An oxygen-sensing mechanism for angiosperm adaptation to altitude

Flowering plants (angiosperms) can grow at extreme altitudes, and have been observed growing as high as 6,400 metres above sea level1,2; however, the molecular mechanisms that enable plant adaptation specifically to altitude are unknown. One distinguishing feature of increasing altitude is a reduction in the partial pressure of oxygen (pO2). Here we investigated the relationship between altitude and oxygen sensing in relation to chlorophyll biosynthesis—which requires molecular oxygen3—and hypoxia-related gene expression. We show that in etiolated seedlings of angiosperm species, steady-state levels of the phototoxic chlorophyll precursor protochlorophyllide are influenced by sensing of atmospheric oxygen concentration. In Arabidopsis thaliana, this is mediated by the PLANT CYSTEINE OXIDASE (PCO) N-degron pathway substrates GROUP VII ETHYLENE RESPONSE FACTOR transcription factors (ERFVIIs). ERFVIIs positively regulate expression of FLUORESCENT IN BLUE LIGHT (FLU), which represses the first committed step of chlorophyll biosynthesis, forming an inactivation complex with tetrapyrrole synthesis enzymes that are negatively regulated by ERFVIIs, thereby suppressing protochlorophyllide. In natural populations representing diverse angiosperm clades, we find oxygen-dependent altitudinal clines for steady-state levels of protochlorophyllide, expression of inactivation complex components and hypoxia-related genes. Finally, A. thaliana accessions from contrasting altitudes display altitude-dependent ERFVII activity and accumulation. We thus identify a mechanism for genetic adaptation to absolute altitude through alteration of the sensitivity of the oxygen-sensing system.

Flowering plants (angiosperms) can grow at extreme altitudes, and have been observed growing as high as 6,400 metres above sea level 1,2 ; however, the molecular mechanisms that enable plant adaptation specifically to altitude are unknown. One distinguishing feature of increasing altitude is a reduction in the partial pressure of oxygen (pO 2 ). Here we investigated the relationship between altitude and oxygen sensing in relation to chlorophyll biosynthesis-which requires molecular oxygen 3and hypoxia-related gene expression. We show that in etiolated seedlings of angiosperm species, steady-state levels of the phototoxic chlorophyll precursor protochlorophyllide are influenced by sensing of atmospheric oxygen concentration. In Arabidopsis thaliana, this is mediated by the PLANT CYSTEINE OXIDASE (PCO) N-degron pathway substrates GROUP VII ETHYLENE RESPONSE FACTOR transcription factors (ERFVIIs). ERFVIIs positively regulate expression of FLUORESCENT IN BLUE LIGHT (FLU), which represses the first committed step of chlorophyll biosynthesis, forming an inactivation complex with tetrapyrrole synthesis enzymes that are negatively regulated by ERFVIIs, thereby suppressing protochlorophyllide. In natural populations representing diverse angiosperm clades, we find oxygen-dependent altitudinal clines for steady-state levels of protochlorophyllide, expression of inactivation complex components and hypoxia-related genes. Finally, A. thaliana accessions from contrasting altitudes display altitude-dependent ERFVII activity and accumulation. We thus identify a mechanism for genetic adaptation to absolute altitude through alteration of the sensitivity of the oxygen-sensing system. Around 25% of the Earth's land surface, containing at least 30% of plant species diversity 4 , is mountainous. Although the altitude at which an individual plant grows may never change, it is critical that individuals (and populations) are adapted to survive at that altitude, and this is an important component of plant ecology 4 . Altitude and latitude have been considered collectively to impart a syndrome of developmental and physiological characteristics linked mostly to climatic adaptation (including dwarfism, resistance to ultra-violet light, low temperature tolerance, flowering time and others 4,5 ). However, no consistent trait has been associated with specific environmental components of altitude, such as pO 2 , which might reveal mechanisms underlying direct altitudinal adaptation. Here we investigate the idea that adaptation to altitude involves direct sensing of oxygen concentration across altitudinal ranges. Mammalian adaptation to very high altitude involved mutation of components of the hypoxia-inducible factor (HIF) oxygen-sensing system [6][7][8] (which is mechanistically different to the plant oxygen-sensing pathway 9 ), indicating the importance of matching metabolism with altitude. Oxygen sensing in plants through Met1-Cys2 ERFVII transcription factors is mediated by the PCO branch of the PROTEOLYSIS 6 (PRT6) N-degron pathway 10,11 . Following Met1 removal, the amino-terminal cysteine is oxidized by PCOs 12 using molecular O 2 , arginylated by arginyl transferase 13 and recognized by the E3 ligase PRT6 for ubiquitin-mediated destruction (Fig. 1a).
We reasoned that plant biochemical pathways that require oxygen may be subject to evolutionary pressure in relation to altitude. Tetrapyrrole synthesis, which leads to chlorophyll, is dependent on ambient O 2 at several points (Extended Data Fig. 1a). Following germination, during etiolated growth in the dark, the chlorophyll biosynthesis intermediate protochlorophyllide (Pchlide) accumulates because angiosperms possess only a light-activated chloroplast enzyme for Pchlide reduction 14 (light-dependent NADPH-protochlorophyllide oxidoreductase (L-POR, hereafter POR)). We set out to determine whether differences in pO 2 with altitude could influence the flux through the tetrapyrrole pathway through oxygen sensing and be a target for evolutionary adaptation. We first investigated whether the ambient O 2 concentration regulates steady-state Pchlide levels in plant species representing diverse angiosperm clades, A. thaliana (rosid), Solanum lycopersicum (asterid), Papaver somniferum (basal dicot) and Brachypodium distachyon (monocot) (Fig. 1b). This showed that decreasing ambient O 2 levels from 27% (hyperoxia) to 5% (hypoxia) decrease Pchlide levels in etiolated seedlings. To show that this is not just the result of hypoxia-related flux restrained by O 2 -requiring enzymes of the pathway we also analysed Article steady-state Pchlide levels for the A. thaliana flu mutant. FLU directly inhibits the first committed enzyme of tetrapyrrole synthesis, glutamyl tRNA reductase (GluTR)-the major form of which is encoded by HEMA1-thereby inhibiting synthesis of 5-aminolevulinic acid (ALA), the precursor of all tetrapyrroles 15 , which results in prevention of accumulation of free Pchlide (Extended Data Fig. 1a). The proposed mechanism for FLU activity involves POR, Pchlide, GluTR and CHL27 (a component of the tetrapyrrole synthesis enzyme Mg-protoporphyrin monomethylester cyclase 3 ); in the dark, free Pchlide (which cannot be converted to chlorophyllide without light) binds POR as part of a complex with CHL27 and FLU. This inactivation complex enables FLU to interact with and inhibit GluTR activity, reducing ALA synthesis and therefore Pchlide levels 16,17 . Under mild hypoxia, compared with normoxia and hyperoxia, steady-state Pchlide levels were unrestrained in the flu mutant, indicating that the observed oxygen-associated Pchlide levels in different species are regulated by oxygen sensing (Fig. 1b, Extended Data Fig. 1b-d). In A. thaliana accession Col-0 (originally collected less than 100 m above sea level (a.s.l.)) grown at 48 m a.s.l. (pO 2 = 21.2 kPa), this decrease was largely abolished in the absence of all 5 ERFVII transcription factors (using the pentuple related to apetala (rap) and hypoxia responsive erf (hre) mutant 18 rap2.12 rap2.2 rap2.3 hre1 hre2 (hereafter referred to as erfVII)) ( Fig. 1c), demonstrating that oxygen sensing via ERFVIIs is required for this response. A Cys-to-Ala mutation (Cys2Ala) (which removes the Cys N-degron) in all ERFVIIs significantly reduced Pchlide levels in etiolated seedlings, similar to prt6 seedlings (in which all ERFVIIs are stable) (Fig. 1d). Mutation of either PRT6 or ERFVII genes led to opposite stable changed steady-state Pchlide levels (lower in prt6, higher in erfVIIs); mutation of RAP2.3 had the largest effect, indicating that all ERFVIIs contribute to this function and that RAP2.3 may have a predominant role (Extended Data Fig. 2a, b). A prt6 flu double mutant showed high levels of Pchlide equivalent to those in the flu mutant, indicating that stabilized ERFVIIs act upstream of FLU (Extended Data Fig. 2c). Because hypoxia suppressed Pchlide levels, we analysed the role of ERFVIIs in regulating expression of genes encoding components of the inactivating complex-including FLU, HEMA1, CHL27 and POR (which is encoded by three genes, two of which (PORA and PORB) are expressed in etiolated seedlings 19 )-and CHLM, which encodes a chlorophyll synthesis enzyme previously shown to be regulated by oxygen sensing 18 . Accumulation of FLU transcripts was positively regulated by ERFVIIs via PRT6, whereas CHL27, PORA, PORB and CHLM were negatively regulated, and HEMA1 was not regulated, by this pathway (Extended Data Fig. 2d). The amount of FLU RNA increased strongly with increasing hypoxia, a response that was abolished in the erfVII mutant, whereas expression of tetrapyrrole synthesis genes showed the opposite trend (Fig. 1e, Extended Data Fig. 2e). Reflecting the transcript data, FLU protein levels were enhanced by hypoxia via ERFVIIs, whereas POR protein accumulation was repressed by hypoxia through ERFVIIs (Extended Data Fig. 2f). Oxygen-controlled repression of FLU-orthologous RNA accumulation was conserved in S. lycopersicum and P. somniferum (Fig. 1f).
In A. thaliana, the C-terminally haemagglutinin (HA)-tagged ERFVIIs HRE2 and RAP2.3 associated with the FLU gene in a region containing an evolutionarily-conserved ERFVII binding site 20 , adjacent to the initiating ATG (similar binding was previously observed for HRE2 at the FLU locus during the response of light-grown seedlings to hypoxia 20 ). HRE2-HA and RAP2.3-HA also showed differential binding to two genes (PCO1 and CRU1) previously shown to be regulated by ERFVIIs. HRE2 similarly associated with ATG-proximal regions of CHL27 and CHLM genes, and   also with gene regions of PORA and PORB but not with HEMA1 ( Fig. 1g, Extended Data Fig. 3). These data demonstrate that O 2 regulation of Pchlide synthesis occurs via oxygen sensing of the PCO N-degron pathway through ERFVII-regulated control of expression of components of the GluTR inactivation complex, in particular the negative regulator FLU.
Because pO 2 decreases with altitude ( Fig. 2a), and could thus affect steady-state Pchlide levels, we investigated the relationship between Pchlide levels and altitude. We analysed Pchlide in accessions of A. thaliana (collected from Eurasia and Africa), Solanum habrochaites (collected from South America), Solanum cheesmaniae (collected from the Galapagos Islands) and B. distachyon (collected from Turkey) from natural populations growing at different altitudes from sea level to more than 3,000 m a.s.l. with different latitudinal ranges ( Fig. 2a, b, Extended Data Fig. 4a, b, Supplementary Table 1a). When analysed at 48 m a.s.l. (pO 2 21.2 kPa), all species showed a positive correlation between altitude of accession collection and Pchlide steady state level. Furthermore, the amount of Pchlide in A. thaliana and S. habrochaites was dependent on the ambient O 2 concentration, and hypoxia resulted in greater reduction of Pchlide levels in accessions from higher altitudes (Fig. 2b, c). In A. thaliana, the steady-state Pchlide level was not related to submergence tolerance (Extended Data Fig. 4c, Supplementary Table 1c). As a result of increased Pchlide (which produces singlet oxygen 15 under light), following transfer to light, dark-grown seedlings of accessions collected at higher altitude accumulated substantially more reactive oxygen species (ROS) at 48 m a.s.l. than those collected at a lower altitude; this effect was dependent on the ambient O 2 level and functional PRT6 (Extended Data Fig. 5). These results show that there is a relationship between Pchlide accumulation and the altitude of accession collection, suggesting adaptation of oxygen sensing and/or downstream signalling through the PCO N-degron pathway that fine-tunes steady-state Pchlide level to local atmospheric O 2 levels, potentially avoiding damaging light-associated singlet-oxygen ROS production. Consistent with such a relationship, we found that accumulation of FLU transcript in A. thaliana grown at 48 m a.s.l. (pO 2 = 21.2 kPa) was strongly influenced by the altitude at which the accession was collected (that is, there was less FLU transcript in high-altitude accessions) and by ambient O 2 level (Fig. 2d). There was a similar relationship between altitude of collection and FLU expression for S. habrochaites (FLU expression was lower in high-altitude accessions) (Fig. 2e). Conversely, the amounts of PORA, PORB and CHLM (but not CHL27 or HEMA1) transcripts were higher in accessions from higher elevations, and lower at 15% oxygen compared with those grown in 21% oxygen (Extended Data Fig. 6a). POR protein accumulated to higher levels in the high-altitude A. thaliana accession Sha (originally collected in Tajikistan at 3,400 m a.s.l, pO 2 = 13.9 kPa) than in the low-altitude Col-0, and by introgressing the prt6-1 transfer-DNA insertion mutation from Col-0 through eight back-crosses into Sha (prt6 Sha ), we showed that this increase was controlled through the oxygen-sensing pathway; conversely FLU protein accumulation was increased in prt6 Sha compared with Sha (Extended Data Fig. 6b). In contrast to the observed relationship between the steady-state Pchlide level and elevation in wild species, an altitudinal cline of cultivated Chenopodium quinoa (quinoa, recently domesticated in the high Andes 21 ) did not show altitude-dependent Pchlide accumulation (Fig. 2f).
Reduced pO 2 is just one altitude-dependent parameter: others include atmospheric pressure. We therefore assessed steady-state Pchlide levels and FLU expression in etiolated seedlings at two sites located at extreme altitudes (site SB: 52.829809° N −1.249732° E, 48 m a.s.l., pO 2 = 21.2 kPa; and site ESPE: −0.312917° N −78.445157° E, 2,479 m a.s.l., pO 2 = 15.7 kPa) (Extended Data Fig. 7). The Pchlide level was much lower in the high-altitude accession Sha when it was grown at   Article the high-altitude site ESPE compared with the low-altitude site SB, whereas in the erfVII mutant Pchlide level remained similar at both sites (Extended Data Fig. 8a). Steady state levels of Pchlide in an A. thaliana altitudinal cline were reduced at ESPE, compared to SB at pO 2 21.2 kPa, and were similar to accumulation at SB in 15% ambient O 2 (Fig. 2g). The amount of FLU transcript was also increased in plants grown under ambient O 2 concentration at ESPE compared with SB, particularly in high-altitude accessions, and was similar to the amount of FLU transcript in plants grown at SB under 15% O 2 (Fig. 2d, h), whereas the amounts of PORA, PORB and CHLM transcripts were lower in the plants grown under atmospheric O 2 concentration at ESPE compared with SB (Extended Data Fig. 8b). These data demonstrate that O 2 is the major component sensed by altitudinal clines when controlling steady-state Pchlide levels. They also indicate that high-altitude populations have adapted to lower ambient O 2 through increased O 2 sensitivity, and that this has occurred in phylogenetically distant angiosperm species. Reduced accumulation of FLU transcripts in high-altitude accessions at pO 2 21.2 kPa further suggests that they exhibit lower ERFVII activity compared with low-altitude accessions. In summary, the sensitivity to O 2 determines downstream Pchlide level, which is tailored to the local ambient pO 2 via regulation of FLU expression mediated by ERFVIIs.
Since the analyses carried out at different altitudes showed that the sensitivity to atmospheric O 2 -as indicated by accumulation of Pchlide and its regulatory component FLU-increases with altitude, we examined the expression of classical hypoxia-associated genes. The expression of ADH1, PDC1, PGB1 and SUS4 transcripts (members of the core 49 conserved hypoxia-induced genes 22 that are regulated by ERFVIIs through the PCO N-degron pathway 23 ) was highly dependent on both the altitude of accession collection and ambient oxygen level in both A. thaliana and S. habrochaites (Fig. 3a, Extended Data Figs. 8c, d). This indicates that altitude adaptation is not restricted to Pchlide accumulation and is a central conserved feature of oxygen sensing in angiosperms. Furthermore, it suggests that-similar to FLU regulation-at high oxygen levels, ERFVIIs are more active in low-altitude accessions than in high-altitude accessions. We next tested whether ERFVIIs are active in etiolated seedlings of the low-altitude accession Col-0 at pO 2 21.2 kPa. Pchlide level in Col-0 was intermediate between those in prt6 and erfVII mutants (Fig. 1d), indicating ERFVII repressive activity. Increasing PCO2 activity in Col-0 led to increased levels of Pchlide, suggesting that active ERFVIIs can be destabilized, but this over-accumulation was inhibited by removal of PRT6 activity (Fig. 3b).
To understand the genetic mechanisms involved in altitude adaptation through oxygen sensing, we investigated the influence of components of the ERFVII-PCO N-degron pathway in the high-altitude A. thaliana accession Sha. We transformed wild-type and Cys2Ala stable versions of the Col-0 ERFVIIs RAP2.3 and HRE2 (expression driven by their own promoters) into Sha. In prt6 Sha , RAP2.3(Cys2Ala)-expressing Sha and HRE2(Cys2Ala)-expressing Sha (but not in Sha expressing wild-type Col-0 RAP2.3), FLU, PORA and PORB expression were markedly affected, resulting in reduced Pchlide (Fig. 3c, d). In addition, expression of hypoxia-related genes was enhanced in prt6 Sha and RAP2.3(Cys2Ala)-expressing Sha (Extended Data Fig. 9a, b). This indicates that constitutively stabilized ERFVIIs reduce Pchlide as effectively in Sha as in Col-0. Therefore, components downstream of stabilized ERFVIIs, including FLU function, are unaltered in Sha (as otherwise altering upstream components would not affect the high Pchlide level in Sha). Expression of components of the PCO branch of the PRT6 N-degron pathway were not significantly different between Col-0 and Sha, and the sequences of Col-0 and Sha FLU genes were identical (Extended Data Fig. 9c, d). Etiolated seedlings from Sha × erfVII reciprocal crosses showed low Pchlide steady-state levels, indicating that ERFVII activity (in repressing Pchlide levels) in Sha has the potential to be as strong as that in Col-0 (Extended Data Fig. 10). This result also suggests the presence of dominant repressor(s) of oxygen sensing in the low-altitude accession Col-0. We further analysed ERFVII activity in Col-0 and Sha by measuring HRE2-HA occupancy on FLU and hypoxia-related genes (which have been shown to bind HRE2 (ref. 20 )). At a pO 2 of 21.2 kPa, HRE2-HA occupancy of these genes was high in the prt6 mutant in both Sha and Col-0 backgrounds (in which HRE2 is stabilized), whereas in the wild type, HRE2-HA displayed lower occupancy of these genes in Sha than   in Col-0 (Fig. 3e). Occupancy increased in Sha and Col-0 when grown in 15% ambient O 2 . Thus, at high ambient oxygen ERFVII occupancy is higher in the low-altitude genetic background, but HRE2 activity in both low-and high-altitude genetic backgrounds responds to hypoxia. Finally, Western blot analysis showed that at pO 2 21.2 kPa, HRE2-HA abundance was higher in Col-0 than in Sha, but treatment with the proteasome inhibitor bortizomib resulted in equivalent markedly higher HRE2-HA abundance in both accessions (Fig. 3f).
Here we demonstrate that altitude adaptation involves genetic modifications of the sensitivity to atmospheric O 2 through oxygen sensing, mediated by ERFVII accumulation and activity. We show that this adaptation, through ERFVII regulation, influences two distinct features: hypoxia-related gene expression and steady-state levels of Pchlide (mainly via oxygen-regulated expression of FLU and POR). Prevention of free Pchlide accumulation is a result of the regulated rate of ALA synthesis, mediated by the POR-Pchlide-CHL27 complex that triggers FLU inactivation of GluTR 24 . Coupling of POR expression to the ambient oxygen concentration may enable matching of POR protein to Pchlide levels, allowing POR to bind to free Pchlide. It may be ecologically important to match tetrapyrrole flux (which requires molecular O 2 ) to the ambient O 2 concentration to provide the most effective Pchlide level once seedlings arrive at the soil surface and chlorophyll synthesis commences in the light. Altitude adaptation involves fine-tuning the activity of the oxygen-sensing system, which acts like a rheostat measuring altitude (Fig. 3g), perhaps through negative regulation of the PCO N-degron pathway. Oxygen sensing is transduced by ERFVIIs to influence outputs including expression of hypoxia-related genes, steady-state Pchlide levels (through regulation of ALA synthesis by inactivation complex components) and potentially other biochemical pathways that require molecular oxygen. Altitude adaptation enables decoding of the ambient oxygen level (determined by the local pO 2 ) to provide equivalent outputs at different altitudes, resulting in, for example, equivalent Pchlide levels in a low-altitude (high pO 2 )-adapted accession grown at low altitude to those in in a high-altitude (low pO 2 )-adapted accession grown at high altitude.
Although this study only investigated etiolated seedlings, other stages of development (including analogous stages such as subsurface rhizome-derived etiolated shoots) may also be subject to a similar adaptive mechanism, as the steady-state Pchlide level is an important regulator of chlorophyll synthesis and FLU has been shown to influence the chlorophyll supply in light conditions 24 . Notably, regulation of tetrapyrrole synthesis by oxygen sensing also occurs in cyanobacteria, which share a common ancestor with chloroplasts 25 . Previous studies have shown that oxygen sensing is an important feature of skotomorphogenesis 18 . Here we demonstrate that oxygen sensing during this important developmental stage is associated with genetic adaptation to altitude. We show that local ambient absolute O 2 concentration regulates expression of hypoxia-related genes and steady-state levels of Pchlide-a biochemical intermediate of chlorophyll biosynthesis-in altitudinal clines of diverse species through the oxygen-sensing system. This provides a general mechanism for adaptation to absolute elevation that is likely to be conserved throughout angiosperms. In addition, as this mechanism appears to not have been selected in breeding of quinoa, it may represent an untapped trait for crop improvement at unadapted altitude 26 . It also represents a component that deserves investigation in relation to plant ecological adaptation. The relevance of this mechanism will be of increasing importance as global warming leads to displacement of wild and crop plants to higher altitudes 27 .

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-022-04740-y.

Methods
No statistical methods were used to predetermine sample size. Experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Plant materials
Information on A. thaliana, S. habrochaites, S. cheesmaniae, B. distachyon P. somniferum and C. quinoa accessions (Supplementary Table 1 ('Lauren's grape') seeds were obtained from Mr. Fothergill's seeds, UK, and tomato Solanum lycopersicum (Marmande) were obtained from Sutton Seeds, UK. The Col-0 accession prt6-1 T-DNA insertion was introgressed through eight back-crosses into the Sha accession following BASTA resistance conferred by the transgene. The 35S:HRE2-3×HA transgene 10 from Col-0 was introgressed into Sha through five back-crosses following BASTA resistance conferred by the transgene, genotyping across all five chromosomes was used to confirm absence of detectable Col-0 genomic DNA away from both transgenes.

Growth conditions for etiolated seedlings
As previously described 18 , surface-sterilized seeds were plated on 0.5× MS media (including 1% w/v sucrose) and chilled for 4 days at 4 °C before being exposed to constant white light at 20 °C for 8 h to activate germination. Subsequently, unless indicated otherwise, plates were incubated in darkness at 20 °C for 5 days. For experiments using different oxygen levels, open plates were placed in a methacrylate chamber (1,120 × 180 × 270 mm) (Epica) and flushed until equilibrium with water-saturated premixed gas combinations (BOC) and left for 5 days in the dark. Oxygen levels in the chamber were measured at the beginning and end of the experiment using an oxygen meter attached to the outlet pipe (KANE 250 Compact Flue Gas Analyzer-Kane International).

Generation of transgenic plants
Individual C2A and WT ERFVII transgenes, including 2 kbp of sequence upstream of the initiating ATG and introns in accession Col-0 were described previously 18 . These were transformed into A. thaliana accession Sha as previously described 33 .

Biochemical analyses
Measurements of Pchlide and ROS. Pchlide was assayed from cotyledons isolated from etiolated seedling as described 34 . Cotyledons (or whole coleoptiles of B. distachyon) of etiolated seedlings were homogenized in 1 ml ice-cold 80% (v/v) acetone overnight at 4 °C. Extractions were vortexed and centrifuged, 800 μl was aliquoted into a fresh tube and 200 μl was used to measure the relative fluorescence at room temperature (excitation: 440 nm; emission: 550-750 nm) using either a Varioskan Flash (Thermo Fisher Scientific) for measurements at SB, or BioTek Cytation 5 Multi-Mode Reader (for measurements at ESPE). To account for differences in cotyledon size between accessions, images of 20 representative cotyledons for each accession in each experiment were taken using a Leica MZ75 and the area was measured using Fiji (https://imagej.net/). Reported Pchlide values represent the florescence (either from 550 to 700 nm or at 636 mn (arbitrary units)) per cotyledon in a 1 mm 2 area. ROS were detected in 5 days old etiolated seedlings 1 day after transfer to light, 10 μM 2′,7′-dichlorodihydrofluorescein diacetate (H 2 DCFDA) was incubated with seedlings for 30 min and then washed in 10 mM MES, 0.1 mM CaCl 2 , pH 6, for 1 h at 22 °C. Dye excitation was at 480 nm and emitted light was detected at 535-550 nm with a Leica DM5000 B. ROS and chlorophyll quantification was carried out using Fiji.
ChIP, gene expression and protein analyses. ChIP was performed as described 35 . Chromatin was extracted from ~3.5 g of etiolated seedlings. Ten microgrammes of anti-HA (Sigma, H3663-200UL) was used for immunoprecipitation. EUKARYOTIC TRANSLATION INITIATION FACTOR 4A1 (EIF4A1, AT3G13920) and ACTIN 2 (ACT2, AT3G18780) were used as negative targets. CRUCIFERINA (CRU1, AT5G44120) and PLANT CYSTEINE OXIDASE 1 (PCO1, AT5G15120) served as a known positive targets for RAP2. 3 and HRE2 (refs. 20,29 ). Chromatin was purified using QIAquick PCR Purification Kit (Qiagen). Quantitative PCR values of the immunoprecipitation product were normalized against the no-antibody samples. Data presented shows the average of three independent biological repeats. Oligonucleotide primers for ChIP are listed in Supplementary Table 2. For quantitative real time (QrtPCR), RNA was extracted from etiolated seedlings using a RNeasy mini kit (Qiagen) and DNase I on-column digestion (Sigma). First strand cDNA was synthesized using a qScript cDNA Synthesis Kit (Quantabio) from 0.8 μg RNA. rtPCR was performed using PerfeCTa SYBR Green FastMix (Quantabio) and the BioRad CFX96 qPCR system. Relative values were normalized over housekeeping genes in the appropriate species. Data presented shows the average of three independent biological repeats. Oligonucleotide primers used for rtPCR are shown in Supplementary Table 2.
Western blots were carried out as previously described 36 . The mouse anti-HA antibody (Sigma-Aldrich) was used at a concentration of 1:1,000 dilution, anti-POR (Agrisera AS05 067-10) was used at 1:4,000 dilution, and anti-FLU (obtained from B. Grimm) was used at 1:2,000 and the secondary antibody, goat anti-mouse IgG1 horseradish peroxidase conjugate (Thermo Fisher Scientific) was used at 1:20,000 dilution. Proteins were extracted from 5-day-old etiolated seedlings under green light as previously described 37 . Bortizomib treatment was carried out in 6-well tissue culture plates (Fisher Scientific); a total reaction volume of 3 ml of 50 μM Bortezomib (Santa Cruz Biotechnology) was added to each well. Negative controls contained an equivalent volume of DMSO (Sigma-Aldrich) (DMSO was used to reconstitute the Bortezomib powder). Five-day-old etiolated seedlings were transferred to each well by laying the seedlings on top of the solution in each well gently ensuring that the roots of each seedlings were immersed in the reaction solution. The plate was covered with aluminium foil and placed on a flatbed shaker (60 rpm) for 2 h, after which material was dried with paper towel to remove excess water and frozen in liquid nitrogen immediately.

Phylogenetic analyses
Genomic DNA and CDS sequences for FLU, CHLM and hypoxia-related gene orthologues were obtained from KEGG: Kyoto Encyclopedia of Genes and Genomes (https://www.kegg.jp/kegg/). FLU, CHLM genomic sequences were searched manually for potential ERFVII binding sites and HRPE-like sequences 20 .

Statistics
For experimental analysis of Pchlide and quantitative reverse transcription PCR, three independent replicates with different biological material are reported for each experiment. For ChIP, three independent replicates with different biological material were used per experiment. Each experiment was repeated at least twice. For analysis of ROS via staining and microscopy 8-15 seedlings were analysed and images taken for representative samples, experiment was repeated at least three times. In all cases measurements were taken from distinct samples. Differences in ROS content were tested by general linear model (GLM) with two factors (genotype and oxygen) with fixed effects. GLM tests were carried out with SPSS v.27. All graphs were produced using Graphpad software (Version 8), also used to calculate standard deviation of the mean for all samples tested. Relationships between dependent and independent variables were assessed by linear regression analysis using standard parameters for coefficiency of determination (R 2 ) in Graphpad.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability
Full versions of all blots are provided in Supplementary Fig 1. Unique identifiers for genes from all species analysed are listed in the text. Where appropriate, seeds of accessions and transgenic lines are available from the corresponding author. Source data are provided with this paper.