The Chloroplast NAD(P)H Dehydrogenase Complex Interacts with Photosystem I in Arabidopsis*

The chloroplast NAD(P)H dehydrogenase (NDH) complex is involved in photosystem I (PSI) cyclic and chlororespiratory electron transport in higher plants. Although biochemical and genetic evidence for its subunit composition has accumulated, it is not enough to explain the complexes putative activity of NAD(P)H-dependent plastoquinone reduction. We analyzed the NDH complex by using blue native PAGE and found that it interacts with PSI to form a novel supercomplex. Mutants lacking NdhL and NdhM accumulated a pigment-protein complex with a slightly lower molecular mass than that of the NDH-PSI supercomplex; this may be an intermediate supercomplex including PSI. This intermediate is unstable in mutants lacking NdhB, NdhD, or NdhF, implying that it includes some NDH subunits. Analysis of thylakoid membrane complexes using sucrose density gradient centrifugation supported the presence of the NDH-PSI supercomplex in vivo. Although the NDH complex exists as a monomer in etioplasts, it interacts with PSI to form a supercomplex within 48 h during chloroplast development.

The chloroplast NAD(P)H dehydrogenase (NDH) 3 complex is localized to the stroma thylakoids (1) and is involved in chlororespiration and photosystem I (PSI) cyclic electron transport in higher plants (2,3). The NDH complex is essential for preventing over-reduction of the stroma in the pgr5 (proton gradient regulation 5) mutant, which is defective in the main pathway of PSI cyclic electron transport (4). This function suggests that NDH alleviates oxidative stress under certain conditions even in wild-type (WT) plants (3,5). So far, 11 plastid-encoded and 4 nuclear-encoded subunits of chloroplast NDH have been identified, and all subunits are conserved in cyanobacteria. It is generally accepted that chloroplast NDH is more closely related to the cyanobacterial NDH-1 complex than to the mitochondrial complex I in the same species (3,6). Although the nuclearencoded subunits are specific to phototrophs containing NDH, plastid-encoded subunits (NdhA-K) are also conserved in bacterial and mitochondrial complex I (7)(8)(9). Although the threedimensional structure of the NDH complex in any organism has not been established, it is discussed by analogy to bacterial and mitochondrial complex I (3). Electron microscopic analysis has indicated that the NDH-1 complex of the cyanobacterium Thermosynechococcus elongatus has an L-shaped structure that is similar to the structure of bacterial and mitochondrial complex I (10). However, homologues of the other three bacterial subunits that function in NADH oxidation are not found in plant and cyanobacterial genomes (3,8,9). As well as four nuclear-encoded subunits (NdhL-O) (11,12), the thylakoid lumen protein PPL2 (PsbP-like protein 2) and other three novel protein, NDF1, NDF2, and NDF4, were recently shown to associate with the NDH complex (13,14). NDF4 is a ferredoxinrelated protein that may be involved in electron transfer (14). Several candidates for the unidentified subunits have been characterized by genetic analysis in Arabidopsis (15)(16)(17)(18). However, these proteins do not have any motifs suggesting electron donor binding or NAD(P)H oxidation. The electron donor binding module of the NDH complex in chloroplasts and cyanobacteria is still unclear.
Modification of its subunit composition allows the cyanobacterial NDH-1 complex to be involved in multiple functions: respiration, PSI cyclic electron transport, and CO 2 uptake (8,9). Consistent with functional genomic analysis, proteome analyses revealed three distinct NDH-1 complexes in cyanobacteria. According to their molecular masses, these complexes are designated NDH-1L (large), NDH-1M (medium), and NDH-1S (small), with molecular masses of ϳ460, 350, and 200 kDa, respectively (19 -23). The NDH-1M complex consists of a membrane-embedded arm (NdhA-C, E, G, L) and a hydrophilic connecting domain (NdhH-K, M-O). It is believed to be a basic subcomplex present in all cyanobacterial NDH-1 complexes (20 -22). The NDH-1L complex contains NdhD1 and NdhF1 in addition to the components of the NDH-1M complex. It is involved in respiration and cyclic electron flow around PSI. NDH-1S associates with NDH-1M to form the NDH-1MS complex, of about 490 kDa (23). Expression of NDH-1MS can be induced under low CO 2 conditions; it is considered to be involved in CO 2 uptake (21,23). The NDH-1S complex consists of four subunits: NdhD3, NdhF3, CupA, and CupS. Alternatively, NdhD4, NdhF4, and CupB form another NDH-1S complex, which is also associated with NDH-1M (8,20).
Six ndhD and three ndhF genes are present in the genome of Synechocystis sp. PCC 6803, whereas the chloroplast genome contains only a single copy of each, which correspond, respectively, to ndhD1/D2 and ndhF1 of cyanobacteria (3,8,9). This fact suggests that the chloroplast NDH complex is similar to cyanobacterial NDH-1L that functions in respiratory and also probably PSI cyclic electron transport (2,3,8,9). Chromatography was used to show that the NDH complex purified from pea chloroplasts has a molecular mass of about 550 kDa (23). An NDH complex with a similar molecular mass was detected by native PAGE in thylakoids isolated from tobacco (25), oat (26), and barley etioplasts (27). More recently, monomeric (ϳ550 kDa) and dimeric (1000 -1100 kDa) NDH complexes were detected in Zea mays mesophyll and bundle sheath chloroplasts by blue native (BN)-PAGE combined with mass spectrometry (28). Larger NDH complexes have been reported also in Arabidopsis (13,29). During isolation and analysis, the NDH monomer usually splits into 300-and 250-kDa subcomplexes because of its fragile nature (25,29). It seems that a low concentration of detergent is essential for detecting the intact NDH complex that is likely to be present in vivo (12).
Here we separated the Arabidopsis NDH complex by BN-PAGE and sucrose density gradient (SDG) after mild solubilization of thylakoids using a low concentration of n-dodecyl-␤-D-maltoside (DM). Inconsistent with the results from some previous studies, the NDH complex was detected only as a single form with a molecular mass of Ͼ1000 kDa. Further analysis using NDH-deficient mutants indicated that this large protein complex is composed of NDH and a large pigment-protein complex including PSI. We also investigated the formation of this supercomplex during the development of chloroplasts from etioplasts.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions-Arabidopsis thaliana (ecotype Columbia gl1 and Wassilewskija) and tobacco plants (Nicotiana tabacum cv. Xanthi) used in this work were grown in soil under growth chamber conditions (50 mol photons m Ϫ2 s Ϫ1 , 16-h photoperiod, 23°C) for 3-4 weeks. An Arabidopsis mutant defective in the ndhM gene was obtained from the Salk T-DNA collection. The Arabidopsis sig4 -10 mutant was kindly provided by Dr. Silva Lerbs-Mache. For light-induced greening studies, seeds were sterilized and sown on agarsolidified MS medium containing 3% sucrose (30). To ensure synchronized germination, the plates were kept in darkness at 4°C for 2 days and then kept in the chamber with continuous light for the first 24 h. After growth in the dark for 1 week, seedlings were illuminated for 0, 24, or 48 h with continuous white light and then harvested.
Thylakoid Membrane Preparation, BN-PAGE, and Immunoblot Analysis-Chloroplasts were isolated as described (31) and osmotically ruptured in buffer containing 20 mM HEPES/KOH (pH 7.6), 5 mM MgCl 2 , and 2.5 mM EDTA. Thylakoid membranes were pelleted by centrifugation (7700 ϫ g for 3 min) and resuspended in the same buffer.
Total leaf membranes were isolated from etiolated and greening seedlings as described (31). The leaves were ground with a conical plastic rod in a 1.5-ml tube in 500 l of isolation buffer containing 20 mM HEPES/KOH (pH 7.6), 5 mM MgCl 2 , and 2.5 mM EDTA. After centrifugation at 500 ϫ g for 2 min, membranes were pelleted by centrifugation at 17,000 ϫ g for 10 min and then resuspended in the same isolation buffer. The protein contents were determined according to Lowry (33).
SDG Analysis of Thylakoid Protein Complexes-Freshly isolated thylakoid membranes were washed twice with buffer containing 5 mM MgCl 2 , 10 mM NaCl, 0.1 M sucrose, and 25 mM MES/NaOH (pH 6.8) and then solubilized in the same buffer containing 1% DM at a chlorophyll concentration of 1 mg ml Ϫ1 for 10 min on ice. After centrifugation at 17,000 ϫ g for 10 min at 4°C, the cleared supernatant (1 ml) was loaded on the top of a 15-ml linear sucrose gradient (5-40%) prepared with buffer containing 5 mM MgCl 2 , 10 mM NaCl, 0.06% DM, and 25 mM MES/NaOH (pH 6.8). Thylakoid protein complexes were separated by ultracentrifuge for 24 h at 141,000 ϫ g at 4°C in an SW28 rotor (Beckman). The gradients were then fractionated from top to bottom into 48 fractions by an automatic density gradient fraction collector (Advantec model CHD255AA), and equal amounts of alternate fractions were used for further immunoblot analysis.

Separation of NDH Complex by BN-PAGE-For character-
ization of the structure of the NDH complex, we solubilized freshly isolated thylakoids from WT Arabidopsis plants in 1% DM and separated protein complexes by BN-PAGE. After the first-dimensional separation in the presence of Coomassie Brilliant Blue G 250 dye, several major bands were clearly separated, including PSII supercomplexes, PSI ϩ PSII dimer, PSII monomer, CP43-less PSII, and light-harvesting complex II (LHCII) trimer (Fig. 1A). Notably, one higher molecular mass green band (Band I) also appeared at the top of the gel (Fig. 1A).
To assign the position of the NDH complex in the BN-PAGE, we performed second-dimension SDS-PAGE and subsequent immunoblot analysis using specific antibodies against subunits NdhH and NdhL. Both were present at the same position in the BN gel, at a molecular mass of Ͼ1000 kDa (Fig. 1B). Similar results were reported from Arabidopsis (13,29) and maize (28). These results indicate that the high molecular mass forms of the NDH complex may be universally present in plant chloroplasts. The positions of the major forms of PSI and PSII complexes in the BN gel were also analyzed using antibodies against their subunits PsaA and CP47, respectively. In addition to the major spot corresponding to the PSI monomer, a trace amount of PSI occurred in a putative supercomplex at the high molecular mass position, co-localized with NDH subunits in the BN gel.
NDH Complex Interacts with PSI-Recently, we characterized an Arabidopsis crr23 (chlororespiratory reduction 23)/ndhl mutant defective in the ndhL gene and showed that the NDH complex almost completely disappeared (12). BN-PAGE found no major differences in the major complex bands between WT and ndhl (Fig. 2, A and C). However, the higher molecular mass green band (Band I of WT) shifted to a position with a slightly lower molecular mass (Band II) in the mutant; this difference was clearer after staining with Coomassie Brilliant Blue (Fig. 2,  A and C). The most straightforward interpretation is that the complex corresponding to Band II interacts with the NDH complex to form a supercomplex (Band I) in WT plants.
To further investigate the components of Bands I and II, we excised them from the BN gels and immunoblotted them (Fig.  2, A and B). Band III, corresponding to the PSII supercomplex, and Band IV, corresponding to the PSI monomer and the PSII dimer excised from the WT lane, were also included in the analysis. Western blot results clearly show that Band I includes subunits NdhH and NdhL, which were not detected in the other bands and the ndhl mutant (Fig. 2B). We also performed immunoblot assays using antibodies against the subunits of PSI. Both PsaA and PsaD were detected in both Bands I and II. A subunit of LHCI, Lhca3, was also present in both bands, but no D1 or cytochrome f was detected in either band (Fig. 2B). These results indicate that NDH and PSI co-localize in the BN gel, forming Band I, which does not contain PSII or the cytochrome b 6 f complex. On the other hand, Band II does not contain NdhH and NdhL and only PSI subunits could be detected in Band II, suggesting this may be an intermediate supercomplex including PSI. The intermediate is stable even in the absence of NdhH and NdhL and was specifically detected in ndhl, suggesting that it forms a supercomplex with NDH in WT. All results imply that NDH and PSI form a supercomplex, present as Band I in the BN gel.
A previous report (1) indicated that NDH and PSI reside on the stroma lamellae. To remove the dominant PSII supercom- plexes and allow better incorporation of the large NDH-PSI supercomplex into the gel, we first separated the stroma and grana lamellae and then analyzed the supercomplex by BN-PAGE. Our results (supplemental Fig. S1) showed that the NDH-PSI supercomplex (Band I) and intermediate (Band II) were specifically detected in stroma lamellae of the WT and ndhl, respectively. The PSII and LHCII complexes are mainly detected in the grana lamellae fraction (supplemental Fig. S1).

The reproducibility of the gel with different isolation methods suggests that the NDH-PSI supercomplex and the intermediate stably exist in vivo.
Like ndhl, the ndhm mutant also lacks the NDH complex because of the absence of the NdhM subunit (11). We used ndhm and two additional mutants, crr6-1 and crr7-1, in which the accumulation of the NDH complex was also impaired (15,16), in further analysis to test the possibility that the NDH complex interacts with PSI. Consistent with the result from the ndhl mutant, Band I again shifted to a position with a slightly lower molecular mass (Band II) in ndhm (Fig. 2C). However, both bands coexisted in the crr6-1 and crr7-1 mutants. This is probably due to a leaky defect in accumulating the NDH complex in both mutants (15,16). As expected, only Band I was detected in the mutant lines complemented by the introduction of the WT genomic CRR6 or CRR7 genes (Fig. 2C, crr6c and crr7c).

Takahashi et al. (34) discovered a large PSI-LHCI-LHCII supercomplex under state II conditions in Chlamydomonas.
To test whether the Band I detected in our results is also dependent on state transition or not, the Arabidopsis stn7 mutant fixed in state I was analyzed by BN-PAGE. Our results showed that only Band I could be detected in stn7 (date not shown), suggesting the formation of Band I is independent of the state transition in Arabidopsis.
SDG Analysis of the Thylakoid Protein Complex-To further test the possibility of the interaction of NDH with PSI, we performed SDG analysis, which facilitates the separation of protein complexes under milder conditions. After centrifugation, 48 equal fractions were divided up and alternate fractions were immunoblotted using specific antibodies. Fig. 3 shows that NdhL and NdhH were fractionated into two major distinct peaks in the gradient. The first peak was present in fractions 25 to 29, and the PSI monomer complex co-migrated in this region. From the migration of the PSI complex, the molecular mass of the NDH complex in this region is about 550 kDa. This complex is widely accepted as the NDH complex monomer (24 -28). The second peak was present in fractions 33 to 39 at a higher molecular mass and may correspond to Band I in the BN gel (Figs. 2 and 3). Besides these two dominant peaks, another small peak appeared in the low molecular mass region (fractions [13][14][15] and is likely to correspond to the NDH subcomplex. Previous research using BN-PAGE and immunoblot analysis detected an NDH subcomplex of about 250 kDa in tobacco (24) and maize (28).
The region corresponding to the second peak of NDH (fractions 33-39) also contains PsaA and PsaD (Fig. 3), suggesting that the fraction contains the supercomplex of NDH and PSI. The low resolution of SDG makes it difficult to conclude that PSI in this region forms a distinct peak from the major PSI monomer peak (fractions [25][26][27][28][29]. To resolve this problem, we also investigated PSI complexes in ndhl by SDG analysis. The level of the predominant peak was identical between ndhl and WT thylakoids. In contrast, the levels of PSI subunits were significantly lower in fractions 33-39 of ndhl than in WT (Fig. 3). No significant differences were detected in the localization of D1 between WT and ndhl (Fig. 3). Consistent with the results of the BN gel ( Figs. 1 and 2), SDG analysis supports the interaction of the NDH complex with PSI to form a supercomplex. However, we could not detect an intermediate supercomplex corresponding to Band II (Fig. 2), probably because of the low resolution of SDG.
Some Membrane-embedded NDH Subunits Are Essential for Stabilizing the Band II Complex-To study the nature of the supercomplex further we analyzed the tobacco ⌬ndhB line, in which the chloroplast ndhB gene was insertionally disrupted (35). Unexpectedly, both Bands I and II were missing in ⌬ndhB (Fig. 4A), which is in contrast to the fact that Band II is stable in ndhl, ndhm, crr6-1, and crr7-1 (Fig. 2C). Two-dimensional  SDS-PAGE also confirmed that the high molecular complex containing PSI corresponding to Band II was missing in ⌬ndhB (Fig. 4B).
Although the lack of NdhM or NdhL subunits only destabilizes the complex corresponding to Band I in Arabidopsis (Fig.  2), both Bands I and II were missing in ⌬ndhB in tobacco (Fig. 4,  A and B). To study the possibility that the discrepancy is due to the difference in plant materials, we also analyzed Arabidopsis crr2-2 defective in the ndhB expression (36), crr4-3 defective in the ndhD expression (37), and the sig4 -10 mutant defective in the ndhF expression (38). Consistent with the results using ⌬ndhB (Fig. 4, A and B), both Bands I and II were missing in these lines (Fig. 4C). The trace level of band I detected in crr2-2 is probably ascribed to the leaky accumulation of the NDH complex in this mutant (36) and the same story is probably true to sig4 -10. We conclude that the membrane-embedded subunits of NdhB, NdhD, and NdhF are essential for accumulating both Band I and II complexes. In contrast NdhL and a soluble subunit of NdhM are required for stabilizing the Band I complex, and Band II complex is stable in the absence of these subunits. Most probably, Band I contains all the subunits of NDH and PSI, whereas Band II contains PSI and some of NDH subunits including NdhB, NdhD, and NdhF. Indeed, proteomics analysis of Band II showed that it includes PSI subunits and some NDH subunits including NdhB, NdhD, and NdhF. 4 Formation of the Supercomplex during Chloroplast Development-Previous reports showed that the NDH complex is already present in etioplasts (27,39). However, the assembly of PSI is light-dependent, and the PSI-LHCI complex is stably assembled only in the chloroplast, which develops at least 24 h after exposure to light (39). Focusing on this phenomenon, we investigated the time course of the supercomplex formation during chloroplast development (Fig. 5). Total membranes of etiolated and greening leaves were solubilized in DM, and protein complexes were separated by BN-PAGE. The BNgel strips were subjected to two-dimensional SDS-PAGE, and immunoblot analysis was performed using antibodies against NdhL (Fig. 5A) and PsaA (Fig. 5B). The NDH complex was present as a 550-kDa complex in etiolated seedlings and comigrated with the PSI monomer, which appeared in greened leaves with 24-or 48-h illumination (Fig. 5). Accumulating information (24 -28) suggests that this complex should be NDH monomer. After 24 h illumination, NDH complex occurred mainly as a monomer, and a small quantity was shifted to the high molecular weight side in the BN gel. Because the spot was also detected by an antibody against PsaA, it corresponds to the NDH-PSI supercomplex (Fig. 5). The NDH-PSI supercomplex was fully assembled after 48 h of illumination during chloroplast development. These results strongly suggest that the NDH complex is present as the monomer in etioplasts and interacts with PSI in chloroplasts. Taking together all these results, we conclude that the NDH complex interacts with PSI in chloroplasts.

DISCUSSION
In chloroplasts of higher plants, several multisubunit protein complexes are embedded in thylakoids and are involved in electron transport through the membrane. Recent studies indicate that these complexes are usually organized into large supercomplexes (40) such as the Cyt b 6 f complex, which forms a dimer of about 220 kDa in vivo (41). In Arabidopsis, four PSII-LHCII supercomplexes could be clearly separated by BN-PAGE ( Fig. 1), consistent with previous proteomic (42) and electron microscopic analyses (43). These reports and our results show that the largest and most abundant PSII supercomplex has a molecular mass of about 1300 kDa, which is slightly smaller than that of the complex corresponding to Band II in our BN gel (Fig. 1). Previous studies also detected the supercomplex including PSI at the top of the BN gel, probably corresponding to Band I, with a larger molecular mass than that of the PSII supercomplexes (29). Furthermore, PSI and NDH subunits (NdhK and NdhH) could be observed in this complex in the silver-stained two-dimensional gel (29). This result and our immunoblot results (Fig. 2B) suggest that Band I corresponds to the supercomplex between PSI and NDH. Band II was specifically detected in the mutants lacking the NDH subunits, ndhl, ndhm, crr6-1, and crr7-1, and it most probably corresponds to the intermediate for the supercomplex. It is also possible that Band II represents a trimeric form of the PSI-LHCI supercomplex with a theoretical molecular mass of 1600 kDa, and Band I corresponds to its further association with the NDH complex. The PSI trimer occurs in cyanobacteria (44), and the PsaL subunit facilitates its formation (45). Although proteomics and single-particle electron microscopy showed the scantness of PSI trimer in higher plants (42,46), this is explained by interference  by the additional PsaH subunit close to PsaL (47). It is still likely, however, that Band II corresponds to the PSI trimer, because it was detected specifically in mutants lacking the NDH complex (Fig. 2).
Unexpectedly, however, Band II was unstable in the mutants lacking NdhB, NdhD, and NdhF. We consider that Band II is an intermediate for the supercomplex including PSI subunits and also at least NdhB, NdhD, and NdhF. It is possible that other membrane-embedded subunits like NdhA, NdhE, and NdhG are also included in Band II. Although NdhL also contains transmembrane domains, it is not essential for stabilizing Band II and thus unlikely to be included in Band II. The idea is consistent with the phenotype of the cyanobacterial ndhl mutant defective in NDH activity but not in the accumulation of the complex (22). It is likely that the position of NdhL is different in the complex from that of the other membrane-embedded subunits. Although we are not sure of the function of CRR6 and CRR7 (15,16), they are likely to be required for stabilizing the soluble subcomplex of NDH, because Band II was detected in the mutant backgrounds.
Burrows et al. (25) showed a stoichiometry of approximately one NDH complex per 50 -100 PSII complexes in tobacco. If we assume the PSI/PSII ratio of about 1:1 in normal conditions, then NDH is present at ϳ1-2% of the PSI on a molar basis. Our BN-PAGE and two-dimensional/immunoblot showed that PsaA in the supercomplex, estimated from the immunoblot signal, was 0.5-5% of the total PsaA in Arabidopsis or tobacco (Figs. 1, 4, and 5). Although it is difficult to make any exact conclusions about stoichiometry, the level of high molecular mass PSI complex is comparable with that of the NDH complex.
BN-PAGE analysis also revealed only the NDH monomer in etioplasts, and that the accumulation of the NDH complex is independent of light (Fig. 5). Consistent with our result, a 580-kDa NDH complex with NADH dehydrogenase activity from etioplast membranes of barley was reported (27). The NdhI subunit of the NDH complex could be detected in pea etioplasts by immunoblot analysis, yet it was not found in silver-stained two-dimensional BN/SDS-PAGE gels (39). These results indicate that the NDH complex is not a dominant protein complex in etioplasts, just as it is not dominant in chloroplasts (25). Rumeau et al. (5) concluded that the NDH complex may act as a proton pump in etioplasts to energize the plastid membrane and/or to favor the biogenesis of photosynthetic complexes. Even in the absence of the NDH complex, however, greening is not affected, a fact that does not support this idea (5). According to our finding (Fig. 5), it is necessary to reconsider the function of NDH in etioplasts in which its PSI partner is not yet present. The NDH complex may switch its function from chlororespiration to PSI cyclic electron transport during chloroplast development. It is also possible that monomeric NDH does not have activity in vivo in etioplasts and is waiting for the assembly of PSI to form an active supercomplex.
Recently, Ma et al. (48) reported a supercomplex of about 1000 kDa with high NADPH dehydrogenase activity in Synechocystis sp. PCC 6803. This supercomplex, composed of several NDH subunits, is absent in the ⌬ndhD1/⌬ndhD2 double mutant and the M55 mutant defective in ndhB. The authors concluded that it represented an NDH-1L dimer. High similarity in structure and function between the cyanobacterial and chloroplast NDH complexes suggests that NDH-1L interacts with PSI also in cyanobacteria. The most straightforward demonstration of the interaction between NDH and PSI is co-immunoprecipitation using antibodies against their subunits. Although our trial was unsuccessful, possibly because of a low abundance of the complex in chloroplasts, the strategy may be possible using cyanobacteria, which accumulate more NDH in cells than chloroplasts.