Genetically encoded fluorescent redox indicators for unveiling redox signaling and oxidative toxicity

Abstract

Redox-active molecules play essential roles in cell homeostasis, signaling, and other biological processes. Dysregulation of redox signaling can lead to toxic effects and subsequently cause diseases. Therefore, real-time tracking of specific redox-signaling molecules in live cells would be critical for deciphering their functional roles in pathophysiology. Fluorescent protein (FP)-based genetically encoded redox indicators (GERIs) have emerged as valuable tools for monitoring the redox states of various redox-active molecules from subcellular compartments to live organisms. In the first section of this review, we overview the background, focusing on the sensing mechanisms of various GERIs. Next, we review a list of selected GERIs according to their analytical targets and discuss their key biophysical and biochemical properties. In the third section, we provide several examples, which applied GERIs to understanding redox signaling and oxidative toxicology in pathophysiological processes. Lastly, a summary and outlook section is included.

Graphic Abstract (For TOC only)

Introduction

Redox-active proteins (e.g., thioredoxin (Trx) and glutaredoxin (Grx)) and small molecules (e.g., various reactive oxygen species (ROS) and reactive nitrogen species (RNS)) are essential for maintaining cellular and physiological homeostasis.^1-5 They are involved in cell signaling and numerous physiological processes, such as cell cycle and differentiation, embryogenesis, inflammation, and tissue regeneration.^1-5 For example, cell growth factors, such as epidermal growth factor (EGF) and platelet-derived growth factor (PDGF), can increase the production of hydrogen peroxide (H₂O₂), a type of ROS, to promote signal transduction of the receptor tyrosine kinases (RTKs).⁶ Another well-known example is that phagocytes can generate ROS/RNS bursts to clear pathogens.^7,8 Redox-active molecules are also crucial for many pathological processes. Dysregulation of the cellular redox states can lead to oxidative stress, which is considered a central mechanism of toxicity in various tissues and organs.⁹ Oxidative stress can cause damage to different biomolecules such as DNA, proteins, and lipids, leading to the initiation and progression of diverse diseases, including cancer, neurodegeneration, respiratory stress, and kidney disorders.^10-17 There is significantly increasing interest and demand in detecting and monitoring redox-active molecules in live cells and organisms.

An extensive array of methods and assays have been developed to detect redox-active molecules and processes. These methods include electron paramagnetic resonance (EPR) spectroscopy,¹⁸ chemiluminescence assays,¹⁹ and fluorescence assays based on either small molecule-based synthetic dyes^20-24 or genetically encoded redox indicators(GERIs).^25-29 GERI-based fluorescence methods have gradually gained popularity for several reasons: First, these indicators can be expressed by cell machinery, providing chances to measure redox signals in the natural cellular contexts while introducing minimal toxicity. Second, using tissue-specific promoters or subcellular localization sequences, GERIs can be readily localized to specific tissues and subcellular compartments, giving superior spatial and temporal resolution over small molecule synthetic dyes. Also, GERIs can be disseminated as plasmids or viral vectors, enabling convenient and broad adaptation by other interested researchers. Moreover, GERIs with compatible colors for different redox parameters have enabled the simultaneous monitoring of multiple redox signals to unveil the interplays between several redox-active molecules or processes.

A classical fluorescent protein (FP) consists of an endogenous chromophore surrounded by a stable 11-stranded β-barrel structure, which shields the chromophore from the external environment.³⁰ The earliest strategy to make GERIs was introducing cysteine pairs into the scaffolds of single FPs (Fig. 1A). The introduced cysteine residues at appropriate locations can reversibly form intramolecular disulfide bonds upon sensing a change in the redox environment.^31,32 The formation of the disulfide bonds leads to a conformational change in the scaffolds of FPs, which consequently modulates the chromophore's photophysical properties. Researchers can thus record these changes by examining fluorescence excitation or emission (i.e., intensity, peak shape or wavelength, or lifetime). To date, a color palette of GERIs has been developed by using this strategy.^31-34 However, these indicators often have limited specificity and respond to the general redox environment changes. To overcome this limitation, alternative GERIs have been created by fusing a redox-sensitive FP with a selected redox-active enzyme (e.g., Grx, Trx, and oxidant receptor 1 (Orp1)) to form a redox relay (Fig. 1B).^35-37 The rationale behind this is that the fused enzyme can improve sensor specificity by reacting with specific redox-active molecules because the redox status of the enzyme is kinetically coupled with the redox status of the nearby, redox-sensitive FP in the same fusion construct.

Fig.1. Illustration of the sensing mechanisms of common GERIs.

(A) A cysteine pair is introduced into FP and the reversible disulfide bond modulates the FP fluorescence. (B) A redox-active enzyme is genetically fused to a redox-active FP and the redox status of the enzyme is kinetically coupled with the redox of the redox-active FP. (C) A cpFP is fused with a redox sensory domain in an optimized configuration with which the redox reaction in the sensory domain is capable of modulating the fluorescence of cpFP. (D) A cpFP is fused with sensory elements capable of ligand binding. (E) GERIs containing special noncanonical amino acids (ncAAs) react with specific ROS or RNS, causing FP fluorescence changes. (F) A redox-sensory element is sandwiched between a FP-based FRET pair and the redox status of the sensory element modulates FRET efficiency.

In addition to the GERIs mentioned above, other GERIs were developed using a different approach, which fuses redox sensory domains with circularly permutated FPs (cpFPs).³⁸ cpFPs are engineered FP variants with a different topology from the wide-type FPs.³⁸ To create a cpFP, the original N- and C- termini of a wild-type FP are connected using a flexible linker, while the new N- and C- termini are created near the chromophore.³⁸ The sensory domains are fused to the cpFP via these new termini, so the spectral properties are more readily modulated by structural rearrangements originated from the interaction or reaction of the redox sensory domains with analytes (Fig. 1C). For example, a family of GERIs for sensing H₂O₂, namely HyPers,^39-43 was developed by fusing OxyR, an H₂O₂-activated transcription factor, to circularly permuted yellow FP (cpYFP) or red FP (cpRFP). Oxidation of these OxyR-based GERIs increases their fluorescence.

In some other examples, the sensory domains do not react with redox molecules but rather bind analytes (Fig. 1D). Moreover, the sensory domain may bind multiple analytes with distinct affinities. With this unique feature, researchers can apply this type of GERIs to measuring the ratio of analytes rather than the absolute concentration changes of a specific analyte. A classic GERI of this type is SoNar for monitoring the nicotinamide adenine dinucleotide (NAD⁺/NADH) redox couple.⁴⁴

Alternatively, GERIs have been developed by incorporating noncanonical amino acids (ncAAs) into FPs using an expanded genetic code (Fig. 1E). Using molecular biology, researchers can integrate codons for ncAAs into the DNA-coding sequence of an FP.⁴⁵ Co-expression of the modified FP gene with an engineered tRNA/synthetase pair in live cells, cultured in the presence of the corresponding ncAA, enables the site-specific incorporation of the ncAA into the FP.⁴⁵ By selecting suitable ncAAs with unique functional groups, the ncAAs may bind to or react with analytes of interest, further modulating the fluorescence of FPs.^45-48

In addition to those above single FP-based GERIs, GERIs involving multiple FPs and often based on Förster resonance energy transfer (FRET) have also been developed. FRET describes an energy transfer process from a donor chromophore to an acceptor chromophore with suitable distance, spectral overlap, and dipole coupling.^49,50 When FRET occurs, the donor's fluorescence emission decreases, whereas the emission from the acceptor increases. The acceptor-to-donor emission ratio changes based on FRET efficiency, which is sensitive to the distance and relative dipole orientation of the donor and acceptor chromophores.⁴⁹ In FRET-based GERIs, a redox sensory element is often sandwiched between a donor FP and an acceptor FP. In response to an analyte, the sensory element undergoes a conformational change. Consequently, the distance and geometry between the donor and the acceptor are altered, and the FRET efficiency between the two FPs changes.⁵¹ (Fig. 1F). One commonly used sensory element is an artificial redox linker (RL) containing adjacent or dispersed cysteine residues between an FP-based FRET pair. The redox of the cysteine pair can alter the distance between FPs and change the FRET efficiencies between them.⁵² In another example, two repeats of the C-terminal cysteine-rich domain of yeast Yap1 transcriptional factor⁵³ were sandwiched between a cyan FPs (CFPs), Cerulean,⁵⁴ and a YFP, Citrine,⁵⁵ resulting in a Redoxfluor indicator.⁵⁶ Oxidation of the cysteine-rich domain decreases the FRET efficiency from Cerulean to Citrine, giving a redox-dependent emission-ratiometric readout. FRET-based GERIs provide ratiometric readouts, which are advantageous because such readouts are less likely affected by variations such as instrumental setups and sensor expression levels. However, FRET-based GERIs usually have larger molecular weights and lower dynamic range (a.k.a. the fold of signal change) than single FP-based indicators. Additionally, the FPs in the same FRET indicator may possess different pH sensitivity, brightness, photostability, and maturation rate, thereby affecting responses and complicating result interpretation.

GERIs by analyte types

In this section, we discuss a list of selected GERIs according to analytical targets. The discussions focus on their key biophysical and biochemical properties. All of these GERIs have been applied to live-cell imaging, and some have been used in multicellular organisms and animals. Examples of these applications are further discussed in the next section, “Selected applications of GERIs”.

1. GERIs for general thiol redox status

GERIs in this category were generated by directly introducing cysteine residues into FP scaffolds (Table 1). These cysteines can form disulfide bonds upon oxidation, while the oxidized disulfide bonds can be reversibly reduced (Figure 1a). Moreover, the selected residue positions are typically close to the chromophore, so the redox reaction alters the fluorescence of FPs. In principle, the introduced cysteines in redox-sensitive FPs can equilibrate with an extensive array of intracellular thiols and disulfides. Because the reduced (GSH) and oxidized (GSSG) glutathione are highly abundant in most live cells and organisms, these redox-sensitive FPs are often considered indicators for the redox of GSH/GSSG. Furthermore, Grx in live cells may catalyze the redox equilibration between redox-sensitive FPs and the glutathione pool.

Table 1.

Representative redox-sensitive FPs.

GERIs	Disulfide Bond	λex, (nm)	λem, (nm)	Midpoint potential E°’ (mV)	Recommended pH control
rxYFP57	C149-C202	512	523	−261	YFP
roGFPs31	C147-C204 or/and C149-C202	400/475 or 400/490	510	−272~-299	pH resistant in the excitation-ratiometric mode
roGFP1-iX61, ERroGFP-S464	C147-C204	395 and 465	505	−229~-246	pH resistant in the excitation-ratiometric mode
rxRFPs33,34	N- and C-terminal cysteines	576	600	−244~-324	pHRFP
rxmRuby268	C148-C200	560	600	−265	mRuby2
Oba-Q series70	C147-C204	370/380/430	425/440/480	−232~-249	pH resistant at physiological pH
Re-Q series75	C147-C204	430/515	480/525	−251~-286	pH resistant at physiological pH
roUnaG78	C56-C64	498	527	−275	pH resistant
FROG/B83	C149-C202	400(Abs)	460/510	-293	NA

1.1. rxYFP

The first developed single FP-based GERI is redox-sensitive yellow FP (rxYFP).⁵⁷ Two cysteine mutations (N149C and S202C) were introduced into YFP. In response to oxidation, the formation of an intramolecular disulfide bond between the two surface residues on the adjacent β-strands of the FP leads to a 2.2-fold fluorescence decrease with peak emission at 523 nm. The redox midpoint potential of rxYFP is −261mV, implying its broad suitability for use in common intracellular compartments.⁵⁷ Because rxYFP is sensitive to pH changes and chloride ions,^58,59 extra caution is required when using this indicator.

1.2. roGFPs

To generate an intrinsically ratiometric redox-sensitive FP that is insensitive to chloride, researchers developed a series of redox-sensitive green FPs (roGFPs) by introducing two cysteine residues into residues on neighboring β-strands of enhanced green FP (EGFP).³¹ Both roGFP1 and roGFP2 have two excitation peaks. The peak at ~ 400 nm corresponds to the chromophore’s protonated form, while the peak at ~ 480 nm represents the chromophore’s anionic form. The oxidation of roGFP1 or roGFP2 favors the protonated neutral chromophore, leading to increased fluorescence excitation at ~ 400 nm (F₄₀₀) and a concomitantly decreased fluorescence excitation at ~ 480 nm (F₄₈₀).³¹ Thus, roGFP1 and roGFP2 can provide an excitation-ratiometric readout (F₄₀₀/F₄₈₀). In addition, the fluorescence intensity of roGFP1 or roGFP2, when excited at wavelengths close to either of the two peaks, is pH-sensitive, but the ratiometric readout (F₄₀₀/F₄₈₀) is less sensitive to pH changes, resulting in a unique advantage of roGFPs over rxYFP.³¹ When using roGFP1 and roGFP2 in actual studies, we must consider several differences between them. First, the primary excitation peak of roGFP1 is at ~ 400 nm, while the predominant excitation peak of roGFP2 is at ~ 490 nm. Moreover, the midpoint redox potential of roGFP1 is −291 mV, different from roGFP2 (−272 mV). Both roGFP1 and roGFP2 are more reducing and show more significant fluorescence changes (a.k.a. larger dynamic range) than rxYFP.^31,60

Using a similar strategy, researchers introduced cysteine pairs to alternative residue sites in EGFP to derive roGFP3 and roGFP4. Meanwhile, a total of four cysteine residues (two pairs) were introduced into EGFP to create roGFP5 and roGFP6.³¹ Among these roGFPs, roGFP3 has the most negative midpoint redox potential (−299 mV), making it suitable for use in highly reducing subcellular compartments such as the cell nucleus and mitochondria;³¹ roGFP2 has the least negative midpoint redox potential (−272 mV), but it is still not adequate for use in oxidizing subcellular compartments such as the endoplasmic reticulum (ER).³¹ Thus, in a subsequent study, single amino acids were inserted into the main chain of roGFP1 at a location adjacent to the active cysteine 147 to lower the thermodynamic stability of the disulfide bond, resulting in a new family of roGFP1-derived GERIs (roGFP1-iX) with midpoint potentials of −229 to −246 mV.⁶¹ A particular mutant, roGFP1-iL, with a midpoint redox potential of −229mV, is best suited for imaging redox dynamics in ER.⁶¹ A similar effort was devoted to modifying roGFP2, but the resultant roGFP2-iL mutant has a much-reduced dynamic range.⁶² Furthermore, because of the low fluorescence of ER-targeted roGFP-iL resulting from inefficient folding, mutations from superfolder GFP (sfGFP)⁶³ were introduced into roGFP-iL to derive ER-targeted roGFP-S4 (ERroGFP-S4), which has much-improved brightness and dynamic range for reliable monitoring of redox change in the ER.⁶⁴

Recently, Nocross et al. fused roGFPs with RFPs to create sensors with partial red fluorescence.⁶⁵ In this case, sensitized red emission was generated via FRET from roGFPs to RFPs while the redox properties of roGFPs were retained.⁶⁵ These new indicators were used along with roGFPs to monitor redox dynamics in multiple cellular compartments simultaneously.

1.3. rxRFPs

Red-shifted GERIs can enhance tissue penetration and provide new capabilities for multiparameter imaging along with GFP-based indicators. Our research group developed the first redox-sensitive RFP (rxRFP) from a circularly permutated RFP, mApple⁶⁶ (cpmApple).³³ The cpmApple fragment, sufficient for autocatalytic chromophore formation, was derived from R-GECO1, a red fluorescence Ca²⁺ indicator.⁶⁷ Unlike rxYFP and roGFPs, in which cysteine pairs were inserted to residues at the surface of the FP β-barrel, rxRFP was developed by adding cysteine residues to the N- and C- termini of cpmApple, followed by subsequent optimization of other residues adjacent to the cysteine residues.³³ The resultant rxRFP has a single excitation peak at 576 nm, a single emission peak at 600 nm, and a 4-fold fluorescence increase upon oxidation from the fully reduced state. In the same study, pHRFP, a red fluorescent pH indicator with a similar pKa to rxRFP, was developed and used as a pH control in live-cell imaging experiments.³³ To further match rxRFP with different midpoint redox potentials of various subcellular compartments, our group developed several rxRFP variants (rxRFP1.X) by mutating the amino acids in proximity to the inserted cysteines residues. rxRFP1.1.-1.4 have midpoint redox potentials ranging from −324 to −244 mV and respond to redox perturbations when localized to suitable subcellular compartments such as the cell nucleus, mitochondria, and ER in mammalian cells.^33,34

1.4. rxmRuby2

Another reported redox-sensitive RFP, rxmRuby2,⁶⁸ was generated by introducing surface-exposed cysteine residues into the β-barrel of mRuby2,⁶⁹ an RFP in the wild-type topology. rxmRuby2 showed fluorescence decrease upon oxidation and fluorescence increase upon reduction and responded to various oxidants and reductants. For example, both GSSG and diamide can oxidize rxmRuby2 in vitro to a higher degree than H₂O₂. On the other hand, only a strong reductant, dithiothreitol (DTT), but not bacterial Trx or nicotinamide adenine dinucleotide phosphate (NADPH), can directly reduce rxmRuby2. It is worth noting that the fluorescence of rxmRuby2 is still sensitive to pH changes, and researchers should include proper controls in their experiments of interest.⁶⁸ rxmRuby2 has been fused with mClover,⁶⁹ a GFP variant. The fusion construct can give out a ratiometric readout and has shown reversible responses to redox changes in live cells.⁶⁸

1.5. Oba-Q and Re-Q series

Oxidation balance sensed quenching proteins (Oba-Q)⁷⁰ are GERIs with fluorescence significantly quenched upon oxidation. Cysteine residues were first introduced into residues 147 and 204 of several CFPs and a blue FP (BFP).⁷⁰ An extra amino acid was then inserted after Cys147, followed by additional mutations around Cys147 for better quantum yield and dynamic range.⁷⁰ In particular, Oba-Qs was derived by introducing three mutations (I146N/S147CD/Q204C) into a CFP named Sirius,⁷¹ while Oba-Qc was generated by introducing three mutations (I146G/S147CE/Q204C) into a mutated CFP (CFP’).⁷⁰ The maximal fluorescence intensity change of Oba-Qs is 6.3-fold, while that of Oba-Qc is only 2.3-fold. The midpoint redox potential values of Oba-Qs and Oba-Qc are −232 mV and −249 mV, respectively, making them suitable for use in the ER, a relatively oxidizing subcellular compartment.⁷⁰ Similarly, introducing mutations (I146G/S147CE/Q204C) into an enhanced BFP, EBFP2,⁷² derived Oba-Qb, an indicator better suited for reducing subcellular compartments due to its more negative redox potential (−263 mV).⁷⁰ The same engineering strategy has been applied to additional CFPs and YFPs, such as mTurquise⁷³ and Venus.⁷⁴ Some of these proteins show fluorescence decrease upon reduction. Such response is opposite to that of Oba-Q, and this new group of proteins was thus named reduction-sensed quenching proteins (Re-Q).⁷⁵ The midpoint redox potentials of this new series of proteins span a broad range from −251 to −286 mV.

1.6. roUnaG

The above-mentioned redox-sensitive FPs were all derived from Aequorea victoria GFP (avGFP) or avGFP-like proteins, and the formation of their chromophores requires molecular oxygen (O₂). Consequently, these GERIs are not well compatible with hypoxic or anoxic conditions. An oxygen-independent GFP, UnaG, was developed in 2013.⁷⁶ The binding of the protein to a cofactor, bilirubin (BR), a catabolic product of heme metabolism, triggers the fluorescence of UnaG.⁷⁷ By introducing a cysteine pair (E56C/D64C) to consecutive β-strands of UnaG, roUnaG is the first oxygen-independent GERI.⁷⁸ The mutant with the double mutations (E56C/D64C) shows slow oxidation kinetics due to backbone constraint, and the introduction of an additional K54T substitution reduced the size of a sidechain close to Cys56 and improved the oxidation kinetics.⁷⁸ roUnaG (UnaG-E56C/D64C/K54T) works well under hypoxia and anoxic conditions and shows an 8-fold fluorescence increase upon oxidation. In addition, the fluorescence of roUnaG is resistant to pH changes, overcoming a significant limitation of GERIs derived from avGFP-like proteins.⁷⁸ With a single excitation and a single emission peak, roUnaG only provides intensiometric readouts, but it has been fused with an RFP, mCherry,⁷⁹ for ratiometric readouts.^77,78

1.7. FROG/B

The wild-type avGFP⁸⁰ has two major absorption peaks at ~ 400 and 480 nm, corresponding to its chromophore’s neutral and anionic forms, respectively. The engineering of avGFP has resulted in variants with only one absorption peak. One such widely known example is EGFP,⁸¹ with a single absorption peak at ~ 480 nm, while other FP variants such as Sapphire⁸² and mKalama⁷² have a single absorption peak at ~ 400 nm at neutral pH. After absorbing ~ 400 nm light, mKalama emits blue light at ~ 460 nm, while Sapphire emits green light at ~ 510nm via a process known as excitation state proton transfer (ESPT). Using a cysteine pair to modulate the efficiency of ESPT, a single-FP-based GERI called FROG/B has recently been developed with dual emission at ~ 460 and 510 nm upon excitation at ~ 400 nm.⁸³ The oxidation of FROG/B enhances ESPT, resulting in a shift of its emission from blue to green. The overall response at pH 7 is ~ 5-fold in terms of the green-to-blue emission ratio, and the response is fully reversible.⁸³ FROG/B has a midpoint redox potential of −293 mV at pH 7.0, lower than rxYFP and roGFP. Thus, FROG/B is well suited for monitoring weak oxidative stress.⁸³ FROG/B is unique due to its emission ratiometric response. It is of note that the green-to-blue emission ratio of FROG/B in the oxidized state is still quite sensitive to pH changes.⁸³

2. GERIs for GSH/GSSG

Although redox-sensitive FPs in the above category are often considered to sense the redox of GSH/GSSH in live cells, their direct equilibration with glutathione is slow in the absence of Grx. Therefore, these redox-sensitive FPs have been fused with Grx to derive a new family of GERIs. Placing a redox-sensitive FP close to Grx generates a redox relay that kinetically couples the redox of the redox-sensitive FP with the redox of the glutathione pool (Figure 1b). The relay between Grx and redox-sensitive FPs increases response kinetics and specificity but does not change their midpoint redox potentials. This fusion strategy is highly generalizable, and in theory, can be applied to any redox-sensitive FP. Below we discuss several selected examples.

2.1. rxYFP-Grx1p

The first redox-relay fusion probe, rxYFP-Grx1p, was developed by the Winther lab.⁸⁴ The coupling of yeast Grx1p with rxYFP significantly increased the oxidation kinetics of rxYFP by GSSG. Since the redox reaction of rxYFP-Grx1p is independent of endogenous Grx, redox measurements can be performed in diverse subcellular compartments and organisms. In addition to faster kinetics, rxYFP-Grx1p provides nearly absolute glutathione specificity due to its low reactivity to other oxidants.⁸⁴ The crystal structure of the fusion protein indicated that the Grx1p and rxYFP do not have specific contact.⁸⁵ Thus, it is speculated that the enhanced thiol-disulfide exchange efficiency between Grx1p and rxYFP might result from a spatial constraint.⁸⁵ The linker limits the diffusion that drives Grx1p and rxYFP apart and enhances their collision in solution, thereby favoring the thiol-disulfide exchange between them.⁸⁵

2.2. Grx1-roGFP2

Grx-roGFP2 is probably the most well-characterized indicator for real-time monitoring of the glutathione redox status.³⁵ The thiol-disulfide exchange between Grx and roGFP2 is achieved through a symmetrical ping-pong process which transfers the disulfide bonds via a monothiol mechanism.⁸⁶ The exchange between glutathione and the indicator happens through three consecutive steps, which are fully reversible. In the first step, the nucleophilic cysteine of Grx1 specifically reacts with GSSG to form a mixed Grx1-glutathione disulfide intermediate. The intermediate subsequently reacts with one of the two thiols on roGFP2, generating S-glutathionylated roGFP2. Last, glutathionylated roGFP2 forms an intramolecular disulfide bond through rearrangement.²⁷ There are two reasons why Grx1-roGFP2 has very high specificity for GSH/GSSG redox pair. First, Grx can specifically respond to glutathione redox, and its specificity is maintained in the fusion probe. Second, Grx1-roGFP2 is not capable of interacting with the Trx system because of the steric constraints.²⁷ The kinetics of electron flux from the reduced probe to GSSG is enhanced at least 100,000 times compared to roGFP2. With its improved dynamic equilibration with GSH/GSSG, Grx-roGFP2 is an excellent tool for studying short-lived and weak oxidative events invisible to non-fused roGFP2.³⁵

2.3. Grx1-roCherry

To create a red GERI for glutathione redox, Shokhina et al. developed roCherry, a redox-sensitive RFP.⁸⁷ Two cysteine residues were inserted at residues 150 and 203 of mCherry⁷⁹ to create the prototypic roCherry. A threonine insertion after Cys150 and an additional point mutation (S151E) were then introduced to increase structural mobility and dynamic range.⁸⁷ Grx1 was then fused to the N-terminus of roCherry via a floppy linker from the Peredox indicator.⁸⁸ The resultant Grx1-roCherry provides intensiometric readouts with a single excitation peak at ~589 nm and a single emission peak at ~610 nm. The probe does not directly respond to H₂O₂ in vitro but shows a detectable response to H₂O₂ in live cells. It was speculated that H₂O₂ could shift the GSH/GSSG ratio in live cells. Grx1-roCherry loses its activity after incubation with 2-acetylamino-3-[4-(2-acetylamino-2-carboxy-ethylsulfanylthiocarbonylamino)phenylthiocarbamoylsulfanyl]propionic acid (2-AAPA), a glutathione reductase (GR) inhibitor.^87,89 Compared to Grx1-roGFP2, Grx1-roCherry is less specific and displays a response to a high concentration of nitric oxide.⁸⁷

3. GERIs for the redox of Trx

Trx is a class of small, 12-kDa antioxidant proteins broadly found in all three domains of life.⁹⁰ Trx facilitates the reduction of its interacting proteins via cysteine thiol-disulfide exchange.⁹¹ There is continuous interest in developing GERIs to monitor the redox status of live-cell Trx.

3.1. TrxRFP

Early attempts to develop GERIs for the redox of Trx by fusing Trx with roGFPs were unsuccessful. The most feasible explanation is that Trx, unlike Grx, cannot form redox relays with roGFPs due to geometric constraints.²⁷ Our research group developed TrxRFP, the first GERI for sensing Trx redox.³⁶ TrxRFP was engineered by directly fusing human Trx1 (hTrx1) with rxRFP through a Gly-Ser-rich linker to create a redox relay, which couples the active cysteines of hTrx1 (C32 and C35) to the active cysteine pair in rxRFP (Figure 1b). Other non-essential Cys residues in hTrx1 were mutated to Ser to avoid crosstalk with the glutathione system.⁹² rxRFP is in a circularly permuted topology with cysteine residues located at the C- and N-termini of the cpFP, possibly allowing an efficient disulfide exchange with hTrx1. TrxRFP can be reduced explicitly by Trx reductase (TrxR) in the presence of NADPH, while rxRFP alone shows no response to TrxR and NADPH. Therefore, it is proposed that the TrxR enzyme system reduces the disulfide in hTrx1, and the thiols are subsequently transferred to rxRFP1 through the redox relay.³⁶ The process is reversible, and the reduced TrxRFP can be oxidized by thioredoxin peroxidase (TPx) in the presence of H₂O_2. TrxRFP does not present a significant response to H₂O₂, GSSG, Grx1, or Trx-related proteins (TRPs), confirming its specificity for Trx redox. TrxRFP1 has been combined with Grx1-roGFP2 to monitor Trx and glutathione redox parallelly. The redox of glutathione and Trx1 can each be disturbed individually without affecting the other in mammalian cells.³⁶ Auranofin, a TrxR reductase inhibitor,⁹³ at low levels in mammalian cells shifts TrxRFP1, but not Grx1-roGFP2, to a more oxidized state.³⁶

3.2. CROST

Change in the redox state of Trx (CROST) is a group of FRET-based GERIs for Trx in photosynthetic bacteria. CROST1 and CROST2 were made by sandwiching fragments derived from Trx-targeted proteins (CP12-2 or CP12) conserved among photosynthetic organisms, between a CFP (mTurquoise Δ11) and a cpYFP (cpVenus).^73,74,94 Elevated Trx oxidation promotes the disulfide bond formation in CP12-2 or CP12, decreasing the FRET donor and acceptor’s distance and enhancing FRET efficiency (Figure 1f).⁹⁴ While CROST1 has a midpoint redox potential of −266 mV at pH 7.5, CROST2 has a more negative midpoint redox potential (−296 mV).⁹⁴ CROST1 and CROST2 display differential responses to different Trx isoforms from plants or cyanobacteria, potentially due to the difference in the midpoint redox potentials. Both indicators have been used for monitoring the redox change of Trx in chloroplasts.⁹⁴

4. GERIs for H₂O₂

H₂O₂ is an important reactive oxygen species and an essential signaling molecule.⁹⁵ Several GERIs for H₂O₂ are now available. Below we discuss a few selected examples (Table 2).

Table 2.

Representative GERIs for H₂O₂.

GERIs	Sensory Domain	Signal	λex, (nm)	λem, (nm)
roGFP2-Orp137	Orp1	GFP	390/480	510-530
roGFP2-Tsa2ΔCR97	Tsa2	GFP	405/488	510
roGFP2-Tpx1.C169S98	Tpx1.C169S	GFP	405/488	510
HyPer39	OxyR (E. coli)	cpYFP	420/500	516
HyPer240	OxyR (E. coli)	cpYFP	420/500	516
HyPer341	OxyR (E. coli)	cpYFP	420/500	516
HyPerRed42	OxyR (E. coli)	cpmApple	575	605
HyPer743	OxyR (N. meningitidis)	cpYFP	400/500	520
NeonOxIrr101	OxyR (E. coli)	cpmNeonGreen	508	520
TriPer104	OxyR (E. coli)	cpYFP	405/488	~ 516
OxyFRET105	Yap1	FRET	435	535/480
PerFRET105	Yap1 and Orp1	FRET	435	535/480
Prx2-based FRET108	Prx2	FRET	488	525/625

4.1. roGFP2-Orp1

A redox relay was created (Figure 1b) by fusing roGFP2 to Orp1, a yeast peroxidase,⁹⁶ resulting in roGFP2-Orp1 that can sense H₂O₂.³⁷ The proximity between roGFP2 and Orp1 converts physiological H₂O₂ signals into measurable fluorescent signals. Oxidation of Orp1 by H₂O₂ first generates an intramolecular disulfide bond between C36 and C82 in Orp1, and the disulfide bridge is subsequently transferred to roGFP2 through a thiol-disulfide exchange. Orp1 can generate a disulfide bridge in roGFP2 using H₂O₂ in a near-stoichiometric way.³⁷ Oxidized roGFP2-Orp1 is reduced by the Trx and Grx/glutathione systems. Thus, the readout of roGFP2-Orp1 is affected by both H₂O₂-induced oxidation and Trx/Grx/glutathione-induced reduction.³⁷ Using a similar strategy, roGFP2 has recently been linked to peroxiredoxins (Prxs) lacking resolving Cys, leading to roGFP2-Tsa2ΔC_R⁹⁷ (2-Cys Prx from S. cerevisiae) and roGFP2-Tpx1.C169S.⁹⁸ These two Prx-based GERIs are sensitive to very low concentrations of H₂O₂ (their limits of detection were estimated to be a few nanomolar). To date, the applications of roGFP2-Tsa2ΔC_R and roGFP2-Tpx1.C169S are mostly limited to yeast.

4.2. HyPers

HyPers are a family of H₂O₂ indicators made by inserting a cpFP into a H₂O₂ sensory domain derived from bacterial transcriptional factor OxyR (Figure 1c).^39–43 Upon oxidation, the active Cys199 of OxyR is repelled by the hydrophobic pocket and forced into proximity with Cys208 to form an intramolecular disulfide bond.⁹⁹ Disulfide bridge formation results in conformational changes of OxyR, which can modulate the fluorescent signal of the inserted cpFP. The oxidation of HyPers can be reversed by cellular enzymes engaged in the thiol-disulfide exchange.^39-43

The first indicator of this family, HyPer, was created by integrating a cpYFP between 205 and 206 sites of Escherichia coli (E. coli) OxyR regulatory domain (Oxy-RD).³⁹ HyPer is an excitation-ratiometric indicator with two excitation peaks at ~420 and ~500 nm and a single emission peak at ~516 nm. Upon oxidation, the fluorescence excited at 420 nm (F₄₂₀) decreases, while the fluorescence excited at 500 nm increases (F₅₀₀). The maximal ratiometric excitation change (F₅₀₀/F₄₂₀) can be up to ~ 3-fold.³⁹ In 2011, an advanced version of HyPer named HyPer2 was developed by introducing a single point mutation A406V to the sensory domain OxyR-RD. HyPer2 displays an improved dynamic range, which is twice of that of HyPer.⁴⁰ Later on, HyPer3 was developed by introducing another point mutation, H114Y, to OxyR-RD. HyPer3 gains an expanded dynamic range compared to HyPer and faster kinetics than HyPer2.⁴¹ The red version of HyPer, HyPerRed, was designed by replacing the cpYFP with cpmApple.⁴² Unlike HyPer, HyPer2, and HyPer3, HyPerRed is an intensiometric indicator whose emission increases by ~ 2-fold upon oxidation by H₂O₂.⁴² Because HyPers are pH-sensitive, a pH control SypHer was also developed.¹⁰⁰ By introducing a point mutation C199S, SypHer is insensitive to H₂O₂ but shows the same pH sensitivity as HyPer.¹⁰⁰

HyPer7, the most updated version of the HyPer family, was developed by inserting cpGFP into an OxyR-RD from Neisseria meningitidis (N. meningitidis).⁴³ HyPer7 is 15- to 17- fold brighter than HyPer3, and its ratiometric readout (F₅₀₀/F₄₀₀) is stable within the physiological pH range. Compared to the previous HyPer indicators, HyPer7 can respond to a relatively low concentration range of H₂O_2, and the response is much more rapid.⁴³ The minimum H₂O₂ concentration needed for a ratiometric increase of 0.1 is approximately 1 μM for HyPer and HyPer3 but only 0.03 μM for HyPer7. The reaction rate of HyPer7 with H₂O₂ is ~ 60 and 86 times faster than those of HyPer and HyPer3.⁴³

4.3. NeonOxIrr and TriPer

Like HyPer family indicators, NeonOxIrr was constructed by fusing a cpmNeonGreen to the truncated OxyR from E.coli.^101,102 NeonOxIrr displayed 5.9-fold higher brightness, 15-fold faster oxidation rate, and 7.6-fold longer reduction half-time than HyPer3.¹⁰¹ Additionally, NeonOxIrr has demonstrated a 2.3-fold dynamic range in response to H₂O₂ with higher stability within the pH range of 5.5-7.5.¹⁰¹ The slow reduction rate of NeonOxIrr can help integrate signals over time; on the other hand, it makes NeonOxIrr unsuitable for real-time monitoring of fast H₂O₂ dynamics.¹⁰¹

HyPer is inadequate to sense H₂O₂ in the thiol-oxidizing ER environment due to its low redox potential. In HyPer, C199 of OxyR reacts with H₂O₂ to form a sulfenic acid, which C208 further attacks to form a disulfide bond. The disulfide bond would be continuously formed in the presence of the ER protein disulfide isomerases (PDIs), depleting all reduced HyPer.¹⁰³ By replacing Ala187 in OxyR with Cys, a tri-Cys indicator, TriPer, was generated.¹⁰⁴ It is speculated that the presence of this additional cysteine creates a diversionary disulfide, C187-C208, keeping a fraction of C199 in its reduced form. Thus, a fraction of the C199 thiolate could still react with H₂O₂ to form a sulfenic acid intermediate, which could be further converted into a low-fluorescence state when excited at 488 nm. As a result, the ER-expressed TriPer responded to H₂O_2, while Hyper did not show any response under the same condition.¹⁰⁴

4.4. OxyFRET and PerFRET

Two novel FRET-based indicators, OxyFRET and PerFRET, were built to measure H₂O₂ generated by NADPH-dependent oxidase (NOX) in live cells.¹⁰⁵ Both indicators have a H₂O₂-sensitive domain inserted between Cerulean⁵⁴ and Venus.⁷⁴ The sensory domains in OxyFRET are N- and C- terminal cysteine-rich domains of Yap1 (nCRD and cCRD), while PerFRET contains Orp1 and Yap1 cCRD.¹⁰⁶ Upon oxidation by H₂O₂, the catalytic cysteine of Orp1 or nCRD forms a disulfide bridge with the cCRD of the Yap1,¹⁰⁷ altering the effective FRET distance between the CFP and YFP (Figure 1f). Interestingly, the F₅₃₅/F₄₈₀ emission ratios of OxyFRET and PerFRET change in opposite directions: oxidation of OxyFRET increases its F₅₃₅/F₄₈₀ emission ratio while the F₅₃₅/F₄₈₀ emission ratio of PerFRET decreases under the same condition.¹⁰⁵ The oxidation of these indicators is rapid and reversible; their ratiometric readouts are relatively insensitive to pH changes. However, these indicators lack complete in vitro characterizations because it has been challenging to purify them.¹⁰⁵

4.5. Peroxiredoxin-2 (Prx2)-based FRET indicator

The exceptionally active cysteine at the N-terminus of human Prx2 can form a disulfide bond with a second cysteine residue located on an adjacent Prx2 monomer, leading to a covalently linked dimer.¹⁰⁸ Prx2 reacts with H₂O₂ faster than other sensing domains such as Orp1 and OxyR, making it more kinetically favored.¹⁰⁸ Based on this mechanism, a fluorescent FRET indicator was developed for monitoring trace H₂O₂ fluctuations. Prx2 was placed between Clover,⁶⁹ a GFP, and mRuby2,⁶⁹ an RFP (Figure 1f). This FRET FP pair was chosen because it was reported to have a higher dynamic range than conventional CFP-YFP FRET pairs.⁶⁹ The resultant Prx2-based indicator is ultrasensitive to the femtomolar-level change of H₂O₂ concentrations.¹⁰⁸

5. GERIs for pyridine dinucleotides

Pyridine dinucleotides are essential for maintaining cellular redox homeostasis. In particular, NADPH provides the reducing power and neutralizes ROS through the glutathione and Trx antioxidant pathways. In addition to its role in protecting cells against oxidative stress, NADPH also serves as the substrate for NADPH oxidases to generate ROS. NADH is another cellular reducing molecule, although, in most organisms, NADH’s function is primarily metabolic. NADH is the primary electron donor for mitochondrial ATP synthesis, and ROS is generated in this process as by-products. NADH and NADPH are linked through their oxidation products, NAD⁺ and NADP⁺, since NAD⁺ kinase (NAK),¹⁰⁹ a well-known enzyme, can convert NAD⁺ into NADP⁺, which is further reduced into NADPH through the pentose phosphate pathway. MESH1 (Metazoan SpoT Homologue 1)¹¹⁰ was recently reported as a cytosolic NADPH phosphatase in mammalian cells, and MESH1 converts cellular NADPH into NADP to promote ferroptosis under stress conditions. Below we discuss a few examples of GERIs that can sense the NADPH and NADH redox couples (Table 3).

Table 3.

Representative GERIS for NADPH and the NADH redox couple.

GERIs	Targets	λex (nm)	λem (nm)	K d a	Signal	Sensory Domain	Recommended pH control
LigA-cpVenus111	NAD+	405/480	530	65 μM	cpVenus	DNA ligase	cpVenus
FiNad113	NAD+/(ATP+ADP) ratio	500	518	14 μM to 1.3 mM	cpYFP	T-Rex	mCherry-cpYFP
Frex114	NADH	421/500	518	3.7μM	cpYFP	B-Rex	cpYFP
FrexH114	NADH	421/500	518	40 nM	cpYFP	B-Rex	cpYFP
Peredox88	NADH/ NAD+ ratio	400	510	<5 nM for NADH	cpT-Sapphire	T-Rex	pH resistant
RexYFP116	NAD+/NADH ratio	490	516	180 nM for NADH	cpYFP	T-Rex	SypHer
Apollo-NADP+117	NADP+	485	535	0.1~20 μM	HomoFRET	G6PD	pH resistant
NADPSor121	NADP+	440	478/526	2 mM	FRET	KPR	pH resistant
SoNar44	NAD+/NADH ratio	420	485	5 μM for NAD+ 0.2 μM for NADH	cpYFP	T-Rex	cpYFP
iNaps125	NADPH	420/500	515	2-120 μM	cpYFP	T-Rex	iNapC

^aK_d: dissociation constant

5.1. LigA-cpVenus

LigA-cpVenus is an NAD⁺ indicator derived from a bacterial DNA ligase, which contains a bipartite NAD⁺-binding domain.^111,112 A cpVenus⁷⁴ was inserted between the two subdomains (Figure 1d). The apparent K_d of LigA-cpVenus to NAD⁺ is ~ 65 μM.¹¹¹ The binding of NAD⁺ results in a decrease of fluorescence in a dose-dependent manner with excitation at 488 nm (F₄₈₈), whereas the fluorescence with excitation at 405 nm (F₄₀₅) remains unchanged.¹¹¹ Thus, an excitation-ratiometric readout (F₄₈₈/F₄₀₅) can be used to normalize sensor expression levels. Tests against other nucleotides and NAD⁺ precursors show that the probe is specific to NAD⁺.¹¹¹ LigA-cpVenus is susceptible to pH change, but the pH influences can be corrected using cpVenus due to their similar pH profiles.¹¹¹

5.2. FiNad

FiNad is another GERI for NAD⁺. A cpYFP was inserted between F189 and L190 of the Thermus aquaticus Rex (T-Rex) monomer through short peptide linkers (Figure 1d).¹¹³ FiNad’s fluorescence increases ~ 7-fold after binding with NAD⁺. It also responds to NADPH, but the response range is far above the physiological concentration of cytosolic and nuclear NADPH.¹¹³ ATP or ADP affects the affinity of FiNad to NAD⁺, and the K_d of FiNad to NAD⁺ ranges from ~14 μM to 1.3 mM, depending on the concentrations of ATP or ADP. Thus, FiNad is actually an indicator for the NAD⁺/AXP ratio, in which AXP is the total pool of ATP and ADP.¹¹³ FiNad has been genetically fused to mCherry,⁷⁹ and the resultant mCherry-FiNad chimera allows a ratiometric readout of NAD⁺ dynamics. The green fluorescence of mCherry-FiNad excited at 485 nm is pH-sensitive, whereas the red fluorescence excited at 590 nm is relatively insensitive to pH changes. When using mCherry-FiNad in conditions with pH fluctuations, a mCherry-cpYFP fusion protein should be used in parallel for correction.¹¹³

5.3. Frex and FrexH

Frex and FrexH are GERIs for monitoring NADH dynamics. They were designed by inserting a cpYFP between a complete Bacillus subtilis Rex (B-Rex) monomer and B-Rex’s NADH-binding domain (Figure 1d).¹¹⁴ Both indicators displayed similar spectral properties as cpYFP with two excitation peaks at 421 and 500 nm and a single emission peak at 518 nm.¹¹⁴ Frex exhibits a 9-fold fluorescence increase upon binding with NADH, while FrexH displays a change in the opposite direction, ~ 3-fold fluorescence turn-off. In addition, both of them showed high selectivity toward NADH. The K_d of FrexH to NADH is ~ 40 nM, 10-fold lower than that of Frex (~ 3.7 μM).¹¹⁴ Due to their pH responses similar to cpYFP, the pH interference can be corrected by using cpYFP in parallel experiments.

5.4. Peredox

Peredox was constructed by integrating a pH-resistant GFP, T-Sapphire,⁸² between two subunits of a T-Rex tandem dimer. It is an indicator for NADH/NAD⁺ ratio because its affinity for NADH effectively decreases with increasing NAD⁺ concentration.⁸⁸ Due to its pH resistance, the fluorescence of Peredox is not affected by pH changes. However, the competitive binding with NAD⁺ or NADH varies substantially with pH. Peredox provides intensiometric readouts with a single excitation peak at 400 nm and an emission peak at 510 nm, but it has been fused with mCherry⁷⁹ or mCitrine,⁵⁵ for ratiometric reading.^88,115 One limitation of Peredox is its exceptionally high affinity to NADH, making it easily saturated in some cellular compartments.⁸⁸

5.5. RexYFP

RexYFP was made by integrating cpYFP into the loop of one subunit of T-Rex, between the nucleotide and DNA binding domain (Figure 1d).¹¹⁶ The molecular weight of RexYFP is 1.5- to 2-fold smaller than Frex and Peredox. NADH binding leads to a fluorescence decrease with 490 nm excitation and a fluorescence increase with 340 nm excitation.¹¹⁶ When tested against various NADH analogs, RexYFP responded to both NADH and NADPH. Thus, it is hard to exclude the NADPH interference when using RexYFP in living cells. Although NAD⁺ does not elicit fluorescence change of RexYFP, it can compete with NADH for binding to RexYFP.¹¹⁶ Due to the greater concentration of NAD⁺ than NADH in live cells, RexRFP should be considered as an indicator for the NAD⁺/NADH.¹¹⁶ Since RexYFP is pH-sensitive, SypHer,¹⁰⁰ a pH indicator with a cpYFP chromophore, was used to correct pH-related artifacts.¹¹⁶

5.6. Apollo-NADP⁺

Apollo-NADP⁺ is based on homo-FRET, which uses the same FP as both the donor and acceptor.¹¹⁷ Homo-FRET decreases steady-state fluorescence anisotropy.¹¹⁸ Glucose-6-phosphate dehydrogenase(G6PD) is an inactive monomer in the absence of NADP⁺. In the presence of high amounts of NADP⁺, it turns into an active homodimer.¹¹⁹ Taking advantage of this dimerization mechanism, Apollo-NADP⁺ was made by sandwiching a catalytically inactive human G6PD between two copies of Venus, a YFP.⁷⁴ NADP⁺ binds Apollo-NADP⁺ at an affinity similar to that of the wild-type G6PD, triggering dimerization and fluorescence anisotropy decrease.¹¹⁷ The probe does not respond to other metabolites, including NADPH, NAD⁺, NADH, and ATP. Apollo-NADP⁺ is a reversible, pH-resistant indicator compatible with many other indicators. The limitation of this indicator is its relatively small (15-20%) dynamic range.¹¹⁷

5.7. NADPsor

NADP⁺ can bind ketopantoate reductase (KPR) as a cofactor.¹²⁰ Hence, NADPSor was generated by inserting the binding element from KPR between a CFP-YFP FRET pair.¹²¹ The binding of NADP⁺ triggers a conformational change of KPR from a relatively closed form to an open state. This structural rearrangement distances the two fluorophores and lowers the FRET efficiency.¹²¹ Thus, the indicator displays an NADP⁺ concentration-dependent decrease in FRET, indicated by a lower fluorescence emission ratio (F₅₂₆/F₄₇₈) at 526 and 478 nm. NADPsor is specific to NADP⁺ with a detection limit of ~1 μM and does not respond to other NADP⁺-like metabolites.¹²¹

5.8. SoNar

SoNar, one of the most promising indicators for monitoring the live-cell NAD⁺/NADH ratio, was constructed based on a truncated T-Rex.⁴⁴ NADH-bound T-Rex exhibits an open conformation while NAD⁺-bound T-Rex shows a closed conformation. Integration of cpYFP into the surface loop of T-Rex couples the NADH/NAD⁺ binding-induced conformational change with cpYFP’s spectral characteristics, enabling SoNar to provide an excitation-ratiometric readout (Figure 1d).⁴⁴ SoNar has two excitation peaks at 420 and 485 nm (F₄₂₀/F₄₈₅). Binding to NADH increases the excitation ratios (F₄₂₀/F₄₈₅) while binding to NAD⁺ decreases the ratio, resulting in a maximal 15-fold fluorescence change in vitro.^44,122 The apparent K_d of SoNar for NAD⁺ and NADH is ~ 5 μM and ~ 0.2 μM, respectively, far below the intracellular NAD⁺ and NADH concentrations, making SoNar a good reporter for the NAD⁺/NADH ratio without being affected by the total NAD(H) pool. ^44,123,124 SoNar responds rapidly even to subtle differences in cellular NAD⁺/NADH.¹²² As a cpYFP-based indicator, the pH effect of SoNar can be corrected using a cpYFP in parallel.

5.9. iNap indicators

iNap indicators (iNaps) were NADPH sensors developed from SoNar. Mutations were introduced to convert T-Rex from an NADH binder to an NADPH binder.¹²⁵ The resultant iNap variants (named iNap1, iNap2, iNap3, and iNap4) have apparent NADPH affinities ranging from 2 μM to 120 μM. iNaps have two excitation peaks at ~420 nm and ~500 nm and one emission peak at 515 nm.¹²⁵ NADPH binding leads to a ~3.5-fold increase in fluorescence excitation at 420 nm and a ~2.5-fold decrease of fluorescence excitation at 485 nm, rendering a ~ 900% excitation-ratiometric change.¹²⁵ As iNaps can give ratiometric readouts, iNaps can be targeted to different cellular compartments to measure the NADPH levels. They have also been applied to measure the change of NADPH levels in cells under oxidative stress.¹²⁵ A pH control, iNapc, was created by introducing four mutations into the binding domain to abolish ligand-binding.¹²⁵

6. GERIs for molecular oxygen (O₂)

The d(UnaG-mOrange) fusion protein for hypoxia and reoxygenation (dUnOHR),¹²⁶ is an indicator consists of an oxygen-independent FP, UnaG,⁷⁷ and an oxygen-dependent orange FP, mOrange.⁷⁹ Under hypoxia, only the fluorescence of UnaG is detectable, while the fluorescence of mOrange is undetectable. Following reoxygenation, the mOrange chromophore matures, and its fluorescence becomes detectable while UnaG fluorescence decreases due to FRET from UnaG to mOrange.¹²⁶ FluBO¹²⁷ is another similar indicator based on FRET between YFP and the oxygen-independent flavin-binding fluorescent protein (FbFP).¹²⁸ The presence of oxygen leads to YFP chromophore maturation and efficient FRET to FbFP. In the absence of oxygen, the chromophore of YFP is not formed, and there is no FRET.¹²⁷ Both dUnOHR and FluBO are based on irreversible oxygen-dependent chromophore maturation, so their responses are irreversible and slow. Additionally, UnaG and FbFP require bilirubin and flavin mononucleotide as cofactors for fluorescence, leading to inaccurate outputs when these cofactors’ levels are limited.¹²⁹

ProCY, developed by our research group, is also a FRET-based indicator responsive to oxygen levels.¹³⁰ Hypoxia-inducible factor (HIF-1α) is a transcription factor regulated by prolyl hydroxylase domain enzymes (PHDs) in an oxygen-dependent manner. Under normal oxygen levels, PHDs can use oxygen to hydroxylate the conserved proline (Pro564) of HIF-1α. As a result, hydroxylated HIF-1α subsequently binds to von Hippel-Lindau (VHL), leading to subsequent polyubiquitination and proteasomal degradation of HIF-1α. Under hypoxia, HIF-1α is stabilized due to the reduced activity of PHDs.^131–133 ProCY was generated by inserting a 22-amino-acid proline-containing peptide (556-577) from HIF-1α and a 10-kDa binding domain of VHL between an ECFP and a yellow FP (YPet).¹³⁴ The HIF-1α derived peptide and the VHL domain were connected by a 21-aa GS rich linker. The interaction between HIF-1α peptide and the VHL domain under different oxygen levels would result in a conformational change that affects distance, orientations, and subsequently, the FRET efficiency between ECFP and YPet (Figure 1f).¹³⁰ ProCY shows a dynamic range of 168% with the FRET ratio change from 2.95 to 1.76 upon proline hydroxylation, making it a reliable tool for monitoring hypoxia.¹³⁰

7. GERIs for Methionine sulfoxide (MetO)

Methionine can be converted to the R- or S-diastereomers of methionine sulfoxide (MetO) by different oxidants when reacted with ROS.¹³⁵ Methionine sulfoxide reductases A (MSRA) and B (MSRB) are the specific enzymes of the S- and R-diastereomers of MetO.^136,137 Reduction of the MetO-containing substrate in MSRA/MSRB can lead to a stable disulfide bond between an MSR and a corresponding Trx in the fusion chimera and thus trigger conformational changes (Figure 1d). Based on this mechanism, two indicators, MetSOx and MetROx, were developed by sandwiching cpYFP between MSRA/MSRB and Trx, enabling the detection for S and R-forms of MetO, respectively.¹³⁸ Reduced recombinant MetSOx shows two excitation peaks at 425 and 505 nm and a single emission peak at ~ 515 nm. Upon reaction with MetO, there is a noticeable increase in fluorescence emission with excitation at 505 nm but not with excitation at 425 nm.¹³⁸ Similarly, reduced MetROx displays two excitation peaks at 410 and 500 nm and a single emission peak at ~ 515 nm. Reaction with MetO results in a turn-on fluorescence change with excitation at 410 nm and a turn-off change with excitation at 500 nm.¹³⁸

8. GERIs for Peroxynitrite (ONOO⁻)

Peroxynitrite (ONOO⁻) is a highly reactive RNS formed from a diffusion-controlled reaction of superoxide and nitric oxide.^139,140 Boronate-containing small molecules^141,142 have been developed as indicators for ONOO⁻, although they often cross-react with other ROS such as H₂O₂ and HOCl.^141,143 Our group used a genetic code expansion technology to introduce an ncAA, p-boronophenylalanine (pBoF), into the chromophores of cpGFPs, and further performed protein engineering on the scaffolds to modulate the reactivity of the introduced new functional group (Figure 1e). Using a circularly permutated sfGFP (cpsGFP),⁶³ we derived pnGFP, with a single excitation peak at 484 nm and a single emission peak at 508 nm.⁴⁶ The indicator switches from a nearly dark state to a fluorescent state after treatment of nanomolar level of ONOO⁻ while showing no visible response to other redox-active molecules at physiologically relevant concentrations.⁴⁶ By identifying an appropriate tRNA/aminoacyl-tRNA synthetase pair for mammalian cells, the probe was expressed in mammalian cells to image ONOO⁻ with minimal interference from other reductants or oxidants.⁴⁶ Recently, we reported a second-generation ONOO⁻ indicator, pnGFP-Ultra, engineered via structure-guided reactivity screening.⁴⁸ pnGFP-Ultra displays a remarkable ~110-fold fluorescence enhancement in response to ONOO⁻. It presents faster chromophore maturation, greater brightness and selectivity, and a larger dynamic range than pnGFP.⁴⁸ With an optimized ncAA incorporation system, the expression level of pnGFP-Ultra in HEK 293T cells has been drastically improved. pnGFP-Ultra has been demonstrated to be a robust tool for ONOO⁻ detection in activated macrophages and primary mouse glia.⁴⁸

In another recent study, the ncAA, thyronine (Thy), was introduced into the chromophores of sfGFP or cpsGFP.¹⁴⁴ The incorporation of Thy led to non-fluorescent proteins, the fluorescence of which could be activated by ONOO⁻ in a concentration- and time-dependent manners.¹⁴⁴ Despite the initial promise, these Thy-based indicators need further improvement for brightness, sensitivity, specificity, and detection limits.

9. GERIs for Hydrogen sulfide (H₂S)

Hydrogen sulfide (H₂S) is a weak reducing agent. Although it may directly neutralize highly reactive ROS and RNS, H₂S executes its antioxidant activity primarily through other indirect pathways. H₂S was shown to induce S-sulfhydration of specific cysteine residues in proteins,¹⁴⁵ upregulate reduced glutathione and Trx1,^146,147 enhance antioxidative enzymes such as superoxide dismutase (SOD),¹⁴⁸ promote the biosynthesis of antioxidative enzymes,¹⁴⁹ and inhibit mitochondrial free radical production.¹⁵⁰ To develop a selective GERI for H₂S, H₂S-reactive ncAA, p-azidophenylalanine (pAzF) was introduced into the chromophores of a cpGFP (Figure 1d). The full-length protein, named cpGFP-Tyr66pAzF, was generated in live cells by co-expressing corresponding tRNA/aminoacyl-tRNA synthetase pairs.⁴⁵ Treatment with H₂S resulted in specific and rapid fluorescence enhancement.⁴⁵ A significantly enhanced version of the H₂S indicator, hsGFP, was later developed based on cpsGFP.⁴⁷ The chromophore-forming Tyr in cpsGFP was replaced with pAzF, followed by optimization of N- and C- terminal residues close to the chromophore. The addition of H₂S to hsGFP decreases the absorbance peak at 391 nm with an accompanying increase at 452 nm.⁴⁷ hsGFP is superior to cpGFP-pAzF in many aspects. Due to the superfolder mutations in hsGFP, the hsGFP indicator shows much-improved folding and maturation. Additionally, the product of hsGFP upon reaction with H₂S is ~ 9-fold brighter than the product of cpGFP-Tyr66pAzF. After incubation with 100 μM H₂S, the fluorescence of hsGFP increases by ~ 5.5-fold, whereas the fluorescence change of cpGFP-Tyr66pAzF is only ~ 0.6-fold.⁴⁷ It was also demonstrated that hsGFP could be readily targeted to various subcellular domains to detect H₂S effectively.⁴⁷ Because hsGFP is an intensiometric indicator, its intensity is sensitive to protein expression levels and instrumental setups. To create a ratiometric H₂S indicator, hsGFP was fused to EBFP2.⁷² In the resultant hsFRET indicator, the reaction product of hsGFP and H₂S acts as a FRET acceptor to EBFP2. Upon reaction, the green-to-blue emission ratio of hsFRET increases from 0.62 to 1.73.¹⁵¹

10. GERIs for organic hydroperoxides (OHPs).

Organic hydroperoxides (OHPs) induce more sustained cellular stress than H₂O_2.¹⁵² While a number GERIs for H₂O₂ have been developed, to our knowledge, OHser (Organic Hydroperoxide Sensor) remains the only indicator for OHPs.¹⁵² OHser was generated by inserting cpVenus,⁷⁴ a circularly permuted form of a YFP variant, into Xc-OhrR,¹⁵³ a transcriptional regulator controlling OHP detoxification in the soil bacterium Xanthomonas campestris (Xc).^152,153 Upon oxidation by OHPs, two intersubunit disulfide bonds are formed, breaking the α5 helix in OhrR into two distinct helices connected by a loop. This helix-to-loop conversion results in a significant conformational change, leading to the fluorescence change of cpVenus (Figure 1c).¹⁵² OHser displays a 2-fold response to cumene hydroperoxide (CHP) and tert-butyl hydroperoxide, but almost no response to H₂O₂. It is sensitive to pH changes, so cpVenus should be used in parallel as a control.¹⁵²

11. GERIs for Nitric Oxide (NO)

Nitric oxide (NO) is a crucial gasotransmitter. In addition to its well-known role in inactivating cyclic guanosine monophosphate (cGMP)-dependent signaling pathways,¹⁵⁴ NO can execute its biological functions through ONOO⁻, which is made from a diffusion-controlled combination of NO and superoxide, or through S-nitrosylation (-SNO), a crucial post-translational modification of sulfhydryl residues in proteins.¹⁵⁵ An amplifier-coupled fluorescent indicator for NO, namely NOA-1, was created by individually fusing the α- and β-subunits of the NO receptor protein sGC with a genetically encoded cGMP indicator named CGY.^156,157 Upon binding with NO, the spontaneously formed sGCα-CGY/sGCβ-CGY heterodimer can generate cGMP from GTP at a ~ 400-fold faster speed. The generated cGMP binds to CGY to cause a fluorescence change, showing as a ratiometric readout.^156,157 Another family of NO indicators is geNOps.¹⁵⁸ A bacteria-derived NO-binding domain, GAF,¹⁵⁹ was fused with various FPs to generate chimeras. The addition of NOC-7, a potent NO donor,¹⁶⁰ quickly reduces the fluorescence of geNOps by ~7-18%. There is a linear correlation between the NO concentration and NO-induced fluorescence quenching, and the response is fully reversible.¹⁵⁸ However, geNOps require high levels of Fe²⁺ for its expression, and it is often needed to soak geNOp-expressing cells with a Fe²⁺ booster solution before imaging. This is an obvious limitation of these geNOp indicators.¹⁵⁸

Selected applications of GERIs

Redox-active molecules, such as Trx, Grx, and H₂O₂, are essential for normal cell functions. For example, EGF promotes RTK autophosphorylation and activation, which further triggers cell proliferation and differentiation through the ERK kinase pathway.¹⁶¹ NOX, activated by EGF, induces H₂O₂ production to deactivate tyrosine phosphatases through cysteine sulfenylation (Cys-SOH), thereby augmenting the EGF signal transduction through RTKs.³ Furthermore, H₂O₂ has been shown to directly enhance the kinase activity of EGF receptor (EGFR) by sulfenylation of EGFR’s catalytic cysteine.¹⁶² Production of H₂O₂ by NOX is, in fact, essential for ERK activation and cell spread.¹⁶³ On the other hand, the GSH and Trx antioxidant systems can reduce sulfenic cysteines back to thiolates, terminating EGF signaling.^3,6 Combining TrxRFP with Grx-roGFP2 reveals that EGF indeed increases the oxidation of both Trx and GSH/GSSG.³⁶

Dysregulation of redox signaling can cause oxidative stress, which shifts cells towards disease states.¹⁶⁴ Antioxidants have been explored to reduce oxidative stress and boost antioxidant defense, but dietary antioxidants have failed to yield consistent and satisfactory clinical outcomes.¹⁶⁵ Some studies have shown that certain nutritional antioxidants can even increase cancer progression by downregulating the expression of antioxidant genes.¹⁶⁶ Also, unsuccessful treatment of neurodegenerative diseases with generic antioxidants indicates that some levels of oxidation may be needed for normal signaling and neurogenesis.^167,168 Because redox signaling is tissue-dependent and spatially regulated, generic antioxidants may not be effective for disease treatment. Meanwhile, a recent study has shown that an organelle-specific NOX inhibitor can substantially reduce airway inflammation and slow down viral replication.^165,169 Therefore, the spatial information of redox signaling in diseases would be crucial for developing effective antioxidant treatments. GERIs provide a convenient means to access the spatiotemporal information of redox signaling, thus offering opportunities to better understand diseases and develop new treatments (Fig. 2). Here, we discuss several examples, which applied GERIs to studies related to oxidative signaling and diseases.

Fig.2. Acute and chronic oxidative stress induced by various factors can lead to a large variety of diseases.

Environmental exposure to toxins, drugs, and aging can result in elevated oxidative stress. Unregulated oxidative stress can cause cellular damage through DNA, protein, lipid damages, which exacerbate oxidative stress and contribute to the initiation and progression of diseases.

1. Using GERIs to study ROS signaling

The number of available GERIs was relatively limited in the first decade of this century. An early study combined roGFP1 with SOD expression and showed that the H₂O₂ level was elevated during cancer progression.¹⁷⁰ Another study combined roGFP1 with small molecule dyes, revealing that mitochondrial inhibition in neurons generates superoxide, H₂O₂, and free radicals and causes neuronal oxidation.¹⁷¹

After the H₂O₂-specific sensor HyPer was developed, the H₂O₂ level can be closely monitored. In one example, HyPer was used to evaluate the roles of H₂O₂ in C. elegans development.¹⁷² The HyPer biosensor revealed that endogenous H₂O₂ levels were elevated during the early larval stage but rapidly decreased in the reproductive and fertile phases.¹⁷² Moreover, the sensor provided an adequate spatial resolution to show the strongest fluorescence signal in the head region.¹⁷² Although the underlying mechanism needs further studies, the available results strongly suggest that the accumulation of H₂O₂ is tissue-dependent.

Similarly, a transgenic Xenopus line expressing HyPer also exhibited a fluorescence increase during fertile phase.¹⁷³ HyPer combined with a red genetically encoded fluorescent calcium biosensor, R-GECO,⁶⁷ revealed that peroxide production correlates with the Ca²⁺ influx in mitochondria after fertilization.¹⁷³ Inhibition of the mitochondrial calcium uniporter (MCU) diminishes the production of peroxide, and the peroxide production quickly recovers after the removal of the MCU inhibitor.¹⁷³ Further studies showed that the cell cycle in the embryos’ early cleavage stages was regulated through mitochondrial ROS oscillation. Inhibition of mitochondrial ROS production results in cell-cycle arrest, and the Cdc25C phosphatase may be a key regulator during the process.¹⁷³

The role of redox signaling in regenerative processes has also been increasingly studied. HyPer was used to map ROS production during vertebrate tail regeneration in Xenopus tadpoles. Several X. laevis transgenic lines were developed to express HyPer ubiquitously. After tail amputation, a remarkable increase in intracellular H₂O₂ was detected using HyPer, and such high intracellular H₂O₂ concentrations are maintained throughout the entire tail regeneration process.¹⁷⁴

HyPer has also been used to visualize H₂O₂ production caused by photodynamic therapy.¹⁷⁵ Interestingly, in drug-resistant cancer cells, ROS levels are often low due to the upregulation of antioxidant defenses such as GSH, demonstrating the vital role of redox molecules in cancer progression and treatment.¹⁷⁶ H₂O₂ is a trigger for the expression of pro-metastatic and proangiogenic factors. Using HyPer, Satoka and Hara-Chikuma show that aquaporin-3 (AQP3) imports the extracellular peroxide produced by NOX2, and the increase of intracellular peroxide favors cancer cell migration.^177,178

GERIs are valuable tools for revealing redox transients in subcellular compartments. roGFP2 has been targeted to the cytosol or mitochondria to assess the oxidative stress in the murine lung.¹⁷⁹ The sensor showed that lung hyperoxia, which refers to the state of having excess oxygen, causes over-oxidation of the mitochondria but not the cytosol in neonatal alveolar cells. Oxidative stress subsequently induces NOX expression and cell death.¹⁷⁹ While hyperoxia can cause cytotoxicity in the lung, some levels of oxidation are essential for regulating its normal function. By using roGFP2, Desireddi et al. were able to show that oxidation induced by hypoxia, which refers to the state of having inadequate oxygen, triggers Ca²⁺ import to promote pulmonary vasoconstriction.¹⁸⁰

GERIs have been applied to studying redox signaling in neurodegenerative diseases, such as Alzheimer’s disease (AD). Amyloid β (Aβ) aggregation is associated with AD. To investigate the onset of Aβ neurotoxicity, the researchers expressed cytosolic roGFP2-Orp1 or Grx1-roGFP2 in the Aβ fruit fly models to monitor the H₂O₂ levels and the redox potential of GSH (E_GSH) in parallel.¹⁸¹ Increased cytosolic E_GSH was observed in neurons upon onset of Aβ aggregation, suggesting that E_GSH is the driving force for the Aβ-mediated neurotoxicity in AD.¹⁸¹ More specifically, this redox change was only observed in neurons but not in glial cells, implying that neurons are susceptible to Aβ-regulated redox stress.¹⁸¹ The study performed in Drosophila Aβ aggregation models suggested that glutathione redox potential plays a pivotal role in AD. In another study, roGFP2 was targeted to mitochondria in neurodegenerative or mitochondrial encephalomyopathy Drosophila models with ATPase deficiency or electron transport chain (ETC) deficiency. roGFP2 revealed an elevated oxidation state at a later age in diseased fly models with reduced lifespan and increased degeneration.¹⁸² Moreover, mitochondria-targeted antioxidants improved locomotor function in early pathogenesis and extended the fly longevity, indicating the involvement of oxidative stress in the disease progression.¹⁸²

Grx1-roGFP2 has been used to study transgenic amyotrophic lateral sclerosis (ALS) and spinal cord injury in mice models. In physiological conditions, mitochondrial contraction, a process in neurons that protects mitochondria from acute stress, is accompanied by reversible redox changes.¹⁸³ Grx1-roGFP2 revealed that mitochondrial contraction is preceded by transient mild mitochondrial oxidation caused by stressors.¹⁸³ Next, the membrane depolarization through potassium (K⁺) or sodium ion (Na⁺) fluxes further reduces membrane potential, consequently increasing the mitochondrial matrix’s pH and accelerating the ETC.¹⁸³ The ETC generates ROS, which oxidizes the mitochondria to activate the uncoupling proteins (UCPs), dissipating the proton gradient to protect mitochondria from prolonged damage.¹⁸³ In the ALS and spinal cord injury models, mitochondria in neurons do not experience contraction. Thus, mitochondrial oxidation in these neurons becomes persistent, causing permanent mitochondrial permeability.¹⁸³

roGFP1 has also been expressed in cultured rat hippocampal neurons or organotypic slices for monitoring the alteration of neuronal redox states during mitochondrial inhibition and neuronal stimulation. Neuronal stimulation by K⁺ consistently increases the oxidation of neurons, while neuronal stimulation by glutamate can increase or decrease oxidation depending on the glutamate levels.¹⁷¹ In another study, the cytosolic or mitochondria-localized roGFP1 was stably expressed in transgenic mice under the Thy1 promoter to ensure cell-type-specific expression in excitatory neurons.¹⁸⁴ Neuron activation by either electrical stimulation or K⁺-mediated depolarization altered the neuronal redox states indicated by roGFP1.¹⁸⁴ The inhibition of the mitochondrial respiratory chain in acute brain slices shifts roGFP1 to a more oxidized state.¹⁸⁴ In these examples, roGFP1 provided a robust ratiometric readout for neuronal oxidation state unaffected by pH changes.¹⁷¹

GERIs have been utilized to investigate the role of redox in cardiac dysfunction. For example, D-amino acid oxidase (DAAO), a protein that can produce H₂O₂ from the oxidation of D-amino acids, was used as a chemogenetic tool to induce H₂O₂ production. At the same time, HyPer was fused with DAAO to monitor H₂O₂ production.¹⁸⁵ Using the cardiac troponin T (cTnT) promoter, the fusion protein, HyPer-DAAO, was specifically targeted to the rat heart to induce H₂O₂ production. HyPer revealed that DAAO stably expressed in the rat heart induced H₂O₂ production when supplementing D-alanine .¹⁸⁵ Rats with H₂O₂ generation showed cardiac dysfunction such as systolic dysfunction (reduced heart contraction) and dilated cardiomyopathy (weakened and enlarged heart muscle) without any signs of fibrotic remodeling.¹⁸⁵ Although overexpression of redox-active enzymes was observed in rat hearts, GSH concentration was lowered while the overall glutathione (GSH/GSSG) concentration was elevated, indicating the presence of oxidative stress in the heart.¹⁸⁵

2. Using GERIs to decipher drug-induced oxidative toxicity

GERIs have been applied in high-throughput drug screening to evaluate the oxidative toxicity of various drugs and drug candidates. SoNar, an indicator for the ratio of NAD⁺/NADH, was used to screen > 5500 metabolism-perturbing anticancer drug candidates for cytotoxicity.⁴⁴ In the study, SoNar was stably expressed in a human lung cancer cell line, H1299, and the effect of the drug candidates on the cytosolic NAD⁺/NADH redox states was then assessed. Interestingly, most compounds decreasing the cellular NAD⁺/NADH ratio showed no or minimal impact on cell toxicity. In contrast, a significant fraction of the compounds that increased the ratio presented apparent cell toxicity. A protein kinase B (Akt) inhibitor KP372-1, which increased the NAD⁺/NADH ratio, exhibited the most potent cell cytotoxicity at the nanomolar concentration range. Furthermore, a collective use of SoNar, HyPer, and roGFP revealed that KP372-1 is an NAD(P)H dehydrogenase quinone 1 regulator, which mediates the NAD(P)H-dependent redox cycling.^44,186 This compound thus kills cancer cells by generating ROS. Using roUnaG, a later study further confirmed this finding that KP372-1 exerts its anticancer function by creating severe oxidative stress.¹⁸⁷

In another study, stable cancer cell lines expressing mitochondrial-localized roGFP2(mt-roGFP) and nuclear H2B-mCherry were utilized as tools for high-throughput drug screening.¹⁸⁸ Mt-roGFP was explicitly used to screen for drugs capable of perturbing mitochondrial redox because mitochondrial oxidation is a marker for cytotoxicity. H2B-mCherry was used as a control to mark nuclear structures. A library of cancer drugs was successfully screened through high-throughput imaging and segmentation of pre- or post-drug-treated cells.¹⁸⁸ The study demonstrated that cisplatin, a chemotherapy medication, can cause mitochondrial damage and cytotoxicity. Moreover, a later study used HyPer2 to further demonstrate that cisplatin could generate H₂O₂ in a time- and dose-dependent manner in cancer cells.¹⁸⁹ Thus, GERIs have provided a promising way to visualize drug-induced toxicity due to their ease of use, adaptability for different cell types, and compatibility with high-throughput screening.

3. Using GERIs to decipher age-induced oxidative toxicity

Aging increases the risk for many diseases as well as morbidity and mortality from diseases.¹⁹⁰ ROS accumulation and the depletion of other endogenous redox-active molecules such as NAD⁺ have been associated with aging.^191,192 For example, mCherry-FiNad showed a decline of NAD⁺ level as mice age.¹¹³ Model animals with compromised antioxidant defense systems often exhibit features of premature aging such as mitochondrial dysfunction and DNA damage.¹⁹³

Using Grx1-roGFP2 and roGFP2-Orp1, Albrecht et al. found that the aging-dependent oxidation in Drosophila was restricted to particular regions (e.g., midgut enterocytes), and the redox states of H₂O₂ and Grx were not coupled.¹⁹⁴ Interestingly, higher levels of H₂O₂ were observed in Drosophila with longer lifespans.¹⁹⁴ In another study using C. elegans, the H₂O₂ level, measured by HyPer, increases as aging occurs.¹⁹⁵ A restricted diet could artificially extend the lifespan of C. elegans and delay aging-dependent oxidation.¹⁹⁵ A recent study has further elucidated the relationship between cellular redox states and C. elegans aging using Grx1-roGFP2. An elevated oxidation level in the early developmental state is associated with a longer lifespan.¹⁹⁶ The study showed that redox signaling could lower the trimethylation of lysine 4 in histone H3 (H3K4me3), resulting in increased stress resistance and more reduced cellular states at a later timepoint.¹⁹⁶

Similarly, experiments in yeasts have shown aging-dependent oxidation. Both the H₂O₂ level measured by roGFP2-Orp1 and the redox potential of the cytosolic GSH/GSSG pool measured by Grx1-roGFP2 increase as yeasts age in aerobic conditions.¹⁹⁷ However, although the H₂O₂ level is still associated with the replicative lifespan of yeasts under anaerobic conditions, the cytoplasmic GSH/GSSG redox system does not seem to play a role in lifespan.¹⁹⁷

4. Using GERIs to decipher environmental exposure-induced oxidative toxicity

It is well recognized that the environment is a critical contributing factor to human health. Exposure to toxic environmental substances, such as heavy metals, pollutants, and chemicals, can cause oxidative stress, skewing the cellular redox balance and elevating ROS levels. For example, iron, copper, and chromium can generate free radicals through Fenton chemistry, while cadmium can cause oxidative stress by inhibiting redox enzymes and depleting glutathione.¹⁹⁸ Subsequently, when the level of ROS exceeds the antioxidant capacity, it can lead to tissue and DNA damage.¹⁹⁹ Cellular and DNA damage further increases ROS levels through inflammatory response, leading to cytotoxicity and even cancer.²⁰⁰

roGFP2 has been applied in E. coli to study the level of oxidative stress resulted from H₂O₂, menadione, sodium selenite, zinc pyrithione, and other heavy metals.²⁰¹ Interestingly, the heavy metals have resulted in oxidation increases smaller than other chemical stressors in the bacterial system.²⁰¹ In another study, a redox biosensor created by immobilizing roGFP2-expressing E. coli in a k-carrageenan matrix was applied to study metalloid-induced oxidative stress.²⁰² Oxidative stress, indicated by a rapid excitation increase of roGFP2 at 400 nm and an excitation decrease at 490 nm, was observed after the exposure to metalloids.²⁰² Arsenite and selenite are two representative metalloid pollutants in the aquatic environment, and they can enter the human body after bioaccumulation. Both can cause cell cytotoxicity by inducing oxidative stress.²⁰² This immobilized redox biosensor showed concentration-dependent fluorescence change in response to both arsenite and selenite metalloids. However, different limits of detection (LODs) were observed, which may be explained by different cytotoxicity levels of these two types of metalloids.²⁰²

Conclusions and outlook

ROS are typical examples of “the dose makes the poison” – a toxicological phenomenon discovered by Paracelsus in 1538.²⁰³ When ROS are generated at low levels, they carry out crucial signaling functions. However, when exceeding the physiological level, ROS can cause toxicity and even irreversible damages. ROS concentrations largely contribute to the level of oxidative toxicity, but the spatiotemporal regulation of various redox molecules is now found to play an equally important role. This review article summarized a selected panel of GERIs, which form a toolkit for studying redox signaling dynamics in various biological systems. When applying these indicators in studies, it is necessary to consider different parameters, including the probe sensitivity, specificity, detection limit, and dynamic range. In addition, artifact-inducing factors, such as pH, temperature, and interactions with interfering ions, should also be excluded using appropriate controls. Moreover, it should be taken into account that overexpression of GERIs in cells may result in buffering effects, limiting the magnitude of redox changes in response to stimuli. In many cases, the selection of the most suitable GERI depends on the particular research question, but as a general rule, GERIs with appropriate midpoint redox potentials should be selected to ensure their proper functions. Sometimes, a head-by-head comparison may be needed to identify the best GERI for a particular application.

While GERIs specific for Grx, Trx, and H₂O₂ have been developed, indicators specific for other redox-active molecules are still in high demand. ROS such as superoxide and hydroxyl radicals exhibit unique toxicity, but the detection of these radicals is difficult due to their short lifetime. Having probes specific for them would be advantageous for understanding ROS-induced toxicity as these radicals can directly damage DNA, react with protein, or modify lipid functions. Other redox enzymes such as Ero1 (ER oxidoreductase 1) and PDI are critical for oxidative protein folding,²⁰⁴ unfolded protein response (UPR),²⁰⁵ and cell apoptosis. GERIs specific to these redox enzymes are currently lacking but needed.

In general, FPs with higher brightness and more robust folding and chromophore maturation should be more amenable to redox-sensitive modifications and sensor engineering. Also, indicators derived from FPs with better photophysical properties are more likely to inherit these favorable features. With the appearance of more and more FPs possessing improved properties, it would be desirable to develop GERIs based on these new FPs. For example, the development of mScarlet, a red FP with better maturation, brightness, and photostability, may lead to the generation of better, red fluorescent redox indicators.²⁰⁶ GERIs with new colors, especially those with longer excitation and emission wavelengths, would benefit redox studies in living organisms. First, illuminating cells with longer wavelengths can lower cytotoxicity. Second, the ‘optical window’ of tissues is also in the far-red and near-infrared spectral range in which efficient endogenous absorbers are lacking.²⁰⁷ GERIs based on far-red and near-infrared FPs will likely become available, resulting in better tissue and animal imaging tools. Moreover, the development of GERIs in various colors can provide multi-color imaging tools to unravel the interplay and crosstalk between different signaling pathways.

GERIs in imaging modalities other than fluorescence would also be valuable tools. Recent studies have generated a panel of luciferase-luciferin pairs independent of the ATP cofactor,^208-213 resulting in a platform for further developments of ATP-independent, bioluminescent GERIs. These efforts will significantly facilitate our understanding of redox processes under opaque tissue because bioluminescence imaging can reach deeper than fluorescence imaging.

The advances in GERIs will continue to accelerate our understanding of redox signaling, which can, in turn, lead to the development of new indicators. The application of these new tools would further expand our knowledge. As more is known, intervention strategies may be developed to restore cellular redox balance and treat diseases.

There are continuous needs for advanced imaging techniques and/or indicators with better photostability, brightness, and dynamic range. Currently, wide-field or confocal microscopy is often used for imaging redox signals in living systems. Super-resolution imaging could be an exciting new tool for investigating redox signaling with better spatial resolution. In one example, super-solution imaging of cellular redox was demonstrated using HyPer2 and stimulated emission depletion(STED) microscopy.²¹⁴ However, super-resolution methods such as STED and photo-activated localization microscopy (PALM or STORM) require high light doses, resulting in substantial photobleaching and cytotoxicity.²¹⁵ Other super-resolution methods, such as structured illumination microscopy (SIM)²¹⁶ and super-resolution radial fluctuations (SRRF)²¹⁷, cause less photobleaching and may be promising for imaging various GERIs. With advances in both microscopy and indicators, the spatiotemporal regulation of redox signaling and oxidative toxicity can be better understood in the future.

Biographies

Biographies

Yu Pang is currently a graduate student in the Department of Chemistry at the University of Virginia (UVa). Before joining UVa in 2018, she obtained her B.S. in Pharmacy from China Pharmaceutical University in 2013 and M.S. degree in Pharmaceutical Science from Temple University in 2016. Her current research interest is developing genetically encoded, red and far-red fluorescent biosensors to study redox signaling in live cells and animals.

Hao Zhang is currently a graduate student in the Department of Chemistry at UVa. Hao Zhang received his B.S. in Biology and Chemistry from Washington College in 2017, and next, attended Dartmouth College as a graduate student in Chemistry. In 2019, he moved to UVa to continue his Ph.D. studies. His research interests focus on developing red fluorescent probes for redox signaling and kinase sensors for understanding insulin signaling and insulin resistance.

Huiwang Ai received his B.S. from Tsinghua University, Ph.D. from the University of Alberta with Robert E. Campbell, and postdoctoral training at Scripps Research with Peter G. Schultz. He became an assistant professor at the University of California, Riverside in 2011. In 2017, he joined UVa as a tenured associate professor of Molecular Physiology and Biological Physics, and Chemistry. His lab uses interdisciplinary approaches, such as protein engineering, fluorescence and bioluminescence imaging, and chemical biology, to study complex biological systems, focusing on the development of novel molecular biosensors to peer into cells and organisms to understand biosignaling and communications.