Pyruvate Formate-lyase and a Novel Route of Eukaryotic ATP Synthesis in Chlamydomonas Mitochondria*

Pyruvate formate-lyase (PFL) catalyzes the non-oxidative conversion of pyruvate to formate and acetyl-CoA. PFL and its activating enzyme (PFL-AE) are common among strict anaerobic and microaerophilic prokaryotes but are very rare among eukaryotes. In a proteome survey of isolated Chlamydomonas reinhardtii mitochondria, we found several PFL-specific peptides leading to the identification of cDNAs for PFL and PFL-AE, establishing the existence of a PFL system in this photosynthetic algae. Anaerobiosis and darkness led to increased PFL transcripts but had little effect on protein levels, as determined with antiserum raised against C. reinhardtii PFL. Protein blots revealed the occurrence of PFL in both chloroplast and mitochondria purified from aerobically grown cells. Mass spectrometry sequencing of C. reinhardtii mitochondrial proteins, furthermore, identified peptides for phosphotransacetylase and acetate kinase. The phosphotransacetylase-acetate kinase pathway is a common route of ATP synthesis or acetate assimilation among prokaryotes but is novel among eukaryotes. In addition to PFL and pyruvate dehydrogenase, the algae also expresses pyruvate:ferredoxin oxidoreductase and bifunctional aldehyde/alcohol dehydrogenase. Among eukaryotes, the oxygen producer C. reinhardtii has the broadest repertoire of pyruvate-, ethanol-, and acetate-metabolizing enzymes described to date, many of which were previously viewed as specific to anaerobic eukaryotic lineages.

tein, oxidizes pyruvate, yielding 2 mol of reduced ferredoxin, 1 mol of CO 2 , and 1 mol of acetyl-CoA per mol of pyruvate (1). By contrast, PFL does not oxidize pyruvate, but uses a radical-based homolytic mechanism (2), yielding 1 mol of formate and 1 mol of acetyl-CoA per mol of pyruvate. Both PFO and PFL are oxygen-sensitive enzymes. The evolutionary origins of such enzymes for the anaerobic lifestyle in eukaryotes remains heavily debated. Either (i) the eukaryote common ancestor was a strict aerobe and anaerobic biochemistry in eukaryotes was acquired secondarily via lateral gene transfer (3,4) or (ii) it was a facultative anaerobe, with its descendant lineages having undergone specialization and differential loss (5)(6)(7)(8).
PFL plays a central role in anaerobic glucose fermentation of a number of obligatory or facultative anaerobic bacteria. The Escherichia coli PFL system has been investigated in detail. Following the shift from aerobic to anaerobic conditions, pfl transcription is induced 12-15-fold (9,10), whereas PFL protein levels increase by about 5-10-fold (11). PFL is converted post-translationally from an inactive to a catalytically active form by a 20-kDa iron-sulfur protein called pyruvate formate lyaseactivating enzyme (PFL-AE). PFL-AE introduces a radical on the ultimate glycine residue of the PFL, in an S-adenosyl-L-methionine-and flavodoxin-dependent reaction (12). The radical-containing protein is extremely sensitive to oxygen: exposure of activated E. coli PFL to air results into its cleavage in two polypeptides of ϳ82 and 3 kDa (2,13). Fragmentation of activated PFL by oxygen has also been reported in the ruminal bacterium Streptococcus bovis (14) and in the lactic acid bacterium Lactococcus lactis (15). Oxygenolytic cleavage of PFL appears to be a drastic means to inactivate an enzyme that would produce less energy than its aerobic counterpart, pyruvate dehydrogenase. Some bacteria possess a mechanism for the reversible deactivation of PFL. In E. coli, PFL deactivase activity is harbored by the bifunctional enzyme aldehyde/alcohol dehydrogenase (ADHE), which catalyzes the sequential conversion of acetyl-CoA (a product of PFL activity) into acetaldehyde and ethanol. Quenching of the radical present on activated PFL by ADHE (16,17) occurs via a yet unknown mechanism. However, ADHE does not always exhibit the PFL deactivase activity, as shown in L. lactis (15).
Reports on eukaryote PFL are limited to a few anaerobic protists and some green algae. PFL activity has hitherto been measured in two amitochondriate eukaryotes, the chytridomycetes Neocallimastix sp. L2 and Piromyces sp. E2 (18). In these anaerobic fungi, PFL was localized to the hydrogenosomes (18), which are anaerobic, hydrogen-producing mitochondria (8). Upon a shift to dark anaerobic conditions, a few unicellular photosynthetic algae such as Chlamydomonas reinhardtii and Chlorella fusca ferment their starch into formate, ethanol, and acetate in a molar ratio of 2:1:1, whereas the production of H 2 and CO 2 remains low (19). The pattern of fermentation end products together with the sensitivity of formate production to oxygen were interpreted as the * This work was supported by grants from the Deutsche Forschungsgemeinschaft (to W. M. and A. Atteia), the CNRS-Dé partement des Sciences de la Vie, and the Fondation Rhô ne-Alpes (to A. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  (47). At its C terminus, C. reinhardtii PFL exhibits a typical glycine radical signature (Prosite PS00850, (STIV)XR(IVT)(CSA)GYX(GACV)); *, glycine residue that is activated into an organic

Typical Eubacterial Enzymes in C. reinhardtii Mitochondria
presence of a PFL-like protein in the algae (19,20). Neither PFL activity nor PFL genes have been reported in animals or plants.
Here we report the expression and compartmentalization of the PFL system (ADHE, PFL, and PFL-AE) in C. reinhardtii with mass spectrometry, immunological, and molecular techniques. Furthermore, using mass spectrometry on the soluble fraction of isolated mitochondria, backed by the available C. reinhardtii genome sequence, we report the occurrence of a phosphotransacetylase-acetate kinase (PTA-ACK) pathway, which is yet undescribed but, as we show, apparently not unique among eukaryote genomes investigated to date.

MATERIALS AND METHODS
Strains and Culture Conditions-C. reinhardtii wild-type strain 11.32a and cell wall-less strain 83.82 (collection of algae Göttingen) were maintained on Tris acetate phosphate (TAP) medium (21) solidified with 1.5% (w/v) agar. TAP medium and TAP medium supplemented with 34 mM acetate, adjusted to pH 7.2 with KOH (H3 medium), were used for liquid cultures. For anaerobic cultures, cells were transferred into a Falcon tube that was introduced in a jar containing AnaerocultA (Merck). Anaerobiosis was typically achieved within 30 min. The jar was placed on a rotary shaker, under light or dark conditions. Following incubation under either aerobic or anaerobic conditions at 22°C, the algal cells were harvested by centrifugation at 2,000 ϫ g for 5 min and immediately frozen. XL1-Blue MRF' E. coli strain (Stratagene) was grown anaerobically on LB medium in the presence of 0.4% (w/v) glucose and 20 mM MOPS, pH 7.0.
Library Screening for C. reinhardtii cDNAs-Specific probes for C. reinhardtii ADHE, PFL, and PFL-AE were obtained by amplification reactions, using as template a cDNA library in ZAPII phagemid (Stratagene) made with mRNAs isolated from cells grown in light with 5% CO 2 . The primers used were: ADHE, 5Ј-GCCACCCCCCATGCTGAGGTG-3Ј and 5Ј-GTTGATCTTGGAGAAGAACTC-3Ј; PFL, 5Ј-GACGCGGGC-ATCAACGTCCAG-3Ј and 5ЈCATGGTGTCGTGGAAGGTGCG-3Ј; PFL-AE, 5Ј-GTTTTCGGAAACGTGCATTCA-3Ј; and 5Ј-CTCGG-CGCAGATGACGGGAAC-3Ј. The resultant PCR products were cloned in pGEM-T Easy vector (Promega), and sequenced. Specific probes were non-radioactively labeled using a DNA-digoxigenin labeling kit (Roche), and used to screen the same cDNA library. Isolated cDNAs were excised from the phages and retrieved in pBluescript SK(Ϫ) (Stratagene).   Protein Overexpression and Antibody Production-A partial sequence of C. reinhardtii PFL cDNA (coding for Leu 236 -Val 677 ; tPFL) was amplified by PCR using oligonucleotide primers containing the BamHI and HindIII restriction sites (underlined) as follows: 5Ј-GACGGATCCCTGTACAG-CACGGTGCGC-3Ј, and 5Ј-GTCAAGCTTCACCCACTCGGCGA-TCTCGTC-3Ј. The PCR product was cloned in pGEM-T Easy (Promega) and recloned in the BamHI/HindIII sites of the overexpression vector pQE30 (Qiagen). Following the same strategy, the nucleotide sequence corresponding to the atypical C-terminal extension (Val 496 -Lys 574 ) of subunit ␤ of C. reinhardtii mitochondrial ATPase (22) was amplified by PCR and cloned into the expression vector pET15b (Novagen). The primers used were: 5Ј-GAATTCCATATGGTGGAGAAGGCCGACAAGCTG-3Ј, and 5Ј-ATGCTGCTCGAGTTACTTCTTGGGCAGGGGCAC-3Ј. The resultant constructs were introduced into XL1 Blue MRF' or BL21 E. coli strains to produce the recombinant proteins. His-tagged proteins were purified under denaturing conditions using Ni-NTA matrix (Qiagen), as recommended by the supplier. Antibodies against tPFL were produced at Eurogentec (Leuven, Belgium), and antibodies against the C-terminal extension of subunit ␤F 1 -ATPase were produced at Charles River Laboratories.
RNA Blot Analysis-Total RNA from C. reinhardtii cells was isolated with NucleoSpinRNAII (Macherey-Nagel) and transferred onto Hybond N ϩ nylon transfer membrane (Amersham Biosciences). RNA was analyzed by gel blot hybridization in 6ϫ SSC, 5ϫ Denhardt's solution, 0.5% (w/v) SDS, 100 g/ml denatured sheared herring sperm DNA, and 50% (v/v) formamide, at 42°C. Following hybridization, membranes were washed twice in 2ϫ SSC, 0.5% (w/v) SDS at 48°C, followed by 2 washes in 0.2ϫ SSC, 0.5% (w/v) SDS, at 48°C. The PFL probe was a 450-bp amplified PCR fragment corresponding to nucleotides 1571-2031 of the isolated PFL cDNA, PFL-AE probe was a 550-bp fragment released after PstI digestion of PFL-AE cDNA, and ADHE probe was a 1.5-kb fragment (corresponding to the 5Ј-end of the open reading frame) released after digestion of ADHE cDNA with SacI. Probes were purified with QIAquick gel extraction columns (Qiagen) and labeled with [ 32 -P]dCTP by random priming.
Isolation of C. reinhardtii Chloroplasts and Mitochondria-Cell wallless C. reinhardtii cells were grown on TAP medium to late exponential phase, harvested at 2,000 ϫ g in a GSA rotor (Sorvall) for 8 min, and resuspended in breaking buffer (0.25 M sorbitol, 50 mM Tris, 50 mM Mes, 10 mM MgCl 2 , 1 mM MnCl 2 , 2 mM EDTA, pH 7.2, KOH). Cell suspension was passed through a cell disruptor (BioNeb, Glas-Col, Terre Haute, IN) with N 2 gas pressure at 20 p.s.i. Cell lysate was centrifuged shortly to 5,000 rpm in a GSA rotor (Sorvall). The pellet was used to prepare intact chloroplasts (23). Mitochondria, in the supernatant, were pelleted at 11,000 rpm for 10 min in SS34 rotor (Sorvall) and further purified on Percoll gradient (24).
Protein Analysis-Frozen cells were thawed and resuspended in 50 mM dithiothreitol and 50 mM Na 2 CO 3 . For protein concentration determination, an aliquot of cells was precipitated with 80% (v/v) acetone and resuspended in 0.4% (w/v) SDS. Protein content was determined using the BCA Assay reagent kit (Pierce). For protein gels, cells were solubilized in 2% (w/v) SDS and 1 mM ␤-mercaptoethanol and heated at 90°C, for 2 min. Proteins were separated by SDS-PAGE using a 10% acrylamide (w/v) gel and subsequently transferred onto Hybond C nitrocellulose membranes (Amersham Biosciences) for immunodetection. Blots were incubated for 1 h with primary antibodies as follows: 1:5000 for anti-C. reinhardtii PFL (this work), 1:20000 for anti-C. reinhardtii ␤F1-ATPase (this work), and 1:50000 for anti-C. reinhardtii LHC proteins (Dr. O. Vallon, IBPC, France). Immunodetection was carried out using the BM chemiluminescent protein blot kit (Roche Diagnostics) accord-ing to the manufacturer's instructions. To reprobe the blots, the Restore Western blot Stripping Buffer (Pierce) was used. TMBZ/H 2 O 2 method was used (25) for in-gel cytochrome detection. Dual color precision plus protein standards (Bio-Rad) and the Benchmark TM pre-stained protein ladder (Invitrogen) were used to estimate molecular mass.
Mass Spectrometry and Protein Identification-Mitochondria were resuspended in 50 mM MOPS, pH 7.4, in the presence of 0.1 mM phenylmethylsulfonyl fluoride and 1 mM ⑀-amino caproic acid to a protein concentration of 15 mg/ml and sonicated 4 times for 10 s. Soluble and membrane components were fractionated by ultracentrifugation. The soluble fraction was run on a 12% (w/v) acrylamide SDS-PAGE. Gel pieces were excised from the Coomassie Blue-stained gel and subjected to tryptic digestion (26). Samples were injected into a nanoLC system directly coupled to a QTOF Ultima mass spectrometer (Waters). MS and MS/MS data were acquired and processed automatically using Masslynx 4.0 software. Data base searching was performed with Mascot 2.0, using NCBInr, the JGI C. reinhardtii version 2.0 gene models and EST databases. Proteins identified with at least 2 peptides showing scores higher than 40 were validated automatically. Peptides with scores between 18 and 40 were checked manually to confirm or cancel Mascot suggestion.
Sequence Analysis-Expressed sequence tag (EST) clones of C. reinhardtii were obtained from the EST databases. A draft of the Chlamydomonas genome is available. Sequence alignments were done with ClustalW 1.82 and refined manually. Motif search was done using the Prosite Data base of the protein family and domains. Predictions for intracellular targeting were performed using Predotar version 1.03, Tar-getP version 1.01, and PSORT.
Networks-Homologues were retrieved by BLAST searches against sequenced genomes and the nonredundant protein data base at GenBank. Sequences were aligned using ClustalW (27). Gapped positions were excluded from phylogenetic analysis. A protein LogDet distance matrix (28) for each alignment was calculated with LDDist (29). Splits were determined by Neighbor-Net (30), a variant of the Neighbor-Joining algorithm (31), and plotted as planar graphs with Splitstree (32). For calculating the LogDet distances among ACK sequences, amino acids were recorded as Dayhoff classes, yielding six possible character states instead of 20 (7).

RESULTS
C. reinhardtii Expresses PFL, PFL-AE, and ADHE-A mass spectrometry proteomics survey of highly purified Chlamydomonas mitochondria revealed numerous peptides matching PFL data base entries (see below). Searching the C. reinhardtii EST databases identified a number of clones for PFL along with homologues for its radical-activating (PFL-AE) and putative deactivating (ADHE) enzymes. These ESTs were used to produce specific DNA probes and to isolate full-length cDNAs for the proteins of interest. The sequence of the longest PFL cDNA clone (3379 bp) contains an open reading frame of 2559 bp coding for a putative protein of 852 amino acids. The inferred PFL amino acid sequence shares extensive similarity with glycyl radical enzymes, an emerging class of anaerobic enzymes. These enzymes use a radical chemistry for carbon-carbon bond formation or cleavage (33)(34)(35). Fig. 1A shows sequence alignment of predicted C. reinhardtii PFL with different members of the larger PFL family (34). The highest sequence identity with prokaryote enzymes was found with the well characterized PFL of E. coli (56%), whereas the identity with other members of the family of PFL-related enzymes was significantly lower: 21% with the glycerol dehydratase from Clostridium butyricum and only 11% with benzylsuccinate synthase from Thauera aromatica. C. reinhardtii PFL exhibits the two adjacent cysteinyl residues (Fig. 1A, Cys 513 -Cys 514 ), which are present in all PFLs shown to catalyze homolytic pyruvate cleavage (2). As compared with bacterial PFLs, C. reinhardtii PFL exhibits a long N-terminal extension (Fig. 1A), suggestive of organelle targeting. This extension contains many Met residues, and computer-based subcellular localization predictions vary according to the assumed intiation Met.
The C. reinhardtii PFL-AE cDNA clone (2007 bp) contains an open reading frame of 978 bp coding for a protein of 326 amino acids. The predicted protein exhibits a radical activating enzyme signature (Prosite PS01087) (Fig. 1B), characteristic of all radical S-adenosylmethionine enzymes (36). The highest identity was found with PFL-AE homologues (35% identity with E. coli), whereas the activating enzymes of C. butyricum glycerol dehydratase and T. aromatica benzylsuccinate synthase were more distantly related (23 and 18% identity, respectively). C. reinhardtii PFL-AE exhibits a long N-terminal extension (Fig. 1B) that is predicted to serve as a chloroplast targeting peptide by PSORT, Predotar, and TargetP.
The isolated C. reinhardtii ADHE cDNA (3804 bp) has an open reading frame of 2865 bp encoding a precursor protein of 954 amino acids, sharing 62% identity with ADHE from the cyanobacterium Thermosynecchococcus elongatus (Q8DM94). Conservation with eukaryote enzymes is to some extent lower with 49% identity with Piromyces sp. ADHE (Q6WJD5) and 53% identity with ADHE of its nonphotosynthetic counterpart Polytomella sp. (Q70YJ9) (37). C. reinhardtii ADHE has a long N-terminal extension predicted to target the protein to the mitochondrion.
Effect of Anaerobiosis and Darkness on ADHE, PFL, and PFL-AE mRNA Levels-Kreuzberg (19) reported that formate production in C. reinhardtii increased during the first hours of anaerobiosis, reaching its highest levels after 6 h. Accordingly, transcript levels for ADHE, PFL, and PFL-AE in cells exposed to aerobiosis and to 6 h of anaerobiosis were compared by RNA blot analysis. PFL transcripts were hardly detectable in cells grown in aerated cultures (Fig. 2). Oxygen removal increased PFL mRNA levels, which were highest in cells kept in darkness (Fig. 2). PFL-AE transcripts were low in cells grown in aerated cultures and were not significantly influenced by anaerobiosis, in contrast to PFL transcripts (Fig. 2). No ADHE mRNAs could be detected in the RNA samples used to follow PFL and PFL-AE transcripts (not shown).
Effect of Anaerobiosis and Darkness on PFL Levels-Truncated C. reinhardtii PFL protein (tPFL; Leu 236 -Val 677 of the precursor protein) was expressed in E. coli and used for antibody production. The produced antibody recognized C. reinhardtii PFL as a protein of ϳ78 kDa as well as PFL from E. coli and Neocallimastix frontalis (supplemental Fig. I). Protein blots were carried out to determine the effect of 6 h of anaerobiosis on PFL levels. As revealed by immunoblotting, no significant changes in PFL levels occurred under anaerobiosis in either light or dark conditions (Fig. 3A). The antiserum detected a single band of ϳ78 kDa in protein extracts from light-exposed cells (with or without  . Immunoblots to compare PFL levels in C. reinhardtii cells exposed to an oxygen-depleted environment. A, effect of a short anaerobiosis on PFL levels. A culture of C. reinhardtii wild-type strain cells grown on TAP medium to 2 ϫ 10 6 cells/ml was divided in three aliquots. Two aliquots were transferred to an anaerobic jar (ϪO 2 ) under light (L) or dark (D) conditions. The third aliquot, maintained in aerobic conditions (ϩO 2 ), was used as control. After 6 h of incubation under agitation, the cells from two independent experiments (Series I and II) were harvested, and prepared for SDS-PAGE analysis. Proteins (40 g) were run on a 10% (w/v) acrylamide gel, transferred to nitrocellulose, and probed with anti-PFL antiserum or anti-light harvesting complex protein antiserum. Note that the ratio between the two PFL bands varied from preparation to preparation. B, a culture of C. reinhardtii wild-type cells grown on TAP medium (2 ϫ 10 6 cells/ml) was divided into six aliquots. Cells in the aliquots were harvested, and resuspended in fresh culture medium as indicated. Cells were then exposed to either light (L) or darkness (D), in the presence (ϩO 2 ) or absence (ϪO 2 ) of oxygen for 24 h. Cells were then pelleted and resuspended in gel loading buffer for protein analysis. Proteins, separated on a 10% (w/v) acrylamide SDS-PAGE, were either stained with Coomassie Blue (CBB) or transferred to nitrocellulose for further immunodetection using anti-C. reinhardtii PFL antiserum (anti-PFL). APRIL 14, 2006 • VOLUME 281 • NUMBER 15 oxygen), but detected two bands of close molecular mass in protein extracts from cells maintained in anaerobiosis in the dark.

Typical Eubacterial Enzymes in C. reinhardtii Mitochondria
The effects of prolonged darkness or anaerobiosis on PFL accumulation were also investigated. TAP-growing cells were harvested in their exponential growth phase, resuspended in fresh culture medium, and incubated for 24 h in the conditions of interest. The medium for resuspension was either TAP medium (TAP) or a TAP-derived medium that contains 3-fold more acetate (H3). Cells were then harvested and prepared for protein analysis. Protein blots showed the presence of PFL in all cells analyzed, although two forms of close M r were observed. The larger form of ϳ78 kDa was found in cells from aerated cultures, regardless of the amount of acetate or the illumination, whereas the smaller form of ϳ75 kDa was detected in cells kept under anaerobic conditions, in continuous darkness or continuous light (Fig. 3B). PFL levels appeared to be slightly lower in cells grown on H3 medium relative to the cells grown on TAP medium (Fig. 3B). Further work will determine whether acetate influences PFL steady-state levels in the algae growing under aerobic conditions. Intracellular Localization of PFL in C. reinhardtii-PFL localization protein was investigated by protein blot analysis. TAP-grown C. reinhardtii cells were harvested, and disrupted by nebulization. Chloroplasts and mitochondria were fractionated by differential centrifugation and further purified on Percoll gradients. To ascertain the purity of the organelle fractions, each was tested for the enrichment of specific marker proteins. Heme staining showed a low contamination of chloroplasts by mitochondrial cytochrome c 550 , whereas chloroplast cytochrome f and cytochrome b 6 were not detected in mitochondrial fractions (Fig. 4B). Furthermore, immunoblot analyses revealed that purified mitochondria lacked detectable amounts of light-harvesting complex proteins, whereas chloroplasts contained traces of mitochondrial F 0 F 1 -ATPase (␤F 1 -ATPase) (Fig. 4C). These data indicated that purified mitochondria were devoid of chloroplasts while chloroplasts were slightly contaminated with mitochondria. The same protein blots probed for PFL revealed the occurrence of the protein in both organelles (Fig. 4C). The electrophoretic mobility of the chloroplast and mitochondrial PFL proteins differs by ϳ3 kDa.
Protein Identification by Mass Spectrometry-Protein blot analysis of soluble and membrane fractions of C. reinhardtii mitochondria indicated that PFL is a soluble protein (not shown). Therefore, the soluble mitochondrial fraction was run on one-dimensional SDS-PAGE, discrete bands were excised from the gel (Fig. 5A) and treated for protein identification by tandem mass spectrometry (26). In the 60 -75-kDa region, 22 peptides that matched the predicted PFL sequence were identified. These peptides cover a large part of the full-length PFL, although no peptide corresponding to residues Met 1 -Lys 107 was identified (Fig.  5B). The coverage of the PFL sequence is comparable with that obtained for the mitochondrial aconitase, indicating that PFL is an abundant protein.
Proteome analysis also uncovered peptides matching proteins that form the PTA-ACK pathway, which in a number of bacteria reversibly interconverts acetyl-CoA and acetate (38). ACK (EC 2.7.2.1) phosphorylates acetate to acetyl phosphate, which is then converted into acetyl-CoA by phosphotransacetylase (PTA, EC 2.3.1.8). Three tryptic peptides that match a putative PTA were identified in the 70 -85-kDa range (Fig. 5A). All peptides are specific to annotated PTA1 (JGI C_870001), whereas no peptide corresponding to a second putative PTA, annotated PTA2 (C_330071), was obtained. The molecular mass range from PTA1 is 10 -20 kDa higher than the mass of 60 kDa calculated for annotated PTA1, suggesting that the gene model may be incomplete. In the 38 -42-kDa range, six tryptic peptides that match an ACK sequence were identified. In the C. reinhardtii genome, two genes are annotated as acetate kinase. Whereas one of the six tryptic peptides is common to both ACK1 (C_170112) and ACK2 (C_330070), the other five are specific to ACK2 (Fig. 5C). The molecular mass of ACK2 (Fig. 5A) is in agreement with the mass of 43 kDa calculated from the gene model. No peptide specific to PFL-AE or ADHE were identified in the fraction of soluble proteins, suggesting either that these proteins are present in very low amounts in the sample analyzed or that they are localized to another subcellular fraction.

Occurrence of PFL and PFL-AE in a Photosynthetic
Eukaryote-Oxygen-sensitive formate production in C. reinhardtii cells under dark, anaerobic conditions had suggested the activity of pyruvate formatelyase (19), an enzyme rare among eukaryotes. The present identification of PFL peptides in C. reinhardtii, along with cDNAs for PFL and PFL-AE, indicate the existence of a PFL system in the algae. Identified C. reinhardtii PFL and PFL-AE are well conserved with their counterparts in bacteria and amitochondriate protists.
In its natural habitats, such as soil and fresh water ponds, C. reinhardtii is exposed to anaerobiosis even more so in darkness. Short anaerobiosis in the dark (6 h) led to a clear increase of PFL transcript levels, but not of protein levels. Ambient levels of PFL in the algae may ensure readiness for immediate adaptation to fermentative conditions, without the need for de novo protein synthesis. More detailed studies will be required to fully describe the regulation of the PFL at RNA, protein, and activity levels.
Compartmentalization of the PFL System-Anaerobic formate production in C. reinhardtii was previously measured in fractions enriched in chloroplasts and mitochondria (20), which suggested the presence of a PFL in both organelles. Here, immunoblots revealed the occurrence of PFL in chloroplasts and mitochondria purified from aerobically grown C. reinhardtii cells. Library screening identified several PFL cDNAs corresponding to a single mRNA, and searching the draft of the C. reinhardtii genome sequence and the EST databases provided no evidence for two PFL genes and only one PFL transcript was detected in RNA blots. This may suggest that PFL is dual targeted to chloroplasts and mitochondria. Dual targeting is not uncommon in higher plants (39) but has not been previously reported in the green algae.
The ϳ3-kDa difference observed between the chloroplast and the mitochondrial PFL forms might be explained by differential processing of the cytosolic precursor. Alternatively, the 3-kDa difference might be explained by differential activation and radical-induced cleavage at the C-terminal end. However, it is not currently known whether C. reinhardtii PFL may be subject to oxygenolytic cleavage as its counterparts in bacteria (2,14,15). In our mitochondrial proteome analysis, one peptide that matches with the C-terminal sequence of PFL was identified. This peptide stems from (i) a protein of high molecular mass (ϳ60 -75 kDa) and (ii) a region of the PFL that is C-terminal to the glycine residue to be activated into a glycyl radical. In E. coli, PFL is a homodimer that is post-translationally activated by introduction of a radical on the ultimate Gly residue of only one of the two monomers (2,15). Whether only one or both PFL subunits are activated in C. reinhardtii is yet unknown.
The rationale behind the differential compartmentalization of PFL in C. reinhardtii is not obvious. Indeed, PFL produces formate, a toxic (but freely diffusible) end product. Plants, which do not possess PFL, have a formate dehydrogenase that oxidizes formate into CO 2 in the presence of NAD ϩ . Plant formate dehydrogenase is localized to the mitochondria (40). No biochemical or molecular evidence for the presence of a formate dehydrogenase has yet been reported in the green algae, which would agree with a role of PFL in core energy metabolism and the status of formate as a genuine end product.
Recycling Coenzyme A in C. reinhardtii Mitochondria-In E. coli, acetyl-CoA resulting from PFL activity is converted into either ethanol or acetate (38). Acetate dissimilation via the PTA-ACK pathway generates one ATP molecule per acetyl-CoA but does not consume reducing equivalents. Ethanol, produced by ADHE, reoxidizes two molecules of NADH per acetyl-CoA without ATP production. In the anaerobe eukaryotes Giardia and Entamoeba, fermentative production of ethanol is catalyzed by ADHE, whereas acetate is produced by ADP-forming acetyl-CoA synthase (EC 6.2.1.13) (1) (Fig. 6D). The ethanol to acetate ratio depends upon oxygen tension (1). In this sense, the ACK-PTA route, which produces ATP and acetate from acetyl-CoA, P i , and ADP in most bacteria would be functionally equivalent to the single enzyme acetyl-CoA synthase in amitochondriate eukaryotes.
Predicted C. reinhardtii ADHE exhibits features typical of ADHE from amitochondriate eukaryotes and bacteria. Under the conditions investigated here, i.e. light versus dark and aerobiosis versus anerobiosis, we failed to detect ADHE transcripts and protein (supplemental Fig. II). In this respect, the green algae differs from its colorless counterpart Polytomella sp. where ADHE is a major protein of the mitochondrial matrix in aerobically grown cells (37) (Fig. 6C). The physiological conditions that lead to expression and activity of ADHE in the green algae are not known. Whether C. reinhardtii ADHE is involved in the anaerobic production of ethanol, and concomitant regeneration of NAD ϩ , remains to be determined.
Several peptides from a PTA-ACK pathway were identified in the soluble fraction of mitochondria isolated from aerobically grown algae. These peptides were specific to PTA1 and ACK2 but distinct from PTA2 and ACK1 whose genes are adjacent to HYDA2 (C_330072) encoding the chloroplast hydrogenase of the algae (41). This gene organization as well as the lack of peptides specific to PTA2 and ACK1 in isolated mitochondria suggest that these proteins are located to the chloroplast (Fig. 6). Phylogenetic networks for C. reinhardtii ACK1 (Fig.  7) and PTA2 (supplemental Fig. III) indicate a common ancestry for both enzymes with the corresponding homologues from the oomycete Phytophthora sojae, suggesting that the two-step pathway that reversibly converts acetyl-CoA into acetate was present in the common ancestor of oomycetes and plants (42). An ACK but no PTA was found in the amitochondriate eukaryote Entamoeba histolytica genome sequence (Fig. 7). The PTA-ACK system is widespread among prokaryotes (38). Notwithstanding the activity measurements by Kreuzberg et al. (20), ACK and PTA represent novel and typically eubacterial enzymes among eukaryotes. C. reinhardtii is capable of growing heterotrophically on acetate and it is commonly accepted that acetate is assimilated by an acetyl-CoA synthase and metabolized to triose following entry into the glyoxylate cycle (21,43). The mitochondrial PTA-ACK pathway in C. reinhardtii may represent an alternative route for acetate assimilation.
Pyruvate Degrading Systems in the Green Algae C. reinhardtii-PFL Versus PFO-A gene coding for a putative pyruvate:ferredoxin oxidoreductase (1303 amino acids) was identified in the C. reinhardtii genome (C_140055). Search of EST databases identified only two PFO clones, which correspond to a conserved N-terminal domain (clone BP093588; His 241 -Ala 305 ) and to the predicted C terminus (BQ821311; Lys 1216 -His 1303 ). The PFO sequence inferred from the gene model exhibits four highly conserved PFO motifs including two [4Fe-4S] ferredoxins, and the iron-sulfur binding region signature (Prosite PS00198) (Cys 766 -Cys 777 and Cys 822 -Cys 833 ), which participate in electron transfer to ferredoxin (44). C. reinhardtii PFO shares 50% identity with PFOs of the strictly anaerobic, sulfate-reducing bacterium Desulfovibrio vulgaris Hildenborough (YP_012236), and the anaerobic parasite E. histolytica (EAL51636) and 49% identity with mitochondrial pyruvate: NADP ϩ oxidoreductase (PNO) of Euglena gracilis (Q94IN5). In the phylogenetic network, C. reinhardtii PFO shares common ancestry with other eukaryote PFO homologues (supplemental Fig. IV).
In contrast to Euglena pyruvate:NADP ϩ oxidoreductase (44), C. reinhardtii PFO does not exhibit a C-terminal NADPH-generating domain, suggesting that the algal enzyme probably accepts electrons from ferredoxin like its homologues from amitochondriate protists (1,45). In C. reinhardtii, ferredoxin is present in the stroma where it donates electrons to ferredoxin:NADPH oxidoreductase but also to a hydrogenase (under anaerobic conditions) (46). We thus propose that C. reinhardtii PFO is localized to the stroma where under anaerobic conditions its activity might be coupled to that of the hydrogenase via ferredoxin (Fig. 6A). Under anaerobiosis in the dark, C. reinhardtii cells produce large amounts of formate, whereas production of CO 2 is low (19), which might indicate that under these conditions PFO is poorly involved in the survival of the algae. The anaerobic enzymes PFL and PFO in the green algae represent an unexpected biochemical and evolutionary link between energy metabolism in amitochondriate and mitochondriate protists.
C. reinhardtii possesses genes for pyruvate dehydrogenase, PFL, and PFO. This 3-fold pyruvate-metabolizing repertoire is not uncommon in prokaryotes (38) but it is hitherto unique among eukaryotes. Together with the presence of ADHE and HYD, the enzymatic repertoire typical of eukaryote anaerobes (1) is almost completely present in C. reinhardtii. The only exception is that ADP-forming acetyl-CoA synthase is lacking, whereby the functionally equivalent PTA-ACK route is present. The large variety of metabolic abilities evidenced in this work are likely helping the photosynthetic algae to optimize its metabolism to the ever changing environmental conditions.