- © 2017 American Society of Plant Biologists. All Rights Reserved.
Abstract
Plant mutants for genes encoding subunits of mitochondrial complex I (CI; NADH:ubiquinone oxidoreductase), the first enzyme of the respiratory chain, display various phenotypes depending on growth conditions. Here, we examined the impact of photoperiod, a major environmental factor controlling plant development, on two Arabidopsis (Arabidopsis thaliana) CI mutants: a new insertion mutant interrupted in both ndufs8.1 and ndufs8.2 genes encoding the NDUFS8 subunit and the previously characterized ndufs4 CI mutant. In the long day (LD) condition, both ndufs8.1 and ndufs8.2 single mutants were indistinguishable from Columbia-0 at phenotypic and biochemical levels, whereas the ndufs8.1 ndufs8.2 double mutant was devoid of detectable holo-CI assembly/activity, showed higher alternative oxidase content/activity, and displayed a growth retardation phenotype similar to that of the ndufs4 mutant. Although growth was more affected in ndufs4 than in ndufs8.1 ndufs8.2 under the short day (SD) condition, both mutants displayed a similar impairment of growth acceleration after transfer to LD compared with the wild type. Untargeted and targeted metabolomics showed that overall metabolism was less responsive to the SD-to-LD transition in mutants than in the wild type. The typical LD acclimation of carbon and nitrogen assimilation as well as redox-related parameters was not observed in ndufs8.1 ndufs8. Similarly, NAD(H) content, which was higher in the SD condition in both mutants than in Columbia-0, did not adjust under LD. We propose that altered redox homeostasis and NAD(H) content/redox state control the phenotype of CI mutants and photoperiod acclimation in Arabidopsis.
Complex I (CI; NADH:ubiquinone oxidoreductase; EC 1.6.5.3), the first enzyme of the respiratory chain of most eukaryotes, including plants, couples electron transfer to proton translocation through the inner mitochondrial membrane (Klodmann et al., 2010). CI is an L-shaped multimeric enzyme of around 1 MD in size composed of a matrix-faced peripheral arm carrying the NADH-oxidizing activity (N module), a connecting module (Q module transferring electrons to the quinone-binding site), and a hydrophobic intramembrane arm carrying the proton translocation activity (P module). Eukaryotic CI comprises more than 40 subunits, of which 14 also are present in the ancestral bacterial enzyme (core subunits; Efremov et al., 2010). They are encoded by either mitochondrial genes (e.g. NAD) or by nuclear genes that might be present in multiple copies (see Table 1 in Meyer et al., 2011). Mutations in both mitochondrial and nuclear genes have been characterized in various plant species. Mutants devoid of NAD subunits were characterized in maize (Zea mays; ncs2, deleted for NAD4; Marienfeld and Newton, 1994) and in Nicotiana sylvestris (CMSI and CMSII, deleted for NAD7; Pla et al., 1995). Mutations in genes controlling NAD cis/trans-processing have been described in N. sylvestris, maize, and Arabidopsis (Arabidopsis thaliana; for review, see Colas des Francs-Small and Small, 2014), and insertion mutants lacking nucleus-encoded subunits have been characterized in Arabidopsis (Meyer et al., 2009; Kühn et al., 2015; Soto et al., 2015). Impaired holo-complex assembly/activity are common features of all plant mutants characterized so far, but putative intermediate assembly forms have been observed in several cases (Gutierres et al., 1997; Karpova and Newton, 1999; Perales et al., 2005; Pineau et al., 2008; Meyer et al., 2009, 2011; Kühn et al., 2015). A general respiratory impairment, measured as either oxygen consumption of leaf discs or CO2 emission by attached leaves in the dark (dark respiration), has not been observed, and in organello experiments showed the induction of nonphosphorylating NAD(P)H dehydrogenases (Rasmusson et al., 2008) in all cases analyzed (Marienfeld and Newton, 1994; Sabar et al., 2000; Meyer et al., 2009; Keren et al., 2012). Furthermore, both the capacity of the alternative oxidase (AOX) pathway, which bypasses the cytochrome oxidase (COX) pathway, and the AOX protein content were increased in all mutants.
Plant CI mutants have various phenotypes, from mild to severe or even near lethal (Kühn et al., 2015); in addition to a slow-growing phenotype, morphological distortions and male sterility were reported in maize NCS2 (Karpova and Newton, 1999) and in N. sylvestris NMS1 and CMSII (De Paepe et al., 1990). Seedlings of Arabidopsis nMat1/4 (Keren et al., 2012; Cohen et al., 2014) and ndufv1 (Kühn et al., 2015) mutants barely survive unless grown in vitro with Suc supplementation. Differential impact of the mutations also was observed at the physiological and metabolic levels. Decreased photosynthetic activity is not a general effect and was reported in the N. sylvestris CMSII and NMS1 mutants (Sabar et al., 2000; Dutilleul et al., 2003a; Priault et al., 2007) and in the Arabidopsis ca2 cal2 mutant only (Soto et al., 2015). In the N. sylvestris mutants, limited carbon supply resulting from higher photorespiration is likely to contribute to the slow-growth phenotype (Priault et al., 2006b). Although similar metabolic alterations were observed in N. sylvestris (Dutilleul et al., 2005; Szal et al., 2008; Djebbar et al., 2012) and Arabidopsis CI mutants (Meyer et al., 2009; Keren et al., 2012) as increased levels of amino acids, ATP, and NAD(H), noteworthy differences were reported recently between Arabidopsis mutants (Kühn et al., 2015). Moreover, an increase in total reactive oxygen species (ROS) content was reported for many Arabidopsis CI mutants (Meyer et al., 2009; Keren et al., 2012; Soto et al., 2015) but not in the CMSII mutant (Dutilleul et al., 2003b). The reasons for these inconsistencies are unknown at present. Complementation studies, performed for CMSII (Pineau et al., 2005), opt43 (de Longevialle et al., 2007), ndufs4 (Meyer et al., 2009), and mutants in CA2/CAL2 genes encoding CA/CAL γ-carbonic anhydrase subunits (Sunderhaus et al., 2006; Soto et al., 2015), indicated that additional mutations are not involved in the altered phenotype in these cases. It has been proposed that the presence of traces of holo-CI might attenuate mutant deficiencies (Kühn et al., 2015). Alternatively, it is possible that assembly intermediates might interfere with normal oxidative phosphorylation or have additional functions, as suggested by Keren et al. (2012).
The phenotype of several CI mutants is affected by growth conditions. CMSII plants showed altered responses to environmental conditions such as light (Priault et al., 2006a), nitrogen, and CO2 (Pellny et al., 2008; Hager et al., 2010) and were fully male sterile under very low illumination only (De Paepe et al., 1990). They were differentially affected by drought stress (Galle et al., 2010; Djebbar et al., 2012). Also, the growth defect of the Arabidopsis ca2 cal2 mutant was rescued under nonphotorespiratory conditions (Soto et al., 2015). However, despite its importance for plant development and productivity, the photoperiod response in CI mutants has not been investigated yet. Plants can be classified into day neutral, short day (SD), and long day (LD) on the basis of the minimum duration of light per day necessary to trigger the transition from a vegetative growth phase to a reproductive state. Arabidopsis is a facultative LD species that blooms earlier in LD (12 h or greater daylength) than in SD, and photoperiod affects morphological (leaf thickness and thylakoid organization) and physiological (carbon assimilation-related parameters and stomatal conductance) parameters, redox state, and stress responses (Robbins and Pharr, 1987; Gibon et al., 2004, 2009; Becker et al., 2006; Lepistö and Rintamäki, 2012; Queval et al., 2012). Although it has been reported that daylength (16 h versus 12 h) impacts on the phenotype of the ABA overly-sensitive 5 mutant affected in nad2 splicing (Liu et al., 2010) and that photoperiodic control was altered in cal mutants (Wang et al., 2012), Arabidopsis CI mutants have been fully characterized in LD only, and the link between CI and photoperiod remains to be studied.
Here, we examined the impact of different photoperiod regimes on two different Arabidopsis CI mutants: a new insertion mutant interrupted in both ndufs8.1 and ndufs8.2 genes encoding the NDUFS8 subunit belonging to the core CI (Efremov et al., 2010) and the previously characterized ndufs4 mutant (Meyer et al., 2009; Kühn et al., 2015). Under greenhouse LD condition, both ndufs8.1 and ndufs8.2 single mutants were indistinguishable from Columbia-0 (Col-0) at both the biochemical and phenotypic levels. In contrast, the ndufs8.1 ndufs8.2 double mutant was devoid of holo-CI assembly/activity and displayed a growth-retardation phenotype similar to that of ndufs4. Although growth was less affected in ndufs8.1 ndufs8.2 than in ndufs4 under SD conditions, both mutants had an impaired growth response when transferred from SD to LD. We further provide insights into metabolomics and biochemical alterations associated with the acclimation to LD and discuss how mitochondrial CI activity might interact with photoperiod acclimation in Arabidopsis.
RESULTS
Molecular and Biochemical Characterization of Single and Double Mutants of the NDUFS8 Subunit
Mutant lines carrying a T-DNA insertion in the ndufs8.1 (At1g16700) and ndufs8.2 (At1g79010) genes (Col-0 accessions) were obtained from the Salk Institute collection. Both genes are composed of six exons (Fig. 1A), and their proteins share 94% similarity. By sequencing T-DNA left border insertion PCR products, we localized the insertion sites in the first intron and in the third intron in ndufs8.1 and ndufs8.2, respectively. Knockout mutations were confirmed by PCR and reverse transcription (RT)-PCR using ndufs8.1 and ndufs8.2 primer combinations: no ndufs8.1 and ndufs8.2 transcripts could be detected in the corresponding mutants (Fig. 1, B and C). In Col-0 seedlings and leaves, NDUFS8.2 was found to be about 2.5 times more expressed than NDUFS8.1 (Fig. 1C). No changes for NDUFS8.2 transcripts were detected in the ndufs8.1 mutant or for NDUFS8.1 transcripts in the ndufs8.2 mutant, indicating no adjustment at the gene expression level. The ndufs8.1 ndufs8.2 double mutant, recovered from crossing single mutants with the ndufs8.2 mutant used as the male donor, lacked both NDUFS8.1 and NDUFS8.2 transcripts (Fig. 1, B and C). In both single and double mutants, DNA and RNA patterns were stably maintained over five selfing generations; thus, all biochemical and physiological analyses were performed on S3 to S5 offspring.
Molecular characterization of single and double insertion mutants affected in the two Arabidopsis genes encoding the NDUFS8 subunit of CI. A, Gene structures and sites of insertion of T-DNA in ndufs8.1 (At1g79010) and ndufs8.2 (At1g16700) genes (arrows). B, Electrophoresis of RT-PCR products using NDUFS8.1- and NDUFS8.2-specific primers in the wild type (Col-0), ndufs8.1 and ndufs8.2 single mutants, and the ndufs8.1 ndufs8.2 double mutant. C, RT-quantitative PCR (qPCR) analysis of NDUFS8.1 and NDUFS8.1 genes in the wild type (Col-0), the ndufs8.1 and ndufs8.2 single mutants, and the ndufs8.1 ndufs8.2 double mutant, using ACTIN2 (ACT2) as a reference. Data are means + se of three independent measurements.
In order to see whether the expression of ndufs8.1 and ndufs8 genes was required for CI assembly and activity, leaf proteins were solubilized with digitonin, a detergent preserving the association of respiratory chain supercomplexes, in particular the I-III2 supercomplex (Pineau et al., 2008), and subjected to blue native (BN)/PAGE. In-gel NADH dehydrogenase activity, revealed by NDH/nitroblue tetrazolium (NBT) staining, was observed around 1 MD (which corresponds to CI size) in the wild type and in ndufs8.1 and ndufs8.2 single mutants but was not detectable in the ndufs8.1 ndufs8.2 double mutant (Fig. 2, left). Moreover, a signal around 1,500 kD corresponding to supercomplex I-III2 (Pineau et al., 2008) was observed in single mutants as in the wild type but not in the double mutant. Similarly, the immunostaining obtained using antiserum directed against NAD9 (mitochondrion-encoded CI subunit located in the peripheral arm; Klodmann et al., 2010) was observed in the wild type and single mutants but not in the double mutant (Fig. 2, middle). By contrast, the antibody directed against the COX2 subunit of cytochrome oxidase (complex IV) gave a similar signal in all cases (Fig. 2, right). Two-dimensional BN/SDS-PAGE of total root membrane proteins solubilized with β-dodecylmaltoside, which dissociates respiratory supercomplexes, failed to reveal a clear CI pattern in the double mutant, despite the presence of trace amounts of CI subunits (Supplemental Fig. S1). For example, the 76-kD subunit, which is part of the NADH dehydrogenase module located on the peripheral arm, was hardly distinguishable in the mutant.
CI assembly/activity in Col-0 and single and double mutants for the NDUFS8 subunit. Membrane proteins extracted from leaves of Col-0 (lanes 1), single ndufs8.1 (lanes 2) and ndufs8.2 (lanes 3) mutants, and the ndufs8.1 ndufs8.2 double mutant (lanes 4) were solubilized with digitonin that preserves the assembly of supercomplexes and resolved by BN/PAGE on 4%/13% gradient acrylamide gels. After migration, gels were stained for CI activity using NDH/NBT or blotted on polyvinylidene difluoride membranes for immunodetection studies. Left, In-gel NDH/NBT-stained leaf proteins. CI and the CI/III supercomplex (purple signals and red arrows), detected in Col-0 and in both ndufs8.1 and ndufs8.2 single mutants, are absent in the double mutant. The PSI complex that originated from thylakoid membranes is indicated by the arrowhead (Pineau et al., 2008). Middle, The anti-NAD9 immunosignal, revealed at the level of CI in the wild type and in both ndufs8.1 and ndufs8.2 single mutants, is not detected in the ndufs8.1 ndufs8.2 double mutant. Right, The accumulation of complex IV using anti-COX2 antibody (bottom of the gel) is similar in all genotypes.
In spite of the impaired CI activity, leaf total oxygen consumption in the dark was similar in ndufs8.1 ndufs8.2 and Col-0 (Fig. 3A). The cytochrome versus AOX partitioning, determined using oxygen isotope discrimination (Guy et al., 1989; Ribas-Carbo et al., 1995), showed a slight increase in the oxygen consumption rate by the AOX pathway in the mutant. This was associated with a marked increase in AOX capacity (Fig. 3B) and protein content (Fig. 3C). Although the anti-CA2 antibody is partly aspecific, a mitochondrion-specific CA signal was observed in comparable amounts in both mutants and the wild type (Fig. 3C). Similarly, the signal corresponding to the NAD9 subunit (synthesized inside mitochondria) was found in both genotypes.
Respiratory pathways and mitochondrial proteins in Col-0 and the ndufs8.1 ndufs8.2 mutant. A, Respiration rates of Col-0 and the ndufs8.1 ndufs8.2 double mutant were determined by oxygen discrimination. Measurements were performed on plants grown in controlled chambers under 12-h/12-h day/night. vt, Total oxygen uptake; vcyt, COX activity; valt, AOX activity; τ, partition to the AOX pathway. Data are means + se of measurements performed on at least three different plants. Asterisks indicate significant differences between the wild type and the mutant. B, AOX capacity. The percentage of cyanide (KCN)-resistant respiration in leaf tissues was determined as described by Florez-Sarasa et al. (2009). Data are means + se of measurements performed on at least three different plants. The asterisk indicates a significant difference between Col-0 and the mutant. C, Immunodetection of mitochondrial proteins on total leaf membranes (tot mb; left) and mitochondrial membranes (mt mb; right); asterisks indicate mitochondrion-specific immunosignals. The experiment was performed at least three times with similar results.
Hence, these results indicate that, as expected from undetectable levels of NDUFS8.1 and NDUFS8.2 transcripts, the NDUFS8 subunit is not synthesized (or only in trace amounts) in the double mutant, resulting in CI misassembly. Despite marked disturbances in the composition of respiratory complexes, including the accumulation of AOX proteins, no abnormalities of mitochondrial (ultra)structure could be observed by electron microscopy in ndufs8.1 ndufs8.2 (Supplemental Fig. S2), in contrast to what was observed in nMat mutants (Keren et al., 2012; Cohen et al., 2014).
Development of Single and Double Mutants for the NDUFS8 Subunit under LD Condition
Under greenhouse conditions (i.e. natural illumination, supplemented by 16 h of lighting with fluorescent tubes), seedling development, rosette morphology, and adult-stage size of ndufs8.1 and ndufs8.2 single mutants were similar to those of the wild type (Supplemental Fig. S3). That is, the lack of any of the two NDUFS8 proteins appeared to be fully compensated for at all developmental stages under greenhouse conditions. However, the development of in vitro germinated seedlings was delayed slightly in both single mutants (Supplemental Fig. S3D).
In contrast to single mutants, the growth of the ndufs8.1 ndufs8.2 double mutant was delayed markedly under both greenhouse and growth chamber (16-h/8-h day/night at 100 µmol m−2 s−1 photosynthetically active radiation [PAR]) conditions. Reduced development of in vitro germinating seedlings could be partly alleviated by Suc supplementation, as reported previously for other Arabidopsis CI mutants (Keren et al., 2012; Kühn et al., 2015). Leaves were somewhat more round shaped in ndufs8.1 ndufs8.2 than in Col-0, although not otherwise malformed. Bolting was delayed for about 2 weeks, whereas leaf size at the mature rosette stage, height and leaf dimensions of adult plants, structure of the inflorescence, flower morphology, and number of siliques were unaffected (Supplemental Fig. S3B). Pollen was fully fertile, as tested by Alexander dye staining (Supplemental Fig. S3C). At both rosette and adult stages, dimensions and aspect of ndufs4 plants were similar to those of ndufs8.1 ndufs8.2, despite a lower germination rate (Supplemental Fig. S3C) and a slightly lower growth rate of young plantlets (Supplemental Fig. S4B). Taken together, and considering the absence of phenotypic alterations in single mutants, these observations provide compelling evidence that the slow-growth phenotype of the ndufs8.1 ndufs8.2 double mutant is caused by the high reduction in CI activity.
Growth Enhancement Associated with the Transfer from SD to LD Conditions Is Similarly Compromised in Both ndufs8.1 ndufs8.2 and ndufs4 Mutants
To determine how the photoperiod regime influenced the slow-growth phenotype when CI was deficient, ndufs8.1 ndufs8.2 and ndufs4 mutants were grown in controlled chambers (100 µmol m−2 s−1 PAR) in the SD condition (8-h/16-h day/night) up to the eight- to nine-leaf stage (stage referred to as SD0 hereafter) and then either transferred to LD (16-h/8-h day/night) or maintained in SD under the same light intensities. In what follows, LD plants and SD plants refer to plants that were transferred to LD or maintained under SD, respectively. The number of days after the transition is denoted as SD3 and LD3 (after 3 d), SD6 and LD6 (after 6 d), and so on.
Surprisingly (and in marked contrast to the LD condition; Supplemental Fig. S3), ndufs4 plants grew less rapidly than ndufs8.1 ndufs8.2 plants in the SD condition (Supplemental Fig. S4). Young rosette plants were more compact in ndufs4 than in the two other genotypes, and necrotic points could be seen in some long-term SD-grown plants (inset in Supplemental Fig. S4A).
However, despite their developmental difference, the growth acceleration of ndufs8.1 ndufs8.2 and ndufs4 plants following their transfer to LD was similarly lower than in Col-0, for which enhanced growth under LD was already obvious at day 3 after transfer (Fig. 4A). The LD6-to-SD0 and LD6-to-SD6 ratios of shoot biomass were lower for both mutants compared with the wild type (Fig. 4B). In LD12 plants, the increase in all parameters of mature leaves (shoot biomass, leaf mass area and thickness, and dimensions of palisade cells) was lower in ndufs8.1 ndufs8.2 than in Col-0 (Supplemental Fig. S5). Hence, despite the lower growth rate of ndufs4 under the SD condition, the response of both mutants to their transfer to the LD condition was similarly impaired compared with the wild type. This suggests that the altered photoperiod response of growth is associated with reduced levels of CI activity but not with the absence of a specific CI subunit or another genetic alteration.
Growth comparisons of Col-0, ndufs8.1 ndufs8.2, and ndufs4 plants in the SD-to-LD transfer experiment. A, Col-0 and mutant plants (ndufs8.1 ndufs8.2 and ndufs4) were initially grown in controlled rooms under SD to achieve a similar development (eight- to nine-leaf stage; SD0), then transferred to LD for 3 d (LD3) or 6 d (LD6). B, Histograms of fresh weight (mg) of SD0, SD6, and LD6 plants in Col-0 (black bars), ndufs8.1 ndufs8.2 (gray bars), and ndufs4 (white bars). Data are means + se of measurements on at least eight different plants in all genotypes. Differences between mutant and Col-0 values are much higher in the LD than in the SD condition. Asterisks indicate significant differences between mutants and Col-0 according to Student’s t test. C, LD-SD ratios of plant biomass (fresh weight) in Col-0 (black bars), ndufs8.1 ndufs8.2 (gray bars), and ndufs4 (white bars) plants. Asterisks indicate significant differences between mutants and Col-0.
The Leaf Metabolome Responds Differently to Photoperiod in Wild-Type, ndufs8.1 ndufs8.2, and ndufs4 Plants
Previous studies in N. sylvestris (Dutilleul et al., 2003b, 2005; Hager et al., 2010) and Arabidopsis CI mutants (Meyer et al., 2009; Keren et al., 2012; Cohen et al., 2014; Kühn et al., 2015) reported alterations in metabolic signature under standard LD conditions, with higher contents in tricarboxylic acid intermediates and amino acids. Here, we compared wild-type and mutant metabolic profiles during transfer from SD to LD by untargeted hydrophobic interaction liquid chromatography (HILIC)-quadrupole time of flight (qTOF)-mass spectrometry (MS) using leaves sampled at day 6 in the middle of the light cycle. This technique provides a signature of polar metabolites, in particular organic and amino acids, sugars, and other metabolites involved in central metabolism (Supplemental Methods S1; Paglia et al., 2012; Pétriacq et al., 2016). Accurately detected mass-to-charge features (error = 0.4 ppm) in negative (ESI−; 17,326 ions) and positive (ESI+; 18,341 ions) ionization modes were integrated using XCMS (Smith et al., 2006). The resulting ion intensities were subjected to unsupervised principal component analysis (Fig. 5A) to obtain a metabolic overview between genotypes and photoperiod regimes. Principal component analysis showed complete separation of both genotypes and photoperiods in ESI− and ESI+. In SD, all three genotypes were rather distinct, and ndufs4 seemed to show the most distinct metabolic phenotype. In LD, while ESI− analysis showed a close grouping of ndufs8.1 ndufs8.2 with ndufs4 compared with the wild type, ESI+ analysis revealed overlapping profiles between the two CI-deficient mutants. Furthermore, untargeted HILIC profiles were poorly impacted by the SD-to-LD transfer in both mutants, especially for ESI+ (Fig. 5A). Metabolic relationships between samples also were visualized by hierarchical clustering analysis in order to reveal metabolic similarities between genotypes and photoperiod (Fig. 5B). Clearly, both CI mutants were rather similar in LD, and ndufs4 displayed the greater metabolic variation in SD. Hence, untargeted metabolomics support the outcome of growth analysis: both CI mutants are more similar in LD than in SD (Supplemental Figs. S4 and S5) and show limited response to SD-to-LD transfer (Fig. 4).
Untargeted metabolomics by HILIC-qTOF-MS. Multivariate analyses of the number of anions (ESI−) and cations (ESI+) detected by HILIC-qTOF-MS from Col-0 (squares), ndufs8.1 ndufs8.2 (triangles), and ndufs4 (circles) plants grown in SD (white symbols) or LD (gray symbols; n = 3). A, Unsupervised principal component analysis displaying the overall metabolic trends between samples. Variances are given in parentheses. PC1 and PC2, Principal components 1 and 2. B, Hierarchical clustering analysis showing metabolic relationships between genotypes and photoperiods (single linkage, tree sorted by size).
In order to identify metabolites driving the differences in LD acclimation between the wild type and mutants, we carried out targeted metabolomics by gas chromatography (GC)-MS at 6 d (SD6 and LD6). In SD, both mutants had elevated contents in tricarboxylic acid intermediates (citrate, succinate, fumarate, and malate), glycerol-3-phosphate (a redox shuttle metabolite), and most amino acids compared with the wild type (Fig. 6). In most cases, the increase was higher in ndufs4 than in ndufs8.1 ndufs8.2, and 2-oxoglutarate was detected in ndufs4 only. While the SD-to-LD transition was associated with a slight increase in succinate and malate and in some amino acids (Ala, Val, Arg, Pro, γ-aminobutyrate, and Lys) in ndufs8.1 ndufs8.2, it induced a decrease in these metabolites in ndufs4. As a result, metabolite pools of tricarboxylic acid intermediates and most amino acids were similar and higher than in the wild type in both ndufs8.1 ndufs8.2 and ndufs4 under LD. This is consistent with untargeted metabolomics, showing lower metabolic differences between the mutants under LD than under SD (Fig. 5), possibly reflecting their similar growth rates under the LD condition (Supplemental Fig. S4).
GC-MS determination of metabolites in Col-0, ndufs8.1 ndufs8.2, and ndufs4 plants. Leaves were sampled from SD6/LD6 plants at the middle of the day period. Analyte contents are expressed in relative units. Black columns, Col-0; gray columns, ndufs8.1 ndufs8.2; white columns, ndufs4. GABA, γ-Aminobutyrate; 2-OG, 2-oxoglutarate.
To get precise data on the timing and day/night effect of photoperiod-dependent metabolomics, GC-MS profiling was performed on wild-type and ndufs8.1 ndufs8.2 plants at days 3 and 6 after transfer, both at the middle of the light period and at the end of the dark period at day 6. Forty metabolites were significantly (ANOVA, P < 0.01) affected by the genotype (Fig. 7). These included sugars (Glc and Fru), tricarboxylic acid intermediates (citrate, malate, succinate, and fumarate), and amino acids (Gly, Ser, Ala, Gln, Glu, and Gln), which were in larger amounts in the mutant regardless of the time point considered, the light condition, and the photoperiod (Supplemental Figs. S7–S9). Thirty-three analytes were significantly affected by the photoperiod, partly overlapping with genotype-significant ones (e.g. malate and succinate; Supplemental Fig. S10). From a supervised orthogonal partial least-squares-discriminant analysis (r2 = 0.99 and q2 = 0.89) in each genotype, the effect of each feature in explaining the discrimination was quantified using a Volcano plot representing the variable importance for the projection against the coefficient along axis 1 (loading score; Supplemental Fig. S11, A and B). Using ANOVA, 18 analytes were found to be significant for the genotype-photoperiod interaction (Supplemental Fig. S11C), including hexoses (Glc and Fru), tricarboxylic acid derivatives (succinate, malate, and citramalate), and amino acids (Ser, Asn, Glu, and Gln). HPLC quantitation of amino acids revealed larger overall differences between ndufs8.1 ndufs8.2 and Col-0 under LD than under SD (Supplemental Fig. S12). There was a substantial effect of LD on amino acids derived from photorespiration and/or glycolysis (Ser, Gly, Ala, Val, Trp, and Leu) and 2-oxoglutarate (Glu, Gln, Arg, and Orn) in the mutant but a limited effect on oxaloacetate-derived amino acids (Asp, Asn, Thr, Lys, and Met).
Heat map and hierarchical clustering (cosine correlation) of metabolites found to be significant with respect to genotype in a two-way ANOVA (GC-MS metabolomics). Col-0 (col) and ndufs8.1 ndufs8.2 (mut) leaves were sampled from SD3/LD3 plants at the middle of the light period and from SD6/LD6 plants both at the middle of the light period and at the end of the night period (dark). Metabolomics analyses were carried out three times (i.e. three biological replicates). Relative metabolite contents are represented as mean-centered values with a color scale (blue, low content; red, high content). Numbers close to metabolite names refer to individual analytes associated with the metabolite of interest.
Finally, we examined whether the marked differences in tricarboxylic acid derivatives between the wild type and ndufs8.1 ndufs8.2 were associated with changes in transcripts associated with enzymes believed to be rate-limiting steps for the tricarboxylic acid pathway in the light (Tcherkez et al., 2009, and refs. therein; Araújo et al., 2012). Except for lower 2-OXOGLUTARATE DEHYDROGENASE2 (OXO2) and ACONITASE3 transcripts under SD and OXO2 and FUMARASE (FUM) transcripts under LD (Supplemental Fig. S13), few significant differences were observed between the wild type and mutants. Regarding the photoperiod effect, the accumulation of FUM transcripts in LD wild-type leaves was consistent with that of fumarate. Although a comprehensive study would require an analysis of all transcripts of tricarboxylic acid-related enzymes, these results suggest that the impact of photoperiod on tricarboxylic acid cycle gene expression is reduced in the ndufs8.1 ndufs8.2 mutant.
In summary, our metabolomic analyses (both untargeted HILIC-qTOF-MS and targeted GC-MS) show that, as for growth phenotype (Fig. 4), ndufs8.1 ndufs8.2 and ndufs4 display a rather similar metabolic phenotype in LD compared with SD (Figs. 5 and 6). Also, despite differences in tricarboxylic acid derivatives and some amino acids upon transfer from SD to LD, there is limited metabolic adaptation to LD in both mutants, in agreement with their impaired growth response.
Photoperiod Modulates Alterations of Carbon and Nitrogen Assimilation-Related Parameters in the ndufs8.1 ndufs8.2 Mutant
The remodeled metabolomes of CMSII (Sabar et al., 2000; Dutilleul et al., 2003a; Priault et al., 2006a, 2007) and ndufs4 (Meyer et al., 2009) mutants were associated with alterations in carbon and nitrogen assimilation under the LD condition. Here, we found that although net CO2 assimilation (A) was lower in the ndufs8.1 ndufs8.2 mutant than in the wild type under elevated light and was stimulated by LD in both genotypes (light response curves shown in Supplemental Fig. S14A), it was similar in all cases at growth PAR (100 µmol m−2 s−1; see inset). Likewise, the response of A to intercellular CO2 mole fraction (Ci) at moderate light (300 µmol m−2 s−1 PAR) and its stimulation by the LD condition did not differ significantly between the two genotypes (Supplemental Fig. S14B). Interestingly, the relationship between intracellular CO2 mole fraction at the carboxylation sites and Ci was the same in the two genotypes (Supplemental Fig. S14C), suggesting that internal conductance for CO2 (dissolution-diffusion) was similar. Also, we did not detect differences in chloroplast number and thylakoid organization (distribution or ultrastructure) between the wild type and ndufs8.1 ndufs8.2 (Supplemental Fig. S6) in either the LD or SD condition. Moreover, the electron flux to oxygenation and leaf glycolate oxidase activity did not differ significantly between genotypes and photoperiods (Supplemental Fig. S14D), indicating a similar photorespiration rate in all cases. Nevertheless, apparent carboxylation efficiency (the initial slope of the A/Ci relationship; inset in Supplemental Fig. S14A), Rubisco capacity, and total leaf ATP contents were higher in the mutant than in the wild type in SD but were not stimulated by LD (Fig. 8A). Also, chlorophyll a/b ratio, stomatal conductance, and dark respiration (CO2 evolution in darkness) did not differ significantly between genotypes in SD and increased in LD in the wild type only. Hence, physiological parameters showed that growth impairment in ndufs8.1 ndufs8.2 did not come from gas-exchange alteration (photosynthetic or photorespiratory effects) in either SD or LD, but photoperiod acclimation of photosynthetic parameters appeared to be compromised in the mutant.
Analysis of growth-related parameters in Col-0 and ndufs8.1 ndufs8.2. Leaves were sampled from Col-0 and ndufs8.1 ndufs8.2 plants at the middle of the light period on day 12. A, Carbon exchange-related parameters determined on leaves of 2- to 3-month-old rosette plants at the same developmental stage. Carboxylation efficiency (Ce; µmol CO2 m−2 s−1), stomatal conductance for CO2 (gs; mol m−2 s−1) calculated at growth illumination, Rubisco capacity (in vitro-measured maximum activity; nmol CO2 min−1 mg−1 protein), chlorophyll a/b ratios, night respiration (Rn; μmol CO2 m−2 s−1), and total leaf ATP (nmol g−1 fresh weight [FW]) are means + se of three to six measurements on different plants. Different letters indicate significant differences according to Student’s t test. B, Nitrogen assimilation-related parameters. Total leaf free amino acids (µmol g−1 fresh weight) determined from HPLC quantification (Supplemental Fig. S12), soluble proteins (mg g−1 fresh weight), total nitrogen content (%), NR capacity (maximum activity; µmol NO2 h−1 mg−1 protein), RT-qPCR analysis of the major nitrate reductase gene (NR2; expression relative to ACT2), and nitrate contents (µmol g−1 fresh weight) are means + se of three to six measurements on different plants. Different letters indicate significant differences according to Student’s t test. C, RT-qPCR analysis of TOR2/LST8 genes of the nutrient-dependent TOR pathway using ACT2 as a reference. Data are means + se of three to six measurements on different plants. Different letters indicate significant differences according to Student’s t test. D, RT-qPCR analysis of CCA1/LHY clock regulators. Data are means + se of six measurements on different plants. Different letters indicate significant differences according to Student’s t test.
In contrast to carbon assimilation, nitrogen assimilation was markedly affected in the ndufs8.1 ndufs8.2 mutant (Fig. 8B). Despite an increase in the mutant of total amino acids (i.e. 20% higher than in the wild type under SD and about 2-fold higher under LD) and protein contents (under SD), total nitrogen content (%) was reduced slightly under both photoperiods (Fig. 8B). Moreover, maximum nitrate reductase (NR) activity, transcript levels of the major nitrate reductase gene (NR2), and nitrate contents were lower in the mutant than in the wild type under SD. Since nitrogen assimilation was not affected significantly by photoperiod in the mutant whereas it was lower in LD than in SD plants in the wild type (also observed by Gibon et al. [2009]), this resulted in lesser phenotypic differences under LD. In addition, we examined the expression of TOR2 and LST8, two key genes of the TOR pathway, which were found previously to be involved in growth and photoperiod adaptation (Moreau et al., 2012). Both genes were up-regulated in LD compared with SD in both ndufs8.1 ndufs8.2 and the wild type, thus suggesting that the accelerated growth rate in LD might benefit from the TOR pathway independently of CI activity (Fig. 8C), although posttranscriptional changes might be involved. We also measured transcript levels of CCA1 and LHY clock regulators, which are important for growth and flowering (for review, see Nagel and Kay, 2012). Remarkably, levels of both transcripts were significantly higher in the mutant than in the wild type under SD but lower under LD due to their stimulation by LD in the wild type (Fig. 8D). Although these expression profiles would require a comprehensive analysis of the full diurnal cycle and other clock regulators, they suggest that the genetic control of the circadian rhythm might be altered in ndufs8.1 ndufs8.2 plants. Taken together, these results show that carbon and nitrogen metabolism are less affected by the SD-to-LD transfer in ndufs8.1 ndufs8.2 than in the wild type, in possible relation to an altered circadian rhythm in the mutant.
The Photoperiodic Regime Differentially Modulates Redox Homeostasis in the Wild Type and the ndufs8.1 ndufs8.2 Mutant
The signaling of the photoperiod response has been proposed to be under redox control in Arabidopsis (Lepistö and Rintamäki, 2012). ROS accumulation has been observed previously in various Arabidopsis CI mutants, including ndufs4 (Meyer et al., 2009; Keren et al., 2012; Soto et al., 2015), but not in the N. sylvestris CMSII mutant (Dutilleul et al., 2003b). In this work, the high heterogeneity between individual illuminated leaves prevented us from demonstrating any consistent differences in ROS content between ndufs8.1 ndufs8.2 and the wild type under both photoperiods, using NBT and diaminobenzidine staining of superoxide and hydrogen peroxide, respectively (Supplemental Fig. S15). Therefore, we used spin-trapping electron paramagnetic resonance (EPR) spectroscopy. A typical EPR comparison of SD and LD plants is shown in Figure 9A, indicating no difference between wild-type and mutant illuminated leaf discs (during 60 min) of SD plants. In both genotypes, the ROS content was low after a dark period (60 min), reflecting the absence of photosynthesis. In the wild type, the amount of apoplastic ROS diffusing to the medium was visibly lower in LD than in SD, as reported previously (Michelet and Krieger-Liszkay, 2012). By contrast, there was no clear photoperiod effect on ROS amounts in the ndufs8.1 ndufs8.2 mutant. We then determined leaf endogenous ROS content using luminol chemiluminescence, both at the middle of the light period and at the end of the dark period (Fig. 9B). In good agreement with the EPR results, the luminescence signal was significantly lower in LD than in SD in wild-type illuminated leaves, while it was not markedly different in the mutant. Therefore, although the ROS content of illuminated leaves was lower in ndufs8.1 ndufs8.2 than in the wild type under SD, they were similar in both genotypes under LD. As observed in the light, ROS contents of both genotypes were lower in LD than in SD at the end of the dark period. However, in contrast to the middle of the light period, dark ROS content tended to be higher in the mutant than in the wild type, maybe reflecting altered mitochondrial metabolism.
ROS content and expression levels of antioxidant enzymes in Col-0 and ndufs8.1 ndufs8.2 plants maintained under SD and LD conditions. A, Typical EPR spectroscopy results, determined on leaf discs sampled from SD12/LD12 plants, as described by Michelet and Krieger-Liszkay (2012). B, Total soluble ROS pools (hydrogen peroxide [H2O2] equivalents) detected by luminol chemiluminescence in leaves sampled from SD12/LD12 plants at the middle of the light period (left) and at the end of the dark period (right). Data are means + se of measurements on four to six different plants. Different letters indicate significant differences according to Student’s t test. FW, Fresh weight. C, RT-qPCR analysis of redox enzymes of leaves sampled from SD12/LD12 plants at the middle of the light period. ACT2 was used as a reference. Data are means + se of measurements on three to six different plants. Different letters indicate statistical differences according to Student’s t test.
We further examined transcript levels of oxidative stress markers in different cell compartments: mitochondrial alternative oxidase AOX1a (Saisho et al., 1997) and external NAD(P)H dehydrogenase NDB2 (Michalecka et al., 2003; Yoshida and Noguchi, 2009), chloroplastic ferritin FER1 (op den Camp et al., 2003), cytosolic ascorbate peroxidase APX1 (Koussevitzky et al., 2008), and peroxisomal catalase CAT2 (Noctor et al., 2007). In illuminated SD leaves, the expression levels of antioxidant enzymes were not higher in the ndufs8.1 ndufs8.2 mutant than in the wild type, and they were increased under LD in both genotypes, except for CAT2 mRNAs, which were increased only in the wild type (Fig. 9C). Although NDB2 transcripts accumulated proportionally more in the mutant than in the wild type, AOX1a was similar in both genotypes, regardless of the photoperiod. This was unexpected, since the AOX protein was found to be more abundant in the mutant under greenhouse conditions (Fig. 3C). In order to examine whether AOX1a transcript accumulation might respond to light intensity, SD and LD plants were transferred to 200 µmol m−2 s−1 PAR for 3 d. Under these conditions, AOX1a transcripts appeared to be markedly higher in the mutant than in the wild type in LD, while CAT2 levels were less abundant (Supplemental Fig. S16A), showing a crossed effect of illumination and photoperiod. When plants were grown in the greenhouse in LD, AOX1a and NDB2 mRNAs also were more abundant in the mutant than in the wild type at the middle of the light period, in contrast to other redox markers (Supplemental Fig. S16B). In addition, transcript levels of AOX1a, NDB2, APX1, and BONZOI-ASSOCIATED PROTEIN1, a marker of oxidative stress in the chloroplastic compartment (op den Camp et al., 2003), were considerably lower at the end of the dark period than at the middle of the light period in both genotypes (Supplemental Fig. S16B).
Previous work has shown that the redox homeostasis of CMSII plants in the LD condition was associated with the activation of the enzymatic antioxidant system (Dutilleul et al., 2003b). Similarly, we found higher detoxification activities in ndufs8.1 ndufs8.2 than in wild-type plants under both photoperiods (Supplemental Fig. S17). Both CAT (the major hydrogen peroxide-detoxifying enzyme in plant leaves; Willekens et al., 1997) and nonchloroplastic ascorbate peroxidase (i.e. cytoplasmic cAPX) activities were significantly higher in the mutant under SD, whereas glutathione reductase (GR; a key enzyme of the ascorbate-glutathione cycle; Foyer and Noctor, 2011) was higher under LD. In both genotypes, antioxidant activities were hardly affected by photoperiod, except cAPX, which was clearly inhibited under LD. In addition, the mutant had higher levels of anthocyanins (involved in redox control in Arabidopsis; Vanderauwera et al., 2005) than in the wild type under both photoperiods, especially during the first 3 d following transfer from SD to LD (Supplemental Fig. S17). Hence, our results indicate that the ndufs8.1 ndufs8.2 mutant does not exhibit a general oxidative stress in either SD or LD, likely because of the induction of antioxidant activities. However, the photoperiod and light/dark responses appeared to be altered, possibly reflecting some differences in the subcellular distribution of ROS, in particular mitochondrial ROS accumulation.
Photoperiod Differentially Modulates Soluble Antioxidants and Redox Cofactors in Wild-Type, ndufs8.1 ndufs8.2, and ndufs4 Plants
In addition to the high activity of antioxidant enzymes in CMSII (Dutilleul et al., 2003b; Vidal et al., 2007), CI mutants contain more soluble redox buffers than the wild type in the LD condition (Dutilleul et al., 2005; Kühn et al., 2015). Here, we found that total leaf glutathione (reduced and oxidized forms) and ascorbate (reduced and oxidized forms) were higher in both ndufs8.1 ndufs8.2 and ndufs4 mutants than in the wild type under both SD and LD, accompanied by slightly higher oxidation (Fig. 10A). Noticeably, levels of both redox buffers were higher in ndufs4 than in ndufs8.1 ndufs8.2 in SD. Transfer to LD resulted in a 30% glutathione increase in the wild type and ndufs8.1 ndufs8.2, whereas levels were unchanged in ndufs4. No significant differences in ascorbate content between SD and LD conditions were observed independently of genotype.
Pools of foliar redox buffers and cofactors in Col-0, ndufs8.1 ndufs8.2, and ndufs4 plants. Leaves were sampled from SD6/LD6 plants at the middle of the day period. Black columns, Col-0; dark gray columns, ndufs8.1 ndufs8.2; white columns, ndufs4. A, Redox buffers. Left, Total leaf content of reduced and oxidized (top, light gray) forms of glutathione (nmol g−1 fresh weight [FW]) and ascorbate (µmol g−1 fresh weight). Redox state (% of oxidized forms) is indicated above the histograms. Data are means + se from at least four extracts from different plants. Asterisks indicate significant differences (P < 0.05) between wild-type and mutant total pools according to Student’s t test; circles indicate significant differences between SD and LD values for the same genotype. Right, LD-SD ratios in the three genotypes. B, Redox cofactors. Left, Total leaf content of oxidized and reduced (top, light gray) forms of NAD(H) and NAD(P)H (nmol g−1 fresh weight). NAD(P)+-NAD(P)H ratios are indicated above the histograms. Results are means + se of eight extracts from different plants. Asterisks indicate significant differences (P < 0.05) between wild-type and mutant total pools according to Student’s t test; the circle indicates a significant difference between SD and LD values for the same genotype. Right, LD-SD ratios in the three genotypes.
In contrast to redox buffers, NAD(H) content was higher in ndufs8.1 ndufs8.2 than in the wild type under the SD condition only, whereas ndufs4 displayed higher contents under both photoperiods (Fig. 10B). A clear buildup in total leaf NAD(H) was observed under LD in the wild type but not in the mutants. Under both photoperiods, the NAD pool was more reduced (larger NADH-NAD+ ratio) in the mutants than in Col-0. Interestingly, mitochondrial NAD+ levels determined in Percoll-purified leaf mitochondria were not significantly affected by the photoperiod and were higher in the mutant than in the wild type under both SD and LD (Fig. 11A). Total leaf NADP(H) contents also were higher in the mutants than in the wild type under SD, and a slight photoperiod effect was observed in the wild type only (Fig. 10B).
Pools of mitochondrial NAD+ and transcriptional analysis of NAD+ biosynthetic genes. In all experiments, leaves were sampled from SD12/LD12 Col-0 and ndufs8.1 ndufs8.2 plants at the middle of the day period. A, NAD+ (nmol g−1 fresh weight [FW]) contents of mitochondrial preparations. Results are means + se of three extracts from different plants. B, RT-qPCR analyses of NAD+ biosynthetic genes were carried out using ACT2 as a reference. Results are means + se of at least six extracts from different plants. Different letters indicate significant differences according to Student’s t test.
In order to determine whether NAD+ contents might be regulated at the transcriptional level, we performed RT-qPCR analyses of genes involved in either de novo NAD+ biosynthesis or NAD+ recycling (Noctor et al., 2006): Asp oxidase (AO; the first enzyme of the de novo pathway in plants; Katoh et al., 2006), quinolinate synthase (QS), quinolinate phosphoribosyltransferase (QPT), nicotinate mononucleotide adenylyltransferase (NaMNAT), NAD synthetase (NADS), and nicotinate phosphoribosyltranferases (NaPT2/4). We also examined the expression of poly-ADP-ribose polymerase2 (PARP2) that uses NAD+ as a substrate in stress responses (Schreiber et al., 2006). There were considerable differences in gene expression between genotypes and photoperiods (Fig. 11B). Under SD, the high leaf NAD+ content of the mutant (compared with the wild type) was accompanied by an accumulation of AO, QPT, NADS, NaPT2/4, and PARP2 transcripts. In Col-0, NADS, QS, NaMNAT, and NaPT2/4 transcripts were more abundant in LD than in SD, thereby matching the higher NAD+ content. In contrast, AO, NADS, QPT, NaPT2/4, and PARP2 transcript levels were lower in the mutant in LD than in SD.
Hence, the loss of CI activity in both mutants is associated with (1) the accumulation of redox cofactors under SD conditions and (2) impaired redox acclimation to the LD condition.
DISCUSSION
Lack of the NDUFS8 Subunit Results in Holo-CI Misassembly and Remodeling of the Respiratory Chain
The NDUFS8 CI subunit is highly conserved from bacteria to eukaryotes, including plants (Klodmann et al., 2010). It is located in the Q module and binds two 2S-4Fe clusters involved in electron transfer to ubiquinone (Efremov et al., 2010). NDUFS8 was found to be required for holoenzyme assembly and activity in bacteria and humans (Procaccio and Wallace, 2004). In contrast to other species investigated (except Brassica rapa), Arabidopsis NDUFS8 is encoded by two genes, NDUFS8.1 and NDUFS8.2 (Fig. 1A) expressed in different tissues (Schmid et al., 2005; Qin et al., 2009), but their respective roles have not been established so far. We found that NDUFS8.2 is about 2.5 times more expressed than NDUFS8.1 in Col-0 seedlings and rosette leaves (Fig. 1C). Despite the lack of visible compensation at the transcript level, neither single mutant showed any apparent difference in CI assembly/activity (Fig. 2), indicating that posttranscriptional/translational controls occurred or, alternatively, that NDUFS8 amount was not limiting. These results indicate that the two NDUFS8 subunits have similar roles in CI assembly, forming complexes able to display normal NADH dehydrogenase activity. Interestingly, although no phenotypic alterations were observed in single mutants grown under greenhouse conditions, in vitro-germinated seedlings display a slightly reduced growth rate, suggesting that the presence of both subunits is beneficial at some development stages, in particular before autotrophy. Further investigations are required to establish potential differences in the composition of the CI complex in different tissues and environments.
In contrast to the single mutants, the ndufs8.1 ndufs8.2 double mutant has undetectable levels of both transcripts (Fig. 1C) and lacks detectable holo-CI assembly/activity. This clearly indicates the absence (or only residual amounts) of the NDUFS8 subunit in the double mutant and its central role in holo-CI assembly in plants. Although traces of holo-CI could not be detected using immunochemistry using anti-NAD9- and anti-CA2 antiserum or NDH/NBT in-gel assays in ndufs8.1 ndufs8.2, very low levels of polypeptides potentially corresponding to CI subunits can be distinguished using two-dimensional BN/SDS-PAGE (Supplemental Fig. S1), suggesting that substoichiometric amounts of the complex might indeed be present. Furthermore, subcomplexes might be assembled, as reported for many CI mutants lacking either peripheral or membrane subunits in plants (Karpova and Newton, 1999; Brangeon et al., 2000; Pineau et al., 2008; Meyer et al., 2009; Kühn et al., 2015; Soto et al., 2015) and humans (Vogel et al., 2007). A specific mitochondrial CA signal was as abundant in ndufs8.1 ndufs8 than in Col-0 mitochondrial proteins (Fig. 3C), suggesting that subcomplexes associated with the CA domain (possibly assembly intermediates; Wang et al., 2012) might be present in this mutant, as reported previously in ndufs4 (Kühn et al., 2015).
Besides potential traces of holo-CI, the induction of alternative respiratory enzymes might be crucial to sustain respiration in ndufs8.1 ndufs8.2. In addition to complex II (succinate dehydrogenase), nonphosphorylating alternative NAD(P)H dehydrogenases (type II dehydrogenases), located on the inner and outer surfaces of the inner mitochondrial membrane, are essential for plant growth and metabolism (Liu et al., 2009; Wallström et al., 2014). Their activity depends on plant metabolic status, such as NADH and Ca2+ concentration (Rasmusson et al., 2008). The stimulation of both internal and external enzymes was reported in the N. sylvestris CMSII and NMS1 mutants (Sabar et al., 2000), and induction of external enzymes was reported in maize NCS2 (Marienfeld and Newton, 1994) and in ndufs4 (Meyer et al., 2009). Similar to CMSII and ndufs4 mutants (Meyer et al., 2009; Kühn et al., 2015), total respiration measured either as oxygen consumption (Fig. 3A) or as CO2 emission (Fig. 8A) was not altered in ndufs8.1 ndufs8.2 LD leaves, indicating that the electron flux through alternative NAD(P)H dehydrogenase(s) was stimulated in the mutant. Consistently, we found an accumulation of mitochondrial NAD(H) in ndufs8.1 ndufs8.2 (Fig. 11A), as would be expected considering the low affinity for NADH of these enzymes (Møller, 2001). Also, as reported for other CI mutants, including ndufs4 (Sabar et al., 2000; Karpova et al., 2002; Meyer et al., 2009; Keren et al., 2012; Cohen et al., 2014), increased AOX protein content and capacity were observed in ndufs8.1 ndufs8.2 (Fig. 3, B and C). Interestingly, AOX accumulation was accompanied by reduced aconitase transcript levels (Supplemental Fig. S13) and increased content in citrate (Figs. 6 and 7; Supplemental Fig. S9), a well-known AOX activator (Vanlerberghe and McIntosh, 1996). High electron partitioning to AOX could have allowed the reoxidation of excess NADH in the mitochondrial matrix caused by CI impairment, while the COX pathway is essentially controlled by ATP (Florez-Sarasa et al., 2007). Here, since alternative pathways are nonphosphorylating (not coupled to H+ translocation), the ATP yield of mitochondrial electron transport is expected to be reduced in ndufs8.1 ndufs8.2, as reported in CMSII (Sabar et al., 2000) and ndufs4 (Meyer et al., 2009). In summary, there is an induction of alternative respiratory pathways in ndufs8.1 ndufs8.2 similar to most CI mutants investigated so far. However, in contrast to CMSII (Priault et al., 2007; Vidal et al., 2007), the in vivo electron partition to AOX is increased in ndufs8.1 ndufs8.2.
The Large Reduction of CI Activity Results in Similar Phenotypes in ndufs8.1 ndufs8.2 and ndufs4 Mutants Grown under the LD Condition
Plant CI mutants have different phenotypes, from mild (growth retardation only, as in ndufs4) to severe (morphological defects, as in N. sylvestris CMSII and NMS1 and maize ncs2) or even near lethal (nMat and ndufv1; see introduction). Here, we found that the ndufs8.1 ndufs8.2 double mutant displays a mild phenotype, without phenotypic alterations at the adult stage in flowering conditions (LD; Supplemental Fig. S3). The slow-growing phenotype is clearly related to holo-CI highly reduced contents in the double mutant, as both single mutants had CI assembly/activity and growth rates identical to the wild type. This also indicates the absence of unidentified insertions in the mutants, at least insertions with phenotypic consequences. Despite very slight differences in germination (+30%) and seedling growth rates (+20%), ndufs8.1 ndufs8.2 plants were similar to ndufs4 plants at subsequent developmental stages (Supplemental Figs. S3 and S4). The mild phenotype of the ndufs4 mutant has been suggested to result from persisting trace amounts of CI activity (Kühn et al., 2015). Although we were unable to detect an NDH/NBT signal around 1 MD in ndufs8.1 ndufs8.2 BN/PAGE (Fig. 2), low levels of CI activity cannot be excluded, since substoichiometric amounts of putative CI subunits are detected by two-dimensional BN/SDS-PAGE (Supplemental Fig. S1). Nevertheless, it is not possible to estimate accurately possible differences in substoichiometric holo-CI levels between ndufs8.1 ndufs8.2 and ndufs4 mutants due to the limited sensitivity of the detection methods (discussed in Keren et al., 2012).
As for other CI mutants, the lower efficiency of the respiratory chain (discussed above) might explain the slow-growth phenotype of ndufs8.1 ndufs8.2 LD plants. Similar to nMat, ndufs4, and ndufv1 mutants (Keren et al., 2012; Kühn et al., 2015), Suc supplementation improves the in vitro development of ndufs8.1 ndufs8.2 seedlings (Supplemental Fig. S3D), suggesting that photosynthetic ATP might compensate for lower mitochondrial ATP production. Nevertheless, as reported previously for CMSII (Szal et al., 2008; Djebbar et al., 2012) and ndufs4 (Meyer et al., 2009), total leaf ATP of illuminated leaves was not affected in ndufs8.1 ndufs8.2 (Fig. 3B), indicating either the activation of alternative mechanisms of ATP production as substrate-level phosphorylation at the level of mitochondrial enzymes, as proposed by Kühn et al. (2015), or reduced rates of cellular processes resulting in a lower ATP consumption. Regardless of the exact specific mechanism(s) involved, the growth retardation phenotype of CI mutants in the LD condition is unlikely to be related to energy limitation.
Oxidative stress also might be involved in the reduced growth of CI mutants. Accordingly, ROS accumulation has been reported to occur in several Arabidopsis CI mutants, including ndufs4 (Meyer et al., 2009), in possible relation to increased mitochondrial ROS generation by the remodeled respiratory chain. High levels of AOX proteins and capacities in all CI mutants investigated so far are in line with this hypothesis. However, like CMSII and NMS1 mutants (Dutilleul et al., 2003b), ROS accumulation could not be observed in illuminated leaves of ndufs8.1 ndufs8.2 using various methods (Fig. 9; Supplemental Fig. S15), likely due to the overactivation of antioxidant enzymes (Supplemental Fig. S17). Therefore, a marked oxidative stress is unlikely to have impeded growth rates under LD in this mutant, although such an effect cannot be excluded in ndufs4.
Metabolic perturbations associated with respiration reorchestration might account for the slow-growth phenotypes of ndufs8.1 ndufs8.2 and ndufs4 mutants. In fact, mitochondria are known to play a key role in the nitrogen/carbon balance (Foyer et al., 2011), which is affected in CMSII, with a marked increase in amino acids plausibly caused by high NAD(H) content (Dutilleul et al., 2005; Hager et al., 2010; Djebbar et al., 2012). Similarly, we found here that, despite lower values for parameters associated with nitrate assimilation (i.e. NR transcripts, NR activity, and NO3− content), amino acids accumulated in both mutants, with the notable exception of Asp (which appeared to be decreased except at the end of the night; Fig. 6; Supplemental Fig. S9). However, we did not observe a significant NAD(H) buildup in ndufs8.1 ndufs8.2 leaves compared with the wild type in LD (Fig. 10B), suggesting that an increase in NAD(H) content per se is not the cause of amino acid accumulation in the mutants. Rather, amino acid accumulation clearly paralleled that of carbon skeletons generated by tricarboxylic acid cycle activity (citrate, succinate, fumarate, and malate) in both mutants. This suggests that, in the light, when a partial tricarboxylic acid cycle is believed to operate (Tcherkez et al., 2009; Sweetlove et al., 2010), the metabolism in mutants is associated with (1) a stimulation of both cytosolic and mitochondrial malate dehydrogenases (whereby oxaloacetate is reduced rapidly to malate) because of over-NADH(H) (Heldt et al., 2005; Tomaz et al., 2010) and (2) an increased phosphoenolpyruvate carboxylase activity (O’Leary et al., 2011), as reported in CMSII mutants (Dutilleul et al., 2005), at the expense of Asp accumulation. In the night, the operation of the full (cyclic) tricarboxylic acid cycle, and thus the lower reduction of oxaloacetate to malate, would account for the high Asp observed in the ndufs8.1 ndufs8.2 mutant.
Hence, our results show a similar impact of reduced CI activity in both ndufs8.1 ndufs8.2 and ndufs4 mutants at developmental, physiological, and metabolic levels in relation to NAD(H) overreduction.
The Growth Retardation Phenotype Is More Pronounced in ndufs4 Than in ndufs8.1 ndufs8.2 Plants Grown in the SD Condition, in Possible Relation to Higher Oxidative Stress
All plant CI mutants characterized so far display growth retardation in LD, associated with various physiological and metabolic alterations, but their phenotype in SD has never been documented. Here, we show that ndufs8.1 ndufs8.2 and ndufs4 plants had a retarded growth phenotype under both LD and SD conditions (Supplemental Fig. S4). However, in contrast to LD, the retardation phenotype was more marked in ndufs4 than in ndufs8.1 ndufs8.2 plants under SD, sometimes associated with necrotic spots. Although the accumulation of total soluble redox buffers, glutathione and ascorbate, was observed in mutants under both SD and LD, redox buffers were in significantly higher amounts in ndufs4 compared with ndufs8.1 ndufs8.2 under SD (Fig. 10A), strongly suggesting enhanced oxidative stress. This effect might have originated from a less activated mitochondrial antioxidant system in ndufs4 compared with ndufs8.1 ndufs8.2. In fact, in contrast to ndufs4 (Meyer et al., 2009), NDB2 and AOX1a transcripts accumulated in ndufs8.1 ndufs8.2 either under our standard illumination condition (100 µmol photons m−2 s−1; Fig. 9C) or in high light (Supplemental Fig. S16). Redox differences between ndufs4 and ndufs8.1 ndufs8.2 might result from the nature and/or location of the lacking CI subunit. In ndufs4, a large accumulation of the CA domain, which was proposed to be involved in plant CO2 transport (Zabaleta et al., 2012) and carbon metabolism (Soto et al., 2015; Fromm et al., 2016), might result in increased mitochondrial ROS production, as reported in animals (Price et al., 2012). NDUFS4 and NDUFS7 are located in different CI modules (N and Q, respectively), and in human cell lines, mitochondrial metabolism was differentially affected in mutants affected in the two modules, in relation to the differential accumulation of assembly intermediates (Leman et al., 2015).
Untargeted metabolomics showed that the reduced growth of ndufs4 compared with ndufs8.1 ndufs8.2 SD plants was accompanied by much larger differences in global metabolism under SD than under LD (Fig. 5), in particular by higher levels of many amino acids (Fig. 6), as reported previously for low-nitrogen-adapted plants (Tschoep et al., 2009). It is interesting that the metabolomics signature of ndufs4 SD leaves has some similarities to that of plants under anoxic stress (de Sousa and Sodek, 2002; Sweetlove et al., 2010). Similarity to the anoxic response also can be found in the citrate accumulation, low level of aconitase transcripts, and increased AOX capacity in ndufs8.1 ndufs8.2 (Figs. 3 and 6; Supplemental Fig. S13), as shown by Gupta et al. (2012). Also, the CMSII mutant displays an up-regulation of fermentation pathways (Shah et al., 2013).
Overall results indicate that redox state is more compromised in ndufs4 in the SD than in the LD condition, possibly resulting in the redirection of tricarboxylic acid fluxes toward amino acid synthesis at the expense of oxidative phosphorylation, thereby explaining the lower growth rates.
Growth and Metabolic Acclimation to LD Depend on CI Activity
The rapid growth acceleration observed in Col-0 plants after their transfer to LD (already visible at day 3; Fig. 4) was clearly reduced in the ndufs8.1 ndufs8.2 mutant. Moreover, physiological changes associated with the SD-to-LD transition in the wild type (such as increased leaf mass area, palisade cell width, stomatal conductance, carboxylation efficiency, dark respiration, and total ATP and decreases in nitrogen assimilation and total ROS levels) were not observed (or were highly reduced) in the mutant (Fig. 8; Supplemental Fig. S5), thereby showing an impaired acclimation to LD. Despite marked growth differences between ndufs8.1 ndufs8.2 and ndufs4 under SD conditions, the growth response of both mutants, taken at a similar developmental stage (nine-leaf stage), to the SD-to-LD transition was comparably impaired compared with the wild type (Fig. 4). Untargeted metabolomics (HILIC-qTOF-MS) and GC-time of flight analyses confirmed that metabolism was less affected in both mutants than in the wild type upon SD-to-LD transfer (Fig. 5). The differential photoperiod effect on tricarboxylic acid intermediates, amino acids, and redox buffers in mutants (Figs. 6 and 10A) likely reflected their different metabolic and redox states under SD (as discussed above). That is, it is unlikely that the altered content in respiratory metabolites was the cause of the impaired photoperiod acclimation.
Rather, we found a clear correlation between the lack of NAD(H) accumulation and impaired LD acclimation in both ndufs8.1 ndufs8.2 and ndufs4 mutants. In fact, in contrast to the wild type, NAD(H) contents (1) were high under the SD condition, as would be anticipated considering the high Km(NADH) values of alternative NADH dehydrogenases, and (2) did not increase significantly in both mutants under LD (Fig. 10B). Presumably, NAD+ accumulation in wild-type plants transferred to LD came from a complex set of interactions between biosynthesis, oxidation by the mitochondrial electron transport chain (higher dark respiration), and degradation. NADH+ consumption under LD is suggested by the decreased levels of PARP2 transcripts (Fig. 11B). Also, in ndufs8.1 ndufs8.2, there was higher expression of several transcripts encoding NAD+ biosynthesis enzymes under SD and a clear depressing effect of LD, suggesting the repression of NAD+ neosynthetic/recycling pathways under LD.
Overall, our data show that the physiological and metabolic reorchestration accompanying LD acclimation in Arabidopsis is similarly compromised in both ndufs8.1 ndufs8.2 and ndufs4 mutants, thereby strongly suggesting that it results from reduced CI activity.
Signaling Mechanisms Possibly Involved in the Impaired LD Acclimation of CI Mutants
Changes in the composition of mitochondrial enzymes in CI mutants are expected to generate increased ROS by the mitochondrial electron transport chain. In fact, ROS were reported previously to accumulate in several plant CI mutants (Meyer et al., 2009; Keren et al., 2012) and in human cell lines lacking CI peripheral arm subunits (Verkaart et al., 2007; Miwa et al., 2014), where they were produced by matrix intermediates (Leman et al., 2015). Although total ROS levels were similar in ndufs8.1 ndufs8.2 and the wild type in the middle of the light period, they were higher in the mutant than in the wild type at the end of night period (Fig. 9B), thus suggesting increased generation of mitochondrial ROS by the respiratory chain. A general role of mitochondrial ROS in plant signaling is well documented (Huang et al., 2016). In humans, mitochondrial ROS were reported to induce pyruvate dehydrogenase activity by inactivating pyruvate dehydrogenase kinase2 (Hurd et al., 2012), thereby supporting a pathway through which mitochondrial ROS may regulate respiratory metabolism. Therefore, a specific signaling role for mitochondrial ROS in the photoperiod response is possible.
Also, glutathione levels and redox state have been reported to regulate daylength transcriptional responses in the cat2 mutants (Queval et al., 2007). Here, we found that, although antioxidants were higher in both mutants, they were differentially affected by photoperiod: glutathione and ascorbate accumulated under LD in the wild type and in ndufs8.1 ndufs8.2 but not in ndufs4, maybe because of the high content in SD (Fig. 10A). Therefore, antioxidants per se are unlikely to be involved in the impairment of the photoperiod response in CI mutants. In contrast, NAD(H) increased in LD in the wild type but not in both mutants, thereby correlating to growth rates in the three genotypes (see above). NAD+ is known to be involved in the control of Arabidopsis growth (Hashida et al., 2009). In addition to its well-established role in redox homeostasis and oxidative phosphorylation, a considerable body of evidence indicates that NAD+ is a crucial signaling molecule (Pétriacq et al., 2012, 2013, 2016; Sassone-Corsi, 2012) driving mitochondrial oxidative metabolism in mammals (Mouchiroud et al., 2013; Rey and Reddy, 2013). In animals, the inhibition of nicotinamide phosphoribosyl transferase, involved in nicotinamide recycling into NAD+, has been found to impede glycolysis and tricarboxylic acid cycle activity and lead to ATP depletion (Tan et al., 2013).
A slight but consistent overreduction of NAD(H) was observed in both Arabidopsis mutants (NAD+/NADH around 7%–8% under all conditions, compared with 10%–11% in the wild type; Supplemental Fig. S10B), as found previously in CMSII (Hager et al., 2010). Reduction levels of pyridine nucleotides are believed to exert a control on tricarboxylic acid fluxes (Igamberdiev and Gardeström, 2003; Araújo et al., 2012), for example by controlling mitochondrial MDH activity (Tomaz et al., 2010). This effect could explain the lack of significant respiratory increase in ndufs8.1 ndufs8.2 under LD (Fig. 8A). Overreduction of the NAD pool in mutants might stem from the induction of alternative NAD(P)H dehydrogenase activities (Liu et al., 2009; Rasmusson and Wallström, 2010; Wallström et al., 2014) and/or the activation of redox shuttles (Shen et al., 2006). A redox effect may have then resulted in alterations of the circadian clock (Stangherlin and Reddy, 2013; Shim and Imaizumi, 2015). Indeed, NAD+ is a clock-regulated metabolite in animals (Peek et al., 2013), and alteration of clock regulator expression during LD acclimation (Fig. 8D) and the dramatic increase in anthocyanins (Supplemental Fig. S17), which are regulated by cryptochromes (Ahmad et al., 1995), might reflect perturbation of the circadian clock in the ndufs8.1 ndufs8.2 mutant. This would be similar to what was reported in mutants affected in promoter regions of genes encoding mitochondrial proteins (Giraud et al., 2010). Interestingly, diurnal rhythms of AOX expression are highly perturbed in the CMSII mutant, possibly reflecting an impaired circadian cycle (Dutilleul et al., 2003b). Conversely, it has been shown that Arabidopsis mutants affected in circadian clock regulation exhibit dramatic changes in the mitochondrial metabolome, in particular tricarboxylic acid derivatives (Fukushima et al., 2009). Thus, we hypothesize that an impairment in the circadian clock occurred in the CI mutants examined here, thus leading to metabolic perturbations (Peek et al., 2012; Haydon et al., 2015), and limited growth enhancement under LD conditions. Nevertheless, the specific signaling mechanisms linking CI to circadian clock regulation need further investigation.
CONCLUSION
In summary, our study provides compelling evidence that photoperiod influences the phenotype of Arabidopsis CI mutants. Despite the lower growth of ndufs4 in SD (possibly related to a higher oxidative stress), ndufs8.1 ndufs8.2 and ndufs4 showed an impaired acclimation to LD after growth in SD, associated with a differential reorchestration of metabolism. We propose that highly reduced CI activity (i.e. oxidation of mitochondrial NADH) affects photoperiod acclimation via mitochondrial ROS and/or NAD(H) content and redox state.
MATERIALS AND METHODS
Plant Material
Arabidopsis (Arabidopsis thaliana) seeds of the SAIL_227F03 line carrying a T-DNA insertion in the At1g16700 gene (further referred to as NDUFS8.1) and of the SALK_062179 line carrying an insertion in the At1g79010 gene (further referred to as NDUFS8.2) were selected using the T-DNA Express Arabidopsis gene-mapping tool (http://signal.salk.edu/cgi-bin/tdnaexpress; Alonso et al., 2003) and ordered from the Nottingham Arabidopsis Stock Centre. Both mutant lines come from a T-DNA-mutagenized population in the Col-0 Arabidopsis ecotype. Primers used are listed in Supplemental Table S1. For initial characterization of single and double mutants, plants were grown in greenhouses under a 16-h photoperiod, at a day/night temperature regime of 23°C/17°C, under natural illumination supplemented with artificial lighting as described by Vidal et al. (2007). In addition, seeds of the ndufs4 mutant were provided by E.H. Meyer. For LD/SD comparisons, wild-type and ndufs8.1 ndufs8.2/ndufs4 seeds were sown at 1-week intervals, and seedlings were grown in controlled chambers in SD (8-h/16-h day/night) at 100 μmol m−2 s−1 PAR to the eight- to nine-leaf stage (referred to as SD0). After this time point, rosette plantlets were either transferred to LD (16 h/8 h) or maintained under SD conditions under the same illumination conditions. Li-Cor fluorometer measurements were performed on well-developed leaves of 2- to 3-month-old rosette plants (minimum 2 cm2 area). In all cases, experiments were carried out on mutant and Col-0 plants of similar development.
RNA Isolation and RT-PCR Analysis
Total RNA was prepared using the Trizol reagent (Invitrogen) following the manufacturer’s recommendations. RNA (1 µg) was treated with RQ1 RNase-Free DNase (Promega) and reverse transcribed using random hexamers and the SuperScript III first-strand kit (Invitrogen) following the manufacturer’s recommendations. The LightCycler 480 detection system (Roche Applied Science) was used to perform quantitative real-time PCR. Relative mRNA abundance was calculated using the comparative delta-cycle threshold method and normalized to the corresponding ACT2 (At3g18780) gene levels. The sequences of primers used in this study are listed in Supplemental Table S1.
CO2-Exchange Measurements
Responses of A to photon flux density (light curves) and internal CO2 molar fraction (A/Ci curves) performed under ambient (21%) oxygen content were measured on attached leaves with an open infrared gas analysis system equipped with a leaf chamber fluorometer (Li-Cor 6400-40; Li-Cor) as described by Priault et al. (2006a). Leaves were dark adapted for at least 30 min before determining dark respiration.
Oxygen Respiratory Measurements
Oxygen isotope discrimination experiments were performed as described by Florez-Sarasa et al. (2007) under controlled conditions (12-h/12-h photoperiod), except that plants were grown under 80 µmol m−2 s−1 and oxygen-isotope fractionation calculations were made as described by Guy et al. (1989) and Ribas-Carbo et al. (1995). The AOX pathway capacity was measured on leaf tissues as described by Florez-Sarasa et al. (2009).
Preparation of Crude Leaf and Root Membrane Extracts, BN Electrophoresis, Western-Blot Analyses, and Determination of In-Gel NADH Dehydrogenase Activity
Extraction of total and mitochondrial proteins, BN/PAGE, two-dimensional BN/SDS electrophoresis, determination of in-gel NADH dehydrogenase activity of CI, and protein silver staining were performed as described previously (Pineau et al., 2008). Gels were electroblotted onto nitrocellulose membranes for SDS-PAGE and onto polyvinylidene difluoride membranes for BN/PAGE. Immunodetections were performed using wheat (Triticum aestivum) anti-NAD9 antibody (a gift from J.M. Grienenberger), anti-CA2 antibody directed against the C-terminal half of the mitochondrial CA2 (At1g47260) from Arabidopsis (a gift from E. Zabaleta), and mouse monoclonal Sauromatum guttatum anti-AOX antibody (a gift from A.H. Millar). Immunosignals were visualized by enhanced chemiluminescence according to the manufacturer’s instructions (Roche Diagnostics).
Minipreparations of Percoll-Purified Leaf Mitochondria
One gram of fresh material was used to obtain crude leaf mitochondrial preparations, as described by Vidal et al. (2007). Mitochondria were further purified on a three-layer Percoll gradient (13%, 25%, and 45% [w/v] Percoll), performed in 2-mL Eppendorf tubes, using the protocol designed for the purification of Nicotiana sylvestris pollen mitochondria (De Paepe et al., 1993). The pyridine redox state could not be determined in mitochondrial extracts due to the oxidation of preparations during Percoll purification.
Rubisco Radioisotopic Assay
Total Rubisco activity was determined from the rate of 14CO2 incorporation into acid-stable compounds and subsequent liquid scintillation counting of 14C, according to Seemann and Sharkey (1986). Details are given in Supplemental Methods S1.
Antioxidant Activities
All enzymes were extracted from washed leaves in 50 mm potassium phosphate, pH 7.5, and measurements of activities of glycolate oxidase (EC 1.1.3.1), CAT (EC 1.11.1.6), APX (EC 1.11.1.11), and GR (EC 1.6.4.2) were as described by Streb et al. (1997). For APX activity, only the cytoplasmic form was determined.
Pigment Contents
Chlorophyll a/b contents were measured from 80% acetone extracts with extinction coefficients reported by Porra (2002).
Determination of Soluble Antioxidants in Leaf and Mitochondrial Extracts
Oxidized and reduced forms of glutathione, ascorbate, NAD, and NADP were measured in total leaf extracts or Percoll-purified mitochondrial preparations, by plate-reader assay, as described by Queval and Noctor (2007) and modified by Pétriacq et al. (2012). The pyridine redox state could not be determined in mitochondrial extracts due to their oxidation during Percoll purification (Vidal et al., 2007).
Untargeted Metabolomics by HILIC-MS
Polar metabolites were detected by HILIC coupled to qTOF-MS using a method from Paglia et al. (2012) that was modified by Pétriacq et al. (2016). Details are given in Supplemental Methods S1.
Targeted Metabolic Profiling by GC-MS
GC coupled to time-of-flight MS profiling and quantitative analysis of amino acids by HPLC were performed as described in detail by Noctor et al. (2007) and Tcherkez et al. (2010) using the isotopic facility structure of the Plateforme Metabolisme Metabolome. Details are given in Supplemental Methods S1.
ATP contents were quantified using the ENLITEN ATP Assay System Bioluminescence Detection Kit (Promega) using the manufacturer’s instructions.
ROS Determination
In situ ROS detection was performed using NBT and diaminobenzidine stains, detecting superoxide and hydrogen peroxide, respectively, as described by Dutilleul et al. (2003b). Room-temperature spin-trapping EPR spectroscopy was carried out to measure apoplastic ROS, as described by Michelet and Krieger-Liszkay (2012). Details are given in Supplemental Methods S1. Luminol chemiluminescence was performed on leaves as described by Pétriacq et al. (2016). Results were calibrated using hydrogen peroxide as the ROS source.
NR Activity and Nitrate Determination
NR activity was determined as described by Fresneau et al. (2007) on desalted extracts purified with a NAP-5 column (Sephadex G-25; GE Healthcare). Nitrate ion contents were determined as described by Cataldo et al. (1975). Details are given in Supplemental Methods S1.
Anthocyanin Determination
Anthocyanin content was determined according to Vanderauwera et al. (2005) and expressed as optical density at 530 nm mg−1 fresh weight.
Electron Microscopy
Leaf samples were processed as described by Hawes and Satiat-Jeunemaitre (2001). Details are given in Supplemental Methods S1.
Statistical Methods
Significant differences (P < 0.05) were calculated using Student’s t test.
Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers At1g16700 and At1g79010 and in Supplemental Table S1.
Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Resolution by two-dimensional BN/SDS-PAGE of root membrane proteins of Col-0 and the ndufs8.1 ndufs8.2 double mutant.
Supplemental Figure S2. Electron microscopy of mitochondria in Col-0 and ndufs8.1 ndufs8.2 leaves.
Supplemental Figure S3. Phenotypes of ndufs8.1 and ndufs8.2 single mutants, the ndufs8.1 ndufs8.2 double mutant, and the ndufs4 mutant.
Supplemental Figure S4. Col-0, ndufs8.1 ndufs8.2, and ndufs4 plants grown in SD and LD conditions.
Supplemental Figure S5. Dimensions of leaf cells of Col-0 and ndufs8.1 ndufs8.2 plants in SD and LD conditions.
Supplemental Figure S6. Micrographs of chloroplast ultrastructure in Col-0 and ndufs8.1 ndufs8.2 plants.
Supplemental Figure S7. Changes in the contents of sugars and tricarboxylic acid derivatives in Col-0 and ndufs8.1 ndufs8.2 plants during the day/night cycle analyzed using GC-MS.
Supplemental Figure S8. Changes in the contents of amino acids in Col-0 and ndufs8.1 ndufs8.2 plants during the day/night cycle analyzed by GC-MS (part 1).
Supplemental Figure S9. Changes in the contents of amino acids in Col-0 and ndufs8.1 ndufs8.2 plants during the day/night cycle analyzed by GC-MS (part 2).
Supplemental Figure S10. Heat map and hierarchical clustering (cosine correlation) of metabolites found to be significant with respect to photoperiod (SD/LD) in a two-way ANOVA (GC-MS metabolomics).
Supplemental Figure S11. Metabolomic pattern in Col-0 and ndufs8.1 ndufs8.2 plants and photoperiod-genotype interaction.
Supplemental Figure S12. HPLC quantitation of amino acids in Col-0 and ndufs8.1 ndufs8.2 leaves.
Supplemental Figure S13. Transcriptional analysis of tricarboxylic acid-related enzymes in Col-0 and ndufs8.1 ndufs8.2 leaves.
Supplemental Figure S14. Gas-exchange measurements of Col-0 and ndufs8.1 ndufs8.2 plants maintained under SD and LD conditions.
Supplemental Figure S15. In situ detection of leaf superoxide and hydrogen peroxide in Col-0 and ndufs8.1 ndufs8.2 plants grown under LD and SD conditions.
Supplemental Figure S16. Transcriptional analysis of redox enzymes in Col-0 and ndufs8.1 ndufs8.2 plants under various illumination conditions.
Supplemental Figure S17. Cellular antioxidant activities and anthocyanin contents of Col-0 and ndufs8.1 ndufs8.2 mutant plants maintained under SD and LD conditions.
Supplemental Table S1. Sequences of primers and accession numbers of genes used in this study.
Supplemental Methods S1. Supplemental Materials and Methods.
Acknowledgments
We thank Béatrice Satiat Jeunemaître for help in ME analyses; Sophie Chailloux, Dorothée Thominet, and Ouardia Layoune for excellent technical help; A.H. Millar for the anti-AOX antibody, J.M. Grienenberger for the wheat anti-NAD9 antibody, and E. Zabaleta for the Arabidopsis anti-CA2 antibody; and E.H. Meyer for the gift of ndufs4 seeds.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Bertrand Gakière (bertrand.gakiere{at}u-psud.fr).
R.D.P. conceived the original screening and research plans; B.G., G.T., R.B., and G.C. supervised the experiments; P.P., B.G., L.G., P.B., F.Gu L.d.B., J.H., I.F.-S., C.L., G.Q., J.M., T.R., C.F., M.L.-S., S.B., A.D., B.P., and A.K.-L. performed most of the experiments; F.Gi. and C.M. provided technical assistance; M.R.-C. and G.T. designed the experiments and analyzed the data; R.D.P. conceived the project and wrote the article with contributions of all the authors; B.G., G.C., and G.T. supervised and complemented the writing.
↵1 This work was supported by the Université Paris-Sud, the Centre National de la Recherche Scientifique, the Commissariat à l’Energie Atomique Saclay, the Institut Jean Pierre Bourgin, the Spanish Ministry of Science and Innovation, and the French Agence Nationale de la Recherche (Génoplante program grant no. GNP0508 and project Jeunes Chercheurs under contract no. 12–0001–01).
↵2 Present address: Labogena, Domaine de Vilvert, 78352 Jouy-en-Josas cedex, France.
↵3 Present address: Central Metabolism Group, Molecular Physiology Department, Max Planck Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476 Potsdam-Golm, Germany.
↵4 Present address: Institut Sylvo-Pastorale, BP n 345, 8110 Tabarka, Tunisia.
↵5 Present address: LBMCE, Unité Mixte de Recherche 8226, Sorbonne Université, UPMC Université Paris 06, Centre National de la Recherche Scientifique, Institut de Biologie Physico-Chimique, 75005 Paris, France.
↵[OPEN] Articles can be viewed without a subscription.
Glossary
- ROS
- reactive oxygen species
- SD
- short day
- LD
- long day
- Col-0
- Columbia-0
- RT
- reverse transcription
- BN
- blue native
- NBT
- nitroblue tetrazolium
- PAR
- photosynthetically active radiation
- HILIC
- hydrophobic interaction liquid chromatography
- qTOF
- quadrupole time of flight
- MS
- mass spectrometry
- GC
- gas chromatography
- EPR
- electron paramagnetic resonance
- qPCR
- quantitative PCR
- Received September 28, 2016.
- Accepted November 16, 2016.
- Published November 16, 2016.
REFERENCES
