Transcription factor RD26 is a key regulator of metabolic reprogramming during dark-induced senescence
Summary
- Leaf senescence is a key process in plants that culminates in the degradation of cellular constituents and massive reprogramming of metabolism for the recovery of nutrients from aged leaves for their reuse in newly developing sinks.
- We used molecular–biological and metabolomics approaches to identify NAC transcription factor (TF) RD26 as an important regulator of metabolic reprogramming in Arabidopsis thaliana.
- RD26 directly activates CHLOROPLAST VESICULATION (CV), encoding a protein crucial for chloroplast protein degradation, concomitant with an enhanced protein loss in RD26 overexpressors during senescence, but a reduced decline of protein in rd26 knockout mutants. RD26 also directly activates LKR/SDH involved in lysine catabolism, and PES1 important for phytol degradation. Metabolic profiling revealed reduced γ-aminobutyric acid (GABA) in RD26 overexpressors, accompanied by the induction of respective catabolic genes. Degradation of lysine, phytol and GABA is instrumental for maintaining mitochondrial respiration in carbon-limiting conditions during senescence.
- RD26 also supports the degradation of starch and the accumulation of mono- and disaccharides during senescence by directly enhancing the expression of AMY1, SFP1 and SWEET15 involved in carbohydrate metabolism and transport. Collectively, during senescence RD26 acts by controlling the expression of genes across the entire spectrum of the cellular degradation hierarchy.
Introduction
Leaf senescence is a highly coordinated developmental process that leads to the disintegration of photosynthetically active leaves and the recovery of nutrients for their re-use in sinks. In aging leaves, metabolism is largely shifted from anabolism to catabolism and nutrient remobilization which finally sacrifices leaves for the sake of recycling C and N sources (Guo et al., 2004; Buchanan-Wollaston et al., 2005; Mueller-Roeber & Balazadeh, 2014). In senescing leaves, intracellular organelle dismantling initiates from chloroplasts and ends with mitochondria and the nucleus (Gan, 2007; Taylor et al., 2010; Thomas, 2013). About 70% of the leaf protein is located in chloroplasts, and Rubisco and Chla/b binding protein are the largest reservoirs of recoverable N in vegetative tissues (Morita, 1980; Makino & Osmond, 1991; Ishida & Yoshimoto, 2008). Autophagy, senescence-associated vacuoles (SAVs) and CHLOROPLAST VESICULATION (CV)-mediated pathways are three known mechanisms involved in the degradation of chloroplast proteins in senescing leaves (Otegui et al., 2005; Ishida & Yoshimoto, 2008; Carrión et al., 2013; Wang & Blumwald, 2014). The latter is an intraplastidic degradation pathway, triggering vesiculation of chloroplast proteins. CV-induced vesicles are then delivered to proteolytic vacuoles for nutrient recovery (Wang & Blumwald, 2014). Destabilization of photosystems and degradation of Chla/b binding protein releases free Chl and toxic catabolites derived from it; their further degradation is thus a critical process (Hörtensteiner, 2006, 2009; Sakuraba et al., 2012).
In senescing leaves with low photosynthetic efficiency heterotrophic metabolism prevails and it can be anticipated that this will ultimately result in C starvation. However, at the same time senescence-related processes are highly energy consuming and an active mitochondrial respiration is a critical necessity. Therefore, besides sugars, catabolism of proteins, lipids and Chl provides alternative substrates for the TCA cycle and hence allows continued operation of mitochondrial respiration (Ishizaki et al., 2005; Kunz et al., 2009; Araújo et al., 2011; Hörtensteiner & Kräutler, 2011). Furthermore, catabolism of lysine, branched chain and aromatic amino acids (BCAAs and AAAs) and phytol in senescing leaves by isovaleryl-coenzyme A (CoA) dehydrogenase (IVDH) and 2-hydroxyglutarate dehydrogenase (D2HGDH) additionally directly channels electrons to the mitochondrial electron transport chain (mETC) (Araújo et al., 2010). The degradation of γ-aminobutyric acid (GABA) is another alternative pathway which supports the maintenance of mitochondrial respiration during senescence. GABA accumulates under stress conditions and in senescing leaves mainly through glutamate metabolism. The GABA shunt converts GABA to succinate which functions both as a TCA cycle intermediate and as an electron donor to the mETC (Kinnersley & Turano, 2000; Watanabe et al., 2013; Michaeli & Fromm, 2015).
Although massive changes in central metabolism during developmental or stress-induced senescence have been reported in several broad or targeted metabolite profiling studies (e.g. Diaz et al., 2005; Wingler et al., 2006; Brychkova et al., 2008; Veyres et al., 2008; Araújo et al., 2010; Watanabe et al., 2013), regulation of these changes is poorly understood, particularly at the transcriptional level. That said it is clear that both the initiation of senescence and its integration into global plant development and physiology are regulated by transcription factors (TFs), including NAC TFs (Guo & Gan, 2006; Kim et al., 2009, 2013; Balazadeh et al., 2011; Yang et al., 2011; Wu et al., 2012; Garapati et al., 2015b). One such TF is ATAF1 from Arabidopsis thaliana, a positive regulator of senescence (Garapati et al., 2015b). ATAF1 induces a carbon starvation transcriptome which includes expression changes of starch metabolic and autophagy-related genes, through unknown mechanisms. However, it directly binds to the promoter of TREHALASE1 (TRE1) thereby affecting sugar metabolism (Garapati et al., 2015a).
The NAC factor JUNGBRUNNEN1 (JUB1) negatively regulates senescence, while it positively affects tolerance to various abiotic stresses (Wu et al., 2012; Ebrahimian-Motlagh et al., 2017). Metabolite profiling revealed elevated levels of proline and trehalose in JUB1 overexpressors (Wu et al., 2012; Shahnejat-Bushehri et al., 2017). However, it is currently unknown whether genes affecting the metabolite levels are directly controlled by JUB1.
RD26, a member of the ATAF subfamily of TFs in A. thaliana (Jensen et al., 2010), has previously been identified as a positive regulator of developmental and dark-induced senescence (Li et al., 2016) and jasmonic acid-induced Chl degradation (Zhu et al., 2015), while it is also involved in abiotic and biotic stress signaling (Fujita et al., 2004; Tran et al., 2004; Zheng et al., 2012; Ye et al., 2017). Here, we report RD26 as an important regulator of metabolic reprogramming during senescence by controlling the expression of genes across the entire spectrum of the cellular degradation hierarchy.
Materials and Methods
General methods
Standard molecular techniques were performed as described (Sambrook & Russell, 2001). Unless otherwise indicated, chemicals and reagents were obtained from Sigma-Aldrich (Munich, Germany), Merck (Darmstadt, Germany), Invitrogen (Darmstadt, Germany) and Roche (Basel, Switzerland). All oligonucleotides (labeled and unlabeled) were obtained from MWG Eurofins (Ebersberg, Germany). Restriction enzymes were obtained from New England Biolabs (Ipswich, MA, USA). Molecular biology kits were purchased from Qiagen, Macherey-Nagel (Bethlehem, PA, USA), New England Biolabs, Invitrogen, Ambion and Thermo Fisher Scientific. DNA sequencing was performed by MWG Eurofins or LGC Genomics (Hoddesdon, UK). The Arabidopsis Information Resource (TAIR; http://www.Arabidopsis.org) was used to obtain sequences; sequence analyses were performed using online tools available at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). Oligonucleotide sequences are given in Supporting Information Table S1.
Growth conditions
A. thaliana (L.) Heynh., accession Col-0, was used as wild-type (WT) and as background for all transgenic lines. Arabidopsis seeds were germinated and seedlings were grown on half-strength Murashige–Skoog (MS) medium (Murashige & Skoog, 1962) supplemented with 7% agar and 1% sucrose; for selection of transgenic lines suitable antibiotics or herbicides were added. For growth under long-day (LD) conditions, plants were cultured in a climate chamber at 16 h light (100 μmol m−2 s−1, 20°C, 60% relative humanity) and 8 h dark (16°C, 75% relative humanity). For growth under day-neutral (DN) conditions, the light period was reduced to 12 h. Einheitserde GS90 (Gebrüder Patzer, Sinntal, Germany) was used for growing plants in soil.
Constructs and transformation
PCR products used for cloning were amplified using Phusion high-fidelity DNA polymerase (New England Biolabs). Constructs were either generated by cut-ligation protocols using DNA restriction and ligation enzymes, or by using the GATEWAY method. Strategies for the generation of each construct and primer sequences are listed in Table S1. The correctness of cloned fragments was confirmed by DNA sequencing. Constructs were transformed into Agrobacterium tumefaciens and used to transform Arabidopsis using the floral deep method (Clough & Bent, 1998).
Isolation of T-DNA insertion lines
RD26 T-DNA insertion lines rd26-1 (SALK_063576) and rd26-2 (SALK_083756) were identified from the sequence-indexed library of insertion mutations (SALK) using the Arabidopsis gene mapping tool (http://signal.salk.edu/cgi-bin/tdnaexpress), and the seed stocks were obtained from NASC (http://arabidopsis.info). Plants homozygous for the T-DNA insertion were identified by PCR-based genotyping using SALK universal T-DNA primer (LBb1.3) and a gene-specific RP primer. To check for the presence of the WT allele, gene-specific primers (RP and LP) were used (Table S1). Quantitative reverse transcription PCR (qRT-PCR) and end-point PCR (using relevant cDNA as template) were performed to check transcript abundance and the presence of intact transcript, respectively.
Microarray analysis
Microarray experiments were performed as described (Balazadeh et al., 2010). Two-week-old MS-grown RD26-IOE seedlings or fully expanded leaf no. 11 of soil-grown plants (at 38 d after sowing, DAS) were transferred to liquid MS medium 24 h before estradiol (EST) treatment to minimize the effect of environmental change. Seedlings or leaves were treated for 2 or 5 h with 10 μM EST or ethanol (0.1%, v/v) in mock treatments, and RNA was extracted for expression profiling using Affymetrix ATH1 microarrays. Preparation of probes and microarray hybridizations (two independent biological replicates for each condition) were performed by ATLAS Biolabs (Berlin, Germany). Microarray expression data are available from the NCBI Gene Expression Omnibus (GEO) repository (www.ncbi.nlm.nih.gov/geo/) under accession number GSE100654.
Quantitative real-time PCR
Total RNA was isolated from c. 100 mg of frozen whole seedlings or mature leaf samples using an RNeasy Plant Mini Kit (Qiagen). cDNA synthesis and qRT-PCR were performed as reported (Caldana et al., 2007; Balazadeh et al., 2008). Absence of genomic DNA contamination was verified by PCR using primers that anneal to an intron of a control gene (At5g65080; for primer sequences see Table S1). Real-time qPCR was performed with SYBR Green (Applied Biosystems) and gene-specific primers, using an ABI PRISM 7900HT sequence detection system (Applied Biosystems). ACTIN2 (At3g18780) served as the reference gene for data analysis. Data were analysed using the comparative Ct method. Briefly, the Ct value of each gene was normalized to the Ct value of ACTIN2, giving the delta Ct value (ΔCt). The level of expression was expressed as the difference between an arbitrary value of 40 and ΔCt, so that a high 40-ΔCt value indicates high gene expression level.
In RD26-IOE seedlings, RD26 expression was induced by treatment with 10 μM EST or ethanol (0.1%, v/v) in mock treatments.
Binding site selection assay
In vitro binding site selection was performed using the CELD system with the pTacRD26-CELD6xHis construct, using a biotin-labeled double-stranded oligonucleotide, which contained a 30-nucleotide random sequence (Xue, 2005). RD26-selected oligonucleotides were cloned and sequenced. The DNA binding activity of RD26-CELD was determined using methylumbelliferyl β-d-cellobioside (MUC) as substrate (Xue, 2002). A biotin-labeled double-stranded oligonucleotide without an RD26 binding site was used as a control for background activity.
Electrophoretic mobility shift assay
Recombinant RD26-CELD fusion protein was expressed in Escherichia coli XL1-Blue harboring the pTacRD26-CELD6xHis plasmid. Recombinant protein was purified using Protino Ni-IDA 150 packed columns (Macherey-Nagel). Aliquots of the elution fractions were analyzed by SDS-PAGE and Coomassie staining. The concentration of RD26-CELD protein was determined by the Bradford assay (Bradford, 1976).
Electrophoretic mobility shift assays (EMSA) were performed as described (Wu et al., 2012) using the Odyssey Infrared EMSA kit (Li-Cor, Bad Homburg, Germany). Thirty to 50 bp of the promoter regions of the putative targets, including the RD26 putative binding site, were used in the EMSA studies. 5′-DY682-labeled oligonucleotides (Table S1) were obtained from Eurofins MWG Operon. Gel electrophoresis was performed using 6% retardation gel (Invitrogen) and DY682 signal was detected using the Odyssey Infrared Imaging System (Li-Cor).
Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP)/qPCR was performed with 16-d-old 35S:RD26-GFP seedlings grown under LD conditions on agar medium, or with RD26:RD26-GFP/rd26-1 seedlings grown in the same way but additionally kept for 3 d in the dark (19 DAS) to enhance transgene expression. Arabidopsis Col-0 plants served as negative controls. ChIP was performed as previously described (Kaufmann et al., 2010) using anti-green fluorescent protein (anti-GFP) antibody to immunoprecipitate protein–DNA complexes. The ChIP experiment was run in two biological replications with three technical replications per assay. qPCR was used to test binding of RD26 to its binding site within the promoters of the target genes. The primers flanked the RD26 binding site. Primers annealing to promoter regions of an Arabidopsis gene (At2g22180) lacking an RD26 binding site were used as negative control. qPCR data were analyzed as described (Wu et al., 2012).
Analysis of leaf metabolites by GC-MS and GC
Metabolite extraction for GC-MS was performed by a method modified from that previously described (Roessner-Tunali et al., 2003). LD grown plants were kept in continuous darkness at 21 DAS and the whole rosette of two Arabidopsis plants were harvested as each biological replicate at the indicated times and immediately frozen in liquid nitrogen before storage at −80°C. Then 50 mg of each sample was homogenized using a ball mill precooled with liquid nitrogen and extracted in 1.4 ml of methanol, and 60 μl of internal standard (0.2 mg ribitol ml−1 water) was subsequently added as a quantification standard. The extraction, derivatization, standard addition and sample injection were as described previously (Lisec et al., 2006). The GC-MS system comprised a CTC CombiPAL autosampler, an Agilent 6890N gas chromatograph and a LECO Pegasus III TOF-MS device running in EI+ mode. Metabolites were identified in comparison to database entries of authentic standards (Kopka et al., 2005). Chromatograms and mass spectra were evaluated using Chroma TOF1.0 (Leco, St Joseph, MI, USA) and TagFinder 4.0 software (Luedemann et al., 2008). Fatty acids were measured by GC as previously described (Chen et al., 2013).
Other methods
The relative Chl content of rosette leaves was measured in nondestructive conditions by using a SPAD analyzer (Konika Minolta, Tokyo, Japan). Alternatively, Chl was extracted and spectrophotometrically measured when leaf destruction was acceptable as described (Gechev et al., 2013). Total proteins from Arabidopsis seedlings were extracted by a phenol-based method as described (Sedaghatmehr et al., 2016). The protein concentration was determined with the BCA Protein Assay kit (Thermo Fisher Scientific). Starch was determined enzymatically in the pellets obtained after ethanol extraction of soluble sugars (Hendriks et al., 2003). Ion leakage measurements were performed in leaves 4–10 as described (Guo & Gan, 2006). The location of RD26-GFP fusion protein in transgenic plants was analyzed by confocal fluorescence microscopy (Eclipse E600 microscope; Nikon, Tokyo, Japan). Histochemical GUS staining assays were performed as described (Plesch et al., 2001).
Results
RD26 activates senescence-associated genes
RD26 shows enhanced expression during senescence and it triggers developmental and dark-induced senescence when overexpressed under the control of the constitutive Cauliflower Mosaic Virus (CaMV) 35S promoter (hereafter, RD26Ox), while senescence is delayed in two rd26 knockout mutants (rd26-1 and rd26-2; see Materials and Methods) (Balazadeh et al., 2008; Li et al., 2016; Figs S1, S2). Here, we sought to identify the gene regulatory network (GRN) that this TF controls. To this end, we expressed RD26 in transgenic Arabidopsis plants under the control of an EST-inducible promoter (RD26-IOE), observing leaf senescence after only 48 h of EST treatment (Fig 1a,b), accompanied by increased expression of the senescence marker genes SAG12 and SAG13 in EST-treated plants (Fig. 1c), confirming that RD26 rapidly induces senescence.
Next, we induced RD26 expression by EST treatment (5 h) in 14-d-old RD26-IOE seedlings and performed transcriptome analysis using Affymetrix ATH1 microarrays. Fifty-six genes were induced by at least two-fold upon EST induction (Table S2), compared to mock-treated samples, of which 39 (70%) are senescence-associated genes (SAGs) and 24 (43%) are dark-induced genes (DINs) (Fig S3a; Table S2; Mueller-Roeber & Balazadeh, 2014). Interestingly, most of the genes were induced after only a 2-h EST treatment, albeit less strongly (Table S2). Furthermore, the vast majority of the genes highly coexpressed with RD26 in public datasets were SAGs and/or DINs: Genome-wide, 3716 (15%) of the 24 000 genes represented by the Affymetrix ATH1 microarray are SAGs, and 3230 genes (13%) are DINs (Mueller-Roeber & Balazadeh, 2014). By contrast, 93% and 74%, respectively, of the RD26-coexpressed genes are SAGs and DINs. This observation strongly supports a function of RD26 in the control of senescence (Fig. S3b,c; Table S3).
We next tested the expression of 31 SAGs with known or proposed functions by qRT-PCR after 5 h of EST induction, using two new biological replications, and found that all genes were induced by RD26 (Fig. 1d; Table S2). Finally, we tested the expression of the 31 SAGs in three RD26Ox lines and the rd26-1 and rd26-2 knockout mutants, and observed elevated expression in overexpressors and reduced expression in the mutants (Fig. 1d; Table S2).
RD26 binding site
We performed CELD-based in vitro binding site selection to determine DNA motifs to which RD26 binds (Xue, 2002, 2005). We used biotin-labeled double-stranded oligonucleotides with a 30-nt random sequence and after the seventh round of selection recovered 45 unique double-stranded oligonucleotides bound by RD26-CELD protein, which we further evaluated for their binding activity (Table 1). Sequence alignment revealed two cis-elements as RD26 binding sites, namely CGTr(n5-6)YACGyhayy (RD26-BSI) and rgwnDnY(n8-9)YACGtmwcy (RD26-BSII). Although the binding site sequences show some variability, sequence alignments revealed CGT and YACG as two critical core elements, separated by a highly variable linker region, of RD26-BSI. Similarly, RD26-BSII contains two elements with a longer flexible part. Therefore, the RD26 binding site is bipartite, with YACG in the downstream part of both, RD26-BS1 and RD26-BSII (Table 1).
RD26 directly regulates degradation of chloroplast proteins through CV
CV encodes a protein that mediates the turnover of chloroplast proteins by vesiculation and their delivery to vacuoles for subsequent proteolysis during senescence (Wang & Blumwald, 2014). CV expression was induced 3.7-fold after EST (5 h) treatment in RD26-IOE, and four- to 23-fold in RD26Ox lines, while its transcript abundance was reduced 1.8-fold in rd26 mutants (Fig. 2a). The promoter of CV contains an RD26 binding site −385 bp upstream of the translation start codon (ATG) (Fig. S4), suggesting it is a direct downstream target of RD26. To test this possibility, we performed an EMSA, which demonstrated binding of RD26 to a c. 40-bp double stranded DNA fragment containing the binding site of RD26 (Fig. 2b). Physical interaction of RD26 with the CV promoter fragment was significantly reduced in the presence of unlabeled promoter fragment (competitor) (Fig. 2b). We then examined whether RD26 binds to the CV promoter in vivo using two types of transgenic plants: one expressing RD26-GFP (RD26 fused to GFP) from the CaMV 35S promoter (35S:RD26-GFP), and one expressing RD26-GFP from the native RD26 promoter in the rd26-1 mutant background (RD26:RD26-GFP/rd26-1). Nuclear localization of RD26-GFP was confirmed by confocal microscopy (Fig. S5a). Furthermore, expression of CV was enhanced in both lines compared to the relevant controls (WT for 35S:RD26-GFP, and rd26-1 for RD26:RD26-GFP/rd26-1; Fig. S5b), confirming that RD26-GFP faithfully operates as a TF. We then performed ChIP assays on 35S:RD26-GFP seedlings (grown under LD conditions on agar medium, 16 DAS), and on RD26:RD26-GFP/rd26-1 seedlings (same growth conditions, but incubated in the dark for another 3 d to induce RD26:RD26-GFP expression, 19 DAS) and observed strong enrichment of CV promoter fragments in the precipitated chromatin of both lines using primers spanning the RD26 binding site (Fig 2c). We also observed in vivo binding of RD26 to the promoter of ClpD (Fig. 2c), which was suggested earlier (Tran et al., 2004), but not yet shown. ClpD encodes a chloroplast caseinolytic protease (Clp) regulatory subunit (Rosano et al., 2011). Its promoter harbors an RD26 binding site (Fig. S4) that physically interacts with RD26 in EMSA experiments (Fig. 2b). ClpD expression is elevated in RD26Ox and RD26-IOE plants (after EST induction), compared to controls, but reduced in rd26 mutants (Fig. 2a).
Considering the induction of two chloroplast protein degradation pathways (CV and ClpD) by RD26, we next determined total protein content in seedlings incubated for 3 d in darkness or under control conditions (LD, 19 DAS). While protein content decreased by 25% over the dark incubation period in WT, it decreased by 40–55% in RD26Ox lines, but by only 15% and 11% in the rd26-1 and rd26-2 knockout mutants (Fig S6), suggesting that RD26 triggers protein degradation during senescence by activating CV and ClpD.
RD26 alters primary metabolites during senescence
Within the set of RD26-affected genes were several with roles in primary metabolism (Table S2), indicating that RD26 reprograms metabolic pathways during senescence. To test this hypothesis further, we transferred 21-d-old plants (grown under LD conditions) to continuous darkness for up to 10 d and performed metabolic profiling of rosette leaves using GC-MS (Lisec et al., 2006). As reported, RD26Ox plants senesced faster than WT and rd26 plants (Li et al., 2016; Fig S1), which was accompanied by a higher expression of senescence marker gene SAG13 in RD26Ox plants, but reduced expression in rd26 mutants (Fig. 2a). Notably, SAG13 is a direct target gene of RD26 (Fig. 2b,c). After 10 d in darkness, RD26Ox plants were virtually dead and therefore not included in the metabolite profiling. Results indicate massive changes in central metabolism in all lines during extended darkness (Figs 3, 4, S7), although we observed different patterns of metabolic alterations in the RD26 transgenic lines compared to WT. In general, RD26Ox lines exhibited earlier metabolic responses to the dark treatment (see next section).
RD26 overexpression leads to precocious accumulation of free amino acids in darkness
Protein degradation during dark-induced and developmental senescence leads to the accumulation of most amino acids (Araújo et al., 2010; Watanabe et al., 2013). We observed an accumulation of free amino acids during extended darkness (Fig. 3). Gln and Asn, which carry two N atoms, are the main sources of N remobilization during senescence; both amino acids accumulate highly in senescing leaves, while contents of Glu and Asp, the closest amino acids to them but carrying a sole N atom, are unaffected or only slightly altered during senescence (Watanabe et al., 2013). Consistently, we observed a high level of accumulation of Gln and Asn which was intensified gradually during dark treatment in all lines, while the level of Glu and Asp did not change dramatically (Fig. 3). The accumulation of free Gln during extended darkness was higher in all three RD26Ox lines than in WT, and its relative level increased 2.3- and 1.6-fold in RD26Ox lines at 3 d after transfer to darkness (DAT) and 6 DAT, respectively, compared to the level in WT at the same time points. Similarly, a higher level of free Asn was observed in RD26Ox lines compared to WT at 3 DAT.
AAAs (Trp, Phe) and His accumulated highly during dark treatment in all lines (Fig. 3). For Trp and His, RD26Ox lines displayed a significantly higher level of accumulation than WT during dark treatment. The higher accumulation of free amino acids in RD26Ox lines might be linked to the higher rate of protein degradation in these lines (Fig. S6).
The levels of Arg and Cys were significantly lower in RD26Ox lines than in WT at 21 DAS (0 DAT; Fig 3). Although dark treatment dramatically increased the level of both amino acids in all lines, RD26Ox lines still exhibited a lower level of Cys accumulation than WT at both 3 and 6 DAT. Gly and Pro were the only amino acids whose levels decreased strongly during dark treatment in all lines (Fig. 3). RD26Ox lines showed significantly lower levels of Gly at both 0 DAT (21 DAS) and 6 DAT.
RD26 overexpression enhances accumulation of TCA cycle intermediates and represses GABA accumulation
The relative levels of the TCA cycle intermediates citrate, malate, fumarate and 2-oxoglutarate were generally increased during dark-induced senescence, but the increases were higher in RD26Ox plants (Fig. 4). GABA highly accumulated during extended darkness in all lines. However, GABA accumulation in RD26Ox lines was reduced compared to WT at 6 DAT. GABA catabolism through the GABA shunt provides succinate and as such acts as an alternative source of energy for mitochondrial respiration. The lower level of GABA in RD26Ox lines might be explained by a higher activity of the GABA shunt in these plants, providing more energy under C-starvation conditions.
Araújo et al. (2010) showed that catabolism of lysine, BCAAs and AAAs under C-limiting conditions can directly provide electrons to the mETC, sustaining the operation of the TCA cycle. Similarly, catabolism of phytol and free fatty acids provides more substrate for mitochondrial respiration as an alternative pathway (Kunz et al., 2009). RD26 upregulates several genes with a function in the aforementioned catabolic processes (Fig. 1d). Therefore, the higher upregulation of the TCA cycle in RD26Ox plants during dark treatment compared to WT might be linked to an enhanced catabolism providing more substrate for mitochondrial respiration.
RD26 triggers lysine degradation by directly controlling LKR/SDH expression
Lys also accumulated in all lines during dark treatment (Fig. 3). However, in contrast to most amino acids, accumulation of Lys is repressed in RD26Ox lines compared to WT during dark-induced senescence. The relative level of Lys in RD26Ox #50 plants were three-fold lower than in WT at 3 DAT. Following the RD26 transcript level, relative levels of Lys in RD26Ox #50, #21 and #42 plants were 6.3-, 2.7- and 1.3-fold, respectively, lower than in WT at 6 DAT, while in rd26-1 they were 1.4- and 1.5-fold higher than in WT at 3 and 6 DAT, respectively (Fig. 3). Although senescence is accelerated in RD26Ox plants compared to WT, the less prominent accumulation of Lys suggests a higher rate of Lys degradation in these plants.
LKR/SDH and IVDH, two genes with functions in amino acid catabolism, were among the RD26 early responsive genes (Fig. 2a). LKR/SDH encodes two enzymes responsible for Lys degradation. LKR/SDH expression is induced in senescing leaves as well as by abscisic acid (ABA) treatment and during sucrose starvation (Stepansky & Galili, 2003; Buchanan-Wollaston et al., 2005). In addition to amino acid degradation as part of the overall degradation processes occurring during senescence, Lys catabolism provides electrons for the mETC in dark-induced senescence and during other C starvation stresses (Araújo et al., 2010). LKR/SDH transcript abundance increased 2.8-fold in the RD26-IOE line 5 h after EST treatment and was around seven-fold more highly expressed in RD26Ox lines compared to EV, while LKR/SDH transcript level was slightly reduced (1.5-fold) in rd26 mutants (Fig. 2a).
The LKR/SDH promoter harbors an RD26 binding site −1173 bp upstream of the translational start codon (Fig. S4). EMSA revealed physical binding of RD26-CELD fusion protein to the RD26 motif (Fig. 2b), and ChIP-qPCR confirmed binding of RD26 in planta to the LKR/SDH promoter, both in 35S:RD26-GFP and in RD26:RD26-GFP/rd26-1 plants, in accordance with elevated expression of LKR/SDH (Fig. S5b), demonstrating that LKR/SDH is a direct downstream target of RD26 (Fig. 2c).
RD26 affects amino acid metabolism by inducing IVDH
BCAAs (Leu, Ile and Val) accumulated highly during dark treatment in all lines. However, Leu and Ile, among others, showed different patterns of accumulation in RD26Ox lines compared to WT. The levels of BCAAs increased continually in WT plants in the dark, while Leu and Ile showed a biphasic pattern in RD26Ox lines: at 3 DAT Leu and Ile accumulated to higher levels in RD26Ox than in WT, while this prominent accumulation was repressed in RD26Ox compared to WT at 6 DAT.
IVDH encodes isovaleryl-coenzyme A dehydrogenase, which catalyses the breakdown of BCAAs, Lys and phytol to help maintain mitochondrial respiration by providing electrons during dark-induced senescence (Daschner et al., 2001; Araújo et al., 2010). Overexpression of RD26 enhances IVDH expression by three- to nine-fold, while expression is slightly reduced (1.7-fold) in rd26 mutants compared to WT (Fig. 2a). Therefore, it is reasonable to assume that in early stages of dark-induced senescence (3 DAT) higher protein degradation in RD26Ox plants leads to more prominent accumulation of Leu and Ile compared to WT, while higher levels of IVDH expression in RD26Ox plants might trigger higher rates of Leu and Ile degradation, resulting in a lower level of their accumulation in RD26Ox plants than WT at later stages (6 DAT; Fig 3).
RD26 regulates catabolism of phytol and free fatty acids by direct induction of PES1
Chl catabolism during senescence and stresses leads to the accumulation of phytol and free fatty acids. PES1 encodes an enzyme with acyltransferase activity which converts these toxic intermediates to phytyl esters. Furthermore, as an alternative pathway during stress or senescence, catabolism of phytol and free fatty acids provides more substrate for mitochondrial respiration. Thus, PES1 is strongly induced during senescence and stresses (Araújo et al., 2010; Lippold et al., 2012). PES1 expression increased by about three-fold in the RD26-IOE line (5 h of EST treatment) and two- to five-fold in RD26Ox lines (compared to WT; Fig. 2a). Furthermore, UDP-GLUCOSYL TRANSFERASE 76E12 (UGT76E12) and GLUTATHIONE S-TRANSFERASE TAU 3 (GSTU3), predicted to be involved in fatty acid catabolism, are strongly induced in RD26-IOE (5 h of EST treatment) and RD26Ox plants, and repressed in rd26 mutants (Table S2). The PES1 promoter harbors an RD26 binding site starting −905 bp upstream of the translation start codon, and by EMSA we demonstrated physical binding of RD26-CELD protein to the respective PES1 promoter fragment (Fig. 2b). RD26 also binds in planta to the PES1 promoter after dark treatment, as shown by ChIP-qPCR (Fig. 2c).
We also studied changes in the composition of fatty acids during dark treatment in all lines. In accordance with previous studies (e.g. Yang & Ohlrogge, 2009; Araújo et al., 2010) palmitic acid (16 : 0), linolenic acid (18 : 3) and linoleic acid (18 : 2) are the major fatty acid components in Arabidopsis leaves grown in LD conditions (0 DAT; Fig. S8). Extended darkness led to a decline of all fatty acid classes in all lines, although linolenic acid and linoleic acid decreased much more than palmitic acid, consistent with the reported reduction in the ratio of unsaturated to saturated fatty acids during developmental or dark-induced senescence (Yang & Ohlrogge, 2009; Araújo et al., 2010). Despite acceleration of dark-induced senescence in RD26Ox plants, there were no significant differences in fatty acid composition compared to WT.
RD26 induces sugar metabolism-related genes during senescence
Sugar metabolism was also altered in RD26Ox plants compared to WT during dark-induced senescence. Although the relative levels of fructose and maltose increased in all lines, the increases were much higher in RD26Ox plants (Fig. 4). The relative level of fructose was 49-fold higher in RD26Ox plants than in WT at 6 DAT, and the level of maltose was 5.5-fold higher. Similarly, trehalose, a disaccharide that accumulates to higher levels in early-senescence mutants than controls (Veyres et al., 2008), accumulated to higher levels in RD26Ox plants than WT at 3 DAT (Fig. 4). We also enzymatically determined starch in 19 DAS LD-grown seedlings or by Lugol staining in 32 DAS DN-grown plants at mid-day (as a further experimental condition) and observed a lower level of starch in RD26Ox lines than in WT (Fig S9a,b); under DN conditions, we observed a tendency for an elevated starch content in rd26 mutants (Fig. S9b).
SUGAR-PORTER FAMILY PROTEIN 1 (SFP1) and ERD SIX-LIKE 1 (ESL1) encode monosaccharide transporters whose expression increases during leaf senescence, accompanied by an accumulation of monosaccharides (Quirino et al., 2001; Buchanan-Wollaston et al., 2005). SWEET15 encodes a sucrose efflux transporter primarily expressed in senescing tissues and showing enhanced expression in 35S:RD26-GFP and RD26:RD26-GFP/rd26-1 plants (Fig. S5b); its overexpression in transgenic plants accelerates senescence (Seo et al., 2011). ALPHA-AMYLASE1 (AMY1) encodes an α-amylase presumably involved in starch mobilization in senescing leaves (Doyle et al., 2007; Jie et al., 2009). Our data show enhanced expression (two- to 28-fold) of all four genes in RD26-IOE (5 h of EST induction) and RD26Ox lines, while expression of those genes was reduced in rd26 mutants (Figs 1d, 2a, S10a).
RD26 binding sites are present in the promoters of all four genes (Fig. S4). We chose the RD26 binding sites of the SFP1, SWEET15 and AMY1 promoters for EMSA and ChIP-qPCR assays and demonstrated in vitro and in planta binding of RD26 to all three of them (Fig. 2b,c).
Discussion
During leaf senescence, central metabolism is shifted from nutrient assimilation to the catabolism of macromolecules and the remobilization of stored nutrients, needed for the successful establishment of new sinks (including developing seeds). This metabolic reprogramming is of immense importance for achieving high crop yields and is due to massive changes in the transcriptome. Here, we reveal a key role of NAC factor RD26 as a regulator of metabolome reprogramming during senescence.
Up to 80% of N and 30% of C in senescing leaves are remobilized towards sinks for providing their high nutrient demands, whereby chloroplasts in leaves are the main sources of N and C (Havé et al., 2016). Although chloroplasts themselves contain various proteases, evidence indicates that massive proteolysis during senescence occurs in vacuoles (Himelblau & Amasino, 2001; Otegui et al., 2005; Kato & Sakamoto, 2010; Breeze et al., 2011; Havé et al., 2016). Besides autophagy (Ishida & Yoshimoto, 2008; Wada et al., 2009; Izumi et al., 2010) and degradation by virtue of SAVs (Otegui et al., 2005; Carrión et al., 2013), the CV-mediated pathway is involved in the degradation of both thylakoid and stromal proteins (Wang & Blumwald, 2014; Xie et al., 2015). Here, we identified RD26 as a positively acting transcriptional regulator of CV expression. Like RD26 overexpression, elevated expression of CV leads to precocious senescence (Wang & Blumwald, 2014), and both genes are highly induced by abiotic stresses. Notably, overexpression of RD26 reduces the level of total leaf protein compared to WT, suggesting an involvement of RD26 in the degradation of chloroplast proteins by promoting vesiculation and proteolysis of these proteins by CV induction. Furthermore, RD26 seemingly affects proteolysis by directly inducing ClpD, a regulatory subunit of the Clp family. The Clp protease complex consists of two multi-subunit components: a catalytic core and a chaperone complex (Roberts et al., 2012). Chaperone subunits ClpD and ClpC1/2 are subunits responsible for substrate recognition, binding and unfolding (Rosano et al., 2011; Bruch et al., 2012). Despite constitutive expression of several catalytic units, ClpD shows a dynamic pattern of expression, induced by age and dark-induced senescence (Andersson et al., 2004; Lin & Wu, 2004; Roberts et al., 2012). We found that ClpD and ClpX are the only members of the large Clp family (at least 23 members in Arabidopsis; Adam et al., 2001) that are among the 3716 senescence-induced genes (Mueller-Roeber & Balazadeh, 2014). Therefore, ClpD may be involved in the recruitment of the Clp complex for degradation of chloroplastic proteins in senescing leaves.
Given the higher rates of protein degradation in senescent leaves, various free amino acids accumulate which are further metabolized, or transported to the sinks. Consistently, we observed high levels of free Gln, Asn, Trp, His, Met, Lys, Leu, Ile, Val, Arg, β-Aln, Phe, Tyr and GABA during extended darkness in all lines, while Thr, Ser, Glu and pyroglutamate (pGlu; a chemical produced by Glu and Gln metabolism and by glutathione degradation; Kumar & Bachhawat, 2012) accumulated only modestly in tandem (Figs 3, 4, S7). Nevertheless, the rate of accumulation of many amino acids in RD26Ox lines was significantly different from that in WT during dark treatment. The higher and more rapid accumulation of Gln and Asn, the two main amino acid sources of N remobilization during senescence, and His and Trp in RD26Ox lines suggest a regulatory role for RD26 in triggering remobilization of organic N during senescence due to enhanced protein degradation triggered by elevated RD26 expression.
Dismantling of the photosynthetic apparatus and remobilization of sugars to sinks during senescence limits the availability of carbohydrates for respiration. In such conditions, proteins, lipids and Chl act as alternative, albeit less efficient, substrates (Ishizaki et al., 2005; Kunz et al., 2009; Araújo et al., 2011). Amino acids can be either converted to pyruvic acid or acetyl-CoA before entering the TCA cycle, or they might directly enter the TCA cycle after being converted to one of its intermediates, for example 2-oxoglutarate (Araújo et al., 2011). In addition, degradation of Lys, BCAAs and AAAs directly channels electrons to the plant's ubiquinol pool within the mETC. Electrons are mainly provided by the action of IVDH, using isovaleryl-CoA as an intermediate catabolite of these amino acids. In accordance with previous studies (Gibon et al., 2006; Fahnenstich et al., 2007; Araújo et al., 2010) we confirmed the accumulation of Lys, BCAAs and AAAs during dark-induced senescence. These amino acids show higher levels of accumulation in ivdh mutant plants than WT during dark treatment, in accordance with the key role of this enzyme in their catabolism (Araújo et al., 2010). Somewhat in contrast to the promotion of senescence in RD26Ox plants is the repression of the dark-induced accumulation of Lys, Leu and Ile in these plants (compared to WT), which suggests a role for RD26 in triggering their catabolism. The LKR/SDH bifunctional enzyme specifically catalyzes Lys degradation, producing acetyl-CoA and glutamate (Zhu et al., 2000). We demonstrate that RD26 positively regulates expression of both IVDH and LKR/SDH, of which at least the latter is a direct target. Therefore, we conclude that RD26 promotes degradation of Lys and BCAAs through the induction of LKR/SDH and IVDH during senescence. Interestingly, expression of LKR/SDH, IVDH and RD26 is induced and inhibited by sugar starvation and resupply, respectively (Daschner et al., 2001; Stepansky & Galili, 2003; Osuna et al., 2007). Therefore, RD26 appears to participate in favor of catabolism of amino acids, providing alternative sources of energy in senescing leaves.
Our results also suggest a positive regulatory role of RD26 on phytol degradation. The catabolism of phytol released from Chl degradation has previously been suggested to supply electrons to the mETC, via IVDH (Ishizaki et al., 2005; Araújo et al., 2010). In Arabidopsis, free toxic phytol is converted to tocopherol and fatty acid phytyl esters. The conversion to phytyl ester is catalyzed by phytyl ester synthase, encoded by PES1 and PES2 (Lippold et al., 2012). We demonstrate that RD26 directly regulates PES1 expression. Notably, the two genes GSTU3 and UGT76E12 with a predicted role in the catabolism and β-oxidation of fatty acids are also positively regulated by RD26 (Table S2). Taken together, RD26 accelerates the degradation of Chl catabolite intermediates, providing a further mechanism to feed metabolites into the alternative respiration pathway during senescence.
GABA is a nonproteinogenic amino acid and its accumulation reflects an alteration of N/C balance in the favour of C. During C starvation, the GABA shunt converts GABA to succinate which functions both as a TCA cycle intermediate and as an electron donor to the mETC (Kinnersley & Turano, 2000; Fait et al., 2008; Michaeli & Fromm, 2015). Notably, GABA-T and SSADH, constituent enzymes of the GABA shunt, are induced in RD26Ox plants (Fig. S10b).
Collectively, the upregulation of the TCA cycle in RD26Ox plants during extended darkness might be linked to the impact of RD26 on promoting the catabolism of Lys, BCAAs, phytol and GABA, all of which provide alternative substrates for mitochondrial respiration under C starvation conditions. The higher accumulation of TCA cycle intermediates in RD26Ox plants during darkness, compared to WT, might indicate their yet incomplete utilization due to massive supply from the upstream degradative processes. Furthermore, accumulation of TCA cycle intermediates during dark incubation has previously been suggested as a symptom of higher respiration (Araújo et al., 2010, 2011).
In senescing leaves the reduction of starch levels is accompanied by an accumulation of mono- and disaccharides, which are then transported to the sinks as a mechanism of nutrient remobilization (Quirino et al., 2001; Rolland et al., 2006; Yamada et al., 2010). RD26 is induced by C starvation, and several genes with a role in carbohydrate metabolism are regulated by RD26. SWEET15 is a member of the SWEET sucrose efflux transporter family; its overexpression enhances leaf senescence and retards plant growth (Seo et al., 2011; Chen et al., 2015). RD26 directly induces SWEET15 expression, leading to phenotypes similar to those of SWEET15 overexpressors. In addition, RD26 directly induces SFP1, which encodes a monosaccharide transporter. The induction of SFP1 in senescing leaves is paralleled by an accumulation of monosaccharides (Quirino et al., 2001). RD26 also elevates expression of ESL1, which encodes a tonoplast-located facilitated diffusion transporter for monosaccharides highly induced during senescence (Yamada et al., 2010). Collectively, our data indicate strongly that RD26 functions to favor sucrose transport during senescence. Furthermore, RD26 controls carbohydrate catabolism by elevating the expression of AMY1, AMY2, SUS1 and SUS4 (Fig. S10a). AMY and SUSs proteins are involved in starch and sucrose catabolism, respectively (Doyle et al., 2007; Jie et al., 2009; Santaniello et al., 2014).
An important aspect with respect to the gene expression changes induced by RD26 is whether such changes occur primarily by the action of RD26, or more indirectly through physiological changes triggered by an elevated expression of RD26. As reported here, changes in gene expression occurred as soon as 2–5 h after induction of RD26 by EST in the RD26-IOE plants, and several of the genes activated by RD26 induction are direct targets of the TF. To determine whether RD26-incuced transcriptome reprogramming occurs before or after physiological changes typically observed in senescing tissues, that is, a decline in protein and starch content, we induced RD26 expression in RD26-IOE seedlings for 0, 5, 12, 24 and 48 h and determined the levels of protein and starch. As seen in Fig. S11(a), a decline of protein content became detectable only 12 h after start of the EST treatment, while a decrease in starch content became evident after 24 h (Fig. S11b). These data clearly demonstrate that RD26-induced changes in gene expression (after EST induction) occur well before changes of senescence-related physiological parameters, as expected for a TF regulating a senescence-associated GRN.
The fact that RD26 controls multiple catabolic genes highlights the importance of transcriptional control for achieving the orderly rearrangement of metabolism during senescence by which a photosynthetically active leaf decays to allow the recovery of nutrients for growth of newly developing organs. Importantly, RD26 controls the expression of genes across the entire spectrum of the cellular degradation hierarchy, from dismantling chloroplasts (through CV and ClpD) and degrading starch down to amino acids and the induction of phytol catabolism, which is essential for mitochondrial energy gain during the late stages of senescence when chloroplasts are dysfunctional (Fig. 5). RD26 thus has a key function during leaf senescence: it orchestrates the degradation of the main energy-producing organelle in leaves, the chloroplast, with feeding of carbon into the TCA cycle/mitochondrial respiratory chain. RD26 therefore represents the first regulator to exert such a wide-ranging and crucial control over metabolic reprogramming during senescence.
Acknowledgements
We thank the University of Potsdam and the Max Planck Institute of Molecular Plant Physiology for supporting our research. I.K. thanks the Iran Ministry of Science and the Chamran University of Ahvaz for financial support. We thank Eugenia Maximova (MPI of Molecular Plant Physiology) for help with confocal microscopy, and Karin Koehl (MPI of Molecular Plant Physiology) and Christiane Schmidt (University of Potsdam) for plant care. We are grateful to Tsanko Gechev (University of Potsdam) for assisting in framing the AbioSen project. This research was supported by the Deutsche Forschungsgemeinschaft (FOR 948; MU 1199/14-2, BA4769/1-2, and ERA-CAPS ‘AbioSen’, MU 1199/16-1).
Author contributions
B.M-R. and S.B. initiated the research and supervised the group; I.K., B.M-R. and S.B. designed the experiments; I.K. performed the experiments and interpreted the data; T.T. and A.R.F. performed the metabolomic profiling; G-P.X. performed the CELD experiments; M.S. contributed to generating constructs; I.K. and B.M-R. wrote the manuscript with contributions from S.B.
