Reduction of plastid-localized carbonic anhydrase activity results in reduced Arabidopsis seedling survivorship.

Carbonic anhydrase (CA; EC 4.2.1.1) catalyzes the interconversion of CO2 and HCO3(-) and is a major protein constituent of the C3 higher plant chloroplast where it is presumed to play a role in photosynthetic carbon assimilation. In this study, we have used both RNA antisense and gene knockout lines to specifically reduce the activity of the chloroplast betaCA1 polypeptide (At3g01500) in the model plant Arabidopsis (Arabidopsis thaliana). Although able to germinate, seedling establishment of transgenic plants is significantly reduced relative to wild-type plants when grown at ambient levels of CO2. Growth at elevated (1,500 microL L(-1)) CO2 or on plates supplemented with sucrose restores seedling establishment rates to wild-type levels. Seed from wild-type and transgenic plants exhibited no significant differences in seed protein, lipid content, or reserve mobilization during seedling growth. betaCA1-deficient seedlings do, however, exhibit reduced capacity for light-dependent 14CO2 assimilation prior to the development of true leaves. The small number of surviving seedlings able to grow and develop are phenotypically similar to wild-type plants, even when subsequently grown at subambient levels of CO2. Microarray analysis of mature leaves of betaCA1-deficient plants shows some differences in transcript abundance, particularly with genes involved in ethylene signaling and response. The data suggest that reduced levels of seedling establishment by betaCA1-deficient plants could be the result of poor cotyledon photosynthetic performance at the onset of phototrophic growth and prior to the development of true leaves.

Carbonic anhydrase (CA; EC 4.2.1.1) is a zinc metalloenzyme that catalyzes the reversible hydration of CO 2 . With numerous isoforms, and activity found in all plants, animals, and microorganisms that have been examined, the ubiquity of its distribution implies that it plays diverse but essential roles in many biological processes (Hewett-Emmett, 2000). In terms of biochemistry, the specific requirement for CA activity is apparent. Although the uncatalyzed hydration/dehydration reactions occur, they can be sufficiently slow at physiological pH values and temperatures that the rate of interconversion limits enzymatic reactions and transport processes that require a specific inorganic species as a substrate. In addition, although equilibrium concentrations of inorganic carbon species are established by the pH, catalysis to equilibrium by CA ensures that diffusion of CO 2 across membranes, and diffusion of all inorganic carbon species within the aqueous cellular environment, is not limited by flux rates between species.
The centrality of CA in many biological processes is also demonstrated by the existence of at least five distinct CA orthologs (designated a, b, g, d, and e) of this enzyme, potentially representing independent evolutionary origins. Although four of these proteins have type isoforms that have been associated with a specific group of organisms (a, vertebrates; b, prokaryotes; g, archaebacteria; and e, chemilithotrophs), genomic analyses have shown the presence of isoforms of more than one CA ortholog within a single organism (for example, in the cyanobacteria [Soltes-Rak et al., 1997] as well as higher plant species). In the C 3 plant Arabidopsis (Arabidopsis thaliana), partial genome analysis identified at least 14 genes likely to encode CA isoforms representing the a, b, and g families (Moroney et al., 2001). Most recently, EST and gene expression analyses based on the full genome have shown that three of eight a-type sequences are expressed in Arabidopsis, whereas all six of the b-type sequences are represented in the EST database (Fabre et al., 2007). In addition, Parisi et al. (2004) identified a family of three g-type sequences in Arabidopsis that are all expressed.
In higher plants, little is known about expression and function of aCAs. Reverse transcription-PCR analysis using Arabidopsis RNA identified broad tissuespecific patterns of expression for aCA1 (At3g52720), aCA2 (At2g28210), and aCA3 (At5g04180; Fabre et al., 2007), and aCA1 has been shown to be localized to the chloroplast stroma following transport through the secretory pathway and N-glycosylation (Villarejo et al., 2005). No specific roles for these proteins have been described. In Arabidopsis, members of the gCA family are localized to the mitochondria and appear to play a role in respiration. Disruption of At1g47250, one of three nucleus-encoded gCA isoforms, results in a reduction in the abundance of mitochondrial complex I and supercomplex I and III 2 and a concomitant reduction in respiratory activity (Perales et al., 2005). Most plant research, however, has focused on the characterization and assignment of function for the more abundant b-family isoforms. In C 4 plants, a mesophyll cell-localized bCA is required for the hydration of CO 2 to provide HCO 3 2 for phosphoenolpyruvate carboxylase (PEPCase), the initial carboxylation reaction of this CO 2 -concentrating pathway (Von Caemmerer et al., 2004). Transgenic Flaveria with mesophyll cell CA activities below 10% of wild-type levels had low rates of CO 2 assimilation and grew poorly at ambient levels of CO 2 . In an analogous but nonphotosynthetic role, bCA isoforms have been localized to nodules of legumes where they are presumed to catalyze the synthesis of HCO 3 2 for use by PEPCase in its anaplerotic role as the source of C-4 acids for amination reactions or bacteroid carbon catabolism (Flemetakis et al., 2003). In addition, the synthesis of nodule localized C-4 acids (with the coordinated expression of CA and PEPCase) has also been postulated as part of the mechanism to control gas diffusion within this tissue (Atkins et al., 2001). Other nonphotosynthetic roles for the higher plant enzyme include a requirement for a plastid localized bCA isoform identified in dark-grown cotton (Gossypium hirsutum) seedlings that appears to provide HCO 3 2 for CoA carboxylase in lipogenesis (Hoang et al., 1999). Expression of this CA was correlated with the period of storage lipid accumulation in maturing embryos, and inhibition of CA activity reduced incorporation of radiolabeled acetate into lipids (Hoang and Chapman, 2002).
The majority of CA activity in C 3 higher plants, however, is found in photosynthetic tissue, primarily leaves. In earlier studies, two isoforms of a b-type CA were identified: a highly abundant chloroplast-localized enzyme and a less well-expressed cytosolic form (Kachru and Anderson, 1974;Fett and Coleman, 1994;Rumeau et al., 1996). In Arabidopsis, both forms are nucleus encoded and the chloroplast-localized enzyme (bCA1) contains a transit peptide, which is removed following entry into the plastid, whereas the cytosolic isoform (bCA2) is unprocessed following translation (Fett and Coleman, 1994). Like other chloroplast-localized proteins involved in photosynthesis, the expression of bCA1 is light regulated and/or chloroplast development dependent, whereas the expression of bCA2 is not (Fett and Coleman, 1994). A recent highly detailed study of patterns of expression of the six bCA isoforms in Arabidopsis confirmed the tissue and cellular localization of bCA1 and bCA2, but also showed that expression of other bCA isoforms occurred in above-ground tissue (Fabre et al., 2007). A role for the cytoplasmic localized CA activity in the leaf has not been identified; however, bCA2 (and other cytoplasmic isoforms) may be required for synthesis of HCO 3 2 for use by anaplerotic enzymes, such as PEPCase, to speed the solubilization of gaseous CO 2 from leaf air spaces and to assist in the diffusion of CO 2 /HCO 3 2 across the cytosol to the chloroplast. Proposed photosynthetic roles for the highly abundant stromal enzyme bCA1 have included facilitation of CO 2 movement across the chloroplast envelope and maintenance of maximal rates of CO 2 and HCO 3 2 diffusion through the stroma by rapidly equilibrating C i speciation . Additional roles may include provision of CO 2 to Rubisco by catalyzing the dehydration of HCO 3 2 in the alkaline stroma in close proximity to the principal CO 2 -fixing enzyme. In support of this role, it has been shown that at least a portion of a chloroplast bCA1 protein and activity is associated with the Rubisco-containing Calvin cycle enzyme complex (Jebanathirajah and Coleman, 1998). An additional proposed role for chloroplastic CA activity includes modulation of the stromal pH in which catalyzed CO 2 / HCO 3 2 interconversion could protect against lightinduced pH transients.
In addition to the stromal-localized bCA, several studies also suggest that another CA activity is associated with the thylakoids where it is presumed to provide bicarbonate for PSII activity (Stemler, 1997). In both C 3 and C 4 plants, membrane-bound CA activity, distinct from the stromal bCA1 isoform, can be measured in PSII complex preparations (Lu and Stemler, 2002;Pronina et al., 2002). Data have also been presented that describe CA activity associated with the PSII OEC33 polypeptide from pea (Pisum sativum); however, this protein exhibits no sequence similarity to any known CA family and a specific role for this activity in PSII has not yet been identified (Lu et al., 2005). Recently, an Arabidopsis bCA5-GFP fusion protein was shown to be targeted to chloroplasts, although a specific association with thylakoids was not identified (Fabre et al., 2007).
Attempts to show that the chloroplastic CA activity in C 3 higher plant leaves plays a role in photosynthesis have included both antisense RNA and chemical inhibitor studies in which the effect of reduced leaf CA activity on photosynthesis was examined. Oxygen electrode studies following treatment of leaf segments with the CA inhibitor ethoxyzolamide resulted in a 30% to 50% decline in rates of O 2 evolution at external CO 2 concentrations approaching the compensation point (Badger and Pfanz, 1995). Increasing external CO 2 concentrations reduced the level of inhibition. In contrast, antisense RNA studies in which the foliar levels of CA have been reduced to 1% to 2% of wildtype activity generated mature tobacco (Nicotiana tabacum) plants that displayed almost no discernable phenotype and only a small reduction in photosynthesis Price et al., 1994). Gas exchange measurements indicated that chloroplast CO 2 concen-trations were reduced by approximately 20 mmol mol 21 , a decline that would not have a profound impact on the rate of CO 2 fixation Williams et al., 1996). In an attempt to further examine the role of higher plant CAs in photosynthesis and other metabolic processes, we have extended specific CA activity modification studies to Arabidopsis, the model plant system from which we have previously characterized the dominant stromal chloroplastic and cytosolic CA isoforms (Fett and Coleman, 1994).

RESULTS
Antisense RNA targeted against bCA1 (At3g01500) and isolation of a homozygous bca1 knockout line (Salk_0106570) were both used to specifically reduce the activity of the major foliar, chloroplast-localized isoform of CA in C 3 plants. The gene organization, location of PCR primers used for characterization, and position of the T-DNA insertion in the Arabidopsis bCA1 gene are shown in Figure 1A. Activity measurements identified a number of antisense lines with reduced CA levels, and a single line showing the lowest expression of the bCA1 protein was selected for further study. Western-blot analysis of soluble protein isolated from actively growing, intact 10-d-old seedlings grown on one-half-strength Murashige and Skoog (MS) plates, as well as CA activity measurements, showed that both antisense (bca1-AS) and knockout (bca1) lines had significantly altered the expression of bCA1 (Fig. 1, B and C) with similar reductions of bCA1 expression and total CA activity observed in even younger plants, such as 4-and 6-dold seedlings. As shown in Figure 1, the abundance of 25-kD bCA1 protein is significantly reduced in the antisense line and absent in the knockout line. In contrast, the abundance of the cytosolic 28-kD bCA2 polypeptide (At5g14740; Fett and Coleman, 1994) when compared with wild-type levels is unaffected in these two lines. In concert with the western analysis, measurement of seedling total CA activity showed that residual CA activity in bca1-AS was approximately 30% of wild-type levels, whereas bca1 was ,22%. The disproportionate strength of the bCA2 immunosignal, relative to bCA1 and the activity measurements, reflects the use of bCA2 as the antigen for IgG production. During these experiments, it was readily apparent that the majority of the bCA1 antisense and knockout plants, although able to germinate and initiate cotyledons, failed to develop further. Typically, this phenotype included limited expansion of cotyledons, poor root growth, and no production of true leaves when grown on minimal media and ambient levels of CO 2 ( Fig. 2A). Conversely, plating of seeds on one-half-strength MS plates containing 1% Suc resulted in a significant increase in survivorship, with almost all antisense and knockout seedlings continuing to develop ( Fig. 2A). In addition, growth of the reduced CA activity plants on one-half-strength MS plates at high levels of CO 2 (1,500 mL L 21 ) also resulted in improved survivorship and the full establishment of the seedlings, results similar to that observed with Suc. Growth at these levels of CO 2 does result in some stress, as shown by the cupped, anthocyanin-containing leaves; however, the treatment does allow for continued development of almost all of the transgenic plants. It is also possible to rescue many stagnated air-grown plants with reduced bCA1 activity by transferring them to high levels of CO 2 (Supplemental Fig. S1). Figure 2B shows the percentage survivorship of bca1 plants when germinated at elevated (1,500 mL L 21 ), ambient, and subambient (150 mL L 21 ) levels of CO 2 or with Suc supplement of the one-half-strength MS plates. Both high levels of CO 2 and Suc addition generate survivorship percentages that are very similar to wild-type plants. Exposure to subambient levels of CO 2 , although trending lower, did not decrease significantly survivorship of the bCA1deficient lines. Seedling survivorship of the bca1 knockout line was also affected by the light conditions during growth. Low light levels improved survivorship when compared with wild-type plants, whereas higher light levels resulted in decreasing levels of mutant plant survivorship on the plates. (Fig. 2C).
As the onset of seedling stagnation usually occurred within 4 to 6 d postgermination, the availability and mobilization of seed reserves and/or early photosynthetic capacity may be determining survivorship. SDS-PAGE analysis of dry seed and seedling soluble proteins over the first 5 d following imbibition did not reveal any obvious qualitative or quantitative differences in protein profiles between the bca1 knockout line and wild-type plants (Fig. 3A). Analysis of dry seed total lipid content isolated from equal numbers of seeds did not reveal any significant differences in quality or quantity between the bca1 and wild-type lines (data not shown). In addition, storage lipid catabolism was seemingly unaffected by the reduced CA activity. Levels of eicosenoic acid (20:1), a fatty acid specific to storage triacylglycerol (Lemieux et al., 1990), exhibited similar rates of decline when measured in bca1 and wild-type germinating seed and seedlings (Fig. 3B). Germination and growth of bca1 knockout and antisense plants in the dark, in terms of hypocotyl and root length, were also not significantly different from dark-grown wild-type plants, suggesting that mobilization of reserves is unaffected by lower CA activity (data not shown). Transfer of these plants, however, following 4 d of dark growth to the light for 14 d resulted in significant mortality of the CA activityreduced seedlings when grown on minimal medium (Fig. 3C). The provision of 1% Suc in the plates eliminated seedling mortality under these conditions.
As these data suggested that the mutant phenotype is light dependent, it is possible that the photosynthetic capacity of the seedlings has been affected by the reduction of CA activity. Although photosynthetic gas exchange analysis of very young Arabidopsis seedlings is not easily achieved, the relative capacity for light-dependent CO 2 assimilation, as measured by Figure 2. Survivorship of transgenic lines exhibiting reduced CA activity grown in air, high CO 2 , Suc, and high light. A, Seeds of wild type (A), the null bca1 Salk line (B), and the bCA1 antisense line bca1-AS (C). Arabidopsis seeds were germinated on onehalf-strength MS plates and grown for 10 d at 22°C and under continuous illumination (120 mmol m 22 s 21 ) and at ambient levels of CO 2 (air) or at elevated levels of CO 2 (1,500 mL L 21 ; CO 2 ), or germinated and grown on one-half-strength MS plates supplemented with 1% Suc. B, Survivorship (as measured by ability to fully expand cotyledons, initiate leaves, and continue to grow) of wild-type, bca1-AS, and bca1 Arabidopsis seedlings grown as described above with the addition of seedlings germinated and grown at subambient levels of CO 2 (low CO 2 , 150 mL L 21 ). n 5 3, where for each n, 30 seeds for each genotype were assayed under each growth condition. C, The effect of light intensity on survivorship of wild-type and bca1 seedlings germinated and grown for 10 d at ambient levels of CO 2 , 22°C, and a 24-h photoperiod at the indicated light intensity. n 5 3, where for each n, 30 seeds for each genotype were assayed under each light regime. uptake of 14 CO 2 into acid-stable products at ambient levels of total CO 2 , was assessed (Table I). Although exhibiting similar rates of 14 CO 2 assimilation 2 d postimbibition, 3-and 4-d-old bca1 antisense and knockout seedlings had significantly reduced capacity for 14 CO 2 incorporation when compared with wild-type plants.
Rates of 14 CO 2 fixation in the dark were very low (,10% of light-dependent levels) and were independent of genotype.
It is apparent, however, that not all seedlings with reduced or absent bCA1 activity require Suc or high CO 2 for growth and development. As can be seen in Figure 2A, some of the plants are able to continue growing and develop true leaves and will eventually flower and set seed when grown at air levels of CO 2. As shown in Figure 4, surviving bca1 knockout seedlings grown to maturity at ambient levels of CO 2 expressed no bCA1 protein and had low levels of foliar CA activity. Residual activity is approximately 28% of wild-type levels and presumably represents the activity primarily associated with bCA2 as well as other active isoforms. Phenotypically, the bca1 plants cannot be distinguished from wild-type plants. Growth at less than ambient levels of CO 2 (150 mL L 21 ) does result in a decrease in plant size and developmental delay of flowering in wild-type and bca1 plants; however, no obvious phenotypic differences were noted between bca1 and wild-type lines.
Given the morphological phenotypic similarity of mature bca1 knockout and wild-type plants, transcript profiling was undertaken to determine if there were any identifiable differences in transcript levels between the two genotypes. It was possible that compensatory changes in gene expression, including other CA isoforms, could have resulted in plants that are phenotypically similar to wild type. As shown in Table  II, microarray analysis using mature leaf tissue from air-grown plants showed that expression of other CA isoforms in mature leaves was not modified in response to the specific reduction of bCA1. Assuming comparable levels of hybridization efficiency between the Affymetrix ATH1 chip gene-specific probe sets, the microarray data also show that transcripts for both bCA1 and bCA2 proportionally constitute the majority of CA-specific mRNAs in wild-type leaves when compared with other CA isoforms. In contrast to the other CA-encoding genes, microarray analysis revealed that 128 genes displayed significant changes in transcript levels, with 125 genes expressed in bca1 plants at 0.55fold or less than wild-type plants and three genes expressed at levels of 1.8-fold or higher than those found in wild-type plants (Supplemental Table S1). Given the presumptive role(s) of bCA1 in chloroplast metabolism, it was noted that there were no significant differences between wild-type and bca1 plants in expression of genes known to encode proteins with primary roles in photosynthesis, Calvin cycle activity, or Suc/starch biosynthesis and catabolism. There did seem to be an overrepresentation of significantly down-regulated genes that have been identified as Figure 3. A, Coomassie-stained SDS-PAGE gel of soluble proteins extracted from equal aliquots of dry wild-type and bca1 seeds and 1-, 2-, and 5-d-old seedlings grown on one-half-strength MS plates at 22°C and a 24-h photoperiod at 120 mmol m 22 s 21 . Molecular mass markers (M) are in kilodaltons. B, Eicosenoic acid (20:1) levels extracted from seeds and intact 2-, 3-, and 5-d-old wild-type and bca1 seedlings grown as described above. Data are expressed as a percentage of dry seed totals, where 100% represents 1.89 6 0.33 mg eicosenoic acid per wild-type seed and 2.12 6 0.30 per bca1 seed. C, Phenotype of bca1 and wild-type seedlings on one-half-strength MS plates at 22°C after germination and 4 d dark growth followed by 14 d at 120 mmol m 22 s 21 light intensity and a 24-h photoperiod in the absence (2) or presence (1) of 1% Suc.
Three genes showed significant increases in transcript abundance in the bca1 plants (Supplemental Table S1). These were: At1g69530, encoding an expansinlike protein; the plastid gene AtCg00420, encoding ndhJ; and At4g38840, an auxin-responsive gene. The process by which reduced bCA1 activity increases transcript abundance of these three genes is unclear. In addition, no obvious plant phenotypes were observed to be associated with these transcript changes.

DISCUSSION
Both antisense and gene knockout strategies have been used in this study to significantly reduce or eliminate specifically the chloroplast-localized activity of the bCA1 protein in Arabidopsis. The CA activity remaining in these plants would appear to be primarily associated with the cytosolic isoform bCA2, which was unaffected in these transgenic lines, although other bCA isoforms (or other CAs from the other identified gene families, a and g) would also contribute to the total remaining activity in the leaf. Transcript profiling shows that five identified bCA as well as three gCA and at least one aCA isoforms are expressed in the leaf, in agreement with a recent study where reverse transcription-PCR also identified bCA3 and two additional expressed aCAs as leaf transcripts (Fabre et al., 2007); however, bCA1 and bCA2 represent approximately 75% of the total CA-specific transcripts identified by microarray analysis.
From the data presented in Figure 2, it is apparent that survivorship of the bca1-AS and bca1 Arabidopsis seedlings is significantly lower than wild-type plants. That elevated CO 2 is able to rescue these seedlings indicates that the reduced CA activity is likely responsible for the inadequate provision of either CO 2 or HCO 3 2 for an important metabolic process(es). The ability of Suc to also restore survivorship to wild-type levels could suggest that it is availability or mobilization of carbon reserves in the seedlings that is limiting growth and survivorship. Many Arabidopsis mutants with reduced capacity to mobilize lipid reserves exhibit a Suc-dependent phenotype for postgermination growth and development, particularly in the dark (Penfield et al., 2005). Analysis of both dry seed protein and total lipid content of the transgenic lines did not reveal any significant differences, in terms of both quantity and quality, when compared with wildtype seed, suggesting that the availability of seed reserves was not limiting early growth (Fig. 3). Polypeptide profiles of young seedlings, prior to the arrest of development seen in the transgenic lines, were also not significantly different. Levels of eicosenoic acid (20:1), specific to storage triacylglycerol, decreased in the bCA1-reduced lines at the same rate as in wildtype seedlings, suggesting that storage lipid catabolism was unaffected. In addition, dark growth, in terms of extent of hypocotyl elongation, was unaffected in the transgenic lines, whereas the seedling establishment phenotype became readily apparent when these lines were transferred into a light environment in the absence of Suc (Fig. 3). The mobilization of reserves in the early stages of growth could have required CA activity; however, the inhibitory impact of high light levels on seedling establishment in the bCA1-reduced plants and the ability of high CO 2 concentrations to rescue seedling development suggest a photosynthetic rather than metabolic role for the bCA1 isoform at this developmental stage. It is possible that low chloroplastic CA activity restricts CO 2 gas exchange in the thicker tissue of the cotyledons (as compared with true leaves). Arabidopsis cotyledons are known to have lower and more variable stomatal densities relative to true leaves (Geisler and Sack, 2002;Teng et al., 2006) and thus may have increased diffusive resistance for CO 2. Reduced intracellular CO 2 could limit Calvin cycle activity and result in reduced levels of photosynthate for growth, increased levels of photoinhibition (particularly at high light levels), and, ultimately, potential stagnation of seedling growth and development. The measurements of photosynthetic assimilation obtained using 14 CO 2 support the hypothesis of Table I. 14 CO 2 uptake by seedlings in the light Total acid-stable 14 C (as dpm 3 10 23 ) incorporated into intact seedlings growing on one-half-strength MS plates for each line and each time point following a 15-min incubation in the light (300 mmol m 22 s 21 ) at ambient levels of CO 2 and 22°C. Values represent means 6 SEs, where n 5 4 and each sample contains eight seedlings. reduced cotyledon carbon fixation in the bCA1reduced plants, particularly at a point in development where embryo reserves are limited and phototrophy is contributing a larger proportion of carbohydrates for growth. It was anticipated that growth at less than ambient CO 2 (150 mL L 21 ) would reduce survivorship to even lower levels that observed at ambient CO 2 concentrations; however, although trending lower, the differences were not statistically significant. It is possible that levels of CO 2 in the closed plates were somewhat higher than the subambient chamber CO 2 levels, or that a much greater reduction in chamber CO 2 level would be required to impact on survivorship beyond that observed at ambient levels. Ultimately, individual seed variation in available reserves, seedling variation in diffusive resistance, or rate of development may allow some individuals to escape this early stage of growth restriction and subsequently develop leaves. The carbon assimilation capacity of those transgenic plants that survive to produce true leaves does not seem to be compromised by low chloroplastic CA activity, as evidenced by their wild-type growth rate and phenotype at ambient and even subambient CO 2 levels (Fig. 4). The plant phenotypes were comparable over a range of photoperiods, including 16-and 12-h days (data not shown), even as short as 8 h (Fig. 4A). The absence of a phenotype in reduced CA activity plants that are able to establish themselves and continue to grow is similar to what was seen in previous studies. Mature bCA1 antisense tobacco plants in which foliar CA activities were reduced to less than 2% of wild-type activities were reported to be phenotypically similar to wild-type plants with respect to growth and development Price et al., 1994). Short-term, on-line gas exchange analysis showed that reduced chloroplastic CA activity resulted in less discrimination against 13 CO 2 than that obtained for wild-type plants. These data were consistent with a reduction in CA antisense plant chloroplastic CO 2 levels that would have resulted in only a small decline (,5%) in the rate of CO 2 assimilation Williams et al., 1996) and are in agreement with the observation that once past an early developmental stage, reduced levels of CA have little impact on photosynthesis and growth at air levels of CO 2 . The earlier studies with tobacco, however, do not report any observations on survivorship during seedling growth. At least one of these studies included Suc in MS plates for germination and growth of seedlings, which would have masked any effect of reduced bCA1 activity on seedling survivorship .
Microarray analysis did not reveal any compensatory changes in the expression of other CA isoforms or any significant changes in expression patterns of genes involved in photosynthesis and carbon metabolism in mature leaves of the bca1 plants. These data (in addition to the mutant phenotype) suggest that CO 2 / HCO 3 2 exchange is not limiting in the absence of bCA1 activity in the chloroplast or that existing carbon Figure 4. Phenotype of wild-type and bca1 mature plants grown at low and ambient levels of CO 2 . Surviving seedlings grown for 7 d at ambient levels of CO 2 , 22°C, and an 8-h photoperiod at 120 mmol m 22 s 21 were transplanted into pots of sterile Promix saturated with 1.0 g L 21 20:20:20 nutrient solution. Pots were placed in ambient and low (150 mL L 21 ) CO 2 chambers under the same growth conditions. A, Thirty-day-old plants under low and ambient CO 2 conditions. B, Western blot of total soluble proteins from mature leaves of the two lines probed with IgG directed against gel-purified Arabidopsis bCA2 protein. C, CA activity measured in protein extracts from mature leaves of similar age and obtained from 30-d-old plants grown at ambient levels of CO 2 as described above. Values are means 6 SE of samples taken from three individual plants of each genotype and expressed as WA units m 22 . Impact of Carbonic Anhydrase on Seedling Survivorship metabolic capacity is sufficiently plastic to not require a transcriptional response. Recently, the CA isoform bCA5 has been reported to be also localized to the chloroplast and may provide sufficient CA activity for adequate catalysis of plastid CO 2 /HCO 3 2 exchange in mature leaves (Fabre et al., 2007). Transcript levels for this protein, however, are very low relative to bCA1, and the abundance or activity of this specific isoform is unknown.
Microarray analysis indicted that there were no other readily discernible patterns in transcript abundance differences between wild-type and bca1 plants other than the unusual overrepresentation of downregulated ERF transcripts and other ethylene-response and -regulated genes in bca1 leaves. The reasons for this are unclear. Up-regulation of the ERF transcription factors is usually in response to ethylene exposure and/or biotic stresses such as wounding or response to pathogens and results in the regulation of expression of subsets of genes containing GCC boxes within their promoter regions (Fujimoto et al., 2000;Onate-Sanchez and Singh, 2002;McGrath et al., 2005;Yang et al., 2005). Similarly, both MYB73 and MYB77 exhibit increased transcript abundance in response to ethylene as well as other stimuli, including jasmonic acid and salicylic acid (Yanhui et al., 2006). The observation of expression levels of these ERF and MYB genes well below wild-type levels in the bca1 plants could be the result of an attenuation of the basal levels of signaling. It is interesting to note that the final enzymatic step in the ethylene biosynthetic pathway, catalyzed by ACC oxidase, occurs when ACC is oxidized to ethylene, HCN, and CO 2 . In vivo and in vitro ethylene biosynthesis is dependent on the presence of CO 2 (Kende, 1993), and enzyme mechanism studies have shown a requirement for CO 2 /HCO 3 2 in the generation of the productive oxidant in ACC oxidase catalysis (Rocklin et al., 2004). It is possible that reduced intracellular CA activity reduces the availability of the appropriate inorganic species for optimal ACC oxidase catalysis with the resultant lower levels of ethylene and the observed transcriptome response. Alternatively, CA activity may be required for some other aspect of ethylene/jasmonic acid signaling that involves the identified subset of ERF, MYB, and ACS genes. A study of the impact of reduced CA activity on ethylene biosynthesis and other aspects of ethylene/jasmonic acid signaling pathways is under way.
In conclusion, we have demonstrated that bCA1 plays a significant role in seedling establishment in Arabidopsis and that its impact is likely promulgated by maintaining the photosynthetic capacity of the cotyledons through the initial stages of seedling photoautotrophic growth and prior to the development of the first true leaves.

Plant Material and Growth Conditions
Following stratification and imbibition of sterilized seeds for 4 d at 4°C, Arabidopsis (Arabidopsis thaliana) ecotype Columbia was grown in sterilized artificial soil (Pro-Mix BX) saturated with 1 g L 21 of 20:20:20 all-purpose plant fertilizer (PlantProd) or on 0.8% agar, one-half-strength MS plates (supplemented with 1% [w/v] Suc when required) at 22°C and a light intensity of 120 mmol m 22 s 21 unless otherwise indicated, and as described previously (Fett and Coleman, 1994). Chamber CO 2 concentrations were determined by infrared gas analyzers (Horiba), which controlled rates of either CO 2 supplementation or the activity of a soda-lime scrubbing system to achieve the desired CO 2 level. At steady state, CO 2 concentrations could be maintained in the chambers at the desired set points of 150 and 1,500 mL L 21 6 30 mL L 21 . Ambient (nonmodified) CO 2 levels were routinely measured at 410 mL L 21 6 15 mL L 21 .

Molecular Biology
The bCA1 antisense vector was generated by cloning a 1.2-kb EcoRI cDNA fragment (Fett and Coleman, 1994) containing the intact bCA1 (At3g01500) coding region into the binary vector pGA643 (An et al., 1989). Restriction digest and PCR analysis was used to confirm the antisense orientation of bCA1 in pGA643. Agrobacterium tumefaciens strain LBA4404 was transformed by electroporation (Gene-Pulser; Bio-Rad) and transformants selected for growth on Luria-Bertani media containing tetracycline (12 mg mL 21 ) and streptomycin (30 mg mL 21 ). Four-week-old, soil-grown Arabidopsis was transformed by Agrobacterium containing the appropriate constructs using vacuum infiltration (Bechtold et al., 1993), and transgenic seed (T1) was identified by growth on one-half-strength MS plates containing 1% Suc and 25 mg mL 21 kanamycin. T4 seed was produced by three rounds of selfing with the initial 2 weeks of growth of each generation on one-half-strength MS plates containing kanamycin and Suc followed by transfer to pots for seed production. Plants from T3 seed lines were tested for reduced foliar CA activity and the T4 seed generated by individual lines with the lowest measured activity used for all studies.
Seeds for the bCA1 (At3g01500) knockout line were obtained from the Arabidopsis Biological Resource Center (The Ohio State University, Columbus, OH) as a T-DNA insertion line (Salk _0106570) and grown on one-halfstrength MS plus Suc plates as described above. PCR analysis using genomic DNA and forward primer 5#-TGCCTTCGTGGTCCGTAACAT-3# and 5#-TCA-AACCATAAATACAACCGATTTG-3# (specific for At3g01500) and primer LBaI 5#-TGGTTCACGTAGTGGGCCATCG-3# (specific for T-DNA left border) followed by sequence analysis of the products was used to identify those plants carrying the insertion at At3g01500. The PCR program used for amplification was as follows: one cycle of predenaturation at 94°C for 3 min, followed by 35 cycles at 94°C for 30 s, 58°C for 40 s, 72°C for 1 min, and a final extension at 72°C for 10 min. Homozygous lines containing a single T-DNA insert in At3g01500 were obtained by back-crossing with wild-type plants followed by two rounds of self-fertilization and progeny analysis on one-halfstrength MS plates containing 25 mg mL 21 kanamycin and 1% Suc.

RNA Sampling and Microarray Analysis
Discs from the area flanking the midvein of fully expanded leaves (leaf 6) of 20-d-old plants grown at 22°C, air levels of CO 2 , a light intensity of 120 mmol m 22 s 21 , and a photoperiod of 16 h light/8 h dark were harvested from nine individual plants of each genotype with samples from three individuals pooled as biological replicates. Following harvesting, samples were immediately frozen in liquid N 2 and RNA extracted using an RNAeasy Plant Mini kit (Qiagen) according to the manufacturer's instructions. cDNA synthesis, labeling, and hybridization to GeneChip Arabidopsis ATH1 genome arrays was carried out using standard Affymetrix protocols (http://Affymetrix.com) at the Botany Affymetrix GeneChip Facility, University of Toronto. Following scanning (Affymetrix GCS 3000) of the six chips (three biological replicates per genotype), the raw data were normalized using the GeneChip Operating System software with the scaling factor set at 500. Hybridization signals for all genes on each chip were examined, and where the signal level was below background level (declared absent by Affymetrix Microarray Suite), that gene was eliminated from further analysis. Reproducibility between replicate samples was tested by comparing hybridization signals for each gene not declared absent and calculating the least square regression. All replicates for each genotype showed acceptable reproducibility (R 2 . 0.9). All genes (except those declared absent) were used in the Significant Analysis of Microarrays program (Tusher et al., 2001), where the two class unpaired response option was used to compare hybridization signals of the two genotypes (wild type versus bca1) with the measure of significant fold change set at 1.8 and a false discovery rate ,4.5%.

Protein Biology and Biochemistry
Chlorophyll assays, CA assays (Wilbur and Anderson, 1948), and western analysis of soluble proteins obtained from intact seedlings or mature leaf discs were as described previously . For seed protein and lipid analysis, 50 seeds of each Arabidopsis line to be tested were extracted in equal volumes of the appropriate buffer to allow for quantitative as well as qualitative comparisons. Extraction and SDS-PAGE analysis of seed proteins was as described by Keith et al. (1994). Techniques used for the extraction, methanolysis, and gas chromatographic identification and quantification of lipids in each 50-seed aliquot were as described by Khan and Williams (1993) and Zhang et al. (2001).

CO 2 Incorporation by Seedlings
Two-, 3-, and 4-d-old seedlings (postimbibition/stratification) growing on one-half-strength MS plates were exposed to 14 CO 2 (obtained by acidification of 200 mCi of NaH 14 CO 3 stock [SA 56 mCi mmol 21 ]; ICN) for 15 min at ambient levels of CO 2 and O 2 in a sealed glass vessel at 22°C and a light intensity of 350 mmol m 22 s 21 . Eight intact seedlings per sample, with four replicates per line at each time point, were frozen in liquid N 2 , ground to a fine powder, and suspended in 200 mL of acidified water (1 mM acetic acid). Aliquots were added to 5 mL of ACS scintillation mixture (Amersham) and the disintegrations per minute incorporated determined using a Beckman LS 6000IC counter following correction for efficiency of sample counting.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Recovery of air-grown bca1 seedlings following transfer to high CO 2 .
Supplemental Table S1. Microarray analysis of RNA isolated from leaves of bca1 plants.