Encyclopedia
Please note this is a comparison between Version 1 by Stefano Bettati and Version 2 by Lindsay Dong.
The intracellular concentrations of oxygen and reactive oxygen species (ROS) in living cells represent critical information for investigating physiological and pathological conditions. Real-time measurement often relies on genetically encoded proteins that are responsive to fluctuations in either oxygen or ROS concentrations. The direct binding or chemical reactions that occur in their presence either directly alter the fluorescence properties of the binding protein or alter the fluorescence properties of fusion partners, mostly consisting of variants of the green fluorescent protein. Oxygen sensing takes advantage of several mechanisms, including (i) the oxygen-dependent hydroxylation of a domain of the hypoxia-inducible factor-1, which, in turn, promotes its cellular degradation along with fluorescent fusion partners; (ii) the naturally oxygen-dependent maturation of the fluorophore of green fluorescent protein variants; and (iii) direct oxygen binding by proteins, including heme proteins, expressed in fusion with fluorescent partners, resulting in changes in fluorescence due to conformational alterations or fluorescence resonance energy transfer.
- oxygen sensor
- reactive oxygen species
- fluorescent proteins
1. Relevance of O2 and ROS Sensing
The evolution of photosynthesis and the consequent oxygenation of the Earth’s atmosphere about 2.4 billion years ago is one of the most relevant transitions in the history of Life. A minority of organisms have since evolved in anoxic environments. Others have evolved in contact with O2, developing biochemical mechanisms and pathways to manage its toxicity and use it as the final acceptor of electrons in oxidative metabolism and as a substrate in several other reactions [1]. In these organisms, the availability of O2 is a key factor in their survival, and hypoxia, which hinders O2-dependent metabolic reactions, is involved in disease. In humans, hypoxia has been correlated with multiple sclerosis, cancer, heart disease, kidney disease, liver disease, lung disease, and inflammatory bowel disease [2]. To accurately measure O2 concentration inside cells and gain insights into these physiological and pathological conditions, O2 imaging is becoming increasingly important in cell biology.
The great advantage that O2 has afforded to Life is counterbalanced by its toxicity. Indeed, O2 can undergo reactions that produce reactive O2 species (ROS), including hydrogen peroxide (H2O2), superoxide (O2−), singlet O2 (1O2), and hydroxyl radical (●OH). The generation of ROS is a double-edged sword. On the one hand, ROS mediate fundamental processes such as cellular signaling and immune response. On the other hand, their high reactivity can damage biological macromolecules. Therefore, monitoring ROS production in vivo allows the study of physiological reactions as well as cellular oxidative stress and the progression of diseases. As for O2, genetically encoded probes are the tool of choice. Due to their reactivity and the action of specialized enzymes such as superoxide dismutase, the half-life times of ROS are very short, making their direct detection challenging; therefore, the existing genetically encoded fluorescent ROS biosensors rely on the monitoring of H2O2 or the ROS effect on other redox-sensitive molecules, such as glutathione, thioredoxins (Trx), methionine, nicotinamide adenine dinucleotide, and its phosphate form. In particular, scholarswe will focus the discussion on those sensors able to directly measure the cellular H2O2 fluctuations based on the variation of the reduced/oxidized glutathione (GSH/GSSG) or Trx redox couples.
Genetically encoded tools employed for the detection of O2 and ROS are typically based on fluorescence. Fluorescence has several advantages over other spectroscopic and microscopic techniques in biomolecular and cellular studies. This holds in terms of signal specificity, since fluorophores are excited and emit light at characteristic wavelengths, and background signal rejection, especially when two- or multi-photon excitation is exploited. Moreover, fluorescence is endowed with high sensitivity to physicochemical conditions of the microenvironment, since emission intensity and the Stokes shift markedly depend on solvent polarity, temperature and possibly pH, redox state and the presence of quenchers, and to structural dynamics.
2. Biosensors for O2
2.1. O
2
Sensing Mediated by Prolyl Hydroxylases
The transcription factor hypoxia-inducible factor-1 (HIF-1) is the main actor in modulating O2-mediated gene expression levels in all Parahoxozoa. HIF-1 is a αβ heterodimer with the levels of the β subunit being O2-independent, whereas those of the α subunit are O2-dependent [3][10] through its O2-dependent proteolysis [4][11]. In fact, the α subunit is constitutively synthesized but rapidly degraded under normoxic conditions [5][12] as a consequence of enzymatic hydroxylation of conserved proline residues (Pro402 and Pro564) located in its oxygen-dependent degradation domain (ODD) [6][7][13,14]. The enzymes responsible for proteolysis are 2-oxoglutarate-dependent dioxygenases containing a prolyl hydroxylase domain (PHD) [8][15]. Hydroxylated HIF-1α binds to the von Hippel–Lindau tumor suppressor protein (VHL) and is then ubiquitylated by the VHL-E3 ligase complex (Figure 1a).
Figure 1. Structures of O2 sensing proteins. (a) Crystal structure of a hydroxylated HIF-1 α peptide (in blue)—with the hydroxylated proline represented in yellow sticks—bound to the human pVHL/elongin-C/elongin-B complex (in pink) (PDB ID 1lqb) [9][17]. (b) Crystal structure of the Escherichia coli sensor heme domain (Ec DosH). The oxy heme is represented in sticks (PDB ID 1s66) [10][18].
A protein homologous to HIF-1α from Drosophila melanogaster, called Sima (bHLH/PAS domain transcription factor Similar), follows a fate similar to HIF-1α: Fatiga, a prolyl hydrolase, hydroxylates Pro-850 [11][19] on the Sima O2-dependent domain. Under hypoxic conditions, lower O2 availability limits proline hydroxylation. Non-hydroxylated Sima accumulates, translocates into the nucleus, and heterodimerizes with HIF β (Tango) for binding to a hypoxic response element (HRE) (Figure 2a) [12][20].
Figure 2. Schematic representation of selected O2 sensors. (a) Mechanism of O2 sensing by Sima. (b) Fusion protein between Sima ODD domain and FPs. (c) Mechanism of O2 sensing by the ProCY fusion protein. (d) Mechanism of O2 sensing by maturation-dependent FbFP. The hypoxia-tolerant fluorescent protein UnaG fused with: (e) CyOFP1 FP for the development of ratiometric reporter; (f) mOrange2 FP for the development of FRET-FLIM reporter; and (g) mCherry FP for the development of ratiometric reporter. The UnaG bilirubin cofactor is represented in sticks. (h) O2 sensing by ANA-Y biosensor; DosH heme cofactor is represented in sticks. (i) O2 sensor based on H-NOX; the porphyrin cofactor is represented in sticks. (l) mCherry fused with myoglobin as an O2 sensor; myoglobin heme cofactor is represented in sticks. (m) Mechanism of tyrosinase copper protein conjugated with a fluorescent dye to sense O2. (n) Mechanism of hemocyanin copper protein conjugated with a fluorescent dye to sense O2.
The pathway of HIF-1α and Sima has been considered as a suitable sensor for O2. For example, a biosensor based on Sima is composed of a fusion protein including its ODD (692–863 residues) and the green fluorescent protein (GFP) (Figure 2b) [13][21]. The fusion construct (GFP-ODD) is constitutively expressed under the control of the ubiquitin-69E (ubi) promoter.
Another genetically encoded biosensor sensitive towards proline hydroxylation by PHD, called ProCY, has been developed for measuring O2 in vitro and in live mammalian cells. The protein is formed of a 22-amino acid peptide from HIF-1α (residues 556–577, containing Pro564) and a small 10-kDa protein domain derived from VHL (residues 60–154) sandwiched between two FPs—the enhanced cyan FP (ECFP) and a yellow FP YPet—that are a highly optimized FRET couple. Pro564 hydroxylation promotes the interaction between the peptide and the VHL domain, thus causing a conformational change that alters the distance between ECPFP and YPet, giving a measurable change in the FRET signal (Figure 2c) [14][23].
2.2. O
2
Sensors Based on O
2
-Dependent Maturation of GFP Variants
The plethora of FP variants that have been devised over the years to tailor spectroscopic, photochemical, and photophysical properties achieve fluorescence through an O2-dependent “maturation process” (Figure 2d–g) [15][24]. The O2-dependence of the chromophore maturation and fluorescence is sometimes considered a limit to the applicability of FPs as reporter molecules, therefore new classes of O2-independent fluorescent reporters have been developed, based, e.g., on photosensory flavoproteins [16][17][18][25,26,27]. On the other hand, the fact that fluorescence is acquired only after spontaneous auto-oxidation and cyclization of the chromogenic tripeptide highlights the possibility of using it for intracellular O2 sensing, e.g., by exploiting GFP differential expression under the control of O2-responsive E. coli promoters [19][20][28,29], or by devising fusion proteins containing two fluorescent domains able to provide ratiometric O2 sensing upon FRET. For instance, a FRET-based O2 sensor was developed by Potzkei et al. [21][30] via a fusion protein between a GFP variant, the yellow FP (YFP), which requires O2 for maturation of the chromophore, and the hypoxia-tolerant flavin-binding FP (FbFP) (Figure 2d).
Very recently, O2-dependent maturation of FPs (the orange large Stokes shift protein CyOFP1 and mOrange2) fused with the hypoxia-tolerant fluorescent protein UnaG was investigated for the development of ratiometric and FRET-fluorescent lifetime imaging microscopy (FLIM) reporters, respectively (Figure 2e,f) [22][23][32,33]. Different from most GFPs and red fluorescent proteins (RFPs), the GFP UnaG from Japanese eel [24][34] does not acquire fluorescence upon a relatively slow (lifetime from minutes to hours) O2-dependent maturation of a chromogenic triad [25][35] but upon O2-independent binding of a fluorogenic cofactor, the porphyrin metabolite bilirubin.
2.3. O
2
Sensors Based on O
2
-Binding Heme Proteins
Oxygen sensing has been also approached by exploiting heme-binding proteins, which use the iron-containing heme group as the O2 sensing element to be transduced in other cellular signals. These constructs are commonly composed of an N-terminal heme-containing globin domain, whose conformational variation upon O2 binding activates a functional domain that triggers a catalytic domain. This latter can have a diguanylate cyclase (DGC) [26][37] or phosphodiesterase (PDE) [27][38] activity toward cyclic diGMP (c-diGMP) and histidine kinase (HK) activity [28][39]. An example is a genetically encoded O2 biosensor based on the direct O2 sensor DosP from E. coli. This protein contains a heme-binding globin domain called DosH (Figure 1b) and a PDE catalytic domain, which converts cyclic-di-GMP to linear-di-GMP [29][40]. However, based on the crystallographic structures [10][30][18,41], the conformational variation of DosP upon O2 binding or dissociation is not sufficient to develop a suitable FRET signal; to overcome this limitation, DosH has been associated with a fluorescent protein to exploit the heme spectroscopic changes in Soret and Q peaks upon O2 binding shifting from 425 and 560 nm to 414 and 580 nm, respectively [31][42]. The yellow FP Venus—a GFP variant with a suitable spectral overlap with DosH—was conjugated with DosH using an optimized antiparallel coiled-coil linker. When Venus was excited at 500 nm, its fluorescence emission at 527 nm was absorbed by DosH in the O2-free form, resulting in low Venus quantum yield; when DosH was O2-bound, Venus’s fluorescence intensity increased. The change in heme absorption was therefore amplified by a change in fluorescence intensity. This sensor, called ANA-Y (anaerobic/aerobic sensor yellow), was used to sense O2 in the micromolar range (Figure 2h) [32][43]. The heme nitric oxide/oxygen-binding protein (H-NOX) of the thermophilic bacterium Caldanaerobacter subterraneus has been exploited to develop an O2 sensor as a robust protein scaffold also able to bind unnatural heme cofactors. The natural heme was replaced with a Pd(II) or Pt(II) porphyrin, a phosphorescent cofactor (650–800 nm range) that, in the triplet excited state, can interact with molecular oxygen; in particular, Pd(II) porphyrins are more sensitive to low O2 levels and Pd(II) has a larger range of O2 sensitivity. A ratiometric sensor based on H-NOX incorporating Pd(II) or Pt(II) porphyrins was developed by conjugating to the protein an Alexa fluorescent dye that is the O2-independent FRET donor to the porphyrin. The selective excitation of Alexa dye guarantees that the porphyrin emission will only be from FRET, thus minimizing the background signal (Figure 2i) [33][44]. As a different approach, the spectroscopic changes of heme upon O2 binding have been exploited and amplified in a fusion protein (Myo-mCherry), where myoglobin has been fused with mCherry fluorescent protein; spectral changes resulted in a change of FRET, captured as a change in fluorescence lifetime within cells by FLIM (Figure 2l).2.4. O
2
Sensors Based on O
2
-Binding Copper Proteins
The use of type 3 copper proteins to sense O2 has been also approached; these proteins have a binding site with two Cu(I) ions that specifically bind to O2, and the complex, once formed, gives absorption peaks at 340 and 570 nm [34][47]. O2 binding typically also causes Trp fluorescence changes, but the use of fluorescence coming from an exogenous fluorophore with a more efficient and higher-wavelength emission is made necessary to increase sensitivity (Trp has a moderate quantum yield, around 0.14) and to allow the detection in tissues showing significant autofluorescence. An example is given by the tyrosinase from the bacterium Streptomyces antibioticus and hemocyanin from the arthropod Carcinus aestuarii, which have been conjugated at the N-terminus with different fluorescent dyes (Alexa 350, Atto390, Cy3, Cy5, Atto655), showing an absorption overlapping Trp emission, thus allowing the detection as a variation in FRET upon O2 binding (Figure 2m) [35][48]. The detection of a single O2 molecule has been obtained by exploiting the O2 carrier hemocyanin from the tarantula Eurypelma californicum. This protein shows spectral differences when it binds oxygen, and these absorption changes can be used to sense O2 presence. Moreover, to amplify the signal, TAMRA fluorescent dye was conjugated with hemocyanin, and its fluorescence quantum yield was shown to decrease by a factor of two when hemocyanin was oxygenated [36][49]. This was caused by a FRET where TAMRA was the donor and the O2 binding site was the acceptor (Figure 2n) [37][50].3. Biosensors for ROS
3.1. ROS Sensors Based on roFPs
Reduction-oxidation sensitive GFPs (roGFPs) are non-natural variants of A. victoria GFP obtained by substituting surface-exposed residues with cysteine residues appropriately distanced to form disulfide bonds at suitable redox potentials (Figure 3a,b and Figure 4a). The first examples of roFPs consisted of the introduction of an artificial pair of cysteine residues on YFP, giving rise to one of the first classes of redox-sensitive protein-based fluorescent sensors. The first rxYFP was conceived in 2001 by Winther’s research group, when the YFP sequence was modified to mutate Asn149 and Ser202 to two cysteine residues able to form a disulfide bridge under oxidizing conditions (Figure 3a) [38][51].Figure 3. Structure of ROS sensing proteins. (a) Crystal structure of rxYFP in the oxidized form, with the reactive cysteine residues 149 and 202 in orange (PDB 1h6r) [38][51]. The fluorophore is represented in yellow. (b) YFP–glutaredoxin fusion protein (PDB ID 2jad) [39][52], with the YFP domain in green and the glutaredoxin domain in gold. The reactive cysteine residues 149 and 202 are shown in orange. The fluorophore is represented in yellow. (c) Crystal structure of the flavoprotein improved LOV (iLOV) domain (PDB ID 4eet) [40][53]. The FAD moiety is represented in sticks. (d) Crystal structure of human peroxiredoxin in the monomeric form (PDB ID 1qmv) [41][54]. The reactive cysteine residues (51 and 172) are represented in sticks.
Figure 4. Schematic representation of selected ROS sensors. (a) General mechanism of roGFP-based biosensors. (b) Mechanism of biosensors based on a fusion protein between a roGFP2 and a redoxin protein. (c) Mechanism of roGFP-Orp1 biosensor. (d) Mechanism of CROST sensor. (e) Mechanism of Grx1-FROG/B sensor. (f) Representation of the circular permutation mechanism in proteins. (g) Schematic representation of HyPer sensor constructs and their general mechanism as ROS sensors; asterisks represent mutations inserted in the original HyPer construct. (h) Mechanism of TrxRFP1 sensor. (i) Mechanism of OHSer sensor. Mechanism of pHaROS (l) and of its variant GRX1-pHaROS (m). (n) Genetically encoded sensor for H2O2 based on the FRET couple Cerulean Δ11 and Cp173 Venus separated by the two CRDs of Yap1 (oxyFRET). (o) Genetically encoded sensor for H2O2 based on the FRET couple Cerulean Δ11 and Cp173 Venus separated by Orp1 linked with the C-terminal CRD of Yap1 (perFRET). (p) Mechanism of perClover-Prx2-mRuby2 biosensor.
3.2. ROS Sensors Based on roGFP Fusion Proteins
Studies on rxYFP expressed in yeast strains demonstrated that its oxidation is mediated by the sulfhydryl disulfide oxidoreductase glutaredoxins (Grxs) [42][55]. Based on this observation, Winther and colleagues exploited the catalytic activity of the yeast glutaredoxin 1 Grxp1 on rxYFP to design a fusion construct combining the two proteins (Figure 3b) [39][43][52,83]. rxYFP-Grx1p is glutathione-specific and does not rely on host Grxs when expressed in vivo, improving the sensor dynamic features and reaction rates and broadening its possible applications. Due to its redox potential of around −260 mV, rxYFP can find application for redox measurements in reducing cellular compartments but is not suited for reliable tests in secretory compartments [42][55].
Therefore, other fusion variants of roFPs have been developed. Particularly, Grx1-roGFP2 is a fusion protein of roGFP2 with Grx1 (Figure 4b). In comparison with roGFP, the equilibration rate with glutathione was higher by three orders of magnitude [44][45][84,85]. Grx1-roGFP2 allowed live imaging of the glutathione redox potential (EGSH) in several cell subcompartments, allowing for the detection of nanomolar changes in oxidized glutathione within seconds to minutes [45][46][85,86]. Grx1-roGFP2 was also used to probe thiol and redox metabolism in Plasmodium falciparum [47][87]. Several organelle- and cytoskeleton-targeted Grx1-roGFPs have been validated [48][88]. Fusion of roGFP with the NADPH oxidases (Nox) organizer protein p47 phox, a redox-sensitive protein that specifically reacts with Nox, was developed [49][50][89,90]. roGFP2-Orp1 is a fusion protein of roGFP2 with the S. cerevisiae protein oxidant receptor 1 (Orp1), a thiol-peroxidase that controls cellular H2O2 homeostasis by activating the transcription factor Yap1 by oxidation. The fusion protein with roGFP2 was specifically developed for the detection of H2O2 (Figure 4c) [44][84].
More roGFP-based fusion proteins have been developed for the detection of species-specific redox compounds (Figure 4b). Mrx1-roGFP2 is a fusion protein of roGFP2 and mycoredoxin-1 (Mrx-1), an oxidoreductase dependent on mycothiol, an unusual thiol compound found in the actinobacteria that functions as a redox buffer [51][94] (Figure 4c). Brx-roGFP2 is a fusion protein of roGFP2 and bacilliredoxin (Brx), designed for the measurement of the redox potential for bacillithiol, which works in Staphylococcus aureus cells as a glutathione surrogate [52][53][95,96]. Tpx-roGFP2 is a fusion protein of roGFP2 with tryparedoxin (Tpx), a Prxs, for the detection of trypanothione in Trypanosoma [54][64].
Sugiura and colleagues set up a series of sensors based on cyan FP (CFP), Sirius, mTurquois, and Venus mutants [55][56][97,98]. By inserting Cys residue couples inspired to the roGFP structure, the authors obtained two series of redox-sensitive proteins called Oba-Q (oxidation balance-sensed quenching) and Re-Q (reduction-sensed quenching), having different midpoint redox potentials applicable in different cellular environments; they generally were not pH-sensitive in physiological conditions and the different emission wavelengths consented their simultaneous use. As a drawback, however, the signal of Oba-Q and Re-Q proteins was non-ratiometric and their fluorescence intensity was proportional to their expression levels. Taking advantage of the expertise on mTurquois and Venus mutants, the same research group also conceived a FRET-based biosensor, where the two FTs were separated by the N-terminus of the Trx-targeted protein CP12 from A. thaliana or the cyanobacterium Anabaena sp. PCC7120 (Figure 4d) [56][98]. These sensors were called CROST (Change in RedOx State of Trx) 1 and 2, respectively, and could discriminate the Trx redox state directly.
Another type of sensor was then developed to monitor in living cells the glutathione redox state fluctuations through Grx activity. Indeed, Grx-FROG/B, composed of the human Grx1 domain connected by a linker to a mutant GFP, exploits a mechanism called excitation state intramolecular proton transfer (ESIPT) that gives green fluorescence in oxidizing conditions and blue emission in reducing ones, with a midpoint redox potential of −293 mV and a good response at physiological pH values (Figure 4e). The green/blue output of Grx-FROG/B can be registered at the same time by a multiplex detector [57][99].
