Introduction

Nearly all organisms possess functional glucose metabolic pathways, such as the glycolytic and oxidative pentose phosphate pathways (oxPPP), which are localized in the cytosol in non-photosynthetic eukaryotes. Glycolysis converts glucose into pyruvate, producing two moles of ATP and two moles of NADH per one mole of glucose, whereas the oxPPP produces NADPH and pentoses, rather than ATP. The produced pyruvate is converted into citrate, which is further oxidized in the mitochondria by the tricarboxylic acid (TCA) cycle through a series of eight enzyme-catalyzed reactions. In photosynthetic organisms, an additional carbohydrate metabolic pathway known as the Calvin–Benson cycle, which is also referred to as the reductive pentose phosphate pathway, produces the triosephosphate glyceraldehyde-3-phosphate (GAP).

In flowering plants, glycolysis and oxPPP exist in two compartments: the cytosol and plastids (Dennis and Miernyk 1982). A total of 105 enzymes related to glycolysis and oxPPP have been identified in Arabidopsis thaliana (Online Resource Table S 1), and their subcellular localization has been examined by various approaches, including green fluorescent protein (GFP) fusion protein analysis, enzymatic activity measurements in isolated organelles, and proteome analysis. In photosynthetic tissues of Arabidopsis, nearly all glycolytic enzymes, from phosphoglucose isomerase (PGI) to pyruvate kinase (PK) in the reaction process (process ID [ID]: 1–10; Fig. 1a), are present in the cytosol, although several enzymes, from PGI to phosphoglycerate kinase (PGK) (ID: 1–7), and PK to pyruvate dehydrogenase (PDH) (ID: 10 and 11) are also found in plastids (Fig. 1b). Recently, the plastid localization of phosphoglycerate mutase (PGAM, ID: 8) and enolase (ENO, ID: 9) has also been reported in non-photosynthetic tissues (Andriotis et al. 2010). All cytosolic enzymes related to glycolysis in Arabidopsis are associated with the mitochondrial outer membrane (Giegé et al. 2003), a configuration that is considered to allow the direct supply of pyruvate to the mitochondria as a substrate of the TCA cycle. In A. thaliana, plastids contain a complete set of oxPPP enzymes (ID: 12–18), whereas the cytosol only contains five oxPPP-related enzymes (ID: 12–16), and lacks the transketolase (TKT, ID: 17) and transaldolase (TAL, ID: 18) (Fig. 1b). Despite extensive studies, localization analysis of these two pathways in A. thaliana remains incomplete, as the Arabidopsis genome encodes a large number of redundant genes.

Fig. 1
figure 1

Subcellular distribution of glycolytic and oxidative pentose phosphate pathways (oxPPP) in plants and algae. a General glycolytic pathway (red line) and oxPPP (blue line). Numbers between intermediates indicate the reaction process ID, and the corresponding enzymes are shown in the table to the right. Abbreviations of intermediates are shown in Fig. 3. Unabridged enzyme names are shown in Table 1 and Online Resource Table S 1. b Glycolysis and oxPPP reported in the plant Arabidopsis thaliana, green alga Chlamydomonas reinhardtii, and diatom Phaeodactylum tricornutum. Metabolic map of A. thaliana was made from the data listed in Online Resource Table S 1. PGAM and ENO have not been reported in greening chloroplasts in A. thaliana, and the dotted line indicates a reaction step that occurs only in non-photosynthetic tissue. The pathway map of C. reinhardtii and P. tricornutum were drawn from data in the studies by Johnson and Alric (2013) and Gruber et al. (2009), respectively. c Glycolysis and oxPPP estimated from the genome sequence of C. merolae (left) and the pathways verified by experimental results in the present study (right). The transaldolase gene was not found in the C. merolae genome, and no transaldolase activity was detected in cell extracts (Online Resource Fig. S 2)

Full size image

In the green alga Chlamydomonas reinhardtii, subcellular compartmentalization of enzymes related to carbohydrate metabolic pathways has been mainly examined by enzymatic activity and proteome analyses of isolated organelles (reviewed in Johnson and Alric 2013). These studies have demonstrated that the cytosol contains enzymes involved in the latter half of glycolysis (ID: 6–10) and the first half of oxPPP (ID: 12–14), whereas the plastid includes enzymes for the first half of glycolysis (ID: 1–7) and a complete complement of oxPPP enzymes (ID: 12–18) (Fig. 1b). Although localization data of metabolic enzymes are relatively limited in other algae, it was reported that glycolytic enzymes in diatoms are localized to the mitochondria (ID: 6–10) in addition to the cytosol (ID: 1–10) and plastids (ID: 1–11) (Liaud et al. 2000). Thus, it appears that differences in the compartmentation of enzymes lead to diversification of glycolysis and oxPPP pathways in photosynthetic eukaryotes. However, comprehensive localization analysis in individual organisms is necessary to understand cellular carbohydrate metabolism.

Cyanidioschyzon merolae is a unicellular rhodophyte that inhabits hot springs with warm water (up to 50 °C) that is acidified by sulfuric acid (pH 1.5–2.5). C. merolae has a simple cell structure consisting of a single mitochondrion and plastid per cell, and is therefore a suitable model for subcellular localization analyses. In addition, the nuclear (Matsuzaki et al. 2004; Nozaki et al. 2007), plastid (Ohta et al. 2003) and mitochondrial genomes (Ohta et al. 1998) of this alga are completely sequenced. The C. merolae genome is small (16.5 Mbp) and contains a small number of predicted protein-coding genes (4,775 genes) with low redundancy. C. merolae is also readily transformed with polyethylene glycol (PEG) (Ohnuma et al. 2008) and has been successfully used for subcellular localization experiments with transformants expressing 3× hemagglutinin (HA)-tagged and GFP-fused proteins (Ohnuma et al. 2008; Moriyama et al. 2014; Watanabe et al. 2011). Genomic analyses for genes related to carbon metabolism suggest that C. merolae encodes metabolic pathways for floridoside, trehalose, storage glucans, and matrix polysaccharides (Barbier et al. 2005). C. merolae is also predicted to possess a minimal set of metabolic transporters, such as those for triose phosphate, but no plastidic dicarboxylate translocators, which are required for nitrogen assimilation and photorespiration pathway and are conserved in green plants and algae (Barbier et al. 2005; Tyra et al. 2007). Our laboratory has identified a unique pathway for lipid biosynthesis in C. merolae based on the results of genomic and experimental analyses (Sato and Moriyama 2007). However, identification and subcellular localization of enzymes related to central carbohydrate metabolic pathways remain unclear.

In the present study, we searched for the enzymes in C. merolae related to glycolysis, oxPPP, and the TCA and Calvin–Benson cycles, and analyzed their subcellular localization by the observation of GFP- and 3× hemagglutinin (HA)-fusion proteins. This comprehensive approach revealed several similarities and differences in the type and localization of enzymes involved in carbohydrate metabolic pathways of photosynthetic eukaryotes.

Materials and methods

Culture conditions and transformation of C. merolae

Cells of C. merolae strain 10D (Toda et al. 1998) were inoculated into 100 mL 2× Allen’s medium (pH 2.5; Minoda et al. 2004) in 200-mL flasks. Flasks were shaken under continuous light provided by three fluorescent tubes (30 µmol m−2 s−1) at 40 °C. Cultured C. merolae cells were transformed by the PEG-method, as described by Moriyama et al. (2014), and transformed cells were cultured overnight at 40 °C with aeration by 1 % CO2.

Overexpression and purification of Anabaena Xfp protein

The full-length sequence of the xfp gene (all2567) was amplified by PCR from the Anabaena sp. PCC 7120 genome (laboratory stock) using the primer pair InF_pThioA_AnaXfp-F (5′-CCTCTAGAGTCGACCTGCAGTACTTTAGCAAGTCCTCCACA-3′) and InF_pThioA_AnaXfp-R (5′-ACCCTGTACGATTACTGCAGCTAATACGGCCATTGCCAGT-3′). The underlined sequences correspond to the ends of PstI-linearized pThioHisA vector (Invitrogen, Carlsbad, CA, USA) and are required for the In-Fusion cloning reaction (Clontech Laboratories, Mountain View, CA, USA). The PCR product was cloned into PstI-digested pThioHisA using the In-Fusion reaction, as directed by the manufacturer (Clontech Laboratories). Overexpression in Escherichia coli TOP10 and purification of recombinant protein were performed as described previously (Moriyama et al. 2008).

Preparation of protein extract

For measurement of transaldolase activity, soluble protein was extracted from C. merolae cells. C. merolae cells were grown to mid-log phase (OD750 = 0.8) and collected by centrifugation at 1,500g for 10 min. The cell pellet was resuspended in protein extraction buffer (100 mM Tris–HCl pH 8.0, 50 mM KCl, 5 mM MgCl2, 1 mM EDTA, and 3 mM dithiothreitol) and was then disrupted by sonication. The supernatant was obtained by centrifugation of the lysed cell suspension at 18,000g for 15 min.

Transaldolase assay

Because erythrose 4-phosphate (E4P) was not commercially available, it was produced enzymatically from fructose 6-phosphate (F6P) by the phosphoketolase reaction with purified Xfp (Meile et al. 2001). The phosphoketolase reaction was performed in 4 mL Xfp reaction mixture (33.3 mM KH2PO4 [pH 6.5], 25 mM F6P, 1 mM thiamine pyrophosphate, and 150 μg Xfp recombinant protein) at 30 °C for 3 h. In this reaction, equimolar amounts of acetyl-phosphate and E4P were produced. The concentration of acetyl-phosphate was measured spectrophotometrically as ferric acetyl hydroxamate by the procedure of Racker (1962).

Transaldolase activity was measured photometrically by following the A340 change of NADH according to a modified procedure of Soderberg and Alver (2004). Briefly, 2 mL transaldolase reaction mixture [50 mM HEPES–KOH (pH 7.6), 5 mM MgCl2, 1 mM EDTA, 46 units of triosephosphate isomerase (TPI; Sigma-Aldrich, St. Louis, MO, USA), 2 units glycerol 3-phosphate dehydrogenase (Sigma-Aldrich), 200 μM NADH, C. merolae protein extract, and 100 μL Xfp-reaction mixture] containing a final concentration of 150 μM E4P and 1 mM F6P was reacted at 30 °C and monitored at A340 for 1 min. As a positive control, yeast transaldolase (Sigma-Aldrich) was added to the mixture instead of C. merolae protein extract. In this assay, transaldolase activity produces glyceraldehyde 3-phosphate (GAP) and sedoheptulose 7-phosphate (S7P) from F6P and E4P, and the produced GAP is then converted to dihydroxyacetone phosphate (DHAP) by TPI. DHAP is then reduced to glycerol phosphate by glycerol 3-phosphate dehydrogenase with concomitant oxidation of NADH to NAD+. The decrease in A340 was monitored during the reaction as a measure of transaldolase activity.

Construction of GFP- and 3× hemagglutinin (3× HA)-fusion genes

Carbohydrate metabolism-related genes and their promoter sequences were amplified by PCR with specific primer sets (Online Resource Table S 2). For the construction of GFP-fusion genes, the cauliflower mosaic virus 35S promoter was first excised from the sGFP (S65T) vector (Chiu et al. 1996) by digestion with XbaI. PCR-amplified DNA fragments were then inserted into the promotorless sGFP (S65T) vector using the appropriate restriction enzymes or the In-Fusion HD cloning kit (Clontech Laboratories).

For overexpression of GFP fusion genes, the amplified DNA fragments were inserted into pCG1 vector, which contains the apcC promoter of C. merolae, sGFP, and NOS terminator (Watanabe et al. 2011). For the construction of 3× HA-fusion genes, the target DNA fragments were inserted into pBSHAb-T3′ vector using the appropriate restriction enzymes or the In-Fusion cloning kit (Ohnuma et al. 2008).

For the construction of N-terminal fusion of 3× HA-tag with citrate synthases expressed by the apcC promoter, first, the apcC promoter, which was amplified by PCR with C. merolae genome as a template with primers (5′-ATGCCCGCGGTTACAATACCGATAGATGAGTTTCG -3′ and 5′-GCATCTGCAGGGATCCTCTAGAGGTCAACGAACGA-3′; underlined sequences are restriction sites of PacII and PstI, respectively), was inserted into pBSHAb-T3′ vector using PacII and PstI. The DNA fragments of citrate synthases were inserted into pBSHAb-T3′ vector containing apcC promoter by the In-Fusion cloning kit.

Microscopic examination of C. merolae cells

For observation of GFP fluorescence, transformed C. merolae cells were centrifuged at 200g for 10 min, and then examined using a BX-60 fluorescence microscope (Olympus, Tokyo, Japan) equipped with a narrowband filter cube (U-MNIBA; Olympus). Immunofluorescence detection was performed essentially according to Nishida et al. (2004). In the primary antibody reaction, anti-GFP or anti-HA-tag monoclonal antibodies diluted to 1/200 with immunoreaction enhancer solution (Can Get Signal Immunostain Solution B; Toyobo, Osaka, Japan) were reacted with cells for 1 h. As the secondary antibody, anti-mouse IgG monoclonal antibody labeled with Alexa fluor 488 (Invitrogen) was diluted to 1/200 with immunoreaction enhancer solution and reacted for 30 min. For immunofluorescence detection using the Tyramide Signal Amplification (TSA) kit (Invitrogen), the primary antibody (anti-GFP antibody) and secondary antibody (anti-mouse monoclonal antibody conjugated with horseradish peroxidase) were diluted to 1/200 and 1/100, respectively. The immunolabelled cells were observed under a fluorescence microscope equipped with a cube U-MNIBA for Alexa fluor 488, and a cube U-MWIG (Olympus) for the detection of chlorophyll autofluorescence. Nomarski differential interference images were also recorded. All microscopic images were captured using a digital camera (model DP-70; Olympus) and processed by Photoshop (Adobe, San Jose, CA, USA).

Results

Enzymes related to carbohydrate metabolism in the C. merolae genome

Enzymes related to glycolysis, oxPPP, and the TCA and Calvin–Benson cycles in C. merolae were identified in the KEGG pathway (http://www.genome.jp/kegg/pathway.html) and Gclust databases (http://gclust.c.u-tokyo.ac.jp/; Sato 2009) are listed in Table 1. The analysis revealed that the C. merolae genome encodes complete sets of enzymes for glycolysis, and the TCA and Calvin–Benson cycles. However, transaldolase (TAL, ID: 18), which is an oxPPP enzyme, was not found in the genome. Among the identified enzymes involved in carbohydrate metabolism, four enzymes, PDH-E1 alpha and beta, RBCL, and RBCS, were encoded by the plastid genome, and three subunits of succinate dehydrogenase, SDH2, SDH3, and SDH4, were encoded by the mitochondrial genome. These enzymes were therefore speculated to be localized to the respective organelles.

Table 1 List of enzymes related to glycolysis, oxidative pentose phosphate pathway, Calvin–Benson cycle, and TCA cycle in C. merolae and summary of localization analysis
Full size table

In the genome database of C. merolae (http://merolae.biol.s.u-tokyo.ac.jp/), five gene products, CML036C, CML037C, CML059, CMM231C, and CMS195C, are annotated as 6-phosphogluconate dehydrogenases (6PGDH). CML059C (6PGDH-2), CMM231C (6PGDH-3), and CMS195C (6PGDH-4) contain a full-length 6PGDH domain, whereas CML036C and CML037C are homologous to the C- and N-terminal regions, respectively, of 6PGDH. The open reading frames (ORFs) of CML036C and CML037C were found to overlap, with the initiation codon of CML036C predicted to be located in the latter half of CML037C. To determine whether CML036C and CML037C were different proteins, the region upstream of CML036C was sequenced, revealing that the 477th cytosine base of CML037C had been erroneously included in the genome sequence due to a sequencing error (Online Resource Fig. S 1a). Removal of this base resulted in a shift of the reading frame of CML037C to be in-frame with CML036C. In addition, alignment analysis of C. merolae 6PGDHs suggested that the start codon of CML037C/CML036C was 219 bp upstream of the first methionine of CML037C (Online Resource Fig. S 1). The amino acid sequence of CML037C/CML036C (6PGDH-1) was nearly identical with that of CML059C (99 %) and CMM231C (99 %) (Online Resource Fig. S 1b).

Subcellular localization of the identified metabolic enzymes was predicted using the computer programs TargetP (http://www.cbs.dtu.dk/services/TargetP/; Emanuelsson et al. 2007) and Wolf PSORT (http://www.genscript.com/psort/wolf_psort.html; Horton et al. 2007) (Table 1). In addition to estimation of localization to plastids and mitochondria, Wolf PSORT program can find peroxisomal targeting sequence (PTS), estimating the existence of PTS1 in two citrate synthases [CS-1 (CMA040C) and CS-2 (CMJ293C)]. We also performed alignments of the amino acid sequences to determine whether the enzymes possessed an N-terminal extension for organellar targeting (data not shown). Based on the results of these analyses, the subcellular localization of the central carbohydrate pathways in C. merolae were predicted (Fig. 1c left).

Measurement of transaldolase activity in C. merolae cells

A gene encoding a known transaldolase was not found in the C. merolae genome. Because it is possible that C. merolae has a transaldolase with low homology to previously characterized transaldolases, protein extracts from C. merolae cells were assayed for transaldolase activity. Transaldolase catalyzes the reversible conversion of GAP and sedoheptulose 7-phosphate (S7P) to erythrose 4-phosphate (E4P) and fructose 6-phosphate (F6P). Here, transaldolase activity was estimated by measuring the production of GAP from E4P and F6P. Although a positive control (yeast transaldolase) showed high activity in this assay, no activity was detected in the protein extract of C. merolae cells (Online Resource Fig. S 2). However, when yeast transaldolase was added to the reaction mixture containing the cell extract, transaldolase activity was detected, suggesting that the C. merolae cell extract did not inhibit transaldolase activity.

Localization analysis of enzymes related to carbohydrate metabolism in C. merolae

Cyanidioschyzon merolae is a unicellular alga with a single plastid and mitochondrion, and reproduces by binary fission (Fig. 2a). To verify the metabolic pathways predicted from the genome sequence (Fig. 1c left), the subcellular localization of 65 nuclear genome-encoded proteins expressed in C. merolae cells as GFP or HA-tag fusion proteins was examined. First, C. merolae cells transformed to express full-length enzyme-GFP fusion proteins driven by their own promoter were examined for GFP fluorescence. If the GFP fluorescence was too weak to be reliably observed, immunofluorescence was detected using an anti-GFP antibody. If necessary, the immunofluorescence signal was amplified by the tyramide signal amplification (TSA) method. If the localization of GFP-fusion could not be determined, 3× HA-tagged fusion proteins were constructed and detected by immunofluorescence using anti-HA-tag antibody. In cases when the signal still could not be observed, the putative transit peptide-GFP fusion proteins were overexpressed using the C. merolae apcC gene promoter (Watanabe et al. 2011).

Fig. 2
figure 2

Localization of GFP-fusion proteins in C. merolae cells. a Fluorescence images with 4′,6-diamidino-2-phenylindole (DAPI) staining of an interphase cell (left) and a dividing cell (right). Nuc nucleus, Mt mitochondrion, Pt plastid. This figure was adopted from Moriyama et al. (2014) with permission of the publisher. b Fluorescence microscopic images of transiently transformed C. merolae cells. Representative images of enzymes localized to the cytosol (PGM-2), plastid (FBA-2), both the cytosol and plastid (PGK), mitochondrion (MDH-1), and the peroxisome (catalase, Imoto et al. 2013) are shown. FBA-2, PGK, MDH-1, and catalase were observed with GFP fluorescence. PGM-2 was observed by immunofluorescence with anti-GFP antibody. Other images of GFP- and 3× HA-fusion protein analyses are shown in Online Resource Figure S 3. DIC, Normarski differential interference contrast images; Chlorophyll, chlorophyll autofluorescence; Merge, merged images of green fluorescence and chlorophyll autofluorescence; Cyto cytosol

Full size image

Figure 2b shows examples of subcellular localization of the examined proteins to the cytosol (PGM-2), plastid (FBA-2), and mitochondrion (MDH-1), the dual localization of PGK enzyme to the cytosol and plastid, and the peroxisome (catalase, Imoto et al. 2013). Images of localization analysis for other carbohydrate metabolism-related proteins are shown in Online Resource Figure S 3. Among the 65 enzymes examined, 26, 17, 15, and 3 were localized to the cytosol, plastid, mitochondrion, and peroxisome, respectively. Four enzymes, PGK, 6-phosphogluconolactonase (PGL), ribose-5-phosphate isomerase (RPI), and triosephosphate isomerase (TPI), were found to be localized to both the cytosol and plastid. The localization results for each enzyme are shown as merged images next to the respective reaction in the carbohydrate metabolic pathways in C. merolae (Fig. 3).

Fig. 3
figure 3

Central carbohydrate metabolic pathways in C. merolae based on subcellular localization analysis. Merged images of green fluorescence by the observation of GFP- or HA tagged-fusion proteins, chlorophyll autofluorescence (red fluorescence), and Normarski differential interference contrast images (gray scale) are overlaid on the arrow for the corresponding enzymatic reaction on the metabolic map. Black arrows show the glycolysis pathway in the cytosol and plastid. Blue arrows with a dashed line show oxPPP and green lines and arrows indicate the Calvin–Benson cycle. The TCA cycle is shown in mitochondrion with black arrows. Color-coding of enzyme names indicates the construction of fusion genes and detection methods used for determination of subcellular localization. The description of the color codes is shown in the lower right-hand table. Single and double asterisks indicate enzymes encoded by the plastid and mitochondrial genomes, respectively. Triple asterisks indicate enzymes with N-terminal sequences that have high homology to that of 6PGDH-1 (no localization analysis was performed for these two enzymes). Abbreviations of intermediates: 1,3BPG 1,3-bisphoglycerate, 2OG 2-oxoglutarate, 2PGA 2-phosphoglycerate, 3PGA 3-phosphoglycerate, 6PG 6-phosphogluconate, AcCoA acetyl-CoA, CIT citrate; DHAP dihydroxyacetone phosphate, E4P erythrose 4-phosphate, F1,6BP fructose 1,6-bisphosphate, F6P fructose 6-phosphate, FUM fumarate, G1P Glucose 1-phosphate, G6P glucose 6-phosphate, GAP glyceraldehyde 3-phosphate, GLC glucose, GLP glucono 1,5-lactone 6-phosphate, ICIT isocitrate, MAL malate, OAA oxaloacetate, PEP phosphoenolpyruvate, PYR pyruvate, R5P ribose 5-phosphate, Ru5P, ribulose 5-phosphate, RuBP ribulose 1,5-bisphosphate, S1,7BP, sedoheptulose 1,7-bisphosphate, S7P sedoheptulose 7-phosphate, SUC succinate, SucCoA succinyl-CoA, Xu5P xylulose 5-phosphate. Unabridged enzyme names are shown in Table 1

Full size image

Glycolysis in C. merolae

Subcellular localization analysis revealed that a complete set of glycolytic enzymes, from glucokinase to pyruvate kinase (ID: 1–10), exists in the cytosol of C. merolae (Fig. 3). Pyrophosphate (PPi)-dependent phosphofructokinase (PFP) and NADP+-dependent-non-phosphorylating GAP dehydrogenase (GAPN) are cytosolic enzymes in plants. The genome of C. merolae encodes one PFP and one GAPN gene, whose protein products were localized to the cytosol, as expected. A. thaliana has two NAD+-dependent GAPDHs (GAPC1 and GAPC2), which are localized to both the nucleus and cytosol (Holtgrefe et al. 2008). Nuclear-localized GAPDHs are speculated to function in stress signal transduction due to their DNA-binding ability. C. merolae was predicted to encode two GAP dehydrogenases (GAPC-1 and -2). GAPC-1 showed typical localization to the cytosol, whereas the GAPC-2-GFP fusion protein was observed as granules in the cytosol. For both GAPCs, nuclear localization was not observed; however, under stressful culture conditions, such as in medium containing H2O2 or nitric oxide (NO), GAPC-1 and -2 may be localized to the nucleus, as occurs in animals (Hara et al. 2006). Plants have alternative pathways for phosphoenolpyruvate (PEP) metabolism in the cytosol. PEP carboxylase (PEPC) catalyzes the irreversible conversion of PEP and HCO3 − to oxaloacetate (OAA) and orthophosphate (Pi). Conversely, OAA is converted to PEP by ATP-dependent PEP carboxykinase (PEPCK) and can be also converted to malate by malate dehydrogenase (MDH), the product of which is a substrate of malic enzyme (ME) to yield pyruvate. The enzymes related to the PEP pathway are conserved in C. merolae and are localized to the cytosol (Fig. 3).

Seven glycolytic enzymes, from PGI to PGK, were localized to the plastid (ID: 1–7) (Fig. 3). Furthermore, two pyruvate kinases (PKs) and pyruvate dehydrogenase E2 and E3 (PDH-E2-3 and LPD[E3]-2) were also targeted to the plastid (ID: 10 and 11), although phosphoglycerate mutase (PGAM, ID: 8) and enolase (ENO, ID: 9) were observed in the cytosol, not in the plastid. These findings are supported by the fact that two E1 subunits of PDH, alpha and beta, are encoded by the plastid genome in C. merolae. The Arabidopsis genome encodes three NADP+-dependent GAPDHs (GAPA1, GAPA2, and GAPB), which function as Calvin–Benson cycle enzymes and are localized to plastids (Muñoz-Bertomeu et al. 2009). In addition to these GAPDHs, Arabidopsis has two NAD+-dependent GAPDHs (GAPCp1 and 2) that are also localized to plastids (Muñoz-Bertomeu et al. 2009). The two GAPCps are mainly expressed in roots, where the enzymes play an important role in serine synthesis. In C. merolae, GAPC was localized to the cytosol, and therefore GAPA was the only plastid-localized GAPDH.

oxPPP and Calvin–Benson cycle in C. merolae

Oxidative pentose phosphate pathways is an alternative pathway to glycolysis for the oxidation of glucose, but this pathway mainly leads to the generation of NADPH and pentoses, rather than ATP. C. merolae has two glucose-6-phosphate dehydrogenases (G6PDHs, ID: 12). Alignment analyses with G6PDH orthologs and computational modeling for subcellular localization predicted the plastid localization of the two C. merolae G6PDHs. Additionally, C. merolae encodes a single 6-phosphogluconolactonase (PGL, ID: 13) and ribose-5-phosphate isomerase (RPI, ID: 15), both of which are predicted to be localized to the plastid (Table 1). Therefore, we speculated that most components of the oxPPP in C. merolae exist in the plastid (Table 1; Fig. 1c left). However, localization analysis revealed that one G6PDH (G6PDH-1) was localized to the cytosol, but not to the plastid. PGL and RPI were dually localized to the plastid and the cytosol (Fig. 3). This indicates that the components of oxPPP in the cytosol of C. merolae are the same as those found in A. thaliana.

In plant plastids, oxPPP largely overlaps with the Calvin–Benson cycle. In C. merolae, nearly all of the enzymes related to the Calvin–Benson cycle were localized to the plastid, as expected. C. merolae has two sedoheptulose-1,7-bisphosphatases (SBPs), which were previously annotated based on sequence similarity and phylogenetically classified (Matsuzaki et al. 2004). In the present analysis, SBP-1 was localized to the plastid, whereas SBP-2 was observed in the cytosol. Cytosolic SBP was also reported in the diatom Phaeodactylum tricornutum, but its function is unknown (Gruber et al. 2009).

TCA cycle in C. merolae

A complete set of TCA cycle enzymes, including PDH, which catalyzes the conversion of acetyl-CoA from pyruvate, were clearly localized to the mitochondrion of C. merolae (Fig. 3). In the present analysis, all mitochondrially localized proteins were enzymes related to the TCA cycle, and no enzymes related to glycolysis or oxPPP were found to be localized to the mitochondrion. The C. merolae genome encodes four citrate synthase (CS) genes; CS-4 was localized to the mitochondrion, whereas CS-1, CS-2, and CS-3 were observed in the cytosol when the constructs of CS-GFP fusion genes were introduced (Online Resource Fig. S 3 c, d). Since CS-1 and CS-2 are predicted to contain PTS1 at the C-terminus, 3× HA-tag-CS fusion genes (also including CS-3) expressed by the apcC promoter were constructed. CS-1, CS-2, and CS-3 were localized to the peroxisome (Fig. 3). Two types of isocitrate dehydrogenase, an NAD+-dependent type (IDH) and NADP+-dependent type (ICDH), are encoded by the C. merolae genome. In C. merolae, two IDHs were localized to mitochondria, and ICDH was localized to the cytosol, as is found in plants (Mhamdi et al. 2010). In plants, NAD+-dependent MEs are present in mitochondria, where they are speculated to function in the coordination of carbon and nitrogen metabolism during the night (Tronconi et al. 2008). NAD+-ME is widely conserved in plants and algae, although it was reported that this enzyme is not essential for the normal growth of C3 plants, such as Arabidopsis and tomato (Jenner et al. 2001; Tronconi et al. 2008). C. merolae was found to have cytosolic NADP+-ME, but not NAD+-ME.

Discussion

In the present study, we identified the enzymes related to central carbohydrate metabolism, glycolysis, oxPPP, and the Calvin–Benson and TCA cycles in the genome of C. merolae, and deduced their subcellular localization using targeting prediction programs and alignment analysis. These in silico analyses indicated that the C. merolae cytosol contains fragments of the glycolytic pathway and oxPPP (Fig. 1c), and lacked five enzymes. The C. merolae genome contains a single gene encoding TPI, PGK, PGL and RPI, which were predicted to be localized to the plastid. However, subcellular localization analysis revealed that these enzymes are localized to both the plastid and cytosol. Notably, all of these enzymes harbor a second methionine that is positioned near the start methionine of the bacterial and eukaryotic cytosolic enzymes used in the alignment (Online Resource Fig. S 4), suggesting that the C. merolae enzymes might be translated from two different methionine residues, yielding plastid- and cytosol-localized proteins, respectively. The fifth enzyme with unexpected localization is G6PDH. C. merolae encodes two G6PDHs harboring an N-terminal extension sequence; G6PDH-2 was localized to the plastid, whereas G6PDH-1 was unexpectedly localized to the cytosol. Taken together, these results demonstrate the importance of performing localization experiments to complement in silico analysis. In particular, it is difficult to predict the dual localization of a protein to the cytosol and an organelle. In the present study, most GFP- and HA-fusion genes, with the exception of six genes (Online Resource Fig. S 3e, f), were expressed by the native promoter of the target gene rather than an overexpression promoter to examine localization of enzymes translated from native start codon(s) including non-AUG start codon, an approach that appears to allow the detection of dual localization.

oxPPP in the plastid of C. merolae

Transaldolase is an oxPPP enzyme that catalyzes the reversible conversion of GAP and S7P to F6P and E4P. Although this enzyme is conserved in almost all organisms, the sequence identity among transaldolases is low, and the enzyme family is classified into five subfamilies (Samland et al. 2012). A. thaliana has two TALs, which belong to subfamilies 2 and 3, and are predicted to be localized to plastids (Kleffmann et al. 2004; Zybailov et al. 2008). Potato TAL and two tomato TALs have also been isolated and examined for tissue-specific expression and enzymatic activity (Moehs et al. 1996; Caillau and Quick 2005), but mutational analysis of TAL has not been reported. The TAL gene is not encoded by the C. merolae genome and was also not found in the EST sequence of Porphyridium purpureum strain CCMP 1328 (Bhattacharya et al. 2013) or the draft genome sequence of P. purpureum strain NIES 2140 (Tajima et al. unpublished data). However, homologs of the TAL gene was found in the sequenced genomes of other red algae, including Galdieria sulphuraria (Schönknecht et al. 2013; XP_005706833.1), Calliarthron tuberculosum (Chan et al. 2011; IDg12772t1), Chondrus crispus (Collén et al. 2013; XP_005719053.1), and Pyropia yezoensis (Nakamura et al. 2013; contig_22126_g5464). These facts suggest that the TAL gene is not uniformly distributed in red algae. To investigate if an unknown enzyme with transaldolase activity was expressed in C. merolae, the transaldolase activity of protein extracts from C. merolae cells was measured, but no activity was detected (Online Resource Fig. S 2). However, it is possible that C. merolae has cellular transaldolase activity, but that the enzyme(s) may require a cofactor and/or a coenzyme for activity that was not present in the reaction mixture.

Because C. merolae plastids lack TAL, it is speculated that S7P is produced under dark conditions by the activity of transketolase (TKT) and accumulates without further consumption. Under light conditions, S7P is produced in the Calvin–Benson cycle by SBP from sedoheptulose-1,7-bisphosphate (S1,7BP) and is consumed by TKT, indicating that the absence of TAL does not affect metabolic flow. It was previously shown that E. coli has an alternative pathway for S7P consumption that does not utilize TAL (Nakahigashi et al. 2009). In this pathway, S7P is converted by glycolytic ATP-dependent phosphofructokinase (PFK) into S1,7BP, which is further converted into E4P and DHAP by the glycolytic enzyme fructose-1,6-bisphosphate aldolase (FBA) (Nakahigashi et al. 2009). The authors also showed that recombinant PFKs of E. coli and the lactic bacterium Lactococcus lactis, which does not possess a tal gene, utilize S7P as a substrate to yield S1,7BP through the activity of a phosphotransferase. The conversion of S7P into S1,7BP by a PPi-dependent phosphofructokinase (PFK) was also reported in the parasitic amoeba Entamoeba histolytica (Susskind et al. 1982). Based on these reports, it is presumed that S7P is converted to S1,7BP by PFK-2 in C. merolae plastids, which intrinsically lack the TAL gene, and S1,7BP is then converted to E4P and DHAP by FBA-2 under dark conditions, as is observed in transaldolase mutants in E. coli (Nakahigashi et al. 2009, Fig. 4).

Fig. 4
figure 4

A proposed model of the oxidative pentose phosphate pathway in the plastid of C. merolae. Details of this model are explained in the “Discussion” section. Black arrows show the conserved reactions of oxPPP in C. merolae. Red arrows show the putative reaction steps that are catalyzed by glycolytic PFK and FBA in C. merolae, which inherently lacks the TAL gene. Gray arrows with a dashed line indicate reactions catalyzed by TAL, which is conserved in most organisms

Full size image

Carbohydrate metabolic pathways in photosynthetic eukaryotes

The present findings for C. merolae demonstrate that the cytosol contains the entire glycolytic pathway (ID: 1–10) and the first part of oxPPP (ID: 1–16), whereas the plastid contains an incomplete glycolytic pathway (ID: 1–7 and 10–11) and an entire oxPPP (ID: 12–17) that differs from other organisms and is presumably complemented by PFK (ID: 2) and FBA (ID: 4). In addition, C. merolae has sets of enzymes of the Calvin–Benson and TCA cycles in the plastid and mitochondrion, respectively, as expected. The subcellular distribution of these pathways in C. merolae is consistent with that in the photosynthetic tissue of A. thaliana (Fig. 1). Compared with C. merolae and A. thaliana, C. reinhardtii has less duplication of glycolysis- and oxPPP-related enzymes between the plastid and cytosol. However, as the compartmentation analyses of carbohydrate metabolic enzymes in C. reinhardtii were based on proteome analysis and measurement of enzymatic activity using isolated organelles and cytosol, it is possible that less abundant enzymes were not be detected. Thus, C. reinhardtii may have similar pathways to those in C. merolae and A. thaliana, although further studies are needed to confirm this speculation. It was reported that the subcellular distribution of metabolic pathways in diatoms are unique compared with photosynthetic eukaryotes harboring primary plastids (Kroth et al. 2008; Gruber et al. 2009; Fig. 1). Glycolysis and oxPPP are predicted to be incompletely localized to both the cytosol and plastids. In addition to the cytosol and plastid, enzymes of the lower half of the glycolysis pathway, GAPDH, PGK, PGAM, ENO, and PK (ID: 6–10), also exist in the mitochondria of diatoms (Fig. 1) (Liaud et al. 2000). Mitochondria-associated glycolytic enzymes were also reported in A. thaliana. These enzymes are located on the mitochondrial outer membrane, but are not found in the matrix; this spatial configuration is considered to allow the direct supply of pyruvate as a respiratory substrate (Giegé et al. 2003). In the present localization analysis in C. merolae, an association with the mitochondrion was not observed for any cytosol-localized glycolytic enzymes. The failure to detect such an association may have been caused by artifacts resulting from the use of fusion proteins or the immunodetection and cell fixation procedure. It is also possible that the extremely small cell size and volume of C. merolae would obviate the need for glycolytic enzymes to bind the mitochondrion for the efficient metabolic flow of glycolysis through the TCA cycle.

Plants have a glyoxylate cycle that is mainly localized to the glyoxysome (peroxisome), and consists of five enzymes: glyoxysomal malate synthase, malate dehydrogenase, citrate synthase, isocitrate lyase, and cytosolic aconitase (Graham 2008). In this cycle, acetyl CoA produced by the β-oxidation of fatty acids is finally converted into succinate. Although C. reinhardtii also utilizes the glyoxylate cycle (Plancke et al. 2014), we did not detect genes encoding malate synthase or isocitrate lyase in the C. merolae genome. Additionally, the peroxisomal localization of malate dehydrogenase or the cytosolic localization of aconitase were not observed. These results suggest that C. merolae does not possess a glyoxylate cycle. In Arabidopsis, a route for moving peroxisomal acetyl-CoA to the mitochondria was reported (Pracharoenwattana et al. 2005). In the route, oxaloacetate is imported into peroxisomes from mitochondria, which is converted to citrate by peroxisomal citrate synthase using acetyl-CoA produced from β-oxidation, then citrate is exported to the mitochondria. In C. merolae, three peroxisomal citrate synthases, CS-1, CS-2, and CS-3 might function in the route as in Arabidopsis.

Author contribution

TM and NS conceived this study. All authors designed and evaluated the experiments. TM performed the experiments. TM and NS wrote the manuscript.