Identification of transcription factors that regulate ATG8 expression and autophagy in Arabidopsis

ABSTRACT

Autophagy is a conserved catabolic process in eukaryotes that contributes to cell survival in response to multiple stresses and is important for organism fitness. In Arabidopsis thaliana, the core machinery of autophagy is well defined, but its transcriptional regulation is largely unknown. The ATG8 (autophagy-related 8) protein plays central roles in decorating autophagosomes and binding to specific cargo receptors to recruit cargo to autophagosomes. We propose that the transcriptional control of ATG8 genes is important during the formation of autophagosomes and therefore contributes to survival during stress. Here, we describe a yeast one-hybrid (Y1H) screen for transcription factors (TFs) that regulate ATG8 gene expression in Arabidopsis, using the promoters of 4 ATG8 genes. We identified a total of 225 TFs from 35 families that bind these promoters. The TF-ATG8 promoter interactions revealed a wide array of diverse TF families for different promoters, as well as enrichment for families of TFs that bound to specific fragments. These TFs are not only involved in plant developmental processes but also in the response to environmental stresses. TGA9 (TGACG (TGA) motif-binding protein 9)/AT1G08320 was confirmed as a positive regulator of autophagy. TGA9 overexpression activated autophagy under both control and stress conditions and transcriptionally up-regulated expression of ATG8B, ATG8E and additional ATG genes via binding to their promoters. Our results provide a comprehensive resource of TFs that regulate ATG8 gene expression and lay a foundation for understanding the transcriptional regulation of plant autophagy.

Abbreviations: ABRC: Arabidopsis biological resource center; AP2-EREBP: APETALA2/Ethylene-responsive element binding protein; ARF: auxin response factor; ATF4: activating transcription factor 4; ATG: autophagy-related; ChIP: chromatin immunoprecipitation; DAP-seq: DNA affinity purification sequencing; FOXO: forkhead box O; GFP: green fluorescent protein; GO: gene ontologies; HB: homeobox; LD: long-day; LUC: firefly luciferase; MAP1LC3: microtubule associated protein 1 light chain 3; MDC: monodansylcadaverine; 3-MA: 3-methyladenine; OE: overexpressing; PCD: programmed cell death; qPCR: quantitative polymerase chain reaction; REN: renilla luciferase; RT: room temperature; SD: standard deviation; TF: transcription factor; TFEB: transcription factor EB; TGA: TGACG motif; TOR: target of rapamycin; TSS: transcription start site; WT: wild-type; Y1H: yeast one-hybrid.

Introduction

Autophagy, literally meaning self-eating, is a fundamental cellular catabolic process conserved in eukaryotic organisms. Eukaryotic cells employ macroautophagy/autophagy to facilitate degradation of unwanted cytoplasmic contents, abnormal protein aggregates or dysfunctional organelles in lysosomes (in animals) or vacuoles (in yeast and plants) for recycling [1]. Autophagy constitutes a primary protective mechanism that allows cells to survive when faced with multiple stresses and is important for organism fitness [2]. Autophagy is dependent on a set of ATG proteins, initially identified in yeast. The process of autophagy begins with the formation of a double-membrane cup-shaped structure (a phagophore), which expands to form a double-membrane vesicle called an autophagosome. This process involves at least 4 molecular components, the ATG1 (autophagy-related 1)-ATG13 (autophagy-related 13) kinase complex for induction, the class III phosphatidylinositol 3-kinase (PtdIns3K) complex for nucleation, and 2 ubiquitin-like conjugation systems that modify ATG12 (autophagy-related 12) and ATG8 for phagophore expansion and maturation [3]. Upon completion of the autophagosome, it docks and fuses with the vacuole for cargo degradation. The resulting breakdown products are released back into the cytosol to maintain nutrient and energy homeostasis [3].

Among the numerous ATG proteins, a central role belongs to the ubiquitin-like protein ATG8, which functions in autophagosome formation, mediating membrane tethering, elongation and fusion [4,5]. Upon autophagy activation, ATG8 undergoes lipidation to generate a membrane-bound ATG8-phosphatidylethanolamine conjugate that localizes to growing phagophores and completed autophagosomes. ATG8 proteins are therefore often used as reliable autophagosome markers to assess the induction and progression of autophagy [6,7]. ATG8 is also important in cargo recognition through specific interactions with autophagy receptors during selective autophagy [8]. The receptors bind both to ATG8 and to ubiquitinated proteins, organelles or even pathogens which are targeted for degradation, leading to their incorporation into autophagosomes [9]. Upon closure of the autophagosome and fusion with lysosomes or vacuoles, the ATG8 protein in the outer autophagosomal membrane is recycled, whereas that in the inner membrane is degraded together with the cargo [10]. Upregulation of ATG8 is therefore necessary to provide sufficient protein for sustained autophagy. ATG8 transcript and protein levels increase following the induction of autophagy [11,12]. Regulation of ATG8 expression is therefore potentially an important point of regulation of the autophagy pathway.

Recently, the network of TFs that regulate ATG gene expression in mammalian cells and yeast has begun to be elucidated. A number of TFs have been reported to be transcriptional regulators of the ATG8/MAP1LC3 (microtubule associated protein 1 light chain 3) family as well as additional ATG genes, including the FOXO (forkhead box O) family [13-16], E2F1 (E2F transcription factor 1) [17-19], ATF4 (activating transcription factor 4) [20-22], CEBPB/CEBP β (CCAAT enhancer binding protein beta) [23], DDIT3 (DNA damage inducible transcript 3)/CHOP (C/EBP homologous protein) [24], GATA1 (GATA binding protein 1) [25], JUN (Jun proto-oncogene, AP-1 transcription factor subunit) [26-28], TFEB (transcription factor EB) [29], SREBF2/SREBP2 (sterol regulatory element binding transcription factor 2) [30], ZKSCAN3 (zinc finger protein with KRAB and SCAN domains 3) [31], Ume6 (unscheduled meiotic gene expression) complex [32] and Pho23 (PHOsphate metabolism) [33]. The activation and nuclear translocation of these TFs enables them to act as crucial regulators of autophagy, allowing a sustained level of cellular autophagy flux [34,35]. For example, TFEB, which is a master regulator of lysosomal biogenesis [36], resides in the cytoplasm under basal conditions, but rapidly translocates to the nucleus during nutrient starvation, upregulating core autophagy genes and enhancing autophagy [29]. Together with post-transcriptional regulation (mainly via miRNAs) and post-translational modifications (including phosphorylation, ubiquitination and acetylation), autophagy is tightly regulated at multiple levels in response to homeostatic perturbations [34,37]. Autophagic response to stress has been proposed to be biphasic, in which the rapid induction of autophagy within minutes or hours of exposure to stress involves post-translational protein modifications and is followed by a protracted autophagic flux that relies on a collection of TFs that upregulate ATG gene expression [37].

In the model plant Arabidopsis thaliana, ATG8 proteins are encoded by 9 genes that have been grouped into 3 clusters based on amino acid sequence alignment [38,39]. The first cluster includes 4 members (ATG8A, ATG8C, ATG8D and ATG8F), the second 3 members (ATG8B, ATG8E and ATG8G), and the third 2 (ATG8H and ATG8I) [38,40]. Different ATG8 genes have distinct spatial and temporal expression patterns in different tissues [40,41], suggesting that the different ATG8 proteins may serve different functions and be regulated by distinct factors. However, information on the transcriptional regulators of ATG genes in plants is limited. The tomato TF HSFA1A (class A heat shock factor 1A) regulates autophagy and contributes to drought tolerance through activating ATG10 (autophagy-related 10) and ATG18F (autophagy-related 18F) genes [42], and Arabidopsis WRKY33 (WRKY DNA-binding protein 33) regulates autophagy during plant defense [43]. Moreover, silencing of tomato WRKY33 genes reduced heat-induced ATG gene expression and autophagosome accumulation [44]. Other than these examples, most of the regulators of plant autophagy identified so far work at the posttranslational level. For example, the TOR (target of rapamycin) signaling complex [45-47] and Snf1-related protein kinase 1 (SnRK1) [48,49] both control autophagy through protein phosphorylation-dependent signaling pathways.

A library of 1956 Arabidopsis TFs has been constructed and used in the generation of several large-scale gene-regulatory networks [50-55]. Here, taking advantage of this TF library, we performed Y1H screens using 4 ATG8 promoters and identified distinct TF-promoter interactions, a subset of which were validated in Arabidopsis protoplasts. One bZIP (basic leucine-zipper protein) TF, TGA9, was shown to upregulate expression of ATG8B and ATG8E as well as additional ATG genes by binding to the TGA motifs in their promoters. Overexpression of TGA9 activated autophagy under both sucrose starvation and osmotic stress conditions, indicating that TGA9 is a positive regulator of autophagy, and enhanced starvation tolerance in seedlings. Activation of autophagy by TGA9 was repressed in atg5 mutant protoplasts or in the presence of the autophagy inhibitor 3-methyladenine (3-MA), indicating that it depends on the typical autophagy machinery. Our results provide an analysis of TF-ATG8 promoter interactions and lay a foundation for understanding the transcriptional control of autophagy in plants.

Results

Distinct expression patterns of Arabidopsis ATG8 genes under different abiotic stresses

The ubiquitin-like protein ATG8 is required for the formation of autophagosomes and for cargo selection [4,8]. To study the transcriptional regulation of ATG8, we first examined the expression patterns of the 9 Arabidopsis ATG8 genes in response to different abiotic stresses (Figure 1). Upon sucrose starvation, ATG8B gene expression was strongly induced, compared to subtle changes for other genes (Figure 1(a)). Most ATG8 genes were significantly induced by nitrogen starvation, with the exception of ATG8E and ATG8F (Figure 1(a)). Under mannitol-induced osmotic stress and NaCl-induced salt stress, all 9 ATG8 genes were significantly up-regulated to varying degrees, and their expression in osmotic stress was higher than that in salt stress (Figure 1(b)). We also analyzed the relative expression patterns of ATG8 genes under fixed-carbon starvation of adult plants, for which data were extracted from the published RNA-sequencing dataset GSE93420 [56]. After 5 d of fixed-carbon starvation in darkness, ATG8A, ATG8B and ATG8H displayed greater induction than ATG8D, ATG8E and ATG8I, while the remaining genes did not change significantly in expression (Figure 1(c)). We therefore hypothesized that transcriptional regulation of the ATG8 genes may contribute to regulation of autophagy under different conditions.

Figure 1.

Expression of ATG8 genes under abiotic stress. (a) 7-d-old Arabidopsis seedlings were transferred to ½ MS medium with sucrose (control), without sucrose (-SUC) and without nitrogen (-N) for 3 d, and the transcript level for each ATG8 gene was quantified by real-time qPCR. (b) 7-day-old Arabidopsis seedlings were transferred to ½ MS medium with 0.35 M mannitol or 0.16 M NaCl for 6 h, and the transcript level for each ATG8 gene was quantified by real-time qPCR. (c) Relative expression of ATG8 genes in 4-week-old plants after 5 d dark treatment to cause fixed-carbon starvation. Data were extracted from the published RNA-sequencing dataset GSE93420 [56]. Expression of each gene was normalized to expression in control conditions and represented as the mean of 3 biological replicates, except for fixed-carbon starvation which had 2 biological replicates. Error bars indicate SD. Differences are significant at p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***) by Student’s t test.

Out of the 9 Arabidopsis ATG8 genes, we selected 4 representatives for initial analysis of transcriptional regulation, ATG8A, ATG8B, ATG8E and ATG8H. These include genes from each of the 3 clusters, and also genes with distinct expression patterns that may therefore have distinct mechanisms of regulation. ATG8A and ATG8E not only encode the 2 most commonly used markers for autophagosomes through fusion with fluorescent tags [57-59], but were also significantly induced by both osmotic and salt stress and specific starvation conditions. ATG8B was the only gene strongly induced by all types of starvation. ATG8H was the most highly upregulated gene under most stress conditions, has lower similarity to other ATG8 family members, and lacks the additional C-terminal residues that are usually present and are proteolytically removed to expose a C-terminal glycine for lipidation [38].

Y1H screen for regulators of ATG8 genes identifies diverse TFs

To identify TFs that regulate ATG8 gene expression, promoter fragments of Arabidopsis ATG8 genes (Figure S1(a) and Figure 2(a)) were used as baits in a Y1H approach and screened against a library of 1956 Arabidopsis TFs according to the schematic [51] shown in Figure 2(b) (Excel S1). Initially we screened 4 short overlapping promoter fragments of ~400 base pairs (bp) in length for ATG8A as suggested in the literature [51,60]. The intergenic region between ATG8H and the gene immediately upstream is very short, and this entire region was therefore screened as the potential promoter (Figure S1(a)). We identified 13–50 TF-interactions per fragment for the ATG8A promoter, for a total of 134 interactions with 81 unique TFs. Over half of these TFs were specific for one single promoter fragment (57%), and some TFs interacted with 2 (27%), 3 (10%) or all 4 (6%) fragments. The first 3 fragments (F1–F3) produced the most interactions, while F4 has the fewest (Figure S1(b,c)). Eighty-three percent of F1 TFs and 77% of F4 TFs also interacted with either F2 or F3 (Figure S1(d)). The ATG8H promoter interacted with 32 TFs, 30 of which overlapped with TFs from the ATG8A promoter screen (Figure S1(e)). Network analysis indicated that TFs from AP2-EREBP (APETALA2/ethylene-responsive element binding proteins), HB (homeobox), GeBP (GLABROUS1 enhancer binding protein) and ARF (auxin response factor) families were enriched in both ATG8A and ATG8H screens, while WRKY family TFs were specifically enriched in interaction with the ATG8A promoter (Figure S2; Figure S3(a,b)). A reproducibility of 62% and 47% was obtained between 2 biological screens for the ATG8A and ATG8H promoters respectively (Table S1).

Figure 2.

Y1H screen to identify regulators of ATG8B and ATG8E gene expression. (a) Promoter schematic of ATG8B and ATG8E genes. Red lines are fragments used for the Y1H screen. TSS indicates transcription start site. (b) Experimental design of Y1H assays. The library of TFs and ATG8 gene promoter fragments were introduced into yeast mating strains, diploids selected, and OD₆₀₀ and β-galactosidase reporter activity measured. (c) Venn diagram of the number of TFs identified for ATG8B and ATG8E promoters. (d) Comparisons between ATG8-Y1H data and DAP-seq data. DAP-seq TF binding profiles for 336 TFs in the TF library were used to assess the overlap between ATG8 promoter binding and Y1H targets identified in our screen. Overlap significance is shown as – log10(p-value) as calculated in GeneOverlap by Fisher’s exact test.

From the above analysis, almost 90% of the TFs in the ATG8A promoter screen could be identified from F2 or F3 (Figure S1(d)). We therefore considered that we could screen for additional promoter interactions more efficiently using single fragments, and we used a fragment of ~650 bp in length starting from the transcription start site (TSS) for the ATG8B and ATG8E promoters (Figure 2(a)). The ATG8B and ATG8E promoter fragments yielded 108 and 163 TF interactions, respectively, 73 of which were shared by both (Figure 2(c)). Taking advantage of the Plant Cistrome Database [61], we analyzed the overlap between candidates from the Y1H screen and DNA affinity purification sequencing (DAP-seq), in which genomic DNA binding to affinity-tagged in vitro-expressed TFs is assessed by next-generation sequencing [61,62]. Significant overlap was found between the TFs identified in the Y1H screen as interacting with ATG8B and ATG8E promoters and the TFs identified by DAP-seq as binding to these promoters, with respective p-values of 3.94E-12 and 7.31E-10 (Figure 2(d); Excel S2). Network analysis indicated that TFs from bZIP and NAC (NAC domain containing protein) families were enriched in both ATG8B and ATG8E screens, while WRKY family TFs were specifically enriched in interaction with the ATG8E promoter (Figures 3 and 4(a,b)). The reproducibility between 2 biological replicates was also increased with the new screening approach, being 88% and 82% respectively for ATG8B and ATG8E (Table S1).

Figure 3.

TF-promoter interaction network. The ATG8B and ATG8E promoters are indicated by yellow rectangles; interacting TFs are ovals color-coded based on different families. C2H2: Cys(2)His(2) zinc finger domain family; GeBP: GLABROUS1 enhancer binding protein family; HB: homeobox family; NAC: NAC-domain containing protein family; bZIP: basic leucine-zipper protein family; WRKY: WRKY DNA-binding protein family. The definitions and gene IDs are provided in Excel S1 for all the TFs shown in the figure.

Figure 4.

Enrichment of TF families and validation of Y1H candidates in Arabidopsis protoplasts. (a) Enrichment for TF families that bind to the ATG8B promoter. (b) Enrichment for TF families that bind to the ATG8E promoter. Families shown are statistically significant (p < 0.05), calculated using Fisher’s exact test. (c) Validation of Y1H candidates by transiently co-expressing each TF with an ATG8B promoter-luciferase fusion. (d) Validation of Y1H candidates by transiently co-expressing each TF with an ATG8E promoter-luciferase fusion. Candidates were selected based on TF family enrichment. The definitions and gene IDs are provided in Excel S1 for all the TFs shown in the figure. Dual-luciferase (REN and LUC) activities were monitored for each reaction. Data were normalized to a vector control for each promoter and shown as the average value ± SD of 3 biological replicates. Differences are significant at p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***) by Student’s t test.

To confirm the biological relevance of these TFs in plant cells, we performed a rapid validation assessment in Arabidopsis protoplasts. We chose representative TFs for validation based on TF family enrichment. A plasmid overexpressing a TF driven by a 35S promoter (35S::TF) and a plasmid carrying an ATG8 promoter-driven firefly luciferase (LUC) reporter construct were co-introduced into protoplasts for transient expression. The renilla luciferase (REN) reporter under a 35S promoter is an internal control in the same vector as the LUC reporter. Most of the TFs tested for each promoter significantly altered relative promoter activity, either activating or repressing LUC expression (Figure S3(c,d) and Figure 4(c,d)).

TGA9 binds to ATG8 promoters

We chose the bZIP family member TGA9 (AT1G08320) as an initial candidate for further analysis, as this TF binds to both ATG8B and ATG8E promoters, gives relatively strong activation of the promoters in protoplasts upon overexpression (Figure 4(c,d)), and has a characterized binding motif within promoters. TGA TFs bind to TGACG motifs [63], of which there are 1 and 2 respectively in the ATG8B and ATG8E promoter fragments used in the Y1H screen (Figure 5(a)). We mutated each TGA motif to GTAATG, and tested TGA9 binding via the Y1H assay. Reporter activation indicated that mutation of the TGA motif in the ATG8B promoter completely abolished LUC induction by TGA9, compared with the 66–fold induction in the wild-type (WT) control (Figure 5(b)). Similarly, compared with the 62–fold LUC induction in the WT control, mutation of an individual TGA motif in the ATG8E promoter caused a substantial reduction in LUC induction, while mutation of both motifs abolished LUC induction (Figure 5(b)). This suggests that TGA9 activated ATG8B or ATG8E expression through binding to the TGA motifs. To further analyze this binding, we expressed green fluorescence protein (GFP)-tagged TGA9 in mesophyll protoplasts (Figure S4(a,b)) and confirmed its localization to the nucleus (Figure S4(a)). We performed Chromatin Immunoprecipitation (ChIP) with anti-GFP antibody [64] and observed an approximately 5–fold enrichment for TGA9 binding to regions containing the TGA motifs in ATG8B and ATG8E promoters (Figure 5(c)), confirming that TGA9 binds to these promoters in plant cells.

Figure 5.

Characterization of the TGA9-ATG8B and TGA9-ATG8E promoter interaction. (a) Schematic diagram of the promoter regions (~2000 bp upstream of start codon) of ATG genes. TGA motifs (TGACG) are indicated by blue triangles, and their directions indicate sense or antisense strands. Red arrows are TSS. P represents primer pairs for ChIP-qPCR. Green boxes indicate fragments used for the Y1H screen. (b) Binding of TGA9 to promoter regions and mutated promoter regions of ATG8B and ATG8E in yeast. Bars represent the fold induction of β-galactosidase activity, shown as the average value ± SD of 3 biological replicates. (c) Binding of TGA9 to the promoter regions of ATG8B and ATG8E in Arabidopsis protoplasts. ChIP assays were performed with 35S::GFP-TGA9 and 35S::GFP-GUS as a control after transient expression in protoplasts. (d) Binding of TGA9 to the promoters of selected ATG genes in plants. ChIP assays were performed with 35S::GFP-TGA9 transgenic seedlings and WT (Col) as a control. For (C) and (D), sonicated chromatin was immunoprecipitated with either anti-GFP or anti-IgG antibody. Immunoprecipitated DNA was quantified by real-time quantitative PCR with primers specific for TGA-binding motifs in the promoters. Primers located at −3000 ~ −4000 bp upstream of the promoters were used as negative controls (pro-Neg) for (C). ACTIN2 is a negative control for (D). Fold induction was normalized to anti-IgG and wild-type controls. Data represent means ± SD from 3 biological replicates. For all panels, differences are significant at p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***) by Student’s t test.

To further analyze the function of TGA9 in plants, we generated 2 transgenic Arabidopsis lines (designated lines 4 and 7) expressing GFP-TGA9 under the constitutive 35S promoter. The GFP fusion was found in the nucleus in seedlings as expected, and expression of the full-length fusion protein was confirmed by immunoblot (Figure S4(c,d)). Analysis of the DAP-seq database identified a number of additional ATG genes that were potential targets of TGA9 [61]. We analyzed the promoters of these ATG genes, and found that each contained 1–3 TGA motifs within ~2000 bp upstream of the start codon (Figure 5(a)). We examined whether representative TGA motifs (Figure 5(a)) were bound by GFP-TGA9 in the transgenic seedlings via ChIP quantitative polymerase chain reaction (ChIP-qPCR) using anti-GFP antibody. Regions containing the TGA motifs in the promoters of ATG1A, ATG3, ATG5, ATG8A, ATG8B, ATG8E, ATG8F, ATG8G, ATG13B, ATG18A and ATG18H were all enriched to different degrees in TGA9 transgenic plants compared with non-transgenic controls, with the ATG8B promoter showing the highest enrichment (Figure 5(d)). Together, these data indicate that TGA9 binds to TGA motifs in the ATG8B and ATG8E promoters both in yeast and in plants. TGA9 can also bind to promoter regions containing TGA motifs of additional ATG genes, indicating that it may play a more general regulatory role in the autophagy pathway.

TGA9 activates ATG gene expression under both sucrose starvation and osmotic stress

To assess the function of TGA9 in autophagy, we employed 2 β-estradiol-inducible TGA9 overexpressing (OE) lines OE436 and OE438 [65], and a T-DNA insertion mutant. As tga9-1 and tga9-2 mutants have already been described [66], we named this mutant tga9-3. No full-length transcript was produced in the tga9-3 mutant, although a partial transcript upstream of the insertion site could be detected (Figure S5(b)). Expression of TGA9 was induced by β-estradiol 2.83– and 21.97–fold in OE436 and OE438 lines, respectively (Figure S5(c)).

We first analyzed expression of ATG8 family members and additional ATG genes in these lines by qPCR under sucrose starvation and mannitol-induced osmotic stress, commonly used stresses to activate autophagy in Arabidopsis [47,49,67]. For sucrose starvation, the OE438 line showed increased expression of ATG8B under both control and starvation conditions, and increased ATG8E expression only under control conditions. OE436 had increased ATG8B expression after starvation. A decrease in ATG8E expression was seen in the tga9-3 mutant but no change was seen in ATG8B expression (Figure 6(a)). Expression analysis for additional ATG genes showed that the more highly expressing line OE438 also significantly upregulated ATG8A, ATG8F, ATG8G, ATG1A, ATG5, ATG13B and ATG18H, while most of the tested ATG genes were not affected in the tga9-3 mutant, with the exceptions of ATG1A, ATG3 and ATG13B (Figure 6(a)). After osmotic stress, OE436 and OE438 lines had increased expression of the majority of the tested ATG genes, with OE438 having stronger activation, in accordance with the stronger TGA9 expression compared with OE436 (Figure 7(a)). Similarly, 4 ATG genes (ATG8B, ATG8D, ATG8G and ATG1A) were down-regulated in the tga9-3 mutant, with others being unchanged (Figure 7(a)). These expression data suggest that TGA9 overexpression upregulates ATG8B, ATG8E and additional ATG genes, and the degree of upregulation correlates with TGA9 expression. However, reduction of TGA9 expression in the tga9-3 mutant had a more limited effect on ATG gene expression, potentially due to redundancy with other TGA family members.

Figure 6.

Activation of autophagy by TGA9 under sucrose starvation. (a) Expression of selected ATG genes in 2 TGA9 OE lines and the tga9-3 mutant after sucrose starvation for 3 d. (b) MDC-labeled autophagosomes in roots of seedlings of the indicated genotypes under control or sucrose starvation conditions. White arrows indicate MDC-labeled puncta. Scale bar: 20 μm. (c) Quantified autophagosome numbers per unit area in the seedlings shown in (B) and shown as means ± SD from 3 biological replicates, with 10–30 images per replicate. (d) mCherry-ATG8E-labeled autophagosomes in mesophyll protoplasts from the indicated genotypes with or without TGA9 expression under control or sucrose starvation conditions. White arrows indicate mCherry-ATG8E-labeled autophagosomes. Scale bar: 10 μm. (e) The percentage of protoplasts with 3 or more mCherry-ATG8E-labeled autophagosomes in samples from (D). Bars indicate means ± SD from 3 biological replicates, with 100 protoplasts per sample per replicate. For all panels, different letters indicate significant differences at p < 0.05 by Student’s t test.

Figure 7.

Activation of autophagy by TGA9 under mannitol-induced osmotic stress. (a) Expression of selected ATG genes in 2 TGA9 OE lines and the tga9-3 mutant after osmotic stress for 6 h. (b) MDC-labeled autophagosomes in roots of seedlings of the indicated genotypes under control or osmotic stress conditions. White arrows indicate MDC-labeled puncta. Scale bar: 20 μm. (c) Quantified autophagosome numbers per unit area in the seedlings shown in (B) and shown as means ± SD from 3 biological replicates, with 10–30 images per replicate. (d) mCherry-ATG8E-labeled autophagosomes in mesophyll protoplasts from the indicated genotypes with or without TGA9 expression under control or osmotic stress. White arrows indicate mCherry-ATG8E-labeled autophagosomes. Scale bar: 10 μm. (e) The percentage of protoplasts with 3 or more mCherry-ATG8E-labeled autophagosomes in samples from (D). Bars indicate means ± SD from 3 biological replicates, with 100 protoplasts per sample per replicate. For all panels, different letters indicate significant differences at p < 0.05 by Student’s t test.

TGA9 activates autophagy under both sucrose starvation and osmotic stress

Next, we assessed the effect of increased and decreased TGA9 expression on autophagy under sucrose starvation and osmotic stress. Initially, autophagy was detected by staining roots with monodansylcadaverine (MDC) to fluorescently label acidic vesicles, primarily autophagosomes, followed by counting labeled puncta [57]. As expected, an increase in MDC-labeled puncta was seen in all genotypes under starvation (Figure 6(b)) and osmotic stress (Figure 7(b)). OE436 and OE438 lines had a more than 2–fold higher basal level of autophagy than WT under control conditions (Figures 6(c) and 7(c)), and after stress treatment the autophagosome numbers were also higher in the overexpression lines than in WT. In the tga9-3 mutant, no change was seen in autophagosome numbers under control conditions, but a significant reduction in autophagy activation by stress was observed (Figures 6(c) and 7(c)).

To confirm the effect of TGA9 on autophagy, TGA9 was overexpressed by transient expression in protoplasts from WT and tga9-3 mutant plants, or by β-estradiol induction in OE436 and OE438 lines, with co-expression of mCherry-ATG8E as an autophagosome marker [68]. Transfected protoplasts were subjected to either sucrose starvation or mannitol treatment. Both stresses induced autophagosome formation in WT with vector control samples. Transient overexpression of TGA9 significantly increased the percentage of protoplasts with active autophagy by 2–fold under non-stress control conditions, and upon stress treatment. Autophagy induction by stress was significantly reduced in tga9-3 mutant protoplasts, and expression of TGA9 in the tga9-3 mutant protoplasts rescued this phenotype (Figures 6(d,e) and 7(d,e)), confirming that it is due to loss of TGA9. β-estradiol-induced TGA9 expression also significantly activated autophagy under both control and stress conditions (Figures 6(d,e) and 7(d,e)). These results indicate that TGA9 overexpression significantly increased the level of autophagy under both control and stress conditions, while the extent of autophagy activation by stress was reduced in the tga9-3 mutant.

Since the T-DNA insertion in tga9-3 is located in the sixth intron (Figure S5(a)) and the insertion did not completely block autophagy (Figures 6(b-e) and 7(b-e)), we confirmed this phenotype using an additional T-DNA insertion line, tga9-2 [66]. The T-DNA insertion in tga9-2 is located in the sixth exon, within the bZIP domain (Figure S5(a)). Similar to the tga9-3 mutant, transcript corresponding to the region upstream of the T-DNA insertion could be detected but the full-length transcript was not present (Figure S5(b)). We measured autophagy activation in the tga9-2 mutant in both sucrose starvation and osmotic stress. Both MDC staining and transient expression of the mCherry-ATG8E autophagosome marker in protoplasts indicated significantly reduced activation of autophagy under both stress conditions compared with that in WT plants, and this phenotype could be complemented by introduction of a TGA9 transgene (Figure S6). Both tga9-2 and tga9-3 mutants therefore have a similar phenotype with respect to autophagy.

To test whether the activation of autophagy by TGA9 is dependent on the autophagy core machinery, we transiently overexpressed TGA9 in atg5 mutant protoplasts, and subjected them to either sucrose starvation or osmotic stress. Autophagy was not activated under any condition in the atg5 protoplasts and overexpression of TGA9 had no effect, indicating that TGA9-induced autophagy is dependent on the ATG5 gene (Figure 8(a,b)). Consistent with the observed autophagy occurring via the canonical machinery, the autophagy inhibitor 3-MA, which blocks the formation of autophagosomes via the inhibition of class III PtdIns3K [69], reduced the autophagy activity in protoplasts overexpressing TGA9 by about 50% in both control and sucrose starvation conditions (Figure 8(c)). Similar effects were observed in protoplasts expressing TGA9 induced by β-estradiol in the OE438 line (Figure 8(d)). Similarly, 5 mM 3-MA strongly repressed the activation of autophagy in both WT and OE438 seedlings under sucrose starvation as seen by MDC staining (Figure 8(e)).

Figure 8.

Activation of autophagy by TGA9 overexpression is compromised in atg5 mutant protoplasts or by 3-MA. (a, b) TGA9 was transiently overexpressed in WT and atg5 mutant protoplasts, with co-expression of mCherry-ATG8E, followed by sucrose starvation for 2 d (a) or osmotic stress for 6 h (b). The percentage of protoplasts with 3 or more mCherry-ATG8E-labeled autophagosomes was determined by epifluorescence microscopy. (c) The percentage of protoplasts with 3 or more mCherry-ATG8E-labeled autophagosomes when transiently overexpressing TGA9 in WT with or without 5 mM 3-MA under sucrose starvation for 2 d. (d) The percentage of protoplasts with 3 or more mCherry-ATG8E-labeled autophagosomes when TGA9 expression was induced by β-estradiol in OE438 with or without 5 mM 3-MA under sucrose starvation for 2 d. For A-D, bars indicate means ± SD from 3 biological replicates, with 100 protoplasts per sample per replicate. (e) Quantified MDC-labeled puncta per unit area in seedlings of WT and OE438 with or without 5 mM 3-MA after sucrose starvation for 3 d. Data are shown as means ± SD from 3 biological replicates, with 10–30 images per replicate. For all panels, different letters indicate significant differences at p < 0.05 by Student’s t test.

As autophagy is a critical process in plant tolerance of nutrient deficiency, we assessed the extent to which TGA9 is important for tolerance to long-term sucrose starvation. 4-day-old seedlings from the indicated genotypes were placed in the dark in the absence of sucrose for the indicated number of days, and then were placed back in the light for a 7-day-recovery period to assess survival. OE438 seedlings had significantly increased survival compared with WT after 12 d starvation, while the tga9-2 and tga9-3 mutants appeared similar to WT (Figure 9(a,b), and data not shown). These results suggested that plants overexpressing TGA9 can tolerate a longer starvation period than WT, potentially due to an increase in autophagy. We then examined whether this starvation tolerance phenotype could be suppressed by 3-MA, and is therefore likely to be related to autophagy. 3-MA significantly reduced the percentage of surviving seedlings for each line after 8 d starvation, while all lines failed to recover after 9 d starvation (Figure 9(c,d)). The increased tolerance of the OE438 line to starvation was almost completely lost in the presence of 3-MA, with only a small difference observed between OE438, WT and tga9-2 (Figure 9(c,d)). An atg5 mutant was included as a control; survival was dramatically reduced in this mutant after 8 d starvation even in the absence of 3-MA, and less than 10% atg5 seedlings recovered after 9 d starvation, at which time WT, OE438 and tga9-2 almost fully recovered (Figure 9(c,d)). The atg5 mutant was still somewhat sensitive to 3-MA, as also reported in mammalian cells [70], which could potentially point to additional effects of the chemical. However, this would not account for the loss of the starvation tolerance of the OE438 line in the presence of 3-MA (Figure 9(b) compared with Figure 9(d)). Our results suggest that the effect of TGA9 overexpression on starvation tolerance is at least partially dependent on an active autophagy pathway.

Figure 9.

Effects of alterations in TGA9 expression on survival after long-term sucrose starvation. (a) Seedling phenotype after long-term sucrose starvation in the dark followed by 7 d recovery in the light. (b) Percentage of surviving seedlings from (A). (c) Seedling phenotype after 0 d, 8 d and 9 d sucrose starvation with or without 2 mM 3-MA in the dark followed by 7 d recovery in light. (d) Percentage of surviving seedlings from (C). For all panels, seedlings remaining green or with new growth emerging are considered as surviving. Data indicate means from 3 biological replicates and error bars indicate SD. Differences are significant at p < 0.05 (*) by Student’s t test.

Discussion

Autophagy is a conserved catabolic process that directs degradation of cytoplasmic material in vacuoles or lysosomes. Most of the core machinery of autophagy was identified with the discovery of the ATG genes, while the regulation of these genes remains less well defined. Transcriptional regulation of autophagy-related genes is critical to either activate or repress autophagy in mammalian cells and yeast [35,71,72]. However, the transcriptional regulation of plant autophagy is still largely unexplored. High-throughput Y1H assays are promoter-centered and can provide potential interactors for a genomic region of interest [51]. Traditional Y1H screens using a cDNA library built from particular tissues or plants grown under specific conditions can be limited in the identification of some types of TFs due to low expression under those particular conditions [60]. We took advantage of a TF-centered Y1H screen, using TF-specific libraries instead of cDNA libraries, to identify TF candidates that transcriptionally regulate Arabidopsis ATG8s. Gene ontology (GO) term enrichment analysis for TF candidates for ATG8A and ATG8H promoters mainly identified developmental processes such as stamen and carpel formation, shoot and root meristem growth, and xylem cell differentiation (Table S2). This is in accordance with previous observations of high-level expression of ATG8A and ATG8H in floral organs [41], implying a role for autophagy during floral organogenesis or rapid senescence after fertilization. TF candidates for ATG8B and ATG8E promoters also participate in developmental processes include floral organ identity, meristem cell division and root development. However, GO term analysis also indicated enrichment in biological processes responsive to stresses and hormones (Table S2). For example, these include bZIP family members from both ATG8B and ATG8E promoter screens involved in salicylic acid-mediated systemic acquired resistance, cellular response to glucose stimulus, and abscisic acid-activated signaling pathway. The 4 promoters share 19 TFs in common, all of which are related to developmental processes, indicating a conserved role of ATG8 family members in maintaining basal autophagy levels during development. By contrast, the TFs that differ between promoters are consistent with the differential regulation of ATG8 genes in response to various stresses.

In plants, bZIP TFs play important regulatory roles in pathogen defense, light and stress signaling, seed maturation and flower development [73]. The TGA motif-binding subclade contains 10 members: TGA1-TGA7, PERIANTHIA (PAN), and TGA9-TGA10 [73]. Seven of them (TGA1-TGA7) function in plant defense through NPR1 (nonexpresser of PR genes 1)-dependent or -independent pathways [74-78]. A subset (TGA2, TGA5 and TGA6) are involved in the activation of a general broad-spectrum detoxification network upon xenobiotic stress [79]. TGA1, TGA3–7 and TGA9 were identified in our screens, and most of them activate ATG8B and ATG8E upon transient expression. TGAs may therefore regulate autophagy during plant pathogen defense or programmed cell death (PCD). One function of autophagy during pathogen invasion is an NPR1-dependent cytoprotective role, in which autophagy rescues uninfected cells from unrestricted runaway cell death [80-82]. The precise role of defense-related TGAs as potential regulators of autophagy in disease resistance remains to be determined.

Among the TGA candidates, TGA9 shows strong activation of ATG8B and ATG8E expression in a transient assay and positively regulates autophagy during either control or stress conditions. PAN and TGA9/10 control flower developmental processes through nuclear interaction with CC-type glutaredoxin family proteins ROXY1/AT3G02000 or ROXY2/AT5G14070 [66,83]. TGA9 contributes to tapetal development and anther dehiscence, processes involving PCD and degradation [66]. Autophagy functions in rice anther development and tapetal degradation during pollen maturation [84,85], processes vital for normal pollen development. Transcriptional control by several key TFs and involvement of reactive oxygen species (ROS) have been reported to play regulatory roles in tapetal PCD in rice and Arabidopsis [86-88]. Unlike rice autophagy mutants, Arabidopsis atg mutants have normal pollen development [84]. However, this does not rule out the involvement of autophagy in the regulation of pollen development via an upstream transcriptional network, since we have identified several TFs (e.g., TGA9 [66], JAG (JAGGED) [89], ANT (aintegumenta) [90], PDF2 (protodermal factor 2) [91], HDG2 (homeodomain glabrous 2) [91], ASL1 (asymmetric leaves 2-like 1) [92]) involved in flower development.

Although there are no prior reports of a function for TGA9 in sucrose starvation or osmotic stress responses, genetic redundancy likely masks such roles [66]. TGA9 is redox-regulated [66] and is involved in ROS-mediated responses to bacterial flg22 (flagellin 22 peptide) [93], suggesting that TGA9 may participate in stress responses. We provide evidence that TGA9 positively regulates responses to sucrose starvation and osmotic stress through activating autophagy, probably due to its regulation of ATG gene expression. DAP-seq data showed that a subset of ATG genes were among the targets of TGA9, including ATG1A, ATG1C, ATG2, ATG3, ATG4A, ATG5, ATG8B, ATG8E, ATG8G, ATG9, ATG11, ATG13B, ATG18A and ATG18H. Most of their promoters contain 1–3 TGA motifs within 2 kb upstream of the start codon. We tested a number of these and found that the TGA motifs in their promoters were enriched in ChIP-qPCR from GFP-TGA9 expressing seedlings compared with WT, and the genes were upregulated transcriptionally upon overexpression of TGA9. Overexpression of TGA9 leads to strong activation of autophagy under both control and stress conditions, both upon transient expression and in transgenic seedlings. The activation of autophagy by stresses is significantly repressed in tga9-2 and tga9-3 mutants, but is not completely blocked. In addition, the mutants had a mild impact on ATG gene expression, and similar responses to long-term sucrose starvation compared with WT. It is possible that the partial transcripts detected in both the tga9-2 and tga9-3 mutants produce truncated proteins that retain some function. However, the T-DNA insertions are within the bZIP domain and we therefore consider it more likely that the weak phenotype is due to redundancy with other family members or to a role for TGA9 in modulating the extent of autophagy rather than being absolutely required for activation of autophagy. Identifying the partners of TGA9 and exploring the mechanisms of binding to promoters will help further define its role in regulating autophagy.

Materials and methods

Plant materials and growth conditions

Arabidopsis thaliana seeds of WT (Col-0), OE436 (ABRC (Arabidopsis Biological Resource Center), CS2101436), OE438 (ABRC, CS2101438), atg5 [41] and tga9 mutant (ABRC, tga9-2: Salk_091349; tga9-3: Salk_141618C) genotypes were sterilized with 25% (v:v) household bleach and 0.1% (v:v) Triton X-100 (Fisher Scientific, BP151) for 20 min, followed by 5 washes with sterile water. Sterilized seeds were sown on ½ strength MS solid medium (Murashige and Skoog vitamin and salt mixture [Caisson Laboratories, MSP01], 0.5% [w:v] Sucrose [Sigma-Aldrich, S0389], 2.4 mM MES [Sigma-Aldrich, M3671], pH 5.7, and 0.6% [w:v] phytoagar [Caisson Laboratories, PTP01]) and stratified at 4°C for 2–3 d before transfer to long-day (LD) conditions (16 h light/8 h dark) at 22°C. For β-estradiol inducible lines, 4-day-old seedlings were transferred to ½ strength MS medium supplemented with 10 µM β-estradiol (Sigma-Aldrich, E8875) for 3 d to induce gene expression before any treatment. Adult plants for protoplasts were grown in soil in a growth chamber with 50% humidity at 20–23°C under LD conditions.

Plasmid construction and transgenic plant generation

The promoter sequences (Figure 2(a) and Figure S1(a)) of 4 ATG8 genes were amplified and cloned into the pGreenII 0800-LUC double reporter (LUC and REN) vector [94]. To generate the constructs for use in Y1H, smaller fragments (300–660 bp, see Figure 2(a) and Figure S1(a)) were subsequently cloned into the pLacZi vector. Mutations in promoter regions were generated by PCR-based mutagenesis. All primers used for constructs are listed in Table S3.

The coding sequences for each TF used in this study were collected from the TF collection [60] and subsequently cloned into the binary pGWB412 (N-FLAG tag) or pGWB406 (N-GFP tag) destination vector [95] using BP and LR Gateway reactions for overexpression in protoplasts or plants.

A construct containing GFP-TGA9 driven by the 35S promoter in the pGWB406 vector was introduced into Agrobacterium tumefaciens strain GV3101, which was used to transform Arabidopsis plants (Col-0) by the floral dip method [96]. Transgenic lines were selected on ½ strength MS solid medium with 50 mg L⁻¹ kanamycin (Fisher Scientific, BP906-5). Transgene expression was analyzed by immunoblotting using anti-GFP antibody, and expression of GFP was also assessed in seedlings of transgenic lines by confocal microscopy.

Y1H screens and analysis

The pLacZi constructs with different promoter fragments were linearized with NcoI and used to transform the yeast YM4271 strain (mating type ‘a’) using the LiAc method [60]. The library of 1956 TFs was introduced into the yeast Yα1867 strain (mating type ‘α’) [97] in a 96-well format and several glycerol stock copies were stored at −80°C. High-throughput Y1H screens were performed by mating YM4271-promoter strains (MATa) with Yα1867-TF strains (MATα) according to published methods [51,60]. After mating, the diploid cells were cultured in SD-Trp-Ura medium for 2 d. The optical density (OD) was read as absorbance at 600 nm using a multi-mode plate reader (Eppendorf AF2200, Hamburg, Germany). β-galactosidase activity was assessed using a commercially-available luminescent β-galactosidase substrate Beta-Glo (Promega, E4740), which is cleaved to release D-luciferin as a firefly luciferase substrate [98]. The LUC values were first normalized to the culture absorbance value at 600 nm and the ratio was normalized to the value obtained from the control pDEST22 for each plate. Binding threshold cut-off values for each promoter fragment were set at 2-fold above the reporter activity of the pDEST22 empty plasmid control, as described in published screens [99]. For each promoter fragment, 2 biological replicate screens were performed. Afterwards, the candidates were selected and mated with the promoter strains for further confirmation, with another 3 biological repeats. Finally, candidates that passed the threshold cutoff in at least 3 replicates were considered as high-confidence in this study.

To determine the enrichment by family for each promoter fragment, the total number of hits in a TF family that passed the threshold cutoff was compared to the total number of TFs in that family that are present in the TF collection library. Enrichment was calculated using Fisher’s exact test. GO enrichment analysis was performed online according to the Gene Ontology Consortium (http://geneontology.org/page/go-enrichment-analysis). The TF-ATG8 promoter network was visualized using Cytoscape version 3.6.0 [100].

Comparison of Y1H data with DAP-seq

To compare Y1H hits with DAP-seq TF binding data, DAP-seq target genes (fraction of reads in peaks (FRiP) ≥5%) [61] for 349 TFs were downloaded (http://neomorph.salk.edu/dev/pages/shhuang/dap_web/pages/index.php). 336 of these TFs were present in the Y1H TF library and used for subsequent analysis. The GeneOverlap package (version 1.16.0) in R (version 3.4.1) was used to compare TF lists from individual ATG8 promoters with TFs binding the corresponding ATG8 promoters. Additionally, a combined list of TFs binding to any of the 9 ATG8 promoters in the DAP-seq data was generated and used for comparisons. Statistical significance of the intersection between TF lists from the Y1H and DAP-seq data was calculated using Fisher’s exact test in GeneOverlap. The resulting matrix of p-values was – log10 transformed and plotted using the ComplexHeatmap R package (version 1.14.0) with the row and column orders determined by hierarchical clustering [101].

Transient assays in Arabidopsis protoplasts

Leaves from 4-week-old Arabidopsis plants grown under LD conditions were collected for protoplast isolation. The ‘Tape-Arabidopsis Sandwich’ technique was employed to peel leaves [102], and the remaining procedures followed a previously described protocol [103]. Protoplasts were resuspended to a final concentration of 3.0–3.5 × 10⁵ ml⁻¹ in MMg solution (400 mM mannitol [Sigma-Aldrich, M9647], 15 mM MgCl₂ [Sigma-Aldrich, M4880], 4 mM MES, pH 5.7).

Plasmid DNA was prepared using the GenElute^TM Plasmid Midiprep (Sigma-Aldrich, NA0200) or Maxiprep kits (Sigma-Aldrich, NA0310) according to the manufacturer’s instructions, and final DNA concentration was adjusted to 1 µg per 4 kb of DNA per µl. For the dual-luciferase assay, small-scale transformation was carried out in a sterile round-bottom 96-well plate. 3 µg TF-overexpression or control DNA and 3 µg pGreenII 0800-LUC double reporter DNA were introduced into 40 µl protoplasts by adding 46 µl of PEG solution (40% PEG4000 [Sigma-Aldrich, 81240], 200 mM mannitol and 100 mM CaCl₂ [Sigma-Aldrich, C7902]). Protoplasts were mixed by gently pipetting 10 times with cut tips and incubated for 20 min at room temperature (RT). 150 µl W5 solution was added and mixed by pipetting 5 times to stop the transformation. Transfected protoplasts were collected by centrifugation at 100 x g for 2 min in a swinging bucket rotor and resuspended in 100 µl W5 solution (154 mM NaCl [Sigma-Aldrich, S3014], 125 mM CaCl₂, 5 mM KCl [Fisher Scientific, BP366], 2 mM MES, pH 5.7). After 16 h incubation in darkness at RT, protoplasts were collected and the dual-luciferase reporter assay system (Promega, E1910) was used to measure the activity of firefly LUC and renilla REN sequentially using a Berthold Centro LB960 luminometer with injectors (Berthold Technologies, Bad Wildbad, Germany). The ratio of LUC:REN was calculated and the relative ratio normalized to the control vector was used as the final measurement.

For autophagosome induction and observation, large-scale transformation was carried out in a sterile 1.5–ml centrifuge tube. 20 µg TF-overexpression or control vector and 20 µg mCherry-ATG8E plasmid DNA were introduced into 300 µl protoplasts by adding 340 µl of PEG solution. After transformation, protoplasts were washed and incubated in 2 ml of W5 solution. For sucrose starvation treatment, protoplasts were incubated in W5 solution without sucrose or with 0.5% (w:v) sucrose as control at RT for 2 d. For mannitol treatment, protoplasts were incubated in W5 solution with or without 0.35 M mannitol at RT for 6 h. For protoplasts isolated from TGA9 OE lines, 10 µM β-estradiol was added to the W5 solution to induce gene expression after transformation, with the same amount of ethanol for the control. When appropriate, 3-MA (Sigma-Aldrich, M9281) was added to the W5 solution to a final concentration of 5 mM; corresponding volumes of water were separately added as solvent controls. Protoplasts were observed by epifluorescence microscopy (Carl Zeiss Axio Imager.A2, Carl Zeiss, Jena, Germany) using a TRITC filter, and protoplasts with more than 3 visible autophagosomes were counted as active for autophagy [47]. A total of 100 protoplasts were observed per treatment per genotype, and the percentage of protoplasts with active autophagy was calculated and averaged from 3 independent experimental replicates.

Autophagy detection in Arabidopsis roots by MDC staining

For sucrose starvation of Arabidopsis seedlings of mutants or overexpressing lines, 7-day-old seedlings were transferred to ½ MS media with or without sucrose for an additional 3 d. At the same time, 3-MA was added to both the sucrose starvation and MS control medium to a 5 mM final concentration when appropriate. Sucrose starvation plates were kept in the dark. For mannitol treatment, 7-day-old seedlings were transferred to liquid ½ MS medium with 0.35 M mannitol for 6 h [47]. After treatment, roots were stained with 0.05 mM MDC (Sigma-Aldrich, 30432) for 15 min, and then washed 2 times with phosphate-buffered saline (8% [w:v] NaCl, 0.2% [w:v] KCl, 1.4% [w:v] Na₂HPO₄ [Sigma-Aldrich, S5136], 0.24% [w:v] KH₂PO₄ [Fisher Scientific, P285], pH7.4) to remove excess MDC [57]. MDC-stained seedlings were observed by epifluorescence microscopy (Carl Zeiss Axio Imager.A2, Carl Zeiss, Jena, Germany) using a 4ʹ,6-diamidino-2-phenylindole (DAPI)-specific filter. 10–30 representative photographs in the root elongation zone were photographed per treatment, and the number of fluorescent puncta in each image was counted and averaged. Three individual biological replicates were performed. Representative confocal microscopy images were taken using a Leica SP5 × MP confocal/multiphoton microscope system with a 63 × 1.4 oil immersion objective (Leica Microsystems, Wetzlar, Germany) with excitation and emission at 488 nm and 509 nm for GFP, and 575 nm and 650 nm for mCherry.

Stress treatment, RNA extraction and real-time qPCR

For nutrient stress induced by sucrose or nitrogen starvation, 7-day-old Arabidopsis seedlings were transferred onto solid ½ MS medium with sucrose (control), without sucrose (sucrose starvation) or without nitrogen (nitrogen starvation). For sucrose starvation, the plates were wrapped with aluminum foil and grown in the dark for 3 d. For control and nitrogen starvation, seedlings were grown in the light for another 3 d. For mannitol and salt treatments, 7-day-old seedlings were transferred to liquid ½ MS medium with 0.35 M mannitol or 0.16 M NaCl for 6 h [47]. Whole seedlings were sampled and frozen in liquid nitrogen for RNA extraction. Total RNA was extracted using an RNeasy Plant Mini Kit (QIAGEN, 74904), and genomic DNA contamination was removed using RNase-free DNase Set (QIAGEN, 79254) in column during RNA extraction according to the manufacturer’s protocols. The first strand cDNA was synthesized with an iScript^TM cDNA Synthesis Kit (BioRad, 1708891). Real-time PCR was performed using SYBR^TM Green PCR Master Mix (Applied Biosystems, 4309155) on the Stratagene Mx4000 Multiplex PCR System (Stratagene, La Jolla, CA, USA) under the following conditions: initial denaturation at 95°C for 3 min, followed by 40 cycles of PCR (denaturing at 95°C for 20 s; annealing at 54°C for 30 s; extension at 72°C for 20 s). The relative gene expression was determined by applying the 2^−∆∆CT method with ACTIN2 as endogenous control. All treatments were performed with 3 individual biological replicates. The primers used for qPCR are listed in Table S3.

Chromatin immunoprecipitation (ChIP)

ChIP using Arabidopsis mesophyll protoplasts expressing TGA9 fused with a GFP tag was performed as described previously [64] with some modifications. Transient expression was performed as described above and ten transformations combined into one replicate. After 16 h, expression of GFP was assessed by fluorescence microscopy and representative images were taken by confocal microscopy. Full-length protein expression was confirmed by immunoblotting using anti-GFP antibody. Protoplasts transiently expressing GFP-TGA9 were fixed in 1% formaldehyde (Fisher Scientific, F79-1) for 8 min, and then crosslinking quenched with 2 M glycine (Sigma-Aldrich, G8898) for 5 min. Protoplasts were collected by centrifugation at 500 x g for 5 min and washed twice with W5 buffer. The protoplast pellets were used to isolate nuclei and chromatin.

ChIP using GFP-TGA9 overexpressing line 7 seedlings was performed as previously described [104] with modifications. Briefly, 3 g of 2-week-old WT and transgenic seedlings were fixed in 1% formaldehyde for 15 min, and crosslinking quenched with 2 M glycine for 10 min. Samples were rinsed 3 times with water and ground in liquid nitrogen to a fine power, which was subsequently used to isolate nuclei and chromatin.

5 µg of anti-GFP antibodies (Invitrogen, A11122) or IgG (Millipore, 12–370) as control were used to immunoprecipitate chromatin, which was collected with 50 µl Dynabeads protein A (Invitrogen, 10001D). The chromatin was sheared with a Sonic Dismembrator (Fisher Scientific, Model 100). The enrichment of specific DNA sequences was examined by qPCR with primers from the indicated regions (Table S3). Fold enrichment was normalized to IgG and wild-type controls. Results were derived from 3 biological replicates.

Long-term sucrose starvation for survival tests

Four-day-old seedlings from each genotype were transferred to ½ MS medium without sucrose but containing 10 µM β-estradiol, with or without 2 mM 3-MA. Plates were wrapped with aluminum foil and grown in the dark for the indicated days. For each indicated time, plates were put back in the light for a 7-day-recovery period to assess survival. Seedlings with green leaves and new growth were considered as having survived. For each replicate, 25–40 seedlings were used for each genotype per treatment. Data were derived from 3 biological replicates.

Statistical analysis

All experiments in this study were performed with at least 3 biological replicates. Statistical analyses were performed using the Student’s t test or Fisher’s exact test. The data were expressed as the mean ± standard deviation (SD), and p < 0.05 were considered statistically significant.

Funding Statement

This work was supported by the National Institutes of Health under Grant number 1R01GM120316-01A1 to DCB and YY, and by the Walter E. and Helen Parke Loomis fund to DCB.

Acknowledgments

We thank Dr. Justin W Walley for pGWB406 and pGWB412 vectors, Chris Minion for the pDONR221 vector and Margaret Carter for assistance with confocal microscopy.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary data

Supplemental data for this article can be accessed here.