Validation of Assays for Reactive Oxygen Species and Glutathione in Saccharomyces cerevisiae during Microgravity Simulation

We have previously reported on the use of molecularly-barcoded yeast deletion series to evaluate the effects of microgravity on living cells in a systematic, quantitative, and unbiased manner. We used the complete collection of ~4,800 homozygous and ~1,100 heterozygous yeast deletion strains and obtained genome-wide sensitivity profiles by comparing strain fitness for clones grown in spaceflight versus the ground (Nislow et al., 2015). The effects of spaceflight on yeast were compared to the effects of thousands of drugs previously analyzed systematically with the same deletion series (Lee et al., 2014). We found the effects of spaceflight have high concordances with the effects of DNA-damaging agents and changes in redox state (Lin et al., 2008; Nislow et al., 2015). We also identified unique spaceflight-specific genes and gene pathways affecting cell survival (Nislow et al., 2015).

Yeast grown in spaceflight showed a profile very similar to that of yeast treated with diallyl disulfide profile, an agent that increases the production of the enzyme glutathione S-transferase (GST) that binds electrophilic toxins in cells (Lin et al., 2008; Nislow et al., 2015). Overloading the cell with inhibitory doses of diallyl disulfide reveals genes required for survival in the presence of increased reactive oxygen species (ROS) (Lin et al., 2008).

Others have observed perturbation of glutathione during spaceflight. During the FOTON-M3 mission, spaceflight induced an enormous extracellular release of glutathione from S. cerevisiae cells, changed the distribution of bud scars, and activated the high osmolality glycerol and cell integrity/protein kinase C (PKC) pathways, as well as protein carbonylation (Bradamante et al., 2010).

These results show promise for the use of microgravity as a laboratory for the study of drug pathways since low redox potential state mirrors the electrophilic conditions of mitochondria, where many drugs are metabolized (Blackman et al., 2012; Cap et al., 2012a). To pursue this aspect of microgravity, we asked whether the reduced redox state developed during spaceflight could be reproduced and modulated in ground-based simulations. Unfortunately, assays for redox status and its major cellular determinant, glutathione, are diverse, often cell-type-specific, and an accepted probe set for yeast studies does not exist (Hedley and Chow, 1994; Kalyanaraman et al., 2012; Sebastià et al., 2003). To address this need, we have validated fluorescent probes for glutathione and reactive oxygen status in the yeast S. cerevisiae to open the door for mechanistic and pathway studies of microgravity and drug metabolism. Our approach focuses on intracellularly-trapped fluorescent probes to access the sensitivity of fluorometry and flow cytometry applications. The fluorescent probes were compared to biochemical and untrapped probe alternatives (Hedley and Chow, 1994; Kalyanaraman et al., 2012; Lewicki et al., 2006).

DC-FDA and mBCL are intracellularly trapped fluorescent probes that formed the starting reagents for our validation (Hedley and Chow, 1994; Jakubowski and Bartosz, 2000). DC-FDA is often reviewed as the best measure of cellular redox status, but a variety of redox molecules contribute to the assay findings (Jakubowski and Bartosz, 2000). Similarly there is no selective dye for glutathione, but mBCL is amongst the most specific of available dyes for thiols, and glutathione is by far the most abundant intracellular thiol (Hedley and Chow, 1994).

To measure dye uptake and conversion, yeast were mixed with 100 to 800 μM mBCL, or 2.5 to 20 μM DC-FDA, and fluorescence was measured every 10 min for 8-12 h in a plate reader. Kinetic uptake curves show that uptake and conversion of both dyes is time and dose dependent (Figure 1A and Figure 1D). Conversion of dye is also dependent on the quantity of yeast (Figure 1C and Figure 1F). Changes measured within these dye doses and times reflect changes without artifacts due to dye saturation. The fluorescence of DC-FDA alone increases very slightly with prolonged incubation, whereas mBCL is extremely stable (Figure 1B and Figure 1E). Because of this background, fluorescence data from the microplate reader is typically presented at net relative fluorescent units calculated by subtracting the fluorescence signal from dye alone, from the total fluorescence signal.

Figure 1.

To evaluate whether the dyes were retained intracellularly, we incubated a series of different yeast clones cultured under different conditions with 400 μM mBCL or 10 μM DC-FDA, and then measured the fluorescence in individual cells by flow cytometry at one and eight hours. During that time interval, the mean channel fluorescence (mcf) of DC-FDA decreased by 49 + 3% (SEM nine samples), suggesting efflux of some dye (Figure 2A). However, the mcf for mBCL increased by 89 + 10% (SEM nine samples), suggesting continued dye loading and or dye conversion by the intracellular GSH (Figure 2B). To evaluate toxicity from DC-FDA or mBCL, cell death was measured by fluorescence of PI using flow cytometry. Viability in the 10 samples at one hour ranged from 86% to 98%. After eight hours of exposure to mBCL, viability had increased by 16.3 + 3.8% over baseline (data not shown). After eight hours of incubation with DC-FDA, viability had increased by 12.7 + 3.0% (mean + SEM; data not shown). The increase in viability over eight hours suggested the yeast may have proliferated during the staining period. Based on these results, we selected 400 μM (final) as the optimum concentration for mBCL staining, and 10 μM (final) as the optimum concentration for DC-FDA; endpoint reads were performed at four hours after dye addition.

Figure 2.

To verify mBCL fluorescence reflected intracellular glutathione concentrations, we compared it to a biochemical assay that relies upon the reaction between cellular GSH extracted with perchloric acid and a highly selective fluorogenic probe, i.e., NDA (Lewicki et al., 2006). The standard curve for exogenous purified GSH assayed by NDA fluorescence is linear, embodied by the equation y=22.019x + 37.004 with R²=0.996 (Figure 3A). The fluorescence of GSH in cell extracts detected with NDA can be completely inhibited by addition of NEM in a dose-dependent manner (Figure 3B). NEM was able to inhibit 75% of the fluorescence generated by NDA mixed with purified GSH (Figure 3C).

Figure 3.

The biochemical NDA-based assay and the fluorometric mBCL assay showed excellent correlation when applied to the same samples. Figure 4 shows the fluorescence for GSH in intact yeast, detected with mBCL dye (horizontal axis), versus the fluorescence in the biochemical assay for GSH in perchloric acid extracts from the same yeast samples (vertical axis). Values plotted are the geomean + SEM of three replicates for the biochemical GSH assay and six replicates for the mBCL assay. GSH detected with mBCL dye versus the fluorescence in the biochemical assay for GSH is linear, embodied by the equation y=0.0448x + 132.91 with R²=1. Note the different scales and baselines used to clarify the trend by exaggeration.

Figure 4.

To determine the relationship between related dyes for fluorescent assay of glutathione and ROS, we assayed a series of 84 samples with a panel of reagents: mBCL for low molecular weight thiols, including glutathione; DC-FDA for hydroxyl, peroxyl, and other reactive oxygen species; Amplex Red for released hydrogen peroxide; DHE for reactive oxygen species and superoxide; and MitoSOX for mitochondrial superoxide production. Figure 5 illustrates the cellular location of the products detected by each reagent. Table 1 summarizes correlation coefficients between the different assays.

Figure 5.

Results from Table 1 suggest DC-FDA and mBCL are not only not concordant, but even trend towards an inverse relationship. The next experiments were designed to verify that DC-FDA (used to detect reactive oxygen species) and mBCL (used to detect glutathione) are measuring different substances in the cells. Figure 6 shows the intensity of the two dyes when used to stain WT and a battery of deletion strains implicated in microgravity and/or redox responses (Cap et al., 2012b; Nislow et al., 2015): Sok2 (defective ammonia production), Sfp1Δ (deleted shear stress promoter), and Msn4Δ (deleted shear stress promoter) after seven days of culture as colonies on YPD agar. These clones were selected for the present validation studies in anticipation of future experiments to define the signaling pathways for redox changes in microgravity. The non-linear scatter pattern verifies the two dyes are measuring different targets in the yeast.

Figure 6.

Because fluorescent probes are easy to measure, have a broad dynamic range, and are commonly used in diverse cell types, there is a tendency to believe they are reliable, sensitive, and specific. In reality, each fluorescent probe must be validated for the application and cell type to which it is applied—the cellular uptake mechanisms are often poorly defined or entirely unknown, and as fluorescent probes respond to physiological conditions, quench at different rates in response to different substrates, and suffer spectral overlap when used in combination with other probes, their utility for assays requires validation (Smith et al., 2012).

Assays of reactive oxygen species tend to measure the general status of the cell with contributions from a variety of molecules, including: superoxide, hydroxyl radicals, and various peroxide and hydro peroxides (Kalyanaraman et al., 2012). Due to their near instantaneous metabolism in the cell, H₂O₂ species are challenging to measure and we must instead focus our efforts on the physiological milieu of secondary products (Kalyanaraman et al., 2012). Oxidation of H₂DCF to its fluorescent form requires the presence of either cytochrome c, or of both redox-active transition metals and H₂O₂ (Karlsson et al., 2010). When staining yeast grown in liquid YPD with DC-FDA, the protocol should include washing, but not so much as to deplete transition metals (Karlsson et al., 2010).

Several lines of evidence suggest that DC-FDA is an excellent indicator of general redox status in yeast. It is taken up and cleaved to its fluorescent form with a linear time course, allowing for substantial dye loading and producing a broad dynamic range for measurements. The dye has low acute toxicity and is well retained during one to four hours, although some does leak out after prolonged incubations, e.g., eight hours. Others have noted this slow leakage, but also concluded the leak is slow enough to allow quantitative measurements (Jakubowski and Bartosz, 2000). By using DC-FDA and a microplate reader, one captures a snapshot of ROS in all of its compartments — the mitochondria, the cytoplasm, and ROS that was released into the extracellular medium. By collecting the microplate data at four hours, we are favoring the detection of intracellularly converted DC-FDA. This data can be complemented by the use of flow cytometry, which detects only cell-associated fluorescence while free dye is diluted away in the sheath fluid.

As much of the reactive oxygen in cells is thought to be generated in mitochondria, it was encouraging that DC-FDA measurements correlated with the MitoSOX assay for mitochondrial superoxide production. Unlike MitoSOX, the DC-FDA correlation with DHE was also predictable, as DHE reports the cytosolic contribution of superoxide to reactive oxygen status. Unlike DC-FDA, DHE undergoes minimal oxidation by cell-free peroxide (Kalyanaraman et al., 2012) and thus is more specific for intracellular ROS. In contrast, when 10-acetyl-3,7-dihydroxyphenoxazine (Amplex Red reagent) in combination with horseradish peroxidase was utilized to detect hydrogen peroxide (H₂O₂) and/or peroxidase activity; there was no detectable correlation with the DC-FDA assay. This is not surprising given the extremely short life of H₂O₂ and the prolonged loading time of DC-FDA (Kalyanaraman et al., 2012). It should be noted that in some systems the rate of DC-FDA oxidation depends on the concentration of glutathione, which is an alternative target for reactive oxygen species (Jakubowski and Bartosz, 2000). mBCL also measures a variety of molecules as it reacts with several low molecular weight thiols including: glutathione, N-acetylcysteine, mercaptopurine, peptides, and plasma thiols (Lewicki et al., 2006). Intracellular reduced GSH is by far the most abundant intracellular thiol and plays a key role in protecting cells from toxicity as it maintains intracellular redox status conjugating with electrophilic xenobiotics and free radicals and detoxifying reactive peroxides (Jakubowski and Bartosz, 2000). Several lines of evidence suggest mBCL is an excellent indicator of thiols — predominantly glutathione — in yeast. It is taken up and forms a fluorescent conjugate with a linear time course, allowing for substantial dye loading, which produces a broad dynamic range for measurements. The dye is well retained and is not acutely toxic to the cells, even during prolonged dye loading for hours. The mBCL signal was NEM sensitive and correlated well with the biochemical assay of Lewicki et al. (2006). As glutathione plays a major role in cytosolic redox buffering, it was predictable that it should negatively correlate with the cytosolic DHE superoxide measures and Amplex Red hydrogen peroxide measures, but not mitochondrial superoxide production. In comparative studies of glutathione probes for flow cytometry use in neurons, mBCL has the lowest background and correlates well with glutathione depletion (Hedley and Chow, 1994). The cell wall of yeast prevents known glutathione depletion reagents from entering the cells, meaning depletion studies are not feasible in yeast (Hedley and Chow, 1994).

Several non-fluorescent methods — such as High Performance Liquid Chromatography (HPLC) or Nuclear Magnetic Resonance (NMR) — are available for measuring GSH and oxidized glutathione (GSSH). They have some disadvantages, including the need to generate derivatives, lack of sufficient sensitivity to allow detection in small samples, and the inability to conveniently measure both GSH and GSSH (Lakritz et al., 1997). Alternatives — such as high-performance liquid chromatography-electrochemical methods — reproducibly measure both GSH and GSSH (Lakritz et al., 1997). The fluorescent methods, by allowing application of flow cytometry, allow quantitative detection of subpopulations of cells and co-localization of markers. Development of redox-sensing fluorescent proteins (redox probe proteins) has enabled live imaging of the physiological redox state within a cell (Oku and Sakai, 2012). These include single fluorophore redox probes — such as roGFP and rxYFP proteins — and double fluorescence resonance energy transfer (FRET)-based redox probes (Oku and Sakai, 2012). Redox probe proteins will likely be very useful for microgravity studies when appropriate analysis hardware is available during flight or on ground-based simulators.

The plethora of fluorescent reagents for reactive oxygen species and glutathione can be bewildering and makes head-to-head comparisons of all the alternatives impractical (Hedley and Chow, 1994; Jakubowski and Bartosz, 2000; Kalyanaraman et al., 2012). Understanding that these reagents measure the physiological milieu of ROS and diverse thiols — rather than specific individual molecules — we conclude that in yeast, mBCL and DC-FDA are suitable for fluorometric and flow cytometry studies. Both dyes have low background fluorescence, predictable loading, good retention, and have low acute toxicity towards S. cerevisiae. This report defines the way forward for future studies of redox potentials in yeast under real and simulated microgravity conditions.