Proteome-wide tagging with an H₂O₂ biosensor reveals highly localized and dynamic redox microenvironments

Construction and Screening of a HyPer7/SypHer7 Fusion Library.

To create a library of yeast strains in which HyPer7 is fused to each individual gene product, the SWAp tag (SWAT) approach was used (13, 14). Briefly, the coding sequence for HyPer7 was inserted at the C-terminus of each individual open reading frame (ORF) to create a gene locus encoding a fusion protein with a 15-amino acid linker, ORF-RTLQVDGGGSGGGGS-HyPer7 (Fig. 1A). The resulting HyPer7 fusion protein library contains 5366 individual strains. In addition, a matching control library in which each protein is fused to the oxidation-insensitive HyPer7 mutant HyPer7(C121S), named SypHer7, was generated. Both libraries were verified at random by PCR and immunoblotting (SI Appendix, Fig. S1 A and B).

To demonstrate that fused HyPer7 is functional in principle, we treated selected library strains with antimycin A, an inhibitor of the electron transport chain (ETC) known to induce mitochondrial H₂O₂ generation. HyPer7 fused to ETC subunits was oxidized almost immediately, whereas cytosolic or nuclear fusion proteins responded after a lag phase, showing that protein-specific HyPer7 tagging provides spatiotemporal resolution (SI Appendix, Fig. S1C). For whole proteome measurements, the combined HyPer7/SypHer7 fusion protein library was arrayed and grown on agar plates containing synthetic complete defined (SCD) medium with 2% glucose (Glc) for 24 h (SI Appendix, Fig. S1D).

We previously reported that the fluorescence ratio of SypHer7 is higher than that of reduced HyPer7 (15). This was an unexpected observation because SypHer7 cannot form the disulfide bridge in the OxyR domain and therefore should mimic the fully reduced state of HyPer7. The biophysical reason for the higher-than-expected SypHer7 baseline ratio is unclear, but it can be assumed that the C121S mutation alters the conformational dynamics in such a way that the fluorophore protonation equilibrium is shifted toward deprotonation (i.e., a higher fluorescence ratio). We therefore asked whether the SypHer7 baseline shift applies to our libraries. An analysis of reference strains expressing the probes as free (unfused) proteins in the cytosol confirmed that SypHer7 exhibits a higher ratio than HyPer7 (Fig. 1B). The same phenomenon is seen when the ratio distributions of HyPer7 and SypHer7 fusion proteins are compared (Fig. 1C). As a consequence, direct ratio comparisons between HyPer7 and SypHer7 fusion proteins are not meaningful. We therefore chose a different approach to compare HyPer7 and SypHer7 responses in a stringent manner.

Unfused cytosolic HyPer7 is almost completely reduced (15). Thus, we first identified HyPer7 fusion proteins exhibiting ratios above the 99% percentile of the unfused HyPer7 ratio distribution (Fig. 1 D, Top). These fusion proteins are most likely to contain (partially) oxidized HyPer7. However, there is also the theoretical possibility that a fusion-protein-specific structural perturbation increases the ratio without oxidation. To identify potential oxidation-independent ratio effects, we identified SypHer7 fusion proteins with ratios above the 99% percentile of the unfused SypHer7 ratio distribution (Fig. 1 D, Bottom). We then excluded all proteins from analysis that showed an elevated ratio in both HyPer7 and SypHer7 fusions. Interestingly, these proteins mostly function in ubiquitination and proteolysis, suggesting that proximity of HyPer7/SypHer7 to these machineries can perturb the fluorophore in a redox-independent manner (SI Appendix, Fig. S2A and Dataset S1). The remaining set of HyPer7 fusion proteins we considered to be (at least partially) oxidized, as it strictly conformed to our two criteria, namely an increased ratio relative to unfused HyPer7 (Fig. 1 E, Upper) without an increase of the corresponding SypHer7 fusion protein relative to unfused SypHer7 (Fig. 1 E, Lower).

Applying the above criteria, growth on glucose-containing SCD medium for 24 h led to the identification of 43 oxidized HyPer7 fusion proteins (Fig. 1F and Dataset S2). Not unexpectedly, the majority of these proteins is mitochondrial. Numerous proteins are part of the ETC, including complex II (Sdh1, Sdh4), complex III (Qcr2, Qcr9), complex IV (Cyt1), and complex V (Atp18, Atp7). We also identified both subunits of the E1 component (Pda1, Pdb1) of the pyruvate dehydrogenase complex, a known source of H₂O₂ (16). Other proteins are associated with sites or processes that have been linked to H₂O₂ generation, for example, components of the peroxisome (Pex11, Pex25) and the heme biosynthetic pathway (Hem15, Mcx1) (17).

Redox Changes Caused by Carbon Source Switching.

Having identified H₂O₂-sensing HyPer7 fusion proteins in cells growing on glucose for 24 h, we then evaluated the influence of changing metabolic conditions in a metabolic switch experiment. Cells were first grown on SCD for 24 h, measured, and then transferred (i.e., repinned) to plates containing raffinose (Raf) (SCR). After 24 h, SCR plates were measured and cells again transferred to SCD plates, to be measured after another 24 h (Fig. 2A).

Using the procedure and criteria defined above, we identified oxidized fusion proteins for the three conditions (Fig. 2B and Dataset S2): Upon shifting from Glc to Raf, more fusion proteins were found to be oxidized (n = 43 vs. n = 117). Following the back-shift to Glc, again fewer fusion proteins were found to be oxidized (n = 48). The same trend was seen across different cellular compartments (Fig. 2C). The oxidation of 7, 77, and 6 fusion proteins was specific to primary Glc growth, Raf growth, and secondary Glc growth, respectively (Fig. 2B). Numerous fusion proteins (n = 24), mainly mitochondrial, were found to be oxidized under all conditions (Fig. 2D). Other fusion proteins (n = 27), including many cytosolic proteins, were only seen to be oxidized under Raf (Fig. 2E). However, these fusion proteins were not measurable under Glc. This is not surprising because the expression of many proteins depends on the carbon source. Therefore, not all fusion proteins can be measured across all conditions. Thus, for further analysis, we specifically focused on those proteins for which we could obtain HyPer7 and SypHer7 ratios across all three conditions (n = 846). We only considered HyPer7 ratio changes if there was no concurrent change in the corresponding SypHer7 ratio. The transition from Glc to Raf decreased the HyPer7 ratio of 104 fusion proteins while increasing the ratio of just two (Fig. 2 F, Left). The return from Raf to Glc increased the HyPer7 ratio of 85 fusion proteins while decreasing the ratio of just two (Fig. 2 F, Right). Comparing primary and secondary Glc growth, only 6 out of 838 fusion proteins differed in their ratio (SI Appendix, Fig. S2B). Ratio changes of example fusion proteins are shown in Fig. 2G. In conclusion, while growth on Raf led to the detection of additional pro-oxidative protein environments (Fig. 2E), many fusion proteins shifted towards a more reductive environment under Raf (Fig. 2F).

Redox Changes Caused by Nutrient Depletion.

We then asked how nutrient depletion influences the redox state of individual HyPer7 fusion proteins. We compared strains grown on SCD for 24 h and 48 h (SCD_24h vs. SCD_48h). Colony growth on minimal medium (SD) was slower and therefore only allowed measurements after 48 h. We therefore compared strains grown on SCD (24 h and 48 h) with those grown on SD at 48 h. When grown on SCD, fewer fusion proteins were oxidized after 48 h (n = 25) than after 24 h (n = 48) (Fig. 3A and Dataset S2). Eight proteins were oxidized at both time points, and 17 proteins were uniquely oxidized after 48 h. Comparing SCD to SD, many more fusion proteins were found to be oxidized on SD (Fig. 3A). At 48 h, there was no overlap between SCD and SD. SCD at 24 h and SD at 48 h shared only 7 oxidized fusion proteins. Interestingly, many more cytosolic fusion proteins are oxidized on SD relative to SCD (Fig. 3B). The comparison of HyPer7 ratios across the three conditions revealed dynamic behavior in both directions (Fig. 3C). Comparing ratios between SCD at 24 h and SCD at 48 h, 720 were unchanged, 15 increased, and 26 decreased. Comparing ratios between SCD at 24 h and SD at 48 h, 300 were unchanged, 31 increased, and 8 decreased. Comparing ratios between SCD and SD, both at 48 h, 139 were unchanged, 46 increased, and 17 decreased. Individual examples of fusion proteins exhibiting HyPer7 redox differences are shown in Fig. 3D. Additionally, we investigated the impact of low glucose (0.2%). Relative to growth on 2% glucose, unfused cytosolic HyPer7 was more oxidized (SI Appendix, Fig. S3A) and 417 out of 677 measured fusion proteins showed a shift toward increased oxidation (SI Appendix, Fig. S3B). This observation is in line with the notion that the regeneration of reducing equivalents is limited under low glucose, thus affecting probe redox states more globally.

Medium Composition Determines Protein-Specific H₂O₂ Availability.

The above data suggested that individual proteins are exposed to different local concentrations of H₂O₂ under different growth conditions. To address this notion in further detail, we selected 119 strains for which we had already observed elevated oxidation and/or redox changes in one of the previously described experiments. We grew these strains on 16 different media to cover a broad range of differing metabolic growth conditions, including differences in amino acid availability and alternative nitrogen sources. Relative to reference conditions (growth on SCD), we observed increases and decreases in oxidation, depending on the fusion protein and the specific medium (Fig. 3E). As expected, the majority of the observed differences were changes toward less oxidation because the strains were selected on the basis of having shown (partial) oxidation under reference conditions. Differences in amino acid availability affected individual fusion proteins in a relatively specific manner (Fig. 3E, conditions 1 to 8). In contrast, differences in carbon source had a much broader impact on fusion protein redox states. While only a few fusion proteins were more reduced under low glucose (Fig. 3E, condition 9), many were more reduced under pro-respiratory growth conditions (Fig. 3E, conditions 11 to 14), as previously seen in the Glc-Raf-Glc switch experiment (Fig. 2F) and in agreement with previous reports showing that pro-respiratory media lead to lowered H₂O₂ levels (9). The change of the nitrogen source from ammonium sulfate to glutamate broadly impacted the redox state of many fusion proteins (Fig. 3E, condition 15). While most fusion proteins were more reduced under alternative metabolic conditions, HyPer7-fused Pda1, part of the pyruvate dehydrogenase (PDH) complex, became more oxidized under increased Glc availability (Fig. 3E, condition 10). Increased glucose flux may increase activity of the PDH complex, potentially leading to increased production of H₂O₂, as reported previously (16, 18).

Genes Influencing Protein-Specific HyPer7 Oxidation.

The above results revealed fusion protein–specific differences in local H₂O₂ availability. In principle, there are several possibilities as to how these differences may arise. For example, the specific fusion protein may be localized in proximity to a H₂O₂ source whose activity depends on certain metabolic conditions. Alternatively, the fusion protein may be localized to a microenvironment that either recruits or excludes oxidizing or reducing enzymes. To identify proteins that determine or modulate the redox state of individual fusion proteins of interest, we selected a panel of 24 enzymes with known (or suspected) roles in either H₂O₂ generation or removal, including various oxidases and reductases. Furthermore, we selected a panel of 59 HyPer7 fusion proteins that we observed to be oxidized under at least one of the conditions described above (Figs. 1–3). We crossed the 59 fusion protein strains with the 24 deletion strains using an SGA-based approach and then measured ratios on SCD, SCR, and SD plates (Fig. 4A). First, we looked at enzymes known to generate H₂O₂. For example, Pox1, the peroxisomal fatty-acyl coenzyme A oxidase, is a source of H₂O₂ inside peroxisomes (19). Deletion of POX1 led to a significant decrease in the oxidation of peroxisomal fusion proteins Pex11-HyPer7 and Pex25-HyPer7 but not of other fusion proteins within the test sample (Fig. 4B). We then asked about the influence of superoxide dismutases. The lack of cytosolic Sod1 significantly increased HyPer7 oxidation in five (SCD) and six (SCR) fusion proteins, with an overall pro-oxidative trend (SI Appendix, Fig. S4 A and B). However, the Pre2-HyPer7 fusion protein was significantly more reduced in the absence of Sod1, under all three metabolic conditions. Among all tested deletions, the lack of Sod1 had the most significant effect on Pre2-HyPer7 (SI Appendix, Fig. S4C). Likewise, the lack of Sod2 decreased HyPer7 oxidation in several fusion proteins (9 and 10, in Glc and Raf, respectively) (SI Appendix, Fig. S4 D and E), suggesting that Sod2 may act as a localized H₂O₂ generator, as previously proposed (20). However, the observed effects could also be caused more indirectly, and compensation effects cannot be excluded.

We then looked at enzymes known to scavenge H₂O₂. Prx1 is the principal mitochondrial thiol peroxidase, and its deletion is expected to increase H₂O₂ availability within the mitochondrial matrix. Indeed, the lack of Prx1 led to a significant increase in oxidation of seven fusion proteins, all of them localized to the mitochondrial matrix (Fig. 4C). IMS-localized fusion proteins, e.g., Ccp1-HyPer7, did not respond to the absence of Prx1. The lack of the cytosolic peroxiredoxin Tsa1 led to increased oxidation of several fusion proteins, while some others were more reduced (SI Appendix, Fig. S5A). The pro-oxidative effect of Tsa1 deficiency was more pronounced on Raf, mostly affecting cytosolic fusion proteins (including Ola1, Arc1, Gis2, Rps7b, Mdh2, Tma17, Pho13, Cub1, and Get4) (SI Appendix, Fig. S5B). For Rps7b, the lack of Tsa1 led to oxidation across all three conditions, more than any other deletion strain that was tested (SI Appendix, Fig. S5C).

In sum, these results reconfirmed that HyPer7 fusion proteins reflect local H₂O₂ availability. We then asked about the role of less-characterized enzymes. The yeast NADPH oxidase homologue (Yno1/Aim14) has been proposed to contribute to H₂O₂ generation (21, 22). On SCD medium, four proteins were less oxidized in the absence of Yno1 (SI Appendix, Fig. S6A). However, the influence of Yno1 was more obvious on SD medium, where 10 fusion proteins were less oxidized (SI Appendix, Fig. S6B). The lack of D-arabinono-1,4-lactone oxidase (Alo1) led to increased oxidation of numerous fusion proteins (SI Appendix, Fig. S6C), suggesting that Alo1 may have a previously unrecognized role in limiting H₂O₂ availability. Alo1 generates both H₂O₂ and D-erythroascorbic acid. The latter is acting as an antioxidant that may counteract H₂O₂ generation by scavenging superoxide (23). A graphical overview of the tested combinations of fusion proteins and deletions, across three different growth conditions, is shown in SI Appendix, Fig. S7.

Growth Conditions.

Yeast strains to be transformed were grown on YEPD (yeast extract peptone dextrose) [1% w/v yeast extract (Gerbu), 2% w/v peptone (Formedium), and 2% w/v glucose (Sigma)]. Selection was performed on YEPD plates solidified with 2% agar (Millipore) containing either G418 (Sigma) or hygromycin (Invitrogen). Screening was performed using the following media: 1. SCD (synthetic complete dextrose) [0.17% w/v yeast nitrogen base without amino acids, with 0.5% w/v ammonium sulfate (Difco, BD), 2% w/v glucose (Sigma), Kaiser complete amino acid mix (Formedium)]. 2. SCR (synthetic complete Raf) (glucose replaced by 2% w/v Raf; Formedium). 3. SD (synthetic defined) [0.17% w/v yeast nitrogen base without amino acids, with 0.5% w/v ammonium sulfate (Difco, BD), 2% w/v glucose (Sigma), supplemented with essential amino acids (Leu, His, Met, Lys; Sigma) and Ura (Sigma)]. 4. SC with low glucose: 0.2% w/v glucose. Colonies were grown at 30 °C for either 24 or 48 h as indicated. Additional metabolic conditions tested: SC with high glucose: 4% w/v glucose; SCD with custom drop-out Kaiser amino acid mixtures (−QE, −FWY, −IV; Formedium); SD with amino acid drop-in media (+P, +Q, +E, +QE; Sigma). Alternative carbon sources: raffinose (Formedium), galactose (Sigma), sucrose (Sigma), and fructose (Sigma) were used at the indicated concentrations. Alternative nitrogen sources: YNB without amino acids and without ammonium sulfate (Formedium) was supplemented with either monosodium glutamate (MSG; Sigma) or proline (Sigma).

Library Generation.

The C-SWAT acceptor library and the donor strain (YMaM639), containing the SceI expression module and the haploid selection cassette, were used as described (13). The plasmid pMAM484 was used to generate the donor vectors: Plasmids pMAM_HyPer7 and pMAM_SypHer7 containing the HyPer7 and SypHer7 coding sequences, respectively, were generated using the BamHI and SpeI cloning sites. The plasmids were verified by sequencing (Euroscarf-GATC) and subsequently transformed into the donor strain using the lithium acetate protocol. The C-SWAT library was generated as described previously, using Singer Rotor technology (13, 25). Briefly, the C-SWAT acceptor library (arrayed in 384 colony format) was mated with the donor strain containing either the HyPer7 or SypHer7 donor plasmid. Following 2 d of mating, the library was plated on selection plates for diploid selection, and this step was repeated twice. Strains were left to sporulate for 7 d at 23 °C and subsequently plated on haploid selection media. Mat a strains were selected using the HIS cassette (26). The strains were then plated on galactose-containing medium, to induce SceI endonuclease expression and promote recombination. The strains were then selected for hygromycin resistance, indicative of correct recombination between the acceptor library and the donor plasmid. Selection was repeated twice. Finally, colonies were grown on SCD media plates. The strains were stored in 96-well plates at −80 °C (95 plates per library). Given the well-characterized efficiency of library generation and the antibiotic selection steps, we performed random PCR analysis on a subset of strains. For this, we used a reverse primer within the HyPer7/SypHer7 sequence and a forward primer within the ORF (27).

Control and Reference Strains.

Nontransformed BY4741 and BY4742 strains were used as autofluorescence background controls. To generate strains expressing genetically integrated unfused HyPer7 or SypHer7, the respective coding sequences were cloned from p415 (15) into pRS303H (28), using Gibson assembly. Resulting plasmids were validated by sequencing (Euroscarf-GATC). Following digestion, the linearized plasmid was transformed into BY4741 and plated on hygromycin-containing plates for selection. Strains were validated by PCR.

Screening Procedures and Conditions.

For the primary screen, strains were arrayed in 1536 colony format. Each plate contains 60 pairs of matching HyPer7/SypHer7 strains, representing 60 ORFs, each in triplicate colonies. In addition, each plate includes a fixed set of 59 pairs of HyPer7/SypHer7 reference reporter strains (expressing fusion proteins chosen to represent different subcellular locations, redox environments, and expression levels), and one pair expressing unfused cytosolic HyPer7/SypHer7. Furthermore, each plate contains a total of 240 nonfluorescent parental strain colonies, evenly distributed over the whole plate, to serve as autofluorescence background controls, and to allow for the correction of position-related variations (spatial normalization), as potentially caused by gradients in agar thickness, temperature, humidity, and oxygen availability (SI Appendix, Fig. S1D). Plate borders are filled with strains expressing unfused HyPer7 and SypHer7 but excluded from the measurements due to variation in nutrient availability and thus colony size. In total, the library consists of 95 plates with 1536 colonies each. The plates are freshly repinned to SCD plates, and after 24 h of growth in a static incubator (30 °C), they are repinned to the measurement plates and allowed to grow for 24 h. Colonies were then repinned to plates containing SCR medium, grown for 24 h and measured (“SCR_24h”). Colonies were then again repinned to SCD medium plates, grown for 24 h, and measured (“SCD_24h_R”). In addition, the library was measured after growth on SCD plates for 48 h (“SCD_48h”) or on SD medium for 48 h (“SD_48h”). Measurements were performed in an excitation ratiometric manner using a fluorescence plate reader (see below). All measurements were independently repeated, yielding two complete data sets for each metabolic condition. For the extended metabolic switch experiment, a similar approach was used. Of note, 119 fusion proteins were selected and arranged in a similar format as described above, with 3 colonies representing one HyPer7 fusion protein and 3 colonies representing its SypHer7 counterpart, arranged next to each other together with nonfluorescent strains. Strains expressing unfused cytosolic HyPer7/SypHer7 were also included. The arrayed plate was repinned several times on SCD for 24 h. Each plate was measured in triplicate for each condition.

Synthetic Genetic Array Analysis.

To combine the expression of HyPer7 fusion proteins with gene deletions, 59 fusion proteins were selected on hygromycin resistance plates and then expanded to ensure they lost KanMX resistance. Gene deletions of 24 proteins (ahp1, aim14, alo1, cta1, ctt1, dot5, erv2, fms1, gpx1, gpx2, grx1, hyr1, jhd2, lpd1, pox1, prx1, sod1, sod2, trx1, tsa1, tsa2, yhb1, ylr456w, ypr172w) marked with KanMX resistance in the BY4742 background were taken from the deletion collection (Euroscarf). The 59 HyPer7 fusion protein strains were arrayed in the middle of a 96-well plate. Each deletion strain was arrayed in a separate 96-well plate. The plates were then mated with the HyPer7 fusion protein plate. Diploids were selected using KanMX and hygromycin and then sporulated in a static incubator for 7 d at 23 °C. Haploid selection was performed as described above, followed by selection for the correct antibiotic resistance markers. The plates were then repinned and arrayed onto a 1536 plate. The 1536 plate was divided into 60 squares of 16 colonies each. Each square of 16 colonies contained four colonies of a wild-type fusion protein strain, four colonies of the fusion protein strain in the deletion background, four colonies of BY4742 autofluorescence background control, and four colonies of the respective deletion in BY4742. Each 1536 colony plate had a total of 59 such squares. As we assessed 24 different deletions, this led to 24 different 1536 plates. These plates were grown several times in SCD for 24 h at 30 °C, before they were pinned on plates containing SCD. SCD plates were measured after 24 h and then repinned to SCR plates or SD plates. As before, SCR plates were grown for 24 h at 30 °C, and SD plates for 48 h at 30 °C. Each was done in duplicates. The plates were measured as described. Like before, plate border areas were filled with HyPer7 expressing strains but excluded from measurement.

Plate Reader Measurements.

All plates were measured with a PHERAstar FSX plate reader (BMG Labtech) at 30 °C. Excitation/emission wavelengths were 405/520 nm and 488/520 nm, with a bandwidth of 10 nm.

Data Processing and Analysis.

Data processing, analysis, plotting, and reporting were done using R within RStudio. Data handling, processing, and cleaning were done using R packages of the tidyverse collection (29): janitor, here, writexl, and readxl. Base R and the rstatix package were used for statistical testing. Following import, data were checked for completeness. Invalid measurements such as overflow or extreme blank values, i.e., any values above or below 6 x IQR (interquartile range) of all nonfluorescent measurements, were removed. A spatial normalization procedure was performed for each plate, and the predicted autofluorescence background values were used for blank correction. The algorithm for spatial normalization was developed by Silvia Calderazzo and Nicholas Schreck (DKFZ Biostatistics Core Facility). Reference strains were used to exclude obvious differences between plates. To obtain a reliable signal to noise ratio, a threshold was applied in addition to the removal of values below blank. A per-plate threshold defined by the 99th percentile blank value of each plate was used for both channels. Only strains (i.e., fusion proteins) fulfilling this criterion for at least 2 out of 3 technical replicates in both HyPer7 and SypHer7 libraries, and in both independent screens, were considered for further analysis. Having excluded all strains not meeting these quality criteria, we calculated the conventional fluorescence emission ratio (485 nm/400 nm) for each colony. Statistical tests were performed as indicated. Values were log2 transformed for the t test and ANOVA. Packages used for reading, tidying, and analysis of data included ggplot2 from the tidyverse (29). Packages viridis, ggrepel, ggpubr, and cowplot were used for plotting. Preprocessed data are included with this manuscript (Dataset S3). Dataset S1 lists all fusion proteins that showed elevated ratios in both HyPer7 and SypHer7, in both repeats. Similarly, Dataset S2 lists fusion proteins matching the criteria defined for oxidized HyPer7 in both repeats.

Cell Lysis, Gel Electrophoresis, and Immunoblotting.

Yeast cells were lysed by bead beating in PBS containing complete Protease Inhibitor Cocktail (Roche) and N-ethylmaleimide. Lysates were run on 4 to 12% gradient NuPage gels (ThermoFischer). Probe expression was visualized by immunoblotting with a polyclonal anti-GFP antibody (PABG1, Chromotek). Detection of phosphoglycerate kinase with an anti-Pgk1 antibody (Life Technologies) served as a loading control.

Antimycin A Treatment.

Selected yeast strains from the HyPer7 library and a strain expressing unfused cytosolic HyPer7 were grown in SCD until mid exponential phase (OD₆₀₀ = 0.6 to 1). Cells were harvested, resuspended in MES-TRIS buffer (pH 6), and dispensed into a 96-well black plate with a clear flat bottom (Falcon). Antimycin A diluted in ethanol was added to a final concentration of 50 μM (0.5% final ethanol concentration). The HyPer7 response was measured at 30 °C using a PHERAstar FSX plate reader (BMG Labtech). Excitation/emission wavelengths were 405/520 nm and 488/520 nm, with a bandwidth of 10 nm.