Metabolic changes associated with dark-induced leaf senescence in Arabidopsis nadk2 mutants

ABSTRACT Arabidopsis NADK2 (NAD kinase 2) is a chloroplast-localized enzyme involved in NADP+ synthesis, which acts as the final electron acceptor in the photosynthetic electron transfer chain. The NADK2-deficient mutant (nadk2) was used to analyze the effect of NAD(P)(H) unbalance in the dark-induced leaf senescence. During senescence, WT plants and nadk2 mutants showed a similar reduction in chlorophyll content. NAD(P)(H) quantification showed that the amount of total NAD(P)(H) decreased on the day 7 in WT but on the day 3 in nadk2. The phosphorylation ratio (i.e. NADP(H)/NAD(H)) decreased on day 1 in WT. In contrast, the nadk2 showed lower phosphorylation ratio at 0 day and no change throughout the aging process. Metabolome analysis showed that the metabolic profiles of both WT plants and nadk2 mutants subjected to dark-induced senescence adopted similar patterns as the senescence progressed. However, the changes in individual metabolites in the nadk2 mutants were different from those of the WT during dark-induced senescence.


Introduction
In all organisms, NAD(P)(H) are ubiquitous electron mediators that transfer electrons between oxidized forms (NAD + and NADP + ) and reduced forms (NADH and NADPH). NADP + and NADPH are typically used in anabolic reactions such as photosynthesis and lipid synthesis, while NAD + and NADH are typically used in catabolic reactions 1 . NAD kinase (NADK) catalyzes the phosphorylation of NAD + to NADP + , which modulates a variety of metabolic pathways by altering the phosphorylation ratio (NADP(H)/NAD(H)) 2 . Arabidopsis has three different types of the NADK proteins, namely, NADK1 (AT3G21070), NADK2 (AT1G2164), and NADK3 (AT1G78590) 3,4 . NADK1 is located in the cytosol 5 , while NADK3 is located in peroxisomes. Both NADK1 and NADK3 are involved in oxidative stress response 5 . The chloroplast-localized NADK2 plays a critical role in energy transduction in the photosynthesis [6][7][8] . In previous studies, it has been shown that altering the NADP + /NAD + ratio affected the metabolism in both rice and Arabidopsis 9,10 . For example, in rice plants overexpressing NADK2 has increased the NADP + / NAD + ratio, and as a result, it has increased the oxidative stress resistance and amino acid accumulation 9,11 . Nonetheless, despite the presence of multiple NADKs that have different subcellular locations and enzymatic properties, their precise physiological role in plants remains largely unknown.
Chloroplast-localized NADK2 is responsible for supplying NADP + , the final electron receptor in the photosynthetic electron transfer chain. Under continuous light conditions, T-DNA insertion mutant of NADK2 (nadk2) had smaller leaves with pale green color due to a reduction in chlorophyll 6 . Under short-day light conditions, nadk2 mutants showed more severe growth inhibition and more reactive oxygen species (ROS) accumulation such as hydrogen peroxide compared to the WT plants 12 .
Leaf senescence is accompanied by extensive metabolic transformation, from biosynthesis to degradation. Numerous environmental and developmental factors contribute to leaf senescence, including aging, darkness, hormones, drought, high salinity, and temperature extremes 13 . Dark-induced senescence is used as a model system for studying natural senescence, as it promotes chlorophyll degradation and protein catabolism 14,15 . NAD(P)(H) are involved in both catabolism and anabolism. The phosphorylation ratio of NAD(P)(H), which is regulated by NADK, greatly affects NAD(H)mediated catabolism and NADP(H)-mediated anabolism. In this study, we focused on leaf senescence, and clarified how nadk2 mutants undergo dark-induced senescence using metabolome analysis.

Plant materials and growth conditions
In this study, Arabidopsis thaliana (ecotype Columbia) was used as the wild type. According to Takahashi et al. 7 , a nadk2 mutant was obtained from the T-DNA Express Collection at the Salk Institute Genomic Analysis Laboratory (http://signal. salk.edu). In continuous light conditions (70 µmol m −2 s −1 ), Arabidopsis seeds were sown directly in Jiffy 7 peat pellets (Jiffy Products International AS, Norway). To induce senescence, two leaves per WT plant and nadk2 mutant were covered with aluminum foil and kept under dark conditions for 1, 3, and 7 d.

Measurement of chlorophyll contents
The chlorophyll (a+b) content of the plants was measured by spectrophotometric method. UV absorbances of chlorophyll elution in N, N-dimethylformamide was measured at wavelengths of 647 nm and 664 nm using UV-Vis spectrophotometer (Pharmacia Biotech Ultrospec® 3000 CT, Uppsala, Sweden) and pigment concentrations were calculated according to the method described by Ceusters et al 16 .

Assays of NAD(P)(H) contents
To measure NAD(P)(H), leaves (20-30 mg) were boiled in 200 µL of 0.2 N HCl (to extract of NAD + and NADP + ) and 0.2 N NaOH (to extract of NADH and NADPH) for 2 min. Samples were homogenized with a handy homogenizer (S-203, AS ONE, Osaka, Japan) and then centrifuged at 15,000 g for 10 min at 4°C after sonication (5 sec three times). For NAD + and NADP + , 15 µL of 0.2 M NaH 2 PO 4 (pH 5.6) and 120 µL of 0.2 N NaOH were added. For NADH and NADPH, 15 µL of 0.2 M HEPES/KOH pH 8.0 and 120 µL of 0.2 N HCl were added. NAD(P)(H) contents were measured using a cycling assay as described by Ishikawa et al 17,18 .

Metabolome analysis
Metabolite extraction was performed according to Miyagi et al. 2020 19 with minor modification. Approximately 100 mg of frozen leaves were ground by Shake Master Neo Ver. 1.0 (Bio Medical Science, Tokyo, Japan), and 0.15 mL of methanol was added to inactivate enzymes. After adding 0.15 mL of 100 μM piperazine-1,4-bis (2-ethanesulfonic acid) (PIPES) and 100 μM methionine sulfone solution as internal standards, the sample was then centrifuged at 15,000 g for 5 min (4°C). The supernatant was then centrifuged (12,000 g, 30 min, 4°C) and ultrafiltrated using a 3 kDa cutoff spin column (Merck, Darmstadt, Germany). The filtrate was used for CE-MS analysis.
Metabolites (organic and amino acids) were quantified using CE-triple quadrupole-MS (CE-QqQ-MS) system (CE, 7100; MS, G6420A; Agilent Technologies, Santa Clara, CA, USA) with multi-reaction monitoring (MRM) mode as described by Miyagi et al. 2020 19 . Quantitative accuracy was assessed using Agilent MassHunter Software and known contents of reference substances (approximately 50 primary metabolites).

Statistical analysis
Multivariate analyses were performed using IBM SPSS v22.0 (IBM, NY, USA). Principal component analysis (PCA) was done using the correlation matrix (cases = each plant; variables = metabolites). The average linkage method (between groups) and the squared Euclidean distance were used in hierarchical clustering analysis (HCA). A heatmap was created using Microsoft Excel 2010 (Microsoft, Redmond, WA, USA). In these analyses, the contents of each metabolite were normalized by the Z-score.

Dark-induced senescence in WT plants and nadk2 mutants
Chloroplast-localized NADK (NADK2) is responsible for providing NADP + , a receptor for electrons in the photosynthetic electron transfer chain. The NADK2 knockout mutant (nadk2) produces small rosette and pale green leaves as demonstrated in previous studies 6,10,12 . Under short-day light conditions, nadk2 mutant showed more severe growth inhibition and accumulated more ROS than in the WT plants 12 .
We also examined the ability of normal dark-induced senescence in the nadk2 mutant with an abnormal NAD(P)(H) balance. When Arabidopsis leaves were covered with aluminum foil, they turned white in 7 d (Figure 1a). In chlorophyll content analysis, nadk2 mutants showed lower chlorophyll levels than the WT plants at day 0 as described previously 6,10,12 , and the chlorophyll levels similarly decreased from day 0 to day 7 in both strains (Figure 1b). These data suggested that the WT and nadk2 leaves similarly undergo dark-induced senescence. However, gene expression associated with senescence (SEN1, SAG13, and SAG12) showed the process of WT and nadk2 differed (Figure 1c). After senescenceinduction, these genes were strongly upregulated as reported previously 20 . The nadk2 showed twofold increase than those of WT in SEN1 and SAG13 expression (day 1-day 7 in SEN1 and day 3-day 7 in SAG13). In contrast, SAG12 expression in the nadk2 was 20% of the WT at day 7. The SAG12 encodes a cysteine protease and SAG13 encodes a short-chain alcohol dehydrogenase 20 . The SEN1 gene is encoding unknown functional protein of Arabidopsis and has also been used as a marker to characterize the senescence-associated response 20 . These changes in senescence-related gene expression in the nadk2 plants support the importance of NAD(P)(H) balance in the process of dark-induced leaf senescence.
A previous study reported that the phosphorylation ratio (i.e., (NADP(H)/NAD(H)) decreased in nadk2 mutants under normal growth conditions 10 . Thus, the NAD(P)(H) contents were measured in senescence-induced leaves of WT plants and nadk2 mutants (Figure 1d). The results showed that the amount of NADP + decreased in WT plants at day 7. However, NADP + content in nadk2 mutants was lower than that in WT plants at day 0 to day 3. A gradual increase in NAD + was observed in WT plants from day 0 to day 3 but decreased by day 7. However, NAD + showed a decreasing trend from day 0 to day 7 in nadk2 mutant. NADPH contents in both strains were not significantly different at any time point. NADH content in the nadk2 mutant was considerably lower than that of the WT plants on day 1. As the senescence progressed, total NAD(P)(H) contents decreased on the day 7 in WT but decreased on the day 3 in nadk2. The phosphorylation ratio (i.e., NADP(H)/NAD(H)) decreased in WT on day 1. In contrast, the nadk2 showed lower phosphorylation ratio at 0 day, and it was not changed at all the time points examined. There were no clear changes in the redox ratio (i.e., (NAD(P) H/NAD(P)) of WT plants and nadk2 mutants (Figure 1d).
NADK2 regulates the phosphorylation ratio of NAD(P) (H), which in turn maintains NAD(H) levels during catabolism and NADP(H) during anabolism. The defect in NADK2 caused difference in the decrease in the NAD(P)(H) contents  and phosphorylation ratio of WT plants and nadk2 mutants but not the redox ratio of NAD(P)(H) balance during darkinduced senescence. We therefore examined what metabolite changes are seen during dark-induced senescence in the nadk2 mutant.

Nadk2 mutants showed altered metabolite profiles during dark-induced senescence
The metabolomic alterations resulted by dark-induced senescence were analyzed in WT and nadk2 mutant plants using CE-MS. Forty-nine metabolites involved in fundamental metabolic  processes such as glycolysis, TCA cycle, Calvin cycle and the pentose phosphate pathway were detected in non-senescent and dark-induced senescent leaves. PCA and HCA were used to visualize the differences of metabolic alterations between WT and nadk2 mutants (Figure 2a, b). Two principal components, PC1 (31%) and PC2 (17%), together accounted for 48% of variance in the metabolite dataset (Figure 2a). The results of the PCA showed that the metabolic patterns of the WT plants and nadk2 mutants were quite different in the non-senescent leaves at day 0, as reported previously 10,21 . In the nonsenescent leaves, different metabolic profiles were observed at day 1, 3, and 7. As opposed to that, in leaves subjected to dark-induced senescence, the metabolic profiles of the WT plants and nadk2 mutants moved toward the first quadrant ( Figure 2a). The WT plants and nadk2 mutants were highly different in their initial metabolic states, but after the induction of dark-induced senescence, the metabolic profiles of both the WT plants and nadk2 mutants were almost similar except several individual metabolites (Figures 3-5). Lactate was significantly increased in the non-senescent leaves of nadk2 mutants at day 0 and in dark-induced senescent leaves of nadk2 mutants at day 1. Malate level at day 7 and citrate level at both day 3 and day 7 in senescent leaves of nadk2 mutants were lower than that of in WT. The succinate levels in senescent leaves were also lower in nadk2 mutants at day 7 ( Figures 3a, and 5).
Numerous amino acids showed an increase in dark-induced senescent leaves (Figures 4 and 5). In particular, a marked increase was observed in tryptophan (Trp), isoleucine (Ile), valine (Val), asparagine (Asn), phenylalanine (Phe) and histidine (His) in the  nadk2 mutants compared to the WT plants at day 3. However, cysteine (Cys), threonine (Thr), and alanine (Ala) were lower in nadk2 mutants than WT at day 3. Focusing on dark-induced senescence day 7, asparagine (Asn) levels were increased in the nadk2 mutants, while methionine (Met), Cys, Thr, leucine (Leu), tyrosine (Tyr), Ile and Val levels were lower in the nadk2 mutants. It has been reported that the proteins in senescent leaves are degraded into their constituent amino acids, which are then transported elsewhere in the plant 23 . As shown in Figure 5, at the later stage of senescence (day 3 and day 7), most sugar phosphates involved in Calvin cycle showed no difference between WT and nadk2 plants. Conversely, amino acids and organic acids showed large differences, suggesting that metabolism related to protein degradation and respiration has been affected in nadk2 mutants.
According to Law et al. 24 , gene encoding for proteins involved in primary energy production (respiration, fermentation, and oxidation), amino acid, lipid, or nucleotide catabolism, sulfur metabolism, and shikimate metabolism are all overexpressed during senescence in darkened leaves. Thus, senescence is not considered to be a passive death step, but an important metabolic process that requires active metabolism. Consequently, the reduced phosphorylation ratio in the NAD(P)(H) balance may affect a wide range of metabolic pathways.
There was a trend toward reduced amounts of sugar phosphates in senescent leaves compared to non-senescent leaves, whereas the intermediates of the TCA cycle showed a rather complex pattern. This complex pattern may reflect the role of these intermediates as alternative respiratory substrates during the progression of senescence 25,26 . Previous studies have shown that the metabolism of C and N is markedly reduced in nadk2 mutants 9,10 . Differences in metabolic profiles in non-senescent leaves may be responsible for the differences in the intermediate processes and final products in dark-induced senescent leaves.

Conclusion
In this study, we investigated changes in NAD(P)(H) balance and metabolites in nadk2 mutants during dark-induced leaf senescence. The results showed that individual metabolites of Arabidopsis WT plants and nadk2 mutants responded differently to the induction of senescence. In particular, nadk2 mutants showed large differences in the amino acid contents produced from protein-degradation during senescence. Further studies will be important to clarify the effects of differences in the NAD(P)(H) balance on the degradation and translocation of substances in dark-induced senescence leaves in Arabidopsis WT plants and nadk2 mutants.