ATI1 (ATG8-interacting protein 1) and ATI2 define a plant starvation-induced reticulophagy pathway and serve as MSBP1/MAPR5 cargo receptors

ABSTRACT

Reticulophagy, the selective autophagy of endoplasmic reticulum (ER) components, is known to operate in eukaryotes from yeast and unicellular algae to animals and plants. Thus far, only ER-stress induced reticulophagy was reported and analyzed in plants. In this study we characterize a reticulophagy pathway in Arabidopsis thaliana that is triggered by dark-induced starvation but not by ER stress. This pathway is defined by the previously reported ATG8-interacting proteins, ATI1 and ATI2. We further identified the ER-localized MSBP1 (Membrane Steroid Binding Protein 1) as an ATI1- and ATI2-interacting protein and an autophagy cargo, and show that ATI1 and ATI2 serve as its cargo receptors. Together, these findings expand our knowledge on plant responses during energy deprivation and highlight the role of this special type of reticulophagy in this process.

Abbreviations: AGO1: ARGONAUTE 1; ATI: ATG8-Interacting Protein; BiFC: Bimolecular Fluorescence Complementation; BR: brassinosteroid; conA: concanamycin A; DMSO: dimethyl sulfoxid; DTT: dithiothreitol; ER: endoplasmic reticulum; GFP: green fluorescent protein; MAPR: Membrane-Associated Progesterone Binding Protein; MSBP: Membrane Steroid Binding Protein; SD: standard deviation; SE: standard error; TM: tunicamycin; TOR: target of rapamycin; Y2H: yeast two-hybrid.

Introduction

Macroautophagy (hereafter referred to as autophagy) is a conserved eukaryotic catabolic mechanism for the removal and recycling of damaged or unneeded cellular components in response to developmental or environmental cues. Autophagy involves sequestration of cellular cargo in de novo-formed double-membrane phagophores that finally close to form specialized vesicles called autophagosomes. The autophagosomes are then delivered to the lytic organelle, the vacuole in plants, for degradation [1,2]. Unlike the ubiquitin-proteasome system, autophagy can handle large protein complexes and aggregates, as well as non-proteinaceous cargo such as nucleic acids, lipid bodies, and entire organelles [3]. The autophagy process requires the regulated and sequential involvement of a set of AuTophaGy-related (ATG) core proteins that are mostly conserved across eukaryotic organisms [3–6]. Autophagy can be a highly regulated selective turnover mechanism, targeting specific cellular components under specific cellular conditions. Recognition and recruitment of specific cargo to autophagosomes are mediated by cargo receptors. These receptors can recognize and bind the cargo directly or indirectly, and tether it to the forming autophagosome/phagophore through interaction with core autophagy proteins [3,7–9]. The ubiquitin-like ATG8 protein family plays an important role in selective autophagy. Soluble ATG8 is conjugated to the membrane lipid phosphatidylethanolamine (PE) on the expanding phagophore and is involved in autophagosome formation, trafficking and fusion with the vacuole. Additionally, ATG8 proteins function as hubs for selective autophagy routes, as they interact with multiple autophagy cargo receptors, mainly through an ATG8-interacting motif (AIM) or a newly identified ubiquitin-interacting motif (UIM) found on the receptors [3,4,10,11]. In mammalian cells, multiple cargo receptors have been identified in recent years, allowing for greater mechanistic understanding of selective autophagy processes and their impact on cellular functions and pathologies [8,12–14]. However, plants lack orthologs to many of the mammalian cargo receptors, and although organelle-specific autophagy, as well as other types of selective autophagy were clearly demonstrated in plants, our mechanistic understanding of these processes and the selective cargo receptors that are required for them is still limited [3,15]. Examples of plant selective autophagy receptors include NBR1 (Next to BRCA1 gene 1), a functional hybrid of mammalian SQSTM1/p62 and NBR1 autophagy receptors, which targets ubiquitinated protein aggregates in plant stress responses [16,17] and virus particles [18], the proteasome subunit RPN10, which targets inactive ubiquitinated 26S proteasomes [19] and TSPO (Tryptophan-rich Sensory Protein) that can target plasma-membrane based aquaporins [20].

The endoplasmic reticulum (ER) is a complex membrane-bound compartment in eukaryotic cells, and its varied cellular functions require rapid adjustment of its shape, size and protein and lipid content in response to changing cellular needs. The involvement of the ER in autophagic processes is multifaceted. The ER is a site for the nucleation of phagophores and a source for autophagosomal membranes [21,22]. However, the ER is also the target of a selective type of autophagy, termed reticulophagy, that is induced in yeast, mammalian cells and plants in response to ER-stress and starvation [22–26]. The Atg8-interacting ER proteins Atg39 and Atg40 were identified as the cargo receptors that mediate reticulophagy in yeast. Atg39 localizes to the perinuclear ER, whereas Atg40 is enriched in the cortical and cytoplasmic ER, suggesting that each protein is needed for the degradation of a different ER subdomain [25]. In mammals, six ER-resident autophagy receptors were identified to date: RETREG1/FAM134B (reticulophagy regulator 1/family with sequence similarity 134, member B) [27], the long isoform of RTN3 (RTN3L) [28], CCPG1 (cell cycle progression 1) [29], SEC62 [30], ATL3 (atlastin GTPase 3) [31] and TEX264 (testis expressed gene 264) [32,33]. The different receptors seem to differ in their function and ER subdomain localization. FAM134B regulates starvation-induced turnover of ER sheets, while RTN3L and ATL3 are important for starvation-induced degradation of ER tubules [27,28,31]. CCPG1 and SEC62 are involved in ER stress-induced reticulophagy [29,30]. The newest addition to the cohort of mammalian reticulophagy receptors, TEX264, is ubiquitously expressed and was suggested to be a major contributor to starvation-induced reticulophagy [32,33]. In addition, CALCOCO1 and EPR1 have been recently identified as cytosolic reticulophagy receptors that associate with ER-resident proteins to recycle ER tubules [34,35].

In Arabidopsis thaliana, treatment of seedlings with the ER-stress inducing compounds tunicamycin (TM) and dithiothreitol (DTT) induces reticulophagy and delivery of ER fragments to the vacuole [24]. Induction of autophagy was also seen in Chlamydomonas exposed to TM and DTT treatments [36]. The ER stress regulator IRE1b (Inositol Requiring 1–1b) is required for reticulophagy induction in Arabidopsis. However, the known downstream target of IRE1b, the transcription factor BZIP60, is not required [24]. IRE1 was shown to participate in IRE1-dependent decay of mRNAs (RIDD), in which mRNAs are degraded by IRE1 upon ER stress. As some of these mRNAs were shown to inhibit autophagy upon overexpression, it was suggested that IRE1b might regulate the degradation of mRNAs that interfere with autophagy induction [37]. Recently, the Arabidopsis homolog of Sec62, ATSEC62, and the maize reticulon proteins RTN1 and RTN2 were identified as plant reticulophagy receptors that play a role in alleviating ER stress [38,39]. Additionally, an evolutionarily conserved cytosolic ATG8-binding protein, CDK5RAP3/C53, was shown to be activated by ribosome stalling and to play a role in ER stress tolerance in both Arabidopsis plants and HeLa cells [40]. However, our mechanistic understanding of reticulophagy in plants is still limited.

The plant-specific transmembrane ATI1 and ATI2 proteins (ATG8-Interacting Protein 1 and 2) were identified in a yeast two-hybrid screen for Arabidopsis ATG8f-interacting proteins [41]. They were shown to localize to dark-induced ER and chloroplast-associated bodies (ATI-bodies) that are transported to the vacuole [41,42]. Although ATI-bodies are distinct from autophagosomes, the delivery of the ATI1 proteins to the vacuole requires the autophagy machinery, suggesting that ATI-bodies eventually associate with autophagosomes [42]. ATI-bodies were shown to deliver chloroplast-targeted green fluorescent protein (GFP) to the vacuole and ATI1 can bind both stromal and membrane-bound chloroplast proteins [42]. Thus, it was suggested that the ATI proteins are chlorophagy cargo receptors [43]. However, their ER-related function is not clear. Recently, ATI1 and ATI2 were shown to be involved in an endogenous autophagic degradation pathway of ER-associated AGO1 (ARGONAUTE 1) that is significantly induced following expression of the viral suppressor of RNA silencing P0 protein [44]. As ATI1 directly interacts with AGO1, it was suggested that it might function as selective cargo receptor for ER-localized AGO1. In this study, we show that ATI1 and 2 define a novel type of dark-induced reticulophagy pathway in Arabidopsis. We further show that ER-localized MSBP1 (Membrane Steroid Binding Protein 1) interacts with the ATI proteins and is degraded by autophagy. Notably, ATI1 is required for the autophagic turnover of MSBP1, supporting the role of the ATI proteins as autophagy cargo receptors involved in selective reticulophagy.

Results

ATI1 is involved in reticulophagy in response to carbon starvation but not to ER-stress

To look at the possible involvement of the ATI proteins in reticulophagy, we used a transgenic line co-expressing an ER marker (mCherry-HDEL) and ATI1 fused to GFP (ATI1-GFP) [41]. Leaves of 4- to 5-week-old plants were treated with the V-ATPase inhibitor concanamycin A (conA), which stabilizes autophagic bodies by raising the vacuolar pH. The plants were either left under regular light conditions, darkened for 24 h, or treated with the glycosylation inhibitor tunicamycin (TM) to induce ER stress [45,46]. ATI1-labeled bodies or the ER marker were rarely observed in the vacuoles of leaves under regular light conditions (Figure 1A, upper panel). However, in accordance with previous results [41,42], numerous ATI1-labeled bodies were observed in the vacuole lumen after 24 h dark treatment (Figure 1A, middle panel). Puncta labeled by the ER marker were also clearly visible in the vacuole following dark and conA treatment, suggesting that reticulophagy took place under these conditions. Interestingly, following dark treatment, the ER marker was highly colocalized with ATI1 in the vacuole (Figure 1A, middle panel). Furthermore, the ER marker was surrounded by the membranal ATI1 that generated ring-like structures (Figure 1A, inset), suggesting that ATI1 is involved in reticulophagy in response to carbon starvation, and that ER components are delivered to the vacuole with ATI1. In contrast, although TM treatment increased the expression of several unfolded protein response (UPR) target genes as expected for induction of ER-stress (Figure 1B,C), it did not result in appreciable increase in ATI1-labeled or HDEL-labeled bodies (Figure 1A, lower panel).

Figure 1.

ER-derived cargo and ATI1 are co-delivered to leaf vacuoles in response to carbon starvation but not to ER stress. (A) Leaves of ATI1-GFP; mCherry-HDEL co-expressing plants were infiltrated with conA and either darkened (dark, middle panel) or treated with TM (TM, lower panel) for 24 h. Control plants were left under regular light conditions and treated with DMSO only (light, upper panel). Representative confocal images of the vacuolar focal plane of the leaves show ATI1-GFP in green and mCherry-HDEL in grayscale. In the overlay image (merged), colocalization of the ATI1-GFP green signal and the mCherry-HDEL red signal results in a yellow signal. Magnification of the area in the white rectangle is shown in the inset. Dark treatment results in the delivery of ATI1-labeled bodies carrying ER-marker to the vacuole . White arrowheads point to puncta where GFP and mCherry signals overlap. Chlorophyll autofluorescence is shown in magenta. Scale bars: 10 μm. (B and C) ATI1-GFP; mCherry-HDEL co-expressing plants were either darkened or left under regular light conditions, and their leaves were infiltrated with TM for 24 h. The upregulation of the well-established unfolded protein response target genes ATERDJ3A, ATERDJ3B (B) and BIP3 (C) was analyzed by quantitative RT-PCR. Data represent means ± SD for each condition (n = 3). The obtained data were normalized against the expression of RNA Helicase (AT1G58050.1) and compared to the light condition. (D) Representative blot of triplicate samples of leaves taken from plants treated as in B and C. The release of free GFP was monitored by analysis of total protein extracts with anti-GFP antibodies. Similar loading is shown by the stained level of ribulose bisphosphate carboxylase small subunit (loading). (E) Quantification of the ratio of free GFP to total GFP (free + fused to ATI1) based on the relative band intensities in the presented blot demonstrate that ATI1 is degraded via autophagy in response to dark but not to TM treatment. Bars represent the mean ± SE (n = 3)

To support our image-based results, we performed western blot-based autophagic flux analyses. Vacuolar degradation of proteins tagged with GFP was shown to result in loss of the tagged protein and release of the relatively stable free GFP [2,4]. Thus, increase in the free GFP: total GFP ratio is indicative of increased autophagic turnover of the tagged protein. Indeed, increased free GFP release from ATI1-GFP was observed following dark treatment of leaves (Figure 1D,E). However, no increase in free GFP: total GFP ratio was observed following incubation of leaves with TM (Figure 1D,E).

ER-stress was previously shown to trigger reticulophagy and delivery of HDEL-labeled bodies to the vacuoles of Arabidopsis root cells [24]. However, we did not observe delivery of the ER marker to the vacuoles of leaves upon TM treatment and induction of ER-stress (Figure 1A, lower panel). To further investigate this point, we repeated our imaging in roots of Arabidopsis seedlings. ATI1-GFP/mCherry-HDEL seedlings were transferred from solid growth media to liquid media and either darkened or treated with TM in the presence of conA. As expected, few ATI1-GFP or HDEL-labeled puncta were observed in the roots of conA-treated control seedlings (Figure 2A, upper panel). Upon dark treatment, many ATI1-labeled as well as HDEL-labeled bodies were observed in root cells. The ER marker was often colocalized with ATI1 (Figure 2A, middle panel), suggesting that similar to leaves, ATI1 is involved in carbon starvation-induced reticulophagy in the roots of Arabidopsis plants. Interestingly, in contrast to leaves and in agreement with previous studies [24], mCherry-HDEL-labeled puncta were clearly visible following TM treatment, supporting ER-stress induced reticulophagy in the roots (Figure 2A, lower panel). However, in roots as in leaves, there was no accumulation of ATI1-GFP bodies in response to ER-stress (Figure 2A, lower panel). Western blot analyses of ATI1-GFP and mCherry-HDEL autophagic turnover in seedlings supported the imaging results. As in leaves (Figure 1D), the free GFP: total GFP ratio significantly increased upon dark, but not TM treatment (Figure 2B, lower panel). The ratio of free mCherry: total mCherry also increased substantially in response to dark. The response to TM was more moderate, probably due to the fact that ER stress-induced reticulophagy occurred in roots, but not in green tissues (Figure 2B, upper panel). Taken together, our results demonstrate tissue-specific response to ER stress, and suggest that ATI1 is involved in carbon starvation-induced, but not in ER stress-induced, reticulophagy.

Figure 2.

ATI1 is not involved in ER-stress induced reticulophagy in the roots. (A) ATI1-GFP; mCherry-HDEL seedlings were either darkened for 24 h (dark, middle panel) or left under regular light conditions and TM added for the last 5 h (TM, lower panel). Control plants were left in the light and treated with DMSO only (light, upper panel). ConA was added for 5 h prior to imaging. Representative confocal images of the vacuolar focal plane of root elongation zone cells show ATI1-GFP in green and mCherry-HDEL in grayscale. In the overlay image (merged), colocalization of the ATI1-GFP green signal and the mCherry-HDEL red signal results in a yellow signal. Magnification of the area in the white rectangles is shown in the merged insets. Both carbon starvation and ER stress result in the delivery of ER marker to the vacuole. However, ATI1-labeled bodies are induced only following the dark treatment. Yellow arrows point to puncta where GFP and mCherry signals overlap, red arrows point to mCherry-HDEL puncta. Scale bars: 10 μm unless otherwise indicated. (B) Representative blots of total protein extracts from seedlings darkened or treated with TM as above and analyzed by anti-GFP or anti-mCherry antibodies. As in leaves, the autophagic flux of ATI1-GFP is increased upon dark, but not TM treatment. The autophagic flux of mCherry-HDEL is increased in response to both treatments. Quantification of the ratio of free mCherry to total mCherry (free + fused to HDEL) based on the relative band intensities is presented below the blot. Similar loading is shown by the stained level of ribulose bisphosphate carboxylase small subunit (loading)

The Target of Rapamycin (TOR) signaling pathway negatively regulates autophagy in yeast, mammals and plants [47]. Recently, TOR was shown to regulate autophagy induced by nutrient, salt or osmotic stress in Arabidopsis plants. However, autophagy induced by ER or oxidative stress was found not to be regulated by TOR signaling [48]. In accordance, treating seedlings stably expressing ATI1-GFP under its native promoter [42] with the active-site TOR inhibitor AZD8055 [49,50] increased the number of ATI-bodies in seedlings in the light, and even more so in the dark (Fig. S1). These results support the involvement of TOR signaling in the ATI1 autophagic pathway.

The ATI proteins interact with MSBP1

To identify possible ER membrane-related protein interactors of the ATI proteins, a split-ubiquitin yeast two-hybrid (Y2H) system was used. ATI2, as a representative, was fused to the C-terminal fragment of ubiquitin along with an artificial transcription factor, and used as bait against a commercial library of Arabidopsis cDNA fused to the modified N-terminal fragment of ubiquitin. Several positive cDNA clones were detected in the Y2H screen, and were further analyzed by sequencing of the corresponding plasmids. Among the positive cDNA clones, MAPR4 (Membrane Associated Progesterone Receptor 4) appeared 5 times. One on one Y2H assay verified that MAPR4 interacted with both ATI1 and ATI2 (Figure 3A). The MAPR family is widespread in eukaryotes, and its members were shown to have varied cellular functions. Arabidopsis contains four MAPR family members [51]. The best characterized is MSBP1 (MAPR5) that was shown to be involved in the signal transduction of the plant hormone brassinosteroid (BR) and in the redistribution of another major plant hormone, auxin. Moreover, MSBP1 was recently found to have a role in organizing and stabilizing cell wall-related lignin biosynthetic P450 enzymes on the ER [52–55]. Y2H assays demonstrated that similar to MAPR4, MSBP1 also interacts with both ATI1 and ATI2 (Figure 3B). To confirm these results, in vivo BiFC assays were used [56]. Fusion proteins linking MSBP1 with the N-terminal fragment of the marker Enhanced Yellow Fluorescent Protein (EYFP; MSBP1-nEYFP), and ATI1 or ATI2 with the C-terminal fragment of the EYFP protein (ATI-cEYFP), were transiently co-expressed in Nicotiana benthamiana leaves. Interaction between the ATI proteins and MSBP1 is expected to bring the two halves of EYFP to close proximity and generate yellow fluorescence. Indeed, strong EYFP fluorescence resulted from the co-expression of MSBP1-nEYFP and ATI1 or ATI2-cEYFP, confirming the interaction between MSBP1 and the ATI proteins (Figure 3C). Similar results were obtained with MAPR4 (Fig. S2A).

Figure 3.

MAPR4 and MSBP1 interact with both ATI1 and ATI2. Split-ubiquitin Y2H assay showing the interaction of MAPR4 (A) or MSBP1 (B) with ATI1 and ATI2. MAPR4 or MSBP were fused to the C-terminal fragment of ubiquitin and ATI1 or ATI2 was fused to the modified N-terminal fragment of ubiquitin. The large T antigen (LargeT) and the endogenous ER protein Alg5 were used as negative controls. While all bait and prey couples can grow on nonselective media (SD-LT), the interaction between the ATI proteins and MAPR4 or MSBP1 brings the two halves of ubiquitin together and allows growth on selective media lacking histidine and adenine (SD-LTHA). (C) Confocal imaging of a BiFC assay involving co-expression of MSBP1-nEYFP with either ATI1 or ATI2 fused to cEYFP in N. benthamiana leaves (top panel). The yellow fluorescence signal indicates an interaction between the two proteins. The co-expression of ATI1 or ATI2-cEYFP with unfused nEYFP (middle panel) or the co-expression of MSBP1-nEYFP with unfused cEYFP (bottom panel) did not show any fluorescence. Scale bars: 50 μm. (D) In vitro pull down with MSBP1 and the chloroplast protein 2CPA (2-Cys peroxiredoxin A) shows that both proteins can interact with the C-terminal region of ATI1. Bacterial lysates containing recombinant protein were mixed and pulled down with glutathione magnetic agarose beads. Bound proteins were visualized by immunoblotting with anti-HIS and anti-MBP antibodies. Input proteins were also visualized with anti-HIS antibodies

ATI1 and ATI2 are transmembrane proteins with a long N-terminal region that contains the functional ATG8-interacting motif (AIM), and a shorter C-terminal part that is predicted to be lumenal [41,57]. To better define the cargo-binding region of ATI, we performed in vitro pull downs using either the N- or C-terminal parts of ATI1. As shown before [57], ATI1 N-terminal region interacted with ATG8 (Fig. S2B), while the C-terminal region was required for the interaction with both MSBP1 and the previously identified chloroplastic cargo of ATI1, 2CPA (2-Cys Peroxiredoxin A) [42] (Figure 3D).

MSBP1 is degraded by autophagy

The interactions observed between MSBP1 and the ATI proteins suggest the involvement of autophagy in MSBP1 turnover. To further characterize the role of autophagy in MSBP1 degradation, we produced stable lines overexpressing mCherry-tagged MSBP1. Under regular growth conditions or in the absence of conA, very little mCherry signal was observed in the vacuoles of transgenic leaves. However, following dark incubation and treatment with conA, numerous MSBP1-mCherry-labeled puncta resembling autophagic bodies were clearly visible in the vacuole (Figure 4A). To confirm that the MSBP1-mCherry-labeled vacuolar puncta are autophagic bodies, we produced a line co-expressing MSBP1-mCherry with the GFP-ATG8f marker that decorates autophagosomes from their formation to their lytic destruction in the vacuole [2,58]. Upon dark and conA treatment, MSBP1-mCherry was indeed highly colocalized with GFP-ATG8f autophagic bodies in the vacuoles of both leaves and roots (Figure 4B and S3A). Furthermore, when MSBP1-mCherry was expressed in autophagy-deficient atg5-1 mutant plants [59], no MSBP1-mCherry-labeled puncta were observed under dark treatment and the addition of conA (Figure 4C). In agreement with the fluorescence microscopy results, the ratio of released free mCherry to MSBP1-mCherry substantially increased upon dark treatment in wild type plants, suggesting an increase in the autophagic flux. In contrast, very little free mCherry was observed in the atg5-1 mutants, even following dark treatment (Figure 4D). Taken together, our results suggest that during dark-induced starvation, MSBP1 is delivered to the vacuole for degradation by autophagy.

Figure 4.

MSBP1 is associated with autophagic bodies and is degraded by autophagy. (A) Representative confocal images of the vacuolar focal plane of leaves from plants stably expressing MSBP1-mCherry that were transferred for 16 h to liquid medium supplemented with sucrose in the light, or without sucrose in the dark with or without treatment with conA. MSBP1-mCherry-labeled puncta accumulate in the vacuole following dark treatment in the presence of conA. Scale bar: 10 μm. (B) Representative confocal images of the vacuolar focal plane of leaves stably co-expressing MSBP1-mCherry and GFP-ATG8f following 16 h of dark and treatment with ConA. MSBP1-mCherry colocalizes with GFP-ATG8f in autophagic bodies. Scale bars: 10 μm. (C) Representative confocal images of the vacuolar focal plane of leaves from plants stably expressing MSBP1-mCherry in either Col-0 or atg5-1 mutant backgrounds that were incubated for 16 h in the dark and treated with conA as in B. MSBP1-mCherry-labeled autophagic bodies do not accumulate in the vacuoles of atg5-1 mutant leaves. Scale bar: 25 μm. (D) Leaves from plants stably expressing MSBP1-mCherry in either Col-0 or atg5-1 mutant backgrounds were darkened for 16 h, and release of free mCherry was monitored by analysis of total protein extracts with anti-mCherry antibodies. A representative blot is presented. The ratio of free mCherry to total mCherry (free + fused to MSBP1) increased in the Col-0, but not in atg5-1 background, following dark treatment. Similar loading is shown by the stained level of RBCL (ribulose bisphosphate carboxylase large subunit; loading). (E) Phenotypes of MSBP1-mCherry-overexpressing lines in the background of Col-0 wild type plants or the atg5-1 mutant. The phenotype of MSBP1 overexpression is exacerbated in the atg5-1 background. Scale bars: 1 cm. (F) Overexpression of MAPR3/MSBP2 and MAPR4, but not MAPR2 shows early senescence phenotype. Scale bars: 2 cm

Interestingly, the atg5-1 autophagy-deficient mutant background exacerbated the phenotype of MSBP1 overexpression. As previously observed, overexpression of MSBP1 in wild type plants led to reduced cell expansion resulting in smaller leaves [53,55]. The overexpressing plants also showed early senescence phenotype (Figure 4E). Similar phenotypes were observed in plants overexpressing untagged MSBP1 (Fig. S3B), and the severity of the phenotype was correlated with the level of overexpression of MSBP1 (Fig. S3C). When MSBP1-mCherry was overexpressed in the atg5-1 autophagy-deficient mutant, the observed phenotype was much more severe, and most plants died before fruiting (Figure 4E).

Among the four members of the Arabidopsis MAPR family, MSBP1, MAPR3/MSBP2 and MAPR4, but not MAPR2, are predicted to have a transmembrane domain at the N-terminal of the protein [51]. Similar to MSBP1, plants overexpressing mCherry-tagged MAPR3/MSBP2 and MAPR4 had smaller leaves and showed early senescence phenotype, and the early senescence phenotype was aggravated in the background of the atg5-1 mutant (Figure 4F). Similar phenotypes were not observed in plants overexpressing MAPR2, though the levels of the expressed proteins were similar in the different lines (Figure 4F and S4B), suggesting a different function for the non-membranal member of the family.

Similar to MSBP1, dark treatment induced accumulation of mCherry-labeled puncta in the leaf vacuoles of MAPR2-, MAPR3/MSBP2- and MAPR4-mCherry-expressing plants, while no fluorescence was observed in the vacuoles of mCherry-MAPR-expressing atg5 mutants (Fig. S4A). Autophagic flux analysis supported the conclusion that other MAPR family members, including the non-membranal MAPR2, are targeted by autophagy (Fig. S4B).

MSBP1 is ER localized and is delivered to the vacuole with an ER marker

A recent study demonstrated that MSBP1 as well as MAPR3/MSBP2 are predominantly localized on the ER [52]. Indeed, when observed under a confocal fluorescence microscope, mCherry-labeled MSBP1 showed a reticulated pattern that colocalized with the ER marker GFP-HDEL, confirming its ER localization (Figure 5A, left panels). Following carbon starvation and conA application, MSBP1 also colocalized with the ER marker in autophagic bodies found within the vacuole lumen, suggesting that it is degraded by dark-induced reticulophagy (Figure 5A, right panels).

Figure 5.

MSBP1-mCherry colocalizes with GFP-HDEL and ATI1-GFP in the ER and vacuoles. (A) Representative confocal images of root cells of seedlings stably co-expressing MSBP1-mCherry and GFP-HDEL show MSBP1-mCherry in red, GFP-HDEL in green and the overlay indicating colocalization in yellow (merged). 7-d-old seedlings were transferred to liquid medium and incubated 16 h in the dark without sucrose. MSBP1-mCherry colocalizes with GFP-HDEL on the ER (left panel), and following treatment with ConA also in vacuolar puncta (right panel). (B) Following similar treatment of seedlings stably co-expressing ATI1-GFP and MSBP1-mCherry ATI1 (in green) is colocalized with MSBP1 (in red) on the ER (left panel) and on puncta in the vacuole (right panel). Magnification of the area in the white rectangle is shown on the right. Scale bars: 5 μm

ATI1 is required for the autophagic degradation of MSBP1

As MSBP1 interacts with the ATI proteins (Figure 3) and both are targeted by reticulophagy in response to carbon starvation (Figures 1 and 5A), we wanted to look at the involvement of the ATI proteins in MSBP1 autophagic degradation. MSBP1-mCherry-expressing lines were crossed with ATI1-GFP- or ATI2-GFP-expressing lines [41]. Seedlings co-expressing MSBP1-mCherry and ATI1- or ATI2-GFP were moved to growth medium without sucrose and kept in the dark for 24 h. Under these conditions, MSBP1-mCherry colocalized with both ATI1- and ATI2-labeled bodies in the cytosol (Figure 5B and S5, left panels). Following treatment with conA, MSBP1 colocalized with ATI1 and ATI2 in the vacuole as well (Figure 5B and S5, right panels). Furthermore, bodies co-labeled by ATI1 proteins and MSBP1 could be seen moving along the ER network and then entering the vacuole (Movie S1). To examine whether the ATI proteins are required for MSBP1 autophagic degradation, we produced a double knockout mutant of ATI1 and ATI2. CRISPR-Cas9-based technology was used to introduce an ati2 knockout mutation (as described in [44]) in the background of a homozygous SALK T-DNA insertion line (SALK_136644). MSBP1-mCherry was then expressed in the ati1;ati2 mutant line. While numerous MSBP1-mCherry-labeled autophagic bodies were visible in the vacuole of WT (Col-0) plants under carbon starvation and conA treatment, such bodies were significantly less visible in the ati1;ati2 background (Figure 6A, upper panel and Figure 6B), suggesting that the ATI proteins are required for efficient autophagic degradation of MSBP1. In contrast, there was no significant difference in the number of MAPR2-mCherry-labeled autophagic bodies in the ati1;ati2 background (Figure 6A, middle panel and Figure 6B). An unrelated ER membranal protein, RTNLB3 (Reticulon-like protein B3) seemed to be only mildly affected by carbon starvation (Figure 6A, lower panel and Figure 6B). Here again, there was no significant difference in the number of RTNLB3-mCherry-labeled autophagic bodies in the ati1;ati2 background. Autophagic flux analysis supported the imaging results and demonstrated a significantly reduced autophagic flux of MSBP1, but not of MAPR2 or RTNLB3, in the background of the ati1;ati2 mutant (Figure 6C and S6A). In accordance, Y2H assays showed that MAPR2 and RTNLB3 do not interact with the ATI1 and ATI2 proteins (Fig. S6B). Taken together, these results demonstrate that the ATI proteins are required for the selective autophagy of MSBP1.

Figure 6.

ATI is required for efficient autophagic degradation of MSBP1. (A) Representative confocal images of leaves from plants stably expressing MSBP1-mCherry (upper panel), MAPR2-mCherry (middle panel) or RTNLB3-mCherry (lower panel) in the background of either Col-0 wild type or ati1;ati2 double knockout plants. Plants were incubated in the dark with conA for 16 h. Scale bar: 10 μm. (B) Quantification of mCherry-labeled puncta per section (0.005 mm²) from Col-0 wild type or ati1;ati2 mutant plants. The number of dark-induced MSBP1-mCherry-labeled puncta is reduced in ati1;ati2 knockout plants. No significant difference in the number of labeled puncta is observed for MAPR2-mCherry or RTNLB3-mCherry. Values represent mean ± SE (n = 32). Asterisks indicate statistically significant difference compared to Col-0) (P < 0.01, Student’s t-test). NS, non-significant. (C) Leaves from plants expressing MSBP1-mCherry, MAPR2-mCherry or RTNLB3-mCherry in the background of either wild type (Col-0) or ati1;ati2 mutants were dark treated, and release of free mCherry was monitored by analysis of total protein extracts with anti-mCherry antibodies. A representative blot in which 3 repeats were loaded for each sample is shown. Arrowheads mark the location of the tagged proteins and of free-mCherry bands. Similar loading is shown by the stained level of RBCL (loading)

Discussion

The essential role of the ER in maintaining cellular homeostasis through the biosynthesis and transport of macromolecules, as well as its role in cellular signaling, requires constant adaptation of its size and activity to environmental and developmental cues. The identification and characterization of several mammalian ER-resident autophagy cargo receptors indicated that selective reticulophagy is a major ER quality control mechanism under different abiotic and biotic stresses [3,60]. The unique metabolic needs of plants are likely to require specific solutions to the maintenance of ER homeostasis. However, the characterization of the first plant reticulophagy cargo receptors is very recent, and our understanding of plant selective reticulophagy mechanisms is still limited [38–40]. In this study, we characterized a carbon starvation-induced selective reticulophagy pathway in Arabidopsis plants that involves the transmembrane ATI proteins. These plant-specific proteins were previously shown to associate with the ER, and are delivered to the vacuole upon carbon starvation in an autophagy machinery-dependent mechanism [37]. However, a direct link to reticulophagy was not demonstrated. Here we show that following dark treatment, the plant equivalent to starvation, ATI1-labeled bodies that carry an ER marker are transported to the vacuole, demonstrating the involvement of ATI1 in reticulophagy (Figures 1 and 2). Unlike other recently identified plant reticulophagy pathways, the ATI1 reticulophagy pathway is not induced in response to ER stress [38–40] (Figures 1 and 2). As generally shown for nutrient starvation-induced autophagy in plants [48], TOR signaling seems to be involved in ATI1 autophagy (Fig. S1). In contrast, ER stress-induced autophagy was shown to be TOR-independent [48]. Thus, our findings support the conclusion that ER stress and starvation-induced reticulophagy are mechanistically distinct. Intriguingly, we demonstrated that ER stress-induced reticulophagy is tissue-specific as it is induced in roots, but not in leaves [24] (Figures 1 and 2). This observation might indicate spatial-specific expression of ER stress-related reticulophagy receptors.

The MAPR family of proteins is found in yeast, mammals and plants and share a similar non-covalent heme/steroid-binding domain that is related to cytochrome b5. MAPR proteins have a surprisingly diverse array of cellular functions including regulation of cytochrome P450, steroidogenesis, vesicle trafficking, progesterone signaling and mitotic spindle and cell cycle regulation [61]. Our study identifies the plant MAPR family members MSBP1 and MAPR4 as ATI-binding proteins and as cargo of carbon starvation-induced reticulophagy (Figures 3 and 4 and S2-S4). MSBP1 is delivered to the vacuole with the ATI proteins (Figure 5 and S5), and its efficient autophagic degradation requires functional ATI proteins (Figure 6). Taken together, our results suggest that the ATI proteins function as selective reticulophagy cargo receptors to MSBP1. MSBP1 is the best characterized MAPR family member in Arabidopsis. It was initially suggested to localize to the plasma membrane and to endocytic vesicles, and was shown to mediate the cycling of PIN2, a member of the PIN family of auxin efflux facilitators in the root [54], and to inhibit brassinosteroid (BR) signaling by enhancing the endocytosis of BAK1, a BR signal co-receptor [53,55]. The inhibitory role of MSBP1 in BR signaling was also demonstrated in tobacco plants [62]. However, a recent paper suggested an additional function for MSBP1 and MAPR3/MSBP2. They were shown to reside predominantly on the ER, and to function as scaffold proteins to monolignol biosynthetic P450 enzymes, stabilizing their activity, thereby specifically controlling phenylpropanoid–monolignol branch biosynthesis [52,55]. The ER localization of MSBP1 is supported by our results (Figure 5). Furthermore, we demonstrate that MSBP1 is specifically targeted by reticulophagy as it is colocalized with an ER marker and with the ER-localized ATI proteins up to the vacuole (Figure 5 and S5). The identification of MSBP1 and the other MAPR family members as a target for autophagic degradation (Figure 4 and S4) is supported by a recent proteomics study showing that both MSBP1 and MAPR3/MSBP2 are more abundant in atg5 Arabidopsis mutants, though no change is seen in their transcription levels [63]. As MSBP1 was shown to serve as a physical scaffold to monolignol P450 monooxygenases, the whole complex might be directed to degradation through ATI-mediated reticulophagy under carbon starvation. Lignification requires a significant investment of fixed carbon and energy. Therefore, selective degradation of the lignin biosynthetic machinery under carbon starvation might be used to cope with energy deprivation. Finally, MSBP1 might be more than just a cargo of the ATI proteins and might actually be involved in the regulation of reticulophagy. PGRMC1, a mammalian MAPR family member, was shown to bind LC3 and to promote autophagy [64]. In Arabidopsis, MAPR3/MSBP2 was recently shown to interact with the ER stress regulator IRE1b and with GAAP1 and GAAP3, two Bax inhibitor-1-like plant factors that play a role in plant survival under ER stress. Though the mechanistic details are not clear, these interactions were suggested to be involved in autophagy regulation in response to ER stress [65].

Similar to the known yeast and mammalian reticulophagy cargo receptors, the ATI proteins are ER membranal proteins that harbor an ATG8-Interacting Motif (AIM) or LC3-Interacting Region (LIR) and can bind members of the ATG8 family [41,57,60,66]. The known ER receptors were suggested to mediate selective autophagy of ER sub-regions based on their sub-ER localization [60]. Thus, the yeast Atg39 ER cargo receptor localizes to the perinuclear ER and induces autophagic sequestration of part of the nucleus, while yeast Atg40 is involved in the autophagy of the cytosolic and cortical ER [25]. In mammals, RETREG1/FAM134B was suggested to regulate the turnover of ER sheets, while RTN3L and ATL3 are important for degradation of ER tubules [27,28,31]. Though the sub-ER localization of the ATI protein is unknown, Arabidospis lignol p450 monooxigenases were shown to bind reticulons [67], a family of proteins that are needed to form the ER tubular network, most probably by stabilizing the high curvature of the tubules [68]. Hence, the association of the ATI proteins with MSBP1 might suggest that they are localized to ER tubules. While it is not clear how and if specific ER cargo is selected by the mammalian ER receptors, our results, together with recent results from Michaeli et al. [44], demonstrate that the ATI proteins directly bind specific proteins (MSBP1, MAPR4 and AGO1) and travel with them to the vacuole. Although other members of the MAPR family are also targeted by autophagy under carbon starvation (Fig. S4), the ATI pathway is selective, as the ATI proteins are not required for the autophagic degradation of a non-membranal member of the MAPR family, MAPR2 (Figure 6). Moreover, another ER protein, RTNLB3, seems generally less amenable to carbon starvation-induced reticulophagy suggesting selectivity within the ER (Figure 6).

Mammalian ER cargo receptors were suggested to function as a bridge between the ER and autophagosomal membranes, possibly through the intrinsically disordered regions (IDRs) that are found in all of them [33]. Additionally, RETREG1 and RTN3L might also function in reshaping or fragmenting the ER membranes in preparation for their engulfment by autophagosomes [27,28]. The plant-specific ATI proteins also harbor an N-terminal IDR that contains their functional AIM motif [57]. However, autophagy involving the ATI proteins seems to have a distinct mechanism. Rather than direct association of cargo receptors on ER membranes with phagophores, this type of reticulophagy seems to involve first the formation of a distinct ATI-body, and only then fusion or engulfment by autophagosomes [41,42]. The ATI proteins are also unique in that they seem to be involved in the turnover of two organelles. ATI proteins were previously shown to be involved in the autophagy of chloroplast components in response to carbon starvation and during senescence. ATI1 binds several chloroplast proteins, and chloroplast-associated ATI-bodies (termed ATI-PS bodies) deliver chloroplast cargo to the vacuole [41,42]. Our results demonstrate that the C-terminal region of ATI1 interacts with both MSBP1 and the chloroplast cargo protein 2CPA (2-Cys peroxiredoxin A) (Figure 3 and S2). In addition, some morphological differences were observed between ATI-PS and ATI-ER bodies [41,42]. Thus, it seems possible that ATI ER-associated bodies and ATI-PS bodies are distinct entities. Future research will be required to clarify the mechanistic relationship between these two pathways.

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana Col-0 ecotype (the Arabidopsis Biological Research Center [ABRC], CS60000) as well as N. benthamiana (for BiFC assays) were used in this study. The T-DNA insertion line ati1 (SALKseq_136644) was obtained from ABRC and its genotype was confirmed by PCR analysis. The ati2 null mutant was generated by CRISPR-Cas9 editing [44], and the two lines were crossed to create the ati1;ati2 double mutant. The Arabidopsis mutant line atg5-1 (SAIL_129B079), and the transgenic lines GFP-HDEL, mCherry-HDEL, GFP-ATG8f, ATI1-GFP, ATI2-GFP, pATI1:ATI1-GFP and ATI1-GFP/mCherry-HDEL were previously described [41,42,44,59,69]. For MSBP1, MAPRs and RTNLB3 expression, wild type plants or the appropriate lines were floral-dipped with Agrobacterium tumefaciens strain GV3101 harboring the appropriate binary vector [70]. For seedlings experiments, seeds were surface sterilized using 3% bleach, sown on 1/2 MS medium (Duchefa Biochemie, M0221) containing 1% sucrose (Bio-Lab ltd, 001922059100) with the proper antibiotic selection, incubated in the dark at 4°C for 2–3 d and grown under long-day conditions (16 h light/8 h dark).

Plasmid construction

To construct the plasmids used for the Y2H assays, BiFC assays and plant transformation, fragments containing the full-length ORFs were digested and ligated into respective vectors. For Y2H assays, the pDHB1 and pPR3-N vectors (Dualsystems Biotech, P01005) were used for bait and prey constructs, respectively. The different genes were ligated with the plasmids following SfiI digestion. For MSBP1, MAPRs or RTNLB3 expression or BiFC plasmids construction, amplified DNA fragments were inserted into the appropriate pSAT vectors carrying 2x35S promoter sequence and either full-length mCherry or the N-terminal or C-terminal of EYFP (ABRC, CD3-1081, CD3-1075 or CD3-1071, respectively) [56,71] using the In-Fusion kit (Clontech, 639648) according to the manual instructions. Modified pART27 or pBART binary vectors (The Arabidopsis Information Resource, 6530624586 or 6530780616) were produced by inserting into the NotI restriction site a multicloning site containing 13 unique restriction endonuclease recognition sites taken from the pPZP-RCS2 (ABRC, CD3-1057) [56]. All generated pSAT expression cassettes were transferred to these modified pART27 or pBART binary vectors using either the AscI (for pSAT1) or the I-SceI (for pSAT4a) sites [56]. For in vitro affinity-isolation assays, the cDNA encoding 2CPA without its putative transit peptide (the first 65 amino acids), and the cDNA encoding MSBP1 without its N-terminal transmembrane domain (the first 40 amino acids), were inserted into modified pET41a (Novagen, 70556) which harbors a 6xHis tag, a TEV cleavage site and GST using BamHI and XhoI.

All the genes, plasmids, primers and the cleavage sites are listed in Table S1.

Split-ubiquitin Y2H assays and BiFC

The Y2H screen was performed using the DUALhunter kit (Dualsystems Biotech) two-hybrid system. The ATI2 coding sequence was cloned into the yeast bait vector pDHB1 (Dualsystems Biotech), and co-introduced with a commercial pNubG-X Arabidopsis cDNA library (Dualsystems Biotech, P02210) to the NMY51 yeast strain (Dualsystems Biotech, P04005). The NMY51 yeast strain contains the reporter genes ADE2, HIS3, and lacZ in its genome. The reporter genes are expressed only when the two-hybrid interaction combines the Cub and NubG parts of the yeast ubiquitin protein, resulting in the release of the LexA-VP16 transcription factor, which in turn enter to the nucleus and activates the three reporters. Positive colonies were selected on medium lacking the amino acids tryptophan, leucine, histidine, and aminodecanoic acid (SD -LTHA) supplemented with 10 mM 3-aminotriazole (3-AT; Sigma-Aldrich, A8056), a competitive inhibitor of the HIS3 gene product. The chosen clones were separately sequenced. For one-on-one Y2H, the ORFs of MSBP1, MAPR4, MAPR2 or RTNLB3 were inserted into the prey vector pPR3-N.

For analyzing in planta interactions between ATI1 or ATI2 and MSBP1 or MAPR4, different Agrobacterium strains harboring each of the following plasmids were used: pART27-ATI1-cEYFP, pART27-ATI2-cEYFP, pART27-MSBP1-nEYFP or pART27-MAPR4-cEYFP and the negative control plasmids pART27-nEYFP or pART27-cEYFP. The appropriate plasmid pairs were transiently co-transformed in Nicotiana benthamiana leaves following their infiltration, as previously described [72].

Confocal microscopy

A Nikon A1 confocal microscope system was used in this study. Generally, samples were put between two microscope glass coverslips (No. 1 thickness) in an aqueous environment. For N. benthamiana transient expression, a small piece (0.5 cm²) of the injected leaf was placed between two glass coverslips as described above, and the epidermis cells were analyzed. For image acquisition, either the x20 or the x60 objectives (numerical apertures of 0.75 and 1.20, respectively) were used. GFP fluorescence images were taken using 488 nm laser excitation and the emission was collected via the 525-nm filter. The mCherry images were taken using 561-nm laser excitation and emission was collected through the 595-nm filter. YFP images were taken with the same settings as for GFP, and chlorophyll autofluorescence was imaged using the 640-nm laser and collected with the 700-nm filter. Time-lapse images were all composed of images taken using line sequence.

Image analysis

Acquired images were analyzed using either the NIS-elements AR imaging software or ImageJ. Counting of labeled puncta was performed on 2D images taken in the vacuolar focal plan of the cell. Counting was done either manually or using the find maxima function of ImageJ. Several plants were imaged in the same experiment, and experiments were repeated 2–4 times resulting in an average of 25–30 analyzed images.

Treatments with chemicals

For ConA (Sigma-Aldrich, 27689) treatment, seedlings were grown in ½ MS agar Petri plates without sucrose for 6 d. Then, several whole seedlings of each of the examined lines were transferred into wells of 24-well plates containing ConA diluted in liquid ½ MS medium. As a control, the same number of seedlings was subjected to the same treatment using DMSO (used to dissolve conA). Seedlings were incubated in the liquid media at 23°C with gentle shaking. For experiments with ATI1-GFP/mCherry-HDEL, seedlings were incubated with 1 μM conA for 24 h, while for experiments with the MSBP1-expressing lines, seedlings were incubated with 2 μM conA overnight. Following incubation, the seedlings were directly examined by confocal microscopy. For leaf infiltration, newly matured rosette leaves were excised and infiltrated with 2 μM ConA with a needleless syringe, and then wrapped with aluminum foil in a box paved with wet paper and incubated overnight at 23°C in the dark.

For dark and TM treatments of ATI-GFP/mCherry-HDEL mature plants, leaves of 4- to 5-week-old plants were infiltrated with a needleless syringe with either 1 μM ConA or with 15 μg/ml TM (stock solution of 10 mg/ml in DMSO; Sigma-Aldrich, D8418). Leaves of control plants were infiltrated with the appropriate amount of DMSO alone. The plants were left for 24 h under regular light conditions or in the dark, as detailed in the figure legends, and then the treated leaves were either imaged by a confocal microscope or collected for protein extraction and immunoblotting.

For AZD8055 treatment, seedlings were grown as detailed above, and then incubated in liquid media supplemented with 0.5 μM ConA for 24 h under regular light conditions or in the dark as detailed in the figure legend. AZD8055 (5 μM; A₂S, A8214) was added for the last 6 h, followed by confocal imaging.

Autophagy flux assays

Arabidopsis leaves were ground to a fine powder in liquid nitrogen and proteins were extracted using extraction buffer (100 mM Tris-HCl, pH 7.5, 1% SDS, 5 mM EDTA and 1× protease inhibitor cocktail [Sigma-Aldrich, P8340]). Following centrifugation at 15,000 x g for 10 min at 4°C, the supernatant was transferred to a new tube and quantified by BCA protein assay (Pierce, 23,227). Reducing sample buffer was added, and the protein extract was separated on reducing SDS-PAGE and probed with commercial anti-RFP or anti-GFP antibodies (Chromotek, 6G6 or Abcam, ab290, respectively) to detect mCherry-tagged MSBP1 or GFP-tagged ATI1. Band quantification was done using ImageJ.

In vitro affinity isolation

Recombinant proteins were produced using E. coli strain Rosetta2(DE3)pLysS (Novagen, 71,403) following induction with 300 μM IPTG (Sigma-Aldrich, I6758) and overnight incubation at 18°C. Pelleted cells were resuspended in lysis buffer (100 mM sodium phosphate, pH 7.2, 300 mM NaCl, 1 mM DTT) containing protease inhibitors (Roche, 04693132001) and sonicated.

Normalized E. coli clarified lysates were added to 10 μl of glutathione magnetic agarose beads (Pierce™ Glutathione Magnetic Agarose Beads, Thermo Scientific™, 78,602), previously equilibrated with wash buffer (100 mM sodium phosphate, pH 7.2, 300 mM NaCl, 1 mM DTT, 0.01% [v:v] IGEPAL [Sigma-Aldrich, I8896]). Beads were washed 5 times with 1 ml of wash buffer. Bound proteins were eluted by adding 50 μl Laemmli buffer. Samples were analyzed by western blotting using commercial anti-MBP (Sigma Aldrich, M1321), anti-HIS6 (Sigma Aldrich, H1029) or anti-GST HRP-conjugated antibodies (GE Healthcare, RPN1236).

RNA analysis

Total RNA was extracted from newly matured leaves using a NucleoSpin RNA kit (Macherey-Nagel, 740,949) according to the manufacturer’s instructions and then treated with DNase I. cDNAs were prepared from 2 μg DNase-free total RNA using an oligo(dT)15 primer and a SuperScript II reverse transcriptase kit (Thermo Fisher Scientific, 18,064–022) following the manufacturer’s instructions. Quantitative real-time PCR was conducted using Fast SYBR Green Master Mix (Thermo Fisher Scientific, 4,385,612) and the appropriate primers in an StepOnePlus Real-Time PCR thermocycler (Applied Biosystems, CA, USA). RNA Helicase (AT1G58050.1) was used as a reference gene. The primer sequences used for RT-qPCR can be found in Table S1.

Statistical analysis

Experimental results were analyzed by the Excel software and presented as mean ± SE for the indicated (n) independent experiments or samples. Data were further analyzed with unpaired, two-tailed Student’s t-test for two group comparisons. p values are represented as * for p ≤ 0.05 and ** for p ≤ 0.01.

Funding Statement

This work was supported by Israel Science Foundation (grant 612/16) and by an Israel’s Council for Higher Education Planning and Budgeting Committee (VATAT) postdoctoral fellowship to J.W. G.G. is the incumbent of the Bronfman Chair of Plant Sciences at The Weizmann Institute of Science.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.