1. Cell Biology
  2. Plant Biology

SnRK1-triggered switch of bZIP63 dimerization mediates the low-energy response in plants

  1. Andrea Mair
  2. Lorenzo Pedrotti
  3. Bernhard Wurzinger
  4. Dorothea Anrather
  5. Andrea Simeunovic
  6. Christoph Weiste
  7. Concetta Valerio
  8. Katrin Dietrich
  9. Tobias Kirchler
  10. Thomas Nägele
  11. Jesús Vicente Carbajosa
  12. Johannes Hanson
  13. Elena Baena-González
  14. Christina Chaban
  15. Wolfram Weckwerth
  16. Wolfgang Dröge-Laser
  17. Markus Teige  Is a corresponding author
  1. University of Vienna, Austria
  2. University of Würzburg, Germany
  3. Max F. Perutz Laboratories, University of Vienna, Austria
  4. Instituto Gulbenkian de Ciência, Portugal
  5. University of Tübingen, Germany
  6. Universidad Politécnica de Madrid, Spain
  7. Utrecht University, Netherlands
  8. Umeå University, Sweden
  9. University of Natural Resources and Life Sciences, Austria
https://doi.org/10.7554/eLife.05828
Share this article
Cite this article
  1. Andrea Mair
  2. Lorenzo Pedrotti
  3. Bernhard Wurzinger
  4. Dorothea Anrather
  5. Andrea Simeunovic
  6. Christoph Weiste
  7. Concetta Valerio
  8. Katrin Dietrich
  9. Tobias Kirchler
  10. Thomas Nägele
  11. Jesús Vicente Carbajosa
  12. Johannes Hanson
  13. Elena Baena-González
  14. Christina Chaban
  15. Wolfram Weckwerth
  16. Wolfgang Dröge-Laser
  17. Markus Teige
(2015)
SnRK1-triggered switch of bZIP63 dimerization mediates the low-energy response in plants
eLife 4:e05828.
https://doi.org/10.7554/eLife.05828
  2. Download BibTeX
  3. Download .RIS

Abstract

Metabolic adjustment to changing environmental conditions, particularly balancing of growth and defense responses, is crucial for all organisms to survive. The evolutionary conserved AMPK/Snf1/SnRK1 kinases are well-known metabolic master regulators in the low-energy response in animals, yeast and plants. They act at two different levels: by modulating the activity of key metabolic enzymes, and by massive transcriptional reprogramming. While the first part is well established, the latter function is only partially understood in animals and not at all in plants. Here we identified the Arabidopsis transcription factor bZIP63 as key regulator of the starvation response and direct target of the SnRK1 kinase. Phosphorylation of bZIP63 by SnRK1 changed its dimerization preference, thereby affecting target gene expression and ultimately primary metabolism. A bzip63 knock-out mutant exhibited starvation-related phenotypes, which could be functionally complemented by wild type bZIP63, but not by a version harboring point mutations in the identified SnRK1 target sites.

https://doi.org/10.7554/eLife.05828.001

eLife digest

Organisms need to adjust their metabolism in response to changing environmental conditions to ensure that they balance their energy intake with the demands of growth and reproduction. In plants, an enzyme called SnRK1 plays a crucial role in responses to starvation in two ways: by altering the activities of enzymes involved in metabolism and by regulating the expression of genes. To perform the second job, SnRK1 is thought to control the activity of proteins called transcription factors—which alter the expression of genes by binding to DNA—but it is not known which ones.

SnRK1 has ‘kinase’ activity, that is, it can alter the activities of other proteins by adding small molecules called phosphates to them. It has been suggested that a group of transcription factors called the bZIP proteins may be regulated by SnRK1. Two bZIP proteins work together to switch on a gene, and the combination of bZIP proteins that interact can influence which genes are switched on. Here, Mair et al. studied the role of a bZIP protein called bZIP63 during starvation in the plant Arabidopsis.

The experiments show that bZIP63 is involved in controlling responses to starvation. Furthermore, its activity is regulated by SnRK1, which adds phosphates to three specific locations on the protein. These phosphates alter the ability of bZIP63 to interact with other bZIP proteins, leading to changes in gene expression during starvation.

Mair et al. triggered starvation in Arabidopsis plants by keeping the plants in darkness for several days. The leaves of normal plants turn yellow in response to starvation, but the leaves of mutant plants that lacked bZIP63 remained green. In contrast, plants containing higher amounts of this bZIP protein showed the opposite effect and their leaves turned yellow much more quickly than normal plants. The mutant plants that lacked bZIP63 could be rescued by the normal protein, but not by another version of the protein that SnRK1 is unable to add phosphates to.

These data suggest that SnRK1 regulates bZIP63 activity to alter metabolism in response to starvation. Mair et al. propose a model in which the ability of bZIP63 to interact with other bZIPs is normally rather low. However, when the plants are starved, SnRK1 adds phosphates to bZIP63, which increases its ability to bind to other bZIP proteins and leads to changes in gene expression. The bZIP proteins are also found in animals; therefore a future challenge is to find out whether these proteins are also regulated in a similar way.

https://doi.org/10.7554/eLife.05828.002

Introduction

Flexibility in the regulation of gene expression is crucial for all organisms to adjust their metabolism to changing growth conditions. Particularly under stress, available energy resources need to be balanced between defense and growth. The SUCROSE NON-FERMENTING RELATED KINASE 1 (SnRK1) in plants and its orthologs, the sucrose-non-fermenting 1 (Snf1) kinase in yeast and the AMP-dependent protein kinase (AMPK) in mammals, are well-known and crucial master regulators of energy homeostasis. SnRK1 is involved in the regulation of plant metabolism, development, and stress response (Polge and Thomas, 2007; Baena-González and Sheen, 2008), Snf1 is required for the switch from fermentative to oxidative metabolism in the absence of glucose (Hedbacker and Carlson, 2008), and AMPK regulates glucose, lipid, and protein metabolism, mitochondrial biogenesis, and feeding behavior in animals (Hardie et al., 2012). They are generally activated under energy starvation conditions and trigger metabolic reprograming to slow down energy-consuming processes and turn on pathways for alternative energy production in order to survive the stress conditions (Hardie, 2007; Tome et al., 2014). This happens, in two ways: by direct phosphorylation and modulation of the activity of key enzymes in nitrogen, carbon, or fatty acid metabolism (Sugden et al., 1999; Kulma et al., 2004; Harthill et al., 2006), and by massive transcriptional reprogramming (Polge and Thomas, 2007; Baena-González and Sheen, 2008; McGee and Hargreaves, 2008). Especially in plants, the latter aspect, the regulation of transcription, is still poorly understood. In Arabidopsis protoplasts, transient overexpression of AKIN10, a catalytic subunit of the SnRK1 complex, resulted in a transcriptional profile reminiscent of various starvation conditions and led to the identification of 1021 putative SnRK1 target genes (Baena-González et al., 2007). However, the transcription factors mediating the transcriptional response of SnRK1 to energy starvation are still unknown. Based on reporter gene activation assays in protoplasts (Baena-González et al., 2007) and modelling of microarray data (Usadel et al., 2008), some members of the C/S1 group of basic leucine zipper (bZIP) transcription factors (TFs)—foremost bZIP11 and bZIP1 from the S1 group—were speculated to be involved in this process. Yet, a direct regulation of these bZIPs by SnRK1 has never been shown.

bZIP proteins form a large and highly conserved group of eukaryotic TFs. They bind the DNA as dimers and are characterized by a basic region for specific DNA binding and a leucine zipper for dimerization (Deppmann et al., 2006; Reinke et al., 2013). They are involved in a multitude of cellular processes, including cell proliferation and differentiation, metabolism, stress response, and apoptosis (Mayr and Montminy, 2001; Jakoby et al., 2002; Motohashi et al., 2002; Rodrigues-Pousada et al., 2010; Tsukada et al., 2011). The diversity and flexibility of transcriptional regulation by bZIP TFs can at least partially be attributed to their potential to form variable dimer combinations, which bind to different consensus target sites (Deppmann et al., 2006; Tsukada et al., 2011). While the leucine zipper determines the possible dimer combinations (Deppmann et al., 2006; Reinke et al., 2013), the actual in vivo dimer composition is further influenced by factors such as protein availability, binding of regulatory proteins, or post-translational modifications (Kim et al., 2007; Schuetze et al., 2008; Lee et al., 2010). Since the initial discovery that the mammalian bZIP cAMP response element binding protein (CREB) is regulated by reversible phosphorylation, many bZIP TFs were reported to be phosphorylated (Holmberg et al., 2002; Schuetze et al., 2008; Tsukada et al., 2011). However, particularly in plants, the functional consequences of these phosphorylation events often remained unclear. For example, it has been known for several years that abscisic acid (ABA)-dependent phosphorylation of some ABA-responsive element binding proteins (AREBs) by SnRK2 kinases increases their transcriptional activity (Furihata et al., 2006), yet the underlying mechanism of this activation is still unknown. It is also surprising that, while many examples for phosphorylation-dependent regulation of bZIP activity, DNA-binding, subcellular localization, stability, and interaction with regulatory proteins are known (Schuetze et al., 2008; Tsukada et al., 2011), reports on the regulation of dimerization are scarce. So far, only three publications (Kim et al., 2007; Guo et al., 2010; Lee et al., 2010) showed compelling evidence for phosphorylation-dependent changes in bZIP dimerization in animals. Still, even in these cases it is often not entirely clear whether bZIP phosphorylation affects dimerization directly or indirectly by enhancing DNA binding.

bZIP63 is a member of the C-group of Arabidopsis bZIPs, which was proposed to play a role in energy metabolism, seed maturation, and germination under osmotic stress (Jakoby et al., 2002; Correa et al., 2008; Veerabagu et al., 2014). Its transcriptional profile indicates that bZIP63 could be involved in the (energy) starvation response, as transcription and mRNA stability are repressed by sugars and ABA and mRNA levels increase in the night and even more during extended night treatments (Matiolli et al., 2011; Kunz et al., 2014). A small set of potential target genes for bZIP63 has been identified, including genes involved in amino acid metabolism (ASN1/DIN6 = ASPARAGINE SYNTHETASE 1, ProDH = PROLINE DEHYDROGENASE), energy starvation response (DIN10 = RAFFINOSE SYNTHASE 6), and senescence (SEN1 = SENESCENCE 1) (Baena-González et al., 2007; Dietrich et al., 2011; Matiolli et al., 2011; Veerabagu et al., 2014). The C-group bZIPs form a dimerization network with the S1-group in plants, in which bZIP63 can interact with all members (Ehlert et al., 2006; Kang et al., 2010). Three of its dimerization partners from the S1-group—bZIP1, bZIP11, and bZIP53—were shown to be important metabolic regulators, especially under energy starvation conditions, and to regulate the expression of ASN1 and ProDH as well (Hanson et al., 2008; Dietrich et al., 2011; Ma et al., 2011). Furthermore, bZIP1 was recently confirmed as a transcriptional master regulator of the rapid response to nutrient signals controlling mainly genes involved in amino acid metabolism and cell death/phosphorus metabolism as primary targets (Para et al., 2014). In that study the authors also speculate about a posttranslational modification of bZIP1 or its binding partners.

Here we show that bZIP63 is an important metabolic regulator, especially under stress/starvation conditions, and that bZIP63 is phosphorylated at multiple sites in vivo in a sugar- and energy-dependent manner. In an unbiased approach, we identified SnRK1 as one of the kinases responsible for bZIP63 phosphorylation and found that it targets three highly conserved serine residues in the N- and C-terminus of bZIP63. Moreover, we demonstrate that the phosphorylation of these sites is crucial for bZIP63's dimerization and activity in planta and propose a molecular model for a phosphorylation-triggered switch of bZIP63 dimerization partners, which ultimately regulates metabolic reprogramming.

Results

bZIP63 controls dark-induced senescence and primary metabolism

To better understand the role of bZIP63 in the plant we first tested whether bZIP63 has a similar phenotype as its dimerization partners bZIP1 and bZIP11. Prolonged darkness was shown to induce increased chlorophyll loss in plants overexpressing bZIP1 (Dietrich et al., 2011). Therefore, we incubated a bZIP63 knock-out (ko), two independent overexpressor (ox) lines (Figure 1—figure supplement 1), and their respective wild types (wt) in the dark and determined the percentage of the green leaf area as a measure for chlorophyll content (Figure 1; Figure 1—figure supplement 2A). This method was preferred over direct chlorophyll measurements (Figure 1—figure supplement 2B) as it is not affected by the water loss in senescing leaves. While no differences were observed before dark treatment, significant differences were visible after 9 days in darkness (Figure1—figure supplement 2C–E). Similar to the bZIP1 ox, the bZIP63 ox lines (ox#2 and ox#3) had a significantly higher percentage of yellow leaf area than the wt. In contrast, bzip63 plants displayed a stay-green phenotype (Figure 1A,B; Figure1—figure supplement 2F). RT-qPCRs of different senescence marker genes confirmed that the phenotype is due to dark-induced and not to natural senescence (Figure 1—figure supplement 3). Transcriptional responses in dark-induced senescence show clear similarities to starvation-induced senescence in cell suspension culture, which likely results from carbon depletion in both systems (Buchanan-Wollaston et al., 2005). We therefore tested whether addition of sugar could rescue this phenotype by performing the dark-induced senescence experiment with seedlings grown on agar plates with and without addition of sugar. Indeed, we found that the phenotype could be rescued by the addition of glucose (Figure 1C,D Figure 1—figure supplement 4), supporting the suggested role of bZIP63 in energy/carbon starvation response.

Figure 1 with 4 supplements see all
Download asset Open asset
bZIP63 mutants have a phenotype in dark-induced senescence.

(A) and (B) Dark-induced senescence phenotype of 4.5 week-old soil grown plants. Comparison of a bzip63 line in the Wassilewskya (Ws-2), and two bZIP63 ox lines in the Columbia (Col-0) background, after 9 days in darkness. (A) Representative leaf series. (B) Box-and-whiskers plot of the total green leaf area of eight biological replicates as determined with ImageJ. See Figure 1—figure supplement 1 for molecular characterization of the bZIP63 mutant lines, Figure 1—figure supplement 2 for a scheme of the determination of the green leaf area, quantitative chlorophyll measurements, controls and green area of individual leaves, Figure1—source data 1 for the used ImageJ macro, and Figure 1—figure supplement 3 for expression of senescence marker genes in wt and bzip63. (C) and (D) Sugar rescue of the dark-induced senescence phenotype. Seedlings were grown for 12 days on ½ MS agar containing 0.5% sucrose, transferred to ½ MS agar containing 0% or 2% glucose, and grown for another 6 days before incubation in the dark for 7 days. (A) Representative seedlings. (B) Box-and-whiskers plot of the chlorophyll content of 8 seedlings per line and condition. See Figure 1—figure supplement 4 for pictures and chlorophyll content of seedlings before dark incubation. p-values from T-tests between mutants and wt < 0.05, < 0.01, and < 0.001 are indicated by *, **, and ***, respectively.

https://doi.org/10.7554/eLife.05828.003
Figure 1—source data 1

ImageJ macro for determination of leaf area and green leaf area.

https://doi.org/10.7554/eLife.05828.004

The notion that several hetero-dimerization partners of bZIP63, including bZIP1 and bZIP11, are important metabolic regulators under starvation conditions (Dietrich et al., 2011; Ma et al., 2011) prompted us to perform an unbiased metabolomics analysis of bzip63 and ox#3 plants and their respective wt lines. Leaves of 5 week-old plants were harvested after 6 hr of light and extended night and analyzed for changes in the primary carbon and nitrogen metabolism using gas chromatography coupled to mass spectrometry (Figure 2; Figure 2—figure supplement 1 and Figure 2—source data 1). Intriguingly, almost all amino acid levels were increased in the bZIP63 ko and decreased in the ox plants. This effect was even more pronounced after the extended night treatment. The biggest differences were observed for proline and the entire glutamate family. This accumulation of amino acids, particularly of proline, is striking in view of the observed senescence phenotype as it has been suggested that proline serves as an alternative energy source, especially under low carbon conditions (Szabados and Savouré, 2010; Szal and Podgórska, 2012).

Figure 2 with 1 supplement see all
Download asset Open asset
bZIP63 mutants have an altered primary metabolism.

Metabolic phenotype of 5 week-old soil grown plants after 6 hr light (L) and extended night (E). Log-2 fold changes of metabolite levels in ko and ox compared to their respective wt, displayed on a simplified map of the central primary metabolism. Values are means of five biological replicates. p-values from T-tests between mutants and wt < 0.05, < 0.01, and < 0.001 are indicated by *, ** and ***, respectively. For more details including mean values and SD see Figure 2—figure supplement 1 and Figure 2—source data 1.

https://doi.org/10.7554/eLife.05828.009
Figure 2—source data 1

Excel table of relative metabolite levels in bZIP63 mutants and p-values from T-tests.

Table of relative metabolite levels in Ws-2, bzip63, Col-0 and ox#3 after 6 hr in light or extended night, which were used to calculate the log2-fold changes shown in Figure 2 and Figure 2—figure supplement 1. Mean values and SD of five biological replicates are given as fold changes to the corresponding wt. p-values from T-tests between wt and mutant are listed.

https://doi.org/10.7554/eLife.05828.010

bZIP63 is phosphorylated at multiple sites in an energy-dependent manner

Kirchler et al. (2010) showed that bZIP63 can be phosphorylated in vitro by crude Arabidopsis extracts. To test whether bZIP63 is also phosphorylated in vivo, we treated total leaf protein extracts of ox#3 plants, expressing GFP-tagged bZIP63, with lambda protein phosphatase (λPP) and separated treated and untreated extracts on 2D gels by isoelectric focusing (IEF) in the first, and SDS-PAGE in the second dimension (Figure 3A). λPP treatment induced a clear shift of bZIP63 towards the basic region of the IEF strip, thus indicating dephosphorylation of the protein.

Figure 3 with 2 supplements see all
Download asset Open asset
bZIP63 is phosphorylated at multiple sites in an energy-dependent manner in vivo.

(A) 2D gel western blots (αGFP) of lambda phosphatase (λPP)-treated protein extracts from adult soil-grown plants expressing bZIP63-GFP (ox#3). (B) Phos-tag gel western blots showing the in vivo phosphorylation state of bZIP63 in seedlings after 6 hr extended night in the presence (+) or absence (−) of 1% sucrose and in 5 week-old soil-grown plants after 6 hr light (L) or extended night (EN). Plants either expressed 35S::bZIP63-GFP (ox#3) or a genomic fragment of bZIP63 (GY11) with a YFP-tag. Recombinant bZIP63-YFP was used as a nonphosphorylated control. Numbered arrowheads on the right mark the position of each observed bZIP63 band for easy reference with other figures (see Figure 3—figure supplement 1 for a comparative image of all Phos-tag western blots). For more information on the genomic line see Figure 7 and Figure 7—figure supplement 1. (C) Identification of in vivo phosphorylation sites by immunoprecipitation (IP) and tandem mass spectrometry (MS/MS). An exemplary western blot of the IP is shown. The scheme at the bottom shows the positions of the identified in vivo phosphorylation sites and the total sequence coverage reached with each proteolytic enzyme (grey bars). Asterisks mark the identification of a phospho-site. For more information on samples and (phospho-) peptides see Figure 3—figure supplement 2. IEF, isoelectric focusing; CBB, coomassie brilliant blue; BD, basic domain; ZIP, leucine zipper; LTQ, linear ion trap quadrupole; chym., chymotrypsin.

https://doi.org/10.7554/eLife.05828.012

As bZIP63 expression is strongly regulated by the day/night cycle and by sugars (Matiolli et al., 2011; Kunz et al., 2014), we investigated its phosphorylation status under these conditions, applying the Phos-tag technique to enhance phosphorylation-induced mobility shifts in 1D SDS-PAGE (Kinoshita et al., 2006). Comparison of protein extracts from seedling cultures after 6 hr extended night in the presence or absence of sucrose or leaves harvested after 6 hr of light or extended night revealed strong differences in the phosphorylation patterns of bZIP63 (Figure 3B; Figure 3—figure supplement 1). Compared to the recombinantly expressed (not phosphorylated) bZIP63-YFP, the majority of plant-expressed bZIP63-GFP appeared in several slower migrating forms, thus indicating multiple and differentially phosphorylated forms. In total, we were able to distinguish eight different forms on the Phos-tag gels, depending on the conditions (Figure 3B; Figure 3—figure supplement 1). However, it is important to note, that not every phosphorylation event results in an equal shift. In the light, two strong bands were visible, possibly reflecting the two spots on the 2D gel (Figure 3B, labelled as band 6 and band 7, respectively). Under extended night conditions, an additional band appeared (labelled as band 8), indicating increased phosphorylation of bZIP63. Notably, the ratio of the most phosphorylated form of bZIP63 (band 8) to the two less phosphorylated forms (band 6 and 7) was different between the ox (ox#3) and a genomic complementation line of bzip63 (GY11, detailed description follows later). In the genomic complementation line, the most phosphorylated form of bZIP63 was the strongest band under extended night conditions, which might be a result of the different bZIP63 amounts in relation to the endogenous kinase(s), and likely better reflects the phosphorylation state of the endogenous protein. Interestingly, in seedlings grown in liquid culture, the extended night-triggered phosphorylation of bZIP63 could be abolished by the addition of 1% of sucrose (Figure 3B). In fact, phosphorylation was even lower here than in the light. These data indicate that bZIP63 has a basal level of phosphorylation in the plant and undergoes hyper-phosphorylation under starvation conditions. In contrast, addition of external sugars leads to a reduced phosphorylation status of bZIP63.

In order to identify the in vivo phosphorylated residues, bZIP63-GFP was immunoprecipitated from total leaf extracts using bead-coupled anti-GFP antibodies, subjected to proteolytic digest and analyzed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) (Figure 3C). To achieve maximum sequence coverage of the protein we combined proteolytic digests from four proteolytic enzymes (trypsin, chymotrypsin, LysC, and subtilisin). This approach resulted in a total sequence coverage of 93.6% and the identification of several phospho-peptides, indicating that bZIP63 is phosphorylated at up to seven serines (S29, S59, S102, S160, S261, S294, and S300) in vivo (Figure 3C; Figure 3—figure supplement 2). Two of these sites—S29 and S300—were also found in a recent phospho-proteomics study (Umezawa et al., 2013). Notably, only three of the seven sites—S29, S294, and S300—were found by tryptic protein digest, underpinning the advantage of alternative proteolytic digests for phospho-peptide identification in targeted proteomics.

SnRK1, CDPKs and CKII are potential upstream kinases of bZIP63

To identify potential upstream kinases of bZIP63, we performed in-gel kinase assays with plant protein extracts from roots of hydroponically grown Arabidopsis plants or root cell culture using recombinantly expressed bZIP63 as substrate. Root material was chosen because in leaf extracts the large amount of Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) interfered with some of the kinase signals and poses a considerable problem during MS-based protein identification. Three strong bands of about 40, 50, and 55 kDa, respectively, were visible on the autoradiogram (Figure 4A), indicating that at least three kinases can phosphorylate bZIP63 in vitro. These bands were not visible on a gel without substrate, excluding the possibility that they originate from kinase auto-phosphorylation (Figure 4—figure supplement 1). To reduce the sample complexity and to enrich low abundant bZIP63-binding proteins before kinase identification, we affinity purified plant protein extracts on immobilized bZIP63. The eluted fractions were tested in an in-gel kinase assay for kinase activity towards bZIP63 and loaded on a normal SDS-PAGE gel without substrate for kinase identification. Bands corresponding in molecular weight (MW) to the signal from the in-gel kinase assay were excised from the gel, digested with trypsin and analyzed by LC-MS/MS (Figure 4B,C). In total, 27 protein kinases and kinase complex subunits were identified in four independent experiments (Figure 4—source data 1; Figure 4—source data 2). From those, proteins which did not match the expected MW or for which it had been shown experimentally that they are not localized in the nucleus were excluded. Only proteins which were identified in more than one sample with at least one proteotypic peptide were considered to be high confidence candidates, resulting in six protein kinases and two kinase regulatory subunits (Figure 4C; Figure 4—source data 1; Figure 4—source data 2). We identified the two main catalytic subunits of SnRK1, AKIN10 and AKIN11, as well as the regulatory subunit SNF4. Several members of the calcium dependent protein kinase (CDPK) family were found, but only CPK3 was identified with high confidence. In addition, two catalytic (CKA1 and CKA2) and one regulatory subunit (CKB1) of casein kinase II (CKII) were found, as well as Casein kinase 1-like protein 2 (CKL2). The SnRK1 kinases, CDPKs, and CKL2, correspond in MW to the two upper bands, while the lower band corresponds to the CKII kinase subunits. As SnRK1 was previously reported to enhance the activity of several C/S1 group bZIPs (Baena-González et al., 2007) and was suggested to be activated under energy starvation conditions (Baena-González and Sheen, 2008)—which could explain the observed hyper-phosphorylation of bZIP63 in extended night—we focused our further analysis on AKIN10 and AKIN11.

Figure 4 with 1 supplement see all
Download asset Open asset
Several different kinases can phosphorylate bZIP63.

(A) In-gel kinase assay with a root protein extract from hydroponically grown wild type plants and bZIP63 as substrate. Arrows indicate the positions of potential bZIP63 kinases. (B) Scheme of the kinase identification process. (C) In-gel kinase assay with samples from affinity purification of a root protein extract with bZIP63 and bZIP63 as a substrate (left) and a list of catalytic and regulatory (*) kinase subunits identified with high confidence (right). The list also contains the gene identifier (AGI), molecular weight (MW), and number of samples in which the protein was found. For controls and a list of all identified kinases and kinase peptides see Figure 4—figure supplement 1 and Figure 4—source data 1, 2. CBB, Coomassie brilliant blue.

https://doi.org/10.7554/eLife.05828.015
Figure 4—source data 1

Overview over the kinases identified by LC-MS/MS after affinity purification with bZIP63.

https://doi.org/10.7554/eLife.05828.016
Figure 4—source data 2

Excel table containing a detailed overview over the identified kinases and analyzed samples as well as all peptides found for the kinase subunit.

The tab ‘kinase overview’ contains a table showing in which samples each kinase subunit was identified, including the number of peptides, proteotypic peptides, MASCOT (Matrix Science) score, and expected molecular weight. The other tabs list all peptides found for the kinase subunits in each sample and state whether the peptide is proteotypic (Y) or not.

https://doi.org/10.7554/eLife.05828.017

The SnRK1 kinase AKIN10 interacts with and phosphorylates bZIP63 in vivo

To confirm that AKIN10 and AKIN11 phosphorylate bZIP63, we first performed an in-gel kinase assay with protein extracts of wt and akin10 seedlings in the presence of EGTA, to reduce the signal from CDPKs of the same MW (Figure 5A). In both root and leaf extracts of akin10 plants one band at the expected MW of AKIN10 nearly disappeared (see Figure 5—figure supplements 1–3 for characterization of the akin10 line). As AKIN11 has approximately the same MW as AKIN10, the remaining signal likely originates from AKIN11. In vitro kinase assays, with equal amounts of both kinases, showed that both kinases can phosphorylate bZIP63. However, AKIN10 phosphorylates bZIP63 much stronger than AKIN11 does (Figure 5B). Addition of the SnRK1 upstream kinase SnAK2 increased the activity of AKIN10 and AKIN11 about sevenfold (see also Crozet et al., 2010), but had no effect on the ratio between the signal intensities (Figure 5—figure supplement 4). In vivo interaction assays with the identified SnRK1 complex subunits supported the findings of the kinase assays. In both yeast two-hybrid (Y2H) (Figure 5C) and bimolecular fluorescence complementation (BiFC) assays (Figure 5D) AKIN10 and the regulatory subunit SNF4 interacted strongly with bZIP63, to a comparable level as measured for the bZIP63 homo-dimer, which was used as positive control (Walter et al., 2004). In contrast, AKIN11 and the two regulatory subunits AKINβ1 and AKINβ2 showed almost no interaction with bZIP63 (Figure 5C,D; Figure 5—figure supplement 5). However, it has to be considered that AKIN10 and AKIN11 are part of a trimeric complex including SNF4 (Emanuelle et al., 2015), which is neglected in these two assays. It is therefore still possible that AKIN11 interacts indirectly with bZIP63 via the regulatory subunit of the SnRK1 complex and plays a minor role in bZIP63 phosphorylation in the plant.

Figure 5 with 7 supplements see all
Download asset Open asset
The SnRK1 kinase AKIN10 phosphorylates bZIP63 and interacts with bZIP63 in vivo.

(A) In-gel kinase assay with protein extracts from wt and akin10 plants and bZIP63 as a substrate (top), western blot against AKIN10 (αAKIN10, middle), and Coomassie brilliant blue stain (CBB, bottom). Asterisks mark the position of AKIN10. For characterization of the akin10 line see Figure 5—figure supplements 1–3. (B) In vitro kinase assay with recombinant AKIN10/AKIN11 and bZIP63 as a substrate. See also Figure 5—figure supplement 4 for kinase assays including the SnRK1 upstream kinase SnAK2. (C) and (D) Interaction of SnRK1 subunits with bZIP63. Homo-dimerization of bZIP63 was used as a positive control. (C) Yeast two-hybrid (Y2H) assay with auto-activation in grey and interaction with bZIP63 in blue. Bars represent means ± SD of eight biological replicates. For Y2H with AKINβ1 and β2 see Figure 5—figure supplement 5. (D) Laser scanning microscopy images of bimolecular fluorescence complementation (BiFC) in transiently transformed Nicotiana tabacum leaves (top). Arrowheads indicate the position of the nucleus. Size bar = 20 µm. Expression of the fusion proteins was verified by western blots (αHA, αFlag, bottom). (E) Phos-tag gel western blots showing the in vivo phosphorylation state of bZIP63 in 4–5 week-old soil grown plants after 6 hr light (L) or extended night (EN) in the presence and absence of AKIN10 alone or both AKIN10 and AKIN11. Plants overexpressing bZIP63-YFP in the akin10 line were compared to plants overexpressing bZIP63-GFP in the wt background (ox#3) (left). Additionally, AKIN10 and AKIN11 were knocked down (akin10/11) by virus-induced gene silencing (VIGS) in plants expressing genomic bZIP63 with a YFP-tag (GY9 line). See Figure 5—figure supplement 7 for images and a western blots of the VIGS plants. Recombinant bZIP63-YFP was used as a nonphosphorylated control. Numbered arrowheads on the right mark the position of each observed bZIP63 band for easy reference with other figures (see Figure 3—figure supplement 1 for a comparative image of all Phos-tag western blots). A comparison of the phosphorylation state of bZIP63 in seedlings can be found in Figure 5—figure supplement 6.

https://doi.org/10.7554/eLife.05828.019

To verify that SnRK1 plays an important role in the in the in vivo phosphorylation of bZIP63 we compared the phosphorylation state of bZIP63 in plants overexpressing bZIP63-GFP or -YFP in the wt and akin10 background, respectively. As SnRK1 has been suggested to act as a major regulator in the energy deprivation response, we again compared leaf protein extracts after 6 hr light and extended night and found that the hyper-phosphorylated form (band 8) of bZIP63, observed in the wt in extended night, was much weaker in the akin10 background (Figure 5E). The same effect was observed in seedling cultures after 6 hr of extended night (Figure 5—figure supplement 6). To see whether the weak phosphorylation remaining in the akin10 mutant would be further reduced in an akin10/11 double mutant, we employed virus-induced gene silencing (VIGS) to knock down AKIN10 and AKIN11 in a genomic bzip63 complementation line (GY9). Resulting plants showed strongly reduced growth and accumulation of anthocyanins (Figure 5—figure supplement 7), as previously described in Baena-González et al. (2007). Like in the akin10 line, there was no difference in bZIP63 phosphorylation between akin10/11 and control plants in light, but the hyper-phosphorylated form of bZIP63 (band 8) in the extended night was now almost completely gone (Figure 5E). This shows clearly, that AKIN10 and—to a lower extent—AKIN11 are the major kinases responsible for bZIP63 hyper-phosphorylation under starvation conditions.

Additional evidence for in vivo phosphorylation of bZIP63 by SnRK1 emerges from a recent phosphoproteomics study in which one of the in vivo phosphorylation sites in bZIP63 (S300) was found to be more abundant in an AKIN10 ox line and less abundant in the ko line after extended night treatment, respectively (Nukarinen et al., unpublished).

AKIN10 phosphorylates three conserved and functionally important serine residues in bZIP63

Next, to elucidate which of the seven in vivo phosphorylation sites can be phosphorylated by AKIN10, we performed in vitro kinase assays using the wt version of bZIP63 and different serine to alanine (S/A) mutants as substrates (Figure 6A; Figure 6—figure supplements 1, 2). Differences in phosphorylation were observed for proteins with mutations in S29, S294, and S300. In detail, the signal from S29A was strongly decreased in full length bZIP63 and completely gone in the N-terminal peptides that appeared as lower MW degradation products in these assays. The S300A mutation led to an even stronger decrease of the signal, comparable to the S294/300A double mutant. Even though the S294A single mutant showed a similar signal as the wt, all three serines had to be mutated to completely abolish phosphorylation, suggesting that all of them are in vitro targets for AKIN10, with S294 being the weakest. Importantly, all three sites match the SnRK1 consensus sequence (Huang and Huber, 2001) (Figure 6B).

Figure 6 with 4 supplements see all
Download asset Open asset
AKIN10 targets three highly conserved and functionally important serine residues in bZIP63.

(A) In vitro kinase assay of wt and S/A mutants of GST-tagged bZIP63 with AKIN10. Positions of full length (FL) and N-terminal fragments of bZIP63 are marked by black arrows. The scheme on the right shows the position of the in vivo phosphorylation sites and the in vitro target sites of AKIN10 (red asterisk) on bZIP63. See Figure 6—figure supplements 1, 2 for controls and Phos-tag gel of kinase assays, respectively. (B) Conservation of phosphorylation sites in bZIP63. Sequences of bZIP63 homologues from eight species were aligned with ClustalΩ. The scheme on top indicates the positions of the in vivo phosphorylation and AKIN10 target sites on bZIP63. The histogram below shows the sequence identity (red/black) and similarity (grey) to A. thaliana bZIP63. Red bars represent the in vivo phosphorylation sites. Below, the alignment of the sequence surrounding the AKIN10 target sites and the SnRK1 consensus motif (Huang and Huber, 2001) are shown. The grey/black shading indicates the degree of conservation, phosphorylation sites are in red. For alignment of non-AKIN10 target sites and full sequence alignment see Figure 6—figure supplement 3, for sequences in fasta format see Figure 6—source data 1. (C) Promoter activation assays in protoplasts with an ASN1/ProDH promoter-driven GUS reporter. Activation by bZIP63 wt and S/A mutants without (grey) or with (blue) co-transformation of AKIN10 is shown. Bars are means ± SD of 4 biological replicates, given in % of the activity of wt bZIP63 with AKIN10. Horizontal dashed lines indicate the signal in the control without bZIP63. Letters indicate significant differences as determined by ANOVA and pairwise T-testing (p < 0.05). See Figure 6—figure supplement 4 for a western blot control. CBB, Coomassie brilliant blue.

https://doi.org/10.7554/eLife.05828.027
Figure 6—source data 1

Sequences of the bZIP63 homologues.

Text file containing the sequences of the bZIP63 homologes used for ClustalΩ alignment in Figure 5B and Figure 5—figure supplement 3 in fasta format.

https://doi.org/10.7554/eLife.05828.028

A comparison of Arabidopsis bZIP63 with orthologs from eight other plant species, ranging from mosses to higher plants, showed that the three putative AKIN10 target sites are highly conserved throughout evolution (Figure 6B; Figure 6—figure supplement 3 and Figure 6—source data 1). As the origin of SnRK1 dates back even further, to the common ancestor of plants and animals (Bayer et al., 2014), it is likely that phosphorylation of bZIP63 by AKIN10 poses an ancient and important regulatory mechanism. We therefore set out to test the functional relevance of AKIN10-mediated phosphorylation of bZIP63. To this end, we tested the transcriptional activity of bZIP63 in protoplast-based promoter activation assays using the ASN1 or ProDH promoter fused to the beta galactosidase (GUS) reporter (Figure 6C; Figure 6—figure supplement 4). In both cases, co-transformation of wt bZIP63 and AKIN10 strongly induced reporter gene expression, while transformation of bZIP63 alone was not sufficient for significant induction. Transformation of AKIN10 alone also led to a weak induction of the reporters, which could be explained by the action of endogenous bZIPs. Mutation of S29 or all three AKIN10 target sites on bZIP63 to alanine reduced the reporter activation almost to background level. In contrast, mutation of the two C-terminal serines, S294 and S300, had only a weak negative effect on ASN1 and no significant effect on ProDH activation. Taken together, our data suggest that AKIN10 phosphorylates bZIP63 at up to three conserved sites, namely S29, S294, and S300 and phosphorylation by AKIN10, especially at S29, is crucial for bZIP63 TF activity.

The AKIN10 target sites play an important role for bZIP63 function in planta

To determine the impact of bZIP63 phosphorylation in planta, we transformed the bzip63 mutant with genomic constructs of bZIP63 containing either the wt sequence (GY lines), or a S/A mutation of S29, S294, and S300, respectively (GAY lines) with a C-terminal YFP tag (Figure 7A; Figure 7—figure supplement 1A–C). The wt construct showed strong phosphorylation in the light (bands 6 and 7) and an even stronger phosphorylation in extended night (band 8), as well as reduced phosphorylation in the presence of sucrose (bands 1, 2, and 4) (Figure 7B; Figure 7—figure supplement 1D). In contrast, the S/A construct showed only weak phosphorylation (only bands 1 and 3 visible) and most importantly no difference between all tested conditions. This indicates that S29, S294 and S300 are the major in vivo phosphorylation sites on bZIP63, which are also responsible for the observed condition-dependent shift in bZIP63 phosphorylation.

Figure 7 with 2 supplements see all
Download asset Open asset
The bzip63 phenotype can be complemented by wt bZIP63, but not by bZIP63 harboring S/A mutations of the AKIN10 target sites.

(A) Genomic complementation constructs. Exons are green, introns black. See Figure 7—figure supplement 1 for characterization of the complementation lines. (B) Phos-tag gel western blots (αGFP) showing the in vivo phosphorylation state of bZIP63 in the complementation lines after 6 hr light (L) or extended night (EN) in 5 week-old soil-grown plants. Recombinant bZIP63-YFP was used as a nonphosphorylated control. Numbered arrowheads on the right mark the position of each observed bZIP63 band for easy reference with other figures (see Figure 3—figure supplement 1 for a comparative image of all Phos-tag western blots). (C) and (D) Dark-induced senescence phenotype of 4.5 week-old soil-grown plants after 9 days in darkness. (C) Leaves 3–7 of one representative plant per line. (D) Box-and-whiskers plot of the total green leaf area of eight biological replicates. Letters indicate significant differences as determined by ANOVA and pairwise T-testing (p < 0.05). See Figure 7—figure supplement 2 for untreated plants and green leaf area of individual leaves. (E) and (F) Metabolite profile. (E) Hierarchical clustering of log-2 fold changes of metabolite concentrations compared to wt. Values are means of five biological replicates. (F) Principal component analysis (PCA). PC1 is plotted against PC2. The proportion of variance in % is indicated. The red line surrounds bzip63 and GAY samples, the green line wt, GY, and ox#3 samples. For relative metabolite concentrations and PCA loadings see Figure 7—source data 1. (G) Relative expression of potential bZIP63 target genes in 5 week-old plants during early extended night as determined by RT-qPCR. Values are means ± SD of four biological replicates and are given as fold change compared to Ws-2 at 0 hr (left) or 4 hr (right). p-values from T-tests between mutants and wt < 0.05, < 0.01, and < 0.001 are indicated by *, ** and ***, respectively. Letters indicate significant differences as determined by ANOVA and pairwise T-testing (p < 0.05). CBB, Coomassie brilliant blue.

https://doi.org/10.7554/eLife.05828.033
Figure 7—source data 1

Excel table containing the relative metabolite levels of the complementation lines and the PCA loadings.

The tab ‘metabolites normalized to wt’ contains the relative metabolite levels, which were used to calculate the fold changes shown in Figure 7E, as well as p-values from statistical tests. Values are the mean ± SD of five biological replicates given as fold change to the corresponding wt. The tab ‘PCA—variance and loadings’ contains the SD, proportion of variance, and loadings of all principal components from the PCA analysis shown in Figure 7F.

https://doi.org/10.7554/eLife.05828.034

Next, we tested whether complementation of the observed bzip63 phenotypes depends on the bZIP63 phosphorylation status, using two independent GY and GAY lines. The dark-induced senescence phenotype of bzip63 was complemented in the GY lines, but not in the GAY lines (Figure 7C,D; Figure 7—figure supplement 2). After 9 days in darkness, the ko and GAY lines showed visibly less chlorosis and had a higher percentage of green leaf area as compared to the wt and GY lines. Metabolite profiling of leaves harvested after 6 hr of light also revealed marked differences between the GY and GAY lines (Figure 7E,F; Figure 7—source data 1). The metabolite profile of the GAY lines was similar to that of bzip63 plants. In contrast, in the GY lines the metabolic changes between mutant and wt were mostly weaker than in bzip63 or even resembled those observed for ox#3 (Figure 7E). In a principal component analysis the GAY lines grouped together with bzip63, while the GY lines were closer to the two wt lines and the ox (Figure 7F).

To test the effect of the S/A mutation on the expression of bZIP63 target genes, we performed RT-qPCR of ASN1, DIN10, and ProDH (Figure 7G)—three suggested AKIN10 target genes (Baena-González et al., 2007). The expression of all three genes increased steadily during a 4 hr extended night treatment, but the increase was delayed in the bzip63 mutant as compared to the wt. At the 4 hr time point we could observe a clear difference between wt and ko. We therefore chose this time point to quantify ASN1, DIN10, and ProDH transcripts in one GY and GAY line. As expected, the GY line had the same or even more transcript than the wt for all three genes, while the GAY line had significantly lower expression levels, similar to bzip63.

In summary, expression of wt bZIP63 but not of the S/A mutant—which cannot be phosphorylated by AKIN10—in the bzip63 background led to complementation of the bzip63 phenotypes. Together, these experiments demonstrate that phosphorylation of bZIP63 at the AKIN10 target sites is essential for the function of bZIP63 in the plant.

AKIN10-mediated phosphorylation affects bZIP63 dimerization

Our findings show that bZIP63 phosphorylation, at residues distant from the central bZIP domain, strongly regulates its activity. As we did not observe any changes in localization in the bZIP63 mutants and also no change in DNA-binding activity (of the bZIP63 homodimer) towards a C-box motif (GACGTC) as canonical bZIP target site (data not shown), we suspected that the observed effect on transcription would be due to changes in dimerization preferences of bZIP63. Therefore, we tested the effect of AKIN10-mediated phosphorylation on bZIP63 homo- and hetero-dimerization with bZIP1 and bZIP11. Both bZIP1 and bZIP11 are metabolic regulators and mediate transcription of ASN1 and ProDH (Hanson et al., 2008; Dietrich et al., 2011). In protoplast two-hybrid (P2H) assays, the addition of exogenous AKIN10 resulted in a clear enhancement of dimerization in all cases—homo-dimerization as well as hetero-dimerization with bZIP1 and bZIP11 (Figure 8A; Figure 8—figure supplement 1A). Note, that the very strong effect observed with bZIP11 in these assays is misleading because bZIP11 is a much stronger activator of transcription as compared to bZIP1 and 63 (see also ‘Discussion’). From that we concluded that the phosphorylation of bZIP63 by AKIN10 is required for dimerization and set out to test the effect of S/A mutations of the AKIN10 target sites on bZIP63 homo- and hetero-dimerization with bZIP11 where we had observed the strongest effect before (Figure 8B; Figure 8—figure supplement 1B). In both cases, the signal was reduced to about 30–40% of the signal obtained from dimerization with wt bZIP63 when S29 or all three serines were mutated to alanine. Mutation of one or two of the C-terminal sites decreased bZIP63 homo-dimerization weakly but had no visible effect on bZIP63-11 dimerization, indicating that these sites play, at most, a minor role in regulation of dimer formation.

Figure 8 with 2 supplements see all
Download asset Open asset
AKIN10-mediated phosphorylation of bZIP63 promotes its dimerization.

(A) and (B) Protoplast two-hybrid (P2H) assays. (A) Interaction of bZIP63 fused to the Gal4-AD (activation domain) with bZIP63, bZIP1, and bZIP11 fused to the Gal4-BD (binding domain) without and with co-transformation of AKIN10. Bars represent the mean normalized GUS activity ± SD of 3–4 biological replicates. p-values from T-tests < 0.05 and < 0.01 are indicated by * and **, respectively. (B) Interaction of AD-bZIP63 (left) or AD-bZIP11 (right) with wt and S/A mutants of BD-bZIP63 without and with co-transformation of AKIN10. Values are given in % of the signal with wt bZIP63 and AKIN10. Letters indicate significant differences as determined by ANOVA and pairwise T-testing (p < 0.05). For western blots for (A) and (B) see Figure 8—figure supplement 1. (C) In vitro kinase assay of bZIP63, 1, 2, 11, 44, and 53 with AKIN10 and the SnRK1 upstream kinase SnAK2. For a kinase assay with the bZIPs and SnAK2 alone see Figure 8—figure supplement 2. (D). Phos-tag gel western blots (αGFP) showing the in vivo phosphorylation state of S29 in bZIP63 after 6 hr light (L) or extended night (EN). 5 week-old soil-grown plants of two genomic bzip63 complementation lines harboring different S/A mutations were used. In line GAY14 (left) none of the AKIN10 target sites (S29/294/300) can be phosphorylated, while in the S294/300A (right) line S29 can still be phosphorylated. Numbered arrowheads on the right mark the position of each observed bZIP63 band for easy reference with other figures (see Figure 3—figure supplement 1 for a comparative image of all Phos-tag western blots). The likely phosphorylation state of the bands in the western blot is shown on the right. X stands for one of the non-mutated serines (59, 102, 160, or 261). CBB, Coomassie brilliant blue.

https://doi.org/10.7554/eLife.05828.037

To exclude the possibility that the observed effects on dimerization are due to phosphorylation of the hetero-dimerization partner rather than bZIP63 itself, we tested whether AKIN10 is able to phosphorylate any of the S1 group bZIPs. To increase the phosphorylation efficiency of AKIN10 we included SnAK2 in the reactions. In contrast to bZIP63, none of the S1 group bZIPs were phosphorylated by AKIN10 (Figure 8C; Figure 8—figure supplement 2). Together, these data indicate that AKIN10-mediated phosphorylation of bZIP63—especially at S29—strongly enhances its ability to form homo- as well as hetero-dimers.

To further substantiate the relevance of S29 phosphorylation in vivo, we compared the phosphorylation patterns of the S29/S294/S300A mutant with a S294/S300A mutant in the light and after extended night treatments in the genomic complementation lines (Figure 8D). Compared to the triple mutant, that did not show any change in the phosphorylation pattern in response to starvation, bZIP63 phosphorylation was clearly increased in the S294/S300A double mutant in the extended night. As S29 is the only AKIN10 target site left in this version, it must indeed be S29 that is phosphorylated by AKIN10 under starvation conditions.

As we saw that AKIN10-dependent phosphorylation promotes dimerization of bZIP63 with all tested partners we wanted to analyze whether phosphorylation has an effect on the dimerization partner preference. Because protoplast two-hybrid assays only allow to analyze the effect of AKIN10 on a single dimer we set up a multicolor BiFC approach (Waadt et al., 2008) in Nicotiana tabacum leaves with bZIP1 and 11 as alternative interaction partners for bZIP63. For this, we expressed all three bZIPs (1, 11, and 63) from the same plasmid and co-transformed AKIN10 (Figure 9A). bZIP11 was fused to CFPN, bZIP1 to VENUSN, and bZIP63 to CFPC. Accordingly, the bZIP63-11 hetero-dimer generates a blue (cyan) signal and the bZIP63-1 hetero-dimer a yellow signal, whereas the bZIP63 homo-dimer cannot be detected. To compare the dimerization preference of phosphorylated and non-phosphorylated bZIP63 we used wt and a S29/294/300A mutant of bZIP63. Quantification of the ratio of the VENUS/CFP signal in more than 100 nuclei clearly showed a shift towards the VENUS emission when wt bZIP63 was compared to the S/A mutant. This demonstrates that phosphorylation by AKIN10 triggers the preferential formation of bZIP63-1 over bZIP63-11 dimers in a competitive in vivo assay.

Figure 9
Download asset Open asset
Phosphorylation of bZIP63 shifts its dimerization preferences.

(A) Multi-color bimolecular fluorescence complementation (BiFC) in transiently transformed N. tabacum leaves to test the effect of bZIP63 phosphorylation on its dimerization preference. A cassette containing bZIP63 (wt or S29/294/300A), bZIP11, and bZIP1—tagged with the C-terminal moiety of CFP and the N-terminal moieties of CFP and VENUS, respectively—was co-transformed with mCherry-tagged AKIN10. CFP (bZIP63-11) and VENUS (bZIP63-1) fluorescence was detected on a confocal laser scanning microscope and quantified for nuclei showing co-expression of AKIN10. Top: scheme of the multi-color BiFC construct and principle. Bottom center: representative microscopy pictures. Size bar = 50 µm. Bottom right: box-and-whiskers plot of the VENUS/CFP ratio of 115–118 nuclei, normalized to the median of the S29/294/300A bZIP63 construct. T-test p-value was < 0.001, as indicated by ***. (B) Model of the regulation of bZIP63 dimerization and activity by AKIN10. Energy deprivation triggers activation of AKIN10, which phosphorylates S29 on bZIP63. This leads to increased formation of specific bZIP63 dimers and altered expression of dimer-specific genes.

https://doi.org/10.7554/eLife.05828.040

Discussion

bZIP63 is an important metabolic regulator in the starvation response

Here we show that bZIP63 plays an important role in the energy starvation response and metabolic regulation. This is in accordance with the sugar/energy-dependent expression of bZIP63 (Matiolli et al., 2011; Kunz et al., 2014), as well as with the fact that several of its proposed target genes are involved in the low-energy response and metabolism (Matiolli et al., 2011; Veerabagu et al., 2014). Furthermore, three members of the S1-group of plant bZIPs—hetero-dimerization partners of bZIP63 (Ehlert et al., 2006; Kang et al., 2010)—have also been linked to energy starvation response and metabolism. Inducible bZIP11 ox lines exhibit a severe dwarf phenotype (Hanson et al., 2008) and a metabolic profile resembling that of carbon starved plants (Ma et al., 2011), while overexpression of bZIP1 and bZIP53 results in enhanced dark-induced senescence and reduced levels of proline and branched-chain amino acids (Dietrich et al., 2011).

Similar to the bZIP1 ox, bZIP63 ox plants showed increased chlorosis after 9 days of darkness, while the bZIP63 ko displayed a clear stay-green phenotype under these conditions. In contrast, neither the single, nor the double ko of bZIP1 and 53 showed reduced dark-induced senescence (Dietrich et al., 2011). This suggests that other bZIPs can take over the function of bZIP1 in starvation-induced leaf yellowing, while bZIP63 plays a more unique role.

Looking at primary metabolism, we found that misregulation of bZIP63 expression has a strong effect, especially on amino acids, which was further enhanced under starvation conditions. In line with the finding that bZIP63 is a positive regulator of ProDH and ASN1 (Matiolli et al., 2011; Veerabagu et al., 2014, and this work: Figures 6C, 7G) and the changes in proline levels in bZIP63 mutants reported by Veerabagu et al. (2014), we measured the strongest differences in proline and the entire glutamate family as well as in aspartate and asparagine levels in the bZIP63 ko and ox. Notably, there is a strong correlation between the amino acid profiles of the bZIP63 and the bZIP1 ox lines (Dietrich et al., 2011). A similar accumulation of amino acids during dark-induced senescence has frequently been observed and was attributed to enhanced protein degradation (reviewed in Araújo et al., 2011). Particularly under low carbon conditions or during senescence, alternative energy sources need to be used in plant cells, and the important role of proline and branched-chain amino acids in this process has been highlighted in a number of studies (reviewed in: Szabados and Savouré, 2010; Szal and Podgórska, 2012). Thus also the altered amino acid levels could contribute to the observed dark-induced senescence phenotype of the bZIP63 mutants.

bZIP63 function is regulated by SnRK1-dependent phosphorylation

We found that bZIP63 is highly phosphorylated in Arabidopsis. By applying different proteolytic digests we were able to identify seven in vivo phosphorylated serine residues, distributed all over the protein. While exogenous sucrose decreased the global phosphorylation level of bZIP63, extended night treatment further increased its phosphorylation, thus supporting the idea that bZIP63 plays a role in energy signaling. Moreover, we found that the SnRK1 kinase AKIN10, which was proposed to be a central regulator of transcription in starvation response (Baena-González and Sheen, 2008), is the major kinase responsible for the starvation-induced hyper-phosphorylation of bZIP63. Therefore, bZIP63 presents the first TF acting as direct target of SnRK1 in the transcriptional energy deprivation response. In vitro, AKIN10 phosphorylated three highly conserved and functionally important residues in the N- and C-terminus of bZIP63—S29, S294, and S300, respectively. Reporter activation assays in protoplasts revealed that these sites—especially S29—are essential for AKIN10-dependent induction of ASN1 and ProDH by bZIP63. Remarkably, the phosphorylation patterns of different bZIP63 mutants on Phos-tag gels pointed towards S29 as the main target site for AKIN10 under extended night conditions. The importance of bZIP63 phosphorylation at the SnRK1 target sites for its in planta function was further underlined by complementation of the bzip63 mutant with genomic bZIP63 constructs. Wt, but not a S29/S294/S300A mutant of bZIP63, could complement the metabolic and senescence phenotypes. Likewise, also the strong delay in extended night-triggered induction of ASN1, DIN10, and ProDH in bzip63 plants was complemented with wt bZIP63, but not the (triple) S/A mutant. The relatively small difference in ProDH expression in plants, as compared to the protoplast assay, is probably due to redundancy within the C/S1 group bZIPs, as it was shown by Dietrich et al. (2011), that only ko of multiple members of the C/S1 group leads to a strong reduction in ProDH and ASN1 expression.

Since the discovery that AKIN10 can activate several members of the C/S1 network (Baena-González et al., 2007), it has been speculated (Baena-González and Sheen, 2008; Usadel et al., 2008), but never shown experimentally, that they are downstream targets of SnRK1. Our data provide compelling evidence that bZIP63, but none of the S1 group bZIPs, is a bona fide in vivo target of SnRK1 in low-energy signaling.

Phosphorylation of bZIP63 alters its dimerization preferences

Differential dimerization is a well-known mechanism for changing the target recognition site, and thereby the target genes of bZIP TFs (Tsukada et al., 2011). For the S1-group bZIP1 is has been shown that dimerization with C-group bZIP10 or bZIP63 affects its in vitro binding to ACGT-based motifs differently (Kang et al., 2010). The notion that different C/S1 dimers have different target genes is further supported by a recent transcriptomics study in protoplasts. Overexpression of four C/S1 group bZIPs (bZIP1, bZIP10, bZIP11, and bZIP63), individually or in combination of two, revealed overlapping but distinct gene expression patterns (Ma, 2012). This means, that although they regulate a core set of common genes, such as ASN1 and ProDH (Baena-González et al., 2007; Dietrich et al., 2011; Ma et al., 2011), different dimers also have distinct functions, and switching of dimerization partners can have a considerable impact on gene expression.

Our data indicate that dimerization and activity of bZIP63 strongly depend on its phosphorylation status. In general, the dimerization of bZIP63 with bZIP1, bZIP11, and bZIP63, was boosted by AKIN10-mediated phosphorylation. Like in the promoter activation assays, the main influence on dimerization came from S29 phosphorylation, while phosphorylation of S294 and S300 showed a mild effect at most. Incidentally, this enhanced dimerization with bZIP63 would explain why AKIN10 was found to activate the S1 group bZIPs (bZIP1, 2, 11, and 53) in protoplast assays (Baena-González et al., 2007), although they do not present direct targets of AKIN10.

At first sight, the phosphorylation triggered boost of dimerization seemed to be strongest for hetero-dimerization with bZIP11. However, it has to be considered that bZIP11 is a very strong activator of transcription as it harbors an activation domain (Ehlert et al., 2006) and was recently also shown to recruit the histone acetylation machinery (Weiste and Droege-Laser, 2014). The same is not the case for bZIP1 and bZIP63. As the readout of the protoplast two-hybrid interaction assay is transcription-based it is therefore not possible to quantitatively compare the effect of phosphorylation on the formation of different dimers and to draw conclusions about the in vivo dimerization preference. Therefore we used a multi-color BiFC assay with bZIP1 and bZIP11 as alternative partners for bZIP63 to test the effect of bZIP63 phosphorylation by AKIN10 on its dimerization preference. The three bZIPs were co-expressed with AKIN10 in N. tabacum leaves and the formation of bZIP63-1 and bZIP63-11 dimers was quantified. Comparison of wt and S/A mutated bZIP63 revealed that, phosphorylation of bZIP63 shifts the dimerization preference from bZIP11 towards bZIP1. This trend fits well to the observation that both TFs are strongly upregulated in the night and extended night, while bZIP11 is downregulated under these conditions (see Figure 5—figure supplement 3). Furthermore, increased expression of bZIP63 and bZIP1 under conditions when bZIP63 is phosphorylated might further enhance the observed shift in the dimerization towards bZIP63-1 dimers.

Based on the data presented in this study, we propose a simplified model for the regulation of bZIP63 dimerization by AKIN10 (Figure 9B). When bZIP63 is not phosphorylated, its capacity to dimerize with other bZIPs is rather low. Under starvation conditions, bZIP63 is phosphorylated at S29 by AKIN10, favoring the formation of bZIP63 homo- and specific hetero-dimers, particularly of bZIP63-1 dimers. This ultimately results in the induction of a different set of target genes and thereby mediates the transcriptional reprogramming of metabolism. However, it is clear that in the plant additional factors like bZIP expression, stability and interaction with other components add more complexity to the situation.

Surprisingly, to date only a small number of papers have reported an influence of phosphorylation on bZIP dimerization (Kim et al., 2007; Guo et al., 2010; Lee et al., 2010). Moreover, to our knowledge this is the first time that phosphorylation outside the bZIP domain was shown to affect dimerization with different partners in a distinct way. While there are numerous reports on phosphorylation-mediated changes in bZIP activity in animals, plants, and yeast, in many cases the underlying mechanism is still unknown. We therefore believe that this novel mechanism of phosphorylation-triggered switch of bZIP dimerization partners could play a substantial role in the regulation of bZIP TF activity in all higher organisms, and should be further addressed in future studies.

Materials and methods

Plant lines

Request a detailed protocol

The lines ox#2 and ox#3 are bZIP63 ox lines in the Col-0 background, expressing bZIP63.2 with a C-terminal GFP tag under the control of the 35S promoter. Generation of these plant lines was previously described in Veerabgu et al. (2014). Overexpression was confirmed by RT-qPCR of bZIP63 mRNA (Figure 1—figure supplement 1E).

The bZIP63 ko line (bzip63) in the Ws-2 ecotype is a T-DNA insertion line. Pool number CSJ1 (NASC ID: N700001) from the Arabidopsis Knockout Facility (AKF) (Sussman et al., 2000) was screened for a T-DNA insertion in bZIP63 and homozygous plants were selected using Kanamycin. Sequencing of the flanking regions revealed that the T-DNA is inserted in the first exon at position 76 (Figure 1—figure supplement 1A). The ko was confirmed by PCR and RT-qPCR of bZIP63 (Figure 1—figure supplement 1B–D). The same line was used by Veerabgu et al. (2014).

For the complementation lines (GY9 + GY11 = wt, GAY4 + GAY14 = S29/294/300A, S294/300A), a genomic fragment containing bZIP63 and 2 kb of upstream sequence was obtained by PCR on Col-0 genomic DNA and ligated into pCRBlunt (Invitrogen, Austria). For the S/A constructs, first S294 and S300 were mutated to alanine by mutagenesis PCR and the plasmid was subsequently used to mutate S29 to alanine (see Table 1 for primers). Correct sequences were confirmed by restriction digests and sequencing. The genomic fragments were then cloned KpnI/NotI into modified pBIN19, containing a BASTA resistance for plant selection and a C-terminal YFP tag, before transformation into the Agrobacterium tumefaciens strain GV3101 (pMP90). Homozygous bzip63 plants were transformed with the floral dip method and selected for positive transformation events by spraying seedlings with 200 mg/l BASTA solution (Bayer, Germany). Transgene expression in GY and GAY lines was tested using RT-qPCR, western blots with an antibody against GFP, and epifluorescence microscopy (Figure 7—figure supplement 1A–C). The S294/300A shown in Figure 8D was at the time of submission still heterozygous and in the T2 generation. Therefore, expression of the transgene was checked by epifluorescence microscopy (not shown), but no quantification by RT-qPCR or western blotting was done.

Table 1

List of primers

https://doi.org/10.7554/eLife.05828.041
GeneForward primerReverse primer
Primers for cloning
genomic bZIP63GGTACCAAAACTATAAATTTCTTGTAGGACAGTGTTGCGGCCGCCCTGATCCCCAACGCTTCGAATACG
bZIP63.2GGGCCCATGGAAAAAGTTTTCTCCGACGAAGAAATCTCCTTGCGGCCGCCCTGATCCCCAACGCTTCGAATACG
AKIN10.1+3GGGCCCATGGATGGATCAGGCACAGGCAGTAGCGGCCGCAGAGGACTCGGAGCTGAGCAA
AKIN11.1+2GGGCCCATGGATCATTCATCAAATAGGCGGCCGCAGATCACACGAAGCTCTGTAA
SNF4GGGCCCATGTTTGGTTCTACATTGGAGCGGCCGCAAAGACCGAGCAGGAATTGGAA
AKINβ1.1GGGCCCATGGGAAATGCGAACGGCAAAGCGGCCGCACCGTGTGAGCGGTTTGTAG
AKINβ2.1+2GGGCCCATGTCTGCTGCTTCTGATGGTGCGGCCGCACCTCTGCAGGGATTTGTAG
bZIP1CGATGGGCCCCATGGCAAACGCAGAGAAGCGATGCGGCCGCTGTCTTAAAGGACG
bZIP2AAAACCATGGCGTCATCTAGCAGCACTGCGGCCGCTATACATATTGATATCATTAG
bZIP11TAGGGCCCATGGAATCGTCGTCGTCGGGAATGGCGGCCGCAATACATTAAAGCATCAGAAG
bZIP44.1CAGGGCCCATGAATAATAAAACTGCTGCGGCCGCAACAGTTGAAAACATC
bZIP53AAAACCATGGGGTCGTTGCAAATGCAAACTGCGGCCGCTGCAATCAAACATATCAGCAG
AKIN10 VIGSGGAATTCACTTTTTCAGCTCAGAAAATTTTGGGGGTACCCCCTCGAGCCACTCCTGATATTATCTGCTG
AKIN11 VIGSCCTCGAGGTTTCTGTATATTTCTTCGCTCGGGGTACCCAGTACTCTACACCAGATATTATC
BiFC MCS1TCTAGAGGGCCCATGGGCCACAGACACAGCCCGGGGCGGCCGCACATTCTGCGTC
BiFC MCS2AAAAGGATCCGGGCCCATGGGCCACAGACACAGCAAGTCCAAATCCTCCGCCCGGGGCGGCCGCACATTCTGCGTC
multi-color BiFC bZIP1TCGGTACCAAGCTTAGCTTGCATGCCTGCAGGCAGGATCCAGCTGGCGAAAGGGGGATGTGCTG
multi-color BiFC bZIP11CAGGATCCTTAGCTTGCATGCCTGCAGGTCCCTAGGCCTAAGCTTCGCTATTACGCCAGCTGGCGAAAGG
Mutagenesis
bZIP63S29ACGTCGTTGAATCGCTCGGCCGCCGAATGGGCATTCAATCACGTCATTCCATTAACCGACCAGTG
bZIP63S59AGTGTGGTGTTTCCGTCTCCGCTCCTCCTAATGTTCCTGCACACGCCGTCGTAGATTCTCCGTCGTC
bZIP63S102AGATACTTCTGGTAGAGCTGACAATGGTGGAGCGTATCCTGAGGTTTGATGAAAGTTC
bZIP63S160ACTGATTCTCTATTAGCTAGCATCCTTTTAACGATCAGCTAGACGGTCCAGAAGAAG
bZIP63S261ACAGAGACATCAAATGCTCCAGACACTACAAGCTTGTAGTGTCTGGAGCATTTGATGTCTCTG
bZIP63S294AGAACAGAACAGCTGCCATGCGTAGAGTTGAGTGTTCATCTTGCACCCTATCAAGGC
bZIP63S294/300AGAACAGAACAGCTGCCATGCGTAGAGTTGAGGCCTTGGAACATCTGCAGTGTTCATCTTGCACCCTATCAAGGC
AKIN10K48MGGTTGCTATCATGATCCTCAATCGTCGGCAACCTTATGTCCTGTCAATGC
qPCR
bZIP63GAAGAAATCTCCGGTAACCATCACGATTCTCCGTCGTCTGCAGC
bZIP1AACGCGGGTCTTAGATCGGAGAAGTCAGCGTTAAACTCGTCGTAGCAA
bZIP11TCGTCAGGATCGGAGGAGAGTGATCGTCTAGGAGCTTTTGTTTCTTC
CAB2TCAATCTTTTGAATTCGAGTGAGATCCACCACAAACACAAACCTAC
SEN1CAGAGTCGGATCAGGAATGGATTATGATTTCATCGTGTTTCC
SAG103AGCTCGAGTGCTGGGATGCGGATTCACAGATCCTTCCT
YLS3GACATCACTAAGTGCCCTGCTACTGTTTCGTTCAGACCTTTAGC
AKIN10ACTGGATTTGCAGAGAGTACAAGGTCCTCAGAGGACTCGGAGCTGAGCA
AKIN11GCTCGTAACTTTTTCCAGCAGATTCAGGTCTCTATGGACAACCA
ASN1GTCGCAAGATCAAGGCTCTGAAGTCTTTGTCAAGGAAAGG
DIN10GCTTGTATTGCCTGATGGAATCTTTAGCAAGCTGACACC
ProDHCGCCAGTCCACGACACAATTCACGAATCAGCGTTATGTGTTGCG
BCAT2AAGGTTATCAGGTAGTAGAGAAGGTTCCTGATATGTGATAGTGCC
AA-TP family proteinGTTCTTGGGATCAACTTCTACAGAACCATTTACATTCCGTAGGAC
TBP2TGCAAGAAAGTATGCTCGGACATGAGCCTACAATGTTCTG
MHF15GTTTCCTGAGCTTCTCCACTGGTCGCTTCATCTTGAG
Characterization of plant lines (PCR)
bZIP63TTGCGGCCGCCCTGATCCCCAACGCTTCGAATACGAACCATGGATAATCACACAGCTAAAGACATTGG
bzip63 RBTGGCGAATGAGACCTCAATTGCGAGCTTT
bzip63 LBCATTTTATAATAACGCTGCGGACATCTAC
AKIN10CCAGCATAATAGAGAACGAAGCGTCCGGTTTAGTATTCAGAGG
akin10 LBATATTGACCATCATACTCATTGC

The AKIN10 ko line (akin10, GABI_579E09) in the Col-0 ecotype is a T-DNA insertion line from the GABI KAT collection (Kleinboelting et al., 2012). For this line, two T-DNA insertions were suggested, one in AKIN10 and another one in IMS2 (2-ISOPROPYLMALATE SYNTHASE 2, At5g23020). The insertion in AKIN10 in front of the last exon was confirmed by PCR, but the second insertion in IMS2 could not be detected and was thus assumed not to be present (Figure 5—figure supplement 1A–C). The ko of AKIN10 was further confirmed by RT-qPCR and western blotting, respectively. No significant amount of transcript or protein could be detected (Figure 5—figure supplement 1D,E). The akin10 line was further tested for obvious growth and developmental phenotypes by comparing leaf number, fresh and dry weight, water content, and flowering time in soil-grown wt and mutant. Only a slight delay in flowering was observed (Figure 5—figure supplement 2A,B). Expression of selected AKIN10 target genes after 6 hr of light or extended night was tested by RT-qPCR. Some genes showed a slight, but not dramatic reduction in expression (Figure 5—figure supplement 2C).

For the line expressing bZIP63-GFP in akin10, bZIP63.2 was amplified from Col-0 cDNA (see Table 1 for primers) and cloned ApaI/NotI into modified pBIN19 containing the UBI10 promoter, a BASTA resistance for plant selection, and a C-terminal YFP tag. Homozygous akin10 plants were transformed and selected like the GY and GAY lines.

Plant growth

Request a detailed protocol

Arabidopsis seeds were surface sterilized with chlorine gas before sowing on soil or growth medium and then vernalized at 4°C for 2 days. Plants were grown in a growth chamber in a 12 hr light/12 hr dark regime with day temperatures between 20 and 22°C and night temperatures between 18 and 20°C and a light intensity of 60–150 μmol/m2s unless specified otherwise. Plants grown under short or long day conditions were cultivated with 8 hr or 16 hr of light per day, respectively. The soil mixture consisted of 4 parts Huminsubstrat N3 (Neuhaus, Germany), 1 part perlite (Gramoflor, Germany), and the fertilizer Osmocote (Substral/Scotts, Germany) according to manufacturer's instructions. For hydroponic cultures, plants were grown with their roots in light-tight box filled with liquid ½ Hoagland medium (2 mM Ca(NO3)2, 0.25 mM K3PO4, 3 mM KNO3, 1 mM MgSO2, 45 μM NaFeIII EDTA, 5 μM H3BO3, 1 μM MnCl2, 0.15 μM ZnSO4, 0.1 μM CuSO4, 7 nM MoO3, 4.5 nM Co(NO3)). Seedling cultures in liquid medium were grown in ½ Gamborg medium (Duchefa, Harlem, The Netherlands) with or without 0.5% sucrose and the medium was exchanged after 7 days. Seedlings on plates were grown on ½ MS (Duchefa, Harlem, The Netherlands) with 0.7 g/l plant agar (Duchefa, Harlem, The Netherlands) and a pH of 5.8. The root cell suspension culture was grown in MS medium containing 30 g/l sucrose and 2.5 µM 2,4D at 22°C in the dark under constant shaking. The medium was exchanged every 7 days by transferring ⅓ to ½ of the culture to a fresh flask and addition of fresh medium.

Dark-induced senescence

Request a detailed protocol

4.5 week-old soil-grown plants were incubated in the dark—in a box with tubes allowing for gas exchange—for 9 days. Before and after incubation the true leaves of 4–8 representative plants were harvested, stuck on white paper with double sided tape, scanned with a flatbed scanner without color correction at a resolution of 600 dpi, and saved as TIFF files. Images were then imported in ImageJ (FIJI) (Schindelin et al., 2012) and the total and green leaf area of each leaf were quantified using the built-in threshold and color threshold function, respectively. The green leaf area in % was then calculated by dividing the green area of a leaf or the whole plant by the respective total area.

See Figure 1—source data 1 for the ImageJ macro for semi-automatic image processing. A scheme of this method can be found in Figure 1—figure supplement 2A. Alternatively, chlorophyll content was determined according to Porra et al. (1989). Total rosettes or seedlings were weighed and frozen in liquid nitrogen. Rosettes were crushed crudely with a spatula to increase extraction efficiency. Chlorophyll was extracted by adding 1–5 ml of a mixture of 80% Acetone and 20% 12.5 mM Hepes KOH pH 8.2 and incubation in the dark for at least 12 hr. Absorbance at 646.6 nm, 663.6 nm, and 750 nm was measured in a quartz cuvette and the chlorophyll content in µg chlorophyll/mg fresh weight was determined as follows: (12.5*(A663.6 − A750) − 2.55*(A646.6 − A750) + 20.31*(A646.6 − A750) − 4.91 (A663.5 − A750))/FW. For sugar rescue assays seeds were germinated on ½ MS agar plates containing 0.5% sucrose and grown in a 12 hr light/12 hr dark cycle for 12 days before transfer to ½ MS agar plates containing 0% or 2% glucose. After 6 additional days seedlings were either harvested or put in a dark box for 7 days. The chlorophyll content in µg/mg FW of individual seedlings was determined as described above.

Metabolic profiling

Request a detailed protocol

Metabolites were extracted from leaves of 5 week-old plants, derivatized and measured as described in Naegele et al. (2014) with minor variations. Approximately 80 mg frozen and ground plant material were extracted with 1 ml –20°C cold MeOH:chloroform:H2O (2.5:1:0.5) by mixing and incubation on ice. The supernatant after centrifugation was mixed with 400–500 µl H2O and centrifuged. For the experiment shown in Figure 2 (2) the polar phase was split into 2 aliquots, spiked with 1 µg C13 labeled Sorbitol (Campro Scientific, Berlin, Germany) and dried. For the experiment shown in Figure 7 (7) the polar phase was not split and 2 µg of Sorbitol were added. For derivatization, metabolites were first dissolved in 10 µl (2) or 30 µl (7) pyridine containing 40 mg/ml methoxyamine hydrochloride by 90 min incubation at 30°C. Then, 40 µl (2) or 120 µl (7) of N-methyl-N-trimethylsilyltrifluoroacetamid (MSTFA; Macherey–Nagel, Düren, Germany), spiked with 60 µl/ml of an Alkane Standard Mixture C10–C40 (Fluka, Vienna, Austria), were added and the samples were incubated for 30 min at 37°C. GC–MS measurements were carried out on an Agilent 6890 gas chromatograph coupled to a LECO Pegasus 4D GCxGC-TOF mass spectrometer (LECO, USA). Injection volume was 1 µl. In the GC step, the initial oven temperature was 70°C, which was held for 1 min, followed by a 9°C/min temperature increase until the final temperature of 350°C was reached, which was held again for 8 min. Metabolites were measured in splitless mode (2 and 7) and alternatively also in split mode with a split ratio of 5 (7). In the MS step the data acquisition rate was set to 20 spectra/s, the detector voltage to 1550 V and the mass range to 40–600 m/z. Raw data were processed with the LECO Chroma-TOF software (LECO, USA). Peak areas were normalized to the area of the internal standard and to the fresh weight before statistical analysis. Outliers, as determined by Grubb's test, were removed from the dataset. For Hierarchical clustering, log-2 transformed fold change values were imported into MeV (MultiExperimentViewer, version 4.9.0, Saeed et al., 2003) and clustering was done using the standard settings with gene tree optimization, Pearson correlation, and average linkage clustering. For the PCA plot, normalized data were imported into R (RStudio, version 0.98.507), missing values were replaced with the k nearest neighbor (knn) method using the ‘impute.knn()’ function and data were Z-transformed. PCA analysis was done using the ‘prcomp()’ function and scores for PC1 and PC2 were plotted against each other.

Electrophoresis and western blotting

Request a detailed protocol

For 2D gels, proteins were extracted from 5–7 week-old soil-grown ox#3 plants which were grown in short day. 4 ml of 1× lambda phosphatase (λPP) buffer (NEB, Frankfurt am Main, Germany), including cOmplete protease inhibitor (Roche, Vienna, Austria), were added to 2 ml frozen and ground plant material, followed by vortexing and centrifugation. 0, 30 or 50 µg of λPP were added to 1.5 ml of the supernatant, followed by 15 min incubation at 30°C. Proteins were extracted with phenol (Carl Roth, Karlsruhe, Germany), precipitated with ammonium acetate and resuspended in 1× rehydration stock solution (7 M urea, 2 M thiourea, 2% (wt/vol) CHAPS, 2% IPG buffer (GE Healthcare, Vienna, Austria), 2.8 mg/ml DTT, Bromphenol blue). Protein extracts were then applied to 7 cm Immobiline DryStrips (GE Healthcare, Vienna, Austria) over night and separated by IEF on an IPGphor (GE Healthcare, Vienna, Austria) according to the manufacturer's instructions. For second dimension separation, the strip was incubated in SDS equilibration buffer (6 M urea, 75 mM TrisCl, 29.3% glycerol, 2% SDS, Bromphenol blue, pH 8.8) with 10 mg/ml DTT and then without DTT for 15 min each, followed by standard SDS PAGE and western blotting.

Phos-tag gel electrophoresis was done according to the manufacturer's instructions. Proteins were separated on SDS PAGE gels containing 8% SDS, 25 µM Phos-tag (WAKO, Neuss, Germany) and 50 µM MnCl2 with an amperage of 15 mA/gel for 1.25 hr. Before semi dry blotting, the gels were incubated in transfer buffer containing 1 mM EDTA, followed by washing with transfer buffer without EDTA. Recombinantly expressed bZIP63-YFP was used as a size marker for the nonphosphorylated fusion protein. For the expression, bZIP63 was amplified from cDNA (see Table 1 for primers) and cloned NcoI/NotI into pTWIN (NEB, Frankfurt am Main, Germany) containing YFP, to create a c-terminal YFP fusion. The protein was expressed in Escherichia coli (ER2566 strain) and purified according to the manufacturer's instructions. Total plant proteins were extracted with phenol. For light/extended night comparison, rosettes of 5 week old plants were collected after 6 hr of light or extended night. For ± sucrose comparison, seedlings were first germinated and grown for 1 week in liquid ½ Gamborg medium containing 0.5% sucrose, followed by 1 week in medium without sucrose. For treatment, at the onset of light 1% suc was added to half of the cultures and all cultures were kept in the dark for 6 additional hours. For Phos-tag gels from kinase assays, the reactions were mixed with 2× Laemmli sample buffer and loaded on the gel.

Western blotting was done by semi dry transfer onto a PVDF membrane, antibody incubation, and detection with an ECL kit following standard procedures. The following primary antibodies were used: Anti-GFP (Roche, Vienna, Austria; or ChromoTek, Munich, Germany), Anti-AKIN10 (Agrisera, Sweden), Anti-AMPKalpha-pT172 (Cell Signaling, Leiden, The Netherlands), Anti-Flag (Sigma–Aldrich, Vienna, Austria), Anti-HA High Affinity (Roche, Vienna, Austria), Anti-HA (Santacruz, Heidelberg, Germany), Anti-GAL4 DNA BD (Sigma–Aldrich, Vienna, Austria), polyclonal peptide antibodies against bZIP63: peptide antibodies against an N-terminal (EKVFSDEEISGNHHWSVNGM) and a C-terminal (SLEHLQKRIRSVGDQ) peptide were raised in rabbit and affinity purified (Davis biotechnology, Regensburg, Germany). The bZIP63 antibodies were tested on protein extracts from wt, bZIP63 ko and ox plants. While the antibodies recognized recombinant protein and bZIP63 in the ox extracts well, they failed to detect endogenous levels of bZIP63, possibly due to the low abundance of the protein (not shown). The following HRP-coupled secondary antibodies were used: Anti-mouse IgG, Anti-rabbit IgG, Anti-rat IgG (GE Healthcare, Vienna, Austria).

Immunoprecipitation of bZIP63-GFP

Request a detailed protocol

To identify in vivo phosphorylation sites on bZIP63, bZIP63-GFP was immunoprecipitated from ox#3 seedlings grown on ½ MS agar plates or leaves of mature soil-grown ox#3 plants, harvested at different time points in the light cycle and in extended night (see Figure 3—figure supplement 2A for growth and harvesting conditions). In one experiment, leaves were infiltrated with H2O containing 100 µM of the proteasome inhibitor MG-132 (Calbiochem/Merck Millipore, Vienna, Austria) 6 hr before harvesting. Protein extracts were prepared by mixing frozen and ground plant material with an equal volume of cold extraction buffer (25 mM TrisCl, 10 mM MgCl2, 15 mM EDTA, 150 mM NaCl, 1 mM DTT, 1 mM NaF, 0.5 mM Na3VO4, 15 mM β-glycerophosphate, 0.1% Tween20, Complete protease inhibitor, pH 7.5), followed by centrifugation. The supernatant was then incubated with protein A Sepharose CL-4B (GE Healthcare, Vienna, Austria), which had been pre-incubated for 2.5 hr in extraction buffer with Anti-GFP antibody (Roche, Vienna, Austria), or with GFP-Trap_A beads (ChromoTek, Munich, Germany) for 1 hr at 4°C. The beads were washed 2–3 times with extraction buffer and alternatively twice with wash buffer (50 mM TrisCl, 250 mM NaCl, 0.1% NP–40, 0.05% sodium deoxycholate, pH 7.5). Finally, the beads were resuspended in 1× Laemmli sample buffer, boiled for 5 min at 95°C, and centrifuged. The supernatant was separated by SDS PAGE and bands were excised for LC-MS/MS analysis.

Identification of proteins and in vivo phosphorylation sites by LC-MS/MS

Request a detailed protocol

For the identification of phosphorylation sites, bZIP63-GFP was immunoprecipitated from leaves of ox#3 plants (see ‘Immunoprecipitation of bZIP63-GFP’). For the identification of kinases, root protein extracts were affinity purified with recombinant GST-bZIP63 (see Figure 4B for a scheme and the methods section ‘Kinase assays’ for detailed description of the affinity purification).

Proteins were first separated by SDS PAGE. Bands of interest were then excised from Coomassie-stained gels and gel sections were chopped, washed with 50 mM ammonium bicarbonate (ABC, pH 8.5), and dried with acetonitrile (ACN). Disulfide bonds were reduced by incubating in 200 µl of 10 mM DTT for 30 min at 56°C. DTT was washed off and cysteines were alkylated by incubation with 100 µl of 54 mM iodoacetamide for 20 min at RT in the dark. Gel pieces were dried with ACN, then swollen in 10 ng/µl trypsin (recombinant, proteomics grade, Roche, Vienna, Austria) in 50 mM ABC and incubated over night at 37°C. For higher sequence coverage of bZIP63 alternative proteases were used: LysC (MS grade, WAKO, Neuss, Germany) at 37°C overnight, subtilisin (Sigma–Aldrich, Vienna, Austria) at 37°C for 0.5–2 hr, chymotrypsin (sequencing grade, Roche, Vienna, Austria) at 25°C for 4 hr. Digestion was stopped by adding formic acid to a final concentration of approximately 1% and peptides were extracted by sonication. Peptides were separated on an UltiMate 3000 HPLC system or on a U3000 nano HPLC (both Dionex, Thermo Fisher Scientific). Digests were loaded on a trapping column (PepMap C18, 5 µm particle size, 300 μm i.d. × 5 mm, Thermo Fisher Scientific), equilibrated with 0.1% trifluoroacetic acid (TFA), and separated on an analytical column (PepMap C18, 3 μm, 75 μm i.d. × 150 mm, Thermo Fisher Scientific) by applying a 60 min linear gradient from 2.5% up to 40% ACN with 0.1% formic acid, followed by a washing step with 80% ACN and 10% trifluoroethanol (TFE) on the U3000 HPLC. The UltiMate 3000 HPLC was directly coupled to a linear ion trap (LTQ, Thermo Fisher Scientific), which was operated in a data-dependent MS3 method for the phosphorylation analysis. One full scan (m/z: 450–1600) was followed by maximal 4 MS/MS scans. If in the MS/MS scan a fragment corresponding to a neutral loss from the precursor of 98, 49, or 32 Th was observed among the top 8 peaks, a MS3 scan was triggered. Fragmentation energy was set at 35%, Q-value at 0.25, and the activation time at 30 ms. High resolution measurements were acquired on an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific), equipped with a nanoelectrospray ionization source (Proxeon, Thermo Fisher Scientific). The electrospray voltage was set to 1500 V. The mass spectrometer was operated in the data-dependent mode: 1 full scan (m/z: 350–1800, resolution 60,000) with lock mass enabled was followed by maximal 12 MS/MS scans. The lock mass was set at the signal of polydimethylcyclosiloxane at m/z 445.120025. Monoisotopic precursor selection was on, precursors with charge state 1 were excluded from fragmentation. The collision energy was set at 35%, Q-value at 0.25, and the activation time at 10 ms. Fragmented ions were set onto an exclusion list for 60 s. When ETD (electron transfer dissociation) was applied, the top 6 peaks from the full scan where fragmented with CID (collision-induced dissociation) and subsequently with ETD. For ETD, the energy parameters were as for the CID except the activation time was set to 80 or 120 ms.

Data interpretation: raw spectra for the kinase identification were interpreted by Mascot 2.2.04 (Matrix Science) using Mascot Daemon 2.2.2. Spectra were searched against the Arabidopsis thaliana entries in the nr-database with the following parameters: the peptide tolerance was set to 2 Da, MS/MS tolerance was set to 0.8 Da, carbamidomethylcysteine was set as a static modification, oxidation of Met as variable modification. Trypsin was selected as the protease allowing two missed cleavages. Mascot score cut-off was set to 30, except for the low abundance sample 1, where the cut-off was set to 20. For the phosphorylation analysis of purified bZIP63, either Mascot or Sequest (Proteome Discoverer 1.2; Thermo Scientific) were used. The search was extended to the phosphorylation of Ser, Thr, and Tyr. High resolution data were searched with 3 ppm precursor mass tolerance. Proteolytic specificity was defined according to the digest. Results were manually validated including comparison of the fragmentation pattern and the relative retention of the nonphosphorylated counterpart. Site localization was checked by manual inspection at the spectrum level in the first place and was confirmed by the site-localization algorithm PhosphoRS (Taus et al., 2011).

Kinase assays

Request a detailed protocol

In vitro kinase assays were performed with GST-tagged recombinant proteins. The cDNA of bZIP63.2, bZIP1, bZIP2, bZIP11, bZIP44.1, bZIP55, AKIN10.1/3, and AKIN11.1/2 was obtained by PCR (see Table 1 for primers) and cloned ApaI/NotI or NcoI/NotI into pGEX-4T. SnAK2 was in pDEST15 (Crozet et al., 2010). An inactive version (K/M) of AKIN10 and non-phosphorylatable (S/A) versions of bZIP63 were created by mutagenesis PCR using the primers listed in Table 1. The proteins were expressed in E. coli (ER2566 or BL21 strain), purified using Glutathione Sepharose 4B (GE Healthcare, Vienna, Austria) according to the manufacturer's instructions, and stored at −80°C in GST elution buffer containing 10–25% glycerol.

Kinase assays were performed by incubating the kinase and substrate for 20–30 min in kinase reaction buffer (20 mM Hepes, 20 mM MgCl2, 100 µM EGTA, 1 mM DTT, 50 µM ATP, pH 7.5) at room temperature. For radioactive assays, 1 µCi γ-32P-labeled ATP (NEN/PerkinElmer, Waltham, MA, USA) was added in each reaction. The reactions were then separated by SDS PAGE and exposure on a Storage Phosphor Screen (GE Healthcare, Vienna, Austria) or Phos-tag gel electrophoresis and Western blotting, respectively.

For in-gel kinase assays, bZIP63 with an N-terminal 6xHis tag was used as a substrate. To construct the expression vector, bZIP63.2 was amplified from cDNA (see Table 1) and first cloned ApaI/NotI into pRSETa-QM (Invitrogen, Germany). From there the expression cassette was excised by SalI digest, followed by a fill in with DNA Polymerase I (Klenow fragment, NEB, Frankfurt am Main, Germany) and XbaI digest. The cassette was ligated with pTXB3 (NEB, Frankfurt am Main, Germany), which was first digested with BamHI, followed by a fill in and XbaI digest. The protein was expressed in E. coli (ER2566 strain) and purified over a HiTrap column (GE Healthcare, Vienna, Austria) according to the manufacturer's instructions. Plant protein extracts for the in-gel kinase assays were made from different plant material. For the identification of bZIP63 kinases, proteins were extracted either from roots of 8 week-old plants that were grown in hydroponic culture in short day and collected in the light phase (shown in Figure 4A,C), from root cell suspension culture, or from seedlings grown on ½ MS agar plates, which were harvested in the dark phase. 4 independent experiments were conducted. For specifications on the plant material used in each of the experiments and the kinases identified please refer to Figure 4—source data 2. For the comparison of bZIP63 phosphorylation with wt and akin10 plant extracts, roots and leaves of 2 week-old seedlings grown in liquid culture in a 12 hr light/12 hr dark cycle and collected after 4 hr of extended night were used. Extraction was done by mixing the frozen and ground plant material with an equal volume of cold protein extraction buffer (25 mM TrisCl, 15 mM EGTA, 10 mM MgCl2, 75 mM NaCl, 1 mM NaF, 0.5 mM NaVO3, 15 mM beta-glycerophosphate, 0.1% Tween20, 1 mM DTT, Complete protease inhibitor, pH 7.5), followed by centrifugation. Protein amounts were determined by Bradford assay. For affinity purification of bZIP63-binding proteins, GST-tagged bZIP63 was expressed in E. coli and the cell lysate of up to 1l culture in GST binding buffer (50 mM TrisCl, 20 mM MgSO4, 2 mM DTT, 5 mM EDTA, 0.5% Tween20, pH 8) was loaded on an equilibrated GSTrap FF column (GE Healthcare, Vienna, Austria). The column was then washed with 5 ml cold GST binding buffer and protein extraction buffer, respectively, and 2–5 ml of total root protein in cold protein extraction buffer were loaded. Subsequently, proteins were eluted from the column by repeated washing with 5 ml cold protein extraction buffer with increasing salt concentrations, and concentrated with Amicon Ultra Centrifugal Filter Units (Millipore). 8–20 µg total protein extracts and up to 40 µg affinity purified proteins were loaded on standard SDS PAGE gels containing 1 mg/ml substrate (6× His-bZIP63) in the separating gel and run at 4°C with 20 mA per gel. The gel was then incubated 3 times for 20 min, respectively, at room temperature in each of the following buffers: wash buffer I (50 mM TrisCl, 20% isopropanol, pH 8), wash buffer II (50 mM TrisCl, 1 mM DTT, pH 8), and denaturation buffer (50 mM TrisCl, 1 mM DTT, 6 M guanidinium HCl, pH 8). Subsequently, the gel was incubated in renaturation buffer (50 mM TrisCl, 0.05% Tween 20, pH 8) at 4°C for 12–18 hr. In this period, the buffer was exchanged 10 times after at least 30 min. After renaturation, the gel was incubated 2 times for 30 min, respectively, in kinase buffer (20 mM HEPES, 20 mM MgCl2, 50 µM CaCl2 or 500 µM EGTA, 1 mM DTT, 0.05% Tween 20, pH 7.5), followed by 30 min incubation in kinase reaction solution (kinase buffer containing 50 µM ATP and 100 µCi γ-32P-labeled ATP). Finally, the gel was washed 2 times for 15 min, respectively, in 5% TCA and several times in 5% TCA containing 1% sodium pyrophosphate, until the wash solution was only weakly radioactive. The dried gels were exposed on a Storage Phosphor Screen and the signal was recorded on a Typhoon 8600 (GE Healthcare, Vienna, Austria).

Y2H assay

Request a detailed protocol

AKIN10.1/3, AKIN11.1/2, AKINβ1, AKINβ2, SNF4, and bZIP63.2 were amplified from cDNA (see Table 1 for primers) and cloned ApaI/NotI or NcoI/NotI into pBTM117 and bZIP63.2 was cloned into pACTIIJ to generate N-terminal fusions with the LexA-BD and the GAL4-AD, respectively. The yeast strain L40 (MATα hisΔ200 trp1-900 leu2-3.112 ade2 LYS2::(lexA op)4HIS3 URA3::(lexA op)8lacZ Gal4 gal80) was transformed with empty pACTIIJ or bZIP63.2-containing pACTIIJ in combination with different pBTM117 vectors. Freshly grown L40 was mixed gently with 1 ml transformation mix (800 µl 50% PEG 3600, 100 µl 2 M LiAc, 100 µl 1 M DTT, 10 µl bacterial RNA [10 µg/µl]), to get a cloudy suspension. 2.5 µg of each plasmid were added to 125 µl transformation mix, followed by 20 min incubation at 30°C and 44°C, respectively, addition of 1 ml H2O, and 1 min centrifugation at 3500×g. The cells were resuspended in a small volume and plated on SD medium without Leu and Trp to select for successful transformation. Single colonies were inoculated in SD medium without Leu and Trp and proteins were extracted from 2 ml of an over-night culture. The yeast cells were resuspended in 200 µl enzyme lysis buffer (25 mM TrisCl, 20 mM NaCl, 8 mM MgCl2, 5 mM DTT, 0.1% NP-40, pH 7.5), 200 µl glass beads (0.4–0.6 mm diameter) were added, the cells were frozen in N2, thawed, and broken by vigorous shaking on a Vibrax at 4°C for 20 min. The supernatant after 10 min centrifugation was transferred to a fresh tube and kept on ice. The protein concentration of the extract was determined by Bradford assay. The GUS activity was determined by mixing 50 µl of the extract with 650 µl Z-buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 10 mM MgSO4, 0.25% beta mercaptoethanol) and 100 µl ONPG (4 mg/ml), and incubating for up to 10 min at room temperature. The reaction was stopped by adding 400 µl 1 M Na2CO3 and the extinction at 420 nm was measured on a photometer. The GUS activity in U/mg protein was calculated as follows: (A420 × 24 × 1000)/(45 × incubation time [min] × protein concentration [mg/ml]).

BiFC and multi-color BiFC

Request a detailed protocol

For BiFC and multi-color BiFC interaction studies in N. tabacum, bZIP63.2, AKIN10.1/3, AKIN11.1/2 and SNF4 were cloned into modified pBIN19 vectors, containing either the N- or C-terminal moieties of the split CFP system as N-terminal fusions as described in Waadt et al. (2008) (see Figure 5D for scheme). To generate BiFC plasmids for ApaI/NotI and NcoI/NotI cloning, CPK3, harboring XbaI, NcoI, and ApaI (for plasmids with MCS [multi cloning site] 1) or ApaI and NcoI (for plasmids with MCS 2) in the terminus and NotI and SmaI in the C-terminus was generated by PCR from cDNA (see Table 1 for primers). It was then cloned XbaI/SmaI or ApaI/SmaI into the different BiFC casettes in pUC19 from Waadt et al. (2008). The cassettes were then cloned into pBIN19 with HindIII/EcoRI and CPK3 was exchanged for other genes using ApaI and NotI. A. tumefaciens (AGL1 strain) was transformed with the resulting plasmids by electroporation and further used for transient transformation of tobacco leaf epidermis. For transformation, 5 ml agrobacterium overnight cultures were filled to 50 ml with fresh LB medium containing 50 µg/ml Kanamycin and 10 µg/ml Gentamicin and grown for 4 hr at 30°C. The cells were pelleted by centrifugation at 3500×g, resuspended in LB containing 150 µM acetosyringone, and grown for another 2 hr. The cultures were then pelleted again and resuspended in 5% sucrose solution to reach a final OD600 of 2. For co-infiltration, equal volumes of agrobacteria suspensions, containing the respective constructs, were mixed and infiltrated into the leaves of 5 week-old plants. 48 hr after infiltration equally sized leaf sections were analyzed for their CFP fluorescence signal with an LSM510 confocal laser scanning microscope (Zeiss) and the corresponding ZEN software (Zeiss). The same settings were used for each construct to allow comparison of the signal intensities. To verify that the fusion proteins were expressed and to determine the relative amount of each interaction partner, proteins were extracted from the leaf sections used for microscopy and subjected to Western blot analysis with antibodies against the Flag (N-terminal CFP moiety) and HA (C-terminal CFP moiety) epitopes.

Multi-color BiFC was done as described in Waadt et al. (2008) with some modifications. bZIP11 fused to the N-terminal moiety of CFP (CFP-N), bZIP1 fused to the N-terminal moiety of VENUS (VENUS-N), and bZIP63 (wt or S29/294/300A) fused to the C-terminal moiety of CFP (CFP-C) were expressed from one plasmid to have co-expression at equal amounts in all transformed cells (for a scheme of the construct see Figure 9A). To generate the construct, bZIP1 and bZIP11 were first amplified from cDNA by PCR (see Table 1 for primers) and cloned ApaI/NotI into pBIN19 containing a c-terminal fusion of CFP-C and CFP-N, respectively (plasmids described above). From there, 35S::bZIP1 was cloned HindIII/NotI into pUC19 containing CPK3 with a c-terminal VENUS-N fusion (described above) to exchange the shorter 35S primer in the VENUS construct. Then, the two cassettes including the 35S promoter, bZIP1 or 11 and the VENUS-N or CFP-N tag were amplified by PCR introducing KpnI/HindII in the front and BamHI in the back of the bZIP1 cassette and BamHI in the front and HindIII in the back of the bZIP11 cassette (see Table 1 for primers). The PCR products were ligated into pCRBlunt (Invitrogen, Germany) and the bZIP1 cassette was then cloned KpnI/BamHI in front of the bZIP11 cassette. Finally, the combined cassette with bZIP1 and 11 was cloned HindII into pBIN19 containing bZIP63 wt or S29/294/300A (mutated by PCR) with an n-terminal CFP-C fusion. AKIN10 was amplified from cDNA by PCR (see Table 1 for primers) and cloned ApaI/NotI into pBIN19 containing a c-terminal mCherry tag. A. tumefaciens (AGL1 strain) was transformed with the resulting plasmids by electroporation and further used for transient transformation of tobacco (N. tabacum) leaf epidermis as described above. 48 hr after infiltration equally sized leaf sections were analyzed for their VENUS, CFP, and mCherry fluorescence signal with a TCS SP5 DM-6000 confocal laser scanning microscope (Leica) and the corresponding Leica software. CFP, VENUS, and mCherry were excited with 458 nm (Ar laser), 514 nm (Ar laser), and 574 nm (white light laser), respectively. The fluorescence was detected at 461–495 nm, 520–550 nm, and 600–615 nm, respectively. To avoid bleed-through of the CFP signal into the VENUS channel CFP and VENUS were excited and detected separately. Pictures were taken with a 20× air objective and the same settings for both combinations (wt bZIP63 and S29/294/300A). The pinhole was set to 1.5. To determine the relative VENUS/CFP signal ratio, images were processed in the LasX software from Leica. Nuclei showing a CFP/VENUS as well as mCherry signal were selected and the signal intensity was determined with the histogram tool. For each nucleus a ratio of the background corrected pixel sum from VENUS and CFP was built. All values for one leaf were divided by the median of the ratio from the bZIP63 S29/294/300A construct to allow comparison between experiments. In total, six leaves from two infiltrations were analyzed with similar results.

Virus-induced gene silencing (VIGS)

Request a detailed protocol

2 week-old GY9 plants were infiltrated (akin10/11) or not (control) with an AKIN10-AKIN11 silencing construct according to the method described by Burch-Smith et al. (2006). For the AKIN10-AKIN11 construct two gene-specific fragments corresponding to AKIN10 and AKIN11 were cloned in tandem into the TRV-based vector pYL156a. First, a 480 bp AKIN10 fragment was amplified from cDNA using the AKIN10 VIGS primers (see Table 1 for primers) and cloned into the pYL156/pTRV2 vector using EcoRI/KpnI. A 503-bp AKIN11 fragment (Baena-González et al., 2007) was thereafter amplified from cDNA using the AKIN11 VIGS primers (see primer list) and cloned into the pYL156/pTRV2 vector harboring the KIN10 fragment using XhoI/KpnI. 2 weeks after infiltration the AKIN10-AKIN11 knock-down plants displayed visible growth defects and anthocyanin accumulation. Silencing of AKIN10 and AKIN11 in these plants was confirmed by western blotting with an antibody against phosphorylated AMPK alpha (Anti-AMPKalpha-pT172) before use in Phos-tag gels (see Figure 5—figure supplement 7). The VIGS experiments were repeated 4 times and 3 biological replicates were analyzed in each experiment with consistent results.

Protoplast transformation for P2H and promoter activation assays

Request a detailed protocol

Protoplast were obtained from 3 week-old soil-grown wt Arabidopsis plants and transformed according to the guide method (Yoo et al., 2007) with small modifications. Leaves were harvested 1 hr after onset of the light phase, cut into tiny stripes, and digested for 30 min under vacuum and 3 hr at atmospheric pressure, respectively, with enzyme solution (1.25% [wt/vol] Cellulase R-10, 0.3% Macerozyme R-10, 400 mM mannitol, 20 mM KCl, 10 mM CaCl2, 20 mM MES, pH 5.7). The protoplast suspension was filtered on a metal net to remove leaf debris and washed twice with 10 ml of W5 solution (2 mM MES, 154 mM NaCl, 125 mM CaCl2, 5 mM KCl, pH 5.7). Protoplasts were resuspended in 10 ml of W5, incubated on ice for at least 1 hr and subsequently resuspended to a final concentration of 1 × 105 cells/ml in MMg buffer (4 mM MES, 400 mM mannitol, 15 mM MgCl2, pH 5.7). Protoplasts were then co-transformed with 10 µg each of up to three effector plasmids, 7 µg of a reporter plasmid, and 3 µg of a 35S::NAN transfection control reporter plasmid (Kirby and Kavanagh, 2002) for normalization. For P2H assays, the effector plasmids were pHBTL containing bZIP63.2, bZIP1, or bZIP11 with an N-terminal GAL4-AD or GAL4-BD fusion (described in detail in Ehlert et al., 2006) and alternatively pHBTL containing AKIN10.1/3. The reporter plasmid contained GAL-UAS4::GUS (beta galactosidase) (Ehlert et al., 2006). For promoter activation assays, effector plasmids were pHBTL containing HA-tagged bZIP63.2 or AKIN10.1/3. pBT10 containing proASN1::GUS or proProDH::GUS (Dietrich et al., 2011) was used as a reporter plasmid. For transformation, 200 μl of the protoplast suspension were gently mixed with the DNA and 220 μl of PEG (40% PEG 4000, 200 mM mannitol, 100 mM CaCl2) were added, followed by gentle mixing, and 10 min incubation at room temperature. The protoplasts were then washed by addition of 800 µl W5 and 1 min centrifugation at 300×g. The supernatant was removed and the protoplasts were incubated for 16 hr in 200 µl of WI solution (4 mM MES, 500 mM mannitol, 20 mM KCl, pH 5.7) in the growth chamber in order to not affect their diurnal circle. GUS and NAN enzyme assays were performed according to Kirby and Kavanagh (2002) and the relative activity of GUS and calculated as GUS/NAN. The expression of the effector constructs was confirmed by Western blot analysis.

Sequence alignment of bZIP63 homologues

Request a detailed protocol

Homologues of bZIP63 in 8 plant species were identified by blasting the protein sequence of bZIP63.2 on the Phytozome webpage (http://phytozome.net). The protein sequences were aligned with ClustalΩ (http://www.ebi.ac.uk/Tools/msa/clustalo/) using the default settings and the Clustal output file was imported into GeneDoc (http://www.psc.edu/biomed/genedoc) for visualization and minor adjustments of the alignment. The identity matrix for the alignment was calculated using the PID3 method in the SIAS webmask (http://imed.med.ucm.es/Tools/sias.html).

Gene identifiers of the bZIP63 homologues (numbers are the Phytozome10 references): Medicago truncatula (Medtr7g115120), Glycine max (Glyma.10G162100), Populus trichocarpa (Potri.013G040700), Vitis vinifera (GSVIVG0102179000), Zea mays (GRMZM2G007063), Oryza sativa (LOC_Os03g58250), Selaginella moellendorffii (270282), Physcomitrella patens (Phpat.009G02690).

RT-qPCR

Request a detailed protocol

RNA was extracted with phenol from total rosettes of 4.5 week-old soil-grown plants at the indicated time points. 500 µl RNA extraction buffer (1% SDS, 10 mM EDTA, 200 mM NaAc, pH 5.2) and 500 µl acidic phenol (pH 4, Carl Roth, Karlsruhe, Germany) were added to 100 mg frozen and ground plant material, followed by 1 min vortexing and 10 min centrifugation. The aqueous phase was extracted twice with an equal volume of PCI (Phenol:Chloroform:Isoamylalcohol (25:24:21), Carl Roth, Karlsruhe, Germany) by vortexing for 30 s and 2 min centrifugation, and twice with chloroform. Then, the RNA was precipitated by adding ⅓ volume 10 M LiCl and incubating at 4°C for at least 2 hr, followed by 15 min centrifugation at 4°C. The pellet was washed once with 2.5 M LiCl and then with 70% EtOH, dried and resuspended in 50 µl RNAse free H2O. 6.5 µg RNA were treated with 2u of RQ1 DNAse (Promega, Mannheim, Germany), according to the manufacturer's instructions and precipitated again for at least 2 hr with ⅓ volume of 10 M LiCl to remove the DNAse from the solution. The solution was then centrifuged for 15 min, washed once with 70% EtOH, dried and resuspended in RNAse free H2O. 1.5 µg RNA were reverse transcribed using M-MLV RT [H-] Point Mutant (Promega, Mannheim, Germany) according to the manufacturer's instructions and diluted 1:4 with RNAse free H2O. 2 µl of cDNA (15 ng/µl) or H2O (for the no template control) were added to 18 µl of a PCR reaction mix containing 0.25u DreamTaq polymerase (Thermo Scientific, Vienna, Austria), 1× Dream Taq buffer, additional 3 mM MgCl2, 100 µM dNTPs, 400 nM of each primer, and 0.4× SYBR Green I nucleic acid gel stain (Sigma–Aldrich, Vienna, Austria). The qPCR was performed on a Mastercycler ep realplex2 (Eppendorf) in white 96-well twin.tec real-time PCR plates (Eppendorf, Vienna, Austria) with the following program: 1 cycle with 3 min at 95°C, followed by 40 cycles with 20 s at 95°C, 20 s at 59/60°C, and 12 s at 72°C. 2–3 technical replicates were used. A 20 min melting curve was added in the end. For data evaluation, raw data were imported into LinRegPCR (version 2014.02, Ramakers et al., 2003), where the PCR efficiency was checked and the N0 (staring concentration of cDNA) was calculated with a common baseline setting for each primer pair. Samples excluded by LinReqPCR were not used. In case only two technical replicates were used, samples for which the dCq between the two technical replicates was bigger than 0.5 were also excluded from the calculations. The mean N0 of the technical replicates was calculated for each sample and primer pair. The resulting mean N0 of the test genes was then normalized by dividing by the mean N0 of the reference genes. For qPCR of AKIN10, AKIN11 (Figure 5—figure supplement 1D); bZIP63, bZIP1, bZIP11 (Figure 5—figure supplement 3); ASN1, DIN10, and ProDH (Figure 7G) TBP2 (Tata binding protein 2) was used as a reference gene. For pPCR of bZIP63, (Figure 1—figure supplement 1D,E, Figure 7—figure supplement 1A); CAB2, SEN1, SAG103, and YLS3 (Figure 1—figure supplement 3B); ASN1, ProDH, BCAT2, AA-TP family protein, and DIN10 (Figure 5—figure supplement 2C) the geometric mean of the values for TBP2 and MHF15 (Thioredoxin superfamily protein) were used. The normalized N0 values were then used to calculate the mean and SD. Outliers were determined with the Grubb's test and removed.

Statistical tests

Request a detailed protocol

Statistical tests were performed in excel or R. Equality of variances was tested with Levene's test. This was followed by unpaired t-tests or multivariate statistics. In case of equal variance ANOVA followed by Bonferroni or TukeyHSD corrected pairwise comparison was chosen. In case of unequal variance Welch corrected ANOVA was applied, followed by Games Howell test of all samples or pairwise comparison of samples with equal variance.

Gene identifiers of Arabidopsis thaliana genes

Request a detailed protocol

bZIP63 (At5g28770), bZIP1 (At5g49450), bZIP11 (At4g34590), bZIP2 (At2g18160), bZIP44 (At1g75390), bZIP53 (At3g62420), AKIN10/SnRK1.1 (At3g01090), AKIN11/SnRK1.2 (At3g29160), AKINβ1 (At5g21170), AKINβ2 (At4g16360), SNF4/AKINβγ (At1g09020), SnAK2/GRIK1 (At3g45240), ASN1/DIN6 (At3g47340), DIN10 (At5g20250), ProDH/ERD5 (At3g30775), BCAT2 (AT1G10070), AA-TP (amino acid transporter) family protein (AT2G40420), TBP2 (At1g55520), MHF15 (At5g06430), CAB2 (At1g29920), SEN1/DIN1 (At4g35770), SAG103 (At1g10140), YLS3 (At2g44290).

References

  45. Book
    1. Ma J
    (2012)
    Reprogramming of metabolism by the Arabidopsis thaliana bZIP11 transcription factor. PhD thesis
    Utrecht University.

Decision letter

  1. Thorsten Nürnberger
    Reviewing Editor; University of Tübingen, Germany

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for choosing to send your work entitled “SnRK1-triggered switch of bZIP63 dimerization mediates the low-energy response in plants” for consideration at eLife. Your full submission has been evaluated by Ian Baldwin (Senior Editor), a Reviewing Editor and three peer reviewers, and the decision was reached after discussions between the reviewers. Based on our discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

While the reviews acknowledge the overall quality of your work, questions remain about the validity of one of the major tools used in your study (the akin10 mutant). As the use of this mutant is crucial for major conclusions of your work, comprehensive characterization of this mutant is required before further consideration of your submission. We would welcome re-reviewing this manuscript if the major concerns listed in the reviews below were convincingly addressed.

Reviewer #1:

Plant SnRK1 kinases are structural and functional homologs of yeast Snf1 and animal AMP-activated kinases that play a central role in sensing energy depletion and maintaining cellular energy homeostasis. In this paper, a multinational consortium set out to demonstrate that Arabidopsis SnRK1, carrying the AKIN10 catalytic subunit, phosphorylates in vivo the C-group bZIP63 transcription factor. bZIP63 has been identified to control transcription of some SnRK1-regulated target genes, involving ASPARGINE SYNTHASE 1 (ASN1/DIN6), PROLINE DEHYDROGENASE (PDH1/PRODH), RAFFINOSE SYNTHASE (DIN10) and SENESCENCE 1 (SEN1). To justify the statements in the Abstract and Introduction that bZIP63 is a key regulator of starvation response, the paper starts with limited characterization of a bzip63 T-DNA insertion mutant, and two bZIP63-ox overexpression lines.

The authors use a phenotypic assay by placing plants for 9 days into the dark and monitoring rosette leaf senescence/bleaching by ImageJ scanning of green versus bleached/yellow leaf areas. Whereas in wild type (wt) the first two leaves (numbered 3 and 4 in Figure 1) show bleaching, they remain greenish in the bzip63 mutant (defined as stay-green phenotype), and 3-5 more leaves undergo bleaching in the bZIP63-ox plants, suggesting enhanced senescence response.

When grown in the light, the bzip63 and bZIP63-ox plants do not differ from wt.

Next, the authors compare metabolites entering or derived from glycolysis and citrate cycle in light-grown (L) and dark exposed (elongated night, EN) plants and find that the levels of “most” free amino acids is increased in the bzip63 mutant and decreased in bZIP63-ox plants, whereas sugar levels are stated to show an opposite trend of changes. Based on these data the authors conclude that bZIP63 is a key regulator of starvation responses (see Abstract and Introduction).

When inspecting the metabolite profiling data, it is clear that the discussion of the results is rather superficial. For example, raffinose levels are decreased, whereas trehalose levels are increased in both mutant and ox lines, similarly to glucose in the dark. In general, we do not learn much from the metabolomics data. The actual conclusion is that “misregulation of bZIP63” (in the subsection “bZIP63 controls dark induced senescence and primary metabolism”) has an effect on metabolism.

On the other hand, we do not find any additional data (such as quantitative measurement of chlorophyll levels, degradation products, or other senescence markers, such as enzymes or transcripts), which would provide a further insight documenting how alteration of bZIP63 levels modulate the starvation-induced and/or senescence responses. Whether 9 days of dark treatment can be considered as starvation or general stress response remains a matter of discussion.

By performing 2D and Phos-tag gel westerns (Figure 2), the authors detect multiple phosphorylated (lambda phosphatase sensitive) isoforms of bZIP63-GFP immunoprecipitated from overexpressing plants. On the Phos-tag gel, they show a rather weak band termed “supershift” of bZIP63, which appears in dark-treated plants, and detect somewhat weaker bZIP63-GFP bands with altered pattern plants grown with 1% sucrose. They use the latter result, without explicit explanation or demonstration, to suggest that the observed phosphorylation changes reflect inhibition of SnRK1 by 1% sucrose.

By mass spectrometry analysis, seven phosphorylated residues were then identified in the immonoprecipitated bZIP63-GFP protein. Subsequent in-gel kinase tests indicate that bZIP63 is phosphorylated by multiple protein kinases showing apparent molecular masses of 50-60 and 35-38 kDa. In Figure 2A, it should be clearly indicated that the plant extract used for in-gel kinase assays derived from hydroponic root cultures (see “Kinase assays”, third paragraph), along with the conditions used for root culture in the case of the experiment in Figure 2A.

To isolate the kinases detected in the in-gel assay, GST-bZIP63 was immobilized on GST-Sepharose and plant protein extracts made from root cultures or root-derived cell suspension (as the latter is maintained in the presence of sucrose, please specify which of these sample is shown in Figure 2) were subjected to affinity binding followed by NaCl elution. The eluted proteins resolved between 30 and 60 kDa were isolated from in-gel kinase PAGE (if I understood well), and identified by mass spectrometry. From 28 candidates only 7 kinases (including the SnRK1 catalytic AKIN10, AKIN11 and activating subunit SNF4, as well as casein kinase subunits CKA1, CKA2, CKB1 and CKL2) were identified by 2-9 peptides. It should be noted in the Materials and methods how the excess of His6-bZIP63 was filtrated out during the mass spectrometry analysis.

In addition, it would be worth knowing whether the GST-alone matrix had retained any of these kinases.

Furthermore, αAKIN10 western showing an enrichment of AKIN10 on the GST-bZIP63 matrix versus GST would be more convincing.

From here on, the authors concentrate on the analysis of bZIP63 phosphorylation by the SnRK1 catalytic subunits AKIN10, which according to Baena-Gonzalez et al. and Jossier et al. (Plant J 59: 316, 2009), is at least 10-20 times more active/abundant compared to its homolog AKIN11 in Arabidopsis. Figure 3B illustrates that GST-AKIN11 purified from E. coli also phosphorylates bZIP63 less efficiently compared to GST-AKIN10 in vitro. It is strange that no autophosphorylation of AKIN10 was ever detected by the authors either in normal or in-gel kinase assays. Any explanation?

As a key argument, the authors use the akin10 GABI_579E09 T-DNA insertion mutant to demonstrate that AKIN10 is indeed responsible for in vivo phosphorylation of bZIP63. Unfortunately, according to the reviewer's knowledge, this mutant has not yet been characterized by others, so, to a certain extent, this must be performed by the authors in this paper.

In Figure 4A, they show that one of the phosphorylated bZIP63 bands around 55-57 kDa is missing in their in-gel kinase assay with akin10 mutant extracts, and they fail to detect AKIN10 by western with the Agrisera αAKIN10 antibody in roots, but observe a lower Mw weakly cross-reacting protein band in leaf extracts. According to these results, the GABI_579E09 insertion mutant appears to produce no immunologically detectable AKIN10. Yet, the authors could successfully cultivate the mutant, which means that it must be fertile. If AKIN10 is a central regulator of starvation responses, and shows an order of magnitude higher activity compared to AKIN11 in Arabidopsis, based on the published striking AKIN10 RNAi senescence phenotype by Bhaena-Gonzalez et al., one would expect a dramatic growth phenotype. Consequently, it is absolutely necessary to show the phenotype of the akin10 mutant both in the light and dark that we can compare that to that of the bzip63 mutant.

Furthermore, the analysis of transcription of bZIP63 target genes should provide further clear evidence that post-translational regulation of bZIP63 is indeed impaired in the GABI_579E09 T-DNA insertion mutant. This is because Figure 4E does not really show a very convincing change in the Phos-tag pattern of phosphorylated bZIP63 despite a lower level of “supershift” indicated by asterisks (see Figure 4E and Figure 4–figure supplement 3).

Figure 4C to D show yeast 2-hybrid and Agrobacterium infiltrated N. tabacum protein-protein interaction assays indicating that that unlike AKIN11, the AKIN10 catalytic and SNF4 activating subunits of SnRK1 interact with bZIP63. Given the high level of amino acid homology between AKIN10 and AKIN11, it is surprising that bZIP63 fails to bind AKIN11 and provide a possible hint about the location of possible interaction sites. Did you map these?

On the other hand, it is remarkable that the kinase-substrate interaction is so strong that this can be clearly detected by both assays. However, as these are heterologous assays based on overexpression of proteins, it is absolutely necessary to present co-immunoprecipitation data using the later described bZIP63 GENE construct (and not 35S ox lines) and αAKIN10 western from wt and akin10 mutant plants.

According to Figure 4–figure supplement 1, the putative upstream activating kinase SnAK2 enhances the activity of AKIN10 (and may be that of AKIN11). SnAK2 is supposed to phosphorylate the T-loop of both AKIN10 and AKIN11. Yet, we do not see enhanced phosphorylation of GST-AKIN10/11 in this figure. Any explanation?

As GST-AKIN10/11, GST-bZIP63 and GST-SnAK2 have closely similar masses, it would be necessary to resolve them in a lower % of SDS-PAGE.

In vitro phosphorylation of bZIP63 by AKIN10 and AKIN11 looks convincing in Figure 4–figure supplement 1B, so this approach should have provided a simple means to determine both AKIN10 and AKIN11 phosphorylation sites comparatively in bZIP63 by mass spectrometry. Yet, in Figure 5, a complicated intuitive combination of alanine replacements of in vivo observed seven serine phosphorylation sites is used for demonstration that GST-AKIN10 probably phosphorylates only S29, S294 and S300 residues of bZIP63.

The referred “consensus SnRK1 sites” in Figure 5B require a corresponding reference either in the text or figure legend.

The promoter activation assays in protoplast support the conclusion that S29 is the critical AKIN10 phosphorylation site, but in combination with the S294A and S300A amino acid exchanges, AKIN10-S29A shows lower transactivation of ASN1/DIN6-GUS and ProDH-GUS reporters. Whereas sequential phosphorylation might indeed trigger threshold-dependent conformational change of TFs, it is premature to conclude anything about possible order of phosphorylation sites (in the subsection “AKIN10 phosphorylates three conserved and functionally important serine residues in bZIP63”).

In the subsection “The AKIN10 target sites play an important role for bZIP63 function in planta” it is stated that: “The phosphorylation status of the wt construct after 6h of light and extended night and in the presence or absence of sugar was similar to the one observed in ox#3 plants (Figure 6B; Figure 6–figure supplement 1E).” Concerning the reproducibility and similarity of Phos-tag gels, would it be possible to run on the same gel the reactions shown in Figure 2B and/or Figure 4E with Figure 6B, as well as Figure 4–figure supplement 3 and Figure 6–figure supplement 1E, to have an idea about the identity/mobility of phosphorylated bands in the respective reactions? Probably, it would also be beneficial to compare the Phos-tag assays on the same gel with the reactions shown in Figure 5–figure supplement 2. This would be necessary because the Phos-tag gel depicted in Figure 6B looks dramatically different for the GY11 line in L and EN compared to the 35S:bZIP63-GFP line in Figure 2B (i.e., despite the authors' statement).

In the same subsection, I do not understand the logic of the following sentence: “Notably, hyper-phosphorylation of bZIP63 in the extended night was even stronger in the GY lines than in ox#3, probably due to lower bZIP63 expression.” Do the authors suggest that AKIN10 activity is limiting in terms of bZIP63 phosphorylation even if AKIN10 was supposed to be activated by extended night?

With the limitations mentioned above concerning the necessary parameters of bzip63 mutant characterization, the lack of complementation by the triple S/A bZIP63 mutant gene is documented. As the mutant gene constructs were generated by PCR of 2+xkb genomic DNA fragments, the Materials and methods should mention that all constructs were verified by sequencing.

In Figure 6G, it is rather surprising that the GY11 wt complementing line shows significantly higher induction of ASN1 and DIN10. In the fold-induction curve and column diagram to the right, the standard error of Ws-2 is huge at 4h of EN, thus no significant difference between wt and bzip63 mutant can be claimed. In this figure part, either the curves shown to the left should include all four samples (Ws-2, bzip63, GY11 and GAY14) or all samples at the four time points should be presented as column diagrams. Given the data presented in Figure 6–figure supplement 1, I wonder why did the authors use the low expressing GY11 line for comparison with GAY14. From Figure 6–figure supplement 1C it is also apparent that the GAY4 line shows no expression of triple S/A mutant bZIP63-YFP, so only one bzip63 mutant transformed with the triple phosphor-site mutant of bZIP63 was analysed.

Ehlert et al. reported that, when overexpressed in plant protoplasts, bZIP63 can heterodimerize with members of the S-group, and in those assays bZIP63 homodimers were barely detectable. Kang et al. (2010) reported later that bZIP63 also interacts with the Arabidopsis response regulator ARR18 involved in cytokinin signaling, whereas alone (i.e. by interacting with S-group partners in heterodimers), bZIP63 functions as negative regulator of osmotic stress responses and seed germination.

Several heterodimerization partners of bZIP63 are well characterized in terms of their transcription regulatory targets. Therefore, it is apparent that their regulatory functions might be modified by bZIP63, where heterodimerization capability of bZIP63 could be controlled post-translationally, e.g. by phosphorylation through SnRK1 and potentially other protein kinases. This idea should be introduced clearly, with some necessary references to the functions of suspected bZIP63 dimerization partners at the beginning of the manuscript and discussed in this term at the end. For example, the observed effect of bzip63 mutation on the levels of free amino acids could be related to the function of its heterodimerization partner bZIP1. So, probably citing a recent paper from Para et al. (Proc Natl Acad Sci USA. 111: 10371, 2014) may be helpful for discussion of metabolite changes observed in the bzip63 mutant.

Evidently, the bzip63 mutation could influence the function of bZIP1, i.e. if we assumed that the bZIP1 functions as heterodimer with bZIP63. It is therefore interesting that overexpression of AKIN10 in protoplasts has only a slight effect on heterodimerization of overexpressed Gal-AD-bZIP1 and Gal-BD-bZIP63. While this artificial protoplast dimerization assay in Figure 7A indicates over 60-fold enhancement of Gal-AD-bZIP11+Gal-BD-bZIP63 dimerization, at the level of GAL-UAS4:GUS reporter, about 15-fold increase is observed, which is reduced to about 2-3-fold by triple mutation of the identified S29/S294/S300 bZIP63 phoshorylation sites. In case of Figure 7B and C, I could not fully understand why the authors used two different (i.e. GAL-UAS4:GUS and GAL-UAS4:LUC) reporters in the dimerization and reporter activation assays. This correlative evidence between the dimerization and reporter activation assay should definitely be strengthened by performing similar dimerization assays with bZIP63 carrying the S29A/S294/S300 phospho-site replacements. This appears to be obligatory.

Although the Discussion takes care to announce that “to our knowledge this is the first time that phosphorylation outside the bZIP domain was shown to affect dimerization with different partners in a distinct way”, it is rather flat. Somewhere, it should mention that the current study did not answer the question of whether the observed phosphorylation sites would modify the cellular location (i.e. nuclear import) and stability of bZIP63, in addition to its heterodimerization, which should be anyhow confirmed by co-IP studies using native gene constructs.

Unfortunately, the majority of the Discussion includes high-flying general statements, such as, “Here we show that bZIP63 plays an important role in the energy starvation response and metabolic regulation…”. If this is so, the reader will ask whether transcription of bZIP63 and its heterodimerization partners is controlled by AKIN10 and is altered in the akin10 mutant (i.e. a necessary but missing control).

Can the enhanced senescence caused by bZIP63-ox compensated by glucose or sucrose supply? Is there any evidence for enhanced ROS production in the bZIP63-ox lines or change or red-ox regulation in the bzip63 mutant?

“We measured strong differences in proline and asparagine levels in the bZIP63 ko and ox […] like altered carbon or nitrogen assimilation or protein turnover” (in the subsection “bZIP63 is an important metabolic regulator in the starvation response”). So, how would bZIP63 control these mechanisms? Through controlling genes involved in N or C signaling or uptake?

You say: “We found that the SnRK1 kinase AKIN10, which was proposed to be a central regulator of transcription in starvation response (Baena-Gonzalez and Sheen, 2008), is the major kinase responsible for the starvation-induced hyper-phosphorylation of bZIP63. Therefore, bZIP63 presents the first TF acting as direct target of SnRK1 in the transcriptional energy deprivation response” (in subsection “bZIP63 function is regulated by SnRK1-dependent phosphorylation”). Yes, but the hyper-phosphorylation of bZIP63 under extended night (called “starvation”, though it was 9 days of dark treatment) caused the appearance of only a weak band marked by asterisks on the images of Phos-tag gels, which represents the only evidence supporting this statement.

In the same subsection, you state: “Reporter activation assays in protoplasts revealed that these (i.e., AKIN10 phosophorylation) sites are essential for AKIN10-dependent induction of ASN1 and ProDH by bZIP63”. This may be, but AKIN10 alone also activates these reporter genes, probably through other factors, which according to this paper are probably not the S-group bZIPs (i.e. as they appear not to be phosphorylated by AKIN10). It would be fair to mention this somewhere.

Still in the same subsection, the authors say: “The relatively small difference in ProDH expression in plants, as compared to the protoplast assay, is probably due to redundancy within the C/S1 group bZIPs…”. Does this sentence predict that other C-group bZIPs could be SnRK1 substrates? Would it be not necessary to discuss potential conservation of observed AKIN10 sites in other bZIPs?

In the same part of the text, the authors state: “Our data provide compelling evidence that bZIP63, but none of the S1 group bZIPs, is a bona fide in vivo target of SnRK1 in low energy signaling.” This should clearly be demonstrated by identical Phos-tag gel patterns of bZIP63 isolated from sucrose-treated or akin10 mutant plants and that of the bZIP63 S29A/S294A/S300A protein obtained from bzip63 plus/minus sugar or the akin10 mutant.

Reviewer #2:

In this manuscript, Mair et al. report the function of bZIP63 as a key regulator in plant starvation response. Furthermore, using a combination of biochemical and omics approaches, the authors demonstrate that bZIP63 is regulated by protein phosphorylation and the kinase involved is AKIN10, an AMPK/Snf1/SnRK1 member. Phosphorylation of bZIP63 by AKIN10 changed its dimerization preference, thereby affecting target gene expression and ultimately plant primary metabolism.

This report identifies a missing link between energy sensing and transcriptional reprogramming of metabolism in plants under starvation conditions. The topic is of interest to a wide audience. The data are of high quality, clearly laid out, and mostly convincing. I only have one suggestion that should clarify the Phos-tag results and further enhance the quality of the paper:

Phos-tag analysis of in vitro phosphorylated bZIP63 samples allows the attribution of each shifted band to the phosphorylation of a specific Ser residue (Figure 5–figure supplement 2). In contrast, the same analyses of in vivo samples showed variable banding patterns that are not easy to reconcile (Figure 2B, Figure 4E, Figure 4–figure supplement 3, and Figure 6B). This could be a result of phosphorylation of bZIP63 by other kinase(s) on S59, S102, S160, and S261, four sites that are not phosphorylatable by AKIN10. I would suggest the authors to determine whether the mutation of these four Ser residues to Ala will affect the function of bZIP63. If it does, the data would reveal that bZIP63 is regulated by multiple kinase pathways to carry out its function in the energy transition response. If it does not, the data would demonstrate that AKIN10-phosphorylation on S29, S294 and S300 is the major phosphorylation event that controls the activity of bZIP63. By performing Phos-tag analysis of bZIP63-S59A/S102A/S160A/S261A, the authors can provide more convincing results about the in vivo phosphorylation of bZIP63 by AKIN10 kinase.

Reviewer #3:

The authors demonstrate low energy stress-dependent SnRK-catalyzed direct phosphorylation of bZIP63 transcription factor as prerequisite for differential bZIP homo/hetero dimerization. Their comprehensive clear data provide for the first time mechanistic insight in how transcriptional reprogramming of gene subsets in a given pathway is driven by a conditional differential hetero/homodimerization of key transcriptional regulators.

The manuscript is well written, clear in its argumentation, and statements are well backed by an almost overly thorough analysis and a large amount of data of high quality. I would like to comment on two aspects:

1) Because SnRK and bZIP are well known proteins in plant biology, and because phosphorylation of bZIPs has been reported before, the true novelty in the manuscript is in my opinion the phosphorylation-dependent differential dimerization leading potentially to a change in transcriptional specificity plus activity toward distinct target genes (summarized in Figure 7D – model). Data in Figure 7A rely on a protoplast two-hybrid assay, allowing a transcriptional readout (equal protein amount in transfected cells has not been included here). To more prominently support the key message of the manuscript will have to address:

i) Differential dimerization on protein level: conducting for example transient co-localization studies based on (multi-color) fluorescent proteins/BiFC (Waadt et al., 2008), or using biochemical means to compare phosphorylation-dependent bZPI63/bZIP63 with bZIP63/bZIP11 formation, and in case even include bZIP S29D (or other) phospho-mimic constructs.

ii) Show differential target gene specificity: Figure 7 contains as marker for transcriptional activity a gene which is target to all three bZIPs (1, 11, 63). Even more convincing will be phosphorylation/heterodimerization-dependent activity toward a distinct bZIP11- or bZIP63-controlled gene, respectively.

2) The primary identification of “in vivo” phosphorylation sites is conducted on bZIP-GFP overexpressing lines on material harvested in both light and prolonged dark. Therefore, basic constitutive bZIP phosphorylation (and the involvement of other kinases than AKIN10) cannot be excluded. Ideally, one may ask for the degree of conditional in vivo phosphorylation (in particular at S29 of endogenous bZIP) performing MRM/SRM phosphoproteomics.

https://doi.org/10.7554/eLife.05828.042

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

We have carefully read the reviewer’s comments and found that although some points that are mentioned are valid and we agree that these have to be addressed, we also saw that some parts of our manuscript were not written clearly enough and have been misunderstood. Therefore we would like to respond here directly to two issues that we think are the major issues for all three reviewers and ask furthermore for concrete feedback regarding the resubmission of a revised version of this work. To this end, we outline the experiments and further clarifications we would propose for a revision.

1) The akin 10 mutant:

We understand the necessity of a comprehensive characterization of the akin10 mutant, which we used for our study, and we apologize for not including that in our manuscript. This mutant has been used – and confirmed – by three groups in our consortium independently. Author response image 1 shows clearly that this mutant (GABI KAT GABI_579E09) is indeed a valid akin10 null mutant. The T‐DNA insertion was indicated at the very 3’ end of the AKIN10 gene directly in front of the last exon (Author response image 1A). We could indeed confirm the insertion at this position using specific primer combinations (Author response image 1B). The Gabi homepage indicates that there might be a second insertion in another gene (IMS2, At5g23020; Author response image 1C). This was tested by PCR using gene‐specific primers against the flanking region of the second insertion but no insertion could be detected (Author response image 1D).

We checked the expression level of AKIN10 protein by Western blotting using two antibodies: firstly the AKIN10 specific antibody from Agrisera (AS 10919), and secondly an antibody against the phosphorylated T‐loop threonine (T172) of AMPKα, the mammalian AKIN10 ortholog (Cell Signaling #2531), that specifically recognizes the phosphorylated T‐loop threonine in AKIN10 and AKIN11 (T175 and T176, respectively; Baena‐González et al., 2007). The position of the peptide recognized by each antibody is indicated in Author response 1A. Notably both antibodies bind to regions far before the T‐DNA insertion in the AKIN10 sequence. As shown in Author response image 1E, the AKIN10 antibody recognizes a single protein at the expected MW (59‐61 kDa) in wt extracts, but not in the akin10 mutant. The anti Thr 172 antibody detects a double band of 61 and 58 kDa in the wt, and only the lower band (58 kDa = AKIN11) in the akin10 line. These data were completely independently obtained in the Baena‐ Gonzalez lab as well as in our lab.

As reviewer 1 raised also concerns regarding a potential growth phenotype of the akin10 mutant, we include here also a phenotypical growth analysis of the akin10 mutant in Author response 1F and 1G. Author response image 1F shows that the akin10 mutant displays normal growth with respect to leaf number, fresh and dry weight as well as water content. As can be seen in Author response image 1G the akin10 mutant displays a little delay in flowering as compared to the wt, but it is flowering and produces viable seeds without problem. This is also in perfect agreement with the data on the AKIN10 RNAi line reported by Baena‐Gonzalez et al., 2007 (Nature 448 (7156):938‐42). The two different AKIN10 RNAi lines described in that paper showed also normal growth. It is the akin10/11 double mutant that is apparently not viable and when an akin10/11 knockdown is generated through VIGS, the plants show premature senescence that precludes flowering. It is also shown there that AKIN10 overexpression alters inflorescence development.

Author response image 1
Download asset Open asset
Characterization of the akin10 mutant line GABI_579E09.

(A‐D) Mapping of the T‐DNA insertion site. (A) The top scheme represents the genomic locus of AKIN10 (AT3G01090). UTRs are grey, exons are green boxes. The first exon, which is only present in splicing form 2 (sf2), is shown in light green. By sequencing of the flanking regions the T‐DNA insertion was mapped to position 2583‐2593 (11pb deletion), just before the last exon. Primers used for the genotyping PCR shown in (B) are indicated. The scheme below represents the AKIN10 protein with the kinase domain (KD) in yellow and the kinase associated (KA) domain in blue. The position of K48 (ATP binding; mutated to M in the inactive version) and T175 (target for SnAK2 in the T‐loop), as well as the approximate binding sites of the antibodies against AKIN10 (Agrisera; AS 10919) and AMPKα‐pT172 (Cell Signaling; #2531) are indicated. (B) Genotyping PCR with primers binding in the AKIN 10 sequence (fwA and rvA) and the T‐DNA left border (LB). (C) The scheme represents the genomic locus and protein of IMS2 (AT5G23020), a potential second T‐DNA insertion site. Sequencing of the T‐DNA flanking region by GABI KAT (www.gabi.kat.da) produced a sequence with partial homology to IMS2. However, this insertion was not confirmed. The proposed position of the T‐DNA is indicated. (D) Genotyping PCR with primers binding in the IMS2 sequence (fwI and rvI) and the sequencing primer (seq) used by GABI KAT. (E) Western blots against AKIN10 in Col‐0 wt and akin10 plants. Antibodies against AKIN10 (top) and AMPKα‐pT172, which recognizes both AKIN10 and AKIN11 phosphorylated in the T‐loop, (bottom) were used. The position of AKIN10 and phospho‐AKIN10/11 is indicated. Left: 5 week‐old plants grown on soil in a 12h/12h light/dark cycle. Total proteins were extracted from rosettes harvested after 6h of light (L) or extended night (EN). Right: 2 week‐old seedlings in liquid ½ MS medium. Total proteins were extracted from seedlings treated with 6h of extended night in the presence (+) or absence (‐) of 1% sucrose. (F) Comparison of 8 week‐old wt and akin10 plants. The plants were grown under short day conditions (8h/16h light/dark). The number of leaves, fresh weight (FW), dry weight (DW), and the ratio between FW and DW was determined. The bars represent the mean ± SD of 19‐20 replicates. T‐tests did not produce significant differences between wt and mutant. (G) Mild flowering phenotype of akin10. Col‐0 wt and akin10 plants were grown under SD conditions and the number of leaves at bolting time were counted. The bars represent the mean ± SD of 15 and 11 replicates, respectively. The p‐value of an unpaired t‐test with unequal variance was < 0.05 as indicated by *.

https://doi.org/10.7554/eLife.05828.044

We also tested the expression levels of AKIN10 target genes in response to six hours extended night treatments in this akin10 line (Author response image 2). After six hours of extended night (EN), we observed a clearly impaired induction of ASN1 (DIN6) and ProDH. DIN10 (raffinose synthase) was not affected under this conditions, but this time point might already be a bit too late, as we observed clear changes in the induction of all three genes already starting after two hours EN in the bzip63 mutant (Figure 6G). Therefore we selected an earlier time point (4 h) for the analysis of bZIP63 target gene expression in the complementation lines. In general, it is important to consider that AKIN10 and AKIN11 seem to act in a dynamic system, in which AKIN11 can take over the function of AKIN10 after some time. This becomes evident at different levels: firstly, the viability of the akin10 mutant; secondly, the level of target gene expression (double knockout is required to abolish induction in response to starvation); thirdly, the remaining background in target phosphorylation (i.e. bZIP63) as, for example, seen in the phos‐tag analysis (see Author response image 3).

Author response image 2
Download asset Open asset

Expression levels of the proposed AKIN10/bZIP63 target genes ASN1, DIN10, and ProDH in wt and the akin10 mutant line in light (L) and extended night (EN). Col‐0 wt and akin10 plants were grown on soil in a 12h/12h light/dark regime for 5 weeks and harvested after 6h of light and extended night. RNA was extracted from whole rosettes, reverse transcribed and used for qPCR of ASN1 (ASPARAGINE SYNTHASE 1; AT3G47340), DIN10 (DARK INDUCIBLE 10; AT5G20250), and ProDH (PROLINE DEHYDROGENASE; AT3G30775). TBP2 (TATA BINDING PROTEIN 2; AT1G55520) and MHF15 (Thioredoxin superfamily protein; AT5G06430) were used as reference genes. Raw data were analyzed with LinRegPCR (version 2014.02, Ramaker et al., 2003). The mean N0 values from two technical replicates were normalized using the geometric mean of the values for TBP2 and MHF15. The bars represent the mean ± SD of 3 biological replicates normalized to the expression in wt in light. P‐values from t‐test < 0.05 are indicated by *.

https://doi.org/10.7554/eLife.05828.045

2) In vivo phosphorylation status of bZIP63 detected in phos‐tag gels:

Reviewer 1 and 2 raised several questions about the phosphorylation status of bZIP63 as detected by Western blotting of phos‐tag gels. As reviewer 1 pointed out, so far no transcription factor has rigorously been shown to represent a direct phosphorylation target of SnRK1 in plants. Therefore, we think that the phos‐tag data, which show the in vivo phosphorylation status of bZIP63, are crucial for this manuscript and we feel that we need to explain these better in different conditions, treatments and of AKIN10.

In order to better illustrate the different phosphorylation status and to allow a better comparison of the different experiments, we prepared a figure that shows a direct comparison of all phos‐tag gels from the different experiments (Author response image 3).

As control for the non‐phosphorylated bZIP63, the recombinant protein is shown in the very left panel. For all other samples we have now labelled the bands from 1‐7 according to an increase in their phosphorylation status. This is the maximum of different isoforms we could identify on different phos‐tag gels and corresponds to the seven identified in vivo phosphorylation sites from the combined MS analyses. Of course, not all isoforms are always visible on the gel as they depend on the phosphorylation level, and furthermore, the phosphorylation status is also highly dynamic, which complicates the picture. Nevertheless, the main point we want to make in the manuscript – and we think this is clearly evident from the figures – is that the level of bZIP63 phosphorylation changes in response to the different treatments and is furthermore depending on AKIN10.

Author response image 3
Download asset Open asset

Comparison of bZIP63 phosphorylation on phos‐tag gels in response to different treatments and in different genetic backgrounds. GFP/YFP‐tagged bZIP63 was detected by western blotting with an antibody against GFP (top). The Coomassie Brilliant Blue (CBB) stained membranes are given below. Recombinant bZIP63‐YFP is shown as a control for non‐phosphorylated bZIP63. On the left, phosphorylation of bZIP63 after 6h of light (L) or extended night (EN) is shown. Proteins were extracted from whole rosettes of 5 week‐old plants grown on soil in a 12h/12h light/dark cycle. On the right, phosphorylation of bZIP63 in response to sugar treatment is depicted. Seedlings were grown for 1 week in liquid ½ MS in the presence of 0.5% sucrose, followed by 1 week without sucrose. They were then treated with 6h of EN in the presence (+) or absence (‐) of 1% sucrose and harvested. The plant lines used are two bZIP63 over‐expressor lines – 35S::bZIP63‐GFP in wt and UBI10::bZIP63‐YFP in akin10 – and two genomic complementation lines expressing YFP‐tagged bZIP63 – wt bZIP63 (GY11) and bZIP63 harboring S/A mutations in all three AKIN10 target sites (AY14) – under its endogenous promoter in the bzip63 background. All bands (except unspecific bands and degradation bands in the two blot on the right) are indicated by black and grey dots. The color of the dot indicates the strength of the band in the blot. The scheme on the right shows the numbers assigned to each band in the other figures.

https://doi.org/10.7554/eLife.05828.046

In the first panel from the left, it can be seen that, in the light, bZIP63 is already highly phosphorylated – which is not surprising considering seven identified in vivo phosphrylation sites and several potential kinases. However, the majority of bZIP63 is present in bands 5 and 6. In response to starvation (6 hrs extended night EN), band 7 appears as higher phosphorylated form. This new band 7 is barely detectable in the EN samples from the akin10 mutant. This might be considered as a small difference but these data should be seen in the overall context. In the genomic complementation, lines bZIP63 are almost exclusively detected in bands 5 and 6 in the light (so the same as before), and now most of the bZIP63 protein is shifted to the most phosphorylated form in band 7 in the extended night (EN). Strikingly, in the line complemented with the version of bZIP63 in which all three AKIN10 target sites are mutated to an alanine (S/A), ZIP63 is detected only in the bands 1 and 2 (so almost non‐phosphorylated) under all conditions. This makes clear that 1) the 3 mutated sites are the main in vivo phosphorylation sites and 2) the increased phosphorylation in EN happens at one of these sites.

As reviewer 1 pointed out, the band pattern in this line (GAY14) is not the same as in the akin10 line. This is indeed true, however, considering again that we identified more than one kinase as potential upstream regulators of bZIP63 this is not surprising and does not mean that AKIN10 does not phosphorylate bZIP63 in vivo. As mentioned before, one likely explanation for the differences in phosphorylation is that AKIN11 might take over part of the function of AKIN10 in the akin10 mutant. This would also explain the lack of a strong phenotype in the akin10 line (see Figure 1 and Baena-González et al., 2007) compared to the akin10 akin11 double ko. Although AKIN11 does not interact well with bZIP63 in the Y2H and BiFC assays, it might still interact with bZIP63 via SNF4 and form a stable complex in vivo. Moreover, other kinases (like the identified CKII or CDPKs) could also phosphorylate one or more of the three target sites of AKIN10. We will discuss this aspect more clearly in a revised version.

Reviewer #1:

Plant SnRK1 kinases are structural and functional homologs of yeast Snf1 and animal AMP-activated kinases that play a central role in sensing energy depletion and maintaining cellular energy homeostasis. In this paper a multinational consortium set out to demonstrate that Arabidopsis SnRK1, carrying the AKIN10 catalytic subunit, phosphorylates in vivo the C-group bZIP63 transcription factor. bZIP63 has been identified to control transcription of some SnRK1-regulated target genes, involving ASPARGINE SYNTHASE 1 (ASN1/DIN6), PROLINE DEHYDROGENASE (PDH1/PRODH), RAFFINOSE SYNTHASE (DIN10) and SENESCENCE 1 (SEN1). To justify the statements in the Abstract and Introduction that bZIP63 is a key regulator of starvation response, the paper starts with limited characterization of a bzip63 T-DNA insertion mutant, and two bZIP63-ox overexpression lines.

Molecular characterization and description of the bZIP63 ko and ox lines was provided in Figure 6–figure supplement 6A-C as well as in the Methods section on plant lines in the previous version of the manuscript. To make the information more accessible, we moved the characterization of the plant lines to a new figure in the beginning of the manuscript (Figure 1–figure supplement 1) and included some additional information on the characterization of the bzip63 line, as well as some plant pictures.

Furthermore, we have now also included a number of different assays to clearly show, that the observed phenotype is indeed dark-induced senescence (DIS) caused by energy depletion (see qPCR data) and that this phenotype can be rescued by addition of sugar.

The authors use a phenotypic assay by placing plants for 9 days into the dark and monitoring rosette leaf senescence/bleaching by ImageJ scanning of green versus bleached/yellow leaf areas. Whereas in wild type (wt) the first two leaves (numbered 3 and 4 in Figure 1) show bleaching, they remain greenish in the bzip63 mutant (defined as stay-green phenotype), and 3-5 more leaves undergo bleaching in the bZIP63-ox plants, suggesting enhanced senescence response.

We would like to correct the statement of the reviewer. As indicated in the text, all leaves used for senescence analysis were true leaves. Cotyledons were not included. Therefore the leaves numbered with 3 and 4 are not leaf 1 and 2 but really leaves 3 and 4.

When grown in the light, the bzip63 and bZIP63-ox plants do not differ from wt.

Next, the authors compare metabolites entering or derived from glycolysis and citrate cycle in light-grown (L) and dark exposed (elongated night, EN) plants and find that the levels of “most” free amino acids is increased in the bzip63 mutant and decreased in bZIP63-ox plants, whereas sugar levels are stated to show an opposite trend of changes. Based on these data the authors conclude that bZIP63 is a key regulator of starvation responses (see Abstract and Introduction).

When inspecting the metabolite profiling data, it is clear that the discussion of the results is rather superficial. For example, raffinose levels are decreased, whereas trehalose levels are increased in both mutant and ox lines, similarly to glucose in the dark. In general, we do not learn much from the metabolomics data. The actual conclusion is that “misregulation of bZIP63” (in the subsection “bZIP63 controls dark induced senescence and primary metabolism”) has an effect on metabolism.

We agree with this and have intensively studied the literature and found a number of relevant links of our metabolite data to starvation and senescence. We have therefore completely re-written this part and are fully confident that the data now do clearly support the DIS phenotype and the role of the AKIN10-bZIP63 module in regulating primary metabolism in the low-energy response.

On the other hand, we do not find any additional data (such as quantitative measurement of chlorophyll levels, degradation products, or other senescence markers, such as enzymes or transcripts), which would provide a further insight documenting how alteration of bZIP63 levels modulate the starvation-induced and/or senescence responses.

We now include quantitative chlorophyll measurements of plants before and after 9 days in darkness (Figure 1–figure supplement 2B). Results are similar to the “green leaf area” data presented throughout our manuscript, which justifies the use of green leaf area as a crude measure for chlorophyll content. In this context we would like to mention that senescing leaves often have reduced water content. Therefore, when normalizing the chlorophyll content to fresh weight this can lead to overestimation of the chlorophyll content in strongly senescing plants. We thus believe that the “green leaf area” method produces more reliable data for estimating the chlorophyll content during dark-induced senescence.

Further, to confirm that the phenotype we observe is related to dark-induced and not age-induced senescence we did RT-qPCR of marker genes for dark-induced and natural senescence (Figure 1–figure supplement 3). Early time points up to day 4 were chosen as transcriptional response to darkness is faster than visible chlorophyll loss. CHLOROPHYLL A/B BINDING PROTEIN 2 (CAB2), as a marker for active photosynthesis was quickly downregulated, while SENESCENCE 1 (SEN1) and SENESCENCE ASSOCIATED GENE 103 (SAG103), were strongly upregulated. Both genes are known to be regulated by dark-induced senescence. In contrast, YELLOW-LEAF-SPECIFIC 3 (YLS3), which is induced by natural senescence, was not increased. We could not observe any strong differences in expression of these genes between wt and bzip63 plants, indicating that bZIP63 is not involved in the signaling pathways upstream of these senescence markers.

Moreover, as mentioned above, the delayed senescence in the bzip63 plants could be related to the metabolic phenotype we observed. Amino acids can serve as alternative energy under starvation conditions (Szabados and Savouré, 2010, Proline: a multifunctional amino acid; Szal and Podgórska, 2012, The role of mitochondria in leaf nitrogen metabolism). Therefore, the increased amount of total amino acids in the bZIP63 ko could lead to a delayed onset of visible senescence effects. We have discussed this possibility in the revised version of our manuscript.

Whether 9 days of dark treatment can be considered as starvation or general stress response remains a matter of discussion.

Dark induced senescence was suggested to be a result of sugar/carbon depletion (Buchanan-Wollaston et al., 2005, Comparative transcriptome analysis reveals significant differences in gene expression and signalling pathways between developmental and dark/starvation-induced senescence in Arabidopsis). We therefore tested whether the dark-induced senescence phenotype is energy related and could be complemented by the addition of sugars. We found that glucose can indeed almost completely rescue the dark-induced senescence phenotype of bZIP63 mutant in seedlings (Figure 1C-D). This points towards an involvement of bZIP63 in sugar/starvation signaling. Also, seedlings grown on plates with sugar showed enhanced growth and reduced or no chlorophyll loss as compared to the control plates without sugar, indicating that energy/carbon depletion is the major reason for senescence under these conditions.

By performing 2D and Phos-tag gel westerns (Figure 2), the authors detect multiple phosphorylated (lambda phosphatase sensitive) isoforms of bZIP63-GFP immunoprecipitated from overexpressing plants. On the Phos-tag gel, they show a rather weak band termed “supershift” of bZIP63, which appears in dark-treated plants, and detect somewhat weaker bZIP63-GFP bands with altered pattern plants grown with 1% sucrose. They use the latter result, without explicit explanation or demonstration, to suggest that the observed phosphorylation changes reflect inhibition of SnRK1 by 1% sucrose.

Although it is tempting to suggest a connection between the reduced phosphorylation of bZIP63 in the presence of sucrose and the activity of SnRK1, we do not claim at any point in the text that this is the case. We merely suggest that sugar feeding counters the starvation resulting from cultivation in extended dark conditions.

However, it is possible that some of the difference in bZIP63 phosphorylation between seedlings incubated in the presence and absence of sucrose is due to altered SnRK1 activity. T6P levels increase and decrease in parallel to sucrose levels and are rapidly depleted in the presence of sucrose (Smeekens et al., 2010, Sugar signals and molecular networks controlling plant growth; Lunn et al., 2014, Trehalose metabolism in plants) and it has been shown that in seedlings and young leaves T6P has an inhibitory effect on SnRK1 activity (Zhang et al., 2009, Inhibition of SNF1-related protein kinase 1 activity and regulation of metabolic pathways by trehalose-6-phosphate).

By mass spectrometry analysis, seven phosphorylated residues were then identified in the immonoprecipitated bZIP63-GFP protein. Subsequent in-gel kinase tests indicate that bZIP63 is phosphorylated by multiple protein kinases showing apparent molecular masses of 50-60 and 35-38 kDa. In Figure 2A, it should be clearly indicated that the plant extract used for in-gel kinase assays derived from hydroponic root cultures (see “Kinase assays”, third paragraph), along with the conditions used for root culture in the case of the experiment in Figure 2A.

We assume the reviewer means 3A in the original manuscript (Figure 4A in the revised manuscript). We have specified the use of roots and root cell culture in the figure legend as well as in the main text. The growth conditions are described in the Methods section under “plant growth”. We further added more detailed information on the plant material used in each of the four affinity purification/kinase identification experiments that were conducted in the excel spreadsheet containing information on which kinase was identified in which sample (first tab of Figure 4–source data 1).

To isolate the kinases detected in the in-gel assay, GST-bZIP63 was immobilized on GST-Sepharose and plant protein extracts made from root cultures or root-derived cell suspension (as the latter is maintained in the presence of sucrose, please specify which of these sample is shown in Figure 2) were subjected to affinity binding followed by NaCl elution. The eluted proteins resolved between 30 and 60 kDa were isolated from in-gel kinase PAGE (if I understood well), and identified by mass spectrometry. From 28 candidates only 7 kinases (including the SnRK1 catalytic AKIN10, AKIN11 and activating subunit SNF4, as well as casein kinase subunits CKA1, CKA2, CKB1 and CKL2) were identified by 2-9 peptides. It should be noted in the Materials and methods how the excess of His6-bZIP63 was filtrated out during the mass spectrometry analysis.

The reviewer misunderstood the use of in-gel PAGE for protein identification. Proteins were isolated from a non-substrate gel. This is shown in the scheme in Figure 4B and described in the Methods subsection “Identification of proteins and in vivo phosphorylation sites by LC-MS/MS”. As this seems not to be clear enough, we clarified the use of a normal SDS PAGE gel for MS analysis in the main text. As the reviewer pointed out correctly, using a gel with substrate would indeed strongly hamper identification of other proteins by MS/MS, if not prevent it completely.

In addition, it would be worth knowing whether the GST-alone matrix had retained any of these kinases.

Unfortunately, we never tested whether any of the kinases phosphorylating bZIP63 binds to a GST-coupled column. Affinity purification was employed in the first place to reduce the complexity of the sample and at the same time enrich for bZIP63 binding proteins before MS analysis. For kinase identification we cut out only those regions giving a signal in the in-gel kinase assay and considered only kinases with the correct molecular weight as a further restrictive step.

It is possible that one of the kinases identified by MS binds to GST and not bZIP63. However, for the “hot candidates” among the kinases we conducted additional assays (Y2H, BiFC) to confirm their interaction with bZIP63 and we tested phosphorylation of bZIP63 by these kinases. Results for SnRK1 subunits can be found (among others) in Figure 5 (including Figure 5–figure supplement 4 and 5) and Figure 6 (including Figure 6–figure supplement 1 and 2). We are therefore confident that AKIN10 and AKIN11 were not retained by GST, but by bZIP63.

Furthermore, αAKIN10 western showing an enrichment of AKIN10 on the GST-bZIP63 matrix versus GST would be more convincing.

We did not do a western blot as we did not have an antibody against AKIN10 at the time we undertook the affinity purifications (2008). However, we are routinely doing proteomics analysis in our department and SnRK1 peptides are usually not detected in total protein extracts with shotgun proteomics. As we repeatedly found a large number of peptides for AKIN10 and AKIN11 in our affinity purification samples, this is a very good indication that these kinases have been enriched.

From here on, the authors concentrate on the analysis of bZIP63 phosphorylation by the SnRK1 catalytic subunits AKIN10, which according to Baena-Gonzalez et al. and Jossier et al. (Plant J 59: 316, 2009), is at least 10-20 times more active/abundant compared to its homolog AKIN11 in Arabidopsis. Figure 3B illustrates that GST-AKIN11 purified from E. coli also phosphorylates bZIP63 less efficiently compared to GST-AKIN10 in vitro. It is strange that no autophosphorylation of AKIN10 was ever detected by the authors either in normal or in-gel kinase assays. Any explanation?

We assume the reviewer means 4B in the original manuscript (Figure 5B in the revised manuscript).

In fact, auto-phosphorylation was detected. However, it is quite weak compared to substrate phosphorylation and can therefore not be seen well on the autoradiography.

As a key argument, the authors use the akin10 GABI_579E09 T-DNA insertion mutant to demonstrate that AKIN10 is indeed responsible for in vivo phosphorylation of bZIP63. Unfortunately, according to the reviewer's knowledge, this mutant has not yet been characterized by others, so, to a certain extent, this must be performed by the authors in this paper.

Unfortunately we were not aware of the fact that this mutant had never been described before. We had of course characterized the akin10 mutant line before use. We added a detailed molecular characterization of the mutant in Figure 5–figure supplement 1. Additionally we show some data on the phenotypic appearance of the mutant and expression of selected AKIN10 target genes in the mutant in light and extended night (Figure 5–figure supplement 2).

In Figure 4A, they show that one of the phosphorylated bZIP63 bands around 55-57 kDa is missing in their in-gel kinase assay with akin10 mutant extracts, and they fail to detect AKIN10 by western with the Agrisera αAKIN10 antibody in roots, but observe a lower Mw weakly cross-reacting protein band in leaf extracts. According to these results, the GABI_579E09 insertion mutant appears to produce no immunologically detectable AKIN10. Yet, the authors could successfully cultivate the mutant, which means that it must be fertile. If AKIN10 is a central regulator of starvation responses, and shows an order of magnitude higher activity compared to AKIN11 in Arabidopsis, based on the published striking AKIN10 RNAi senescence phenotype by Bhaena-Gonzalez et al., one would expect a dramatic growth phenotype. Consequently, it is absolutely necessary to show the phenotype of the akin10 mutant both in the light and dark that we can compare that to that of the bzip63 mutant.

The reviewer confuses the plant lines described in Baena-Gonzalez et al. (2007, A central integrator of transcription networks in plant stress and energy signaling). The mutant showing the strong phenotype in the Baena-Gonzalez paper is an akin10/akin11 double mutant (VIGS of AKIN11 in AKIN10 RNAi background). They also showed akin10 and akin11 single mutants, which did not have a striking phenotype. Single mutants of AKIN10 (RNAi) were not smaller than wt and flowered normally, which is probably due to the action of AKIN11. Therefore, the lack of a strong phenotype in our akin10 T-DNA insertion line is not in contrast to the data published in Baena-Gonzalez, but rather supports the findings described there. A molecular and phenotypic characterization of the GABI-KAT akin10 line can be found in Figure 5–figure supplement 1-2.

Furthermore, the analysis of transcription of bZIP63 target genes should provide further clear evidence that post-translational regulation of bZIP63 is indeed impaired in the GABI_579E09 T-DNA insertion mutant.

Not many bZIP63 target genes are known. We tested the expression of ASN1, DIN10, and ProDH in the akin10 line after 6h of light and extended night (Figure 5–figure supplement 2C). ASN1 and ProDH showed reduced expression under both conditions and in extended night, respectively. Expression of the genes is not completely abolished, but this is not surprising considering that AKIN11 seems to have redundant functions (as can be seen from the phenotype of single and double mutants).

This is because Figure 4E does not really show a very convincing change in the Phos-tag pattern of phosphorylated bZIP63 despite a lower level of “supershift” indicated by asterisks (see Figure 4E and Figure 4–figure supplement 3).

It is true that there is no difference in bZIP63 phosphorylation in wt and akin10 background in the light and in extended night. Only the upper band is reduced (Figure 5E, left) and we can see how this may seem disappointing at first sight. However, we identified more than one kinase and it can therefore not be expected that the phosphorylation of bZIP63 vanishes completely when one kinase is missing. We do see a clear reduction in the phosphorylation of bZIP63 under extended night conditions in the akin10 mutant line, which tells us that AKIN10 phosphorylates bZIP63 only under specific conditions. The fact that only one band is vanishing and not all means that probably only one site is less phosphorylated, while others, which are targeted by other kinases, are not.

As it is possible that AKIN11 phosphorylates bZIP63 stronger in the absence of AKIN10, we now knocked down both AKIN10 and AKIN11 using VIGS and looked at the bZIP63 phosphorylation pattern. We saw that again in light there was no difference but the reduction of the “supershift” band in extended night was even stronger in these plants (Figure 5E, right). In our eyes, this confirms that both SnRK1 kinases phosphorylate bZIP63 under extended night conditions. Thus, we are fully confident that with the additional Phos-tag data, particularly those from the genomic constructs and the VIGS experiment, the effect is absolutely convincing.

Figure 4C to D show yeast 2-hybrid and Agrobacterium infiltrated N. tabacum protein-protein interaction assays indicating that that unlike AKIN11, the AKIN10 catalytic and SNF4 activating subunits of SnRK1 interact with bZIP63. Given the high level of amino acid homology between AKIN10 and AKIN11, it is surprising that bZIP63 fails to bind AKIN11 and provide a possible hint about the location of possible interaction sites. Did you map these?

No, we did not map the interaction sites. We observed a very weak interaction with AKIN11 in tobacco and no detectable interaction in yeast. This probably reflects the affinity of the kinase subunit alone to the bZIP and fits to the observed reduction in phosphorylation of bZIP63 by AKIN11. However, in vivo AKIN11 probably still plays a role in bZIP63 phosphorylation. SnRK1 forms a trimeric complex, which always includes SNF4 (Emanuelle et al., 2015, SnRK1 from Arabidopsis thaliana is an atypical AMPK). As SNF4 interacts strongly with bZIP63, it might act as a link to bring bZIP63 and AKIN11 together.

On the other hand, it is remarkable that the kinase-substrate interaction is so strong that this can be clearly detected by both assays. However, as these are heterologous assays based on overexpression of proteins, it is absolutely necessary to present co-immunoprecipitation data using the later described bZIP63 GENE construct (and not 35S ox lines) and αAKIN10 western from wt and akin10 mutant plants.

We tried co-IPs from the genomic complementation lines. However, we could not successfully pull down AKIN10. One possible explanation for this is that, as the reviewer stated above, interactions between a kinase and its substrate are mostly not very strong and often not very long lived. We can therefore expect that only a fraction of bZIP63 is bound to AKIN10 at any given time and that there is not enough bound AKIN10 to be detected in the western blot. This analysis is furthermore complicated because bZIP63 is expressed at very low levels and very hard to detect (already in plant extracts), therefore it is again not surprising that we could not detect it in a co-IP where most likely only a fraction will bind very transiently to the kinase.

According to Figure 4–figure supplement 1, the putative upstream activating kinase SnAK2 enhances the activity of AKIN10 (and may be that of AKIN11). SnAK2 is supposed to phosphorylate the T-loop of both AKIN10 and AKIN11. Yet, we do not see enhanced phosphorylation of GST-AKIN10/11 in this figure. Any explanation?

The phosphorylation of AKIN10 and AKIN11 in the presence of SnAK2 is enhanced, as can be seen in Figure 5–figure supplement 4C. However, the amount of AKIN10 and SnAK2 used in this kinase assay is much lower than the amount of bZIP63. Therefore, the phosphorylation band of AKIN10 is weak in relation to the bZIP63 band and cannot be seen very well on the gel. In Figure 5–figure supplement 4A AKIN10 and AKIN11 are always together with SnAK2, so the increased AKIN10 phosphorylation cannot be seen in this figure. Comparing phosphorylation intensities between figures is not possible.

As GST-AKIN10/11, GST-bZIP63 and GST-SnAK2 have closely similar masses, it would be necessary to resolve them in a lower % of SDS-PAGE.

The bands are close together, but still distinguishable. Only SnAK2 is very close in size to bZIP63. We provide all controls to show that the substrate, and not the kinases, is highly phosphorylated.

To better separate the bands, it would be necessary to run the gels longer, which we would like to avoid as otherwise radioactive nucleotides run into the running buffer.

In vitro phosphorylation of bZIP63 by AKIN10 and AKIN11 looks convincing in Figure 4–figure supplement 1B, so this approach should have provided a simple means to determine both AKIN10 and AKIN11 phosphorylation sites comparatively in bZIP63 by mass spectrometry. Yet, in Figure 5, a complicated intuitive combination of alanine replacements of in vivo observed seven serine phosphorylation sites is used for demonstration that GST-AKIN10 probably phosphorylates only S29, S294 and S300 residues of bZIP63.

Kinase assays with non-phosphorylatable mutants of potential kinase target sites are a standard technique to pin down the likely kinase target sites. We do not think that this strategy is complicated and unusual.

Furthermore, as was mentioned in our manuscript, it is extremely difficult to cover the total protein with MS as some areas do not produce suitable tryptic peptides. Therefore an MS based approach was not our first choice.

The referred “consensus SnRK1 sites” in Figure 5B require a corresponding reference either in the text or figure legend.

The reference was added in the figure legend.

The promoter activation assays in protoplast support the conclusion that S29 is the critical AKIN10 phosphorylation site, but in combination with the S294A and S300A amino acid exchanges, AKIN10-S29A shows lower transactivation of ASN1/DIN6-GUS and ProDH-GUS reporters. Whereas sequential phosphorylation might indeed trigger threshold-dependent conformational change of TFs, it is premature to conclude anything about possible order of phosphorylation sites (in the subsection “AKIN10 phosphorylates three conserved and functionally important serine residues in bZIP63“).

We removed speculations about the order of phosphorylation.

In the subsection “The AKIN10 target sites play an important role for bZIP63 function in planta“ it is stated that: “The phosphorylation status of the wt construct after 6h of light and extended night and in the presence or absence of sugar was similar to the one observed in ox#3 plants (Figure 6B; Figure 6–figure supplement 1E).” Concerning the reproducibility and similarity of Phos-tag gels, would it be possible to run on the same gel the reactions shown in Figure 2B and/or Figure 4E with Figure 6B, as well as Figure 4–figure supplement 3 and Figure 6–figure supplement 1E, to have an idea about the identity/mobility of phosphorylated bands in the respective reactions? Probably, it would also be beneficial to compare the Phos-tag assays on the same gel with the reactions shown in Figure 5–figure supplement 2. This would be necessary because the Phos-tag gel depicted in Figure 6B looks dramatically different for the GY11 line in L and EN compared to the 35S:bZIP63-GFP line in Figure 2B (i.e., despite the authors' statement).

Unfortunately, it is impossible to run all samples on one gel due to technical limitations, i.e. loading capacities. As the identity of the bands on the Phos-tag gels seem to be too confusing, and comparison between figures is difficult, we have first aligned all Phos-tag gels and put them in a supplementary figure for convenient comparison (Figure 3–figure supplement 1).

Further, we marked all bands and labeled them persistently with numbers from 1-8 in each of the figures according to an increase in their phosphorylation status. This is the maximum of different isoforms we could identify on different Phos-tag gels and corresponds to the nonphosphorylated form plus the seven identified in vivo phosphorylation sites from the combined MS analyses. Of course, not all isoforms are always visible on the gel as they depend on the phosphorylation level and, furthermore, the phosphorylation status is also highly dynamic, and not every phosphorylation causes a similar strong shift, which complicates the picture. Nevertheless, the main point we want to make in the manuscript – and we think this is clearly evident from the figures – is that the level of bZIP63 phosphorylation changes in response to the different treatments and is furthermore depending on AKIN10.

In the same subsection, I do not understand the logic of the following sentence: “Notably, hyper-phosphorylation of bZIP63 in the extended night was even stronger in the GY lines than in ox#3, probably due to lower bZIP63 expression.” Do the authors suggest that AKIN10 activity is limiting in terms of bZIP63 phosphorylation even if AKIN10 was supposed to be activated by extended night?

The expression of bZIP63 in the genomic line is of course lower than in the ox lines. Therefore the ratio between bZIP63 and AKIN10 is also lower in these plants, which means that there is more AKIN10 per bZIP63 in the wt situation, which corresponds to the Phos-tag gel, where more bZIP63 is shifted to the hyper-phosphorylated form.

With the limitations mentioned above concerning the necessary parameters of bzip63 mutant characterization, the lack of complementation by the triple S/A bZIP63 mutant gene is documented. As the mutant gene constructs were generated by PCR of 2+xkb genomic DNA fragments, the Materials and methods should mention that all constructs were verified by sequencing.

The constructs were of course verified by sequencing. We have added this information in the Materials and methods section.

In Figure 6G, it is rather surprising that the GY11 wt complementing line shows significantly higher induction of ASN1 and DIN10.

The line is somewhat over-expressing (see Figure 7–figure supplement 1A), which could explain the higher levels of ASN1.

In the fold-induction curve and column diagram to the right, the standard error of Ws-2 is huge at 4h of EN, thus no significant difference between wt and bzip63 mutant can be claimed.

We did the appropriate statistical tests and found significant differences. Looking at standard deviations alone is not sufficient to deduce significant differences. Also, the error bars are the standard deviations and not standard errors.

In this figure part, either the curves shown to the left should include all four samples (Ws-2, bzip63, GY11 and GAY14) or all samples at the four time points should be presented as column diagrams.

The two halves of the figure serve different purposes. We first did a time course (left) with wt and ko lines to determine the best time point for comparison of the gene expression. This was done with only two lines to reduce sample numbers. The 4h time point was chosen to test also the complementation lines.

Given the data presented in Figure 6–figure supplement 1, I wonder why did the authors use the low expressing GY11 line for comparison with GAY14.

The lower expressing line was chosen, because bZIP63 expression is more similar in these lines (Figure 7–figure supplement 1A).

From Figure 6–figure supplement 1C it is also apparent that the GAY4 line shows no expression of triple S/A mutant bZIP63-YFP, so only one bzip63 mutant transformed with the triple phosphor-site mutant of bZIP63 was analysed.

The RT-qPCR in Figure 7–figure supplement 1A shows clearly that bZIP63 is expressed in this line. The levels are just lower than in the other lines, but in fact expression in this line is most similar to the wt line. The signal in the western blot is therefore also weaker and can only be seen in the extended night sample.

Ehlert et al. reported that, when overexpressed in plant protoplasts, bZIP63 can heterodimerize with members of the S-group, and in those assays bZIP63 homodimers were barely detectable. Kang et al. (2010) reported later that bZIP63 also interacts with the Arabidopsis response regulator ARR18 involved in cytokinin signaling, whereas alone (i.e. by interacting with S-group partners in heterodimers), bZIP63 functions as negative regulator of osmotic stress responses and seed germination.

This was reported by Veerabagu et al. (2014, The Interaction of the Arabidopsis Response Regulator ARR18 with bZIP63 Mediates the Regulation of PROLINE DEHYDROGENASE Expression).

Several heterodimerization partners of bZIP63 are well characterized in terms of their transcription regulatory targets. Therefore, it is apparent that their regulatory functions might be modified by bZIP63, where heterodimerization capability of bZIP63 could be controlled post-translationally, e.g. by phosphorylation through SnRK1 and potentially other protein kinases. This idea should be introduced clearly, with some necessary references to the functions of suspected bZIP63 dimerization partners at the beginning of the manuscript and discussed in this term at the end. For example, the observed effect of bzip63 mutation on the levels of free amino acids could be related to the function of its heterodimerization partner bZIP1. So, probably citing a recent paper from Para et al. (Proc Natl Acad Sci USA. 111: 10371, 2014) may be helpful for discussion of metabolite changes observed in the bzip63 mutant.

We agree, and have introduced that reference in the Introduction and indeed re-written the entire discussion of the metabolite data and highlighted the similarities and complementary aspects with bZIP1. We are confident that in the revised version the entire story is indeed much more stringent and fits nicely together also with respect to the other published data on bZIP1.

Evidently, the bzip63 mutation could influence the function of bZIP1, i.e. if we assumed that the bZIP1 functions as heterodimer with bZIP63. It is therefore interesting that overexpression of AKIN10 in protoplasts has only a slight effect on heterodimerization of overexpressed Gal-AD-bZIP1 and Gal-BD-bZIP63. While this artificial protoplast dimerization assay in Figure 7A indicates over 60-fold enhancement of Gal-AD-bZIP11+Gal-BD-bZIP63 dimerization, at the level of GAL-UAS4:GUS reporter, about 15-fold increase is observed, which is reduced to about 2-3-fold by triple mutation of the identified S29/S294/S300 bZIP63 phoshorylation sites. In case of Figure 7B and C, I could not fully understand why the authors used two different (i.e. GAL-UAS4:GUS and GAL-UAS4:LUC) reporters in the dimerization and reporter activation assays. This correlative evidence between the dimerization and reporter activation assay should definitely be strengthened by performing similar dimerization assays with bZIP63 carrying the S29A/S294/S300 phospho-site replacements. This appears to be obligatory.

We agree that mixing the LUC and GUS system is not optimal and have therefore repeated the P2H assays of bZIP63 with bZIP1, 11, and 63 as interaction partners in the GUS system (Figure 8A).

While the general trend (addition of AKIN10 leads to stronger dimerization) remained the same, the magnitude of the change – especially for bZIP11 – is lower and fits better to the assays shown in Figure 8B. It could be that the LUC system is more sensitive than the GUS system, which led to the discrepancies in the fold change between interaction in the absence and presence of AKIN10. Moreover, we have also noticed that in the previous version we have been mistaken by the strong effect on the interaction with bZIP11. As we discuss now, bZIP11 is a strong activator of transcription and bZIP1 and 63 not, which makes of course a big difference in these assays. This has been corrected and we did also include the requested point mutations, which confirm the strong effect of S29 on the dimerization.

Although the Discussion takes care to announce that “to our knowledge this is the first time that phosphorylation outside the bZIP domain was shown to affect dimerization with different partners in a distinct way”, it is rather flat. Somewhere, it should mention that the current study did not answer the question of whether the observed phosphorylation sites would modify the cellular location (i.e. nuclear import) and stability of bZIP63, in addition to its heterodimerization, which should be anyhow confirmed by co-IP studies using native gene constructs.

We did not observe an effect of S/A mutation on the localization of bZIP63 (also not with other serines than 29, 294, and 300). It is therefore not likely that phosphorylation alters the subcellular localization of bZIP63. We have stated this now clearly in the manuscript.

Concerning DNA binding, we tested whether S/A or S/D mutation of the N- and C-terminal serines would have an effect on binding to the C-box consensus motif as it was shown for S160, 164, and 168 in the DNA binding domain (Kirchler et al., 2010, The role of phosphorylatable serine residues in the DNA-binding domain of Arabidopsis bZIP transcription factors). Again, we did not see any difference between wt and mutant bZP63 in these assays, which we also mention in the text.

Unfortunately, the majority of the Discussion includes high-flying general statements, such as, “Here we show that bZIP63 plays an important role in the energy starvation response and metabolic regulation…”. If this is so, the reader will ask whether transcription of bZIP63 and its heterodimerization partners is controlled by AKIN10 and is altered in the akin10 mutant (i.e. a necessary but missing control).

We tested whether bZIP63, bZIP1, and bZIP11 are differentially expressed in wt and akin10 plants by RT-qPCR (Figure 5–figure supplement 3). For bZIP63 and bZIP1, no significant difference between the lines could be observed, neither in the light nor in the extended night. For bZIP11 we observed a mild increase in akin10 plants. However, the difference was less than 2-fold. It is therefore unlikely that AKIN10 regulates the transcription of these genes.

Can the enhanced senescence caused by bZIP63-ox compensated by glucose or sucrose supply?

To answer this question we did dark-induced senescence experiments in seedlings on plates with and without sugar. Seedlings were first grown on plates containing 0.5% sucrose and transferred to plates containing 0 or 2% glucose, grown for some days and put in the dark for 7 days. We found that sugar can indeed rescue the DIS phenotype in the ox to some extent (Figure 1C and 1D).

Is there any evidence for enhanced ROS production in the bZIP63-ox lines or change or red-ox regulation in the bzip63 mutant?

Due to time constraints this question could not be addressed.

“We measured strong differences in proline and asparagine levels in the bZIP63 ko and ox […] like altered carbon or nitrogen assimilation or protein turnover” (in the subsection “bZIP63 is an important metabolic regulator in the starvation response”). So, how would bZIP63 control these mechanisms? Through controlling genes involved in N or C signaling or uptake?

As already mentioned before, the Discussion has been rewritten to emphasize that these higher amino acids levels do indeed fit very nicely to starvation and have also been observed in dark-induced senescence. One explanation might be the increased turnover (degradation) of proteins and their role as alternative energy source, particularly under low-carbon conditions. This is now discussed in the manuscript.

You say: “We found that the SnRK1 kinase AKIN10, which was proposed to be a central regulator of transcription in starvation response (Baena-Gonzalez and Sheen, 2008), is the major kinase responsible for the starvation-induced hyper-phosphorylation of bZIP63. Therefore, bZIP63 presents the first TF acting as direct target of SnRK1 in the transcriptional energy deprivation response” (in subsection “bZIP63 function is regulated by SnRK1-dependent phosphorylation”). Yes, but the hyper-phosphorylation of bZIP63 under extended night (called “starvation”, though it was 9 days of dark treatment) caused the appearance of only a weak band marked by asterisks on the images of Phos-tag gels, which represents the only evidence supporting this statement.

This is a misunderstanding. Phos-tag gels were done after 6h of extended night not after 9 days of darkness. The extended night treatment represents a starvation condition for the plant.

Further, although the hyper-phosphorylation of bZIP63 is sometimes not so strong in the ox plants, it is very strong in the genomic complementation lines, which likely better reflect the phosphorylation state of bZIP63 in the wt. We think that the reduced phosphorylation of this band in akin10 as well as snrk1 VIGS plants after 6h of extended night can be seen as good evidence that bZIP63 is a direct target of SnRK1.

In total, we provide evidence for this statement by:

(i) identifying AKIN10 as kinase for bZIP63 in an unbiased approach after affinity purification;

(ii) a missing band in in-gel kinase assays with extracts from the akin10 mutant;

(iii) (various) direct interaction and phosphorylation assays;

(iv) a clearly reduced appearance of the hyper-phosphorylated form in the akin10 mutant;

(v) even clearer results of this effect after VIGS of AKIN10 and AKIN11;

(vi) starvation (low-energy) related phenotypes of the bzip63 mutants, which cannot be complemented by a bZIP63 version lacking the AKIN10 target sites;

(vii) a missing shift in response to EN treatments in the S29/S294/S300 mutant.

In the same subsection, you state: “Reporter activation assays in protoplasts revealed that these (i.e., AKIN10 phosophorylation) sites are essential for AKIN10-dependent induction of ASN1 and ProDH by bZIP63”. This may be, but AKIN10 alone also activates these reporter genes, probably through other factors, which according to this paper are probably not the S-group bZIPs (i.e. as they appear not to be phosphorylated by AKIN10). It would be fair to mention this somewhere.

The activation of the reporter genes by AKIN10 and a possible explanation had been mentioned in the text (“Transformation of AKIN10 alone also led to a weak induction of the reporters, which could be explained by the action of endogenous bZIPs”).

Still in the same subsection, the authors say: “The relatively small difference in ProDH expression in plants, as compared to the protoplast assay, is probably due to redundancy within the C/S1 group bZIPs…”. Does this sentence predict that other C-group bZIPs could be SnRK1 substrates? Would it be not necessary to discuss potential conservation of observed AKIN10 sites in other bZIPs?

No, this sentence does not imply that other C-group bZIPs are SnRK1 targets (although it is possible). We rather wanted to underline that several bZIPs have been found to regulate ProDH and therefore bZIP63 might not be necessarily required for ProDH expression.

In the same part of the text, the authors state: “Our data provide compelling evidence that bZIP63, but none of the S1 group bZIPs, is a bona fide in vivo target of SnRK1 in low energy signaling.” This should clearly be demonstrated by identical Phos-tag gel patterns of bZIP63 isolated from sucrose-treated or akin10 mutant plants and that of the bZIP63 S29A/S294A/S300A protein obtained from bzip63 plus/minus sugar or the akin10 mutant.

We do not agree with this statement. We believe that bZIP63 is phosphorylated by several different kinases, which makes the phosphorylation pattern more difficult to interpret.

As analysis of the akin10 and akin10/11 mutants showed that bZIP63 is phosphorylated by more than just AKIN10, as phosphorylation was not abolished completely in these mutants (Figure 5C).

Furthermore, the S29/294/300A mutant has a much lower phosphorylation than bZIP63 in both of these mutants (Figure 7B). This indicates that one or more of the sites can be phosphorylated by another kinase than AKIN10 and AKIN11.

Sugar treatment as well, could affect the activity of other kinases than AKIN10. It can therefore not be concluded that sugar treatment and akin10 ko should have the same effect.

Taken together, we think that the phosphorylation pattern of bZIP63 is quite complex and depends on different factors (including kinases with potentially overlapping target sites) and cannot be simplified as the reviewer suggests.

Reviewer #2:

Phos-tag analysis of in vitro phosphorylated bZIP63 samples allows the attribution of each shifted band to the phosphorylation of a specific Ser residue (Figure 5–figure supplement 2). In contrast, the same analyses of in vivo samples showed variable banding patterns that are not easy to reconcile (Figure 2B, Figure 4E, Figure 4–figure supplement 3, and Figure 6B).

As the comparison of different Phos-tag gels was difficult for several reviewers, we have aligned all Phos-tag gels in one figure (Figure 3–figure supplement 1) and introduced a numbering system to enable easier comparison of band patterns between figures (as has been mentioned before).

This could be a result of phosphorylation of bZIP63 by other kinase(s) on S59, S102, S160, and S261, four sites that are not phosphorylatable by AKIN10. I would suggest the authors to determine whether the mutation of these four Ser residues to Ala will affect the function of bZIP63. If it does, the data would reveal that bZIP63 is regulated by multiple kinase pathways to carry out its function in the energy transition response. If it does not, the data would demonstrate that AKIN10-phosphorylation on S29, S294 and S300 is the major phosphorylation event that controls the activity of bZIP63.

We do believe, as the reviewer suggested, that other kinases also phosphorylate bZIP63. However, as we focused on SnRK1 regulation of bZIP63 in this manuscript, we have not analyzed the functionality of these other sites. This would be a time-consuming process and is well beyond the focus of this manuscript. We have therefore not included additional data on the non-AKIN10 target sites.

For S160 it has been shown that it can potentially reduce DNA binding of bZIP63 when phosphorylated (Kirchler et al., 2010, The role of phosphorylatable serine residues in the DNA-binding domain of Arabidopsis bZIP transcription factors), but for the other sites, no data on functionality are available.

By performing Phos-tag analysis of bZIP63-S59A/S102A/S160A/S261A, the authors can provide more convincing results about the in vivo phosphorylation of bZIP63 by AKIN10 kinase.

We feel that the relatively low gain of information for this line does not justify the large amount of work and time needed to create this line (∼ 1 year). It would be wrong to conclude that only AKIN10 would phosphorylate the remaining three sites in this mutant, which would again complicate the interpretation of the results. This is based on the fact, that AKIN10 ko leads to the vanishing of only one band, while mutation of the three sites leads to a strong reduction in phosphorylation. In fact, in the S29/294/300A mutant, only two bands remain (the non-phosphorylated form and a presumably single phosphorylated form). Therefore, the phosphorylation pattern of S29+294+300 would again be a combined effect from more than one kinase.

To further analyze the phosphorylation of bZIP63 in an akin10/akin11 background, we added Phos-tag gels from VIGS ko of these two kinases (Figure 5E). Also here, the only difference to wt plants was the vanishing of the hyper-phosphorylation in extended night. We therefore believe that the major phosphorylation of bZIP63 by AKIN10 happens in extended night and that maybe only one site is strongly phosphorylated by AKIN10 in vivo. This could be S29, as we showed with a new S294/300A mutant that S29 is phosphorylated in extended night, but not in the light (Figure 8D).

Reviewer #3:

1) Because SnRK and bZIP are well known proteins in plant biology, and because phosphorylation of bZIPs has been reported before, the true novelty in the manuscript is in my opinion the phosphorylation-dependent differential dimerization leading potentially to a change in transcriptional specificity plus activity toward distinct target genes (summarized in Figure 7D – model). Data in Figure 7A rely on a protoplast two-hybrid assay, allowing a transcriptional readout (equal protein amount in transfected cells has not been included here).

We have added a western blot to show the amounts of bZIP1, bZIP11, bZIP63, and AKIN10 in the P2H assay shown in Figure 8A (Figure 8–figure supplement 1A).

To more prominently support the key message of the manuscript will have to address:

i) Differential dimerization on protein level: conducting for example transient co-localization studies based on (multi-color) fluorescent proteins/BiFC (Waadt et al., 2008), or using biochemical means to compare phosphorylation-dependent bZPI63/bZIP63 with bZIP63/bZIP11 formation, and in case even include bZIP S29D (or other) phospho-mimic constructs.

We did pick up this suggestion by the reviewer, particularly as we noticed that the protoplast two-hybrid data do not allow a conclusion on preferred dimerization of different dimers because that assay is based on a transcriptional readout and leads to over-estimation of the effect on bZIP11 as mentioned before and also discussed in the revised manuscript. Therefore – as suggested by the reviewer – we have conducted a multi-color BiFC experiment to analyze the dimerization preferences of bZIP63 in a comparative in vivo assay (Figure 9A).

As exemplary dimerization partners for bZIP63, we have chosen bZIP1 and bZIP11. It is important to note here that of course many other bZIPs will also affect dimerization under the ‘real’ conditions in the wt. All three dimerization partners (bZIP1, bZIP11, and bZIP63) were expressed under control of the 35S promoter from the same plasmid to ensure that all transformed cells express all three partners. A scheme of the BiFC experiment is shown in Figure 9A.

Dimerization of bZIP63 with bZIP1 reconstructs the VENUS construct and leads to ‘yellow’ fluorescence. Dimerization of bZIP63 with bZIP11 reconstructs the CFP construct and leads to ‘cyan’ fluorescence. Homo-dimerization of bZIP63 or dimerization of any of the three partners with an endogenous (untagged) bZIP does not lead to a signal.

To see if there is a difference between bZIP63 dimerization with bZIP1 and 11, we co-transformed AKIN10 and compared the resulting signals obtained with the wt and the S29/294/300A version of bZIP63. This experiment allowed us to study the dimerization preference of phosphorylated and non-phosphorylated bZIP63. Accordingly, the bZIP63-11 hetero-dimer will generate a blue (cyan) signal and the bZIP63-1 hetero-dimer a yellow signal, whereas the bZIP63 homo-dimer cannot be detected. A quantification of the ratio of the VENUS/CFP signal obtained from more than 100 nuclei did clearly show that the co-transformation of AKIN10 did shift the signal towards the VENUS emission, thus showing that AKIN10 does preferentially trigger the formation of bZIP63-1 hetero-dimers in a competitive in vivo assay.

At first glance, this seems to stand in contrast to the data obtained in the protoplast two-hybrid assays, but as already pointed out, these assays cannot be compared. As we have been mistaken by not realizing this problem before, we had also to revise our model accordingly (Figure 9B). However, we would like to emphasize that the multicolor BiFC experiment is indeed a first in vivo evidence for the proposed phosphorylation triggered shift in dimerization, mediated by AKIN10. We are well aware that other factors (protein stability, expression levels etc.) will also strongly influence this process, but in our view there is currently no experimental possibility to address these points as well in one assay.

ii) Show differential target gene specificity: Figure 7 contains as marker for transcriptional activity a gene which is target to all three bZIPs (1, 11, 63). Even more convincing will be phosphorylation/heterodimerization-dependent activity toward a distinct bZIP11- or bZIP63-controlled gene, respectively.

Yes, we agree, this would indeed be a nice experiment and we would like to do that. Unfortunately, we do not know yet any target gene, which would be specific for only one of the different bZIP dimers. Microarray data are available for bZIP1 and bZIP11, but not for bZIP63. Moreover, also from these data, such genes cannot be deduced as the other different dimerization partners are still present. Maybe ChIP experiments would be a solution here, but that would clearly be beyond the scope of this study.

2) The primary identification of “in vivo” phosphorylation sites is conducted on bZIP-GFP overexpressing lines on material harvested in both light and prolonged dark. Therefore, basic constitutive bZIP phosphorylation (and the involvement of other kinases than AKIN10) cannot be excluded. Ideally, one may ask for the degree of conditional in vivo phosphorylation (in particular at S29 of endogenous bZIP) performing MRM/SRM phosphoproteomics.

To address this question we generated a genomic complementation line expressing the S294/300A version of bZIP63 in the ko background. In this line, S29 can be phosphorylated, but none of the other potential AKIN10 target sites. Due to the short time we had for the revision, this plant line is still heterozygous and cannot be used for phenotypic analysis at this point. However, we could use the line for Phos-tag gels and we here we saw indeed a striking effect: Compared to the triple mutant, that did not show any change in the phosphorylation pattern in response to starvation, the S294/S300 double mutant did still show a strong shift in the extended night. As S29 is the only AKIN10 target site left in this version, it must indeed be S29 that is phosphorylated by AKIN10 under starvation conditions. We think therefore we did add really significant novel information, and more is simply not possible in the few months we had for the revision.

https://doi.org/10.7554/eLife.05828.043

Article and author information

Author details

  1. Andrea Mair

    Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    Contribution
    AM, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.

  2. Lorenzo Pedrotti

    Pharmaceutical Biology, Julius-von-Sachs-Institute, University of Würzburg, Würzburg, Germany
    Contribution
    LP, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  3. Bernhard Wurzinger

    Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    Contribution
    BW, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  4. Dorothea Anrather

    Mass Spectrometry Facility, Max F. Perutz Laboratories, University of Vienna, Vienna, Austria
    Contribution
    DA, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  5. Andrea Simeunovic

    Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    Contribution
    AS, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  6. Christoph Weiste

    Pharmaceutical Biology, Julius-von-Sachs-Institute, University of Würzburg, Würzburg, Germany
    Contribution
    CW, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  7. Concetta Valerio

    Instituto Gulbenkian de Ciência, Oeiras, Portugal
    Contribution
    CV, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  8. Katrin Dietrich

    Pharmaceutical Biology, Julius-von-Sachs-Institute, University of Würzburg, Würzburg, Germany
    Contribution
    KD, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  9. Tobias Kirchler

    Department of Plant Physiology, Center for Plant Molecular Biology, University of Tübingen, Tübingen, Germany
    Contribution
    TK, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  10. Thomas Nägele

    Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    Contribution
    TN, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.

  11. Jesús Vicente Carbajosa

    Centro de Biotecnología y Genómica de Plantas, Universidad Politécnica de Madrid, Madrid, Spain
    Contribution
    JVC, Analysis and interpretation of data, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6332-1712

  12. Johannes Hanson

    1. Department of Molecular Plant Physiology, Utrecht University, Utrecht, Netherlands
    2. Department of Plant Physiology, Umea Plant Science Center, Umeå University, Umea, Sweden
    Contribution
    JH, Analysis and interpretation of data, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.

  13. Elena Baena-González

    Instituto Gulbenkian de Ciência, Oeiras, Portugal
    Contribution
    EB-G, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.

  14. Christina Chaban

    Department of Plant Physiology, Center for Plant Molecular Biology, University of Tübingen, Tübingen, Germany
    Contribution
    CC, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.

  15. Wolfram Weckwerth

    Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    Contribution
    WW, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.

  16. Wolfgang Dröge-Laser

    Pharmaceutical Biology, Julius-von-Sachs-Institute, University of Würzburg, Würzburg, Germany
    Contribution
    WD-L, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.

  17. Markus Teige

    1. Department of Ecogenomics and Systems Biology, University of Vienna, Vienna, Austria
    2. Department of Applied Genetics and Cell Biology, University of Natural Resources and Life Sciences, Vienna, Austria
    Contribution
    MT, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    markus.teige@univie.ac.at
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7204-1379

Funding

Austrian Science Fund (FWF) (P23435)

  • Andrea Mair
  • Bernhard Wurzinger
  • Markus Teige

European Commission (ITN MERIT GA 264474)

  • Lorenzo Pedrotti
  • Andrea Simeunovic
  • Thomas Nägele

Deutsche Forschungsgemeinschaft (HA2146/8-2)

  • Tobias Kirchler

Fundação para a Ciência e a Tecnologia (Portuguese Science and Technology Foundation) (PTDC/BIA-PLA/3937/2012)

  • Concetta Valerio

Fundação para a Ciência e a Tecnologia (Portuguese Science and Technology Foundation) (UID/Multi/0455½013)

  • Elena Baena-González

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Lena Fragner (University of Vienna) for help with the metabolomics measurements and Klaus Harter (University of Tübingen) and Sjef Smeekens (Utrecht University) for critical comments on the manuscript. We further thank the Core Facility Cell Imaging and Ultrastructure Research (CIUS, University of Vienna) for providing access and support to the Leica confocal laser scanning microscope. This work was funded by the FP7 Marie Curie ITN MERIT (GA 264474) (LP, AS, TN), the Austrian Science Fund (FWF) projects AP19825 (AM, BW, MT) and AP23435 (AM, MT), the Deutsche Forschungsgemeinschaft (DFG) project HA2146/8-2 (CC), and the Fundação para a Ciência e a Tecnologia projects PTDC/BIA-PLA/3937/2012 and UID/Multi/04551/2013 (EBG).

Reviewing Editor

  1. Thorsten Nürnberger, University of Tübingen, Germany

Publication history

  1. Received: November 30, 2014
  2. Accepted: August 10, 2015
  3. Accepted Manuscript published: August 11, 2015 (version 1)
  4. Accepted Manuscript updated: August 18, 2015 (version 2)
  5. Version of Record published: September 3, 2015 (version 3)

Copyright

© 2015, Mair et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 5,863
    Page views
  • 1,744
    Downloads
  • 158
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Andrea Mair
  2. Lorenzo Pedrotti
  3. Bernhard Wurzinger
  4. Dorothea Anrather
  5. Andrea Simeunovic
  6. Christoph Weiste
  7. Concetta Valerio
  8. Katrin Dietrich
  9. Tobias Kirchler
  10. Thomas Nägele
  11. Jesús Vicente Carbajosa
  12. Johannes Hanson
  13. Elena Baena-González
  14. Christina Chaban
  15. Wolfram Weckwerth
  16. Wolfgang Dröge-Laser
  17. Markus Teige
(2015)
SnRK1-triggered switch of bZIP63 dimerization mediates the low-energy response in plants
eLife 4:e05828.
https://doi.org/10.7554/eLife.05828

Categories and tags

Research organism

Further reading

    1. Cell Biology

    Intermediate filament network perturbation in the C. elegans intestine causes systemic dysfunctions

    Florian Geisler, Sanne Remmelzwaal ... Rudolf E Leube
    Research Article

    Intermediate filaments (IFs) are major components of the metazoan cytoskeleton. A long-standing debate concerns the question whether IF network organization only reflects or also determines cell and tissue function. Using C. elegans, we have recently described mutants of the MAPK SMA-5 which perturb the organization of the intestinal IF cytoskeleton resulting in luminal widening and cytoplasmic invaginations. Besides these structural phenotypes, systemic dysfunctions were also observed. We now identify the IF polypeptide IFB-2 as a highly efficient suppressor of both the structural and functional deficiencies of sma-5 animals, by removing the aberrant IF network. Mechanistically, perturbed IF network morphogenesis is linked to hyperphosphorylation of multiple sites throughout the entire IFB-2 molecule. The rescuing capability is IF isotype-specific and not restricted to SMA-5 mutants but extends to mutants that disrupt the function of the cytoskeletal linker IFO-1 and the IF-associated protein BBLN1. The findings provide strong evidence for adverse consequences of the deranged IF networks with implications for diseases that are characterized by altered IF network organization.

    1. Cell Biology

    Annexin A6 mediates calcium-dependent exosome secretion during plasma membrane repair

    Justin Krish Williams, Jordan Matthew Ngo ... Randy Schekman
    Research Article Updated

    Exosomes are an extracellular vesicle (EV) subtype that is secreted upon the fusion of multivesicular bodies (MVBs) with the plasma membrane. Exosomes may participate in intercellular communication and have utility as disease biomarkers; however, little is known regarding the physiological stimuli that induce their secretion. Ca2+ influx promotes exosome secretion, raising the possibility that exosomes are secreted during the Ca2+-dependent plasma membrane repair of tissues damaged by mechanical stress in vivo. To determine whether exosomes are secreted upon plasma membrane damage, we developed sensitive assays to measure exosome secretion in intact and permeabilized cells. Our results suggest that exosome secretion is coupled to Ca2+-dependent plasma membrane repair. We find that annexin A6 (ANXA6), a well-known plasma membrane repair protein, is recruited to MVBs in the presence of Ca2+ and required for Ca2+-dependent exosome secretion, both in intact and in permeabilized cells. ANXA6 depletion stalls MVBs at the cell periphery, and ANXA6 truncations localize to different membranes, suggesting that ANXA6 may serve to tether MVBs to the plasma membrane. We find that cells secrete exosomes and other EVs upon plasma membrane damage and propose that repair-induced secretion may contribute to the pool of EVs present within biological fluids.

    1. Cell Biology
    2. Neuroscience

    Interaction between Teneurin-2 and microtubules via EB proteins provides a platform for GABAA receptor exocytosis

    Sotaro Ichinose, Yoshihiro Susuki ... Hirohide Iwasaki
    Research Article

    Neurons form dense neural circuits by connecting to each other via synapses and exchange information through synaptic receptors to sustain brain activities. Excitatory postsynapses form and mature on spines composed predominantly of actin, while inhibitory synapses are formed directly on the shafts of dendrites where both actin and microtubules (MTs) are present. Thus, it is the accumulation of specific proteins that characterizes inhibitory synapses. In this study, we explored the mechanisms that enable efficient protein accumulation at inhibitory postsynapse. We found that some inhibitory synapses function to recruit the plus end of MTs. One of the synaptic organizers, Teneurin-2 (TEN2), tends to localize to such MT-rich synapses and recruits MTs to inhibitory postsynapses via interaction with MT plus-end tracking proteins EBs. This recruitment mechanism provides a platform for the exocytosis of GABAA receptors. These regulatory mechanisms could lead to a better understanding of the pathogenesis of disorders such as schizophrenia and autism, which are caused by excitatory/inhibitory (E/I) imbalances during synaptogenesis.