The starch-deficient plastidic PHOSPHOGLUCOMUTASE mutant of the constitutive crassulacean acid metabolism (CAM) species Kalanchoë fedtschenkoi impacts diel regulation and timing of stomatal CO2 responsiveness

Kalanchoë fedtschenkoi ‘Hamet et Perrier’ plants were propagated clonally from the same original accession obtained from the Royal Botanic Gardens, Kew, by Malcolm Wilkins (Wilkins, 1959).

The rPGM transgenic lines of K. fedtschenkoi were generated by introducing a double-stranded RNA (hairpin RNA) binary construct, designed to target the silencing of the endogenous plastidic PGM gene according to the construct design, assembly and plant transformation methods described previously for both K. fedtschenkoi and Kalanchoë laxiflora (tetraploid) (Dever et al., 2015; Hartwell et al., 2016; Boxall et al., 2017; Boxall et al., 2020). The PGM gene family in the diploid Kalanchoë genomes decoded to date [namely K. fedtschenkoi (two accessions), K. laxiflora (two accessions) and Kalanchoë gracilipes (one accession)] consists of three members. One plastidic PGM (orthologue of JGI Phytozome accession number Kaladp0008s0557.1), characterized by the presence of a chloroplast transit peptide at its N-terminus and through pairwise identity to the Arabidopsis thaliana plastidic PGM of 82.4 % amino acid level identity, was silenced in K. fedtschenkoi using RNAi in this work. In addition, the Kalanchoë genomes contain two cytosolic PGM isogenes (JGI Phytozome accession numbers Kaladp0057s0141.1 and Kaladp0059s0263.1), which were not targeted by the plastidic PGM-specific RNAi construct used here.

Briefly, a 458 bp fragment of the K. fedtschenkoi PGM gene (orthologue of JGI Phytozome accession number Kaladp0008s0557.1) was amplified with high-fidelity PCR using KOD Hot-Start DNA polymerase (Merck) and the primers KfPGM RNAiF 5ʹ- CACCTCAGAGGTCTTCTTTCACGTTCAGACT-3ʹ and KfPGM RNAiR 5ʹ- AAATTTCCAGCCTGTAGGAACCTCATA-3ʹ. The PCR product was TOPO-cloned directionally into the pENTR/D Gateway-compatible entry vector (Life Technologies). The entry clone was recombined using LR Clonase II enzyme mix (Life Technologies) into the Gateway destination vector intron-containing hairpin RNAi binary vector pK7GWIWG2 (II) (Karimi et al., 2002). Constructs were confirmed by DNA sequencing before transformation into Agrobacterium tumefaciens strain GV3101. Agrobacterium-mediated stable transformation of K. fedtschenkoi was achieved using the tissue culture-based method described by Dever et al. (2015).

Wild-type and RNAi plants were propagated by the growth of new plantlets on the leaf margin. After 8 weeks, the plantlets were transferred to plastic pots of 127 mm diameter containing John Innes No. 2 compost (J. Arthur Bower’s) and perlite (3:1). Growth conditions were set at 25 °C/19 °C (day/night) and a diurnal photosynthetic photon flux density (PPFD) of 250 µmol m⁻² s⁻¹ at plant height, with a 12 h photoperiod. All measurements were made using CAM-performing leaves (leaf pair 6) of 12-week-old plants. Epidermal tissue and ground mesophyll were harvested separately over a 24 h day/night cycle, using three biological replicates for each time point. K. fedtschenkoi is an amphistomatic species; therefore, the epidermal tissue of both abaxial and adaxial leaf surfaces were harvested together as described by Abraham et al. (2020). All samples were immediately snap frozen in liquid nitrogen and stored at −80 °C until their evaluation.

For determining the content of titratable acids, aliquots of methanol extracts were titrated against NaOH (0.05 mm) to a neutral endpoint and expressed as millimoles of H⁺ per gram fresh weight (fwt), as described by Cushman et al. (2008a). Starch was extracted from the leaf mesophyll as described previously (Haider et al., 2012), and its content in mesophyll was measured as glucose equivalents using the colorimetric phenol/sulphuric acid test described by DuBois et al. (1956). To determine starch content in guard cells, the peels were fixed (50 % v/v methanol, 10 % v/v acetic acid) and stained with Lugol’s iodine solution as described by Horrer et al. (2016). Soluble sugars were measured in both tissues using the high-pressure ion chromatography (HPIC) technique (Thermo Scientific Dionex), and the amount of sugars (in micromoles per gram fresh weight) was calculated based upon standards of glucose, fructose, sucrose and maltose. Malate content in both mesophyll and guard cell-enriched epidermis was determined by the enzymatic method developed by Hohorst (1970).

Net CO₂ uptake (in micromoles of CO₂ per square metre per second), stomatal conductance (in moles of H₂O per square metre per second) and transpiration (in millimoles of H₂O per square metre per second) were determined by gas exchange analysis. Three biological replicates of each genotype were evaluated during a 24 h day/night cycle, using the LI-6400XT Portable Photosynthesis System (LI-COR Biosciences). The ambient CO₂ concentration was set at 400 µmol CO₂ mol⁻¹ air, and the light and temperature were set to track the conditions established in the growth chamber. Integrated instantaneous water-use efficiency (WUEinst) was measured by calculating the area under the curves of net CO₂ uptake and leaf transpiration at three different times: during the night period, during the light period and over 24 h. To determine stomatal responsiveness to [CO₂] at night, stomatal conductance was determined under low (50 µmol CO₂ mol⁻¹ air) [CO₂] administered at different times of the dark period in wild type and rPGM1a. Additionally, both wild type and rPGM1a were exposed to low (50 µmol CO₂ mol⁻¹ air) and elevated (1600 µmol CO₂ mol⁻¹ air) CO₂ concentrations at mid-day (phase III), and towards the end of the photoperiod (phase IV). Finally, based on pore length and stomatal density, the maximum theoretical stomatal conductance (g_max) was estimated using the following equation according to Lawson et al. (2018):

where Dw is diffusivity of water vapour in air at 25 °C (0.0000249 m² s⁻¹), v is molar volume of air (0.0245 m³ mol⁻¹), SD is stomatal density (stomata per square metre), pamaxis maximum stomatal pore area (in square metres) calculated as an ellipse, pd is stomatal pore depth (in metres) considered to be equivalent to the width of a turgid guard cell, and π2pamaxπ is the ‘end correction’ that takes into account the influence of diffusion shells from outside the end of the stomatal pore (Lawson et al., 2018).

Real-time qPCR was used to confirm silencing of Kf-pgm1 and to investigate the transcript abundance of several genes encoding enzymes involved in starch, sugar and malate metabolism and which have been implicated in stomatal regulation. These genes included the ATP-binding cassette malate transporter (Kf-ABCB14), the sucrose synthase 1 (Kf-SUSY1) and the sugar transporter (Kf-STP1). RT-qPCR measurements were made to assess differences in abundance and day/night timing of maximal gene expression in guard cell-enriched epidermis and mesophyll cells in rPGM1a and wild type. To determine the relative quantity, the 2^−ΔΔCt method (Livak and Schmittgen, 2001) was performed, normalizing the data against the cDNA calibrator and the K. fedtschenkoi thioesterase/thiol ester dehydrase-isomerase (Kf-TEDI) superfamily protein (Phytozome Kaladp0068s0118.1), as reported by Boxall et al. (2017). In addition, no-template controls (NTC) were added to confirm the absence of contamination.

To determine significant differences, an ANOVA and LSD (least significant difference) post hoc test were performed. The data obtained from gas exchange experiments were analysed with the non-parametric Mann–Whitney U-test, based upon its non-normality and the non-homogeneity determined with the Shapiro–Wilk test and Levene’s test, respectively. All the statistical analyses were done using SPSS software (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0; IBM, Armonk, NY, USA).

Our first aim was to isolate a starch-deficient RNAi line of K. fedtschenkoi that could be used to examine how a lack of starch impacted on gas exchange characteristics and stomatal responses to [CO₂]. Six independent RNAi lines for which the single copy of the plastidic PGM gene had been targeted for silencing with a hairpin RNA transgene were tested for leaf starch content and titratable acids measured at the start (dawn) and end (dusk) of the photoperiod. Lines rPGM1a, rPGM1c and rPGM2c were found to have the lowest leaf starch contents at both dawn and dusk, and the overnight accumulation of titratable acids in these lines was also reduced compared with wild type (Fig. 1). Consequently, rPGM1a, rPGM1c and rPGM2c were evaluated further in order to select the best line for this study. Accumulation of soluble sugars, which is a phenotype consistent with the starch deficiency resulting from the loss of plastidic PGM activity in the Arabidopsis thaliana mutant (Caspar et al., 1985; Hanson and McHale, 1988), was measured in these three RNAi lines and compared with wild type. Of the three RNAi lines analysed, line rPGM1a showed the most significant accumulation of soluble sugars measured at the end of the photoperiod (Fig. 2). Both rPGM1a and rPGM1c also showed a complete lack of starch in the guard cells, whereas rPGM2c showed variations in the starch content of guard cells (Fig. 3). Lines rPGM1a and rPGM1c were subsequently compared with wild type in terms of diel patterns of leaf gas exchange. Both lines showed lower rates of nocturnal net CO₂ uptake compared with wild type, whilst daytime net CO₂ uptake was substantially higher in the RNAi lines compared with wild type, where the absence of net CO₂ uptake for most of the light period indicated complete stomatal closure (Fig. 4). Of these two RNAi lines, rPGM1a showed the largest reduction in nocturnal net CO₂ uptake, combined with elevated light period net CO₂ uptake relative to wild type, and was thus taken forward for further analyses.

In rPGM1a, the lower net CO₂ uptake during the night and the unaffected stomatal conductance compared with wild type (Fig. 5A) suggest that the lower CAM activity of rPGM1a was attributable to a deficiency in PEP carboxylation, probably because of a lack of substrates and not owing to reduced stomatal opening. In addition, during the day period, the rPGM1a line was not able to close the stomata like the wild type (Fig. 5B). The integrated CO₂ uptake and leaf transpiration over a 24 h period indicated that instantaneous WUEinst was significantly higher (P ≤ 0.05) in wild type compared with rPGM1a during the light period but not at night (Table 1). Additionally, anatomical measurements of stomatal size and density (Supplementary data Fig. S1) showed that the maximum theoretical stomatal conductance (g_max) was significantly higher (P ≤ 0.05) in rPGM1a compared with wild type (451.383 ± 9.384 and 355.823 ± 15.429 mmol m⁻² s⁻¹, respectively), implying that starch deficiency conferred an anatomical predisposition towards potentially higher stomatal conductance.

In the wild type, starch content increased in the mesophyll over the course of the photoperiod and was steadily and almost completely depleted during the night (Fig. 6). Starch content was negligible in mesophyll of rPGM1a, confirming the starch-deficient phenotype of this line, as indicated by the initial screen described above (Figs 1A and 6A). The guard cell starch content in the wild type, represented as starch granule area (in square micrometres), showed no significant change in content over the course of the light period. There was significant (P ≤ 0.05) net depletion of guard cell starch during the first 1 h of darkness, but thereafter the guard cell starch increased gradually over the course of the night (Fig. 6B).

Diel changes in malate content in both genotypes represented the typical CAM turnover of this organic acid, with diurnal degradation owing to decarboxylation and nocturnal accumulation as a product of PEP carboxylation by PEPC (Fig. 7). Diel turnover of malate in the mesophyll was significantly higher (3.5-fold higher) in wild type compared with rPGM1a (Fig. 7A). Nocturnal accumulation and daytime mobilization of malic acid were also evident in the guard cell-enriched epidermis (Fig. 7B). However, on a fresh weight basis, diel turnover of malate in wild type was more than eight times higher in the mesophyll cells compared with the epidermal peels. In the epidermis of rPGM1a, malate accumulated over the first half of the night, in line with changes in malate content in the wild-type epidermis. However, over the second half of the night, malate content continued to increase in the wild-type epidermis, but no further net increase in malate content was observed in the epidermis of rPGM1a. The most marked differences in diel changes in epidermal malate content were observed during the photoperiod (Fig. 7B). In wild-type epidermis, malate content declined over the 12 h light period, whereas in rPGM1a, malate content increased over the first half of the photoperiod, peaking in the middle of the day, before decreasing over the remaining 6 h of the photoperiod.

Diel soluble sugar content was significantly different (P ≤ 0.05) between wild type and rPGM1a. Glucose and fructose contents in both mesophyll and epidermis increased during the day in rPGM1a, reaching maximum values at the end of the photoperiod that were ~7-fold higher than those in wild type (Fig. 8A–D). Sucrose was present in lower amounts compared with glucose and fructose, and although sucrose contents were comparable between wild type and rPGM1a, there were marked differences between genotypes and between mesophyll and epidermis in terms of diel turnover. In wild type, sucrose was depleted during the day and then accumulated over the subsequent night in both mesophyll and epidermis. In rPGM1a, sucrose accumulated in the mesophyll over the first few hours of the photoperiod and was subsequently depleted over the remainder of the photoperiod and over the night, whereas sucrose in the epidermis increased steadily over the photoperiod and was subsequently degraded during the following night (Fig. 8E, F).

Stoichiometric analysis was performed to determine the relationship between nocturnal degradation of carbohydrates and malate accumulation during phase I in both the mesophyll and guard cell-enriched epidermis and to compare the starchless phenotype of rPGM1a and the lack of significant starch degradation in wild-type guard cells. Starch, soluble sugars and malate were quantified at dawn and dusk (Table 2). The breakdown of mesophyll starch is measured as glucose equivalents, where each mole of glucose potentially produces two moles of PEP via glycolysis, and each mole of PEP is carboxylated to produce one mole of malate. PEP excess or deficit was calculated as the difference between the amount of PEP required for nocturnal malate accumulation minus the PEP produced from glycolytic breakdown of all the different carbohydrate sources. The calculations suggested that nocturnal malate formation in wild-type mesophyll was derived from breakdown of starch and hexoses (PEP excess of 3.036 µmol). Regarding rPGM1a, the higher accumulation of soluble sugars in the mesophyll was probably responsible for nocturnal PEP formation by glycolysis, as a compensation for the absence of starch. However, the lower malate turnover overall in rPGM1a together with the deficit in PEP (20.230 µmoles) indicate the importance of starch turnover for CAM activity. Given that starch in guard cell-enriched epidermis was measured as starch granule area, it was not possible to calculate the relationship with PEP turnover in this tissue.

In order to establish whether starch deficiency affects stomatal responsiveness to [CO₂] at night, we compared stomatal responses to low (50 µmol CO₂ mol⁻¹ air) [CO₂] administered at different times of the dark period in wild type and rPGM1a (Fig. 9). The data indicate that although starch deficiency did not, in general, curtail stomatal opening in response to low [CO₂], the responsiveness of stomatal conductance (i.e. the speed and magnitude of opening) to low [CO₂] differed between genotypes at different points in the night. Thus, stomatal responsiveness to low [CO₂] increased as the night progressed in wild type, but the opposite was seen in rPGM1a, where stomatal responsiveness to low [CO₂] declined over the course of the night (Fig. 9).

In terms of stomatal behaviour during the photoperiod, reduced nocturnal malate accumulation in rPGM1a might consequently result in a lower internal C_i during malate decarboxylation, such that the C_i was insufficient to mediate stomatal closure. Thus, we predicted that exposing leaves of the starch-deficient line to high [CO₂] during the day would result in complete closure of stomata. Both wild-type and rPGM1a plants were exposed to low (50 µmol CO₂ mol⁻¹ air) and elevated (1600 µmol CO₂ mol⁻¹ air) CO₂ concentrations at mid-day (phase III), and towards the end of the photoperiod (phase IV), in order to determine whether starch deficiency impacted the daytime stomatal response to [CO₂]. Data revealed that in phase III, when wild-type stomata were closed, changes in the external atmospheric CO₂ concentration did not have any effect on stomatal conductance (Fig. 10A). In rPGM1a, exposure to low [CO₂] in phase III resulted in a sustained increase in stomatal conductance, followed by stomatal closure within 11 min when plants were exposed to elevated [CO₂]. However, stomata of rPGM1a did not remain closed under elevated CO₂ and partly re-opened after 16 min (Fig. 10A). Later in the photoperiod during phase IV, both wild type and rPGM1a opened stomata in response to low [CO₂], with the speed and magnitude of opening being more pronounced in the starch-deficient plants. Exposure to high [CO₂] resulted in stomatal closure in both genotypes. However, in rPGM1a, an inability to maintain complete stomatal closure under elevated [CO₂] was again observed, as stomata started to re-open after 14 min under elevated [CO₂] (Fig. 10B).

As an additional way of testing how stomata of wild type and rPGM1a responded to altered daytime C_i, plants were exposed to CO₂-free air during the night to curtail nocturnal malate synthesis and then released into ambient [CO₂] (i.e. 400 µmol CO₂ mol⁻¹ air) at the start of the photoperiod (Fig. 11). The gas exchange profiles obtained were compared with those measured in control conditions of ambient [CO₂] given during the night and photoperiod. Data showed that after a night in CO₂-free air, both genotypes fixed CO₂ during the first hours of the day, and this was followed by a reduction in net CO₂ uptake, indicating stomatal closure. Importantly, the data indicate that curtailing nocturnal malate accumulation delayed, but did not prevent, daytime closure of stomata in wild type (Fig. 11A). In the starch-deficient rPGM1a, curtailing nocturnal malate accumulation by a night in CO₂-free air had a negligible impact on subsequent daytime gas exchange (Fig. 11B). Together, these experiments suggest that factors other than changes in C_i regulate diurnal stomatal behaviour in CAM plants and that starch metabolism has important implications for daytime stomatal regulation.

A significant reduction in transcript abundance of Kf-PGM1 in both mesophyll and guard cell-enriched epidermis of rPGM1a sampled at dawn and dusk confirmed the RNAi-directed downregulation of this gene. In wild type, Kf-PGM1transcript levels were significantly higher in the mesophyll compared with the epidermis (Supplementary data Fig. S2A). Regarding malate active transport, the ATP-binding cassette transporter ABCB14, for which the encoded protein is located in the guard cell plasma membrane, mediates malate uptake from the apoplast in Arabidopsis thaliana (Lee et al., 2008). In K. fedtschenkoi, transcript abundance of Kf-ABCB14 was highest in epidermal peels of the wild type at the beginning of the night, whereas in rPGM1a this transcript was most abundant at the beginning of the day (Supplementary data Fig. S2B). Given the higher diurnal accumulation of sugars in rPGM1a, the transcript abundance of the sugar transporter protein 1 (STP1) was also evaluated in both genotypes. In Arabidopsis thaliana, this transporter, located in the plasma membrane of guard cells, is involved in the import of monosaccharides to sustain osmoregulation on stomatal opening (Stadler et al., 2003). The highest transcript abundance of Kf-STP1 was noted at dawn in the guard cell-enriched epidermis of rPGM1a, where the level of this transcript was significantly higher than that in wild type (Supplementary data Fig. S2C). If activity follows transcript abundance, the data imply enhanced import of sugars to guard cells of the starch-deficient plants at the start of the photoperiod. Additionally, given the sucrolytic activity of sucrose synthases (SUSY) in the conversion of sucrose into fructose and UDP-glucose (UDPG), and the importance of sucrose turnover on stomatal metabolism, the regulation of Kf-SUSY1 and Kf-SUSY3 was also determined in both lines. No significant difference in Kf-SUSY3 expression was found between genotypes (data not shown). The peak in Kf-SUSY1 transcript was restricted to the beginning of the photoperiod, with differences between genotypes. In wild type, transcript abundance of Kf-SUSY1 was higher in the mesophyll (Supplementary data Fig. S2D). In contrast, in rPGM1a this transcript was highly abundant in the guard cell-enriched epidermis, suggesting that guard cells of rPGM1a have a unique metabolic machinery compared with mesophyll and/or that regulation of sucrose metabolism differs between mesophyll and epidermis in the absence of significant plastidic PGM activity.

Starch degradation plays a central role in regulating the CAM cycle by providing the sugars that are metabolized through glycolysis to PEP, the substrate for PEPC in the mesophyll to assimilate atmospheric CO₂ in the dark period (Ceusters et al., 2021). The aim of the present study was to assess whether starch degradation could also have a key role in CAM by being important for the correct regulation of nocturnal stomatal opening via the provisioning of energy and osmolytes to increase guard cell turgor at night. Previous work using the facultative CAM species M. crystallinum indicated that mutants deficient in phosphoglucomutase (PGM) were CAM and starch deficient (Cushman et al., 2008a; Haider et al., 2012), but the impact on stomatal behaviour was not assessed. To address this question, we used RNAi to generate PGM-deficient transgenic lines of the model constitutive CAM species, K. fedtschenkoi. We screened six independent RNAi lines for CAM activity and starch content and found two independent lines (rPGM1a and rPGM1c) that lacked starch in the leaf ground mesophyll and guard cells and that showed reduced nocturnal net CO₂ uptake and malate accumulation relative to the wild type. Line rPGM1a had the most extreme phenotype in terms of starch deficiency and reduced CAM activity, but nocturnal stomatal conductance was comparable to that of wild type. These data support the view that reduced CAM in lines with reduced plastidic PGM activity, which leads to starch-deficient plants, is attributable to a limitation in the nocturnal supply of sugars for PEP synthesis, rather than a consequence of diffusional limitation to atmospheric CO₂ uptake into the leaves for nocturnal carboxylation (Cushman et al., 2008a; Borland et al., 2016). Thus, the production of ATP and osmolytes required for nocturnal stomatal opening in K. fedtschenkoi is not reliant on starch degradation in the guard cells or mesophyll cells.

Malate is generally accepted as the predominant anion during stomatal opening and closing in C₃ photosynthesis species (Fernie and Martinoia, 2009) and has also been proposed as the main osmolyte responsible for nocturnal stomatal opening in CAM plants (Lee, 2010). Our data support this hypothesis, with nocturnal accumulation of malate measured in guard cell-enriched epidermis of both wild-type and starch-deficient K. fedtschenkoi. Furthermore, the accumulation of malate was severalfold higher (on a molar basis) compared with that of soluble sugars. The source of this guard cell malate is still a matter of debate. Previous proteomics analysis of guard cell-enriched epidermis of K. fedtschenkoi indicated the presence of phosphorylated (active) PEPC at night (Abraham et al., 2020), indicating that malate could potentially be produced directly in the stomatal complex using PEP generated from the degradation of guard cell starch and/or sugars. Clearly, malate accumulation in the epidermal tissue of rPGM1a did not require starch degradation, and there was limited net breakdown of guard cell starch over the course of the night in wild-type K. fedtschenkoi, although malate content increased steadily in the guard cell-enriched epidermis throughout the night. Glycolytic processing of soluble sugars in the epidermis to provide sufficient PEP for night-time synthesis of malate via PEPC appeared to be stoichiometrically feasible in the starch-deficient rPGM1a but not in wild type (Table 2). Such findings suggest that in wild-type K. fedtschenkoi, malate was transported from underlying mesophyll cells into the guard cells, rather than synthesized in situ.

In Arabidopsis, the ATP-binding cassette transporter ABCB14, which is located in the guard cell plasma membrane, mediates malate uptake from the apoplast and also coordinates the response of stomata to changes in internal CO₂ concentration (Lee et al., 2008). Transcript abundance of the K. fedtschenkoi ABCB14 orthologue was significantly enriched in wild-type epidermis at the start of the night, which, if transcript abundance coordinates transporter activity, would be consistent with the hypothesis that nocturnal import of malate to guard cells promotes stomatal opening at night. Accumulation of malate in the mesophyll and its transport to guard cells via ABCB14 could be an important hub that connects CAM photosynthesis in the mesophyll with stomatal behaviour over the diel cycle. A contrasting scenario was revealed for the starch-deficient rPGM1a plants, in which Kf-ABCB14 transcripts were barely detected at night, but instead increased in abundance at the start of the photoperiod. This was coincident with the sustained net accumulation of malate in guard cell-enriched epidermis of the mutant line over the first half of the light period, at a time when malate was declining in wild-type epidermis (Fig. 7). The earlier suggestion that malate could be synthesized directly within the guard cells of the rPGM1a plants using soluble sugars to provide a substrate for nocturnal anapleurotic CO₂ fixation, presents the hypothesis that night-time import of malate to guard cells in the starch-deficient line is quantitatively less important than in wild type. Ultimately, genetic manipulation of Kf-ABCB14 in K. fedtschenkoi using RNAi and/or CRISPR-Cas9-mediated mutagenesis (Liu et al., 2019) will be necessary to establish whether this gene plays a similar role to the Arabidopsis orthologue in terms of mediating malate import to the K. fedtschenkoi guard cells.

A key trigger for nocturnal opening of stomata in CAM is believed to be driven by reduced C_i when PEPC activity increases at dusk (von Caemmerer and Griffiths, 2009; Males and Griffiths, 2017). Stored products in the guard cells and signals from the mesophyll are believed to influence stomatal response to changes in C_i in K. fedtschenkoi (Santos et al., 2021). We hypothesized that starch deficiency and the resultant changes in sugar and malate homeostasis in both mesophyll and guard cells would impact on nocturnal responses to [CO₂] in this CAM species. Our data indicated that starch degradation was not required to facilitate stomatal opening in response to low external [CO₂] administered at night, and the magnitude and speed of this response was not reliant on starch degradation per se. However, although stomatal responsiveness (i.e. the speed and magnitude of opening) to low external [CO₂] increased over the course of the dark period in wild type, in the starch-deficient plants stomata became less responsive to low external [CO₂] as the night progressed. Circadian gating of stomatal responsiveness to CO₂ has been discussed previously within the context of CAM (Borland et al., 2014; Hartwell, 2006). Stomatal responses to [CO₂] in K. fedtschenkoi appear to be influenced by the presence of the mesophyll, but are not mediated solely through changes in C_i (although these clearly play an important role; Santos et al., 2021). In CAM, it appears that some other unknown diffusible mesophyll signal coordinates stomatal behaviour with mesophyll demands for CO₂ (von Caemmerer and Griffiths, 2009). The present study with starch-deficient RNAi lines of K. fedtschenkoi that displayed reduced nocturnal CO₂ fixation supports the view that genetic manipulations that disrupt the diel cycle of the accumulation and turnover of malate and other primary metabolites have a profound influence on the temporal control and optimization of the CAM-associated rhythms of CO₂ fixation and stomatal conductance (Dever et al., 2015; Boxall et al., 2017; Boxall et al., 2020). Further work is required to determine whether and how accumulation of malate in the mesophyll and its transport to guard cells connects CAM photosynthesis in the mesophyll with stomatal behaviour over the diel cycle, whilst maintaining CO₂ responsiveness over duration of the night.

The substantial elevation in C_i that accompanies the decarboxylation of malate in the photoperiod is believed to be a key driver for stomatal closure in the light (Males and Griffiths, 2017). Curtailed daytime closure of stomata was noted in two starch-deficient lines (rPGM1a and rPGM1c), and it was hypothesized that this was a consequence of limited malate decarboxylation in these lines. To test this hypothesis, wild type and rPGM1a were exposed to CO₂-free air overnight to abolish nocturnal malate accumulation, with the prediction that both genotypes would increase stomatal conductance the following day. The data showed that although wild-type plants increased stomatal conductance for the first few hours of the photoperiod after a night in CO₂-free air, they still showed complete stomatal closure during phase III. The rPGM1a plants showed little change in daytime stomatal conductance after a night in CO₂-free air. Thus, as surmised for stomatal behaviour at night, C_i is not the only factor influencing the daytime behaviour of CAM stomata (von Caemmerer and Griffiths, 2009). It is noteworthy in this context that Lefoulon et al. (2020) reported that guard cell anion channel activity in K. fedtschenkoi CAM leaves assayed in the light period tracked the temporal pattern of transcript abundance cycling of the genes encoding the channel proteins. Thus, the data presented here, taken together with those reported previously by von Caemmerer and Griffiths (2009) and Lefoulon et al. (2020), are consistent with the proposal that the circadian clock regulates the transcript oscillations of guard cell anion channel genes and through this mediates stomatal closure in the light period in Kalanchoë leaves regardless of metabolic or genetic interventions that reduce nocturnal CO₂ fixation and vacuolar malate accumulation during the preceding dark period.

Starch deficiency was found to have profound impacts on the responsiveness of stomata to low and elevated [CO₂] administered during the day. The wild-type stomata showed no (phase III) or limited (phase IV) stomatal opening in response to low [CO₂], whereas the starch-deficient plants showed very marked stomatal opening, particularly in phase IV, when stomatal conductance values were >3.5 times higher than that measured for wild type. Moreover, although stomata in rPGM1a closed in response to elevated external [CO₂] (i.e. 1600 µmol CO₂ mol⁻¹ air), stomata did not remain closed under this treatment. Together, the data indicate that starch biosynthesis is required for sustained daytime stomatal closure in K. fedtschenkoi. Substantial deposits of starch were found in the guard cells of K. fedtschenkoi and, as reported previously, this starch was not mobilized during the day as is the case in Arabidopsis (Abraham et al., 2020). Such findings support the hypothesis of a crucial role for starch synthesis/accumulation in daytime stomatal closure in CAM.

A comparison of stomatal responses to [CO₂] in mutants of Arabidopsis lacking starch in both mesophyll and guard cells with mutants lacking starch only in the mesophyll indicated that starch synthesis specifically in the guard cells was crucial for mediating [CO₂]-induced stomatal closure (Azoulay-Shemer et al., 2016). Guard cell starch synthesis is thought to play an essential role in CO₂-induced stomatal closure by acting as a sink for C skeletons coming from malate degradation via gluconeogenesis during guard cell osmotic adjustment (Azoulay-Shemer et al., 2016). Starch might also act as a sink for sugars produced via the Calvin cycle, and the disruption of starch synthesis increases sucrose content, which, together with its degradation products, fructose and glucose, is involved in stomatal osmoregulation (Lawson et al., 2014; Azoulay-Shemer et al., 2015). In the present study, the inability of rPGM1a to convert osmolytes such as malate and sucrose into insoluble starch granules in the guard cells in addition to the mesophyll could explain the incomplete daytime stomatal closure in these starch-deficient plants. The significantly higher soluble sugar content noted in guard cell-enriched epidermal peels and mesophyll from rPGM1a support the hypothesis that soluble sugars increased guard cell turgor pressure and curtailed complete closure of stomata during the day in the starch-deficient plants. Future testing of this hypothesis will require a more targeted genetic approach to manipulate starch metabolism in the CAM guard cell independently of that in the mesophyll and thereby establish the functional significance of the diel rescheduling of starch turnover in CAM guard cells, as reported here and previously (Abraham et al., 2020).

The engineering of the inverted stomatal rhythm of CAM into non-CAM species offers the potential to increase WUE of important crops for human consumption and thus the productivity of arid environments. At night, CAM stomata are thought to open in response to reduced C_i caused by the consumption of CO₂ by PEPC and the accumulation of malate (Males and Griffiths, 2017). According to Niechayev et al. (2019), in order to engineer the CAM stomatal rhythm successfully into non CAM, it is necessary that the host species have guard cells fully responsive to changes in [CO₂] that allows the nocturnal stomatal opening and the diurnal closure. A first approach in the modification of Arabidopsis stomatal conductance has been made by Lim et al. (2019), who overexpressed genes implicated in both the carboxylation and decarboxylation modules of the facultative CAM species M. crystallinum. Transgenic lines overexpressing the carboxylation enzymes phosphoenolpyruvate carboxylase 1 (PEPC1), NAD⁺ malate dehydrogenase (NAD⁺ MDH), NADP malate dehydrogenase (NADP-MDH) and phosphoenolpyruvate carboxylase kinase (PPCK1), respectively, showed increased stomatal conductance. In contrast, lines overexpressing the decarboxylation module enzymes, such as NAD-malic enzyme 1 and 2 (NAD-ME1 and NAD-ME2) and NADP-malic enzyme (NADP-ME), showed decreased stomatal conductance and showed depletion of the content of organic acids.

Apart from the carboxylation and decarboxylation modules, engineering starch turnover in C₃ guard cells with an increased content during the day followed by degradation at beginning of the night, as observed here and previously (Abraham et al., 2020), will confirm whether the reprogramming of starch metabolism in guard cells is crucial for the inverted stomatal rhythm in CAM plants. In the same way, based on the different starch degradation pathways in the mesophyll that CAM and C₃ follow (Ceusters et al., 2021), a rerouting to the phosphorolytic pathway in C₃ species will contribute to understanding how this pathway affects stomatal behaviour as a possible provider of substrates in the synthesis of nocturnal amino acids that can act as osmolytes in the guard cells. To achieve this, it is also necessary to elucidate whether differences in both guard cells and mesophyll starch turnover in CAM are under circadian regulation and whether this altered clock control has to be engineered into C₃ plants to allow the inverted stomatal rhythm.

In conclusion, we have shown that the nocturnal opening of stomata in the constitutive CAM species K. fedtschenkoi is not reliant upon nocturnal starch degradation, as evidenced by comparable stomatal conductance and opening in response to low [CO₂] administered at night in wild-type and starch-deficient plants of K. fedtschenkoi. Starch synthesis was required for stomatal closure in the light, potentially acting as a sink for soluble sugars and/or malate, thereby promoting a reduction in guard cell turgor, as illustrated in our proposed model (Supplementary data Fig. S3). This hypothesis was supported by marked accumulation of soluble sugars in both the mesophyll and guard cell-enriched epidermis of rPGM1a during the day. Data also implicate the importance of malate accumulation in the mesophyll and its transport into guard cells, which appears necessary for driving nocturnal stomatal opening and for connecting CAM photosynthesis in the mesophyll with stomatal behaviour. Together, the findings reported here indicate that the nocturnal degradation and daytime synthesis of starch in this CAM species regulates sugar and malate homeostasis between mesophyll and guard cells and within the guard cells per se. These processes, in turn, have profound implications for the diel control of stomatal conductance and responsiveness to [CO₂] across the diel CAM cycle.