In Planta Mutagenesis of Src Homology 3 Domain-like Fold of NdhS, a Ferredoxin-binding Subunit of the Chloroplast NADH Dehydrogenase-like Complex in Arabidopsis

Background: NdhS is a ferredoxin-binding subunit of chloroplast NADH dehydrogenase-like (NDH) complex. Results: Mutagenesis of the Src homology 3-like domain of NdhS revealed that its positively charged pocket is required for high affinity binding of ferredoxin. Conclusion: The positive charge of Arg-193 plays a critical role in electrostatic interaction with ferredoxin. Significance: Our results provide insights into the evolution of the electron input module of chloroplast NDH. Chloroplast NADH dehydrogenase-like (NDH) complex mediates cyclic electron transport around photosystem I and chlororespiration in angiosperms. The Src homology 3 domain (SH3)-like fold protein NdhS/CRR31 is an NDH subunit that is necessary for high affinity binding of ferredoxin, indicating that chloroplast NDH functions as a ferredoxin:plastoquinone oxidoreductase. However, the mechanism of the interaction between NdhS and ferredoxin is unclear. In this study, we analyzed their interaction in planta by using site-directed mutagenesis of NdhS. In general, binding of ferredoxin to its target proteins depends on electrostatic interaction. In silico analysis predicted the presence of a positively charged pocket in the SH3-like domain of NdhS, where nine charged residues are highly conserved among plants. Systematic alteration of these sites with neutral glutamine revealed that only arginine 193 was required for high NDH activity in vivo. Further replacement of arginine 193 with negatively charged aspartate or glutamate or hydrophobic alanine significantly decreased the efficiency of ferredoxin-dependent plastoquinone reduction by NDH in ruptured chloroplasts. Similar results were obtained in in vivo analyses of NDH activity and electron transport. From these results, we propose that the positive charge of arginine 193 in the SH3-like domain of NdhS is critical for electrostatic interaction with ferredoxin in vivo.

Light reactions of photosynthesis in the thylakoid membrane of chloroplasts convert light energy into NADPH and ATP. The process consists of two types of electron transport: linear electron transport and cyclic electron transport around photosystem I (CET). 2 In linear electron transport, photosystem II (PSII) oxidizes water to dioxygen in its manganese cluster, and the electrons derived are transported to photosystem I (PSI) via plastoquinone (PQ), the cytochrome b 6 f complex, and plastocyanin. PSI reduces feredoxin (Fd), an iron-sulfur electron carrier protein, and Fd donates electrons to ferredoxin:NADP ϩ oxidoreductase (FNR) to reduce NADP ϩ to NADPH for CO 2 fixation by the Calvin-Benson cycle. Proton translocation from the stroma to thylakoid lumen, coupled with the Q-cycle in the cytochrome b 6 f complex and water oxidation in PSII, produces the proton motive force to produce ATP by ATP synthase. In contrast, CET is driven solely by PSI, and electrons are recycled from Fd to PQ (1). Consequently, CET generates ⌬pH without reducing NADP ϩ . Two physiological roles of CET have been proposed (2)(3)(4): 1) regulation of the ATP to NADPH production ratio in photosynthesis and 2) acidification of the thylakoid lumen to induce the thermal dissipation of absorbed excess photon energy in PSII, which can be monitored as a qE component of non-photochemical quenching (NPQ) of chlorophyll fluorescence. Acidification of the thylakoid lumen down-regulates the cytochrome b 6 f complex to prevent overreduction of PSI, especially under fluctuating light conditions (5). In angiosperms, backflow of electrons from Fd to PQ operates via two redundant pathways, namely a PGR5 (proton gradient regulation 5)/PGRL1 (PGR5-like photosynthetic phenotype 1) complex-dependent antimycin A(AA)-sensitive pathway (6 -9) and an NADH dehydrogenase-like complex (NDH)-dependent antimycin A-insensitive pathway (10 -12).
The complete sequencing of plastid genomes has revealed 11 genes, ndhA to ndhK, that encode proteins homologous to subunits of mitochondrial NADH dehydrogenase (complex I) (13). Chloroplast NDH is more similar to cyanobacterial NDH-1, which is believed to be the origin of chloroplast NDH, than to * This work was supported by Grants 22114509 and 22247005 from the Min-mitochondrial NADH dehydrogenase present in the same species (14). Extensive genetic and proteomic analyses have revealed that the chloroplast NDH complex is composed of five subcomplexes, i.e. the A, B, membrane, lumen, and electron donor-binding subcomplexes (15,16). Chloroplast NDH is further associated with PSI to form a Ͼ1-MDa supercomplex via minor light-harvesting complex I (LHCI) proteins (15,16). Recently, the fine tertiary structure of complex I of the bacterium Thermus thermophilus was resolved (17)(18)(19). The bacterial complex I forms an L-shaped structure consisting of a peripheral arm and a membrane-embedded domain. The top of the peripheral arm is equipped with three subunits, Nqo1 to -3, involved in the oxidation of NADH (17). Although the L-shaped skeleton is conserved in chloroplast and cyanobacterial NDH, the complex does not include subunits corresponding to Nqo1 to -3. For this reason, the nature of the electron donor and electron donor-binding subunits has long been unclear in chloroplast and cyanobacterial NDH.
A novel NDH subunit, NdhS/CRR31 (chlororespiratory reduction 31), has been identified by proteomic and genetic analyses of the NDH-PSI supercomplex in Arabidopsis thaliana (12). NdhS is a thylakoid peripheral protein, and its C-terminal region forms the Src homology 3 domain (SH3)-like fold that is known to be involved in protein-protein interaction (20). NdhS interacts with subcomplex A of NDH complex via the J-proteins NdhT/CRRJ and NdhU/CRRL (12), forming an electron donor-binding subcomplex in chloroplast NDH (14). NDH with high affinity for Fd has been reconstituted in ruptured chloroplasts isolated from Arabidopsis null allele ndhs-1 (crr31-1) by the addition of recombinant NdhS (12). From these results, we propose that the chloroplast NDH complex functions as a PGR5/PGRL1-independent, Fd-dependent PQ reductase and that it should be renamed as NADH dehydrogenase-like complex (12).
Chloroplast Fds are small, two-iron two-sulfur proteins that serve as electron donors in various metabolic pathways other than photosynthesis, such as nitrogen and sulfur assimilation (21)(22)(23). The surface net charge of Fd is highly negative, and this property plays a crucial role in electrostatic interaction with its target proteins, such as FNR, nitrite reductase, and sulfite reductase (24). To address the mechanisms underlying the interaction between NdhS and Fd, we conducted site-directed mutagenesis of nine highly conserved charged amino acids in the SH3-like domains from land plant NdhS homologs and analyzed the function of NdhS variants in planta.

EXPERIMENTAL PROCEDURES
Plant Materials and Growth Conditions-A. thaliana (Columbia) was grown in soil in a growth chamber (50 mol of photons m Ϫ2 s Ϫ1 , 16-h photoperiod, 23°C) for 3-4 weeks. Biochemical and physiological analyses were performed on plants before the initiation of bolting. ndhS-1/crr31-1 was provided by GABI-Kat, and ndhS-1 pgr5 was produced as reported previously (12).
In Planta Site-directed Mutagenesis of NdhS-The wild-type genomic sequence containing NdhS cloned into pDONR/Zeo (Invitrogen) was used as a template for site-directed mutagenesis (12). NdhS has no intron sequences. Site-directed mutagenesis of triplet codons for the positively or negatively charged amino acids in the SH3-like domain (K178Q, R193Q, R193K, R193H, R193D, R193E, R193A, E204Q, D209Q, R210Q, K232Q, R222Q, and K224Q) was done with a QuikChange site-directed mutagenesis kit (Stratagene). All of the mutated sequences were confirmed by sequencing and then transferred to the binary vector pGWB-NB1 by LR Clonase reaction (Invitrogen). Agrobacterium tumefaciens C58 was transformed by electroporation with the resultant plasmids, and the bacteria were used to transform Arabidopsis ndhs-1 and ndhs-1 pgr5 mutants by using the floral dip method (25). Transformed plants were selected on solidified 1/2 Murashige-Skoog medium containing 7.5 g ml Ϫ1 BASTA, and insertion of the BASTA resistance gene (bar) into the genome was verified by PCR. The T 2 generation of transgenic lines was used for physiological and biochemical analyses.
In Vivo Chlorophyll Fluorescence Analyses-Chlorophyll fluorescence was measured with a pulse-amplitude modulation portable chlorophyll fluorometer, MINI-PAM (Walz), as described before (12). Three or four plants of each genotype were analyzed, and average values and standard deviations were calculated. Minimum fluorescence at open PSII centers in the dark-adapted state (F o ) was excited with a weak measuring light (wavelength 650 nm) at a photon flux density (PFD) of 0.05-0.1 mol of photons m Ϫ2 s Ϫ1 . A saturating pulse of white light (800 ms, 8000 mol of photons m Ϫ2 s Ϫ1 ) was applied to determine the maximum fluorescence at closed PSII centers in the darkadapted state (F m ) and during actinic light (AL) illumination (F m Ј). The steady-state fluorescence level (F s ) was recorded during AL illumination (5-1000 mol of photons m Ϫ2 s Ϫ1 ). The maximum quantum yield of PSII was calculated as F v /F m . NPQ was calculated as (F m Ϫ F m Ј)/F m Ј. The quantum yield of PSII (⌽ PSII ) was calculated as (F m Ј Ϫ F s )/F m Ј (26). Electron transport rate (ETR) was calculated as ⌽ PSII ϫ PFD (mol of photons m Ϫ2 s Ϫ1 ). qL, the fraction of open PSII center (27), was calculated as (28). The transient increase in chlorophyll fluorescence after the AL was turned off was monitored as described in the legend for Fig. 2 (10). Intensity of the measuring light was 1 mol of photons m Ϫ2 s Ϫ1 .
In Vitro Assay of Fd-dependent PQ Reduction-In vitro assay of Fd-dependent PQ reduction was performed as described previously (12). Intact chloroplasts (20 g of chlorophyll ml Ϫ1 ) were osmotically ruptured in 50 mM HEPES-NaOH (pH 8.0) for each assay. NADPH (0.25 mM) and the indicated concentrations of spinach Fd (Sigma) were added, and the increase in apparent minimum chlorophyll fluorescence (F o Ј) was recorded with a MINI-PAM. Fluorescence levels were normalized against F m levels. To measure NDH-dependent PQ reduction, 5 M AA (Sigma) was added to the assay to inhibit PGR5-PGRL1-dependent PQ reduction activity, as reported previously (29).
SDS-PAGE and Immunoblot Analyses-Intact chloroplasts were purified from the leaves of 3-4-week-old plants, as described previously (6). The purified chloroplasts were suspended in 20 mM HEPES-KOH (pH 7.6) containing 5 mM MgCl 2 and 2.5 mM EDTA. The insoluble fraction containing thylakoids and envelopes was separated from the stromal frac-tion by centrifugation for 5 min at 15,000 ϫ g. The chlorophyll concentration was determined as described previously (30). Proteins in the insoluble fraction separated by 12.5% (w/v) SDS-PAGE were electrotransferred onto polyvinylidene fluoride membranes. The antibodies were added, and the protein-antibody complexes were labeled by using an ECL Plus Western blotting detection kit (GE Healthcare). Chemiluminescence was detected with a lumino-image analyzer LAS3000 (FUJIF-ILM) and analyzed with MultiGauge Version 3.0 software (FUJIFILM). Antibody against FNR was purchased from Agrisera.
Homology Modeling and Calculations of Surface Net Charge and Hydrophobicity-Homology models of the SH3-like domain in NdhS variants were constructed by using Swiss-Model, with a Synechocystis sp. PCC 6803 NdhS homolog (Protein Data Bank (PDB): 3C4S) as a template. The surface net charges of the homology models were calculated with PDB2PQR Server version 1.7 by using the Poisson-Boltzmann equation (31). PROPKA was used to assign the protonation state at pH 8.0 (32). The nature of the pocket on the protein surface was predicted by using Q-SiteFinder (33). The hydrophobicity of the surface was calculated in accordance with the values on the hydrophobicity scale of Kyte and Doolittle (34). All images of homology models were produced by using UCSF Chimera 1.5.3r (35).
Accession Numbers-Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank TM /Euro-  Fig. 1A shows multiple alignments of the C-terminal region of NdhS homologs from land plants and cyanobacteria. Although the ocean cyanobacterium UCYN-A does not encode genes for PSII subunits or Calvin-Benson cycle enzymes, ndhS is conserved as well as other genes for NDH subunits of A and membrane subcomplexes (36). In the cyanobacterium Synechocystis sp. PCC 6803 (S. 6803), the function of NdhS as a subunit of the NDH complex is conserved (37). By homology modeling on the basis of the tertiary structure of S. 6803 NdhS (PDB: 3C4S), we revealed that Arabidopsis NdhS had a conserved SH3-like domain composed of five ␤-sheets (Fig. 1, A and B) (10). The SH3-like domain was sandwiched between land plant-specific domains (corresponding to Glu-57-Leu-168 and Arg-222-Pro-250 of AtNdhS). The N-terminal extension is not essential for the function of NdhS (12). Several positively or negatively charged amino acids were highly conserved among the SH3-like domains; in particular, Arg-193, Lys-198, and Glu-204 were completely conserved among cyanobacteria and land plants (note that this numbering refers to AtNdhS) (Fig. 1A). Asp-209 was also conserved, except in the case of UCYN-A. Lys-178, Arg-210, Arg-222, Lys-224, and Lys-234 were specifically conserved in land plants. Prediction of the protein surface on the basis of models produced by the PDB2PQR server (31) and Q-SiteFinder (33) indicated the presence of a positively charged pocket in the SH3-like domain in both Arabidopsis and S. 6803 NdhS (Fig. 1, C and E). The pocket of AtNdhS was likely to be more positively charged than that of S. 6803 NdhS. Two conserved positively charged amino acids, Arg-193 and Lys-198, were localized to the edge of this pocket in AtNdhS (Fig. 1, B  and D). The volume of the positively charged pocket in AtNdhS was estimated to be 86 Å 3 by the Q-SiteFinder. Prediction by using eF-surf and the Pocket-Finder Pocket Detection server also suggested that the pocket around Arg-193 was positively charged and hydrophilic in S. 6083 NdhS (data not shown).

Identification of Conserved Charged Amino Acids and Positively Charged Pocket of the SH3-like Domain of NdhS-
Electrostatic interaction plays a crucial role in the binding of Fd to target proteins (24). Because our previous study indicated that the N-terminal extension of AtNdhS is not essential for its function in vivo (12), for the mutagenesis, we focused on nine highly conserved charged amino acids at the C-terminal region, summarized in Fig. 1A.
Site-directed Mutagenesis of Nine Conserved Charged Amino Acids in the SH3-like Domain in Planta-An Arabidopsis ndhs-1/crr31-1 mutant is a knock-out allele of the ndhS gene (12). To assess the contribution of the nine charged residues conserved in the SH3-like domains, Lys-178, Arg-193, Lys-198, Glu-204, Asp-209, Arg-210, Arg-222, Lys-224, and Lys-232 were replaced by neutral Gln, and the mutated versions of the ndhS gene were introduced into the Arabidopsis ndhs-1 mutant.
Chloroplast NDH mediates electron flow from Fd to PQ ( Fig.  2A). After AL is turned off, NDH still reduces PQ in the dark. This non-photochemical reduction of PQ by NDH in vivo can be monitored as a transient increase in chlorophyll fluorescence after AL is turned off (F o Ј rise) (10,38). As reported previously (12), ndhs-1 did not show an F o Ј rise, indicating that NdhS is required for efficient operation of NDH in vivo. Expression of the wild-type and truncated form (NdhS ⌬57-168 ) of NdhS in ndhs-1 led to recovery of the rise in F o Ј, indicating that the N-terminal extension is not essential for NdhS function (Fig. 2B) (12). In a series of Gln scannings, only R193Q significantly decreased the F o Ј rise (Fig. 2B). Immunoblot analyses indicated that this replacement did not affect NdhS stability (Fig. 2C). Except for R193Q, none of substitutions of conserved amino acids with Gln affected the F o Ј rise or NdhS stability (data not shown). Our antibody does not detect NdhS ⌬57-168 (12). Point mutations or a lack of NdhS did not affect the accumulation of NdhL, a subunit of subcomplex A of the NDH complex (Fig. 2C), as reported previously (12). Of the nine conserved positively charged residues, only Arg-193 was necessary for NdhS function.
Mutations of Arg-193 Alter the Surface Net Charge of the Pocket of SH3-like Domain-Arg-193 was included in the positively charged pocket that could form an Fd-binding site via electrostatic interaction (Fig. 1). To predict the role of Arg-193 in formation of the positively charged pocket, the surface net charges of homology models of the SH3-like domain were calculated in NdhS R193 variants (Fig. 3). Substitution of Arg-193 with positively charged Lys did not affect the surface net charge in NdhS R193K . The positive charges of the pockets of NdhS R193H , NdhS R193Q , and NdhS R193A were predicted to be decreased. Furthermore, upon substitution with the negatively charged residues Asp and Glu, the surface net charges of NdhS R193D and NdhS R193E were predicted to be neutral and slightly negative, respectively. Arg-193 was predicted to form a hydrophilic patch on the edge of the pocket (data not shown). In NdhS R193A , the hydrophilic patch was predicted to be hydrophobic.

Positive Charge at the Position of the 193rd Amino Acid is Essential for High Affinity Binding of Fd to the NDH Complex-
To test for a link between the positively charged surface of the SH3-like domain of NdhS and NDH activity, we analyzed Fd-dependent PQ reduction activity in ruptured chloroplasts isolated from ndhs-1 mutant plants accumulating NdhS variants by monitoring the increase in chlorophyll fluorescence due to PQ reduction (Fig. 4A). Because the PGR5-PGRL1-dependent pathway contributes mainly to PQ reduction activity, a subtle change in PQ reduction activity was observed in the original ndhs-1 plants and ndhs-1 plants accumulating NdhS variants in the absence of AA, except in the case of pgr5 (Fig.  4, B and C).  DECEMBER 20, 2013 • VOLUME 288 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 36331
NdhS is not essential for NDH activity but is required for high affinity binding of Fd to NDH (12). To evaluate the function of NdhS, we monitored the Fd concentration dependence of PQ reduction activity (Fig. 4, D-L) in the presence of AA. PQ reduction activity was almost saturated at 3 M Fd in the wild type (Fig. 4D). In ndhs-1, however, even in the presence of 20 M Fd, PQ reduction activity was slightly lower than that in the wild type with 1 M Fd (Fig. 4E). Introduction of NdhS R139K almost fully complemented PQ reducing activity, as in plants that expressed wild-type NdhS (Fig. 4, F and G). However, introduction of NdhS R139H and NdhS R139Q only partly complemented PQ reducing activity (Fig. 4, H and I). Introduction of NdhS R139A , NdhS R139E , or NdhS R139D did not complement PQ reducing activity at all (Fig. 4, J-L), although the plants accumulated wild-type levels of NdhS proteins (Fig. 2C). The level of FNR, whose reverse reaction is required in the assay system, was not affected in any genotypes (Fig. 2C). Similar results were obtained by analyzing a different set of measurement using independent transgenic lines (supplemental Fig. S1).
Physiological Significance of Arg-139 in Photosynthesis-In Arabidopsis, chloroplast NDH is dispensable under growth chamber conditions but is essential in the pgr5 mutant background (4). To evaluate the physiological significance of the positive surface charge of the SH3-like domain, the ndhs-1 pgr5 double mutant was transformed by using NdhS variants in which the light intensity dependence of ETR, NPQ, and 1-qL was analyzed. The ndhs-1 defect scarcely affected NPQ induction, even under the pgr5 mutant background (data not shown). However, ETR and 1-qL were drastically affected in the ndhs-1 pgr5 double mutant (Fig. 5, A-D), as reported previously (12). For representative results, we selected two points of light intensity (95 and 692 mol of photons m Ϫ2 s Ϫ1 ). Introduction of the wild-type ndhS into the double mutant partially restored both parameters (Fig. 5, A-D). Introduction of NdhS R193K also restored ETR and 1-qL, but the levels were slightly lower than those in the lines accumulating wild-type NdhS (statistically significant depending on the light intensity and parameter), suggesting that NdhS R193K was also slightly less efficient than the wild-type NdhS in vivo. In the lines accumulating NdhS R193E , in which the pocket was slightly negatively charged, ETR and 1-qL were more severely affected than in the line accumulating NdhS R193K , consistent with the results in the ruptured chloroplasts (Figs. 4 and 5). Notably, even NdhS R193E restored the phenotypes in the double mutant significantly; these variants still retained the function of high affinity binding to Fd. The lines accumulating NdhS R193H , NdhS R193Q , NdhS R193A , and NdhS R193D had intermediate phenotypes between those of the lines accumulating NdhS R193K and those accumulating NdhS R193E , although the differences were not significant between some of the variants (Fig. 5, A-D). Consistent with the results in ruptured chloroplasts (Fig. 4), the positive surface charge of the SH3-like domain was required for the efficient function of NdhS in vivo.

DISCUSSION
A key to elucidating the function of NdhS in the chloroplast NDH complex was the finding that the predicted tertiary structure of the C-terminal domain of NdhS forms an SH3-like domain that is similar to that of PsaE, which forms the Fd-binding site of PSI with PsaC and PsaD (12,39). In cyanobacteria, PsaE was reported to be involved in CET in unspecified manner (40). On the basis of our in silico prediction, again, we discovered a putative positively charged pocket in the SH3-like domain of NdhS (Fig. 1). Among nine charged residues that are conserved among NdhS proteins, only Arg-193 was essential for efficient operation of NDH in vivo (Figs. 2 and 5) and in ruptured chloroplasts (Fig. 4). In silico prediction of the tertiary structure and surface net charge of a protein is a powerful tool for clarifying the function of the protein, especially when the information is combined with that from in vitro mutagenesis.
Arg-139 is a critical site for the high affinity binding of NDH to Fd. As observed in the interaction between FNR and Fd, electrostatic interaction is likely to play an important role (22). A similar case was observed in Arg-39 of PsaE, the SH3-like domain of which also forms the site for the docking of PSI to Fd (41). As observed in NDH activity both in vivo (Figs. 2 and 5) and in ruptured chloroplasts (Fig. 4) (41). Most likely, NdhS is involved in high affinity binding with Fd via the same molecular mechanism as with PsaE.
Even when the acidic amino acids Asp or Glu were substituted for Arg-139, the lines showed higher NDH activity than the ndhs-1 null allele (Figs. 2 and 5). This contrasts with the fact that similar substitutions in Arg-39 of PsaE increase the K d value even higher than that in the psaE null allele (41). In NdhS, other amino acid residues also contribute to the high affinity binding of NDH to Fd, although eight of the charged residues conserved in NdhS proteins do not have this role.
PsaE serves as high affinity Fd-docking protein of PSI, but it is not essential for photoreduction of Fd by PSI (42,43). Besides PsaE, PsaC and PsaD also contribute to Fd binding (39). Similarly, NdhS is required for high affinity binding of NDH to Fd. We do not eliminate the possibility that electron transport from Fd is partly affected in the ndhs mutant, but NdhS is not essential for NDH activity (12). Complex I is believed to have originated from [NiFe]-hydrogenase, which accepts electrons from Fd rather than NADH (44). In this enzyme, the subunit EchF is involved in oxidation of Fd in Methanosarcina barkeri. EchF is structurally similar to NdhI in chloroplast NDH, implying that NdhI directly accepts an electron from Fd. If this is the case, NdhS may be closely localized to NdhI to form the Fd-binding site. How does Fd still interact with NdhI in the ndhS mutant (12)? In genuine NADH dehydrogenase in Escherichia coli, a subunit corresponding to NdhI accepts electrons from an NADH-binding module consisting of NuoE, -F, and -G (44). It is unlikely that the L-shaped skeleton of the NADH dehydrogenase-related

Ferredoxin-binding Site of Chloroplast NDH
complex possesses a site for high affinity binding to Fd. In Campylobacter jejuni, complex I lacks subunits corresponding to NuoE and NuoF. In the absence of these proteins, CJ1574 protein, which is not conserved in complex I in other organisms, mediates electron flow from reduced flavodoxin to the complex (45,46). As in C. jejuni, the remaining NDH activity in the ndhs-1 mutant may depend on additional accessory NDH subunits that form the Fd-binding site, as well as on NdhS (12). In Arabidopsis, NdhS forms the electron donor-binding subcomplex with the J protein NdhT/ CRRJ and the J-like protein NdhU/CRRL. In contrast to NdhS, NdhT and NdhU are essential for NDH activity (12). It is also possible that unknown factors are still lacking in the model (14).  A representative line from each construction was selected, and three independent T 2 seedlings were analyzed. Columns with the same letters are not significantly different (Tukey-Kramer test, p Ͻ 0.05). Error bars indicate mean Ϯ S.D. E, immunodetection of NdhS, NdhL, FNR, and Cytf in chloroplast membranes. Membrane proteins corresponding to 1 g of chlorophyll were loaded. NdhL, FNR, and Cytf were detected as controls for the NDH complex and thylakoid membrane protein, respectively. Cytf, cytochrome f.