Endoplasmic reticulum–bound ANAC013 factor is cleaved by RHOMBOID-LIKE 2 during the initial response to hypoxia in Arabidopsis thaliana

Significance

Waterlogging and flooding due to extreme weather events are detrimental to plant growth, yield, and survival. Both waterlogging and flooding severely impair aerobic metabolism, requiring a swift initiation of adaptation responses. Yet, the initial mechanisms regulating the response to low oxygen stress remain largely uncharacterized. We identified here an endoplasmic reticulum membrane-bound transcription factor that is released upon hypoxia and translocates to the nucleus to initiate transcriptional reprogramming required for tolerance to low-oxygen stress. Moreover, the release of the transcription factor is likely due to a currently unknown mitochondrial retrograde signal that promotes cleavage by a rhomboid-like protease. Our findings reveal a fundamental role for rhomboid proteases in abiotic stress signaling.

Abstract

Aerobic reactions are essential to sustain plant growth and development. Impaired oxygen availability due to excessive water availability, e.g., during waterlogging or flooding, reduces plant productivity and survival. Consequently, plants monitor oxygen availability to adjust growth and metabolism accordingly. Despite the identification of central components in hypoxia adaptation in recent years, molecular pathways involved in the very early activation of low-oxygen responses are insufficiently understood. Here, we characterized three endoplasmic reticulum (ER)–anchored Arabidopsis ANAC transcription factors, namely ANAC013, ANAC016, and ANAC017, which bind to the promoters of a subset of hypoxia core genes (HCGs) and activate their expression. However, only ANAC013 translocates to the nucleus at the onset of hypoxia, i.e., after 1.5 h of stress. Upon hypoxia, nuclear ANAC013 associates with the promoters of multiple HCGs. Mechanistically, we identified residues in the transmembrane domain of ANAC013 to be essential for transcription factor release from the ER, and provide evidence that RHOMBOID-LIKE 2 (RBL2) protease mediates ANAC013 release under hypoxia. Release of ANAC013 by RBL2 also occurs upon mitochondrial dysfunction. Consistently, like ANAC013 knockdown lines, rbl knockout mutants exhibit impaired low-oxygen tolerance. Taken together, we uncovered an ER-localized ANAC013-RBL2 module, which is active during the initial phase of hypoxia to enable fast transcriptional reprogramming.

One consequence of climate change is an increased frequency of waterlogging and flooding events due to excessive rainfall (1). A meta-analysis of the impact that waterlogging has on global crop yield revealed an average decrease of 33% (2), urging for a better understanding of low-oxygen tolerance in plants to protect current and future crop yields. Flooding and waterlogging cause a considerable slow-down of the oxygen exchange between the environment and the plant (3). Especially roots rapidly experience an anoxic soil environment upon waterlogging, causing a shift from aerobic respiration to less efficient anaerobic metabolism to produce energy (4). Oxygen limitation results in rapid impairment of mitochondrial ATP production since NADH is no longer efficiently oxidized (5, 6). The shift to anaerobic metabolism is among others initiated through transcriptional reprogramming, which ensures glycolytic ATP production and restoration of the NADH/NAD⁺ ratio (7, 8). In Arabidopsis, the transcriptional response to low-oxygen stress is characterized by the induction of hypoxia core genes (HCGs) responding in all plant tissues (9). Among the HCGs are ALCOHOL DEHYDROGENASE 1 (ADH1) and PYRUVATE DECARBOXYLASE 1 (PDC1) involved in anaerobic metabolism. It was recently demonstrated that a subset (~70%) of the HCGs contain a so-called hypoxia-responsive promoter element (HRPE) recognized by the group VII ETHYLENE-RESPONSE FACTORs (ERF-VII) transcription factors RELATED TO APETALA 2.2 (RAP2.2) and RAP2.12 (10). Furthermore, RAP2.2 and RAP2.12 were shown to act redundantly with the constitutively expressed RAP2.3 (11, 12). Importantly, ERF-VII transcription factors are targets of the N-degron pathway that, under aerobic conditions, direct ERF-VII proteins to proteasomal degradation (13–15). As the HRPE is absent in a substantial proportion of HCGs (10), it suggests that other transcriptional regulators may act during the initial response to low-oxygen stress.

Restraints on mitochondrial respiratory activity during stress provoke retrograde response signaling pathways leading to transcriptional reprogramming and adaptation (16). Genes either responding to short-term treatment with respiratory inhibitors or showing altered expression in plants with modified levels of mitochondrial proteins contain in their promoters the mitochondrial dysfunction motif (MDM) (17). This cis-element was shown to be recognized by five NO APICAL MERISTEM/ARABIDOPSIS TRANSCRIPTION ACTIVATION FACTOR/CUP-SHAPED COTYLEDON (NAC) transcription factors, including ANAC013 and ANAC017, which both contain a C-terminal transmembrane domain (TMD). Knockdown of ANAC013 impairs the induction of some MITOCHONDRIAL DYSFUNCTION STIMULON (MDS) genes after 6 h of antimycin-A treatment (17) but not for instance the expression of ALTERNATIVE OXIDASE 1a (AOX1a). ANAC017 was originally identified as REGULATOR OF AOX1A 2 (RAO2), which is required for the induction of AOX1a upon antimycin-A application (18). Thus, ANAC transcription factors that recognize the MDM cis-element appear to overlap in their functions only partially. In addition, ANAC017 has been implicated in chloroplast retrograde signaling (19, 20) and ER-reductive stress signaling (21). Interestingly, tolerance to submergence stress is decreased upon loss of ANAC017 (22). Under submergence, ANAC017 mainly regulates genes involved in chloroplast function/photosynthesis and the response to oxidative stress but does not seem to be a regulator of common HCGs (22). How MDM-binding ANAC factors are activated and released during stress is only poorly understood. However, it is suspected that intramembrane rhomboid-like proteases (RBLs) are necessary for cleavage and thus release of ANAC transcription factors anchored at the ER membrane (18).

Here, we demonstrate that a subset of HCGs having the MDM cis-element in their promoters are regulated by ER-anchored ANAC transcription factors, suggesting a role for ANACs in the signaling of mitochondrial dysfunction during hypoxia stress in Arabidopsis. We show that from the three analyzed ANACs, i.e., ANAC013, ANAC016, and ANAC017, only ANAC013 translocates during the initial phase of hypoxia to the nucleus. Upon nuclear accumulation, ANAC013 associates with the promoters of important HCGs such as ADH1, PDC1, and PHYTOGLOBIN1 (PGB1). Moreover, the release of ANAC013 from the ER membrane under hypoxia requires the activity of the RHOMBOID-LIKE 2 (RBL2) protease. Notably, RBL2 also mediates ANAC013 upon pharmacologically induced mitochondrial dysfunction. In line with the action of RBLs on ANAC013, rbl mutants display a hypoxia-sensitive phenotype, similar to that of ANAC013 knockdown lines. In conclusion, we provide evidence that early translocation of ANAC013 during the initial phase of hypoxia is initiated by rhomboid proteases and allows for rapid activation of adaptation responses.

Results

The ANAC-Bound MDM Motif Is Present in Most HCG Promoters.

The mitochondrial dysfunction motif (MDM) is a cis-element recognized by ANAC013, ANAC016, ANAC017, ANAC053, and ANAC078 (17). Since ANAC053 and ANAC078 are not involved in mitochondrial retrograde signaling but implicated in transcriptional regulation of the 26S proteasome (23, 24), these factors were excluded from further analyses. An in silico promoter analysis of all 49 HCGs revealed that 31 (63%) contain at least one copy of the ANAC-bound MDM within the first 2 kb upstream of the transcriptional start site (TSS) (Fig. 1A and Dataset S1). Twenty-three genes (45%) contain both the MDM and HRPE motif, the latter which is recognized by ERF-VII factors (SI Appendix, Fig. S1). Usage of public DNA-affinity purification sequencing (DAP-SEQ) data (25) revealed direct interactions between ANACs and HCG promoters. Specifically, ANAC013, ANAC017, and the very similar ANAC016 were found to associate with 13, 12, and 21 HCG promoters in vitro, respectively (Fig. 1B). Seven HCGs are bound by all three ANACs (Fig. 1B). One indication for a gene to have a role during oxygen limitation would be a transcriptional response to hypoxia, though it is not a mandatory prerequisite. While ANAC013 was significantly upregulated under hypoxia (1% oxygen), ANAC016 and ANAC017 were not differentially expressed (Fig. 1C). A time series confirmed ANAC013 induction between 30 min and 4 h of hypoxia in seedlings (Fig. 1D). Tissue-specific analysis of ANAC013 expression using the promoter-GUS reporter system showed under normoxic conditions the strongest expression in the first leaf pair. Hypoxia visibly increased ANAC013 expression in the entire shoot and to some extent in the root (Fig. 1E).

Fig. 1.

ANAC transcription factors are potential regulators of hypoxia core genes. (A) 63% of the hypoxia core genes (HCG) contain at least one MDM motif within the first 2 kb upstream of the transcriptional start site. (B) The MDM motif is recognized by ANAC013, ANAC016, and ANAC017 for which available DAP-seq data revealed their association with 13, 21, and 12 HCG promoters, respectively. (C) A heatmap depicting the expression of the three ANAC genes under low-oxygen stress. Shown are data obtained from the following experiments: Leaves treated for 4 h with 1% oxygen (26), seedlings kept for 9 h under anoxic conditions (27), and roots treated for 5 h with 1% oxygen (28). (D) qRT–PCR analysis of ANAC013 expression during hypoxia. Biological triplicates were statistically analyzed using one-way ANOVA with the post hoc Tukey HSD test. Different letters indicate significant differences (P < 0.05). (E) In planta analysis of ANAC013 promoter activity during hypoxia using a GUS reporter. Three-week-old plants were treated with 1% oxygen for the indicated time, followed by GUS staining and imaging.

ANACs Bind and Activate HCGs Containing the MDM.

To verify that ANAC013, ANAC016, and ANAC017 act on a proportion of HCGs in vivo, promoters of three HCGs containing the MDM motif and two lacking the motif were fused to the firefly luciferase reporter gene and tested in transactivation assays (Fig. 2A). Each ANAC (in its soluble form lacking its TMD) was able to activate the promoters of ADH1, PGB1, and PDC1, which all contain at least one MDM copy. In contrast, no activation of HYPOXIA RESPONSE ATTENUATOR 1 (HRA1) and HYPOXIA RESPONSE UNKNOWN PROTEIN 7 (HUP7) promoters was observed, consistent with the lack of the corresponding cis-element (Fig. 2A). Activation of the ADH1 and PGB1 promoters with each two MDM copies was substantially reduced by introducing mutations in one of the two MDM sequences (M1). Complete lack of activation for both promoters was achieved in case of mutation of the MDM closest to the TSS (M2) or when both MDMs were mutated (Fig. 2 B and C). This mutational analysis indicates that a functional MDM motif is required for induction of HCGs by ANACs. In addition, electrophoretic mobility shift assays (EMSAs) were conducted with each ANAC (lacking its TMD) in combination with DNA probes matching the two MDM copies in ADH1 and PGB1, respectively (Fig. 2D). All three ANAC transcription factors were able to bind MDM motifs from both promoters. Adding an unlabeled competitive DNA probe let the observed band shift vanish (Fig. 2D). Taken together, HCGs possessing the MDM motif in their promoters are bound and activated by ANAC013, ANAC016, and ANAC017.

Fig. 2.

ANAC013, ANAC016, and ANAC017 bind and activate promoters of hypoxia core genes. (A) The impact of ANAC013, ANAC016, and ANAC017 on the activity of hypoxia core gene (HCG) promoters, either containing or lacking the MDM cis-element, was tested in a transient transactivation assay. Promoters containing the MDM, including pADH1, pPGB1, and pPDC1 were activated by ANAC013, ANAC016, and ANAC017, while those lacking the MDM, like pHRA1 and pHUP7, were not responsive. Site-directed mutagenesis of the two MDM copies in either pADH1 (B) or pPGB1 (C) promoter impairs activation by all three ANACs. For A, B, and C, n = 4 and the experiments were repeated three times. Shown are normalized relative luminescence values, whereby different letters above boxes indicate significantly different groups as determined by one-way ANOVA followed by the post hoc Tukey test (P < 0.05). (D and E) EMSA experiment showing association of ANAC013, ANAC016, and ANAC017s with the MDM motifs present in pADH1 (D) and pPGB1 (E). Labeled DNA probes cover the MDM motif as indicated and were incubated with in vitro expressed ANAC proteins. Competitor experiments indicate specific binding to the MDM-containing DNA probes.

ANAC013 Translocates to the Nucleus during the Initial Phase of Hypoxia Stress.

Under nonstress conditions, the ERF-VII transcription factor RAP2.12 is stored at the plasma membrane through complex formation with ACYL-COA BINDING PROTEIN (ACBP) (14, 29). Hypoxia induces RAP2.12 dissociation and enables the stabilization of soluble RAP2.12 protein, leading to nuclear accumulation of the transcription factor (14, 29). Similarly, ANAC013, ANAC016, and ANAC017 are anchored at the ER membrane in the absence of stress; however, this is directly achieved through their C-terminal TMDs (30 and SI Appendix, Fig. S2). To examine whether these ANACs, like RAP2.12, translocate to the nucleus upon hypoxia (1% O₂), their subcellular localization was investigated in transiently transformed tobacco leaves. GFP-tagged ANAC013 appeared in the nucleus within 2 h of hypoxia and was still detectable after 4 h of stress (Fig. 3A). In contrast, ANAC016 and ANAC017 remained at the ER and were not translocating after 2 and 4 h of stress (Fig. 3A). Nuclear appearance of ANAC013 was further examined in a detailed time-series experiment, revealing that it accumulates in the nucleus within 1.5 h of hypoxia and remains in the nucleus for up to 7 h of stress (SI Appendix, Fig. S3). As ANAC013 is implicated in mitochondrial stress responses (17), the impact of antimycin-A under aerobic conditions on the localization of ANAC013 was examined. Antimycin-A treatment likewise induced nuclear translocation of ANAC013 (SI Appendix, Fig. S4). Release of membrane-bound ANAC transcription factors is tightly controlled to ensure rapid transcriptional responses to incoming stimuli (31). To study a potential cleavage of the early translocating ANAC013 under hypoxia, an ANAC013-reporter was designed, consisting of a truncated transcription factor protein that, instead of the NAC DNA binding domain, contains a GFP protein which provided an N-terminal nuclear localization sequence (NLS; Fig. 3B and SI Appendix, Fig. S5A). After extraction and blotting, the full-length ANAC013-reporter (~49 kDa) was detectable under normoxia and hypoxia (3 h, 1% oxygen) (Fig. 3C). In addition, hypoxia-treated plants furthermore displayed a processed reporter fragment (~37 kDa) that matches the size difference of 12 kDa previously reported for the proteolytically processed ANAC013 (17). Additional immunoblots with the reporter (SI Appendix, Fig. S5 B and C) indicate that cleavage of ANAC013 occurs within 60 min of hypoxia.

Fig. 3.

ANAC013 translocates to the nucleus under short-term hypoxia. (A) Confocal laser scanning microscopy (CLSM) images of GFP-tagged ANAC013, ANAC016, and ANAC017 proteins fluorescence in tobacco leaves before and after hypoxia stress (1% oxygen) for the time indicated. From left to right, images of GFP fluorescence in green, mCherry-stained endoplasmic reticulum (ER) in red, a brightfield image, and the overlay of all three images in the fourth column, respectively. The right columns show enlargement of the nucleus and surrounding ER. Representative results for each N-terminally GFP-tagged NAC transcription factor are shown. In contrast to ANAC016 and ANAC017, only ANAC013 shows translocation to the nucleus within 2 h of hypoxia stress (1% oxygen). (Scale bars, Left panels: 20 µm, Right panels: 5 µm.) (B) A synthetic ANAC013 reporter was generated by replacing the N-terminal DNA-binding domain (DBD) of ANAC013 with GFP containing a nuclear localization sequence (violet). Importantly, the ANAC13 reporter contains the transmembrane domain (T) of the full-length protein. Numbers indicate start and end positions of all domains. Blue arrow indicates potential cleavage site according to De Clercq et al. (17). (C) Immunoblotting of the synthetic ANAC13 reporter shows that it is processed upon hypoxia stress yielding a ~37 kDa fragment (blue arrow) as based on De Clercq et al. (17). The reporter was detected by using anti-GFP antibody.

Consistent with the observations made in the transient system, stable Arabidopsis GFP-ANAC013 lines showed translocation of ANAC013 after 2 h and 4 h of hypoxia (Fig. 4A and SI Appendix, Fig. S6A), indicating that ANAC013 participates in the initial response to low-oxygen stress. Moreover, no translocation of ANAC017 during the initial phase of hypoxia was observed in stable Arabidopsis lines expressing GFP-ANAC017 (SI Appendix, Fig. S6A), validating the transient approach. Interestingly, we found that the here applied hypoxia treatment induces the expression of several MDS genes (SI Appendix, Fig. S6B), indicating a potential involvement of mitochondrial stress signals. Furthermore, GFP-ANAC013 translocation was, next to 1% oxygen, also observed at oxygen concentrations of 4% and 8%, while at 12%, translocation of ANAC013 was prevented (SI Appendix, Fig. S6C). This indicates that the release of ANAC013 occurs already under relatively mild hypoxic conditions.

Fig. 4.

Identification of the ANAC013 gene regulatory network under hypoxia. (A) Arabidopsis plants stably expressing a full-length GFP-ANAC013 construct were exposed to 1% oxygen for either 2 or 4 h. Confocal laser scanning microscopy (CLSM) images of GFP-tagged ANAC013 (green), nuclear DAPI-stain (blue), and brightfield (BF) and an overlay image are shown for a specific leaf area together with a close-up of the nucleus. (Scale bars, Left overview panels: 20 µm, Right nucleus panels: 5 µm.) (B) Distribution of enriched peak areas in relation to gene positions in the chromosome for both normoxic (N) and hypoxic (H) conditions. (C) Number and overlap of upstream hits under normoxia and hypoxia. (D) The MDM motif is overrepresented in the identified binding peaks under hypoxia as determined by HOMER motif analysis. (E) Gene ontology (GO) term enrichment analysis of genes with upstream hits under hypoxia. Analysis was performed with AgriGOv2.0 and results are shown as −log₁₀(P-value). Source data underlying Fig. 4E are provided as supplemental file. (F) Genome browser view of selected HCGs (ADH1; PGB1; LBD41; PDC1; and HRE2) and the general stress gene SAG14. Enrichment of the selected regions is shown for normoxia (light blue) and hypoxia (green). Locations of ANAC013-ChIP peaks and MDM cis-element are depicted as violet box and red line, respectively.

ANAC013 Binds to HCGs In Planta and Is Required for Low-Oxygen Tolerance.

ChIP-SEQ was performed in order to reveal in planta the gene regulatory network (GRN) controlled by ANAC013 under hypoxia. Since ANAC013 moves to the nucleus within 2 h of hypoxia in stable Arabidopsis lines (Fig. 4A), the ChIP-SEQ experiment was consequently performed after 2 h of hypoxia and compared with control conditions. In total, 2 h of hypoxia yielded 1,606 peaks, while under control conditions, 458 peaks were detected (Dataset S2), suggesting that ANAC013 executes its function primarily under hypoxia. Annotation of the peaks revealed that for hypoxic conditions, 512 (32%) peaks were found in promoter regions, while 96 (20%) of the binding events detected under control conditions occurred within the first 2-kb region upstream of the TSS (Fig. 4 B and C). Interestingly, only three hits within genes were found under both treatments, encoding two hypothetical proteins (AT2G11010 and AT4G31875) and GL2-INTERACTING REPRESSOR 1 (GIR1 and AT5G06270), the latter of which is involved in root hair development (32). A regulatory motif enrichment analysis yielded an overrepresentation of the MDM motif in the ANAC013-bound regions under hypoxia (P < 1e⁻²), suggesting that ANAC013 binds to target genes under hypoxia via the MDM (Fig. 4D). Moreover, a gene ontology (GO) enrichment analysis for the hypoxia-specific GRN downstream of ANAC013 revealed genes involved in the response to low oxygen levels and those participating in energy metabolism (Fig. 4E and Dataset S3). Enrichment of these GO terms support a role for ANAC013 during hypoxia. Indeed, several hypoxia-specific upstream peaks attributed to HCGs, including the metabolic genes ADH1, PDC1, and PGB1 but also the transcription factor genes LOB DOMAIN-CONTAINING PROTEIN 41 (LBD41) and HYPOXIA RESPONSIVE ERF 2 (HRE2) (Fig. 4F and Dataset S2). Furthermore, expression analysis for the 512 ANAC013-specific target genes under hypoxia showed that at least 165 genes are differentially expressed during the initial phase of hypoxia (SI Appendix, Fig. S7 and Dataset S4).

Interestingly, the GO enrichment analysis identified genes involved in the response to osmotic stress, salt stress, and oxidative stress (Fig. 4E). Fitting to this, many of the ANAC013 target genes under hypoxia are differentially expressed under salt stress (145 genes), osmotic stress (132 genes), and heat stress (151 genes) (SI Appendix, Fig. S8 and Dataset S4). This is exemplified by the ANAC013-target gene SENESCENCE-ASSOCIATED GENE 14 (SAG14; Fig. 4F), a gene involved in cell wall modifications (33) being differentially expressed under all abiotic stress conditions analyzed, including hypoxia (Dataset S4). As ANAC013 itself is induced by these abiotic stresses (34), ANAC013 might act as an early signaling hub during multiple stresses.

The association of ANAC013 with HCG promoters during hypoxia suggests a participation in low-oxygen stress adaptation. Hence, in a next step, we tested the contribution of ANAC013 to low-oxygen stress tolerance. Here, previously generated knockdown and overexpression lines for studying ANAC013 were used (17; SI Appendix, Fig. S9). Knockdown of ANAC013 significantly reduced seedling tolerance to anoxia, demonstrated by a lower survival score as compared to the wild type (Fig. 5 A and B). A detailed phenotypic analysis of ANAC013 knockdown roots under hypoxia revealed a shorter relative length of lateral roots, reduced later root number and less lateral roots per cm of primary root under hypoxia as compared to the stressed wild type (Fig. 5 C−E and SI Appendix, Fig. S10). At a later developmental stage, i.e., at the age of 5 wk, ANAC013 knockdown plants exhibited lower biomass accumulation after submergence stress (Fig. 5 F and H and SI Appendix, Fig. S11). However, at both the seedling and vegetative stage, full-length overexpression of ANAC013 did not impact on stress tolerance (Fig. 5 A, B, F, and H and SI Appendix, Fig. S11). Under nonstress conditions, ANAC013 amiR and OX lines did not differ from wild type in shoot weight (Fig. 5G and SI Appendix, Fig. S11). Fitting to the observed sensitive stress phenotype upon ANAC013 knockdown, several tested HCGs harboring an MDM, i.e., ADH1, PDC1, LBD41, and SUCROSE SYNTHASE 1 (SUS1), were less induced in the ANAC013 knockdown line than in the wild type or in the ANAC013 overexpression line under stress (Fig. 5I). HRA1 lacking an MDM was equally responding to hypoxia in all genotypes (Fig. 5I). Taken together, ANAC013 provides low-oxygen stress tolerance by directly regulating MDM-containing HCGs.

Fig. 5.

ANAC013 is required for low-oxygen stress tolerance. (A) Phenotypes of artificial micro RNA (amiRNA) and overexpression lines of ANAC013 and wild-type (WT) seedlings after 9 h anoxia treatment followed by 3 d of recovery. (Scale bar: 1 cm.) (B) Survival score of ANAC013 lines and wild-type (WT) seedlings after recovery of the anoxia treatment. n = 5 (with each 15 seedlings/genotype). Different letters above the boxes indicate significant difference to the wild type (WT) (one-way ANOVA, P < 0.05). (C) Representative pictures of root growth phenotypes of wild-type (WT) and two ANAC013 amiRNA lines. Six-day-old light-grown seedlings were exposed to either 4 d of normoxic (21% O₂) or hypoxic (2% O₂) conditions in the dark. (Scale bar: 1 cm.) (D) Relative lateral root length was calculated by comparing the total length of LRs under hypoxic conditions to the total length of wild-type LR under normoxic conditions in percent. Averages ± SE were calculated from three independent experiments as described in Fig. 5C. Different letters above the boxes indicate significant difference to the wild type (one-way ANOVA, P < 0.05, n = 22 to 28). (E) Number of elongated LR of wild-type and two ANAC013 amiRNA lines under normoxic and hypoxic conditions for 4 d as described in Fig. 5C. Averages ± SE were calculated from three independent experiments. Different letters above the boxes indicate significant difference to the wild type (WT) (one-way ANOVA, P < 0.05, n = 22 to 28). (F) Phenotypes of ANAC013 amiRNA and overexpression line after submergence. (Scale bar: 2 cm.) (G and H) Fresh weight of wild-type (WT), ANAC013 amiRNA #2, and ANAC013 OX #8 plants under control conditions (G, with six plants) or after exposure to dark submergence for 4 d followed by 5 d reoxygenation (H, with 15 plants). Different letters above bars indicate significantly different groups as determined by one-way ANOVA followed by the post hoc Tukey test (P < 0.05). (I) Expression of HCGs after 2 h of hypoxia treatment in the wild type (WT), ANAC013 amiRNA #2, and ANAC013 OX. Boxes indicate relative change in expression as compared to normoxic samples (n = 5), as determined by qRT-PCR. Different letters above boxes indicate significantly different expressed genes [one-way ANOVA followed by the post hoc Tukey test (P < 0.05)].

ANAC013 Release from the ER Requires Rhomboid-Like Protease Activity.

ER-bound ANACs are considered potential targets of rhomboid-like proteases (RBLs) (18); however, a proof-of-principle for such an involvement in ANAC processing, especially under hypoxia, has not been provided yet. Rhomboid proteases were discovered as regulators of Drosophila epidermal growth factor (EGF) receptor signaling and were soon found to be present in all kingdoms of life (35). In the current study, to prove chemically that RBLs do play a role in hypoxia signaling in Arabidopsis, the serine protease inhibitor N-p-Tosyl-L-phenylalanine chloromethyl ketone (TPCK) was used. Its application in animals and plants, including Arabidopsis, has been reported to efficiently block rhomboid protease activity (18, 36). Wild-type seedlings sprayed with 100 µM TPCK prior to the anoxia treatment exhibited a reduced stress tolerance (Fig. 6A). HCGs harboring an MDM and showing impaired responses to hypoxia in ANAC013 knockdown lines were likewise lower induced in the TPCK-treated wild type under stress (Fig. 6B), suggesting an involvement of RBLs in the ANAC013-dependent low-oxygen response cascade.

Fig. 6.

Rhomboid-like proteases mediate ANAC013 translocation during hypoxia. (A) The effect of the rhomboid protease inhibitor TPCK on the survival rate of seedlings exposed to 9 h of anoxia. Data represent the survival score with n = 5 (15 seedlings/plate). The asterisk indicates significant difference between the hypoxia-stressed wild type and hypoxia-stressed wild type treated with TPCK (one-way ANOVA, P < 0.05). (B) Response of HCGs to 2 h of hypoxia in wild-type seedlings that were mock- or TPCK-treated prior to stress exposure. Boxes indicate relative change in expression as compared to control samples (n = 5), as determined by qRT-PCR. Different letters above boxes indicate significantly differently expressed genes [one-way ANOVA followed by the post hoc Tukey test (P < 0.05)]. (C) Alignment of the transmembrane domains of the rhomboid substrate Spitz and of ANAC013. To determine the importance of residues within this domain for proteolytic release of ANAC013, double (TMD2L) and triple mutations were introduced (TMD3L). Identical or chemically similar residues are shown in blue, while the exchanged residues are underlined with red. (D) Translocation analysis of ANAC013 using the full-length wild-type (WT) sequence and those containing the mutated transmembrane domain, TMD2L and TMD3L, respectively. Images show localization of GFP-ANAC013 in tobacco leaves under control and hypoxic conditions (3 h 1%O₂). From left to right, images of GFP fluorescence in green, mCherry-stained endoplasmic reticulum (ER) in red, a brightfield image, and the overlay of all three images in the fourth column, respectively. The right columns show enlargement of the nucleus and surrounding ER. (E) BiFC assays in tobacco leaves to identify potential interactions between ANAC013 and the selected RBL proteins. The left side the Top, middle, and Lower panels shows the results obtained with VYNE-RBL1:VYCE-ANAC013, VYNE-RBL2:VYCE-ANAC013, and VYNE-RBL6:VYCE-ANAC013, respectively. (F) Top panels represent the positive control (PC) MLO1-VYNE:CaM1-VYCE, Lower panels represent the negative control (NC) MLO1-VYNE:GPA-VYCE. YFP fluorescence is depicted in yellow, ER-CFP or PM-CFP marker is depicted in light blue. (D-F) (Scale bars, overview panels: 20 µm, nucleus panels: 5 µm.)

In Drosophila, the protease RHOMBOID-1 (Rho-1) cleaves the TMD of its target protein Spitz, which is an EGF growth factor, to promote Spitz release through the secretory pathway (37). Sequence comparison of the Spitz TMD with the TMDs of all 14 Arabidopsis membrane-anchored ANACs revealed the highest similarity of Spitz with ANAC013, ANAC016, and ANAC017 (SI Appendix, Fig. S12). This suggests that ANAC013, ANAC016, and ANAC017 are potentially proteolytically processed by unknown RBLs sharing similar substrate preferences with Rho-1. For Spitz, it was shown that substituting several TMD-specific amino acid residues with helix-stabilizing residues (i.e., leucine) prevents cleavage by Rho-1 (38). In analogy, we exchanged multiple TMD residues in the homologous ANAC013 TMD (Fig. 6C). Mutating two nonconserved residues in ANAC013 (2TMD2L; residues F512L and W513L) did not affect transcription factor translocation after 2 h of hypoxia (Fig. 6D). Additional mutation of the conserved Cys residue (3TMD2L; residues F512L, W513L, and C509L), however, prevented nuclear accumulation of ANAC013 (Fig. 6D). Similar translocation dynamics, i.e., an abolished nuclear localization in case of three TMD-specific mutations including C509L was observed for the ANAC013-reporter (SI Appendix, Fig. S13).

Arabidopsis contains 20 RBLs with differing predicted subcellular localizations (39, SI Appendix, Fig. S14). Of them, RBL1, RBL2, and RBL6 are putatively ER-localized and, based on their amino acid sequence, group together with Rho-1 from Drosophila (SI Appendix, Fig. S14). Moreover, RBL2 was previously shown to be able to cleave the Rho1-substrate Spitz in a heterologous system (40). Therefore, we decided to characterize RBL1, RBL2, and RBL6 in terms of their potential role in processing ANAC013 under hypoxia. In contrast to ANAC013, none of the selected RBL genes showed a response during the initial phase of hypoxia (SI Appendix, Fig. S15). Subsequently, a potential colocalization of ANAC013 with one or multiple RBLs was examined, which is a prerequisite for a protease in order to act on its substrate. Like ANAC013, RBL1, RBL2, and RBL6 fused to GFP were solely present at the ER under normoxia (SI Appendix, Fig. S16). As anticipated, all RBLs remained ER-anchored under hypoxia (SI Appendix, Fig. S16). In order to perform protein–protein interaction studies, it was necessary to determine which RBL terminus is in close proximity to the cytosolic N-terminal part of the ANAC013 protein. To do so, the number of TMDs and the orientation of the N- and C-termini were determined for all three RBL proteins using the TMHMM-2.0 webserver (41). This analysis revealed that all three RBLs have a similar topology, containing seven TMDs, a cytosolic N terminus, and a C terminus that is present in the ER lumen (SI Appendix, Fig. S17). In case of RBL1, the N-terminal cytosolic tail is twice as long as for RBL2 and RBL6. Following the predicted topology, each RBL was, like ANAC013, N-terminally tagged to study protein–protein interaction in a bimolecular fluorescence complementation assay (BiFC). Interestingly, only an ER-specific interaction between ANAC013 and RBL2 was found, while no interaction was detected for ANAC013 with either RBL1 or RBL6 (Fig. 6E and SI Appendix, Fig. S18). Thus, ANAC013 release during hypoxia is potentially enabled by RBL2. Unexpectedly, also ANAC016 and ANAC017 were found to be able to interact with RBL2 (SI Appendix, Fig. S19), suggesting that during hypoxia, a specific signal is required that enables the processing of the substrate.

RBL2 Enables ANAC013 Release from the ER under Hypoxia.

To reveal the relevance of RBLs for hypoxia tolerance and their potential role in ANAC013 release during hypoxia, single rbl1, rbl2, and rbl6 mutants were identified. In addition, a double mutant (rbl2/rbl6) for the homologous RBL2 and RBL6 was generated to ensure that the effects of RBL2 knockout are not masked by potential overlapping RBL6 activity (SI Appendix, Fig. S20). The importance of RBL2 for ANAC013 release under hypoxia was tested by introducing GFP-ANAC013 into the homozygous rbl2-2/rbl6-1 mutant (Fig. 7 A and B). Nuclear accumulation of GFP-ANAC013 as observed in the wild type after 2 h of hypoxia was impaired in the rbl2-2/rbl6-1 line (Fig. 7A). Hypoxia treatment of wild type expressing GFP-ANAC013 resulted in a 100% nuclear localization (Fig. 7B). In contrast, in the rbl2-2/rbl6-1 line, only 6% of the nuclei had a GFP signal under stress (Fig. 7B). Moreover, in single rbl2 mutants GFP-ANAC013 failed to translocate to the nucleus during hypoxia (SI Appendix, Fig. S21), indicating that RBL2 is primarily responsible for the processing of ANAC013. As ANAC013 functions in mitochondrial retrograde signaling (17), we determined if the here-identified rhomboid proteases are required for the release of ANAC013 upon antimycin-A treatment. As with hypoxia, antimycin-A failed to induce nuclear accumulation of GFP-ANAC013 in the rbl2-2/rbl6-1 mutant background, while in the wild type this occurred (SI Appendix, Fig. S22A). Moreover, the expression of several hypoxia and/or mitochondrial stress genes was also impaired in the double-mutant upon hypoxia (SI Appendix, Fig. S22 B and C). ANAC013 expression in rbl2-2/rbl6-1 under aerobic conditions corresponded to wild-type levels (SI Appendix, Fig. S22D). According to the impaired nuclear ANAC013 accumulation in rbl2-2/rbl6-1, we hypothesized that this line but also potentially single rbl mutants possess a decreased low-oxygen tolerance. Single mutants for RBL2 and RBL6 but also RBL1 were less tolerant than the wild type when subjected to anoxia at the seedling stage (Fig. 7 C and D). Interestingly, the degree of reduced anoxia tolerance observed for the double-mutant rbl2-2/rbl6-1 was comparable to that of the single mutants (Fig. 7 C and D). As for RBL2, only a single viable T-DNA insertion line could be identified, and several independent rbl2-2/35S:RBL2 complementation lines were generated (SI Appendix, Fig. S23). The complementation lines display a wild type-like anoxia tolerance (Fig. 7 E and F).

Fig. 7.

RBL2 is required for ANAC013 translocation during hypoxia. (A) Analysis of ANAC013 translocation dynamics in the wild type (WT) and the rbl2-2/rbl6-1 double mutant during hypoxia. From left to right, images of GFP fluorescence in green, DAPI-stained nuclei in blue, a brightfield image, and the overlay of all three images in the fourth column, respectively. The right columns show enlargement of the nucleus. (Scale bars, overview panels: 10 µm, nucleus panels: 5 µm.) (B) Quantification of GFP-ANAC013 nuclear occurrence in the wild type and rbl2-2/rbl6-1 mutant under control conditions and hypoxia. Twenty DAPI-stained nuclei (in four leaves each) were counted in terms of the presence or absence of a GFP signal. Different letters above boxes indicate significantly differently expressed genes [one-way ANOVA followed by the post hoc Tukey test (P < 0.05)]. (C) Phenotypes of rbl mutants and wild-type (WT) seedlings after 9 h anoxia treatment followed by 3 d of recovery. (Scale bar: 1 cm.) (D) Survival scores of rbl mutants and wild-type (WT) seedlings after recovery of the anoxia treatment. n = 5 (with 15 seedlings/genotype). Different letters above boxes indicate significant difference to WT (one-way ANOVA, P < 0.05). (E) Complementation of the anoxia sensitivity of the rbl2-2 mutant by introducing the RBL2 CDS driven by a 35S promoter. Seedlings were placed for 9 h under anoxic conditions in the dark and allowed to recover for 3 d under standard growth conditions. (F) Survival scores for rbl2-2 and its complementation lines and the wild type (WT). n = 5 (with 15 seedlings/genotype). Different letters above the boxes indicate significant difference to WT (one-way ANOVA, P < 0.05). (G) panel (i) depicts localization of ANAC013 under control conditions. ANAC013 is tethered to the ER membrane and forms a complex with RBL2. panel (ii) depicts translocation of ANAC013 to the nucleus upon low-oxygen stress. Reduced availability of oxygen limits mitochondrial functioning resulting in a retrograde signal that promotes cleavage of ANAC013 by RBL2 at the ER. The N-terminal fragment including the DNA binding domain of ANAC013 will translocate to the nucleus and bind to the MDM motif present in its target genes, including HCGs. Activation of HCGs by ANAC013 promotes adaptation responses to low-oxygen stress to ensure plant survival.

Discussion

As oxygen depletion has a rapid and profound impact on many cellular processes, it is of vital importance to swiftly counteract and compensate the low oxygen–induced energy crisis. Our results reveal that the ER membrane–associated transcription factor ANAC013 is readily released upon hypoxia to initiate nuclear transcription of HCGs (Fig. 4A and SI Appendix, Fig. S6A). Activation of HCGs by ANAC013 occurs through recognition of the MDM motif, which is implicated in mitochondrial retrograde signaling (17 and Fig. 4D). Although the additionally examined ANAC016 and ANAC017 show transactivation activity toward HCG promoters (Fig. 2 A–C) and recognize the MDM motif (Fig. 2 D–E), we found that they do not translocate during the initial phase of hypoxia (Fig. 3A and SI Appendix, Fig. S6A). ANAC017 participates, like ANAC013, in mitochondrial retrograde signaling (18, 42) and was suggested to repress programmed cell death (PCD) through balancing reactive oxygen species (ROS) homeostasis (43). That ANAC017, which positively regulates submergence tolerance (22, 44), is not released during the initial low-oxygen stress phase indicates a role rather during prolonged hypoxia stress and/or reoxygenation. Next to rapid translocation of ANAC013 after the onset of hypoxia, ANAC013 is the only factor showing transcriptional upregulation during this early stress phase (Fig. 1 C–E), though a transcriptional upregulation is not mandatory for a transcription factor in order to be relevant under hypoxia. Since ANAC013 binds to its own promoter (17), a feedforward mechanism under hypoxia is likely in order to replenish the pool of ANAC013 at the ER. Such a mechanism is especially relevant when plants are exposed to short-term stress, as it reestablishes the initial sensing mechanism.

Through ChIP-SEQ analysis, we identified the ANAC013-specific GRN consisting of at least 512 target genes under hypoxia (Dataset S2). GO term analysis revealed an enrichment of genes involved in the response to oxygen levels (Fig. 4E). Among those are the HCGs ADH1, HRE2, LBD41, PDC1, PGB1, and SUS4, which all contain at least one MDM but also one HRPE cis-element within their promoters (Dataset S1). HRE2 is an ERF-VII transcription factor, which is de novo synthesized during hypoxia stress (5, 13), and which, according to the current study, acts downstream of ANAC013 under hypoxia. Previously, HRE2 was shown to be induced in an ANAC013-dependent manner by antimycin-A, a compound blocking mitochondrial complex III (17), suggesting that ANAC013 and HRE2 are linked to the regulation of mitochondrial dysfunction during hypoxia. Several of the identified ANAC013 targets are also targets of ERF-VII transcription factors (10, 14). In the current study, we found ER-tethered ANAC013 to be released early under hypoxia, i.e., after 2 h of treatment (1% oxygen) in Arabidopsis (Fig. 4A). It may therefore be speculated that ANAC013 cooperatively activates HCGs with ERF-VII transcription factors. Transcription factors acting on a same subset of stress genes are well known for cold and heat stress and allow for careful adjustment of the stress response (45), but they are more or less unexplored for hypoxia stress. Although ANAC013 is known for its role in mitochondrial retrograde signaling, only recently it was shown that altered mitochondrial activity through overexpression of the inner-mitochondrial membrane protein UNCOUPLING PROTEIN 1 (UCP1) limits the activity of the N-degron pathway required for ERF-VII protein degradation (46). Potentially, a cross-talk exists between mitochondrial signals triggered through overexpression of UCP1 and those signals regulating the release of ANAC013.

We found that plants with reduced ANAC013 expression levels show a decreased anoxia and submergence tolerance and an impaired induction of multiple MDM-containing HCGs under stress (Fig. 5 A, B, and F–I and SI Appendix, Fig. S11). Interestingly, overexpression of full-length ANAC013 did not oppositely improve stress plant performance at multiple developmental stages, but resembled the wild-type performance (Fig. 5 A, B, and F–H and SI Appendix, Fig. S11), which may be explained by the fact that ANAC013 protein in either case has to be stored first at the ER before its release is initiated. Processing of ANAC013 at the ER may hence represent a bottleneck, limiting the amount of freed protein capable of entering the nucleus under hypoxia. Fitting to ANAC013 being expressed in root tissues (Fig. 1E), lowered ANAC013 levels impact on the number and length of lateral roots produced under hypoxia (Fig. 5 C–E). Soil waterlogging and partial shoot submergence are known to affect root morphology, which occurs actively as part of the plant adaptation response (4). Our ChIP-SEQ analysis revealed that ANAC013 binds to important root regulatory genes under hypoxia (Dataset S2), including LATERAL ROOT STIMULATOR 1 and ROOT GROWTH FACTOR 10 (47, 48). Many of the here identified ANAC013 target genes are not solely responsive to hypoxia, but like ANAC013, also differentially expressed during salt, osmotic, or heat stress (Dataset S4). As mitochondria are impacted not only under hypoxia, but also during other abiotic stresses (49), it is likely that ANAC013 represents an early response factor during multiple abiotic stresses.

Next to the role of ANAC013 during hypoxia stress, we reveal that release of ANAC013 depends on the rhomboid-like protease RBL2 (Fig. 7 A and B and SI Appendix, Fig. S21). The requirement of rhomboid proteases for ANAC013 release from the ER was previously suggested based on rhomboid inhibitor assays in combination with antimycin-A treatment (17, 18). However, direct interactions or genetic proof for RBLs participating in ANAC protein release were lacking. Here, we show that chemical inhibition of rhomboid proteases reduced low-oxygen tolerance in Arabidopsis (Fig. 6A) and impaired the transcriptional response of those HCGs harboring the MDM (Fig. 6B). The TMD of ANAC013 shows sequential features similar to those of the Drosophila rhomboid-substrate Spitz (50). It was furthermore shown that the insertion of helix-stabilizing residues within the TMD of rhomboid substrates abolishes subsequent cleavage (38). In the current study, we demonstrate that introducing three helix-stabilizing leucine residues in the TMD of ANAC013 prevents its nuclear accumulation under hypoxia (Fig. 6 C and D), reminiscent of an intramembrane protease-specific cleavage mechanism. Based on their similarity to Rho-1 from Drosophila (SI Appendix, Fig. S14) and their ER-localization shared with ANAC013 (SI Appendix, Fig. S16), RBL1, RBL2, and RBL6 were examined in terms of a potential involvement in hypoxia-induced ANAC013 release from the ER. We found that RBL2 is able to physically interact with ANAC013 at the ER, while no interactions with RBL1 and RBL6 were detected (Fig. 6E). Of note, RBL2 was previously shown to be capable of cleaving the Drosophila substrate Spitz (40). Fitting to this, primary loss of function of RBL2 prevents translocation of ANAC013 under hypoxia (Fig. 7 A and B and SI Appendix, Fig. S21), indicating that ANAC013 is a substrate of RBL2. Thus, RBL2 mediates low-oxygen responses through releasing ANAC013 from the ER. A mitochondrial dysfunction signal appears to be required for initiating ANAC013 release by RBL2 since ANAC013 translocation upon antimycin-A treatment is impaired in the rbl2-2/rbl6-1 double mutant (SI Appendix, Figs. S4 and S22). The fact that RBL2 and ANAC013 already interact at the ER under normoxic conditions (Fig. 6E) suggests a model in which both proteins form a signaling hub existing prior to the onset of stress, which is activated by a hypoxia-specific (mitochondrial) signal resulting in ANAC013 release (Fig. 7G). As also ANAC016 and ANAC017 can interact with RBL2 (SI Appendix, Fig. S19), it suggests that a specific signal is perceived by the substrate, i.e., ANAC013 and not the protease itself, that allows for substrate cleavage under hypoxia.

Besides RBL2 representing the first example for rhomboid proteases acting in hypoxia signaling, our data clearly suggest more rhomboids and substrates to participate in the response to low-oxygen stress. First, ANAC013, ANAC016, and ANAC017 all interact with RBL2 (SI Appendix, Fig. S19) and share a very similar TMD (SI Appendix, Fig. S2), suggesting that also ANAC016 and ANAC017 are most likely rhomboid-dependently released from the ER. Since ANAC017 is a positive regulator of submergence tolerance (22, 44), it will be of great interest to explore during which hypoxia phase ANAC017 accumulates in the nucleus and which rhomboid protease is involved herein. Second, we provide evidence that not only RBL2 but also RBL1 and RBL6 are important for low-oxygen tolerance (Fig. 7 C–F). It is expected that revealing the role of additional rhomboid proteases in hypoxia and identification of their substrates will lead to a step-stone change in our understanding of low-oxygen signaling and adaptation in plants.

Methods

Plant Materials.

Arabidopsis thaliana ecotype Columbia-0 (Col-0) was used throughout this study. The 35S:GFP-ANAC013 line, pANAC013:GUS, and ANAC013 overexpression and knockdown lines were described previously (17). Here, amiRNA#2 and amiRNA#5 correspond to previously published ANAC013-miR-3 and ANAC013-miR-5, respectively, and OX#6 and OX#8 correspond to 35S:ANAC013-4 and 35S:ANAC013-6, respectively. The following T-DNA insertion lines were identified and used during this study: rbl1-1 (GABI_055E09), rbl1-2 (SALK_001269), rbl2-1 (SALK_105389), rbl2-2 (SALK_059847), rbl6-1 (SAIL_511_B11), and rbl6-2 (SALK_111653). T-DNA lines were verified by PCR genotyping using primers listed in Dataset S5. The GFP-ANAC013, GFP-ANAC017, and the ANAC013 reporter construct (see below for cloning procedure) were transformed using the floral-dip method (51). Complementation of rbl2-2 mutant was achieved by cloning the CDS of RBL2 into pENTR-D after amplification by PCR. The entry clone was recombined with pH7WG2, and the obtained destination clone was used to transform the rbl2-2 mutant by the floral-dip method. ANAC013 in pH7FWG2 was used to transform rbl2-2/rbl6-1 double mutant for translocation studies.

Accession Numbers.

Information on the genes studied can be found at The Arabidopsis Information Resource (http://www.arabidopsis.org) using the following accession numbers: ANAC013, AT1G32870; ANAC016, AT1G34180; ANAC017, AT1G34190; RBL1, AT2G29050; RBL2, AT1G63120; RBL6, AT1G12750.

Cloning of Constructs.

Detailed cloning strategies are described in SI Appendix, Materials and Methods.

Growth Conditions and Stress Assays.

A detailed description of growth and stress assays is provided in SI Appendix, Materials and Methods.

RNA Extraction and (Semi)-qRT-PCR Analysis.

Detailed information is given in SI Appendix, Materials and Methods.

Transactivation Assays.

Transcription factor sequences in p2GW7 and target promoters in p2GWL7 were cotransformed into Arabidopsis mesophyll protoplasts (52, 53) together with the normalization vector containing 35S:RLUC. After incubation overnight, the protoplasts were lysed and luminescence was determined using the Dual-Luciferase Reporter Assay System (Promega) in a SYNERGY MY (BioTek) platereader system.

Chromatin Immunoprecipitation (ChIP), Sequencing, and Analysis.

A detailed description of the ChIP assay and data analysis is provided in SI Appendix, Materials and Methods.

Immunopurification and Western Blotting.

Detailed information on protein extraction and western blotting is provided in SI Appendix, Materials and Methods.

GUS Staining.

Plants carrying the pANAC013:GUS construct were stained as described previously (54).

Supplementary Material

Appendix 01 (PDF)

Dataset 01 (XLSX)

Dataset 02 (XLSX)

Dataset 03 (XLSX)

Dataset 04 (XLSX)

Dataset 05 (XLSX)

Data, Materials, and Software Availability

All raw sequencing data have been deposited in the European Nucleotide Archive resource, https://www.ebi.ac.uk/ena/browser/home  (accession number PRJEB52547) (55).