The γ-carbonic anhydrase subcomplex of mitochondrial complex I is essential for development and important for photomorphogenesis of Arabidopsis.

Complex I (NADH:ubiquinone oxidoreductase) is the entry point for electrons into the respiratory electron transport chain; therefore, it plays a central role in cellular energy metabolism. Complex I from different organisms has a similar basic structure. However, an extra structural module, referred to as the γ-carbonic anhydrase (γCA) subcomplex, is found in the mitochondrial complex I of photoautotrophic eukaryotes, such as green alga and plants, but not in that of the heterotrophic eukaryotes, such as fungi and mammals. It has been proposed that the γCA subcomplex is required for the light-dependent life style of photoautotrophic eukaryotes, but this hypothesis has not been successfully tested. We report here a genetic study of the genes γCAL1 and γCAL2 that encode two subunits of the γCA subcomplex of mitochondrial complex I. We found that mutations of γCAL1 and γCAL2 in Arabidopsis (Arabidopsis thaliana) result in defective embryogenesis and nongerminating seeds, demonstrating the functional significance of the γCA subcomplex of mitochondrial complex I in plant development. Surprisingly, we also found that reduced expression of γCAL1 and γCAL2 genes altered photomorphogenic development. The γcal1 mutant plant expressing the RNA interference construct of the γCAL2 gene showed a partial constitutive photomorphogenic phenotype in young seedlings and a reduced photoperiodic sensitivity in adult plants. The involvement of the γCA subcomplex of mitochondrial complex I in plant photomorphogenesis and the possible evolutionary significance of this plant-specific mitochondrial protein complex are discussed.

The light-dependent development of plants, or photomorphogenesis, has been extensively studied in Arabidopsis (Arabidopsis thaliana). It is now clear that several nuclear photosensory receptors, such as red/ far-red light receptor phytochromes, blue light receptor cryptochromes, and LOV domain F-box proteins ZEITLUPE (ZTL) and FLAVIN-BINDING, KELCH REPEAT, F-BOX1 (FKF1) mediate the light-dependent modulation of gene expression to alter various physiological aspects to regulate photomorphogenesis (Franklin and Quail, 2010;Kami et al., 2010;Liu et al., 2011b;Li et al., 2012). Those photoreceptors interact with regulators of transcription or protein degradation to affect gene expression. For example, phytochromes interact with the basic helix-loop-helix (bHLH) transcription factor PHYTOCHROME-INTERACTING FACTORS (PIFs) in response to red/far-red light to trigger degradation of the PIF proteins and alter the transcription of the PIF target genes (Ni et al., 1998;Huq and Quail, 2002;Bauer et al., 2004;Leivar and Quail, 2011); cryptochromes interact with the bHLH transcription factor CRYPTOCHROME-INTERACTING BHLH1 and the SUPPRESSOR OF PHYTOCHROME A1/COP1 E3 ubiquitin ligase complex in response to blue light to regulate transcription and protein degradation, respectively (Yang et al., 2000;Wang et al., 2001;Liu et al., 2008Liu et al., , 2011aLian et al., 2011;Zuo et al., 2011); and ZTL and FKF1 interact with transcription regulators such as TIMING OF CAB EXPRESSION1, CYCLING DOF FACTOR1, GIGANTEA, and CONSTANS (CO) to modulate the transcription or protein degradation of the target genes and physiological responses to light (Más et al., 2003;Imaizumi et al., 2005;Kim et al., 2007;Sawa et al., 2007;Ito et al., 2012). Many photoreceptor target genes are regulators or enzymes of various metabolic processes, especially the energy metabolism in chloroplasts and mitochondria . For example, Arabidopsis cryptochromes regulate the transcription of the nucleus-encoded chloroplast s factor, SIG5, to regulate plastid transcription and development to affect photomorphogenesis (Ruckle et al., 2007;Ruckle and Larkin, 2009). However, it is less clear how photoreceptors affect mitochondrial development or whether mitochondrial proteins affect photomorphogenesis.
The respiratory complex I (NADH:ubiquinone oxidoreductase or NADH dehydrogenase) is the entry point of electrons to the mitochondrial respiratory electron transport chain. Despite the similar function in energy metabolism, complex I of photoautotrophic organisms exhibits different subunits and structure features in comparison with its counterpart of heterotrophic organisms (Brauna and Zabaleta, 2007;Hunte et al., 2010;Klodmann et al., 2010). Notably, complex I of plants and algae contains a g-carbonic anhydrase (gCA) subcomplex that is absent from complex I of fungi and animals (Dudkina et al., 2005;Sunderhaus et al., 2006;Brauna and Zabaleta, 2007;Bultema et al., 2009;Efremov et al., 2010;Hunte et al., 2010;Klodmann et al., 2010;Klodmann and Braun, 2011;Fig. 1A). In Arabidopsis, the gCA subcomplex contains three carbonic anhydrase subunits, gCA1 (At1g19580), gCA2 (At1g47260), and gCA3 (At5g66510), and two gCA-like subunits, gCAL1 (AT5G63510) and gCAL2 (AT3G48680). The gCA and gCAL genes are widely found in plant genomes but not in animal genomes examined thus far (Brauna and Zabaleta, 2007). Therefore, it was proposed that the gCA subcomplex may play a role in the light-dependent life style of plants (Brauna and Zabaleta, 2007). Several Arabidopsis mutants have been reported that affect proteins of complex I; these mutants showed growth retardation, abnormal leaf morphology, and poor germination (de Longevialle et al., 2007;Keren et al., 2012). A tobacco (Nicotiana tabacum) complex I-specific mutant, CMSII, exhibits decreased photosynthesis at atmospheric Figure 1. gCAL1 and gCAL2 are mitochondrial proteins of the plant-specific gCA subcomplex of complex I. A, Diagram depicting the gCA subcomplex (black module indicated by arrows), the mitochondrial complex I of green algae (Polytomella spp.) and plants (Arabidopsis; Dudkina et al., 2005;Sunderhaus et al., 2006), but not in bacteria (Escherichia coli; Morgan and Sazanov, 2008), fungi (Neurospora spp.; Guénebaut et al., 1997), or mammals (bovine; Grigorieff, 1998). The structural outlines of complex I are redrawn from the published structure of the respective complex I. B to E, Confocal images showing the subcellular localization of the gCAL1-YFP and gCAL2-YFP (B-D) or gCAL1-GFP and gCAL2-GFP (E) fusion proteins in hypocotyl cells of 5-d-old seedlings (B and C) or protoplasts isolated from 3week-old plants (E). CS16264 is a mitochondrial marker protein. Propidium iodide (PI) was used to stain the cell wall. BF, Bright field. Bars = 20 mm (B and C) and 25 mm (D). The boxed areas (B-D) are enlarged to show details. [See online article for color version of this figure.] CO 2 levels but not at higher CO 2 levels (Sabar et al., 2000;Dutilleul et al., 2003), demonstrating a role of complex I in light-dependent energy metabolism. However, because those complex I mutants affect either multiple complex I proteins or proteins not specific to photoautotrophic organisms, how the plant-specific subunits of complex I, especially that of the gCA subcomplex, might affect plant development remains unknown. Indeed, no photomorphogenic phenotype has been reported for those tobacco or Arabidopsis mutants (Sabar et al., 2000;Dutilleul et al., 2003;de Longevialle et al., 2007;Keren et al., 2012). The T-DNA insertion mutations of Arabidopsis genes encoding the gCA subunits have been reported previously, but none of those mutants showed visible phenotypic alterations, making it difficult to directly test the physiological function of the gCA subcomplex (Perales et al., 2005). For example, mutants impaired in the gCA2 or gCA3 genes exhibited morphologic phenotypes indistinguishable from that of the wild-type plants (Perales et al., 2005). A suspension culture derived from the gca2 mutant showed clearly reduced growth rate and respiration, but it remains unclear whether such a defect is dependent on light and how the phenotype of the suspension culture is directly related to specific aspects of the development of whole plants.

gCAL1 and gCAL2 Are Mitochondrial Proteins
We isolated clones corresponding to the Arabidopsis gCAL2 gene from a yeast two-hybrid screen in our search for proteins that interact with the blue light receptor CRY2 (Liu et al., 2008;Zuo et al., 2011;X. Yu, unpublished data). The Arabidopsis genome encodes three closely related gCAs, gCA1, gCA2, and gCA3, and two gCA-like proteins, gCAL1 and gCAL2, which together form the gCA subcomplex of the mitochondrial complex I Klodmann et al., 2010;Klodmann and Braun, 2011;Supplemental Fig. S1). The amino acid sequences of the three gCA subunits contain all critical and evolutionarily conserved residues in the gCAs, whereas the two gCAL subunits lack two critical His residues that are conserved in the carbonic anhydrases and, therefore, are not likely to be catalytically active Klodmann et al., 2010;Supplemental Fig. S1). gCA1, gCA2, and gCA3 are more closely related to each other (approximately 75% identity) than they are to gCAL1 and gCAL2 (approximately 30% identical; Supplemental Fig. S1). The gCAL1 and gCAL2 genes, which share approximately 90% amino acid sequence identity, appear to express constitutively without obvious responses to light or circadian rhythm (Supplemental Fig. S2). The gCAL1 and gCAL2 proteins have been previously identified in the Arabidopsis mitochondrial fraction by systematic mass spectrometry analyses (Sunderhaus et al., 2006;Klodmann et al., 2010). To verify the subcellular localization, we prepared and examined transgenic plants expressing gCAL1-YELLOW FLUORESCENT PROTEIN (YFP) and gCAL2-YFP proteins. We found that the gCAL1-YFP and gCAL2-YFP fusion proteins accumulate exclusively in mitochondria in the transgenic plants ( Fig. 1, B-E). Given that Arabidopsis CRY2 performs all the known functions and undergoes light-dependent protein modifications in the nucleus (Yu et al., 2007), gCAL2 seems unlikely to be directly involved in the function or regulation of CRY2, and our initial yeast two-hybrid result is most likely an often-encountered artifact in such experiments. However, we cannot completely exclude the remote possibility that CRY2 and gCAL2 may interact in the cytosol before they are imported into their respective organelles. Because of a potential involvement of these mitochondrial proteins in light-dependent plant growth and development (Brauna and Zabaleta, 2007), we continued our study to investigate the physiological functions of gCAL1 and gCAL2. gCAL1 and gCAL2 Are Essential Genes with Overlapping Functions We first collected and analyzed a complete set of T-DNA insertion mutants of the genes encoding each of the five subunits of the Arabidopsis gCA subcomplex, including gca1, gca2, gca3, gcal1, and gcal2 (Supplemental Figs. S3 and S4). None of the monogenic mutants or the ca1ca3 double mutant showed easily discernible phenotypic alternation (Supplemental Fig. S4). These results are consistent with previous reports on phenotypes of ca2 and ca3 mutants . Transgenic plants overexpressing the gCAL1 and gCAL2 genes also showed no apparent phenotypic alterations (X. Yu, unpublished data).
We reasoned that the gCAL1 and gCAL2 genes may have redundant functions, given that they share over 90% sequence identity (Supplemental Fig. S1), and prepared the double mutant by a conventional genetics method. However, we failed to identify the cal1cal2 double mutant after genotyping over 500 F2 progeny derived from reciprocal crosses of gcal1 and gcal2 mutants (X. Yu, unpublished data). Because plants homozygous for one gCAL gene but heterozygous for the other gene (c1c1C2c2 or C1c1c2c2) showed a normal phenotype, we suspect that the c1c1c2c2 double mutant may suffer from embryonic defects. To test this possibility, we examined young siliques derived from parents that are homozygous for the mutation of one gCAL gene but heterozygous for the mutation of the other gene (c1c1C2c2 or C1c1c2c2). We found that young siliques derived from the c1c1C2c2 or C1c1c2c2 parents contain approximately 20% to 25% colorless ovules ( Fig. 2A, left), which appeared to turn deep brown at the later stage of silique development ( Fig.  2A, right). Those colorless/deep-brown ovules were apparently delayed in embryogenesis to various extents, in comparison with that of the normal green ovules (Fig. 2, B and C). A statistical analysis demonstrated that the colorless/deep-brown phenotype of the c1c1C2c2 parents (23.91%) exhibits the 3:1 segregation ratio (Table I), indicating that those abnormal ovules are the double mutant progeny of the c1c1C2c2 parent. The colorless/deep-brown phenotype of the C1c1c2c2 parents (19.94%) appears to deviate from the 3:1 segregation (Table I). These results suggest that the cal2 mutation has a nearly normal penetrance for the colorless/deep-brown ovule phenotype, whereas the cal1 mutation may have a less than normal penetrance. Surprisingly, in contrast to a common embryonic defective mutant, we did not observe an abnormal number of aborted embryos in the mature siliques of the c1c1C2c2 and C1c1c2c2 plants ( Fig. 2A, right). We reasoned that the colorless/deep-brown ovules might complete their development to set seeds, but the double mutant seeds may fail to germinate. This explains why we were unable to find the double mutant in the F2 plants. To test this interpretation, we Figure 2. The gCAL1 and gCAL2 subunits of the gCA subcomplex is essential for Arabidopsis development. A, The immature siliques of wild-type (WT) and rescued mutant (35S::CAL1/c1c1c2c2) parents that show uniform and green-colored ovules, and the siliques derived from C1c1c2c2 or c1c1C2c2 hemizygote parents that show segregating colorless ovules indicated by arrowheads (left). The mature siliques from the wild-type and 35S::CAL1/c1c1c2c2 parents show uniformly yellow seeds, but those from the c1c1C2c2 hemizygote parent show segregating deep-brown seeds indicated by arrowheads (right). No dark-brown seed was found in the mature siliques of the C1c1c2c2 hemizygote parents, but no double homozygous progeny were identified. Bars = 0.5 mm. B, Embryos of the segregating progeny of the c1c1C2c2 hemizygote. Ovules from a young silique of the c1c1C2c2 hemizygote parent were cleared with Hoyer's solution and observed with a differential interference contrast microscope. The segregating white ovules arrested at various stages (c1c1c2c2) and the pale-green wild-type-looking ovule at the mature cotyledon stage from the same young silique are shown. Bars = 50 mm. C, Seeds (left) and embryos (right) of the nongerminating seeds (ngs) harvested from the C1c1c2c2 and c1c1C2c2 hemizygote parents. For the wild type, C1c1c2c2 and c1c1C2c2 seeds were placed on MS medium for 7 d. The nongerminating seeds were collected first for photographic record (left) and then dissected using a microscope. The dissected embryos corresponding to the seed (left) are shown on the right. D, Genotyping of the nongerminating seeds. Genomic DNA was isolated from the cal1 mutant, the cal2 mutant, and nongerminating seeds derived from the c1c1C2c2 parent. Genomic PCR was performed using the primers designed to amplify the wild-type gCAL1 (CAL1) and gCAL2 (CAL2) genes or the gcal1 (cal1) and gcal2 (cal2)  examined the germination rate of the seeds harvested from the selfed c1c1C2c2 and C1c1c2c2 parents (Table II). In contrast to the wild-type seeds that showed the germination rate of higher than 99%, about 20% to 25% of seeds derived from the selfed c1c1C2c2 and C1c1c2c2 parents failed to germinate (Table II). As a control, the seeds resulting from the reciprocal crosses between the wild-type and the c1c1C2c2 and C1c1c2c2 plants showed normal germination rates (Supplemental Table S1), suggesting that the nongerminating seeds derived from selfed c1c1C2c2 and C1c1c2c2 parents are the double mutants. Similar to the colorless/deep-brown ovule phenotype, the cal2 mutation had a nearly normal penetrance for the nongermination phenotype, whereas the cal1 mutation had a less than normal penetrance for the germination phenotype (Supplemental Table S1).
We confirmed that the colorless ovule and nongerminating seed phenotypes both result from the c1c1c2c2 double mutation by genotyping and complementation tests (Figs. 2D and 3; Tables I and II). We first collected the nongerminating seeds from the progeny derived from selfed c1c1C2c2 parents and genotyped those seeds. Indeed, the nongerminating seeds derived from the c1c1c2C2 hemizygote parents contained T-DNA inserts in both the gcal1 and gcal2 mutant loci, and they lacked the intact gCAL1 and gCAL2 genes (Fig. 2D). For the complementation assay, we transformed the C1c1c2c2 plants with the 35S:: CAL1 transgene and genotyped the transgenic individuals (Basta resistant) in the T2 generation (Fig. 3). All T2 Basta-resistant individuals grew normally and showed normal ovules and seed germination (Tables I  and II). We randomly selected some T2 Basta-resistant individuals to identify those that showed homozygous gcal1 and gcal2 mutations (c1c1c2c2; Fig. 3A). In contrast to our previous failure in identifying any c1c1c2c2 individuals homozygous for both the gcal1 and gcal2 mutations in over several hundred progeny of the C1c1c2c2 parent, we readily identified individuals homozygous for both mutations (c1c1c2c2) in the Basta-resistant transgenic progeny of the C1c1c2c2 parent (Fig. 3A). We then examined the existence of the 35S::CAL1 transgene in plants of the c1c1c2c2 genotype and found that all c1c1c2c2 double mutant plants also contained the 35S:: CAL1 transgene (Fig. 3B). This result confirmed that the c1c1c2c2 double mutant was rescued by the 35S::CAL1 transgene. We conclude that the cal1cal2 double mutation caused the defective embryogenesis and nongerminating seed phenotypes and that the gCAL1 and gCAL2 genes are essential for plant development.

Decreased Expression of gCAL1 and gCAL2 Altered Plant Light Responses
Since the monogenic gcal1-1 and gcal2-1 mutants showed no phenotype alteration whereas the c1c1c2c2 double mutant failed to germinate, we sought to generate intermediate phenotypes to facilitate functional analysis using a "knockdown" approach. We prepared gCAL2 RNA interference (RNAi) transgenic lines in the wild type or the cal1 mutant backgrounds, which are referred to as c2i or c1c2i knockdown lines, respectively. Both knockdown lines showed similar phenotypes (Figs. 4 and 5; Supplemental Fig. S5; data not shown), but we focused on the c1c2i knockdown lines for the following analyses (Figs. 4-7). The c1c2i knockdown lines showed reduced expression of the gCAL1 and gCAL2 genes without affecting the mRNA expression of the three gCA genes of the gCA subcomplex ( Fig. 5A; Supplemental Fig. S6). The phenotypic alterations of the c1c2i knockdown plants are observed in multiple independent transgenic lines (Figs. 4-7), arguing that the phenotypic alterations described are due to the reduced expression of the gCAL1 and gCAL2 genes. Consistent with the observation that the c1c1c2c2 double mutant failed to germinate, the c1c2i knockdown lines showed delayed germination (Fig. 4). After germination, the c1c2i plants appeared to grow normally, except that they exhibited smaller stature ( Fig. 5; Supplemental Fig. S5). By the time the wild-type plants developed four true leaves, the c1c2i plants had only two true leaves, and the leaves of the c1c2i plants remained markedly smaller than Table I. Approximately 20% to 25% of ovules of the c1c2C2c2 and C1c1c2c2 hemizygote parents show defective embryogenesis The young siliques of the indicated genotypes were dissected using a microscope to examine the ovules. The numbers of normal (green) and defective (colorless) ovules were calculated and are shown. The frequency of the colorless ovules of c1c1C2c2 (23.91%) is consistent with the 3:1 segregation ratio. The frequency of colorless ovules observed in young siliques of the C1c1c2c2 hemizygote parent (19.94%) deviates from the 3:1 segregation, indicating a less than perfect penetrance of the ycal mutation. The frequency of colorless ovules observed in young siliques derived from the 35S::CAL1/c1c1c2c2 transgenic plant (1.53%) is similar to that of the wild-type control (0.35%), indicating rescue of the colorless ovules of the c1c1c2c2 double mutant by the 35S::CAL1 transgene.  Fig. S5). It remains unclear whether the smaller stature of the c1c2i plants results from delayed germination or retarded growth or both. In addition to the effect on the organ and plant size, decreased expression of the gCAL1 and gCAL2 genes also affects leaf morphology and leaf senescence (Supplemental Fig. S7). The c1c2i seedlings grown in the dark exhibited typical etiolated phenotypes similar to the wild-type seedlings, including elongated hypocotyls, unexpanded cotyledons, and normal apical hook, although the c1c2i seedlings had slightly shorter hypocotyls than the wild-type control seedlings (Fig. 6A). However, when grown under light, the c1c2i seedlings developed a more pronounced shorter hypocotyl phenotype, and they also showed shorter petioles (Fig. 6, A and B). A fluence-rate analysis using different monochromatic lights demonstrates that c1c2i seedlings are hypersensitive to light in a wavelengthindependent manner (Fig. 6B). This result demonstrates a functional association between the mitochondrial gCA subcomplex and photomorphogenesis. Light-grown c1c2i seedlings also showed increased anthocyanin accumulation (Fig. 6A). The gene encoding the key enzyme of anthocyanin synthesis, chalcone synthase, was expressed higher in c1c2i than in the wild-type seedlings (Fig. 6C). Similar to the hypocotyl phenotype, CHS expression was markedly exaggerated in light-grown c1c2i seedlings relative to the etiolated c1c2i seedlings.
In addition to the altered photomorphogenic development of young seedlings, decreased expression of gCAL1 and gCAL2 also affects photoperiod-dependent reproductive development of the knockdown plants (Fig. 7). When grown in long-day photoperiods, the c1c2i plants clearly exhibited a late-flowering phenotype measured by both "days to flower" and "rosette leaf number" (Fig. 7A). Although the c1c2i plants grown in short-day photoperiods also took longer to flower than did the wild-type control as measured by days to flower, the c1c2i and wild-type plants developed the same number of rosette leaves at the time of flowering in short days (Fig. 7B). This result argues that the apparent late-flowering phenotype of the c1c2i plants grown in the short-day photoperiod is due to a retardation of growth rather than a delay in floral initiation per se (Koornneef et al., 1991). Consistent with the notion that gCAL1 and gCAL2 expression affects photoperiodic flowering, genes closely associated with the photoperiodic control of floral initiation, such as CO, SOC1, and FT, showed markedly reduced mRNA expression in the c1c2i plants, whereas another floweringtime gene, FLC, which is an autonomous pathway gene not directly involved in the photoperiodic control of flowering time, showed normal expression in the c1c2i plants (Searle and Coupland, 2004;Sung and Amasino, 2005;Bäurle and Dean, 2006;Fig. 7C). Importantly, the development-dependent change of mRNA expression Table II. Approximately 20% to 25% of seeds of the c1c2C2c2 and C1c1c2c2 hemizygote parents failed to germinate Seeds harvested from individual siliques of the indicated genotype were germinated on MS plates for 7 d, and the frequency of nongerminating seeds was calculated on the basis of a single silique. The frequency of nongerminating seeds harvested from the c1c1C2c2 hemizygote parents (24.58) is consistent with the 3:1 segregation ratio. The frequency of nongerminating seeds harvested from the C1c1c2c2 hemizygote parent (20.31%) deviates from the 3:1 segregation, indicating a less than perfect penetrance of the gcall mutation. The frequency of nongerminating seeds harvested from the 35S::CAL1/c1c1c2c2 transgenic plant (0.32%) is similar to that of the wild-type control (0.39), indicating rescue of the colorless ovules of the c1c1c2c2 double mutant by the 35S::CAL1 transgene. for the photoperiodic pathway genes FT and SOC1, but not the autonomous pathway gene FLC, is almost abolished in the c1c2i plants (Fig. 7C). Given the essential role of gCAL1 and gCAL2 in plant development , this result argues that the developmental impacts of these two mitochondrial proteins affect the photoperiodic pathway more than the autonomous pathway that controls floral initiation. Taken together, we conclude that gCAL1 and gCAL2 play important roles in light-dependent plant growth and development.

DISCUSSION
We have demonstrated in this report the physiological role of a plant-specific gCA subcomplex of mitochondrial complex I in Arabidopsis. We found that gCAL1 and gCAL2 are not only essential for plant development, they are also involved in the light regulation of seedling development and photoperiodic control of flowering. The association of gCAL1 and gCAL2 with light-dependent plant development appears to provide a functional and evolutionary explanation for why the gCA subcomplex is found only in the mitochondrial complex I of photoautotrophic organisms. On the other hand, our study also raised several questions that remain to be further investigated.
It remains unclear how mitochondrial proteins gCAL1 and gCAL2 affect the light regulation of plant growth. The mRNA abundances of the nuclear genes gCAL1 and gCAL2 are not apparently regulated by light (Supplemental Fig. S2). It is possible that light may affect the protein modification or abundance of gCAL1 and gCAL2. Our attempts to raise antibodies against the gCAL1 and gCAL2 proteins were unsuccessful, so these hypotheses cannot be directly tested at present. Alternatively, the gCA subcomplex may affect the composition and activity of complex I, which in turn affects energy metabolism in response to light, such as photosynthesis. Indeed, the c1c2i plants exhibited a modest but statistically significant (P , 0.05) reduction in the photochemical efficiency of PSII, especially at a relatively high fluence rate of light (Supplemental Fig.  S8). This result is consistent with the previous finding that the mitochondrial complex I is important for photosynthesis (Sabar et al., 2000;Dutilleul et al., 2003). The trivial explanation of a change in the composition of light harvesting and photosystem stoichiometry is unlikely, because the chlorophyll a/b ratios are unchanged and blue native gel electrophoresis of thylakoid membranes does not reveal any differences in the formation of supercomplexes of photosystems and light-harvesting complexes (Supplemental Fig. S9). There remains the possibility that the biogenesis or maintenance (repair  cycle) of PSII is impacted, which is consistent with the exacerbated phenotype with a higher light fluence (Supplemental Fig. S8). The reduced photochemical efficiency of PSII of the c1c2i plants may result from an effect on photorespiration that occurs partly in the mitochondrion. It is known that complex I affects photorespiration (Sabar et al., 2000;Dutilleul et al., 2003), and it has also been proposed that the gCA subcomplex may play a role in photorespiration (Brauna and Zabaleta, 2007). The gCA subunits (gCA1-3) of the gCA subcomplex may, by converting CO 2 to HCO 3 2 , facilitate the recycling of CO 2 released at the decarboxylation step of photorespiration to the chloroplast. Therefore, we investigated whether the gCA subcomplex may affect photorespiration by testing whether the c1c2i phenotype might be rescued by a high concentration of CO 2 . It is well known that a high CO 2 concentration can suppress photorespiration and photorespiration-dependent phenotypes of photorespiration-defective mutants (Somerville and Ogren, 1980;Foyer et al., 2009;Peterhansel et al., 2010). However, increased CO 2 concentration failed to rescue the reduced PSII efficiency of the c1c2i plants (Supplemental Fig. S9). Consistent with this result, the c1c2i plants grown at a high level of CO 2 (5%) also showed a similar morphological phenotype as that grown in normal air (approximately 0.03% CO 2 ; Supplemental Fig. S10). Therefore, the reduced PSII efficiency and growth defects of the c1c2i plants may not be simply explained by the defective photorespiration. It should be noted that our result does not necessarily imply a lack of function of the gCA subcomplex in photorespiration, because all three gCA genes are apparently expressed normally in the c1c2i mutant (Supplemental Fig. S5).
The biochemical function that gCAL may play in complex I is not clear at present, nor is it clear why the reduced level of gCAL1 and gCAL2 proteins affect photomorphogenesis. As expected, gCAL2 showed no carbonic anhydrase activity (Supplemental Fig. S11A). On the other hand, it is known that the gCAs contain the hexapeptide repeats ([LIV]-X2-[STAV]-X) that form a lefthanded b-helix structure (Gaedeke et al., 2001;Perales et al., 2004;Ferry, 2010). Sequence analysis and structural modeling indicate that both gCAL proteins contain the hexapeptide repeats and the left-handed b-helix folds (Supplemental Fig. S11B). The left-handed b-helix structure is widely found in members of the acyltransferase superfamily (Raetz and Roderick, 1995). This superfamily contains enzymatically unrelated proteins, including acetyltransferases, such as Ser acetyltransferases of bacteria and mitochondrial Ser acetyltransferases of Arabidopsis ;  Pye et al., 2004;Haas et al., 2008). The left-handed b-helix structure constitutes the active site of the acetyltransferases (Pye et al., 2004). Interestingly, gCAL2 appears to exhibit a protein acetyltransferase activity in vitro (Supplemental Fig. S11C). The in vivo substrate of gCAL2 remains unclear at present. Our attempts to test a possible acetyltransferase activity with other subunits of complex I was unsuccessful, which may be due to the difficulties in the purification of native subunits of this membrane complex. Given that fatty acids of seed lipids are converted to acetyl-CoA during germination for the synthesis of compounds essential for germination and early seedling growth, a potential acetyltransferase activity of gCAL2, which utilizes acetyl-CoA as a substrate in vitro, may partially explain the germination defect of the gacl1gcal2 double mutant. It is conceivable that the energy metabolism is important for photomorphogenesis in general, and certain malfunctions in mitochondria may adversely affect photomorphogenesis. Specifically, it is tempting to speculate that gCAL proteins may acetylate other mitochondrial proteins in response to the changing light environment and acetyl-CoA concentration to indirectly affect plant photomorphogenesis. This proposition, however, remains to be further investigated.

MATERIALS AND METHODS
Most Arabidopsis (Arabidopsis thaliana) gca and gcal mutants and all transgenic lines described in this study are from the Columbia accessions, except the gcal2 mutant, which is from the Wassilewskija accession. All the mutants were obtained from the Arabidopsis Biological Resource Center. Plant transformation and genetic, photomorphogenic, DNA, and RNA analyses were as described previously (Yu et al., 2007;Liu et al., 2008;. For the embryonic phenotype analyses, young (green) and mature (brown) siliques (15-20 per sample) were collected from 3-to 4-week-old plants, opened with a needle using a dissecting microscope, and examined with the microscope. To measure germination rates, seeds from individually collected siliques were grown in Murashige and Skoog (MS) medium for 7 d under continuous light before counting the germinating seeds. Additional methods and primers (Supplemental Table S2) used in the genomic and quantitative PCR can be found in the Supplemental Materials and Methods S1.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Sequence analysis of the gCAL and gCA proteins.
Supplemental Figure S2. Light-independent mRNA expression of gCAL1 and gCAL2.
Supplemental Figure S3. Genotyping of the gcal1 and gcal2 mutants.
Supplemental Figure S4. Phenotypes of adult plants of the indicated genotypes.
Supplemental Figure S5. mRNA expression of the gCA genes are not affected in the c1c2i knockdown plants.
Supplemental Figure S6. Analyses of transgenic lines expressing the CAL2-RNAi construct in the wild-type background (c2i).
Supplemental Figure S7. Morphological defects of leaves of the c1c2i knockdown plants.
Supplemental Figure S8. The growth retardation phenotype of the c1c2i knockdown plants is not rescued by the high-CO 2 growth condition.
Supplemental Figure S9. Reduced expression of the gCAL1 and gCAL2 genes modestly affects the maximum quantum efficiency of PSII.
Supplemental Figure S10. Normal photosynthetic apparatus of the c1c2i mutant plants.
Supplemental Figure S11. Possible biochemical functions of the gCAL proteins.
Supplemental Table S2. Oligonucleotide primers used in this study.
Supplemental Materials and Methods S1.