Carbonic anhydrases (CAs) are Zn-containing metalloenzymes that catalyse the reversible hydration of CO2. We investigated the αCA and βCA families in Arabidopsis, which contain eight αCA (AtαCA1-8) and six βCA genes (AtβCA1-6). Analyses of expressed sequence tags (ESTs) from The Arabidopsis Information Resource (TAIR) database indicate that all the βCA encoding sequences, but only three of the AtαCA, are expressed. Using semi-quantitative PCR experiments, functional CA genes were more strongly expressed in green tissue, but strong expression was also found in roots for βCA3, βCA6 and αCA2. Two αCA genes were shown to respond to the CO2 environment, while the others were unresponsive. Using the green fluorescent reporter protein gene fused with cDNA sequences coding for βCAs, we provided evidence that βCAs were targeted to specific subcellular compartments: βCA1 and βCA5 were targeted to the chloroplast, βCA2 and βCA3 to the cytosol, βCA4 to the plasma membrane and βCA6 to the mitochondria. The targeting and the pattern of gene expression suggest that CA isoforms play specific roles in subcellular compartments, tissues and organs. The data indicate that other CA isoforms than the well-characterized βCA1 may contribute to the CO2 transfer in the cell to the catalytic site of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco).
As the interconversion of carbon dioxide and bicarbonate is slow and thus appeared limiting in many physiological processes in which these inorganic carbon species are used, the existence of a catalysed conversion has been inferred long ago by physiologists and biochemists. Firstly purified from red blood cells (Meldrum & Roughton 1933; Stadie & O'Brien 1933), carbonic anhydrase (CA) [enzyme class (EC) 18.104.22.168], a Zn metalloenzyme that catalyses this reaction with exceptionally high efficiency, has subsequently been identified in the three domains of life: archaea, bacteria and eukarya.
On the basis of their amino acid sequence, CA proteins are grouped in four distinct classes: α, β, γ and δ, which share no sequence similarities and appear to have evolved independently (Hewett-Emmett & Tashian 1996; Tripp, Smith & Ferry 2001; So et al. 2004). Initially, the existence of a fifth class named ε, whose representatives have been discovered in marine cyanobacteria and some chemolithoautotrophs has been proposed (So et al. 2004), but the recent description of the crystal structure of CsoS3, the carboxisome shell-localized ε-type CA from the bacterium Halothiobacillus neapolitanus (Sawaya et al. 2006), has revealed that the enzyme is a variant β-type CA suggesting that the ε class should be withdrawn from the CA nomenclature.
The four classes have differing distributions: in animals, all enzymes so far discovered belong to the α class, while other eukaryotes encode α, β and γ classes of CA. To date, the δ class is restricted to the marine diatom Thalassiosira weissflogii(Roberts, Lane & Morel 1997; Tripp et al. 2001) and representatives of the ε class have only been discovered in marine cyanobacteria and some chemolithoautotrophs
CA has been involved in a broad range of biochemical processes that involve carboxylation or decarboxylation reactions including photosynthesis and respiration. CA also participates in pH regulation, inorganic carbon transport, ion transport, and water and electrolyte balance (Tashian 1989; Henry 1996; Smith & Ferry 2000).
In plants and algae, all known CAs belong to the α, β and γ classes with the β class predominating (reviewed by Moroney, Bartlett & Samuelsson 2001). In algae, the role of CA in concentrating inorganic carbon for efficient photosynthetic fixation has been studied most (reviewed in Badger & Price 2003). In the model microalga Chlamydomonas reinhardtii, CA biochemistry and function has been extensively explored leading to a better understanding of the CO2-concentration mechanism (CCM), which contributes to the acclimation of this microalga to CO2-limiting growth conditions. Six different CA isoforms have been characterized in the extracellular, mitochondrial and chloroplastic compartments. Periplasmic CA activity underlain by two αCA isoforms (CAH1 and CAH2, Fujiwara et al. 1990; Fukuzawa et al. 1990) has been implicated in the supply of inorganic carbon for photosynthesis. CAH1 whose synthesis is induced under low CO2 conditions, canaccelerate the equilibrium between CO2 and HCO3– so that CO2 at the cell surface can diffuse across the membrane surface (Moroney, Husic & Tolbert 1985; Van & Spalding 1999). Another αCA localized in the thylakoid lumen (CAH3, Karlsson et al. 1995) and a βCA in the chloroplast stroma (CAH6, Mitra et al. 2004) catalyse the CO2/HCO3– equilibrium within their respective subcompartment. CAH3, which is required for growth of the algal cells at ambient levels of CO2 (Funke, Kovar & Weeks 1997; Karlsson et al. 1998), has been proposed to increase generation of CO2 from bicarbonate and consequently CO2 availability at the catalytic site of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) in the pyrenoid, assuming that CO2 diffuses through the thylakoid membrane (Raven 1997). CAH6 is mainly located in an area around the pyrenoid. It possibly converts CO2 diffusing out into bicarbonate, increasing the HCO3– pool in the stroma thereby retaining inorganic carbon within the chloroplast (Mitra et al. 2004). Both mitochondrial βCAs, whose synthesis is induced at low CO2, appear to be part of the CCM as well. Their proposed function is to convert the CO2 produced in the mitochondria by respiration and photorespiration into HCO3–, thereby limiting the CO2 leakage through the plasmalemma (Raven 2001). Their role in anaplerotic assimilation catalysed by phosphoenolpyruvate carboxylase has also been proposed (Giordano et al. 2003).
Available information on CA in plants originates largely from biochemical and functional studies that focus on the role of the main CA isoforms in the efficiency of photosynthesis. In this context, C4 and C3 plants must be distinctly considered. C4 plants contain a CCM based on a cell anatomical differentiation and metabolic coordination within the leaf (Hatch 1987). Most of the CA is located in the cytosol of the mesophyll cells (Ku, Kano-Murakami & Matsuoka 1996) and is required to supply adequate level of HCO3– to phosphoenolpyruvate carboxylase (Hatch & Burnell 1990; Von Caemmerer et al. 2004), which produces C4 acids. The C4 acids diffuse to and are decarboxylated in the bundle sheath cells providing CO2 to Rubisco. In Flaveria bidentis, three cDNAs encoding putative βCAs (CA1, CA2 and CA3) have been recovered from a leaf cDNA library (Von Caemmerer et al. 2004). A recent study using an antisense strategy to modulate the expression of CA3, the main putative cytosolic CA, has shown that CA activity is essential to maximize C4 photosynthesis (Von Caemmerer et al. 2004). In C3 plants, where it represents about 5% of soluble protein in the stroma, the chloroplastic βCA is initially assumed to be highly important for optimizing CO2 availability to Rubisco at the alkaline stromal pH. Its role has been investigated in tobacco using an antisense strategy (Majeau, Arnoldo & Coleman 1994; Price et al. 1994). Antisense plants exhibiting about 1–10% of the wild-type CA level surprisingly have not shown an evident growth phenotype and only minor effects on mesophyll CO2 transfer and photosynthesis (Price et al. 1994). Another βCA isoform has been identified in the cytosol of C3 plants (Fett & Coleman 1994). Although relatively abundant, as its activity represents 13% of total CA activity estimated in potato leaf (Rumeau et al. 1996), its role has not been reported. Occurrence of CA in plants has been extended to perennial plants as a βCA encoding cDNA has been identified in hybrid aspen (Populus tremula × tremuloides, Larsson et al. 1997). Its molecular characterization confirms that both the structure and activity of hybrid aspen CA resemble βCAs from annual plants.
Over the last decade, data afforded by various genomes sequencing have revealed the multiplicity of CA isoforms in plants. When about 70% of the Arabidopsis thaliana genome sequences were available, Moroney et al. (2001) reported the existence of expressed sequence tags (ESTs) corresponding to six, four and three genes that align with α, βand γCA sequences, respectively. Two of the genes encoding βCAs have been initially described by Fett & Coleman (1994); one encodes a putative cytosolic isoform; the other, whose expression is light regulated and dependent on chloroplast development, is expected to code for the main chloroplastic isoform. Recently, a family of five γCA genes has been characterized (Parisi et al. 2004). The gene products are targeted to the mitochondria (Millar et al. 2001) where they associate to complex I (Perales et al. 2005; Sunderhaus et al. 2006). It has been demonstrated that at least one of the γCA is important for complex I assembly (Perales et al. 2005). Recently, the product of an αCA gene, CAH1 (Villarejo et al. 2005), has been characterized. Intriguingly, CAH1 is transported through the secretory pathway and is N glycosylated before entering the chloroplast. Its role has not yet been assessed.
The multiplicity of CA isoforms in a single C3 species underlines CA importance of and even suggests that new functions for this enzyme may emerge from in-depth investigations. In this context, the present work investigated the significance of CA redundancy in the C3 model plant A. thaliana, and addressed basic questions on the determination of the number of CA gene sequences, their organ-specific expression and the localization of their encoded proteins.
MATERIALS AND METHODS
Arabidopsis thaliana (ecotype Columbia-0) plants were potted in soil and grown in controlled environments under an 8h photoperiod of 300 µmol m–2 s–1. Day/night temperature and relative humidity were 22/18 °C and 60/50%, respectively. For tissue expression assays, tissues were harvested on silique ripening plants (growth stage 8, Boyes et al. 2001). For the CO2 experiments, plants were grown in CO2-controlled environmental chambers with CO2 concentration set at 150, 360 or 1000 μL L−1. The general design of the chambers, including the temperature and hygrometry control and the CO2 regulation, was extensively described (Fabreguettes et al. 1994).
Total RNA was extracted from 100 mg of frozen sample using RNeasy Plant Minikit (Qiagen, Courtaboeuf, France) and was treated with RNAse free DNAse set (Qiagen) to eliminate any genomic DNA contamination. First-strand cDNA synthesis was performed with 1 µg total RNA using Omniscript RT kit (Qiagen) and oligo(dT) as primer. PCR conditions were 25 cycles of 94 °C for 45 s, 55 °C for 45 s and 72 °C for 45 s. Reactions were run in a Gene Amp PCR system 2700 cycler (Applied Biosciences, Courtaboeuf, France). The following specific primer pairs were designed in different exon sequences; thus, the cDNA could be distinguished from genomic DNA: AtβCA1 (At3g01500), 5′-AATCTTCTCTCCAGAAACTC-3′ and 5′-ACCCTTTGCGAGCTCACCGTA-3′; AtβCA2 (At5g14740), 5′-TTCAGCCACTTCAAACTTGAA-3′ and 5′-TGGTAAACAGATTTTGACCCA-3′; AtβCA3 (At1g23730) 5′-CCAAGAACAGGATCAAGCAGGA-3′ and 5′-AAGGCAAAGGCAGGGGTAGTCT-3′; AtβCA4 (At1g70410), 5′-TTTCATGTTCTGCTGCGCTA-3′ and 5′-TGGAAATTCAAGATGTGAGAT-3′; AtβCA5 (At4g33580), 5′-ATTGATCCAGATTCGAAGTA-3′ and 5′-GTAGTGTTCAAAGTCATCCAT-3′; AtβCA6 (At1g58180), 5′-TCGTCGTCTAGTCTCTGCAA-3′ and 5′-CTGAACCGGGGTAACGAGAT-3′; AtαCA1 (At3g52720), 5′-ACTAGTGAACCACGTCTGTAA-3′ and 5′-AGTAGTGAGTGAACCAATGTAT-3′; AtαCA2 (At2g28210), 5′-AATAGAGGCCATGATATGATGCTG-3′ and 5′-AGTAGTAAGTGATCCAATGTAT-3′; AtαCA3 (At5g04180), 5′-AACCGTGGATTCGACATGAAGGTT-3′ and 5′-AGTCGTGAGTGAACCTCTGTAT-3′; AtαCA4 (At4g20990), 5′-ACACGCTCATTCTGAAGCG-3′ and 5′-CCCTTTGAGTCTTGAACCGG-3′; AtαCA5 (At1g08065), 5′-AAGAAAGGAGAGAAGGGGCC-3′ and 5′-AGTTTACTATGCCATGTTGG-3′; AtαCA6 (At4g21000), 5′-CGTTTATGCTCGTGAAACG-3′ and 5′-ACGCCTTCGTGCATGGAGG-3′; AtαCA7 (At1g08080), 5′-TGCTTCAAGTCACGGAGAAG-3′ and 5′-GTGCACTATGCGCTTGTTGG-3′; AtαCA8 (At5g56330) 5′-ACTAGGATGGGTCTTAGTCC-3′ and 5′-CGGATAGTGACACCAATACC-3′.
As control, RT-PCR was performed using actin gene-specific primers (At2g37620), 5′-AAAATGGCTGATGGTGAAGACA-3′ and 5′-GATGGTTATGACTTGTCCAT-3′.
RT-PCR analysis of CA transcripts in different plant tissues and in leaves of plants exposed to varied CO2 concentrations were repeated four and two times, respectively.
Green fluorescent protein (GFP) fusion construction and Arabidopsis stable transformation
Total leaf RNA extraction, cDNA synthesis and PCR were carried out as for transcript analysis. All constructs were made using Gateway cloning technology (Invitrogen, Cergy-Pontoise, France) according to the manufacturer's instructions. The cDNA of each of the six AtβCA genes was amplified using primers (see below) allowing the addition of attB recombination sites and was cloned into a pDONR221 vector using a Gateway BP Clonase enzyme mix technology (Invitrogen). One entry clone was fully sequenced before subsequent cloning in frame with an N-terminal enhanced green fluorescent protein (EGFP) gene in the binary Gateway destination vector pB7FWG2 (Plant Systems Biology, VIB-Ghent University, Belgium) (Karimi, Inze & Depicker 2002) by using a Gateway LR Clonase enzyme mix (Invitrogen). The pB7FWG2 vector allows fusion of EGFP with the CA C terminus, and expression of the resulting cDNA is placed under the control of the cauliflower mosaic virus 35S promoter and of the 3′ untranslated transcribed region of a nopaline synthase gene. The binary constructs were introduced into the Agrobacterium tumefaciens strain C58, and the resulting bacterial cultures were used to transform Arabidopsis ecotype Columbia-0 by the standard flower dip method (Clough & Bent 1998).
Whenever necessary, protoplasts were isolated in a solution of 1% cellulases R10 (Serva, Paris, France) and 0.25% macerozyme R10 (Serva) dissolved in 2-morpholinoethanesulfonic acid (MES)-KOH buffer (50 mM, pH 5.6) containing 8 mM CaCl2 and 400 mM mannitol (O'Neill et al. 1993). Following overnight digestion, protoplasts were collected by centrifugation at 400 g for 3 min.
Images of green GFP fluorescence and chlorophyll red autofluorescence were monitored using a confocal laser scaning microscope (LCTCS-SP2; Leica, Rueil-Malmaison, France) with a 488 nm argon laser excitation. Objective HCX Plan Apo CS 40X Oil NA1.40 (Leica) was used. The fluorescence signals were collected from 500 to 535 nm for GFP and from 680 to 720 nm for chlorophyll. For mitochondria detection, just before observation, small leaf fragments (1 mm2) wereinfiltrated with the dye MitoTracker Red (CM-H2XROS; Molecular Probes, Invitrogen). Excitation was performed at 561 nm using a laser diode and fluorescence collected from 590 to 610 nm.
Identification and analysis of the Arabidopsis αCA and βCA gene families
To identify the Arabidopsis αCA and βCA gene families, an extensive search of the Arabidopsis genomic sequences was performed using BLAST and FASTA algorithms and using diverse αCA and βCA sequences as initial query sequences. Eight αCA and six βCA encoding sequences were retrieved. Table 1 shows the list of the CA genes indicating their identification numbers (AGI number) and where applicable, previous gene designation. To determine the proteins encoded by the AtαCA and the AtβCA genes, the gene structure and the exon–intron organization had first to be confirmed. Where possible, reverse transcriptase PCR products were sequenced and compared to cDNA sequences retrieved from The Arabidopsis Information Resource (TAIR), the MIPS and the Riken web sites (http://www.arabidopsis.org/, http://mips.gsf.de/proj/thal/db/index.html and http://www.gsc.riken.go.jp/indexE.htm as of January 2006). With this combined approach, we identified the AtαCA1-3 and AtβCA1-6 genes unambiguously. We were not able to amplify any transcript of AtαCA4-8; therefore, AtαCA4-8 exon–intron splice sites were predicted using GeneScan (http://genes.mit.edu/GENSCAN.html) and based on the most favourable similarity to the characteristic protein structure. αCA and βCA amino acid sequences were deduced from the cDNA sequences. Figures 1 and 2 show multiple sequence alignments of Arabidopsis αCA and βCA gene products, respectively. The sequence of the extensively characterized human isoenzyme II αCA and pea βCA (Swiss Prot accession numbers P00918 and P17067, respectively) were aligned in parallel for comparison to the Arabidopsis αCA and βCA sequences, respectively. The sequence alignments show that the CA sequences contain the key features of AtαCAs and AtβCAs. The αCA sequence alignment reveals that all of the AtαCA protein sequences contain the three Zn-liganded histidine residues and most of the residues predicted to form the hydrogen-bond network to Zn-bound solvent molecules in mammalian αCA isoenzymes (Fig. 2). The amino acid residues that have been shown either to constitute Zn ligands (one histidine and two cysteine residues) or to orientate towards Zn (Kimber & Pai 2000; Mitsuhashi et al. 2000) appeared highly conserved in the AtβCA sequences. These findings suggest that the products of all the Arabidopsis αCA and βCA genes are functional enzymes. Considerable sequence identity exists between the different protein members of the AtαCA and AtβCA family sequences. However, several CA sequences contain distinct N-terminal extensions that can correspond to target peptides suggesting that the different isoforms are addressed to specific cell compartments.
Table 1. Overview of the Arabidopsis thaliana αCA and βCA gene families
AtαCA and AtβCA expression pattern in different organs
A search of the TAIR database (http://www.arabidopsis.org/ as of January 2006) indicated that ESTs (Table 1) have been deposited for three of the eight αCA and all of the six βCA encoding genes signifying that the βCA genes are functional. The lack of ESTs for AtαCA4, AtαCA5, AtαCA6, AtαCA7 and AtαCA8, although a cDNA sequence was released for AtαCA4 (GeneBank reference AY122907), might indicate that these genes were pseudogenes or that they were expressed at very low levels and/or that the conditions used in generating ESTs data were limiting and did not include the conditions under which these genes were expressed or that the corresponding gene transcripts were not stable. To determine the organ-specificity expression pattern of each member of the AtαCA and AtβCA gene family, RT-PCR analysis was performed on RNA isolated from leaves, stems, flowers, siliques and roots of 8-week-old plants. Figure 3 shows agarose gel analyses of RT-PCR products. Transcript for actin was used as internal control. To avoid detecting DNA contamination, primers spanning an intron were designed. The correct identities of the PCR products were confirmed by sequencing. For AtβCA6, three different products were obtained following amplification with gene-specific primers (Fig. 3); however, identity of the major band as AtβCA6 product was confirmed. No PCR product was obtained following amplification with AtαCA4-8 specific primers as expected according to the lack of ESTs in the databases (data not shown). The AtβCA1, AtβCA2, AtβCA4 and AtβCA5 genes were mainly strongly expressed in the aboveground tissues with almost no expression in roots. Transcripts of AtβCA3 and AtβCA6 were detected in all parts of the plant but at a lower level in roots. As expected from the ESTs data, only the expression of AtαCA1, AtαCA2 and AtαCA3 was detected in Arabidopsis plants. The transcript of AtαCA1 was detected in all organs except roots, while AtαCA2 is expressed only in stems and roots, and AtαCA3 expression is restricted to flowers and siliques.
Effects of CO2 levels on the accumulation of CA mRNA
We tested if changes in environmental CO2 concentration altered AtαCA and AtβCA gene expression. Plants were grown either at low (150 ppm) or high (1000 ppm) CO2 concentrations. Gene expression levels were quantified by RT-PCR in leaves and compared to plants grown under air CO2 concentration (360 ppm). The transcript for actin was used as internal control. As shown in Fig. 4, the expression of most AtβCA genes was insensitive to both CO2 conditions. Only the expression of AtαCA6 appeared slightly but significantly enhanced when plants were grown under high CO2 conditions. Contrary to AtαCA1 whose transcript level remained unaffected by CO2 regime, AtαCA2 and AtαCA3 expression was stimulated when the plants were grown in low CO2 concentration.
Localization of encoded βCAs
To uncover the possible potential subcellular localization of the products of AtCA genes expressed in leaves, we searched for different signatures specific to cellular compartments using TargetP, Predotar and Psort softwares (http://www.cbs.dtu.dk/services/TargetP/, http://urgi.infobiogen.fr/predotar/french.html and http://www.psort.org/, respectively) for AtβCAs (Table 2). The results predicted AtβCA1 and AtβCA5 to have chloroplastic target peptides. AtβCA6 was predicted to be targeted to mitochondria. Prediction indicated no preferred localization subcompartment for AtβCA2, AtβCA3 and AtβCA4. In order to determine the subcellular localization of the CA encoded by the six AtβCA genes, we used GFP as a reporter. Individual cDNA encoding the full protein sequence were cloned in frame with the GFP. The resulting transgenes were expressed under the control of the 35S promoter in stably transformed Arabidopsis plants. Expression of the fusion proteins was analysed using a confocal laser scanning microscope (Fig. 5). In mesophyll cells of transgenic plants, the GFP signal fused to AtβCA1 or AtβCA5 co-localized with chlorophyll autofluorescence, indicating its association with chloroplast. In plants expressing AtβCA2-GFP and AtβCA3-GFP, GFP fluorescence was emitted in the cytosol (Fig. 5). To allow a better evidence that the GFP fluorescence originated from the cytosol, protoplasts from plant expressing AtβCA2-GFP and AtβCA3-GFP were used (Fig. 6). For AtβCA4-GFP, GFP fluorescence was monitored in epidermal cells (Fig. 5), where the signal appeared as a thin line at the cell periphery, suggesting a localization in the plasma membrane. Protoplasts prepared from transgenic plants expressing AtβCA4-GFP supported that the localization of AtβCA4 was restricted to the plasma membrane (Fig. 6). In plants expressing the AtβCA6-GFP construct, fluorescence was localized to small particules, which we supposed to be mitochondria. Leaves fragments were infiltrated with MitoTracker (Molecular Probes); merging GFP and MitoTracker fluorescence confirmed mitochondria targeting of AtβCA6 (Fig. 7).
Ch, chloroplast; O, outside; EW, elsewhere; Cy, cytosol; M, mitochondria.
The plant model A. thaliana contains genes encoding CAs from the α, β and γ classes. Whereas the family of genes encoding the recently discovered γCAs (Alber & Ferry 1994) has been characterized (Parisi et al. 2004; Perales et al. 2005; Sunderhaus et al. 2006), paradoxically, data available on the genes encoding αCA and βCA are scarce. In the past, Fett & Coleman (1994) were the first who characterized two transcriptionally active βCA genes, CA1 and CA2, encoding the two main CAs in Arabidopsis located in the chloroplast and cytosol, respectively. When 70% of the Arabidopsis genome was sequenced, Moroney et al. (2001) identified four αCA and five βCA gene sequences. Recently, Villarejo et al. (2005) isolated a cDNA containing an open reading frame (ORF) that encodes an αCA (CAH1). Here, based on the full genome sequence, we update the list of Arabidopsis CA genes and report the existence of eight αCA and six βCA genes in Arabidopsis (Table 1). AtβCA1 and AtβCA2 correspond to CA1 and CA2 described by Fett & Coleman (1994), while AtαCA1 codes for CAH1.
The structure and catalytic mechanism of the αCA and βCA have been well-documented. CAs from both classes contain one Zn per enzyme unit and likely have a common Zn-hydroxide mechanism for catalysis (Silverman & Lindskog 1988; Bracey et al. 1994). However, the Zn ligands in βCAs are quite different from those of αCAs. In αCAs, nitrogen atoms of three histidines directly coordinate the Zn (Hakansson et al. 1992), and a water molecule acts as a fourth ligand. A threonine and a glutamate residue interact directly through the bound water. These essential aminoacids are well conserved in all AtαCA gene products, suggesting that encoded αCAs are functional isoenzymes (Fig. 1). Similarly, in βCAs, Zn is coordinated by two cysteine and one histidine residues (Bracey et al. 1994; Kimber & Pai 2000) and other well-conserved residues participate in the catalytic function (Fig. 2). In the AtβCA sequences, essential residues appear well conserved as seen in the AtαCA family, suggesting that all βCA isoenzymes are functional isoenzymes too.
Three αCA and six βCA genes are expressed in Arabidopsis
To gain insight into where and under what conditions AtCAs function, we initiated a study to characterize the expression of the members of the αCA and βCA gene families. The Arabidopsis EST resource is a very useful source for information about gene expression, as large data sets have been derived from numerous libraries representing different tissues and different experimental treatments (The NCBI dbEST database release from 18 August 2006 included 622 972 ESTs, http://www.ncbi.nlm.nih.gov/dbEST/). A preliminary bioinformatics survey of ESTs indicates that the different genes were not similarly expressed. Whereas the products of all members of the βCA family were represented, the lack of ESTs for five out of the eight αCA genes suggest that only three of them (AtαCA1, AtαCA2 and AtαCA3) may be functional, although the lack of EST does not rule out AtαCA4-8 to be expressed at very low level and/or in very specialized cells. Surprisingly, a single cDNA encoding AtαCA7 has been isolated from an expanded Arabidopsis library (Yamada et al. 2003), and AtαCA4, the protein product of At4g20990, was identified using a mass-spectrometry proteomic approach in thylakoid membranes of Arabidopsis (Friso et al. 2004). These results demonstrate that, although extensive, the ESTs data are not exhaustive and suggest that AtαCA4 and AtαCA7 transcripts are highly unstable and/or that gene expression is very low or restricted to particular tissues, developmental stages or environmental conditions. For similar reasons, we cannot conclude as whether AtαCA5 and AtαCA6 are functional, although the presence of the essential amino acids involved in catalysis in the corresponding protein products would suggest it.
CA expression is organ specific
Because information about organ expression patterns for a gene often provides clues about its function, we performed RT-PCR experiments on flowering plants grown under control conditions. The expression of three of eight AtαCA encoding genes (only AtαCA1-3) could be detected by RT-PCR performed using RNA from various organs. All attempts to amplify the products of AtαCA4-8 failed, consistent with the fact that no ESTs have yet been documented for these genes. This result confirms either the low abundance of AtαCA4-8 transcripts in the different organs of the plant or their low stability. The six genes encoding βCAs are significantly expressed in aboveground tissues, and their expression is faint in roots with the exception of AtβCA3 and AtβCA6 that show large expression in roots. If we assume that the product of any of these genes is involved in photosynthetic carbon fixation, we expect its expression to occur preferentially in green tissues. The preferential expression of AtCA1-2-4-5 in green tissue suggests their products are interesting candidates for Ci transfer in photosynthetic processes. Expression pattern of AtαCA1-3 encoding genes appears to vary more in between organs as compared with AtβCA genes. Interestingly, whereas AtaCA1 is broadly expressed in aboveground tissues, expression of AtαCA2 and AtαCA3 is restricted to conducting organs and flowers, respectively, suggesting rather specific functions for these two AtαCAs.
Expression of AtCA in response to CO2 differs
We examined the effect of long-term exposure to different CO2 concentrations on the expression of CA genes in Arabidopsis leaves. We detect no change in the transcript abundance of the βCA gene family except a faint increase at high CO2 for AtβCA6. The AtβCA6 transcript whose product is most likely targeted to mitochondria increased in leaves of plants grown at high CO2 concentration, suggesting that this isoform is important to the acclimation of plants to high CO2 conditions. In Chlamydomonas, mitochondrial βCAs are induced at high CO2 concentration by increasing the NH4+ concentration in the growth medium, suggesting that the mitochondrial CA is involved in supplying HCO3– for anaplerotic assimilation catalysed by phosphoenolpyruvate carboxylase (Giordano et al. 2003). The proposed function for mitochondrial CA of Chlamydomonas may be transposed to AtβCA6. Interestingly, AtαCA2 and AtαCA3, whose mRNA could be detected in leaves at ambient or high CO2 concentration, appeared expressed at low CO2 concentration, implicating a specific function when CO2 is limiting. Previously, two other published studies performed on pea and on Arabidopsis plants, grown at different CO2 concentrations, described opposing results. In pea grown at a concentration of 1000 ppm CO2, Majeau & Coleman (1996) reported a decrease of βCA transcript abundance in contradiction with the increase of βCA mRNA observed by Raines et al. (1992) in Arabidopsis rosettes grown in 660 ppm CO2. Opposite results may be due either to different species-specific responses to high CO2 or to differing CO2 concentrations used in both studies. However, at that time, the possibility that differenttranscripts cross hybridize with a single probe was not considered.
βCAs are located in different subcellular compartments
The simultaneous expression of the six AtβCA genes in aboveground tissues of Arabidopsis can be linked to the specific requirement of each corresponding protein product in specific cell compartments. In order to address this point, we have determined their localization using GFP fusion targeting experiments. The results confirmed the plastid and cytosolic localization of AtβCA1 (CA1) and AtβCA2 (CA2), respectively (Fett & Coleman 1994). Intriguingly, AtβCA2 was detected by mass-spectrometry in the plasma membrane of Arabidopsis leaf cells (Alexandersson et al. 2004). In the plasma membrane of tobacco, the AtβCA2 ortholog seems to be primarily concentrated in the lipid rafts (Mongrand et al. 2004). These small detergent-insoluble micro domains appear to recruit a specific set of plasma membranes among which CA and aquaporins are preponderant (Mongrand et al. 2004). AtβCA2 localization using GFP fluorescence imaging did not allow us to discriminate between cytosol and plasma membrane; thus, we cannot exclude simultaneous targeting of AtβCA2 to both subcompartments. Interestingly, the association of aquaporin and AtβCA2 in the lipid rafts suggests that this protein may be involved in CO2 membrane transport as proposed recently (Uehlein et al. 2003). In the past, Utsunomiya & Muto (1993) described a CA activity that was located at the cytosolic face of the plasma membrane and was sustained by a protein immunologically related to the main chloroplastic CA isoform. The product of AtβCA2 in Arabidopsis represents the protein identified by Mongrand et al. (2004) in the lipid rafts and can correspond to the CA whose activity was described by Utsunomiya & Muto (1993). When fused to GFP, the product of AtβCA4 was exclusively targeted to the plasma membrane; this protein could also contribute to the CA activity at the plasma membrane. AtβCA4 has been identified in the chloroplast envelope and the tonoplast using mass spectrometry (Froehlich et al. 2003; Endler et al. 2006) but not in the plasma membrane proteomic data, in contradiction to our experimental results. In the envelope and tonoplast proteome analyses, the authors raised the difficulty of preparing Arabidopsis pure chloroplast envelopes and tonoplasts and could not overcome contamination with other membranes. The three other βCAs, AtβCA3-5-6, appeared targeted to the cell compartments cytosol, chloroplast and mitochondria, respectively. While two βCAs have been identified in the mitochondrial matrix of C. reinhardtii (Eriksson et al. 1996; Geraghty & Spalding 1996), no corresponding CA has been reported in algae or higher plants.
In conclusion, this is the first study in C3 plants to argue for molecular redundancy of βCAs in the cytosol and the chloroplast, and for a specific βCA localized in the plasma membrane. Additional studies on the localization of AtαCA gene products are likely to increase this complexity of CA distribution. Work is under way to further resolve the expression and the localization of the αCA family. We will consider the significance of CA in the cytosol and the plasma membrane for the understanding of the transfer of inorganic carbon through the plasma membrane and the cytosol to the catalytic site of Rubisco in the chloroplast. Specific functions of CA activity in various tissues are also required to be addressed.
We thank the Groupe de Recherche Appliquée en Phytotechnologie (GRAP) team for technical assistance with controlled growth chambers and Mohamed Barakat and Laboratoire de Biologie et Dévelopement des Plantes for valuable technical assistance with the confocal microscopy. I. Reiter and B. Genty acknowledge the financial support provided through the European Community's Human Potential Programme under contract HPRN-CT-2002-00254 (Stressimaging).