findings highlight, in Kalancho€ e, anion current characteristics that are quantitatively similar to those of the SLAC1-like channels of Vicia, Nicotiana and Arabidopsis (Keller et al., 1989;Schroeder & Keller, 1992;Schmidt & Schroeder, 1994;Grabov et al., 1997;Chen et al., 2010;Wang & Blatt, 2011). These characteristics extend to the voltage-dependent conductance of the current, its slow (de)activation kinetics, and Cland NO 3 À permeabilities. Thus, both thermodynamic and kinetic considerations suggest that the Kalancho€ e anion current is similarly important for solute efflux and stomatal closure.
The archetypal anion channels found in Vicia, tobacco and Arabidopsis are enhanced by low millimolar extracellular Mal (Hedrich & Marten, 1993;Wang & Blatt, 2011). Thus, we anticipated that Mal might well be important as a signal in Kalancho€ e stomatal control. The critical question, therefore, was whether Mal affected the anion channels to promote their gating in Kalancho€ e, as it does in these other species. Surprisingly, unlike the anion channels found in Vicia, tobacco and Arabidopsis, the Kalancho€ e anion current was not activated, but was reversibly inhibited, by millimolar external Mal. This effect was evident throughout the light period. It was also clearly demonstrable earlier in Phase III of the light period when the activity of the channels would need to be highest to facilitate stomatal closure and when Mal release and its possible leakage from the mesophyll might be expected. In short, even if apoplastic Mal does increase during the daytime in Kalancho€ e leaves, the enhanced activity of the guard cell anion current cannot be explained on the basis of its stimulation by Mal, because the action of the acid anion is to suppress the current.

Anion channel activity follows the daytime decline in transcript abundance
By contrast, we found that the transcript abundance of both the Kalancho€ e SLAC1 and ALMT12 orthologues displayed a diurnal cycle that roughly approximated the temporal pattern for the anion current activities we were able to resolve. Notably, the Kalancho€ e SLAC1 transcript abundance rose during the night to a maximum at the start of the daylight period and declined steeply thereafter. This temporal pattern roughly matched the higher anion channel activity we recorded early in Phase III of the light period and its decline to near-zero values by Phase IV, at the end of the daylight period, when the stomata open. The Kalancho€ e ALMT12 transcript constituted a smaller proportion of transcript at all times throughout the diurnal cycle (Fig. 4). Its abundance, too, showed a pronounced maximum, albeit roughly 4 h before the start of the daylight period.
We suspect that the earlier peak in ALMT12 transcript anticipates the time, early in the daylight period, when anion channel activity is most important to promote stomatal closure. Accordingly, the 4-h gap between maximum ALMT12 transcript abundance and the start of the daylight period may be seen to reflect a lag commensurate with mRNA processing and translation, and with anion channel assembly and delivery to the plasma membrane. A similar, 4-h temporal lag is known to occur between maximal KfPPCK1 transcript abundance and KfPPCK-mediated PPC phosphorylation (Hartwell et al., 1996. In other words, a 4-h lag is a feature familiar among CAM transcriptional cycles and consistent with anticipation of the period of stomatal closure. There is an additional reason to expect ALMT12 transcript abundance might precede the time period in which the channel activity was most needed. In Arabidopsis, ALMT12 is associated with the fast-inactivating or Rapid (R-) type anion current, QUAC1 (Meyer et al., 2010;Mumm et al., 2013). A significant feature of the ALMT12 current is its voltage dependence, which ensures ALMT12 activation only when the membrane is depolarised. As a consequence, the channel protein may reside at the membrane in abundance in advance of the daylight period, but until the membrane depolarisessuch as might follow on a decline in H + -ATPase activity and enhanced SLAC1 currentthe ALMT12 channel will remain silent and will not contribute to the total membrane current at rest (Jezek & Blatt, 2017). Thus, again, ALMT12 transcript abundance anticipates a period of high anion channel activity, but it does not indicate directly the timing of this period.
It is equally significant that we observed overall the same temporal pattern for both anion channel transcripts in the wild-type and in the Kalancho€ e rNAD-ME1 and rPPDK1 RNAi mutant lines (Dever et al., 2015). As in the wild-type, this pattern in the Kalancho€ e rNAD-ME1 and rPPDK1 mutants and its parallel to the temporal decline in anion channel activities (Fig. 6) adds support to the idea that it is the transcriptional cycle of these genes that drives the cycle in anion currents evident in Kalancho€ e. These lines generate a reduced amount of Mal at dawn compared with the wild-type, and Mal turnover in the light period is greatly reduced (Dever et al., 2015). We stress here that comparisons within each genotype are informative, whereas comparisons between the three genotypes are not. Indeed, it is not surprising that the anion currents recorded, for example in the rPPDK1 mutant (Fig. 6), might be similar to those of the wild-type Kalancho€ e even if KfALMT12 abundance was altered relative to the wild-type (Fig. 4). The rNAD-ME1 and rPPDK1 mutants are likely to have other effects on cellular homeostasis that impact on gene translation and/or channel regulation. For example, one might reasonably argue that, with the reduction in Mal accumulation and turnover, there is less Mal in the cytosol as well as less released to the apoplast and available for uptake by the guard cells. Any one of these factors, in turn, could influence post-translational control of the anion current and, as a result, moderate the current even if the total population of anion channels differs from that of the wild-type Kalancho€ e. What is important is that the decline in anion current in the mutants follows qualitatively with transcript abundance and, again, argues against any role of apoplastic Mal in regulating Kalancho€ e stomatal movements.
In conclusion, we find that the anion channel current of Kalancho€ e fedtschenkoi guard cells exhibits many of the characteristics of the SLAC1 family of anion channels known in guard cells of Vicia, Nicotiana and Arabidopsis. The Kalancho€ e anion current follows a diurnal pattern of activity consistent with variations in transcript abundance for the SLAC1 orthologue. The current shows a high level of activity during the daylight period especially earlier in Phase III of the CAM cycle, consistent with a role in facilitating the closed state of the stomata during this time. Unlike the archetypical S-type channels, the Kalancho€ e current is markedly suppressed by apoplastic Mal at low millimolar concentrations. These characteristics lead us to conclude that transcript availability for translation of the channel protein and its turnover is likely to determine the activity of the anion current in Kalancho€ e guard cells and that apoplastic Mal is not a metabolic signal promoting the anion current for stomatal closure for CAM.  Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews and Tansley insights.
Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication 'as ready' via Early View -our average time to decision is <26 days. There are no page or colour charges and a PDF version will be provided for each article.
The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for