Transgenic Perturbation of the Decarboxylation Phase of Crassulacean Acid Metabolism Alters Physiology and Metabolism But Has Only a Small Effect on Growth  

Abstract

Mitochondrial NAD-malic enzyme (ME) and/or cytosolic/plastidic NADP-ME combined with the cytosolic/plastidic pyruvate orthophosphate dikinase (PPDK) catalyze two key steps during light-period malate decarboxylation that underpin secondary CO₂ fixation in some Crassulacean acid metabolism (CAM) species. We report the generation and phenotypic characterization of transgenic RNA interference lines of the obligate CAM species Kalanchoë fedtschenkoi with reduced activities of NAD-ME or PPDK. Transgenic line rNAD-ME1 had 8%, and rPPDK1 had 5% of the wild-type level of activity, and showed dramatic changes in the light/dark cycle of CAM CO₂ fixation. In well-watered conditions, these lines fixed all of their CO₂ in the light; they thus performed C₃ photosynthesis. The alternative malate decarboxylase, NADP-ME, did not appear to compensate for the reduction in NAD-ME, suggesting that NAD-ME was the key decarboxylase for CAM. The activity of other CAM enzymes was reduced as a consequence of knocking out either NAD-ME or PPDK activity, particularly phosphoenolpyruvate carboxylase (PPC) and PPDK in rNAD-ME1. Furthermore, the circadian clock-controlled phosphorylation of PPC in the dark was reduced in both lines, especially in rNAD-ME1. This had the consequence that circadian rhythms of PPC phosphorylation, PPC kinase transcript levels and activity, and the classic circadian rhythm of CAM CO₂ fixation were lost, or dampened toward arrhythmia, under constant light and temperature conditions. Surprisingly, oscillations in the transcript abundance of core circadian clock genes also became arrhythmic in the rNAD-ME1 line, suggesting that perturbing CAM in K. fedtschenkoi feeds back to perturb the central circadian clock.

Crassulacean acid metabolism (CAM) is a metabolic adaptation of photosynthetic metabolism that can dramatically reduce water loss and thus favors efficient carbon capture and growth in semi-arid and arid environments (Borland et al., 2009, 2014). CAM has evolved independently many times and is found in diverse groups of dicots, monocots, and ferns (Smith and Winter, 1996; Silvera et al., 2010, 2014; Christin et al., 2014).

CAM plants open their stomata at night and undertake primary atmospheric CO₂ fixation in their photosynthetic leaf mesophyll cells via phosphoenolpyruvate carboxylase (PPC) and malate dehydrogenase (MDH), generating malate, which is stored as malic acid in the vacuole. At dawn, malic acid is released from the vacuole and is decarboxylated by one of three decarboxylases: either NAD-malic enzyme (ME) and/or NADP-ME (ME-type plants), or phosphoenolpyruvate carboxykinase (PCK; PCK-type plants; Dittrich, 1976; Holtum et al., 2005)_. Liberated CO₂ concentrates behind closed stomata and is refixed by Rubisco in the Calvin cycle; the by-product pyruvate is recycled to stored starch via pyruvate orthophosphate dikinase (PPDK), gluconeogenesis, and the starch synthesis pathway (Borland et al., 2009). CAM plants can be categorized as either constitutive/obligate or facultative/inducible. Constitutive CAM species undergo a one-way developmental progression to CAM and maintain CAM even under well-watered conditions (Winter et al., 2008). Facultative CAM species can switch from C₃ to CAM in response to the environment, with the flexibility to return to C₃ when favorable conditions return (Winter and Holtum, 2014)

Both primary and secondary CO₂ fixation are localized within individual leaf mesophyll cells, and thus require strict temporal regulation of enzyme activities, membrane transporters, and their regulatory proteins to prevent futile cycling. It has long been proposed that the optimized temporal control of CAM is achieved via regulation by the circadian clock (Wilkins, 1992; Wyka et al., 2004; Boxall et al., 2005; Hartwell, 2005, 2006). CAM plants exhibit endogenous circadian rhythms of CO₂ fixation in constant light and temperature (LL) conditions due to the periodic activity of the primary carboxylase phosphoenolpyruvate carboxylase (PPC) and secondary carboxylase Rubisco (Wilkins, 1992). The activity of PPC is modulated via dark-phase protein phosphorylation of an N-terminal Ser that renders the enzyme less sensitive to feedback inhibition by its end product malate, and more sensitive to positive effectors such as phosphoenolpyruvate (PEP), Glc-6-P, and triose phosphate (Nimmo et al., 1986, 1987). Dark phosphorylation is catalyzed by a dedicated, circadian clock-controlled protein kinase called phosphoenolpyruvate carboxylase kinase (PPCK; Nimmo et al., 1987; Carter et al., 1991; Hartwell et al., 1999, 2002; Nimmo, 2000; Boxall et al., 2005). The nocturnal increase in PPCK activity results from clock-controlled transcription and translation of the kinase (Hartwell et al., 1999).

Although understanding of the regulation of primary nocturnal CO₂ fixation by PPC is relatively well developed, malate decarboxylation and the subsequent light-period steps of CAM, and their regulation, are less well studied, even in physiologically and biochemically well-studied constitutive/obligate CAM species such as Kalanchoë fedtschenkoi (Cook et al., 1995). High activities of both mitochondrial NAD-ME and chloroplastic/cytosolic NADP-ME have been detected in leaf extracts of various ME-type CAM species (Dittrich, 1976; Spalding et al., 1979; Christopher and Holtum, 1996), but the relative importance of the two forms of ME to malate decarboxylation in the light period during CAM is poorly understood.

ME-type CAM plants are widely reported to possess high activities of PPDK, which converts pyruvate to PEP during the light period (Sugiyama and Laetsch, 1975; Ramachandra Reddy and Das, 1982). In K. fedtschenkoi, approximately two-thirds of PPDK activity was located in the cytosol, with one-third in the chloroplast (Kondo et al., 2000), but little information exists as to the importance of the enzyme for efficient CAM or its regulation during the CAM cycle. In C₃ and C₄ species, PPDK activity is regulated via phosphorylation/dephosphorylation by a unique and specific dual-activity kinase/phosphatase known as PPDK-regulatory protein. Phospho-PPDK is inactive, whereas dephospho-PPDK is active (Chastain et al., 2002; Chastain and Chollet, 2003). To our knowledge, PPDK phosphoregulation during the daily CAM cycle has not been reported previously.

In this article, we investigated the importance of NAD-ME and PPDK to the light-period malate decarboxylation phase of CAM using transgene silencing (a hairpin RNA/RNA interference [RNAi] approach) in the obligate CAM model species K. fedtschenkoi. Independent stable RNAi lines lacking either enzyme performed mostly C₃ photosynthesis, but still displayed some features of CAM in terms of gas exchange. We also demonstrated that a reduction in NAD-ME activity had major effects on light/dark (LD) and circadian clock-controlled phosphorylation and inactivation/activation of PPDK and PPC. Lastly, arrhythmicity of gene regulation in the core molecular circadian oscillator is demonstrated as a consequence of reduced NAD-ME activity.

RESULTS

Initial Screening of NAD-ME and PPDK  RNAi Lines of K. fedtschenkoi

K. fedtschenkoi does not produce viable seed due to a failure of the seed to achieve dehydration tolerance in the final stages of its development (Garcês et al., 2007). For this reason, all data presented here are for primary transformants, which were propagated clonally via leaf plantlets and/or stem cuttings. Primary transformants were grown to the six-leaf stage and screened for changes in leaf cell sap pH and starch accumulation using leaf discs sampled from leaf pair 6 (LP6) at the end of the dark and light periods (Cushman et al., 2008). Wild-type K. fedtschenkoi displayed the expected diurnal fluctuation in acidity; leaves were acidic at dawn, due to nocturnal CO₂ fixation and vacuolar accumulation of malic acid, and were more neutral at dusk due to decarboxylation of malate throughout the light period. Leaf disc screening identified both NAD-ME and PPDK  RNAi lines because they had low leaf pH as a result of malic acid retention at the end of the light period. Similarly, both the NAD-ME and PPDK  RNAi lines also accumulated less starch than the wild type by the end of the light period (screening data not shown).

Lines that differed from the wild type in terms of dusk/dawn starch content and/or cell sap pH were screened for down-regulation of the transcript abundance of the endogenous target genes with reverse transcriptase (RT)-PCR using primers outside the region targeted with the hairpin RNA construct. The transcript abundance of the target gene was reduced for three NAD-ME  RNAi lines (rNAD-ME1, rNAD-ME2, and rNAD-ME3) and one PPDK  RNAi line (rPPDK1; Fig. 1A). Line rNAD-ME1 also showed a reduction in the level of the α-subunit transcript (Fig. 1A).

Transgenic lines with down-regulation of the target gene were selected for detailed characterization of the impacts of reduced NAD-ME or PPDK on phenotypes associated with CAM.

Biochemical Characterization

Immunoblotting using antisera specific to the α- and β-subunits of NAD-ME confirmed a large reduction in the level of the β-subunit protein in all three rNAD-ME lines (Fig. 1B). The α-subunit of NAD-ME was also reduced in these lines, but to varying degrees. rNAD-ME2, which had the smallest reduction in NAD-ME activity (approximately 15% of wild-type activity; Fig. 1C), had the highest level of the α-NAD-ME, whereas rNAD-ME1, which had the greatest reduction in NAD-ME activity (8% of wild-type activity; Fig. 1C), had the lowest level of the α-subunit (Fig. 1B).

Immunoblots using an anti-maize (Zea mays) PPDK antibody demonstrated a complete absence of detectable PPDK protein in leaf tissue from line rPPDK1, whereas the wild type possessed a clear band of the correct M  _r in both the light and dark (Fig. 1B).

Several important enzymes of the CAM pathway were assayed in the transgenic lines. All three rNAD-ME lines were confirmed to have reduced activities of NAD-ME (Fig. 1C). Line rPPDK1 possessed only 5% of the wild-type PPDK activity (Fig. 1C).

These dramatic reductions in the activity of either NAD-ME (reduced by 92% in line rNAD-ME1) or PPDK (reduced by 95% in line rPPDK1) relative to the wild type also led to reduced activity of other CAM enzymes. PPDK activity, as well as being reduced in the rPPDK1 line, was also reduced to a similar low activity level in the rNAD-ME lines (6% of wild-type activity; Fig. 1C). However, unlike the situation in the rPPDK1 line, which lacked detectable PPDK transcript or protein, the PPDK gene transcript and protein were present in the rNAD-ME lines as determined with RT-PCR (Fig. 1A) and immunoblotting. Similarly, NAD-ME activity was reduced in the rPPDK1 line, but was still present at 65% of the wild-type activity (Fig. 1C). PPC activity was reduced in all of the RNAi lines, with the most marked reduction, to approximately 20% of wild-type PPC activity, in rPPDK1 (Fig. 1C). The reduction in PPC activity in the rNAD-ME lines was less marked. Line rNAD-ME1, which had the greatest reduction in NAD-ME activity, possessed approximately 61% of the wild-type PPC activity (Fig. 1C).

The activity of the second malate decarboxylase, NADP-ME, was assayed for all lines using both LP3 and LP6 (Fig. 1C). Wild-type K. fedschenkoi leaves possessed higher NADP-ME activity in LP3 relative to LP6 for all plants tested, despite significantly higher levels of CAM, as judged by nocturnal CO₂ fixation and malic acid accumulation, in LP6 relative to LP3, at least for the wild type. This difference between LP3 and LP6 was particularly marked for the rPPDK1 line. In both rNAD-ME1 and rNAD-ME2, as well as rPPDK1 leaves, activity was slightly greater than the wild type for LP3; for LP6, NADP-ME activity was only greater for the rNAD-ME1 leaves.

Growth Analysis

Dry weight vegetative yields of rNAD-ME1, rPPDK1, and the wild type were measured for plants grown in a greenhouse using 16-h-light/8-h-dark cycles and well-watered conditions (Fig. 2A). Student’s t tests showed that the above-ground dry weight was significantly lower in the rNAD-ME1 and rPPDK1 plants compared with the wild type, with t = 2.15 (12 degrees of freedom), P = 0.02658, and t = 5.68 (12 degrees of freedom), P = 0.00007, respectively. The average percentage of dry weight reduction of rNAD-ME1 was 12.5% (n = 7) and the average percentage of dry weight reduction of rPPDK1 was 34% (n = 7) compared with the wild type.

Malate and Starch Levels

Malate and starch represent the key reciprocating pools of metabolites that vary over the LD cycle in starch-accumulating CAM species such as K. fedtschenkoi. To characterize any major perturbations in CAM-associated metabolism, both metabolites were quantified in the transgenic lines, using samples collected 1 h before dawn and dusk (Fig. 2, B and C).

The wild type, rNAD-ME1, and rPPDK1 accumulated malate in the dark. Wild-type plants accumulated the most malate by dawn (77 µmol g fresh weight⁻¹, se 4.3; Fig. 2B). Line rNAD-ME1 accumulated 83% of this, whereas rPPDK1 only accumulated 46% of the wild-type level of malate by dawn (Fig. 2B). At the end of the light period, 83% of the dark-accumulated malate had been decarboxylated in the wild type, such that the dawn/dusk Ɗ malate was 64 µmol g fresh weight⁻¹. By comparison, only 33% of the dawn level of malate was decarboxylated in rNAD-ME1, and 40% in rPPDK1, yielding Ɗ malate values of 21 and 14 µmol g fresh weight⁻¹, respectively. At the end of the light period, wild-type CAM leaves (LP6) accumulated over twice the amount of starch found in rNAD-ME1 leaves, and three times more than rPPDK1 leaves (Fig. 2C). All plants had very low levels of starch at the end of the dark period (Fig. 2C).

Diurnal Regulation of the Transcript Abundance of CAM Genes

In addition to the expected reduction in the steady-state transcript level for each respective gene targeted with the RNAi transgene (e.g. low PPDK transcript level in rPPDK1; Fig. 1A), both the rNAD-ME1 and rPPDK1 lines showed remarkable similarity in the reduction of the transcript abundance of other CAM-associated genes when assayed over the 12-h-light/12-h-dark cycle. In particular, both lines displayed a general reduction in the transcript levels of KfPPC1_CAM, KfNAD-ME_b1, KfNAD_ME_a1 (in line rNAD-ME1), KfPPDK, and GLUCAN WATER DIKINASE (KfGWD; Supplemental Fig. S1, A–D). In both rPPDK1 and rNAD-ME1, the transcript abundance of the GLUCOSE 6-PHOSPHATE: PHOSPHATE TRANSLOCATOR2 (KfGPT2) was reduced in the light relative to the wild type, in particular for the rPPDK1 line (Supplemental Fig. S1E).

LD oscillations in the transcript abundance of PPCK, which encodes the circadian clock-controlled protein kinase responsible for dark-period phosphorylation of PPC, showed no significant difference from the wild type in either RNAi line (Supplemental Fig. S1F). Both lines had a characteristic diurnal cycle of PPCK transcript levels, reaching a peak in the dark period consistent with previous findings (Hartwell et al., 1999). Similarly, the core circadian oscillator genes, TIMING OF CHLOROPHYLL A/B BINDING PROTEIN1 (KfTOC1) and CIRCADIAN CLOCK ASSOCIATED1 (KfCCA1_LHY2), plus a control clock output gene, LIGHT HARVESTING CHLOROPHYLL A/B BINDING PROTEIN (KfCAB1), oscillated with a similar phase and amplitude in both the wild type and RNAi lines under 12-h-light/12-h-dark conditions (Supplemental Fig. S1, G–I).

PPC Phosphorylation

Dark-period phosphorylation of PPC in K. fedtschenkoi, driven by circadian oscillations in the transcript abundance and activity of PPCK (Hartwell et al., 1999), has been shown to cause an increase in the K  _i of PPC for feedback inhibition by malate, making the enzyme less sensitive to malate inhibition in the dark period (Nimmo et al., 1984). The phosphorylation state of PPC was investigated using immunoblotting with an anti-phospho-PPC antibody (Feria et al., 2008). PPC protein levels were also measured on replicate immunoblots using an anti-PPC antibody to ensure even loading of protein in each lane (Fig. 3A).

PPC activity in the RNAi lines was lower than that in the wild type (Fig. 1C). To correct for the reduced level of PPC activity in rNAD-ME1, the amount of protein extract loaded onto the SDS-PAGE gels used for immunoblotting was increased, so that the amount of PPC protein in the wild type and rNAD-ME1 were comparable (Fig. 3A, right). The degree of dark phosphorylation of PPC in rNAD-ME1 was clearly reduced compared with the wild type (Fig. 3A, left). Phosphorylated PPC was only detected at a low level at 10 h into the dark period in rNAD-ME1, whereas PPC phosphorylation in the wild type was detected at high levels at both 6 and 10 h into the dark period (Fig. 3A).

The rPPDK1 line showed an 80% reduction in PPC activity (Fig. 1C). However, because of the limited volume of the well on the SDS-PAGE gel, it was only possible to double the amount of protein loaded onto the gel for the comparison between the wild type and rPPDK1 (Fig. 3A, right). For rPPDK1 leaves, a normal LD pattern of PPC phosphorylation was observed (Fig. 3A, left). Taking into account the reduced loading of PPC onto the rPPDK1 SDS-PAGE gel, we estimated that the phosphorylation level of PPC in rPPDK1 was very similar to PPC phosphorylation in the wild type (Fig. 3), although we cannot rule out that phosphorylation commenced earlier in the wild type.

Phosphorylation of PPDK

Enzyme assays indicated that the activity of PPDK was as low in the rNAD-ME lines as it was in the rPPDK1 line, at around 6% of wild-type activity (Fig. 1C). However, while the rPPDK1 line had very low transcript and protein abundance for PPDK (Fig. 1, A and B), the rNAD-ME1 line showed good levels of KfPPDK transcript abundance (although slightly reduced compared with the wild type; Fig. 1A; Supplemental Fig. S1C). The PPDK protein was also readily detected using immunoblotting in leaves of rNAD-ME1 sampled over a 12-h-light/12-h-dark time course (Fig. 3B). The transcript abundance of the PPDK-REGULATORY PROTEIN1 (KfPPDK_RP1) varied over the 12-h-light/12-h-dark time course and peaked during the dark in wild-type leaves (Supplemental Fig. S1J). However, in the rNAD-ME1 and rPPDK1 lines, transcript oscillations had a lower amplitude, with the KfPPDK-RP1 transcript present throughout the light period as well as during the dark (Supplemental Fig. S1J). Immunoblots of the wild type and rNAD-ME1 sampled over a 12-h-light/12-h-dark cycle challenged with antibodies to phosphorylated PPDK confirmed this (Fig. 3B), showing phosphorylation of PPDK at the end of light and throughout the dark for the wild type but constitutive PPDK phosphorylation throughout the 24-h cycle in the rNAD-ME1 line.

Gas Exchange Characteristics in LD Cycles

Using a multicuvette, gas-switching, infrared gas analyzer system, we investigated the 24-h pattern of CO₂ exchange over 7 d in 12-h-light/12-h-dark cycles for well-watered, whole young plants (LP8 stage) of the wild type, rNAD-ME1, and rPPDK1. Wild-type plants displayed the four classic phases of CAM (Osmond, 1978) with marked net fixation of atmospheric CO₂ occurring during the dark period (phase I of CAM: stomata open, primary atmospheric CO₂ fixation via PPC; Fig. 4A). Both rNAD-ME1 and rPPDK1 displayed a markedly different 24-h CO₂ exchange profile compared with the wild type (Fig. 4A). Both RNAi lines had no significant phase I nocturnal CO₂ fixation, and instead displayed a net loss of CO₂ for much of the dark period. However, the RNAi lines did display a decrease in the loss of respired CO₂ during the middle of the dark period (Fig. 4A) and significant CO₂ fixation in the light, presumably directly via Rubisco. Instead of the three readily distinguished light period phases of CAM observed in the wild type (phases II–IV; Osmond, 1978), the RNAi lines showed a merging of phases III and IV, with a slight dip in the middle of the 12-h-light period. The wild type, rNAD-ME1, and rPPDK1 all showed a pronounced phase II peak of CO₂ fixation just after dawn. This peak was prolonged by 2 h in the rPPDK1 line, but phase II had a similar duration to the wild type in rNAD-ME1. It is worthwhile to note that when full CAM leaves (LP6) were detached from 10-week-old mature plants and measured in isolation with their petiole in distilled water using the gas exchange system, strong CAM was observed in wild-type LP6 characterized by dark CO₂ fixation (peak 4 µmol m⁻² s⁻¹) and stomatal closure throughout the light period (Supplemental Fig. S2). Total net uptake of CO₂ was calculated for the dark and light periods from the area under the curves in Figure 4A. Wild-type CO₂ fixation was greater than both rNAD-ME1 and rPPDK1 in both the light and dark, with the RNAi lines failing to achieve any net CO₂ fixation in the dark (Supplemental Fig. S3).

Detached LP6 from mature rNAD-ME1 and rPPDK1 plants showed a large reduction in dark CO₂ fixation relative to the wild type, but did achieve net fixation of CO₂ (0.5–1 µmol m⁻² s⁻¹) for the latter approximately 6 h of the 8-h-dark period (Supplemental Fig. S2). In the light, detached leaves of both rPPDK1 and rNAD-ME1 displayed a prolonged phase II-like bout of CO₂ fixation lasting several hours into the light period; this was particularly pronounced in the rNAD-ME1 line (Supplemental Fig. S2).

CO₂ fixation under both LD and LL cycles was compared for detached LP6 from the three rNAD-ME lines using plants entrained under 16-h-light/8-h-dark cycles (Supplemental Fig. S4). rNAD-ME1 performed the least nocturnal CO₂ uptake under LD, and its endogenous rhythm in LL dampened to arrhythmia very rapidly. Line rNAD-ME2 showed the highest level of dark CO₂ fixation under LD, whereas rNAD-ME3 was intermediate between the other two lines. Under LL free-running conditions, rNAD-ME2 and rNAD-ME3 maintained weak, low-amplitude rhythms of CO₂ fixation that were phase delayed relative to the robust, high-amplitude oscillations observed for the wild type (Supplemental Fig. S4).

Phenotypic Characterization of RNAi Lines in Well-Watered and Drought-Stressed Conditions

In light of the high water-use efficiency of CAM (Borland et al., 2014), it was important to understand the physiological responses of the RNAi lines to drought stress. CO₂ exchange was measured for drought-stressed whole plants. Wild-type drought-stressed plants displayed strong CAM, with pronounced dark CO₂ fixation (phase I), but only phase III, with no phase II and IV, in the light (Fig. 4B). rNAD-ME1 and rPPDK1 achieved detectable net CO₂ fixation only during phase II after lights on, but this was greatly reduced compared with well-watered plants (Fig. 4B). Phase IV CO₂ fixation at the end of the light period, which is characteristically observed in well-watered wild-type plants, was absent in both the RNAi lines and the wild type.

In order to dissect the contribution of older leaves (full CAM in the wild type) and younger C₃ leaves to the overall diurnal CO₂ fixation patterns observed in Figure 4A for rNAD-ME1 and rPPDK1, entire young plants possessing eight leaf pairs were placed in the gas exchange cuvettes. Gas exchange of the whole plants was recorded for several days before the youngest leaves (LP2–LP4) were removed from two plants, leaving two control plants with all of their leaves attached. Gas exchange data were area corrected according to the area of leaf tissue in the cuvette at both stages of the experiment (pre- and post-defoliation; Fig. 5).

The data revealed that for both rNAD-ME1 and rPPDK1 plants, the CO₂ uptake of CAM leaves alone (LP5–LP8) was reduced during the light compared with plants with all of their leaves. This revealed that the younger leaves (LP2–LP4) contributed the majority of the C₃-type fixation of atmospheric CO₂ in the light period detected for the whole plants (Fig. 5).

CAM plants achieve high water-use efficiency, allowing them to grow productively in arid and semi-arid environments (Borland et al., 2009). We subjected both the wild-type and the RNAi lines to 39 d of drought in order to investigate the advantages of a fully functional CAM pathway in wild-type plants. Multiple clones of all lines were grown from plantlets under identical conditions to the LP6 stage. One-half of the plants were maintained well watered as controls, and the other one-half were subjected to drought stress. Both sets of plants were maintained in the same growth cabinet such that the only environmental variable was water supply.

All of the control, well-watered plants grew well. There was very little observable phenotypic difference between the wild type and the RNAi lines, consistent with the small but significant differences in dry weight yield observed in Figure 2A.

After 39 d of drought, one of the most obvious differences between the wild type and the RNAi lines was the color and positioning of the leaves. Wild-type leaves had developed a distinct purple-tinge due to the presence of anthocyanins (in keeping with the common name of K. fedtschenkoi [Lavender Scallops]), and were closed up (in-rolled shape) around the stem (Supplemental Fig. S6). Both RNAi lines maintained green leaves throughout the drought treatment with no visible purple tinge developing. Furthermore, rNAD-ME1 held its leaves in an open position rather than closed up to the stem. Leaf color was quantified by extracting associated anthocyanin pigments using methanol:1% (v/v) HCl. Extracts were measured at 530 nm (Neff and Chory, 1998). After drought stress, wild-type leaves contained 23-fold more anthocyanin than the well-watered plants (Fig. 6A). By contrast, anthocyanin content increased only 4-fold and 1.7-fold in response to drought stress in rNAD-ME1 and rPPDK1, respectively. Under drought-stress conditions, the wild type produced more than 5 times the amount of anthocyanin accumulated by rNAD-ME1, and 30 times the amount in rPPDK1. rPPDK1 also had lower levels of anthocyanin than the wild type in well-watered conditions (Fig. 6A).

Gas Exchange Characteristics under Circadian Free-Running Conditions

K. fedtschenkoi has been studied for many years as an example of a CAM plant that exhibits a robust circadian rhythm of CO₂ fixation in both LL, normal air, and constant dark, CO₂-free air conditions (reviewed by Wilkins, 1992; Hartwell, 2006). The exchange of CO₂ in LL conditions was measured for whole young plants (eight-leaf stage; Fig. 6, B and C). The wild-type plants displayed the classic LL CAM gas exchange circadian rhythm in both well-watered and drought-stressed conditions (Fig. 6, B and C). In well-watered conditions, both rNAD-ME1 and rPPDK1 displayed a weak, low-amplitude rhythm of CO₂ exchange that was phase delayed relative to the wild-type rhythm (Fig. 6B). However, drought-stressed plants of both RNAi lines displayed an initial peak of CO₂ fixation, which dampened rapidly to arrhythmia after the first 24 to 48 h in LL conditions (Fig. 6C). Although well-watered rNAD-ME1 maintained a high level of CO₂ fixation (approximately 1.5 µmol m⁻² s⁻¹) in LL that oscillated with a low-amplitude rhythm (Fig. 6B), the drought-stressed plants collapsed toward very low levels of arrhythmic net CO₂ fixation after 3 d in LL (approximately 0.5 µmol m⁻² s⁻¹ for rPPDK1 and approximately 0.1 µmol m⁻² s⁻¹ for rNAD-ME1; Fig. 6C). Thus, although drought-stressed rNAD-ME1 and rPPDK1 performed only a tiny phase II burst of net CO₂ fixation under LD conditions (Fig. 4B), they achieved positive net CO₂ fixation, with low-amplitude oscillations, throughout the LL circadian free-run experiment (Fig. 6C).

Phosphorylation of PPC under Circadian Free-Running Conditions

The circadian rhythm of CAM CO₂ exchange in LL conditions (Fig. 6, B and C) for the wild type has been shown to correlate tightly with circadian clock control of the phosphorylation state of PPC and its K  _i for malate (Hartwell et al., 1996, 1999; Hartwell, 2005). CO₂ exchange in LL conditions for rNAD-ME1 and rPPDK1 showed, at best, low-amplitude rhythmicity in well-watered plants (Fig. 6B). We therefore investigated the circadian regulation of the phosphorylation state of PPC and the transcript levels of PPCK under LL conditions. In wild-type plants, PPC was phosphorylated during the initial dark period and then displayed a robust circadian rhythm of phosphorylation throughout 3 d under LL (Fig. 7, A and B). The period of the rhythmic phosphorylation of PPC matched the well-documented, temperature-dependent, short-period rhythm of gas exchange observed in K. fedtschenkoi (Anderson and Wilkins, 1989). In the RNAi lines, PPC was phosphorylated in the last dark period before entering LL conditions, although at a reduced level in the rNAD-ME1 line consistent with Figure 3A (Fig. 7, A and B). Once under LL conditions, the PPC phosphorylation state ceased to oscillate in both the rNAD-ME1 and rPPDK1 lines (Fig. 7, A and B). This loss of the circadian rhythm of PPC phosphorylation correlated with arrhythmia (rNAD-ME1) or dampened oscillations (rPPDK1) of the rhythm of KfPPCK transcript levels in LL (Fig. 7, A and B).

The loss of circadian rhythmicity of gas exchange, PPC phosphorylation, and KfPPCK1 transcript levels (Figs. 6, B and C, and 7, A and B) led us to question whether the central molecular oscillator continued to oscillate robustly under LL conditions in the RNAi lines. We monitored the oscillation of the transcript abundance of the core circadian clock gene KfTOC1 (Strayer et al., 2000), as a readout of rhythmicity in the core oscillator (Fig. 7C). Although KfTOC1 transcript levels oscillated with a robust rhythm in both the wild type and rPPDK1 line under LL conditions, in rNAD-ME1, KfTOC1 transcript levels displayed an initial in-phase peak at 18 to 22 h in LL followed by arrhythmia for the remainder of the LL period (Fig. 7C).

DISCUSSION

Testing CAM Gene Function Using Transgenic K. fedtschenkoi

A CAM- and starch-deficient plastidic phosphoglucomutase mutant of Mesembryanthemum crystallinum, which failed to perform CAM due to a lack of a sufficient carbohydrate pool to supply nocturnal PEP, has been reported (Cushman et al., 2008). However, it has not been possible previously to silence individual steps in the CAM pathway in a targeted manner due to the lack of a transformable CAM species. This has undoubtedly hampered efforts to understand the molecular-genetic basis for CAM and its circadian clock control. We developed a simple and efficient stable transformation system for K. fedtschenkoi, and silenced two key enzymes that function in the light-period malate decarboxylation phase of CAM (namely mitochondrial NAD-ME and cytosolic/chloroplastic PPDK). Our goal was to understand the relative importance of these enzymes to the malate decarboxylation phase of CAM in the light period, and to investigate the impact of perturbed CAM on growth, stress tolerance, physiology, metabolism, and circadian clock control within a constitutive CAM species.

Physiological Consequences of the Loss of NAD-ME or PPDK in a Constitutive CAM Plant

Under well-watered conditions, net nocturnal fixation of atmospheric CO₂ (phase I CAM) was abolished in the absence of NAD-ME or PPDK (Fig. 4A). However, a reduction in the escape of respired CO₂ was detected in the middle of the dark period (Fig. 4) and this may underlie the reduced but detectable level of nocturnal malate accumulation in both rNAD-ME1 and rPPDK1 (Fig. 2B). By contrast, rNAD-ME1 and rPPDK1 performed substantial fixation of atmospheric CO₂ in the light period, although this did show evidence of phases II and III of CAM (Fig. 4A). However, in the young whole plants used for the gas exchange analysis shown in Figure 4A, the wild type fixed more CO₂ over each 24-h period than either RNAi line (Supplemental Fig. S3). This reduced net fixation of CO₂ in both transgenic lines relative to the wild type correlated with a reduction in above-ground dry biomass of both lines (Fig. 2A). Although the RNAi lines fixed CO₂ throughout the light period, they did display a dip in CO₂ fixation in the middle of the light period. This may be due to a residual level of malate decarboxylation generating a small internal supply of CO₂, or could be due to an underlying circadian clock-controlled midday dip in stomatal aperture, as suggested previously for Kalanchoë daigremontiana (von Caemmerer and Griffiths, 2009). The higher concentrations of malate in the light in the leaves of the RNAi lines may contribute to stomatal regulation due to the proposed role of malate as a mesophyll to guard cell signal involved in the regulation of stomatal aperture (Araújo et al., 2011). Furthermore, malate is an important counter-anion in guard cells during stomatal opening, and it may be that the associated metabolism of malate in guard cells involves NAD-ME and/or PPDK. Thus, we cannot rule out knock-on effects of these 35S promoter RNAi constructs on guard cell metabolism of malate, in addition to the major effects of these RNAi constructs on CAM in the mesophyll cells of the leaf.

Whole young plants were used to monitor CO₂ uptake. In the wild type, these include both young C₃ and older CAM leaves. The contribution of mature leaves to the daily carbon gain of the plant was investigated by removing the young C₃ leaves (LP2–LP4; Fig. 5). Removal of C₃ leaves resulted in reduced CO₂ fixation for much of the light period (Fig. 5). Thus, younger leaves (LP2–LP4) contributed a large proportion of the light period CO₂ fixation observed in both transgenic lines. We therefore conclude that C₃ photosynthesis in young leaves made the greatest contribution to the daily carbon gain supporting the growth of the rNAD-ME1 and rPPDK1 lines.

Due to the importance of CAM in enhancing plant water-use efficiency, we investigated the influence of drought stress on the daily gas exchange of the RNAi lines. Gas exchange measurements using young whole plants that had been drought-stressed highlighted the remarkable plasticity of CAM in K. fedtschenkoi. The drought-stressed wild-type plants displayed significant phase I nocturnal CO₂ uptake (peak 3 µmol m⁻² s⁻¹), and closed their stomata throughout the light period (phase III; Fig. 4B). In rNAD-ME1 and rPPDK1, a small but measureable period of phase I CO₂ fixation was recorded in the dark period (peak approximately 0.2 µmol m⁻² s⁻¹). After a very minor phase II at dawn, light-period CO₂ fixation was not detected in either RNAi line. Thus, the drought-stressed RNAi lines achieved only minimal net CO₂ fixation over the 24-h cycle.

Drought caused wild-type leaves to develop a purple color, whereas leaves of both RNAi lines remained green. These visual observations were confirmed quantitatively by anthocyanin measurements (Fig. 6A). Anthocyanin biosynthesis requires substrate from the shikimate pathway, which uses PEP as a substrate (Hibberd and Quick, 2002; Brown et al., 2010). Thus, the failure to decarboxylate malate in rNAD-ME, or convert pyruvate to PEP in rPPDK1 may starve the shikimate pathway of substrate, preventing anthocyanin biosynthesis in response to drought. It is clear from these findings that global metabolomics profiling of these RNAi lines will likely reveal fascinating insights into the interconnected metabolic pathways that operate alongside CAM, and share or compete for a number of metabolic intermediates.

Biochemical and Molecular Changes in the RNAi Lines

The three independent rNAD-ME lines all showed large reductions in the steady-state transcript level of the targeted β-NAD-ME gene (KfNAD_ME1_b1), indicating strong silencing of the endogenous transcript (Fig. 1A). Furthermore, the β-subunit of NAD-ME was not detected on immunoblots (Fig. 1B), and NAD-ME activity was reduced to 8% to 15% of wild-type levels (Fig. 1C). In addition, two rNAD-ME lines also displayed substantial reductions in the α-subunit (Fig. 1B). These results revealed that reduction of the β-subunit may have caused down-regulation of the level of the α-subunit, perhaps via some form of metabolic feedback, or that the region of the gene targeted by the hairpin RNA construct was sufficiently highly conserved in the α-subunit gene for the hairpin RNA to mediate degradation of both transcripts.

The latter possibility of cross silencing of both transcripts is certainly plausible, because the region targeted by the hairpin RNA construct shares 69.6% nucleotide sequence identity between the major α-NAD-ME and β-NAD-ME genes expressed in CAM leaves (Supplemental Fig. S5). However, the α-subunit was less strongly down-regulated in line rNAD-ME2, even though the β-subunit was not detected on immunoblots (Fig. 1B). The expression level of the hairpin RNAi transgene in this line may have been sufficient to drive the degradation of all β-NAD-ME transcripts, for which it has a perfect sequence match, but only a small percentage of α-NAD-ME transcripts.

The higher NAD-ME activity of line rNAD-ME2 suggests that the α-subunit detected on the immunoblot may have been able to form an active α-α homodimer. This is consistent with the findings of Tronconi et al. (2008), who reported that an Arabidopsis (Arabidopsis thaliana) transfer DNA insertion mutant lacking β-NAD-ME-retained residual NAD-ME activity that was presumed to be due to the α-α homodimer. This was supported by CO₂ uptake data that showed rNAD-ME2 performed the most nocturnal CO₂ uptake (peak 1 µmol m⁻² s⁻¹ compared with the wild-type peak >3 µmol m⁻² s⁻¹), whereas rNAD-ME1 achieved the lowest level of dark CO₂ fixation (peaking at <0.5 µmol m⁻² s⁻¹; Supplemental Fig. S4).

The transcript abundance and activity of several key CAM enzymes, in particular PPC and PPDK, were reduced in the RNAi lines (Fig. 1C; Supplemental Fig. S1, A and C). This suggests that down-regulation of light-period malate decarboxylation by NAD-ME, or pyruvate recycling by PPDK, led to feedback regulation of several other key steps in the CAM pathway. This metabolic feedback was manifested both at the level of gene transcript and protein abundance, and enzyme activity. The identity of the metabolite(s) that mediate this feedback control and the way(s) in which they are sensed and interpreted will be important areas for future research, and will likely provide valuable insights into the regulatory signals that maintain high levels of key CAM enzymes such as PPC and PPDK in the mesophyll cells of the CAM leaf.

PPDK activity was reduced to less than 10% of its wild-type level in the rNAD-ME lines (Fig. 1C), whereas changes in the level of PPDK transcripts and protein were much smaller (Fig. 3B; Supplemental Fig. S1C). PPDK was phosphorylated throughout the LD cycle in rNAD-ME1, suggesting that the low detected activity was due to inactivation of PPDK via phosphorylation (Fig. 3B). It is possible that constitutive phosphorylation of PPDK in rNAD-ME1 was due to the elevation of KfPPDK-RP1 transcript levels in the light at a time when the wild-type KfPPDK-RP1 transcript level reached its daily trough (Supplemental Fig. S1J). However, detailed understanding of the feedback regulation of PPDK in rNAD-ME1 will require a much more complete knowledge of the localization and regulation of PPDK during CAM, and the global metabolic consequences of reducing NAD-ME activity. In particular, it will be important to understand the mechanism by which PPDK-RP is regulated to mediate the dusk-phased phosphorylation and inactivation of PPDK observed in the wild type (Fig. 3B).

This inactivation of PPDK in the dark during CAM in the wild type (Fig. 3B) is consistent with PPDK phosphorylation playing a pivotal role in the LD temporal coordination of CAM, and correlates with the LD dephosphorylation/phosphorylation of PPDK during C₄ photosynthesis (Burnell and Chastain, 2006; Chen et al., 2014). However, to our knowledge, this is the first report of LD phosphoregulation of PPDK in a CAM species. This adds a second key layer of LD regulation to the CAM pathway, in addition to the well-studied, circadian clock-controlled nocturnal phosphorylation of PPC by PPCK (Hartwell et al., 1999).

It is important to note that not all of the genes investigated showed changes in transcript abundance in the RNAi lines relative to the wild type. In particular, the transcript oscillations of core circadian clock and clock-controlled output genes including KfTOC1, KfCCA1, KfCAB1, and KfPPCK1 were not perturbed relative to the wild type in 12-h-light/12-h-dark conditions (Supplemental Fig. S1, F–I). This suggests that the down-regulation of the transcript levels of several known CAM-associated genes is a specific effect that could be the result of metabolic feedback regulation of gene expression and/or transcript stability.

As several ME-type CAM species are known to have high activities of both NAD-ME and NADP-ME, including several Kalanchoë spp., it was possible that the rNAD-ME lines would rescue themselves by increasing NADP-ME activity. In wild-type plants, NADP-ME activity was approximately 6-fold lower than NAD-ME activity for LP6 (Fig. 1C). Kondo et al. (2000) reported NADP-ME activity at approximately one-half of the activity of NAD-ME for K. fedtschenkoi. By contrast, Dittrich (1976) reported equal activity for both NAD-ME and NADP-ME in K. fedtschenkoi. These earlier articles did not always distinguish the age of the leaves used for their activity measurements, and thus it is possible that leaves of different ages were pooled together. Consistent with our findings, Winter and Smith (1996) mentioned unpublished work that showed a decline in NADP-ME activity in K. daigremontiana as leaves mature to CAM. Although NADP-ME activity and transcript abundance was increased slightly in two of the rNAD-ME lines and in the rPPDK1 line (Fig. 1C; Supplemental Fig. S1K), this change was clearly insufficient to rescue substantial malate decarboxylation in the light. Our data suggest that, relative to NAD-ME, NADP-ME plays at most a minor role in malate decarboxylation during full CAM in mature leaves of K. fedtschenkoi.

Impacts on LD and Circadian Control of CAM and the Central Circadian Clock

Circadian clock control of the phosphorylation state of PPC is a key mechanism for optimizing the temporal coordination of CAM (Carter et al., 1991; Borland et al., 1999; Hartwell et al., 1999). The phosphorylation state of PPC can be measured using anti-phospho-PPC antibodies to probe immunoblots (Pacquit et al., 1995; Ueno et al., 2000). Using this approach, dark phosphorylation of PPC in mature leaves of rNAD-ME1 was reduced relative to the wild type (Fig. 3A). LD oscillations in the transcript abundance of PPCK were not altered in either RNAi line (Supplemental Fig. S1F). This suggests that the dark phosphorylation of PPC may have been inhibited by the high malate concentrations present in rNAD-ME1 in the early part of the dark period (Wang and Chollet, 1993). Furthermore, Borland et al. (1999) found that in K. daigremontiana leaves, the circadian control of dark phosphorylation of PPC by PPCK could be overridden by manipulating the level of dark CO₂ fixation and the resulting level of malate.

In the rPPDK1 line, dark phosphorylation of PPC reached similar levels to the wild type, but showed a clear phase delay (Fig. 3A). rPPDK1 had approximately double the wild-type malate level at dusk, and accumulated only approximately 50% of the wild-type malate level by dawn (Fig. 2B). Thus, high malate in the first six hours of the dark period may also inhibit PPCK and the dark-period phosphorylation of PPC in rPPDK1 (Wang and Chollet, 1993). The high level of phosphorylation of PPC achieved in the second half of the dark period in rPPDK1 did not allow the plants to achieve net fixation of CO₂ in the dark (Fig. 4A). It may thus represent a futile effort to drive dark CO₂ fixation in the rPPDK1 line.

In LL, the free-running rhythm of phosphorylation of PPC in the wild type (Fig. 7, A and B) correlated with gene transcript abundance rhythms of KfPPCK1 and KfTOC1 (Fig. 7). This wild-type rhythm of phosphorylation had a period shorter than 24 h (Fig. 7) that correlated well with the short-period of the CO₂ fixation rhythm that is well characterized for K. fedtschenkoi under LL conditions (Fig. 6; Supplemental Fig. S4; Hartwell, 2006). The peak of PPC phosphorylation was phased to the end of the subjective light period by the third day in LL conditions, whereas the first phospho-PPC peak under LL was phased to the end of the subjective dark period, consistent with the dark-period peak in PPC phosphorylation under LD cycles (Fig. 7, A and B). The correlation between KfPPCK and KfTOC1 transcript rhythms in LL and the phosphorylation state of PPC suggest close coupling between the central circadian oscillator (CCA1-TOC1 core loop), the circadian control of CAM  PPCK, and the PPC phosphorylation state (Fig. 7, A and B).

rPPDK1 lacked a rhythm in the phosphorylation state of PPC in LL, despite detectable rhythms of KfPPCK1 and KfTOC1 transcript levels over the 72 h of the LL time course (Fig. 7B).

rNAD-ME1 showed reduced PPC phosphorylation at the dark/light transition, consistent with reduced phosphorylation of PPC in rNAD-ME1 shown in Figure 3A, and a complete absence of PPC phosphorylation in LL (Fig. 7A). This lack of phosphorylation correlated with arrhythmic KfPPCK1 and KfTOC1 transcript levels after around 30 h in LL. This effect on KfTOC1 rhythmicity indicates that failure to decarboxylate malate in the light could be feeding back to perturb the core circadian oscillator in rNAD-ME1. The molecular basis for the circadian control of CAM has been the subject of extensive debate. Early theories favored an underlying oscillator based on malate oscillations and rhythmic changes in the properties of the tonoplast membrane (Wilkins, 1992; Lüttge, 2002). However, more recently, a number of authors have argued for a molecular-genetic oscillator based around a CCA1/LHY-TOC1 core gene loop, as has been shown to underpin circadian rhythmicity in Arabidopsis (Wyka et al., 2004; Boxall et al., 2005; Hartwell, 2005; von Caemmerer and Griffiths, 2009). Our data (Fig. 7) suggest that a major perturbation in core CAM metabolism in rNAD-ME1 can block rhythmicity of the central clock gene KfTOC1. This reveals that the maintenance of robust rhythmicity in the CAM leaf may require cross talk between the metabolism of CAM and the CCA1/LHY-TOC1 molecular oscillator. It may be that the coupling between CAM and the core molecular clock shares some of its signaling mechanisms with the recently demonstrated ability of sugars to entrain the Arabidopsis circadian clock (Haydon et al., 2013).

CONCLUSION

K. fedtschenkoi is established here as a constitutive CAM species that is readily amenable to stable genetic transformation using an Agrobacterium tumefaciens-mediated tissue culture-based method. Hairpin RNA binary constructs were used to reduce the activity of two key CAM enzymes, NAD-ME and PPDK, both required for the decarboxylation of malate in the light period to release CO₂ for secondary refixation via Rubisco. Reducing either enzyme to less than 10% of wild-type activity in independent transgenic lines led to a failure to perform nocturnal primary net atmospheric CO₂ fixation via PPC, a key characteristic of CAM. Furthermore, the silenced lines performed all of their net carbon gain in the light, mainly via CO₂ fixation by young leaves performing C₃-like photosynthesis. Despite the dramatic reduction in their ability to perform CAM, the silenced lines grew well and displayed only small but significant reductions in dry biomass yield under optimal growth conditions. Even under drought conditions, where CAM might be expected to be a key survival strategy due to its intrinsically high water-use efficiency, the RNAi lines performed well in terms of survival, possibly due to an osmoprotectant role of their constitutively high [malate]. However, CO₂ fixation at the whole plant level was curtailed in the RNAi lines under drought, whereas the drought-stressed wild type continued to fix considerable amounts of CO₂ in the dark. Under all but the most extreme drought-stress conditions, we find that CAM is dispensable for efficient growth in the constitutive CAM species K. fedtschenkoi. One of our most surprising findings was that free-running rhythms of the core circadian clock gene KfTOC1 collapsed to arrhythmia in the rNAD-ME1  RNAi line, suggesting some form of metabolic feedback regulation between CAM-associated metabolism, possibly malate, and the core circadian clock. The rNAD-ME1 line also had low PPDK activity due to constitutively high phosphorylation of PPDK, which may, at least in part, have been mediated by high levels of KfPPDK-RP1 transcripts throughout the LD cycle. These transgenic lines are valuable tools for further elucidating the role of CAM in high water-use efficiency and drought tolerance, and for understanding the interaction between organic acid metabolism and both the circadian clock and the reverse stomatal control required for CAM.

MATERIALS AND METHODS

Plant Materials

Unless otherwise stated, Kalanchoë fedtschenkoi ‘Hamet et Perrier’ plants were propagated clonally from stem cuttings and leaf margin adventitious plantlets using the same clonal stock originally obtained from the Royal Botanic Gardens, Kew, by Malcolm Wilkins (Wilkins, 1959). Plants were grown in 1.63-L square pots (13 × 13 × 13 cm) using a mix of John Innes no. 3 compost containing one-third Perlite plus Osmocote slow-release fertilizer applied at the manufacturer’s recommended level. Plants were initially grown in greenhouse conditions with supplementary lighting (16 h of light, approximately 250 µmol m⁻² s⁻¹; 23°C/8 of h dark at 18°C). Prior to all experiments, plants were pre-entrained for 7 d in 12 h of light (450 µmol m⁻² s⁻¹), 25°C, 60% humidity, and 12 h of dark, 15°C, 70% humidity in a Snijders Microclima MC-1000 climate-controlled plant growth cabinet (Snijders Scientific). Leaf pairs were numbered according to age, where LP1 was the youngest pair of leaves flanking the shoot apical meristem and leaf age increased down the stem. Mature plants had around 10 to 12 leaf pairs and had been grown under greenhouse conditions for at least 8 to 10 weeks prior to entrainment in the Snijders growth cabinets. Plants grew more rapidly through summer months due to increased quantity and quality of natural sunlight hours in the greenhouse; thus, the plant developmental stage was always judged by the number of leaf pairs, rather than relying on the chronological age of the plants. Plants took longer to reach 10 to 12 leaf pairs in the winter months when growth was more reliant on the supplementary greenhouse lighting.

Time-Course Experiments

For LD time-course experiments, opposite pairs of LP6 were collected every 4 h over a 12-h-light/12-h-dark cycle, starting 2 h (2 am) after the lights came on at 12 am. For LL free-running circadian time-course experiments, plants were entrained in 12-h-light/12-h-dark conditions as above, and representative control light and dark samples were collected at 10 am and 10 pm, respectively, prior to release into LL conditions. The LL conditions were as follows: light, 100 µmol m⁻² s⁻¹; temperature, 15°C; and humidity, 70%. LP6 pairs were sampled every 4 h from three individual (clonal) plants, starting at 2 am (2 h after lights on). All sampling involved immediate freezing of leaves in liquid nitrogen. Samples were stored at −80°C until use.

Generation of Transgenic K. fedtschenkoi Lines

Full-length complementary DNA (cDNA) clones for KfNAD_ME_b1 and KfPPDK were identified via TBLASTN searching the corresponding Arabidopsis (Arabidopsis thaliana) amino acid sequences against a de novo sequenced and assembled transcriptome for K. fedtschenkoi (J. Hartwell, unpublished data). Intron-containing hairpin RNAi constructs were designed to target silencing of the genes encoding the β-subunit of mitochondrial NAD-ME (KfNAD_MEb1, K. fedtschenkoi ortholog of At4g00570) and PPDK (KfPPDK, ortholog of At4g15530; Cushman and Bohnert, 1999; Parsley and Hibberd, 2006). For each gene, a short fragment (396 bp for KfNAD_ME_b1 and 382 bp for KfPPDK) was amplified from CAM leaf cDNA using high-fidelity PCR with KOD Hot Start DNA Polymerase (Merck). The amplified fragments spanned the 3′ end of the coding sequence and extended into the 3′ untranslated region. Primers used for cloning are listed in Supplemental Table S1. PCR products were cloned into the pENTR/D Gateway-compatible entry vector via directional TOPO cloning (Life Technologies) and recombined into the intron-containing hairpin RNAi binary vector pK7GWIWG2(II) as described by Karimi et al. (2002) using LR Clonase II enzyme mix (Life Technologies). Constructs were confirmed by DNA sequencing and introduced into Agrobacterium tumefaciens strain GV3101 using the freeze-thaw method (Höfgen and Willmitzer, 1988).

A. tumefaciens-mediated stable transformation of K. fedtschenkoi was carried out using a tissue culture-based procedure. Approximately 1 month prior to transformation, K. fedtschenkoi plantlets were propagated axenically to generate sterile explants. Mature leaves were washed in 70% (v/v) ethanol for 30 s and then surface sterilized in 10% (v/v) NaOCl with 1% (v/v) Tween 20 for 10 min, and rinsed five times with sterile water. Leaf margins were excised, cut into 2-cm-long pieces, and placed on Murashige and Skoog 30 (MS30) media (Murashige and Skoog medium with Gamborg B5 vitamins; Melford), 3% (w/v) Suc (Sigma), pH 5.8, 0.8% (w/v) Phytoagar (Melford) containing 1 mg L⁻¹ thidiazuron (TDZ), and 0.2 mg L⁻¹ indole-3-acetic acid (IAA). All plant hormones were sourced from Duchefa (via Melford). The cut edge of the leaf margin was placed in contact with the growth medium. Two to three plantlets developed on each section of leaf margin. The size of the plantlets used for transformation was around 0.5 cm.

Plant transformation was performed as follows. A. tumefaciens strain GV3101 cells transformed with the relevant hairpin RNA construct in pK7GWIWG2(II) were grown overnight in 10 mL of Luria-Bertani liquid medium containing 100 mg L⁻¹ streptomycin and 300 mg L⁻¹ spectinomycin with shaking at 28°C. Cells were collected by spinning cultures for 10 min at 1,932g in a Sorvall Legend LT benchtop centrifuge using a swing out rotor. Pellets were resuspended in liquid MS30 medium to optical density at 600 nm = 0.1. Acetosyringone was added to the culture to a final concentration of 100 μm, and the resuspended A. tumefaciens culture was maintained at room temperature for at least 2 h before dipping plant tissues in the A. tumefaciens suspension.

The axenic leaf margin plantlets generated as described above were separated from their mother leaf and sliced finely in a petri dish using a sharp razor blade to generate the explants for transformation. Explants were cocultivated with the A. tumefaciens culture for 1 h in the sterile laminar flow bench with occasional gentle agitation of the explants within the A. tumefaciens suspension. Subsequently, the liquid culture was poured off and the explants were blotted dry using sterile filter paper. Individual pieces of the plant tissue were placed on cocultivation medium (MS30, 1 mg L⁻¹  TDZ, 0.2 mg L⁻¹  IAA, and 100 μm acetosyringone) for 48 h at 20°C under 16 h of light (approximately 100 µmol m⁻² s⁻¹) and 8 h of dark. The explants were subsequently transferred to regeneration medium (MS30 with 1 mg L⁻¹  TDZ and 0.2 mg L⁻¹  IAA plus 100 mg L⁻¹ kanamycin and 200 mg L⁻¹ carbenicillin) to select for positive transformants and kill off the remaining A. tumefaciens. Once healthy green callus had formed after approximately 4 to 6 weeks, it was transferred to shoot induction medium (MS30 with 1 mg L⁻¹ benzylaminopurine and 0.2 mg L⁻¹  IAA plus 100 mg L⁻¹ kanamycin and 200 mg L⁻¹ carbenicillin). Shoots that developed after another 3 to 4 weeks were detached from the callus when large enough to handle, and were placed on rooting medium (MS30 plus 50 mg L⁻¹ kanamycin and 200 mg L⁻¹ carbenicillin). When roots developed, the transgenic plants were transferred to soil (John Innes no. 3 compost with one-third Perlite and Osmocote fertilizer at the manufacturer’s recommended level). All tissue cultures were grown in a controlled environment growth room under 16 h of light (100 µmol m⁻² s⁻¹) and 8 h of dark at 20°C.

High-Throughput Leaf Acidity and Starch Content Screens

Leaf acidity (as a proxy for leaf malate content) and leaf starch content were screened with leaf disc stains using chlorophenol red and iodine solution at both dawn and dusk as described by Cushman et al. (2008). For each transgenic line, leaf discs were sampled in triplicate at 1 h before dawn and 1 h before dusk and stained in a 96-well plate format.

Net CO₂ Exchange

Gas exchange measurements were performed using a six-channel, custom-built infrared gas analyzer system (PP Systems), which allowed the individual environmental control (CO₂/water) and measurement of rates of CO₂ uptake for each of six gas exchange cuvettes with measurements collected every 10 min. The six cuvettes were housed in a controlled environment growth chamber (Snijders Microclima MC-1000; Snijders Scientific) whose clock settings and conditions determined the light intensity and temperature regime in the gas exchange cuvette. Gas exchange parameters were then measured using an infrared gas exchange system based on a CIRAS-DC analyzer (PP Systems). Gas exchange parameters were calculated using SC-DC software (PP System). Either whole, rooted, young plants with eight leaf pairs or detached LP6 with their petiole in distilled water were placed in the gas exchange cuvettes. The contribution of soil respiration and soil moisture, or dissolved gases and water vapor from the distilled water for the detached leaf experiments, to the environment in the cuvette was minimized by wrapping multiple layers of parafilm around the rim of the pot, taking care to seal around the stem/petiole. During the 12-h-light period, illumination was 250 µmol m⁻² s⁻¹ with 60% humidity at 25°C and in the dark period humidity was 70% and temperature 15°C. Free-running circadian rhythms of gas exchange were measured under LL conditions of light (100 µmol m⁻² s⁻¹), temperature (15°C), and humidity (70%). LL CO₂ exchange measurements began at dawn and continued for at least 7 d. At the end of the gas exchange experiments, leaves were removed and scanned to calculate leaf area using ImageJ freeware image analysis software (ImageJ, http://rsbweb.nih.gov/ij/). The CIRAS SC-DC software (PP Systems) had a built-in feature to recalculate the photosynthetic rate based on the leaf areas. All experiments were repeated at least three times using three separate individual young plants (eight leaf pairs), or detached LP6 from three separate clonal plants, and representative gas exchange traces are shown. The wild type, rNAD-ME1, and rPPDK1 were compared in neighboring gas exchange cuvettes during each run such that the data are directly comparable between each line. For example, many runs included two cuvettes containing the wild type, two cuvettes containing rNAD-ME1 and two with rPPDK1.

Drought-Stress Experiments

For the drought-treatment gas exchange experiments, clonal leaf margin plantlets from the wild type, rNAD-ME1, and rPPDK1 were grown to the LP8 stage using 150 mL of our standard compost/perlite mix (as described above under “Plant Materials”) and maintained well watered throughout their development. At the beginning of the drought experiments, 150 mL of compost was watered to field capacity (100% compost hydration; compost held an average of 108 mL of water at the start of the drought treatment). Water was withheld for 25 d under our standard 12-h-light/12-h-dark growth conditions in the Snijders Microclima MC-1000 growth cabinet, by which time soil water content was reduced to 5% to 10% of field capacity, based on weight. The number of days of drought was selected based on gas exchange measurements; at 25 d, the wild-type plants continued to perform substantial dark-period atmospheric CO₂ fixation, but CO₂ fixation in the light was greatly reduced relative to well-watered plants (Fig. 4, compare B with A). Twenty-five days was therefore selected as the appropriate period of drought to compare and contrast the wild type against rNAD-ME1 and rPPDK1 based on this strong CAM dark CO₂ fixation phenotype of the wild type.

For the drought-stress experiment to investigate the induction of leaf anthocyanins, mature greenhouse-grown plants (10–12 leaf pairs stage, as described above under “Plant Materials”) growing in 1 L of compost mix were watered to field capacity at the beginning of the experiment before withholding water for 39 d under greenhouse conditions. After 39 d, water content of the compost was reduced to 1% of field capacity based on weight; the plants were therefore under severe drought stress. Thirty-nine days was selected based on the development of a strong purple coloration for the wild-type leaves. Photographs of the plants at the time of sampling are shown in Supplemental Figure S6.

Leaf Malate and Starch Content

LP6 (full CAM in the wild type) from mature plants were sampled into liquid nitrogen at the indicated times and stored at −80°C. Frozen samples were ground to a fine powder in liquid nitrogen and 0.5 g of frozen tissue was mixed with 4 mL of 80% (v/v) methanol and heated in a water bath at 80°C for 1 h with intermittent shaking. Samples were centrifuged at 3,636g for 10 min and the supernatant aspirated and stored on ice. Two further extractions were performed using 2 mL of methanol followed by 1 mL of methanol at 80°C as above. The pooled supernatants were dried using a vacuum drier at 50°C. The dried residue was redissolved in 1 mL of 0.2 m bicine buffer, pH 7.8, and centrifuged to remove any insoluble material. Malate was measured using an enzyme-linked spectrophotometric assay as described by Möllering (1974).

Starch was extracted from the insoluble material left after methanol extraction and assayed as outlined in Smith and Zeeman (2006). The assay involved hydrolyzing the starch to Glc units and measuring Glc with an enzyme-linked spectrophotometric assay (Smith and Zeeman, 2006).

Total RNA Isolation and Semiquantitative RT-PCR

Total RNA was isolated from 100 mg of frozen, ground leaf tissue using the Qiagen RNeasy kit following the manufacturer’s protocol with the addition of 13.5 µL of 50 mg mL⁻¹ polyethylene glycol 20,000 (PEG-20000) to the 450 µL of RLC buffer used for each extraction. cDNA was synthesized from the total RNA using the Qiagen Quantitect RT kit according to the manufacturer’s instructions. The resulting cDNA was diluted 1:4 with molecular biology-grade water prior to use in RT-PCR. PCR reactions were performed using 1 µL of each cDNA sample in a reaction mixture (10 µL) containing 1× Sigma REDTaq ReadyMix PCR reaction mix with MgCl₂ and gene-specific primers according to the method described by Boxall et al. (2005). The housekeeping gene Polyubiquitin10 (KfUBQ10, GenBank accession no. KM114222) was used as a loading control. Quantification of band intensities was achieved using Metamorph image analysis software (Molecular Devices). Normalization factors were calculated based on KfUBQ10 for each sample and used to calculate relative intensities of the gene of interest. Primers used for RT-PCR are listed in Supplemental Table S2.

Immunoblotting

Total protein extracts of K. fedtschenkoi leaves were prepared by grinding samples in 1× SDS sample buffer (1 m Tris-HCl, pH 6.8, 20% [v/v] glycerol, 4% [v/v] 2-mercaptoethanol, and 3% [w/v] SDS) in a precooled pestle and mortar in liquid nitrogen with a small quantity of acid-washed sand. Ground samples plus buffer were placed in a screw-cap microfuge tube and boiled immediately for 5 min. The pH was adjusted to 6.8 and samples were centrifuged for 5 min at full speed in a microfuge. The supernatants were removed and protein concentration was quantified using Bradford Ultra Reagent (Expedeon).

One-dimensional SDS-PAGE and immunoblotting of leaf proteins were carried out following standard methods. Blots were developed using the ECL system (GE Healthcare). Immunoblot analysis was carried out using antisera to PPC (raised against PPC from K. fedtschenkoi, kindly supplied by Hugh G. Nimmo, University of Glasgow; Nimmo et al., 1986), and the phosphorylated form of PPC (raised against a phospho-PPC peptide from barley [Hordeum vulgare] and kindly supplied by Cristina Echevarría, Universidad de Sevilla; González et al., 2002; Feria et al., 2008), PPDK and phosphorylated PPDK (raised against maize [Zea mays] PPDK and kindly supplied by Chris J. Chastain, Minnesota State University, Moorhead; Chastain et al., 2000, 2002), and α- and β-NAD-ME (raised against the Arabidopsis proteins, supplied by Maria F. Drincovich, Universidad Nacional de Rosario; Tronconi et al., 2008).

Dry Weight Growth Measurements

Mature plants were grown from developmentally synchronized clonal leaf plantlets in greenhouse conditions (described above) for 138 d, harvested as above-ground tissues, and dried in an oven at 60°C until they reached a constant weight. Weights were measured to two decimal places in grams using a fine balance. Seven individual clonal plants were sampled for each line.

Enzyme Assays

For all enzyme assays, frozen leaf tissue was ground in liquid nitrogen with a small quantity of acid-washed sand and the relevant enzyme specific extraction buffer (approximately 1 g tissue to 3 mL of extraction buffer). Extracts were filtered through Miracloth (Merck), and a sample of the filtrate was taken for chlorophyll estimation, before being centrifuged at full speed in a microfuge. The supernatant was either used directly for enzyme assays, or was first desalted using a Sephadex-G25 desalting column for the NAD-ME assay. For PPDK, leaves were detached and placed with their petiole in distilled water, and illuminated at >1,000 µmol m⁻² s⁻¹ for 2 h prior to harvesting into liquid nitrogen as described by Sugiyama and Laetsch (1975) for Kalanchoë daigremontiana, and Chastain et al. (2002) for C₄ and C₃ species (maize, rice [Oryza sativa], Vicia faba, spinach [Spinacia oleracea], and Flaveria trinervia).

The PPC extraction buffer comprised 100 mm Tris-HCl, pH 8.0, 2 mm EDTA, 10 mm  l-malate, 1 mm benzamidine hydrochloride, 1 mm dithiothreitol (DTT), and 2% (w/v) PEG-20000, ground with 100 mg g⁻¹ tissue NaHCO₃ and 200 mg g⁻¹ tissue polyvinylpolypyrrolidone (PVPP). The assay comprised 50 mm Tris-HCl, pH 7.8, 5 mm MgCl₂, 2 mm  PEP, 0.2 mm NADH, 10 mm NaHCO₃, 5 U MDH (from pig heart; Roche Life Sciences).

The NAD-ME extraction buffer comprised 100 mm HEPES-KOH, pH 7.0, 2 mm MnCl₂, 5 mm  DTT, 1% polyvinylpyrrolidone-40, 1 mm EDTA, 2% PEG-20000, and 0.5% Triton X-100. The desalting buffer comprised 100 mm HEPES-KOH, pH 7.0, 2 mm MnCl₂, 5 mm  DTT, and 1 mm EDTA. The assay comprised 50 mm HEPES-KOH, pH 7.6, 1 mm EDTA, 1 mm  DTT, 5 mm  l-malate, 5 mm NAD, 25 µm NADH, 1 U pig heart MDH (Roche Life Sciences), 100 µm acetyl CoA, and 5 mm MnCl₂; 25 µm NADH was included to minimize the interference of MDH in the assay and reduce the chance of overestimation of NAD-ME activity (Hatch et al., 1982).

The PPDK extraction buffer comprised 50 mm HEPES-KOH, pH 8.2, 5 mm  DTT, 0.2 mm EDTA, 2% PEG-20000, 2.5 mm K₂HPO₄, 2.5 mm pyruvate, and 200 mg g⁻¹ tissue PVPP. The assay comprised 50 mm Tris-HCl, pH 8.0, 5 mm  DTT, 10 mm MgCl₂, 1.25 mm pyruvate, 0.25 mm NADH, 2.5 mm NaHCO₃, 2.5 mm K₂HPO₄, 1 U pig heart MDH, 1.25 mm ATP, and 6 mm Glc-6-P.

The NADP-ME extraction buffer comprised 200 mm Tricine-NaOH, pH 7.6, 1 mm EDTA, 2% polyvinylpyrrolidone-40, 2 mm  DTT, 1 mm benzamidine hydrochloride, 2% PEG-20000 plus 50 mg g⁻¹ tissue NaHCO₃, and 200 mg g⁻¹ tissue PVPP. The assay comprised 50 mm HEPES-KOH, pH 7.0, 0.1 mm EDTA, 0.5 mm NADP⁺, 5 mm  l-malate, 5 mm  DTT, and 0.5 mm MgCl_2.

Chlorophyll and Anthocyanin Assays

Chlorophyll content was determined as described by Arnon (1949). Anthocyanin content was estimated by extracting 100 mg of liquid nitrogen ground leaf tissue in methanol:1% (v/v) HCl, followed by chloroform extraction, and measurement of the absorbance of the upper aqueous phase at 530 and 657 nm as described in Neff and Chory (1998).

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers KfPPCK1 (AF162662), KfPPC1_CAM (KM078709), KfPPDK (KM078710), KfNAD_MEa1 (KM078711), KfNAD_MEb1 (KM078712), KfNADP_ME1 (KM078713), KfGPT1 (KM078714), KfGPT2 (KM078715), KfTOC1 (KM078716), KfCCA1_LHY1 (KM078717), KfCCA1_LHY2 (KM078718), KfCAB1 (KM078719), KfPPCK2 (KM078720), KfGWD1 (KM078721), KfPPDK-RP1 (KM078722), KfPPDK_RP2 (KM078723), KfGI (KM078724), KfLOS2 (KM078725), KfTOC1_2 (KM078726), and KfUBQ10 (KM114222).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Representative diel cycles of gene transcript abundance for genes associated with CAM and the circadian clock in the wild type and transgenic lines rNAD-ME1 and rPPDK1.

• Supplemental Figure S2. Forty-eight-hour LD gas exchange profile for the wild type, rNAD-ME1, and rPPDK1 detached leaf pair 6 (full CAM leaves) from plants pre-entrained in 16-h-light/8-h-dark cycles.

• Supplemental Figure S3. Total net CO₂ fixation for the wild type, rNAD-ME1, and rPPDK1 calculated from the area under the curve in Figure 4A.

• Supplemental Figure S4. LD and LL gas exchange profiles for the wild type, rNAD-ME1 and rPPDK1 using detached LP6 from plants previously entrained in 16-h-light/8-h-dark cycles.

• Supplemental Figure S5. Nucleotide alignment of the region of the KfNAD_ME_b1 gene used for the hairpin RNAi binary construct with the corresponding region of the KfNAD_ME_a1 gene.

• Supplemental Figure S6. Visible phenotypes of the wild type compared with rNAD-ME1 and the wild type compared with rPPDK1 after 39 d of drought stress treatment imposed by withholding water.

• Supplemental Table S1. Primers used to generate the KfNAD_ME_b1 and KfPPDK1 hairpin RNAi binary constructs for A. tumefaciens-mediated transformation of K. fedtschenkoi.

• Supplemental Table S2. Primers used to amplify gene specific fragments of CAM and circadian clock-associated genes using semiquantitative RT-PCR.

ACKNOWLEDGMENTS

We thank Hugh Nimmo (University of Glasgow), Cristina Echevarria (Universidad de Sevilla), Maria Drincovich (Universidad Nacional de Rosario, Argentina), and Chris Chastain (University of Minnesota-Moorhead,) for sharing antibodies, as well as Carol Johnstone (University of Glasgow) and Jonathan Foster (University of York) for early work on the development of the stable transformation method for K. fedtschenkoi, Noémie Mayer for the contribution to initial cloning and transformation experiments that led to the production of the NAD-ME and PPDK  RNAi lines of K. fedtschenkoi, Ben Wareham (University of Exeter) for developing the Excel macro used to calculate the area under the gas exchange curves, and Martin Mortimer (University of Liverpool) for advice on the Student’s t tests used to analyze the growth data.

Glossary

 
• ME

  malic enzyme

 
• PPDK

  pyruvate orthophosphate dikinase

 
• CAM

  Crassulacean acid metabolism

 
• PPC

  phosphoenolpyruvate carboxylase

 
• MDH

  malate dehydrogenase

 
• PCK

  phosphoenolpyruvate carboxykinase

 
• PEP

  phosphoenolpyruvate

 
• PPCK

  phosphoenolpyruvate carboxylase kinase

 
• RNAi

  RNA interference

 
• cDNA

  complementary DNA

 
• MS30

  Murashige and Skoog 30

 
• TDZ

  thidiazuron

 
• IAA

  indole-3-acetic acid

 
• PEG-20000

  polyethylene glycol 20,000

 
• DTT

  dithiothreitol

 
• PVPP

  polyvinylpolypyrrolidone

LITERATURE CITED

Author notes