Absence of carbonic anhydrase in chloroplasts affects C₃ plant development but not photosynthesis

Significance

Carbonic anhydrase enzymes located in chloroplast stroma have been hypothesized to facilitate photosynthesis in C₃ plants because they catalyze a reaction involving bicarbonate and CO₂, a substrate of the carbon-fixing enzyme RuBisCO. To test this possibility, tobacco mutants completely lacking chloroplast stromal carbonic anhydrase activity were produced by CRISPR/Cas9 mutagenesis. The plants displayed normal photosystem II activity and CO₂ assimilation but also abnormal development and increased reactive oxygen species and stromal pH. We conclude that chloroplast carbonic anhydrase does not play a direct role in providing CO₂ for carbon fixation. Instead, as is also true in microorganisms, carbonic anhydrase is necessary to supply bicarbonate for biosynthetic processes.

Abstract

The enzyme carbonic anhydrase (CA), which catalyzes the interconversion of bicarbonate with carbon dioxide (CO₂) and water, has been hypothesized to play a role in C₃ photosynthesis. We identified two tobacco stromal CAs, β-CA1 and β-CA5, and produced CRISPR/Cas9 mutants affecting their encoding genes. While single knockout lines Δβ-ca1 and Δβ-ca5 had no striking phenotypic differences compared to wild type (WT) plants, Δβ-ca1ca5 leaves developed abnormally and exhibited large necrotic lesions even when supplied with sucrose. Leaf development of Δβ-ca1ca5 plants normalized at 9,000 ppm CO₂. Leaves of Δβ-ca1ca5 mutants and WT that had matured in high CO₂ had identical CO₂ fixation rates and photosystem II efficiency. Fatty acids, which are formed through reactions with bicarbonate substrates, exhibited abnormal profiles in the chloroplast CA-less mutant. Emerging Δβ-ca1ca5 leaves produce reactive oxygen species in chloroplasts, perhaps due to lower nonphotochemical quenching efficiency compared to WT. Δβ-ca1ca5 seedling germination and development is negatively affected at ambient CO₂. Transgenes expressing full-length β-CA1 and β-CA5 proteins complemented the Δβ-ca1ca5 mutation but inactivated (ΔZn-βCA1) and cytoplasm-localized (Δ62-βCA1) forms of β-CA1 did not reverse the growth phenotype. Nevertheless, expression of the inactivated ΔZn-βCA1 protein was able to restore the hypersensitive response to tobacco mosaic virus, while Δβ-ca1 and Δβ-ca1ca5 plants failed to show a hypersensitive response. We conclude that stromal CA plays a role in plant development, likely through providing bicarbonate for biosynthetic reactions, but stromal CA is not needed for maximal rates of photosynthesis in the C₃ plant tobacco.

The metalloenzyme carbonic anhydrase (CA, Enzyme Commission [EC] 4.2.1.1) catalyzes the reversible hydration of carbon dioxide with bicarbonate and protons with carbonic acid as an intermediate (Fig. 1). The uncatalyzed reactions are quite slow, with effective rate constants of ∼0.15 s⁻¹ for the forward reaction and 50 s⁻¹ for the reverse reaction; at equilibrium, dissolved CO₂ is favored over bicarbonate by a factor of 340. CAs have one of the fastest catalytic cycles and increase the rate by which carbon dioxide and bicarbonate reach equilibrium by about seven orders of magnitude (1).

Fig. 1.

Scheme illustrating carbonate equilibria, the reactions catalyzed by CA, and their coupling to the fixation of atmospheric CO₂ in C₃, C₄, and CAM photosynthesis. Rate constants are in pure water at 25 °C.

Vascular land plants contain α-, β-, and γ- CAs (2), which function in a myriad of physiological roles depending on the plant and the type of photosynthesis it carries out (3, 4). In C₃ photosynthetic plants, dissolved CO₂ is the substrate for the enzyme RuBisCO, which feeds inorganic carbon into the Calvin cycle. In C₄ and Crassulacean acid metabolism (CAM) photosynthesis, inorganic carbon is ultimately fixed from dissolved CO₂ and RuBisCO but is initially fixed from bicarbonate via the enzyme phosphoenolpyruvate carboxylase (5). Considering the location of CA function between dissolved CO₂ and bicarbonate, one might anticipate very different photosynthetic functions for CAs in C₃ versus C₄ and CAM plants (6, 7).

Research on CAs in C₃ plants has resulted in the assignment of several roles to the enzymes. The chloroplasts of Arabidopsis thaliana (hereafter Arabidopsis) are known to contain at least two stromal CAs: β-CA1 and β-CA5 (8). In Arabidopsis, the chloroplast-localized β-CA1 is the most highly expressed CA in leaf tissue, with nearly 50 times the amount of RNA-sequencing reads and 13 times the number of expressed sequence tags compared to β-CA5 (9). Tobacco β-CA1 is a salicylic acid (SA)–binding protein and participates in plant defense. Silencing of β-CA1 in tobacco resulted in a reduction in the hypersensitive response (HR) during a Pto-avrPto interaction (10), and, more recently, multiple β-CAs have been shown to play a role in the perception of SA in Arabidopsis (11). Mutating both the chloroplast-localized β-CA1 and plasma membrane–localized β-CA4 in Arabidopsis caused a reduction in stomatal CO₂ response (12, 13). In another study of the β-ca1ca4 Arabidopsis mutant, researchers observed an intolerance to oxidative stress, suggesting that CAs play a role in cell death homeostasis during light stress (14). Transgenic tobacco plants carrying an antisense construct that reduced stromal CA expression to 5% of the wild-type (WT) level had altered lipid biosynthesis, though plant morphology appeared to be normal (15). Furthermore, cytoplasmic CAs like β-CA2 may also play an important function in certain cellular biosynthesis pathways, as demonstrated in Arabidopsis β-ca2ca4 mutants, which showed a significant decrease in aspartate that was rescued when plants were grown in elevated CO₂ levels (9).

Although CAs in plants can account for as much as 20% of total soluble leaf protein (3), one study has suggested a limited role of CAs in C₃ photosynthesis. In Nicotiana tabacum (tobacco, a C₃ plant), reduction of chloroplast CA activity by 99% caused no significant reduction in the CO₂ assimilation rate (16). However, CAs have a very high catalytic rate so that even as little as 1% remaining CA could potentially be adequate for many of its functions in the stroma. Nevertheless, this study suggests the possibility that most CO₂ that enters the leaf from the atmosphere is used directly by RuBisCO without passing through a bicarbonate pool (17).

Interest in the role of stromal CAs has recently been reawakened because of efforts to install CO₂-concentrating mechanisms into C₃ plant chloroplasts (18, 19). In cyanobacteria, CA is located within the carboxysome, where bicarbonate is converted to CO₂ for use by RuBisCO (20, 21). If CA is introduced by transgene expression into the cyanobacterial cytoplasm, the CO₂-concentrating mechanism cannot function (21). Incorporating an operational carboxysome within chloroplasts will require removal of all CA in the stroma (22), which is the functional equivalent of the cytoplasm of cyanobacteria. However, the consequences of a complete lack of stromal CA activity have not previously been investigated in vascular plants.

Here, we identify β-CA1 and β-CA5 as the tobacco chloroplast stromal CAs and use the CRISPR/Cas9 system to generate knockout lines, thus removing all CA activity from the stroma. We discovered that the absence of stromal CA activity had unexpected detrimental effects on leaf, floral bud, and seed development and increased the reactive oxygen species (ROS) and increased the stroma pH. We demonstrate that the developmental phenotypes can be reversed by growing the mutant plants in high (9,000 ppm) CO₂, which is known to result in increased nonenzymatic production of bicarbonate. Importantly, measurements of photosynthesis and gas exchange did not distinguish the double mutant from WT, indicating that rapid equilibration of dissolved CO₂ and bicarbonate pools is not required for high steady-state rates of C₃ photosynthesis.

Results

Tobacco CA Enzymes β-CA1 and β-CA5 Localize to the Chloroplast Stroma.

The coding sequences (CDS) of three candidate tobacco CA genes were selected based on TargetP sequence-based predictions (β-CA1 and β-CA5) and previous localization studies of CA homologs in Arabidopsis (α-CA1). A series of YFP fusion experiments were carried out to determine which of these tobacco CA enzymes localize to the chloroplast stroma. In Arabidopsis, β-CA1 and β-CA5 have been observed to localize to the chloroplast stroma (8). β-CA1 in Arabidopsis and tobacco are 75 to 76% similar, and the β-CA5 proteins are 78 to 83% similar.

Tobacco YFP-fused β-CA1 and β-CA5 were both found to be localized to the chloroplast stroma (Fig. 2A). There are four predicted isoforms for β-CA5 in tobacco (accession numbers XP_016482110, XP_016482109, XP_016446542, and XP_016446541), two for each of the two tobacco progenitor Nicotiana sylvestris and Nicotiana tomentosiformis alleles, but reverse transcription of tobacco leaf RNA yielded only one isoform for each (see Materials and Methods).

Fig. 2.

Localization and CRISPR-generated mutations of tobacco chloroplast CAs. (A) Confocal imaging of CA-YFP constructs transiently expressed in N. tabacum mesophyll cells. βCA1-YFP localized to the chloroplast stroma and cytoplasm. βCA5-YFP localized to the chloroplast stroma. Magenta: Chlorophyll autofluorescence, excitation 633 nm. Green: YFP, excitation 514 nm. (Scale bars: 10 µm.) (B) Single mutant lines Δβ-ca1 and Δβ-ca5 show no observable phenotype in their gross morphology compared to WT. The double mutant, Δβ-ca1ca5, produces pale leaves with symptoms of necrosis. Plants were grown in 9,000 ppm CO₂ for 4 wk and then transferred to ambient CO₂ for 2 wk before being imaged. White arrows indicate Δβ-ca1ca5 leaves that developed in high CO₂. (C) CA activity of whole leaf homogenate at ambient CO₂ (Bars: SE). There is no significant difference detectable between WT activity and Δβ-ca5 activity, with four replicates for each assay (t test, P = 0.68), while Δβ-ca1 and Δβ-ca1ca5 differ, with P = 0.0001 (t test).

The tobacco α-CA1 is predicted to be a secretory pathway protein with a transmembrane sequence near the N terminus, but it is reported to be localized to the chloroplast stroma in Arabidopsis (23). YFP-fused tobacco α-CA1 did not localize to the chloroplast stroma and instead appeared to be on the plasma membrane, which is consistent with a software prediction (SI Appendix, Fig. S1). Taken together, YFP localization indicates that β-CA1 and β-CA5 are the CA enzymes in the stroma of tobacco chloroplasts.

Targeting Tobacco β-ca1 and β-ca5 with CRISPR/Cas9.

The CRISPR/Cas9 system was used to generate mutations in the tobacco β-ca1 and β-ca5 genes to create three transgenic lines: Δβ-ca1, Δβ-ca5, and the double knockout Δβ-ca1ca5. Tobacco is an allotetraploid (a hybrid of N. sylvestris and N. tomentosiformis), so there are two homologs for each of the CA genes. The single guide RNAs (sgRNAs) were designed to target the sequences of both forms of the CA homologs (SI Appendix, Fig. S2).

A Δβ-ca1 mutant line contains a 2-bp deletion in the messenger RNA (mRNA) sequence of one homolog of the β-ca1 gene and a 52-bp deletion in the other (SI Appendix, Fig. S3). Both deletions created a frame-shift mutation, resulting in an early stop codon in the coding sequences. A Δβ-ca5 mutant contains 2-bp and 31-bp deletions in the mRNA sequence of the two β-ca5 homologs, both of which result in frame-shift mutations. The β-ca1 homologs in the double mutant Δβ-ca1ca5 are composed of a 1-bp insertion or a 2-bp deletion, and the β-ca5 homologs both have large, 31-bp deletions between the two sgRNA target sites (SI Appendix, Fig. S3). The sequencing data demonstrates the creation of frame-shift mutations in the targeted CA genes in all the CRISPR/Cas9-generated transgenic lines.

The gross morphology of Δβ-ca1 and Δβ-ca5 is similar to WT tobacco when the plants are grown at ambient CO₂ concentrations (Fig. 2B). In contrast, Δβ-ca1ca5 displays a dramatic developmental phenotype, which results in pale and shriveled leaves. The plants in Fig. 2B were first grown at 9,000 ppm CO₂ for 4 wk before being transferred to ambient CO₂ for 2 wk and then imaged. The white arrows highlight Δβ-ca1ca5 leaves, which developed in high CO₂ before being transferred.

CA Activity in Mutant Lines.

The CA activity of the lines was measured in whole leaf homogenate from plants growing at ambient CO₂ in 16 h daylight, using leaves that had expanded under high CO₂ conditions. We used the CA activity assay described in ref. 11 with modifications (see Materials and Methods). This assay measures the activity of CAs in all subcellular locations of the tissue used, not only activity within the chloroplasts. Δβ-ca1 and Δβ-ca1ca5, the two lines in which β-CA1 was mutated, show greatly decreased levels of CA activity (Fig. 2C), while CA activity in Δβ-ca5 showed no reduction. As expected, most of the CA activity found in leaves appears to come from the highly expressed β-CA1 enzyme, with the Δβ-ca1 transgenic line showing an 87% reduction in CA activity compared to WT (Fig. 2C).

Morphological Characteristics of Δβ-ca1ca5 Transgenic Tobacco.

The gross morphological and developmental phenotypes of the Δβ-ca1ca5 double mutant were observed at ambient CO₂ concentrations (about 415 ppm; Figs. 2B and 3). When Δβ-ca1ca5 mutants were transplanted into soil and grown in 9,000 ppm CO₂, their morphology (leaf color and form) appeared to mirror WT tobacco (Fig. 3B). Leaves of Δβ-ca1ca5 that were able to fully expand at high CO₂ retained their WT-like morphology when they were transferred back to ambient CO₂ concentrations (Fig. 3C, green arrow). These leaves did not become necrotic and remained green as the plant transitioned into flowering. Leaves that did not finish expanding in high CO₂ displayed necrotic lesions in their still-developing photosynthetic tissues (Fig. 3C, blue arrow), and leaves that had initiated their development in ambient CO₂ also exhibited the mutant phenotype (Fig. 3 C, red arrow and D). When first emerging at ambient CO₂ concentrations, Δβ-ca1ca5 leaves do not show any obvious phenotypes or gross morphological differences from WT (Fig. 3G, leaf 1). As leaf development progresses, however, necrotic lesions form and spread throughout the photosynthetic tissue. The midrib and primary veins do not display a cell-death phenotype (Fig. 3G, leaves 6 and 7). Despite the death of mesophyll cells, the leaves remain attached to the stem of the tobacco plants. The pattern of necrosis (appearing first at the base, Fig. 3D and SI Appendix, Fig. S5) matches the pattern in which leaves transition from sink-to-source tissue (tip to base) (24). This pattern is consistent with the lack of necrosis in tissues like the midrib and primary veins.

Fig. 3.

Δβ-ca1ca5 mutants have a severe developmental phenotype rescued by high CO₂. (A) Δβ-ca1ca5 shoot at ambient CO₂ growing on 3% sucrose media. (B) Δβ-ca1ca5 transplanted to soil and grown at 9,000 ppm CO₂. (C) Δβ-ca1ca5 mutant allowed to mature for 4 wk at 9,000 ppm CO₂ before being transferred to ambient CO₂ (at white arrow) and allowed to grow for four more weeks. Green arrow: example of a leaf that fully expanded at high CO₂. Blue arrow: leaf that was developing at the time of transfer. Red arrow: leaf that budded and developed at ambient CO₂. (D) An expanding high-CO₂–grown Δβ-ca1ca5 leaf after being transferred to ambient CO₂. (E and F) Abscission of flowering buds in Δβ-ca1ca5 mutants at ambient CO₂. Purple arrows show two of the abscission zones. (G) Development of mutant phenotype in leaves developing at ambient CO₂ (Left to Right: youngest to oldest). Numbers indicate node position from the top of the plant.

At ambient CO₂, Δβ-ca1ca5 plants also exhibit a unique flowering phenotype. Before the emergence of petals, the developing flower buds experience an early termination event, separating from the stem at the abscission zone (Fig. 3 E and F). A small number of buds eventually develop into mature flowers at ambient CO₂ and produce seeds via natural self-pollination.

Free Fatty Acid Distribution Is Different between WT and Δβ-ca1ca5 Leaves.

Previously, acetyl incorporation into lipids, which is catalyzed by acetyl-CoA carboxylase using bicarbonate as a substrate, was found to be reduced in antisense β-CA1 tobacco in which CA was present at only 5% of WT levels (15). Here, we investigated the free fatty acid (FFA) makeup of WT and Δβ-ca1ca5 leaves (one young leaf from each of the first three nodes) grown on sucrose media in ambient CO₂. Five fatty acid classes were detected by mass spectroscopy (Fig. 4A and SI Appendix, Fig. S4). A significant difference between the two lines was observed. Unsaturated FFAs 18:2, 18:3, and 18:4 were up-regulated in Δβ-ca1ca5 leaves, whereas saturated FFAs 16:0 and 18:0 were higher in WT leaves. Overall, the total area under the peak for measured fatty acids was lower by about 30% in Δβ-ca1ca5 when compared to WT.

Fig. 4.

Analysis of CA mutant seeds. (A) FFA analysis on ambient CO₂-grown leaves of WT tobacco and the Δβ-ca1ca5 mutant line (P ≤ 0.001). Nomenclature for FFA is number of carbons::number of saturated bonds. (B) Sample comparison of T1 Δβ-ca1ca5 seeds with different growth conditions of the T0 mutant flowers. (Scale bars: 0.2 mm.) (C) Average mass of seeds from WT and Δβ-ca1ca5 plants grown under high or ambient CO₂ concentrations (Bars: SD). (D) Average size of seeds produced by WT and Δβ-ca1ca5 plants grown under high or ambient CO₂ concentrations (Bars: SE). (E) Germination rate of T1 seeds derived from Δβ-ca1ca5 T0 plants grown in different conditions. x-axis indicates germination conditions; a, ambient in soil; h, high CO₂ in soil; ms, ambient in sucrose media. (Top) Letter a or h indicates conditions during seed development. (F) Germination rates of WT seeds and T1 Δβ-ca1, Δβ-ca5, and Δβ-ca1ca5 seeds in soil at 415 ppm CO₂. Each point represents 31 seeds placed in soil (for C–E: h = 9,000 ppm CO₂, a = 415 ppm CO₂; ms, sucrose media). (G) WT and T1 Δβ-ca1ca5 seedlings (from T0 grown in ambient), of the same age postsowing, on 3% sucrose media in 415 ppm CO₂. Data for C and D come from 1,161 WT seeds, 838 high-CO₂ Δβ-ca1ca5 seeds, and 238 ambient-CO₂ Δβ-ca1ca5 seeds

Seed Development and Germination Is Severely Affected in Δβ-ca1ca5.

Given that bicarbonate is required for fatty acid synthesis, a reduction in its availability due to the absence of stromal CA would be expected to affect synthesis of lipids in embryos and seeds. Earlier, Hoang and Chapman (15) reported that inhibitors of CA affected incorporation of acetyl into lipids in cotton embryos. We investigated the effect of a complete absence of CA in the tobacco mutant on seed filling and germination.

Seeds produced by Δβ-ca1ca5 in ambient CO₂ had significant morphological alterations and often contained large indentations in their coats, while the morphological phenotype of seeds from high CO₂-grown Δβ-ca1ca5 plants is similar to WT (Fig. 4B). The high CO₂ Δβ-ca1ca5 plants produced heavier seeds, averaging 0.068 mg/seed compared to a mass of 0.038 mg/seed produced from plants in ambient CO₂ (Fig. 4C). In both atmospheric conditions, however, the mutant seeds weighed less than WT seeds (P = 0.00002 and P = 0.0007 for low and high CO₂ respectively, t test), which had an average mass of 0.081 mg, indicating the 9,000 ppm CO₂ growth could not completely restore biosynthetic reactions utilizing bicarbonate. To observe whether both mass and seed size differed, the two-dimensional areas of WT and Δβ-ca1ca5 seeds were measured. The seeds from ambient CO₂-grown mutants had an average area of 0.29 mm², which was a statistically significant difference compared to the WT seed area of 0.32 mm² (Fig. 4D, P = 0.0002, t test). There was no significant difference in the average area of WT seeds 0.32 mm² compared to the seeds of 9,000 ppm CO₂-grown Δβ-ca1ca5 plants, at 0.32 mm² and 0.31 mm², respectively (Fig. 4D, P = 0.52).

The germination rates of T1 seeds from the two single knockout lines, Δβ-ca1 and Δβ-ca5, and the double mutant Δβ-ca1ca5, all produced in plants grown at ambient CO₂ (as in Fig. 3 D and E), were measured in soil under ambient CO₂ concentrations and a long day (16 h) photoperiod (Fig. 4 E and F). After 7 d, 97% of Δβ-ca1 seeds and 100% of WT seeds germinated. Germination of seeds from Δβ-ca5 was delayed, with 74% of seeds germinating by day 7 and 90% by day 13. Only 3% of Δβ-ca1ca5 seeds germinated after 7 d; 15 d were required for the seeds to reach a germination rate of 29%.

Seeds from Δβ-ca1ca5 plants produced in high CO₂ resulted in substantial improvement in seed germination rates, regardless of the CO₂ concentration in which the seeds were sowed. When grown on soil, these seeds had a 72% germination rate at ambient CO₂ and a 67% germination rate at high CO₂ (Fig. 4E). Seeds produced by the Δβ-ca1ca5 plant grown at ambient CO₂ had germination rates of 29% and 22% when sown in soil and placed in 415 ppm CO₂ and 9,000 ppm CO₂ concentrations, respectively. When placed on 3% sucrose media at 415 ppm CO₂, the seeds from low CO₂–grown Δβ-ca1ca5 plants had a germination rate of 28% (Fig. 4E). Thus, the growth conditions of the T0 Δβ-ca1ca5 parent had a strong influence on the germination rate of the ensuing seeds. Restoration of bicarbonate amount at 9,000 ppm evidently provided sufficient substrate for biosynthesis in developing embryos and seeds.

The T1 Δβ-ca1 and Δβ-ca5 seedlings continued to develop normally under ambient CO₂ after germination (like their T0 parents). In contrast, the growth and development of T1 Δβ-ca1ca5 seedlings in ambient CO₂ arrested shortly after germination (Fig. 4G), irrespective of whether seeds developed in high CO₂ or ambient. Notably, the Δβ-ca1ca5 seedlings (Fig. 4G) failed to produce true leaves at ambient CO₂ even when grown on media containing 3% sucrose.

CA Mutants Show No Difference in Photosynthetic Capacity.

Chloroplast CA has long been hypothesized to play a role in provision of dissolved CO₂ for RuBisCO. If true, then such a role would likely be most pronounced when demand for CO₂ is highest, during photosynthesis at saturating light. To study the effect of removing CA from the chloroplast stroma, we imaged in vivo chlorophyll fluorescence from whole plants as leaves made the transition from limiting (7 µmol photons ⋅ m⁻² ⋅ s⁻¹) to saturating (790 µmol photons ⋅ m⁻² ⋅ s⁻¹) light. Because leaf development in the Δβ-ca1ca5 double mutant is severely compromised, all measurements were made on intact, fully expanded leaves from plants grown at 9,000 ppm CO₂ and transferred to ambient CO₂ for 24 h before measurement. Φ_PSII, the efficiency with which absorbed photons are used to drive photosystem (PS) II electron transport, was measured at 1-min intervals beginning 10 s after the transition to saturating light. We observed that the kinetics of changes in Φ_PSII following the transition from limiting to saturating light (Fig. 5A) were nearly identical in WT, Δβ-ca1, and Δβ-ca1ca5 mutants, indicating that photosynthesis was not impaired in these mutants compared to WT. The values of Φ_PSII were consistently about 10% lower in the Δβ-ca5 mutant than in the other three genotypes.

Fig. 5.

Photosynthetic measurements. (A) Kinetics of Φ_PSII changes in response to a transition from steady-state low light (7 μmol photons ⋅ m⁻² ⋅ s⁻¹) to saturating light (790 μmol photons ⋅ m⁻² ⋅ s⁻¹). Data are averages over 18 to 60 areas of interest per genotype; fits are to a single exponential rise (r² = 0.987, 0.951, 0.969, and 0.978 for WT, Δβ-ca1, Δβ-ca5 and Δβ-ca1ca5 respectively). x-axis is minutes after transitioning to a saturating light intensity. (B) Left axis: PSII and NPQ efficiency (phi) between WT and mutant tobacco lines grown in high (9,000 ppm) CO₂ for 6 wk. The reduction in NPQ in the double mutant Δβ-ca1ca5 when compared to WT tobacco is significant (*P = 0.02, Student's t test). Right axis: Fv/Fm values show no significant difference between the four lines. The nominal Fv/Fm value of 0.8 is shown as a dotted line. Error bars: SD. (C and D) A/C_i curves measured at 1% O₂ and ambient O₂. For 1% O₂, data are from one leaf from four different plants in each genotype. Ambient data are from three plants for each WT and double mutant.

Under optimal photosynthetic conditions, there is a complementary relationship between the yields of photochemical (Φ_PSII) and nonphotochemical (Φ_NPQ) quenching, with the sum of the two being equivalent to the dark-adapted value of F_v/F_m (25). While there were no significant differences among Φ_PSII values measured at light saturation between the mutants and WT tobacco, there was a significant decrease in Φ_NPQ in the Δβ-ca1ca5 double mutant (Fig. 5B).

Φ_PSII measures the fraction of absorbed photons that are used to drive total electron transport through PS II. Although a useful indicator of PS II activity and total photosynthesis, it is subject to two types of errors. In all plants, measurement of Φ_PSII assumes that all incident photons are absorbed or that at least a similar fraction of incident photons are absorbed among the samples being compared. In C₃ plants, there is the further complication that a significant fraction of electron flow through PS II can be due to photorespiration rather than net CO₂ fixation, particularly under light-saturated conditions (26). To eliminate these potential errors, we measured CO₂ uptake under saturating light with O₂ reduced to 1%, which eliminates photorespiratory losses without affecting photosynthesis (27). The assimilation rate (A) versus internal CO₂ (C_i) curves exhibit very similar behavior for WT and the Δβ-ca1 and Δβ-ca1ca5 mutants (Fig. 5C). This relationship was also observed at ambient O₂ (Fig. 5D). Importantly, A/C_i measurements did not distinguish WT and the double mutant at either ambient or saturating CO₂ (Fig. 5D and SI Appendix, Fig. S5), indicating that photosynthesis was not affected when CA activity was eliminated from the chloroplast stroma.

pH Is Increased in Δβ-ca1ca5 Chloroplasts.

In humans, CA plays a crucial role in the blood’s bicarbonate buffer (28), and while plant CAs have been hypothesized to play a role in regulating the pH of the chloroplast stroma (29), the level of bicarbonate at equilibrium in the stroma might be too low (∼1 mM at equilibrium) for it to act as an effective buffer system (2). To investigate possible changes in pH in the CA mutant lines, the pH-sensitive GFP pHluorin2 was altered with the addition of a chloroplast transit peptide (from the Arabidopsis RecA protein) to its N terminus (SI Appendix, Fig. S6). This construct was then transiently expressed in WT and mutant CA plants grown at ambient CO₂ and observed to localize in chloroplasts (Fig. 6B). While the exact pH cannot be determined with this method, it can give insights into the relative stromal pH of the mutants compared to the WT, which is expected to have a stromal pH of ∼8 (30).

Fig. 6.

Chloroplast stromal pH. (A) Ratio (colored bars) of RecAcTP-pHlourin2 signals excited by 405-nm (light gray) and 488-nm (dark gray) lasers measured on a confocal microscope. Emission signals measured in relative fluorescent units ranging from 0 to 255. Bars: SE. (B) Confocal images of chloroplast-targeted pHlourin2 in transformed WT tobacco. 633 nm: Chlorophyll A autofluorescence. 405 nm: pHlourin2. 488 nm: pHlourin2. (Scale bars: 10 µm.)

The ratio between the emission signals of pHluorin2 excited with the 405-nm laser and the 488-nm laser (405/488) was used to determine the relative pH (31). A high 405 nm/488 nm ratio observed in WT chloroplasts indicates a basic environment, which is consistent with the predicted WT stromal pH of about 8 (Fig. 6A). The pH of the single mutant lines Δβ-ca1 and Δβ-ca5 were similar to WT, but Δβ-ca1ca5 showed an increase in the 405/488 ratio, indicating a higher pH and more basic cellular environment in the chloroplast stroma of the mutant growing in ambient CO₂ than WT (Fig. 6A). The altered pH could have an effect on catalysis by enzymes whose pH optimum is less basic, further disturbing biosynthesis.

ROS Production Is Increased in Δβ-ca1ca5 Leaves.

We assayed the level of ROS in WT and mutant lines (Fig. 7A). Previous studies have shown plant CAs to play a role in preventing the creation of ROS (10, 14). Leaves from the first node of tobacco plants grown in ambient CO₂ were homogenized, and the homogenate was suspended in a 5-µM CellROX Green solution. CellROX Green undergoes a conformational change when oxidized, resulting in increased fluorescence. Δβ-ca1ca5 leaves displayed a greatly increased fluorescent signal in comparison to leaves of Δβ-ca1, Δβ-ca5, and WT (Fig. 7A). There was also a slight increase in the fluorescent signal in Δβ-ca5 over Δβ-ca1 and WT. The increased ROS and resultant cellular damage may play a role in the abnormal leaf phenotype of the double mutant grown at ambient CO₂.

Fig. 7.

ROS production. (A) Spectrofluorometer emission spectra of CellROX Green–treated homogenized leaf samples. Signal intensity is in relative fluorescent units (RFU). (B) Confocal imaging of WT and mutant leaves grown in ambient CO₂ and treated with 5 μM CellROX Green. Magenta, chlorophyll autofluorescence; green, CellROX Green (Scale bars: 10 µm.) Leaves assayed are from the first node of 10-wk-old tobacco plants grown for 8 wk at 9,000 ppm CO₂ and 2 wk at ambient CO₂.

The location of the ROS signals was determined in the mature leaves that had fully developed on plants grown in 9,000 ppm CO₂ and were then transferred to ambient CO₂ for 1 wk. The ROS signals found in Δβ-ca1ca5 mesophyll cells appear to be primarily associated with the chloroplast (Fig. 7B), indicating the origins of the increased ROS measured in the Δβ-ca1ca5 plant (Fig. 7A).

Complementation of Δβ-ca1ca5 Mutants.

To confirm that the phenotypes observed in Δβ-ca1ca5 mutants were not a result of off-target mutations caused by the CRISPR/Cas9 system, a series of complementation experiments were performed. For these experiments, Δβ-ca1ca5 T1 plants lacking the Cas9 transgene were identified (SI Appendix, Fig. S7). Transgene constructs in which the β-CA1 CDS and β-CA5 CDS were put under the control of the 35S promoter were transformed into cas9⁻ Δβ-ca1ca5 T1 plants (Fig. 8A). Both Δβ-ca1ca5_35S::β-CA1 and Δβ-ca1ca5_35S::β-CA5 complemented the mutant phenotype and produced plants with a gross morphology comparable to WT at ambient CO₂ concentrations (Fig. 8B). This result establishes that the phenotype observed in Δβ-ca1ca5 tobacco is caused by lack of stromal CA activity and not by an off-target mutation.

Fig. 8.

Complementation of CRISPR-generated Δβ-ca1ca5 mutant with different forms of CA proteins. (A) Construct diagrams for complementation of Δβ-ca1ca5 double mutant tobacco. cTP: chloroplast transit peptide (62 amino acids); TERM: terminator. (B) Morphology of plants grown in high CO₂ then transferred to ambient CO₂. (C,Top) Ponceau stain showing the RuBisCO large subunit signal. (Bottom) Immunoblot of WT, mutant, and complemented lines leaf protein extracts with anti-CA1 antibody. (D) CA activity assay of WT, Δβ-ca1ca5, and complemented lines (*CA assay done on the T0 generation). Each assay was performed with four samples. (Bars: SE.) (E) Spectrofluorometer emission spectra of CellROX Green–treated homogenized leaf samples. Signal intensity is in relative fluorescent units (RFU). (F) Inoculation of TMV onto fully expanded leaves of WT and transgenic lines.

The WT phenotype could not be restored to cas9⁻Δβ-ca1ca5 plants expressing altered forms of the β-CA1 enzyme. We generated a construct in which two of the residues that bind zinc in β-CA1 were mutated to adenine (C152A and D152A; shortened hereafter as ΔZn) to make a catalytically inactive form of β-CA1 (Fig. 8A). The zinc ion located at the active site cavity of β-CAs is essential for the nucleophilicity of the enzyme and thus its catalytic activity (32). We also generated a construct in which the chloroplast transit peptide (identified as the first 62 amino acids) was removed from β-CA1. ΔZn-βCA1 and Δ62-βCA1 were both driven by a 35S overexpression promoter and transformed into cas9⁻Δβ-ca1ca5 plants. Neither of these constructs resulted in transgenic plants with normal morphology; instead, the plants showed the characteristic, necrosis-like phenotype at ambient CO₂ (Fig. 8B).

An immunoblot of extracted proteins using an anti-CA1 antibody showed that the mutant and full-length forms of β-CA1 were present in the three transgenic lines used in the complementation experiments (Fig. 8C), indicated by a positive band at ∼28 kDA in leaf homogenate of WT, Δβ-ca5, Δβ-ca1ca5_35S::β-CA1, Δβ-ca1ca5_35S::Δ62-βCA1, and Δβ-ca1ca5_35S::ΔZn-βCA1 tobacco plants.

CA activity assays determined that cas9⁻Δβ-ca1ca5 plants expressing the cytoplasm-localized βCA1 (Δ62-βCA1) and full-length β-CA1 or β-CA5 had substantially increased CA activity compared to the double mutant (Fig. 8D), signifying that Δ62-βCA1 is still catalytically active. Both Δβ-ca1ca5_35S::Δ62-βCA1 and Δβ-ca1ca5_35S::ΔZn-βCA1 had heightened ROS levels like Δβ-ca1ca5, while Δβ-ca1ca5_35S::β-CA1 showed a decrease in ROS production (Fig. 8E).

As previously reported (10), β-CA1 is an SA-binding protein that plays a role in the hypersensitive defense response of tobacco. Mature leaves of WT and CA mutant lines, grown at high CO₂, were treated with 5 µM suspended tobacco mosaic virus (TMV) and left in a growth chamber containing ambient CO₂. After 3 d, WT and Δβ-ca5 plants displayed the characteristic necrosis at the site of infection, indicating an HR (Fig. 8F). As expected from previous work (10, 11), neither Δβ-ca1 nor Δβ-ca1ca5 produced an HR when inoculated with TMV (Fig. 8F). However, Δβ-ca1ca5_35S::ΔZn-βCA1 plants display the characteristic programed cell death of an HR despite their minimal CA activity levels (Fig. 8F). This result suggests that ΔZn-βCA1 is still able to bind to SA and contribute to plant defense even with a mutated catalytic center. These experiments indicate that both the location of β-CA1 (in the chloroplast stroma) and its catalytic activity are necessary to complement the Δβ-ca1ca5 phenotype. In contrast, an inactive β-CA1, while not restoring WT phenotype, can elicit the HR.

Discussion

Previous experiments that studied single knockouts of β-CA1 did not eliminate all CA enzymatic activity from the chloroplast stroma because no lines with both stromal CAs mutated were examined (2, 10, 16). In Arabidopsis, a slow growth and sterility phenotype was observed when β-CA5 alone is knocked out (11), suggesting that activity of the other chloroplast CAs in Arabidopsis cannot fully compensate for the absence of β-CA5 during development. In contrast, the single mutants in tobacco exhibited normal phenotype. However, eliminating expression of both chloroplast stromal CAs resulted in a striking defect in leaf and floral bud development as well as impaired seed formation. This report presents the complete disruption of CA activity in the chloroplast stroma and provides insight into its function in C₃ plants.

The role of CAs in C₃ plant photosynthesis has not been established, unlike their role in the C₄ CO₂-concentrating mechanism and photosynthesis (7). The majority of CA activity in C4 plants is present in the cytosol of mesophyll cells (6), and little or no activity is present in bundle sheath cells (33), the site of carbon fixation. C₄ plants with reduced cytosolic CA activity show reduced photosynthetic assimilation, especially at decreasing CO₂ levels (34). In C₃ plants, the dominant CA activity is present in stroma, and it was postulated that removal of stromal CAs will significantly reduce mesophyll conductance (35). In previous measurements on an antisense-CA1 mutant in tobacco (16), a 99% reduction in CA activity showed no measurable reduction in RuBisCO activity nor CO₂ uptake, and the plants exhibited normal development. However, because of the high catalytic rate of CA, even a small amount of CA activity could possibly provide sufficient turnover and result in the lack of any morphological phenotype. To further clarify the role of CA in C₃ photosynthesis, we measured CO₂ fixation rates under ambient (21% O₂) and nonphotorespiratory conditions (1% O₂) and found no significant difference in assimilation rate at either ambient or saturating CO₂ in the double mutants compared to WT (Figs. 5 C and D and SI Appendix, Fig. S5). Likewise, measurements of Φ_PSII under ambient CO₂ and O₂ conditions showed no significant differences between WT and the Δβ-ca1 and Δβ-ca1ca5 mutants (Fig. 5A). Together these data indicate that stromal CA is not required to provide a sufficient flux of CO₂ from the atmosphere to RuBisCO under either light-saturated steady-state or transient (low light to saturating light) conditions. While we cannot rule out that stromal CA does not affect photosynthetic efficiency in every possible environmental condition, we can conclude that its primary role is not in enhancing carbon fixation.

The similarity between CO₂ fixation rates and Φ_PSII measurements between WT and the mutants at ambient O₂ also suggests that photorespiratory losses are small under the experimental conditions and are not influenced by the removal of stromal CA activity.

Even though a complete lack of chloroplast CA activity has not been previously reported in plants, absence of CA activity has been reported in bacterial (36–38) and yeast mutants (39, 40). Typically, the CA-less mutants grow poorly at ambient CO₂ levels, but the phenotype can be alleviated by using high CO₂ levels (∼5%) or complementing with heterologous CA. The poor growth phenotype in microorganisms is known to be caused by deficiency of bicarbonate, which is an important precursor for carboxylation reactions in several biosynthetic pathways (2, 36, 41). In a fast-growing microbe like Escherichia coli, it was calculated that the bicarbonate requirement is 10³- to 10⁴-fold greater than what could be provided by uncatalyzed hydration of CO₂ (36). In plants, chloroplasts are the primary site for many of these pathways, leading to the logical speculation that CAs are essential to support nonphotosynthetic carboxylation pathways, which generally have a high K_m for C_i at current ambient CO₂ levels (2). It was calculated that even with CAs, such pathways are operating at about 0.5 to 17% of HCO₃⁻ saturation, which would make such reactions staggeringly slow in the absence of CA and with competition for CO₂ by RuBisCo (2). As the uptake of bicarbonate by chloroplasts is four orders of magnitude lower than CO₂ diffusion, as shown by a study with intact chloroplasts (17), and no bicarbonate transporters have been functionally characterized in the chloroplast envelope (42), the pathways requiring bicarbonate in chloroplasts are highly dependent on CA. The mutant phenotype of Δβ-ca1ca5 leaves grown at ambient CO₂ (Fig. 3A) is lost when the leaves are allowed to fully expand under high (9,000 ppm) CO₂ (Fig. 3B) in a manner similar to CA-free unicellular organisms. Note that growth under high CO₂ will, at equilibrium, elevate the stromal CO₂ and bicarbonate levels by about a factor of 20, possibly increasing the rate of formation and concentration of bicarbonate enough to support bicarbonate-dependent carboxylation pathways.

A model for the movement of inorganic carbon from the atmosphere to RuBisCO in C₃ plants is shown in Fig. 9, showing the two theoretical alternatives for moving inorganic C across the chloroplast envelope membranes. Pathway (1) is the direct pathway in which CO₂ crosses the envelope either by passive or facilitated diffusion; aquaporins have been shown to contribute to CO₂ transport into the chloroplast (17). Pathway (2) would utilize passive or active transport of bicarbonate into the chloroplast, with cytoplasmic and stromal CAs catalyzing the interconversion of dissolved CO₂ and bicarbonate. However, there is no evidence for bicarbonate transport into vascular plant chloroplasts (42). As inorganic carbon is not transported in bicarbonate form, removal of CAs should have minimal impact on mesophyll conductance (g_m). We estimated g_m from the assimilation data using the calculator described by Sharkey (43). Although this method gives assimilation-weighted estimates, it is useful in comparing plants in an experiment, and we found no significant difference in estimated g_m values of WT and Δβ-ca1ca5 (SI Appendix, Fig. S5C). Within the chloroplast, we propose that CAs provide HCO₃⁻ for consumption by bicarbonate-dependent carboxylation pathways by maintaining HCO₃⁻ at equilibrium concentration.

Fig. 9.

A simple model for the movement of inorganic C from the atmosphere to RuBisCO in C₃ photosynthesis. CO_{2 (g)}: gaseous (atmospheric) CO₂; CO_{2 (d)}: dissolved CO₂. 1: passive or facilitated diffusion of CO₂ across the chloroplast envelope membranes; 2: facilitated diffusion or active transport of bicarbonate across the chloroplast envelope membranes.

In a detailed study of a yeast CA-knockout line (41), the authors attempted to complement specific bicarbonate-dependent pathways by nutritional supplementation. Even though some specific supplementations improved growth, a full nutritional supplementation was not able to completely restore the growth of the CA mutant to the level of the reference strain. The authors suggested that the failure to rescue growth resulted from the inability of the yeast to take up sufficient fatty acids. Similar experiments cannot be performed with plant cells, given the well-known lack of uptake of many types of exogenously supplied nutrients from cell culture. Likewise, a large-scale analysis of metabolite differences between the chloroplast CA-less mutant and WT would be predicted to produce a bewildering array of changes due to the large number of pathways affected by limitation of bicarbonate.

Nevertheless, some information about disrupted metabolism following limiting of CA and bicarbonate has been obtained through analysis of fatty acids and lipids. Metabolites whose concentration could be affected by inadequate bicarbonate include malonyl-CoA and its derivatives, which are used in the synthesis of lipids (44). Malonyl-CoA and fatty acid synthesis depend on the enzyme acetyl-CoA carboxylase (K_m 0.9 to 2.5 mM), which uses bicarbonate to carboxylate its substrate (2, 45–47). When CA activity in tobacco leaves is suppressed using inhibitors (like ethoxyzolamide) or antisense RNA silencing, the incorporation of acetyl into total lipids is greatly reduced (15). We observed alterations in the FFA profiles of young WT and Δβ-ca1ca5 leaves (Fig. 4A and SI Appendix, Fig. S4A). The double mutant had a significant reduction in the amount of 16:0 and 18:0 FFA, both of which arise in the chloroplast before being elongated and further modified in other organelles. Previously, it had been observed that suppressing CA activity resulted in improper incorporation of acetyl into lipids of developing cotton (Gossypium hirsutum) embryos and an increase in seedling lethality in Arabidopsis (15, 48). We detected morphological defects and a similar drop in the germination of seeds in the Δβ-ca1ca5 double mutant (Fig. 4). Importantly, Hoang and Chapman (15) demonstrated that inhibition of plastid CA activity resulted in the reduced incorporation of acetyl into lipids in antisense CA tobacco lines.

Further evidence for the need of CAs for biosynthesis comes from considering the disruption of the sink-to-source transition in the double mutant. Normal leaf development involves a metabolic transition of the leaf from sink to source; this transition does not occur uniformly in the leaf but begins at the tip and ends at the base of the leaf (49). In parallel with this developmental pattern, we observe that when developing Δβ-ca1ca5, leaves are moved from high CO₂ to ambient CO₂, areas of the leaf closest to the tip do not display the mutant phenotype, while areas closer to the leaf base develop the characteristic lesions (Fig. 3 C, blue arrow and D). Interestingly, leaves that develop completely at ambient CO₂ show a different pattern in which necrosis begins at the tip and spreads to the base (Fig. 3G). Thus, the part of an expanding leaf that has completed the sink-to-source transition (closer to the tip) at the time it is transferred from high to ambient CO₂ is less likely to exhibit the mutant phenotype than the base of the leaf, which has not completed this transition.

We observed a rise in chloroplast ROS, which may be a contributor to the double mutant’s leaf phenotype when grown at ambient CO₂. The increase in ROS in the double mutant is consistent with our observation of a significant decrease in F_v/F_m and in the yield of nonphotochemical quenching (Φ_NPQ) in the Δβ-ca1ca5 mutant (Fig. 5B). Under most conditions, the yields of photochemical and nonphotochemical quenching are complementary—for every incremental change in Φ_PSII, there is a complementary change in Φ_NPQ (50). A breakdown in this complementarity, as seen in the double mutant, necessarily leads to increased production of ROS. Although newly budded leaves of the Δβ-ca1ca5 double mutant in ambient CO₂ appear like WT in gross morphology, the leaf does exhibit higher ROS localized to the chloroplast (Fig. 7 A and B), which may set the pathway to necrosis in motion.

β-CA1 has been shown to bind the plant hormone SA (10) and is involved in the perception of SA levels in plants (11). When Slaymaker et al. (10) silenced β-CA1 in N. tabacum, they observed an absence of an HR when the leaves were transiently transformed by Agrobacterium expressing Pto and avrPto. We also observed a lack of HR in Δβ-ca1 and Δβ-ca1ca5 plants (Fig. 8F). Even though Δβ-ca1ca5 plants expressing a zinc-binding site mutated β-CA1 (ΔZn-βCA1) were unable to complement the developmental mutant phenotype at ambient CO₂ concentrations (Fig. 8B), expression of the mutant protein was able to complement HR defect (Fig. 8 C and F). This finding implies that the protein’s ability to bind SA has not been compromised by the zinc-binding site mutations at its catalytic core. These plants can induce programmed cell death in response to a pathogen but still demonstrate leaf necrosis at ambient CO₂, which indicates that the ΔZn-βCA1 has no CA activity. Thus, chloroplast CA activity can be removed from a plant without removing its ability to respond to viruses through an HR.

Inorganic carbon concentrations at ambient conditions are not expected to play a significant role in pH buffering in the chloroplast (2). While the pH measurements we performed were not quantitative, they do provide insight into the relative sizes of the bicarbonate and CO₂ pools when the absence of stromal CA prevents rapid equilibration of the two pools. The Δβ-ca1ca5 mutant showed an increased ratio between the 405-nm and 488-nm emission signals, indicating a more basic stroma compared to WT and the single mutants Δβ-ca1 and Δβ-ca5 (Fig. 5A). Our observation that photosynthesis is unaffected in the Δβ-ca1ca5 mutant suggests that any disruption of stromal pH in the mutant is not sufficient to alter the rate of photosynthesis.

In summary, CRISPR-generated knockout lines in tobacco demonstrate that CA activity in the chloroplast stroma is required for normal plant and seed development at ambient CO₂ concentrations but not for supplying CO₂ for photosynthesis. Removing CA activity from the chloroplast stroma results in large necrotic lesions in developing leaf tissue and reduced germination of seeds following growth in ambient CO₂. Both mutant phenotypes can be prevented by growing Δβ-ca1ca5 plants in high CO₂ concentrations of 9,000 ppm. These phenotypes and their relief in high CO₂ point to underlying defects in metabolism, which is starvation of bicarbonate-dependent enzymes in absence of CA activity. We show that such a defect is also accompanied by elevated levels of ROS, which may play a role in the manifestation of necrotic lesions.

Typically, algal and bacterial systems that rely on HCO₃⁻ transporter–based CO₂-concentrating mechanisms either have highly regulated CA activity in stroma or no CAs in the cytoplasm, respectively. Schemes to produce a functional carboxysome or artificial pyrenoid in chloroplasts propose the incorporation of bicarbonate transporters on the chloroplast envelope membrane along with the removal of stromal CAs (18, 51, 52). We provide evidence that the HR could be maintained in plants engineered to lack CA, provided that a catalytically inactive CA is incorporated. CA catalytic activity is not necessary for photosynthesis but to provide bicarbonate for biosynthetic pathways, which would be rendered functional by a bicarbonate transporter. Thus, our work supports the feasibility of an important step in the engineering of a CO₂-concentrating mechanism into a C₃ plant.

Materials and Methods

Transformation Constructs to Produce YFP Fusion for CA Localization or Complement Mutants.

The TargetP online prediction program (53) was used to identify candidate CA proteins that might be located in the chloroplast. Production of complementary DNA, cloning into vectors with YFP at the carboxyl terminus, and transient expression are described in SI Appendix, Supplementary Materials and Methods, where details of cloning of complementation constructs can also be found.

Design and Assembly of CRISPR/Cas9 Constructs.

The CCTop online program was used to identify and select CRISPR/Cas9 sgRNA target sites in the β-ca1 and β-ca5 genes (54). Further details can be found in SI Appendix, Supplementary Materials and Methods.

Agrobacterium-Mediated Transformation of Tobacco.

Transformation of tobacco generally followed the protocol described in Sparkes et al. (55). Further information can be found in SI Appendix, Supplementary Materials and Methods.

Production of Anti-CA1 Polyclonal Antisera.

The peptide SLPADGSESTAFIEC was synthesized and conjugated to keyhole limpet hemocyanin and used to immunize two rabbits by Pocono Rabbit Farm and Laboratory, Inc. One of the protein A–purified antisera reacted with proteins from WT on immunoblot at the expected molecular mass for β-CA1 and did not detect a protein in Δβ-ca1 or Δβ-ca1ca5 tissue but did detect β-CA1 in Δβ-ca5 plants, as expected.

Growth Chamber Conditions.

Plants were grown under 100 μmol/m²/s photosynthetically active radiation with a 16-h photoperiod in ambient and high (9,000 ppm) CO₂ growth chambers. Growth room temperatures were maintained at an average of 22 °C with 50% relative humidity during daylight conditions for the ambient CO₂ chamber and 64% relative humidity for the high CO₂ chamber. Plants were germinated and grown in pots containing LM-111 All Purpose Mix (Lambert), watered daily in the morning by hand, and kept in free-draining flats.

CA Activity Assay.

The CA activity assay was carried out as previously described (11) with slight modifications. Further information can be found in SI Appendix, Supplementary Materials and Methods.

ROS Assay.

Tissue was labeled with the fluorescent dye CellROX Green (Thermofisher Scientific, Catalog No. C10444) and examined in a spectrofluorometer and a confocal microscope. Further information can be found in SI Appendix, Supplementary Materials and Methods.

pH Measurements with pHlourin2.

The plasmid pME-pHluorin2 was a gift from David Raible, University of Washington, Seattle, WA (Addgene plasmid No. 73794; http://n2t.net/addgene:73794; RRID: Addgene_73794) (31, 56). The pHlourin2 protein sequence was altered with the addition of the chloroplast transit peptide from the Arabidopsis RecA DNA recombination family protein (Genbank accession number NM_001084375.1) to its N terminus.

Chlorophyll Fluorescence and Gas Exchange Measurements.

Chlorophyll fluorescence was measured on intact plants using WALZ Imaging PAM M-series chlorophyll fluorometer (Heinz Walz GmbH). The LI-6800 Portable Photosynthesis System (Li-Cor Biosciences) was used for assay of net CO₂ fixation and internal (leaf) CO₂ concentration.

Assay of Leaf FFAs.

Young leaves from the first three nodes of WT tobacco and the Δβ-ca1ca5 mutant were ground into powder in liquid nitrogen, extracted, and assayed by mass spectrometry as described in SI Appendix, Supplementary Materials and Methods.

TMV Inoculation

Inoculation with TMV was performed similarly to the method of Guo et al. (57) as described in SI Appendix, Supplementary Materials and Methods.

Data Availability

DNA sequence data have been deposited in GenBank (accession no. MN153507.1). All study other data are included in the article and/or SI Appendix.