Quantification of light-induced miniSOG superoxide production using the selective marker, 2-hydroxyethidium

Abstract

Genetically-encoded photosensitizers produce reactive oxygen species (ROS) in response to light. Transgenic expression of fusion proteins can target the photosensitizers to specific cell regions and permit the spatial and temporal control of ROS production. These ROS-generating proteins (RGPs) are widely used for cell ablation, mutagenesis and chromophore-assisted light inactivation of target proteins. However, the species produced by RGPs are unclear due to indirect measures with confounding interpretations. Recently, the RGP mini “Singlet Oxygen Generator” (miniSOG) was engineered from Arabidopsis thaliana phototropin 2. While miniSOG produces singlet oxygen (¹O₂), the contribution of superoxide (O₂^•−) to miniSOG-generated ROS remains unclear. We measured the light-dependent O₂^•− production of purified miniSOG using HPLC separation of dihydroethidium (DHE) oxidation products. We demonstrate that DHE is insensitive to ¹O₂ and establish that DHE is a suitable indicator to measure O₂^•− production in a system that produces both ¹O₂ and O₂^•−. We report that miniSOG produces both ¹O₂ and O₂^•−, as can its free chromophore, flavin mononucleotide. miniSOG produced O₂^•− at a rate of ~4.0 μmol O₂^•−/min/μmol photosensitizer for an excitation fluence rate of 5.9 mW/mm² at 470 ± 20 nm, and the rate remained consistent across fluences (light doses). Overall, the contribution of O₂^•− to miniSOG phenotypes should be considered.

Introduction

Photosensitizers produce reactive oxygen species (ROS) in response to light [1]. Reactive-oxygen-species-generating proteins, or RGPs, are a class of genetically-encoded photosensitizers [1]. These include SuperNova [2], KillerRed [3], KillerOrange [4], and miniSOG [5]. A RGP has the capability to generate different types of ROS including superoxide (O₂^•−) and singlet oxygen (¹O₂). mini “Singlet Oxygen Generator” (miniSOG) is unique in that it generates a relatively large quantum yield of ¹O₂ [5]. This ROS production is attributed to its chromophore flavin mononucleotide (FMN), a well-known ¹O₂-generating photosensitizer [6, 7]. miniSOG has been used for a variety of applications, such as electron microscopy [5], cell death [8, 9], mutagenesis [10] and target protein inactivation [11, 12]. While miniSOG was successful for these applications, the ROS responsible remained unknown as the ¹O₂ yield was debated [13, 14].

The disparity between ¹O₂ yields with different detection methods led to the hypothesis that O₂^•− may be a species produced by miniSOG [13]. Pimenta et al. measured O₂^•− production using the fluorescence of dihydroethidium (DHE) oxidation products, a nonspecific measure of O₂^•−. The fluorescence of DHE oxidation products can be the result of the O₂^•− specific product, 2-OHE⁺, and the nonspecific product, E⁺. These resulting DHE oxidation products are indistinguishable via fluorescence alone, and require HPLC separation to measure O₂^•− production specifically [15, 16]. Moreover, although based on fluorescence, there are conflicting results and no clear consensus on whether or not ¹O₂ can react with DHE to form E⁺ [17–19]. Overall, our goal was to clarify the impact of ¹O₂ on DHE-oxidation products and confirm if miniSOG generates O₂^•− by measuring the formation of the O₂^•− specific DHE oxidation product, 2-OHE⁺.

Thus, we measured O₂^•− generated by miniSOG using HPLC separation of the DHE oxidation products to specifically detect O₂^•−. We characterized the measurement system using Rose Bengal, a chemical photosensitizer that generates both O₂^•− and ¹O₂ [20–22]. Detailed DHE oxidation product analysis demonstrates that ¹O₂ does not react with DHE. Under conditions where miniSOG makes ¹O₂, it also produces O₂^•− at a flux that is consistent across fluences (light doses).

Materials and Methods

Singlet oxygen detection using singlet oxygen sensor green

¹O₂ was measured using singlet oxygen sensor green (SOSG; Molecular Probes). SOSG has a weak blue fluorescence but upon reaction with ¹O₂ exhibits a strong green fluorescence [23]. SOSG (1μM) baseline fluorescence was measured in a cuvette containing a photosensitizer, Rose Bengal (RB, 2.5 μM; Sigma) or Deuteroporphyrin (DP, 2.5 μM; Frontier Scientific), and SOSG buffer (SB; 120 mM KCl, 25 mM sucrose, 5 mM MgCl₂, 5 mM KH₂PO₄, 1 mM EGTA, 10 mM HEPES, 0.1 mM DPTA, pH 7.3). Where indicated, 20 mM azide or 800 units/mL superoxide dismutase (SOD, Sigma) was present (as illustrated in scheme 1). Temperature was held constant at 25°C. The fluorescence (Ex 488 nm; Em 525 nm; slit width 5 nm) was recorded for 1 minute with constant stirring before and after illumination (560 ± 20 nm, 10.6 mW/mm²) for 0–30 minutes. The change in fluorescence intensity (post minus pre-illumination) was calculated. Experiments using FMN (10 μM) or purified miniSOG (10 μM) were illuminated at 470 ± 20 nm (5.9 mW/mm²). To avoid spectral overlap with FMN, SOSG fluorescence was excited at 504 nm (Em 525 nm; slit width 5 nm). Structures of Rose Bengal and FMN are shown in Supplemental Fig. 1.

Scheme 1. Overview of ROS detection methods.

ROS generators were used to produce superoxide (O₂^•−) and/or singlet oxygen (¹O₂). 2-hydroxyethidum (2-OHE⁺) is a O₂^•−-selective marker, while ¹O₂ was detected using Singlet Oxygen Sensor Green (SOSG). Superoxide dismutase (SOD) converts O₂^•− to hydrogen peroxide (H₂O₂), which can be removed by catalase (CAT) to water. Both uric acid and azide quench ¹O₂, while azide additionally quenches the triplet state of FMN. Abbreviations: X (Xanthine), XO (Xanthine oxidase), RB (Rose Bengal), FMN (flavin mononucleotide).

Light intensity was measured using a calibrated thermopile detector (818P-010-12, Newport Corporation, Irvine, CA) connected to an optical power meter (1916-R, Newport Corporation). For all experiments in which the illumination was varied, the intensity of light or fluence rate (watts/area) was held constant. The total fluence (joules/area) was altered by varying the illumination time.

Quantification of xanthine/xanthine oxidase superoxide production using cytochrome c

Xanthine oxidase (XO) catalyzes the oxidation of xanthine (X) to uric acid. During this reaction XO generates O₂^•− or H₂O₂ via either a 1 or 2 electron reduction, respectively. The rate of X/XO O₂^•− formation was measured as the rate of SOD-sensitive cytochrome c reduction. Briefly, at ambient O₂, XO (0.1 units/mL) and X (1 mM) were added to a 0.1 cm cuvette containing cytochrome c (800 mM). The rate of cytochrome c reduction was monitored at 550 nm and calculated using an extinction coefficient of 18.7 M⁻¹ cm⁻¹ in the presence or absence of SOD (800 units/mL) [24]. The X/XO reaction predominately generates H₂O₂ in a pH and oxygen concentration-dependent manner [25]. The percentage of O₂^•− generated by X/XO was calculated by dividing the O₂^•− generation by the total electron flux of X to uric acid by XO, as previously described by Kelley et al. [25].

Superoxide detection using HPLC separation of DHE oxidation products

Xanthine/xanthine oxidase (X/XO) system

O₂^•− was measured using dihydroethidium (DHE; Thermo Fisher Scientific) followed by HPLC separation of the resulting oxidation products [16]. Upon oxidation, DHE forms red fluorescent products, a O₂^•− specific product, 2-OHE⁺, and a nonspecific product, E⁺, which must be separated by HPLC due to overlapping fluorescence spectra [16, 26]. Structures of DHE, 2-OHE⁺ and E⁺ are shown in Supplemental Fig. 1. In 1 mL of PBS containing DPTA (DPBS; 7.78 mM Na₂HPO₄, 2.2 mM KH₂PO₄, 0.1 mM DPTA, pH 7.4 at 37°C) and DHE (50 μM), 0.1 units/mL XO and 1 mM X were added to generate O₂^•−. Where indicated, 800 units/mL SOD was present (as illustrated in scheme 1). The solution was incubated at 37°C for 0–20 minutes (as indicated) to generate increasing amounts of O₂^•−, after which the reaction was stopped with an equal volume of acidified methanol (200 mM HClO₄ in methanol). The solution was incubated at −20˚C for 30 minutes and then centrifuged at 17,000 × g, 4°C for 20 minutes. Next, an equal volume of sample was combined with 1 M K⁺PO₄⁻ (pH 2.6). Again, the sample was incubated at −20˚C for 30 minutes and then centrifuged at 17,000 × g for 10 minutes. Samples were separated on a Polar-RP column (Phenomenex, 150 × 2 mm; 4μm) run on Shimadzu HPLC with fluorescence detection (RF-20A). From 0 to 15 minutes the detector was on low sensitivity (channel 1: Ex: 358 nm, Em: 440 nm; channel 2: Ex: 490 nm, Em: 596 nm). After 15 minutes, the sensitivity switched to high (channel 1: Ex: 490 nm, Em: 567 nm; channel 2 remained constant) with a constant flow rate of 0.1 mL/min. Two mobile phases were used (A, 10% ACN with 0.1 %TFA in water; B, 60% ACN with 0.1 %TFA in water) using the following gradient: 0 min, 40% B; 5 min, 40% B; 25 min, 100% B; 30 min, 100% B; 35 min, 40% B; 40min, 40% B. Concentrations of DHE, 2-OHE⁺ and E⁺ were measured using standard curves. DHE and E⁺ standards were commercially available. The 2-OHE⁺ was made using X/XO generated O₂^•−. The resulting mixture was separated using HPLC and the fraction containing 2-OHE⁺ was collected and lyophilized to a dry powder. 2-OHE⁺ was confirmed using mass spectrometry (URMC Mass Spectrometry Resource Laboratory; m/z = 330.16; the intervals between isotopic peaks = 1.0).

Photosensitizers

Superoxide production from photosensitizers was measured as described above with the following modifications. DHE (50 μM) and RB or DP (2.5 μM) were illuminated (560 ± 20 nm, 17 mW/mm²) in assay buffer (120 mM KCl, 25 mM sucrose, 5 mM MgCl₂, 5 mM KH₂PO₄, 1 mM EGTA, 10 mM HEPES, 0.1 mM DPTA, pH 7.3) for 0–5 minutes. As a control, one cuvette was also incubated in the dark. Where indicated, 20 mM azide or 800 units/mL SOD or 4,200 units/mL of catalase (CAT, 1 mg/mL; from bovine liver or from Corynebacterium glutamicum as indicated, Sigma) or hydrogen peroxide (H₂O₂, Sigma) was present. The procedure was repeated with 10 μM free FMN (Sigma) or purified miniSOG. Samples were illuminated at 470 ± 20 nm (5.9 mW/mm²).

miniSOG Purification

Recombinant miniSOG with an N-terminal histidine tag was expressed and purified from BL21DE3pLYS cells. Briefly, the coding region of miniSOG (courtesy of Drs. Roger Tsien and Yishi Jin, University of California, San Diego) was inserted in pRSET B using BamHI and EcoRI. The plasmid was electroporated into competent BL21pLyS cells and plated on ampicillin plates (100 μg/mL). A single colony was grown to OD₆₀₀ of 0.5 and miniSOG expression was induced using IPTG (100 μM). Cells were grown in the dark and lysed after 4 hours of expression with lysis buffer (3% SDS, 50 mM Tris, pH 8.0) and protease inhibitors (Pierce). His-tagged miniSOG was then allowed to bind to Ni-NTA agarose beads (Qiagen) overnight; fluorescence was monitored to ensure protein binding. miniSOG was eluted with 100 μM imidazole and desalted with a PD-10 column (GE Healthcare). The sample was concentrated using a centrifugal filter with a 10 KDa cutoff (Amicon) and stored in 10% glycerol PBS buffer, at 4°C. To ensure that miniSOG ROS generation was only due to miniSOG bound FMN, ROS measurements were taken immediately after concentrating, thus ensuring any free FMN would be removed. The entire purification process was performed in the dark and the protein was protected from light. The resulting protein was characterized by measuring the extinction coefficient, fluorescent quantum yield using fluorescein as standard (ΦF = 0.95) [27], and the absorption and emission spectra. Since the purified miniSOG contained both mature (FMN-containing) and immature (void of FMN) protein, the concentration of mature functional protein was determined using the extinction coefficient of flavin as previously described [13]. Briefly, miniSOG was denatured to release FMN and the concentration of free FMN (ε=12,500 ± 500 M⁻¹ cm⁻¹ at 450 nm) [28, 29] was determined using a standard curve. Since a mature photoactivatable miniSOG contains a single FMN as its chromophore, the concentration of FMN was assumed to equal the concentration of mature miniSOG [5, 13, 14].

Results

Detection of O₂^•− using HPLC separated DHE-oxidation products

First, we characterized the HPLC separation and fluorescent detection of DHE oxidation products using the X/XO generated O₂^•− (Fig. 1). Importantly, the X/XO reaction does not generate ¹O₂ and moreover, a reaction product, uric acid, is an efficient ¹O₂ scavenger [30]. Thus, we attribute changes in DHE oxidation products to O₂^•− and resulting dismutation products. We observed an incubation-time-dependent increase in 2-OHE⁺ (Fig 1A) with a minimal impact on the E⁺ component. SOD, which scavenges O₂^•− 1,000 times faster than it can react with the DHE [16], significantly decreased the amount of detectable 2-OHE⁺, while the E⁺ component was unaffected (Fig. 1B). We attempted to mimic the inhibitory effect of SOD on 2-OHE⁺ formation using the chemical O₂^•− scavenger, TEMPO, but found that TEMPO converted all of the DHE to E⁺, thereby preventing any possible 2-OHE⁺ formation (Supplemental Fig. 2). This is consistent with a previous result, which observed an increased fluorescence in DHE products in the presence of TEMPOL [31]. Our findings confirm the E⁺, not 2-OHE⁺, contribution to the observed fluorescence signal [19]. We conclude that the HPLC separation of DHE oxidation products can measure O₂^•− production via 2-OHE⁺ formation in an SOD-sensitive manner.

Fig. 1.

Characterization of DHE oxidation products derived from xanthine/xanthine oxidase generated ROS. (A) Representative HPLC traces of the fluorescence versus retention time separation of xanthine/xanthine oxidase generated 2-OHE⁺ and E⁺. Xanthine (X, 1mM) and xanthine oxidase (XO, 0.1 units/mL) were incubated for increasing amounts of time to generate O₂^•− in the presence of DHE (50 μM). The resulting DHE oxidation products were separated via HPLC. Using fluorescence detection, peaks at 22.0 and 24.5 minutes correspond to 2-OHE⁺ and E⁺, respectively. DHE was not limiting. (B) Peaks were integrated and quantified using a standard curve. The addition of SOD (800 U/mL) to X/XO reduced the 2-OHE⁺ component while not effecting E⁺. (C) 2-OHE⁺ formation is dependent on the concentration of O₂^•−. Using the cytochrome c reduction assay, X/XO O₂^•− production was determined to be 60.5 ± 8.7 μM/min. XO and X were incubated for increasing amounts of time in the presence of DHE as in A, B to generate 2-OHE⁺ as a function of time. The ratio of O₂^•− to 2-OHE⁺ was found to be logarithmic (y=3.3838ln(x)–5.4326; R² = 0.97). Inset is 2-OHE⁺ formation at the lower O₂^•− concentrations. All data are means ± SEM, N > 4. ** p < 0.01 versus 5-minute incubation (2-way ANOVA with Tukey correction).

We sought to quantify the superoxide production rate of miniSOG, however, the ratio of 2-OHE⁺ formation to O₂^•− is not linear. Given a fourfold difference in X/XO incubation time (5 min vs. 20 min), the 2-OHE⁺ concentration only doubled (Fig. 1B). A recent report demonstrated that the stoichiometry of O₂^•− molecules to 2-OHE⁺ formation varies with the concentration of O₂^•− [15]. For low concentrations of O₂^•−, the O₂^•− to 2-OHE⁺ ratio is 2:1 and increases nonlinearly with the concentration of O₂^•− [15]. The increase in the O₂^•−:2-OHE⁺ ratio is hypothesized to be the result of increased O₂^•− dismutation [15]. Therefore, in order to equate 2-OHE⁺ formation with O₂^•−, we first determined the O₂ ^•− production capabilities of X/XO using cytochrome c reduction. Secondly, we incubated X/XO with DHE for various incubation times and measured 2-OHE⁺.

XO produces both H₂O₂ and O₂^•− in a pH and oxygen dependent manner [25]. The rate of O₂^•− production was measured via cytochrome c reduction. Under conditions that mimic DHE experiments, the O₂^•− flux of X/XO was 60.5 ± 8.7 μM/min. This rate demonstrates that 30.3% of total XO electron flux results in O₂^•− and is consistent with a previous report [25]. Next, we measured 2-OHE⁺ formation as a function of time to mimic photosensitizer O₂^•− formation (Fig. 1C). The cumulative amount of 2-OHE⁺ generated during a period was plotted as a function of cumulative O₂^•− generated. The stoichiometry of O₂^•− to 2-OHE⁺ is in agreement with previous reports [15] and follows a logarithmic relationship. Using X/XO O₂^•− generation as a standard, we can quantify O₂^•− production using 2-OHE⁺ formation.

Next, to determine whether DHE is suitable for measuring photosensitizer-generated ROS, we measured DHE oxidation products resulting from the illumination of Rose Bengal (RB). Since RB is known to generate large quantities of ¹O₂ and to a lesser amount O₂^•− [6, 7, 22], it served as a control to determine if ¹O₂ has confounding effects on the DHE oxidation products. Additionally, DHE is light-sensitive and can undergo photo-oxidation [32]. The illumination of Singlet Oxygen Sensor Green (SOSG) or DHE alone did not prevent the detection of photosensitizer-generated ¹O₂ or O₂^•− (see figure legends for each wavelength). When RB was illuminated, ¹O₂ was detected (Fig. 2 A) and we found the rapid formation of DHE-oxidation products (Fig. 2B). Notably, both the 2-OHE⁺ and E⁺ components were insensitive to the ¹O₂ scavenger azide (Fig. 2C, p = 0.16; p = 0.13; 5 min light vs 5 min light + azide for 2-OHE⁺ and E⁺, respectively), which suggests that DHE is insensitive to ¹O₂. Consistent with the X/XO data, SOD specifically reduced RB generated 2-OHE⁺, suggesting we can detect O₂^•− despite the presence of ¹O₂. Similar results were found using another chemical photosensitizer, deuteroporphyrin (DP, Supplemental Fig. 3). Collectively, these results demonstrate that the formation of the O₂^•−-selective DHE oxidation product 2-OHE⁺ is SOD-sensitive and that azidesensitive ¹O₂ does not impact its formation, as seen in scheme 1.

Fig. 2.

Detection of Rose Bengal (RB) O₂^•− production by DHE. (A) RB was illuminated (560 ± 20 nm, 10.6 mW/mm²) for 5 minutes in the presence of Singlet Oxygen Sensor Green (SOSG). ¹O₂ production was determined by the change in SOSG fluorescence intensity. The ¹O₂ scavenger, azide (20 mM) or O₂^•− scavenger superoxide dismutase (SOD, 800 U/mL) was present where indicated. Azide or SOD in the presence of SOSG in the dark alone had no effect. SOSG illuminated for 5 minutes in the absence of photosensitizer yielded a change in fluorescence of −6.28 ± 1.4 a.u. Since RB has a large ¹O₂ quantum yield, we wanted to determine if we were still able to detect O₂^•− with dihydroethidium (DHE). (B, C) RB (2.5 μM) was illuminated under the same conditions for increasing amounts of time in the presence of DHE (50 μM). Peaks, 2-OHE⁺ and E⁺, were integrated and quantified using a standard curve. The formation of the O₂^•− specific product, 2-OHE⁺, is sensitive to SOD, as the addition of SOD (800 U/mL) reduced the 2-OHE⁺ component while not effecting the E⁺ component. Azide (20 mM) had no effect on either component, 2-OHE⁺ or E⁺. DHE illuminated for 5 minutes resulted in 258 ± 186 pmol/mL 2-OHE⁺ and 607 ± 423 pmol/mL E⁺. All data are means ± SEM, N > 4. ** p < 0.01 versus 5-minute light (2-way ANOVA with Tukey correction).

miniSOG ¹O₂ generation

Given that the HPLC separation of DHE oxidation products can measure O₂^•− in the presence of ¹O₂, we next sought to determine if miniSOG produces O₂^•−. We purified miniSOG and the resulting protein had an extinction coefficient of 15.15 ± 0.9 × 10³ M⁻¹ cm⁻¹, fluorescence quantum yield of 0.35 ± 0.017, and an expected molecular weight (Supplemental Fig. 4) as previously reported [5, 13]. miniSOG is a small protein that encapsulates a flavin mononucleotide (FMN), which acts as the photosensitizer. In order to understand its ROS production, we therefore compared miniSOG to its free chromophore, FMN. Both FMN and miniSOG generate ¹O₂ in response to light [5]. We first determined if miniSOG was functional by detecting ¹O₂ via Singlet Oxygen Sensor Green (SOSG).

We found that both FMN and miniSOG behaved similarly, with SOSG fluorescence steadily increasing with illumination time until a peak value was reached (Fig. 3A, C). Notably, FMN produced a larger amount of detectable ¹O₂ at a faster rate than miniSOG (99.5 a.u./min versus 24.6 a.u./min, FMN vs. miniSOG, respectively). This supports the lower published ¹O₂ quantum yield of miniSOG [13, 14]. However, both reached asymptotic levels following prolonged irradiation (Fig. 3A, C). Azide prevented the FMN- and miniSOG-mediated SOSG fluorescence increase, suggesting a ¹O₂-dependent mechanism (Fig. 3B, D). As expected, fluorescence was not lost with SOD (Fig. 3B, D). These results demonstrate that the purified miniSOG is functional and, like FMN, can generate ¹O₂.

Fig. 3.

Singlet oxygen sensor green (SOSG) detection of flavin mononucleotide (FMN) and miniSOG generated ¹O₂. (A) Free flavin mononucleotide (FMN) and (C) purified miniSOG were illuminated (470 ± 20 nm; 5.9 mW/mm²) for the indicated time in the presence of SOSG. ¹O₂ production was determined by the change in SOSG fluorescence intensity. Note: SOSG was excited at 504 nm. Where indicated, the ¹O₂ scavenger, azide (20 mM) or O₂^•− scavenger, SOD (800 U/mL) were present when (B) free FMN or (D) miniSOG were illuminated for 5 minutes. SOSG illuminated for 5 minutes yielded a change in fluorescence of −1.53 ± 0.3 a.u. All data are means ± SEM, N > 4. *p<0.05, **p<0.01 versus 5-minute light (2-way ANOVA with Tukey correction).

miniSOG O₂^•− generation

Similar to RB, FMN generates both ¹O₂ (Fig. 3) and O₂^•− in response to light [6, 7]. However, the precise mechanism of light-induced miniSOG ROS production was unclear. We hypothesized that miniSOG could produce both ¹O₂ and O₂^•− in a manner similar to its free photosensitizer, FMN. Alternatively, miniSOG’s protein encapsulation of the FMN could alter its ROS production properties to favor a particular photosensitization mechanism. Thus, we next measured the O₂^•− production of FMN and miniSOG using HPLC separation of DHE oxidation products, as seen in scheme 1. FMN and miniSOG were incubated with DHE and illuminated for various amounts of time (Fig. 4A, C). Both FMN and miniSOG produced 2-OHE⁺, but miniSOG did so to a lesser extent. SOD significantly reduced the 2-OHE⁺ components in both systems (p < 0.001, Fig. 4B, D). While azide had no effect on miniSOG-mediated 2-OHE⁺ formation (p=0.22, 5 min light vs 5 min light + azide, Fig. 4D), azide significantly reduced the 2-OHE⁺ component in the FMN system (Fig. 4B). We hypothesize that the reduction is due to azide directly interacting with the excited state of FMN [33] rather than scavenging ROS. Light excites FMN to its triplet state, which can then interact with O₂ to form either O₂^•− or ¹O₂. Azide can prevent the generation of ROS by interacting with FMN’s excited state, thus preventing FMN’s ability to react with O₂ [33]. Azide quenches ¹O₂ with K_Q ~ 2 to 4 × 10⁸ M⁻¹s⁻¹; however, azide additionally quenches the triplet state of FMN at a rate of ~ 2 to 5 ×10⁹ M⁻¹s⁻¹ [7]. Thus, the decrease in FMN light-dependent 2-OHE⁺ formation in the presence of azide may be the result of azide preventing ROS formation rather than scavenging it (Fig. 4B). In support of this hypothesis, uric acid, a different ¹O₂ scavenger significantly decreased FMN ¹O₂ production with no effect on O₂^•− generation (Supplemental Fig. 5). Azide did not have the same impact on 2-OHE⁺ generation for miniSOG as it did with free FMN (Fig. 4D).

Fig. 4.

Detection of FMN and miniSOG light-dependent O₂^•− generation using HPLC separation of DHE oxidation products. (A) free FMN (10 μM) or (C) purified miniSOG (10 μM) were illuminated (470 ± 20 nm; 5.9 mW/mm²) for increasing amounts of time in the presence of DHE (50 μM). The resulting 2-OHE⁺ or E⁺ peaks were integrated and quantified using a standard curve. Where indicated, the ¹O₂ scavenger, azide (20 mM), O₂^•− scavenger, SOD (800 U/mL) were present when (B) free FMN or (D) miniSOG were illuminated for 5 minutes. DHE illuminated for 5 minutes resulted in 50 ± 17 pmol/mL 2-OHE⁺ and 673 ± 142 pmol/mL E⁺. All data are means ± SEM, N > 4. *p<0.05, ** p <0.01 versus 5-minute light (2-way ANOVA with Tukey correction.

In order to quantify miniSOG O₂^•− rates, we used the ratio of O₂^•−:2-OHE⁺ calculated from the X/XO system (Fig. 1C). A previous report had demonstrated that miniSOG’s ¹O₂ quantum yield is light dose-dependent [14]. Thus, we determined the O₂^•− flux-dependence on light dose (Supplemental Fig. 6). miniSOG’s O₂^•− remained constant at ~4.0 μmol O₂^•−/min/μmol photosensitizer for an excitation fluence rate of 5.9 mW/mm² at 470 ± 20 nm and was independent of light dose (Supplemental Fig. 6).

A previous study utilized catalase-sensitive DHE fluorescence to elude to miniSOG-mediated O₂^•− formation [13]. Catalase decomposes hydrogen peroxide (H₂O₂) to water and oxygen and has no direct effect on O₂^•−. Moreover, H₂O₂ does not directly react with DHE [19]. Thus, we examined the impact of H₂O₂ and catalase on DHE oxidation products. We found that H₂O₂ increased E⁺ formation in a catalase-dependent manner and the addition of catalase alone increased E⁺ formation through an unknown mechanism (Supplemental Fig. 7). Impurities in the H₂O₂ solution or the catalase preparation may contribute the increased E⁺ formation. However, catalase from Corynebacterium glutamicum yielded a finding similar to catalase from bovine liver (DHE alone, 112 ± 30 pmol/mL 2-OHE⁺ and 1583 ± 423 pmol/mL E⁺; DHE plus C. glutamicum catalase, 317 ± 31 pmol/mL 2-OHE⁺ and 7405 ± 1397 pmol/mL E⁺). Despite the impact of catalase on E⁺ generation (Supplemental Fig. 7), we sought to confirm if catalase could decrease miniSOG generated 2-OHE⁺ formation [13]. The addition of catalase to the FMN or miniSOG photosensitization reaction had no effect on the 2-OHE⁺ component (Fig. 5B, C) and is contrary to a previous report [13]. We suggest that the difference can be attributed to analyzing fluorescence alone as opposed to in combination with HPLC separation.

Fig. 5.

H₂O₂ scavenger catalase (CAT) does not impact FMN or miniSOG generated O₂^·−. (A) Free FMN (10 μM) or (B) purified miniSOG (10 μM) were illuminated (470 ± 20 nm; 5.9 mW/mm²) for 5 minutes in the presence of DHE (50 μM) and CAT (1 mg/mL). DHE illuminated for 5 minutes (no photosensitizer) resulted in 73 ± 37 pmol/mL 2-OHE⁺ and 642 ± 37 pmol/mL E⁺. CAT had no significant effect on either the 2-OHE⁺ or E⁺ component (2-way ANOVA with Tukey correction). All data are means ± SEM, N > 4.

Discussion

Genetically-encoded photosensitizers contain a chromophore, which when illuminated can generate ROS. The miniSOG chromophore, FMN, is relatively well-studied, has a ¹O₂ quantum yield of 0.51 and generates O₂^•− [5, 34]. However, by encapsulating this chromophore, miniSOG’s ROS-producing capabilities may be altered. miniSOG’s ¹O₂-generating capabilities are reported elsewhere [5, 13, 14]; however, it was unclear if it could produce O₂^•−, since previous studies relied on indirect measures of O₂ ^•−and used catalase as a scavenger [13].

While DHE is a commonly used O₂^•− detector, fluorescence alone does not indicate the presence of O₂^•− due to the formation of a nonspecific oxidation product [15, 16]. Additionally, catalase, an enzyme that removes H₂O₂ from the system, should not impact O₂^•− (scheme 1). Lastly, it was unknown whether or not ¹O₂ could impact the oxidation of DHE [17–19]. Consequently, a report that miniSOG generates O₂^•− based on catalase-sensitive fluorescence spectra was inconclusive [13]. We improve upon these measurements by isolating the O₂^•−-specific product by HPLC separation. The data presented herein demonstrate that despite the presence of both ¹O₂ and O₂^•−, we found no evidence that ¹O₂ impacted the O₂^•−-selective DHE oxidation product (Fig. 2C). We demonstrate that miniSOG generates O₂ ^•− in an SOD-dependent and azide/catalase-independent manner. miniSOG’s O₂^•− production rate was lower than its free chromophore and insensitive to light dose. Interestingly, while azide directly interacted with the excited state of FMN to prevent ROS formation, azide did not have the same effect on miniSOG. This is consistent with a lack of access to FMN in the protein or a shift in the triplet state energy of bound FMN, rendering the quenching by azide less efficient. Together, these data demonstrate that the surrounding protein can influence ROS flux from miniSOG.

The ¹O₂ quantum yield of miniSOG is lower than that of FMN [13, 14]. One of the contributing factors to the difference in yield is miniSOG’s protein encapsulation of FMN. Ruiz-Gonzalez et al. monitored the time-resolved phosphorescence of miniSOG-generated ¹O₂ with increasing fluence (0–400 J/cm²) and found a ~10-fold increase in ¹O₂ quantum yield [14]. The authors hypothesize that this increase in yield is due to inactivation of ¹O₂ quenching amino acid residues in miniSOG [14]. Thus, the ¹O₂ quantum yield of miniSOG ranges from 0.03 to ~0.30 and is directly related to fluence [13, 14]. In contrast to ¹O₂, we found no change in the O₂^•− production rate with increasing light dose (0–177 J/cm², Supplemental Fig. 6), suggesting that O₂^•− is able to freely diffuse out of the protein.

Each ROS has unique chemical properties, which may permit preference towards a biological target [35, 36]. For example, O₂^•− has a high reaction rate with iron-sulfur clusters, while ¹O₂ by comparison has a more indiscriminate reactivity [35, 36]. miniSOG is used in various applications ranging from electron microscopy to photoablation [5, 8–12]. Understanding the type and amount of biologically available ROS produced by miniSOG can improve the experimental design and interpretation of the results.

Conclusion

Genetically-encoded photosensitizers that produce ROS are important tools for redox biology. Proteins such as miniSOG, offer spatial and temporal control over ROS production [1]. However, a complete study of their ROS-generating capability is limited and hampers further development of the optogenetic toolbox. Our results demonstrate that, in addition to ¹O₂, miniSOG generates O₂^•− in response to light, much like its FMN chromophore. The O₂^•− production rate was determined using HPLC separation and fluorescence detection of the O₂^•−-selective product, 2-OHE⁺. These results addressed the confounding interpretations using nonspecific readouts of fluorescence alone. The contribution of each ROS to the overall output of miniSOG is light dose-dependent and should be considered in future studies.

Highlights.

• Oxidation of dihydroethidium establishes miniSOG-mediated O₂^•− generation

• HPLC separation of O₂^•−-selective marker, 2-hydroethidium quantified miniSOG O₂^•−

• miniSOG O₂^•− production rate was consistent over a range of fluences