A redox-mediated modulation of stem bolting in transgenic Nicotiana sylvestris differentially expressing the external mitochondrial NADPH dehydrogenase.

Cytosolic NADPH can be directly oxidized by a calcium-dependent NADPH dehydrogenase, NDB1, present in the plant mitochondrial electron transport chain. However, little is known regarding the impact of modified cytosolic NADPH reduction levels on growth and metabolism. Nicotiana sylvestris plants overexpressing potato (Solanum tuberosum) NDB1 displayed early bolting, whereas sense suppression of the same gene led to delayed bolting, with consequential changes in flowering time. The phenotype was dependent on light irradiance but not linked to any change in biomass accumulation. Whereas the leaf NADPH/NADP(+) ratio was unaffected, the stem NADPH/NADP(+) ratio was altered following the genetic modification and strongly correlated with the bolting phenotype. Metabolic profiling of the stem showed that the NADP(H) change affected relatively few, albeit central, metabolites, including 2-oxoglutarate, glutamate, ascorbate, sugars, and hexose-phosphates. Consistent with the phenotype, the modified NDB1 level also affected the expression of putative floral meristem identity genes of the SQUAMOSA and LEAFY types. Further evidence for involvement of the NADPH redox in stem development was seen in the distinct decrease in the stem apex NADPH/NADP(+) ratio during bolting. Additionally, the potato NDB1 protein was specifically detected in mitochondria, and a survey of its abundance in major organs revealed that the highest levels are found in green stems. These results thus strongly suggest that NDB1 in the mitochondrial electron transport chain can, by modifying cell redox levels, specifically affect developmental processes.


INTRODUCTION
amounts varying significantly from WT in at least one transgenic line (Fig. 3). However, with the exception of citramalate and maltose, which were lowered in at least one overexpression line but not in S8, the changes in overexpressors and the suppression line were not consistent with the genetic modification. For example, shikimate was significantly increased in lines S2, S6 and S8.
In order to account also for plant-to-plant variation in the analysis, we determined Pearson coefficients and significances for the correlation of both the NADPH/NADP +ratio and the absolute NADPH level to the quantified metabolites. Fig. 4A displays the stem metabolites that correlated significantly to NADP(H) as well as those that correlated to these metabolites in turn. Eleven metabolites correlated to NADPH/NADP + and/or NADPH, including hexose phosphates and sugars, intermediates of the citric acid cycle and ascorbate metabolism, and amino acids. Consistent with the analysis of Fig. 3, positive correlation to glutamate and trehalose and negative to 2-oxoglutarate and ascorbate were observed. An obvious stem specific pattern seen was that, 2-oxoglutarate, galactonate and ascorbate correlated negatively to the NADPH/NADP + -ratio, but only to very few of the other metabolites analyzed.
Leaves displayed a correlation pattern highly different from stems (Fig. 4B).
Hexose phosphate, especially glucose-6-phosphate was positively correlated to NADP (H) in both organs. However, raffinose was positively correlated to NADP(H) in stem but negative in leaves, and the opposite pattern was seen for ascorbate. In leaves, 20 additional metabolites correlated to NADPH/NADP + and/or NADPH, including intermediates from both primary and intermediary metabolism. The results thus display that the integration of NADP(H) into the metabolic systems is completely different in stem and in leaf.

The overexpression and suppression of St-NDB1 induce up-and down-regulation of flowering-associated genes
To determine the mode of influence of the observed redox changes on bolting, we analyzed gene expression in stems from the transgenic plants by realtime RT-PCR. The transcript levels for St-NDB1 and Ns-NDB1 were consistent with previous results for leaves (Liu et al., 2008), confirming the overexpression of St-NDB1 in lines S2 and S6 as well as suppression of both genes in line S8 in stem tissue (Fig. 5). On searching of databases, we found two floral induction-associated MADS box genes in N. sylvestris (Jang et al., 1999;Jang et al., 2002;Gallego-Giraldo et al., 2007). The Ns-MADS2 transcript was 3-fold higher in the overexpressor lines (S2 and S6) than in the suppressor line (S8), with WT being intermediate (Fig. 5). A similar but weaker trend was observed for Ns-MADS1. We also used primers against Nt-NFL2, a homolog of Arabidopsis LEAFY present in N. tabacum but originating from the N. sylvestris parent (Kelly et al., 1995).
Transcript levels for NFL2 were lower in S8 as compared to the other lines. The expression of Ns-MADS1 and Ns-MADS2 was significantly correlated to the NADPH/NADP + -ratio (Fig. 5B). This suggests that a lower stem NADPH/NADP + -ratio promotes expression of floral phase transition genes, inducing the observed phenotype.
To investigate if any of the metabolites correlating to NADPH levels ( Fig. 4A) were likely to mediate the effect, we additionally performed Pearson analyses for these, which revealed correlation of Ns-MADS1 and Ns-MADS2 to D-xylose, but not to the other metabolites ( Fig. 5B).
Gibberellins (GAs) induce bolting and floral phase transitions in plants, including the rosette plant N. sylvestris (Lee and Zeevaart, 2005). Also, ascorbate and 2oxoglutarate, which were observed to vary in the stems of the transgenic lines, are cofactors in GA biosynthesis. We therefore analyzed gene expression with primers against N. tabacum GA2OX3 and GA2OX5, which are induced by GA in N. tabacum (Gallego-Giraldo et al., 2008). Both genes were unresponsive in the transgenic lines, except for a higher GA2OX3 expression in S8 (Fig. 5). This indicates that GA levels were unchanged in the overexpressors, but may be elevated in line S8, which would have partially counteracted the late bolting phenotype observed in S8. The relative unresponsiveness of GA-induced genes was consistent with that S8 bolted significantly later than S6 also when grown with the GA synthesis inhibitor paclobutrazol. For four plants, the average bolting time with paclobutrazol was 56.7±0.6 and 52.9±1.2 days for S8 and S6, respectively. Thus, the observed phenotypic change was not mediated by GA.

NADP(H) reduction varies temporally and spatially in the stem apex
To further investigate the importance of cellular NADPH reduction level during bolting, we compared the apical 10 mm parts of WT stems at different stages (Fig. 6A). A 3-fold decrease in the NADPH/NADP + -ratio was observed during the growth of the stem from 15 mm (i.e. just before bolting) to 60 mm length, whereas NADH/NAD + -ratio was unchanged. In the same developmental span the total amounts of NADP(H) in the apex increased, whereas the NAD(H) amount was little affected. The spatial distribution of nucleotides and reduction levels was investigated in 50-60 mm high stems that were dissected before extraction. NAD(H) total amount and reduction levels were relatively similar over the investigated length (Fig. 6B). However, the total amounts of NADP(H) were concentrated to the apex, whereas the segment 20-40 mm below the apex contained only a small part of the total NADP(H). In the same extracts, the NADPH/NADP + -ratio displayed somewhat higher values in the basal parts.

St-NDB1 is highly abundant in mitochondria purified from green stems
We wanted to investigate if the observed effects of NDB1 in the stem were mirrored in the expression pattern of the encoding gene. The St-NDB1 transcript is unresponsive to several treatments in potato leaves (Svensson and Rasmusson, 2001;Svensson et al., 2002;Geisler et al., 2004), and the Arabidopsis ortholog (At-NDB1; At4g28220) is stably expressed upon stress treatments of cell cultures (Clifton et al., 2005). Comparison of the Arabidopsis col-0 arrays present in Genevestigator (https://www.genevestigator.ethz.ch), revealed expression of At-NDB1 in most tissues, including the apical meristem. For 1357 arrays, the standard deviation of the signal was 35% of average signal intensity for the At-NDB1, but 101% for both the energy-bypass genes At-NDB2 and At-AOX1a. Thus, the At-NDB1 transcript appears relatively unresponsive over a wide range of stresses and conditions.
We previously reported the use in western blotting of an antiserum specifically detecting St-NDB1, but not N. sylvestris homologs (Michalecka et al., 2004), and which is suitable for analyzing the St-NDB1 protein distribution in its species of origin. Since a recent targeting analysis using a C-terminal peptide of the Arabidopsis NDB1 suggested that NDB1 may reside also in peroxisomes (Carrie et al., 2008), we also compared purified potato tuber mitochondria and peroxisomes (Fig. 7A). The peroxisomes had high catalase activities, as previously observed (Struglics et al., 1993), and no mitochondrial contamination. Antibodies against a glyoxysomal malate dehydrogenase detected a 37 kDa band in peroxisomes, and a larger band in mitochondria, consistent with previous observations (Gietl et al., 1996). However, even at film exposure times where the St-NDB1 antibodies gave a strong signal in mitochondria, no NDB1 signal could be detected in peroxisomes (Fig. 7A). We therefore conclude that St-NDB1 resides in mitochondria and not peroxisomes in potato, consistent with previous results using less specific antibodies against the conserved C-terminus of St-NDB1 (Rasmusson and Agius, 2001), and that the phenotype observed in this investigation was due to enzymatic changes in the mitochondrial electron transport chain.
To determine organ distribution of St-NDB1 in potato, mitochondria were purified in triplicates from seven potato organs, including sink and source tubers, etiolated and green stem, young and mature leaves, and flower buds. The mitochondria were analyzed by western blotting, comparing St-NDB1 to other ETC proteins (Fig. 7B). Consistent between preparations, St-NDB1 was most abundant in green stems, with lower levels being detectable in source and sink tubers and mature leaves. A weak band was repeatedly detected in young leaves, but not in etiolated stems and flower buds.
Immunoprobing was also carried out with antibodies against the alternative oxidase and the type II NAD(P)H dehydrogenase St-NDA1, the latter of which most likely also detects a second potato NDA protein (Svensson and Rasmusson, 2001;Svensson et al., 2002;Geisler et al., 2004). These immunosignals were for both proteins strongest in mature leaves, with the NDA1 signal showing a smaller difference between organs, possibly due to the detection of differently regulated NDA-type proteins in a single band. The uncoupling protein was relatively evenly distributed, but somewhat elevated in tubers and stems. The 78 kDa and NAD9 subunits of complex I, the proton-pumping NADH dehydrogenase of the ETC, were evenly distributed in mitochondria from different organs, consistent with their role in basic energy production. To estimate whether mitochondria were differently prevalent in the organs, we analyzed the copy number ratio of the mitochondrial NAD9 and the nuclear NDA1 gene by realtime PCR. The results indicate a higher mitochondrial DNA copy number per nuclear DNA in sink tubers and flower buds as compared to other organs. If normalized in this respect, the highest levels of St-NDB1 would be present in green stems and sink tubers, but the difference between green stems and leaves is unchanged. The results therefore strongly suggest that St-NDB1 levels are specifically elevated in green stem mitochondria, consistent with a specific function in stems. Also, the organ distribution of St-NDB1 deviates substantially from that of energy bypass pathways as well as proteins of the energy-conserving part of the potato ETC. We present here three consistent lines of evidence suggesting that the NDB1 external NADPH dehydrogenase is important for stem development and specifically accelerates bolting: (1) The NADPH/NADP + -ratio is lowered 3-fold during bolting and onset of stem elongation, (2) The overexpression and suppression of NDB1 accelerates and retards bolting, respectively, coupled to specific changes in stem NADPH/NADP + -ratio and expression of flowering induction-associated genes, and (3) The investigated St-NDB1 protein is specifically highly abundant in stem mitochondria of the transgene donor species, potato.

DISCUSSION
St-NDB1 is an external calcium-dependent NADPH dehydrogenase that is theoretically able to modulate the NADPH/NADP + -ratio in the cytosol irrespective of the cellular energy charge (Michalecka et al., 2004). Consistently, the N. sylvestris plants overexpressing St-NDB1 displayed decreased leaf NADPH/NADP + -ratios when grown at 200 µmol m -2 s -1 (Liu et al., 2008). The absence of such an effect in leaves at higher light ( Fig. 2), could be due to NDB1 being inactive (e.g. calcium availability being insufficient). Alternatively, a larger photon flux may increase export of chloroplast reductant via a triose phosphate shuttle (Krömer, 1995), compensating for an elevated mitochondrial NADPH oxidation. In contrast, the stem NADPH/NADP + -ratio varies between the genotypes, suggesting that NDB1 is active under these conditions and affecting the NADPH/NADP + -ratio beyond what other reactions can compensate for. The simplest interpretation of the phenotype is therefore that higher levels of external mitochondrial NADPH dehydrogenase will speed up a decrease in NADPH/NADP + -ratio that normally takes place in the WT apex during bolting.

Integration of NDB1 and NADP(H) in the cellular metabolic systems
phosphate dehydrogenase was attributed to an elevated glucose-6-phosphate dehydrogenase activity (Rius et al., 2006). The glucose-6-phosphate level is a marker for carbon status in Arabidopsis, and is located at the connection point between growth, respiration and carbon storage (Stitt et al., 2007). Thus, the correlation between glucose-6-phosphate and NADPH reduction level appears to be a general property rather than a specific one directly involved in the mediation of bolting.
Apart from the hexose phosphates, the correlation of NADPH to metabolites is highly different in stem and in leaf. Differences in individual metabolites may be consequences of the NADPH/NADP + -ratio being outside the response range of a critical enzyme in one organ, variations in isoenzymes expressed, or switches between metabolic pathways using the metabolites in question. Particularly marked differences between leaf and stem were seen for sugars. We further observed that the stem NADPH/NADP + -ratio specifically correlates (positively or negatively) to a small set of metabolites, three of which (2-oxoglutarate, galactonate and ascorbate), show little correlation to other metabolites in the data set. In contrast, the leaf NADPH/NADP + couple appears to interact with a large number of metabolites that represent numerous pathways (Fig. 4). This is consistent with NADP(H) having a regulatory role in stems (as suggested by the effect on bolting) whereas a general role in metabolic redox transfer would dominate in the leaves.
In stems, we observed a positive correlation of the NADPH/NADP + -ratio to glutamate and a negative one to 2-oxoglutarate and citrate (Fig. 3-4), metabolites that are central intermediates connecting the citric acid cycle and nitrogen assimilation. Thus, the glutamate/2-oxoglutarate redox pair responds to the NADPH reduction level. A potentially NADPH-mediated functional connection between the oxidative pentose phosphate pathway and glutamate synthase was reported for barley and pea roots (Esposito et al., 2003;Bowsher et al., 2007). However, an NADPH effect on the glutamate/2-oxoglutarate redox pair being mediated by an NADP(H)-glutamate dehydrogenase (Purnell et al., 2005) cannot presently be excluded. The similarity in response between 2-oxoglutarate and citrate may also be derived from an NADP(H) effect on the formation or removal of these metabolites, for example via the control of citric acid cycle proteins by thioredoxin, which is reductively activated by NADPH (Balmer et al., 2004). Overall, the observations here emphasize the metabolic differences between leaves and stem, and provide in vivo support for an influence of NADP(H) redox homeostasis on sugar metabolism, the pentose phosphate pathway and glutamate metabolism. metabolites, there is little insight into how the effects are mediated. Our investigation identifies a pattern of relatively few compounds correlated to NADPH levels in the stems of the transgenic lines. These are candidates for mediating, individually or in combination, the NADP(H)-imposed redox regulation of floral integrator genes and consequentially bolting. Especially, the correlation of MADS gene expression to xylose suggests an involvement of sugar metabolism, complementing the previously known sugar effects on shoot meristems (Francis and Halford, 2006). The changes in ascorbate content are also particularly interesting given that the terminal step of ascorbate biosynthesis appears to be physically (Millar et al., 2003) and functionally (Nunes-Nesi et al., 2005) linked to the ETC. Therefore, ascorbate synthesis may be influenced downstream of NDB1 in the electron transport path, whereas the majority of the other metabolic changes described here are most likely a direct consequence of the modified NADPH/NADP + ratio.
An option worth considering would be that NADPH and/or NADP + are directly sensed in the cell. This scenario is not without precedence since sensors for NADH/NAD + such as Streptomyces coelicolor Rex (Brekasis and Paget, 2003) and the human CtBP protein (Zhang et al., 2002), have been long identified. Furthermore, human HSCARG was recently reported to be a sensor for NADPH/NADP + , and via protein-binding to decrease NO formation in response to lowered NADPH/NADP + -ratios (Zhao et al., 2008).
When the results of these and our current study are considered together, it seems reasonable to postulate that an NADPH-sensing system also exists in plants. Identifying the exact mechanism linking the NADPH/NADP + redox couple will probably require considerably more research effort. However, irrespective of which of these models is ultimately correct the results we present here clearly demonstrate that manipulation of the mitochondrial external NADPH dehydrogenase St-NDB1 has profound effects on stem NADPH/NADP + -ratio, and as a consequence floral integrator genes and bolting time.
Thus, NADPH constitutes a connection between redox signaling and floral phase transitions, both of which integrate multiple influences from the plant environment (Bernier and Perilleux, 2005;Foyer and Noctor, 2005). male sterility traits involve chimeric genes in the mitochondrial genome that modify expression of flower organ identity genes, leading to conversions of floral organs without affecting the vegetative plant body (Linke and Börner, 2005;Carlsson et al., 2008). In a second class of genetic alterations, N. sylvestris and Zea mays genotypes lacking subunits of energy-conserving ETC complexes display retardations regarding growth, leaf shape, photosynthesis and chloroplast greening (Marienfeld and Newton, 1994;Gutierres et al., 1997;Karpova et al., 2002;Dutilleul et al., 2003;Noctor et al., 2004). Thirdly, repression of citric acid cycle proteins has resulted in variable effects. Examples in tomato include a decrease in stomatal function, photosynthesis and growth rate in fumarase antisense plants and an increase of photosynthesis and growth rate in plants suppressed for malate dehydrogenase (Nunes-Nesi et al., 2005;Nunes-Nesi et al., 2007). Arabidopsis suppressed for pyruvate dehydrogenase kinase displayed decreased vegetative growth and early flowering (Zou et al., 1999). In a fourth category, genetic modifications of energy bypass pathways in Arabidopsis have uncovered an importance of the uncoupling protein At-UCP1 for optimal photosynthesis and growth rate (Sweetlove et al., 2006), whilst the alternative oxidase gene At-AOX1a is essential for optimal growth at lower temperatures (Fiorani et al., 2005) and combined light and drought stress (Giraud et al., 2008). In summary, plants modified for enzymes of the mitochondrial respiratory pathways have previously displayed phenotypes of modified photosynthesis and growth, but not specific developmental changes without showing a modified biomass accumulation.
In combination, the specific bolting phenotype, the lack of an effect on the biomass accumulation seen here and at lower light (Liu et al., 2008) and the protein distribution in potato organs, suggest that St-NDB1 deviates in physiological role from ETC proteins of glass vial and 750 µL chloroform and 1500 µL water were added. The mixture was centrifuged for 15 min at 4000 rpm. The upper polar phase was dried in vacuo and stored at -80 °C. Metabolite analysis was performed by GC-TOF-MS as previously described (Lisec et al., 2006) and the relative metabolite contents were evaluated by comparison to libraries housed in the Golm Metabolome Database (Kopka et al., 2005;Schauer et al., 2005).

Transcript quantification
Total RNA isolation, DNase treatment, cDNA synthesis and realtime RT-PCR was carried out using the RNeasy Plant mini kit (Qiagen), DNase I (New England Biolabs) RevertAid H minus first strand cDNA synthesis kit (Fermentas) and GoTaq DNA polymerase (Promega), respectively, but otherwise as previously described (Svensson and Rasmusson, 2001;Svensson et al., 2002;Geisler et al., 2004). Primer pairs used were as

Data treatment
Comparisons of values for significant differences were made using Student's t-test in       Abbreviations: 3CQ, cis-3-caffeolyl quinate; DHAsc, dehydroascorbate; Fructose-6-P, fructose-6-phosphate; GABA, gamma-amino butyrate; Glucose-6-P, glucose-6-phosphate; GlyAld-3-P, glyceraldehyde-3-phosphate.   In (A) purified potato tuber mitochondria from cv. Desiree (DM) and cv. Bintje (BM) and peroxisomes from cv. Bintje (BP; two independent preparations) were analyzed for content of NDB1. As controls, the antibody against glyoxysomal malate dehydrogenase (gMDH) detected bands of 37 and 44 kDa in peroxisomes and mitochondria, respectively, and catalase activity is denoted for each preparation in µmol min -1 mg -1 protein. In (B), mitochondria were purified from seven organs of potato plants (cv. Desirée) in three independent preparation sets. Twenty µg protein per lane was subjected to western blotting using antibodies against different mitochondrial proteins. Antibody designations and apparent molecular masses are denoted to the right. Results from one representative www.plantphysiol.org on August 25, 2017 -Published by Downloaded from Copyright © 2009 American Society of Plant Biologists. All rights reserved.