An Improved HRPE-Based Transcriptional Output Reporter to Detect Hypoxia and Anoxia in Plant Tissue

Oxygen levels in plant tissues may vary, depending on metabolism, diffusion barriers, and environmental availability. Current techniques to assess the oxic status of plant cells rely primarily on invasive microoptodes or Clark-type electrodes, which are not optimally suited for experiments that require high spatial and temporal resolution. In this case, a genetically encoded oxygen biosensor is required instead. This article reports the design, test, and optimization of a hypoxia-signaling reporter, based on five-time repeated hypoxia-responsive promoter elements (HRPE) driving the expression of different reporter proteins. Specifically, this study aimed to improve its performance as a reporter of hypoxic conditions by testing the effect of different untranslated regions (UTRs) at the 5′ end of the reporter coding sequence. Next, we characterized an optimized version of the HRPE promoter (HRPE-Ω) in terms of hypoxia sensitivity and time responsiveness. We also observed that severe oxygen deficiency counteracted the reporter activity due to inhibition of GFP maturation, which requires molecular oxygen. To overcome this limitation, we therefore employed an oxygen-independent UnaG fluorescent protein-coupled to an O2-dependent mCherry fluorophore under the control of the optimized HRPE-Ω promoter. Remarkably, this sensor, provided a different mCherry/UnaG ratiometric output depending on the externally imposed oxygen concentration, providing a solution to distinguish between different degrees of tissue hypoxia. Moreover, a ubiquitously expressed UnaG-mCherry fusion could be used to image oxygen concentrations directly, albeit at a narrow range. The luminescent and fluorescent hypoxia-reporters described here can readily be used to conduct studies that involve anaerobiosis in plants.


Introduction
The study of low oxygen (hypoxia) conditions has attracted growing attention in recent years across several fields of biology, including plant science [1][2][3]. In plants, hypoxia is a well-characterized stressful condition associated with submergence or waterlogging, since it restricts respiratory metabolism. Despite its apparent negative consequences for efficient energy production, hypoxia has been observed to occur as a chronic endogenous condition in several different tissues. These include tissues with obvious restriction in gas diffusions, such as tubers, fruits, and seeds, but chronic hypoxia was also recently measured in meristematic tissues [4][5][6]. For example, hypoxia-induced gene expression was shown to occur in conjunction with the early stages of lateral root primordia development, a five-time repeat of the HRPE element through the combination of O 2 -dependent and independent reporter proteins, providing a means to detect hypoxia and anoxia in plant tissue.

Plant Materials
Arabidopsis thaliana Columbia-0 seeds were used as a wild-type ecotype. The HRPE:GUS-GFP lines were previously described [7]. The HRPE-Ω:GUS-GFP and HRPE-ADH:GUS-GFP lines were newly generated as described in the cloning section. Wildtype Nicotiana benthamiana was used for transient transfection.

Plants Growth Conditions
Arabidopsis thaliana seeds for in vitro cultivation were sown on half-strength, agarized Murashige and Skoog medium and stratified for 48 h at 4 • C in the dark. Seeds were then germinated in short-day conditions (12 h light 12 h dark, 23 • C, 50% relative humidity, 100 µmol/m 2 /s light intensity). Seven-days old seedlings were used for GUS staining and GFP imaging following hypoxic treatment. Nicotiana benthamiana seeds were germinated on moisturized filter paper and then transferred to a soil-perlite mixture (3:1 ratio). Plants were then grown in long-day conditions (18 h light, 6 h dark, 23 • C, 50% relative humidity, 100 µmol/m 2 /s light intensity). Leaves of four-weeks old plants were used for transient transfection.

Hypoxia Treatments
Hypoxia treatments were performed by placing the plants in a Gloveless Anaerobic chamber (COY). Mixing of nitrogen gas and atmospheric air was performed to reach the indicated oxygen concentration for each experiment. A Pyroscience FireStingO2 (FSO2-2) oxygen meter, together with OXSP5 sensor spots, were used to confirm the desired oxygen concentration inside the glovebox.

Constructs Cloning and Assembling
HRPE promoter variants (Table S1) were de novo synthesized as gateway entry vectors by GeneArt service (Thermo-Fisher Scientific, Waltham, MA, USA). Starting from the previously described Hypoxia-responsive promoter element (HRPE) [7], we introduced the 5 -leader sequence (called Ω) of tobacco mosaic virus (TMV) [33] downstream of the HRPE promoter, generating the HRPE-Ω. Similarly, the 5 UTR (-254 upstream of the ATG) sequence of At1g77120 (ADH1) was placed downstream of the HRPE promoter, generating the HRPE-ADH. Both promoter units were designed flanked by attL sites for subsequent application in gateway cloning. For transient experiments, HRPE entry vectors were recombined in the pGreen800GW destination vector [34,35] by gateway cloning. The HRPE-Ω:GG (GUS-GFP) and HRPE-ADH:GG constructs were realized by gateway cloning using the pH7GWFS7 [36] destination vector.
Transcriptional units encoding the Pp2FbFP, iLov, and UnaG fluorophores (Table S2) were designed as DNA strings carrying an additional CACC sequence at the 5 -end for immediate subcloning into the pENTR/D-TOPO ® vector (Thermo-Fisher Scientific). For protoplast transfection, all entry fluorophore vectors were recombined into the p2GW7 [36] destination vector using gateway cloning. Gateway reactions were performed using the Gateway™ LR Clonase™ II Enzyme mix (Thermo-Fisher Scientific).

Fluorescence Microscopy
Seven-days old seedlings were kept for 16 h at either 1%, 2.5%, or 21% v/v oxygen concentration and then used for GFP imaging. Imaging was performed with a Leica THUNDER imager model organism using bandpass filters for GFP (excitation: 470/40 nm, emission: 525/50 nm) and RFP (excitation: 546/10 nm, emission: 605/70 nm). Confocal laser scanning microscopy was performed using a Zeiss airyscan 800. Fiji was used to quantify UnaG and mCherry fluorescence intensity [38]. Each data point represents the average mCherry/UnaG ratio at the nuclei and cytosol.

Statistical Tests and Data Representation
One and two-way analysis of variance (ANOVA) were performed using GraphPad Prism 7.0. Boxplot limits represent the 25th and 75th percentiles of each dataset. The whiskers extend to the lowest and highest data point. The central line represents the median. Histograms represent the mean ± the standard deviation.

Histochemical GUS Staining
Histochemical GUS staining of seedlings expressing the HRPE variants fused to GUS-GFP, was performed by four hours or overnight incubation with GUS staining solution (100 mM buffer phosphate, 0.1% Triton X-100, EDTA pH 8 10 mM, potassium ferrocyanide 0.5 mM, potassium ferricyanide 0.5 mM, X-Gluc 200 mM) and then cleared in several washes of 70% (v/v) ethanol. Images were taken using the THUNDER imager model organism (Leica microsystems).

RT-qPCR
To assess mRNA levels of GFP, GUS, and PCO1 (At5g15120), seven-day old Arabidopsis thaliana seedlings were grown vertically on agarized half-strength MS plates. Full seedlings were harvested after four hours of 21% or 0% O 2 treatments. Total RNA extraction, subsequent DNase treatment, cDNA synthesis, and RT-qPCR analysis were performed as described previously [28].

Transient Transfection of Nicotiana Benthamiana
Leaves of four-weeks old Nicotiana benthamiana were used for agroinfiltration and transient transfection. Agrobacterium tumefaciens cultures (strain GV3101) were grown overnight in LB media, selection antibiotics (50 ug/mL), and 20 µM acetosyringone. The cultures were then pelleted, and the bacteria were resuspended in a 200 µM acetosyringone MMA solution (MS 5 g/L, MES 1.95 g/L, sucrose 20 g/L, pH 5.6), reaching a final OD600 of 0.4. The abaxial sides of the third, fourth, and fifth leaves of Nicotiana benthamiana plants were infiltrated with the Agrobacterium solution using a 5 mL syringe. Following infiltration, plants were kept in the dark for 4 h and then moved to standard long day growing conditions as described before. Disks cut out of infiltrated leaves were used after 48 h for confocal imaging or protein extraction.

Reporter Activity Assay
Four-weeks old Nicotiana benthamiana plants were grown for 48 h in long-day conditions after agroinfiltration. Leaves were then cut into disks, placed into 6-well plates filled with water and subjected to hypoxic treatment. Following treatment, leaf disks were harvested in liquid nitrogen and used for protein extraction following the Dual Luciferase Reporter (DLR) Assay System (Promega, Madison, WI, USA). Disks were ground in 400 µL Passive Lysis Buffer and then diluted 1:300 in the same buffer. Samples were vortexed, and 6 µL of protein extract was used for DLR assay. We used 30 µL of Luciferase Assay ReagentII (Promega) for firefly luciferase activity, and 30 µL of Stop and Glo (Promega) to quench firefly activity and induce renilla luciferase activity. All luciferase activity readings were performed using a Lumat LB 9507 luminometer (Berthold Technologies, Oak Ridge, TN, USA).

Optimisation and Characterization of HRPE-Based Hypoxia Reporters
We first set out to improve the dynamic range of the original HRPE-based reporter by improving the translation of its reporter protein. Previous reports highlighted how hypoxia generally inhibits the plant translation machinery, due to ATP shortage, while specific proteins are still selectively produced [36,37]. Thus, we reasoned that we could improve the translation of reporter genes under oxygen limitation by including the untranslated region (UTR) present at the 5 of hypoxia-inducible mRNAs or using a strong viral UTR. We selected the 5 UTR of Arabidopsis thaliana alcohol dehydrogenase (ADH1, AT1g77120) mRNA and the Ω-leader region present in the 35S promoter of the Tobacco Mosaic Virus (CaMV). We fused either new version of the HRPE promoters to a firefly luciferase (FLUC) reporter (HRPE-ADH:FLUC and HRPE-Ω:FLUC) and used a renilla luciferase (RLUC) driven by a 35S CaMV promoter as normalization control (Figure 1a). When transiently transfected in Nicotiana benthamiana leaves, the HRPE-Ω:FLUC showed a stronger increase in luminescence signal after five hours of hypoxia treatment, as compared to the original HRPE or the HRPE fused to the 5 UTR of ADH1 ( Figure 1b).
Next, we characterized the O 2 -sensitivity and the response-time of the HRPE-Ω construct. HRPE-Ω:FLUC signal was observed within two hours of hypoxia treatment, but was highest after four hours ( Figure 1c). Significantly increased HRPE-Ω:FLUC luminescence was observed progressively at O 2 concentrations of 5% and 1% v/v in air, while 10% O 2 showed a mild induction, although undetectable by parametric statistics (Figure 1d). This indicates that the strength of hypoxia-responses in plants depends on the length and severity of the hypoxic condition. Taken together, these data demonstrate that the HRPE-Ω reporter can be applied as a hypoxia signaling output reporter within two hours of treatment and is activated at a range of 0-5% O 2 .

Stable in Planta Expression of HRPE Reporters
We stably introduced the HRPE variants driving a chimeric GUS-GFP reporter protein (HRPE:GG, Figure 2a) in Arabidopsis thaliana plants. Strong overnight hypoxia (1% O 2 ) treatments led to an activation of HRPE promoter activity in all tissues, while a milder treatment (2.5% O 2 ) induced GUS activity primarily in the shoot apex, young primordia, and in the root (Figure 2b,c). In aerobic conditions, GUS staining was observed in the shoot apex of each reporter line, although only for HRPE-ADH and the original HRPE variant when GUS staining was performed overnight (Figure 2b, supplementary Figure S1). This is in line with the hypoxic status of this tissue [7]. Confocal microscopy imaging of GFP in 7-day old root tips of HRPE-Ω:GG plants showed that this tissue does not activate HRPE in aerobic conditions, while hypoxia led to the induction of GFP signal (Figure 2c). Remarkably, patchy patterns of green fluorescence were observed in hypoxic root tips, and a comparable signal was observed in HRPE-ADH:GG and the original HRPE:GG lines, indicating that it is not an artifact induced by the omega 5 UTR or the genomic position of the transgene (Figure 2d). the Tobacco Mosaic Virus (CaMV). We fused either new version of the HRPE promoters to a firefly luciferase (FLUC) reporter (HRPE-ADH:FLUC and HRPE-Ω:FLUC) and used a renilla luciferase (RLUC) driven by a 35S CaMV promoter as normalization control (Figure 1a). When transiently transfected in Nicotiana benthamiana leaves, the HRPE-Ω:FLUC showed a stronger increase in luminescence signal after five hours of hypoxia treatment, as compared to the original HRPE or the HRPE fused to the 5′ UTR of ADH1 (Figure 1b).  To investigate the activity of HRPE-Ω:GG under anoxia, we shortened the treatment time to four hours to avoid cell death. Anoxia treatments did not lead to an increase in GFP in root tips, and GUS staining revealed heterogenicity in HRPE-Ω:GG reporter activity under this condition (Figure 2e,f). RT-qPCR analysis of GUS and GFP transcripts in HRPE-Ω:GG plants revealed a strong increase in GUS and GFP mRNA upon anoxia, which was comparable to the induction of the endogenous hypoxia-inducible PCO1 transcript (Figure 2g). Therefore, while the HRPE-Ω shows robust activation upon anoxia, the aberrant induction of GUS activity and GFP fluorescence at anoxia hints at impaired translation or maturation of the reporter.

Generation of Anoxia Sensors
The O 2 -dependent maturation of GFP limits the usage of HRPE-Ω:GG to conditions at which the reporter is induced (>5% O 2 ), but also sufficient oxygen is available for GFP to fluoresce. Indeed, while anoxia treatments permitted variable, but detectable GUS activity, GFP fluorescence was completely impeded when driven by the HRPE-Ω (Figure 2e,f). Moreover, it is plausible that GFP fluorescence is at least partially affected at hypoxic conditions, leading to an under-appreciation of reporter activity. To circumvent this drawback of GFP, we investigated the possibility of using O 2 independent fluorophores, which have been characterized in vitro and in vivo in metazoans or bacteria [18,19]. Among these are the flavin mononucleotide binding cyan-green fluorophores FbFP, iLOV, and the bilirubin dependent green fluorescent UnaG protein. To test their potential application in plants, we first observed their fluorescent signal in transiently transformed Arabidopsis thaliana mesophyll protoplasts using a constitutive 35S: promoter. In protoplasts, detectable fluorescence was observed for Pp2FbFP and UnaG, but not for iLOV (Figure 3a). Among these, UnaG showed the strongest fluorescence, which matches reports from publicly available databases (Fpbase.com). RT-qPCR analysis of GFP, GUS, and PCO1 expression in HRPE-Ω:GG seedlings exposed to 21 and 0% O2. UBQ10 was used as a housekeeping gene. The expression level of each gene was calculated relative to its expression at 21% O2. Hypoxia treatments were performed overnight. Anoxia treatment was performed for 4 h. Two-sided t-test. Stars indicate a statistical significant difference (**** p value < 0.0001).

Generation of Anoxia Sensors
The O2-dependent maturation of GFP limits the usage of HRPE-Ω:GG to conditions at which the reporter is induced (>5% O2), but also sufficient oxygen is available for GFP to fluoresce. Indeed, while anoxia treatments permitted variable, but detectable GUS activity, GFP fluorescence was (g) RT-qPCR analysis of GFP, GUS, and PCO1 expression in HRPE-Ω:GG seedlings exposed to 21 and 0% O 2 . UBQ10 was used as a housekeeping gene. The expression level of each gene was calculated relative to its expression at 21% O 2 . Hypoxia treatments were performed overnight. Anoxia treatment was performed for 4 h. Two-sided t-test. Stars indicate a statistical significant difference (**** p value < 0.0001). Next, we generated hypoxia and anoxia reporters based on HRPE-Ω driving a fusion of UnaG and mCherry (HRPE-Ω:UnaG-mCherry, Figure 3a). mCherry requires oxygen for maturation, and therefore, should not fluoresce under anoxic conditions, which instead does not impair UnaG fluorescence. In this manner, we expected to distinguish between tissue anoxia, which should lead to exclusively green UnaG fluorescence, and moderately hypoxic tissue, showing green and orange fluorescence. To test this hypothesis, HRPE-Ω:UnaG-mCherry was transiently transformed in Nicotiana benthamiana leaves, and treated with different oxygen concentrations. At oxygen concentrations of 2% and 5% we could observe UnaG and mCherry fluorescence, while no signal was detected under 21% O2, confirming that the HRPE-Ω:UnaG-mCherry reporter drives expression in response to hypoxia (Figure 3c). In line with the requirement of oxygen for the maturation of mCherry, a lower ratio of mCherry/UnaG was observed at 2% versus 5% O2, while anoxic treatment only led to UnaG fluorescence and no detectable mCherry signal (Figure 3c,d). This shows that the HRPE-Ω:UnaG-mCherry variant can be used to detect tissue anoxia, while its ratiometric UnaG/mCherry output can be used to infer the actual O2 concentration. Based on these observations, we reasoned that a direct, and hypoxia signaling independent, O2 sensor could be generated using the same UnaG-mCherry fluorescent pair, but employing a constitutive UBQ10 promoter. Remarkably, a linear relationship between the mCherry/UnaG ratio and the O2 concentration was found at 0.5-5% O2 (Figure 4a,b). Instead, no significant difference was observed between 5% and 21% O2. Therefore, the ratiometric output of pUBQ10:UnaG-mCherry can be used to detect hypoxia, but it is not suitable for imaging of the oxic status of well-oxygenated tissue. Next, we generated hypoxia and anoxia reporters based on HRPE-Ω driving a fusion of UnaG and mCherry (HRPE-Ω:UnaG-mCherry, Figure 3a). mCherry requires oxygen for maturation, and therefore, should not fluoresce under anoxic conditions, which instead does not impair UnaG fluorescence. In this manner, we expected to distinguish between tissue anoxia, which should lead to exclusively green UnaG fluorescence, and moderately hypoxic tissue, showing green and orange fluorescence. To test this hypothesis, HRPE-Ω:UnaG-mCherry was transiently transformed in Nicotiana benthamiana leaves, and treated with different oxygen concentrations. At oxygen concentrations of 2% and 5% we could observe UnaG and mCherry fluorescence, while no signal was detected under 21% O 2 , confirming that the HRPE-Ω:UnaG-mCherry reporter drives expression in response to hypoxia (Figure 3c). In line with the requirement of oxygen for the maturation of mCherry, a lower ratio of mCherry/UnaG was observed at 2% versus 5% O 2 , while anoxic treatment only led to UnaG fluorescence and no detectable mCherry signal (Figure 3c,d). This shows that the HRPE-Ω:UnaG-mCherry variant can be used to detect tissue anoxia, while its ratiometric UnaG/mCherry output can be used to infer the actual O 2 concentration. Based on these observations, we reasoned that a direct, and hypoxia signaling independent, O 2 sensor could be generated using the same UnaG-mCherry fluorescent pair, but employing a constitutive UBQ10 promoter. Remarkably, a linear relationship between the mCherry/UnaG ratio and the O 2 concentration was found at 0.5-5% O 2 (Figure 4a,b). Instead, no significant difference was observed between 5% and 21% O 2 . Therefore, the ratiometric output of pUBQ10:UnaG-mCherry can be used to detect hypoxia, but it is not suitable for imaging of the oxic status of well-oxygenated tissue. Nicotiana benthamiana leaves after treatment with various O2 concentrations. mCherry maturation is O2-dependent, while UnaG is not. Images display the mCherry/UnaG ratio, corresponding to the tissue O2 concentration. (b) Quantification of mCherry/UnaG ratios at different O2 concentrations. The green, orange, and blue data points represent the 21% O2 control for 0.5%, 2%, and 5% O2, respectively. Crosses represent the means. The linear regression was calculated using the ratios at 0.5%, 2%, and 5% O2. Two-way analysis of variance (ANOVA) followed by Tukey post hoc test. Letters indicate a statistical significant difference (p value < 0.05).

Discussion
In this article, we reported the optimization and characterization of a hypoxia signaling responsive reporter, which was used to detect hypoxia and anoxia in vivo. Fusing a 5′ Ω-UTR to the end of the five-times repeat of the HRPE sequence improved the signal output under hypoxia, while the use of the ADH1 5′ UTR did not. The latter was unexpected since hypoxic conditions significantly induce ADH1, and its mRNA is known to be selectively translated at such conditions [36]. The increased hypoxia output conferred by the Ω-UTR likely represents a constitutive increase in translation efficiency, which may be able to overcome the general downregulation of translation associated with hypoxia [36]. The HRPE-Ω promoter sequence was found to drive expression of a reporter within two hours from the onset of hypoxia. This is slightly delayed compared with previous reports where hypoxia-responsive transcripts were found to be significantly increased within one hour of treatment [39,40]. This lag may represent the time required for translation and folding of the FLUC reporter protein after the onset of hypoxia, or it may hint at a different responsiveness of Nicotiana benthamiana as compared to Arabidopsis thaliana. The time required to observe a detectable signal of HRPE-Ω may be decreased through the use of a NanoLuc luminescent reporter, which is of smaller size and brighter, compared to FLUC [41]. The selection of a more rapidly maturing fluorophore is also a valid alternative to tackle this aspect. An extensive analysis of a collection of commonly used fluorophores revealed that while eGFP belongs to the rapidly maturating The green, orange, and blue data points represent the 21% O 2 control for 0.5%, 2%, and 5% O 2 , respectively. Crosses represent the means. The linear regression was calculated using the ratios at 0.5%, 2%, and 5% O 2 . Two-way analysis of variance (ANOVA) followed by Tukey post hoc test. Letters indicate a statistical significant difference (p value < 0.05).

Discussion
In this article, we reported the optimization and characterization of a hypoxia signaling responsive reporter, which was used to detect hypoxia and anoxia in vivo. Fusing a 5 Ω-UTR to the end of the five-times repeat of the HRPE sequence improved the signal output under hypoxia, while the use of the ADH1 5 UTR did not. The latter was unexpected since hypoxic conditions significantly induce ADH1, and its mRNA is known to be selectively translated at such conditions [36]. The increased hypoxia output conferred by the Ω-UTR likely represents a constitutive increase in translation efficiency, which may be able to overcome the general downregulation of translation associated with hypoxia [36]. The HRPE-Ω promoter sequence was found to drive expression of a reporter within two hours from the onset of hypoxia. This is slightly delayed compared with previous reports where hypoxia-responsive transcripts were found to be significantly increased within one hour of treatment [39,40]. This lag may represent the time required for translation and folding of the FLUC reporter protein after the onset of hypoxia, or it may hint at a different responsiveness of Nicotiana benthamiana as compared to Arabidopsis thaliana. The time required to observe a detectable signal of HRPE-Ω may be decreased through the use of a NanoLuc luminescent reporter, which is of smaller size and brighter, compared to FLUC [41]. The selection of a more rapidly maturing fluorophore is also a valid alternative to tackle this aspect. An extensive analysis of a collection of commonly used fluorophores revealed that while eGFP belongs to the rapidly maturating fluorophores and shows 90% fluorescence intensity within 62.8 ± 6.6 min at 32 • C, mGFPmut3d achieves this almost four-times as fast [42].
HRPE-Ω was activated when the external O 2 concentration dropped below 5%, which correlates with the accumulation of nuclear RAP2.12 and likely reflects the affinity of PCOs for oxygen [38,39]. Interestingly, in young HRPE-Ω:GG plants, HRPE-Ω GUS activity was primarily found at 2.5% O 2 in the root and shoot apex region, while 1% O 2 treatment led to an increase of the reporter in all tissues, hinting at a different sensitivity of these tissues to hypoxia. While the SAM is chronically hypoxic, reducing the external oxygen concentration likely accentuates this further, leading to a stronger increase in HRPE-Ω:GG activity, compared to the rest of the shoot [7]. Roots are more likely to experience hypoxic conditions, due to waterlogging, and may, therefore, respond more sensitively to mild hypoxic conditions. Curiously, hypoxia led to a patchy GFP signal in the root tip with significant differences in GFP intensity between cells. Although the distribution of the GFP signal appeared random rather than following a defined pattern or a gradient, this may hint at a different sensitivity to hypoxia of cells or at altered efficiency of GFP translation, depending on their state within the cell cycle [43].
Almost all genetically encoded biosensors rely on fluorescent proteins that undergo an oxygen-dependent maturation step to fluoresce, and this prevents their utilization under strongly oxygen limiting conditions. Similarly, the catalysis of luciferin requires oxygen, and therefore, in vivo analysis of pHRPE:FLUC activity, i.e., by spraying plants with luciferin, under strong oxygen limiting conditions, is not expected to result in detectable reporter activity. Here, we tested a previously identified O 2 -independent fluorescent protein, UnaG, and found that it was able to produce detectable fluorescence in plants subjected to anoxic conditions. This enables its use as reporter proteins in plants, in particular for the study of chronically hypoxic tissues, such as meristems and galls. Indeed, whilst mCherry was found to display reduced fluorescence at <2% O 2, the signal emitted by UnaG was confirmed as O 2 -independent (Figure 3c), indicating that the latter would be a more suitable reporter when performing experiments at O 2 concentrations below the 2% threshold. Based on these results, we generated an HRPE-Ω reporter driving an UnaG-mCherry fusion protein. Indeed, the superiority of this sensor compared to the GFP based one was apparent by its striking UnaG fluorescence under anoxia, compared to no detectable GFP signal. Likewise, no signal for mCherry was observed at 0% O 2 , and mCherry maturation was also negatively affected at 2% O 2 , but not at 5% O 2 , as indicated by a lower ratio of mCherry/UnaG intensity. Thus, while HRPE-Ω provides inducibility at 5% O 2 , its ratiometric output can also be employed to infer the actual oxygen concentration. Moreover, when expressing the mCherry/UnaG pair using a constitutive UBQ10 promoter, we observed a linear relationship between mCherry/UnaG and an O 2 concentration range of 0.5 to 5%. Therefore, while not suitable as a sensor for moderate to high O 2 levels, it can be used to detect strong hypoxia when also providing a means to detect spatial differences in the O 2 concentration within chronic hypoxic tissue, such as meristems.
While the HRPE-based sensors described here should prove as a useful and robust tool to detect hypoxia and anoxia in tissue and to quantify hypoxic responses, one should bear in mind that they act primarily as an output of hypoxia signaling, not actual O 2 levels. Recent reports found that ethylene, nitric oxide, and ATP levels impact the plant oxygen-sensing pathway, and this should, therefore, be taken into consideration when interpreting hypoxia-signaling output reporters [40,[44][45][46]. On the flip side, a combination of the here described HRPE-based reporters together with direct O 2 concentration measurements and the pUBQ10:UnaG-mCherry sensor may be an elegant strategy to disentangle hypoxia signaling from the tissue O 2 concentration.
Author Contributions: Experimental design and conceptualization were done by P.P. and D.A.W., G.P., S.I., E.D.M. and D.A.W. performed the experiments. D.A.W. wrote the manuscript. P.P. commented and corrected the manuscript. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.