Developmental control of hypoxia during bud burst in grapevine

Dormant or quiescent buds of woody perennials are often dense and in the case of grapevine ( Vitis vinifera L.) have a low tissue oxygen status. The precise timing of the decision to resume growth is difficult to predict, but once committed, the increase in tissue oxygen status is rapid and developmentally regulated. Here, we show that more than a third of the grapevine homologues of widely conserved hypoxia ‐ responsive genes and nearly a fifth of all grapevine genes possessing a plant hypoxia ‐ responsive promoter element were differentially regulated during bud burst, in apparent harmony with resumption of meristem identity and cell ‐ cycle gene regulation. We then investigated the molecular and biochemical properties of the grapevine ERF ‐ VII homologues, which in other species are oxygen labile and function in transcriptional regulation of hypoxia ‐ responsive genes. Each of the 3 VvERF ‐ VIIs were substrates for oxygen ‐ dependent proteolysis in vitro, as a function of the N ‐ terminal cysteine. Collectively, these data support an important developmental function of oxygen ‐ dependent signalling in determining the timing and effective coordination bud burst in grapevine. In addition, novel regulators, including GASA ‐ , TCP ‐ , MYB3R ‐ , PLT ‐ , and WUS ‐ like transcription factors, were identified as hallmarks of the orderly and functional resumption of growth following quiescence in buds.


| INTRODUCTION
The transition from heterotrophic to autotrophic metabolism is a defining feature of vegetative organogenesis in higher plants. The initial events during this transition are rapid and accompanied by considerable changes in the cellular and extracellular environment, notably light and oxygen-dependent cues, which function as powerful signals influencing cell function and fate. Bud burst in woody perennial species is an important ecological and agricultural context of this transition. Much of the global supply of fruits and nuts are generated from perennials and reliant on the successful timing and orchestration of bud burst, as it helps to synchronize flowering, fruit set, and harvest.
Several studies have modelled ecological requirements for bud burst (e.g., Chuine, 2000;Chuine & Cour, 1999;Linkosalo, Hakkinen, & Hanninen, 2006;Nendel, 2010), monitored the effect of particular chemical stresses (Halaly et al., 2008;Lavee & May, 1997;Ophir et al., 2009) or hormones (Aloni & Peterson, 1997;Aloni, Raviv, & Peterson, 1991;Lavee & May, 1997;Rinne et al., 2011) on the degree of latency, or carried out detailed molecular analysis of changes during the preceding stages of dormancy onset and release (Mazzitelli et al., 2007;Rohde et al., 2007;Ruttink et al., 2007). However, few studies have investigated the physiology in relation to transcriptional change during the course of bud burst. In particular, few studies have investigated the physiology of tissue oxygen status, and accompanying oxygen-dependent metabolism and signalling. Tissue oxygen status has emerged as an important context for the regulation of developmental transitions (Considine et al., 2017) and acclimation to stress (Voesenek & Bailey-Serres, 2015), as both an essential substrate for aerobic metabolism and important medium of post-translational modification and signal transduction.
Grapevine is among the most well-studied woody perennials. The mature proleptic bud of grapevine is a complex organ comprising multiple meristems of differing organogenic states. Metabolism reaches a minimum during dormancy and begins to increase with hydration prior to bud burst, then accelerates rapidly with the onset of bud burst (Gardea et al., 1994;Meitha et al., 2015). It is commonly accepted that a chilling requirement to satisfy dormancy (endodormancy) precedes the rise in metabolism and onset of bud burst (Lavee & May, 1997).
However, bud burst can be induced by a number of chemical and physical stresses, and indeed commercial practice commonly exploits one or other of these. Oxidative stress is a common theme among responses to these stresses, for example, inhibition of mitochondrial respiration by azide, application of hydrogen peroxide, heat shock, or hypoxia (see Or, 2009). Treatment with heat shock or cyanamide (commonly used in production) resulted in similar induction of antioxidant, glycolytic and fermentation enzymes, metabolism, and gene expression (Halaly et al., 2008;Ophir et al., 2009;Or, Vilozny, Eyal, & Ogrodovitch, 2000). Hypoxia and other inhibitors of mitochondrial respiration also elevated the levels of hydrogen peroxide (Vergara, Parada, Rubio, & Pérez, 2012), whereas hypoxia or cyanamide induced the expression of a number of conserved hypoxia-responsive genes as well as a homologue of FLOWERING LOCUS T ) and α-AMYLASE (Rubio, Donoso, & Perez, 2014). More recently, cyanamide was shown to increase expression of a number of homologues of cyclins and cyclin-dependent kinases (CDK; Vergara, Noriega, Parada, Dantas, & Pérez, 2016), reduce levels of abscisic acid (Vergara, Noriega, Aravena, Prieto, & Pérez, 2017), and promote the level of nitric oxide in the bud prior to bud burst (Sudawan, Chang, Chao, Ku, & Yen, 2016). Although most of these studies were conducted in stress conditions, the weight of evidence and practicality to industry suggest an important role for oxygen-dependent metabolism and signalling in the natural process of bud burst.
A signalling role and mechanism for cellular hypoxia was recently established in plants (Gibbs et al., 2011(Gibbs et al., , 2014Licausi et al., 2011). Despite the prevailing focus on stress-induced hypoxia in the literature, regulated hypoxia has emerged as an important developmental cue, essential for meiosis (Kelliher & Walbot, 2012), seed germination (Gibbs et al., 2014), and photomorphogenesis (Abbas et al., 2015), and likely to be important in preserving genome integrity through maintaining quiescence within the stem cell niche of meristems (Considine et al., 2017). The bud is enclosed by lignified bracts, a structure resembling a seed coat. We have previously shown that the core of the quiescent bud (postdormant) of grapevine has a low tissue oxygen status (internal partial pressure of oxygen, pO 2 < 3 kPa ≈ 40 μM [O 2 ]; Meitha et al., 2015). During the early stages of bud burst, pO 2 increased in a regulated spatial manner and there was a spatial and temporal shift in the localization of superoxide, from meristematic tissues to provascular tissues (Meitha et al., 2015). Although the outer scales are a barrier to oxygen diffusion, the pO 2 was clearly regulated by a balance of primary metabolism, as the pO 2 minimum was peripheral to the bud core.
However, physiological hypoxia cannot be defined by an objective measure of pO 2 alone (Sasidharan et al., 2017). A number of conserved metabolic and transcriptional responses have been observed among species in response to physiologically hypoxic conditions. These include restriction of oxidative phosphorylation and increased starch and sucrose catabolism, substrate-level phosphorylation (glycolysis) and fermentation, and activation of the γ-aminobutyric acid (GABA) shunt (Voesenek & Bailey-Serres, 2015).
Many of the conserved responses to hypoxia are transcriptionally regulated, in particular by Group VII ethylene response factors (ERF-VII; Gibbs et al., 2011;Licausi et al., 2011;Gibbs et al., 2014).
Oxygen-sensitive ERF-VIIs are targeted for proteolysis in the presence of oxygen and nitric oxide, as a function of their N-terminal motif (N-degron), a process termed N-end rule proteolysis. The consensus motif of the N-degron in ERF-VIIs, MCGGAI/L, is crucial for O 2 / . NO-dependent proteolysis. The degradation process is initiated by the cleavage of methionine from the N-terminus by a METHIONINE AMINOPEPTIDASE (MetAP), exposing cysteine for oxidation by the PLANT CYSTEINE OXIDASE1 or 2 (PCO1/2), a step that requires both oxygen and nitric oxide (Gibbs et al., 2014;Weits et al., 2014;White et al., 2017). Oxidized cysteine is then arginylated by an ARGINYL tRNA TRANSFERASE (ATE), which triggers interaction with E3 ligases (N-recognin) for ubiquitination and proteasomal degradation. In conditions of low oxygen and nitric oxide, the ERF-VIIs are stabilized and effect transcriptional regulation. For example, the stabilization of RELATED TO APETALA2.12 (RAP2.12), one of five Arabidopsis thaliana (arabidopsis) ERF-VIIs, promotes transcription of several hypoxia responsive genes (Kosmacz et al., 2015).
The RAP2.12 transcription factor directly binds to a specific motif in the promoter region of hypoxia responsive genes, consisting of a 12bp cis-regulatory sequence (5′-AAACCA[G/C][G/C][G/C]GC-3′), known as the hypoxia-responsive promoter element (HRPE;Gasch et al., 2016). The motif presents 1-3 times in the promoter region of 39 out of 49 core hypoxia-responsive genes of arabidopsis (Mustroph et al., 2009) and is adequate for the activation of the AtPCO1 and LATERAL ORGAN BOUNDARIES PROTEIN41 (AtLBD41) RAP2.2 and RAP2.12 in arabidopsis (Gasch et al., 2016). Nevertheless, not all plant ERF-VIIs are substrates for N-end rule proteolysis, as the position of a downstream lysine and a tertiary structure that exposes the N-terminus also determines ability to function via the N-end rule (Gibbs et al., 2011).
The aim of this study was to investigate the role of oxygendependent signalling in coordinating the effective resumption of transcription and metabolism during bud burst in grapevine. We demonstrate that the rapid development of a physiologically hypoxic state is reflected in genome-wide transcriptional regulation during the first 24 hr of the resumption of growth in the postdormant buds.
This was evident in the enrichment of numerous conserved hypoxiaresponsive gene homologues, as well as genes with an HRPE, within 24 hr, during which time tissue oxygen status remained low, as respiration, sugar consumption, and starch hydrolysis increased.
Further, we demonstrate that the Vitis ERF-VII transcription factors are substrates for oxygen-dependent N-end rule proteolysis in vitro, providing strong evidence for a developmental role of hypoxia signalling during bud burst in grapevine. Moreover, the transcriptional and physiological patterns were in harmony with the orderly resumption of an autotrophic capacity for growth and metabolism, even in the absence of light. This study provides a new benchmark for understanding the developmental regulation of bud burst in perennial plants. Moreover, these data implicate a facultative light requirement for bud burst in grapevine, which we explore in an accompanying manuscript (Signorelli, Agudelo-Romero, Meitha, Foyer, & Considine, 2018).

| Plant material and physiological assays
Unless otherwise stated, water used throughout the study was Milli-Q® water (MQW, Merck-Millipore, Bayswater, Australia) and chemicals were analytical grade from Sigma-Aldrich (Castle Hill, Australia). Grapevine (Vitis vinifera var. Crimson Seedless) canes with quiescent buds intact were sampled from a vineyard in Western Australia (33.694°S, 115.102°E) in mid-winter and stored at 4°C in the dark until required (4 months). This period of chilling was similar to that used to study dormancy release in raspberry (Mazzitelli et al., 2007) and enables the experimental approximation of a temperate-climate winter. Prior to the experiment, the stored canes were removed from 4°C and buds from Nodes 5-7 were cut into intact single-node cuttings (explants), with approximately 10 mm cane above the node and 40 mm cane below. The explants were then planted in moist vermiculite and maintained in growth conditions with a constant temperature of 23°C, in constant darkness or dark/light treatment (12/12 hr dark/light photoperiod, 150 μmol quanta·m −2 ·s −1 , LED tubes #108D18-V10, Everfine, Hangzhou, China) and deionized watered to field capacity (i.e., substrate water-holding with drainage), for a period of 3, 24, 72, or 144 hr. The dark/light regime commenced with 6 hr dark, followed by alternating 12 hr light and dark cycles, such that the 24, 72, and 144 hr samples were taken in the middle of the light period.
There were no visible differences in development between these time points, irrespective of light condition ( Figure S1). Both time points were classified at the bud-swell stage of EL2/3 (modified Eichorn-Lorenz [EL] scale; Coombe, 2004), that is, prior to the emergence of leaf tips, which indicates bud burst sensu stricto (EL4). In these conditions, buds reached 50% bud burst (EL4) within 7-10 days, with minimal difference between treatment conditions; however, a trend towards acceleration of bud burst in the presence of light was evident (data not shown). For metabolite and transcript analysis, buds were excised from the cane immediately upon sampling, directly submerged in liquid nitrogen and stored at 80°C until required.
Bud respiration (three biological replicates of four buds each) were performed as described previously (Meitha et al., 2015). Internal partial pressure of oxygen (pO 2 ; three replicates of one bud each) was measured as described for buds without removing the outer scales in Meitha et al. (2015). It is worth mentioning that the degree of noise in the pO 2 data is due to structural heterogeneity of the bud. For this reason, we presented the raw data as well as the spline fit and 95% confidence intervals. Analysis of internal pO 2 was not reliable at 144 hr in our hands and thus not presented. Moisture content was measured on the same buds used for respiration (three replicates of four buds each) as follows: clean excised buds, without agar were weighed fresh (FW) then dipped in liquid nitrogen and stored at −80°C or directly transferred into a freeze dryer for >48 hr before dry weight (DW) was recorded. Hydration was expressed as the percentage of water mass (g) per 100-g FW.

| Sugar, starch, and chlorophyll analysis
The buds for metabolite analyses were the same as those used for respiration and hydration (three biological replicates of four buds each). For sugars and starch, extraction was performed as previously described (Gomez, Rubio, & Augé, 2002) with minor modifications and in a smaller scale. Freeze-dried buds were ground in an gently agitated at 4°C for 20 min to disperse. After 10 min of sonication at 60°C, the tubes were allowed to cool down on ice for 2 min. A two-liquid phase separation was obtained after 3 min of 13,000 × g centrifugation at 4°C for 3 min. The polar supernatant was used for the subsequent processes of soluble sugars assay, whereas the pellet was washed in 100% (v/v) methanol, dried in a speed-vac and retained for starch quantification as previously described (Gomez, Rubio, & Lescourret, 2003). For GC, bud extracts were derivatized by mixing with 20 mg/ml methoxyamine hydrochloride in pyridine at 65°C for 120 min. The derivatives were injected in split mode (10:1) into a GC (Agilent Technologies, 7890 GC System, Mulgrave, Australia) with a 30-μm capillary injection column (VP5-MS, 0.25 mm diameter and 0.25 μm film). Injection temperature was 280°C, and oven ramp was 325°C, held for 3.5 min, 6°C/min ramp to 215°C, held for 1 min, 40°C/min ramp to 320°C, held for 22 min. Helium, the carrier gas, was at a constant flow rate of 0.9 ml/min. Glucose, sucrose, and fructose were identified and peaks quantified (Chemstation Quantitation Process Program; Agilent Technologies and Agilent MassHunter Workstation Software for Quantitative Analysis) by comparison with authentic standards and the internal standard.
For chlorophyll analysis, snap-frozen buds were ground to a powder in a 1.5-ml tube with a micropestle. Chlorophyll was extracted as pheophytin by addition of 0.2 N hydrochloric acid (Sigma-Aldrich), in a mass per volume ratio of 10 mg FW/100 μl at room temperature. After centrifugation at 14,000 × g for 8 min at 4°C, the supernatant was separated for pheophytin quantification as described previously (Queval & Noctor, 2007). The pellet was kept to determine the dry weight. Chlorophyll was expressed as microgram pheophytin per g DW.
For statistical comparison of respiration, moisture content, sugar, starch, and chlorophyll, data were analysed by ANCOVA using a second-order polynomial to fit the time-response relationship (R-Core Team). Where the treatment factor was statistically significant, the residual standard error was used to estimate the 95% confidence limits for the predicted means.

| RNA extraction, library construction, and sequencing
Samples were ground in liquid nitrogen to a fine powder. Total RNA extraction was performed using the Spectrum Plant Total RNA kit with an on-column DNase treatment according to the manufacturer's instructions (Sigma-Aldrich), followed by an isopropanol/acetate precipitation. The quality and integrity of the isolated RNA was tested using a NanoDrop 100 spectrophotometer (Thermo-Scientific, Scoresby, Australia) and agarose gel electrophoresis. Only
Resulting reads were aligned to the whole 12X V1 Vitis vinifera PN40024 reference genome (Jaillon et al., 2007) using Kallisto (Bray et al., 2016). Gene expression profiling was performed using edgeR (Robinson, McCarthy, & Smyth, 2010) and limma (Ritchie et al., 2015) Bioconductor packages. The counts matrix obtained from Kallisto was read in edgeR, and the quality of the replicates was checked using Pearson's correlation (0.90-0.98). Raw data were then normalized using the trimmed mean of M values (TMM) method (Table S1) and the log 2 transcripts-per-million (logTPM) was obtained. A linear model (limma) was then applied to do a (Table S2). P values were corrected for multiple testing using the Benjamini-Hochberg's (1995) method (FDR P ≤ 0.01). Data were then filtered considering a fold change (FC|2.5|).

| Functional enrichment analysis and principal component analysis (PCA)
Three lists of genes differentially expressed from each comparison, up-and down-regulated genes and both (up-plus down-regulated genes) were used to perform a functional enrichment analysis using transcripts housed in the grapevine 12X V1 gene predictions (Grimplet et al., 2012). Significant enrichment was considered for P values (P ≤ 0.01) after Benjamini and Hochberg correction for multiple testing (Benjamini & Hochberg, 1995). Finally, the covariance matrix of the PCA was built using the prcomp() function of R. Eigenvalues and eigenvectors were calculated using the factoextra library of R. BLAST tool from CRIBI was used (blastn; http://genomes.cribi.

| Cis-regulatory motif analysis and BLAST
unipd.it/grape/), where all the sequences with an e-value ≤35 were selected to further analysis.

| Plasmid construction, in vitro protein translation, and immunodetection
In vitro assays to determine the capacity of VvERF-VII proteins to act as substrates for N-end rule proteolysis was performed as previously   (Considine, 2018;Signorelli et al., 2018). Light incidence is required for bud outgrowth in many species (Leduc et al., 2014). Although grapevine buds apparently do not require light to burst, light incidence does have a strong influence on inflorescence initiation and fruitfulness within buds (Buttrose, 1970). In this study, we were primarily interested in the role of light as a substrate that promotes photosynthetic oxygen synthesis. The incidence of light has a strong influence on primary metabolism and tissue oxygen tension in photosynthetic seeds (Borisjuk & Rolletschek, 2009), and as hypoxia was a key focus of this study, we hypothesized that the degree of hypoxia (low bud oxygen tension) would be directly influenced by light. As such, explants were grown in continuous darkness (D) or a dark/light photoperiod (DL; 12/12 hr). There were no visible phenotypic differences between these two time points, irrespective of light condition; all experimental buds were at the bud-swell stage (EL2/3), prior to bud burst (EL4; Coombe, 2004; Figure S1). The increase in moisture content and respiration followed a similar pattern, with significant developmental effects but no significant impact of light (p ≤ .05; Figure 1a). Changes in chlorophyll content were not statistically significant, although there appeared to be an interaction between treatment (±light) and time, such that the net chlorophyll content was only maintained in the presence of light.
Temporal changes in sugar and starch contents were significant, but the effect of light was only significant in interaction with time and generally minor. Glucose and fructose concentration appeared to peak at 72 hr, followed by a decline, which was also reflected in the decline in starch content (Figure 1a). Tissue oxygen status (pO 2 ) provides a spatial resolution of oxygen metabolism. The bud is relatively hypoxic at the earliest stages of bud burst (3 hr), followed by a considerable oxygenation within 24 hr (Meitha et al., 2015). In the previous report, however, buds were grown in continuous darkness only. Here, we show that there was negligible difference in the pO 2 minimum of the profile, or the pO 2 at the core of the bud at 72 hr in the DL compared with D conditions (Figure 1b).
Taken together, these data show that the resumption of primary metabolism is rapid during bud burst, with an increase in respiration and starch hydrolysis within 72-144 hr. We earlier hypothesized that contrasting the presence and absence of light would significantly affect oxygen-dependent physiology and metabolism. Of the measures we assayed, only chlorophyll content appeared to be affected and this was not significant, and thus, we conclude that the resumption of metabolism during bud burst in grapevine was unaffected by the presence of light.

| Transcriptome analysis
We performed short-read RNA sequencing in order to further investigate the dynamics of transcriptional regulation during bud burst over 6 days, which precedes leaf emergence (0, 3, 24, 72, and 144 hr).
A dendrogram and PCA of the normalized TPM data demonstrated that replicates of each condition clustered together and that the first and second components discriminated conditions by time and treatment, accounting for approximately 40% of the variance collectively ( Figure S4a  Three different functional enrichment analysis were performed for each comparison, where "All" (up-plus down-regulated genes, grey), "Up" (up-regulated genes, purple) and "Down" (down-regulated genes, green) make reference to the list of genes assessed A HYPOXIA RESPONSIVE2 (HRE2, also named ERF071) and two ALCO-HOL DEHYDROGENASE (ADH) homologues were also specifically regulated in this way, as described in further detail below.
A number of common pathways were implicated in the subse-  (Voesenek & Bailey-Serres, 2015) as well as stress-induced metabolism during the release of bud dormancy or quiescence (Considine & Considine, 2016;Or, 2009 indicating the development of a hypoxic state early during bud burst.
VIN3, described above has also been shown to be required for acclimation to hypoxia, as it is for the vernalization response in winter annuals (Bond, Wilson, Dennis, Pogson, & Finnegan, 2009). Strikingly, by 24 hr, expression of two homologues of the SUCROSE NONFERMENTING1-RELATED KINASE1 (SnRK1) catalytic subunit AKIN gamma were elevated, which is consistent with the earlier identification of a SnRK1homologue during dormancy release in grapevine buds . SnRK1 is a major hub of energy homeostasis (Baena-González, Rolland, Thevelein, & Sheen, 2007) and shown to be induced in hypoxic conditions, acting as a positive regulator of sugar response elements in the promoter region of α-AMYLASE genes, enabling substrate supply to glycolysis and fermentation (Lee, Chen, & Yu, 2014). In addition, in our data, four homologues of CIPK (CBL-

INTERACTING PROTEIN KINASE) genes, which positively regulate
SnRK1-mediated starvation responses in plants, were induced at 3 and/or 24 hr. This is also compatible with the strong evidence for GA signalling, which can also promote amylase expression. In contrast, there was no clear direction of the regulation of trehalose-6phosphate abundance, which is a major counterpart to SnRK1 in energy homeostasis and signalling, as genes coding for the synthesis and catabolism of trehalose-6-phosphate were differentially regulated over this period.
The plant hormone ethylene has been widely implicated in acclimation to hypoxia (Voesenek & Bailey-Serres, 2015). A number of homologues of ethylene synthetic and signalling genes were differentially regulated in the first 24 hr, including an ACC SYNTHASE, five ACC OXIDASE transcripts, and two homologues of the ETHYLENE OVERPRODUCER1 (ETO1), encoding a negative regulator of ACC SYN-THASE (Table S4). Additionally, a number of AP2/EREBP (APETALA2/ ETHYLENE-RESPONSIVE ELEMENT BINDING PROTEIN) homologues were regulated, described further below. Although the ACC SYNTHASE levels were reduced at 3 hr, so too was an ETO1, whereas three of the ACC OXIDASE transcripts were elevated. These transcripts were unchanged at 24 hr, indicating that expression levels were maintained, and hence suggesting that ethylene synthesis was elevated in the first 24 hr of bud burst. Together, these data provide strong evidence for the transcriptional activation of a state of metabolic control that is typical of hypoxia-induced responses.

| Perturbed redox metabolism and signalling suggests increased redox activity early during bud burst
Redox metabolism and signalling have also been implicated in crosstalk or direct regulation of hypoxia, metabolic, hormone, and cell-cycle regulating pathways in plant development (Considine et al., 2017;Considine & Foyer, 2014). Several studies highlighted in the introduction have demonstrated induction of hydrogen peroxide as well as numerous classical antioxidant enzymes during the stress-induced release of bud dormancy or quiescence (Considine & Considine, 2016;Or, 2009). In our data, an increase in cellular oxidation was suggested in the first 3 hr and further at 24 hr, by the elevated expression of two ALDEHYDE DEHYDROGENASE, a number of PEROXIDASE, including three ASCORBATE PEROXIDASE, and a number of THIOREDOXIN-like homologues (Table S4 and Figure S5). Nevertheless, a homologue of THIOREDOXIN h, previously shown to be induced by cyanamide in grapevine buds (Halaly et al., 2008), as well as three  Table S4). To gain further insight to the prevailing transcriptional control during bud burst, we carried out enrichment for known cis-regulatory motifs in the 2.5-kb upstream region of the DEGs. We retained focus on the first 24 hr (3/0 hr and 24/3 hr) in darkness. Several motifs were commonly enriched throughout the first two transitions such as the Ebox and three bZIP motifs (bZIP16, bZIP68, and AREB; Figure 3a; Table S5). The prevalence of these binding motifs is understandable: E-box motifs are most well understood in the establishment of a circadian rhythm (Seitz, Voytsekh, Mohan, & Mittag, 2010); AREB motifs are known to function in dehydration stress (Urano et al., 2017), whereas the bZIP16 and bZIP68 binding motifs are involved in the redox-dependent regulation of gene expression during photomorphogenesis (Shaikhali et al., 2012). Also of interest, although less significantly enriched, were binding motifs for ELONGATED HYPO-COTYL (HY5) and PHYTOCHROME-INTERACTING FACTORs (PIF), again highlighting the constitutive activation of autotrophic development (Table S5).
Particular binding motifs were enriched in a stage-specific manner and cell cycle genes in the shoot apical meristem (Simonini et al., 2012;Aguilar-Martínez et al., 2015). The enrichment of CAMTA1 motifs may be a product of the temperature transition at this stage, although calcium signalling and responses were among the highly regulated transcripts at 3 hr (Table S4), and have previously been implicated in the release of bud dormancy in grapevine (Pang et al., 2007).
At 24 hr, fewer binding motifs were uniquely enriched, approximately half the number found at 3 hr; however, three MYB3R motifs were considerably enriched. Plants possess only a small number of three-Myb-repeat proteins (MYB3R), and many of those in arabidopsis have been shown to directly regulate canonical mitotic proteins (Haga et al., 2011). Knockout and biochemical studies showed that the arabidopsis MYBR1 and MYB3R4 act as transcriptional activators of G2/M phase genes, whereas MYBR3 and MYB3R5 act as repressors, possibly functioning to promote the endocycle and differentiation (Haga et al., 2011;Kobayashi et al., 2015

| Conserved hypoxia-responsive genes and representation of an HRPE suggest oxygen-dependent signalling is important during bud burst
Previous studies have identified a set of hypoxia-responsive genes conserved across phyla, which corresponds to 49 in arabidopsis (Mustroph et al., 2009). In addition, a unique plant HRPE was recently identified and shown to function in oxygen-dependent signalling (Gasch et al., 2016). We used BLAST to identify a set of 122 Vitis homologues of the 49 arabidopsis hypoxia-responsive genes and queried the expression profiles of our transcript data, using the existing FC and FDR criteria. Forty-four of the 122 Vitis homologues (36%) were differentially expressed in the first 24 hr (Figure 4, upper panel).
More than half of these were elevated at 3 hr, and many were unchanged or only moderately repressed at 24 hr (relative to 3 hr).
Many of the elevated transcripts at 3 hr have already been described, particularly with regard to glycolysis, fermentation, and hypoxia, including HRE2, but notable also were four homologues of PLANT CYSTEINE OXIDASE, which function upstream of the ERF-VII in O 2 / .
NO-dependent N-end rule proteolysis (Weits et al., 2014). The HRPE is a specific and recently defined plant cis-regulatory motif that is not present in the databases used by HOMER. We identified 252 genes in the Vitis genome with one or more HRPE in the upstream promoter region. Of these, 55 were present in the total core set of 4,022 DEGs, and 47 of these were differentially expressed in the first 24 hr, that is, nearly 20% of the whole HRPE-containing transcriptome ( Figure 4). This included nine of the 44 Vitis homologues of conserved hypoxia-responsive genes (not duplicated in the lower panel of Figure 4). Furthermore, many of the HRPE-containing genes that were differentially regulated at 3 hr were not further changed at 24 hr. Although cautious interpretation is warranted, these data suggest that oxygen-dependent signalling is most important within 3 hr of the resumption of growth during bud burst; however, there is no striking repression of these, suggesting that hypoxia-dependent gene expression may be maintained throughout the first 24 hr. This is consistent with the persistence of low oxygen status during the first 24 hr in the bud, followed by oxygenation (Meitha et al., 2015). The presence of an HRPE in one of the induced GASA-like transcripts is particularly interesting given the prominence of these transcription factors in data already presented, as well as the redox regulation of several members of this family in arabidopsis (Sun  (Mustroph et al., 2009) are shown in the upper panel of the heatmap (adjacent the yellow column). DEGs containing one or more hypoxia-responsive promoter element (HRPE; Gasch et al., 2016) in the 2.5-kb upstream promoter region (refer methods) are shown in the lower panel of the heatmap (adjacent the green column). There were 197 genes containing one or more HRPE in the Vitis reference genome (Grimplet et al., 2012). Genes marked with an asterisk are among the homologues of arabidopsis hypoxia-responsive genes that contain one or more HRPE motif. The gene marked with a black box (■) is a homologue of the arabidopsis Group VII ETHYLENE RESPONSE FACTOR (ERF-VII) HRE2 (also named ERF071)

| VvERF-VIIs are substrates for oxygen-dependent N-end rule proteolysis in vitro
The N-terminal cysteine (Cys2) is a tertiary destabilizing residue in ERF-VIIs, which are subject to N-end rule proteolysis (Gibbs et al., 2011(Gibbs et al., , 2014. Nevertheless, presence of Cys2 does not adequately predict N-end rule proteolysis; the rice ERF-VII SUB1A was previously shown to evade this regulation, despite bearing Cys2, for example (Gibbs et al., 2011). We thus examined the stability of the VvERF-VIIs using an in vitro translation system, using full-length wild-type (MC-) constructs and with those where the Cys2 was mutated to alanine (MA-), which attenuates N-end rule proteolysis. In addition, we exam-  contributed to the conception of the study and drafting the results.