Identification of NdhL and Ssl1690 (NdhO) in NDH-1L and NDH-1M Complexes of Synechocystis sp. PCC 6803*

The subunit compositions of two types of NAD(P)H dehydrogenase complexes of Synechocystis sp. PCC 6803, NDH-1L and NDH-1M, were studied by two-dimen-sional blue-native/SDS-PAGE followed by electrospray tandem mass spectrometry. Fifteen proteins were observed in NDH-1L including hydrophilic subunits (NdhH, -K, -I, -J, -M, and -N) and hydrophobic subunits (NdhA, -B, -E, -G, -D1, and -F1). In addition, NdhL and a novel subunit, Ssl1690 (designated NdhO), were shown to be components of this complex. All subunits men-tioned above were present in the NDH-1M complex except NdhD1 and NdhF1. NdhL and Ssl1690 (NdhO) were homologous to hypothetical proteins encoded by genomic DNA in higher plants, suggesting that chloroplast NDH-1 complexes contain related subunits. Diagnostic sequence motifs were found for both NdhL and NdhO homologous proteins. Analysis of ndhL deletion mutant (M9) revealed the presence of assembled NDH-1L and NDH-1M complexes, but these complexes appear to be functionally impaired in the absence of NdhL. Both NDH-1 complexes were absent in the molecular two-dimen- Millipore), visualized Coomas- R-250. protein four gels excised, destained. of Protein for Mass Spectrometry Silver-stained gels

The subunit compositions of two types of NAD(P)H dehydrogenase complexes of Synechocystis sp. PCC 6803, NDH-1L and NDH-1M, were studied by two-dimensional blue-native/SDS-PAGE followed by electrospray tandem mass spectrometry. Fifteen proteins were observed in NDH-1L including hydrophilic subunits (NdhH, -K, -I, -J, -M, and -N) and hydrophobic subunits (NdhA, -B, -E, -G, -D1, and -F1). In addition, NdhL and a novel subunit, Ssl1690 (designated NdhO), were shown to be components of this complex. All subunits mentioned above were present in the NDH-1M complex except NdhD1 and NdhF1. NdhL and Ssl1690 (NdhO) were homologous to hypothetical proteins encoded by genomic DNA in higher plants, suggesting that chloroplast NDH-1 complexes contain related subunits. Diagnostic sequence motifs were found for both NdhL and NdhO homologous proteins. Analysis of ndhL deletion mutant (M9) revealed the presence of assembled NDH-1L and NDH-1M complexes, but these complexes appear to be functionally impaired in the absence of NdhL. Both NDH-1 complexes were absent in the ndhB deletion mutant (M55).
The proton-pumping NADH:ubiquinone oxidoreductase catalyzes the electron transfer from NADH to ubiquinone linked with proton translocation across the membrane (1). Subunit composition of the enzyme varies among organisms. In complex I originating from mammalian mitochondria, 45 different proteins were discovered (2,3). In bacteria, the corresponding complex NDH-1 consists of 14 different polypeptides. Homologues of these 14 proteins are found among subunits of the mitochondrial complex I, and therefore bacterial NDH-1 might be considered as a model proton-pumping NADH dehydrogenase with a minimal set of subunits (4). Escherichia coli NDH-1 readily disintegrates into 3 subcomplexes: a water-soluble NADH dehydrogenase fragment (NuoE, -F, and -G), the connecting fragment (NuoB, -C, -D, and -I), and the membrane fragment (NuoA, -H, -J, -K, -L, -M, -N) (5).
In cyanobacteria and their descendents, chloroplasts of green plants, the subunit composition of NDH-1 remains obscure. The genes for eleven subunits NdhA-NdhK, homologous to the NuoA-NuoD and NuoH-NuoN of the E. coli complex, have been found in the genome of Synechocystis sp. PCC 6803 (hereafter Synechocystis 6803) (9). Further, Synechocystis 6803 has a family of 6 ndhD genes and a family of 3 ndhF genes. It was proposed (7,8) that multiple forms of NDH-1, which vary in NdhD/NdhF composition, exist in cyanobacteria. Genetic approach suggests that NDH-1 containing NdhF1 together with either NdhD1 or NdhD2 is involved in cyclic electron flow passing electrons from PSI back to the plastoquinone pool whereas the enzymes containing NdhF3/NdhD3 or NdhF4/ NdhD4 play important roles in CO 2 uptake (9, 10). Further, based on studies of the Synechocystis 6803 mutant RKb, Ogawa (11) suggested that the Ssr1386 protein also belongs to cyanobacterial NDH-1 and designated it NdhL.
Attempts to isolate intact NDH-1 complexes from cyanobacteria have met with serious difficulties (12,13). Recently Prommeenate et al. (14) purified the largest NDH-1 (460 kDa) from Synechocystis 6803 using His-tagged NdhJ. Both hydrophilic (NdhH, -I, -J, and -K) and hydrophobic (NdhA, -B, -C, -D1, -F1, and -G) subunits were identified in this complex. Moreover, it was demonstrated that the cyanobacterial NDH-1 complex contains additional subunits, NdhM and NdhN, compared with the minimal set of the bacterial enzyme. However, NdhL has not been detected in any of purified preparations of cyanobacterial NDH-1. Thus, the complete subunit composition of NDH-1 from organisms capable of photosynthesis still remains uncertain.
Recently we reported two multisubunit complexes, NDH-1L and NDH-1M, which represent distinct NDH-1 complexes in the thylakoid membrane of Synechocystis 6803 (15). 1 NDH-1L was shown to be essential for photoheterotrophic cell growth, whereas expression of NDH-1M was a prerequisite for CO 2 uptake and played an important role in growth of cells at low CO 2 . Here we report the subunit composition of these two complexes. Fifteen proteins were discovered in NDH-1L including NdhL, a new component of the membrane fragment, and Ssl1690 (designated as NdhO), a novel peripheral subunit. NDH-1M comprised the same subunits as NDH-1L except NdhD1 and NdhF1. We also show that deletion of NdhL does not prevent the assembly of the NDH-1L and NDH-1M complexes although it is known to severely hamper their function.

EXPERIMENTAL PROCEDURES
Cell Culture Conditions-Synechocystis 6803 glucose tolerant strain (WT), 2 the ndhL gene inactivation mutants M9 (17) and the ndhB gene inactivation mutants M55 (18) were grown in BG-11 medium (19) at 32°C under 50 mol of photons m Ϫ2 s Ϫ1 with gentle agitation, in conditions of high CO 2 (3% CO 2 in air) or low CO 2 (air level) at pH 7.5. For mutant strains, the appropriate antibiotics were supplemented. * 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  Isolation of Crude Thylakoid Membranes-The cells were harvested at the logarithmic phase, washed twice with 20 ml of washing buffer (50 mM Hepes-NaOH, pH 7.5, 30 mM CaCl 2 ). The thylakoids were isolated according to Gombos et al. (20) as follows. The cells were suspended in 2 ml of isolation buffer (50 mM Hepes-NaOH, pH 7.5, 30 mM CaCl 2 , 800 mM sorbitol, 1 mM ⑀-amino-n-caproic acid), supplemented by the same volume of glass beads and disrupted by vortexing eight times at the highest speed for 1 min at 4°C with 1-2 min cooling on ice between runs. The crude extract was centrifuged at 3000 ϫ g for 5 min to remove the glass beads and unbroken cells. Membranes were pelleted by centrifugation at 17,000 ϫ g for 20 min and resuspended in storage buffer (50 mM Tricine-NaOH, pH 7.5, 600 mM sucrose, 30 mM CaCl 2 , 1 M glycinebetaine).
Separation of NDH-1 Protein Complexes and Their Subunits-The blue-native PAGE (BN-PAGE) of Synechocystis 6803 membranes was performed basically as described earlier (21) with modifications for cyanobacteria (15). Membranes were washed with 330 mM sorbitol, 50 mM BisTris, pH 7.0, 250 g/ml Pefabloc, and subsequently suspended at the final concentration of 20 g of protein/l in 20% glycerol w/v, 25 mM BisTris pH 7.0, 10 mM MgCl 2 , 0.002 units/l RNase-Free DNase RQ1 (Promega, Madison, WI). The samples were incubated on ice for 10 min, and the equal volume of 3% n-dodecyl-␤-D-maltoside in the same buffer was added. Solubilization was performed by incubation on ice for 10 min and followed at room temperature for 20 min. Insoluble material was removed by centrifugation at 18,000 ϫ g for 15 min. Solubilized membranes were mixed with 1:10 volume of 0.1 M EDTA and 1:10 volume of sample buffer (5% Serva blue G, 200 mM BisTris, pH 7.0, 75% sucrose, 1 M ⑀-amino-n-caproic acid) and directly applied to 0.75 mm-thick 5-12.5% acrylamide gradient gel (Hoeffer Mighty Small mini-vertical unit) in amounts of ϳ150 g of protein per well. Electrophoresis was performed at 4°C with a gradual increase of voltage (50 V for 30 min, 75 V for 30 min, 100 V for 1 h, 125 V for 30 min, 150 V for 1 h, 175 V for 1 h, and 200 V for 1 h). For electrophoresis in the second dimension, a lane of the BN gel was cut out and incubated in Laemmli SDS sample buffer supplemented with 5% ␤-mercaptoethanol and 6 M urea for 1 h at 23°C. The lane was then laid onto a 1-mm thick 14% SDS-PAGE gel with 6 M urea (22). Prestained protein markers, broad range (New England BioLabs, Beverly, MA) were used for estimation of apparent molecular masses of proteins in SDS/PAGE. The proteins were visualized by silver staining (23).
N-terminal Protein Sequencing-Proteins separated by two-dimensional BN/SDS-PAGE were electrotransferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore), and visualized with Coomassie Brilliant Blue R-250. The corresponding protein spots from four gels were excised, combined, and destained. Sequencing was performed by Dr. Jukka Hellman at the University of Turku using the Edman degradation method.
Preparation of Protein Samples for Mass Spectrometry Analysis-Silver-stained protein spots were excised from gels and digested with Trypsin Gold (Promega) according to Shevchenko et al. (24). Spots with low intensity of staining were collected from several gels (up to six) and combined. After digestion, peptides were eluted by 1-h washing of gel pieces first with 5% formic acid, then with 5% formic acid, 50% acetonitrile, and finally with 5% formic acid, 80% acetonitrile. The extracts were combined, dried, and stored at Ϫ20°C until further use.
Identification of Proteins by MALDI-TOF-Eluted and dried protein digests were purified from salts using self-made reverse-phase POROS R3 (Perseptive Biosystems) microcolumns according to Gobom et al. (25). Peptides were eluted from a column directly onto the MALDI plate with ␣-cyano-4-hydroxycinnamic acid (10 mg/ml) in 60% acetonitrile, 0.3% trifluoroacetic acid. MALDI-TOF analysis was performed in reflector mode on the Voyager-DE PRO mass spectrometer (Applied Biosystems, Foster City, CA). Calibration of spectra was based on masses of trypsin autodigestion products (842.510, 1045.564, and 2211.105 Da). Proteins were identified by searching in the NCBI data base using Mascot (www.matrixscience.com), or MS-Fit in ProteinProspector (prospector.ucsf.edu). The search parameters allowed for carbamidomethylation of cysteine, one miscleavage of trypsin, and 50 ppm mass accuracy.
Liquid Chromatography-ESI MS/MS-Tandem mass spectrometry was performed on API QSTAR (Applied Biosystems) equipped with nano electrospray source (Protana, Toronto, Canada) and connected in-line with the nano-HPLC system and the autosampler (Ultimate, Switchos and Famos) (LC Packing, Amsterdam, Netherlands). Eluted and dried protein digests were dissolved in 12 l of 2% formic acid, centrifuged for 10 min at 12,000 ϫ g, and transferred into an autosampler vial. Aliquots (5 l) of samples were loaded onto a C18 PepMap, 5 m, 1 mm ϫ 300 m I.D. nano-precolumn (LC Packing), desalted for 1.5 min, and subjected to reverse-phase chromatography on a C18 PepMap, 3 m, 15 cm ϫ 75 m I.D. nanoscale LC column (LC Packing). The gradient of 5-50% acetonitrile in 0.1% formic acid was applied for 50 min with the flow rate of 0.2 l/min. The acquisition of MS/MS data was performed on-line using the fully automated IDA feature of the Analyst QS software (Applied Biosystems). The acquisition parameters were 1 s for TOF MS survey scans and 2-3 s for the product ion scans of two most intensive doubly or triply charged peptides. The major trypsin peptides were excluded from MS/MS acquisition. Analyses of MS/MS data were performed with Analyst QS software followed by protein identification by Mascot with search parameters allowing for carbamidomethylation of cysteine, one miscleavage of trypsin, oxidation of methionine, and 200 ppm mass accuracy.

Separation of NDH-1L and NDH-1M Complexes and Their
Subunits from Synechocystis 6803-Membrane protein complexes isolated from Synechocystis 6803 cells were separated by BN-PAGE in the first dimension followed by denaturing SDS-PAGE in the second direction. Sections of silver-stained gels containing NDH-1L and NDH-1M complexes are presented in Fig. 1. The identification of the major photosynthetic complexes, PSI, PSII, and ATP synthase, has been described in detail by Herranen et al. (15). Approximate molecular masses of NDH-1L and NDH-1M appeared to be ϳ490 and 350 kDa, respectively, based on the mobility of photosynthetic complexes in BN-PAGE.
In WT cells the relative amounts of NDH-1 complexes varied depending on growth conditions. The abundant NDH-1M complex was the characteristic feature of cells grown photoautotrophically under air level of CO 2 (Fig. 1A). In WT cells grown in high CO 2 (3%) the relative amount of NDH-1M drastically decreased. NDH-1L, conversely, was slightly up-regulated in high CO 2 -grown cells (Fig. 1B).
Protein patterns of NDH-1L and NDH-1M closely resembled each other. Fifteen protein spots were resolved in NDH-1L by SDS-PAGE (Fig. 1C). The subunit composition of NDH-1M was more difficult to access in WT cells since the complex overlapped with the PSI monomer and partially with the PSII monomer. The better resolution of NDH-1M subunits was obtained using the membrane preparation from the PSI-less mutant of Synechocystis 6803 (26) despite the relatively low amount of NDH-1M compared with NDH-1L (data not shown). The diffuse spot in NDH-1L moving at the level of CP43 (PsbC) of PSII corresponded to two unresolved proteins, and the subunits of NDH-1L were numbered as shown in Fig. 1C. The distinct difference between NDH-1L and NDH-1M was the absence of spots 2 and 8 in the NDH-1M complex.
To exclude the possibility that NDH-1L overlapped with an unknown complex, the membrane preparation of Synechocystis 6803 strain M55 (⌬ndhB) (18) grown at high CO 2 was analyzed under the same conditions. Both NDH-1L and NDH-1M complexes were absent in this strain (Fig. 1D), and none of the spots depicted in Fig. 1C was detected, demonstrating that all 15 proteins indeed belonged to the NDH-1L complex.
Mass Spectrometry Identification of NDH-1 Subunits-Earlier NDH-1L and NDH-1M were recognized as NDH-1 complexes since MALDI-TOF analysis demonstrated the presence of NdhH, -K, -I, and -J subunits in both of them (15). Other spots failed MALDI identification, most probably because of the scarce amounts of proteins in small subunits or because of hydrophobic properties of Ndh subunits with several transmembrane regions. Here we have used a combination of MALDI-TOF and ESI-QTOF for identification of separated proteins and results are summarized in Table I. Spots 1, 5, 6, and 7 corresponded to NdhH, -K, -I, and -J, respectively. The proteins did not possess transmembrane regions, and they were efficiently digested by trypsin providing many peptides to ensure the reliable identification by MALDI-TOF MS alone (15). ESI MS/MS analysis was nevertheless performed to exclude the possibility of the existence of comigrating unknown proteins. Spot 10 corresponded to one of the additional (compared with minimal bacterial complex) subunits Sll1262 (NdhN) recently identified by Prommeenate et al. (14) whereas spot 11 represented Slr1623 (NdhM), another additional subunit originally found by Berger et al. (12). Spots 3, 4, 9, 13, and 15 corresponded to NdhB, -A, -G, -E, and -L, respectively. For these subunits, only a few tryptic peptides were seen in MALDI-TOF spectra, most probably because transmembrane proteins are not effectively cleaved by trypsin and/or highly hydrophobic peptides are lost during sample preparation. However, results obtained by ESI MS/MS allowed the identification of these proteins. For NdhE and NdhL, assigned on the basis of the sequence of a single peptide, the complete series of y ions were obtained allowing unambiguous reading of the whole peptide sequence. Further, the corresponding peptides were also detected in MALDI-TOF spectra (data not shown). The MS/MS spectrum of the NdhL peptide is shown in Fig. 2, the peptide appeared to be modified (formaldehyde adduct of Trp residue) during the silver staining procedure. The NdhL was identified in spot 15 originating from the NDH-1M complex of cells grown under carbon-limiting conditions when the complex is most abundant. Under no growth conditions the NDH-1L complex was expressed as strongly as NDH-1M at low CO 2 , and therefore the amount of the NdhL protein from this complex was not sufficient for analysis.
There was a clear difference between the protruding spot 2/3 in NDH-1L and spot 3 in NDH-1M. Peptides of only the NdhB  predicted molecular mass of the NdhF1 (74.4 kDa) and the apparent molecular mass (20 kDa), spot 8 was N-terminally sequenced after blotting of the protein from two-dimensional BN/SDS-PAGE onto a polyvinylidene difluoride membrane. The sequence obtained (AMXVXXL-) demonstrated that the NdhF1 was cleaved between Gly 482 and Ala 483 , and spot 8 was formed by the C-terminal fragment of NdhF1. To further investigate the occurrence of NdhF1 protein in the NDH-1L complex, we made an attempt to diminish the probable unspecific proteolysis of NdhF1 by including a mixture of protease inhibitors (leupeptin, pepstatin, and EDTA) instead of pefablock in all steps from disruption of cells to the SDS-PAGE loading buffer during preparation of the thylakoid samples. We could not significantly prevent the proteolysis of NdhF1 and the C-terminal part of NdhF1 (spot 8) was still prominent. However, an additional spot of ϳ60 kDa was observed under these conditions (Fig. 3). Mass spectrometry analysis revealed the peptides typical for spot 8 in the spectrum of this protein (data not shown). Spot 14 was identified as the unknown protein Ssl1690 by both MALDI-TOF MS and ESI-MS/MS. Spot 12 failed both techniques. However, the only peak observed in MALDI-TOF spectrum (besides trypsin peaks) with m/z of 654.43 might belong to NdhC subunit (marked with brackets in the Fig. 1C). High hydrophobicity of NdhC (3 membrane-spanning regions) might explain the lack of other tryptic peptides suitable for mass spectrometry analysis.
Sequence Analysis of Ssl1690 and NdhL and Search for Homologous Proteins in Other Organisms-NdhL of Synechocystis 6803 encoded by ssr1386 contains 80 amino acids. Two transmembrane helices were found in this protein by TMHMM ver. 2.0 (27). The BLAST search performed against NCBI data base demonstrated the presence of homologous sequences in other cyanobacteria. The alignment of cyanobacterial NdhL sequences (Fig. 4) revealed 16 identical and 16 conserved amino acid residues. In order to find related proteins in green plants, the Synechocystis 6803 NdhL sequence was compared using the FASTA algorithm (28) to the MIPS Arabidopsis genome scaffold (29), the TIGR rice genome data base (30) and to assembled plant unigene sequences from the Sputnik data base (31). No unambiguous sequence matches were identified, but a number of related plant proteins did show a weak similarity over a restricted area of between 53 and 88 amino acids. The candidate match sequences from plants including the unknown Arabidopsis protein, At1g70760 and the rice protein, 9633.t02493, all showed sequence conservation (Fig. 4). The proteins from green plants are larger than cyanobacterial homologues. The chloroplast location was predicted for these Arabidopsis and rice nuclear-encoded proteins by Predotar (32). The alignment of the sequences using DiAlign (33) revealed that there is a motif conserved between the cyanobacteria and the plants. The sequences from the cyanobacteria, the protein sequences from both the Arabidopsis and rice genomes and the plant EST sequence-derived peptides were used to identify possible diagnostic domain description patterns using the Pratt software (34). The best descriptive domain identified was YX 0, 1 LX 2, 3 P(A/L/P/V) Comparison of this pattern against the SwissProt and TrEMBL databases using the Prosite tools resulted in identification of representative sequences from Synechocystis 6803, Arabidopsis, and Prochlorococcus marinus. By utilizing EST data we have characterized this domain within asterids, rosids, caryophyllids, and monocots. Thus, the pattern was present within both cyanobacterial and plant sequences, and no evidence for this domain was found within other organisms.
The novel Ndh subunit, Ssl1690, consists of 72 amino acids. Topology prediction recognized Ssl1690 as a hydrophilic protein since no transmembrane helices were found in the sequence by TMHMM (27). The BLAST search against NCBI data base showed that homological sequences with reliable E values (e-04) could be found only among cyanobacteria. The first non-cyanobacterial hit was detected for the At1g74880 Arabidopsis protein of 158 amino acids. According to Predotar (32), At1g-74880 is located in chloroplasts. Further, proteins homologous to At1g74880 could be found in other plant species. The alignment of the cyanobacterial Ssl1690 homologues with Arabidopsis and rice is shown in Fig. 5. Among these sequences and twelve additional ones from Solanaceae, Asteraceae, and monocots the following conserved motif was found: KKG

NDH-1L and NDH-1M Complexes Are Both Present in M9 (⌬NdhL) but Absent in M55 (⌬NdhB) Synechocystis 6803 Mutants-
The occurrence of NDH-1L and NDH-1M complexes was compared in two Synechocystis 6803 mutants, M9 (⌬NdhL) (17) and M55 (⌬NdhB) (18), which show a very similar phenotype. Both strains grew photoautotrophically in high CO 2 conditions and also in an air level of CO 2 at elevated pH, and the efficiency of their growth was comparable to that of WT (Table II). However, contrary to the WT, the growth of M9 was greatly diminished at pH 7.5, air level of CO 2 , and M55 died under the same conditions. Moreover, both M55 and M9 were unable to grow photoheterotrophically in the presence of glucose plus DCMU (Table II).
Despite the similarity in phenotype, the two-dimensional BN/SDS-PAGE demonstrated the striking difference in the occurrence of NDH-1 complexes in these two mutants. In M9 both NDH-1L and NDH-1M complexes were present in amounts comparable to those in WT (Fig. 1E) in contrast to M55 where both NDH-1 complexes were absent (Fig. 1D). For M9, the complete segregation of the NdhL mutation was confirmed (data not shown).

DISCUSSION
The NDH-1 complex of cyanobacteria and the plastidial NDH-1 complex of green plants function not only as a respiratory/chlororespiratory complex like in a majority of other organisms but also participate in cyclic electron flow around photosystem I (PSI) (35)(36)(37). Based on genome studies, it was predicted that eleven subunits are similar between NDH-1 of E. coli and the enzymes located in thylakoid membranes of photosynthetic organisms. Since the complexity of the enzyme varies between organisms, it was plausible to presume that additional subunits exist in NDH-1 of cyanobacteria and chloroplasts compared with minimal NDH-1 of E. coli.
Traditional chromatographic methods used for purification of NDH-1 from Synechocystis 6803 have been unsuccessful since the enzyme containing both hydrophilic and hydrophobic components appeared to be extremely fragile and present in thylakoid membranes in a low quantity. Moreover, it has been difficult to discriminate between the new subunits of the complex and contaminating proteins (12). Only recently, Prommeenate et al. (14) isolated the NDH-1 complex of a high purity using His-tagged NdhJ. They confirmed that Slr1623 (NdhM), discovered by Berger et al. (12), indeed belongs to the Synechocystis NDH-1 complex, and revealed a novel subunit, Sll1262 (NdhN). However, NdhL has never been found before as a component of isolated NDH-1 complexes.
In our previous proteome studies (15) we described four complexes containing ndh gene products. Among them NDH-1L and NDH-1M showed a multiple Ndh subunit composition, but the majority of protein subunits remain unknown. These two complexes seem to have distinct functions in cyanobacterial cells. 1 To clarify the structural difference, we took a systematic approach to identify the subunits comprising NDH-1L and NDH-1M using ESI MS/MS analysis supplemented by MALDI-TOF mass spectrometry.
Fifteen Subunits of the Synechocystis NDH-1 Complex-In the NDH-1L complex we identified 10 of 11 subunits of Synechocystis NDH-1 (NdhA, -B, -D1, -E, -F1, -H, -I, -J, and -K) that are homologous to the E. coli NDH-1 complex. The 11th subunit, NdhC, was most probably represented by the unidentified spot 12 since the relative mobility of the protein corresponded well to the NdhC subunit identified by Prommeenate et al. (14) by N-terminal sequencing. NDH-1M contained the same subunits as NDH-1L except for NdhD1 and NdhF1.
Intriguingly, we found four additional subunits of Synechocystis NDH-1 that do not have homologues in the bacterial "minimal" complex. The unknown protein, Ssl1690, appeared to be a novel component of NDH-1. We also found NdhL (Ssr1386), thus confirming that this subunit indeed belongs to NDH-1 as was proposed by Ogawa (11). NdhM (Slr1623) and NdhN (Sll1262) were described earlier (12,14). These four additional subunits were found in both NDH-1L and NDH-1M complexes. Since the NDH-1L/NDH-1M ratio varies depending on the strain and/or growth conditions (Fig. 1, A and B), the fact that relative intensities of these additional subunits in NDH-1L and NDH-1M always followed the relative intensities of other subunits in the same complex is the further indication that the proteins belong to NDH-1 of Synechocystis 6803. Moreover, all 15 proteins disappeared from the gel when the M55 strain deficient in ndhB was analyzed (Fig. 1D). Following the subunit nomenclature of cyanobacterial NDH-1, we suggest annotating the previously unknown Ssl1690 protein as NdhO.
NdhF1 was detected in NDH-1L as a short protein comprising the C-terminal part of Slr0844. N-terminal sequencing demonstrated that the cleavage occurred between Gly (482) and Ala (483). The same N-terminal sequence of NdhF1 was obtained for the protein from the NDH-1 complex isolated by Prommeenate et al. (14). The observed cleavage of NdhF1 appeared to be very specific. It is important to note that NdhF3 was not cleaved (15). Comparison of NdhF1, NdhF3, NdhF4, and NuoL from E. coli showed that the sequence around the cleavage site is specific for NdhF1. However, this specific region is not conserved among NdhF1 proteins from other cyanobacteria (data not shown). Inclusion of additional inhibitors allowed diminishing the cleavage to a degree that an additional spot of ϳ60 kDa appeared in the NDH-1L complex (Fig. 3). The presence of the peptides typical for spot 8 in the spectrum of this protein suggested that it might represent the intact NdhF1 subunit. However, the cleaved C-terminal part of NdhF1 was still clearly present in the pattern of NDH-1L.
In the NDH-1L complex, we could see a faint diffuse spot migrating at the level of AtpC (marked by an asterisk in Fig. 1,  A, B, and E). However, attempts to identify this spot have been unsuccessful. The corresponding spot was not detected in NDH-1M, therefore it is possible to speculate that it represented the N-terminal part of NdhF1. Low amounts of the protein and 12 transmembrane helices might explain the absence of peptide peaks suitable for mass spectrometry analysis. The presence in the NDH-1L complex of the intact NdhF1 (ϳ75 kDa) or both C-and N-terminal parts of the protein together with NdhD1 (ϳ60 kDa) corroborates well with the mass difference between NDH-1L and NDH-1M of ϳ140 kDa. It seems highly probable that the cleavage of NdhF1 was a result of proteolysis during sample preparation; however, since the cleavage is specific we cannot exclude the possibility that this processing is functionally or structurally significant.
The NDH-1L complex resembles the complex A isolated by Prommeenate et al. (14) using His-tagged NdhJ. However, the identity of NDH-1L with the complex A remains obscure. It was suggested (14) that the latter contains two hydrophilic subcomplexes comprised of NdhH, -I, -J, and -K. We do not have any reason to make the same conclusion about the NDH-1L complex. The duplication of the hydrophilic subcomplex described by Prommeenate et al. (14) could be caused by the His tag modification of the NdhJ subunit. Further, the ambiguity in small subunits should be noted.
It is important to emphasize that NDH-1M and the complex B described by Prommeenate et al. (14) have different subunit compositions. Complex B consisted of the same subunits as complex A, but contained only one hydrophilic subcomplex. The NDH-1M complex does not include NdhD1 and NdhF1 subunits. Further, we did not detect variation in intensities of hydrophilic subunits between NDH-1L and NDH-1M. In addition, we never observed in our proteome map of thylakoid membrane of Synechocystis 6803 (15) the smaller complexes C and D described by Prommeenate et al. (14).
NdhL Is an Important Functional Component of NDH-1L and NDH-1M Complexes-To date there is no three-dimensional structural information available for complex I of any origin. However, in electron microscopy studies an L-shaped overall structure of NDH-1 was observed (38) with a membrane fragment and a perpendicular peripheral fragment protruding into bacterial cytoplasm.
On the basis of hydrophilic properties of the protein it is possible to speculate that the novel Ndh subunit Ssl1690 (NdhO) belongs to the peripheral arm of the enzyme together with NdhH, -I, -J, -K, -M, and -N. Moreover, Ssl1690 was detected in the proteome of peripheral proteins isolated from thylakoid membranes of Synechocystis 6803 (39). It is intriguing that the protein was detected at two spots with different  apparent pI values (4.8 and 7.9) indicating the presence of a post-translational modification. In contrast to NdhO, the NdhL subunit containing 2 transmembrane helices most probably belongs to the membrane fragment of the enzyme. The hypothetical scheme of cyanobacterial NDH-1 is shown in Fig. 6. It should be noted that the three catalytically active subunits homologous to NuoE, -F, and -G and comprising the NADH dehydrogenase fragment of E. coli NDH-1 still remain unknown for cyanobacterial or chloroplast enzyme. Therefore, it is still uncertain what is the electron donor for cyanobacterial NDH-1. Alternatively to NADH, NADPH (13,35) or ferredoxin (1) might be the electron donor. The situation is similar for the plastidial NDH-1 complex of green plants (36,40). At present it is not clear what roles NdhM, -N, -O, and -L, the "extra" components of cyanobacterial NDH-1, play in the complex. No known sequence motifs have been found in sequences of these subunits that could provide an insight into their function.
Functional studies of M9, the NdhL deletion mutant, demonstrated that the presence of this protein is essential to inorganic carbon acquisition and photoheterotrophic growth of Synechocystis 6803 (17,41). The phenotype of M9 is similar to that of M55, the NdhB deletion mutant (18). Both strains grow photoautotrophically in high CO 2 conditions and do not grow at pH 7.5, air level of CO 2 . Moreover, in both strains the cyclic electron flow around PSI was suppressed (41). The effect of the NdhL mutation was, however, slightly less profound compared with that of NdhB. In low CO 2 conditions, M9 cells did not grow but remained alive whereas M55 cells died. Further, cyclic electron flow was almost completely suppressed in M55 whereas in M9 a rather low cyclic PSI activity could be observed (41). Our results demonstrated that NdhL is not essential to the in vivo integrity of the NDH-1 complex because both NDH-1L and NDH-1M were present in M9. In contrast, NdhB is crucial for complex formation. Because in the absence of NdhL all physiological functions so far demonstrated for Synechocystis NDH-1 were severely impaired, it can be concluded that NdhL is a critical subunit for the function but not for the assembly of NDH-1 complexes in Synechocystis 6803. Therefore, it is likely to be located to the periphery of the membrane fragment of NDH-1 yet in a close proximity to the soluble subcomplex (Fig. 6). The exact role of this small subunit for the activity of NDH-1 remains to be elucidated.

Homologues of Ssl1690 and NdhL Exist in Higher Plants-
The results of BLAST searches demonstrated the presence of Ssl1690 and NdhL homologues in cyanobacteria, but the presence of homologous subunits in higher plants remained ambiguous. This was an important question particularly for NdhL because this protein was shown to be involved in Ci transport (11). In order to distinguish between two hypotheses: (a) NdhL and Ssl1690 are specific for cyanobacterial species, and (b) these subunits have relatives in higher plants despite the lack of significant similarity, we performed a profound bioinformatic analysis of several plant databases. The results showed that for both subunits homologous unknown proteins are present in various plant species. For example, At1g70760 from Arabidopsis and 9633.t02493 from rice are homologous to NdhL, and At1g74880 and 9629.t06877 are homologous to Ssl1690. Moreover, corresponding diagnostic sequence motifs were found for both NdhL and Ssl1690 homologues that are conserved among cyanobacteria and plants but absent in other organisms.
Comparison of cyanobacterial sequences with green plant counterparts showed that proteins considerably differ in size. Plant proteins have N-terminal extensions, and At1g70760 extended also at C termini (the C terminus of the rice protein is not certain because of the splicing ambiguity). A significant part of N-terminal extensions should be cleaved in mature proteins during translocation into chloroplasts, thus increasing similarity with cyanobacterial sequences. It is consistent that At1g74880 and 9629.t06877 are hydrophilic proteins, likewise Ssl1690, while At1g70760 was predicted to have three transmembrane helices (Fig. 4). The second and third helices match the corresponding hydrophobic regions of NdhL. It seems probable that during evolution some additional feature(s) was/were accumulated in NdhL-homologous sequences of green plants in parallel with functionally important preservation of the cyanobacteria-like domain.
Regarding green plants, it is necessary to consider two complexes, the mitochondrial complex I and the NDH-1 complex of chloroplasts. Heazlewood et al. (16) studied the subunit composition of mitochondrial complexes I from Arabidopsis and rice and described several subunits specific for the complex. Importantly, proteins homologous to Ssl1690 and NdhL were not found among them. This is consistent with Predotar results for the plastidial location of these proteins. We conclude that in green plants unknown proteins homologous to Ssl1690 and NdhL might be the components of the poorly studied NDH-1 complex located in chloroplasts. Together with NdhM and NdhN, it makes four nuclear-encoded Ndh subunits discovered in plastidial NDH-1. NdhM has been reported to be homologous to the B13 subunit of the bovine complex I (12,14). The other three proteins seem to be specific for thylakoid-located NDH-1 of photosynthetic organisms.