Photorespiratory metabolism is essential in all oxygenic photosynthetic organisms. In plants, it is a highly compartmentalized pathway that involves chloroplasts, peroxisomes, mitochondria and the cytoplasm. The metabolic pathway itself is well characterized, and the enzymes required for its function have been identified. However, very little information is available on the transport proteins that catalyze the high metabolic flux between the involved compartments. Here we show that the A BOUT DE SOUFFLE (BOU) gene, which encodes a mitochondrial carrier, is involved in photorespiration in Arabidopsis. BOU was found to be co-expressed with photorespiratory genes in leaf tissues. The knockout mutant bou-2 showed the hallmarks of a photorespiratory growth phenotype, an elevated CO2 compensation point, and excessive accumulation of glycine. Furthermore, degradation of the P-protein, a subunit of glycine decarboxylase, was demonstrated for bou-2, and is reflected in strongly reduced glycine decarboxylase activity. The photorespiration defect in bou-2 has dramatic consequences early in the seedling stage, which are highlighted by transcriptome studies. In bou-2 seedlings, as in shm1, another photorespiratory mutant, the shoot apical meristem organization is severely compromised. Cell divisions are arrested, leading to growth arrest at ambient CO2. Although the specific substrate for the BOU transporter protein remains elusive, we show that it is essential for the function of the photorespiratory metabolism. We hypothesize that BOU function is linked with glycine decarboxylase activity, and is required for normal apical meristems functioning in seedlings.
The enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is the central enzyme of carbon dioxide (CO2) fixation for all phototrophic organisms. Carboxylation of ribulose-1,5-bisphosphate (RuBP) yields two molecules of 3-phosphoglyerate (3-PGA), which are converted to triose phosphates using chemical energy in the form of ATP and NADPH provided by photosynthetic light reactions (Calvin, 1962). However, Rubisco also accepts oxygen (O2) as a competitive substrate (Bowes et al., 1971; Ogren, 1984). Oxygenation of RuBP yields only one molecule of 3-PGA and one molecule of 2-phosphoglycolate (2-PG). The latter molecule represents a metabolic dead end, and is inhibitory if it accumulates in high amounts (Kelly and Latzko, 1976). It must therefore be removed from the chloroplast and recycled to 3-PGA in a complex highly compartmentalized pathway named photorespiration. In photorespiration, two molecules of 2-PG are recycled to one molecule of 3-PGA in a series of enzymatic steps distributed across four subcellular compartments (Reumann and Weber, 2006). 2-PG is hydrolyzed to glycolate in the chloroplast (Somerville and Ogren, 1979; Schwarte and Bauwe, 2007), and metabolized to glyoxylate and further to glycine in the peroxisomes (Volokita and Somerville, 1987; Igarashi et al., 2003). Glycine oxidation and serine synthesis take place in the mitochondria (Somerville and Ogren, 1981; Voll et al., 2006; Engel et al., 2007), and serine is shuttled back to the peroxisomes where it is deaminated to form hydroxypyruvate (Liepman and Olsen, 2001) and subsequently reduced to glycerate (Murray et al., 1989; Timm et al., 2008). Glycerate is imported and finally phosphorylated to 3-PGA in the chloroplast (Boldt et al., 2005). The resulting 3-PGA can then re-enter the Calvin–Benson cycle. In summary, two molecules of 2-PG are converted to one molecule each of 3-PGA, CO2 and ammonia (NH3). Hence, in addition to the Calvin–Benson cycle, photorespiration is also interlinked with the nitrogen cycle (Ogren, 1984; Keys, 2006). It is widely accepted that NH3 released during glycine oxidation in the mitochondria enters the chloroplast, where it is reassimilated by the concerted action of glutamine synthetase and glutamate synthase (Wallsgrove et al., 1987; Coschigano et al., 1998). The product, glutamate, is transferred to the peroxisomes where it is utilized to produce glycine from glyoxylate, and hence a new round of nitrogen cycling is initiated.
The basic necessity of photorespiration for all oxygenic photosynthetic organisms (Eisenhut et al., 2008; Bauwe et al., 2012) is apparent in the phenotypes of photorespiration mutants, most of which show stunted growth and leaf defects in normal air. Typically, this so-called photorespiratory phenotype may be relieved in a high-CO2 environment (Somerville and Ogren, 1979).
As outlined above, photorespiration is a prime example of pathway compartmentalization in eukaryotic cells, as enzymes for the carbon and nitrogen cycles are partitioned in almost equal parts between chloroplasts, peroxisomes and mitochondria. Despite the excellent progress in understanding the photorespiratory pathway, to date none of the metabolite transporters that move the photorespiratory intermediates through the cell have been identified (Reumann and Weber, 2006; Eisenhut et al., 2012). So far, only two affiliated transport systems have been identified: the plastidial dicarboxylate (2-oxoglutarate/malate) transporters DiT1 (Kinoshita et al., 2011) and DiT2.1 (Somerville and Ogren, 1983; Renné et al., 2003) are essential for unctional photorespiratory metabolism.
In this paper, we investigated the role of the A BOUT DE SOUFFLE (BOU) protein in photorespiratory metabolism by a reverse genetic approach. Physiological characterization of the knockout mutant bou-2 revealed a photorespiratory growth phenotype and an increased CO2 compensation point. bou-2 seedlings were unable to undergo transition to autotrophy due to impaired meristem activity. Strong accumulation of glycine as a result of reduced glycine decarboxylase (GDC) activity suggested a central role of the mitochondrial membrane protein BOU in transport of an unknown compound that is crucial for GDC activity, which itself is required for photoautotrophic growth and proper function of apical meristems.
Identification of BOU as candidate for a transporter involved in photorespiratory metabolism
The core photorespiratory metabolism is distributed over the three compartments chloroplast, peroxisome and mitochondrion. Therefore, transporters must exist to facilitate coordinated channeling of photorespiratory intermediates and co-factors of photorespiratory enzymes. To identify candidate photorespiratory transporter proteins, we performed co-expression analyses using the ATTED-II database (http://atted.jp/, Obayashi et al., 2007). BOU was among the top 5% of transcripts co-regulated with a set of 14 highly co-expressed photorespiratory genes (Table S1). BOU belongs to the mitochondrial carrier family and was previously hypothesized to function as a carnitine/acylcarnitine carrier (Lawand et al., 2002). Importantly, the BOU gene is strongly co-expressed with the genes GDP2 (At2 g26080), GDP1 (At4 g33010), GDT1 (At1 g11860) and GDH3 (At1 g32470), encoding subunits of GDC, and SHM1 (At4 g37930), encoding the mitochondrial serine hydroxymethyl transferase SHM1 (Figure 1a). Both GDC and SHMT are central to photorespiration and C1 metabolism. Furthermore, BOU promoter activity was observed in cotyledons and leaf tissues, and appeared to be absent from meristematic tissues (Figure 1b–d). BOU gene expression is light-inducible and dependent on photon flux density (Figure 1b–d and Figure S1a,b) (Lawand et al., 2002) and plastid signals (Figure S1c). These observations link BOU gene expression to photosynthetic processes, and led us to investigate the hypothetical function of BOU in photorespiration.
The T-DNA knockout line bou-2 shows a photorespiratory phenotype
To test the hypothesis that BOU is involved in photorespiratory metabolism, we established a homozygous knockout line that harbors a T-DNA insertion in the second exon of the BOU gene, named bou-2 (Figure S2a,b). BOU protein was not detectable in mitochondrial protein extracts from bou-2, indicating complete loss of the protein in the knockout line (Figure S2c).
Next, we analyzed the bou-2 mutant for the hallmark features of a typical photorespiratory mutant (von Caemmerer, 2000; Somerville, 2001; Igarashi et al., 2003; Boldt et al., 2005; Voll et al., 2006; Schwarte and Bauwe, 2007; Timm et al., 2008). Mutants with a photorespiratory phenotype show strongly reduced or fully inhibited growth and chlorotic leaves in ambient air (380 ppm CO2), but are indistinguishable from wild-type (WT) plants in CO2-enriched air (Somerville and Ogren, 1979). As shown in Figure 2(a,b), bou-2 is sensitive to ambient air as demonstrated by chlorotic leaves and impaired growth. These effects are almost fully compensated for when the mutant is grown at 3000 ppm CO2. We performed photosynthetic gas exchange measurements and found considerably elevated CO2 compensation points for the bou-2 mutant (Figure 2c). When grown in an atmosphere enriched with 3000 ppm CO2, the CO2 compensation point of bou-2 plants (136.7 ± 9.9 μl l−1) was almost three times greater than that of WT plants (54.4 ± 0.7 μl l−1). However, 3 days after transfer of the plants to ambient air, the CO2 compensation point of bou-2 was even higher (293.8 ± 35.8 μl l−1), but the value for WT plants did not change significantly (52.8 ± 3.6 μl l−1). As the CO2 compensation point reflects the balance of photosynthetic CO2 uptake and photorespiratory CO2 release (von Caemmerer, 2000), the elevated CO2 compensation point also supported our hypothesis that BOU is involved in photorespiration. The CO2 fixation rates measured at 400 ppm CO2 and saturating light intensity (1500 μmol photons m−1 sec−1) are strongly reduced in bou-2 (36% of WT activity) and almost fully impaired (6.9% of WT activity) in the mutant 3 days after transfer to ambient air (Figure 2d).
Leaf metabolite profiles demonstrate heavily disturbed photorespiratory metabolism in bou-2
For several photorespiratory mutants, it has been demonstrated that one or more intermediates of the pathway accumulate after transfer to ambient air (Somerville, 2001). To corroborate the results of phenotypic characterization of bou-2, we performed metabolite analysis in WT and bou-2 upon a shift from 3000 ppm CO2 to 380 ppm CO2 (Figure 3). The most dramatic change was observed for the amino acid glycine (Figure 3a). The glycine concentration in bou-2 rapidly increased from seven times greater than WT to 17.5 times greater than WT within 17 h. In contrast, bou-2 contained only 60% of the amino acid serine under both elevated and ambient CO2 conditions (Figure 3b). Furthermore, the shift in CO2 concentrations strongly affected the levels of glutamate and 2-oxoglutarate. In bou-2, the glutamate concentration decreased to 30% of the WT level (Figure 3c), whereas 2-oxoglutarate increased tenfold after the shift to ambient CO2 (Figure 3d). Significant changes were also detected for other photorespiratory marker metabolites such as glycolate (Figure 3f) and glycerate (Figure 3g). It is noteworthy that γ-aminobutyrate (GABA) accumulates significantly in bou-2 before (8.5-fold) and after the shift (10.7-fold) (Figure 3e).
As a control, we also analyzed shm1, a mutant that is deficient in mitochondrial SHM1 activity (Voll et al., 2006). With the exception of glycerate, the alterations of metabolite levels in bou-2 and shm1 were similar, suggesting a closely connected function of both proteins in photorespiratory metabolism. The relative concentrations of all quantified metabolites are presented in Table S2.
Molecular complementation of the bou-2 mutant
A bou-2 mutant expressing BOU under the control of the 35S promoter, bou-2/BOU (Figure S3a), germinated and developed normally under ambient air conditions (Figure S3b) and no significant glycine accumulation was observed (Figure S3c), showing that the BOU mutation causes the CO2-dependent phenotype.
GDC activity is strongly reduced in bou-2
We hypothesized that the disturbed glycine/serine homeostasis in bou-2 was caused by a defect in photorespiratory glycine oxidation, either because of a deficiency in glycine uptake into the mitochondria or lack of GDC activity. Therefore, we analyzed in detail the enzymes involved in inter-conversion of these amino acids. Mitochondrial GDC is a multi-enzyme system consisting of the four subunits: P-protein (GDP), H-protein (GDH), T-protein (GDT) and L-protein (mLPD). Furthermore, lipoylation of the H-protein is essential for GDC activity (Douce et al., 2001; Ewald et al., 2007). The second partner essential for glycine to serine conversion is the mitochondrial serine hydroxymethyl transferase (SHM1). We performed protein gel-blot analyses with specific antibodies against all four GDC subunits, SHM1 and lipoic acid (LA) (Figure 4a). Levels of GDH (Figure 4a), GDT, mLPD and SHM1 were not influenced by loss of the BOU protein (Figure S4), but distinct changes were found for GDP. In protein extracts from bou-2 grown at 1% CO2, the amount of GDP was somewhat elevated compared to WT. In bou-2 plants shifted to ambient air, we detected decreased amounts of GDP. Intriguingly, an additional signal was detected using the GDP-specific antibody. This signal probably originated from a degradation product of GDP. This GDP fragment also occurred in WT (Figure S4), but accumulated to a much greater extent in bou-2 after the shift to ambient air (Figure 4a). Lipoylation of GDH and the E2 subunits of pyruvate dehydrogenase was not affected in bou-2, but the amount of lipoylated E2 subunit of α-ketoacid dehydrogenase was increased in protein extracts from bou-2 plants after the shift to ambient air (Figure 4a).
The clear changes in the amount and status of GDP prompted us to measure the activity of the GDC. Therefore, mitochondria were isolated from WT and bou-2 plants grown under 3000 ppm CO2, and GDC activity was determined (Figure 4b). In bou-2, the apparent rate of the GDC reaction was strongly reduced to only 15.2% of that observed for WT. The results suggest that the transport activity of BOU is linked with the function and activity of the GDC.
bou-2 plants have an amino acid accumulation defect at the seedling stage
A striking feature of bou-2 is the accumulation of glycine at the seedling stage (Table 1). Glycine was more abundant in 5-day-old bou-2 seedlings than in WT, as were glutamine, serine, ornithine and citrulline (Table 1). bou-2 and shm1 seedlings grown in vitro for 5–8 days accumulated high glycine levels (Figure S8). This indicates that the bou-2 and shm1 mutations lead to similar photorespiratory metabolism defects as early as the seedling stage.
Table 1. Comparative amino acid content (nmol per mg FW) in 5-day-old seedlings
ud, undetectable; nd, not determined; *Indicates significant statistical difference (Student t-test with p < 0.05). Non standard amino acids: CIT, citrulline, ORN, ornithine, GABA, gamma-aminobutyric acid.
0.33 ± 0.03
0.54 ± 0.06
0.54 ± 0.07
1.64 ± 0.13
0.75 ± 0.06
0.95 ± 0.07
1.31 ± 0.2
0.50 ± 0.02
0.05 ± 0.01
1.22 ± 0.12
0.06 ± 0.01
0.53 ± 0.06
3.79 ± 0.33
20.72 ± 1.58
2.25 ± 0.4
1.07 ± 0.14
0.24 ± 0.05
53.24 ± 3.49
0.05 ± 0.01
0.28 ± 0.03
0.03 ± 0.00
0.19 ± 0.03
0.09 ± 0.03
0.73 ± 0.14
0.01 ± 0.01
0.13 ± 0.2
0.42 ± 0.17
1.63 ± 0.18
3.41 ± 0.28
0.43 ± 0.05
0.21 ± 0.02
0.49 ± 0.1
0.33 ± 0.2
To relate this to altered gene expression in bou-2 seedlings, we analyzed the bou-2 transcriptome and found that 1527 genes categorized according to the Munich Information Center for Protein Sequences (MIPS; http://mips.helmholtz-muenchen.de/proj/funcatDB/) functional catalogue interface were differentially expressed compared to WT (Usadel et al., 2005; Ruepp et al., 2004) (Figure 5, Figure S7 and Table S3). The global changes in gene expression observed would have an impact on the levels of amino acids in planta, particularly of glycine. Genes associated with photorespiratory and amino acid metabolism were de-regulated in bou-2 seedlings (Figure 5a,b and Table S5). Expression of the GDCH1 and GDCT genes, encoding glycine decarboxylase subunits, is down-regulated in bou-2. Other genes involved in glycine metabolism were also de-regulated. Genes encoding an alanine glyoxylate:aminotransferase (AGT3) involved in conversion of alanine and glyoxylate to glycine and pyruvate and a threonine aldolase (THA1) involved in the synthesis of glycine from threonine (Joshi et al., 2006) were both up-regulated. Photorespiratory and amino acid metabolism is thus de-regulated at the gene expression level in bou-2 seedlings.
bou-2 displays growth defects and chlorosis at the seedling stage
As originally demonstrated by Lawand et al. (2002), knockout mutants of BOU (bou-1 mutant) seedlings germinate in continuous light but stop growing just after the pale-green cotyledons have developed unless they are grown on sucrose-containing medium (Lawand et al., 2002). We investigated this growth defect in more detail by comparing bou-2 and shm1 at the seedling stage. Unlike the WT, which has already initiated leaves 5 days after germination, both bou-2 and shm1 stop growing and their development remains arrested (Figure 6a). At 8 days, cotyledons of bou-2 and shm1 are chlorotic (Figure 6a), suggesting a defect in photosynthesis.
This is reflected in the transcriptome of bou-2 seedlings, in which carbon fixation and photosynthetic genes are down-regulated (Figure 5a,b and Table S5), and 36% of the repressed genes are chloroplast-related (Figure S7b). Carbon assimilation genes such as two Rubisco small subunit genes (RBCS1A and RBSC3B), Rubisco activase (RCA), Calvin–Benson cycle genes and three carbonic anhydrase genes were down-regulated in bou-2. By contrast, chlorophyllase ATCHL2 (At5 g43860), the first enzyme in chlorophyll degradation (Tsuchiya et al., 1999), was up-regulated (Figure 5b and Table S5). Given this defect in carbon fixation and metabolism in the bou-2 mutant, we hypothesized that it may not be able to accumulate starch, and indeed starch was not detected in root caps (Figure S5).
The shoot apical meristem (SAM) is defective in bou-2 and shm1
The WT shoot apical meristem (SAM) develops between the cotyledons, has three characteristic external cell layers (L1–L3) and is made up of a central zone, peripheral zones and medullary meristems. Five-day-old WT seedlings had a typical dome-shaped SAM, indicative of cell-cycle activity of SAM peripheral zones, which gave rise to formation of several leaves (Figure 7a). By contrast, 5-day-old bou-2 and shm1 seedlings had a flat and disorganized SAM (Figure 7b,c). In addition, for these two mutants, leaf initiation was either absent or limited to one leaf (Figure 7b,c), indicative of low SAM cell-cycle activity. Eight-day-old WT plants display an enlarged SAM typical of the pre-floral phase (Figure 7d), while the SAM of bou-2 and shm1 appears flat, disorganized and formed of enlarged and vacuolated cells (Figure 7e,f). Thus both bou-2 and shm1 seedlings were unable to sustain cell-cycle activity in their SAM to form new organs.
It has been shown that the cells of the WT Arabidopsis SAM express Shoot meristemless (STM), a transcription factor that defines meristematic cell identity (Long et al., 1996). Expression of an STM promoter–GUS reporter gene was always detected in the bou-2 mutant (Figure S6), suggesting that meristem cell identity is somehow maintained despite the strong meristem disorganization and low cell-cycle activity (Long et al., 1996; Kirch et al., 2003). The maintenance of this meristematic identity explains why 5-day-old bou-2 and shm1 seedlings supplemented with sucrose produce a typical dome-shaped SAM made up of small cells with dense cytoplasm (Figure 7h,i).
Growth from meristems requires photorespiration or high CO2
Growth from meristems occurs by coordinated cell divisions that maintain the meristem and initiate organ primordia. Cell division and expansion in primordia lead to organ formation (Barton and Poethig, 1993; Long et al., 1996; Fletcher, 2002; Weigel and Jürgens, 2002; Kirch et al., 2003). We crossed the CycB1;1::uidA reporter gene, which is specifically expressed in the G2 and mitosis (M) phases of the cell cycle, into the WT and bou-2 backgrounds (Colón-Carmona et al., 1999). WT shoot apices were GUS-stained, revealing that numerous cells are in the G2 and M phases and thus dividing. bou-2 seedlings showed little or no GUS staining in the apical region, so cells were either at other phases of the cell cycle or had exited the cell cycle (Figure 6b). When supplemented with sucrose for 24 h, mutant seedlings displayed GUS staining in shoot apices (Figure 6b). Root growth was also arrested in bou-2 seedlings. Cell division generates cells in the meristematic region of the root. In WT root apices, strong scattered GUS staining confirmed that cells in this region are undergoing mitosis (Figure 6b). However, little or no reporter gene activity was detected in bou-2, but this was reversed if sucrose was provided (Figure 6b).
The effect of 1% CO2 on cell division was also investigated. Scattered dense staining of both leaf primordia and root meristems of CycB1;1::uidA/bou-2 plants showed that cell division resumed 1 day after bou-2 was transferred to 1% CO2 (Figure 6b). The reversion of the mutant phenotype at high CO2 levels indicated that defective photorespiration in bou-2 impairs CO2 assimilation and sugar availability, leading to the arrested growth phenotypes.
In the transcriptome analysis, cell cycle S-phase markers such as ribonucleotide reductase and histone genes HIS2A and HIS4 were strongly repressed in bou-2, but a linker histone gene, HIS1-3, was expressed at a higher level in bou-2 (Table S4) (Chaubet et al., 1996). These results confirm at the molecular level that the bou-2 seedling growth arrest is the consequence of cell division arrest.
Thus the photorespiratory mutants bou-2 and shm1 display a similar phenotype, suggesting that the arrested growth after 5 days is not unique to the bou-2 mutant. The defective photorespiratory pathway in these two mutants has strong developmental consequences as it prevents proper maintenance and function of the meristems after germination.
The BOU gene encodes a six-transmembrane domain protein that belongs to the mitochondrial carrier family (MCF). Eukaryotic MCF carriers share a similar structure, with three 100 amino acid repeats each forming two transmembrane helices. The substrates of MCF proteins range from anionic (ADP/ATP, phosphate) to neutral or cationic molecules, such as basic amino acids, carnitine and lipid-derived organic molecules (Palmieri et al., 2006; Palmieri and Pierri, 2010). There are 58 genes in Arabidopsis that encode putative MCF proteins (Picault et al., 2004 Palmieri et al., 2011). The location of MCF proteins in Arabidopsis is not restricted to mitochondria. MCF proteins may also be located in chloroplast (Ferro et al., 2002; Bedhomme et al., 2005; Bouvier et al., 2006) and peroxisomal membranes (Linka et al., 2008; Bernhardt et al., 2012), as well as in the endomembrane system (Leroch et al., 2008) and the plasma membrane (Rieder and Neuhaus, 2011). Importantly, BOU was shown to reside in the mitochondrial membrane (Lawand et al., 2002). Amino acid similarity analysis groups the BOU protein with carnitine carriers and basic amino acid carriers from yeast and mammals (Lawand et al., 2002; Picault et al., 2004; Toka et al., 2010). It was previously hypothesized that BOU may transport lipid-derived acetylcarnitine to the mitochondria following germination and fatty-acid β-oxidation in the glyoxysome (Lawand et al., 2002). However, several results challenge this hypothesis. First, although carnitine has been associated with lipid metabolism in Arabidopsis, it is not abundant (Bourdin et al., 2007). Second, the carnitine-dependent lipid metabolic pathway described in fungi and animals is different from that operating in plants. In plants, lipid-derived fatty acids are entirely degraded by β-oxidation in the glyoxysome, which produces malate that is transported to the mitochondria (Graham, 2008).
The carbon and nitrogen cycles of the photorespiratory pathway are distributed across three compartments, acting jointly to recycle two 2-PG molecules into one 3-PGA (Tolbert, 1997). A prerequisite for trafficking of the photorespiratory intermediates are numerous transport proteins inserted into the plastid, peroxisomal and mitochondrial membranes. So far, only the two dicarboxylate transporters DiT1 (Kinoshita et al., 2011) and DiT2.1 (Renné et al., 2003), which are both required for photorespiration, have been identified. We provide strong evidence in this study for a role of BOU as a photorespiratory transporter.
BOU is essential for functional photorespiration
Co-expression analysis is a powerful tool for assigning proteins of as yet unknown function to a biological process (Obayashi and Kinoshita, 2010). We found that the BOU gene is strongly co-expressed with genes encoding subunits of the GDC (Figure 1a), a central enzyme system of the photorespiratory pathway. We therefore hypothesized a photorespiration-related function for BOU. This hypothesis was supported by the observation that BOU expression is light-regulated (Figure 1b–d and Figure S1a–c). The physiological analysis of bou-2 presented here clearly associates BOU with photorespiratory metabolism. We detected a typical photorespiratory growth phenotype (von Caemmerer, 2000; Somerville, 2001; Igarashi et al., 2003; Boldt et al., 2005; Voll et al., 2006; Schwarte and Bauwe, 2007; Timm et al., 2008), with inhibited growth and chlorotic leaves in ambient air, but not in CO2-enriched air (Figure 2a,b). The significantly elevated CO2 compensation points for the bou-2 mutant (Figure 2c) probably result from reduced CO2 fixation rates (Figure 2d) and increased respiration as suggested from elevated levels of the TCA cycle intermediates isocitric acid, succinic acid, fumaric acid and malic acid (Table S2). Finally, the CO2-dependent accumulation of photorespiratory intermediates (Figure 3 and Table S2) strongly supports a photorespiration-related function for BOU.
Ineffective photorespiration impairs Arabidopsis seedling development at ambient CO2
In a previous study, it was shown that, when grown in vitro, the bou-1 mutant stops growing at the seedling stage unless sugar is provided in the growth medium, and has low chlorophyll content. This phenotype is more exaggerated in continuous light than under dark/light growth conditions (Lawand et al., 2002). We found a similar phenotype for bou-2 seedlings (Figure 5a).
Arabidopsis seedlings germinate as heterotrophic organisms, using lipids and proteins stored in embryo cells as a source of organic molecules and energy (Martin et al., 2002; Graham, 2008). The ability to switch to autotrophic metabolism requires correct utilization of stored molecules for building functional chloroplasts that are vital for both carbon and nitrogen assimilation, leading to seedling establishment.
Some Arabidopsis mutants are blocked at this critical developmental stage. Storage molecule utilization mutants such as the ped mutant, which is impaired in β-oxidation (Hayashi et al., 1998), or the acyl CoA oxidase (ACX) mutant (Pinfield-Wells et al., 2005), are unable to mobilize fatty acids after germination and stop growing at the establishment stage post-germination (Graham, 2008). Growth of these mutants may be restored by externally added sugar. When environmental conditions are not optimal, a developmental block may also prevent development of a seedling into an autotrophic plant. For example, the stress hormone abscisic acid and hyperosmotic stress trigger growth arrest via ABI3, ABI5 and WRKY2 transcription factors (Lopez-Molina et al., 2001, 2002; Jiang and Yu, 2009).
Under normal growth conditions, light activates the differentiation of chloroplasts. Cotyledons switching from degradation of storage proteins to photosynthetic activity assimilate CO2 and produce sugar. Mutations impairing chloroplast differentiation, light activation and photosynthesis lead to non-photosynthetic cotyledons. Such Arabidopsis mutants also remain arrested at the seedling establishment stage in the absence of externally added sugar (Leon et al., 1998; Chen and Thelen, 2010; Pogson and Albrecht, 2011). In this study, we show that bou-2 seedlings cannot make the transition to being phototrophic as a consequence of malfunctioning photorespiratory metabolism leading to meristem arrest, global gene expression modification and metabolite accumulation. Growth may resume in elevated CO2 concentrations or with added sugar. We compared bou-2 to shm1, another photorespiratory mutant. We found that shm1 stopped growing at the seedling stage, similarly to bou-2 (Figures 5 and 6, and Figures S5–S8), and that SAM activity may be restored by added sugar. The bou-2 and shm1 seedlings display a lack of SAM activity, preventing leaf formation, vegetative development and growth (Fletcher, 2002; Weigel and Jürgens, 2002). It is possible that photosynthesis in seedlings of such strong photorespiratory mutants do not produce sufficient sugar, which is both a metabolic substrate and a signaling molecule that induces cell division at the shoot apical meristem (SAM) (Riou-Khamlichi et al., 2000).
Strongly elevated levels of glycine and other metabolites were detected in bou-2 seedlings and shm1 seedlings (Table 1 and Figure S8). The arrested development of the mutant seedlings may be caused partly by accumulation of metabolites, as recently demonstrated for methylglyoxal, dihydroxyacetonephosphate or glycerol-3-phosphate (Chen and Thelen, 2010).
It would be interesting to determine whether any photorespiratory mutant displays the same arrested growth at the seedling stage. Mutants with an arrested growth phenotype caused by lack of carbon assimilation or reserve mobilization as described above may result in the same disorganized SAM phenotype as described here. Arabidopsis is therefore dependent at a very early developmental stage upon an efficient photorespiration metabolism that allows the plant to switch to autotrophic CO2 fixation metabolism.
BOU activity is critical for GDC stability and function
We observed similar results for the shm1 and bou-2 mutants in terms of growth phenotypes (Figure S7) (Voll et al., 2006), on photosynthetic (Figure 2b) (Timm et al., 2012) and metabolite levels (Figure 3 and Table S2), indicating that bou-2 phenocopies shm1. Thus we hypothesize that both proteins perform closely connected functions in the photorespiratory metabolism. The actual substrate transported by BOU has not yet been identified. However, glycine accumulation in bou-2 prompts two alternate hypotheses. First, BOU functions as a mitochondrial glycine shuttle and the glycine accumulation is caused by impaired glycine uptake into the mitochondria. Second, BOU functions as a mitochondrial carrier for a GDC co-factor and the glycine accumulation is caused by impaired glycine oxidation in the mitochondria. We favor the second hypothesis for the following reasons. Function of BOU as a glycine transporter seems unlikely as it has been demonstrated that glycine diffuses through the membranes of isolated spinach leaf mitochondria if the concentration exceeds 0.5 mm (Yu et al., 1983). Typically, the glycine concentration is around 1 mm in photosynthetically active cells (Yu et al., 1983) and is even higher in bou-2 cells, allowing (facilitated) diffusion processes. Furthermore, we observed that the glycine accumulation was accompanied by increased GABA concentrations in bou-2 plants. As GABA can only be metabolized by the mitochondrial GABA transaminase, and glycine acts as inhibitor for this enzyme (Shelp et al., 2012), this may be taken as indirect evidence for glycine accumulation inside mitochondria. This implies that mitochondrial glycine breakdown by GDC rather than glycine uptake is impaired in bou-2. On the basis of the shm1 phenocopying and especially the strong glycine accumulation, we thus hypothesize that the metabolite transported by BOU is required for proper GDC activity. It has been demonstrated that impaired biosynthesis of the co-factor lipoic acid (Ewald et al., 2007) or tetrahydrofolate (THF) re-cycling in mitochondria causes a decrease in GDC activity (Collakova et al., 2008). Import of malonic acid and pyruvate is a prerequisite for lipoic acid synthesis, and the three precursors glutamate, pABA (para-aminobenzoic acid) and pterin are required for THF synthesis inside the mitochondria (Wada et al., 1997; Rebeille et al., 2007). The strongly reduced GDC activity (Figure 4b), most likely as a result of enhanced degradation of the GDP subunit (Figure 4a) in bou-2, supports this hypothesis.
In summary, although the specific substrate for the BOU transporter protein remains elusive, its essential function in photorespiratory metabolism in both seedlings and adult plants is firmly established. Determination of the substrate of BOU is the subject of ongoing research.
Arabidopsis lines and culture conditions
The Arabidopsis thaliana Heynh. Col-0 line was used as wild-type, and the bou-2 mutant is the GABI-Kat Line number 079D12 of Col-0 (http://www.gabi-kat.de/db/lineid.php, Li et al., 2003; Rosso et al., 2003) (available from the Nottingham Arabidopsis Stock Centre as N370142 and harboring a T-DNA insertion in the second exon of the BOU gene sequence). Seeds were surface-sterilized (20% commercial sodium hypochloride, 80% ethanol) for 10 min, rinsed once in ethanol and dried. Seeds were grown on 0.5× Murashige and Skoog (MS) medium with 0.8% agar at pH 5.8; 1% sucrose was added where indicated. After an overnight imbibition at 4°C to synchronize germination, seedlings were grown at 22°C under continuous light (150μmol m−2 sec−1) or 12 h light/12 h dark cycles. Seedlings were grown in 1% CO2 or a 3000 ppm CO2-enriched atmosphere where indicated. For further physiological characterization, seedlings were transferred to soil and cultivated under the conditions described above.
Crossing of bou and GUS transgenic lines
Heterozygous bou/+ plants were crossed with pCYCB1;1:Dbox-GUS 9 (Colón-Carmona et al., 1999) or STM::uidA plants (Kirch et al., 2003). T2 progeny were tested for GUS histochemical staining and bou phenotype on medium lacking sucrose. Histochemical GUS staining was performed as described previously (Toka et al., 2010).
Samples were fixed for 4 h in 4% paraformaldehyde and 1% glutaraldehyde in PBS at 4°C. After fixation, samples were washed several times in PBS, post-fixed in 1% osmium tetraoxide for 2 h at room temperature, then dehydrated through an ethanol series, and finally embedded in an epoxy resin. After cutting, semi-thin sections (3 mm) were placed on glass slides and stained with Paragon (0.37 g toluidine blue and 0.27 g basic fuchsin in 30% ethanol).
Gas exchange analysis
CO2 compensation points and net photosynthetic rates were measured using an LI-6400XT photosynthesis analyzer (Li-Cor, http://www.licor.com/env/). A/Ci curves (CO2 assimilation in relation to internal CO2 partial pressure) for determination of the CO2 compensation points were recorded at a light intensity of 1500 μmol photons m−2 sec−1, various CO2 concentrations (50–2000 μl l−1), 21% O2, and a leaf temperature of 20°C. Net photosynthetic rates were monitored at 400 μl l−1 CO2 unless otherwise indicated.
WT and mutant plants were grown for 17 days at 3000 ppm CO2 under 12 h light/12 h dark cycles on 0.5× MS medium with 0.8% agar at pH 5.8. Entire rosettes were harvested 1 h before the end of the light period. Samples were taken from plants grown under 3000 ppm CO2 and 17 h after a shift to ambient CO2 (380 ppm CO2). The plant material was snap-frozen in liquid nitrogen. Experiments were set up in a completely randomized manner. Extraction of plant tissues and metabolite analysis was performed as described by Rosar et al. (2012).
Amino acid analysis
Frozen plant tissues were ground in liquid nitrogen. Total free amino acids were extracted in a solution of 2% 5-sulfosalicylic acid as described previously (Ferrario-Méry et al., 1997; Toka et al., 2010).
Protein isolation and immunological studies
Samples for protein isolation were taken at approximately the middle of the light period (5 h after onset of light) from plants grown to developmental stage 5.1 (Boyes et al., 2001) under high-carbon conditions (1% CO2) and 1, 2 and 3 days after transfer to normal air (0.038% CO2). Approximately 100 mg of leaf tissue were homogenized using an MM 400 ball mill (Retsch, http://www.retsch.com) and dissolved in 200 μl extraction buffer containing 50 mm HEPES/KOH pH 7.6, 10 mm NaCl, 5 mm MgCl2, 100 mm sorbitol, 0.1 mm phenylmethylsulfonyl fluoride. After centrifugation (4°C, 10 min, 20 000 g), the protein content of the supernatant was determined as described previously (Bradford, 1976), using BSA as the protein standard. For immunological analysis, 10 μg of total leaf protein was separated by SDS-PAGE and blotted onto a nylon membrane according to standard protocols. Antibodies against photorespiratory proteins were produced and purified from Pisum sativum (GDP, mLPD and SHM1, purified proteins), Solanum tuberosum (GDT, purified protein) and Flaveria trinervia (GDH, recombinant protein). The antibody against lipoic acid was purchased from Calbiochem (http://www.millipore.com/Antibodies).
Analysis of GDC activity
Mitochondria from WT and bou-2 mutant plants were isolated as described by Keech et al. (2005). The mitochondrial pellet was resuspended in 500 μl of 50 mm TES/KOH pH 7.5, 2 mm EDTA, 5 mm MgCl2, 30% v/v glycerol and 0.1% v/v Triton X-100 to permeabilize the membrane and release the matrix proteins. The 100 μl activity assay contained 50 μl mitochondrial preparation and 50 mm TES/KOH (pH 7.5), 1 mm EDTA, 2.5 mm MgCl2, 15% v/v glycerol, 0.01% v/v Triton X-100, 2 mm β-merpatoethanol, 0.03 mm pyridoxalphosphate, 0.5 mm tetrahydrofolate, 1 mm NAD+ and 0.2 mm [14C(U)]-glycine (final concentrations). The reactions were performed in an airtight vial. After 15 min, the reaction was terminated by addition of 1 m sulfuric acid (0.5 ml). GDC activity was linear for at least 30 min. The released [14C]-CO2 was transferred with a stream of nitrogen gas into a new vial filled with 5 ml of a 1 m KOH solution to trap the carbon dioxide as hydrogen bicarbonate (Walker and Oliver, 1986). The released [14C]-CO2 was quantified by liquid scintillation counting with a Beckman LS6000 counter (Beckman Coulter Inc., http://www.beckmancoulter.com/).
Microarray analysis was performed at the Unité de Recherche en Génomique Végétale (Evry, France), using CATMA arrays containing 24 576 gene-specific tags corresponding to 22 089 genes from Arabidopsis (Crowe et al., 2003; Hilson et al., 2004). Two independent biological replicates were used. For each biological repetition and each point, RNA samples were obtained by pooling RNAs from 150 plantlets. Plants were collected at 1.0 developmental growth stage (Lurin et al., 2004) and cultivated without sucrose under continuous light. In order to determine the direct effects of the bou-2 mutation on plant growth at the seedling stage, RNA samples were extracted from 5-day-old plantlets with the mutant phenotype or Col-0 WT plantlets. Eight-day-old bou-2 plantlets were also collected and compared to 5-day-old Col-0 WT plantlets. Total RNA was extracted using an RNeasy purification kit (Qiagen, http://www.qiagen.com) according to the manufacturer's instructions. For each comparison, one technical replication with fluorochrome reversal was performed for each biological replicate (i.e. four hybridizations per comparison). Labeling of cRNAs with Cy3-dUTP or Cy5-dUTP (Perkin–Elmer/NEN Life Science Products, http://www.perkinelmer.com), hybridization to the slides, and scanning were performed as described by Lurin et al. (2004).
Further experimental procedures are described in Methods S1.
This work was supported by a grant from the Deutsche Forschungsgemeinschaft (PROMICS Research Group, WE 2231/8-2) to A.P.M.W. and by the Ministère de l'Enseignement Supérieur et de la Recherche/Centre National de la Recherche Scientifique. We thank Samantha Kurz for technical assistance, Katrin Weber and Elisabeth Klemp for metabolite analysis, and Stéphanie Boutet and Sylvie Citerne (Laboratoire Commun Chimie du Végétal of the Institut Jean-Pierre Bourgin, Institut National de la Recherche Agronomique, Versailles, France) for amino acid analysis. We kindly acknowledge Andrea Bräutigam for helpful discussion. The authors are indebted to Rachel J. Carol (Emendo bioscience editing service, http://emendo.co.uk) for her careful and efficient help in the writing of this manuscript.