Improved red fluorescent redox indicators for monitoring cytosolic and mitochondrial thioredoxin redox dynamics

Abstract

Thioredoxin (Trx) is one of the major thiol-dependent antioxidants in living systems. The study of Trx functions in redox biology was impeded by the lack of practical tools to track Trx redox dynamics in live cells. Our previous work developed TrxRFP1, the first genetically encoded fluorescent indicator for Trx redox. In this work, we report an improved fluorescent indicator, TrxRFP2, for tracking the redox of Trx1, which is primarily cytosolic and nuclear. Furthermore, because mitochondria specifically express Trx2, we have created a new genetically encoded fluorescent indicator, MtrxRFP2, for the redox of mitochondrial Trx. We characterized MtrxRFP2 as a purified protein and used subcellularly localized MtrxRFP2 to image mitochondrial redox changes in live cells.

Graphical Abstract

INTRODUCTION

Thioredoxin (Trx) is a family of redox-active proteins that play vital roles in cellular redox homeostasis.^1,2 While imbalanced cellular homeostasis can cause oxidative stress, Trx can defend against the stress by actively regulating many cellular functional activities, such as transcription, deoxyribonucleotides synthesis, cell proliferation, and apoptosis.^2-4 Trx serves as a key regulator in cellular function, and dysfunction of Trx redox activities can lead to diseases. For example, alterations in Trx activities are found in various neurological disorders,⁵ while overactivation of Trx reductase (TrxR), a Trx regulator, is often observed in cancer proliferation and metastasis.^4,6 Moreover, Trx may impact viral infection. For example, the reaction between Trx secreted from CD4⁺ T cells and the redox-active D2 disulfide of CD4 is essential for the entry of HIV-1 into susceptible cells.⁷ In this context, Trx and TrxR have emerged as promising therapeutic targets for cancer, viral infection, and neurodegenerative diseases.^4-8

The typical Trx system contains Trx, TrxR, and nicotinamide adenine dinucleotide phosphate (NADPH). The electrons are first transferred from NADPH to Trx through TrxR.⁹ Reduced Trx subsequently neutralizes reactive oxygen species (ROS) via peroxiredoxins or induces the reduction of many other oxidized proteins while getting itself oxidized in the process.^9,10 Mammalian Trx antioxidant system includes two major members, the cytosolic Trx1¹¹ and the mitochondrial Trx2,¹² each with a corresponding reductase, TrxR1 and TrxR2, respectively. Both Trx1 and Trx2 are 12-kDa redox enzymes with a conserved redox-active Cys-Gly-Pro-Cys motif.¹³ Trx1 and Trx2 are primarily localized to different subcellular organelles for spatial control of redox signaling. Trx1 is mainly responsible for regulating the redox signaling in the cytosol and nucleus, while Trx2 regulates the mitochondrial redox.¹ Because mitochondria are the primary cellular sources of ROS, inhibition of mitochondrial Trx2 and TrxR2 can remarkably decrease the capacity of cells to scavenge ROS.¹⁴

Despite their essential roles in redox biology, the research tools for tracing Trx redox dynamics remain inadequate. Prevalent approaches, including Western blotting,¹⁵ immunohistochemistry,¹⁶ and mass spectrometry¹⁷ have been widely utilized to study the redox status of Trx1. These methods, which only provide endpoint measurements, are incapable of providing spatiotemporal resolution of redox signals in live cells. Moreover, the time-consuming procedures, including cell lysis and sample preparation, might generate ROS, resulting in artifacts. In the past two decades or so, fluorescent protein (FP)-based indicators have appeared as promising tools for real-time tracking of redox molecules due to their genetic encodability, excellent specificity, and superior capability to provide spatiotemporal information with outstanding signal-to-noise ratios.^18-22 For example, Gutscher et al. genetically fused glutaredoxin-1 (Grx1) to the N-terminus of redox-active green fluorescent protein (roGFP2) to generate Grx1-roGFP2, a fluorescent protein-based biosensor for detecting endogenous glutathione dynamics in live cells.²³ Similar strategy to fuse human Trx1 (hTrx1) with roGFP2 did not yield any effective redox-sensing biosensor.²³ Previously, our group developed the first genetically encoded fluorescent biosensor, TrxRFP1, that can detect the dynamic redox status of Trx in live mammalian cells by fusing hTrx1 with redox-active RFP (rxRFP1). In contrast with roGFP2, which bears the active cysteine residues at the surface of the β-barrel of a green FP (GFP), rxRFP1 has two active cysteines at the N- and C- termini of a circularly permutated red FP (cpRFP).^24-26 In TrxRFP1, Trx1 was placed in proximity to rxRFP1 through a 30-amino-acid Gly-Ser-rich linker. Because of the spatial proximity, the redox status of the active cysteines in Trx1 is kinetically coupled with the redox state of rxRFP1. Such coupling allows TrxRFP1 to monitor the redox status of Trx1.²⁴

In this work, we performed directed evolution with TrxRFP1 using an enzymatic reaction-based screening strategy. We developed an enhanced Trx fluorescent biosensor, TrxRFP2, which maintains the specificity but displays higher responsiveness than TrxRFP1 in live mammalian cells. In parallel, we developed a Trx2-specific red fluorescent redox indicator, MtrxRFP2, by generating and optimizing the redox relay between human Trx2 and rxRFP1.1.²⁷ We localized MtrxRFP2 to the mitochondria of living cells for detecting Trx2 redox.

EXPERIMENTAL SECTION

Engineering of TrxRFP2.

Directed evolution was used to improve the responsiveness of TrxRFP1.²⁴ Briefly, oligos pBAD-F and pBAD-R (Supplementary Information, Table S1) were used to amplify the TrxRFP1 gene in a pBAD/His B vector under an error-prone PCR (EP-PCR) condition. The resultant product was cloned into pBAD/His B predigested with Xho I and Hind III using Gibson Assembly.²⁸ The resulting library was used to transform E. coli DH10B cells via electroporation. Cells were plated on LB agar plates supplemented with 100 μg/mL ampicillin and 0.02% (w/v) L-arabinose. After incubation at 37 °C overnight, the plates were imaged using a customized imaging system equipped with a Dolan-Jenner Mi-LED Fiber Optic light source, appropriate excitation and emission filters in Thorlabs motorized filter wheels, and a QSI 628 CCD camera. Bacterial colonies with medium to high fluorescence were selected and cultured in 96-well deep-well plates containing 2×YT medium supplemented with 100 μg/ml ampicillin and 0.2% (w/v) L-arabinose. Cells were cultured at 30 °C, 250 r.p.m for 24 h and then at 16 °C for additional 48 h. Bacteria cells were then pelleted at 3000×g for 15 minutes and lysed using Bacterial Protein Extraction Reagents (B-PER). Fluorescence responses of cell lysates were measured on a BioTek Synergy Mx Microplate Reader. Fluorescence intensities were quantified and compared between an oxidized state (lysates treated with 10 μM peroxiredoxin TPx1 and 100 μM H₂O₂) and a reduced state (lysates treated with 10 μM TrxR1 and 200 μM NADPH). Two clones with high brightness and the greatest responsiveness were selected for DNA extraction. Their mixture was used as the template for the next round of EP-PCR. A total of seven rounds of directed evolution were performed to derive TrxRFP2.

Protein purification and comparison of TrxRFP1 and TrxRFP2.

Oligos pET-TrxRFP-F and pET-TrxRFP-R (Table S1) were used to amplify TrxRFP1 or TrxRFP2 gene fragment from plasmid pBAD-TrxRFP1 or pBAD-TrxRFP2, respectively. Each amplified fragment was inserted into a pET28a vector predigested with BamH I and Xho I via Gibson Assembly.²⁸ The products were used to transform E. coli DH10B cells. Plasmids were prepared, confirmed by Sanger sequencing, and used to transform BL21(DE3) competent cells. A single E. coli colony was selected for inoculating 5 mL 2×YT medium supplemented with 50 μg/mL kanamycin. After overnight cultured at 250 r.p.m and 37 °C, the culture was used to inoculate 500 mL of 2×YT medium with 50 μg/mL kanamycin. Cells were cultured at 37 °C and 250 r.p.m until the optical density at 600 nm (OD₆₀₀) reached 0.8. Next, 1 mM IPTG was added to induce protein expression at 16 °C for 72 h. Proteins were purified using Ni-NTA agarose beads, and next, subjected to size-exclusion chromatography with 1× phosphate-buffered saline (PBS, pH 7.4) as the eluent. Protein concentrations were determined using an alkali denaturation method.²⁹ The prepared proteins were initially in the fully oxidized state due to the existence of oxygen in the air. To monitor reduction kinetics, oxidized proteins (1 μM) were reduced with 10 μM TrxR1 and 200 μM NADPH in 1× PBS (pH 7.4) at room temperature. To monitor oxidation kinetics, the freshly prepared proteins (20 μM) were first reduced with 100 molar equivalents of dithiothreitol (DTT) in an argon-filled anaerobic chamber at room temperature overnight; next, the reduced proteins were immediately diluted 100 times with 1× PBS (pH 7.4) and oxidized by the addition of 20 μM TPx1 and 200 μM H₂O₂ at room temperature. Both human TPx1 and TrxR1 were recombinantly expressed and purified from E. coli. For technical convenience, the selenocysteine residue in TrxR1 was mutated to cysteine. Kinetic traces were monitored at room temperature using the BioTek Synergy Mx Microplate Reader with excitation and emission wavelengths set at 560 and 610 nm, respectively. The intensity values were subtracted for the background intensity of 1× PBS and then fitted for monoexponential decay (one phase decay) or pseudo-first order association (one phase association) using the equations preset in Graphpad Prism 9.

Modeling of TrxRFP2 with ColabFold.

ColabFold,³⁰ which combines AlphaFold2 with MMseqs,^31,32 was used to predict the 3-dimensional structure of TrxRFP2. Default settings were used. The program generated five structures ranked by pTMscores for each sequence input. The one with the best pTMscore was used to create illustrations and compared with the crystal structure of Trx (PDB:3TRX). Structural presentations were prepared in PyMOL.

Engineering of MtrxRFP2.

Oligos pBAD-Mtrx-F1 and pBAD-Mtrx-R1 (Table S1) were used to amplify the human Trx2 gene using a gBlocks Gene Fragment ordered from Integrated DNA Technologies (IDT). Next, oligos pBAD-Mtrx-F2 and pBAD-Mtrx-R2 were used to amplify a Gly-Ser-rich linker and mutate the C-terminal residues of rxRFP1 from pBAD-TrxRFP1.²⁴ The C-terminus of rxRFP1 was mutated to that of rxRFP1.1 for a lower mid-point redox potential suited for the mitochondrial redox environment.²⁷ These two fragments were assembled in an overlap PCR reaction using pBAD-Mtrx-F1 and pBAD-Mtrx-R2 as the amplification primers. The resultant PCR product was inserted into a pBAD/His B vector predigested with Xho I and Hind III using Gibson Assembly.²⁸ The product was used to transform cells as described above. The plasmid was isolated and confirmed with Sanger sequencing.

Protein purification and in vitro characterization of MtrxRFP2.

The plasmid pBAD-MtrxRFP2 was used to transform E.coli DH10B cells, and a single colony was used to inoculate 5 mL 2×YT medium supplemented with 100 μg/mL ampicillin. After shaking overnight at 250 r.p.m and 37 °C, the culture was diluted with 500 mL of 2×YT medium containing 100 μg/mL ampicillin. OD₆₀₀ was monitored, and 0.2% (w/v) L-arabinose was added to induce protein expression when OD₆₀₀ reached 0.6. Expression was conducted at 30 °C for 24 h and then at 16 °C for another 72 h. The protein was purified using Ni-NTA agarose beads and size-exclusion chromatography as described above. An alkali denaturation method was used to determine the protein concentration.²⁹ To examine the indicator’s specificity, freshly prepared proteins (1 μM) were incubated with the indicated chemicals or enzymes at room temperature for 10 minutes, and fluorescence with 570 nm excitation and 600 nm emission was measured using the BioTek Synergy Mx Microplate Reader. Human TrxR2 used in these assays was recombinantly expressed and purified from E. coli, and the selenocysteine residue was mutated to cysteine.

Construction of mammalian expression plasmids.

Oligos pcDNA3-TrxRFP2-F and pcDNA3-TrxRFP2-R (Table S1) were used to amplify the TrxRFP2 gene from pBAD-TrxRFP2. The resultant product was inserted into a pcDNA3 vector predigested with Hind III and Xba I via Gibson Assembly.²⁸ To construct plasmids for mitochondrial expression, the TrxRFP1 (or MtrxRFP2) gene was amplified with oligos pCS2-Mito-F and pCS2-Mito-R1 (or pCS2-Mito-R2) (Table S1) from the corresponding pBAD bacterial expression plasmids. The PCR products were inserted between Age I and Xba I of a compatible pCS2+ vector, which contains tandem mitochondrial targeting signals derived from cytochrome c oxidase subunit VIII.

Mammalian cell culture, transfection, and live-cell imaging.

HEK (human embryonic kidney) 293T and HeLa cells were cultured and transfected as previously described.²⁴ At 72 h post-transfection, cells in DPBS supplemented with 1 mM Ca²⁺, 1 mM Mg²⁺ were imaged using a Leica DMi8 inverted microscope equipped with a Leica EL6000 light source, a TRITC filter cube (545/25 nm bandpass excitation and 605/70-nm bandpass emission), and a Photometrics Prime 95B sCMOS camera. To study the time-lapse responses of the transfected cells, auranofin (15 μM) was added to the abovementioned imaging buffer, and images were acquired at the indicated intervals. To test the reversibility of the indicators in live cells, additional DTT (10 mM) was added to cells that were pre-treated with auranofin (15 μM). Images were analyzed using the ImageJ³³ software. Background subtraction was performed by setting the rolling ball radius to 50 pixels. Cells were randomly selected, and the intensity means for regions of interest were extracted for further analysis. Fluorescence intensities at different time points (F) were normalized to the value at t = 0 min (F₀), and the F/F₀ ratios were plotted against time.

RESULTS AND DISCUSSION

Engineering and structural modeling of TrxRFP2, an enhanced TrxRFP1 variant

TrxRFP1 is a genetic fusion of hTrx1 and rxRFP1 via a 30-amino-acid Gly-Ser-rich linker.²⁴ Despite the initial probe being adequate for detecting the redox changes of Trx both in vitro and in live mammalian cells, its responsiveness was not fully optimized. Therefore, we performed seven rounds of error-prone PCR-based directed evolution on TrxRFP1. To screen for candidates with better redox-indued fluorescence response, colonies from each library with medium to high fluorescence were selected, and crude protein extracts were screened for redox-induced responsiveness. We carried out enzyme-based assays in 96-well plates. TPx1 and H₂O₂ were used to oxidize the proteins, while TrxR1 and NADPH were used to induce reduction. Fluorescence intensities at the two states were quantified, and intensity ratios were derived. After screening ~20,000 individual colonies on LB agar plates and testing ~700 clones as cell lysates, we identified a promising mutant showing a larger fluorescence response than the parental TrxRFP1. This mutant, named TrxRFP2 (Figure 1A), contains five mutations from TrxRFP1 (A18T, E56G, A66T, N93D, D248V). We compared the reduction and oxidation of TrxRFP1 and TrxRFP2 using enzyme-based assays. Recombinant TrxR1 and NADPH were used to reduce the oxidized proteins, while recombinant peroxiredoxin TPx1 and H₂O₂ were used to oxidize the reduced proteins. Within the first 60 s, the magnitudes of changes caused by reduction were ~3.0-fold and ~4.2-fold for TrxRFP1 and TrxRFP2, respectively, while the oxidation reactions caused fluorescence increases by ~3.5-fold and ~4.5-fold, respectively (Figure 1BC). The residual oxygen in the reaction systems (e.g., pre-dissolved in the buffers or from the air during the measurements) may impact the fluorescence changes and account for the discrepancies in the response magnitudes during reduction and oxidation. Nevertheless, TrxRFP1 and TrxRFP2 were compared in parallel during these experiments. We fitted the fluorescence responses using a monoexponential decay or pseudo-first order association model. The half-time (t_0.5) for TrxRFP2 reduction was determined to be 5.1 s while t_0.5 for TrxRFP1 reduction was 8.8 s. As for the oxidation kinetics, t_0.5 values were 3.9 and 6.8 s for TrxRFP2 and TrxRFP1, respectively. In both reduction and oxidation assays, TrxRFP2 showed faster kinetics and a more significant response magnitude than TrxRFP1.

Figure 1. Mutations and in vitro response kinetics of TrxRFP2.

(A) Primary sequence alignment of TrxRFP1 and TrxRFP2. Residues mutated in TrxRFP2 via directed evolution from TrxRFP1 are shaded in orange. The active-site cysteine residues in human thioredoxin1 (hTrx1) are shaded in blue. The sequences derived from hTrx1 and rxRFP1 are highlighted in blue and red boxes. The key cysteine residues in the rxRFP1 portion for reversible disulfide formation are shaded in red. The unnumbered N-terminal residues (including the His₆ tag) were derived from the pBAD/His B vector. (BC) Comparison of oxidized (B) or reduced (C) TrxRFP1 and TrxRFP2 in response to enzymatic reduction (10 μM TrxR1 + 200 μM NADPH) and oxidation (20 μM TPx1 + 200 μM H₂O₂). Fluorescence intensities were normalized to the values at t = 0 min. Data represent mean ± SD of three technical repeats. (D) TrxRFP2 structural model built with ColabFold (AlphaFold2 using MMseqs2). Left panel: the overall 3-dimensional model highlighting mutations (cyan), key cysteine residues (blue), and the interdomain floppy linker (black). Right panels: local structures of the TrxRFP2 model (magenta) overlaid with the Trx1 crystal structure (PDB: 3TRX) (cyan), highlighting newly formed H-bonds near A18T (top) and A66T (bottom) mutations.

Recently, artificial intelligence and deep learning have been successfully used to predict 3-dimensional protein structures. AlphaFold2, one of the most advanced computational tools, is known for its impressive accuracy.³¹ We used ColabFold,³⁰ which runs a simplified version of AlphaFold2 by using a fast MMseqs2³² search in the input feature generation stage, to predict the structures of TrxRFP2. Five structures ranked by pTMscores (a computed parameter reflecting the global modeling accuracy) were provided by the program. Among these structures, the folding of the individual hTrx1 and rxRFP1 domains is almost identical, but the relative position of the two domains varies. Within three of the five structures (including the one with the best pTMscore shown in Figure 1D), the reactive cysteines in hTrx1 and rxRFP1 are closely positioned, likely promoting the kinetic disulfide exchange between them. TrxRFP2 is different from TrxRFP1 with five mutations. Although it is hard to pinpoint the contributions of each mutation to the enhanced response kinetics of TrxRFP2, we speculate that the A18T and A66T mutations introduce additional H-bonds into the hTrx1 fold (Figure 1D), thereby enhancing thermostability and potentially affecting the redox reactions. In addition, E56G and N93D are close to the reactive cysteines, possibly affecting local structural flexibility and the electrostatic environment. The D248V mutation is within the rxRFP1 fold, and it seems to stabilize a short α-helix. Further studies are needed to better delineate the roles of these mutations.

Chemical-induced redox response of TrxRFPs in HEK 293T cells

We next directly compared TrxRFP2 with TrxRFP1 in live mammalian cells. Both indicators were transiently expressed in human embryonic kidney (HEK) 293T cells. Fluorescent responses induced by oxidation-stimulating chemicals were examined. Auranofin is a selective TrxR inhibitor.³⁴ TrxRFP1 and TrxRFP2-expressing cells presented a notable fluorescent increase when treated with 15 μM auranofin (Figure 2AB). Moreover, cells expressing TrxRFP2 showed a more noticeable fluorescence increase, and the improvement of the dynamic range was ~1.75-fold (Figure 2A-C). Of note, the auranofin treatment did not interrupt intracellular pH, as shown previously.²⁴ No significant change was observed in both TrxRFP1- and TrxRFP2-expressing cells treated with a low-nanomolar concentration of 2-AAPA, a glutathione-reductase inhibitor (Figure 2C).³⁵ Similarly, no apparent fluorescent responses were caused in both groups of cells after 16-h incubation with 20 μM dimethyl fumarate (DMF),³⁶ a GSH redox reagent involved in GSH synthesis and GSH reductase upregulation (Figure 2C). The data suggest that both TrxRFP1 and TrxRFP2 do not respond to chemicals specifically disturbing the glutathione antioxidant system and are specific indicators for the redox of the Trx system. Furthermore, TrxRFP2-expressing cells pre-treated with auranofin reacted quickly with the further added reductant, DTT, confirming the reversibility of TrxRFP2 in live mammalian cells (Figure 2D). We speculate that several reasons may collectively contribute to the improved performance of TrxRFP2 relative to TrxRFP1 in live cells. In addition to the increased response magnitude and kinetics observed in the previously presented biochemical assays, the mutations in TrxRFP2 may enhance its folding, resulting in a larger portion of proteins in live cells responsive to the auranofin treatment.

Figure 2. TrxRFPs in HEK 293T cells in response to oxidation-stimulating chemicals.

(A) Time course of the responses of HEK 293T cells expressing TrxRFP1 (blue) or TrxRFP2 (orange) to auranofin (15 μM). Fluorescence intensities were normalized to the values of each cell at t = 0 min. Data represent mean ± SD of 15 cells from three technical repeats. (B) Time-lapse pseudocolored fluorescence images (F/F₀) of HEK 293T cells expressing TrxRFP1 or TrxRFP2 treated with 15 μM auranofin. Scale bar, 50 μm. (C) Comparison of fluorescence responses of TrxRFP1 (blue bar) and TrxRFP2 (orange bar). Cells were treated with auranofin (15 μM), 2-AAPA (100 nM), or DMF (20 μM) for 180 min. Fluorescence intensities were normalized to the values at t = 0 min in their corresponding groups. Data represent mean ± SD of 12 cells from three technical repeats. P values were determined by two-way ANOVA with Šidák’s multiple comparisons test (****P < 0.0001; n.s, not significant, P > 0.05). (D) Sequential responses of TrxRFP2 in HEK 293T to auranofin (15 μM) and DTT (10 mM). Fluorescence intensities were normalized to the values at t = 0 min. Data represent mean ± SD of 9 cells from three technical repeats.

Development of MtrxRFP2, an indicator for the redox of mitochondrial Trx2

Trx1 and TrxR1 are primarily cytosolic. In contrast, Trx2 and TrxR2 are found in the mitochondria of mammalian cells. We previously fused TrxRFP1 to a mitochondrial targeting sequence (MTS) and successfully used Mito-TrxRFP1 to monitor compartmentalized Trx oxidation. Although the result suggests that the mitochondrial Trx system can cross-react with Trx1 in TrxRFP1, such reaction is likely not kinetically optimal. Using a sensor design strategy similar to TrxRFP1, we genetically linked rxRFP1.1, which has a redox potential suitable for mitochondria, to human Trx2, through a 30-amino-acid Gly-Ser-rich linker. We hypothesized that within this configuration, the redox change of active Cys residues (C31 and C34) in Trx2 could be coupled to the redox status of the inserted active cysteine pair at the C- and N-termini of rxRFP1.1. Such disulfide exchange reaction could be kinetically preferred due to the close distance between Trx2 and rxRFP1.1 in the fusion construct. As a result, the fluorescence of rxRFP1.1 becomes an indicator for the Trx2 redox status. We named the TrxR2 and rxRFP1.1 fusion construct MtrxRFP2 and purified the protein for further characterization. The freshly purified MtrxRFP2, which was oxidized by air, presented identical excitation and emission peaks at 575 nm and 600 nm as rxRFP1.1 (Figure 3A). Dithiothreitol (DTT) reduced MtrxRFP2 quickly, leading to a 3.5-fold fluorescence decrease (Figure 3A). Oxidized MtrxRFP2 showed minimal response to H₂O₂ and GSH at millimolar concentrations (Figure 3B). In addition, MtrxRFP2 in the oxidized state showed a robust, ~ 3.3-fold fluorescence decrease upon the addition of our purified TrxR2 with NADPH as the co-factor (Figure 3B).

Figure 3. In vitro characterization of purified MtrxRFP2.

(A) Excitation (solid line) and emission (dot line) spectra of MtrxRFP2 in oxidized (red) and reduced (grey) states. (B) Fluorescence responses of MtrxRFP2 post 10-min incubation with: 1, PBS; 2, 1 mM H₂O₂; 3, 1 mM GSH; 4, 20 μM purified TrxR2 + 400 μM NADPH. Data represent mean ± SD of three technical repeats. P values were determined by one-way ANOVA with Dunnett’s multiple comparisons test (****P < 0.0001; n.s, not significant, P > 0.05).

Chemical-induced Trx redox changes in the mitochondria of mammalian cells

We next subcellularly localized TrxRFP1 and MtrxRFP2 to the mitochondria to monitor Trx redox changes. A tandem MTS was appended to the N-terminus of TrxRFP1 or MtrxRFP2 (Figure 4A). Expressing the constructs led to the successful localization of these probes in the mitochondrial compartment of cultured human cervical cancer HeLa cells. When treating the cells expressing mitochondrial TrxRFP1 or MtrxRFP2 with auranofin, the fluorescence intensity increased gradually within the monitored 30-min period. In comparison, mitochondrial MtrxRFP2 presented a more robust fluorescence increase than mitochondrial TrxRFP1 (F/F₀, 140% versus 120%) and faster kinetics (Figure 4B, C). These results indicate that MtrxRFP2 is a better indicator than TrxRFP1 for monitoring Trx2 redox dynamic changes in the mitochondria of mammalian cells. This property could result from the mitochondrial-specific sensory domain Trx2 and the fused RFP1.1, which are equipped with a redox potential more suitable for the mitochondria environment. In addition, we examined the reversibility of mitochondrial MtrxRFP2 in HEK 293T cells. Fast reversed responses to DTT were observed in cells pre-oxidized with auranofin (Figure 4D).

Figure 4. Sequences and fluorescence response of mitochondrially localized MtrxRFP2 and TrxRFP1 in mammalian cells.

(A) Domain arrangements and primary sequences of mitochondrial MtrxRFP2 and TrxRFP1. Top: Mito-MtrxRFP2 highlighting the tandem MTS (orange box) and fragments derived from hTrx2 (magenta box) and rxRFP1.1 (crimson box). Bottom: Mito-TrxRFP1 highlighting the tandem MTS (orange box) and fragments derived from hTrx1 (cyan box) and rxRFP1 (red box). (B) Time-lapse pseudocolored fluorescence images (F/F₀) of HeLa cells expressing mitochondrial TrxRFP1 or MtrxRFP2 treated with 15 μM auranofin. Scale bar, 20 μm. (C) Time course of Mito-TrxRFP1 or Mito-MtrxRFP2 fluorescence in HeLa cells treated with auranofin (15 μM). Fluorescence intensities were normalized to the values of each cell at t = 0 min. Data represent mean ± SD of 15 cells from three technical repeats. (D) Time course responses of MtrxRFP2 in HEK 293T to auranofin (15 μM) and DTT (10 mM). Fluorescence intensities were normalized to the values of each cell at t = 0 min. Data represent mean ± SD of 10 cells from three technical repeats.

CONCLUSION

Trx, as a thiol-dependent antioxidant protein, is essential for redox signaling and has been linked to many pathological processes. We previously reported the first genetically encoded biosensor, TrxRFP1. Based on our previous work, we have developed an improved indicator, TrxRFP2, via random mutagenesis and enzyme-based screening. Compared to TrxRFP1, TrxRFP2 showed improved response kinetics and a greater fluorescence response to TrxR inhibition in mammalian cells without appreciable cross-reaction to the GSH system. Due to its specificity and improved dynamic range, TrxRFP2 is a better indicator for redox biology studies.

MtrxRFP2 was created by linking a redox-sensitive rxRFP1.1 and mitochondrial Trx (Trx2). The proximity forced by the linker in the fusion construct indeed facilitated the disulfide exchange between rxRFP1.1 and Trx2. MtrxRFP2 displayed a ~ 3.5-fold dynamic range. We compared subcellularly localized MtrxRFP2 to TrxRFP1 for monitoring mitochondrial Trx redox changes. MtrxRFP2-expressing cells showed larger and faster responses upon auranofin treatment than TrxRFP1-expressing cells. Thus, although mitochondrially localized TrxRFP1 can detect the dynamics of mitochondrial Trx redox, MtrxRFP2 derived from Trx2 is a more robust indicator for monitoring Trx redox dynamics in mitochondria.