The SnRK1 Kinase as Central Mediator of Energy Signaling between Different Organelles^,

Abstract

SnRK1 is a central integrator of energy signaling in different subcellular locations with emerging roles in organellar and hormone metabolism.

The evolutionarily conserved SNF1-related protein kinase 1 (SnRK1) kinase complex is a key regulator in adjusting cellular metabolism during starvation, stress conditions, and growth-promoting conditions. Over the last two decades, extensive genetic evidence for a widespread SnRK1 signaling network has accumulated. It is now well established that SnRK1 is a central integrator of energy signaling. However, little is known about the connections between the cytoplasmic and nuclear-localized SnRK1 and plastids and mitochondria as the main energy-producing compartments in the cell. Here, we review recent findings indicating how SnRK1 affects metabolic adaptation, including plastidial and mitochondrial functions. Special emphasis is put on identified direct targets of SnRK1, which would eventually enable cross talk with organelles. In this context, a number of transcription factors (TFs) are emerging as mediators of SnRK1 signaling, potentially linking SnRK1 activity to organellar functions. Furthermore, many SnRK1 targets act in various hormonal signaling pathways, which are at least partly localized in plastids. With this review, we summarize the current knowledge on SnRK1 organelle interaction and provide ideas on the potential molecular mechanisms governing these interactions.

METABOLIC REPROGRAMMING BY SNRK1 KINASE ACTIVITY UNDER DIFFERENT GROWTH AND STRESS CONDITIONS

Already under optimal growth conditions, plants need to continuously adjust their metabolic balance between autotrophic growth based on photosynthesis during the day and respiration during the night. At daytime, sugars and other metabolites are produced by photosynthesis in chloroplasts and further distributed to be used in other metabolic pathways at different cellular compartments or transported to nonphotosynthetic sink tissues. During the night, starch and sugars supply carbon equivalents for respiration, providing the necessary energy for further growth and metabolic activities. Additionally, metabolic reprogramming is required for plants to adjust their metabolism to diverse biotic and abiotic environmental stimuli. This often leads to a stop of plant growth involving a reduction in ribosomal protein synthesis, and in parallel, accumulation of protective metabolites or defense compounds. This switching of cellular energy metabolism is mediated by the activity of the evolutionarily conserved AMPK/SNF1/SnRK1 kinase complex (Box 1; Crozet et al., 2014; Cutler et al., 2010; Hrabak et al., 2003; Kudla et al., 2010; Baena-González and Sheen, 2008; Broeckx et al., 2016). The heterotrimeric SnRK1 is the plant ortholog of the yeast SNF1 (sucrose nonfermenting 1) kinase and the mammalian AMPK (AMP-activated protein kinase). In Arabidopsis (Arabidopsis thaliana), AKIN10 and AKIN11, the kinase subunits of the SnRK1 complex, were found to regulate the expression of more than 600 target genes in response to starvation or nutrient signals in protoplasts (Baena-González et al., 2007). However, being a protein kinase, direct targets of SnRK1 need to be measured by phosphoproteomics. To this end, we applied a quantitative phosphoproteomic approach combined with proteomics and metabolomics to identify in vivo targets of SnRK1 and to uncover the related metabolic reprograming using SnRK1 mutants under energy starvation (Nukarinen et al., 2016). This study revealed hundreds of changed phosphoproteins and a very pronounced SnRK1-dependent reprogramming of metabolism including sugars.

Besides being the universal fuel of life, sugars also act as important developmental signals. Accordingly, all biological systems have evolved homeostatic mechanisms to regulate their sugar levels. For example, the level of Glc functions as an ancient and conserved regulatory signal controlling gene expression and primary and secondary metabolism, as well as growth and development (Sheen, 2014; Wingler, 2018). Plants have three known Glc-modulated master regulators: HEXOKINASE1, a direct Glc sensor (Moore et al., 2003; Li and Sheen, 2016); the energy-sensing protein kinase SnRK1, which is inhibited by sugars (Baena-González and Sheen, 2008); and the TARGET OF RAPAMYCIN kinase, which is activated by Glc (Xiong et al., 2013). These major regulators of energy metabolism are evolutionarily highly conserved, and the protokinases can be found in all three domains of life (Roustan et al., 2016). In response to energy deficits, the AMPK/SNF1/SnRK1 kinases restore energy homeostasis by switching on ATP-producing catabolic pathways (such as glycolysis and fatty acid oxidation), while in parallel switching off biosynthetic and other nonessential ATP-consuming metabolic and growth processes. SnRK1 plays a central role in the regulation of starch metabolism. In the moss Physcomitrella patens, it was observed that a SNF1a/b double knockout revealed reduced starch mobilization in response to darkness (Thelander et al., 2004). Similarly, Baena-González et al. (2007) found that virus-induced gene silencing of AKIN10/11 impaired starch mobilization also in Arabidopsis during the night. However, it is important to note that SnRK1 can have quite different roles in source and sink tissues. For example, opposite effects have been observed in starch metabolism where overexpression of AKIN10 in Arabidopsis leaves led to reduced Glc-induced starch accumulation in seedlings (Jossier et al., 2009), while SnRK1 overexpression increased starch accumulation in potato (Solanum tuberosum) tubers (McKibbin et al., 2006). A first mechanistic insight into AKIN10-dependent starch mobilization came from the observation that a luciferase reporter gene was specifically activated by AKIN10 coexpression via the α-amylase Amy3-SRC promoter (Lin et al., 2014). Furthermore the tissue-specific differential expression of SnRK1-interacting negative (SKIN) regulator proteins offers a way to explain the seemingly contradictory influence of SnRK1 on starch metabolism in different tissues (Lin et al., 2014). However, in conclusion, it becomes evident that not all modulators of SnRK1 activity under different conditions in different tissues have been identified. Overall, it is clear that SnRK1 can be activated by very diverse abiotic and biotic stress conditions that directly or indirectly cause an energy deficit by affecting photosynthesis, respiration, or carbon allocation, and the activity of the SnRK1 kinase complex is repressed by sugars (Baena-González et al., 2007). Still, the exact mechanisms of this inhibition as well as the identity of other inhibitors remain unclear (Emanuelle et al., 2016). The kinase subunits AKIN10 and AKIN11 are found in the nucleus and the cytosol, and additionally, a chloroplast localization of AKIN10 has been reported (Fragoso et al., 2009) but could not be confirmed (Bayer et al., 2012). Nevertheless, the regulation of photosynthesis is intimately linked to energy metabolism, and accordingly, functional links between AKIN10/11 activity and chloroplast functions have been reported. For example, AKIN10 is activated by 3-(3,4-dichlorophenyl)-1,1-dimethylurea treatment (Baena González et al., 2007), which blocks electron transport at PSII, thus leading to energy deprivation by inhibiting photosynthesis. Moreover, many photosynthetic genes were found to be regulated transcriptionally depending on SnRK1 activity mediated by trehalose-6-P (Zhang et al., 2009). The role of sugar signaling and SnRK1 activity is discussed in detail in another Update review of this issue (Wingler, 2017).

SNRK1 IN METABOLIC AND HORMONE SIGNALING UNDER STRESS CONDITIONS

SnRK1 activation also is triggered by hypoxia, leading to suppression of mitochondrial aerobic respiration resulting in cellular energy deprivation (Im et al., 2014; Cho et al., 2016). Flooding stress or hypoxia leads to a reduced availability of O₂ as the final electron acceptor in the mitochondrial electron transport chain (mETC), causing a rapid decrease in the cellular ATP:ADP ratio (Bailey-Serres and Voesenek, 2008). Cells cope with this energy crisis by relying primarily on glycolysis and fermentation to generate ATP and regenerate NAD⁺, respectively. To overcome this critical energy limitation, plants react to hypoxia and flooding stress by limiting protein synthesis and maintaining translation of a subset of cellular mRNAs, many of which encode enzymes involved in anaerobic metabolism and reactive oxygen species (ROS) scavenging as well as the mobilization of storage compounds such as starch. This also includes amino acid and nitrogen metabolism, for example, at the level of nitrate reductase (NR) activity or amino acids fueling the citric acid cycle. Such conditions could even lead to ATP import by mitochondria, mediated by Ca²⁺-regulated importers (Stael et al., 2011). Following the submergence-induced hypoxia, the subsequent exposure to oxygen during reoxygenation produces a burst of ROS. Recently, it was shown that ROS inhibit SnRK1 kinase activity, which would be consistent with a termination of hypoxia-induced SnRK1 signaling at the beginning of the reoxygenation phase (Wurzinger et al., 2017). This reoxygenation phase also involves a rapid accumulation of jasmonates (JAs) and increased transcript levels of JA biosynthesis genes. Mutants deficient in JA biosynthesis and signaling were found to be sensitive to reoxygenation, which was linked to the function of the basic helix-loop-helix TF MYC2, a key regulator of JA signaling (Yuan et al., 2017). MYC2 overexpression enhanced the tolerance to posthypoxic stress, and myc2 knockout mutants showed increased sensitivity to reoxygenation. In a previous study, it was shown that AKIN10 overexpression resulted in decreased MYC2 protein levels, which could be counteracted by addition of proteasome inhibitor MG132. The authors demonstrated that AKIN10 and MYC2 physically interact, and identified the Ser that when phosphorylated is crucial for proteasomal degradation of MYC2 (Im et al., 2014).

Plants activate ethylene signaling in response to flooding stress to initiate enhanced hypocotyl growth leading to the emergence of shoots (Bailey-Serres and Voesenek, 2008; Sasidharan and Voesenek, 2015). A comparison of transcriptional responses to darkness in air and under submerged conditions revealed a common early transcriptional and posttranscriptional response signature that was conserved primarily across Arabidopsis genotypes (van Veen et al., 2016). The common set of down-regulated genes (shared between negative clusters of ethylene and shade) was highly enriched in photosynthesis-related proteins, thus indicating a transient down-regulation of photosynthesis during enhanced growth to induce accelerated shoot elongation to bring leaf tips from the water layer into the air (Sasidharan and Voesenek, 2015). Overall, this low-energy escape syndrome shows a remarkable overlap with shade avoidance responses (Das et al., 2016). Notably, both ethylene signaling and SnRK1 kinase activity were reported to have a role in flooding-induced hypoxia tolerance in plants. A functional link between ethylene signaling, photosynthesis, and AKIN10 activity was reported recently (Kim et al., 2017a). It was shown that PSII activity is involved in the regulation of ethylene-inducible Arabidopsis hypocotyl growth in the light. A lack of ethylene responsiveness in etr1 mutants causes PSII inefficiency, leading to cellular energy deprivation and activation of AKIN10 expression. This, in turn, suppresses ethylene-inducible hypocotyl growth in the light.

The important regulatory function of SnRK1 for sugar and nitrate metabolism and its connection to abscisic acid (ABA) signaling has been shown in a number of biochemical analyses of SnRK1 mutants (Jossier et al., 2009). Mechanistically, this connection was further underpinned by the identification of two clade A type 2C protein phosphatases, both established repressors of the ABA hormonal pathway, as interactors and negative regulators of the SnRK1 catalytic subunit causing its dephosphorylation (Rodrigues et al., 2013). However, the ABA core signaling pathway is mainly regulated by the activation of SnRK2 kinases resulting in rapid responses, including the regulation of ion channels resulting in stomatal closure, and gene expression via AREB TFs to modulate plant growth to respond to the stress condition (Fujita et al., 2013; Munemasa et al., 2015). Under osmotic stress conditions, ABA also controls the activity of the starch amylases BAM1 and AMY3 in leaves through the AREB/ABF-SnRK2 kinase signaling pathway (Thalmann et al., 2016). Moreover, SnRK1 also has been identified as regulator of the MYC2 TF in the regulation of ABA-responsive elements (Im et al., 2014), and a further functional link between SnRK1 and ABA signaling was provided in a study of Suc responses in apple (Malus domestica), where the overexpression of the C2 domain ABA Insensitive Protein1 (MdCAIP1) was found to cause insensitivity toward ABA (Liu et al., 2017b). This phenotype was abolished in the presence of active MdSnRK1, which was shown to directly phosphorylate MdCAIP1 and thereby render it a target for 26S proteasomal degradation. Another link to JA signaling was provided by the report that MdSnRK1.1 interacted with the MdJAZ18 protein, which acts as a repressor of JA signaling. MdSnRK1.1 phosphorylated MdJAZ18 to facilitate its 26S proteasome-mediated degradation (Liu et al., 2017a). Under conditions of Suc oversupply and without JA, MdJAZ18 binds to the TF MdbHLH3, rendering it inactive toward anthocyanin biosynthesis gene activation. In their study, the authors describe a direct phosphorylation of MdJAZ18 by MdSnRK1, which subsequently leads to degradation of MdJAZ18 by the 26S proteasome (Liu et al., 2017a). All together, these data nicely illustrate the intense cross talk among the ethylene, JA, and ABA signaling pathways and indicate that SnRK1 acts as one key regulator in these interconnections.

SNRK1 AS REGULATOR OF DEVELOPMENTAL TRANSITIONS

Light not only drives photosynthesis and thereby energy metabolism, but also serves as an important signal for many developmental switches during the life cycle of a plant. At the end of the life cycle, plants undergo senescence, which is a coordinated redistribution of resources such as nitrogen from the source (i.e. chloroplasts) to sink tissues such as seeds. Ethylene signaling is a key regulator of senescence and involves a set of TFs including ETHYLENE INSENSITIVE3 (EIN3). AKIN10 directly interacts, phosphorylates, and antagonistically modulates EIN3 activity. Consistently, 3-(3,4-dichlorophenyl)-1,1-dimethylurea treatment of plants slows down senescence progression through destabilization of EIN3 in Arabidopsis (Kim et al., 2017b). Plant leaf senescence involves autophagy (Liu and Bassham, 2012), and recently, AKIN10 was identified as a positive regulator of plant autophagy. Transgenic Arabidopsis lines overexpressing AKIN10 show delayed leaf senescence and increased tolerance to nutrient starvation and abiotic stresses, and a functional autophagy pathway was found to require AKIN10 activity (Baena-González et al., 2007; Chen et al., 2017; Soto-Burgos and Bassham, 2017). Moreover, phosphorylation of ATG1 (AUTOPHAGY RELATED GENE1), the plant ortholog of mammalian ULK1, was enhanced when AKIN10 was overexpressed, indicating that the mechanism of autophagy activation is conserved between plants and animals (Chen et al., 2017). This is consistent with the results that similarly to overexpression of AKIN10 and reduced expression of AKIN10 and AKIN11, decreased and increased trehalose-6-P content also delays and accelerates senescence, respectively (Baena-González et al., 2007; Wingler et al., 2012).

Another level of complexity to these regulatory circuits is added by the circadian regulation of gene expression, which is evident for the nuclear-encoded photosynthesis genes such as chlorophyll-binding proteins (Harmer et al., 2000). Sugar is the major outcome of photosynthesis, and recently, a link between ethylene and sugar signaling and the circadian clock was established by the finding that ethylene shortens the circadian period, depending on Suc and the circadian clock evening element GIGANTEA (GI; Haydon et al., 2017). These findings reveal that Suc affects the stability of circadian oscillator proteins and can mask the effects of ethylene on the circadian system identifying novel molecular pathways for input of sugar to the Arabidopsis circadian network. A functional link between the circadian clock and AKIN10 was found by the observation that elevated AKIN10 expression delayed the peak expression of GI under diurnal conditions and lengthened the clock period specifically under light conditions (Shin et al., 2017). Previously, it also was shown that inhibition of photosynthesis and thereby of the endogenous oscillations in sugar levels mediates a metabolic feedback to the clock through the morning-expressed gene PSEUDO-RESPONSE REGULATOR7 (PRR7), and that prr7 mutants are insensitive to the effects of Suc on the circadian period (Haydon et al., 2013).

Sugar metabolism and the allocation of sugars between tissues are intimately associated with flowering transition in plants. A functional link between the regulation of flowering time, sugar levels, and SnRK1 activity became evident from the observations that overexpression of AKIN10 delayed flowering (Baena-González et al., 2007) and the finding that a loss of TPS1 (trehalose-6-P synthase 1) causes extremely late flowering in Arabidopsis, even under otherwise inductive environmental conditions (Wahl et al., 2013). More recently, a direct link between SnRK1 and the regulation of flowering time was shown in a study reporting that AKIN10 interacts with the INDETERMINATE DOMAIN (IDD)-containing TF IDD8 in the nucleus and phosphorylates IDD8 primarily at two Ser residues, Ser-178 and Ser-182 (Jeong et al., 2015). IDD8 regulates flowering time by modulating sugar metabolism and transport under sugar-limiting conditions in Arabidopsis. The phosphorylated Ser residues in IDD8 reside in the fourth zinc finger domain that mediates DNA binding and protein-protein interactions. The AKIN10-mediated phosphorylation of IDD8 reduced its transcriptional activation ability, but did not affect the subcellular localization and DNA-binding property of IDD8. Similarly, the phosphorylation of the bZIP TF bZIP63 by AKIN10 at two evolutionarily conserved Ser residues was found to regulate its (hetero)dimerization status and thereby target gene specificity upon low-energy stress (Mair et al., 2015), and a phosphorylation of bZIP63 in its DNA-binding domain also was shown to impair its DNA-binding ability (Kirchler et al., 2010).

METABOLIC SIGNALS VIA REGULATION OF ENZYMES AS TARGETS OF SNRK1

As outlined before, the switching of cellular metabolism is a hallmark of SnRK1 signaling and has been clearly demonstrated in vivo by a recent study on SnRK1 mutants under energy starvation combining quantitative phosphoproteomics and metabolomics (Nukarinen et al., 2016). Such a reprogramming of cellular (energy) metabolism by SnRK1 can, in principle, be achieved by two different modes of action: (1) by a direct phosphorylation of metabolic enzymes, resulting in either an altered enzymatic activity or protein stability; or (2) by changing transcript levels of key enzymes for metabolic pathways. In fact, various examples for both possibilities have been reported as recently reviewed by Broeckx et al. (2016). Key metabolic enzymes in the cytosol, such as Suc-P synthase, NR, TPS5, and HMG-CoA reductase, have been identified as direct targets of SnRK1 in vitro (Jossier et al., 2009; Robertlee et al., 2017) and in vivo (Nukarinen et al., 2016). Quantitative phosphoproteomics revealed altered phosphorylation levels of SnRK1 target sites in planta using an artificial miRNA knockdown approach to silence both AKIN10 and AKIN11 (Nukarinen et al., 2016). It should be noted here that those metabolic enzymes also are targeted by other kinases, such as CDPKs or MAP kinases, and therefore seem to present a central hub for regulation and potential cross talk of different signaling pathways (Huber et al., 2002; Chardin et al., 2017). Additionally, the use of untargeted quantitative phosphoproteomics revealed the unexpected identification of several differentially phosphorylated sites in organellar proteins, most prominently the chloroplast, which was at first sight surprising (Table I). However, considering that chloroplasts and mitochondria are the key organelles of energy metabolism in plant cells, such a link comes not completely unexpected. Still, it immediately raises the question of how this connection is functioning at a mechanistic level. As the AKIN10/11 kinases are not localized within these organelles, this connection must be mediated by other signals instead. Of course, sugars or intermediates of energy metabolism would be prime candidates for such a connection. In fact, many changes in metabolite levels have been reported in the context of either SnRK1 function or downstream targets, such as bZIP TFs. The bZIP TF bZIP63 was identified recently as a direct in planta target of SnRK1 under energy deprivation (Mair et al., 2015) and also was found to be differentially phosphorylated in SnRK1 mutants (Nukarinen et al., 2016). bZIP63 is a member of the group C bZIP TFs, which form heterodimers with group S bZIP TFs such as bZIP1, bZIP11, or bZIP53. These bZIP TFs are known for their role in sugar signaling and amino acid metabolism under low-energy stress, salt stress, or during dark-induced senescence (Hanson et al., 2008; Alonso et al., 2009; Weltmeier et al., 2009; Dietrich et al., 2011; Hartmann et al., 2015). In these studies, it was commonly observed that the level of primary carbohydrates, Pro, and branched-chain amino acids (BCAAs) was strongly regulated by different combinations of bZIP TFs. For example, a bzip1 bzip53 double mutant was found to be affected in the coordination of BCAA catabolism under salt stress conditions, indicating a central function of these bZIPs in amino acid breakdown (Hartmann et al., 2015). Also, mutants defective in the bZIP63 TF showed increased levels of many proteinogenic amino acids, particularly under conditions of energy deprivation (Mair et al., 2015). Simultaneously, concentration of the TCA cycle intermediates citrate, 2-oxoglutarate, and malate were found to be significantly increased in bzip63 plants. In addition to effects on amino acid degradation, these findings point to a role of bZIP63 in coordinating subcellular metabolism. The malate shuttle across the chloroplast envelope is essential for the transport of 2-oxoglutarate and Glu between cytosol and chloroplast (Facchinelli and Weber, 2011), and thereby automatically also affects mitochondrial metabolism. Hence, it is interesting to speculate that SnRK1 coordinates the subcellular C/N balance not only by phosphorylating NR, Suc-P synthase, or F2KP, but also via transcriptional control mediated by bZIP63 leading to allocation of organic and amino acids between multiple compartments.

Table I. Mitochondrial and plastid proteins differentially phosphorylated in an AKIN10-dependent manner.

Organelle	Gene Identifier	Annotation
Plastid	AT3G53460a	CP29, chloroplast RNA-binding protein 29
	AT5G35630a	GS2, Gln synthetase 2
	AT3G46780a	PTAC16, plastid transcriptionally active 16
	AT3G03710a	PNP1, polyribonucleotide nucleotidyltransferase
	AT4G15530a	PPDK, pyruvate orthophosphate dikinase
	AT1G67090a	RBCS1A, ribulose bisphosphate carboxylase small chain 1A
	AT2G47450a	CAO, chloroplast signal recognition particle component
	AT4G13670a	PTAC5, plastid transcriptionally active 5
	AT1G68830a	STN7, STT7 homolog STN7
	AT4G02510a	TOC159, translocon at the outer envelope membrane of chloroplasts 159
	AT3G08940b	LHCB4.2 (CP29.2), light-harvesting complex photosystem II
	AT2G46820b	PSAP, PTAC8 photosystem I P subunit
	AT2G35980b	NHL10, NDR1/HIN1-like protein 10
	AT3G02150b	PTF1, plastid transcription factor 1
	AT4G18480b	CHLI1, magnesium-chelatase subunit ChlI-1
	AT4G22890b	PGRL1A, PGR5-like protein 1A
	AT5G20720b	CPN20, 20-kD chaperonin
Mitochondria	AT5G66760a	SDH1-1, succinate dehydrogenase 1-1
	AT5G26860a,b	LON1, ion protease 1
	AT4G27585a	SLP1, SPFH/Band 7/PHB domain-containing membrane-associated protein family
	AT3G25140a	GAUT8, galacturonosyltransferase 8
	AT5G14780b	FDH, FDH formate dehydrogenase

^aProteins found to be significantly (P < 5%) differentially phosphorylated in leaves between AKIN10 knockout, wild-type, and overexpressor plants after 2 h of extended night treatment (Nukarinen et al., 2016).

^bProteins found to be significantly (P < 5%) differentially phosphorylated in leaves between AKIN10 wild-type and knockdown plants at one or more time points within a time series along the transition into extended night (Nukarinen et al., 2016).

During energy-limiting conditions such as extended darkness, alternative substrates are required to fuel mitochondrial respiration. This can be achieved by oxidation of amino acids, which can then either be further oxidized in the tricarboxylic acid cycle or be used to generate electrons that can be directly transferred to the mETC via the electron transfer flavoprotein/ubiquinone oxidoreductase (ETF/ETFQO) system (Araújo et al., 2010). Particularly, the breakdown of BCAAs has recently gained more attention, leading to the conclusion that ETF/ETFQO is an essential pathway to donate electrons to the mETC and that amino acids are alternative substrates to maintain respiration under carbohydrate starvation (Cavalcanti et al., 2017). A functional link to SnRK1 via bZIP TFs has been suggested based on the observation that different bZIP TFs of the groups C and S were found to regulate amino acid breakdown, particularly Pro and BCAAs (Hanson et al., 2008; Dietrich et al., 2011; Hartmann et al., 2015). ETFQO can directly fuel electrons to the mETC and has recently been identified as direct target gene regulated by bZIP63 (Pedrotti et al., 2018), thereby providing the first functional link between SnRK1 and mitochondrial energy metabolism (Fig. 1).

Figure 1.

Different modes of metabolic regulation by SnRK1. SnRK1 regulates cytosolic metabolism directly by phosphorylation of metabolic enzymes and also can indirectly affect organellar metabolism via phosphorylation of TFs in the nucleus that regulate expression of nuclear-encoded organellar proteins. This, in turn, will affect the proteomic state within the organelles. The phosphorylation status of several plastid and mitochondria proteins is altered by yet unknown mechanisms in response to SnRK1 activity changes. TFs directly phosphorylated by SnRK1 are predicted to bind to nuclear-encoded plastid and mitochondria genes, eventually altering their transcription. The thereby altered metabolic state might be directly transduced into plastids and mitochondria, or indirectly via transcriptional regulation.

ANTEROGRADE SIGNALS TO ORGANELLES VIA PHOSPHORYLATION OF TRANSCRIPTION FACTORS BY SNRK1?

Currently, increasing evidence is accumulating for communication networks between organelles that are regulated in part by retrograde (= organelle to nucleus) signals (Chan et al., 2016; Kmiecik et al., 2016; de Souza et al., 2017). In contrast, anterograde (= nucleus to organelle) signals mainly regulate biogenesis of the organelles as well as maintenance, for example, in repair processes during stress situations. This topic is further discussed in another review of this special issue focusing on mitochondrial retrograde signals (Wagner et al., 2018). The finding that bZIP63 is a direct regulator of ETFQO and thereby affects mitochondrial electron transport (Pedrotti et al., 2018) and the identification of several differentially phosphorylated proteins in chloroplasts and mitochondria in AKIN10/11 mutants (Nukarinen et al., 2016) prompted us to screen further for target genes of those TFs, which have been identified as SnRK1 targets. In addition to the already mentioned EIN3, MYC2, IDD8, and bZIP63, we also included here Wrinkled1 (WRI1), a member of the plant-specific APETALA2 family, and the B3 domain TF FUSCA3 (FUS3), for both of which a direct interaction with AKIN10 has recently been shown. WRI1 was found to be a central regulator of oil synthesis and flowering time, and in Brassica, BnWRI1 accelerated flowering and enhanced oil accumulation in both seeds and leaves without leading to visible growth inhibition (Li et al., 2015). WRI1 also is a positive regulator of glycolysis and lipid biosynthesis in Arabidopsis, and recently, phosphorylation of WRI1 by AKIN10 was shown to result in its proteasomal degradation (Zhai et al., 2017). This AKIN10-dependent degradation of WRI1 provides a homeostatic mechanism that favors lipid biosynthesis when intracellular sugar levels are elevated and AKIN10 is inhibited. FUS3, a member of the AFL (ABI3/FUS3/LEC2) subfamily of B3 TFs, is a master regulator of seed maturation and hormonal responses during late embryogenesis and germination (Carbonero et al., 2017). AKIN10 was identified as a FUS3-interacting protein, and it was reported that AKIN10 physically interacts with and phosphorylates FUS3 at its N-terminal region, which delays degradation of FUS3 (Tsai and Gazzarrini, 2012). Through their interaction, SnRK1 and FUS3 were shown to work synergistically in ABA signaling, thereby influencing developmental phase transition and lateral organ growth.

To screen the Arabidopsis genome for binding sites in the promoter region of target genes of those TFs, which are directly targeted by SnRK1, we used the AthaMap Web tool (www.athamap.de; Hehl et al., 2016). To specifically assess the extent of how much SnRK1 anterograde signaling could affect plastidial and mitochondrial proteins, we focused on nuclear-encoded targets of those TFs with a known or clearly predicted localization in plastids or mitochondria according to SUBA 4.0 (http://suba.live; Hooper et al., 2017). This analysis uncovered a remarkable list of nuclear-encoded organellar proteins with important functions in mitochondria and chloroplasts (Supplemental Table S1). These potential targets of the SnRK1-targeted TFs include a number of well-known key regulators for organellar protein import and biogenesis as well as organellar maintenance and repair (ALB3, TIM, TIC, and TOC proteins), or the FtsH proteases, which are required for PSII repair. Moreover, a number of nuclear-encoded subunits of PSI and PSII and components of the light-harvesting systems were among those proteins as well as different Tetratricopeptide repeat (TPR)-like superfamily proteins, which are involved in organellar RNA processing or ribosomal proteins or t-RNA synthetases. Finally, several metabolic enzymes or components involved in electron transport or redox processes also were found, once more supporting the hypothesis that these TFs could indeed mediate an important functional role of SnRK1 signaling to chloroplasts and mitochondria. As the protein kinases inside the chloroplast, for example, the state transition kinases STN7 and STN8 or the chloroplast Casein Kinase II, respond to changes in the metabolic state or redox conditions (Link, 2003; Rochaix, 2013), changes in the electron flow and organellar redox state might also trigger changes in their activity, which would in turn explain the observed differences in phosphorylation levels of organellar proteins in the SnRK1 mutants. A similar scenario has been proposed before for other TFs involved in retrograde signaling (Kmiecik et al., 2016). In fact, evidence for this scenario was provided recently by the finding that ABI4, a central TF involved in different chloroplast retrograde signaling pathways, was phosphorylated by the MAP kinase MPK6 in a Ca²⁺-dependent manner (Guo et al., 2016). Strikingly, MPK6 also was recently found to be regulated by SnRK1 under hypoxia conditions (Cho et al., 2016).

CONCLUSION

The advances of sensitivity in mass spectrometry-based analysis of protein modification combined with targeted enrichment strategies of phosphorylated proteins allowed a quantitative determination of the phosphorylation status of SnRK1 targets in an in vivo context. Combined with the analysis of mutants including inducible double knockout/knockdown mutants of AKIN10/11, this enabled us to assess the level of regulation of metabolism by the SnRK1 kinase under different environmental conditions revealing an altered phosphorylation state of several metabolic enzymes. This explains one part of SnRK1 action in metabolic reprogramming by direct regulation of enzyme activities. The second part, however, the regulation of transcription, is mediated by regulation of TFs, which are very low-abundant proteins. Accordingly, such an analysis will be difficult to achieve using mass spectrometry-based approaches. Nevertheless, for the bZIP TF bZIP63, a difference in phosphorylation level could be observed by this method. As several other TFs have been identified as interactors or direct phosphorylation targets of SnRK1, based mostly on in vitro studies, further studies of those will be needed in the future using complementary approaches, for example, using target genes of those TFs as functional readout (see Outstanding Questions). Hence, the lists of potential targets summarized here can present a valuable starting point for future research. This is particularly true for the third aspect that has been highlighted in this review: the role of SnRK1 in the regulation of organellar functions. Our analysis uncovered a remarkable list of nuclear-encoded chloroplast and mitochondrial proteins with key functions for organellar biogenesis and metabolism. Functional testing of those proteins in SnRK1 mutants will reveal if the hypothesis proposed here is correct.

Finally, it becomes obvious that SnRK1 fulfills different functions in different plant tissues. For example, its involvement in FUS3-dependent ABA signaling during developmental phase transition is restricted due to FUS3 expression in seeds, cotyledons, hypocotyls, leaf primordia, and shoot apical meristems (Tsai and Gazzarrini, 2012), whereas SnRK1 action in autophagy via interaction with ATG1 may very well take place in all tissues of the plant (Chen et al., 2017). These points are of course also due for other kinases, such as CDPKs (Simeunovic et al., 2016) and had also been observed for other examples mentioned in this review, which demonstrate how important it is to interpret SnRK1 connections to other signaling networks in a tissue-dependent context to draw valid conclusions about its functions in planta.

Supplemental Data

The following supplemental materials are available.

• Supplemental Table S1. Predicted targets of TFs physically interacting with AKIN10.