Identification of novel plant cysteine oxidase inhibitors from a yeast chemical genetic screen

Abstract

Hypoxic responses in plants involve Plant Cysteine Oxidases (PCOs). They catalyze the N-terminal cysteine oxidation of Ethylene Response Factors VII (ERF-VII) in an oxygen-dependent manner, leading to their degradation via the cysteine N-degron pathway (Cys-NDP) in normoxia. In hypoxia, PCO activity drops, leading to the stabilization of ERF-VIIs and subsequent hypoxic gene upregulation. Thus far, no chemicals have been described to specifically inhibit PCO enzymes. In this work, we devised an in vivo pipeline to discover Cys-NDP effector molecules. Budding yeast expressing AtPCO4 and plant-based ERF-VII reporters was deployed to screen a library of natural-like chemical scaffolds and was further combined with an Arabidopsis Cys-NDP reporter line. This strategy allowed us to identify three PCO inhibitors, two of which were shown to affect PCO activity in vitro. Application of these molecules to Arabidopsis seedlings led to an increase in ERF-VII stability, induction of anaerobic gene expression, and improvement of tolerance to anoxia. By combining a high-throughput heterologous platform and the plant model Arabidopsis, our synthetic pipeline provides a versatile system to study how the Cys-NDP is modulated. Its first application here led to the discovery of at least two hypoxia-mimicking molecules with the potential to impact plant tolerance to low oxygen stress.

Keywords: Arabidopsis thaliana; ERF-VII; chemical genetics; enzyme inhibitor; high-throughput screening (HTS); hypoxia; low oxygen stress priming; plant cysteine oxidase; yeast.

Figure 1

In vivo search for chemical effectors of the cysteine N-degron pathway with a yeast platform.A, deletion of the S. cerevisiae ERG6 locus with the Delitto Perfetto strategy. B, schematic outline of the Cys-NDP reporter platform devised in this study. A synthetic cysteine N-degron pathway (Cys-NDP) was incorporated in the drug-sensitized pdr5Δ,erg6Δ, to obtain the “syNDP_ΔΔ” strain yeast strain, where proteasomal degradation of Cys-starting peptides is enabled thanks to the heterologous enzyme AtPCO4. This impinges on the native arginine N-degron pathway of yeast (38). The activity of the Cys-NDP pathway can be revealed by the stability of a genetically encoded reporter substrate, DLOR, and measured from its relative luciferase activity (Fluc/Rluc). DLOR harbors a ubiquitin cleavage site, indicated by the black arrowhead, that permits the post-translational generation of a cysteine N-degron. C, microtiter plate set-up adopted for the HTS platform. 200 μl of syNDP_ΔΔ cell suspension at OD₆₀₀ = 0.1 were dispensed in 96-well plate wells and supplemented with 1 μl DMSO (blue), MG132 (orange), cycloheximide (CHX, red), or different test compounds (gray). Clean media was dispensed in one well (white), to be used as a blank for spectrophotometric measurements. D, sample output plate from the screening (corresponding to plate 1). Fold change values indicate the variation in DLOR reporter activity (Fluc/Rluc ratio) from the average basal activity in DMSO-treated wells. Fold changes are represented as a three-colour scale where blue is the minimum fold change recorded on the plate, red is the average fold change from MG132-treated wells and white corresponds to 50% of the latter (values are specified below the color scale). E, growth (OD₆₀₀) profiles, over the total duration of the experiment (6 h), from three selected microcultures treated with CHX (red), compounds 1C2 (cyan) and 1G9 (yellow), or mock-treated cells from the same plate (black line). F, example of yeast generation times calculated during the chemical treatments between the 4 h and 6 h time points (data are from plate 1 of the screening). Data are represented as three-color scale where blue represents average generation times that are not different from the mock treatment, azure white shows generation times that are double that of the mock (taken as mild inhibition of growth) and orange represents generation times that are five times that of the mock (taken as slow cell division). Negative growth rates (associated with flattened growth curves such as CHX or 1G9 in Fig. 1E) were omitted (white wells).

Figure 2

Dissecting the action of Cys-NDP inhibitors on the components of the pathway with yeast DLOR reporters.A, fate of C-, D- and R-DLOR reporters according to Cys-NDP dependent regulation. ATE1 and UBR1 are endogenous Arginyl-tRNA-protein transferase 1 and E3 ubiquitin-protein ligase, respectively, composing the yeast Arg-NDP. B, effect of the selected compounds from Table S3 DLOR stability. Cells were treated with 50 μM of each compound (1/200 dilution in the microcultures) or 0.5% DMSO (v/v) for 6 h, as in the previous screening. Histograms are mean + SD (n = 4) of the fold change of DLOR activity (Fluc/Rluc) over the relative DMSO control treatment. The reference value from the previous screening is reported as a black column for each compound. The dashed line indicates the arbitrary threshold (fold change = 1.5 over DMSO) chosen for substrate stabilization at this stage. Fold change values can be found in Table S4. Two-tailed Student’s t test (n = 4) was performed on the treatments that induced DLOR stabilization; asterisks represent significantly higher Fluc/Rluc ratios for each DLOR version over their respective DMSO controls (two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, with Q = 1%); p < 0.05.

Figure 3

Impact of candidate Cys-NDP inhibiting compounds from the chemical screening on the 28RAPFluc reporter of Arabidopsis.A, luciferase activity (Fluc μg⁻¹ protein) after 6 h application of 100 μM chemicals or 1% v/v DMSO on 5-day-old seedlings grown in liquid media. Data are mean ± SD (n = 4). B, impact of 15 h-long dark anoxia on 7-day-old seedlings pre-treated with 50 μl of 1% DMSO (v/v) or 100 μM chemicals. Chemical treatments were administered 6 h before the stress. Better tolerance was indicated by higher chlorophyll levels at the end of a 6-day-long recovery period, digitally estimated from a higher density of green pixels in the shoot image area. Some DMSO-treated plants were kept unstressed (blue violin plot) for comparison. Individual and median values are shown for each treatment (n = 14–20). C, histochemical β-glucuronidase (GUS) assay on promADH:GUS (55) reporter plants (7 day-old, grown in vertical plates) treated with 100 μM of the indicated chemicals, 1% DMSO in normoxia or 1% DMSO in hypoxia (1% O₂ v/v) for 6 h. D, expression of six hypoxia-inducible marker genes in wild-type seedlings treated as in (C). E, overview of the PCOff pipeline and number of chemicals analyzed (n) that overcame each threshold in yeast (lilac) and plants (green), together with the 2-D structures of 2A10, 4D5, and 4C5, identified as possible hypoxia mimetic molecules, retrieved from their catalog numbers (Table S1) at Otava Chemicals. Box plots represent the median (line) and interquartile range (IQR), whiskers span from minimum to maximum values (n = 4). Statistical significance was determined with a one-way ANOVA followed by a Dunnett’s multiple comparison tests on relative Fluc activity (A, BZ treatment excluded), green pixel density (B) and relative mRNA level (D, hypoxic treatment excluded) (∗∗∗∗p < 0.0001; ∗∗∗, 0.0001 ≤ p < 0.001; ∗, 0.01 ≤ p < 0.05). BZ, bortezomib.

Figure 4

Inhibition of recombinant AtPCO4 activity by Cys-NDP inhibiting compounds.A, effect of 5 candidate Cys-NDP inhibiting compounds on AtPCO4 activity towards a peptide representing Nt-Cys initiating RAP2.12 (RAP_2-17 herein) relative to DMSO controls. RAP_2-17 (500 μM) was reacted with AtPCO4 (0.5 μM) pre-incubated with 1 mM inhibitor or DMSO vehicle only, for 10 min at 25 °C. A non-hit molecule from the screening (6C11) was included as a negative control. Data represent mean ± SD (n = 3), statistical significance determined comparing AtPCO4 activity co-incubated with each candidate compound compared with the negative control, by one-way ANOVA followed by a Dunnett’s multiple comparison test (∗∗∗∗p < 0.0001; ∗∗∗, 0.0001 ≤ p < 0.001). B and C, dose-response curves for 2A10 (B) and 4D5 (C) treatment of AtPCO4 activity; reactions were performed as described for (A) with inhibitor present at concentrations indicated by the logarithmic scale. Data are presented as mean ± SD (n = 3). D, autodock modeled structures of AtPCO4 (PDB 6S7E) (teal) in complex with 2A10 (yellow sticks, left) and RAP_2-8 (pink sticks, right). Both occupy the putative substrate binding site. Metal at the active site is shown by an orange sphere; AtPCO4 residues which line the active/substrate binding site are shown with teal lines. E, dose–response curves for AtPCO4 activity in the presence of increasing concentrations of DMSO, indicated by the logarithmic scale. Data are presented as mean ± SD (n = 3).