Pyruvate:Ferredoxin Oxidoreductase Is Coupled to Light-independent Hydrogen Production in Chlamydomonas reinhardtii*

Background: The pathway(s) of light-independent H2 production in green algae are yet unknown. Results: Pyruvate:ferredoxin oxidoreductase PFR1 and [Fe-Fe]-hydrogenase HYDA1 of Chlamydomonas can be coupled for pyruvate-dependent H2 production. Conclusion: H2 production by green algae in the dark is similar to bacterial PFOR-dependent fermentation. Significance: Understanding the fermentation metabolism of green algae allows insights into plastid bioenergetic pathways. In anaerobiosis, the green alga Chlamydomonas reinhardtii evolves molecular hydrogen (H2) as one of several fermentation products. H2 is generated mostly by the [Fe-Fe]-hydrogenase HYDA1, which uses plant type ferredoxin PETF/FDX1 (PETF) as an electron donor. Dark fermentation of the alga is mainly of the mixed acid type, because formate, ethanol, and acetate are generated by a pyruvate:formate lyase pathway similar to Escherichia coli. However, C. reinhardtii also possesses the pyruvate:ferredoxin oxidoreductase PFR1, which, like pyruvate:formate lyase and HYDA1, is localized in the chloroplast. PFR1 has long been suggested to be responsible for the low but significant H2 accumulation in the dark because the catalytic mechanism of pyruvate:ferredoxin oxidoreductase involves the reduction of ferredoxin. With the aim of proving the biochemical feasibility of the postulated reaction, we have heterologously expressed the PFR1 gene in E. coli. Purified recombinant PFR1 is able to transfer electrons from pyruvate to HYDA1, using the ferredoxins PETF and FDX2 as electron carriers. The high reactivity of PFR1 toward oxaloacetate indicates that in vivo, fermentation might also be coupled to an anaerobically active glyoxylate cycle. Our results suggest that C. reinhardtii employs a clostridial type H2 production pathway in the dark, especially because C. reinhardtii PFR1 was also able to allow H2 evolution in reaction mixtures containing Clostridium acetobutylicum 2[4Fe-4S]-ferredoxin and [Fe-Fe]-hydrogenase HYDA.

Chlamydomonas reinhardtii is a photoautotrophic eukaryote that is equipped with a repertoire of fermentative enzymes allowing the cells to perform a mixed acid type fermentation (1)(2)(3)(4). Of special interest for biotechnological applications is the capability of the cells to generate molecular hydrogen (H 2 ) in the absence of oxygen (O 2 ) (5). H 2 is generated by a highly efficient hydrogenase of the [Fe-Fe] type, HYDA1 (6 -8), which, despite its extreme sensitivity toward O 2 (9 -11), is located in the chloroplast (6). The natural electron donor of HYDA1 is the photosynthetic ferredoxin PETF (7,12), and the highest rates of H 2 evolution are observed in the light (4,13). However, the cells have to be adapted to dark anaerobic conditions to induce hydrogenase activity (7,14). Upon the shift to illumination, H 2 production is only transient, because the hydrogenase enzymes are inactivated by photosynthetically generated O 2 (15), and assimilatory photosynthetic electron sinks, the Calvin cycle above all, are reactivated (16,17). However, a sustained H 2 metabolism in illuminated C. reinhardtii cells is induced by sulfur deprivation (18). Sulfur deficiency results in a strongly reduced photosystem 2 activity, mainly because of photo-damage and inadequate recycling of the D1 core subunit of photosystem 2 (19 -21). Hence, incubated in sealed flasks, sulfur-deficient algae establish hypoxic conditions despite illumination, because O 2 evolution rates drop below respiratory O 2 consumption (18). Anerobiosis elicits hydrogenase gene expression (22,23) and allows the O 2 -intolerant enzyme to be active. However, only the diminution of assimilatory electron sinks caused by cessation of cell division (18,21) allows sustained and relatively high H 2 evolution rates (17). Electrons for photosynthetic H 2 production originate from residual photosystem 2 activity (24 -26) but also from nonphotochemical plastoquinone reduction via plastidic NAD(P)H: plastoquinone oxidoreductase NDA2 (27)(28)(29). Electrons for this photosystem 2-independent, so-called indirect pathway result from oxidative starch and possibly protein degradation (18,24,26). It is generally accepted that in sulfur-deprived C. reinhardtii cells, H 2 generation serves as an alternative electron sink, allowing photosynthetic electron transport and thus energy generation to continue while preventing an over-reduction of the photosynthetic machinery (18,25).
H 2 production in nutrient-deficient green algae, however, is not the only pathway allowing the cells to maintain an energy and redox balance. Rather, the cells accumulate formate and ethanol simultaneously (30 -32). The excretion of nongaseous fermentation products prevails in anaerobic Chlamydomonas cells in the dark (4). The generation of formate, ethanol, and acetate in a ratio of 2:1:1 in dark-incubated algae resembles mixed acid fermentation of prokaryotes like the enterobacterium Escherichia coli (33). The presence of a pyruvate:formate lyase (PFL1), catalyzing the thioclastic cleavage of pyruvate to formate and acetyl-CoA in the eukaryotic alga, was therefore proposed (34) and genetically and biochemically proven (2,30,35). Because C. reinhardtii does not contain genes for a formate-hydrogen lyase complex that is responsible for formatedependent H 2 generation in fermenting E. coli (33), light-independent H 2 production in the green alga was proposed to result from pyruvate:ferredoxin oxidoreductase (PFOR) 2 activity (36). A Chlamydomonas PFOR (PFR1) was identified on the genetic (2) and protein levels, and the protein is located in the chloroplast of the cells (37,38). PFOR enzymes oxidatively decarboxylate pyruvate to yield acetyl-CoA and CO 2 , simultaneously transferring electrons to flavodoxin or ferredoxin (39). Thereby, ferredoxin reduction by plastidic C. reinhardtii PFR1 might couple fermentative pyruvate catabolism to H 2 generation in a way typical for strict anaerobic bacteria of the genus Clostridium (40 -43).
In this study we show that C. reinhardtii PFR1 heterologously produced in E. coli is active in pyruvate-and oxaloacetate-dependent methyl viologen reduction. Moreover, PFR1 enables methyl viologen-or ferredoxin-dependent H 2 production by isolated Chlamydomonas HYDA1, which proves that the long postulated pathway of dark H 2 generation in the green alga is biochemically possible.

EXPERIMENTAL PROCEDURES
Organisms and Growth Conditions-E. coli strain DH5␣ MCR (Novagen) was used for cloning procedures. Heterologous expression of C. reinhardtii PFR1 and ferredoxin encoding cDNAs was done in E. coli BL21 (DE3) ⌬iscR (44,45). E. coli strains were grown according to standard procedures as described before (30). Clostridium acetobutylicum ATCC 824 was used for heterologous synthesis of C. reinhardtii HYDA1 and homologous expression of C. acetobutylicum [Fe-Fe]-hydrogenase HYDA and ferredoxin CAC0303.
C. reinhardtii strain CC124 (137c, mtϪ nit1Ϫ nit2Ϫ) was grown in Tris acetate-phosphate (TAP) medium (46) on a shaker with bottom-up illumination of 100 mol of photons⅐m Ϫ2 ⅐s Ϫ1 at 20°C. For determination of in vivo H 2 production rates in the light or in the dark, C. reinhardtii cultures were grown until they reached a chlorophyll (Chl) content of 15 g⅐ml Ϫ1 , harvested by mild centrifugation (2,000 ϫ g, 3 min, 20°C), and resuspended in fresh TAP medium to reach a final Chl concentration of 110 g⅐ml Ϫ1 . The cell suspension was transferred to a shaded flask and purged with nitrogen gas for 4 h. In vitro hydrogenase activity was determined as described before (47) to ensure the anaerobic induction. Afterward, 2-ml aliquots of the cell culture were withdrawn using a syringe and transferred to sealed and O 2 -free head space bottles. Half of the withdrawn 2-ml aliquots were incubated in the dark and half in the light (100 mol of photons⅐m Ϫ2 ⅐s Ϫ1 ). The H 2 amount in the head space was analyzed after various time points of incubation by gas chromatography as reported previously (35).
Cloning of C. reinhardtii PFR1 cDNA-Total RNA of anaerobic algal cells was isolated according to Philipps et al. 2011 (35), and cDNA was synthesized after DNase digestion (Turbo DNAfree kit; Ambion) using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. The PFR1 coding region was amplified from this cDNA with hot start Pfu DNA polymerase (Stratagene) using oligonucleotides 5Ј-TTGGATCCCCCGCCGCTGCTGGCCGCGCCAC-CAACG-3Ј and 5Ј-TAGATATCACGTCCTTCTATTCTAG-AGTGGCCGCCGCAGCCGCTCT-3Ј. The bold letters indicate restriction sites for BamHI and EcoRV used for cloning. The forward oligonucleotide was generated in a way that the protein encoded by the cDNA would lack the putative chloroplast target sequence indicated by the first VXA amino acid motif typical for cleavage of Chlamydomonas chloroplast targeting sequences (48). The sequence was ligated with vector pASK-IBA37 (IBA GmbH) via BamHI and EcoRV restriction sites. This resulted in construct pIBA37_PFR1, allowing a tightly regulated expression via the anhydrotetracycline inducible tet promoter and in recombinant PFR1 enzyme equipped with an N-terminal His 6 tag.
Heterologous Production and Purification of His-tagged Recombinant PFR1 and Other Proteins-E. coli BL21 (DE3) ⌬iscR (44) was transformed with vector pIBA37_PFR1 by electroporation (49). Precultures were grown shaking in 200 ml of LB medium at 37°C overnight and used to inoculate four 2-liter flasks, each containing 500 ml of Vogel-Bonner minimal medium (50) with 100 g⅐ml Ϫ1 ampicillin, 0.2 M of the O 2 indicator resazurin, and 50 M thiamine hydrochloride. As soon as the cultures had reached an A 600 of 0.6 after growth at 37°C, 5 g of glucose⅐l Ϫ1 was added to the medium, and the cell suspensions were transferred to a 2-liter bottle. Expression of the PFR1 cDNA was induced by adding 0.2 g⅐ml Ϫ1 anhydrotetracycline. The flasks were sealed air tight and incubated at 8°C overnight. Then the suspension was transferred to centrifugation flasks in an anaerobic tent (Toepfer LabSystems), and the cells were harvested by centrifugation (20 min at 7,500 ϫ g, 4°C). All further steps were also conducted under strictly anaerobic conditions (1% H 2 , 99% N 2 ) in the anaerobic tent. The E. coli cell pellet was resuspended in O 2 -free buffer (100 mM Tris-HCl, pH 8.0, 10% (v/v) glycerol), and the cells were lysed by sonication (five times for 30 s; output, 25; Branson Sonifier 250). Cell debris was removed by centrifugation (60 min, 200,000 ϫ g, 4°C). The soluble fraction was filtered through sterile filters (pore size, 0.2 m; Sarstedt AG & Co.) and loaded on a 4-ml gravity flow nickel-nitrilotriacetic acid fast flow column pre-equilibrated with 100 mM Tris-HCl, pH 8.0, 10 mM imidazole, and 0.5 mM thiamine pyrophosphate (TPP). Removal of unspecifically binding proteins was obtained by washing the column with the above mentioned buffer containing increased concentrations of imidazole (40-ml steps with 10 and 20 mM imidazole). His-tagged PFR1 was eluted in 1.2-ml steps using 10 ml of buffer containing 100 mM imidazole. Protein concentration of the main fraction was determined spectrophotometrically (NanoDrop; Peqlab) at ϭ 280 nm, and this protein solution was used for further analysis. The size and purity of the eluted protein fractions were analyzed by denatur-ating SDS-PAGE according to standard techniques (51). PFR1 activity tests were conducted immediately after purification.
Heterologous production of C. reinhardtii [Fe-Fe]-hydrogenase HYDA1, as well as homologous production of HYDA in C. acetobutylicum and subsequent purification via Strep-tag II was done as described before (52,53). Recombinant Chlamydomonas [2Fe2S]-ferredoxins PETF and FDX5 were obtained as published previously (12,54). C. reinhardtii FDX2 was produced accordingly by amplifying the FDX2 coding region without the N-terminal transit peptide encoding sequence using oligonucleotides 5Ј-ATGGTAGGTCTCAGCGCCACTTTA-AGGTCACGTTTAAGACC-3Ј and 5Ј-ATGGTAGGTCTCA-TATCAGAGCTTGGACTCCTGGTCGGT-3Ј (BsaI restriction sites are indicated by bold letters). The C. acetobutylicum ferredoxin CAC0303 encoding region was amplified from the E. coli expression vector pET21c0303 (55) using hot start Pfu DNA polymerase (Stratagene) with oligonucleotides 5Ј-GAA-GGATCCGCATATAAAATAACAGACGCTTGTG-3Ј and 5Ј-AAGCGGCGCCACGGAGCTCGAATTCT-3Ј. The bold letters mark the BamHI and EheI restriction sites used for ligation with the clostridial expression vector pThydA CR1 -C-tag after excision of its insert (52). All further steps (transformation and cultivation of C. acetobutylicum, as well as protein purification) were done as described before (53).
PFR1 Activity Assays Using Methyl Viologen as Artificial Electron Acceptor-The enzymatic activity of recombinant C. reinhardtii PFR1 was determined by following the reduction of methyl viologen spectrophotometrically at ϭ 604 nm (56) using a 96-well plate reader (Beckmann, Paradigm 1113) operated in an anaerobic tent and connected to a PC running multimode analysis software. The molar extinction coefficient of methyl viologen used was ⑀ 604 ϭ 13.6 mM Ϫ1 cm Ϫ1 (57). The standard reaction mixture contained 1.4 M recombinant PFR1, 100 mM Tris-HCl, pH 8.0, 10 mM sodium pyruvate, 2 mM sodium CoA, 5 mM TPP, 10 mM methyl viologen, and 16 mM dithioerythritol in a final volume of 100 l. The reaction was started by adding PFR1 and conducted at room temperature. Absorbance was measured every 30 s until saturation was reached. The value obtained after 6 min, which was in the late linear phase, was used for determining activity. Kinetics were performed varying the concentration of one substrate while keeping the concentrations of all other substrates constant and saturating. The K m and V max values were determined in each reaction and calculated using GraphPad Prism software.
In Vitro Reconstitution of PFR1-coupled H 2 Production-For analyzing the H 2 producing capacity of [Fe-Fe]-hydrogenases upon electron delivery by PFR1-catalyzed ferredoxin reduction, recombinant enzymes and proteins were mixed in various combinations. The standard reaction mixture contained 0.7 M recombinant PFR1, 40 M ferredoxin, 0.01 M hydrogenase, 10 mM sodium pyruvate, 2 mM sodium CoA, 5 mM TPP, and 16 mM dithioerythritol in 200 l of 100 mM potassium-phosphate buffer, pH 6.8. The reactions were carried out in sealed 2-ml reaction vessels. Before incubation, the reaction mixtures were purged with argon for 3 min to reset the system. After incubation for 30 min at 37°C in a shaking water bath, 400 l of the head space were injected in a gas chromatograph (GC-2010 (Shimadzu), equipped with a PLOT fused silica coating mol-sieve column (5 Å, 10 m ϫ 0.32 mm) from Varian) to determine the H 2 concentration.

RESULTS
Recombinant C. reinhardtii PFR1 Has Typical PFOR Activity-As described in the introduction, in vivo H 2 production in C. reinhardtii is higher in the light, because electrons are provided by photosynthetic activity. We compared the light-dependent and -independent H 2 evolution rates of anaerobically adapted Chlamydomonas cell suspensions in a setup moderately modified from those reported before (as in Ref. 35, for example) (Fig. 1). In cells transferred from anaerobic conditions in the dark to illumination, H 2 accumulated to 2.13 nmol of H 2 ⅐g Chl Ϫ1 within the first 10 min, whereas cells kept in the dark produced only 0.05 nmol of H 2 ⅐g Chl Ϫ1 in the same time period (Fig. 1). In the following 50 min, the cells exposed to light produced additional 0.92 nmol of H 2 ⅐g Chl Ϫ1 (plus 43.2%) and shaded cells generated 0.22 nmol of H 2 ⅐g Chl Ϫ1 (plus 437.5%). In both illuminated and dark-incubated C. reinhardtii cell suspensions, H 2 generation rates slowed down in the following hour, because the former evolved 0.078 nmol of H 2 ⅐g Chl Ϫ1 and the latter evolved 0.152 nmol of H 2 ⅐g Chl Ϫ1 (Fig. 1). To analyze whether the low but significant H 2 production in darkadapted algae might be driven by pyruvate oxidation via PFOR activity, Chlamydomonas PFR1 was heterologously produced.
The annotated gene models of the C. reinhardtii PFR1 gene have changed considerably from the first Chlamydomonas genome version to the most recent annotation on Phytozome v8.0, C. reinhardtii v5.3. Although most parts of the primary sequences are the same in the newest gene models (Cre11.g473950.t1.1 and g1910.t2 on Phytozome v8.0, C. reinhardtii v4.3 and v5.3, respectively, and au5.g2553_t1 and SKA_Chlre2_kg.scaffold_62000019 on JGIv4), a region starting at position 940 in the Cre11.g473950.t1.1 protein model is highly variable. We aligned all available PFR1 models, as well as the sequence translated from the PFR1 cDNA amplified in this study with bacterial enzymes and concluded that the cDNA and protein sequences, respectively, obtained here are correct (i.e., amino acids AKKWVLFCARLLTQ starting at position 940 in Cre11.g473950.t1.1 are actually missing). Therefore, we used the protein sequence deduced from our cDNA for the align-ment shown in Fig. 2. The alignments revealed that the C. reinhardtii PFR1 polypeptide sequence contains all sequence motifs known to be essential for PFOR enzyme activity (marked in Fig. 2). It features an N-terminal conserved 2-oxoacid-bind-  FEBRUARY 8, 2013 • VOLUME 288 • NUMBER 6 ing site (YPITP) (58), a C-terminal TPP-binding site (59), and three [4Fe-4S]-cluster-binding signatures, two of which are typical for 2[4Fe-4S]-ferredoxins and one of which is atypical (60, 61) (Fig. 2). The comparisons with bacterial enzymes revealed that the C. reinhardtii PFR1 sequence contains an N-terminal extension that is not homologous to other PFOR proteins. Because PFR1 was shown to be localized in the chloroplast (37), we assumed that the first VXA amino acid motif (starting at position 24 of Cre11.g473950.t1.1) might represent a chloroplast targeting sequence cleavage site (48). Therefore, we excluded the respective region encoding these first 24 residues from the cDNA used for heterologous production of PFR1.

Pyruvate-dependent H 2 Production in Green Algae
After heterologous expression of the truncated PFR1 cDNA in E. coli and subsequent purification of the His-tagged protein via nickel-nitrilotriacetic acid chromatography, a protein of the expected size (144 kDa) could be eluted. Activity assays using pyruvate and CoA as substrates and methyl viologen as artificial electron acceptor resulted in a specific activity of 0.45 Ϯ 0.01 units⅐mg Ϫ1 (1 unit was defined as the conversion of 1 mol of pyruvate or CoA and the reduction of 2 mol of methyl viologen, respectively, per minute) (Fig. 3). The K m values obtained for methyl viologen, CoA, and pyruvate in this assay were 2.3, 0.7, and 1.7 mM, respectively (Fig. 3).
Chlamydomonas PFR1 Allows Pyruvate-dependent H 2 Production-To analyze whether PFR1-catalyzed pyruvate oxidation would allow H 2 production by C. reinhardtii HYDA1, reconstitution assays were performed in which various combinations of electron carriers (methyl viologen, and ferredoxins) were mixed. The combination of recombinant PFR1 and C. reinhardtii HYDA1 in the presence of pyruvate, CoA, and methyl viologen as artificial electron carrier resulted in a H 2 production rate of 400 Ϯ 66 nmol of H 2 ⅐min Ϫ1 ⅐mg PFR1 Ϫ1 , which was only slightly higher than the rate obtained with Chlamydomonas ferredoxin PETF ( Fig. 4A; 338 Ϯ 32 nmol of H 2 ⅐min Ϫ1 ⅐mg PFR1 Ϫ1 ). The K m value of PFR1 for PETF was determined as 2.4 Ϯ 0.34 M (Fig. 4B), which is considerably lower than the K m value of HYDA1 for PETF (20 -30 M) (7,12). We also examined whether two further ferredoxins, FDX2 and FDX5, would allow PFR1-dependent H 2 evolution. FDX2 (62) and especially FDX5 transcripts and FDX5 protein (54,63,64) have been shown to accumulate in anaerobic Chlamydomonas cells, which makes them candidates for being involved in pathways specific for anaerobiosis. Furthermore, both ferredoxin isoforms are localized in the plastid (54, 62) together with HYDA1 and PFR1. Using FDX2 as electron carrier, a PFR1-dependent H 2 evolution rate of 287 Ϯ 15 nmol of H 2 ⅐min Ϫ1 ⅐mg PFR1 Ϫ1 could be observed, whereas no activity was determined using FDX5 (Fig. 4A). Notably, recombinant C. reinhardtii PFR1 also allowed H 2 generation of C. acetobutylicum HYDA in the presence of the clostridial ferredoxin CAC0303 in a rate of 118 Ϯ 35 nmol of H 2 ⅐min Ϫ1 ⅐mg PFR1 Ϫ1 (Fig. 4A).
Using the amount of recombinant C. reinhardtii HYDA1 enzyme as a basis, the PFR1-and methyl viologen-mediated H 2 production rates (80 Ϯ 13 mol of H 2 ⅐min Ϫ1 ⅐mg HYDA1 Ϫ1 ) were only 16% of those determined with sodium diothionitereduced methyl viologen (516 Ϯ 42 mol of H 2 ⅐min Ϫ1 ⅐mg HYDA1 Ϫ1 ; Table 1). However, in the presence of PETF as the electron carrier, the reaction driven by 0.7 M PFR1 reached 42% of the reaction in which sodium diothionite served as chemical electron donor (68 Ϯ 7 versus 160 Ϯ 17 mol of H 2 ⅐min Ϫ1 ⅐mg HYDA1 Ϫ1 ; Table 1). In this reaction mixture, 0.7 M PFR1 was close to saturation because the rates obtained using 0.35 M PFR1 were 61 Ϯ 9, and those in assays containing 0.9 M PFR1 were 72 Ϯ 8 mol of H 2 ⅐min Ϫ1 ⅐mg HYDA1 Ϫ1 .
PFR1 Is Able to Use Oxaloacetate, but Not ␣-Ketoglutarate as a Substrate-Some bacterial PFOR enzymes have been reported to be able to oxidize various substrates such as 2-oxoacid:ferredoxin oxidoreductase from Sulfolobus species strain 7 or Sulfolobus solfataricus P1 (58,65), whereas others such as PFOR from E. coli can only oxidize pyruvate (66). We examined the methyl viologen reducing activity of C. reinhardtii PFR1 in the presence of oxaloacetate and ␣-ketoglutarate. When oxaloacetate was included in the reaction mixture, a specific activity of 0.26 Ϯ 0.03 U⅐mg Ϫ1 was observed, but no methyl viologen reduction could be detected using ␣-ketoglutarate (Fig. 5A). In a reconstitution assay using the same substrates but using the electron carrier PETF and the [Fe-Fe]hydrogenase HYDA1 to examine substrate-dependent H 2 evolution, oxaloacetate was almost as suitable as pyruvate, whereas no H 2 production could be observed in the presence of ␣-ketoglutarate (Fig. 5B).

DISCUSSION
C. reinhardtii has long been known for its complex mixed acid fermentative metabolism, which has more similarities to fermentation of bacteria or strictly anaerobic protists than to plant or animal anaerobic pathways (1,2,4,34). In addition to PFL1, which is mainly known from enterobacteria such as E. coli (33), a cDNA encoding pyruvate:ferredoxin oxidoreduc- tase was identified in C. reinhardtii (2) and shown to accumulate in anaerobic algal cells (30,63). PFR1 was proposed to be involved in dark H 2 production by C. reinhardtii (36,63) in analogy to fermentative H 2 production in clostridiae (40,42,67) or anaerobic hydrogenosome containing protists (68,69). This model was supported by the phenotype of a Chlamydomonas pfl1 mutant strain, which showed higher yields of ethanol, CO 2 , and H 2 generation in the dark (35). The fermentative pattern of the mutant was interpreted in a way that the absence of PFL1 would result in higher pyruvate supply to PFR1, which reduces ferredoxin upon oxidative decarboxylation of pyruvate to acetyl-CoA and CO 2 . In contrast, further allelic pfl1 mutants reported recently do not exhibit higher dark H 2 generation in a different experimental setup (70). Other metabolic pathways providing electrons for H 2 generation in the dark were also discussed, such as ferredoxin reduction by FNR in analogy to the pathways allowing sulfate assimilation in the roots of higher plants (71).
In this study we show that PFR1-driven H 2 production is biochemically plausible, because recombinant C. reinhardtii PFR1 is able to allow H 2 generation by isolated [Fe-Fe]-hydrogenases. Although these results cannot prove that this reaction occurs in living Chlamydomonas cells, they indicate that the PFR1-dependent pyruvate to H 2 pathway can be operable in vivo (Fig. 6). The biochemical properties of the purified PFR1 protein are similar to PFOR enzymes isolated and characterized from other organisms regarding pyruvate-and CoA-dependent methyl viologen reduction (72,73). The K m values for pyruvate and methyl viologen were in the range of the K m values determined for other PFOR enzymes, whereas the K m for CoA was higher (74 -76). Also, the specific activity of our PFR1 preparations were at the lower range when compared with other PFOR enzymes (74,77,78). This might support the physiological data obtained so far, which all speak for PFL1 being the major fermentative enzyme in C. reinhardtii wild type cultures (4,30,35). However, we cannot exclude that the protein solution contained inactive PFR1 enzymes. We observed a marked instability of the enzyme, despite its isolation and examination under strictly anaerobic conditions sufficient for the analysis of the extremely O 2 -sensitive [Fe-Fe]-hydrogenase of C. reinhardtii (9 -11). Instability of isolated PFOR enzymes has been reported before (79), and besides destruction by O 2 (73,80), loss of the TPP factor has been proposed to be one reason for this phenomenon (75).  In reaction mixtures combining PFR1 with C. reinhardtii HYDA1 and ferredoxin PETF, which is the most suitable redox partner for HYDA1 known so far (12), a pyruvate-dependent H 2 production could be observed, showing that the postulated reaction is possible. Notably, PFR1-driven H 2 generation via PETF was only 2.5-fold lower than H 2 generation using sodium dithionite-reduced PETF as an electron donor for HYDA1. This indicates that the capacity of PFR1-coupled H 2 production is quite high, despite the low specific activity of PFR1. We assume that both the low K m value of PFR1 for PETF and the high specific activity of HYDA1 contribute to the high efficiency of the coupled system. The low rates of in vivo H 2 evolution in dark-incubated algal cells are therefore probably limited by PFR1 substrate supply rather than by PFR1 activity.
PFR1-mediated H 2 production by HYDA1 was also possible using FDX2 as an electron carrier (Fig. 6), and the rates obtained were only moderately lower than with PETF. Notably, FDX2 lacks one phenylalanine residue that, in PETF, is essential for proper interaction with HYDA1 (12,81). The capability of HYDA1 to generate H 2 using FDX2 as an electron donor in the PFR1-driven system might therefore indicate a different interaction mechanism. FDX2 was suggested to be specifically involved in nitrite reduction. The protein can hardly be detected in C. reinhardtii cells incubated in ammonium-containing medium but accumulates in cells growing on nitrate and allows a higher catalytic activity of Chlamydomonas nitrite reductase than PETF (62). However, because FDX2 was also as efficient as PETF for FNR catalytic activity and even better regarding affinity (62), FDX2 might in general be used for reactions that can also employ PETF.
The results obtained with FDX2 indicated that the electron transfer reaction coupling PFR1 and HYDA1 might be different from the other electron delivering reactions analyzed so far. Therefore we analyzed whether FDX5 was able to shuttle electrons between PFR1 and HYDA1, although FDX5 is unable to drive H 2 generation activity upon artificial reduction (54). The FDX5 gene is strongly induced upon anaerobiosis (54,63) but also in copper-deficient C. reinhardtii cells (62,64,82). FDX5 is regulated by CRR1 (the copper response regulator 1) under both conditions (64). Notably, it was shown recently that the HYDA1 gene is also activated by the absence of copper (82) and partially regulated by CRR1 (83). Although a connection between FDX5 and HYDA1 might be suggested from these findings, the data presented here confirm that a direct metabolic interaction does not take place.
PFOR-mediated H 2 generation is central to anaerobic energy generation in several strict anaerobes such as amitochondriates (69) and clostridiae (67). Notably, C. reinhardtii was able to allow C. acetobutylicum HYDA activity in the presence of clostridial 2[4Fe-4S]-ferredoxin. This indicates that PFR1 kept the basic features of the evolutionary old PFOR protein (68,84), despite the fact that C. reinhardtii and clostridial PFOR sequences were calculated to be evolutionary distant (85). A similar promiscuity regarding redox partners was observed for other PFOR enzymes, such as Clostridium thermoaceticum (74) and Rhodobacter capsulatus PFORs (86).
Recombinant C. reinhardtii PFR1 had methyl viologen reducing and H 2 driving activity also in the presence of oxaloacetate as substrate, whereas ␣-ketoglutarate did not result in measurable activity. Both metabolites are intermediates of the TCA cycle, whereas the branch via ␣-ketoglutarate and succinyl-CoA is absent in the glyoxylate cycle. An anaerobically operating glyoxylate cycle is active in the alga (87), and the oxidation of malate was suggested to contribute to H 2 photoproduction (88). Pyruvate and oxaloacetate are furthermore end products of the catabolism of several amino acids. It might be assumed that oxaloacetate degradation by PFR1 supports the anaerobic operation of the glyoxylate cycle or parts thereof (Fig.  6). The physiological role of PFR1 in fermenting C. reinhardtii cells might therefore become important during long term fermentation. A 2:1:1 ratio of formate:ethanol:acetate production observed in algae after 4 -6 h of anaerobiosis (4,34,35) is typical for pyruvate:formate lyase activity. However, formate production only prevails in the first hours of anaerobiosis, whereas ethanol and especially CO 2 generation rates increase thereafter, simultaneously to a slowdown of starch degradation (34). Although speculative, a scenario might be envisioned in which PFL1 is mostly responsible for short term fermentation using starch and glucose, respectively, as substrate. In long term anaerobiosis, PFR1 activity would allow Chlamydomonas to utilize acetyl-CoA derived from fatty acids and end products of amino acid catabolism as energy sources (Fig. 6). Furthermore, the coupling to the hydrogenase is a means to dispose of electrons via the nontoxic and highly diffusible H 2 molecule.
Acknowledgments-We are thankful for the gift of pET21c0303 from Laurence Girbal (Laboratoire d'Ingénierie des Systèmes Biologiques et des Procédés, Toulouse, France). We also thank Annika Brünje and Simone Schmidt for technical support during cloning of CAC0303 and FDX2. FIGURE 6. Model of fermentative pathways involved in dark anaerobic H 2 production in C. reinhardtii. In the wild type, PFL1 is the major fermentative enzyme in short term anaerobiosis and cleaves pyruvate into formate and acetyl-CoA. The latter can be reduced to ethanol via bifunctional acetaldehyde-alcohol dehydrogenase (ADH1) (92) or converted to acetate via phosphotransacetylase and acetate kinase (PAT/ACK). In addition to PFL1, PFR1 is capable of pyruvate oxidation. In a pfl1 mutant or in long term anaerobiosis, pyruvate oxidation might be preferably catalyzed by PFR1. PFR1 transfers electrons to ferredoxins and thereby allows H 2 generation via the [Fe-Fe]hydrogenase. In addition to pyruvate, PFR1 is able to use oxaloacetate as a substrate. This might link the oxidation of other substrates such as fatty acids or amino acids to fermentative H 2 production, possibly via an anaerobically operating glyoxylate cycle or parts thereof.